Datasets:

kilian-group
/

arxiv-classifier

Dataset card Data Studio Files Files and versions Community

Subset (4)

default · ~99.6k rows (showing the first 120k)

Split (2)

train · ~72.4k rows (showing the first 93.1k)

paper_id stringlengths 9 16	version stringclasses 26 values	yymm stringclasses 311 values	created timestamp[s]	title stringlengths 6 335	secondary_subfield sequencelengths 1 8	abstract stringlengths 25 3.93k	primary_subfield stringclasses 124 values	field stringclasses 20 values	fulltext stringlengths 0 2.84M
1307.4753	1	1307	2013-07-17T20:00:00	Examining the broadband emission spectrum of WASP-19b: A new z band eclipse detection	[ "astro-ph.EP" ]	WASP-19b is one of the most irradiated hot-Jupiters known. Its secondary eclipse is the deepest of all transiting planets, and has been measured in multiple optical and infrared bands. We obtained a z band eclipse observation, with measured depth of 0.080 +/- 0.029 %, using the 2m Faulkes Telescope South, that is consistent with the results of previous observations. We combine our measurement of the z band eclipse with previous observations to explore atmosphere models of WASP-19b that are consistent with the its broadband spectrum. We use the VSTAR radiative transfer code to examine the effect of varying pressure-temperature profiles and C/O abundance ratios on the emission spectrum of the planet. We find models with super-solar carbon enrichment best match the observations, consistent with previous model retrieval studies. We also include upper atmosphere haze as another dimension in the interpretation of exoplanet emission spectra, and find that particles <0.5 micron in size are unlikely to be present in WASP-19b.	astro-ph.EP	astro-ph	Draft version September 9, 2018 Preprint typeset using LATEX style emulateapj v. 5/2/11 3 1 0 2 l u J 7 1 . ] P E h p - o r t s a [ 1 v 3 5 7 4 . 7 0 3 1 : v i X r a EXAMINING THE BROADBAND EMISSION SPECTRUM OF WASP-19B: A NEW Z BAND ECLIPSE DETECTION George Zhou1, Lucyna Kedziora-Chudczer2, Daniel D.R. Bayliss1, Jeremy Bailey2 Draft version September 9, 2018 ABSTRACT WASP-19b is one of the most irradiated hot-Jupiters known. Its secondary eclipse is the deepest of all transiting planets, and has been measured in multiple optical and infrared bands. We obtained a z band eclipse observation, with measured depth of 0.080 ± 0.029 %, using the 2 m Faulkes Telescope South, that is consistent with the results of previous observations. We combine our measurement of the z band eclipse with previous observations to explore atmosphere models of WASP-19b that are consistent with the its broadband spectrum. We use the VSTAR radiative transfer code to examine the effect of varying pressure -- temperature profiles and C/O abundance ratios on the emission spectrum of the planet. We find models with super-solar carbon enrichment best match the observations, consistent with previous model retrieval studies. We also include upper atmosphere haze as another dimension in the interpretation of exoplanet emission spectra, and find that particles < 0.5µm in size are unlikely to be present in WASP-19b. Subject headings: planets and satellites: atmospheres -- planets and satellites: individual (WASP-19b) 1. INTRODUCTION Recent observations of transiting planet systems have lead to the first in-depth characterisation of exoplanet atmospheres. Observations of the secondary eclipse event, when the planet is blocked by the host star, is the predominant method of measuring the emergent flux of close-in exoplanets. In particular, secondary eclipses observed at multiple wavelength bands have provided first spectral energy distribution of exoplan- ets (Charbonneau et al. 2005, 2008). Similar observa- tions have revealed the presence of molecular absorption features (e.g. Grillmair et al. 2008; Swain et al. 2009) in the emission spectra of hot-Jupiters, and hinted at the diversity of chemical compositions across exoplanet at- mospheres (e.g. Barman 2008; Madhusudhan & Seager 2009). WASP-19b (Hebb et al. 2010) is a 1.17 MJup, 1.39 RJup exoplanet in a 0.79 day prograde orbit (Hellier et al. 2011) which transits a Vmag = 12.3 G dwarf. The equi- librium temperature for the planet is at least 2000 K, making it one of the hottest hot-Jupiters known, and the most favourable target for eclipse observations. The proximity of WASP-19b to the host star also makes it an interesting case study of irradiated atmospheres. In par- ticular, eclipse observations have shown that WASP-19b is inconsistent with the hypothesis that highly irradiated planets exhibit thermal inversion features (Hubeny et al. 2003; Burrows et al. 2007; Fortney et al. 2008), although some exceptions are known (e.g. HD189733b, TrES- 3b, and XO-1b, Charbonneau et al. 2008; Fressin et al. 2010; Machalek et al. 2008). Madhusudhan (2012) pro- posed WASP-19b as a planet hosting a carbon rich at- mosphere, depleted in TiO, which is a primary absorber for inversion layers. In addition, the Carbon-to-Oxygen 1 Research School of Astronomy and Astrophysics, Australian National University, Cotter Rd, Weston Creek, ACT 2611, Aus- tralia; [email protected] 2 School of Physics, University of New South Wales, Sydney, NSW 2052, Australia ratio (C/O) is a potential indicator for the location in the proto-planetary disk where these hot-Jupiters originated (e.g. Lodders 2004; Oberg et al. 2011). The C/O enrichment hypothesis is based upon existing multi-band eclipse observations of WASP-19b, including the ASTEP400 broadband centred at 0.67 µm (Abe et al. 2013), z band (Burton et al. 2012; Lendl et al. 2012), 1.190 µm narrow band (Lendl et al. 2012), H band (Anderson et al. 2010), K band (Gibson et al. 2010), Spitzer 3.6, 4.5, 5.8, and 8.0 µm bands (Anderson et al. 2011), as well as spectrophotometric observations at 1.25 − 2.35 µm by Bean et al. (2013). However, it is dif- ficult to produce a single model that can fit all the mea- surements within their uncertainty constraints. Ground based observations at the 0.1% level remain difficult, and are affected by a range of systematic effects, such as at- mospheric variations, unstable telescope tracking, and detector defects. Independent confirmation observations are required to strengthen the reliability of individual measurements. The depth of the z band eclipse is particularly im- portant in determining the C/O ratio of WASP-19b, a deeper eclipse is indicative of an atmosphere defi- cient in TiO absorption and enriched in C/O abun- dance. New Technology Telescope ULTRACAM obser- vations by Burton et al. (2012) reported an eclipse depth of 0.088 ± 0.019 %, whilst a combined set of observa- tions with EulerCam and TRAPPIST over ten epochs by Lendl et al. (2012) reported a shallower eclipse depth of 0.035± 0.012 %. Whilst these observations are consis- tent at the ∼ 2σ regime, the difference between the two measurements makes it difficult to constrain the atmo- sphere models of WASP-19b. In this study, we present an independent observa- tion of a WASP-19b eclipse event using Faulkes Tele- scope South aimed at confirming its z band secondary eclipse depth. We present a careful treatment of the photometry to achieve near photon-limited lightcurves. To investigate the previous claim of a carbon rich at- 2 Zhou et al. mosphere, we use the VSTAR radiative transfer code (Bailey & Kedziora-Chudczer 2012) and the ensemble of observations to model the atmosphere of WASP- 19b and examine the effects of temperature-pressure profiles and C/O abundance on its emergent spec- trum. One draw back of existing model retrieval studies (Madhusudhan & Seager 2009) is its lack of treatment for non-isotropic scattering. The lack of absorption features in the transmission spectrum in HD189733b (Pont et al. 2008), as well as the weaker than expected detections of sodium in various hot- Jupiters (e.g. Charbonneau et al. 2002; Fortney et al. 2003; Zhou & Bayliss 2012), all point to the importance of clouds and haze in modelling planetary atmospheres. We also exploit the rigorous treatment of Rayleigh scat- tering by VSTAR to investigate the effect of upper at- mosphere haze on the emission spectrum of WASP-19b. 2. DETECTION OF Z BAND ECLIPSE 2.1. Observations We monitored an eclipse of WASP-19b using the 2 m Faulkes Telescope South (FTS), located at Siding Spring Observatory, Australia, on 2012 December 29, from 12:03 -- 15:50 UT, with the expected eclipse occurring dur- ing 13:25 -- 15:02 UT. Observations were performed in the Pan-STARRS z-band, centred at 0.866 µm (Tonry et al. 2012), using the Merope 2K× 2K camera, with 4.7′× 4.7′ field of view, unbinned pixel size of 0.139" pixel−1, read out with 2× 2 bins. 161× 60 s exposures were taken. The seeing on the night previous to the observing sequence was ∼ 1". The telescope was slightly defocused to avoid saturation and to reduce the effect of intra- and residual inter-pixel variations, resulting in point spread functions (PSF) with full width half maximum (FWHM) of ∼ 2". Bias subtraction and flat field corrections were performed using the CCDPROC package in IRAF 3 with the most recent archival calibration frames. An example image is shown in Figure 1. 2.2. Analysis 2.2.1. PSF Variations and Adaptive Aperture Photometry Upon close examination of the images, we find the stel- lar PSF is asymmetric across the image. The distortion and elongation of the PSF is a result of the defocus- ing applied. In addition, the position angle of the elon- gated PSF changes with the rotation of the telescope (Figure 2), as FTS is on an alt-az mount. To further in- vestigate the PSF variations, we create a template PSF from a single exposure taken mid-run, and fit it to the remaining exposures, allowing for rotation and spatial di- lation. We find no significant deviations in the fit resid- uals, with the exception of the initial images taken at high airmass, suggesting that the general shape of the PSF remained constant throughout the night. We performed elliptical aperture photometry on the re- duced images. Compared to conventional circular aper- tures, variable elliptical apertures best account for all of the stellar flux whilst minimising background noise. The 3 IRAF is distributed by the National Optical Astronomy Ob- servatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation Figure 1. FTS z band image of the WASP-19 field. The target, located in the centre of the field, is circled in red; the chosen set of reference stars are circled in white. The size of the circle indicates the size of the background aperture. The column of dead pixels in the top left of the image was masked out for the photometry. Figure 2. PSF variations of the target star over course of the observations, shown at the start (left), middle (centre), and end of the night (right). The solid white lines mark the 0.95, 0.50, and 0.10 peak height contours. The dashed black ellipse marks the photometry aperture used. The crops are 20 pixels in size. ellipse parameters, semi-major axis A, semi-minor axis B, and position angle θ, were measured using Source Extractor (Bertin & Arnouts 1996), and are plotted in Figure 3, along with other relevant global parameters. A and B are the maximum and minimum root-mean-square (RMS) of the spatial profile. The size of the aperture, R, is a scaling factor that maintains the shape and orienta- tion of the ellipse, and is related to the ellipse parameters CXX, CY Y , and CXY by, R2 =CXX(x − ¯x)2 + CY Y (y − ¯y)2 (1) + CXY (x − ¯x)(y − ¯y) cos2 θ A2 + sin2 θ A2 + sin2 θ B2 cos2 θ B2 CXX = CY Y = CXY =2 cos θ sin θ(cid:18) 1 A2 − 1 B2(cid:19) . The lowest out-of-eclipse scatter was achieved using aper- ture sizes of R = 4.2, enclosing ∼ 99% of the flux. The adopted elliptical aperture parameters A, B, θ were de- termined from linear fits in time to the averaged mea- surements from Sourced Extractor. Higher order fits to Broadband emission spectrum of WASP-19b 3 the ellipse parameters were tested, and did not result in significantly different lightcurves or eclipse depths. Ex- posures with HJD < 2456291.03 were discarded from the analysis, since they were taken at high airmass, when the PSF shape varied rapidly. The background was estimated using a 100 pixel di- ameter outer aperture on a background image with all detected sources masked out. Since WASP-19 resides in a relatively crowded field, masking out field stars is es- sential to achieving optimal background subtraction. We note that Burton et al. (2012) followed a similar tech- nique in their analysis. The background count around WASP-19 is plotted in Figure 3 Differential photometry was performed using five ref- erence stars (labelled in Fig. 1), chosen for their lack of nearby neighbours, similar colour indices to the target, and the eventual stability of the lightcurves. A master reference lightcurve (M ) was created by averaging the ensemble of reference stars (Ri), each with errors ∆Ri: M = Xi ci ∆Ri Ri , (2) where weights ci were chosen to minimise the RMS scat- ter of the corrected object lightcurve. The use of weights to minimise the object lightcurve scatter is similar to applying the Trend Filtering Algorithm to the out-of- transit dataset (Kov´acs et al. 2005). To remove uncorre- lated trends in the individual reference star lightcurves, we divided each reference star by a master reference lightcurve made of all other reference stars. Any slow varying residual trends in that reference star were then corrected for by a linear fit. individual outlier points significantly different from other reference stars were removed by sigma clipping. Finally, a lin- ear trend was removed from the target lightcurve by fitting for the out-of-eclipse points. We note that the target lightcurve was treated by the same processes as the reference lightcurves. The ensemble of raw reference lightcurves, as well as the raw target lightcurve, are plot- ted in Figure 3. In addition, 2.2.2. Eclipse model fitting We fit a Mandel & Agol (2002) eclipse model to the FTS lightcurve via a downhill simplex minimisation of the χ2 of fit, followed by a Markov chain Monte Carlo (MCMC) ensemble sampler (emcee implementa- tion, Foreman-Mackey et al. 2012) to determine the un- certainties. For the MCMC routine, we artificially in- flate the photometric uncertainties such that the reduced χ2 = 1. This accounts for the contribution of other systematic effects in addition to the photon-noise un- certainty. The free parameters of the fit are the tran- sit centre t0, depth D, normalised planet orbital radius a/R⋆, and the impact parameter b. The parameters t0, a/R⋆, b are constrained by Gaussian priors based on the joint analysis performed by Anderson et al. (2011). The system period and planet-star radius ratio are fixed to Anderson et al. (2011) values. The fitted lightcurve is shown in Figure 4. The final eclipse depth is 0.080 ± 0.029 %. The is corresponding 2680+140 −180 K. A MARCS model atmosphere spectrum (Gustafsson et al. 2008) was adopted for the host star z band brightness temperature Reduced χ2 and eclipse depth after Table 1 decorrelation External Parameter Reduced χ2 D % None X, Y A A/B θ Airmass Background 3.28 3.25 3.30 3.29 3.28 3.31 3.30 0.080 0.087 0.078 0.078 0.074 0.080 0.071 in the brightness temperature calculation for the planet. The derived brightness temperature agrees well with the ASTEP, 1.6 and 2.09 µm temperatures (Abe et al. 2013; Anderson et al. 2011). The depth can also be derived separately by binning the in- and out-of-eclipse points. In Figure 4, we bin the points according to the predicted ephemeris. The eclipse depth, given by the difference in the sigma clipped mean of the two bins, is 0.083 ± 0.026 %, with the uncertainty taken as the error in the mean of the two bins, added in quadrature. This agrees with the transit depth measured by the model fit. 2.2.3. Correlation to external parameters Most high-precision transit and eclipse photometry to date have been processed with some form of external parameter decorrelation to remove residual systematic trends. This is often done by multiplying the lightcurve with a linear combination of external parameters, such as airmass, position, FWHM (e.g. L´opez-Morales et al. 2010; Burton et al. 2012); occasionally, higher order terms have also been employed (e.g. up to 4th order Lendl et al. 2012). We test for the effectiveness of detrending by simulta- neously fitting for the eclipse and a combination of ex- ternal terms involving X, Y position, semi-major axis A, ellipticity A/B, airmass, and background counts, whilst holding t0 constant. In each case, the removal of a linear trend is also allowed. Analysis is performed over the en- tire lightcurve, since the out-of-eclipse points constitute less than half of the observations, and cannot sufficiently represent the entire dataset. Table 1 shows the reduced χ2 after each minimisation routine. No significant im- provements to the χ2 was achieved from any decorrela- tions. The transit depth also remained roughly indepen- dent of these external parameters. We can also check for time correlated noise in the resid- uals using the β factor diagnostic (Winn et al. 2008). For residuals binned into M bins, with N points per bin, the scatter σN as a function of the noise of the unbinned data σ1 is σN = β . (3) σ1 √Nr M M − 1 For uncorrelated data, β = 1. Our residuals have an average of β = 1.15, suggesting minimal time correlated trends in the residuals. Figure 5 shows the RMS of the residuals as a function of bin widths. The lack of need for any decorrelation can be primarily attributed to the use of variable, elliptical, apertures. 4 Zhou et al. Figure 3. Variations in the target X, Y position, PSF semi-major and minor axis (A, B), ellipticity (A/B), ellipse position angle (θ), airmass, background counts, and normalised raw target (red) and ensemble reference (black, arbitrarily offset by 0.05) fluxes are plotted. Linear fits to the ellipse parameters, used to define the elliptical photometry apertures, are plotted in red. Broadband emission spectrum of WASP-19b 5 sively tested and applied to objects ranging from solar system planets (Chamberlain et al. 2013; Cotton et al. 2012; Kedziora-Chudczer & Bailey 2011) to M-dwarfs (Bailey & Kedziora-Chudczer 2012). It is impossible to obtain a unique model that can best fit the currently available broadband data that only sparsely covers the optical and infrared spectrum. Instead we focus on a discussion of effects observed in a spectrum by changing specific conditions in the planetary atmosphere. Highly irradiated planets, like WASP-19b, have been hypothesised to show thermal inversion in its atmo- spheric profile due to condensation of VO and TiO within a cold trap (Fortney et al. 2008). However the Spitzer IRAC data (Anderson et al. 2011) and near infrared ground measurements at 1.6 and 2.1 µm (Anderson et al. 2010; Gibson et al. 2010) appear to be inconsistent with thermal inversion in the planet's atmosphere. Vari- ous explanations have been proposed for the lack of thermal inversion in some highly irradiated planets, such as the dependency on the presence of a cold trap (Showman et al. 2009; Spiegel et al. 2009), destruction of absorbers by stellar activity (Knutson et al. 2010), dis- equilibrium photochemistry (Zahnle et al. 2009), or the enrichment of C/O that leads to depleted TiO abundance (Madhusudhan 2012). In our modelling we use four different atmospheric pressure -- temperature (P-T) profiles without inversion (Figure 6a): (I) the red profile in Fig.12 of Madhusudhan (2012); (II) a 'hotter' profile, which corresponds to condi- tions discussed by Anderson et al. (2011); (III) a 'cooler' profile which reflects range of temperatures and pres- sures assumed by Bean et al. (2013) to explain their near infrared data; (IV) a 'narrow' P-T profile, with reduced range of temperatures that overlap with the Madhusudhan (2012) model over the range of 0.05-0.5 bar. Our models assume a plane-parallel, stratified atmo- sphere with 25 layers characterised by temperature, pres- sure and mixing ratios of the following molecular and atomic species: H2O, CO, CH4, CO2, C2H2, HCN, TiO, VO, Na, K, H2, He, Rb, Cs, CaH, CrH, MgH and FeH. Mixing ratios of these opacity sources are calculated in chemical equilibrium. Atmospheres of hot-Jupiters like WASP-19b are most likely dominated by H2, which is a source of the H2-H2 and H2-He collisionally induced absorption (CIA) that we included with opacities calcu- lated by Borysow & Borysow (1998). We also considered Rayleigh scattering by H2, He, H in the atmosphere of the planet, and free-free and bound-free absorption from H, H− and H2 −. Our spectral line absorption database is described in detail in Bailey & Kedziora-Chudczer (2012). Table 2 lists the references to the sources of spec- tral lines for absorbers used in our models. A spectrum of the WASP-19, G8V type star was obtained from the STScI stellar atmosphere models by Castelli & Kurucz (2004). Figure 6b shows the model spectra for the four P- T profiles considered above. All models presented in this figure have carbon to oxygen ratio C/O=1.1. Thus they can be easily compared with the red spectrum in Fig.12 of Madhusudhan (2012). The differences between our model and the Madhusudhan (2012) model with the same P-T profile can be attributed to use of line databases which may have varied levels of completeness. Figure 4. Top: The eclipse lightcurve of WASP-19b, with the best fitting model plotted in red. Data binned at 0.015 days are plotted as large red points for clarity. Bottom: Histogram showing the distribution of flux measurements in- (solid) and out-of-eclipse (dashed), with the centroids of the distributions marked by the corresponding vertical lines. Figure 5. The RMS of the residuals as a function of bin width are plotted in red. The dashed line shows the 1/√N drop off expected for an uncorrelated signal. 3. VSTAR ATMOSPHERE MODEL We use the VSTAR line-by-line radiative transfer code (Bailey & Kedziora-Chudczer 2012) to derive a model atmosphere of WASP-19b that fits our measure- ment and the data published previously. VSTAR is a comprehensive atmospheric radiative transfer model incorporating a chemical equilibrium model, an exten- sive database of molecular spectral lines and a full treatment of multiple-scattering radiative transfer us- ing the discrete ordinate method. It has been exten- 6 Zhou et al. List of molecular and atomic absorbers used in the VSTAR modelling with references to the line databases. Table 2 Line Absorbers Reference CH4 CO2 H2O CO HCN C2H2 CaH MgH FeH 2.2.6 in see Bailey & Kedziora-Chudczer (2012) Pollack et al. (1993) Barber et al. (2006) Goorvitch (1994) Harris et al. (2006) Rothman et al. (2009) Weck et al. (2003b) Weck et al. (2003) Dulick et al. (2003); Hargreaves et al. (2010) Burrows et al. (2002) Plez (1998) Plez, B., private communication Skory et al. (2003a); CrH TiO VO K, Na, Rb, Cs Piskunov et al. (1995); Kupka et al. (1999) The model with 'hotter' P-T profile tends to fit bet- ter near infrared (NIR) data from Anderson et al. (2010) and Gibson et al. (2010), while 'cooler' P-T profile pro- duces a spectrum that matches closer to the data ob- tained by Bean et al. (2013). However both these pro- files either overestimate or underestimate absorption ob- served in the Spitzer data between 3.6 and 8 µm. We also found very little difference between the 'narrow' P- T profile and the one from Madhusudhan (2012), with only slightly increased absorption between 1 and 2µm in the 'narrow' P-T profile. In Figure 7 we show the effect of varying C/O ratio on spectra using example of a model with our hottest P-T profile, although qualitative results are the same for other profiles considered here. The spectrum ob- tained for C/O=0.5 is dominated by the oxygen bear- ing molecules with strong H2O bands visible in NIR and far IR, and CO2 and CO bands around 4.5 µm. Around C/O=1 the abundances of carbon and oxygen containing molecules change dramatically by many orders of mag- nitudes. This explains the rapid decrease of H2O ab- sorptions in the spectra when C/O ratio varies between 0.9 and 1.1 in right panel of Figure 7, while only mod- est changes are visible in left panel of Figure 7 between C/O=1.1 and C/O=4.0. Spectra of atmospheres with high content of carbon are dominated by CH4 absorption in addition of molecules such as HCN and C2H2 consid- ered also in Madhusudhan (2012). While strong water absorption bands are absent, the CO features around 2.3 and 4.8 µm become more prominent. Currently available spectral measurements for WASP-19b seem to be more consistent with the atmosphere models which are derived with the C/O ratio higher than solar. Models presented so far in Figure 6 assumed a clear at- mosphere. However recently published data for the hot- Jupiter HD189733b (Pont et al. 2008) indicate a pres- ence of haze in the top layers of its atmosphere. Compo- sition of hazes depends on the abundance and refractory properties of different compounds (Burrows & Sharp 1999). In hot-Jupiters and brown dwarfs suggested con- densates may be formed by highly refractory species such as perovskite (CaTiO3) and corundum (Al2O3), which condense in temperatures close to 1600 K. More abun- dant Si, Mg and Fe elements combine into compounds such as enstatite (MgSiO3) and forsterite (Mg2SiO4) that condense in lower temperatures. Even at the rel- atively lower temperatures at the top of the atmosphere of WASP-19b it is not clear which species could poten- tially exists in a form of haze. In Figure 8 we assume that such a haze exists and it is composed of unknown particulate with refractive index similar to enstatite. Four examples of the model spec- trum are shown for WASP-19b, where the optical depth of a cloud in the top layer of the atmosphere is varied. The particles with a mean size of 0.5 µm are assumed in the left panel. On the right all models are derived with the varied mean size of particles, while the same optical depth τ = 1 at 1.5 µm is assumed. The absorp- tion and scattering properties as a function of wavelength are calculated using Mie theory. At wavelengths compa- rable to the size of cloud particles, scattering processes operate efficiently, which leads to increase of planetary albedo in the corresponding part of its spectrum. On the other hand the added opacity in top layers obscures thermal emission from the planet, which has an effect of lowering the received flux in infrared part of a spec- trum. Differences in particle sizes affect both scattering and absorption properties of the haze. Particles smaller than 0.5 µm appear to generate highly reflective haze at visible wavelengths, which may not be consistent with the measurement from Abe et al. (2013). Observations of secondary eclipses at different wave- lengths are sensitive to different properties of the plane- tary atmosphere. Observations of the flux in z-band can provide a sensitive probe of the C/O ratio in the atmo- sphere of the planet, as shown in Figure 7. In optical spectrum strongly absorbing bands of VO and TiO dom- inate the measured flux in temperatures above 1700 K when the C/O ratio is similar to solar. After VO and TiO start to condense below this temperature, absorp- tion from alkali lines and water bands takes over. How- ever if C/O is higher than solar, alkali lines will be dis- tinct even at lower temperatures due to reduced abun- dance of VO and TiO. On the other hand, in the presence of stratospheric haze Rayleigh scattering may dominate optical spectrum almost entirely as seen in Figure 8. A few more strategically placed photometric data points in optical and near infrared spectrum will help to discrim- inate between these broad conditions of the WASP-19b atmosphere. However more detailed and unique mod- els can only be derived when the amount of photometric data becomes sufficient to break degeneracies in inter- pretation of current spectral features. This is currently rather remote prospect as discussed in Line et al. (2013). 4. DISCUSSIONS We presented an examination of the emission spec- trum of WASP-19b measured in eclipse. Using FTS observations, we measured the z band eclipse depth to be 0.080 ± 0.029 %. This result is in excellent agree- ment with the depth measured by Burton et al. (2012) of 0.088± 0.019 %, and also consistent with the tentative detection of a significant eclipse in the optical ASTEP band by Abe et al. (2013), as well as deep NIR detec- tions by Anderson et al. (2010); Gibson et al. (2010). It Broadband emission spectrum of WASP-19b 7 Figure 6. Left: Four P-T atmosphere profiles described in more detail in Section 3. Right: The WASP-19b VSTAR models corresponding to the P-T profiles shown on the left. C/O ratio of 1.1 was assumed in all four models and molecular absorption due to H2O, CO, CH4, CO2, C2H2, HCN, TiO, VO. The yellow hexigon denotes the data point from the FTS observation reported in this paper, red crosses show data from Bean et al. (2013), while red hexagons mark results of all other observations described in Section 1. Figure 7. Example of WASP-19b atmosphere models with for the 'hotter' P-T profile and changing ratio of C/O. Left: The model with close to solar ratio C/O=0.5 seems to be especially inconsistent with the data from optical observations. Models with C/O ratios above 1 fit better all data except the NIR observations of Bean et al. (2013). Right: Strong changes in water absorption bands across infrared range of the spectrum around the region of C/O=1 are shown, where abundances of oxygen and carbon bearing species vary by orders of magnitude. 8 Zhou et al. is also in 2σ agreement with the measurement made us- ing multiple eclipses from the 1.2 m Euler-Swiss telescope and the 0.6 m TRAPPIST telescope of 0.035 ± 0.012 % (Lendl et al. 2012). From the non-exhaustive set of VSTAR spectra, we find no single model that can fit all of the reported obser- vations. However, when the spectrophotometry measure- ments by Bean et al. (2013) are discarded, the C/O en- riched models present a good fit to the remaining points. The Bean et al. (2013) points are also inconsistent with available photometric J, H, K measurements at the same wavelengths. We also note that our new z band detection is consistent in brightness temperature with the photo- metric near-infrared detections, not the spectrophotom- etry measurement. These difficulties highlight the chal- lenges of transit spectrophotometry observations, espe- cially when WASP-19 is the faintest object targeted with the technique to date. Line et al. (2013) assessed the difficulty of interpreting broadband emission spectra via retrieval techniques, and noted that C/O classifications tend towards a bimodal posterior of 0.5 or 1. This agrees with our assessment that large-scale changes in the spectrum are only appar- ent from C/O of 0.9 to 1.1. Although a quantitative estimate of carbon enrichment in these atmospheres is unlikely, WASP-19b is still more consistent with a super- solar C/O composition. In addition to WASP-19b, Madhusudhan (2012) pointed to XO-1b, CoRoT-2b, WASP-33b, and WASP- 12b as carbon rich candidates. XO-1b is a significantly less irradiated planet that has only been studied in the Spitzer bands (Machalek et al. 2008). WASP-33 is a rapidly rotating F-dwarf that exhibits photometric vari- ability on the hour timescale, for which precision pho- tometry results are difficult to interpret (Smith et al. 2011). Light from WASP-12 was found to be contam- inated by a blended M-dwarf, and the compensated eclipse measurements can be modelled without a carbon rich atmosphere (Crossfield et al. 2012). Only CoRoT- 2b has received as thorough an observational evalua- tion as WASP-19b, with measurements available from the CoRoT optical band to the Spitzer bands. No ex- isting analysis has included all available observations to examine the validity of its carbon enriched claim. This paper uses observations obtained with facilities of the Las Cumbres Observatory Global Telescope. The work was, in part, supported by the Australian Research Council through Discovery grant DP110103167. Facilities: FTS(Merope) REFERENCES Abe, L., Gon¸calves, I., Agabi, A., et al. 2013, ArXiv e-prints Anderson, D. R., Gillon, M., Maxted, P. F. L., et al. 2010, A&A, 513, L3 Anderson, D. R., Smith, A. M. S., Madhusudhan, N., et al. 2011, ArXiv e-prints Bailey, J., & Kedziora-Chudczer, L. 2012, MNRAS, 419, 1913 Barber, R. J., Tennyson, J., Harris, G. J., & Tolchenov, R. N. 2006, MNRAS, 368, 1087 Barman, T. S. 2008, ApJ, 676, L61 Bean, J. L., D´esert, J.-M., Seifahrt, A., et al. 2013, ArXiv e-prints Bertin, E., & Arnouts, S. 1996, A&AS, 117, 393 Borysow, A., & Borysow, J. I. 1998, NASA STI/Recon Technical Report N, 99, 63017 Burrows, A., Hubeny, I., Budaj, J., Knutson, H. A., & Charbonneau, D. 2007, ApJ, 668, L171 Burrows, A., Ram, R. S., Bernath, P., Sharp, C. M., & Milsom, J. A. 2002, ApJ, 577, 986 Burrows, A., & Sharp, C. M. 1999, ApJ, 512, 843 Burton, J. R., Watson, C. A., Littlefair, S. P., et al. 2012, ApJS, 201, 36 Castelli, F., & Kurucz, R. L. 2004, ArXiv Astrophysics e-prints Chamberlain, S., Bailey, J., Crisp, D., & Meadows, V. 2013, Icarus, 222, 364 Charbonneau, D., Brown, T. M., Noyes, R. W., & Gilliland, R. L. 2002, ApJ, 568, 377 Charbonneau, D., Knutson, H. A., Barman, T., et al. 2008, ApJ, 686, 1341 Charbonneau, D., Allen, L. E., Megeath, S. T., et al. 2005, ApJ, 626, 523 Cotton, D. V., Bailey, J., Crisp, D., & Meadows, V. S. 2012, Icarus, 217, 570 Crossfield, I. J. M., Barman, T., Hansen, B. M. S., Tanaka, I., & Kodama, T. 2012, ApJ, 760, 140 Dulick, M., Bauschlicher, Jr., C. W., Burrows, A., et al. 2003, ApJ, 594, 651 Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2012, ArXiv e-prints Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, 1419 Fortney, J. J., Sudarsky, D., Hubeny, I., et al. 2003, ApJ, 589, 615 Fressin, F., Knutson, H. A., Charbonneau, D., et al. 2010, ApJ, 711, 374 Gibson, N. P., Aigrain, S., Pollacco, D. L., et al. 2010, MNRAS, 404, L114 Goorvitch, D. 1994, ApJS, 95, 535 Grillmair, C. J., Burrows, A., Charbonneau, D., et al. 2008, Nature, 456, 767 Gustafsson, B., Edvardsson, B., Eriksson, K., et al. 2008, A&A, 486, 951 Hargreaves, R. J., Hinkle, K. H., Bauschlicher, Jr., C. W., et al. 2010, AJ, 140, 919 Harris, G. J., Tennyson, J., Kaminsky, B. M., Pavlenko, Y. V., & Jones, H. R. A. 2006, MNRAS, 367, 400 Hebb, L., Collier-Cameron, A., Triaud, A. H. M. J., et al. 2010, ApJ, 708, 224 Hellier, C., Anderson, D. R., Collier-Cameron, A., et al. 2011, ApJ, 730, L31 Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011 Kedziora-Chudczer, L., & Bailey, J. 2011, MNRAS, 414, 1483 Knutson, H. A., Howard, A. W., & Isaacson, H. 2010, ApJ, 720, 1569 Kov´acs, G., Bakos, G., & Noyes, R. W. 2005, MNRAS, 356, 557 Kupka, F., Piskunov, N., Ryabchikova, T. A., Stempels, H. C., & Weiss, W. W. 1999, A&AS, 138, 119 Lendl, M., Gillon, M., Queloz, D., et al. 2012, ArXiv e-prints Line, M. R., Wolf, A., Zhang, X., et al. 2013, ArXiv e-prints Lodders, K. 2004, ApJ, 611, 587 L´opez-Morales, M., Coughlin, J. L., Sing, D. K., et al. 2010, ApJ, 716, L36 Machalek, P., McCullough, P. R., Burke, C. J., et al. 2008, ApJ, 684, 1427 Madhusudhan, N. 2012, ApJ, 758, 36 Madhusudhan, N., & Seager, S. 2009, ApJ, 707, 24 Mandel, K., & Agol, E. 2002, ApJ, 580, L171 Mishchenko, M. I., Travis, L. D., & Lacis, A. A. 2002, Scattering, absorption, and emission of light by small particles Oberg, K. I., Murray-Clay, R., & Bergin, E. A. 2011, ApJ, 743, L16 Piskunov, N. E., Kupka, F., Ryabchikova, T. A., Weiss, W. W., & Jeffery, C. S. 1995, A&AS, 112, 525 Plez, B. 1998, A&A, 337, 495 Pollack, J. B., Dalton, J. B., Grinspoon, D., et al. 1993, Icarus, 103, 1 Pont, F., Knutson, H., Gilliland, R. L., Moutou, C., & Charbonneau, D. 2008, MNRAS, 385, 109 Rothman, L. S., Gordon, I. E., Barbe, A., et al. 2009, J. Quant. Spec. Radiat. Transf., 110, 533 Showman, A. P., Fortney, J. J., Lian, Y., et al. 2009, ApJ, 699, 564 Broadband emission spectrum of WASP-19b 9 Figure 8. Model of the WASP-19b atmosphere with P-T profile from Madhusudhan (2012) shown in Figure 6 and C/O=1.1. Left: Including clouds of varied opacity (τ ) at the top of atmospheric layer, with a power-law distribution of particles with effective radius of 0.5 µm and effective variance 0.2 µm as defined in Mishchenko et al. (2002). Right: Varying the mean particle size, whilst assuming τ = 1 at 1.5 µm. 10 Zhou et al. Skory, S., Weck, P. F., Stancil, P. C., & Kirby, K. 2003, ApJS, Weck, P. F., Schweitzer, A., Stancil, P. C., Hauschildt, P. H., & 148, 599 Smith, A. M. S., Anderson, D. R., Skillen, I., Collier Cameron, A., & Smalley, B. 2011, MNRAS, 416, 2096 Spiegel, D. S., Silverio, K., & Burrows, A. 2009, ApJ, 699, 1487 Swain, M. R., Vasisht, G., Tinetti, G., et al. 2009, ApJ, 690, L114 Tonry, J. L., Stubbs, C. W., Lykke, K. R., et al. 2012, ApJ, 750, 99 Kirby, K. 2003a, ApJ, 582, 1059 Weck, P. F., Stancil, P. C., & Kirby, K. 2003b, ApJ, 582, 1263 Winn, J. N., Holman, M. J., Torres, G., et al. 2008, ApJ, 683, 1076 Zahnle, K., Marley, M. S., Freedman, R. S., Lodders, K., & Fortney, J. J. 2009, ApJ, 701, L20 Zhou, G., & Bayliss, D. D. R. 2012, MNRAS, 426, 2483
1105.2321	2	1105	2011-10-19T02:38:51	The Role of Drag in the Energetics of Strongly Forced Exoplanet Atmospheres	[ "astro-ph.EP" ]	In contrast to the Earth, where frictional heating is typically negligible, we show that drag mechanisms could act as an important heat source in the strongly-forced atmospheres of some exoplanets, with the potential to alter the circulation. We modify the standard formalism of the atmospheric energy cycle to explicitly track the loss of kinetic energy and the associated frictional (re)heating, for application to exoplanets such as the asymmetrically heated "hot Jupiters" and gas giants on highly eccentric orbits. We establish that an understanding of the dominant drag mechanisms and their dependence on local atmospheric conditions is critical for accurate modeling, not just in their ability to limit wind speeds, but also because they could possibly change the energetics of the circulation enough to alter the nature of the flow. We discuss possible sources of drag and estimate the strength necessary to significantly influence the atmospheric energetics. As we show, the frictional heating depends on the magnitude of kinetic energy dissipation as well as its spatial variation, so that the more localized a drag mechanism is, the weaker it can be and still affect the circulation. We also use the derived formalism to estimate the rate of numerical loss of kinetic energy in a few previously published hot Jupiter models with and without magnetic drag and find it to be surprisingly large, at 5-10% of the incident stellar irradiation.	astro-ph.EP	astro-ph	The Role of Drag in the Energetics of Strongly Forced Exoplanet Atmospheres Emily Rauscher1 Lunar and Planetary Laboratory, University of Arizona, 1629 East University Blvd., Tucson, AZ 85721-0092, USA and Kristen Menou Department of Astronomy, Columbia University, 550 West 120th St., New York, NY 10027, USA ABSTRACT In contrast to the Earth, where frictional heating is typically negligible, we show that drag mechanisms could act as an important heat source in the strongly-forced atmospheres of some exoplanets, with the potential to alter the circulation. We modify the standard formalism of the atmospheric energy cycle to explicitly track the loss of kinetic energy and the associated frictional (re)heating, for application to exoplanets such as the asymmetrically heated "hot Jupiters" and gas giants on highly eccentric orbits. We establish that an understanding of the dominant drag mechanisms and their dependence on local atmospheric conditions is critical for accurate modeling, not just in their ability to limit wind speeds, but also because they could possibly change the energetics of the circulation enough to alter the nature of the flow. We discuss possible sources of drag and estimate the strength necessary to significantly influence the atmospheric energetics. As we show, the frictional heating depends on the magnitude of kinetic energy dissipation as well as its spatial variation, so that the more localized a drag mechanism is, the weaker it can be and still affect the circulation. We also use the derived formalism to estimate the rate of numerical loss of kinetic energy in a few previously published hot Jupiter models with and without magnetic drag and find it to be surprisingly large, at 5-10% of the incident stellar irradiation. 1. Introduction While the field of atmospheric dynamics is well established and mature, we are currently watching its new expansion into the exotic regimes introduced by extrasolar planets. Much of the same, decades-old theory can be fruitfully applied to exoplanet atmospheres, but important differences exist between these planets and those found in our solar system. This may result in the need to adapt theories to a new context and check that basic assumptions are not applied where they are not valid. 1NASA Sagan Fellow -- 2 -- The strongly-irradiated atmospheres of close-in, extrasolar gas giants ("hot Jupiters") differ from solar system giants in a few important ways: 1) the stellar heating of the atmosphere dominates over internal heating, 2) these planets are expected to be tidally locked into synchronous orbits, meaning that the irradiation pattern is strongly asymmetric, and 3) the long rotational (= orbital) period means that the Coriolis force will have a weaker effect in these atmospheres (compared to Jupiter or Saturn) and atmospheric features should exist on a planetary scale (for a comprehensive review see Showman et al. 2010). One important issue that is currently being investigated is the role of friction or drag in these atmospheres. Atmospheres in the solar system are largely inviscid and estimates for hot Jupiters also imply very high Reynolds numbers (e.g. Li & Goodman 2010); however, there can be other important sources of friction or drag in an atmosphere. In particular, Goodman (2009) questioned whether sources of drag were being correctly accounted for in hot Jupiter numerical models, concerned that a lack of drag could allow the winds to accelerate to unphysical speeds. On the Earth it is well known that surface drag has an important impact on the momentum budget of the atmospheric flow. While hot Jupiter atmospheres will not experience drag from a solid surface, other relevant mechanisms may be at work (e.g., shocks2, instabilities, and magnetic drag, which we review in more detail in Section 3.1). It may be that models without explicit friction are still valid, if physical sources of friction are weak (compared to the radiative forcing) and/or numerical dissipation dominates, but it is necessary to have good estimates of the strengths of these mechanisms to determine if that is the case. In this paper we will discuss the importance of friction in the context of the energetics of atmospheric circulation. We emphasize that drag can have multiple effects on the flow, namely as: 1) a momentum sink, or 2) as a possible source of the heating that drives the flow. (Throughout this paper we will be using the terms friction and drag synonymously. Whether it is frictional dissipation from microscopic interactions, or large-scale drag from other physical processes, the result will still be the irreversible conversion of kinetic energy into localized heating.) While the concerns raised in Goodman (2009) were related to the first point, in this paper we will show that the second point may be of comparable importance in terms of correctly modeling exotic exoplanet atmospheres. In Section 2 we review standard equations for the global energy balance in an atmosphere (from Pearce 1978) and explicitly account for heating from frictional dissipation. We then discuss the contexts in which it is acceptable to neglect friction from the energetics (even if not from the momentum equation). In Section 3 we discuss the difference between these contexts and the case of a strongly-forced atmosphere, with a discussion of the drag mechanisms that have been proposed for hot Jupiters (3.1). We then present equations for calculating the local values of the various energies and their conversion rates in Section 3.2, and in Section 3.3 use these to estimate whether frictional heating should have been explicitly accounted for in previous numerical models. We 2We are loosely referring to shocks as drag, in the sense that they transform bulk kinetic energy into heat. -- 3 -- briefly discuss how these concepts will apply to the special case of an eccentric, close-in gas giant planet in Section 4. Finally, in Section 5 we summarize our main points and emphasize the context in which friction may matter to the energetics of atmospheric circulation. 2. The Standard Theory of Friction in Atmospheric Energetics The total energy in an atmosphere is divided between several forms: the specific kinetic energy ( 1 2 v2, where v is the three-dimensional velocity), the internal energy3 (cvT , where cv is the specific heat and T is the local temperature), and the gravitational potential energy (gz, where g is the If the atmosphere is in hydrostatic equilibrium4 gravitational acceleration and z is altitude). (dP/dz = −ρg, where P is pressure and ρ is density), the ideal gas law (P = ρRT ) and the relation between the specific gas constant and specific heats (R = cp − cv) can be used to show that the integrated internal and gravitational potential energies for a column of air are equal to the integrated total specific enthalpy (cpT ). In studying the energetics of an atmosphere, we want to track the sources and sinks of potential energy5 (through heating and cooling), the kinetic energy loss (through frictional dissipation), and the adiabatic conversion between potential and kinetic energies. This conversion occurs when parcels of fluid flow down gradients created by differential heating in the atmosphere. The concept of "available" potential energy (APE) was first formulated by Lorenz (1955) as the component of energy in an atmosphere that is generated by heating and available for direct conversion into kinetic energy. His definition was in reference to some idealized, APE=0, state with no horizontal temperature/pressure gradients. Pearce (1978) created a new formalism based on sources, sinks, and conversions, in which he recognized that APE is only generated through variations in heating, whereas uniform heating only creates "unavailable" potential energy (UPE). This concept of APE has also been formulated as available exergy or enthalpy (Dutton 1973; Marquet 1991). We find the formalism by Pearce (1978) most useful for our purposes here and in this section will follow his derivations. Given that these are generally unfamiliar concepts in astrophysics, we consider it useful to review them here. If X is a local quantity (per unit mass) in the atmosphere, its global average is denoted by X 3The internal energy of a parcel of air can change through compositional or phase changes (e.g. evaporation or condensation), but in this analysis we assume a constant state. 4This is a common assumption -- and usually justifiably so -- for most atmospheres. For a discussion of hydrostatic equilibrium in hot Jupiter atmospheres see Showman et al. (2008). Throughout this work we will assume that the atmospheres we consider are in hydrostatic equilibrium. 5Note that throughout this paper we will be using the term "potential energy" to refer to thermodynamic potential energy (usually in the form of enthalpy). For an atmosphere in hydrostatic equilibrium the gravitational acceleration is balanced by the vertical pressure gradient; it is not this component of the potential energy that leads to winds, but rather the horizontal gradients of potential energy. -- 4 -- and calculated as: X = 1 MZ ρXdV = 1 MZ ρX ρg dxdydP = 1 gMZ XdxdydP (1) where M is the total mass of the atmosphere, x and y are the horizontal coordinates, and the vertical coordinate has been converted from altitude to pressure by assuming hydrostatic equilibrium and that the gravitational acceleration (g) is constant. Following Pearce (1978), we can then write global expressions for the time derivatives of the specific total potential (etot) and kinetic (eK ) energies of the atmosphere: and detot/dt = −c + q deK /dt = c − d (2) (3) where q is the globally averaged heating rate, c is the rate of conversion from potential to kinetic energy, and d is the rate of dissipation of kinetic energy. (In principle the conversion could go in either direction, but the loss of kinetic energy through dissipation requires a positive net conversion from potential to kinetic energy.) To evaluate what fraction of the total potential energy will be available for conversion into kinetic energy, we begin with the thermodynamic equation: ds/dt = (cp/T )(dT /dt) − Rω/P = q/T where s is the specific entropy and ω = dP/dt. The global average of this equation is: T(cid:17) = ds/dt = d(cid:16) q \ q(cid:18) 1 T(cid:19) + \ T(cid:19)∗ q∗(cid:18) 1 (4) (5) where ∗ denotes the deviation from the global average. We can then define a reference temperature, Tr, by 1/Tr ≡ \(1/T ) and multiply Equation 5 by Tr to obtain: Trds/dt = q + \ T (cid:19)∗ q∗(cid:18) Tr = du/dt (6) The quantity Trds/dt is, by definition, the change in energy associated with a globally uniform change in entropy and therefore defines the time derivative of the specific "unavailable" potential energy (UPE, or u). The potential energy that is available for conversion to kinetic energy (the APE, or a) will be the difference between this and the entire potential energy of the atmosphere (from Equation 2): (7) Through this derivation Pearce (1978) was able to identify the term − \q∗(Tr/T )∗ (the global average encompasses all variables) as the efficiency with which heating acts as a source for APE rather than da/dt = detot/dt − du/dt = −c − \q∗(Tr/T )∗ -- 5 -- UPE, and see that it depends on the spatial correlation between non-uniform heating and a non- uniform temperature structure.6 We now return to the equations for the total potential and kinetic energies (2, 3) to see that a steady state atmosphere (detot/dt = deK /dt = 0) requires: q = c = d (8) In order to explicitly track the frictional heating due to the dissipation of kinetic energy, we separate the heating rate into frictional and non-frictional7 components, q = qf + qrad. We can then equate the global loss of kinetic energy with the global frictional heating (d = qf , an irreversible conversion) and see that Equation 8 implies: q = qf , qrad = 0. (9) The expression qrad = 0 is simply a statement of global radiative balance.8 We can also write the heating deviation as a combination of radiative and frictional components: Taken all together, these equations from Pearce (1978) identify the sources of potential energy in a steady-state atmosphere in global radiative equilibrium: q∗ = q∗ rad + q∗ f . (10) • The only source of UPE is the mean frictional heating. • The source of APE is differential heating, both radiative and frictional. With these equations we can now clearly formulate the context in which frictional heating can be neglected from the atmospheric energetics. If the friction is mostly uniform it will all go into generating unavailable potential energy and will not directly affect the circulation. If there is a good estimate of the amount of non-uniform frictional heating, it can be compared to the 6Throughout this analysis we are implicitly assuming time-averaged values, including the definition of the refer- ence temperature, Tr, as is appropriate for atmospheres with minimal global-scale time variability. However, these equations are also valid for instantaneous values and could be expanded to analyze the energetics of a time-varying heating and temperature structure. 7To be more explicit, we make a distinction between frictional heating -- the increase in molecular kinetic energy from frictional dissipation of the kinetic energy of the flow -- and non-frictional heating, which includes all other sources of heating, including radiation, conduction, and latent heat release. Typically the radiative heating dominates over other non-frictional sources. 8By assuming that the total energy of the atmosphere is conserved (detot/dt = 0) we are neglecting the evolutionary cooling of the planet. The gas giants in our solar system are not in global radiative balance, with all but Uranus radiating more flux than they receive from the Sun. However, here we are considering strongly forced atmospheres, where the stellar irradiation dominates over the internal heat flux and the amount by which the planet is out of global radiative equilibrium should be negligible. -- 6 -- radiative heating to determine whether it can be a significant source of APE. In the case of the Earth, for example, it has long been known that variations in the frictional heating are well below 10% of the variations in radiative heating (Pearce 1978). Thus in the Earth-modeling context the surface friction is an important momentum sink, but can safely be neglected from the energetics without altering the global circulation. We emphasize this point: when the friction is weak, and mostly uniform, it is safe to neglect it from the atmospheric energetics without altering the global circulation. 3. Extension to Strongly Forced Exoplanet Atmospheres We now turn our attention to extrasolar planets, in particular hot Jupiters. While they should not have a solid surface (the main source of friction in a terrestrial atmosphere), there are several drag mechanisms that have been suggested to exist in hot Jupiter atmospheres. Most of these possible processes are significantly non-uniform and may be very strong. However, the radiative forcing is also intense and a careful comparison of the relative strengths is warranted. Before moving on to discuss these atmospheres in more detail, we first point out that there is a complication in defining the atmosphere of the planet, as separate from the interior. Unlike terres- trial planets, there is no clear boundary at a solid surface, but instead a smoother transition between the irradiation-driven (upper) atmosphere, a deeper radiative region that is not directly heated by the stellar insolation, and the interior convective zone at pressures of hundreds or thousands of bars (e.g., Fortney & Nettelmann 2010). This makes it more difficult to define a closed system and calculate global rates of energy generation and exchange, since there will be some interaction with the deeper levels. Although the radiative and advective timescales increase dramatically as we descend deeper into the planet, there may be some nonzero net transfer of momentum or heat; for example, the downward transport of kinetic energy and subsequent return as an extra internal heat flux has been proposed as one mechanism for inflating planetary radii (Showman & Guillot 2002). These exchange rates may be important for describing the evolution of a planet throughout its lifetime, but here we are making the implicit assumption that we are averaging over some time period much less than the age of the planet (and that global-scale variability is minimal, consistent with many numerical models, including the ones analyzed in Section 3.3). There is currently no good prescription for how to model interaction between the upper at- mosphere and deepest layers of hot Jupiters. Numerical simulations typically assume that the modeled atmosphere is a closed box, with energy and momentum conserved. For this analysis we will likewise assume that the upper atmosphere is decoupled from the interior -- that any exchange terms are close to zero and can be neglected. In the case that any exchange with the interior is mostly uniform, then the APE generation should not be much effected. However, given the strongly day-night asymmetric nature of hot Jupiters, it is worth considering this issue more carefully. Once there is a better understanding of how best to treat atmosphere-interior coupling, we will want to extend the formalism presented here to include exchange terms. -- 7 -- 3.1. Discussion of possible drag mechanisms Although many drag mechanisms could be at work in a planet's atmosphere, here we re- view three main physical processes that have been identified as possibly producing significant drag on hot Jupiters: vertical shear instabilities (Goodman 2009; Li & Goodman 2010), shocks (Dobbs-Dixon et al. 2010; Rauscher & Menou 2010), and magnetic drag (Perna et al. 2010a). While numerical simulations are able to capture large-scale instabilities9, the computational limits on resolution for global circulation models could eliminate any instabilities that are triggered on sub-grid scales. Additionally, models that assume hydrostatic equilibrium have no formal ver- tical acceleration (vertical velocity exists, but results from the continuity equation) and as such, these models are unable to capture vertical shear instabilities, which must be parameterized. As an attempt to determine what kind of sub-grid parameterization needs to be included in hot Jupiter global circulation models, Li & Goodman (2010) perform a linear analysis and non-linear sim- ulations of limited regions of a hot Jupiter atmosphere (two-dimensional, with finite extents in longitude and altitude). In their simulations they see weak transient shocks, but the main source of kinetic energy dissipation is through recurring Kelvin-Helmholtz instabilities. Although there is not yet a clear prescription for how to do so, it will be important to determine how to connect processes in localized and global simulations to correctly treat the full range of atmospheric scales on hot Jupiters. Transonic flow is present in most numerical simulations of hot Jupiter atmospheres. While transient shocks may exist in the atmosphere, many models contain a shock (or hydraulic jump10) as a steady feature (Dobbs-Dixon & Lin 2008; Showman et al. 2008, 2009; Dobbs-Dixon et al. 2010; Rauscher & Menou 2010). This is a localized region of strong dissipation and heating that has a markedly different spatial structure than the radiative heating pattern, often appearing on the night side of the planet, a fact that may strongly affect the APE generation. The primitive equation mod- els of Showman et al. (2008, 2009) and Rauscher & Menou (2010) do not include shock-capturing schemes and the heating associated with this feature is consistent with adiabatic heating of the downward flow that is induced by the strong convergence, rather than from any shock-heating. The models in Dobbs-Dixon et al. (2010) do include artificial viscosity and so should be able to cor- rectly treat any generation of APE from shocks. In principle the differences between these models could demonstrate the impact of standing shocks on the energetics of the circulation, but there are other important differences between these modeling approaches and it is difficult to disentangle the 9An example of this is the horizontal shear instability that is triggered in the shallow hot Jupiter model of Menou & Rauscher (2009) and which prevents flow speeds from reaching large values. 10A hydraulic jump is the equivalent of a shock in an incompressible hydrostatically-balanced flow; mass and momentum fluxes are continuous across the jump, but the energy flux is not (cf. Landau & Lifshitz 1959). In models that use the primitive equations of meteorology (instead of the full Navier-Stokes equations), vertical sound waves are filtered out and horizontal ones may or may not be, depending on the boundary conditions imposed. See Kalnay (2002) or Vallis (2006). -- 8 -- physics. Finally, recent work implies that the atmospheres of hot Jupiters may be weakly thermally ionized and subject to interaction with the planetary magnetic field (Batygin & Stevenson 2010; Perna et al. 2010a). In particular, the mostly neutral flow may be dragged by the interaction between free charges and the magnetic field. This mechanism is strongly affected by geometry -- both as relates to the orientation of the magnetic field and due to conductivity differences between the hot dayside and cold nightside -- and Perna et al. (2010a) estimate that the amount of drag could be strong enough to significantly alter the momentum of the flow. The drag is associated with an induced magnetic field, itself sustained by electric currents which will dissipate ohmically in a non-local way. The result of this complex interaction between the fluid and the magnetic field is that the kinetic energy dissipated will not be locally deposited as heating. Nevertheless, the energy may be returned to the flow in a way that alters the energetics of the circulation. A full MHD simulation may be required to fully understand how this process can affect the atmospheric flow. There are also numerical sources of dissipation, although these are usually not well character- ized and will vary from code to code, depending on the exact scheme used. A known issue with global simulations is the build up of noise on small scales and a common technique is to use hyper- dissipation (or hyperdiffusion) to remove power from the smallest scales (see Thrastarson & Cho 2011, for a study of how this affects the kinetic energy spectrum of the flow). Hyperdissipation is applied in order to fix an artificial result of the numerical simulation and in principle should not have an erroneous effect on the APE generation. However, any scheme that alters the wind and/or temperature profiles, as this does, has the potential to change both the physical dissipation, as well as the efficiency with which that would be turned into reheating. 3.2. Local theory of energetics In order to evaluate the importance of various possible drag mechanisms in the context of strongly forced atmospheres we need expressions for the local energetics. In this section we will borrow from the derivation of available potential enthalpy in Marquet (1991), but extended to explicitly track dissipation and frictional heating. For a steady ideal flow, without any net heating/cooling or dissipation of kinetic energy, we can use Bernoulli's theorem to define a set of locally conserved energies (see, e.g. Vallis 2006): D Dt(cid:20) 1 v2 + h + Φ(cid:21) = 0 2 where D/Dt is the material derivative, h = cvT + P/ρ = cpT is the specific enthalpy, and Φ is the gravitational potential energy (= gz for a constant gravitational field).11 Our local equations for the 11In Section 2 we were able to include the gravitational energy as part of the total potential energy through (11) -- 9 -- energetics of the circulation must include these energies, the conversion terms between them, and local sources and sinks that can alter the energy budget locally, but should be globally conserved. Recognizing that not all of the enthalpy will be available for conversion to kinetic energy, we employ a local definition of APE as the available potential enthalpy and define it as a ≡ h − Trs, where we have subtracted off the unavailable component (UPE, u ≡ Trs) that will globally integrate to agree with our previous definition of UPE (Equation 6). Given the definition for specific entropy (s = cp ln[θ/θr], where the potential temperature is θ = T [P0/P ]κ, and κ = R/cp), we can expand out the local definition of APE: a = h − Trs = cpT − cpTr ln(cid:18) T Tr (cid:20) Pr P (cid:21)κ(cid:19) = cpT − cpTr ln(T /Tr) + RTr ln(P/Pr). (12) Values with a subscript r are for some reference state with uniform atmospheric temperature, defined as: ≡Zatm 1 Tr 1 T dm M (13) with an implicit time-average over an appropriate interval (equivalent to the constant Tr from Section 2). The specific unavailable potential enthalpy is: u = Trs = cpTr ln(T /Tr) − RTr ln(P/Pr). (14) For an atmosphere in hydrostatic equilibrium the specific kinetic energy is dominated by the horizontal winds and we write: eK = 1 2 v2, with the understanding that v is only the horizontal velocity (along equipotentials). Finally, we must also track the gravitational potential energy: eG = gz. We now derive expressions for the rates of creation, conversion, and dissipation of the various atmospheric energies. With reference to its definition (Equation 14), the material derivative of the unavailable specific enthalpy is simply: Du Dt = Tr =(cid:18) Tr T (cid:19) q, Ds Dt (15) while from the definition of specific available enthalpy (Equation 12) and the thermodynamic equa- tion (Equation 4) we can write: Da Dt = Dh Dt − Tr Ds Dt integration over the atmosphere, but since we are now defining local terms we must keep track of these components separately. -- 10 -- − Tr T q Tr T q = cp DT Dt RT P = ω + q − 1 ∂t ρ(cid:18) ∂P ρ(cid:18) ∂P ∂t 1 ∂P + ~v · ∇P + w ∂z(cid:19) +(cid:18)1 − + ~v · ∇P(cid:19) − wg +(cid:18)1 − Tr T (cid:19) q T (cid:19) q. Tr = = (16) Here we have expanded out ω ≡ DP/Dt = ∂P/∂t + ~v · ∇P + w(∂P/∂z) and used the hydrostatic assumption: w(∂P/∂z) = −wρg, where w ≡ Dz/Dt is the vertical velocity. For the kinetic energy we take the dot product of the horizontal velocity and the standard horizontal momentum equation, D~v/Dt = −(1/ρ)~∇P − ~f × ~v + ~Ffric, where ~f = (2Ω sin φ)~k is the Coriolis parameter (Ω is the planetary rotation rate and φ is latitude) in the vertical direction (~k), to get: DeK Dt 1 ρ = − (~v · ∇P ) − D, (17) where we have defined D as the irreversible frictional dissipation of kinetic energy (= −~v · ~Ffric). The derivative of the gravitational potential energy is, by definition: D(eG) Dt = g Dz Dt = gw (18) We explicitly split the heating into its frictional (qf ) and non-frictional (mainly radiative, qrad) components and combine Equations 15-18 into a local energy cycle:12 T (cid:21) (qrad + qf ) = (cid:20) Tr = (cid:20)1 − Tr T (cid:21) (qrad + qf ) + (~v · ∇P ) − D = − 1 ρ = wg Du Dt Da Dt DeK Dt DeG Dt 1 ρ (~v · ∇P ) − wg + 1 ρ ∂P ∂t (19) (20) (21) (22) The heating/cooling of the atmosphere (qrad + qf ) acts as a source/sink of APE and UPE, with the local temperature structure (Tr/T ) acting as an efficiency factor for generating one form of potential energy or the other. APE will be generated in regions of the atmosphere that are hotter- than-average (Tr/T < 1, so [1 − Tr/T ] > 0) and heated (q > 0) or cooler-than-average (Tr/T > 1) 12Note that the primary difference between this energy cycle and the one presented in Marquet (1991) is that we have explicitly included the frictional heating as a source of potential energy. This also necessitates the inclusion of UPE in our set of equations. -- 11 -- and cooled (q < 0). Note that the global integral of the APE source term ([1 − Tr/T ]q) is equal to the global source term derived previously in Equation 7 (− T ]∗). The APE is converted to kinetic energy by the term −(1/ρ)(~v · ∇P ), which is equivalent to ~v · ∇(gz) if pressure is used as the vertical coordinate. There will be exchange between APE and gravitational potential energy (wg), which relates to the puffing up of a hydrostatic atmosphere when it is heated. The term (1/ρ)(∂P/∂t) is related to adiabatic expansion and disappears under a steady state assumption. \ q∗[ Tr If dissipated kinetic energy is returned locally as frictional heating, so that D = qf , the sum of Equations (19-22) will be D(u + a + eK + eG)/Dt = qrad, for a steady-state atmosphere. While some of the processes described above may result in a local conversion of kinetic energy to heat, it is important to recognize that this will not always be the case. One example is magnetic drag, which will induce a magnetic field, and the resulting current will be dissipated non-locally. In this case the ohmic heating will not occur where the kinetic energy is dissipated and locally D 6= qf . Non-local reheating can also result when instabilities trigger gravity waves, which then propagate through the atmosphere and deposit heat and momentum where they dissipate (Watkins & Cho 2010). We can now see that the importance of returning the dissipated kinetic energy as a source of localized heating will depend on its strength relative to the radiative heating (qf vs. qrad), but also on its spatial correlation with the efficiency factor 1 − Tr/T . Given a drag mechanism with a known strength and spatial distribution (and assuming localized reheating), we can use these expressions to estimate its effect on the energetics of the circulation. 3.3. Drag and energetics in previously published numerical models Perna et al. (2010a) present three numerical models that attempt to estimate the effect of magnetic drag on (the momentum of) hot Jupiter circulation. Although magnetic interaction between the winds and any planetary magnetic field could result in significant drag, the effect will not be localized heating since the currents sustaining the induced magnetic field will be ohmically dissipated in a non-local way (see Batygin & Stevenson 2010; Perna et al. 2010b, for studies of the related heating). This means that we cannot use these models to estimate the amount of localized heating that results from magnetic drag. However, in the spirit of simplicity, we use the same drag strength as the magnetic drag in these models, and we assume local dissipation, in order to explore the effect of frictional heating on the atmospheric energetics and calibrate the level at which it may become important. We compare the models in Rauscher & Menou (2010) and Perna et al. (2010a), which all have identical set-ups for a generic hot Jupiter (see the papers for details), except for different amounts of Rayleigh drag applied as a term in the momentum equation. The model in Rauscher & Menou (2010) has no drag applied, while the three models in Perna et al. (2010a) have drag strengths that vary with height (stronger higher in the atmosphere). Drag timescales range from 107 − 109 -- 12 -- seconds in the weakest-drag model (PMRa in Table 1), and are a factor of 10 and 100 shorter in the medium- and strongest-drag models (PMRb and PMRc in Table 1). For reference, advective timescales are on the order of 105 seconds and radiative timescales vary from ∼ 3 × 103 to greater than 107 seconds (increasing with depth). The local specific kinetic energy loss is calculated from the prescribed drag timescales as D = (1/2)v2/τdrag. The radiative heating is treated by a Newtonian relaxation scheme, making it simple to determine the value of qrad at each point in the modeled atmosphere. We calculate the mass-weighted13 global integrals of these quantities, as well as the source terms for UPE and APE (from Equations 19 and 20), and present our results in Table 1. Note that our numerical solver does not convert the lost kinetic energy into reheating (in other words, while D is nonzero in Equation 21, qf = 0 in Equations 19 and 20). The values listed for the generation of UPE and APE are (Tr/T )qrad and (1−Tr/T )qrad, respectively, while the values listed as "missing" generation replace qrad with qf = D to determine how much potential energy should have been generated if we did indeed include frictional heating in the models. Before discussing the results, we address one issue of concern with these models, which is that the deepest levels (below ∼10 bar) may have not completely reached a statistically steady state by the end of the 1450 or 500 planet days simulated in Rauscher & Menou (2010) and Perna et al. (2010a). It is likely that these deep levels are still being accelerated through interaction with the upper atmosphere. This could complicate our energetics calculations and so we have rerun and extended all of the runs from those papers to mitigate this issue. We calculate energy rates at 4975, 5000, and 10000 planet days for the Rauscher & Menou (2010) model, and at 5000 days for the Perna et al. (2010a) models. 3.3.1. Global energy rates The first point worth noting is that for all of these models the global integral of qrad 6= 0, meaning that the planet is not in global radiative equilibrium. There are two possible resolutions to this issue. The first is that the atmosphere is still being accelerated by the heating and so has yet to reach a steady state. However, if this were the case, the value of qrad for the Rauscher & Menou (2010) run at 10000 days should be less than at 5000 days, whereas there is not significant difference between them. We have also inspected the time evolution of kinetic and internal energies throughout the runs and we see no evidence for continued acceleration of the flow or heating of the atmosphere. The second explanation, which we prefer, is that the non-zero value of qrad is a measure of the amount of numerical dissipation in our model runs. As can be clearly seen in Equation 8, any amount of dissipation (physical or numerical), d, requires a non-zero average global heating rate, q, in order for there to be a net conversion from APE to kinetic energy to balance its loss. While 13The mass element for each point in the atmosphere is dm = ρdxdydz = −(1/g)dxdydP . -- 13 -- Table 1. Global energetics of various runs RM10a PMRab PMRbb PMRcb Radiative heating, qrad (×1019 W) Generation of UPE (×1019 W) Generation of APE (×1019 W) Drag dissipation, D (×1019 W) Relative to qrad: Missing UPE gen. (×1019 W) Relative to qrad UPE gen.: Missing APE gen. (×1019 W) Relative to qrad APE gen.: Heat engine efficiency, η adjusted for missing APE gen.: 160, 150, 180 -39, -65, -20 200, 210, 200 0 0 0 0 0 0 7% -- 390 -19 410 3.8 1% 7.3 38% -3.4 1% 13.7% 13.6% 430 -45 470 34 8% 68 610 -5.0 620 170 28% 340 150% -34 7% 6800% -170 27% 15.7% 20.7% 14.5% 15.0% aFrom runs identical to the one presented in Rauscher & Menou (2010), except extended out to 4975, 5000, and 10000 planet days. bFrom runs in Perna et al. (2010a), with increasing amounts of drag for PMRa, PMRb, and PMRc, associated with planetary magnetic field strengths of 3, 10, and 30 G, respectively. These results are for simulations run for 5000 planet days. Note. -- All values are global totals, integrated over the atmosphere. The dissipation is only what has been explicitly included in the model and does not account for any numerical dissipation. In these models the dissipated kinetic energy is not re-included as localized heating, but the values we list as "missing" generation show what the heating would have added to the global energies. -- 14 -- physically the non-zero global heating rate will be supplied by the global frictional heating (qf in Equation 9), in our numerical models the frictional heating is not included (qf = 0) and so the global radiative heating, qrad, must be non-zero to compensate for numerical losses (dnum). This interpretation is supported by noting that qrad is higher for the runs with drag applied, because in these cases the increased amount of dissipation (now both numerical and physical, dnum + ddrag) must be balanced by an increased amount of heating (qrad, given that qf remains zero). In order to estimate the kinetic energy loss in our models, we can compare the global dissipation (assuming d = qrad) to the input heating from incident stellar flux. Our radiative forcing is applied through a Newtonian relaxation scheme (for details see Rauscher & Menou 2010), so that there is no well-defined stellar flux incident on the model atmosphere. However, the values chosen for the forcing are derived from a one-dimensional radiative transfer model (Iro et al. 2005) for the hot Jupiter HD 209458b and so it is reasonable to compare our dissipation to a stellar heating rate appropriate for that planet.14 Assuming an incident stellar heating rate of ∼ 3 × 1022 W, our drag-free model's qrad ≃ 1.5 × 1021 W (= dnum) implies a numerical dissipation rate of ∼ 5%. For the models with an increasing amount of physical drag applied (PMRa, b, c) the global numerical dissipation is dnum = qrad − D and apparently dominates over the physical drag loss (D), with a magnitude evaluated as (qrad − D)/(3 × 1022) ≃ 13 − 15%. Without a more detailed analysis it is difficult to know where this numerical loss of kinetic energy occurs in our model, be it evenly throughout the atmosphere or preferentially at regions of strong flow convergence or shear. The amount of kinetic energy lost may also be dependent on the order and magnitude of hyperdissipation used.15 These results imply that our numerical scheme leads to a significant amount of kinetic energy loss for the hot Jupiter atmospheric regime, an issue that will require further attention outside of this paper. As expected, the amount of (physical) kinetic energy loss (D) increases with the amount of drag applied to the atmosphere and it also becomes equal to a greater percentage of the radiative heating (although this is not linear with shorter drag timescales, since the increased drag will also work to decrease wind speeds and reduce the kinetic energy). From the formalism presented above, we know that it is not the ratio of qf /qrad that matters in determining whether the energetics of the circulation will be altered, but the contribution that the frictional heating would make to the generation of APE (Equation 20). We find that in these models the spatial correlation between the atmospheric temperature structure and the kinetic energy dissipation is such that, in all cases, frictional heating would have worked to reduce APE generation, instead of feeding UPE. In these models the effect of frictional heating from drag on the atmospheric flow should be twofold: it will work directly to reduce wind 14The planet also has an internal heat flux, but that is <1% of the stellar heating. 15All of the models shown here were run at T31 (spectral) resolution using hyperdissipation applied as an eighth- order operator on the vorticity, divergence, and temperature fields with a coefficient of 8.54 × 1047 m8 s−1. -- 15 -- speeds and the associated heating would also work to decrease the amount of APE generated and available to drive the winds. The importance of this effect is determined by the ratio of the APE generation that would have come from frictional heating and the APE generation from radiative heating. We find that for the amount of drag applied to models PMRa, PMRb, and PMRc, the relative importance of APE generation from frictional heating is 1, 7, and 27%, respectively. We have demonstrated that -- for this spatial function for the drag -- the amount of drag necessary to significantly affect the energetics of the circulation is something at a strength comparable to the medium-strength drag applied in the models of Perna et al. (2010a). We also know that the energetics would not have been significantly changed by the amount of drag applied to the weakest-drag model, while in the strongest-drag model the missing frictional heating would have significantly reduced the rate of APE generation. The energetic effect of frictional heating can also be evaluated by considering the atmosphere as a heat engine and calculating its efficiency. The standard Carnot efficiency is defined as η = (TW − TC)/TW , where TW and TC are the temperatures of the warm source and cold sink, respectively. We calculate TW and TC for each model as the average temperature of the regions of the atmosphere that are being diabatically heated and cooled, respectively, weighted by the local strength of the heating/cooling. This gives η = 3, 9, 12, and 17% for RM10, PMRa, b, and c, respectively. However, this is not the best estimate of the heat engine efficiency of the atmosphere. For the case of the Earth, the Carnot efficiency gives a value of η ∼ 10%, but a better estimate comes from the actual energy input into the system: the ratio of the frictional dissipation (= APE generation) to incoming solar flux, which gives η ≈ 0.8% (Peixoto & Oort 1992). If we compare the global APE generation to incoming stellar heating (≈ 3 × 1022 W) for our models we get η ≈ 7, 13.7, 15.7, and 20.7% for RM10, PMRa, b, and c, respectively. This indicates that, seen as heat engines, these hot Jupiter atmospheres are at least an order of magnitude more efficient than the Earth's atmosphere. The efficiency increases with the amount of drag applied to the hot Jupiter atmosphere, as the same amount of radiative forcing has to be used more efficiently to generate APE, which feeds the kinetic energy, which is then dissipated at a higher rate. Recalculating the efficiency that would have resulted from including the "missing" APE generation, we get η ≈ 13.6, 14.5, and 15.0% for the PMRa, b, and c models. Since the frictional heating in these models is always working against APE generation, the efficiency is reduced. The statement that frictional heating would have reduced the APE generation by X% is equivalent to saying that the efficiency of the atmosphere as a heat engine would be reduced by X%. While this does provide an estimate of the strength of drag that may energetically affect the atmospheric flow -- assuming local reheating from dissipation -- we note that our result is dependent on the set-up we used, in particular the spatial form used for the drag. As we have emphasized throughout, it is the spatial structure of the atmosphere that dictates the flow energetics, specifically the correlations between temperatures, heating rates, and dissipation rates. Our results here are -- 16 -- dependent on the particular relation between the spatial function we have used for the atmospheric drag and the underlying atmospheric structure. Many of the drag mechanisms mentioned earlier will have very different spatial dependencies. The best way to estimate the energetic importance of any drag mechanism is to apply the Equations (19-22) to a particular atmosphere with some form of drag at work. 3.3.2. Spatial structure of the energetics We can further understand the energetics of the atmosphere by studying its spatial properties. This will allow us to identify the regions of the atmosphere that are the most relevant for APE generation, as determined by the spatial correlation between the atmospheric quantities of interest. Before we proceed, we need to highlight an important nuance in this analysis. Although Equations (19) and (20) give local expressions for UPE and APE generation, note that the local values only have meaning in reference to the global total. The quantity Tr is explicitly a global value and is an inherent part of the definitions of APE and UPE, both of which were originally motivated by the concept of a reference state with APE= 0. The reader should keep in mind that the following plots of APE and UPE generation are to be taken as an indication of which areas of the atmosphere (when integrated over) contribute the most to the global total. In Figure 1 we plot the local values of UPE and APE generation as a function of longitude and pressure (the model's vertical coordinate) for an equatorial slice through the atmosphere. (The energy rates tend to be greatest at the equator.) In the left panel we show the radiative generation for the Rauscher & Menou (2010) model at 10,000 days, while in the right panel we plot the "missing" frictional generation for the strongest-drag model from Perna et al. (2010a) at 5000 days. The pattern of radiative UPE and APE generation is similar for all of the models, although the amplitudes and some of the detailed structure is different. The "missing" UPE and APE generation in the PMRa and PMRb models is weaker and primarily concentrated in the uppermost layers of the atmosphere, on the night side. The longer drag times in these models mean that it is only the highest levels (where the applied drag is the strongest) which are able to significantly dissipate kinetic energy, even though most of the kinetic energy resides in the deeper layers. (See Figure 4 of Rauscher & Menou 2010 for a plot of kinetic energy versus depth; it has a maximum at ∼ 2 bar.) -- 17 -- a) d(UPE)/dt a) d(UPE)/dt ] r a b [ e r u s s e r P ] r a b [ e r u s s e r P 0.01 0.1 1 10 100 0.01 0.1 1 10 100 0 -6.4e+18 -3.2e+18 0.0e+00 3.2e+18 6.4e+18 b) d(APE)/dt -2.8e+18 -1.4e+18 0.0e+00 1.4e+18 2.8e+18 100 200 300 longitude [degrees east of substellar] ] r a b [ e r u s s e r P ] r a b [ e r u s s e r P 0.01 0.1 1 10 100 0.01 0.1 1 10 100 0 1.8e+08 1.9e+17 3.7e+17 5.6e+17 7.4e+17 b) d(APE)/dt -5.3e+17 -4.0e+17 -2.6e+17 -1.3e+17 6.6e+12 100 200 300 longitude [degrees east of substellar] Fig. 1. -- Plots of the generation of available and unavailable potential energy (APE, UPE, in watts), for an equatorial slice through the atmosphere (the equator dominates the energy budget). Left: the UPE and APE generated/lost through radiative heating/cooling in the RM10 model. Right: for the strongest-drag model (PMRc), the UPE and APE that would have been generated/lost through frictional heating, if it had been included and locally equal to the kinetic energy lost from drag. The colors have been scaled so that red/yellow is always positive and green/blue is always negative. See the text for further explanation. From Figure 1 we can see that locally the radiative generation of UPE is much greater than the generation of APE, although we know from Table 1 that it will integrate to a smaller global rate. This is a reflection of the fact that the radiative heating on hot Jupiters is intense and a large amount of energy will be stored in UPE, but by its very nature this will have a small net effect on the atmospheric flow. According to the left panel of Figure 1, most of the UPE is generated on the day side and most of the APE is generated on the night side (with opposite behavior for the loss). This is simple to explain by comparing the signs of the local efficiency factor (Tr/T ) and the radiative heating/cooling (qrad). The calculation of the reference temperature, Tr, is a mass-weighted integral over the atmosphere (Equation 13) and the deepest layers, which contain the most mass, are generally hotter than the upper layers (see Figure 5 of Rauscher & Menou 2010 for a plot of temperature- -- 18 -- pressure profiles for different locations throughout the model). For the models under consideration here, Tr ≃ 1750 − 1800 K,16 and Tr/T > 1 for most of the atmosphere at pressures less than ∼ 10 bar. This means that UPE/APE will be generated where the atmosphere is heated/cooled so that qrad is greater/less than zero (see Equations 19 and 20). In general cool air advected from the night side will be heated on the day side, while flow of hot air from day to night will result in cooling on the night side, although we will discuss deviations from this later. The maximum UPE/APE generation will occur where the heating/cooling is strongest, weighted by Tr/T . The amount of heating and cooling is a function of how much the winds are able to bring gas out of local radiative equilibrium and how quickly the gas can radiatively respond. These properties are themselves both related to the generation of APE and conversion to kinetic energy. The UPE and APE generation that would have resulted from localized frictional heating has a different spatial pattern than the radiative generation. In the right panel of Figure 1 we see that the frictional UPE generation (and APE loss) is primarily on the night side. The majority of the kinetic energy in the atmosphere is found on the night side and since we have adopted a horizontally uniform drag in these models, this results in having most of the dissipation on the night side. If the drag were instead stronger on the day side, for example, we might see a more even pattern of dissipation around the planet. Frictional heating will always be positive (the dissipation of kinetic energy does not lead to cooling) and so in the upper atmosphere (at P < 10 bar, where Tr/T > 1) it will always lead to a decrease in APE. In this set-up the night side has most of the radiative APE generation and most of the frictional APE loss. As a test of how much our answers are affected by the value of the reference temperature, we recalculated Tr by only integrating over the upper layers (P < 10 bar), where the atmosphere is being radiatively heated. This gave Tr ≈ 1600 − 1650 K. We then used this value to calculate the global energy rates (over the entire model atmosphere) and found that our radiative APE and UPE generation remained unchanged, while our "missing" frictional generation was decreased by ∼ 10 − 20%. However, it is not appropriate to restrict the definition of Tr to a partial region of the atmosphere. Although not directly heated, the inert layers (at P > 10 bar) contain a significant fraction of the total kinetic energy, which has been gained through interaction with the upper layers, and so must be included as part of the closed system. The real issue is then where to set to the bottom boundary of the model, as discussed above at the beginning of Section 3. We can also examine the horizontal structure of APE generation. In Figure 2 we plot the UPE and APE generation as a function of latitude and longitude at the 150 mbar pressure level. The left panel is the radiative generation from the Rauscher & Menou (2010) model at 10,000 days and the right panel is the "missing" frictional generation for the strongest-drag model from Perna et al. (2010a) at 5,000 days. The radiative generation at this level in the PMRc model is similar to what we see for the RM10 model, although with the amplitudes slightly decreased and the spatial 16Tr is cooler for RM10 and hotter for the PMR runs. There was a variation of ∼1% in Tr between the days sampled for RM10. -- 19 -- structure more smoothed out. We choose to show this particular level because it contains significant generation of potential energy, from both the radiative forcing and the "missing" frictional heating. a) d(UPE)/dt a) d(UPE)/dt 0 45 90 135 180 -135 -90 -45 60 30 0 -30 -60 0 45 90 135 180 -135 -90 -45 60 30 0 -30 -60 -5.9e+18 -2.9e+18 b) d(APE)/dt 0.0e+00 2.9e+18 5.9e+18 2.2e+12 9.0e+16 1.8e+17 2.7e+17 3.6e+17 b) d(APE)/dt 0 45 90 135 180 -135 -90 -45 60 30 0 -30 -60 0 45 90 135 180 -135 -90 -45 60 30 0 -30 -60 -2.2e+18 -1.1e+18 0.0e+00 1.1e+18 2.2e+18 -1.4e+17 -1.0e+17 -6.9e+16 -3.5e+16 -4.3e+11 Fig. 2. -- Cylindrical projections of the generation of available and unavailable potential energy (APE, UPE, in watts), for a horizontal slice of the atmosphere at 150 mbar, with the substellar point at (0,0). Left: the UPE and APE generated/lost through radiative heating/cooling in the RM10 model. Right: for the strongest-drag model (PMRc), the UPE and APE that would have been generated/lost through frictional heating, if it had been included and locally equal to the kinetic energy lost. The colors have been scaled so that the red/yellow is always positive and the green/blue is always negative. See text for further explanation. As we discussed above, the generation of potential energy depends on the temperature structure (by Tr/T ), spatially correlated with the radiative or frictional heating rate. The radiative heating generally works to make the day side hotter and the night side colder, with a dependence on the advected temperature structure, while the frictional heating depends on the flow structure and wind speeds. In order to more clearly understand the patterns of potential energy generation, we plot the temperature and velocity structures for both models at this pressure level in Figure 3. In agreement with Figure 1, we see that most of the radiative UPE generation (and APE loss) is on the day side, while this is reversed on the night side. This is explained, as before, by the fact -- 20 -- 60 30 0 45 90 135 180 -135 -90 -45 0 0 45 90 135 180 -135 -90 -45 -30 -60 60 30 0 -30 -60 859 974 1090 1205 1320 1435 1551 862 977 1092 1207 1322 1437 1552 Fig. 3. -- Cylindrical projections of temperature (color, in K) and velocity (arrows, where length scales with magnitude) for a horizontal slice through the atmosphere at 150 mbar, with the sub- stellar point at (0,0). Left: for the RM10 model. Zonal (east-west) wind speeds range from -1.6 to 4.3 km s−1 and meridional wind speeds range from -1.7 to 2 km s−1. Right: for the PMRc model. Zonal wind speeds range from -2.5 to 2.6 km s−1 and meridional wind speeds range from -2.8 to 2.8 km s−1. In both models the sound speed at this level ranges from 2.1 to 2.9 km s−1. that the efficiency factor, Tr/T , is greater than 1 at this level and so day side heating increases UPE while night side cooling increases APE. The deviations from this pattern are due to the detailed structure of the advected temperature pattern, scaled by the local efficiency factor (at this level Tr/T ranges between ∼2 in the coldest regions and ∼1.2 in the hottest regions). In general we see eastward advection, so that hot air is pulled from day to night across the east terminator (at 90◦) and cold air is pulled across the west terminator. The boundary between positive and negative generation is not exactly at 90◦ longitude because the air that was heated at the substellar point begins to cool even before it reaches the night side. This effect is also seen at the west terminator, close to the poles, where the flow is westward. The region of reversed sign on the night side (positive UPE, negative APE generation) at ∼ ±45◦ latitude is because the air there is cooler than the night side equilibrium temperature and so is being heated, as prescribed by our Newtonian relaxation scheme. Note that night side heating is not unphysical. Although there is no incident stellar light, there is still flux from the deeper, hotter atmospheric layers, which will prevent the night side temperatures from dropping too low. We also find that the strongest rates of potential energy generation are not at the sub- and anti-stellar points ([0◦, 0◦] and [180◦, 0◦] in longitude and latitude, respectively). On the day side the heating is primarily of cold air advected across the west terminator (-90◦) and at this level the relation between advective and radiative timescales is such that the maximum heating occurs at -- 21 -- ∼ −45◦. On the night side there are two regions of cooling: where hot air has been advected from the day side across the east terminator, and where air has been heated by the shock-like feature at ∼ 135◦. This feature is discussed in more detail in Rauscher & Menou (2010); it is a steady feature that extends across many pressure levels and is associated with hot regions, consistent with adiabatic heating from downward flow induced by the strong horizontal convergence. Note that in Figure 1 there is a region of night side APE loss associated with this feature. While the potential energy generation from both the radiative and frictional heating will be weighted by the local temperature through the efficiency factor Tr/T , the radiative heating is more strongly dependent on the temperature structure, while the frictional heating is dependent on the pattern of wind speeds. The "missing" potential energy generation shown in the right panel of Figure 2 follows the same structure as the kinetic energy at this level. We can see this from the flow pattern in Figure 3, where the strongest winds are on the night side, maximized in an equatorial jet and with significant flow in curving branches at high latitude. 4. A Special Case: Drag in Eccentric Planet Atmospheres Not all of the strongly irradiated close-in gas giants are on circular orbits; there are many known with significant non-zero eccentricities, the most extreme being HD 80606b, with an eccentricity of 0.934 (Moutou et al. 2009). These eccentric planets may be subject to the same drag mechanisms that have been proposed for circular hot Jupiters, but with the added complication that the level of stellar irradiation will change throughout the orbit. During periastron passage there will be intense stellar heating that will drive up the potential energy of the atmosphere, some of which will be converted into kinetic energy and then dissipated. The temperature and velocity structures of the atmosphere will vary throughout the orbit. The radiative APE generation should closely follow the temperature evolution of the atmosphere, while the frictional dissipation will follow the evolution of the kinetic energy. At any point in the orbit there may be a mismatch between the rates of APE generation, conversion to kinetic energy, and dissipation, leading to a temporary build up of APE or kinetic energy. This delay in energy transfer is seen in the three-dimensional model of the eccentric hot Neptune GJ436b (a = 0.02872 AU and e = 0.15, Lewis et al. 2010), where they find that the peak in wind speeds occurs 4-8 hours after periastron passage. The evolving dissipation may then lead to a secondary period of APE generation from the frictional heating and, depending on the delayed rates of energy transfer, this could be late enough after periastron passage that the frictional heating might dominate over the stellar heating. As we can see, steady state assumptions and implicit time-averages are no longer valid for eccentric planet atmospheres and a more detailed analysis of the energetics will require further study. -- 22 -- 5. Summary and Conclusions We have presented a discussion of the standard formalism for understanding atmospheric energetics, explicitly accounting for frictional heating from the dissipation of kinetic energy. The potential energy of the atmosphere is divided into the component that can be converted into kinetic energy (the "available" potential energy, APE) and the component associated with uniform changes in temperature (the "unavailable" potential energy, UPE). Uniform frictional heating will feed the UPE, while differential heating (from any source) will feed APE. Frictional heating can be safely neglected from the energetics of the atmosphere (to some level of precision) if it is: 1) predominantly uniform, or 2) its differential heating is much less than the radiative differential heating and it contributes minimally to the generation of APE. We give local expressions for energy generation, conversion, and dissipation in Equations (19- 22). APE is generated in regions with positive correlations between temperature and heating (when hotter-than-average regions are heated and colder-than-average regions are cooled). In order to determine if frictional heating matters for the energetics of an atmosphere, we must identify possible drag mechanisms and estimate their strengths and spatial variations. We can then take global integrals of Equations (19-22) to determine if the APE generation from frictional heating is a significant fraction of the radiative APE generation. Given the many unknowns related to modeling hot Jupiter atmospheres, one may consider that frictional APE generation is no longer negligible once it exceeds ∼10% of the radiative APE generation. This is equivalent to saying that the heat engine efficiency of the atmosphere will be altered by more than 10%. As an exercise and test of this formalism, we calculated APE and UPE generation rates for numerical models that include drag as a kinetic energy sink but do not return the energy as heat. We used the previously published numerical models by Rauscher & Menou (2010) and Perna et al. (2010a), which have identical set-ups but a range of representative drag strengths. We found that in these models the spatial correlations between temperature and heating rates are such that in all cases the frictional heating would have worked to decrease the global generation of APE (or, equivalently, the atmosphere's heat engine efficiency), at a factor of of 1, 7, and 27% for the models with weak, medium, and strong drag, respectively. This provides an estimate of the drag strength at which frictional heating can significantly alter the energetics of the circulation, with the caveat that this result is strongly dependent on the spatial form we chose for the applied drag. In these models drag timescales were horizontally constant; drag mechanisms that have greater spatial variation can have more effect on the energetics at weaker drag strengths. Through this analysis we were also able to estimate the rate of numerical dissipation in these models. Any atmospheric dissipation of kinetic energy (numerical or physical) must be balanced by a non-zero global heating rate, in order to generate the APE that is converted to kinetic energy (Equation 8). In these models the only source of heating is radiative and so this heating must be non-zero, although it means that the atmosphere is not in global radiative equilibrium. Comparing the net radiative heating rate to a representative value for the incident stellar flux, we estimate -- 23 -- that the rate of numerical kinetic energy loss from these models is substantial, at ∼ 5 − 15%. By calculating local APE generation throughout these models we were able to identify the regions that contributed the most to the global energy rates. The source term for APE in Equation (20) is a function of the local heating rate and an efficiency factor based on the local temperature structure, 1 − Tr/T , where Tr is a mass-weighted harmonic mean of temperature throughout the entire atmosphere (Equation [13]). The deeper layers of the atmosphere, which contain most of the mass, are generally hotter than the upper atmosphere. In these models Tr/T > 1 for pressures less than ∼10 bar, the same region in which all of the stellar radiative heating occurs. This means that at these pressures APE is generated radiatively where the atmosphere is cooled (mostly on the night side) and is lost where the atmosphere is heated (mostly on the day side). In these models most of the kinetic energy resides on the night side and -- with our choice of a horizontally constant drag -- this means that most of the dissipation occurs on the night side. Since frictional dissipation always produces heating (never cooling), in this model frictional heating always decreases the APE at pressures less than 10 bar. Under our assumption that any lost kinetic energy is locally returned as heating, this means that the APE loss due to friction is spatially correlated with the regions of strong kinetic energy and strong radiative APE generation. Finally, we discuss how the energetics will change for planets on eccentric orbits. Since the temperatures and wind structure of the atmosphere should vary throughout the planet's orbit, we expect evolving rates of APE production, conversion, and dissipation of kinetic energy. A mismatch, or time delay, between these energy transfer rates would result in a build up of APE or kinetic energy, and following periastron passage there could be a secondary period of heating from frictional dissipation. Through this analysis we have shown how frictional heating would alter the generation of available potential energy and have argued that if this effect is substantial, it could change the nature of the atmospheric circulation, since the APE is converted into kinetic energy (in the form of winds). However, it is difficult to predict what effect we would see in the temperature and wind structure of the atmosphere. In order to test how big of a change would result from the 27% decrease in APE generation for the PMRc model, for example, we would need to run a model that explicitly included the localized heating from frictional dissipation. While beyond the scope of this paper, this will be an interesting avenue to pursue in future work. In conclusion, we suggest that performing an energetic analysis of atmospheric models for strongly-irradiated planets can be beneficial in several ways. Aside from providing another tool with which to understand these exotic atmospheres, the diagnostics presented here can be used to evaluate whether particular heat sources (such as frictional dissipation) can significantly affect the atmospheric circulation and should be explicitly included in numerical models. Additionally, this analysis provides an estimate of the amount of numerical kinetic energy loss in models. While there should be some amount of physical dissipation in an atmosphere, we need to carefully consider whether the magnitude and spatial properties of the numerical dissipation introduce any erroneous effects, and further analysis is warranted. -- 24 -- We thank Adam Showman and Jeremy Goodman for valuable comments that helped improve the quality of this paper. We thank the anonymous referee for useful feedback. This research began while the authors were in residence at the Kavli Institute for Theoretical Physics, generously supported by the National Science Foundation under Grant No. PHY05-51164. This work was performed in part under contract with the California Institute of Technology (Caltech) funded by NASA through the Sagan Fellowship Program. KM recognizes support from NASA under grant no. PATM NNX11AD65G. REFERENCES Batygin, K., & Stevenson, D. J. 2010, ApJ, 714, L238 Dobbs-Dixon, I., Cumming, A., & Lin, D. N. C. 2010, ApJ, 710, 1395 Dobbs-Dixon, I., & Lin, D. N. C. 2008, ApJ, 673, 513 Dutton, J. A. 1973, Tellus, 25, 89 Goodman, J. 2009, ApJ, 693, 1645 Iro, N., B´ezard, B., & Guillot, T. 2005, A&A, 436, 719 Fortney, J. J., & Nettelmann, N. 2010, Space Sci. Rev., 152, 423 Kalnay, E. 2002, Atmospheric Modeling, Data Assimilation and Predictability, by Eugenia Kalnay, pp. 364. ISBN 0521791790. Cambridge, UK: Cambridge University Press, December 2002 Landau, L. D., & Lifshitz, E. M. 1959, Fluid Mechanics, Oxford: Pergamon Press, 1959 Lewis, N. K., Showman, A. P., Fortney, J. J., Marley, M. S., Freedman, R. S., & Lodders, K. 2010, ApJ, 720, 344 Li, J., & Goodman, J. 2010, ApJ, 725, 1146 Lorenz, E. N. 1955, Tellus, 7, 157 Marquet, P. 1991, Quarterly Journal of the Royal Meteorological Society, 117, 449 Menou, K., & Rauscher, E. 2009, ApJ, 700, 887 Moutou, C., et al. 2009, A&A, 498, L5 Pearce, R. P. 1978, Quarterly Journal of the Royal Meteorological Society, 104, 737 -- 25 -- Peixoto, J. P., & Oort, A. H. 1992, Physics of Climate, by Jos´e P. Peixoto & Abraham H. Oort, New York: American Institute of Physics (AIP), 1992 Perna, R., Menou, K., & Rauscher, E. 2010, ApJ, 719, 1421 Perna, R., Menou, K., & Rauscher, E. 2010, ApJ, 724, 313 Rauscher, E., & Menou, K. 2010, ApJ, 714, 1334 Showman, A. P., Cho, J. Y.-K., & Menou, K. 2010, in Exoplanets, ed. S. Seager, 471 Showman, A. P., Cooper, C. S., Fortney, J. J., & Marley, M. S. 2008, ApJ, 682, 559 Showman, A. P., Fortney, J. J., Lian, Y., Marley, M. S., Freedman, R. S., Knutson, H. A., & Charbonneau, D. 2009, ApJ, 699, 564 Showman, A. P., & Guillot, T. 2002, A&A, 385, 166 Thrastarson, H. T., & Cho, J. Y-K. 2011, ApJ, 729, 117 Vallis, G. K. 2006, Atmospheric and Oceanic Fluid Dynamics, by Geoffrey K. Vallis, pp. 770. Cam- bridge University Press, November 2006. ISBN-10: 0521849691. ISBN-13: 9780521849692 Watkins, C., & Cho, J. Y.-K. 2010, ApJ, 714, 904 This preprint was prepared with the AAS LATEX macros v5.2.
0908.0803	2	0908	2009-08-10T08:44:26	Planet formation in highly inclined binaries	[ "astro-ph.EP" ]	We explore planet formation in binary systems around the central star where the protoplanetary disk plane is highly inclined with respect to the companion star orbit. This might be the most frequent scenario for binary separations larger than 40 AU, according to Hale (1994). We focus on planetesimal accretion and compute average impact velocities in the habitable region and up to 6 AU from the primary.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. binincli7 November 16, 2018 c(cid:13) ESO 2018 Planet formation in highly inclined binaries F. Marzari1, P. Th´ebault2 and H. Scholl3 9 0 0 2 g u A 0 1 . ] P E h p - o r t s a [ 2 v 3 0 8 0 . 8 0 9 0 : v i X r a 1 Dipartimento di Fisica, University of Padova, Via Marzolo 8, 35131 Padova, Italy 2 LESIA, Observatoire de Paris, Section de Meudon, F-92195 Meudon Principal Cedex, France 3 Laboratoire Cassiop´ee, Universit?? de Nice Sophia Antipolis, CNRS, Observatoire de la Cote d'Azur, B.P. 4229, F-06304 Nice e-mail: [email protected] e-mail: [email protected] Cedex, France e-mail: [email protected] Received XXX ; accepted XXX ABSTRACT Aims. We explore planet formation in binary systems around the central star where the protoplanetary disk plane is highly inclined with respect to the companion star orbit. This might be the most frequent scenario for binary separations larger than 40 AU, according to Hale (1994). We focus on planetesimal accretion and compute average impact velocities in the habitable region and up to 6 AU from the primary. Methods.Planetesimal trajectories are computed within the frame of the restricted 3 -- body problem determined by the central star, the companion star and massless planetesimals. Relative velocities are computed and interpreted in terms of accreting or eroding impacts. Results.We first show that, for binary inclinations higher than 10 degrees, planetesimals evolve, at a first approximation, in a gas-free environment. Planetesimal accretion is confined around the central star to a region determined by two main parameters, firstly by the mutual inclination between the binary plane and the disk, and, secondly, by the binary eccentricity. Conclusions. The onset of large mutual inclinations between planetesimals due to the nodal randomization causes an increase in the relative velocity. The chances for a successful planet accumulation process depend on the balance between the timescale for node randomization and that of planetesimal accretion. When the binary semimajor axis is larger than 70 AU, planet formation appears possible even for eccentric binaries (up to 0.4). For lower binary separations the region where planetesimals accumulate into protoplanets shrinks consistently. When the mutual inclination between the binary plane and that of the planetesimal disk is larger than 40◦ the Kozai mechanism strongly inhibits planetesimal accumulation. Key words. Planetary systems: formation; Celestial mechanics; Methods: numerical 1. Introduction Planetary formation in binary systems is a complex issue, since each step of the process can be affected in different ways by the companion perturbations. Recent numerical stud- ies (Marzari & Scholl, 2000; Th´ebault et al., 2004, 2006, 2008, 2009; Paardekooper et al., 2008; Xie & Zhou, 2008) have shown that one stage is particularly sensitive to the presence of the sec- ondary star: the initial accretion of kilometer-sized planetesimals (a review on this topic is given in Haghighipour (2009)). Indeed, the coupled effect of secular perturbations of the companion star and friction due to gas in the nebulae induces a size-dependent phasing of orbits which may lead to high impact velocities. This could slow down or even halt the accretion process even in the terrestrial planet region for a wide range of binary separations, i.e., up to ab ∼ 50 AU for high eccentricity systems (see for ex- ample Figs.8 and 9 in Th´ebault et al., 2006). However, these studies are based on the assumption that the planetesimal disk is coplanar to the stellar orbit. Even if this as- sumption might intuitively seems reasonable, a systematic study by Hale (1994) on binary systems with solar-type components suggests that the spin of the two stars is aligned only for binary systems of 30 -- 40 AU or less. Beyond this distance, the primary's equator, and thus a putative planetesimal disk, appears to be ran- Send offprint requests to: F. Marzari domly inclined respect to the binary planet. As a consequence, the inclination between the binary's orbital plane and the cir- cumprimary disk is a parameter that has to be taken into account when studying planetesimal accretion, at least for systems with ab > 30 − 40 AU. We will focus in this paper on binaries with intermediate sep- arations, i.e. in the 40-100 AU range, exploring the inclination as a free parameter. Similarly to the studies for the coplanar case, the main outcome we are interested on is the impact velocity dis- tribution within the planetesimal population, since this parame- ter controls the fate of planetesimal collisions, either accretion or erosion. For a significant mutual inclination between the bi- nary orbital plane and the disk of planetesimals embedded in the gaseous disk, the forced inclination due to the companion star might be much more effective in increasing the relative veloc- ities and halting planet formation. As the planetesimals decou- ple from the gaseous disk and evolve gravitationally, they would feel the binary perturbation and move into inclined orbits. The perturbations of the companion star leads to a progressive ran- domization of planetesimal node longitudes, starting from the outer region of the disk where the secular periods are shorter. The planetesimal disk gradually evolves into a cloud with an an- gular opening equal to twice the mutual initial inclination of the disk respect to the binary plane. We explore in this paper the effects of the nodal randomization on the mutual relative veloci- ties within the planetesimal swarm and on the accretion process. 2 F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries We also estimate the minimum inclination below which planet formation may occur in spite of the binary inclination. The paper is organized as follows: in Section 2. we show that the planetesimal dynamics perturbed by the companion star keeps the swarm out of the gaseous disk for most of the orbital period. This makes gas drag a negligible perturbation. In Section 3. we describe the numerical model used to compute the plan- etesimals relative velocities. Section 4. is devoted to the anal- ysis of the impact velocities for different binary parameters. In Section 5. we derive limiting inclinations for accretion at differ- ent binary separations. Finally, in Section 6. we summarize our results. 2. Decoupling between gaseous disk and planetesimals Most recent studies of planetesimal accretion in a binary envi- ronment (Marzari & Scholl, 2000; Th´ebault et al., 2006, 2008, 2009; Paardekooper et al., 2008) have focused on the influence of the gaseous component of the disk on particle dynamics. However, the implicit assumption that the planetesimal swarm is embedded in the gas disk is only valid if the disk is copla- nar to the binary orbital plane. In this case, planetesimals feel a steady gas drag and have their orbital evolution significantly af- fected by frictional forces. However, if the companion star is on an inclined orbit with respect to the disk mid -- plane, the situation is dynamically more complicated. Three possible scenarios can be envisaged for the interactions between gas and planetesimals: -- Planetesimals form within the gas disk which remains a long -- lived coherent entity in spite of the binary perturba- tions. Numerical simulations with constant viscosity and a polytropic equation of state performed by Larwood et al. (1996) with an SPH code suggest that a disk perturbed by an inclined companion star maintains a coherent structure if the Mach number is lower than 30. It behaves like a rigid body preceding at a rate ωp given by: (1) 15MsR3 d 32MpD3 cos(im)Ω(R) ωp = − where Ms and Mp are the masses of the secondary and pri- mary star, respectively, D is the radius of the circular orbit of the binary, im is the mutual inclination between the disk and the binary orbit, Ω(R) is the keplerian frequency and Rd is the disk radius. This equation is derived under the simplified assumption that the disk has a constant density, but it is in general a good approximation to more general cases. In this scenario, when the planetesimals reach the size (1 -- 10 km in diameter) for which they evolve under the dominating grav- itational force of the two stars, they leave the disk plane be- cause of the forced component in the inclination. Their orbits move in the binary orbital plane and their nodes circulate at different rates, depending on their semimajor axis. Gas drag is probably not a significant perturbation in this scenario, since it affects planetesimal evolution only in the fraction of time during which they cross the gaseous disk plane. This is clearly illustrated in Fig.1, where we show the projection of the planetesimal positions with respect to the gaseous disk when the inclination of the binary orbital plane is im = 20◦ with respect to the initial disk plane. Planetesimals spend most of their time out of disk where the gas density is neg- ligible. According to our simulations, for im = 30◦ planetes- imals spend on average only 9% of their time within one 4 2 ) U A ( z 0 -2 -4 -10 -5 0 x (AU) 5 10 Fig. 1. Planetesimal positions (red squares) in the x -- z carte- sian plane after 1 × 105 yrs from the beginning of their gravity dominated evolution. The gaseous disk (assuming a scale height h = 0.05(r/AU)(5/4) AU) is shown by green dashed lines. The orbital plane of the binary is in the x -- y plane and the disk is inclined of im = 20◦. scale height of the gaseous disk. This fraction increases to 13% when im = 20◦ and to 27% when im = 10◦. As a conse- quence, we estimate that for im ≥ 20◦ gas friction can be, to a first approximation, completely neglected when computing planetesimals orbital evolution, while the im = 10◦ case ap- pears as a limiting value below which gas friction has to be taken into account. -- The gaseous disk begins to warp and it is disrupted by the binary perturbations. It loses coherence and the gas is dis- persed in space. According to Larwood et al. (1996) such disruption by differential precession might affect extremely thin disks. Also in this case, the planetesimals would evolve in a gas-free environment. If the disk disrupts before the planetesimals detach from the disk then this would be the most significant gas free case, where planet formation would start from a disk of solid material made of small planetesimal precursors which would evolve under gravity only. -- As in the first case, the disk remains coherent but it relaxes to the binary plane on a timescale comparable to the viscous timescale (Larwood et al., 1996). If the process is fast due to a high viscosity of the disk, kilometer-sized planetesimals have not the time to form before the disk relaxes to the bi- nary plane. Planetesimals would then grow when their orbital plane, and that of the disk, are already aligned to that of the binary. In this case any information on the initial inclination would be lost and the system would evolve as a coplanar case (Marzari & Scholl, 2000; Th´ebault et al., 2006, 2008, 2009; Paardekooper et al., 2008). Apart from the case of fast relaxation, which possibly occurs in a minority of cases with very high viscosity, in all other cases we expect no or very weak coupling between the gas disk and the planetesimals orbital evolution for binaries with inclination i ≥ 10◦. Note that this low-i case with gas drag has been recently investigated by Xie & Zhou (2009), who showed that small in- clinations between the binary and a circumprimary disc might favor planetesimal accretion as compared to the fully coplanar case. For our numerical exploration, we will thus make the sim- plifying assumption that planetesimals evolve in a gas -- free en- F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries 3 vironment: the gas -- drag force, which introduces a de-phasing of the planetesimal perihelia, does not come into play as in the 2 -- D case introducing a de-phasing of the planetesimal perihelia. The evolution of the swarm can be described as a pure gravitational N -- body problem (Th´ebault et al., 2006) where the relative im- pact velocity steadily increases because of both the de -- phasing of perihelia and nodes. In the next section we will numerically compute the evolution of planetesimal relative velocities. 3. Numerical procedure Planetesimal trajectories are computed within the frame of the restricted 3 -- body problem made of the central star, the compan- ion star and massless planetesimals. We use the same code as in previous studies of the 2-D case (e.g. Th´ebault et al., 2006, 2008, 2009), since this code is 3-D in essence and can handle out of plane perturbers. As already mentioned, the main param- eter we are interested in is the evolution of the average impact velocities within the population of test planetesimals. To that ef- fect, our code has a build-in close encounter search algorithm, which tracks down at each timestep all 2-body encounters, al- lowing to precisely compute the relative velocity for each colli- sion (a "collision" being defined as a close encounter within an "inflated radius" equal to 3 × 10−4 AU assigned to each parti- cle, see Th´ebault et al. (2009) for more details). The precision we obtain in our relative velocity estimate is of the order of 5m/s at 2AU. These values of h∆vi have then to be interpreted in terms of accreting or eroding impacts. The limit between erosion and accretion is defined by a threshold velocity v∗s1,s2, which de- pends on the respective sizes s1 and s2 of the impactors, as well as on the value of Q∗s1,s2, the threshold energy for catastrophic fragmentation. Unfortunately, the parameter Q∗ is very poorly constrained and estimates found in the literature can differ by up to more than 2 orders of magnitude. We chose here a careful approach and consider that Q∗ is comprised between 2 limiting values for a "hard" and "weak" prescription. This will in turn result in 2 bracketing values for v∗s1,s2 (see the discussion in Th´ebault et al., 2006, for more details). The initial planetesimal swarm is made of 15000 bodies ini- tially set on a 2 -- dimensional disk inclined by an angle i0 with respect to the binary orbital plane. All the bodies in the disk are started on circular orbits with semimajor axis ranging from 0.8 to 6.5 AU. All the nodal lines are parallel since all the bodies are clustered in a disk shape. The mass of the primary and secondary stars are fixed to 1M respectively. The binary's or- ⊙ bital parameters ab, eb are chosen as free parameters. ab ranges from 40 to 100 AU; beyond those values the perturbations of the companion in the initial phases of planetesimal accretion are too weak. The binary eccentricity eb assumes different values from 0 to 0.4. The inclination i0 varies from 0◦ to 40◦. For larger in- clinations the Kozai mechanism strongly inhibits planetesimal accumulation, as we will see in the following. and 0.5M ⊙ Our initial model setup is based on the assumption that ini- tially the planetesimal swarm form a flat disk, in other words that it is a dynamically "quiet" system, with all planetesimals e f ree and i f ree = 0. For a gas rich environement this choice might be justified by the fact that the progenitors of the km-sized planetes- 1. In the imals are coupled to the gas and cannot have large e f ree present case, this "decoupling" is more delicate to define, since it 1 although this issue might be more complex than this simple pic- ture, depending on how planetesimals are formed (see Discussion in Thebault et al.2006) i=10 deg i=20 deg i=50 deg 400000 350000 300000 250000 200000 150000 100000 50000 ) r y ( d o i r e P 0 1 2 3 4 5 6 7 8 9 10 Semimajor axis (AU) Fig. 2. Circulation period of the nodel longitude as a function of semimajor axis for planetesimals started on a disk around the primary star. The companion star has a semimajor axis ab = 50 AU, and eccentricity eb = 0 and different inclinations between the binary orbital plane and the planetesimal disk are consid- ered. The values of the circulation period are computed through direct numerical integration of the equation of motion (3 -- body problem). could either be the consequence of the planetesimals vertical dis- persion around a coherent gas disc (case 1), or the consequence of the gas disc dispersal (case 2). In each case, the relative timing between the arrival of the "initial" kilometre-sized planetesimals and the decoupling from the gas is difficult to pinpoint. In a worst case scenario, we could have an initial orbital distribution where some planetesimals have inclination close to i f orced while others are still around i = 0. This would introduce a high initial free relative velocity component that could not be erased with time (contrary to the gas-rich case, see Fig.10 of Thebault et al.2006). This difficult issue clearly exceeds the scope of this paper, but our results should probably be taken as a lower estimate in terms of inhibition of planetesimal accretion 4. Results 4.1. Dynamicalbehaviour,encountervelocities When the planetesimals feel the binary gravitational pull, their node longitude Ω starts precessing at a rate which strongly de- pends on the individual semimajor axis of the bodies as shown in Fig.2. At the same time the binary perturbations cause a growth of the eccentricity and a de -- phasing of the perihelia. In Fig.3 we illustrate the orbital distribution of the planetesimal swarm at t = 5 × 104 and t = 105 yr when the binary orbital plane is in- clined by 30o respect to the planetesimal disk and the eccentric- ity of the binary orbit is 0.2 (the semimajor axis is 50 AU). The different timescales of perihelia and node circulation are mani- fest in the plot. The different degree of randomization of and Ω makes a difficult task to predict the evolution of the relative impact velocity between the planetesimals. The planetesimal disk moves to the binary orbital plane within one orbital period of the outer planetesimals and grad- ually looses coherence as a disk. The nodes are randomized and the planetesimal Keplerian orbits take them out of the disk plane. In Fig.4 we illustrate this behaviour by plotting the positions of the planetesimals at t = 0, when they are still grouped in a disk, and at t = 1 × 105 yr when the randomization has disrupted their 4 y t i c i r t n e c c E ) g e d ( . g n o L n o i l e h i r e P ) g e d ( e d u t i g n o l e d o N 0.1 0.05 0 360 300 240 180 120 60 0 360 300 240 180 120 60 0 1 1 1 F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries 2 2 2 3 3 4 4 3 4 Semimajor axis (AU) 5 5 5 6 6 6 Fig. 3. Distribution of the planetesimal orbital elements at t = 5 × 104 yr (green dots) and t = 105 yr (red dots). The binary parameters are ib = 30o, eb = 0.2 and ab = 50 AU. t=0 t=105 yr 5 4 3 2 1 0 -1 -2 -3 -4 -5 -10 -8 -6 y(AU) -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 x(AU) 4 6 8 10 Fig. 4. 3 -- D spatial distributions of planetesimals at t = 0 (red dots) and t = 1 × 105 yr (green dots). The randomization of the nodal lines disrupt the coherence of the initial disk starting from the outer regions. The disk becomes an extended cloud. The binary orbital parameters are ab = 50 AU, eb = 0. and i0 = 30◦. Fig. 5. Average impact rate in the planetesimal swarm as a func- tion of time for the binary configuration: ab = 50 AU,eb = 0 and ib = 30o. Different curves refer to increasing radial distances from the primary star. The randomization of the node longitudes lead to a sparse planetesimal configuration leading to a reduction of the encounter rate. bations and it can be taken as representative of the impact rate around a single star. The other, and more crucial effect is the progressive increase of impact velocities, as can be clearly seen in Fig.6. For val- ues of ib ≤ 40o, this increase is due to the combination of the 2 and the large inclination particles small free eccentricities e f r oscillations induced by the inclined companion star. Indeed, the small random horizontal excursion due to e f r brings in contact bodies having both i and Ω values increasingly different with time as the node oscillations get tighter. This effect is of course more pronounced in the outer regions of the disk, where the pull of the companion star is felt more strongly. Note however that the steady h∆vi increase is observed everywhere in the 0.8-6 AU region, it is just the pace of this progressive increase which de- pends on radial distance. For ib > 40o a fully different behaviour is observed and the Kozai oscillations come into play. The eccentricity of the plan- etesimal orbits begin to grow while the inclination decreases in order to keep the action: initial spatial configuration. We adopt this timescale since it is a higher limit for the initial planetesimal accumulation process (e.g. Lissauer, 1993) The randomization of the node longitudes affects the dynam- ics of the planetesimal population in two ways. The first one is that the sparser distribution of the bodies in space leads to a lower impact rate in spite of the growth in the relative velocity. In Fig.5 we plot the impact rate as a function of time and radial dis- tance. It shows a gradually declining trend as the nodes become more randomly distributed. This trend is more marked at larger distances from the primary star where the nodal randomization is faster. After 105 years the impact rate is reduced by 55% at 1 AU and it drops down by 94% when r = 6AU, independently of the planetesimal size. This percentage can be interpreted as the fractional reduction of the impact rate compared to that of a planetesimal swarm around a single star. Around t = 0 the impact rate of our model is not yet affected by the binary pertur- √h = q1 − e2 plcos(Ipl) (2) a constant of motion. The period of oscillation of e2 pl depends on the semimajor axis of the planetesimal orbit and, as a conse- quence, large differences build up in the distribution of orbital eccentricities of the swarm. This leads to much higher impact velocities as is clearly shown in Fig.7. 4.2. Effectonplanetesimalaccretion As illustrated in Fig.6 and discussed above, the increase of im- pact velocities is an unavoidable consequence of the node ran- domization due to the companion star's perturbation that affects 2 The existence of a non-zero e f r component is unavoidable. Indeed, for an unperturbed disc of kilometer-sized planetesimals, equilibrium encounter velocities are of the order of the bodies escape velocities, i.e. a few m.s−1, corresponding to e f r in the 10−5 − 10−4 range F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries 5 all the simulated 0.8 to 6 AU region. However, in spite of this undesired effect, planetesimals might still undergo accretion and form planets. The critical condition is that the timescale for both the mutual velocity growth and impact rate reduction should be long compared to the accretion timescale. More precisely, h∆vi have to stay at a low, accretion-friendly value long enough so that large planetesimals have enough time to build up. When the high velocity regime is reached, the growing objects have reached a v∗s1,s2 value high enough to have accreting impacts de- spite higher h∆vi. Of course, studying this effect in detail would require to follow the evolution of the planetesimal size distri- bution in addition to their dynamical one. This is unfortunately beyond the reach of deterministic N-body codes 3. We shall thus adopt here a simplified criterion and consider the time at which an averaged hv∗i is reached for two cases: a "small planetesi- mals" case with smin = 1km and smax = 10km and a "large plan- etesimals" case with smin = 10km and smax = 50km, assuming that planetesimal sizes follow a centered Gaussian distribution between smin and smax. These two limiting hv∗i values are indi- cated by light and dark grey areas in Fig.6, the width of these areas being due to our careful definition of v∗s1,s2 as being com- prised between two extreme values for hard and weak material (see Section 3). As can be seen in Fig.6, for our example case with ab = 50 AU, eb = 0 and ib = 30◦, the whole system remains accretion- friendly for a population of large ≥ 10km bodies for the dura- tion of the simulation, i.e. 105 years, a conservative timescale for runaway growth. For kilometer-sized planetesimals, the sit- uation is less favorable and the a ≥ 5 AU region becomes hos- tile to kilometre-sized planetesimal accretion after ∼ 5 × 104 years. In these regions, planet growth can proceed only if in a few 104 years planetesimals can grow large enough to have ac- creting impacts in a h∆vi ∼ 50 − 100m.s−1 environment. Note however that even i f planetesimal accretion is possible, it can probably not lead to the same runaway growth as around an un- perturbed single star (Kortenkamp et al., 2001). Indeed, the in- crease of the impact velocity, even if it cannot stop accretion, significantly slows it down by decreasing the value of the grav- itational focusing factor onto growing objects. (see the detailed discussion in Th´ebault et al., 2006). Past this initial planetesi- mal growth stage, it is difficult to predict the evolution of the swarm at farther stages when big planetesimals and planetary embryos will collide at these high relative velocities. Large rela- tive inclinations might be produced within the planetary system. Quintana et al. (2002) have simulated the formation of terres- trial planets in α Centauri and found that planets may indeed be formed with large mutual inclinations if the orbital plane of the binary is inclined respect to that of the planetary embryos. However, their simulations start from a coherent and flat disk of protoplanets, while the growing protoplanets might have al- ready developed significant inclinations by the time they collide, according to our scenario. This suggests that the final phase of planet formation in inclined binaries may need additional inves- tigation. The situation is very different in the ib > 40◦ cases where Kozai oscillations dominate the planetesimal dynamics. As ex- pected, these cases are much more hostile to planetesimal ac- cretion. As can be seen in Fig.7 (for ab = 50 AU, eb = 0 and ib = 40◦), after 105years the impact velocities reach values be- 3 The size evolution of a planetesimal population, under the effects of accreting, cratering and fragmenting impacts, can only be treated in statistical particle-in-a-box codes for which the dynamical modelling is necessarily very simplified Fig. 6. Average impact rate in the planetesimal swarm as a func- tion of time at different radial distances from the primary star. The grey bands show the erosion limit for planetesimals 1-10 km in size (lower band) and 10-50 km (upper band). The initial inclination between the planetesimal disk and the binary orbit is ib = 30◦. Fig. 7. As in Fig.6 for an initial inclination of the binary equal to ib = 40◦. yond the erosion limit for both "small" and "large" planetesi- mals in the whole a ≥ 2 AU region. Like in the non Kozai case, the rate at which h∆vi grows strongly depends on the radial dis- tance. As an example, the ≥ 5 AU region becomes hostile to the accretion of 1-10 km bodies after only a few 103years, whereas it takes almost 105years for this to be true at ∼ 1 AU. Th´ebault et al. (2006) derived analytically the timescale be- fore the onset of large impact velocities between planetesimals in planar eccentric binary systems as a function of the binary pa- rameters. In a gas free environment they computed the degree of perihelia randomization required to give high relative velocities and the time needed to reach it. Even if the inclined case may ap- pear similar because accretion occurs in a gas free environment and the relative velocities grow because of the node randomiza- tion, it is not possible to derive a similar analytical expression. It is a complex task to estimate how the relative encounter veloc- 6 F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries ity depends on the degree of node dispersion since this is a full 3 -- D problem. In addition, in this scenario both nodes and peri- helia, once dispersed, contribute to the grow of the planetesimal relative velocities. 5. Role of the binary configuration In the previous section we have analyzed in details the dynamical and accretional behaviour of a planetesimal population for two specific test binary configurations. For pedagogical purposes, we chose cases with eb = 0 to focus on the effect of the binary incli- nation and, in particular, on the transition to a Kozai dominated regime when ib ≥ 40◦. We now explore how these results depend on all the binary's orbital parameters ab, eb and ib (the mass ra- tio between the 2 stars being kept constant and equal to 0.5). For the sake of clarity, the accretion/erosion scenario for each binary configuration will be summarized by a single simplified param- eter, the semimajor axis al within which accretion is possible for the 1-10km planetesimal population at a threshold timescale of t f ∼ 5 × 104yr for the inner zone ranging from 1 − 3 AU and t f ∼ 1 × 105 yr for the outer zone extending from 4 − 6 AU. We assume that planet formation is possible if h∆vi is lower than v∗s1,s2 for any t ≤ ∆t f . Under this condition larger planetesimals can form and resist to higher velocity impacts. As an example, by inspecting Fig.6 we can say that planetesimals beyond 5 AU reach the erosion regime before 1 × 105 yr while for 4 AU h∆vi is still below the erosion limit. The inner region within 3 AU has always impact velocities lower than v∗s1,s2 when t < 5 × 104 yr. is somewhat difficult and arbitrary. Runaway growth in a planetesimal swarm around a single star is supposed to start after about 104 yrs while after 105 yrs ap- proximately 33% of the disk mass is supposed to be in runaway bodies, according to Wetherill and Stewart (1993). These values cannot be directly applied to the binary case since 1) the bi- nary perturbations increase the relative velocities between the planetesimals accelerating the erosion rate 2) the collisional fre- quency decreases with time because of nodal dispersion. In this scenario it is difficult to derive a reliable value for the ∆t f with- out knowing the details of the planetesimal accretion process. This would be possible only with a statistical code like the planet building code (Weidenschilling et al. (1997)) which, on the other hand, cannot model the perturbations of a companion star. Here we cautiously consider a value of ∆t f for the inner and outer re- gion of the planetesimal disk which is larger than the runaway growth timescale and should somehow be a good estimate for the time required by planetesimals to grow large enough to sustain further accretion into protoplanets. The choice of ∆t f The outcome of this analysis is shown in the form of 2- dimensional maps. In Figs.8 we plot al vs. (ab,ib) for eb = 0.0, eb = 0.2 and eb = 0.4. The outcome for ib = 0◦ is only a ref- erence value since for low inclinations gas drag comes into play and it must be included in the numerical model (for these low inclination case, see the recent study by Xie & Zhou, 2009). It is noteworthy that in the analysis of the data it never happens that accretion is possible beyond 4 AU and is prevented within 3 AU. When the companion star is on a circular orbit (Fig.8a), the effect of inclination is noteworthy only for small values of ab. The secular period of the nodes are short enough to perturb the disk on a timescale comparable to the accretion timescale only for ab ≤ 50AU. For these small separations and eb = 0 planet formation is always possible in the ≤ 3 AU region, except for the Kozai regime at ib ≥ 40◦ Beyond this point the swarm is quickly eroded because of the enhanced relative velocities stirred up by the nodal randomization. At small value of ab there is a strong dependence of al on ib showing that at the origin of the shrinking of the planet formation zone there is the randomization of the nodes. For binary semimajor axes larger than 50 AU, the situation is much more favorable to accretion, which can only be stopped, before t f , in the Kozai regime with ib ≥ 40◦. This means that ab ∼ 50AU is approximately the border value within which secular perturbations alone are efficient enough to affect planet accretion in the non Kozai regime. For more eccentric binary orbits, the randomization of both nodes and perihelia combine destructively and push the limit for accretion at larger values of ab. As an illustration, Fig.8c shows the situation for eb = 0.4. In the black zone the relative veloc- ity is beyond the erosion value even at 0.8 AU from the pri- mary star (the inner limit for our planetesimal population). In these cases the formation of planets, in particular in the outer regions of the disc, is strongly inhibited. For inclinations lower than ∼ 10◦ some accretion is possible within 1 -- 2 AU. However, as already stated, our model for ib ≤ 10◦ is less accurate since gas drag may in this case affect the evolution of planetesimals. For ab larger than 70 AU we retrieve the dependence of al on ib and at ab = 90 AU, planetesimal accumulation can only be perturbed in the Kozai regime. 6. Conclusions In this paper we explore the effect of high (ib ≥ 10◦) binary inclination on the planetesimal accretion process. The main out- comes of our work are: -- The gaseous disk and the planetesimals decouple because of the forced inclination of the companion star. As a con- sequence, planetesimal accumulation should occur in an al- most gas free environment for inclinations approximately larger than 10◦. -- The progressive randomization of the planetesimal node lon- gitudes lead to the dispersion of the planetesimal disk that expand into a cloud of bodies surrounding the star. The sparser configuration leads to a significant reduction in the collisional rate. -- The onset of large mutual inclinations among planetesimals causes an increase of the relative impact velocity that may halt the planet formation process. This effect is particularly strong for ib ≥ 40◦ where the Kozai mechanism comes into play. Below this value, planetesimal accretion might be pos- sible, preferentially in the regions closest to the primary star, depending on the value of ib -- The possibility of planetesimal accumulation depends on the balance between the timescale of node randomization and that of planetesimal accretion. For a binary on circular or- bit, the value of ab around which this balance occurs is around 50 AU. Within this value the secular perturbations are fast enough to induce large relative velocity on a timescale shorter than the typical timescale for planetesimal accumula- tion. Outside this limit planetesimals have probably enough time to growth big enough to sustain high velocity impacts. -- When the binary is on an eccentric orbit, the randomization of nodes and periastra can lead to destructive collisions for binary separations up to 70 AU. -- The dispersion of planetesimals in the nodal longitude, in those cases where the accretion is effective and lead to pro- toplanets, possibly leads to planetary systems a) that form on longer timescales because of the reduction of the accre- tion rate b) on highly mutually inclined orbits. F. Marzari, P. Th´ebault and H. Scholl: Planets in inclined binaries 7 perturbations: the planetesimal disk is coherent during accumu- lation into rocky planets and core of giant planets. In this sce- nario, planets with a significant inclination respect to the binary orbit can form. For intial inclinations larger than ∼ 40o the Kozai cycles may lead, on a long timescale, a planet into a highly ec- centric orbit which, at the same time, is very inclined with re- spect to the equator of the primary. This occurs because dur- ing the cycle the inclination respect to the binary plane is sig- nificantly decreased at the eccentricity peak leading the planet far from the equatorial plane of the primary. This might explain the observed orbit of HD 80606b (Wu & Murray (2003); Gillon (2009); Pont et al. (2009)). Acknowledgements. We thank the referee for his useful comments and sugges- tions that helped to improve the paper. References Book "Planets in Binary Systems", ed. Nadar Haghighipour, to be published by Springer, 2009. Gillon, M., 2009, MNRAS (submitted), 2009arXiv0906.4904 Hale, A. 1994, AJ, 107, 306 Kortenkamp, S.J., Wetherill, G.W., Inaba, S. 2001, Science, 293, 1127-1129. Larwood, J.D., Nelson, R.P., Papaloizou, J.C.B., & Terquem, C., 1996, MNRAS 282, 597 Lissauer, J.J., 1993, ARA&A 31, 129 Marzari F., Scholl H., 2000, ApJ 543, 328 Paardekooper, S.-J., Th´ebault, P., & Mellema, G. 2008, MNRAS 386, 973-988. Pont, F., Hebrard, G., Irwin, J. M., Bouchy, F., Moutou, C., Ehrenreich, D., Guillot, T., Aigrain, S., 2009, A&A (accepted for publication), arXiv0906.5605. S.-J., Th´ebault, P., & Mellema, G. 2008, MNRAS 386, 973-988. Quintana, E.V., Lissauer, J.J., Chambers, J.E., Duncan, M.J., 2002, ApJ 576, 982-996 Th´ebault, P., Marzari, F., Scholl, H., Turrini, D., Barbieri, M.,2004, A&A, 427, 1097 Th´ebault, P., Marzari, F., Scholl, H., 2006, Icarus, 183, 193 Th´ebault, P., Marzari, F., Scholl, H., 2008, MNRAS, 388, 1528 Th´ebault, P., Marzari, F., Scholl, H., 2009, MNRAS, 393, L21 Wetherill, G.W., and Stewart, G.R., 1993, Icarus 106, 190 -- 204. Weidenschilling, S. J., Spaute, D., Davis, D. R., Marzari, F. and Ohtsuki, K., 1997, Icarus 128, 429-455. Wu, Y., & Murray, N. 2003, ApJ. 589, 605-614. Xie J.-W.,Zhou J.-L., 2008, ApJ, 686, 570 Xie J.-W.,Zhou J.-L., 2009, ApJ, 698, 2066 40 50 70 60 80 Semimajor axis (AU) 90 100 40 50 70 60 80 Semimajor axis (AU) 90 100 50 40 ) g e d ( n o i t a n i l c n I 30 20 10 50 40 ) g e d ( n o i t a n i l c n I 30 20 10 50 40 ) g e d ( n o i t a n i l c n I 30 20 10 6 5 4 3 2 1 0 6 5 4 3 2 1 0 6 5 4 3 2 1 0 ) U A ( r ) U A ( r ) U A ( r 40 50 70 60 80 Semimajor axis (AU) 90 100 Fig. 8. Maps showing the limiting values for accretion al as a function of (ab, ib). The top plot refers to the case with eb = 0.0, the middle plot to eb = 0.2, and the lower plot to eb = 0.4. The color coding gives different values of al (in AU), the limiting semimajor axis beyond which planetesimal accretion is possible. Each square of the map refers to the lower value of the labels in the axes. The cases for ib = 0o do not include gas drag so they are only indicative. For binary semimajor axes much larger than those we con- sidered in this paper the nodal longitude randomization becomes much longer. As a consequence, the first stages of planetary formation will probably proceed unaffected by the companion's
1711.07535	1	1711	2017-11-20T20:38:07	1I/2017 U1 (Oumuamua) Might Be A Cometary Nucleus	[ "astro-ph.EP" ]	In this work we find evidence that the object is of cometary origin.	astro-ph.EP	astro-ph	"1I/2017 U1 (Oumuamua) Might Be A Cometary Nucleus" Ignacio Ferrín, Jorge Zuluaga Solar, Earth and Planetary Physics Group & Computational Physics and Astrophysics Group (FACom) Institute of Physics, University of Antioquia Medellin, Colombia [email protected] With the detection of an extra-solar object discovered by the Panstarrs telescope on October 18th 2017, the astrophysics of the minor bodies of our own solar system has jumped to other stars. At the time of writing 14 papers have appeared in a short period of time (30 days), in the Physics Arxiv.org depository, discussing matters like origin (Schneider, 2017; Portegies et al., 2017), photometry (Jewitt et al., 2017; Knight et al., 2017; Bannister et al., 2017; Bolin et al., 2017), spectroscopy (Ye et al., 2017; Masiero, 2017), planetary system formation (Trilling et al., 2017), dark matter (Cyncynates et al., 2017), sending an spacecraft (Hein et al., 2017), consequences of detection (Laughlin and Batygin, 2017), dynamics and kinematics(C. de la Fuente and R. de la Fuente, 2017; Mamajek, 2017), and a detailed model for assessing the origin of interstellar bodies (Zuluaga et al., 2017). Since the photometry has not been able to detect a coma or a tail, the current consensus is that we are in the presence of an asteroid, whose colors are comparable to those of excited objects of the Kuiper belt or less-red Jupiter Trojans (Bannister et al., 2017), consistent with Kuiper belt colors (Masiero, 2017), colors overlaping the mean colors of D-type Trojan asteroids and other inner solar system populations, and inconsistent with the ultra-red matter found in the Kuiper belt (Jewitt et al., 2017). Four of the above papers have to do with photometric properties (Jewitt et al., 2017; Bolin et al., 2017; Bannister et al., 2017; Knight et al., 2017). Knight et al. do not give colors. Bannister et al. present their results in their Table 1 and 2, but it was not possible to derive Table 2 from Table 1, some information is missing and thus we were unable to use their datasets for this investigation. Jewitt et al. give the colors in the BVRI system. Bolin et al. give the colors in the Sloan Digital Sky Survey photometric system. Using the following data by Bolin et al. : In this work we find evidence that the object is of cometary origin. B-V = +0.75±0.05 V-R= +0.45±0.05 The agreement is good except that the errors of Jewitt et al. are much smaller. g – r = 0.41±0.24 r – i = 0.23±0.25 g = 23.5±0.1 r = 23.1±0.1 i = 22.9±0.2 B-V= +0.63±0.49 V-R= +0.40±0.42 R-I= +0.43±0.43 B = g + (0.327±0.047)( g - r ) + (0.216±0.027) V = g - (0.587±0.022)( g - r ) - (0.011±0.013) R = r - (0.272±0.092)( r - i ) - (0.159±0.022) I = i - (0.337±0.191)( r - i ) - (0.370±0.041) we convert from the SDSS photometric system to the BVRI system with the transformation equations by Chonis and Gaskell (2008): obtaining the following BVRI colors of 1I/2017 U1 : which may be compared with those derived by Jewitt et al. We have been accumulating colors of cometary nuclei (Ferrín, 2006) and we have data for 21 comets. The above colors can be plotted on the color-color diagrams shown in Figure 1, a) B-V vs V-R and b) R-I vs V-R. The diagrams shows that colors of cometary nuclei fall inside an irregular ellipsoid, but 70% of them fall on a tilted line that we call the main sequence of cometary nuclei colors, MS. Plotting the above three observed colors of 1I/2017 U1 on the diagrams shows that the values lie on the MS. This suggests that 1I/2017 U1 is a cometary nucleus. Deep imaging by Meech (2017) failed to show a coma, but we can not exclude the possibility that this was an active comet because there were no observations at or near perihelion (the object was discovered +39 days past perihelion). Some low level cometary nuclei have very short periods of activity. As an example 107P/Wilson-Harrington was active for only 35±5 days (Ferrín et al., 2017). One implication of this result is that then we do not know from where the object came. It may have come from the inner planetary region, from a local main The next question is if this is an active or an extinct cometary nucleus. belt, from the nearby region of a local Jupiter or from the Oort cloud of the parent star. Conclusion. The facts: a) both independent datasets agree, b) the three data points lie on the MS, c) the data point with the smallest error lies almost at the center of the distribution, and d) the same result is found in the B-V vs V-R as in the R-I vs B-V diagrams, point to the conclusion that 1I/2017 U1 might be a cometary nucleus. ACKNOWLEDGEMENTS The FACom group is supported by the project "Estrategia de Sostenibilidad 2015 - 2016", sponsored by the Vicerectoría de Investigación of the Universidad of Antioquia, Medellín, Colombia. REFERENCES - Bannister, M. T. et al., 2017. https://arxiv.org/pdf/1711.06214.pdf - Bolin, B. T. et al., 2017. https://arxiv.org/pdf/1711.04927.pdf - Chonis, T. S., Gaskell, C. M., 2008. An. J., 135, 264-267 - Ferrín, I., 2006. Icarus, 185, 523-543. - Ferrín, I., Pérez, M., Rendón, J., 2017. PSS, 137, 52-63. - Jewitt, D., et al., 2017. https://arxiv.org/pdf/1711.05687.pdf - Knight, M., et al., 2017. - Meech, K., 2017. MPC 2017-U183. - Zuluaga, J. I. et al., 2017. To appear in Physics Arxive.org. https://arxiv.org/ftp/arxiv/papers/1711/1711.01402.pdf Figure 1a). The color-color diagram for cometary nuclei. a) B-V vs V-R. The nuclei are located inside an irregular ellipsoid, but 70% of them lie on a tilted line we call the main sequence of cometary nuclei colors, MS. Notice that the two measurements of the colors of 1I/2017 U1 by Jewitt et al. (2017) and Bolin et al. (2017), lie on top of the MS. Notice the large error bars of Bolin et al. (2017) data and the small error bars of Jewitt et al. (2017). The black dot represents the centroid of the distribution. The centroid lies inside the error bars of the Jewitt et al. data point. Figure 1b) R-I vs V-R. The same comment as in Figure 1a) applies. The black dot represents the centroid of the distribution.
1606.02701	1	1606	2016-06-08T19:47:15	H-alpha Variability in PTFO8-8695 and the Possible Direct Detection of Emission from a 2 Million Year Old Evaporating Hot Jupiter	[ "astro-ph.EP", "astro-ph.SR" ]	We use high time cadence, high spectral resolution optical observations to detect excess H-alpha emission from the 2 - 3 Myr old weak lined T Tauri star PTFO8-8695. This excess emission appears to move in velocity as expected if it were produced by the suspected planetary companion to this young star. The excess emission is not always present, but when it is, the predicted velocity motion is often observed. We have considered the possibility that the observed excess emission is produced by stellar activity (flares), accretion from a disk, or a planetary companion; we find the planetary companion to be the most likely explanation. If this is the case, the strength of the H-alpha line indicates that the emission comes from an extended volume around the planet, likely fed by mass loss from the planet which is expected to be overflowing its Roche lobe.	astro-ph.EP	astro-ph	Hα Variability in PTFO8-8695 and the Possible Direct Detection of Emission from a 2 Million Year Old Evaporating Hot Jupiter Department of Physics & Astronomy, Rice University, 6100 Main St. MS-108, Houston, Christopher M. Johns–Krull TX 77005, USA [email protected] Lisa Prato & Jacob N. McLane1,2 Lowell Observatory, 1400 W. Mars Hill Rd., Flagstaff, AZ 86001, USA [email protected], [email protected] David R. Ciardi3 & Julian C. van Eyken NASA Exoplanet Science Institute (NEXScI), Caltech M/S 100-22, Pasadena, CA 91125, USA [email protected], [email protected] Wei Chen2 Department of Physics & Astronomy, Rice University, 6100 Main St. MS-108, Houston, TX 77005, USA [email protected] John R. Stauffer Spitzer Science Center/Caltech, 1200 East California Boulevard, Pasadena, CA 91125, USA [email protected] Charles A. Beichman4 Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Drive, Pasadena, CA 91109, USA [email protected] Sarah A. Frazier – 2 – Department of Physics & Astronomy, Rice University, 6100 Main St. MS-108, Houston, TX 77005, USA [email protected] Andrew F. Boden Division of Physics, Math and Astronomy, California Institute of Technology, 1200 E California Blvd., Pasadena, CA 91125, USA [email protected] Maria Morales-Calder´on Centro de Astrobiolog´ıa, INTA-CSIC, ESAC Campus, P.O. Box 78, E-28691 Villanueva de la Canada, Spain [email protected] Luisa M. Rebull Spitzer Science Center/Caltech, 1200 East California Boulevard, Pasadena, CA 91125, USA [email protected] ABSTRACT We use high time cadence, high spectral resolution optical observations to detect excess Hα emission from the 2 − 3 Myr old weak lined T Tauri star PTFO8-8695. This excess emission appears to move in velocity as expected if it were produced by the suspected planetary companion to this young star. 1Department of Physics & Astronomy, Northern Arizona University, Flagstaff, AZ 86011, USA 2Now at Department of Astronomy, The University of Texas at Austin, Austin, TX 78712, USA 3Visiting Astronomer, Kitt Peak National Observatory, National Optical Astronomy Observatory, which is operated by the Association of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation. 4NASA Exoplanet Science Institute (NExScI), California Institute of Technology, 770 S. Wilson Ave, Pasadena, CA 91125, USA – 3 – The excess emission is not always present, but when it is, the predicted velocity motion is often observed. We have considered the possibility that the observed excess emission is produced by stellar activity (flares), accretion from a disk, or a planetary companion; we find the planetary companion to be the most likely explanation. If this is the case, the strength of the Hα line indicates that the emission comes from an extended volume around the planet, likely fed by mass loss from the planet which is expected to be overflowing its Roche lobe. Subject headings: accretion, accretion disks - line: profiles - stars: atmospheres - stars: formation - stars: magnetic fields - stars: pre–main-sequence - 1. Introduction The number of known and candidate extrasolar planets continues to grow. As of 11 May 2016 there are ∼ 3200 confirmed planets and ∼ 2400 planet candidates1. The vast majority of these known and candidate planets have been discovered around middle-aged main sequence stars, and many of these discoveries have challenged our understanding of planet formation, starting with the discovery of the first extrasolar planet orbiting a Sun-like star (51 Peg b; Mayor & Queloz 1995), the first of the class of "hot Jupiters." The wide variety of extrasolar planetary systems now known has led to increased interest and debate over the processes that lead to planet formation. The core nucleated accretion model (e.g., Pollack et al. 1996; Bodenheimer et al. 2000; Hubickyj et al. 2005; Lissauer & Stevenson 2007) produces Jupiter mass objects slowly as they are built up as a result of collisions of dust and ice particles which stick together and gradually form larger and larger bodies until sufficient mass is obtained in order to gravitationally accrete large amounts of gas from the disk. The competing gravi- tational instability model (e.g., Kuiper 1951; Cameron 1978; Boss 1997, 1998; Durisen et al. 2007) posits circumstellar disks which are massive enough to fragment as a result of their own gravity and form Jupiter mass planets on a much more rapid timescale than is typical in conventional core accretion models. Both models of planet formation find support and difficulties with current observa- tions. For example, the planet metallicity correlation (Gonzalez 1997; Santos et al. 2004; Fischer & Valenti 2005) is often quoted as evidence in support of the core accretion model. On the other hand, the direct imaging discovery of massive planets on wide orbits (e.g. HR8799b, c, & d, Marois et al. 2008) has been taken as evidence that gravitational insta- 1see http://exoplanets.org; http://planetquest.jpl.nasa.gov; http://exoplanetarchive.ipac.caltech.edu/ – 4 – bilities must be important for forming at least some planets that cannot be easily explained by the core accretion scenario (Dodson-Robinson et al. 2009). A chief limitation for the core accretion model is the timescale involved relative to the lifetime of circumstellar accretion disks. Neither the core accretion model nor the gravitational instability model originally pre- dicted the existence of hot Jupiters, leading to the need for a mechanism such as migration to move massive planets from their distant formation sites to final positions in close to the star (e.g. Papaloizou et al. 2007; Levison et al. 2007). As a result, there is a significant need to establish the timescale of planet formation and planet migration. The candidate young hot Jupiter studied here could potentially set important constraints on our understanding of these processes. A desirable way to study the planet formation process, its timescale, and the role of migration and other phenomena, is to search for planets around young stars that are in the process of forming their planetary systems. Detecting this youngest generation (∼ few Myr old) of planets presents special challenges. The youngest stars, many still surrounded by the circumstellar material from which planets are presumed to be actively forming, are mostly located in regions at distances of >100 pc. Thus, these targets are inherently faint and are further obscured and reddened by material local to the star forming region. Young pre-main sequence stars have very strong magnetic fields (e.g., Johns–Krull 2007) and possess large star spots (e.g., Hatzes 1995). This makes detection of extrasolar planets through radial velocity (RV) monitoring difficult because star spots can introduce periodic RV signals that mimic those produced by planetary companions (e.g., Saar & Donahue 1997). Nevertheless, several radial velocity searches for planetary mass companions have been or are currently being conducted around low mass, relatively young stars, including pre-main sequence stars (Esposito et al. 2006; Paulson et al. 2006; Setiawan et al. 2007, 2008; Huerta et al. 2008; Crockett et al. 2011, 2012; Nguyen et al. 2012). To date, these studies have yielded one planet around the 100 Myr old G1−G1.5 V star HD 70573 (Setiawan et al. 2007). A planet has also been claimed around the 10 Myr old classical T Tauri star (CTTSs) TW Hya; however, additional study of this object suggests the RV signal from the putative planet is actually caused by spot induced radial velocity jitter (Hu´elamo et al. 2008). Significant spot induced periodic RV variability has been detected in a few additional young stars (Prato et al. 2008; Mahmud et al. 2011), highlighting the challenges of this technique when applied to young stars. Potential planetary mass objects have recently been found around young stars through direct imaging studies (Neuhauser et al. 2005; Luhman et al. 2006; Lafreni`ere et al. 2008; Schmidt et al. 2008; Ireland et al. 2011; Kraus & Ireland 2012; Delorme et al. 2013; Bowler et al. 2013; Kraus et al. 2014). These objects are typically at orbital separations of ≥ 50 AU with estimated masses of several MJU P . These objects also challenge our models of planet for- – 5 – mation, particularly the core accretion model, as the timescale to form planets at such large distances in a disk is expected to be about an order of magnitude greater than the estimated age of these objects (Pollack et al. 1996). The mass estimates for these objects come from comparing their estimated luminosity and temperature with theoretical evolutionary models. The theoretical models are uncertain at these young ages and the observations required to pin down the luminosity and temperature have a number of challenges, resulting in considerable uncertainty in the final mass estimate for a given object. As an example, the companion to GQ Lup discovered by Neuhauser et al. (2005) has mass estimates that range from 1 MJU P on the low side to ∼ 40 MJU P on the high side (e.g., Neuhauser et al. 2008). As we attempt to advance our observational and theoretical understanding of planet and brown dwarf for- mation, it will be important to obtain strong limits on the mass of potential companions to young stars. Such strong mass constraints are the forte of RV measurements of extrasolar planets, particularly for those with independent constraints on the orbital inclination. Transiting extrasolar planets offer several advantages for the study of sub-stellar mass companions to stars (e.g. Charbonneau et al. 2007). Of primary advantage is that the incli- nation is well characterized allowing for a more certain mass determination. Additionally, the radius and hence density of the planet can be determined, and numerous additional fol- low up observations are possible, at least in principle. Several transit searches for extrasolar planets around young stars have now been performed (Aigrain et al. 2007; Miller et al. 2008; Neuhauser et al. 2011; van Eyken et al. 2011; Cody et al. 2013, 2014). Very recently, a can- didate transiting extrsolar planet candidate has been reported around a low mass young (∼ 3 Myr) star (PTFO8-8695) in the Orion OB1a/25-Ori region (van Eyken et al. 2012). This discovery paper suggests a planet with a mass ≤ 5.5 ± 1.4 MJU P and a radius of 1.91 ± 0.21 RJU P in a 0.45 day orbit around a 0.34 − 0.44 M⊙ M3 (Briceno et al. 2005) weak-lined (non accreting) T Tauri star (WTTS). The discovery observations noted unusual changes in the transit light curve from one observing season to the next, which Barnes et al. (2013) argue could be the result of mutual precession of the stellar rotation axis and the planet's orbital axis resulting from tidal interaction of the planet with an oblate star. The analysis of Barnes et al. suggests a likely planet mass of 3.0 or 3.6 MJU P and radius of 1.64 or 1.68 RJU P de- pending on the assumed mass of the star. Follow-up transit and stellar RV observations by Ciardi et al. (2015) lend support to this hypothesis. Here, we report on high spectral resolution optical observations of PTFO8-8695 densely sampled over a few orbital periods. We clearly detect excess Hα emission that moves in radial velocity as predicted by the expected orbit of the companion, providing further evidence for the existence the planet. The Hα luminosity associated with the planet is almost equal to that coming from the star, indicating that the Hα emission volume is substantially larger than the planet itself. The most likely explanation is that the planet is losing mass at a – 6 – subtantial rate, though at this time we are not able to fully rule out a small amount of accretion related emission from a very low mass disk that may remain around this young star. In §2 we describe the observations of this system, in §3 present our analysis of the data, and in §4 provide a discussion of these results, which are summarized in §5. 2. Observations 2.1. HET and Keck Spectroscopy Included in the discovery paper of van Eyken et al. (2012) is a set of high resolution echelle spectra of PTFO8-8695 taken at the Hobby–Eberly Telescope (HET Ramsey et al. 1998) and at the Keck I telescope. At the HET the High Resolution Spectrograph (HRS Tull 1998) was employed, while the High Resolution Echelle Spectrometer (HIRES Vogt et al. 1994) was used at Keck. The details of the observations and data reduction procedures can be found in van Eyken et al. (2012). The spectral resolution of these observations is ∼ 15, 000 at the HET and ∼ 60, 000 at Keck. These data were used by van Eyken et al. (2012) to study the radial velocity variability of PTFO8-8695. Here, we use these observations to investigate the variability of the Hα emission line. 2.2. McDonald Observatory Observations of PTFO8-8695 were taken at the McDonald Obervatory 2.7 meter Harlan J. Smith telescope with the Robert G. Tull Coud´e echelle spectrograph (Tull et al. 1995) on UT 15 November 2013. A 1.2′′ slit was used with the E2 grating to give a spectral resolution of R∼61,400 (with ∼ 2.05 pixels per resolution element) for all observations. Approximately 50 orders with ∼100 A per order were dispersed across the 2080×2048 Tektronix CCD, covering the wavelength region ∼ 3, 400 − 10, 900 A. Integration times for all PTFO8-8695 observations at McDonald Observatory were 2400 s, and the seeing was ∼2′′ on average. Because of the faintness of the target (V = 16.26, van Eyken et al. 2012), the signal to noise ratio obtained is quite low. Nevertheless, significant information can be extracted from the Hα emission line of this star. Table 1 gives a full log of the PTFO8-8695 observations obtained on 15 November 2013. We also use a spectrum of the dM3e flare star AD Leo as an example of the Hα profile shape of a chromospherically active M star of the same spectral type with strong emission lines. This spectrum was obtained with the same telescope and instrument, but on UT 8 November 1995. For this observation, the CCD was placed at the F1 focus (as opposed to – 7 – the F3 focus for PTFO8-8695). A 0.59′′ slit was used to observe AD Leo, yielding a spectral resolution R ∼ 120, 000 spread accross ∼ 4 pixels. The same CCD was used, resulting in only 19 partial (∼ 23 A) orders, including the one containing the Hα line, being recorded. 2.3. Kitt Peak Observatory Observations of PTFO8-8695 were also taken at the Kitt Peak National Obervatory 4 meter Mayall telescope with the echelle spectrograph on the nights of UT 8 − 10 December 2012. A 1.5′′ slit was used with the 58.5−63◦ grating to give a spectral resolution of R∼25,500 (with ∼ 3.24 pixels per resoultion element) for all observations. The slit length projected to 9.73′′ on the sky. Approximately 21 orders with ∼150 A per order were dispersed across the 2080×2048 Tektronix CCD which was binned by a factor of 2 in the cross-dispersion direction resulting in 2080×1024 images. The observed spectra covered the wavelength range ∼ 5, 500 − 8, 600 A. Integration times for the Mayall observations of PTFO8-8695 observations ranged from 600 to 1200 seconds and were typically taken in groups of three exposures with a Thorium-Argon lamp exposures taken at the begining of each group. The seeing varied during the run but was typically ∼2′′. Again, the signal to noise ratio obtained is relatively low. A full log of the PTFO8-8695 observations made at Kitt Peak is given in Table 2. 2.4. Data Reduction All spectra were reduced with custom IDL echelle reduction routines which have been broadly described by Valenti (1994) and Hinkle et al. (2000). The reduction procedure is quite standard and includes bias subtraction, flat fielding by a normalized flat spectrum, scattered light subtraction, and optimal extraction of the spectrum. The blaze function of the echelle spectrometer is removed to first order by dividing the extracted stellar spectra by an extracted spectrum of the flat lamp. Final continuum normalization was accomplished by fitting a low order polynomial to the blaze corrected spectra in the regions around the lines of interest for this study. For the Mayall spectra, there was room on the CCD where sky spectra are recorded above and below the stellar spectrum. A sky spectrum was extracted ∼ 3′′ above or below (depending on how well centered the star was) the stellar spectrum. The resulting sky spectrum was scaled to match the sky lines away from Hα and was then subtracted from the object spectrum. As shown below, the features of interest for this study are much broader than sky lines, so failure to subtract sky from the McDonald observations should not have any significant impact on the final results. The wavelength solution for the – 8 – McDonald data was determined by fitting a two-dimensional polynomial to nλ as function of pixel and order number, n, for approximately 1800 (for the F3 focus) or 100 (for the F1 focus) extracted thorium lines observed from an internal lamp assembly. The wavelength solution for the Kitt Peak data was determined for the Hα order only and utilized a 3rd order polynomial fit to 13 extracted Thorium lines in this order. 3. Analysis 3.1. HET and Keck Data Figure 1 shows the 4 Hα profiles collected at the HET and the 5 Hα profiles collected at Keck. Each profile is labelled by the relative phase of the suspected planet (phase of zero is mid-transit), and time runs down in the Figure. These profiles were collected over 2 months in early 2011. Only one profile was collected on any given night, so at least 2 orbital cycles occur between any two of the observed profiles. To aid in keeping track of the time elapsed between these observations, the phases include non-zero values for the integer part of the phase which represents how may orbits have occured since the first of these exposures. Many of the profiles in Figure 1 show an essentially symmetric emission profile about line center with a narrow core on top of a broader emission base, similar to that seen in rapidly rotating, chromospherically active dMe stars (e.g. Jones et al. 1996). Other profiles show significantly red- or blue-shifted emission in addition to this centered apparent chromospheric emission. There is not an obvious relationship between the location of this excess emission and the predicted velocity position (shown as the red vertical line) of the candidate planetary companion. Overall, Figure 1 shows that there is substantial Hα line profile variability; however, given the generally large time delay in orbital cycles from one observation to the next, it is difficult to understand the source of this variability without the inclusion of datasets with more dense temporal sampling. 3.2. McDonald Data Figure 2 shows all the observed profiles of PTFO8-8695 obtained on 15 November 2013 at McDonald Observatory. Each profile is identified with the UT time of the midpoint of the exposure. Also shown is the velocity position expected for the planetary companion based on the ephemeris published in van Eyken et al. (2012). There is clear Hα emission in the profile that appears to be moving in velocity space with the expected position of the planet. In addition, there is strong, centrally peaked Hα emission. PTFO8-8695 is a WTTS – 9 – (Briceno et al. 2005; Hern´andez et al. 2007) and as such its Hα emission is expected to be chromospheric in origin, that is, it is believed to be produced by the magnetic activity of the star itself (Bertout 1989). In order to estimate the stellar contribution to the observed line profiles we used two different observed profiles from Figure 2. In the profile taken at UT 5:43, the planetary emission appears to be confined to the red side of the line profile. The profile taken at UT 10:32 is the one in which the planet is expected to be the most blue-shifted, and the excess emission appears to be confined to the blue side of the line profile. The red side of the line profile in both the UT 10:32 and UT 11:17 are nearly identical, further indicating the potential planetary emission is confined to the blue side of the profile. Therefore, to estimate the stellar chromospheric component of the line, we take the blue side of the UT 5:43 profile and combine it with the red side of the UT 10:32 profile to get the final profile shown in Figure 3. We measured the Hα equivalent width of this stellar emission, finding a value of 10.49 ± 0.21 A. Also shown in Figure 3 is the line profile of AD Leo, a dM3e flare star rotationally broadened to the same vsini (80.6 km s−1 van Eyken et al. 2012) observed in PTFO8-8695. Barnes et al. (2013) predict that the apparent vsini of PTFO8-8695 will change by ∼ 13% as the result of precession of the stellar rotation axis. This effects was looked for by Yu et al. (2015) and was not seen, though they only had two observing epochs. The exact value we use for vsini may be slightly off; however, we see no evidence that this is the case. The width of the rotationally broadened AD Leo spectrum is similar to the reconstructed "chromospheric" profile of PTFO8-8695, but is weaker. Multipling the AD Leo emission component by a factor of 2.4 leads to the smooth solid profile in Figure 3 which provides a reasonably good match to the PTFO8-8695 profile, suggesting that the profile presented in Figure 3 is a good representation of the stellar component of the line profile. Figure 4 shows each of the observed profiles from Figure 2 with the stellar profile from Figure 3 (the histogram) subtracted off. The leftover emission from this subtraction process could be entirely due to the planetary companion. Examining Figure 4 shows that the cen- troid of this emission tracks very well the predicted velocity position of the planet. However, there is a slight hint that the Hα emission is not quite at the predicted velocity from the ephemeris. In order to characterize this apparent planetary emission, we measure the Hα equivalent width of each profile in Figure 4, as well as the velocity centroid of the emission and the velocity width of each profile. All these values are reported in Table 1. For the velocity centroid we compute the flux (above the continuum) weighted mean velocity from the data. For the line width, σD, we use a measure of dispersion given by σD = (cid:16)Σ(v − vo)2(Fλ − 1) Σ(Fλ − 1) (cid:17)1/2 where v is the velocity of each channel in the continuum normalized spectrum difference spectrum Fλ from Figure 4, and vo is the previously determined velocity centroid. This – 10 – measure of line width has the advantage of being purely emipirical and does not rely on fitting any particular functional form to the data. In the case of a Gaussian profile, this σD is equal to σ in the standard Gaussian formula. There is some indication that the dispersions of the measured profiles are somewhat larger for the times when the planetary velocity position is changing the most rapidly. This may indicate that there is some smearing of the profile from the relatively long exposure times compared to the expected orbital period. 3.3. Kitt Peak Data The first two nights (8-9 December 2012) of the Kitt Peak observations of PTFO8-8695 are shown in Figures 5 and 6. In the figures, the individual profiles are shown in the black histograms, and the average nightly profile is overplotted in red on each of the individual profiles. In contrast to the McDonald observations shown in Figure 2, there is no clear excess emission observed at the predicted velocity position of the planetary companion to the star. Some weak variation in the Hα line is observed on both nights; however, the velocity positions of these changes appear well within the range expected for the stellar chromospheric emission. These variations are likely caused by spatially localized chromospheric emission features on the surface of this magetically active WTTS, and are probably unrelated to the potential planetary companion. The Hα emission on these two nights, presumably stellar in origin, is somewhat weaker than seen in the McDonald data: the emission equivalent widths are 8.33 ± 0.03 A and 7.93 ± 0.06 A on 8 and 9 December 2012 respectively. Since no excess Hα emission is clearly detected on these nights, we do not attempt to measure any quantities related to this and so no data is reported for the excess equivalent width, velocity centroid, or line width in Table 2. On the third night of the Kitt Peak run, the behavior of the Hα line in PTFO8-8695 became much more active, showing similar behavior to that seen in the McDonald data. The Hα profiles from this night (10 December 2012) are shown in Figure 7, where starting with the fourth exposure (UT 5:53) obvious excess emission was detected in the Hα line that again appears to move with the velocity of the planetary companion predicted by van Eyken et al. (2012). We estimate the stellar chromospheric emission using the profile obtained at UT 7:34 on this night when the planet is predicted to be most red-shifted. To minimize the effects of potential stellar Hα fluctuations such as those seen in Figures 5 and 6, we use this profile to estimate the stellar component of the line since it is close in time to the profiles showing the excess emission. The UT 7:34 profile is reflected about zero velocity to create an estimate of the stellar chromospheric profile. This estimated chromospheric profile has an equivalent width of 8.57 ± 0.12 A. – 11 – In an effort to isolate the non-stellar excess emission, we subtract the estimated chro- mospheric from all the spectra observed on night 3, and the resulting profiles are shown in Figure 8. We then performed the same measurements of excess emission equivalent width, velocity centroid, and line width as described above for the McDonald data. These values are given in Table 2. For the first 2 observations on this night, we do not report the excess emission velocity centroid or line width. Both of these quantities are computed by effectively dividing by the excess equivalent width. Because the excess emission is very weak and not securely detected in these first two observations, the values and uncertainties in the velocity centroid and line width become extremely large and provide no constraints on any of the analysis. 3.4. Revisiting the HET and Keck Data As mentioned in §3.1, the HET and Keck data collected in 2011 do not show any obvious relation to the predicted velocity position of the planet. This can be made more clear by attempting a stellar subtraction similar to that done to produce Figures 4 and 8. For these figures, we were able to use profiles from the same night to represent the stellar component, but in the case of the HET and Keck data there is only one observation per night and there is substantial variation from one observation to the next. Since the profiles on the first two Kitt Peak nights show no apparent excess emission, we take the average of these 21 spectra to represent the stellar emission. This stellar component appears to vary in strength between the McDonald and Kitt Peak observations, and examinations of the profiles in Figure 1 suggest it does here as well. Therefore, we scaled this stellar component so that it matched the observed profiles in Figure 1 as best it could, but such that the stellar profile was never above the observed profile. The implicit assumption is that the observed profile is composed of a stellar (including a chromosphere) component plus potentially an excess emission component on top of this. Once we subtract the scaled stellar profiles, only the excess emission is observed and these profiles are shown in Figure 9. While some profiles appear to subtract nearly to zero (e.g. phase 127.792), there is substantial excess emission in many of them. As mentioned above, in some cases the excess emission component is close in velocity position to the predicted velocity of the planet, while in other cases it is far away and shows no obvious connection to the suspected planet. However, given that substantial time elapsed between each of the spectra shown in Figure 9 (and 1), we can not be certain how these profiles evolved. Below, in §4.3.4, we offer an interpretation that attempts to account for all the observed profile shapes from this star. – 12 – 3.5. Orbital Fits The observations of the radial velocity variations of the excess Hα emission recorded at McDonald Observatory and at Kitt Peak, densely sampled in time, lend themselves to an exploration of possible orbital motion. Figure 10 shows the measured velocity centroids of the excess Hα emission against orbital phase based on the ephemeris in van Eyken et al. (2012). While Barnes et al. (2013) revise the ephemeris of PTFO8-8695, they do not provide a unique solution; the results depend on the assumed stellar mass. However, the ephemerides of Barnes et al. result in a maximum phase offset of 0.026 for the data presented here, which is likely negligible compared to other uncertainties in the measurements and analysis discussed below. Therefore, we use the ephemeris of van Eyken et al. (2012) throughout and plot (dashed line) the predicted velocity curve of the planet in Figure 10 using their orbital parameters, which assume a circular orbit. The value of χ2 for this velocity curve is 607, and the reduce χ2 value is χ2 r = 41. The measured velocity variations generally track the expected orbital motion of the planet, with the McDonald observations somewhat closer to the predicted curve than those from Kitt Peak. We discuss below the possibility that the excess Hα emission includes components not purely in orbit with the planet which might lead to this difference; however, for completeness, we proceed here assuming these velocity variations are the result of orbital motion. The radial velocity observations presented in Figure 10 were obtained in 2012 and 2013, while the ephemeris in van Eyken et al. (2012) is based on transit data taken in 2009 and 2010, so there is some uncertainty in the predicted phasing of the radial velocity observations presented here. Ciardi et al. (2015) analyze additional transit data for PTFO8-8695 includ- ing photometry from the Spitzer satellite. The transits recovered in Ciardi et al. (2015) were offset from the predicted times using the discovery ephemeris, but were all within the original ephemeris uncertainty from van Eyken et al. (2012). We can use the offset to the observed Spitzer transit (data obtained in April 2012) as an estimate of the uncertainty in the ephemeris when comparing the predicted phases of the radial velocity observations in Figure 10. The dash-dot curve in this plot shows the predicted radial velocity curve of the planet using the orbit determined in van Eyken et al. (2012), shifted to match the transit midpoint observed by Spitzer in 2012 (Ciardi et al. 2015). The shift is only 0.026 in phase. This emphemeris fits the Hα RV variations more poorly, giving χ2 = 889 and χ2 r = 60. This may indicate a problem with the Spitzer transit time determination or, as discussed in §4, may be be due to the Hα emitting gas not strictly moving with the candidate planet around this star. Assuming that the excess Hα RV variations shown in Figure 10 result from the orbital motion of the planet around PTFO8-8695, we fit a sine wave (circular orbit) to the RV – 13 – points (solid line in the figure; χ2 = 570 and χ2 r = 41), obtaining a velocity semi-amplitude of 211.9 ± 22.1 km s−1. Clearly, the pure sine wave fit and predicted sine curves (dashed and dash-dot lines) of Figure 10 are not good fits to all the observed RV variations of the excess Hα emission, though they do a fairly good job of fitting the data from McDonald Observatory in 2013. We can obtain a better fit using a full Keplerian curve, allowing the orbit to be eccentric. This fit is shown in the dotted line of Figure 10 (χ2 = 177 and χ2 r = 16). In addition to fitting the RV points, we used the transit time as an additionl constraint. To do so, we assume the transit midpoint of van Eyken et al. (2012) as the observed transit time and assign an uncertainty equal to the phase offset between this original transit time and that determined from the Spitzer observations of Ciardi et al. (2015). The resulting velocity semi-amplitude is Ksini = 268.1 ± 13.4 km s−1 with an eccentricity of e = 0.35 ± 0.12. While such a large eccentricity brings the putative planet closer to the star, the planet is still outside the star at periastron for the largest stellar radius (1.07 R⊙) found by either van Eyken et al. (2012) or Barnes et al. (2013), though nominally the surface of the planet would pass within one planetary radius of the stellar surface. We suggest below that there may be significant non-orbital contribution to the excess Hα RV variations, so a full orbital analysis is likely not warranted. As a result we do not report other parameters of the orbital fit. We note that using equation (18) of Gu et al. (2003) that the planet's orbit should be circularized in 103 − 105 yrs depending on the exact value of the planetary quality factor Q′ p, so we do not expect an eccentric orbit for the suspected planet, unless that eccentricity is being excited by a third body in the system. Below we argue that the excess Hα emission observed in the Kitt Peak spectra may be particularly affected by non-orbital motion, so we perform a full Keplerian fit to the McDonald RVs only, again using the transit time as a constraint. This fit is shown in the dash-triple dot line in Figure 10 and gives Ksini = 196.2 ± 5.6 km s−1 with an eccentricity of e = 0.02 ± 0.05. The total χ2 = 14 which is substantially improved, while χ2 r = 7.0 which is also an improvement, but only modestly so due to the small number of degrees of freedom given only 6 RV data points to which the orbit is fit. 4. Discussion We have detected variable Hα emission from the young transiting planet candidate PTFO8-8695. This object sometimes shows a component of Hα emission that appears to be in excess to the stellar chromospheric emission. At times, the excess Hα appears at a random phase relative to the expected velocity position of the claimed planetary companion (Figure 1 and 9). However, at other times the excess Hα emission appears to move in wavelength as would be expected if it were produced by the suspected planetary companion. This raises the intriguing possibility that the excess Hα emission is associated with the planetary – 14 – companion. In general though, there are at least three potential sources of this variable Hα emission that should be considered. The emission could be associated with 1.) the star itself, 2.) accretion flows from a tenuous disk, or 3.) it could be directly related to a low mass companion. 4.1. A Stellar Source: Magnetic Activity and Flaring PTFO8-8695 is a young T Tauri star, and as such is expected to be very magnetically active. Such stars produce variable chromospheric Hα emission (e.g., Hatzes 1995). If all of the observed Hα emission is taken to be from the star, then the measured emission equiva- lent width when it is active would be ∼ 18 A. This would give PTFO8-8695 a log(LHα/Lbol) = -2.92 which would make it stronger than every other magnetically active M star in the sample of Hawley et al. (1996). Additionally, the rotationally broadened and enhanced chro- mospheric profile of the dMe flare star AD Leo matches well the core of the Hα line from PTFO8-8695 (Figure 3). While there is clearly some variability in this core as seen in Fig- ures 5 and 6, the strong, highly Doppler shifted excess Hα emission seen in Figures 2 and 7 cannot be explained by chromospheric emission on the surface of the star. However, there is a potential that a stellar flare could produce such Doppler shifted emission, at least in principle. Stellar flares in dMe stars often produce significant, nearly symmetric line broadening at the base of Hα. The resulting line profile routinely shows the standard narrow chromospheric emission on top of a very broad (FWHM of a few hundred km s−1) base of emission (e.g. Eason et al. 1992; Jones et al. 1996) reminisent of some of the weaker line profiles shown in Figure 1 (e.g. at phases 4.918, 7.127, 22.699, & 127.174). It is very rare in a flare to see dramatically asymmetric emission with a highly red- or blue- shifted component nearly equal in strength to the central chromospheric emission as seen in the profiles of Figures 2 and 7. T Tauri stars in general are known to flare (e.g., Gahm 1990; Guenther & Ball 1999). A potentially better analog of the type of variable Hα emission expected from chromospheric emission and flaring on PTFO8-8695 is the WTTS V410 Tau, with a vsini = 77.7 km s−1 (e.g., Carroll et al. 2012) compared to the measured vsini = 80.6 ± 8.1 km s−1 for PTFO8- 8695 (van Eyken et al. 2012). The Hα emission equivalent width (< 3 A with a typical value ∼ 1 − 2 A) on V410 Tau (e.g., Hatzes 1995; Fern´andez et al. 2004; Mekkaden et al. 2005) is weaker than seen on PTFO8-8695, though V410 Tau has an earlier spectral type which raises the continuum level without necessarily affecting the strength of the chromospheric emission. V410 Tau has been observed to flare in a number of studies. Outside of flares, the Hα line of V410 Tau is fairly symmetric, relatively narrow, and is similar in shape to – 15 – the chromospheric Hα profiles for PTFO8-8695 seen in Figures 3, 5, and 6 (Hatzes 1995; Fern´andez et al. 2004; Skelly et al. 2010). The Hα line of V410 Tau can grow much stronger and broader during a flare, and also show asymmetries; however, the observed asymmetries seen during flares do not show excess emission with apparent peaks shifted out to greater than ±200 km s−1 (Hatzes 1995; Rice et al. 2011) as seen here in PTFO8-8695. The typical pattern in a flare is for the line to very rapidly (times scale of a few minutes) strengthen and broaden with only a slight asymmetry developing. The strength and width of the line then decay exponentially with a time scale of ∼ 1 hour for strong flares (e.g., Fern´andez et al. 2004). This is not the temporal behaviour observed in PTF08-8695. There is at least one additional piece of evidence against the flaring interpretation for the excess Hα emission seen in PTFO8-8695. Whenever V410 Tau shows flare emission in Hα, significant He I 5876 A emission also appears. This He I line is covered in the echelle formats of both our McDonald and Kitt Peak data. We have searched both datasets for evidence of this emission, including co-adding the spectra when the Hα emission appears stationary (UT 9:44 to 11:17 for McDonald; UT 6:45 to 8:25 for Kitt Peak) to increase the signal to noise. No evidence of He I emission is seen. Lastly, if the observed excess Hα emission seen in Figures 2 and 7 were the result of a stellar flare, it would be a remarkable coincidence that the flare induced asymmetry just happened to appear at and move with the same velocity position in the line profile as that expected for the planetary companion. In particular, the motion shown in Figures 2 and 4 where the excess emission first appears strongly on one side of the line profile and then moves to the other side has not to our knowledge been observed in the Hα emission of flare stars. Flares have been observed on PTFO8-8695 (van Eyken et al. 2012; Ciardi et al. 2015), and while flares on this star likely will produce changes in the strength and shape of the Hα emission line, we believe all the points described above argue strongly against a purely stellar origin. 4.2. A Disk Accretion Source Another possibility is that the Hα emission arises from material accreting onto the star from a tenuous disk that may still surround PTFO8-8695. While this star is classified as a WTTS, it is at the boundary between WTTSs and accreting CTTSs (Briceno et al. 2005), although no dust is evident in its infrared spectral energy distribution, including Spitzer data out to 24 µm (Hern´andez et al. 2007). If we take the average excess Hα equivalent width and estimate an accretion rate using the empirical calibrations in Fang et al. (2009), we find a value of ∼ 3 × 10−10 M⊙ yr−1 which is relatively low compared to the full sample in Fang et al. If there is weak accretion from a tenuous disk, it is probable that such a disk would be detected in infrared emission as even very tenuous disks which feed very low accretion – 16 – rates on the order of 10−11 M⊙ yr−1 produce a detectable near-IR excess (Gillen et al. 2014). This accretion rate is significantly less than the value estimated above, suggesting that if the Hα emission was resulted from disk accretion, PTFO8-8695 should show an IR excess unless there has been substantial grain growth around this 2-3 Myr old star which has removed essentially all the small grains. If PTFO8-8695 is accreting material onto the star from a gas disk devoid of small grains, the excess Hα emission may result entirely from the accreting material whether or not there is a planetary companion present. This accretion related emission would presumably be similar to Hα emission seen in other CTTSs, many of which also have close companions. If there is a low mass companion to PTFO8-8695, accretion from a disk may be through accretion streams such as those proposed by Artymowicz & Lubow (1996) (see also Gunther & Kley 2002). At this time, it is not known if a planetary mass companion can excite accretion streams such as those modeled by Artymowicz & Lubow (1996) and Gunther & Kley (2002). A few CTTSs binaries are thought to potentially be accreting through accretion streams. These include DQ Tau (Mathieu et al. 1997; Basri et al. 1997), UZ Tau E (Jensen et al. 2007), AK Sco (Alencar et al. 2003), KH 15D (Hamilton et al. 2012), and the eclipsing binary system CoRoT 223992193 in NGC 2264 (Gillen et al. 2014). None of these stars shows the type of Hα variations seen in PTFO8-8695 where the accretion related emission appears to move from one side of the line profile to the other as it spirals onto one or both of the stars. This type of line profile behavior is also not seen in the Hα profile variations of single CTTSs in extensive studies of their line profile variability (e.g., Giampapa et al. 1993; Johns & Basri 1995a,b; Johns–Krull & Basri 1997; Oliveira et al. 1998; Alencar et al. 2001), nor is it predicted from theoretical models of magnetospheric accretion such as those shown in Kurosawa & Romanova (2013). While we cannot completely rule out accretion from a tenuous disk as the source of the excess Hα emission observed in PTFO8-8695, we argue that this is not the most likely explanation of the observed emission. Deep mid IR or millimeter continuum observations, or a deep search for close circumstellar disk gas emission (e.g. H2 emission, see France et al. 2012), could shed light on whether there is a tenuous disk around this star feeding accretion onto it. 4.3. A Planetary Companion Source Particularly given the radial velocity variations of the excess emission component of the Hα line, the most likely explanation is that this emission arises from an orbiting companion. – 17 – 4.3.1. Mass Estimates Assuming the companion hypothesis, we can use these measured RV variations with the orbital fits performed above to estimate the mass of the system. van Eyken et al. (2012) did not positively detect RV motion of PTFO8-8695 itself; however, if we adopt their upper limit on the reflex motion of the star (2.13 ± 0.12 km s−1), we can use Kepler's laws to find an upper limit for the total mass in the system of (M∗ + MP )sin3i = 0.456 ± 0.082 M⊙ assuming a circular orbit (solid curve in Figure 10). Alternatively, if we use the full Keplerian solution and again apply the upper limit on the stellar reflex motion of the star from van Eyken et al. (assuming that this is the upper limit for the velocity semi-amplitude, K), we obtain an upper mass limit for the system of (M∗ + MP )sin3i = 0.539 ± 0.047 M⊙. Finally, if we restrict ourselves to the fit to the McDonald only data, we find an upper limit to the mass of (M∗ + MP )sin3i = 0.362 ± 0.018 M⊙. We can then use these total mass estimates with the observed ratio of the companion and stellar RV amplitudes, again adopting the upper limit on the stellar RV amplitude of the star from van Eyken et al., to estimate the mass of the companion. Using the parameters from the circular orbit fit gives MP sin3i = 4.75 ± 0.56 MJU P , while using the parameters from the full Keplerian fit gives MP sin3i = 4.45 ± 0.34 MJU P , and the McDonald only fit giving MP sin3i = 4.07 ± 0.45 MJU P , all clearly in the planetary range. van Eyken et al. (2012) determine an orbital inclination of 61.◦8 ± 3.◦7 which then gives a total mass for the system of 0.666 ± 0.108 M⊙ for the circular orbit, 0.787 ± 0.095 M⊙ for the eccentric orbit, and 0.529 ± 0.027 M⊙ for the McDonald only orbital fit. The planet mass then becomes 6.94 ± 0.92 MJU P for the circular fit, 6.50 ± 0.76 MJU P for the eccentric fit, and 5.95 ± 0.66 MJU P for the McDonald only fit. Again, this is an upper limit to the planet's mass given that the RV variations measured for the star are only upper limits on the reflex motion of the orbit. Indeed, Ciardi et al. (2015) present new K band based RV measurements of the star, estimating a new stellar RV semi-amplitude of K = 0.370 ± 0.333 km s−1, which would lower the estimate of the planet's true mass to ∼ 1.1 MJU P . The true stellar RV variations could be even smaller which would imply a still lower planet mass, though if the planet's mass is too low its Roche radius would become smaller than the planet and it should then be losing substantial mass. The estimates given above for the actual mass of the planet depend on the inclination of the planet's orbit, which is uncertain. van Eyken et al. (2012) determine an orbital incli- nation of 61.◦8 ± 3.◦7; however, Barnes et al. (2013) argue that this inclination changes over time as a result of nodal precession of the orbit. They find the value of the orbital inclination can change from ∼ 25◦ to 90◦ with a period in the range of 300-500 days. Thus, there is the possibility the planet's orbital inclination at the time of the observations presented here was significantly different than the 61.◦8 assumed. Assuming i = 61.◦8 ± 3.◦7, the stellar mass inferred from the circular orbit fit is M∗ = 0.659 ± 0.108 M⊙. The effective temperature – 18 – of PTFO8-8695 is 3470 K (Briceno et al. 2005). Using the pre-main sequence evolutionary tracks of Siess et al. (2000), this corresponds to a mass of ∼ 0.35 M⊙ at an age of 2 Myr [the tracks of Baraffe et al. (2015) give a mass of ∼ 0.32 M⊙ at the same temperature and age]. Thus, our measured mass is larger than the predicted masses by ∼ 3σ. It is well established that theoretical pre-main sequence tracks underestimate the true mass of young stars at these masses (Hillenbrand & White 2004). According to these authors, pre-main sequence tracks routinely predict a mass that is 30% lower than the true mass. Thus, if the true mass is 0.659 M⊙, we might expect evolutionary tracks to give a mass of 0.461 M⊙. With an uncertainty of 0.108 M⊙, the actual predicted masses from the evolutionary tracks are 1.0 – 1.3 σ lower than this value. Thus, while our inferred mass is higher than the predicted mass, the difference is within the expectations given the known systematic trends in the evolutionary tracks at these low masses. No matter what the true inclination is, our measured mass will be larger than the mass inferred from evolutionary tracks. For example, if the inclination is 90◦, the measured mass would be 0.451 M⊙. While this is larger than the mass inferred from the tracks, it is larger by ∼ 30% which is the typical difference found for young stars in this mass range. If the orbital inclination at the time of the spectroscopic observations presented here was as low as 40◦, this would imply true stellar mass of 1.7 M⊙, which corresponds to an effective temperature of ∼ 4670 K, or a spectral type between K3 and K4, using the Siess et al. (2000) evolutionary tracks at 2 Myr. This is very inconsistent with the observed spectral type. Thus, we infer that the orbital inclination at the time of the spectroscopic observations could not be much less than 61.◦8. 4.3.2. Production of Excess Hα Emission If the excess Hα emission does come from the planetary companion, how is it produced? One of the most remarkable aspects of the apparent excess Hα emission is its strength. Tables 1 and 2 show that, when visible, the planet's Hα equivalent width is typically 70% to 80% that from the star. Because the excess equivalent width and the stellar chromospheric equivalent width are measured relative to the same stellar continuum, the Hα luminosity from the planet then reaches 80% that from the star. The 10.5 A emission equivalent width from the star, while stronger than the dM3e star AD Leo as shown in Figure 3, is not atypical of very magnetically active dM3-4e stars, several of which have Hα emission equivalent widths of 10.0 A or more (e.g. Hawley et al. 1996). However, the planet orbiting PTFO8-8695 is much smaller than the star - the apparent area of the planet is ∼ 3.4% that of the star (van Eyken et al. 2012). In order to produce an Hα equivalent width that is ∼ 75% that of the star, the Hα surface flux of the planet would have to be 22 times that of the star. We do not know the effective temperature or spectral type of the young planet orbiting PTFO8- – 19 – 8695; however, it is likely cooler and later in type than the star. Such a dramatic rise in Hα surface flux is not observed in active stars or brown dwarfs with spectral types later than M4 (e.g. Schmidt et al. 2010; West et al. 2011). As a result, it is doubtful the emission from the planet around PTFO8-8695 is produced by magnetic activity. We further argue that the emission from this planet does not arise on its surface at all, but instead either in a confined region around the planet such as in the planetary magneto- sphere, or in a more complicated flow with the planet feeding the material. Similar gas flows have been observed in absorption (e.g. Cauley et al. 2015; Ehrenreich et al. 2015) and have been modeled by a number of investigators (e.g. Matsakos et al. 2015). The radius of the planet is in the range of 1.64 RJU P (Barnes et al. 2013) to 1.91 RJU P (van Eyken et al. 2012). Assuming the planet's rotation is tidally locked with its orbit (the synchronization timescale is estimated to be << 1 Myr using relationships given in Gu et al. 2003), the maximum vsini the planet could have is 18.6 – 21.7 km s−1. However, the average velocity width of the excess Hα emission is 87.3 ± 4.9 km s−1, significantly more than can be accounted for by the rotation of the planet. Instead, we suggest the most likely explanation is that the emission results from mass outflow from the planet. 4.3.3. Mass Loss from the Companion Mass outflow from a hot Jupiter was first detected by Vidal-Madjar et al. (2003) and these observations have been confirmed by several additional studies (e.g., Linsky et al. 2010). Mass loss from hot Jupiters has been studied theoretically by a number of investiga- tors (e.g., Lammer et al. 2003; Baraffe et al. 2004, 2006; Murray-Clay et al. 2009; Lai et al. 2010; Trammell et al. 2011; Adams 2011). van Eyken et al. (2012) noted that this planet's radius is very close to its Roche lobe radius. Using equation (1) of van Eyken et al. (2012) and the best fitting parameters of Barnes et al. (2013) actually places the Roche radius of the planet slightly inside the inferred radius of the planet. Thus, the potential planet orbiting PTFO8-8695 is a prime candidate for significant mass loss. Using the best fit parameters from Barnes et al. (2013), the escape velocity from the planet is 82 – 88 km s−1 which is very comparable to the velocity width of the excess Hα emission apparently associated with the planet, further suggestive that this planet may be losing considerable mass. Murray-Clay et al. (2009) was the first to theoretically study mass loss from a hot Jupiter orbiting a young pre-main sequence star. These authors note that the very high mass loss rates from T Tauri stars will likely completely stifle planetary flows on the day side of the hot Jupiter as the result of the large ram pessure associated with the stellar wind. However, the large mass loss rates used by Murray-Clay et al. (2009) are only really appropriate for CTTSs which are still accreting material from a disk. Stellar wind mass loss rates appropriate for a WTTS such as PTFO8-8695 have not yet been determined. – 20 – The semi-major axis of the suspected companion to PTFO8-8695 is less than ∼ 2 R⊙ while the star itself has a radius of ∼ 1 R⊙ (van Eyken et al. 2012; Barnes et al. 2013). Thus, the planet would be orbiting only about 1 stellar radius above the stellar surface. T Tauri stars are measured to have average surface magnetic fields of 2 – 3 kG (Johns–Krull 2007), so it is likely that the planet is orbiting inside the Alfv´en radius of the star, in which case Murray-Clay et al. (2009) suggest the outflow from both the day and night side of the planet may be suppressed. However, the planetary magnetosphere may help balance the stellar magnetic pressure in order to allow the wind from the planet to get started, and it may also lead to the outflowing material building up substantial optical depth at a size that is several times the nominal planetary radius (e.g., Trammell et al. 2011) which is needed in order to produce the amount of Hα emission seen as discussed above. Once the material escapes the planet and its magnetosphere, it may then be the stellar field that ultimately controls where the material flows. Most models of mass loss from a hot Jupiter find that the flow ultimately gets redi- rected by the stellar wind with the material eventually leaving the system. Here though, the planetary wind would be launched inside the region of space governed by the stellar magnetosphere. In the case of CTTSs, much of the material flowing off the disk near the co-rotation radius is forced by the stellar magnetosphere to travel along the field lines and accrete onto the star (e.g., Bouvier et al. 2007). van Eyken et al. (2012) find that the stel- lar rotation period is locked to the candidate planet's orbital period, so the planet would be feeding material into the stellar magnetosphere at the co-rotation radius which is very analogous to CTTSs accretion models (e.g., Shu et al. 1994). It may then be that material flowing off the planet is accreting onto the star in a magnetospheric accretion flow similar to that in CTTSs. This accreting material might then produce emission at higher redshifted velocities as it accelerates in the gravitational potential well of the star, as modeled for ex- ample by Matsakos et al. (2015). Such higher velocity tails are hinted at in Figures 7 and 8. Material that is lost into a wind may also contribute Hα emission at velocities different from the orbital velocity of the suspected planet. 4.3.4. A Paradigm to Explain the Full Range of Hα Profiles The interplay of material being lost by the planet embedded in the magnetosphere of a young star might explain all the Hα variations observed in PTFO8-8695. Line profiles seen in Figures 1, 5, and 6 suggest that there are times when only emission from the chromo- spherically active star is present. As mentioned above, PTFO8-8695 is known to flare. A strong stellar flare may rapidly strip away the outflowing material from the planet resulting in no excess emission for some time. Due to flares or other dynamo related variability, the planet at times may be in regions dominated by open stellar field lines where a strong stellar – 21 – wind is flowing which stifles the planetary outflow as suggested by Murray-Clay et al. (2009). At such times, there may be no excess Hα emission associated with the planet. At some point after a flare or once the planet moves into a region of closed stellar fields, the plane- tary outflow may be re-established and we observe the excess emission associated with the planet. As the flow continues and material begins to fill the stellar magnetosphere, accrete onto the star, or be lost to a wind, Hα emission will begin to have contributions from gas no longer directly tied to the planet. The radial velocity structure of this gas as viewed from Earth, and the resulting line profile, might then get quite complex and produce the various profile shapes observed in Figure 1 and 9, as well as producing the higher velocity tails seen in Figures 5 and 6. The possibility that some component of the Hα excess emission is not directly tied to the planet cautions that the eccentric Keplerian orbit discussed above may not be real, but simply due to the difficulty in isolating the emission coming directly from the planet. Indeed, since the RV variations traced by the excess Hα emission likely include components not strictly in orbit with the suspected planet, it is difficult to definitively rule out any of the specific orbital fits shown in Figure 10. When the stellar field adjusts itself due to another flare or some dynamo variation, the sequence outlined above could then start all over again. Barnes et al. (2013) suggest that the planet's orbital plane is inclined relative to the stellar equatorial plane, so even though the stellar rotation and planet's orbit may be locked, the planet is still moving in latitude relative to the star and hence is moving through different parts of the stellar magnetosphere. This motion through the stellar magnetic field could produce changes in a planetary flow within even a single orbit, leading to complicated variations in the Hα line profile. To explain the full range of variation in the line profiles observed from PTFO8-8695, we are forced to invoke accretion onto the star. As discussed above, there is no suggestion of an infrared excess in this star, even out to 24 µm. Feeding the accretion flow with material escaping from a planet naturally explains the lack of an infrared excess, and the profile variability in PTFO8-8695 is at times well matched assuming excess Hα emission that is physically associated with the reported planet around this star. Thus, we find the planet scenario outlined above the most likely explanation for the variations we see in the lines. Above, we crudely estimated that the excess Hα emission would imply an accretion rate of 3 × 10−10 M⊙ yr−1 onto the star. If this accretion rate is fed entirely by a planet of ∼ 7 MJU P as suggested above, the lifetime of the planet would be ∼ 2 × 107 yr. Thus, if the candidate planet orbiting PTFO8-8695 is real, we expect it to evolve substantially over the pre-main sequence lifetime of this star. 5. Summary We have used relatively high time cadence, high spectral resolution optical observations to detect excess Hα emission from the 2 − 3 Myr old WTTS PTFO8-8695. At some times – 22 – these high cadence observations show that the excess emission appears to move in velocity as expected if it were produced by the suspected planetary companion to this young star. We have considered the possibility that the observed excess emission is produced by stellar activity (flares), accretion from a disk, or from a planetary companion; we find the planetary companion to be the most likely explanation. Yu et al. (2015) recently examined additional photometry and spectroscopy of this star in an effort to test the planet hypothesis for this system. They do not favor the hot Jupiter hypothesis, instead suggesting that their data point to either starspots, eclipses by circumstellar dust (fed by either a circumstellar disk or a low mass evaporating planet), or occultations of an accretion hotspot as the most likely explanation for the variations they observe. However, these authors did not observe the strong, variable, excess Balmer emission that we discuss here and their alternatives would not account for it. Above, we discuss the difficulties associated with a stellar activity or disk accretion origin for the Hα variations, and we conclude that an evaporating planet is the best explanation for the variations we observe. While no single model may fit all the data on this star, this may be due to the extreme nature of this object as a very rapidly rotating, magnetically active pre-main sequence star. Therefore, we believe the planetary companion hypothesis is still a viable component of this unique system. If the excess Hα emission we see does come from a planetary companion, the strength of the emission indicates that it arises in an extended volume around the planet, likely fed by mass loss from the planet which is expected to be overflowing its Roche lobe. Interpreting the radial velocity variations of the excess Hα emission as coming from the planet, we place an upper limit on the mass of the star as M∗sin3i = 0.535 ± 0.047 M⊙, while the planet's mass would then be MP sin3i = 4.45 ± 0.34 MJU P . While there is evidence that the orbital inclination of this system varies dramatically due to nodal precession (Barnes et al. 2013; Ciardi et al. 2015), the inclination at the time of our spectroscopic observations can not be too low or the stellar mass would become very inconsistent with its spectral type. This leads to an upper limit for the planet's mass of ∼ 7 MJU P . While the observations presented here are highly suggestive that we have directly observed the spectroscopic signature of a mass losing planet in orbit around this young star, these results are primarily based on relatively low signal-to-noise observations. Further high signal-to-noise, high cadence time resolved Hα observations over several predicted orbits will greatly aid in solidifying the picture outlined in this work. In addition, we caution that theoretical work is likely still needed to determine whether such strong emission could really be produced by a young hot Jupiter undergoing mass loss. LP and CMJ-K wish to acknowledge partial support for this research from the NASA Origins of Solar Systems program through grant number 07-SSO07-86 made to Lowell Obser- – 23 – vatory. CMJ-K also wishes to acknowledge partial support for this work from the National Science Foundation through grant number 1212122 made to Rice University. This work was funded in part through the 2013-2014 NAU/NASA Space Grant Undergraduate Research Internship for Jacob N. McLane. Finally, we wish to thank an anonymous referee for several useful comments that improved the original manuscript. REFERENCES Adams, F. C. 2011, ApJ, 730, 27 Aigrain, S., Hodgkin, S., Irwin, J., et al. 2007, MNRAS, 375, 29 Alencar, S. H. P., Johns-Krull, C. M., & Basri, G. 2001, AJ, 122, 3335 Alencar, S. H. P., Melo, C. H. F., Dullemond, C. P., et al. 2003, A&A, 409, 1037 Artymowicz, P., & Lubow, S. H. 1996, ApJ, 467, L77 Baraffe, I., Alibert, Y., Chabrier, G., & Benz, W. 2006, A&A, 450, 1221 Baraffe, I., Selsis, F., Chabrier, G., et al. 2004, A&A, 419, L13 Barnes, J. W., van Eyken, J. C., Jackson, B. K., Ciardi, D. R., & Fortney, J. J. 2013, ApJ, 774, 53 Basri, G., Johns-Krull, C. M., & Mathieu, R. D. 1997, AJ, 114, 781 Bertout, C. 1989, ARA&A, 27, 351 Bodenheimer, P., Hubickyj, O., & Lissauer, J. J. 2000, Icarus, 143, 2 Boss, A. P. 1997, Science, 276, 1836 Boss, A. P. 1998, ApJ, 503, 923 Bowler, B. P., Liu, M. C., Shkolnik, E. L., & Dupuy, T. J. 2013, ApJ, 774, 55 Bouvier, J., Alencar, S. H. P., Harries, T. J., Johns-Krull, C. M., & Romanova, M. M. 2007, Protostars and Planets V, 479 Briceno, C., Calvet, N., Hern´andez, J., et al. 2005, AJ, 129, 907 Cameron, A. G. W. 1978, Moon and Planets, 18, 5 – 24 – Carroll, T. A., Strassmeier, K. G., Rice, J. B., Kunstler, A. 2012, A&A, 548, AA95 Cauley, P. W., Redfield, S., Jensen, A. G., et al. 2015, arXiv:1507.05916 Charbonneau, D., Brown, T. M., Burrows, A., & Laughlin, G. 2007, Protostars and Planets V, 701 Ciardi, D. R., van Eyken, J. C., Barnes, J. W., et al. 2015, ApJ, in press Cochran, W. D., Hatzes, A. P., & Paulson, D. B. 2002, AJ, 124, 565 Cody, A. M., Tayar, J., Hillenbrand, L. A., Matthews, J. M., & Kallinger, T. 2013, AJ, 145, 79 Cody, A. M., Stauffer, J., Baglin, A., et al. 2014, AJ, 147, 82 Crockett, C. J., Mahmud, N. I., Prato, L., et al. 2011, ApJ, 735, 78 Crockett, C. J., Mahmud, N. I., Prato, L., et al. 2012, ApJ, 761, 164 Delorme, P., Gagn´e, J., Girard, J. H., et al. 2013, A&A, 553, L5 Dodson-Robinson, S. E., Veras, D., Ford, E. B., & Beichman, C. A. 2009, ApJ, 707, 79 Durisen, R. H., Boss, A. P., Mayer, L., et al. 2007, Protostars and Planets V, 607 Eason, E. L. E., Giampapa, M. S., Radick, R. R., Worden, S. P., & Hege, E. K. 1992, AJ, 104, 1161 Ehrenreich, D., Bourrier, V., Wheatley, P. J., et al. 2015, Nature, 522, 459 Esposito, M., Guenther, E., Hatzes, A. P., & Hartmann, M. 2006, Tenth Anniversary of 51 Peg-b: Status of and prospects for hot Jupiter studies, 127 Fang, M., van Boekel, R., Wang, W., et al. 2009, A&A, 504, 461 Fern´andez, M., Stelzer, B., Henden, A., et al. 2004, A&A, 427, 263 Fischer, D. A., & Valenti, J. 2005, ApJ, 622, 1102 France, K., Schindhelm, E., Herczeg, G. J., et al. 2012, ApJ, 756, 171 Gahm, G. F. 1990, Flare Stars in Star Clusters, Associations and the Solar Vicinity, 137, 193 Giampapa, M. S., Basri, G. S., Johns, C. M., & Imhoff, C. 1993, ApJS, 89, 321 – 25 – Gillen, E., Aigrain, S., McQuillan, A., et al. 2014, A&A, 562, A50 Gonzalez, G. 1997, MNRAS, 285, 403 Gu, P.-G., Lin, D. N. C., & Bodenheimer, P. H. 2003, ApJ, 588, 509 Guenther, E. W., & Ball, M. 1999, A&A, 347, 508 Gunther, R., & Kley, W. 2002, A&A, 387, 550 Hamilton, C. M., Johns-Krull, C. M., Mundt, R., Herbst, W., & Winn, J. N. 2012, ApJ, 751, 147 Hatzes, A. P. 1995, ApJ, 451, 784 Hawley, S. L., Gizis, J. E., & Reid, I. N. 1996, AJ, 112, 2799 Hern´andez, J., Calvet, N., Briceno, C., et al. 2007, ApJ, 671, 1784 Hillenbrand, L. A., & White, R. J. 2004, ApJ, 604, 741 Hinkle, K., Wallace, L., Valenti, J., & Harmer, D. 2000, Visible and Near Infrared Atlas of the Arcturus Spectrum 3727-9300 A ed. Kenneth Hinkle, Lloyd Wallace, Jeff Valenti, and Dianne Harmer. (San Francisco: ASP) ISBN: 1-58381-037-4, 2000. Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2005, Icarus, 179, 415 Hu´elamo, N., Figueira, P., Bonfils, X., et al. 2008, A&A, 489, L9 Huerta, M., Johns–Krull, C. M., Prato, L., Hartigan, P., & Jaffe, D. T. 2007, ApJ, 678, 472 Ireland, M. J., Kraus, A., Martinache, F., Law, N., & Hillenbrand, L. A. 2011, ApJ, 726, 113 Jensen, E. L. N., Dhital, S., Stassun, K. G., et al. 2007, AJ, 134, 241 Johns, C. M., & Basri, G. 1995, AJ, 109, 2800 Johns, C. M., & Basri, G. 1995, ApJ, 449, 341 Johns–Krull, C. M. 2007, ApJ, 664, 975 Johns–Krull, C. M., & Basri, G. 1997, ApJ, 474, 433 Jones, B. F., Fischer, D. A., & Stauffer, J. R. 1996, AJ, 112, 1562 – 26 – Kraus, A. L., & Ireland, M. J. 2012, ApJ, 745, 5 Kraus, A. L., Ireland, M. J., Cieza, L. A., et al. 2014, ApJ, 781, 20 Kuiper, G. P. 1951, 50th Anniversary of the Yerkes Observatory and Half a Century of Progress in Astrophysics, 357 Kurosawa, R., & Romanova, M. M. 2013, MNRAS, 431, 2673 Lafreni`ere, D., Jayawardhana, R., & van Kerkwijk, M. H. 2008, ApJ, 689, L153 Lai, D., Helling, C., & van den Heuvel, E. P. J. 2010, ApJ, 721, 923 Lammer, H., Selsis, F., Ribas, I., et al. 2003, ApJ, 598, L121 Levison, H. F., Morbidelli, A., Gomes, R., & Backman, D. 2007, Protostars and Planets V, 669 Linsky, J. L., Yang, H., France, K., et al. 2010, ApJ, 717, 1291 Lissauer, J. J., & Stevenson, D. J. 2007, Protostars and Planets V, 591 Luhman, K. L., Wilson, J. C., Brandner, W., et al. 2006, ApJ, 649, 894 Mahmud, N. I., Crockett, C. J., Johns–Krull, C. M., et al. 2011, ApJ, 736, 123 Marois, C., Macintosh, B., Barman, T., et al. 2008, Science, 322, 1348 Mathieu, R. D., Stassun, K., Basri, G., et al. 1997, AJ, 113, 1841 Matsakos, T., Uribe, A., Konigl, A. 2015, A&A, 578, A6 Mayor, M., & Queloz, D. 1995, Nature, 378, 355 Mekkaden, M. V., Pukalenthi, S., Muneer, S., & Bastian, A. B. 2005, Bulletin of the Astro- nomical Society of India, 33, 433 Miller, A. A., Irwin, J., Aigrain, S., Hodgkin, S., & Hebb, L. 2008, MNRAS, 387, 349 Murray-Clay, R. A., Chiang, E. I., & Murray, N. 2009, ApJ, 693, 23 Neuhauser, R., Errmann, R., Berndt, A., et al. 2011, Astronomische Nachrichten, 332, 547 Neuhauser, R., Guenther, E. W., Wuchterl, G., et al. 2005, A&A, 435, L13 Neuhauser, R., Mugrauer, M., Seifahrt, A., et al. 2008, A&A, 484, 281 – 27 – Nguyen, D. C., Brandeker, A., van Kerkwijk, M. H., & Jayawardhana, R. 2012, ApJ, 745, 119 Oliveira, J. M., Unruh, Y. C., Foing, B. H., & MUSICOS 96 Collaboration 1998, Ap&SS, 261, 143 Paulson, D. B., & Yelda, S. 2006, PASP, 118, 706 Papaloizou, J. C. B., Nelson, R. P., Kley, W., Masset, F. S., & Artymowicz, P. 2007, Proto- stars and Planets V, 655 Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62 Prato, L., Huerta, M., Johns-Krull, C. M., et al. 2008, ApJ, 687, L103 Ramsey, L. W., Adams, M. T., Barnes, T. G., et al. 1998, Proc. SPIE, 3352, 34 Rice, J. B., Strassmeier, K. G., & Kopf, M. 2011, ApJ, 728, 69 Saar, S. H., & Donahue, R. A. 1997, ApJ, 485, 319 Santos, N. C., Israelian, G., & Mayor, M. 2004, A&A, 415, 1153 Schmidt, S. J., West, A. A., Hawley, S. L., & Pineda, J. S. 2010, AJ, 139, 1808 Schmidt, T. O. B., Neuhauser, R., Seifahrt, A., et al. 2008, A&A, 491, 311 Setiawan, J., Henning, T., Launhardt, R., Muller, A., Weise, P., Kurster, M. 2008, Nature, 451, 38 Setiawan, J., Weise, P., Henning, Th., Launhardt, R., Muller, A., & Rodmann, J. 2007, ApJ, 660, L145 Shu, F., Najita, J., Ostriker, E., et al. 1994, ApJ, 429, 781 Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593 Skelly, M. B., Donati, J.-F., Bouvier, J., et al. 2010, MNRAS, 403, 159 Trammell, G. B., Arras, P., & Li, Z.-Y. 2011, ApJ, 728, 152 Tull, R. G., MacQueen, P. J., Sneden, C., & Lambert, D. L. 1995, PASP, 107, 251 Tull, R. G. 1998, Proc. SPIE, 3355, 387 Valenti, J. A. 1994, Ph.D. Thesis, University of California, Berkeley – 28 – van Eyken, J. C., Ciardi, D. R., Rebull, L. M., et al. 2011, AJ, 142, 60 van Eyken, J. C., Ciardi, D. R., von Braun, K., et al. 2012, ApJ, 755, 42 Vidal-Madjar, A., Lecavelier des Etangs, A., D´esert, J.-M., et al. 2003, Nature, 422, 143 Vogt, S. S., Allen, S. L., Bigelow, B. C., et al. 1994, Proc. SPIE, 2198, 362 West, A. A., Morgan, D. P., Bochanski, J. J., et al. 2011, AJ, 141, 97 Yu, L., Winn, J. N., Gillon, M., et al. 2015, ApJ, 812, 48 This preprint was prepared with the AAS LATEX macros v5.2. – 29 – Table 1. McDonald Observing Log for 15 November 2013 UT Excess Hα Timea S/Nb Weq (A) Excess Hα Pred. Planet σD (km s−1) RV (km s−1) Phasec RV (km s−1)c Excess Hα 05:43 08:07 08:53 09:44 10:32 11:17 3.2 4.3 4.7 5.7 5.1 4.4 8.8 ± 0.3 9.2 ± 0.2 7.3 ± 0.2 5.6 ± 0.2 8.2 ± 0.2 8.2 ± 0.2 86.5 120.3 104.4 59.3 84.8 80.9 189.6 ± 9.4 34.4 ± 4.0 −58.8 ± 4.2 −156.5 ± 6.1 −175.4 ± 5.9 −190.6 ± 7.7 0.282 0.505 0.577 0.656 0.730 0.800 195.1 -6.2 -98.2 -164.8 -197.1 -189.0 aUT time at the midpoint of the exposure. bSignal-to-noise per pixel in the continuum near Hα. cBased on ephemeris in van Eyken et al. (2012). – 30 – Table 2. Kitt Peak Observing Log UT Ind. Exp. Timea Nexp Time (s) Excess Hα S/Nb Weq (A) Excess Hα Pred. Planet σD (km s−1) RV (km s−1) Phasec RV (km s−1)c Excess Hα 04:33 05:16 06:02 06:46 07:29 08:13 09:14 09:57 10:39 11:21 11:55 04:35 05:23 06:11 06:55 07:42 08:21 09:00 09:47 10:38 11:27 04:25 04:48 05:27 05:53 3 3 3 3 3 3 3 3 3 3 2 3 3 3 3 3 2 3 3 3 3 1 1 1 1 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 600 7.2 6.9 6.8 7.7 7.8 5.8 8.7 7.4 7.9 7.2 6.1 5.5 7.4 6.2 2.6 3.1 6.4 4.1 6.8 5.5 6.8 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 8 Dec 2012 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 9 Dec 2012 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 10 Dec 2012 1200 1200 1200 1200 4.4 0.3 ± 0.2 5.2 −0.1 ± 0.2 1.8 ± 0.2 7.1 7.3 3.6 ± 0.2 · · · · · · 83.4 92.3 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 57.3 ± 50.0 99.4 ± 15.2 0.484 0.551 0.621 0.689 0.756 0.824 0.919 0.986 0.051 0.115 0.169 0.719 0.792 0.867 0.935 0.008 0.068 0.128 0.201 0.280 0.356 0.902 0.931 0.968 0.026 20.0 -62.7 -137.2 -184.7 -199.0 -178.0 -97.0 -17.5 62.7 131.7 173.9 -195.3 -192.2 -147.7 -79.1 10.0 82.5 143.4 189.7 195.6 156.6 -115.0 -83.6 -39.8 32.4 – 31 – Table 2-Continued UT Ind. Exp. Timea Nexp Time (s) Excess Hα S/Nb Weq (A) Excess Hα Pred. Planet σD (km s−1) RV (km s−1) Phasec RV (km s−1)c Excess Hα 06:15 06:45 07:12 07:34 08:03 08:25 08:47 1 1 1 1 1 1 1 1200 1200 1200 1200 1200 1200 1200 8.5 7.2 8.1 7.0 6.0 7.1 7.6 4.7 ± 0.2 6.4 ± 0.2 5.7 ± 0.2 4.7 ± 0.2 3.8 ± 0.2 2.7 ± 0.2 2.6 ± 0.2 111.4 101.6 97.2 88.6 86.4 62.3 69.1 174.0 ± 11.5 272.4 ± 10.9 265.9 ± 10.0 251.7 ± 12.6 245.9 ± 17.2 202.4 ± 16.8 198.8 ± 16.3 0.068 0.102 0.149 0.225 0.269 0.303 0.337 82.5 160.3 185.1 196.6 197.7 188.2 170.1 aUT time at the midpoint of exposure(s). bSignal-to-noise per pixel in the continuum near Hα. cBased on ephemeris in van Eyken et al. (2012). – 32 – 25 20 15 10 5 0 −500 0 Velocity (km/s) 0.499 4.918 7.127 22.699 112.174 127.792 130.040 132.274 134.534 500 l x u F e v i t a l e R Fig. 1.- The continuum normalized Hα profiles of PTFO8-8695 obtained with the HET (top 4 profiles) and with Keck (bottom 5 profiles), as part of the discovery study of van Eyken et al. (2012). The orbital phase (with 0.0 the midpoint of the first transit to precede the first HET observation) at the midpoint of the observation is given with each profile. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. There is clear variability in the Hα emission. – 33 – l x u F e v i t a l e R 20 15 10 5 0 −500 0 Velocity (km s−1) 05:43 08:07 08:53 09:44 10:32 11:17 500 Fig. 2.- The 6 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 15 November 2013 UT at McDonald Observatory are shown with the UT time of the midpoint of each observation given. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit – 34 – Relative Flux − 1 0 1 2 3 4 5 6 − 5 0 0 V e l o c i t y ( k m s − 1 ) 0 5 0 0 Fig. 3.- The black histogram shows our estimate of the stellar (chromospheric) Hα emission produced by the WTTS PTFO8-8695 in the continuum normalized spectrum. The dashed line shows a spectrum of AD Leo (dM3e) rotationally broadened to the same vsini as PTFO8- 8695. The smooth red profile shows the spectrum of AD Leo in which the Hα emission has – 35 – l x u F e v i t a l e R 20 15 10 5 0 −500 0 Velocity (km s−1) 05:43 08:07 08:53 09:44 10:32 11:17 500 Fig. 4.- The 6 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 15 November 2013 UT with the stellar (chromospheric) component subtracted out are plotted, again with the UT time of the midpoint of each observation given. The thick, red vertical line on each profile again marks the expected velocity position of the planetary companion – 36 – 30 25 20 15 10 5 0 l x u F e v i t a l e R −500 0 Velocity (km s−1) 4:33 5:16 6:02 6:45 7:29 8:13 9:14 9:57 10:39 11:21 11:55 500 Fig. 5.- The 11 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 8 December 2012 UT at Kitt Peak Observatory are shown in the black histograms with the UT time (midpoint) of each observation. The smooth red curve shows the average Hα profile from this night. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. No clear excess emission is seen in the line profile on this night. – 37 – 30 25 20 15 10 5 0 l x u F e v i t a l e R −500 0 Velocity (km s−1) 4:35 5:22 6:11 6:55 7:42 8:21 9:00 9:47 10:38 11:27 500 Fig. 6.- The 10 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 9 December 2012 UT at Kitt Peak Observatory are shown in the black histograms with the UT (midpoint) time of each observation. The smooth red curve shows the average Hα profile from this night. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. No clear excess emission is seen in the line profile on this night. – 38 – 30 25 20 15 10 5 0 l x u F e v i t a l e R −500 0 Velocity (km s−1) 4:25 4:48 5:27 5:53 6:15 6:45 7:12 7:34 8:03 8:25 8:47 500 Fig. 7.- The 11 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 10 December 2012 UT at Kitt Peak Observatory are plotted with the UT time (midpoint) of each observation. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. Starting at UT 5:53 there is clear excess Hα emission coincident in velocity space at the radial velocity of the planet. – 39 – 30 25 20 15 10 5 0 l x u F e v i t a l e R −500 0 Velocity (km s−1) 4:25 4:48 5:27 5:53 6:15 6:45 7:12 7:34 8:03 8:25 8:47 500 Fig. 8.- The 11 observed continuum normalized Hα profiles of PTFO8-8695 obtained on 10 December 2012 UT with the stellar (chromospheric) component subracted out are plotted. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. Starting at UT 5:53 there is clear excess Hα emission coincident in velocity space at the radial velocity of the planet. Some weak emission also appears to be present at UT 5:27. – 40 – 25 20 15 10 5 0 −500 0 Velocity (km/s) 0.499 4.918 7.127 22.699 112.174 127.792 130.040 132.274 134.534 500 l x u F e v i t a l e R Fig. 9.- The continuum normalized Hα profiles of PTFO8-8695 obtained with the HET (top 4 profiles) and with Keck (bottom 5 profiles) from Figure 1 after an estimate of the stellar component has been subtracted off. The thick, red vertical line on each profile marks the expected velocity position of the planetary companion detected by van Eyken et al. (2012) from transit observations of this WTTS. – 41 – Radial Velocity (km s−1) − 2 0 0 − 1 0 0 0 1 0 0 2 0 0 0 . 0 0 2 . 0 . 4 P h a s e 0 . 6 0 8 . 1 0 . Fig. 10.- Solid symbols show the measured RV of the excess Hα emission from PTFO8- 8695, including uncertainties, as a function of orbital phase using the ephemeris from van Eyken et al. (2012). Circles show data from McDonald Observatory and squares show data from Kitt Peak National Observatory. The dashed line shows the predicted planetary velocity curve using the ephemeris from van Eyken et al. (2012). The dash-dot line shows the predicted RV curve from van Eyken et al. (2012) shifted in phase to account for the revised transit center epoch determined from Spitzer data by Ciardi et al. (2015). The solid line is a circular orbit fit to all the measured RV points. The dotted line is a full Keplerian fit to all the measured RV points, including the transit midpoint time as a constraint (see text). Finally, the dash-triple dot line is a full Keplerian fit to only the McDonald RV points, still using the transit midpoint time as a constraint.
1602.00751	1	1602	2016-02-02T00:11:40	Effect of Pressure Broadening on Molecular Absorption Cross Sections in Exoplanetary Atmospheres	[ "astro-ph.EP" ]	Spectroscopic observations of exoplanets are leading to unprecedented constraints on their atmospheric compositions. However, molecular abundances derived from spectra are degenerate with the absorption cross sections which form critical input data in atmospheric models. Therefore, it is important to quantify the uncertainties in molecular cross sections to reliably estimate the uncertainties in derived molecular abundances. However, converting line lists into cross sections via line broadening involves a series of prescriptions for which the uncertainties are not well understood. We investigate and quantify the effects of various factors involved in line broadening in exoplanetary atmospheres - the profile evaluation width, pressure versus thermal broadening, broadening agent, spectral resolution, and completeness of broadening parameters - on molecular absorption cross sections. We use H$_2$O as a case study as it has the most complete absorption line data. For low resolution spectra (R$\lesssim$100) for representative temperatures and pressures (T $\sim$ 500K-3000K, P$\lesssim$1 atm) of H$_2$-rich exoplanetary atmospheres we find the median difference in cross sections ($\delta$) introduced by various aspects of pressure broadening to be $\lesssim$1\%. For medium resolutions (R$\lesssim$5000), including those attainable with JWST, we find that $\delta$ can be up to 40\%. For high resolutions (R$\sim$10$^5$) $\delta$ can be $\gtrsim$100\%, reaching $\gtrsim$1000\% for low temperatures (T$\lesssim$500K) and high pressures (P$\gtrsim$1 atm). The effect is higher still for self broadening. We generate a homogeneous database of absorption cross sections of molecules of relevance to exoplanetary atmospheres for which high temperature line lists are available, particularly H$_2$O, CO, CH$_4$, CO$_2$, HCN, and NH$_3$.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1–24 (0000) Printed 3 February 2016 (MN LATEX style file v2.2) Effect of Pressure Broadening on Molecular Absorption Cross Sections in Exoplanetary Atmospheres Christina Hedges∗ & Nikku Madhusudhan† Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge, CB3 0HA, UK Accepted 2016 January 29. Received 2016 January 29; in original form 2015 December 3 ABSTRACT Spectroscopic observations of exoplanets are leading to unprecedented constraints on their atmospheric compositions. However, molecular abundances derived from spec- tra are degenerate with the absorption cross sections which form critical input data in atmospheric models. Therefore, it is important to quantify the uncertainties in molecular cross sections to reliably estimate the uncertainties in derived molecular abundances. However, converting line lists into cross sections via line broadening in- volves a series of prescriptions for which the uncertainties are not well understood. We investigate and quantify the effects of various factors involved in line broadening in exoplanetary atmospheres - the profile evaluation width, pressure versus thermal broadening, broadening agent, spectral resolution, and completeness of broadening parameters - on molecular absorption cross sections. We use H2O as a case study as it has the most complete absorption line data. For low resolution spectra (R(cid:46)100) for representative temperatures and pressures (T ∼ 500K-3000K, P(cid:46)1 atm) of H2-rich exoplanetary atmospheres we find the median difference in cross sections (δ) intro- duced by various aspects of pressure broadening to be (cid:46)1%. For medium resolutions (R(cid:46)5000), including those attainable with JWST, we find that δ can be up to 40%. For high resolutions (R∼105) δ can be (cid:38)100%, reaching (cid:38)1000% for low temperatures (T(cid:46)500K) and high pressures (P(cid:38)1 atm). The effect is higher still for self broadening. We generate a homogeneous database of absorption cross sections of molecules of rele- vance to exoplanetary atmospheres for which high temperature line lists are available, particularly H2O, CO, CH4, CO2, HCN, and NH3. Key words: planetary systems — planets and satellites: atmospheres — methods: laboratory: molecular 1 INTRODUCTION In recent years it has become possible to observe high- precision atmospheric spectra of a variety of exoplanets de- tected via transits, direct imaging and Doppler surveys (see e.g. review by Madhusudhan et al. 2014b). For example, high-precision observations with the HST Wide Field Cam- era 3 (WFC3) in the near-infrared (1.1-1.7 µm) has led to unambiguous detections of H2O in several hot Jupiter at- mospheres (Deming et al. 2013; McCullough et al. 2014; Kreidberg et al. 2014b; Madhusudhan et al. 2014a). De- spite the modest resolution (R∼10-100; depending on spec- tral binning), the high photometric precisions of HST WFC3 spectra have allowed unprecedented constraints on the H2O abundances in these atmospheres (e.g. Kreidberg et al. ∗E-mail: [email protected] †E-mail: [email protected] c(cid:13) 0000 RAS 2014b; Madhusudhan et al. 2014a). On the other hand, it has also become possible to detect molecules such as H2O and CO in atmospheres of hot Jupiters orbiting bright stars using very high resolution (R = 105) infrared Doppler spec- troscopy with large ground-based facilities (Snellen et al. 2010; Brogi et al. 2012). Complementary to short period exoplanets, high-resolution and high-precision spectra have also been reported for young giant planets on large orbital separations discovered via direct imaging from ground lead- ing to detections of several key molecules including H2O, CO and CH4 (Konopacky et al. 2013). While these ob- servational advancements are already leading to detailed constraints on the chemical compositions and physical pro- cesses in exoplanetary atmospheres, the field will be fur- ther revolutionised with upcoming facilities including JWST (R ∼ 1000 − 3000) in space and several large ground-based facilities such as the E-ELT (R ∼ 105). Central to the interpretation of such observations, how- 2 Hedges and Madhusudhan ever, is the accuracy of the fundamental inputs to atmo- spheric models. The opacity contributing to a spectrum due to a given chemical species is proportional to the product of its absorption cross section and the molar abundance of that species. Therefore, the chemical abundances derived from at- mospheric spectra, using standard retrieval codes (e.g. Mad- husudhan & Seager 2009; Madhusudhan et al. 2011a; Line et al. 2013; Benneke & Seager 2012; Lee, Fletcher & Irwin 2012; Waldmann et al. 2015), are directly degenerate with the absorption cross sections. In conventional atmospheric models the absorption cross sections are fixed as fundamen- tal inputs and hence the uncertainties on derived chemical abundances assume no uncertainties in cross sections. How- ever, with improving data quality, both in precision and res- olution, it becomes imperative to examine the uncertainties in these model inputs. Ultimately the precision of a derived molecular abundance will be limited by the accuracy of its wavelength-dependent cross sections as they are assumed as model inputs. Significant progress has been made in recent years to generate molecular absorption line lists for molecules of relevance to exoplanetary atmospheres. While traditionally molecular line databases such as HITRAN (Rothman et al. 2006) provided data for a large compendium of molecules, the data was typically available for terrestrial applications of temperatures (cid:46) 300 K. Since exoplanetary atmospheres that are observable with current instruments span much higher temperatures (∼600 - 3000 K) it has become necessary to generate high temperature line lists for many molecules. Therefore, several new high-temperature molecular line lists have been reported in recent years. For example, the recently revised HITEMP database (Rothman et al. 2010) provides a compilation of theoretical and experimental high temper- ature line lists and line broadening parameters for impor- tant molecules such as H2O, CO, and CO2. More recently, and more extensively, the ExoMol database (Tennyson & Yurchenko 2012) has reported high temperature theoretical line lists for numerous molecules of relevance to exoplan- etary atmospheres, including the largest line lists for CH4 and NH3 known to date (e.g. Yurchenko et al. 2013). Molecular absorption cross sections are generated from transition line lists by incorporating the appropriate line broadening and binning to the required resolution. Several factors can cause line broadening in exoplanetary atmo- spheres, such as thermal Doppler broadening and pressure broadening. However, while detailed high temperature line lists are becoming available for several molecules, there is still a lack of detailed line broadening parameters that are required for generating accurate cross sections. Freedman et al. (2008) and Freedman et al. (2014) report opacity calcula- tions using existing high-temperature line lists and highlight the lack of detailed pressure broadening parameters. In par- ticular, current atmospheric observations are most sensitive for giant exoplanets with H2-rich atmospheres, and hence molecular opacities in models need to incorporate pressure broadening due to H2. However, while latest databases such as HITEMP provide broadening data for self and air broad- ening, H2 broadening data is currently unavailable for most molecules. Thus, high-temperature molecular cross sections for several molecules rely on insufficient H2 pressure broad- ening data (Freedman et al. 2008, 2014; Sharp & Burrows 2008). On the other hand, purely theoretical line lists such as ExoMol are computed in zero pressure conditions due to which pressure broadening data is typically unavailable and hence not included while computing cross sections (e.g. Hill, Yurchenko & Tennyson 2013). Finally, even where broad- ening parameters are available, deriving cross sections from line lists involves a series of numerical considerations which can influence the cross sections. In the present work, we investigate the dependence of molecular cross sections on the various factors involved in computing them from molecular line lists including both thermal and pressure broadening. We discuss each aspect of the construction of the cross sections and where errors can be introduced by comparing different data sources, tem- peratures, pressures, broadening agents, evaluation widths and resolutions. Meaningful comparisons can be difficult to make as cross sections span many orders of magnitude and contain many peaks and troughs in features across wave- lengths. As discussed in section 6.1 we find the median per- centage difference to be a useful measure and representative of the change due to pressure broadening as a whole. To as- sess this we also generate a bank of cross sections with only thermal broadening included for comparison with those in- cluding pressure broadening. Using this method, we create a database of molecular cross sections from a range of available line list sources. The cross sections generated span a wide range of pressures (10−4 - 100 atm) and temperatures (300 - 3500 K) that are relevant for exoplanetary atmospheres and sub-stellar objects. We investigate the cross sections over a wide range in spectral resolution (R=102-105) which reflects current and future instruments for atmospheric characterisation of exo- planets as discussed above. There are a wide array of future instruments that we can prepare for. The James Webb Space Telescope (JWST) is scheduled for launch in 2018 and will host several instruments in the infra-red range. NIRSPEC and MIRI spectrographs will hold significance for the char- acterisation of exoplanet atmospheres. NIRSPEC will en- compass the 0.6-5 µm range using three overlapping bands. MIRI will be a particularly broadband instrument stretching from 5-20 µm in wavelength. NIRSPEC will have a spectral resolution ranging up to R=1400-3600 µm for its highest res- olution grating and MIRI will achieve R ≈ 3000 µm. This is a drastic increase in our current capabilities with HST which warrants an investigation into our current modelling inputs. In the distant future we also anticipate the E-ELT to have R=100,000 and be highly capable of atmospheric characterisation for exoplanets. Based on these instruments we present our tests of cross sections at a range of resolu- tions of R=100, 1000, 3000, 10,000 and 100,000. This gives a representative picture of what we can achieve now and in the future. 2 LINE LIST SOURCES There are several line list repositories available where lists of transitions for molecules can be obtained. Those most rel- evant to atmospheric characterisation and that are publicly available are given in Appendix 10.1 including the number of transitions each list contains. These databases are given in a variety of formats but each source contains the key param- eters (e.g. Einstein coefficients, degeneracies, energy levels, c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 3 etc.) and a method for obtaining the line positions (e.g. in wavenumber, ν) and the line intensities S(T ) for a given tem- perature (T). These line data sources cover a wide spectral range, typically spanning the visible to mid-IR (e.g. ∼0.5 µm to ∼30 µm), although this can vary between the different line lists. Line transitions are uniformly spread in frequency space leading to line lists being given in wavenumber rather than wavelength. Sources also range in their completeness with some containing fewer transitions than others. Lack of completeness leads to less reliable cross section data for two reasons; gaps in the wavenumber coverage cause some fea- tures to be missed from the cross section and lines of lower intensity which can contribute significantly to the cross sec- tion are not represented. As such accurate cross sections require the most complete lists of molecular transitions. One of the largest and most well established of all the repositories for molecular line lists is the HITRAN database, which has been updated every few years (Roth- man et al. 1998, 2013). The HITRAN database has been mainly used for terrestrial applications and predominantly includes molecules of importance for the Earth’s atmo- sphere, with temperatures below ∼300 K. Because of this lower temperature approach these data are less applicable for the most observable exoplanetary atmospheres. For cur- rent atmospheric observations of exoplanets we require spec- troscopic data covering a higher temperature range, e.g. for applications to highly irradiated planets and young directly- imaged planets which have a wide range of temperatures going up to 3000 K. This need has been met by the newer HITEMP database, also known as HITEMP2010 to distin- guish from an earlier version, which contains fewer molecules but many more transitions for each (Rothman et al. 2010). HITEMP currently covers OH, NO, CO, CO2 and H2O which are particularly useful for hot Jupiter atmospheres, accurate up to temperatures of 4000K. More recently, the ExoMol database has begun addressing the deficit of data available for molecules of astrophysical importance at high temperatures (Tennyson & Yurchenko 2012). 3 LINE BROADENING The diversity of physical conditions in exoplanetary atmo- spheres can lead to different types and degree of line broad- ening. Exoplanetary atmospheres span a wide range of tem- peratures (∼400 - 3000 K) and dynamical parameters (e.g. wind speeds, and orbital and spin rotation rates) ranging from tidally locked close-in planets to young giant planets on wide orbital separations. The two prominent sources of line broadening in planetary atmospheres are thermal (Doppler) broadening and pressure (collisional) broadening (Chamber- lain 1978; Mihalas, Auer & Mihalas 1978; Seager & Deming 2010). Thermal Doppler broadening is caused by the line- of-sight thermal velocity distribution of molecules at a given temperature in the planetary atmosphere. Pressure broad- ening is induced by collisions between chemical species with the collision frequency being a strong function of pressure. Other sources of broadening can be prevalent depending on the planetary properties and observing geometry. In prin- ciple, natural broadening due to the intrinsic uncertainty in energy levels is always present but is expected to be negli- gible compared to other broadening mechanisms discussed c(cid:13) 0000 RAS, MNRAS 000, 1–24 above. Further broadening and shifting of spectral lines can be caused by the planetary rotation and strong winds in the planetary atmosphere, especially for close-in hot Jupiters observed in transmission spectra (Spiegel, Haiman & Gaudi 2007; Miller-Ricci Kempton & Rauscher 2012; Showman et al. 2013). Finally, rotational broadening due to the spin of the planet can also be significant, especially for exoplanets that are not tidally locked such as those on wide orbital sep- arations (Snellen et al. 2014). These sources of broadening can be important on a case-by-case basis. 3.1 Broadening Profiles Opacities used in models of exoplanetary atmospheres typ- ically include thermal and pressure broadening of the spec- tral lines wherever the broadening parameters are available. While thermal broadening is straightforward to include (e.g. Hill, Yurchenko & Tennyson (2013)), the parameters for pressure broadening are sparse as relevant approximations need to be made for each molecule considered (Freedman et al. 2014; Freedman, Marley & Lodders 2008), as discussed in section 3.3. In the present work, we consider both ther- mal and pressure broadening and investigate the effect of the assumed pressure broadening parameters on the result- ing absorption cross sections. Under the assumption of a Maxwell-Boltzmann thermal velocity distribution the Doppler broadening takes the form of a Gaussian profile. On the other hand, pressure broaden- ing is represented by a Lorentzian profile. The Doppler and Lorentzian broadening profiles are given, in wavenumbers in cm−1, as fD(ν − ν0) = √ 1 π γG exp(− (ν − ν0)2 γ2 G ) (1) fP (ν − ν0) = 1 π γL (ν − ν0)2 + γ2 L (2) where, νo is the centroid in wavenumbers, γG we de- fine as the Doppler width and γL is the Lorentzian pressure broadening half-width at half-maximum (HWHM) both in units of cm−1. These are given by (see e.g. Hill, Yurchenko & Tennyson 2013; Rothman et al. 1998): (cid:114) (cid:17)n γG = (cid:16) Tref T γL = (3) (4) 2kBT ν0 c (cid:88) m P b γL,b pb where, P is pressure in atm, T is the temperature in Kelvin, Tref is the reference temperature (usually 296K), pb is the partial pressure of the broadener, n is a tempera- ture scaling factor and γL,b indicates the Lorentzian HWHM due to a specific broadening molecule in units of cm−1/atm. Here kB is the Boltzmann constant, m is the mass of the molecule in grams and c is the speed of light in cm/s. Here, (cid:80) signifies the sum over all the broadening parameters for each broadening medium. Pressure broadening is typically harder to evaluate than thermal broadening for multiple reasons. Firstly, as alluded to above, the line-by-line pressure broadening parameters, 4 Hedges and Madhusudhan n(ν) and γ(ν), are typically unavailable for most molecules under the conditions encountered in exoplanetary atmo- spheres, e.g. high temperatures up to ∼3000 K and varied atmospheric compositions such as H2-rich gas giant atmo- spheres (Freedman et al. 2008,2014). We discuss this further in section 3.3. Secondly, the Lorentzian profile contributes a higher percentage to its extensive wings which can result in a significant amount of the intensity being moved into the wings of the line profile. In cases of extreme broadening this can significantly increase the impact of high intensity tran- sitions far from the line centre and influence cross sections from neighbouring low intensity transitions. Therefore, the profile needs to be treated particularly carefully by sampling the Voigt well to approximate the profile shape with low un- certainties. This ensures appropriate normalisation so that no intensity is lost from binning the profile. Figure 1 shows the spread of both the Lorentzian HWHM (γL,b) and the temperature scaling parameter (n) with the intensity for CO, one of the simplest molecules in the database. We can see discrete levels in each param- eter due to the discrete nature of the J quantum number that the pressure broadening values are generated from in Complex Robert-Bonamy (CRB) calculations (Antony et al. 2006; Robert & Bonamy 1979). The spread is quite narrow in this parameter space leading us to believe that a mean value of each parameter could reasonably approximate the results of a detailed line-by-line treatment of pressure broadening. It is useful to understand where each of the two broad- ening mechanisms contributes most significantly in the pressure-temperature (P -T ) space relevant for exoplanetary atmospheres. However, as the two profiles are disparate with a Gaussian profile containing more information in the core compared to the extended Lorentzian wings it is difficult to compare the two with a common metric. Nevertheless, √ Figure 2 shows a comparison between the Doppler HWHM (γG ln2) and the Lorentzian HWHM over the P -T space of interest to give an approximation of where each profile contributes most significantly. As expected, at low pressures thermal (Gaussian) broadening provides a significant con- tribution to the final profile core, whereas at high pres- sures pressure (Lorentzian) broadening is stronger. Closer to the boundary between these two regimes, both broad- ening mechanisms are likely to contribute significantly to core of the profile. Furthermore, due to the extended wings of the Lorentzian it is generally advisable to consider both broadening contributions even when the Lorentzian HWHM is narrow in comparison to the Gaussian. This is done using the Voigt profile discussed below. As discussed in Ngo et al. (2012) it has been shown through comparisons with experimental data on pressure broadening that there is some deviation in reality from the Voigt profile. This is due to the change in velocity of the broadener particle by the collision with the broadening agent which affects the profile shape, the width and the shift of the line. Here we have used only the standard Voigt profile and have not investigated further in terms of profile shape. It would be possible to change to a more sophisticated profile and regenerate molecular cross sections if it were found to be an important factor. These deviations due to the velocity of the particles undergoing collisions are of the order of a few percent and we do not expect a more physically accurate Voigt profile shape to impact our results. Figure 1. Pressure broadening parameters for the CO line list from HITEMP. We see discrete values of each parameter as the CRB formalism uses the discrete J quantum number. Self and air broadening parameters have both been plotted here and show that in general air broadening values are slightly lower than self broadening. 3.2 Evaluating the Voigt Profile The joint contributions due to thermal and pressure broad- ening are modelled using a Voigt profile which is a convolu- tion of the Gaussian and Lorentzian profiles, given as (cid:90) ∞ −∞ (cid:113) fV (ν − ν0) = (cid:48) − ν0)fL(ν − ν (cid:48) (cid:48) . )dν fG(ν (5) The characteristic width of the Voigt function is inves- tigated by Olivero & Longbothum (1977). Using coefficients from this work we define the Voigt width γV with the ap- proximation γV ≈ 0.5346γL + 0.2166γ2 L + γ2 G. (6) This Voigt width is used in later sections to approximate the width of the combination of the two profile types. The issue of how best to evaluate the Voigt profile is a well known problem. The profile must be calculated accu- rately and quickly for a wide range of Lorentzian and Gaus- sian profiles, corresponding to the wide range of tempera- ture and pressure values, for potentially millions of lines of a given molecule. The two parameters used for generating the profile are u = ν − ν0 γG , a = γL γG (7) where, u is the distance from the profile centroid and a is the ratio of the Lorentzian and Gaussian widths (Zaghloul 2007). To be able to calculate the Voigt function accurately u must be evaluated over many orders of magnitude to en- compass the relevant temperature and pressure region. c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 5 916 from Zaghloul & Ali (2011) which is known to provide accurate results. We find this package gives converged pro- files over our required parameter space with fast computa- tional speeds of less than 2 ms per profile. We use a sampling rate of 6 points per Voigt width, which is much finer than our required final resolution. We evaluate the Voigt profile to 500 Voigt widths around the centroid to accurately capture the information in the Lorentzian wings, as the derived cross sections are critically dependent on this evaluation width (discussed in more detail in section 5.2). 3.3 Availability of Broadening Parameters Computing line broadening, as discussed above, requires broadening parameters for each line for both the thermal and pressure broadening components. For thermal broaden- ing of a spectral line of a given molecule at a given tem- perature the Doppler width (γG) is easily calculated from Equation 3. Several recent works have reported such ther- mal broadened cross sections for several molecules, e.g. Bar- ber et al. (2006), Hill, Yurchenko & Tennyson (2013). Fig- ure 3 shows a comparison between cross sections for H2O generated with only thermal broadening and those gener- ated with a full Voigt profile including both thermal and pressure broadening. It is clear that the Voigt profile has a significant effect on the low intensity lines, and increases the overall continuum of the molecular cross sections, especially where pressure is high. Despite their critical importance pressure broadening parameters are not yet readily available for all molecules of relevance to exoplanetary atmospheres (Freedman et al. 2008,2014). As shown in Equation 4, the parame- ters required for computing pressure broadening are the Lorentzian HWHM (γL) for the required broadeners and the temperature-scaling parameter (n) for each line in the line list. These parameters are hard to determine. Theo- retical calculations of pressure broadening parameters are particularly time consuming and not covered for a wide range of molecules or broadening agents, particularly at high temperatures. Such methods have been explored by Gamache et al. (1997); Gamache, Laraia & Lamouroux (2011) and used to generate the HITRAN database, e.g. Complex Robert-Bonamy (CRB) calculations are used as discussed in Gamache, Lynch & Neshyba (1998); Gamache et al. (2012) where values are also verified experimentally. Molecular line lists in the HITRAN data base do contain pressure broadening parameters for self-broadening and air- broadening, but are typically relevant only to low tempera- ture atmospheres (∼300 K). For high temperature exoplan- etary atmospheres, particularly of H2-rich atmospheres that are most observable, pressure broadening data is still scarce. Typically, state-of-the-art ab initio line lists such as ExoMol are computed under zero pressure conditions due to which the pressure broadening line parameters are not available. On the other hand experimental data are also scarce for the conditions relevant for exoplanetary atmospheres. In recent years, significant efforts have been dedicated towards generating pressure broadening parameters for ex- oplanetary applications, particularly with a focus on broad- ening molecules such as H2 and other molecules of interest outside of Earth applications. Useful parameters for impor- tant molecules can be found in works such as (Li et al. 2015) Figure 2. Comparison of widths of line cores of Gaussian vs Lorentzian profiles in pressure-temperature space. The red (blue) region represents P-T space where the HWHM of a Gaussian (Lorentzian) profile is wider than that of a Lorentzian (Gaussian) profile. The Gaussian profile is wider at low pressures and the Lorentzian at high pressures, as expected. The Voigt function is given as fv(ν, γG, γL) = H(a, u) (8) where, as above γG and γL are the widths of the Gaus- sian and Lorentzian widths and ν denotes wavenumber. Here, H(a, u) = 1 π ae−t2 (u − t)2 + a2 dt. (9) (cid:90) ∞ −∞ From formula 7.4.13 in Abramowitz & Stegun (1964) ye−t2 (x − t)2 + y2 dt = πRw(x + iy), (cid:90) ∞ −∞ where (10) (11) (12) (13) (14) w(z) = e −z2 erf c(−iz), where erfc is the complimentary error function erf c(z) = 1 − erf (z), where erf is the error function (cid:90) z 0 erf (z) = 2√ π −t2 e dt. From this we can see that fv(ν, γG, γL) = Rw(u + ia) The function w(z) is known as the Faddeeva func- tion and here is calculated using the Faddeeva package (S. G. Johnson 2012). Several numerical methods have been proposed to com- pute the Voigt profile accurately and efficiently over different regions of parameter space (Schreier 2011). We implement the Voigt profile using a method based on the complex error function or Faddeeva function (S. G. Johnson 2012). This is a fast and accurate method with the relevant libraries publicly available. The Faddeeva package includes Algorithm c(cid:13) 0000 RAS, MNRAS 000, 1–24 6 Hedges and Madhusudhan and (Faure et al. 2013) where pressure broadening coeffi- cients are made available. In this work we use only air and self broadening provided by HITEMP and HITRAN and the PS 1997 list for water from (Partridge & Schwenke 1997), which contains H2 broadening R. Freedman, personal com- munication). We focus on the H2O case as it is the most well studied molecule currently with a large variety of line list sources and three different broadening molecules available to test. Further, H2O is one of the best measured molecules in exoplanet atmospheres to date. We anticipate expanding this work to cover other molecules such as CO with further broadening agents as it is another molecule of interest, par- ticularly in very high resolution Doppler spectroscopy of hot exoplanetary atmospheres (e.g. Snellen et al. 2013). However, despite these sources there is a lack of both experimental and theoretical high-temperature data on H2 broadening parameters for many molecules of interest for exoplanetary atmospheres, e.g. CH4, CO2, NH3, C2H2, HCN, TiO, CO, etc (Madhusudhan 2012). Currently most molecules that are available with pressure broadening infor- mation have only self broadening and air broadening param- eters. This is beginning to be addressed and more H2 broad- ening parameters are becoming available Wilzewski et al. (2016). There has been a great deal of work in making these parameters accurate and accessible which is made difficult by the sheer volume of lines for which broadening parame- ters are needed. Databases such as HITEMP, HITRAN and GEISA give pressure broadening values for each transition for each molecule. Where values are not available we are forced to turn to what is available in the literature from experiments and from other line lists. Where detailed line- by-line pressure broadening values are not available for a molecule we use the mean of pressure broadening values available from other sources. While this approach is not ideal it is currently the only option in some cases. For example in this work the YT10to10 line list for CH4 is highly com- plete and a useful asset to modelling of exoplanetary atmo- spheres, however pressure broadening values have not been calculated. An alternative HITRAN list has pressure broad- ening components but only a small fraction of the molec- ular transitions are covered and the list is only appropri- ate for room temperature applications. We will discuss the consequences of this method of taking the average broaden- ing in section 6.3 and the impact using a mean broadening value has on the final cross section. Table 3 also shows which molecules have detailed broadening available and which only have mean broadening. For each case where detailed broad- ening parameters are available a mean case has also been investigated in order to make a comparison. Figure 3. Comparison of purely thermally-broadened H2O cross sections (blue) with cross sections including both thermal and pressure broadening using a Voigt profile (red) at the native line spacing of 0.01 cm−1. A combination of the two broaden- ing types brings extensive wings from the Lorentzian component which brings up the level of the continuum. ent broadening parameter or source. If a cross section was required between these points interpolation could be used on the cross sections between the nearest pressure and tem- perature values as outlined in Hill, Yurchenko & Tennyson (2013). In this section, we describe the procedure we use to compute cross sections. 4.1 Line Intensities and Partition Functions Generation of cross sections requires the intensity of each transition to be accurately calculated. Most line list databases give Einstein coefficients for each transition with degeneracies and energies for each state. These can be con- verted into line intensities as (Rothman et al. 2013): Si,j (Tref ) = Ai,j i,j Q(Tref ) 8πcν2 gie−hcEj /kB Tref (1−e−hcνi,j /kB Tref ) (15) where Ai,j is the Einstein coefficient for spontaneous emission for the transition between states i and j, gi is the upper state degeneracy, Ej is the lower state energy in cm−1 and νi,j is the transition frequency between i and j, also in cm−1 and finally h is Planck’s constant. Here, Q(Tref ) is the partition function at the required reference tempera- ture. When an intensity is given at a reference temperature, usually 296 K, it can be converted to an intensity at any temperature using 4 GENERATING CROSS-SECTIONS The combination of molecular line parameters and broaden- ing profiles discussed above allows us to compute the molec- ular absorption cross sections which form inputs to exoplan- etary atmospheric models. We compute molecular cross sec- tions over a wide range of pressures (P ) and temperatures (T ) relevant for exoplanetary atmospheres. Our grid in P -T space is shown in Table 1. For each point in P -T each cross section will usually be calculated multiple times using differ- Si,j (T ) = Si,j (Tref ) Q(Tref ) Q(T ) exp(−hcEj /kB T ) exp(−hcEj /kB Tref ) [1 − exp(−hcνi,j /kB T )] [1 − exp(−hcνi,j /kB Tref )] . (16) the intensity of a transition in units of This gives cm−1/(molecule cm−2). As evident from Eqs. 15 and 16 the partition function scales the line intensities. The partition function gives a measure of how many of the molecules of a gas are in the ground state compared with all other states. This ratio of state populations increases with temperature as it becomes c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 7 more likely to find particles in higher energy states. This is intrinsically linked to the energy of each transition which makes the partition function unique for each molecule. The partition function is given by Q(T ) = −Ej /kB T gje (17) (cid:88) j where gj is the lower state degeneracy and Ej is the lower state energy as described in Eq. 15. However, to calcu- late a partition function this way the spectral information of the molecule must be complete. Missing transitions result in an inaccurate partition function which will not be represen- tative, especially at high temperatures. Information on how partition functions are calculated for ExoMol can be found in Tennyson & Yurchenko (2012). Errors in the partition function may cause discrepancies that are small when con- sidering individual cross sections at particular temperatures but may impact significantly when considering full atmo- spheric models. Models of atmospheres involve temperature- pressure profiles with respect to altitude and so require cross sections at many different temperatures to be included in computing the emergent spectrum. This will cause any er- rors in the cross section from the partition function to be- come compounded by many layers of an inhomogeneous at- mosphere. In the present work, partition functions have either been adopted from existing databases or computed using the TIPS code (Fischer et al. 2003). The ExoMol database provides partition functions for use with each molecular line list. The HITRAN database uses the TIPS code as discussed in Fischer et al. (2003) to create partition functions between 70K-3000K. There can be some discrepancies between par- tition functions from different sources, particularly at high temperatures. Some are only available in a small range of temperatures. Where this is the case we use extrapolation to find the values at higher temperatures. In our case this only affected our highest temperature point of T=3500K. Using a line list at a temperature greater than that recommended by the source is possible by using the cor- rect partition function. However, while at low temperatures many transitions have low enough intensity to be discounted without effecting the final result, as temperature increases these low intensity lines begin to contribute more noticeably. Because of this any low transitions that are missed, (either by an intensity cut off that is too low or from being omitted in the source,) can cause errors in the cross section if molec- ular line lists are used above their recommended temper- atures. These temperatures are given in the last column of Table 3. Here we see with the exception of NH3 and CH4 the ExoMol and HITEMP line lists provide coverage up to and beyond the temperatures used in this work. The HITRAN sources, used primarily for earth applications, are only valid at room temperature. Due to their small number of transi- tions, (shown in Table 5) HITRAN line lists are unreliable at high temperatures even when used with the correct parti- tion function. Nevertheless, these have been included in this work for two reasons; firstly to investigate the dependence of completeness of line list sources on the accuracy of cross sections and, secondly, to include useful molecules that are not covered by any other source. c(cid:13) 0000 RAS, MNRAS 000, 1–24 T (K) - - 300 1000 3000 400 1200 3500 500 1400 600 1600 700 1800 800 2000 900 2500 P (atm) 0.0001 0.001 0.01 0.1 1 10 100 Table 1. Pressure-Temperature grid for cross sections. Values between these points could be interpolated by the user to allow a finer grid. These points have been chosen as they represent observable temperatures and pressures of currently known exo- planets. Wavenumber Range [cm−1] Grid spacing ∆ν [cm−1] 10-100 100-1000 1000-10000 10000-30000 10−5 10−4 10−3 10−2 Table 2. Table from (Hill, Yurchenko & Tennyson 2013) giving the staggered spacing of the grid for used for line mapping. Here we present an alternative adaptive grid spacing which is given in Equation 18. 4.2 Cross-Sections from Line Intensities Cross sections are derived from line intensities by first broad- ening with the appropriate profile. This is followed by bin- ning the resultant cross sections to the desired spectral reso- lution. This is a general approach followed by several recent studies (e.g. Hill, Yurchenko & Tennyson 2013), albeit with minor differences in implementation. Here we discuss our implementation. The differences from other works are dis- cussed in section 5. For a given temperature (T ) and pres- sure (P ), computing the cross sections from line intensities involves three steps as follows. Firstly, the Voigt profile (fv) is computed at a high resolution in order to accurately evaluate each individual line profile as described in section 3.2. The spacing of this fine grid, here referred to as the ‘sub-grid’, is given as ∆ν = γv(ν = 500, T, P, m) 6 (18) where, γv(ν = 500, T, P, m) is the HWHM of a Voigt profile at ν = 500 cm−1. We find this prescription, which samples each Voigt width with 6 points, to provide the req- uisite resolution and accuracy with optimal computational speed, as discussed in more detail in section 5. This very fine sampling gives an accurate representation the Voigt function across the whole wavenumber space. This sub-grid spacing is a function of T , P and molecular mass making it specific to the molecule concerned. Secondly, for a given spectral line, the cross section is computed at each point on the sub-grid described above. The cross section σ(ν) of a transition between states i and j at a certain pressure (P ) and temperature (T ) is given by (19) σi,j,P,T (ν) = Si,j,P,T ≈ Si,j,P,T fv(ν) (cid:82) ∞ (cid:82) νi,j + ∆νc −∞ fv(ν) dν fv(ν) 2 νi,j− ∆νc 2 fv(ν) dν 8 Hedges and Madhusudhan in units of cm−1/molecule where, Si,j,P,T is the line intensity and fv(ν) is the Voigt function with broadening parameters corresponding to the line at the given P and T. νi,j is the wave number of the line centre and ∆νc is the extent of the profile to which the line wings are evaluated. For a given line, we use a ∆νc value of 500 Voigt widths, (250 around the line centroid,) which we find to be optimal, as discussed in section 5. The integral is evaluated up to this cut off and normalises the profile. Evaluating up to this cut off effec- tively folds in the intensity from the missing wings that are not evaluated back into the profile, ensuring no intensity is missed. Finally, the high-resolution cross section profile com- puted for each line as described above is binned to a final cross section grid with a coarser spacing for saving on stor- age space. The final cross section grid spacing is still high resolution at 10−2 cm−1 which corresponds to a resolution of R=100,000 or greater depending on wavelength. For in- stance, at a wavelength of 1 µm this spacing corresponds to a spectral resolution (R = λ/dλ = ν/dν) of 106. When a lower resolution cross section grid is desired, this high- resolution grid can be binned down further in frequency- space or wavelength-space as required. For example, for a given resolution R the grid points in wavelength space can be determined following R = λ/dλ which will give a non- linear grid in λ. The mid points between adjacent bins are selected and all values within these bounds are averaged giv- ing the binned down contribution at each wavelength on the grid. In this work as, in discussed in Rothman et al. (2013) and other works, we apply a cut in intensity to only evaluate the significant lines. This provides a reduction in computa- tion time with minimal effect on accuracy as very low inten- sity lines contribute little to the final cross section even at high resolution. We apply a cut off at 10−30 cm−1/(molecule cm−2) in intensity across all line lists apart from the BYTe and YT10to10 line lists for NH3 and CH4 respectively. Due to their large size the cut off was increased to 10−26 cm−1/molecule cm−1. We find that as both these molecules have many complex transitions our results are unaffected by omitting these low lines across all resolutions covered. Unlike the HITRAN database we do not scale our cross sec- tions by any relative abundances of isotopes based on their terrestrial measurements. This gives each cross section with the molecule at 100% abundance. The complete list of molecules used in this work includ- ing all sources for data is given in Table 3. Many different line list sources were chosen, particularly for water, for com- parison to investigate how completeness effects the resulting cross section. 5 OPTIMAL RESOLUTION AND CUT OFF OF BROADENING PROFILE It is important to consider an appropriately high resolution and extent of the line profile in order to accurately sam- ple the contribution from the profile. Both these properties also influence the computational cost. Therefore, it is de- sirable to adopt optimal values for each of the properties which facilitate both a high enough accuracy and a reason- able computation time. Several recent studies have adopted Molecule Source Broadening Max T (K) H2O H2O H2O H2O CO2 CO2 CO CO OH OH NO NO CH4 CH4 NH3 NH3 HCN HCN C2H2 BT2a HITEMPb HITRANc PS (1997)d HITEMPb HITRANc HITEMPb HITRANc HITEMPb HITRANc HITEMPb HITRANc YT10to10e HITRANc BYTef HITRANc Harrisg HITRANc HITRANc Agent Self, Air Self, Air Self, Air H2 Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air Self, Air 3000K 4000K 296K - 4000K 296K 4000K 296K 4000K 296K 4000K 296K 2000K 296K 1600K 296K 4000K 296K 296K a Barber et al. (2006) b Rothman et al. (2010) c Rothman et al. (2013) d Partridge & Schwenke (1997) e Yurchenko & Tennyson (2014) f Yurchenko, Barber & Tennyson (2011) g Harris et al. (2006) Table 3. Molecules used to generate cross section database and all line list sources, including which broadening agents are given. (Note HITRAN is recommended at room temperature of 296K.) Here we choose many different sources to compare the effect that completion has on our final cross sections. different prescriptions for these parameters in the particular context of molecular cross sections for exoplanetary applica- tions (e.g. Hill, Yurchenko & Tennyson 2013; Grimm & Heng 2015). In this section, we systematically investigate the ef- fect of both the profile grid resolution as well as the extent of the profile wings on the cross sections in an attempt to determine optimal values for these parameters. 5.1 Effect of Profile Grid Resolution In the present work we use a grid resolution that is adaptive with the equivalent width of the Voigt profile for a given line profile as given in Eq 6. This approach allows for op- timal computational time while ensuring high accuracy of the cross sections. In this formulation, the grid in frequency space on which we evaluate the Voigt profile, here referred to as the ‘sub-grid’ , is defined by Eq. 18. In this work we use a minimum sampling of 6 points per Voigt width which was found to be sufficiently accurate based on the investigations carried out below. Evaluating this spacing at ν =500 cm−1 gives a conservative estimate of the width of a Voigt profile where the Gaussian component is narrowest. (See Equation 3.) This also corresponds to a wavelength of 20 µm which is the longest wavelength of interest in the present work and c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 9 the upper wavelength limit of the infrared observations of exoplanetary atmospheres for instruments such as MIRI on JWST. The necessary grid spacing can become very wide at high pressures where the equivalent width of the Voigt pro- file becomes large. Therefore, we place an upper-limit of 0.01 cm−1 on the grid spacing. When evaluating Voigt profiles on this grid the range up to which the profile is calculated is given by ∆νc, as shown in Eq 19, and discussed in detail in section 5.2. Our spacing is coarser than that used in some previous studies but is optimised for computational time and accu- racy in computing cross sections. For example, the spacing used in Hill, Yurchenko & Tennyson (2013), as shown in Ta- ble 2 and referred to here as a staggered grid, is finer than the grid employed here. This is in part due to work from Hill, Yurchenko & Tennyson (2013) concerning only ther- mal broadening where their profile is Gaussian and much narrower requiring finer grid spacing to normalise. However using our coarser, adaptive grid does not cause significant er- rors in the final cross sections. Figure 4 shows a comparison between the spacing of our adaptive grid and the staggered grid for a representative temperature of 1000 K. It can be seen that both grids are much finer than the Voigt widths for each pressure case, (upper panel). While a fine grid spacing gives highly accurate profiles using the staggered grid can lead to unnecessarily high resolution, especially in the limit of high pressures where the profiles become inherently very broad. This can be computationally expensive particularly for high pressures. The grid we propose in Eq. 18, referred to here as an adaptive grid as its spacing with changes in pressure, uses fewer points at higher pressures to overcome this problem while preserving the accuracy. Ensuring the profile is evaluated accurately is impera- tive to achieve an accurate normalisation and neither under nor over represent the line intensity contribution at each grid point. For normalising the profiles we compute the area under the curve using a simple trapezium rule. When a pro- file is only sparsely evaluated this approach will tend to a greater area estimation, producing a lower intensity contri- bution from each line profile after normalisation. This results in a small percentage of “missing” intensity. To test the va- lidity of our approach we analyse this amount of missing intensity in a single profile when calculated using the adap- tive grid compared with the finer, staggered grid from Hill, Yurchenko & Tennyson (2013). This is done for each T and P point. In order to test our adaptive grid and how well it ap- proximates the Voigt profile it is compared with the stag- gered grid using a wide cut off of ∆νc=100 cm−1 wavenum- bers for each pressure and temperature case. Here we take one profile per T and P point only and map to each grid. The profiles are generated in the same way as in section 3.2. Any difference in the profiles will effect the integrated area, which is used for normalisation. We consider the staggered grid to be high enough resolution in all cases that it will produce an accurate area estimation. The comparisons have been conducted over a range of wavelengths but here we se- lect a representative wavelength of λ =2µm for illustration. Figure 5 shows the results of this comparison. The difference in the final output of these grids is very small with (cid:46)0.2% of intensity missing at P<1 atm at all temperatures. At P=10 atm a maximum of 2% of the original intensity is missed at c(cid:13) 0000 RAS, MNRAS 000, 1–24 Figure 4. Comparison of grid spacing in our adaptive grid with that in the staggered grid from Table 2 of Hill, Yurchenko & Ten- nyson (2013) for different pressures. Top: The coloured lines show the Voigt profile widths at different pressures and a representa- tive temperature of 1000 K. The staggered grid spacing is shown in dashed line for reference, demonstrating that the grid spacing is well below the Voigt profile width for pressures down to 10−4 atm. Bottom: The coloured lines show the adaptive grid spac- ing we use. For high pressures ((cid:62)1 atm), the minimum spacing is fixed at 10−2 cm−1 as profiles become very broad. While our grid spacing is coarser than the staggered grid spacing for high pressures the resultant effect on the cross sections is small, as shown in Fig. 5. Note that a single value of broadening width and temperature scaling has been adopted for this figure. low temperatures. The largest differences found are ∼10% for pressures of 100 atm, but given that such high pressures are not directly observable for exoplanetary atmospheres the corresponding differences are less of a concern. We find that in such cases the cut off chosen ∆νc is too narrow for the extreme case of P=100 atm. For the 10−4 to 10 atm pressure range we find the adaptive grid to be very accurate, partic- ularly after binning to the final output grid with a spacing of 0.01 cm−1 which corresponds to a spectral resolution of R=105-106 at λ <10 µm. 5.2 Effect of Profile Evaluation Width In order to accurately account for the contribution of the line wings the broadening profile must be evaluated over a wide enough range centred on the line centre, referred to here as a cut off value (∆νc), as discussed in section 4.2. This cut off value determines the extent of evaluation of the profile as well as its normalisation as described in Eq. 19. The choice of ∆νc has already been noted in the field as an important factor in computing cross sections (Sharp & Burrows 2007; Grimm & Heng 2015). A common approach is to apply a cut 10 Hedges and Madhusudhan Figure 5. Comparison of adaptive grid with staggered grid from Table 2 . Voigt profiles are mapped to a fine grid from equation 18. The integrated area under the profile for the adaptive case is compared with that for the staggered case to produce this result. Only pressures of ∼100 atm are affected significantly by the res- olution of the adaptive grid beyond a few percent. Below 1 atm the difference is ≈ 0.2% or less. off in wavenumber, with values ranging between 10 and 100 cm−1, especially for high pressures as discussed by Sharp & Burrows (2007). Another approach is to take a number of Lorentzian widths from the centroid such as in Grimm & Heng (2015), however this does not take into account the full width of the profile after its convolution with the Gaussian component. We employ a cut off in multiples of Voigt widths, given by Eq 18 and implemented in Eq 19. The ∆νc in this approach adapts with both the Lorentzian and Gaussian components ensuring that the wings of the profile are accounted for in an adaptive manner depending on the broadening conditions. Based on our investigation we find 500 Voigt widths to be sufficient for current applications including JWST-like resolutions and VLT applications with small uncertainties when using the standard Voigt profile. Several cut off values have been investigated to establish the optimal balance between accuracy and computational time. As discussed above we use multiples of Voigt widths to establish a cut off that adapts to the specific profile. When using a short cut off lower intensity lines are underestimated by many orders of magnitude due to the lack of additional intensity from the wings of high intensity neighbours. This leads to the continuum being poorly approximated by short cut offs. When the cut off is increased we maintain the same normalised area within the profile. Because of this the pro- file height becomes slightly shorter as the cut off becomes wider. Due to this effect we find that at low cut off values the cores of very strong lines are slightly over estimated, (by approximately 0.2%) at the native spacing of the grid. The effect of this slight overestimation will reduce drastically with resolution. The underestimation in the profile wings is 10% for H2O at the native spacing. However this underestimation is Figure 6. Comparison of cross sections obtained at the high- est resolution using profile evaluation width (∆νc) of 500 Voigt widths versus 10,000 Voigt widths for H2O at 500K and 0.1 atm. Overestimated and underestimated points are shown in red and blue, respectively. We see there are some points overestimated by more than 0.1%, however no points are overestimated by more than 1%. The underestimations average 10% at this high resolu- tion. confined to the lowest intensity transitions with high inten- sity neighbours which are by their nature confined within high intensity features. As such an underestimation in these points is of less consequence to most applications. Due to the profiles being very broad at high pressures where the Voigt profile width is beyond 0.01 cm−1 the spac- ing becomes fixed to 0.01 cm−1 wavenumbers. This also fixes our evaluation width to 60 cm−1 wavenumbers around the centroid. For high pressures, (P> 10 atm), this is an under- estimate, however such pressures are less useful for atmo- spheric modelling in exoplanets. 5.3 Comparison of Evaluation Widths The wings of the profile will have an affect on any neighbour- ing lines. If wings are not evaluated out to a large enough separation the continuum for neighbouring lines will be un- derestimated. This is particularly important in the case of line lists and cross sections as intensities span many orders of magnitude and so line wings from high intensity transitions can be comparable to the peaks of low intensity neighbours. However the cut off also affects the evaluation of the profile. A cut off that is too close to the centroid will provide poor normalisation and will miss some of the intensity contribu- tion. The cut off value ∆νc is fixed at 500 Voigt widths around the centroid. This is designed so that the evaluation width adapts to both the Gaussian and Lorentzian profiles which is particularly important at extremes of pressure or temperature. This gives 250 Voigt widths around the cen- troid in each direction. ∆νc is increased to 1000 Voigt widths around the centroid at pressures of 1 atm and above. To es- tablish the difference between this method and others in the c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 11 Figure 7. Comparison of the Voigt width cut off from Eq 18 with ∆νc set at 500 Lorentzian widths. For pressures of 0.001 atm and below we see that missing intensity can be as high as ∼50% leading to a loss of information at these crucial low pressures. Figure 8. Comparison of the Voigt width cut off from Eq 18 with a cut off set at 25P wavenumbers (up to a maximum of 100cm−1) around the centroid. For pressures of 0.001 atm and below the differences are ∼50% or greater. field we undertake the following comparisons. Firstly we take a single Voigt profile on our fine, adaptive grid given by Eq. 18 and ∆νc of 500 Lorentzian widths as taken from Grimm & Heng (2015). The Voigt profile is calculated in the same way as given in section 3.2. This is then compared with a single Voigt profile on the same grid with a cut off at 500 Voigt widths, our value of ∆νc. The profiles are left unnor- malised. The missing area from the Lorentzian cut off when compared with the method we present is then evaluated as a percentage of the total profile area. The percentage of missing intensity from this compari- son is shown in Figure 7. It can be seen from the Lorentzian and Voigt width equations that the Voigt width should always cover the same or more wavenumbers and so the Lorentzian width cut off has a smaller integrated area than our approach. The Lorentzian width evaluation fails at low pressures as the thermal Gaussian component becomes much stronger at lower pressures. These low pressures are crucial to un- derstanding the upper atmospheres and are the most likely to effect observations. Any intensity missed in the initial evaluation of the profile then leads to inaccuracies to each individual profile which can then be further compounded by binning several such transitions to the final output grid. An underestimation of the true area under the profile results in an incorrect normalisation and an overestimation of the cross section value for each transition. We see in Figure 7 this overestimation can be 50% for pressures of 0.001 atm. Another approach, taken from Sharp & Burrows (2007), is to apply ∆νc of a certain number of wavenumbers from the centroid. Sharp & Burrows (2007) suggest ∆νc = min(25P, 100) cm−1, where P is pressure in atm. This ap- proach also adapts in pressure. However we find this also provides an underestimation of the area under the Voigt profile at low pressures as shown in Figure 8. Based on these parameters we find our combination of a sparse, adaptive grid with a wide cut off to be an im- c(cid:13) 0000 RAS, MNRAS 000, 1–24 provement on current methods in the field in terms of both accuracy and reduction in number of points which require evaluation. We also find our method to give sufficiently ac- curate profile estimates for the final output grid spacing of 0.01 cm−1 and the instrument resolutions given here in later sections. 5.4 Sub-Lorentzian Shapes In this work, as discussed in Section 3, we use only the standard Voigt profile without modification to account for changes in the shape from other processes, (e.g. line-mixing and collisionally induced velocity changes) such as those dis- cussed in Ngo et al. (2012). Edwards & Strow (1991) and Birnbaum (1979) discuss specifically a sub-Lorentzian shape for pressure broadening where the far line wings are mod- ified by an exponential, reducing their contribution. Cur- rently there are many estimates of the distance from the line centroid where this occurs ranging from 1-30cm−1. This sub- Lorentzian shape at the far line wings can be difficult to esti- mate and contributions to the profile at large separations are not well understood. Changes to the broadening shape used here could be made in the future to investigate whether a change in profile shape significantly alters results.Molecular transitions vary greatly in intensity and this may have a greater effect in the wavelength regions where there is a sharp drop in line intensity (’window regions’). For some molecules with less complex structure, such as CO, these can be very sharp changes. In such cases, having no neigh- bouring lines, a sub-Lorentzian may have some effect on the edges of these regions. Other approaches have included a simple cut off at a given wavenumber from the centroid of the line which would be very similar to the tests performed in Section 5.3. We find that changes in this cut off cause small underestima- tions in the lower intensity transitions across wavelength. However we find our metric, discussed in Section 6.1, to be 12 Hedges and Madhusudhan quite general even accounting for this effect. At most reso- lutions many transitions are binned together mitigating this problem and our metric uses a median over a wide band. 6 EFFECT OF PRESSURE BROADENING ON CROSS SECTIONS In this section we systematically investigate the dependence of molecular cross sections on the various parameters and assumptions involved in implementing pressure broadening with a given line list. Our goal here is to both quantify the uncertainties propagated into cross sections due to various choices involved in their generation as well as to define a quantitatively meaningful approach to make those choices. In order to pursue this here we focus on the H2O molecule which has the most complete line list and broadening data currently available for exoplanetary atmospheres. We inves- tigate the dependence of the cross sections on the follow- ing key factors: (a) pressure and temperature, (b) average versus line-by-line treatment of broadening parameters, (c) spectral resolution (i.e. binning) of the cross sections, and (d) broadening agent. Note that the effect of the parameters of the Voigt line profile, namely the profile resolution and extent of the line wings, were investigated in the previous sections and here we adopt the optimal values and method- ology discussed in sections 4.2 and 5. 6.1 Definition of Change Due to Broadening In order to assess the difference that any given aspect of broadening makes to the cross sections we need to formally define the corresponding change quantitatively. Since the cross sections for any molecule can span many orders of mag- nitude over a given spectral range and features very drasti- cally with wavelength defining a robust metric is challeng- ing. For example, a simple metric such as a ’mean difference’ across the entire spectral range available is often unreliable. On the other hand, focusing on lines with maximum error places undue emphasis on the lowest intensity lines which will see the highest fractional change but would be hard to observe. Conversely, focusing on the highest intensity lines is unreliable because they are not representative of the line population. Therefore, we use the median percentage differ- ence across the entire spectral range as our metric of choice to quantify the change in cross sections due to any aspect of broadening. The median percentage change in cross section for each line is computed as (cid:110) σ − σ0 (cid:111) δ = Median σ0 λ × 100 (20) where σ0 and σ are the cross sections before and after in- corporating the particular pressure broadening prescription, and the median is evaluated for cross sections computed over the entire wavelength range of interest, 0.5 - 20 µm. 6.2 Effect of Pressure and Temperature The effect of pressure broadening on cross sections as a func- tion of pressure and temperature is shown in Figure 9. In the region of interest for exoplanet atmospheres, for pressure Figure 9. Cross sections for H2O at various temperatures and pressures using air broadening at a grid spacing ∆ν of 0.01 cm−1 wavenumbers, the output grid of this database. This corresponds to a resolution up to an order of magnitude higher than R=105 below 10 µm. around 10−4 to 1 atm and temperatures of 500 to 2000K) we find notable changes to the cross section as a function of resolution. For low resolutions (R < 100), we find the differ- ences introduced to cross sections from pressure broadening is <1% across all P and T considered in this work. At res- olutions of R=5000, similar to those expected from JWST, and representative hot Jupiter temperatures of T∼1000K we find δ <1% at P=0.1 atm. At the highest resolution of R=105 we find that for temperatures of 1000K and pressures of 0.1 atm δ =60% for H2O in an H2 atmosphere. For lower temperatures of 500K this can increase dramatically giving δ=1000%. Figure 3.2 shows that at low temperatures (T(cid:46)500K) the wide Lorentzian will have more of an effect than the Gaussian profile. From this we expect to see that cross sec- tions, (and full atmospheric models,) at low temperatures are more effected by pressure broadening than those at high temperatures. Figure 10 shows that, even before propagating through an atmospheric model, we can expect cool targets of ∼500K to be more affected by pressure broadening than hotter targets ∼1000K by 100% or more at the highest res- olutions. At resolution of R=105 we can expect the median c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 13 6.3 Line-by-line versus Mean Broadening Parameters As currently there is a lack of line-by-line broadening param- eters for several molecules it is often unavoidable to rely on sparse broadening data when computing pressure broadened cross sections, as discussed in section 3.3. When only sparse data is available, i.e. broadening parameters are available only for a few lines, in this work representative values for the broadening parameters are chosen based on the avail- able data and the mean applied across all the lines. We find that the difference to cross sections when using this method is up to ∼20% at the output grid spacing of 0.01 cm−1 for all pressures. In this section, we investigate the difference such an approach makes to the cross sections overall compared to cases where broadening parameters are available for all the lines. For purposes of demonstration, we use the latest line list of H2O from the HITEMP database which includes line- by-line broadening parameters with air broadening. In one case, we calculate H2O cross sections over a wide range of temperatures and pressures using detailed line-by-line val- ues for the air-broadening parameters. In another case, we adopt constant values for the broadening parameters av- eraged over the entire line list and apply those values for broadening every line. The median percentage difference in the cross sections derived from the two cases over the en- tire line list at the native spacing of the line list with a grid spacing of 0.01 cm−1 is shown in Figure 11. At this resolu- tion, we find that using mean broadening values can result in cross sections that are inaccurate by up to 20% for observ- able pressures (∼0.1 bar) and low temperatures (T (cid:46)500 K) where pressure broadening is strongest. For lower pressures and higher temperatures the effect is less pronounced in a ‘median’ sense. For lower resolutions, the differences reduce. Therefore, when numerous lines are available to calculate representative average values for the broadening parameters as in the present case then the mean treatment of pres- sure broadening is a reasonable approximation to a detailed treatment for low resolution observations. Figure 11 shows this difference to decrease as pressure increases. In high pressure cases the broadening becomes wide enough such that many profiles begin to overlap. This effectively smooths the information and causes the the dif- ferences in profile shapes to be less distinguishable. From Figure 9 this effect is more clear as we see at high pressures much of the information from the individual transitions is lost. This effect occurs at pressures greater than 0.1 atmo- spheres implying that for high pressures of P(cid:38)1 the detailed broadening parameters for each transition may not affect cross sections as much. Contrary to the above scenario where we can construct a reasonable mean from many broadening parameters, de- tailed line-by-line broadening parameters are required to de- rive accurate cross sections, especially from upcoming high- precision and high-resolution observations. In the above case, while the median error across all the lines is low over- all, individual lines with significant deviation from the mean broadening values can result in more than 100% difference in cross sections which are relevant for interpreting high- resolution observations that rely on detecting specific lines (Snellen et al. 2010, 2014). Secondly, for several molecules Figure 10. Median difference between cross sections derived with Gaussian-only broadening and Voigt broadening in the WFC3 band pass for H2O at a resolution of R=100,000. Here we show the H2 broadening case, though self broadening case gives the largest change to the cross section overall from pressure. For low temperatures the difference dramatically increases as the Gaus- sian component of the Voigt becomes narrower. difference between the two broadening cases, (thermal only and pressure and thermal,) for water to be 100% beyond pressures of 0.1 atm at T=500K as shown in Figure 10. This would lead to uncertainties in molecular abundance of the a similar factor. The implication here is that cooler targets are likely to be more effected by pressure broadening. When ob- serving such cool targets with high resolution instruments pressure broadening will potentially limit the precision on abundance measurements. These cooler targets are likely to have complex atmospheric structure and incorporate dif- ferent chemistry in their upper atmosphere, even including broadening from more complex molecules than molecular hydrogen. However for cool targets there may be other fac- tors obscuring observations such as clouds and hazes. In recent years, observational programs have largely fo- cussed on hotter targets, mostly hot Jupiters, as their thick atmospheres and high temperatures lead to stronger spectral features and their short orbital periods make them easier to observe than other targets. Currently observing lower-mass, cooler planets is proving difficult, however pressure broad- ening may affect us more in the future when observations become more sensitive to such planets. At temperatures of 1000K and beyond we still find an effect from pressure broadening though less strong. For these hotter targets we might expect pressure broadening to have an effect in the 0.1-1 atm pressure regime of around 10-100%. As discussed in section 6.5 this depends also on the broadening agent and we find that self broadening for water is much stronger than H2. c(cid:13) 0000 RAS, MNRAS 000, 1–24 14 Hedges and Madhusudhan Figure 11. Effect of mean versus detailed pressure broadening. The curves show comparisons between cross sections obtained using mean values for all the line broadening parameters versus those obtained using detailed line-by-line broadening parameters. These results are for the highest-resolution spacing of 0.01 cm−1, corresponding to a resolution of 105-106 at λ <10 µm. relevant broadening values are available based on experimen- tal data for only a few lines, the average of which may not be representative of the entire line population, leading to larger inaccuracies than found in the above example. Finally, re- cent and upcoming observations are already sensitive to cool and dense planetary and Brown Dwarf atmospheres (e.g. Fraine et al. 2014; Buenzli et al. 2015) with high-precision observations which necessitate accurate line-by-line broad- ening parameters. Appendix 10.1 shows all molecules that are available from various sources with how many lines each contains which here we use as a proxy for how complete a line list is, with those containing only a few thousands of lines being most unreliable. HITEMP, HITRAN and GEISA have line by line pressure broadening parameters generated but only for air and self broadening, with only HITEMP containing values for the high temperatures relevant for exoplanetary atmospheres. In order to address the issue of pressure broad- ening accurately in the future line lists will need to include broadening due to molecular hydrogen, relevant for giant planets, and be complete to high temperatures. 6.4 Effect of Spectral Resolution One of the most important questions that can be answered in this work is how the difference created by pressure broad- ening to molecular cross sections is influenced by spectral resolution. As discussed earlier, we use H2O as our case study and consider cross sections in the HST G141 band- pass (1.1-1.7 µm). Observations of exoplanetary spectra are conducted over a wide range of spectral resolution, ranging from broadband photometric observations and low resolu- tion spectra (R (cid:46) 100), e.g. with HST, to very high reso- Figure 12. Comparison of molecular cross sections in the WFC3 G141 bandpass and the NIRSpec bandpass at HST-like and JWST-like resolutions, respectively. The cross sections are gen- erated using air broadening for all the molecules. Degeneracies between the specific molecule, the abundance of the molecule and the temperature can be harder to break at lower resolutions, how- ever JWST’s improved wavelength and resolution will help break these degeneracies in future. lution spectra with large ground based facilities (R ∼ 105). Resolution can greatly influence how an observed spectrum can be interpreted and can break the degeneracies between different molecules. Here we discuss the effect resolution has on δ as a function of pressure and temperature. We find that at the highest resolutions and lowest temperatures (R=105 and T=500K) pressure broadening can introduce a differ- ence to the final cross section of δ=1000% for P=0.1 atm. For lower resolutions of R=5000, similar to those that will be attainable with JWST, the differences become much smaller. However at low temperatures (T=500K) and high pressures (P=1 atm) a δ of 40% is found for H2 broadening at R=5000. For low resolution spectra (R(cid:46)100) of exoplanets, that are possible with current instruments, for representative exo- planetary temperatures (T=500K-2500K, P(cid:46)1 atm) and H2 rich atmospheres, we find the median difference in cross sec- tions introduced by various aspects of pressure broadening (δ) to be (cid:46)1%. For illustration, Fig. 12 shows molecular cross sections c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 15 of several molecules binned to the spectral ranges and reso- lutions achievable with HST (WFC3 G102 and G141 grisms, R∼100) and JWST (NIRSpec, R ∼3000). We can see at the higher resolution of JWST it is much easier to break the degeneracies and identify molecules. This is also easier in JWST due to the longer spectral range giving more poten- tial to search for molecules. Figure 13 shows δ as a function of P and T. We find that δ increases with increasing resolution, reaching differ- ences up to 100% or higher for R (cid:38) 104, P (cid:38) 0.1 atm, and T (cid:46) 500 K. For example, as shown in the upper panel of Fig. 13 considering a temperature of 500 K, we find that at a nominal pressure of 1 atm, δ can be as high as 100% for R 104 or more. On the other hand, for the highest resolutions possible today of R ∼ 105, δ (cid:38) 100% even for pressures as low as 0.1 atm. Similarly, the lower panel of Fig. 13 shows the varia- tion in δ with temperature for a nominal pressure of 1 atm, showing δ can be very high ((cid:38) 100%) for T < 1000 K for R >104. Consequently, we find that it is very important for atmospheric models to include pressure broadening when interpreting high-resolution spectra (R(cid:38) 104) of exoplanet atmospheres observable with current and upcoming facili- ties (e.g. VLT, Keck, and E-ELT). Otherwise, the derived molecular abundances will be limited by a minimum uncer- tainty of more than 100% due to inaccurate cross sections. On the other hand, we find that for R < 104, δ is reduced reaching a maximum of ∼ 10% for pressures of relevance to exoplanetary atmospheric observations of P (cid:46) 0.1 atm. Figure 14 shows many slices across these plots with both H2 broadening and self broadening, the latter having an even larger effect on the cross sections than H2 broadening (as discussed in section 6.5). While our results indicate that it is important to in- clude pressure broadening in cross sections for interpreting high resolution observations, it is nevertheless advisable to also include the same for low resolution spectra as well. Even though δ is found to be at a maximum of ∼ 10% for R < 104 P (cid:46) 0.1, low resolution spectra at very high precision with HST and JWST could allow retrieval of molecular abun- dances with uncertainties of a few percent, in which case the 10% uncertainty in the cross sections could become a limiting factor. Secondly, it is to be noted again that δ is a metric of differences only in a median sense while individual lines could potentially contribute higher δ than the median value. Finally, while the current analysis focused on H2O with H2 broadening the same with other molecules could in principle lead to higher δ even at low resolutions which future studies need to investigate. 6.5 Effect of Broadening Agent Another important factor in pressure broadening is the pri- mary broadening agent in the planetary atmosphere. As shown in Eq.4, the broadening agent governs the Lorentzian HWHM of the broadening profile. Here we find that for res- olutions of R=100,000 the difference between broadening in an H2 atmosphere and broadening in an H2O atmosphere can be significant for water features such as those in the WFC3 1-1.7 µm bandpass. At low temperatures of 500K we find δH2 ∼1000% and δH2O ∼10,000% for pressures of 0.1 atm at R=100,000. At higher temperatures, which have c(cid:13) 0000 RAS, MNRAS 000, 1–24 Figure 13. Effect of resolution on median difference between H2O cross sections when pressure broadening is included com- pared with a Gaussian-only case across the WFC3 band pass range of 1-1.7 µm. Here we show H2O broadened by H2 though stronger broadening can be achieved using self broadening as dis- cussed in section 6.5. (Note: Here the individual points of the T,P and R grid have been linearly interpolated over for plotting pur- poses. (Hill, Yurchenko & Tennyson 2013) discusses interpolation between temperature and pressure points.) been shown to reduce the δ found in previous sections, we find that for T=1000K δH2 ∼60% and δH2O ∼300% in the same conditions. We find that across pressure, temperature and resolution the δ from self broadening in H2O is on av- erage 4 times greater than that from H2. Molecular line lists containing pressure broadening data, e.g. HITRAN or HITEMP, typically contain data for self and air as the broadening agents, motivated by the ter- restrial applications which HITRAN was originally intended for. However, for giant planetary atmospheres H2 is the dom- inant broadener and is of particular relevance for studying atmospheres at high spectral resolution and photometric precision. Accurate line-by-line H2 broadening data for high temperatures are still elusive for most molecules of interest though a few molecules have data available, particularly for 16 Hedges and Madhusudhan Figure 14. Effect of resolution at different temperatures and pressures on H2O cross sections in the WFC3 bandpass. Median percentage difference between cross sections evaluated including pressure and cross sections evaluated with a Gaussian-only, thermal broadening. At high resolutions, high pressures and low temperatures there is the largest change to cross section. At the highest resolution of R=100,000 the median difference can be more than 1000% for H2 broadening. We see that H2 broadening of H2O is consistently weaker than self broadening. Resolutions of between R=100 and R=100,000 are shown in colours given by the top right panel. H2O (Partridge & Schwenke 1997) and more recently for CO (Li et al. 2015). Here, we investigate the effect of broadening agent on the median accuracy of molecular cross sections for a rep- resentative case. We consider the case of H2O for which we have line-by-line broadening parameters with H2, self, and air as broadening agents (Partridge & Schwenke 1997, R. Freedman - personal communication). To illustrate the dif- ferences made by changing the broadening agent we have used cases where the molecule is broadened only by a par- ticular molecule self, air, or H2 (i.e. the partial pressure is 1 in each case). Figure 15 shows the median percentage differ- ence in cross sections caused by each of the three scenarios compared to Gaussian-only broadening for an illustrative case with resolutions of 104 and 105 and T = 500 K. Figure 15 shows that it is important to carefully choose broadening agents before generating cross sections for dif- ferent planet types. Firstly, self-broadening can cause sig- nificantly higher δ values compared to air or H2 broadening at observable pressures (P ∼ 0.1 - 1 atm). Secondly, the differences between H2 broadening and air broadening are relatively small in the H2O case. Therefore, while modelling H2-rich atmospheres in the absence of any H2 broadening data for H2O molecules, though not ideal, it is more advis- able to use air broadening than self broadening. On the con- trary, when modelling atmospheres of low-mass exoplanets, e.g. super-Earths that can have volatile-rich atmospheres such as H2O-rich or CO2-rich atmospheres, it is important to use cross sections that are generated with the appropri- ate broadener. For example, for H2O-rich atmospheres self- c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 17 broadening of H2O should be considered in the cross sections rather than air broadening. The effect of broadening agent can be substantial de- pending on other parameters. The effect of the broaden- ing agent is naturally strongest in regions of parameter space where pressure broadening is expected to be strongest, namely at high pressures, low temperatures, and high res- olution. Fig. 14 shows the differences between self and H2 broadening for various parameters. The differences begin to become significant at high pressures (P (cid:38) 0.1 atm) for R (cid:38) 10000 and become substantial even at lower pressures for high resolutions. For very high resolution observations at R (cid:38) 105 self broadening can lead to differences of 1000% in cross sections. This has a much greater effect than H2 broadening over a large range of pressures, making it criti- cal to use broadening data with the appropriate broadening agents for interpreting observations at these resolutions. Fi- nally, the differences in cross sections induced by different broadening agents are particularly strong at lower temper- atures (T (cid:46) 1000 K) as the broadening has a greater effect at these temperatures. Given the wide range of temperatures and pressures of exoplanetary atmospheres that are accessible to current and upcoming observations line-by-line broadening parameters are required in molecular line lists for different broaden- ing agents. In particular, there is a critical need for high- temperature ((cid:38) 500 K) H2 broadening data as the most observable atmospheres are those of hot and giant exoplan- ets with H2-rich atmospheres, for which very high resolution spectra (R ∼ 105) are also being reported. In the mean time, it is advisable to use air broadening where available for such atmospheres because while not ideal it provides closer cross section estimates to H2 broadening for H2O and are an im- provement on incorporating no pressure broadening at all. Future studies would also need to investigate if the same is true for other molecules as and when H2 broadening data become available for those molecules. Finally, a realistic at- mosphere will contain many different molecules and so con- tribute broadening from many different species. For smaller planets we expect atmospheres that are more complex, con- taining more massive molecules with high abundance. This will effect the pressure broadening particularly as the partial pressure will no longer be 1 and there will be contributions from many species. In such cases we expect the contribu- tion to vary depending on the abundance of more massive broadening molecules with greater pressure broadening pa- rameters. Our current work gives an estimate for the most extreme cases of H2O or H2 dominated atmospheres. 7 EFFECT OF PRESSURE BROADENING ON TRANSMITTANCE A thorough investigation of the effect of pressure broaden- ing on fully modelled exoplanetary spectra is non-trivial and beyond the scope of the present study. Many factors such as the inhomogeneous P-T structure and composition of the atmosphere will determine the final spectrum. Nevertheless, as a simpler exercise, here we nominally asses the effect of pressure broadening on the transmittance in a fiducial at- mosphere represented by a uniform column of gas. As in other parts of this study we use H2O as the only absorbing c(cid:13) 0000 RAS, MNRAS 000, 1–24 Figure 15. Effect of broadening agent on cross sections. The curves show median difference pressure broadening induces when compared with Gaussian-only cross sections when broadening agent is changed. Here the H2O molecule is considered at R=10,000 and R=100,000 at a temperature of 500 K, chosen since lower temperatures provide the highest effect from pressure broadening. The figure illustrates that self broadening is 400% stronger on average than H2 broadening and that air and H2 broadening are comparable in magnitude. species, and we consider pressure broadening in a H2-rich atmosphere. Cross sections are used in atmospheric codes to deter- mine the resultant intensity transmitted through a column of gas. Usually, the column will undergo some changes in pres- sure, temperature, and number density over the length of the column. These factors combine to give an optical depth τλ where (cid:90) z2 τλ = nσλdz (21) z1 where z is length through the column, n is the abun- dance (number density) of the molecule and σλ is the cross section which is a function of wavelength. The abundance of a molecule is determined by the temperature and pressure 18 Hedges and Madhusudhan of the column by the simple gas law P = nkBT . (22) The optical depth gives a measure of how much inten- sity will be transmitted through a column of gas based on these properties. For a source intensity (I0λ), the resultant intensity (Iλ) at the end of the column is given by Iλ = I0λe −τλ , (23) where, the scale factor e−τλ is the transmittance. In order to make a meaningful comparison between the effects of altering the cross section on emergent intensity as a function of pressure and temperature we choose to fix τ to a single value. Here, we assume the column of gas to be at a given constant pressure and temperature, and hence constant density. The length of the column is allowed to vary in order to contain the same τ regardless of pressure and temperature. As τ is also a function of wavelength its value alters depending on the particular molecular feature. To fix τ we take the value at the peak of the water feature near 1.4 µm in the WFC3 band. If only the cross section σλ is changed, between a pressure broadened case σ1λ and an unbroadened case σ2λ which will each have intensities I1λ and I2λ, we can then find the effect pressure broadening has on the transmitted intensity. This is given as I1λ − I2λ I2λ = e −(cid:82) z2 z1 −(cid:82) z2 z1 nσ1λdz − e e−(cid:82) z2 z1 nσ2λdz nσ2λdz (24) As we assume a gas of the same number density at each pressure and temperature point then we can assume A = ndz = nx (25) (cid:90) z2 z1 where x is some distance scale. Substituting this we find ∆Iλ = I1λ − I2λ I2λ −A(σ1λ−σ2λ)) − 1. = (e (26) The above expression gives the relative change to the intensity, and hence the transmittance, for a given abun- dance and path through a uniform column of gas induced by using cross sections with pressure broadening compared to those without. Using the cross section across the WFC3 bandpass of 1-1.7µm and binning down to a given resolution we can find the difference to the transmittance of the column of gas as a function of wavelength. As discussed above, the length of the column is fixed such that the maximum opti- cal depth of the column in the given bandpass equals a fixed parameter (τ ), for a given density corresponding to a given temperature and pressure. We can then alter τ to investigate the optically thin and optically thick regimes as functions of pressure and temperature. We note that the change induced to transmittance (∆Iλ) across a given bandpass is higher at wavelengths with higher absorption, which are also of the wavelengths of interest to observations. We therefore con- sider the max(∆Iλ) in the WFC3 bandpass as our metric of choice in evaluating the effect of pressure broadening on transmittance in that bandpass. This does not take into ac- count how signal to noise might affect taking such observa- tions as zero transmittance (e−τ ) implies no signal; however, here we consider only values of e−τ (which can have values between 0 and 1) that are greater than 0.01. c(cid:13) 0000 RAS, MNRAS 000, 1–24 Figure 16. Effect of pressure broadening on the transmittance in an idealized atmosphere (see section 7). The contours show the maximum percentage difference in transmittance, i.e. the scale factor e(−nσx), induced by considering cross sections with pres- sure broadening versus those with Gaussian-only thermal broad- ening in the HST WFC3 bandpass. The relative difference is shown for a wide range in key parameters: pressure (P), spec- tral resolution (R), and maximum optical depth in the WFC3 bandpass (τ ); a nominal temperature of 1000 K is chosen for il- lustration but the general temperature-dependence is discussed in section 7. Here the individual points of the T,P and R grid have been linearly interpolated over for plotting purposes (see e.g. Hill, Pressure Broadening in Exoplanetary Atmospheres 19 The effect of pressure broadening on the transmittance in our idealized column of gas is similar to the effect on cross sections discussed in previous sections. Figure 16 shows the fractional difference (∆Iλ) pressure broadening makes to the transmittance as a function of several key parameters: the optical depth (τ ), pressure (P), and resolution (R); a nom- inal temperature of 1000 K is chosen for illustration but the general temperature-dependence is discussed below. At the outset, for low resolutions (R(cid:46)100), ∆Iλ is (cid:46)1% across the HST WFC3 G141 bandpass (1.1-1.7 µm) for almost the entire range of parameters of relevance to exoplanetary at- mospheres, particularly for P< 1 atm, T = 500 - 3000 K, and τ < 5. Naturally, however, ∆Iλ is higher for higher res- olutions. Considering nominal values of τ < 1, P<0.1 atm, and T>1000K, we find a maximum ∆Iλ in the WFC3 band to be 6% for a JWST-like medium resolution of R = 5000, and 75% for a VLT-like very high resolution of R = 105. The ∆Iλ for each resolution increases with increasing pressure and lowering temperature, particularly for very high resolution. For our lowest T of 500 K and τ = 1, for R = 5000 ∆Iλ is (cid:46)12% for P< 0.1 atm, and (cid:46) 65% for P< 1 atm. On the other hand, for the same T and τ , for R = 105 ∆Iλ is (cid:46) 100% for P< 0.1 atm and (cid:46) 2000% for P< 1 atm. As τ increases the difference between the two cases increases as there is more material to modify the in- tensity. However, as τ is increased to very high values the medium becomes optically thick and no light is transmitted in certain wavelength regions. This approach is simplistic as clearly it does not fac- tor in the the changes that could happen within the col- umn in temperature and pressure however this does give us a first approximation of the difference induced by chang- ing cross sections on observations of transmission spectra of exoplanetary atmospheres. In reality, light travels through many layers of an exoplanetary atmosphere, with different temperatures, pressures, and densities, before reaching the observer. The results above will hold for a specific pressure but full spectral models of exoplanetary atmospheres, both for transmission spectra and emission spectra, are required for a comprehensive investigation of the effect of pressure broadening discussed in the present work. 8 CROSS-SECTION DATABASE In this work we present a range of cross sections for H2O, CO2, CO, CH4, NH3 and HCN from a range of sources shown in Table 3. These have been investigated with de- tailed, line-by-line calculations of the Voigt profile with pres- sure and thermal broadening simultaneously included which no other database to date provides. The cross sections span a temperature range of 300K-3500K and pressures of 10−4 to 102 atmospheres. Finally the cross sections have been cre- ated in a variety of resolutions. The data in this work benefits not only from the ad- dition of a further dimension of pressure with an accu- rate broadening profile, but also in being generated uni- formly with the same code across molecules. This ensures low and consistent systematic errors across our data. The full database is represented in Figure 17. c(cid:13) 0000 RAS, MNRAS 000, 1–24 Table 4. Positions of the most prominent features in the absorp- tion cross sections of each molecule. Molecule Feature Position µm H2O CO2 CO HCN NH3 CH4 NO OH 6.61, 5.90, 2.76, 2.67, 2.60 1.87, 1.36, 1.13, 0.95 14.95, 4.23 4.57, 4.32, 2.68, 2.00 14.00, 7.30, 6.93, 4.73, 3.86 3.57, 3.00, 2.50, 1.53 15.96, 12.14, 10.37, 9.23, 6.67 6.15, 3.00, 2.26, 1.95, 1.51 1.22, 1.03 7.70, 7.40, 6.47, 3.42, 3.32 3.21, 2.37, 1.67 5.32, 2.58 4.20, 2.25, 1.55 8.1 Molecular Spectra of Observational Relevance Figure 17 shows cross sections for each molecule that has currently been addressed in this work. Many of these molecules have had cross sections computed for different line lists, broadening molecules and for a mean or detailed ap- proach but here only the most complete line list cases are shown. Each of the cross sections contains strong molecular fea- tures relating to the particular molecule. These features are usually the most ideal regions for observations with low sig- nal to noise. A list of the highest intensity features is given in Table 4 with their representative central wavelengths. 8.2 Temperature Dependence of Cross-Sections There is a strong dependence on temperature in these cross sections where increasing the temperature raises the transi- tion features in the spectrum, reducing the contrast between the peaks and troughs of each band of transitions. Molecules have a different temperature dependence based on their par- tition function. Individual molecules can have a stronger or weaker temperature dependence which can have an affect on our observations as in an atmosphere many molecular fea- tures will be combined. An example of the effect of such a case is shown in Figure 18. CH4 has a stronger temperature dependence than H2O. This leads to changes in their rela- tive contributions to the combined cross section as temper- ature increases. While at low temperatures the two contri- butions are comparable, at very high temperatures the H2O cross section is stronger than the CH4 cross section. While it would still be possible to tell these two contributions apart in high resolution spectra it would be more difficult to as- sess the abundance of CH4 in the presence of H2O. This is a general effect from the partition functions, but there are subtleties depending on the individual transitions. For example, transitions with high values of the total angular momentum quantum number J, and high lower energy lev- 20 Hedges and Madhusudhan Figure 17. Absorption cross sections of molecules in our database for a representative T=1400 and P=0.1 at a resolution of 0.01 cm−1 wavenumbers, using air broadening, with the same relative abundances. Such cross sections have been generated for all the molecules over a wide range of P and T using different sources of line lists and broadening molecules. Sources are listed in Table 3. els, become relatively much stronger at higher temperatures. Such effects can influence the shape of individual bands and could affect the continuum level of low intensity ‘window’ regions in spectra where opacity is low. This is a demonstration of the effect and is not necessar- ily a likely combination of molecules, however cross sections such as those developed here give us a tool to understand at what temperatures molecules are dominant, making them easier to observe. Understanding these features and their behaviour both with wavelength and temperature is crucial when developing ideal band passes for observation, particu- larly when taking telluric lines into account. For simple molecules such as CO there are only a few key bands in the infra-red which will contain spectroscopic information while for more complex molecules there is a contribution across the whole wavelength range. The most prominent features for each molecule are listed in Table 4. The WFC3 G141 bandpass (1.1-1.7 µm) is very well placed to detect and characterize H2O, CH4, and NH3 absorption features in spectra of exoplanets and brown dwarfs at low resolution (R (cid:46) 100). On the other hand, the NIRSpec and MIRI instruments aboard the JWST spanning a wide range of 0.6-28 µm will be able to detect a wide range of molecules at higher resolution. 9 DISCUSSION AND SUMMARY In this work we present a systematic and quantitative inves- tigation of the effects of various aspects of pressure broaden- ing on molecular cross sections for application to exoplane- tary atmospheres. We first use H2O as our primary molecule of choice for this investigation as it has the most complete absorption line data. The factors we investigate include the resolution and evaluation width of Voigt profiles, pressure versus thermal broadening, broadening agent, spectral reso- lution, and completeness of broadening parameters. We in- vestigate in detail the effect of pressure broadening both on the absorption cross sections of H2O under varied condi- tions as well as on the transmittance of a fiducial idealized atmosphere. We use the optimal methods resulting from this investigation to systematically and homogeneously generate a library of pressure-broadened absorption cross sections for a wide selection of molecules of relevance to exoplanetary atmospheres across a wide range of temperature, pressure, and spectral resolution. This study allows us to address the question of the in- accuracies introduced to molecular absorption cross sections from pressure broadening, both in the context of current and future observational capabilities. As new instruments come online with improved specifications we will have access to a wealth of high resolution data on exoplanet atmospheres. The interpretations of these data sets will be impacted by c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 21 profiles affecting the continuum level of low intensity neigh- bours greatly, which can be underestimated when a cut off is too narrow. As discussed in section 5 we present a method of accu- rately evaluating the Voigt profile on a fine grid that pro- duces minimal errors in the final cross section at resolutions of interest. This is achieved with a spacing that is adaptive in temperature and in pressure. The grid we adopt is found to be as accurate as the grid from (Hill, Yurchenko & Ten- nyson 2013) to within 0.2% at pressures of 1 atm or less and gives a vast saving on computational time, particularly for high pressures. We pair this grid with a cut off value, ∆νc, of 500 Voigt widths (raised to 1000 above pressures of 1 atm). ∆νc describes the separation around the wavenumber cen- troid of the line transition up to which the Voigt profile is evaluated to. Having investigated a range of values at differ- ent resolutions we find 500 Voigt widths to be sufficient both to provide good normalisation for the profile and to evalu- ate far into the extensive Lorentzian wings. When compared with other values of ∆νc from literature we find our value to be more accurate ( 10-100%) at low pressures (P= 0.1 - 0.001 atm) due to the Voigt width adapting with both temperature and pressure as the Gaussian and Lorentzian components change. This combination of ∆νc and an adap- tive grid provides low errors for all resolutions discussed in this work up to R=100,000. We find errors of less than 1% (averaging ∼0.2%) in the final cross sections at the peaks of transition features. This increases to 10% within transition features at very low intensity, however we find such transi- tions to be less likely to significantly effect observations and modelling results. Beyond this resolution it may be wise to increase ∆νc and use a finer sampling of the profile. Here we use a standard Voigt profile though works such as Ngo et al. (2012) have shown that a change in velocity of the broadening particle can alter the pressure broadened profile shape. This would likely change the wing shape and alter the continuum from what we present here though we anticipate that difference to be small. We find the effect of pressure broadening to be varied depending on the resolution, pressure and temperature. We choose to measure the change induced to H2O cross sections due to pressure broadening using the median percentage dif- ference over the HST WFC3 bandpass (1.1-1.7 µm). This provides a reasonable estimate of the characteristic differ- ence, though it is possible to induce higher changes for spe- cific lines. Generally, the differences are larger for higher res- olutions, higher pressures, and lower temperatures. For low resolution spectra (R(cid:46)100) of exoplanets, that are possible with current instruments, for representative exoplanetary temperatures (T=500K-2500K, P(cid:46)1 atm) and H2 rich at- mospheres we find the median difference in cross sections in- troduced by various aspects of pressure broadening (δ) to be (cid:46)1%. For higher resolutions (R(cid:46)5000), including those at- tainable with JWST, we find that δ can be up to 40%. On the other hand for very high resolution spectra (R∼105) pres- sure broadening can introduce δ(cid:38)100%, reaching (cid:38) 1000% for low temperatures, (T(cid:46)500K), high resolutions (R∼105) and high pressures (P∼0.1-1 atm). Such a case could be found with instruments such as the VLT and E-ELT if cool H2-dominated targets were observed. For hotter targets of T=2000K this reduces to 15%, though this is a median over wavelength and can be found to be higher ((cid:38)100%) for cer- Figure 18. Comparison of CH4 and H2O across different tem- peratures. Top: comparison of normalised partition functions (Q(T )/Q(T = 296K)). CH4 shows a much stronger temperature dependence than H2O. Bottom: Comparison of cross sections at different temperatures. We see that the contrast between peaks and troughs in both spectra is lowered at high temperatures due to the partition function. However CH4 is is more effected due to the higher partition function values and, at higher temperatures, is weakened compare to H2O. basic model inputs such as cross sections which are directly degenerate with the molecular abundances derived using spectral retrieval methods. The comparisons presented here show in detail the magnitude of the errors we can expect in these fundamental inputs to atmospheric models across a range of parameters. To generate cross sections we follow a prescription of mapping line intensities broadened by an appropriate func- tion to a fine ’sub-grid’ which finely samples the profile of the line. This is iterated over each transition from the source and binned to an output grid with a wider spacing, (lower resolution,) for further use. The lines are broadened by either a Gaussian-only model, (which uses only thermal broaden- ing,) or a Voigt profile, (which combines the Gaussian ther- mal and Lorentzian pressure broadening,) evaluated using the Faddeeva package (S. G. Johnson 2012). When evaluating the Voigt profile on a grid there are two clear sources of error. Firstly the grid spacing may be too wide, causing the evaluation of the contribution to each grid point to be poor, leading also to a poor normalisation and misrepresenting the line transition intensity. Secondly the wings of the profile may be cut off prematurely, leading to small fractions of intensity from the wings being missed. This is aggravated by the range of intensity values which span many orders of magnitude within a narrow wavenum- ber range. This leads to the wings of isolated, high intensity c(cid:13) 0000 RAS, MNRAS 000, 1–24 22 Hedges and Madhusudhan tain features. For spectral resolutions of R=5000, (similar to that achievable with JWST) this reduces to δ=5% for hot targets of T=2000K at P=0.1 atm. From this we can see that even with very high resolu- tions current hot Jupiter targets with temperatures of 800K- 2500K will not be greatly effected by differences induced from pressure broadening with pressures of (cid:46)0.1 atm. A more significant change is found at pressures of 1 atm or above, though current observations of exoplanetary atmo- spheres typically probe pressures above 1 atm (Madhusud- han 2012). Data on cool targets at high resolution is cur- rently a distant future prospect and we are unlikely to be affected by this level of uncertainty in the near future with such targets. However, even with the lower temperature end of hot Jupiter targets (T∼1000K) and modern day instru- ments such as the VLT pressure broadening can cause dis- crepancies in the cross section of 30-200% for H2 dominant atmospheres with P=0.1-1 atm. Molecular cross sections are degenerate with abundance in atmospheric models and any error in cross sections results in an uncertainty in our abundance measurements. From this work we find that for cool targets (T∼500K) at high resolutions (R(cid:38)105) we would expect uncertainties in the abundance measurements of at least 100% purely from the cross section inputs to atmospheric models over those that do not include pressure broadening for H2 dominated atmo- spheres. A true spectrum involves many cross sections from an atmosphere with many layers of temperature and pres- sure and an observation through many optical depths which will compound this difference. For a true estimation of the difference pressure broadening will make to abundances full, rigorous atmospheric models are needed. Cross sections have been created from a variety of sources which span different levels of completeness, i.e. the number of transitions for which line data are available, and temperature validity, i.e. the temperature up to which the intensity values and completeness can be trusted. In this work we focus on H2O as it is both currently detectable in hot Jupiters and well documented in line list sources. We find that our metric of finding the median percentage differ- ence gives good results even when line lists have low com- pleteness, such as the PS line list for H2O. Due to this we are able to make comparisons between cross sections generated from line lists of different sources. Currently significant efforts are being made into obtain- ing data on pressure broadening parameters for molecules in different gasses. However, there are some cases where no pressure broadening data is available for certain molecules. In other cases it may be that a smaller, low temperature line list source such as HITRAN has broadening parameters for fewer transitions but a more accurate and more complete line list from sources such as ExoMol does not. To investi- gate what the best approach is in such situations a mean approach has been tested where broadening parameters are averaged and the mean parameter is applied to each profile. We find that, when taken across a wide wavelength range, the differences between cross sections generated in a detailed manner and those with a mean broadening parameter ap- plied is up to 20%, even at very high resolution (R>105). However when looking in detail in a narrow wavelength band individual lines may be inaccurate at higher pressure due to slight differences in the broadening parameter from the mean. Despite this, using our metric of finding the median percentage difference we find that mean broadening parame- ters are still useful to ascertain the magnitude of the change to the cross section pressure broadening can induce. We also investigated the influence of different broaden- ing agents (self, air and H2) on the cross sections using H2O as a case study. Generally, self broadening is significantly stronger than H2 broadening, by about a factor of 4 on av- erage across our range of pressure, temperature, and resolu- tion. For H2 broadening, which is the dominant component for giant exoplanet atmospheres that are most amenable to spectroscopy, we see a smaller effect on cross sections than that due to self broadening or air broadening. We find in our current investigations using H2O that where only self and air broadening are available, air broadening produces a closer result to H2 broadening. As with other parameters discussed above, the differences induced to cross sections due to the different broadening agents are <1% for low res- olution (R (cid:46) 100). For medium resolutions of R=5000 we find that δ ∼10% at T=500K and P=0.1 for H2O in an H2O atmosphere (i.e. self broadening) where as δ ∼1% for an H2 atmosphere (i.e. H2 broadening) . For hotter targets with higher resolution (T=1000K, R=100,000) we find that δ ∼100% for P=0.1 atm. We find that when looking at the transitions of water in a water dominated atmosphere we expect the effect of pressure broadening to be more pro- nounced implying that pressure broadening is likely to be very important for hotter water-rich targets. A partial pressure of one has been used here, assuming that only one broadening agent is present, though a combi- nation of agents would be more physical, particularly with He included. Further investigation could be undertaken to find at what concentration other agents affect the broaden- ing profile shape. A final investigation has been undertaken to assess the difference including pressure broadening in cross sections makes to the transmitted intensity through a uniform col- umn of gas, as a function of the pressure, temperature, spec- tral resolution, and optical depth. Our investigation focused on an idealized column of H2-dominated gas with H2O as the only absorber in the HST WFC3 G141 bandpass (1.1- 1.7 µm). The results follow the general trends of how each of these parameters influence the cross sections themselves, as discussed above. For low resolutions (R(cid:46)100), we find the relative change in transmittance (∆Iλ) to be (cid:46)1% across the HST WFC3 G141 bandpass (1.1-1.7 µm) for almost the entire range of parameters of relevance to exoplanetary at- mospheres. For representative parameters of τ < 1, P<0.1 atm, and T>1000K, ∆Iλ can be up to 6% for a JWST-like medium resolution of R = 5000, and 75% for a VLT-like very high resolution of R = 105. ∆Iλ can be even higher for higher T, lower T, and larger τ . While for R (cid:46) 5000 the ∆Iλ are still below ∼100%, for very high resolutions (R∼105) ∆Iλ can be as high as ∼2000%. Ultimately, our present work suggests that incorporat- ing pressure broadening to compute molecular cross sections for atmospheric models will be necessary depending on the desired accuracy in molecular abundance estimates retrieved from the spectra. Across all the various factors considered in this work, for low resolution observations (R(cid:46)100) of exo- planetary spectra that are currently possible, e.g. with HST, the median differences in cross section induced due to accu- c(cid:13) 0000 RAS, MNRAS 000, 1–24 Pressure Broadening in Exoplanetary Atmospheres 23 rate pressure broadening is found to be (cid:46)1%. For medium resolutions (R∼5000), similar to those possible with JWST, the differences are expected to be at the (cid:46)40% level. For very high resolution spectra (R∼105), that are possible with cur- rent and future large ground-based telescopes such as VLT and E-ELT, significantly higher differences are possible of 100% or even much larger, depending on other factors dis- cussed above. With atmospheric characterisation becoming an ever more important part of exoplanet research we can begin to see that pressure broadening will impact us in the fu- ture. With medium and high resolution spectra of exoplan- ets, both hot and cool, we can expect our abundance mea- surements to be affected in some way. Beyond that we may be able to detect and characterise the pressure in the at- mospheres of other planets by finding regions of wavelength space particularly affected and using high resolution spectra. This goal would be difficult to achieve even with a wealth of molecular data at our disposal as signal to noise ratios for such data are likely to be low. The structure and dy- namics of a full atmosphere leads to a convolution of many profiles making characterisation of pressure broadening ex- ceedingly difficult. Other factors such as wind speed and Doppler broadening provide further barriers. These will be the ultimate challenges of the future when we will eventually be able to conduct very high resolution spectroscopy of cool low mass exoplanets. ACKNOWLEDGEMENTS CH acknowledges the support from the Science and Technol- ogy Facilities Council (STFC) for PhD funding. We thank the groups that have spent time developing molecular line lists, particularly ExoMol, HITEMP and HITRAN, without which this work would not be possible. We thank Richard Freedman, Iouli Gordon, Jonathan Tennyson, and Sergey Yurchenko for helpful discussions. We would like to thank the anonymous reviewer for a very helpful review. REFERENCES Abramowitz M., Stegun I., 1964, Handbook of Mathemat- Buenzli E., Saumon D., Marley M. S., Apai D., Radigan J., Bedin L. R., Reid I. N., Morley C. V., 2015, ApJ, 798, 127 Chamberlain J. W., 1978, Theory of planetary atmospheres : an introduction to their physics and chemistry Cich M., McRaven C., Lopez G., Sears T., Hurtmans D., Mantz A., 2012, Applied Physics B, 109, 373 Crouzet N., McCullough P. R., Deming D., Madhusudhan N., 2014, ApJ, 795, 166 Deming D. et al., 2013, ApJ, 774, 95 Draegert D. a., Williams D., 1968, Journal of the Optical Society of America, 58, 1399 Dutta J., Jones C., Goyette T. M., Lucia F. C. D., 1993, Icarus, 102, 232 Edwards D. P., Strow L. L., 1991, J. Geophys. Res., 96, 20 Faure A., Wiesenfeld L., Drouin B., Tennyson J., 2013, Journal of Quantitative Spectroscopy and Radiative Transfer, 116, 79 Fischer J., Gamache R., Goldman A., Rothman L., Perrin A., 2003, Journal of Quantitative Spectroscopy and Ra- diative Transfer, 82, 401 , the {HITRAN} Molecular Spec- troscopic Database: Edition of 2000 Including Updates of 2001. Fraine J. et al., 2014, Nature, 513, 526 Freedman R. S., Lustig-Yaeger J., Fortney J. J., Lupu R. E., Marley M. S., Lodders K., 2014, ApJS, 214, 25 Freedman R. S., Marley M. S., Lodders K., 2008, ApJS, 174, 504 Gamache R. R., Hartmann J.-M., 2004, Journal of Quan- titative Spectroscopy and Radiative Transfer, 83, 119 Gamache R. R., Lamouroux J., Laraia A. L., Hartmann J.- M., Boulet C., 2012, Journal of Quantitative Spectroscopy and Radiative Transfer, 113, 976 , three Leaders in Spec- troscopy Gamache R. R., Laraia A. L., Lamouroux J., 2011, Icarus, 213, 720 Gamache R. R., Lynch R., Neshyba S. P., 1998, Journal of Quantitative Spectroscopy and Radiative Transfer, 59, 319 , atmospheric Spectroscopy Applications 96 Gamache R. R., Lynch R., Plateaux J. J., Barbe A., 1997, J. Quant. Spec. Radiat. Transf., 57, 485 Grimm S. L., Heng K., 2015, ArXiv e-prints Gryvnak D. A., 1972 Harris G. J., Tennyson J., Kaminsky B. M., Pavlenko Y. V., ical Functions, 5th edn. Dover, New York Jones H. R. A., 2006, MNRAS, 367, 400 Antony B. K., Gamache P. R., Szembek C. D., Niles D. L., Hill C., Yurchenko S. N., Tennyson J., 2013, Icarus, 226, Gamache R. R., 2006, Molecular Physics, 104, 2791 1673 Barber R. J., Tennyson J., Harris G. J., Tolchenov R. N., 2006, MNRAS, 368, 1087 Benneke B., Seager S., 2012, ApJ, 753, 100 Bernath P. F., 2014, Royal Society of London Philosophical Transactions Series A, 372, 30087 Bernath P. F., 2014, Phil. Trans. R. Soc A, 173 Birnbaum G., 1979, Journal of Quantitative Spectroscopy Imnd V., ????, 91109 Knutson H. A. et al., 2012, ApJ, 754, 22 Konopacky Q. M., Barman T. S., Macintosh B. A., Marois C., 2013, Science, 339, 1398 Kreidberg L. et al., 2014a, Nature, 505, 69 Kreidberg L. et al., 2014b, ApJ, 793, L27 Kurucz R. L., 1992, Rev. Mexicana Astron. Astrofis., 23, and Radiative Transfer, 21, 597 45 Bouanich J.-P., Aroui H., Nouri S., Picard-Bersellini a., Lee J.-M., Fletcher L. N., Irwin P. G. J., 2012, MNRAS, 2001, Journal of molecular spectroscopy, 206, 104 420, 170 Brault J. W., Brown L. R., Chackerian C., Freedman R., Predoi-Cross A., Pine A. S., 2003, Journal of Molecular Spectroscopy, 222, 220 Lemaire V., Babay A., Lemoine B., Rohart F., Bouanich J. P., 1996, 45, 40 Letchworth K. L., Benner D. C., 2007, Journal of Quanti- Brogi M., Snellen I. A. G., de Kok R. J., Albrecht S., Birkby tative Spectroscopy and Radiative Transfer, 107, 173 J., de Mooij E. J. W., 2012, Nature, 486, 502 Li G., Gordon I. E., Rothman L. S., Tan Y., Hu S.-M., c(cid:13) 0000 RAS, MNRAS 000, 1–24 24 Hedges and Madhusudhan Kassi S., Campargue A., Medvedev E. S., 2015, ApJS, 216, 15 Line M. R. et al., 2013, ApJ, 775, 137 Luijendijk S. C. M., 1977, Journal of Physics B: Atomic and Molecular Physics, 10, 1735 Madhusudhan N., 2012, ApJ, 758, 36 Madhusudhan N., Crouzet N., McCullough P. R., Deming D., Hedges C., 2014a, ApJ, 791, L9 Madhusudhan N. et al., 2011a, Nature, 469, 64 Madhusudhan N., Knutson H., Fortney J. J., Barman T., 2014b, Protostars and Planets VI, 739 Tennyson J. et al., 2014, ArXiv e-prints Tennyson J., Yurchenko S. N., 2012, MNRAS, 425, 21 Tennyson J., Yurchenko S. N., 2014a, in IAU Symposium, Vol. 297, IAU Symposium, Cami J., Cox N. L. J., eds., pp. 330–338 Tennyson J., Yurchenko S. N., 2014b, Experimental As- tronomy Tinetti G., Tennyson J., Griffith C. A., Waldmann I., 2012, Royal Society of London Philosophical Transactions Series A, 370, 2749 Waldmann I. P., Tinetti G., Rocchetto M., Barton E. J., Madhusudhan N., Mousis O., Johnson T. V., Lunine J. I., Yurchenko S. N., Tennyson J., 2015, ApJ, 802, 107 2011b, ApJ, 743, 191 Madhusudhan N., Seager S., 2009, ApJ, 707, 24 McCullough P. R., Crouzet N., Deming D., Madhusudhan N., 2014, ApJ, 791, 55 Wilzewski J. S., Gordon I. E., Kochanov R. V., Hill C., Rothman L. S., 2016, Journal of Quantitative Spec- troscopy and Radiative Transfer, 168, 193 Yadin B., Veness T., Conti P., Hill C., Yurchenko S. N., Mihalas D., Auer L. H., Mihalas B. R., 1978, ApJ, 220, Tennyson J., 2012, MNRAS, 425, 34 1001 Miller-Ricci Kempton E., Rauscher E., 2012, ApJ, 751, 117 Ngo N. H., Tran H., Gamache R. R., Hartmann J. M., 2012, Royal Society of London Philosophical Transactions Series A, 370, 2495 Olivero J., Longbothum R., 1977, Journal of Quantitative Spectroscopy and Radiative Transfer, 17, 233 Padmanabhan A., Tzanetakis T., Chanda A., Thomson M. J., 2014, Journal of Quantitative Spectroscopy and Radiative Transfer, 133, 81 Partridge H., Schwenke D. W., 1997, J. Chem. Phys., 106, 4618 Pine A. S., 1992 Plez B., 1998, A&A, 337, 495 Robert D., Bonamy J., 1979, Journal de Physique, 40, 923 Rothman L. S. et al., 2013, J. Quant. Spec. Radiat. Transf., 130, 4 Rothman L. S. et al., 2010, J. Quant. Spec. Radiat. Transf., 111, 2139 Rothman L. S. et al., 1998, J. Quant. Spec. Radiat. Transf., 60, 665 S. G. Johnson, A. Cervellino J. W., 2012, libcerf, numeric library for complex error functions. http://apps.jcns. fz-juelich.de/libcerf Schreier F., 2011, Journal of Quantitative Spectroscopy and Radiative Transfer, 112, 1010 Seager S., Deming D., 2010, ARA&A, 48, 631 Sharp C. M., Burrows A., 2007, ApJS, 168, 140 Showman A. P., Wordsworth R. D., Merlis T. M., Kaspi Y., 2013, Atmospheric Circulation of Terrestrial Exoplanets, Mackwell S. J., Simon-Miller A. A., Harder J. W., Bullock M. A., eds., pp. 277–326 Snellen I., de Kok R., de Mooij E., Brogi M., Nefs B., Al- brecht S., 2011, in IAU Symposium, Vol. 276, IAU Sym- posium, Sozzetti A., Lattanzi M. G., Boss A. P., eds., pp. 208–211 Snellen I. A. G., Brandl B. R., de Kok R. J., Brogi M., Birkby J., Schwarz H., 2014, Nature, 509, 63 Snellen I. A. G., de Kok R. J., de Mooij E. J. W., Albrecht S., 2010, Nature, 465, 1049 Snellen I. A. G., de Kok R. J., le Poole R., Brogi M., Birkby J., 2013, ApJ, 764, 182 Spiegel D. S., Haiman Z., Gaudi B. S., 2007, ApJ, 669, 1324 Tashkun S., Perevalov V., 2011, Journal of Quantitative Spectroscopy and Radiative Transfer, 112, 1403 Yurchenko S. N., Barber R. J., Tennyson J., 2011, MNRAS, 413, 1828 Yurchenko S. N., Tennyson J., 2014, MNRAS, 440, 1649 Zaghloul M. R., 2007, MNRAS, 375, 1043 Zaghloul M. R., Ali A. N., 2011, CoRR, abs/1106.0151 Zaghloul M. R., Ali A. N., 2012, ACM Trans. Math. Softw., 38, 15:1 10 APPENDIX 10.1 Available Line List Sources Available sources for line lists are listed in Table 5 with each molecule that is offered. Some lists such as those from HITRAN, HITEMP and GEISA offer pressure broadening parameters as well as intensity and wavelength for transi- tions. Here we give the number of transitions for each list for each molecule. Where two sources are available it is on the whole better to use one with more transitions as it is more complete, though there is also temperature validity to consider. We have used HITEMP’s water line list as the most accurate in this work as it is an update of the BT2 line list. ExoMol, however, does offer the most complete line lists with the highest temperature validity range so where possible we recommend using their line lists though they do not currently provide pressure broadening parameters. c(cid:13) 0000 RAS, MNRAS 000, 1–24 Molecule HITRAN HITEMP GEISA ExoMol Yueqi Other (Source) Pressure Broadening in Exoplanetary Atmospheres 25 AlO and Isotopes BeH C2 C2H2 C2H4 C2H6 C2HD C2N2 C3H4 C3H8 C4H2 C6H6 CaH CF4 CH and Isotopes CH3Br CH3Cl CH3CN CH3CN CH3D CH3OH CH3OH CH4 ClO ClONO2 CN CO CO2 COF2 CP CrH CS FeH FeH GeH4 H2 H2CO H2O H2O2 H2S HBr HC3N HCl HCN HCOOH HDO HF HI HNC HNO3 HNO3 HO2 HOBr HOCl Kcl and Isotopes LiH MgH and Isotopes N2 N2O NaCl and Isotopes NaD NaH NH NH3 NO - - - 12613 18097 43592 - - - - 124126 - - 60033 - 18692 107642 - 3572 - - 19897 336830 5721 21988 - 1019 169292 168793 - - 5129 - - - 4017 - 142045 126983 36561 3039 180332 11879 2955 62684 - 10073 3161 - - 903854 38804 2177 8877 - - - 1107 33074 - - - - 45302 103710 c(cid:13) 0000 RAS, MNRAS 000, 1–24 - - - - - - - - - - - - - - - - - - - - - - - - - - 113631 11193608 - - - - - - - - - 114241164 - - - - - - - - - - - - - - - - - - - - - - - - - - 115610 - - - 11340 18378 28439 15512 2577 19001 8983 119480 9797 - 291 - 36911 18344 200 - 49237 19897 - 240858 7230 356899 - 13515 413524 70904 - - - - - 824 - 37050 67504 126983 20788 1294 179347 533 81889 62684 - 107 806 5619 669988 - 38804 - 17862 - - - 120 50633 - - - - 5000000 7858 - - - - - - - - - - 26980 - - - - - - - - - 10 ×1010 - - - - - - - - - - - - - - 505000000 - - - - - 34418408 - 700000000 - - - - - - - - 7000000 18982 6716 - - 5000000 167224 79898 - 29082 105079 1138 323 351 - - - 47570 - - - - - - - - - 6000 - 53086 - - - - - - - - - - 195120 - - - - 13825 - 93040 - - - - - - - - - 2588 - - - - - - - - - - - - - - - - - - - 10425 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 573881316 Tashkun & Perevalov (2011) - - - - - - - - - - 204688 Bernath (2014) - - - - - - 296000000 Partridge & Schwenke (1997) - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 26 Hedges and Madhusudhan Molecule HITRAN HITEMP GEISA ExoMol Yueqi Other (Source) NO+ NO2 O O2 O3 OCS OH PH PH3 PN and Isotopes ScH SF6 SiO and Isotopes SO2 SO3 TiH TiO 1206 104223 2 1787 261886 15618 30772 - 22189 - - 2889065 - 72460 10188 - - - - - - - - 41557 - - - - - - - - - - 1206 104223 - 6428 389378 33809 42866 - 20364 - - 92398 - 68728 - - - - - - - - - - 1.68×1010 - 700000 1152826 - 1784965 - 174674257 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - 181080 - 157430 8325354 Bernath (2014) Kurucz (1992) Table 5. A list of all available line list sources for different molecules with the number of lines available in each. In general the more complete line lists are preferable not only as more features are evaluated within the molecular spectrum but because the continuum of low intensity lines is better approximated. c(cid:13) 0000 RAS, MNRAS 000, 1–24
0912.2299	1	0912	2009-12-11T18:13:10	Relative equilibria in the unrestricted problem of a sphere and symmetric rigid body	[ "astro-ph.EP", "physics.class-ph" ]	We consider the unrestricted problem of two mutually attracting rigid bodies, an uniform sphere (or a point mass) and an axially symmetric body. We present a global, geometric approach for finding all relative equilibria (stationary solutions) in this model, which was already studied by Kinoshita (1970). We extend and generalize his results, showing that the equilibria solutions may be found by solving at most two non-linear, algebraic equations, assuming that the potential function of the symmetric rigid body is known explicitly. We demonstrate that there are three classes of the relative equilibria, which we call "cylindrical", "inclined co-planar", and "conic" precessions, respectively. Moreover, we also show that in the case of conic precession, although the relative orbit is circular, the point-mass and the mass center of the body move in different parallel planes. This solution has been yet not known in the literature.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 10 (2009) Printed 11 June 2021 (MN LATEX style file v2.2) Relative equilibria in the unrestricted problem of a sphere and symmetric rigid body Mikhail Vereshchagin1(cid:63), Andrzej J. Maciejewski2† and Krzysztof Go´zdziewski1‡ 1Toru´n Centre for Astronomy, Nicolaus Copernicus University, Gagarin Str. 11, 87-100 Toru´n, Poland 2Institute of Astronomy, University of Zielona G´ora, Podg´orna 50, PL-65 -- 246 Zielona G´ora, Poland Accepted 2009 December 3. Received 2009 December 2; in original form 2009 September 10 ABSTRACT We consider the unrestricted problem of two mutually attracting rigid bodies, an uniform sphere (or a point mass) and an axially symmetric body. We present a global, geometric ap- proach for finding all relative equilibria (stationary solutions) in this model, which was already studied by Kinoshita (1970). We extend and generalize his results, showing that the equilibria solutions may be found by solving at most two non-linear, algebraic equations, assuming that the potential function of the symmetric rigid body is known explicitly. We demonstrate that there are three classes of the relative equilibria, which we call cylindrical, inclined co-planar, and conic precessions, respectively. Moreover, we also show that in the case of conic preces- sion, although the relative orbit is circular, the point-mass and the mass center of the body move in different parallel planes. This solution has been yet not known in the literature. 1 INTRODUCTION In this paper, we consider the dynamical problem of two rigid bod- ies interacting mutually according to the Newtonian laws of gravi- tation. It is well known that if both these bodies are approximated by point-masses, their dynamics are analytically solvable in terms of the integrable Kepler problem. However, even if only one of the bodies is an arbitrary rigid body then the complexity of the model changes dramatically. First of all, we have three more de- grees of freedom required to describe the rotational motion of the rigid body. Moreover, because the gravitational potential of a rigid body, in general case, depends on an infinite number of parame- ters, the two rigid-body dynamics depend also on this infinite set of parameters. Finally, orbital and rotational degrees of freedom are coupled non-linearly, and the resulting system is not analytically solvable. Because of the mathematical complexity involved, many ap- proximate models were studied in the past. For example, under cer- tain assumptions, one can consider the co-planar two rigid-body problem, see e.g., Beletskij (1975); Barkin (1985); Beletskij & Ponomareva (1990); Go´zdziewski (2003). Another way to simplify the problem relies on truncating series expansions of the gravita- tional potential of the rigid body (the potential of mutual interac- tions), see e.g., Markeev (1967b,a, 1985, 1988); Shcherbina (1989). In the so-called satellite (or restricted) models, it is assumed that the rotational motion does not influence the relative, Keplerian orbit of the mass centres of the bodies. In such models, the relative orbit is given parametrically, and only the rotational dynamics need to be investigated; for details, see e.g., Markeev (1985, 1967b, 1975); Maciejewski (1994); Maciejewski & Go´zdziewski (1995). In this work, we focus on particular version of the model in- volving a sphere (a point-mass), by considering that the rigid body is axially symmetric. Obviously, this leads to much simplified form of the gravitational potential. However, in general, it still depends on an infinite number of parameters. For the first time this problem c(cid:13) 2009 RAS was investigated by Kinoshita (1970), and hence it will be called the Kinoshita problem from hereafter. Our aim is to determine all possible classes of the relative equilibria (stationary solutions) in this problem, using a general formalism in Wang et al. (1991) and Maciejewski (1995). Keeping in mind a motivation of the work by Kinoshita, we try to generalize his analysis and to simplify condi- tions for the existence of these equilibria, avoid repetition of "of" assuming only the necessary general form of the gravitational po- tential of the rigid body. In this sense, our paper is also related to the work by Scheeres (2006), who searched for the relative equilibria in the unrestricted two rigid-body problem, and formulated conditions of the equilibria in terms of solutions to two non-linear equations, parameterized by integrals of the total energy and angular momen- tum. Moreover, we stress that our analysis rely on the very basic vectorial form of the equations of motion, and we consider the full, non-restricted model with symmetric rigid body. The assumption of symmetry implies that the results derived for the general, full two- body model cannot be simply "translated" as a particular case, and, in fact, a special reduction of the system which takes into account the symmetry explicitly is necessary. We note that the relative equi- libria of the model with axially symmetric body may be understood as particular periodic orbits in the full two rigid body model con- sidered by Scheeres (2006). This reduction is in fact crucial for a generalization of the results of Kinoshita and to discover a class of equilibria that have been missed in his analysis. Investigations of the unrestricted two rigid body problem, in a version considered in this work, and its particular stationary so- lutions, are interesting because they concern a special case of gen- erally unsolved, classic problem of the dynamics. Moreover, the qualitative analysis of this model may help us to answer for impor- tant, even "practical" questions on the dynamics and coupling of the rotational and orbital motions in many astronomical systems. These are, for instance, binary asteroids, see Scheeres (2006) and references therein. A deep understanding of the qualitative dynam- 2 M. Vereshchagin, A. J. Maciejewski & K. Go´zdziewski ics are important for the attitude determination and control of large artificial satellites orbiting planets and/or irregular natural moons. Recently, the diversity of extrasolar planets discovered in wide dy- namical environments also rises questions on their rotational mo- tion and related long term effects (e.g., Correia et al. 2008). In the context of mathematical complexity of the problem, the relative equilibria are the simplest class of solutions that may be found and investigated analytically, and are helpful to construct local, precise analytic theories of motion in their vicinity in the phase space (e.g., Go´zdziewski & Maciejewski 1998). A plan of this paper is the following. In Sect. 2 we formulate the mathematical model and the equations of motion in the most general, vectorial form are derived. Section 3 is devoted to define the Kinoshita problem. Next, we introduce the relative equilibria, and we perform a global analysis of their existence and bifurcations (Sect. 4). This part relies on particular, geometric reduction of the equations of motion, and is the primary key for our analysis. Fi- nally, we compare the classic results by Kinoshita with the results obtained through the approach introduced in this work. 2 THE EQUATIONS OF MOTION Let us consider the gravitational two rigid body problem. We as- sume that one of the bodies, P is a point mass, or an uniform sphere, with mass m1. The second one, B, is a rigid body with mass m2 (Fig. 1). The mechanical problem in the most general settings has nine degrees of freedom, so the dynamics of the bodies are de- scribed by 18 first-order differential equations. Moreover, the dy- namical system possesses symmetries, i.e., it may be shown that the equations of motion are invariant with respect to a six-dimensional group. This fact is a direct consequence of the very basic laws of Newtonian mechanics. Namely, the equations of motion do not de- pend on particular choice of the inertial reference frame. Thus, we can choose the origin of the inertial frame in an arbitrary point, and this implies that the equations of motion are invariant with respect to the natural action of three dimensional group of translations. Moreover, also the orientation of the inertial frame can be cho- sen arbitrarily. Hence, the equations of motion are invariant with respect to the natural action of the three-dimensional group of ro- tations. The existence of these symmetries makes it possible to reduce the dimension of the phase space of this system and to simplify its analysis. However, there is no unique procedure for such a re- duction. Obviously, it should rely on such a transformation of the initial system of the equations of motion (phase-space variables) that the resulting equations form a lower-dimensional sub-system. In fact, the reduced equations of motion of the two rigid body prob- lem were obtained by Wang et al. (1991) (actually, they considered the motion of a rigid body in the central gravitational field), and, with a simpler and more direct method and under general settings by Go´zdziewski & Maciejewski (1999). These papers are good ref- erences to step-by-step development of the reduced equations of motion, which is skipped here to save space. Before we write down these equations of motion, we need to fix the notation. Components of a vector xxx (as a geometric object) in an inertial reference frame will be denoted by x = [x1,x2,x3]T . Components of the same vector in the rigid body fixed frame we will denote by the corresponding capital letter, i.e., X = [X1,X2,X3]T . Thus, if A is the orientation (attitude) matrix of the body with respect to the inertial frame, then we have x = AX. The scalar product of two vectors xxx and yyy is denoted by xxx·yyy. It can Figure 1. Geometry of the problem. XXX, YYY , ZZZ are units vector coincided with the principal axis of inertia of the rigid body. be calculated in an arbitrary orthonormal frame as follows: xxx·yyy = (cid:104)x,y(cid:105) := 3 ∑ i=1 We shall also write: xiyi = (cid:104)AX,AY(cid:105) = (cid:104)X,Y(cid:105) = 3 ∑ i=1 XiYi. x2 := xxx·xxx = (cid:104)x,x(cid:105) = X 2 := (cid:104)X,X(cid:105). In our exposition, we follow Maciejewski (1995); Go´zdziewski & Maciejewski (1999). The reduced equations of motion describe the relative motion of the bodies with respect to a frame fixed in rigid body B. They have the following form of the so-called Newton- Euler equations: d dt d dt d dt R = R× Ω + P, P = P× Ω− ∂U ∂R , G = G× Ω + R× ∂U ∂R ,  (1) where R is the radius vector directed from the mass centre of B to P, P is the relative linear momentum, G is the angular momentum of B, and Ω = I−1G is its angular velocity; U is the gravitational potential of the body B, and I = diag(A,B,C) stands for its tensor of inertia. We assume that the rigid body fixed frame coincides with the principal axes frame. The geometry of the model is illustrated in Fig. 1. As it was shown in Go´zdziewski & Maciejewski (1999), equa- tions (1) are Hamiltonian with respect to a non-canonical Poisson bracket. They admit two first integrals: the energy integral, H = (cid:104)P,P(cid:105) + 1 2 1 2 (cid:104)G,Ω(cid:105) +U(R), (2) and the modulus of the total angular momentum L, which, with respect to the body frame, is given by: L2 = (cid:104)L,L(cid:105), where L := R× P + G. (3) c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 RGΩPZXYThe total angular momentum with respect to the inertial frame, i.e., (4) is a constant vector. The time evolution of the attitude matrix A is given by the following kinematic equations: l = AL, where, as in Go´zdziewski & Maciejewski (1999), for a vector X = [X1,X2,X3]T , we denote: d dt A = A(cid:98)Ω,  0 −X3 X3 −X2 X2 0 −X1 X1 0  . (cid:98)X := (5) (6) If the rigid body B does not possesses any additional symmetry, then the relative equilibria of the considered two rigid body prob- lem are just the equilibria solutions (critical points) of the reduced equations (1). They were investigated by many authors, see Wang et al. (1991); Scheeres (2006) and references therein. The most in- triguing result of these investigations is the discovery of the so- called non-great-circle, or non-Lagrangian solutions. If the system is in a relative equilibrium, then the relative orbit is circular. How- ever, in a case of non-Lagrangian equilibrium, the orbital planes of the point P and the rigid body B are mutually parallel but do not coincide. These classes of equilibria have been discovered by Abulnaga & Barkin (1979). Later on, they were re-discovered by Wang et al. (1991). 3 THE SYMMETRIC KINOSHITA PROBLEM From hereafter, we shall assume that the rigid body B is dynami- cally symmetric with respect to its third principal axis, i.e., we set I = diag(A,A,C). Moreover, we assume also that the gravity field of B is symmetric with respect to the same axis and that it can be expressed through the following condition: R1 ∂U ∂R2 (R)− R2 ∂U ∂R1 (R) = 0. (7) (We note, for a further reference, that the equatorial plane of the rigid body is defined in such a way, that it contains the mass centre and this body and the symmetry axis is normal to that plane). Under these assumptions, equations (1) have an additional first integral G3. The existence of this first integral is connected with the fact that now the reduced equations (1) have one-dimensional symme- try group. To show this explicitly, we need to introduce additional notations and to state or recall some elementary facts. By SO(3,R) we denote the special orthogonal group, i.e., the group of 3× 3 matrices A, satisfying AAT = E, and detA = 1. Its Lie algebra so(3,R) consists of 3 × 3 antisymmetric matrices. It can be identified with R3 by the following isomorphism: R3 (cid:51) X (cid:55)−→(cid:98)X :=  0 −X3 X3 −X2 X2 0 −X1 X1 0  ∈ so(3,R). Later on, we use several identities which we collect as the following statements (e.g., Maciejewski 1995). Proposition 3.1. For arbitrary X, Y ∈ R3, and A ∈ SO(3,R), the following identities are hold true: (i) (cid:98)XT = −(cid:98)X = (cid:91)(−X). (ii) (cid:98)XX = X× Y = −(cid:98)YX. c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 The two-body problem of Kinoshita revisited 3 (iii) (cid:98)X(cid:98)Y = YXT − XT YE. (iv) (cid:92)X× Y =(cid:98)X(cid:98)Y−(cid:98)Y(cid:98)X. (v) (cid:100)AX = A(cid:98)XAT . (vi) A(X× Y) = A(cid:98)XY = (AX)× (AY), where E denotes the 3× 3 identity matrix. All rotation about the third axis of the body form a subgroup of SO(3,R) isomorphic with SO(2,R). We show that equations (1) possess SO(2,R) symmetry. That is, they do not change its form after the following transformation: [R,P,G] = [A3R∗,A3P∗,A3G∗], (8) where A3 is an arbitrary matrix of rotation about the third axis of the body, i.e., a matrix of the following form: a −b b 0 a 0 0 0 1  where a2 + b2 = 1. A3 = (9) Let us notice that A3I = IA3 (i.e.,I commutes with A3), hence we have, d dt R = A3 d dt R∗ = (A3R∗)× (I−1A3G∗) + A3P∗. R∗ × I−1G∗(cid:105) (cid:104) , This equation may be rewritten as (A3R∗)×(I−1A3G∗) = (A3R∗)×(A3I−1G∗) = A3 where the last equality follows from the sixth identity given in Proposition 3.1. Hence, d dt R∗ = A3 [R∗ × Ω∗ + P∗] , A3 where Ω∗ = I−1G∗, and, finally, R∗ = R∗ × Ω∗ + P∗, d dt and this shows that the first equation in (1) is invariant with respect to variables change (8). In a similar manner, we show that the re- maining two equations have the same property. 4 RELATIVE EQUILIBRIA 4.1 General theory of the relative equilibria Let us recall a formal, geometric definition of a relative equilib- rium of a system with symmetry (see, for instance Marsden & Ratiu 1994), which is crucial to perform our analysis: a relative equilibrium is a solution of the system represented by the phase- space curve which is an orbit of a point under the action of a one- dimensional subgroup of the symmetry group of the system. In our case, the symmetry group is SO(2,R) which is identi- fied with matrices of the form (9). It is a one-dimensional group. Let us put a = cosϕ and b = sinϕ in equation (9). With this param- eterization, we denote elements of SO(2,R) by A3(ϕ). Hence, an orbit of a point [R0,P0,G0]T under the action of SO(2,R) is the following set: (cid:110) [A3(ϕ)R0,A3(ϕ)P0,A3(ϕ)G0] ∈ R9 ϕ ∈ [0,2π) (cid:111) , where R0,P0,G0 are constant vectors. Thus, the relative equilib- rium of the symmetric Kinoshita problem is a solution to equa- tions (1), which has the following general form:  R(t) = A3(Nt)R0, P(t) = A3(Nt)P0, G(t) = A3(Nt)G0, (10) 4 M. Vereshchagin, A. J. Maciejewski & K. Go´zdziewski that Ω(t) = I−1G(t) = where N is the real number. Notice, A3(Nt)Ω0, where Ω0 = I−1G0. In other words, the relative equi- librium is a special periodic solution to the equations of motion (1). From the above formulae we can deduce more useful conclusions on the relative equilibrium: (i) vectors R(t), P(t), G(t) and Ω(t) have constant lengths, (ii) their "third" components R3(t), P3(t), G3(t) and Ω3(t) are constant, (iii) all angles between vectors R(t), P(t), G(t) and Ω(t) are constant. Let us notice that the relative equilibrium (10) does not determine an unique set of [R0,P0,G0]. In fact, instead of [R0,P0,G0], we can take [A(ϕ)R0,A(ϕ)P0,A(ϕ)G0], with arbitrary chosen constant ϕ. One would like to perform a reduction of (1) in such a way that in the reduced system, the relative equilibrium corresponds to an "usual" equilibrium point (i.e., which is understood as the criti- cal point of the equations of motion). It is possible, however, it may lead to singularities. For example, let us introduce three sets of cylindrical coordinates: R = (ρcosϕ,ρsinϕ,R3)T , P = (pcosυ, psinυ,P3)T , Ω = (ωcosψ,ωsinψ,Ω3)T . Using these coordinates, we may write: =(cid:0)Uρ cosϕ,Uρ cosϕ,U3 (cid:1)T , ∂U ∂R where, Uρ = ∂U ∂ρ , and U3 = ∂U ∂R3 . With respect to these new variables, system (1) reads as follows: (cid:3) , υ = p−1(cid:2)P3ωcos(ψ− ϕ) +Uρ sin(ψ− ϕ)− pΩ3 ρ = pcos(υ− ϕ)− zωsin(ψ− ϕ), z = P3 + ρωsin(ψ− ϕ), d dt d dt d dt d p = −P3ωsin(ψ− ϕ)−Uρ cos(υ− ϕ), dt d P3 = pωsin(ψ− υ)−U3, dt d dt d dt d dt ω = A−1(cid:2)zUρ − ρU3 ψ = (Aω)−1(cid:2)(cid:0)zUρ − ρU3 (cid:3)sin(ψ− ϕ), (cid:1)cos(ψ− ϕ)− (A−C)ωΩ3 ϕ = ρ−1 [psin(υ− ϕ) + zωcos(ψ− ϕ)− ρΩ3] , (cid:3) ,  Hence, finally we have one equation less than in (1) because Ω3 is constant. Note that the right hand sides of the equations depend on the difference between the angles but not on the angles themselves. Thus we can introduce the following new set of variables: α = ψ− ϕ, β = υ− ϕ, and then we obtain: ρ = pcos(β)− zωsin(α), d dt d z = P3 + ρωsin(α), dt d p = −P3ωsin(α− β)−Uρ cos(β), dt d P3 = pωsin(α− β)−U3, dt d dt d dt (cid:3)sin(α), ω =(cid:2)zUρ − ρU3 (cid:1)ω−1 cos(α) + (CΩ3− α =(cid:0)zUρ − ρU3 β = p−1(cid:2)P3ωcos(α− β) +Uρ sin(β)(cid:3)− Aω)[psin(β) + zωcos(α)]ρ−1, ρ−1 [psin(β) + zωcos(α)] , ϕ = 0. d dt d dt  The first seven equations in the above set represent a system of equations which flows from the reduction (1) by the symmetry. Now, a relative equilibrium of (1) corresponds to an equilibrium point of the above system. However, this system has singularities when one of variables ρ, ω or p vanishes. Because of this fact it would be very difficult to perform global, qualitative analysis which is our primary goal. 4.2 The geometric reduction of the system A possibility that the reduction may introduce singularities to the equations of motion inspired us and, in fact, forced us to choose a particular approach which we describe below in detail. Clearly, instead of using any local variables, it is much more convenient to derive the necessary and sufficient conditions for the existence of relative equilibria in terms of the original global and non-singular variables (R,P,G). Let us assume that a relative equilibrium is given by (10). Our aim is to find equations determining (R0,P0,G0), and N. At first, we may notice that: A3(Nt) = A3(Nt)(cid:98)N, d dt where N = [0,0,N]T . Thus we may easily derive that: A3(Nt)R0 = A3(Nt)(cid:98)NR0 = A3(Nt)(N× R0). d dt R(t) = d dt On the other hand, we have also: R(t) = R(t)× Ω(t) + P(t) = A3(Nt)(R0 × Ω0 + P0), d dt as the system has SO(2,R) symmetry. So, we have N× R0 = R0 × Ω0 + P0. We proceed in a similar way with other components of the phase variables defining the relative equilibrium, i.e., with P(t) and G(t). Finally, we obtain the following equations: N× R0 = R0 × Ω0 + P0, N× P0 = P0 × Ω0 − ∂U ∂R0 N× G0 = G0 × Ω0 + R0 × ∂U ∂R0 , .  c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 This system can be rewritten in a more compact form as follows: (cid:101)Ω× R0 = P0, (cid:101)Ω× P0 = − ∂U (cid:101)Ω× G0 = R0 × ∂U ∂R0 ,  (11) where, (cid:101)Ω = Ω0 + N, is the instantaneous angular velocity of the ∂R0 , rigid body. System (11) defines nine equations which must be satisfied by ten unknowns. We found that there are several ways to overcome this inconvenience. In our further procedure, we choose the follow- ing approach. We look for the relative equilibria which exist for a given value of the angular momentum of the system, chosen as the model parameter. Because after the elimination of P0, we obtain: Hence, our goal is to find all solutions to the following nonlinear equations: . L2 = L2 = (R× P + G)2 = (R0 × P0 + G0)2 , (cid:17)2 (cid:16)(cid:101)ΩR2 0 − R0(cid:104)(cid:101)Ω,R0(cid:105) + G0 (cid:101)Ω× R0 = P0, (cid:101)Ω(cid:104)(cid:101)Ω,R0(cid:105)− R0(cid:101)Ω2 = − ∂U (cid:101)Ω× G0 = R0 × ∂U (cid:16)(cid:101)ΩR2 0 − R0(cid:104)(cid:101)Ω,R0(cid:105) + G0 (cid:17)2 = L2. ∂R0 ∂R0 , (12a) (12b) (12c) (12d) , In the above equations, L, and the third component of G0 are now fixed parameters of the reduced model. 4.3 Geometric description of the equilibria solutions The equations of motion (1) describe the relative motion of the bod- ies in the rigid body fixed frame. Apparently, this makes it difficult to obtain the geometrical and physical interpretation. In fact, we can deduce all necessary information quite easily. First of all, let us recall that the total angular momentum lll is a constant vector. So, we can choose an inertial frame in such a way that its third axis is directed along that vector. If R(t) = A3(Nt)R0 describes the relative orbit in a relative equilibrium, and rrr(t) is the corresponding equilibrium vector, then we have: lll ·rrr(t) = (cid:104)l,r(t)(cid:105) = (cid:104)AL,AA3(Nt)R0(cid:105) = (cid:104)L,A3(Nt)R0(cid:105). Moreover, the following general relation holds true: L = A3(Nt)(R0 × P0 + G0), so, finally, we have: lll ·rrr(t) = (cid:104)R0 × P0 + G0,R0(cid:105) = (cid:104)G0,R0(cid:105). (13) We can conclude that in a relative equilibrium, the projection of the relative radius vector onto the total angular momentum vector is constant. An important conclusion follows immediately: if this projection is non-zero, then orbits of the point and the rigid body lie in different planes. Let zzz(t) denotes the unit vector along the symmetry axis of the body. Then lll·zzz(t) =(cid:104)L,Z(cid:105) =(cid:104)A3(Nt)(R0 × P0 + G0),Z(cid:105) =(cid:104)R0 × P0 + G0,Z(cid:105), c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 The two-body problem of Kinoshita revisited 5 because Z = [0,0,1]T , and so, A3(Nt)Z = Z. Thus, in a relative equilibrium, the projection of the symmetry axis onto the total an- gular momentum is also constant. Let us introduce the orbital reference frame with axes parallel to rrr(t), rrr(t) and ccc(t) := rrr(t)× rrr(t). Because the relative orbit in a relative equilibrium is circular, this frame is orthogonal. We show that the projection of the symmetry axis zzz(t) onto the axis of this frame is also constant. In other words, we are going to prove that in the relative equilibrium the symmetry axis of the rigid body has a fixed orientation with respect to the orbital reference frame. To shorten notation, we shall write A3 instead of A3(Nt). Let B = AA3. Then we have: B = AA3 + A A3 = A(cid:98)ΩA3 + AA3(cid:98)N. d dt But in a relative equilibrium we have Ω = A3Ω0. By the fifth prop- erty given in Proposition 3.1:(cid:98)Ω = A3Ω0AT B = B(cid:98)(cid:101)Ω, hence we obtain that: 3 , d dt where(cid:101)Ω = Ω0 +N is a constant vector. Now, it is easy to show that: r(t) = AR(t) = BR0, r(t) = B((cid:101)Ω× R0), r(t)× r(t) = B(R0 × ((cid:101)Ω× R0)). Because z(t) = AZ = BZ, and A3Z = Z, we have: rrr(t)·zzz(t) = (cid:104)(cid:101)Ω× R0,Z(cid:105) = (cid:104)P0,Z(cid:105), rrr(t)·zzz(t) = (cid:104)r(t),z(t)(cid:105) = (cid:104)BR0,BZ(cid:105) = (cid:104)R0,Z(cid:105), ccc(t)·zzz(t) = (cid:104)R0 × ((cid:101)Ω× R0),Z(cid:105). The above formulae prove our initial claim. (14) (15) 5 CONDITIONS FOR THE RELATIVE EQUILIBRIA In this section, we show that all relative equilibria can be deter- mined by solutions of certain non-linear system comprising of only two non-linear scalar equations. We shall also prove that there ex- ist three classes of these relative equilibria in the dynamical model which we consider. In order to simplify notation while working with coordinates, we shall skip index "0" when denoting coordinates of vectors R0, P0, G0 from hereafter. Thus, we set R0 = [R1,R2,R3]T , P0 = [P1,P2,P3]T , Ω0 = [Ω1,Ω2,Ω3]T , (cid:101)Ω = [Ω1,Ω2,Ω3 + N]T , G0 = [AΩ1,AΩ2,CΩ3]T , ∂U ∂R0 ∂U ∂R (R0) = [U1,U2,U3]T . := The fact that the body is symmetric implies that: R1U2 − R2U1 = 0, (16) 6 M. Vereshchagin, A. J. Maciejewski & K. Go´zdziewski . (17) with accord to (7). Taking the vector product of both sides of equa- tion (12b) with R0, we obtain: (cid:104)(cid:101)Ω,R0(cid:105)(cid:101)Ω× R0 = R0 × ∂U (cid:104)(cid:101)Ω,R0(cid:105)(cid:104)Z,(cid:101)Ω× R0(cid:105) = 0. ∂R0 Hence, taking into account (16), we have: Thus, we have to consider two different cases. If (cid:104)(cid:101)Ω,R0(cid:105) = 0 then from (17) it follows immediately that: R0 × ∂U or, (cid:104)Z,(cid:101)Ω× R0(cid:105) = 0, and this gives ∂R0 = 0, Ω1R2 − Ω2R1 = 0. In the first case, the gradient of U at point R0 is parallel to the radius vector. According to Scheeres (2006), such a point is called locally central. We shall say that a relative equilibrium is locally central if point R0 is locally central. In our analysis, we assume that the gradient of the potential does not vanish at a considered point R0. Thus, a relative equilib- rium is locally central if and only if (cid:104)(cid:101)Ω,R0(cid:105) = 0. In fact, if R0 is locally central, then from (17) it follows that either (cid:104)(cid:101)Ω,R0(cid:105) = 0, or (cid:101)Ω×R0 = 0. If the second possibility occurs, then, by (12a), P0 = 0, and, as equations (11) imply, the gradient of U at R0 vanishes. But, according with our assumptions, it is impossible. At the final stage of our analysis, it is convenient to use cylin- drical coordinates (ρ,ϕ,R3) instead of (R1,R2,R3). The potential U expressed with respect to the cylindrical coordinates depends only on ρ, and R3; still, as a function of these two arguments it will be denoted by the same symbol U. It will be clear from the context which coordinates are in use. 5.1 Locally central relative equilibria If R0 is locally central, then, as we have shown, (cid:104)(cid:101)Ω,R0(cid:105) = 0, and of course: R0 × ∂U ∂R0 = 0. Thanks to this fact, the basic system of the equations of motion (12) may be simplified considerably, and it reads as follows: Equation (18c) is equivalent to the following two scalar equations: Therefore, either Ω1 = Ω2 = 0, or A(cid:101)Ω3 −CΩ3 = 0. We investigate both cases separately. , ∂U ∂R0 P0 =(cid:101)Ω× R0, R0(cid:101)Ω2 = (cid:101)Ω× G0 = 0, (cid:17)2 (cid:105) (cid:105) Ω2 = 0, 0 + G0 (cid:16)(cid:101)ΩR2 (cid:104) A(cid:101)Ω3 −CΩ3 (cid:104) A(cid:101)Ω3 −CΩ3 Ω1 = 0. = L2.  (18a) (18b) (18c) (18d) 5.1.1 Case Ω1 = Ω2 = 0 (cylindrical precession) If Ω1 = Ω2 = 0 then condition (cid:104)(cid:101)Ω,R0(cid:105) = 0 reduces to (cid:101)Ω3R3 = 0. But (cid:101)Ω3 (cid:54)= 0, otherwise (cid:101)Ω2 = 0, and, as (18b) shows, the gradient of U vanishes and then R3 = 0. Moreover, R1 and R2 do not van- ish simultaneously, otherwise R0 = 0. Hence, we can safely use cylindrical coordinates (ρ,z ≡ 0)1 because ρ > 0. We introduce two functions of ρ given by: f (ρ) := U3(ρ,0), and g(ρ) := Uρ(ρ,0). Now, equations (18b) and (18d) lead to the following system: ρ(cid:101)Ω2 (cid:104) ρ2(cid:101)Ω3 +CΩ3 (cid:105)2 f (ρ) = 0, 3 = g(ρ), = L2 (19) Through appropriate simplification, we find that the above system is equivalent to the following ones: (cid:101)Ω3 = −ε L + εCΩ3 ρ2 , (20) f (ρ) = 0, (L + εCΩ3)2 = ρ3g(ρ), where parameter ε ∈ {−1,1}. The above considerations can be summarized in the following way. In the considered case, the relative equilibrium exists if and only if there exists a solution ρ > 0 of equations: f (ρ) = 0, (L + εCΩ3)2 = ρ3g(ρ). (21) If such a solution exists then we define: R1 = ρcosϕ, R2 = ρsinϕ, α2 = ρ3Uρ, β = α ρ2 , where ϕ ∈ [0,2π] can be chosen arbitrarily. Then the relative equi- librium is determined through:  , G0 = C  0  , N = β− Ω3.  , P0 = β R1 R2 0 −R2 R1 0 R0 = 0 Ω3 From the above formulae it immediately follows that point P and the mass centre of the rigid body B move in one plane. In fact, by (13), we have: lll ·rrr(t) = (cid:104)G0,R0(cid:105) = 0. Moreover, by (15), we have also: rrr(t)·zzz(t) = (cid:104)R0,Z(cid:105) = 0, rrr(t)·zzz(t) = (cid:104)P0,Z(cid:105) = 0. Hence, the axis of symmetry of the rigid body is perpendicular to the plane of the relative orbit, see Fig. 2. We called this kind of the relative equilibrium as the cylindrical precession from hereafter. In the paper of Kinoshita, it is named, after Duboshin (1959), as the float solution. Regarding existence conditions of this solution, let us notice that system (21) is overdetermined, we have two equations for one variable ρ. However, if the body is symmetric with respect to the equatorial plane, then U is an even function of R3. In this instance, we have: f (ρ) = (ρ,0) ≡ 0, ∂U ∂R identically. Thus, in this case, the cylindrical precession exists if and only if the second equation of system (21) has a solution. Moreover, if the potential is such that there exists ρ > 0, for which 1 Note that due to the symmetry, the third angular coordinate φ is irrelevant here. c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 The two-body problem of Kinoshita revisited 7 Figure 3. The plot of f (ρ) crosses ρ-axis for ρ (cid:39) 0.26. The values of parameters are chosen in the following way: µ1 = 0.15, µ2 = 0.3, µ3 = 0.55, where µi = m1/∑3 j=1 m j and mi is mass of the i-th sphere. See the text for details. whenever they exist, are not isolated. In terms of the cylindrical co- ordinates, we may rewrite (24) as follows: L2 (cid:0)R2 3 + ρ2 + A(cid:1)2 = Uρ, (cid:0)R2 3 + ρ2 + A(cid:1)2 = U3, L2 ρ R3  (25) A solution to the above system gives us R0. Because, due to the symmetry, we may choose freely the polar angle, then we can also assume, without any loss of generality that R2 = 0. Vector G0 is perpendicular to R0, and its modulus is fixed by (23). In the generic case, R1R3 (cid:54)= 0, and we can put: G0 = a1A[0,1,0]T + a2 [R3,0,−R1]T , A R0 where a1 and a2 are arbitrary real numbers satisfying the following relation: a2 1 + a2 2 = L2 0 + A)2 . (R2 For each choice of a1 and a2, we have G0, and then the correspond- ing value of P0 is given by (22). Thus, a solution to equations (25) gives us a one-parameter family of the relative equilibria. From the above formulae it immediately follows that point P, and the mass centre of the body B move in one plane. In fact, we have: lll ·rrr(t) = (cid:104)G0,R0(cid:105) = 0. Moreover, as in the first case of the cylindrical precession, we can determine the orientation of the symmetry axis using (15). We have: rrr(t)·zzz(t) = (cid:104)R0,Z(cid:105) = R3, ccc(t)·zzz(t) = (cid:104)R0 × ((cid:101)Ω× R0),Z(cid:105) = −a2R0R1. rrr(t)·zzz(t) = (cid:104)P0,Z(cid:105) = P3 = −a1R1, Figure 2. Cylindrical precession in the inertial reference frame. Shaded bodies show the attitude of the rigid body B and the point P in other mo- ments of time. g(ρ) > 0, then we always may find parameters of the problem such that equations (21) are fulfilled. If the body is not symmetric with respect to the equatorial plane, the first equation of system (21) is not satisfied identically. Thus a question emerges: does there exist an axially symmetric body which is not symmetric with respect to the equatorial plane, and for which equation: U3(ρ,0) = 0, has a solution ρ > 0? To answer this question, let us consider an axisymmetric rigid body composed of three uniform spheres whose centers of masses are placed at the same line. The masses of these spheres are chosen in a way to guarantee that the potential is not symmetric with respect to the equatorial plane. The resulting func- tion f (ρ) is depicted in Fig. 3. Since its graph crosses the ρ-axis for ρ > 0, there exist such points which satisfy the first condition of system (21), indeed. Then the parameters of the system can always be chosen to fulfill the last condition of (21). Thus, the cylindrical precessions may exist in the system with the rigid body, which is not symmetric with respect to its equatorial plane. 5.1.2 Case A(cid:101)Ω3 = CΩ3 (inclined co-planar precession) Condition A(cid:101)Ω3 = CΩ3 implies immediately that Moreover, now (cid:104)(cid:101)Ω,R0(cid:105) = 0, is equivalent to (cid:104)G0,R0(cid:105) = 0, because A(cid:101)Ω = G0. We have even more, because now equation (18b) can be C− A A Ω3. N = rewritten in the following form: 1 A P0 = and equation (18d) reads as: (cid:16)(cid:101)ΩR2 G0 × R0. (cid:17)2 G2 A2 (R2 0 L2 = 0 + A)2. Therefore, equation (18b) can be rewritten as follows: 0 + G0 = L2 0 + A)2 R0 = (R2 ∂U ∂R0 . (22) (23) (24) A solution of this equation gives R0. However, this form is not con- venient for further analysis because solutions to the above equation, c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 r(t)r(t)c(t)zxyl8 M. Vereshchagin, A. J. Maciejewski & K. Go´zdziewski for i = 1,2. (30) Now, let us consider equation (26c). It is easy to see that the third component of this equation is fulfilled identically. Using (28), the first two components can be rewritten in the following form: Ri (cid:16) A(cid:101)Ω3 −CΩ3 (cid:16) A(cid:101)Ω3 −CΩ3 As R1 and R2 do not vanish simultaneously, the above system is equivalent to just one equation: UρU3 = Ri (cid:17) (cid:0)ρUρ + R3U3 (cid:17) UρU3 =(cid:0)ρUρ + R3U3 (cid:2)(cid:0)A + R2 (cid:1)U3 + ρR3Uρ (cid:3)2 (cid:1)(cid:0)ρU3 − R3Uρ (cid:1)(cid:101)Ω3, (cid:1)(cid:0)ρU3 − R3Uρ (cid:1)(cid:101)Ω3. (cid:19)2 (cid:18) ρUρ(cid:101)Ω3 + G3 , + Next, substituting expressions (27) and (28), equation (26d) is rewritten as: (31) L2 = 3 ρUρ + R3U3 Thus, eventually, we obtain the final subsystem that makes it pos- sible to determine the phase-space variables of the relative equilib- rium, in quite a compact form: Finally, we use (30) to eliminate (cid:101)Ω3 from equations (29) and (31).  (32) 3 K =(cid:2)AUρU3 + K(cid:0)R3Uρ − ρU3 [AU3 + R3K]2 , L2U2 3U2 U2 ρ +U2 3 (cid:1)(cid:3)2 3 K = C2Ω2 (cid:16) (cid:17) , where we introduced a new variable: K ≡ ρUρ + R3U3. U3K A solution to the above conditions gives us R0. The remaining equi- librium variables can be determined by equations (27), (28), and (29), respectively. For this solution we can find, after tedious simplifications, that lll ·rrr(t) = (cid:104)G0,R0(cid:105) = (cid:0)R3Uρ − ρU3 Uρ (AU3 + R3K) (cid:1) . According with our assumptions, ρ (cid:54)= 0, and hence Uρ (cid:54)= 0. Thus, the right hand side of the above equation vanishes if either R3Uρ − ρU3 = 0, or AU3 + R3K = 0. In the first case, R0 is locally central point which we excluded from our considerations. In the second instance, from equation (32), we obtain that L = 0, so it is also excluded from our considerations. We conclude that non-locally central relative equilibrium is non-Lagrangian ones. In order to determine the orientation of the symmetry axis in the orbital frame which is specific for this type of relative equilibria, we can easily find that: rrr(t)·zzz(t) = (cid:104)R0,Z(cid:105) = R3, rrr(t)·zzz(t) = 0, ccc(t)·zzz(t) = (cid:104)R0 × ((cid:101)Ω× R0),Z(cid:105) = ρUρ(cid:101)Ω3 . Thus, in the orbital reference frame, the axis of symmetry always lies in the plane formed by the relative position vector and the nor- mal to the orbital plane. We call this kind of the regular preces- sion as the conic precession from hereafter. Notice, that in the con- sidered case, vector ccc(t) is not perpendicular to the orbital planes. Moreover, we have: lll ·zzz(t) = (cid:104)R0 × ((cid:101)Ω× R0) + G0,Z(cid:105) = +CΩ3. ρUρ(cid:101)Ω3 Thus, the axis of symmetry is inclined to the orbital plane. The ge- ometry of this solution is presented in Fig. 5. Unlike the locally central cases, even for bodies with simple potentials, it is difficult to decide whether equations (32) have a solution. In order to jus- tify that the relative equilibria belonging to the conic precession c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 Figure 4. The inclined co-planar precession in the inertial reference frame. Shaded meshes are for an illustration of the attitude and relative positions of the rigid body B and the point P at different moments during the motion of the system. Thus, in general, the symmetry axis of the rigid body is inclined with respect to any axis of the orbital reference frame. We shall call this class of relative equilibria as the inclined, co-planar regular precession, and its geometry is illustrated in Fig. 4. 5.2 Non-locally central case (conic precession) As it has been already established, the non-locally central equilibria are characterized by P3 = R2Ω1−R1Ω2 = 0. It appears that in such an instance, instead of using system (12), it is better to analyse equations in which vector P is not eliminated. Hence, we consider the following system of equations: (cid:101)Ω× R0 = P0, (cid:101)Ω× P0 = − ∂U (cid:101)Ω× G0 = R0 × ∂U ∂R0 (R0 × P0 + G0) = L2. ∂R0 , , (26a) (26b) (26c) (26d) Our aim is to find all solutions of this system which are not locally central. Thus, we look for solutions for which none of variables ρ, R3,(cid:101)Ω3 vanishes. Hence, we can express U1 and U2 in terms of Uρ, i.e., Ui = Ri ρ Uρ, for i = 1,2. Since P3 vanishes, from the first two components of equations (26b) it follows that and P2 = R1 . (27) Uρ(cid:101)Ω3ρ Knowing P1 and P2, from the first two components of equation (26a), we find: , for i = 1,2. (28) Substituting (27), (28) into the third component of equation (26b), we obtain: (cid:1)(cid:101)Ω2 3 = U2 ρ , (29) , Uρ(cid:101)Ω3ρ P1 = −R2 ρ(cid:101)Ω2 ρR3(cid:101)Ω3 3 −Uρ (cid:0)ρUρ + R3U3 Ωi = Ri θr(t)r(t)c(t)zxylThe two-body problem of Kinoshita revisited 9 does not vanish identically. Hence, by the implicit function theo- rem, system (32) has a solution ρ = ρ(Ω3), R3 = R3(Ω3), such that ρ(0) = 0, and R3(0) = R∗ 3. Moreover, such solution is unique. We underline an important fact that, even if the body is sym- metric with respect to the equatorial plane, the described conic pre- cession does exist. In other words, splitting of the orbital planes can be induced not only by an asymmetric mass distribution, it can be also induced by a proper, particular rotation of the rigid body. Remarkably, the first type of solutions has been known since Ab- ulnaga & Barkin (1979), who constructed them for a class of non- symmetric potentials and further analysed their consequences in the motion model of the Earth-Moon system (Barkin 1980). 6 A COMPARISON WITH THE WORK OF KINOSHITA Our analysis presented in the previous sections lead to somewhat different conclusions than those ones in (Kinoshita 1970), and in this section we give an explanation and overview of these discrep- ancies. Kinoshita found three classes of stationary solutions: (i) the "float" case, when the axis of symmetry of the rigid body is always perpendicular to the orbital plane, (ii) the "spoke" case, when the axis of symmetry lies along the relative position vector, (iii) the "arrow" case, when the axis of symmetry lies in the plane formed by tangential and normal to the orbital plane vectors. Now we may show that all these solutions are, in our terminology, locally central and Lagrangian, i.e., the point mass P, and the mass centre of the rigid body B move in one plane. Since for the description of the system there has been chosen the reference frame fixed at mass centre of the point body P, the potential of the body, besides the distance from the mass centre, depends also on the orientation of the rigid body. Following Ki- noshita, we denote a parameter, which describes the orientation, as ν. In this designation, ν is an angle under which the point body P is "seen" from the mass centre of the rigid one B. Note an impor- tant fact that the "float" and "arrow" types of motion correspond to a case when Uν vanishes. Let us determine the physical meaning of that condition in terms of our parameterization. For that reason, we will use an expansion of the rigid body potential through se- ries in terms of the Legendre polynomials, the same as in Kinoshita (1970). This representation of the potential of B is the following: (cid:16) a (cid:17)k r U = − 1 r ∞ ∑ k=0 Jk Pk(ν), (34) where ν is the cosine of the angle between the symmetry axis of the body and the relative radius vector, and Pk is the Legendre polyno- mial. For the "float" and the "arrow" types of motion, Uν vanishes. Hence, Uν = 1 r ∞ ∑ k=0 Jk for a certain value of ν. reads as follows: U = − 1 R0 ∞ ∑ k=0 Jk (cid:16) a (cid:17)k r P(cid:48) k(ν) = 0, (cid:18) a (cid:19)k R0 (cid:18) R3 (cid:19) R0 . Pk In terms of variables used in our paper, the expansion of (34) Figure 5. Conic precession in the inertial reference frame. Shaded bodies show the attitude of the rigid body B and the point P in other time points. class may really exist, we apply the following reasoning. If ρ, as well as Ω3, tends to zero, then the conic precession tends to a par- ticular case of the inclined planar regular precession described in Section 5.1.2. In fact, if ρ = Ω3 = 0, then Uρ = 0, so the second equation in (32) is satisfied identically, and the first one coincides exactly with the equation describing corresponding case of the in- clined co-planar solution. Thus, we can take this particular solu- tion as the zero-th order approximation (with respect to Ω3) of the conic precession solution, and we apply the perturbative approach to solve the system (32). Thus, the zero-th order solution is given through: where R∗ 3 is a solution of the following equation ρ = 0, R3 = R∗ 3, (cid:1)2 R3. L2 U3ρ=0 = 3 (cid:0)A + R2 (cid:12)(cid:12)ρ=0,R3=R∗ (cid:12)(cid:12)ρ=0 = Uρ3 (cid:20) 0 3 J = J12 0 (cid:12)(cid:12)ρ=0 = 0, (cid:21) , Uρ To shorten the notation, we define: U3ρ=0,R3=R∗ Taking into account that = F, Uρρ 3 = V, U33ρ=0,R3=R∗ 3 = W. we can find that the Jacobian at the zeroth order solution has the form of: where (cid:20)(cid:16) 3R∗2 3 (cid:0)A + R∗2 (cid:1)7 J12 = − L6R∗3 (cid:0)A + R∗2 J21 = − L2R∗ 3 3 3 J21 (cid:17) 3 − A (cid:20) (cid:16) (cid:1)4 V A + R∗2 3 (cid:17)3(cid:21) (cid:16) (cid:21) (cid:17)3 − L2A , , (33a) (33b) L2 +W A + R∗2 3 The determinant of this Jacobian, given through: Let us recall that a point is locally central if and only if detJ = −J12J21, c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10 ρU3 − R3Uρ = 0. θr(t)r(t)c(t)zxyl10 M. Vereshchagin, A. J. Maciejewski & K. Go´zdziewski Thus, we have: ρU3 − R3Uρ = ρ R4 0 ∞ ∑ k=0 Jk (cid:18) a (cid:19)k R0 (cid:18) R3 (cid:19) R0 P(cid:48) k = ρ R4 0 Uν = 0. (35) If Uν = 0 then it means that the considered solution is locally cen- tral. Condition Uν = 0 includes all possible locally central cases except of that ones when ρ vanishes. That is why the "arrow" and the "float" type relative equilibria found by Kinoshita (1970), un- der assumption that Uν = 0, are locally central. Moreover, it is easy to see , the "spoke" solution is also locally central. In fact, for this solution we have ρ =Uρ = 0. Thus, all solutions found by Kinoshita are locally central. The "float" equilibrium coincides exactly with our first so- lution as in both of these cases the axis of symmetry is always perpendicular to the orbital plane. The "arrow" and the "spoke" motions are just particular cases of our second solution. Indeed, the second solution, when Ω3 vanishes, becomes the "spoke" motion, i.e., the axis of symmetry lies along the relative position vector; if we have R3 = 0, then it coincides with the "arrow" motion. Quite surprisingly, in the paper of Kinoshita, we did not find any other cases corresponding to the inclined planar solution. It seems that investigating the "arrow" case, Kinoshita assumed for simplicity, that the body B is symmetric with respect to the equatorial plane. Moreover, then he used generally wrong implication saying that Uν = 0 means ν = 0. Let us notice that in our representation ν = 0 is equivalent to R3 = 0. To summarize this Section, Kinoshita (1970) did not find the non-Lagrangian solutions because he implicitly assumed that one can always choose an inertial reference frame in such a way that the relative orbit lies in its (x,y)-plane. In fact, this is equivalent to a priori assumption that all stationary solutions must be Lagrangian ones. 7 CONCLUSIONS In this paper, we performed global, geometric analysis of the sta- tionary motions in the unrestricted problem of a point mass and an axially symmetric rigid body (the Kinoshita problem). Our aim was to determine all possible relative equilibria in this problem. We have shown that three types of stationary motions can be distin- guished. Two of them, the cylindrical precession and the inclined planar motion, are Lagrangian and are characterized by co-planar orbits of the mass centers of the bodies, whereas the third type of solutions, the conic precession, is non-Lagrangian. In the cylindrical precession the axis of symmetry of the rigid body is always perpendicular to the orbital plane. This type of sta- tionary motion was also found by Kinoshita (1970), where it is called the "float" motion. In the inclined planar precession, the axis of symmetry of the rigid body is inclined to the orbit. Two special cases of such kind of motion are found by Kinoshita (1970). In the first case the axis of symmetry lies along the relative position vector, and in the second one lies in the plane formed by the tangent and the normal to the orbit. In Kinoshita (1970) these solutions are named as the "spoke" and the "arrow" motions, respectively. In the conic precession, the axis of symmetry lies in the plane formed by the relative position and the normal to the orbital plane vectors. Moreover, the point and the mass centre of the rigid body move in different parallel planes. This type of stationary motion is completely new in the problem analyzed in this work, although, as we noted above, such solutions have been constructed for specific non-symmetric potentials by Abulnaga & Barkin (1979). We found them thanks to a formulation of the problem through the minimal set of assumptions, basically implying only the form of the gravita- tional potential of the rigid body, besides the analysis of the prob- lem in the very basic settings of the Newtonian dynamics. We show that the determination of the stationary motions in the problem of Kinoshita may be reduced to solving at most two non-linear, algebraic equations. Our results can be applied to an arbitrary axially symmetric body, providing that an explicit form of its potential function is given. In our forthcoming paper (Vereshchagin et al. 2009), which is a direct continuation of this work, we perform the stability analysis of the relative equilibria found here, and we apply our approach to study the dynamics of a few specific models (i.e., choosing explicit form of the gravitational potential of the rigid body). ACKNOWLEDGMENTS We thank Antonio Elipe for the review of the manuscript. This work has been supported by the European Commission through the Astrodynamics Network under Marie Curie contract MRTN-CT- 2006-035151. REFERENCES Abulnaga M. Z., Barkin I. V., 1979, AZh, 56, 881 Barkin I. V., 1985, Kosmicheskie Issledovaniia, 23, 26 Barkin Y. V., 1980, Pis'ma Astronomicheskii Zhurnal, 6, 377 Beletskij V. V., 1975, Moscow Izdatel Moskovskogo Universiteta Pt Beletskij V. V., Ponomareva O. N., 1990, Kosmicheskie Issle- dovaniia, 28, 664 Correia A. C. M., Levrard B., Laskar J., 2008, A&A, 488, L63 Duboshin G. N., 1959, Soviet Astronomy, 3, 154 Go´zdziewski K., 2003, Celestial Mechanics and Dynamical As- tronomy, 85, 79 Go´zdziewski K., Maciejewski A. J., 1998, A&A, 339, 615 Go´zdziewski K., Maciejewski A. J., 1999, Celestial Mechanics and Dynamical Astronomy, 75, 251 Kinoshita H., 1970, PASJ, 22, 383 Maciejewski A. J., 1994, Acta Astronomica, 44, 301 Maciejewski A. J., 1995, Celestial Mech., 63, 1 Maciejewski A. J., Go´zdziewski K., 1995, Celestial Mechanics and Dynamical Astronomy, 61, 347 Markeev A. P., 1967a, Cosmic Research, 5, 318 Markeev A. P., 1967b, Cosmic Research, 5, 457 Markeev A. P., 1975, Cosmic Research, 13, 285 Markeev A. P., 1985, Kosmicheskie Issledovaniia, 23, 323 Markeev A. P., 1988, Soviet Astronomy Letters, 14, 118 Marsden J. E., Ratiu T. S., 1994, Introduction to Mechanics and Symmetry. No. 17 in Texts in Applied Mathematics, Springer Verlag, New York, Berlin, Heidelberg Scheeres D. J., 2006, Celestial Mechanics and Dynamical Astron- omy, 94, 317 Shcherbina G. A., 1989, Kosmicheskie Issledovaniia, 27, 31 Vereshchagin M. D., Maciejewski A. J., Go´zdziewski K., 2009, Stability of the relative equilibriums for axisymmetric sphere- restrcited 2-body problem., Not published yet Wang L. S., Krishnaprasad P. S., Maddocks J. H., 1991, Celestial Mech. Dynam. Astronom., 50, 349 c(cid:13) 2009 RAS, MNRAS 000, 1 -- 10
0905.4680	1	0905	2009-05-28T16:07:25	A transit timing analysis of nine RISE light curves of the exoplanet system TrES-3	[ "astro-ph.EP" ]	We present nine newly observed transits of TrES-3, taken as part of a transit timing program using the RISE instrument on the Liverpool Telescope. A Markov-Chain Monte-Carlo analysis was used to determine the planet-star radius ratio and inclination of the system, which were found to be Rp/Rstar=0.1664^{+0.0011}_{-0.0018} and i = 81.73^{+0.13}_{-0.04} respectively, consistent with previous results. The central transit times and uncertainties were also calculated, using a residual-permutation algorithm as an independent check on the errors. A re-analysis of eight previously published TrES-3 light curves was conducted to determine the transit times and uncertainties using consistent techniques. Whilst the transit times were not found to be in agreement with a linear ephemeris, giving chi^2 = 35.07 for 15 degrees of freedom, we interpret this to be the result of systematics in the light curves rather than a real transit timing variation. This is because the light curves that show the largest deviation from a constant period either have relatively little out-of-transit coverage, or have clear systematics. A new ephemeris was calculated using the transit times, and was found to be T_c(0) = 2454632.62610 +- 0.00006 HJD and P = 1.3061864 +- 0.0000005 days. The transit times were then used to place upper mass limits as a function of the period ratio of a potential perturbing planet, showing that our data are sufficiently sensitive to have probed for sub-Earth mass planets in both interior and exterior 2:1 resonances, assuming the additional planet is in an initially circular orbit.	astro-ph.EP	astro-ph	A transit timing analysis of nine RISE light curves of the exoplanet system TrES-3 N. P. Gibson, D. Pollacco, E. K. Simpson, S. Barros, Y. C. Joshi, I. Todd and F. P. Keenan Astrophysics Research Centre, School of Mathematics & Physics, Queen's University, University Road, Belfast, BT7 1NN, UK [email protected] I. Skillen and C. Benn Isaac Newton Group of Telescopes, Apartado de Correos 321, E-38700 Santa Cruz de la Palma, Tenerife, Spain D. Christian Physics & Astronomy Department, California State University Northridge, Northridge, California 91330-8268, USA M. Hrudkov´a Astronomical Institute, Charles University Prague, V Holesovickach 2, CZ-180 00 Praha, Czech Republic and I. A. Steele Astrophysics Research Institute, Liverpool John Moores University, CH61 4UA, UK ABSTRACT We present nine newly observed transits of TrES-3, taken as part of a tran- sit timing program using the RISE instrument on the Liverpool Telescope. A Markov-Chain Monte-Carlo analysis was used to determine the planet-star radius ratio and inclination of the system, which were found to be Rp/R⋆ = 0.1664+0.0011 −0.0018 and i = 81.73+0.13 −0.04 respectively, consistent with previous results. The central tran- sit times and uncertainties were also calculated, using a residual-permutation algorithm as an independent check on the errors. A re-analysis of eight previ- ously published TrES-3 light curves was conducted to determine the transit times -- 2 -- and uncertainties using consistent techniques. Whilst the transit times were not found to be in agreement with a linear ephemeris, giving χ2 = 35.07 for 15 degrees of freedom, we interpret this to be the result of systematics in the light curves rather than a real transit timing variation. This is because the light curves that show the largest deviation from a constant period either have relatively little out- of-transit coverage, or have clear systematics. A new ephemeris was calculated using the transit times, and was found to be Tc(0) = 2454632.62610 ± 0.00006 HJD and P = 1.3061864 ± 0.0000005 days. The transit times were then used to place upper mass limits as a function of the period ratio of a potential perturbing planet, showing that our data are sufficiently sensitive to have probed for sub- Earth mass planets in both interior and exterior 2:1 resonances, assuming the additional planet is in an initially circular orbit. Subject headings: methods: data analysis, stars: planetary systems, techniques: photometric individual (TrES-3) , stars: 1. Introduction Transit surveys of extrasolar planets have vastly improved our understanding of plan- etary systems in recent years, with a rapid increase in the number of new discoveries1. Transiting systems are particularly important, because, when coupled with radial velocity measurements, they allow the measurement of the mass, radius and density of the planet. While the majority of these systems are Hot Jupiters, neither ground-based transit nor ra- dial velocity surveys have reached the precision required to search for Earth-sized planets. However, Earth-sized planets may be found via high precision ground-based observations through the detection of Transit Timing Variations (TTV). A transiting planet will maintain a constant period whilst orbiting its parent star (ex- cluding tidal effects and general relativity), unless acted on by a third body. Measuring the central transit times allows us to detect perturbations in the period thus revealing the presence of another body in the system, which is the principle of the TTV method (Miralda-Escud´e 2002; Holman & Murray 2005; Agol et al. 2005; Heyl & Gladman 2007). TTV is particularly sensitive to small bodies in resonant orbits, or even exomoons (Kipping 2009) and trojans (e.g. Ford & Gaudi 2006; Ford & Holman 2007), and therefore has the potential to provide the first detection of an Earth-sized body orbiting a main-sequence star 1see http://exoplanet.eu/ -- 3 -- other than our own. Constraining a TTV signal requires many high precision light curves with high cadence. In theory we can measure TTVs to several seconds. However, in practice we are limited by correlated noise in the light curves, which may arise due to effects such as pixel-to- pixel sensitivity variations, temperature fluctuations, or changes in the observing conditions (Pont et al. 2006). Indeed, there may also be non-instrumental effects caused by brightness variations of either the target or comparison stars. We are unable to distinguish any stellar activity from other sources of correlated noise in the light curves, and therefore they have the same detrimental effects as instrumental systematics when making transit observations. Providing that these sources of correlated noise can be kept to a minimum, it is still possible to measure central transit times to better than 10s (see §4). This allows us to probe for the presence of Earth-sized planets in low-order mean-motion resonance, or more massive perturbers in non-resonant orbits (see e.g. Steffen & Agol 2005; Agol & Steffen 2007; Bean 2009). RISE (Rapid Imager to Search for Exoplanets) is a fast camera mounted on the Liverpool Telescope (LT) on La Palma, primarily to obtain high precision light curves of transiting exoplanets (see Steele et al. 2008; Gibson et al. 2008). It was commissioned in February 2008, and observations of several exoplanet systems have been ongoing since then in an effort to detect TTV signals in these systems. Sozzetti et al. (2009) presented eight transits of TrES-3, a G-type dwarf hosting a 1.9 MJ planet in a 1.3 day period (O'Donovan et al. 2007). They concluded that a linear period did not provide a particularly good fit to the central transit times, which indicates that either they underestimated the sytematics in their light curves, or that there is indeed a real TTV indicating a third body in the system. Here, we present a further nine RISE transit light curves of TrES-3, and re-analyse those from Sozzetti et al. (2009) using consistent techniques, in an effort to detect and understand any TTV signal. In §2 we describe the observations and data reduction, and in §3 describe how the light curves are modelled and in particular how the central transit times and uncer- tainties are found. Our results are presented in §4, and we use the transit timing residuals to place upper mass limits on a perturbing planet that could be present in the TrES-3 system without being detected from our observations. Finally, in §5 we summarise and discuss our results. -- 4 -- 2. Observations and data reduction 2.1. RISE photometry Seven full and two partial transits were observed using the LT and RISE from 2008 March 8 to 2008 August 4. The RISE instrument is described in detail in Steele et al. (2008) and Gibson et al. (2008). It consists of a frame transfer CCD which allows for continuous observation with effectively no dead time, a relatively large field of view (9.4 × 9.4 arcmin squared), and a single wide band filter (∼500-700 nm). For all observations, an exposure time of 8s was used with the instrument in 2 × 2 binning mode, giving a scale of 1.1 arcsec/pixel. For the full transits, 1 350 images were obtained resulting in 3 hours of continuous observations, allowing ∼50 mins of observations both before and after the transit event. The images have a typical FWHM of ∼2-4 pixels (∼2.2-4.4 arcsec). The nights were clear for the majority of the observations, except for part of the nights of 2008 July 5 and 2008 August 4, where large scatter due to thin clouds can be seen towards the end of the light curves. A summary of the observations is given in Table 1. Images were first debiased and flat fielded with combined twilight flats using standard IRAF2 routines. Aperture photometry was then performed on the target star and nearby companion stars using Pyraf3 and the DAOPHOT package. In each night different aperture sizes and numbers of comparison stars were used to minimise the out-of-transit RMS. These varied as the conditions and field orientation changed for each night of observations. The flux of TrES-3 was then divided by the sum of the flux from the companion stars (all checked to be non-variable) to obtain each lightcurve. Initial estimates of the photometric errors were calculated using the aperture electron flux, sky and read noise. The lightcurves were then normalised by dividing through with a linear function of time fitted to the out-of- transit data, setting the unocculted flux of TRES-3 equal to 1. The light curves, along with their best fit models and residuals (see §3.1), are shown in Figures 1 and 2. 2IRAF is distributed by the National Optical Astronomy Observatories, which are operated by the As- sociation of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. 3Pyraf is a product of the Space Telescope Science Institute, which is operated by AURA for NASA. -- 5 -- 3. Light curve modelling and analysis 3.1. Determination of system parameters In order to determine the system parameters from the transit light curves, a param- eterised model was constructed as in Gibson et al. (2008). This used Kepler's Laws and assumed a circular orbit to calculate the normalised separation (z) of the planet and star centres as a function of time from the stellar mass and radius (M⋆ and R⋆), the planetary mass and radius (Mp and Rp), the orbital period and inclination (P and i), and finally a central transit time for each lightcurve (T0,n). The analytic models of Mandel & Agol (2002) were then used to calculate the stellar flux occulted by the planet from the normalised sepa- ration and the planet/star radius ratio (ρ) assuming the quadratic limb darkening function Iµ I1 = 1 − a(1 − µ) − b(1 − µ)2, where I is the intensity, µ is the cosine of the angle between the line-of-sight and the normal to the stellar surface, and a and b are the linear and quadratic limb darkening coefficients, respectively. Limb darkening parameters were obtained from the models of Claret (2000). We linearly interpolated the ATLAS tables for Tef f = 5 650 K, log g = 4.4, [Fe/H] = -0.19 and vt = 2.0 kms−1 (from Sozzetti et al. 2009) to obtain limb darkening parameters in both the V and R bands. The average from the V and R bands was then adopted as our theoretical limb darkening parameters. Several tests were performed to examine the effects of the choice of limb darkening parameters on the results, which are described at the end of this section. A Markov-Chain Monte-Carlo (MCMC) algorithm was then used to obtain the best fit parameters and their uncertainties (see e.g., Tegmark et al. 2004; Holman et al. 2006; Winn et al. 2008; Collier Cameron et al. 2007). This consists of calculating the χ2 fitting statistic, χ2 = N Xj=1 (fj,obs − fj,calc)2 σ2 j + (M⋆ − M0)2 σM0 2 , where fj,obs is the flux observed at time j, σj is the corresponding uncertainty and fj,calc is the flux calculated from the model for time j and for the set of physical parameters described above. The second term represents a Gaussian prior placed on M⋆, where M0 and σM0 are the stellar mass and uncertainty as given in Sozzetti et al. (2009). This allows the stellar mass to vary within constraints for each model fit, so that errors in the stellar mass are taken into account when extracting errors from the MCMC distributions. The stellar radius was updated for each choice of M⋆ using the scaling relation R⋆ ∝ M 1/3 , whilst P and Mp ⋆ -- 6 -- were held fixed at their previously determined values, as their uncertainties do not have any significant effect on the output probability distributions. Subsequent parameter sets are then chosen by perturbing small amounts to the previously accepted parameter set and are then accepted with probability exp(−∆χ2/2) at each point in the chain, where ∆χ2 represents the difference in χ2 calculated for the old and new parameter sets. The procedure is the same as that used in Gibson et al. (2008), to which the reader is referred for details. To obtain reliable estimates of parameters and their uncertainties, it is important that the photometric errors are calculated accurately. The photometric errors σj are first rescaled so that the best fitting model for each lightcurve has a reduced χ2 of 1. It is also vital to account for any correlated ("red") noise in the data (see e.g., Pont et al. 2006; Gillon et al. 2006). The same procedure was used as in Gibson et al. (2008), where we evaluated the pres- ence of red noise in each light curve by calculating a factor β (≥ 1) according to Winn et al. (2008) and rescaled the photometric errors by this value. A value for β is determined by analysing the residuals from the best fit model of the lightcurves. Calculating the standard deviation of the residuals σ1, and the standard deviation after binning the residuals into M bins of N points σN , one would expect σN = σ1√Nr M M − 1 in the absence of red noise. This is usually larger by a factor β. However, the value deter- mined for β depends strongly on the choice of averaging time, i.e. M and N. Previously we have used an average of β values in the range 10 -- 35 mins (the approximate time-scale of ingress or egress) to rescale the photometic errors. However, for this analysis we decided to use the maximum value for β in this range in order to be as conservative as possible in determining our resultant errors. Normalisation plays an important role in determining parameters and errors from light curves, and to account for this a further 2 parameters were added to the model for each transit. These were the out-of-transit flux (foot,n) and a time gradient (tGrad,n), which are vital for TTV measurements as these affect the symmetry of the light curve and therefore the central transit times. An airmass correction was not used as previous studies have shown this produces similar results for full transits, but impedes chain convergence for partial transits (Gibson et al. 2008). An initial MCMC analysis was used to estimate the starting parameters and jump functions for ρ, i, T0,n, foot,n and tGrad,n. An MCMC run was then started for all nine light curves, fixing the central transit times to those determined in the initial run. Other parameters, such as the normalisation parameters, that are independent for each light curve were still allowed to vary. Five separate chains with 200 000 points were then computed -- 7 -- with the initial free parameters set by adding a 5σ gaussian random to their previously determined best fit values. The first 20% of each chain was eliminated to keep the initial conditions from influencing the results, and the remaining parts of the chains were merged to obtain the best fit values and uncertainties for each free parameter. The best fit value was set as the modal value of the probability distribution, and the 1σ limits to the values where the integrals of the distribution from the minimum and maximum values were equal to 0.159. To test that the chains had all converged to the same region of parameter space, the Gelman & Rubin statistic (Gelman & Rubin 1992) was then calculated for each of the free parameters, and was found to be less than 0.5% from unity for all parameters, a good sign of mixing and convergence. To check for any errors that may have resulted from a poor choice of limb darkening parameters, the above procedure was repeated, this time allowing the linear limb darkening parameter (a) to vary freely whilst holding the quadratic limb darkening parameter (b) fixed at the theoretical value, as in Southworth (2008). This, however, results in unphysical models of the limb darkening (negative values), as was found in Sozzetti et al. (2009), but not by Gibson et al. (2008) using the same technique for the WASP-3 system. This is probably due to the higher impact parameter of TRES-3, and therefore a higher sensitivity to the limb darkening parameters. As a compromise, the same a priori constraint was imposed on the linear limb darkening as used in Sozzetti et al. (2009), assuming that the limb darkening parameters do not drift from their theoretical values by more than 0.2 (Southworth 2008). This involved adding another term to the χ2 function as follows; χ2 = N Xj=1 (fj,obs − fj,calc)2 σ2 j + (M⋆ − M0)2 2 σM + ( a − a0 0.2 )2, where a0 represents the theoretical limb darkening coefficient. As this causes significant increases in the errors determined for i and ρ, these results were adopted as our final system parameters. A further two checks were performed to test the limb darkening parameters. The first involved repeating this process by replacing the quadratic limb darkening coefficient deter- mined for the combined V+R filter by that obtained for the individual V and R filters. This causes no significant changes to our results. The second check involved allowing each light curve to have its own independently varying (linear) limb darkening coefficient (within the prior constraints), rather than having one set of limb darkening coefficients to describe all the light curves. This again caused no significant changes to our results. It is therefore favourable to have the same set of limb darkening coefficients to describe all of the light curves, as forcing the same transit shape may reveal systematics through small differences in each light curve, that could otherwise be hidden through varying the limb darkening coefficients independently. -- 8 -- 3.2. Central Transit Times In order to calculate the central transit times for a TTV analysis we used the MCMC code, as described above, on each individual light curve, this time keeping the system pa- rameters ρ and i fixed at the best fit values determined in the previous section. Modeling the light curves individually has the advantage of needing much shorter chains, and does not result in underestimated uncertainties. This is because the central transit times are not very sensitive to the physical system parameters, but rather to those parameters that effect the symmetry of the light curves, in particular the normalisation function. The same analysis was done on the light cuves from Sozzetti et al. (2009), so that the central transit times and errors were found using consistent methods. For each light curve, five chains of length 50 000 were computed, and T0 and its uncer- tainty were extracted as before from the probability distribution after merging the chains (again discarding the first 20% of each). The linear limb darkening coefficient, stellar mass, and stellar radius were allowed to vary within the same constraints outlined before. Again, the Gelman & Rubin statistic was used to check for convergence. The systematics were accounted for using the same technique as described in the previous section, by re-scaling the errors of each light curve by a factor β. Values for β are given in Tables 3 and 4. A residual-permutation (RP) or "prayer bead" algorithm (see e.g. Southworth 2008; Gillon et al. 2009) was also used on each of the light curves to determine the errors in the transit times. This is another method commonly used to evaluate the effects of systematic noise on light curves, which often results in larger uncertainties than the MCMC method. The RP method consists of reconstructing the light curve by adding the residuals to the best fit model from the MCMC fit, each time shifting the residuals by a random amount, and performing a new fit on the light curve. 10 000 such fits were performed for each transit, with M⋆, R⋆, i and ρ selected from a Gaussian distribution at the start of each using the stellar parameters and uncertainties from Sozzetti et al. (2009), and the system parameters determined for the combined RISE light curves. The transit times, normalisation parameters, and linear limb darkening co-efficient were allowed to vary freely, using starting points determined randomly within 10σ from the best fit values. Errors in the central transit times were then estimated from the resulting distribution of fits. This method has the advantage that it preserves the actual correlated noise from the light curve, whereas the error re-scaling technique used alongside the MCMC fitting is sen- -- 9 -- sitive to the choice of averaging time. However, a comparison of the two methods showed, that in most cases, the errors from the MCMC fit were larger than those from RP. This is likely due to choosing the maximum value for β in the 10 -- 35 min range to re-scale the photometric errors prior to the MCMC runs, rather than using the average value. For each transit the "worst case" was assumed, i.e. we adopted the error from the RP method only when it produced a larger uncertainty than the MCMC code. Note we always used best fit values from the MCMC fit as the RP method already assumed these transit times when reconstructing each light curve. The methods used to determine each of the timing errors are given in Tables 3 and 4. 4. Results 4.1. System parameters The system parameters derived from the MCMC fits of the RISE transits are shown in Table 2. Sozzetti et al. (2009) undertook a thorough analysis of the stellar properties and radial velocities, and therefore we focus only on the planet parameters that are observable in the light curves, namely the inclination of the orbit and the ratio of the planet to stellar radius. We found i = 81.73+0.13 −0.0018. These are consistent with previously determined values, although with slightly smaller uncertainties, and therefore the planet radius and density (derived from the stellar radius and planetary mass from Sozzetti et al. 2009) are also consistent with previous studies. −0.04, and ρ = 0.1664+0.0011 4.2. Transit ephemeris The central transit times are shown in Table 3 for the RISE light curves, and in Ta- ble 4 for the light curves of Sozzetti et al. (2009), where the transit times were found to be consistent to within ∼ 0.3σ in all cases, and typically to less than 0.1σ. Due to the more rigourous approach used to account for red noise in our analysis, the error bars were found to be ∼ 10 − 40% larger. A new ephemeris was calculated by minimising χ2 through fitting a linear function of Epoch E and Period P to the transit times Tc(E) = Tc(0) + EP, where E = 0 was set to the transit from 2008 June 14 taken with RISE, as it has the smallest uncertainty. The results were Tc(0) = 2454632.62610 ± 0.00006 and P = 1.3061864 ± -- 10 -- 0.0000005. Figure 3 shows a plot of the timing residuals of the RISE and Sozzetti et al. (2009) transits using this updated ephemeris. For the RISE data a straight line fit yeilds χ2 = 13.49 for 7 degrees of freedom, and for the Sozzetti et al. (2009) data gives χ2 = 19.40 for 6 degrees of freedom, much lower than the value of 35.22 found from their analysis, simply because of the larger timing errors. For the combined data set, χ2 = 35.07 for 15 degrees of freedom, and therefore a reduced χ2 of 2.34. This all seems to support the conclusions from Sozzetti et al. (2009), that the uncertainties are underestimated, or that a linear period is not a good fit to the data. As we have been as sceptical as possible regarding the timing errors, this seems to suggest tentative evidence of a third body in the system perturbing the orbit of TrES-3b. However, after closer inspection of the light curves that contribute most to χ2, this conclusion is less convincing. Most of the large contributors to χ2 are partial transits or have very little out-of-transit data. If we remove all of the light curves with less than 20 mins of out-of-transit data either before ingress or after egress (transits E = -332, -319, -29, 23 and 32) this results in a χ2 of 13.53 for 10 degrees of freedom, or a reduced χ2 of 1.35. This is because these transits are much more difficult to normalize due to lack of out-of- transit data, and unseen systematics could certainly cause the normalization gradient to be skewed, therefore effecting the symmetry of the light curves and hence the central transit times. This seems to suggest that transits need at least ∼ 20− 30 mins of out-of-transit data either side of the transit to be useful for transit timing studies, unless a more robust method of normalizing light curves and accounting for the errors is found. The largest remaining contributor to χ2 is transit E = 29 which lies ∼ 2.4σ from the straight line fit. This transit not only has a high level of red-noise (β > 2), but a dip in the residuals from the best fit model is clearly seen around egress. The net effect of this on the model fit would be to "drag" the measurement of the central transit time later, as seen in the transit timing residuals. Removing this transit results in a reduced χ2 < 1, which suggests a constant period. Conclusions supporting a third body in this system would therefore rely on transits with little out-of-transit coverage and/or those with large visible systematics. 4.3. Limits on a second planet in the TrES-3 system. Despite not revealing a significant TTV signal, the data can still be used to place upper mass limits on the presence of a hypothetical second planet in the TrES-3 system. The shape and amplitude of transit timing residuals are dependent on a large number of parameters, such as mass, period, eccentricity and argument of periastron of the perturbing planet. -- 11 -- In order to compute model timing residuals, the equations of motion for a three body system were integrated using a 4th order Runge-Kutta method. The first two bodies were set to represent the star and planet of the TrES-3 system, which was assumed to have an initially circular orbit. Transit times were then extracted when the star and the transiting planet were aligned along the direction of observation, with the third body representing the perturbing planet. The transit times were then fit with a linear function of time and the timing residuals used for comparison with the data. The orbits of the planets and direction of observation were assumed to be coplanar. Ideally we would like to search the parameter space of the perturbing planet completely and set upper mass limits at each point. However, this is not possible given the large amount of computation required to produce a model of timing residuals at each point in such a large parameter space. Therefore, some simplifications and assumptions were made. Firstly, we as- sumed that the amplitude of the timing residuals for a given perturbing orbital configuration are proportional to the mass of the perturbing planet (Agol et al. 2005; Holman & Murray 2005), and verified this by constructing models with a range of perturbing planet masses. Secondly, we assumed that the perturbing planet had a starting eccentricity of 0, as an in- crease in eccentricity generally increases the amplitude of the timing residuals and therefore to set upper masses as a function of period we only need to investigate perturbing planets on circular orbits. This assumption is tested later in this section. Models were created for an Earth-massed perturbing planet with a period ratio dis- tributed from 0.2 to 5.0 (the regime in which relatively small masses may be detected), increasing the sampling around both the interior and exterior 2:1 resonances, where we ex- pect to probe for the smallest masses. For each model produced, the transit times were extracted for a range of observation directions. To calculate the maximum allowed mass for each model, χ2 was calculated by fitting the model residuals to the measured timing residuals from the light curves. The mass of the perturbing planet was increased (or decreased) by scaling the timing residuals until χ2 was increased by a value ∆χ2 = 9 (Steffen & Agol 2005; Agol & Steffen 2007) from that of a constant period (ie timing residuals = 0) which corresponds to a 3σ confidence limit. We then minimised χ2 along epoch only, and then let the mass of the perturber grow again until the maximum allowed mass for each model was determined. This procedure was repeated for the range of observation directions of each model, and the largest upper mass determined was assumed as our upper mass limit for each period ratio. To check that the starting mass of each model had no impact on the mass limits found for each period (i.e. test that residuals are indeed proportional to the perturbing mass), models were re-calculated with the mass of the perturbing planet set as the upper mass -- 12 -- limits found from the χ2 fits. Upper masses limits were then determined as before. This process was repeated twice and was found to make little difference to the final upper mass limits, therefore justifying our assumption. Figure 4 shows a plot of the resulting upper mass limits found as a function of the period ratio. The solid black line shows the upper mass limits found for the 3 body simulations, and the horizontal dashed line represents an Earth-mass planet. The results show that we have probed for masses as low as 0.97 M⊕ and 0.71 M⊕ in the interior and exterior 2:1 resonances, respectively. To test our assumption that perturbers on initially circular orbits will cause the small- est perturbations, and therefore can be used to set upper mass limits on a perturbing mass as a function of period ratio, we created models this time allowing the perturbing mass to have non-zero initial eccentricity. A set of models was created with a period range spanning the exterior 2:1 resonance and the perturbing bodies eccentricity ranging from 0 to 0.15. It was found that generally the amplitude of the signal drops and reaches a minimum be- tween e ∼0.01 and 0.12, before increasing again, and could drop by as much as an order of magnitude. A similar (but smaller) effect was found for the interior 2:1 resonance. This invalidates our assumption, and suggests that to set upper mass limits around resonance at least the period, eccentricity and argument of periastron of the perturbing planet needs to be explored in parameter space, which is beyond the scope of this paper. However, upper mass limits were estimated using this set of models and the same technique as before, and we found that more realistic upper mass limits are ∼ 3 − 4M⊕ and ∼ 10 − 15M⊕ in the interior and exterior 2:1 resonances, respectively. Out of resonance, it was found that the amplitude of TTV signals increases with eccentricity of the perturbing planet, and thus our assumption and upper mass limits are valid. A long term stability analysis was not performed for the 3-body systems. Bean (2009) found that for the CoRoT-1 system, only test particles with period ratios greater than ∼ 1.8 were stable for more than 106 orbits of the transiting planet. CoRoT-1 has a similar G-dwarf host star and a slightly longer period (∼ 1.5 days), which suggests that a similar analysis would prove useful here. Barnes & Greenberg (2006) explore the stability limits in exoplanet systems, and provide an inequality to test whether a system is Hill stable (equation 2). Using this inequality for the TrES-3 system (assuming an Earth-massed perturber), places lower and upper limits on the period ratio of 0.64 and 1.59, respectively. The resulting region not guaranteed to be Hill stable is marked on figure 4 by the grey shading, although stable configurations may still occur in this region. Trojan companions could also exist in stable orbits near the 1:1 resonance. Madhusudhan & Winn (2009) placed a 2σ upper mass limit of 81.3 Earth masses on a trojan in the TrES-3 system by combining transit observations -- 13 -- and radial velocity data. 5. Summary and discussion This paper presents the first transits taken using RISE specifically for a transit timing analysis, consisting of nine light curves of TrES-3. The transits were fit with an MCMC code and the derived system parameters found to be consistent with previous studies. Two different methods were used to determine the errors in the central transit times, trying to take into account the systematics in the light curves. These were scaling the errors bars by a constant prior to MCMC fitting after analysing the residuals, and using a residual permutation algorithm. The largest error found was used for each transit. We have shown that when systematics are kept at a minimum, it is possible to determine transit times to ∼10 seconds - the level of accuracy expected from RISE. Whilst the transit times appear to deviate significantly from a constant period when a χ2 analysis is performed, those that contribute most to the deviations tend to have very little out of transit data or obvious systematics. After removing transits with less than ∼20 mins of out-of-transit coverage either before ingress or after egress, the data are consistent with a constant period, and therefore no conclusive evidence was found for the presence of a second planet in the TrES-3 system. The transit times were then used to place upper mass limits on a perturbing planet as a function of period that could be present in the system yet not detected through our observations. This showed that our observations were sensitive to Earth-mass planets or smaller in the interior and exterior 2:1 resonances when we assume the perturbing mass is on an initially circular orbit. However, larger planets may exist in low eccentricity orbits around the 2:1 resonances, and exploring period, eccentricity and argument of periastron in parameter space is called for to set true upper mass limits, as well as a long term stability analysis. This study highlights the difficulties in attempting to detect a TTV signal, as we need to have complete confidence in the error bars calculated, which may require a more robust method to normalize the light curves and deal with red noise, especially if we are to trust transits with limited out-of-transit data. Confirming a true TTV signal may therefore re- quire some obvious structure in the observed residuals (which we would expect for resonant systems), rather than relying on a larger than expected "scatter" of points. RISE was designed and built with resources made available from Queens University Belfast, Liverpool John Moores University and the University of Manchester. The Liverpool Telescope is operated on the island of La Palma by Liverpool John Moores University in the -- 14 -- Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrofisica de Canarias with financial support from the UK Science and Technology Facilities Council. D.L.P. was supported by a Leverhulme Research Fellowship for the duration of this work. F.P.K. is grateful to AWE Aldermaston for the award of a William Penney Fellowship. We also thank A. Sozzetti, for making his data available for re-analysis, and the referee, D. Fabrycky, for comments which improved the content of this paper. Facilities: Liverpool:2m (RISE) Agol, E., Steffen, J., Sari, R., & Clarkson, W. 2005, MNRAS, 359, 567 REFERENCES Agol, E. & Steffen, J. H. 2007, MNRAS, 374, 941 Barnes, R. & Greenberg, R. 2006, ApJ, 647, L163 Bean, J. L. 2009, ArXiv e-prints Claret, A. 2000, A&A, 363, 1081 Collier Cameron, A., Wilson, D. M., West, R. G., et al. 2007, MNRAS, 380, 1230 Ford, E. B. & Gaudi, B. S. 2006, ApJ, 652, L137 Ford, E. B. & Holman, M. J. 2007, ApJ, 664, L51 Gelman, A. & Rubin, D. B. 1992, Stat. Sci., 7, 457 Gibson, N. P., Pollacco, D., Simpson, E. K., et al. 2008, A&A, 492, 603 Gillon, M., Pont, F., Moutou, C., et al. 2006, A&A, 459, 249 Gillon, M., Smalley, B., Hebb, L., et al. 2009, A&A, 496, 259 Heyl, J. S. & Gladman, B. J. 2007, MNRAS, 377, 1511 Holman, M. J. & Murray, N. W. 2005, Science, 307, 1288 Holman, M. J., Winn, J. N., Latham, D. W., et al. 2006, ApJ, 652, 1715 Kipping, D. M. 2009, MNRAS, 392, 181 Madhusudhan, N. & Winn, J. N. 2009, ApJ, 693, 784 -- 15 -- Mandel, K. & Agol, E. 2002, ApJ, 580, L171 Miralda-Escud´e, J. 2002, ApJ, 564, 1019 O'Donovan, F. T., Charbonneau, D., Bakos, G. ´A., et al. 2007, ApJ, 663, L37 Pont, F., Zucker, S., & Queloz, D. 2006, MNRAS, 373, 231 Southworth, J. 2008, MNRAS, 386, 1644 Sozzetti, A., Torres, G., Charbonneau, D., et al. 2009, ApJ, 691, 1145 Steele, I. A., Bates, S. D., Gibson, N., et al. 2008, Proc. SPIE, 7014 Steffen, J. H. & Agol, E. 2005, MNRAS, 364, L96 Tegmark, M., Strauss, M. A., Blanton, M. R., et al. 2004, Phys. Rev. D, 69, 103501 Winn, J. N., Holman, M. J., Torres, G., et al. 2008, ApJ, 683, 1076 This preprint was prepared with the AAS LATEX macros v5.2. l x u F e v i t l a e R l x u F e v i t l a e R 1.02 1.00 0.98 0.96 0.60 0.62 0.64 1.02 1.00 0.98 0.96 0.56 0.58 0.60 -- 16 -- 0.66 HJD - 2454534 0.68 0.70 0.62 HJD - 2454615 0.64 0.66 0.72 0.74 0.68 0.70 1.02 1.00 0.98 0.96 1.02 1.00 0.98 0.96 1.02 1.00 0.98 0.96 0.56 0.58 0.54 0.56 l x u F e v i t l a e R l x u F e v i t l a e R l x u F e v i t a e R l 0.60 0.62 0.64 HJD - 2454632 0.58 0.60 0.62 HJD - 2454649 0.66 0.68 0.70 0.64 0.66 0.68 0.44 0.46 0.48 0.50 HJD - 2454653 0.52 0.54 0.56 0.58 Fig. 1. -- RISE light curves of TrES-3 taken from 2008 March 8 to 2008 July 5 with their best fit models from the MCMC analysis over-plotted. Residuals from the best fit model are shown below each light curve offset. -- 17 -- 0.66 0.68 HJD - 2454662 0.70 0.72 0.48 0.50 0.52 HJD - 2454670 0.42 0.44 0.46 HJD - 2454674 0.54 0.56 0.58 HJD - 2454683 0.54 0.56 0.48 0.50 0.52 0.60 0.62 l x u F e v i t l a e R 1.02 1.00 0.98 0.96 0.60 0.62 0.64 l x u F e v i t l a e R l x u F e v i t l a e R l x u F e v i t a e R l 1.02 1.00 0.98 0.96 1.02 1.00 0.98 0.96 1.02 1.00 0.98 0.96 0.44 0.46 0.38 0.40 0.50 0.52 Fig. 2. -- Same as Figure 1, for light curves from 2008 July 14 to 2008 August 4. -- 18 -- 2 1 0 -1 -2 -350 -300 -250 -200 -150 -100 -50 0 50 Epoch s n m i / - C O Fig. 3. -- Timing residuals of the RISE transits (triangles) and those from Sozzetti et al. (2009, squares). -- 19 -- 104 103 102 101 100 h t r a E M / s s a M r e p p U 1.0 Period Ratio 5.0 J M / s s a M r e p p U 101 100 10-1 10-2 10-3 0.2 Fig. 4. -- Upper mass limits of a hypothetical 2nd planet in the TrES-3 system as a function of period ratio. The solid black line represents the upper mass found for the 3 body simulations, and the horizontal dashed line represents an Earth-mass planet. The region where an Earth- massed planet is not guaranteed to be Hill stable is marked by the grey shading. -- 20 -- Table 1: Summary of the RISE light curves of TrES-3. Night No. exposures No. comparison Aperture size RMS (residuals) stars (pixels) (mmag) 2008 Mar 8 2008 May 28 2008 Jun 14 2008 Jul 1 2008 Jul 5 2008 Jul 14 2008 Jul 22 2008 Jul 26 2008 Aug 4 1350 1350 1350 1350 1350 825 1350 1125 1350 4 4 7 6 7 6 2 8 2 8 7 5 4 6 7 4 4 4 1.31 1.42 1.00 1.10 3.31 1.81 1.57 1.18 2.60 Table 2: Parameters and 1σ uncertainties for TrES-3 as derived from MCMC fitting of RISE light curves and some further calculated parameters. Parameter Planet/Star radius ratio Orbital inclination Impact parameter Transit duration Transit epoch Period Planet radius Planet massa Planet density Planetary surface gravity Symbol ρ i b Td T0 P Rp Mp ρp log gp Value 0.1664+0.0011 −0.0018 81.73+0.13 −0.04 0.852+0.004 −0.013 1.332+0.024 −0.010 Units deg hours 2454632.62610 ± 0.00006 HJD days 1.3061864 ± 0.0000005 RJ MJ ρJ [cgs] 1.341+0.025 −0.035 1.910+0.075 −0.080 0.792+0.047 −0.042 3.421+0.023 −0.022 aFrom Sozzetti et al. (2009), displayed here for convenience. -- 21 -- Table 3: Central transit times and uncertainties for the RISE photometry including the error source. Epoch Central Transit Time Uncertaintly β b Error Source -75 -13 0 13 16 23 29 32 39 [HJD] 2454534.66243 2454615.64553 2454632.62613 2454649.60634 2454653.52504 2454662.66896 2454670.50630 2454674.42423 2454683.56734 (days) 0.00017 0.00017 0.00011 0.00013 0.00037 0.00034 0.00033 0.00052 0.00018 1.07 1.49 1.25 1.31 1.95 1.56 2.04 2.83 1.00 RP MCMC MCMC MCMC MCMC MCMC RP RP MCMC bRe-scale factor from red noise analysis (See §3.1). Table 4: Central transit times and uncertainties for the light curves of Sozzetti et al. (2009) including the error source. Epoch Central Transit Time Uncertaintly β c Error Source -342 -332 -320 -319 -74 -61 -48 -29 [HJD] 2454185.91040 2454198.97307 2454214.64631 2454215.95210 2454535.96825 2454552.94898 2454569.92910 2454594.74597 (days) 0.00028 0.00033 0.00036 0.00024 0.00023 0.00021 0.00021 0.00028 1.63 1.52 1.58 1.29 1.39 1.52 1.34 1.30 MCMC MCMC MCMC MCMC MCMC MCMC MCMC MCMC cSee Table 3.
1104.1460	1	1104	2011-04-07T23:28:12	On the dynamics and collisional growth of planetesimals in misaligned binary systems	[ "astro-ph.EP" ]	Context. Abridged. Many stars are members of binary systems. During early phases when the stars are surrounded by discs, the binary orbit and disc midplane may be mutually inclined. The discs around T Tauri stars will become mildly warped and undergo solid body precession around the angular momentum vector of the binary system. It is unclear how planetesimals in such a disc will evolve and affect planet formation. Aims. We investigate the dynamics of planetesimals embedded in discs that are perturbed by a binary companion on a circular, inclined orbit. We examine collisional velocities of the planetesimals to determine when they can grow through accretion. We vary the binary inclination, binary separation, D, disc mass, and planetesimal radius. Our standard model has D=60 AU, inclination=45 deg, and a disc mass equivalent to the MMSN. Methods. We use a 3D hydrodynamics code to model the disc. Planetesimals are test particles which experience gas drag, the gravitational force of the disc, the companion star gravity. Planetesimal orbit crossing events are detected and used to estimate collisional velocities. Results. For binary systems with modest inclination (25 deg), disc gravity prevents planetesimal orbits from undergoing strong differential nodal precession (which occurs in absence of the disc), and forces planetesimals to precess with the disc on average. For bodies of different size the orbit planes become modestly mutually inclined, leading to collisional velocities that inhibit growth. For larger inclinations (45 degrees), the Kozai effect operates, leading to destructively large relative velocities. Conclusions. Planet formation via planetesimal accretion is difficult in an inclined binary system with parameters similar to those considered in this paper. For systems in which the Kozai mechanism operates, the prospects for forming planets are very remote.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. ms15378 November 8, 2018 c(cid:13) ESO 2018 1 1 0 2 r p A 7 . ] P E h p - o r t s a [ 1 v 0 6 4 1 . 4 0 1 1 : v i X r a On the dynamics and collisional growth of planetesimals in misaligned binary systems M.M. Fragner1, R.P. Nelson1 & W. Kley2 1 Astronomy Unit, Queen Mary, University of London, Mile End Road, London E1 4NS, U.K. 2 Institut fur Astronomie & Astrophysik, Universitat Tubingen, Auf der Morgenstelle 10, 72076 Tubingen, Germany e-mail: [email protected], [email protected], [email protected] Received/Accepted ABSTRACT Context. Abridged. Many stars are members of binary systems. During early phases when the stars are surrounded by a discs, the binary orbit and disc midplane may be mutually inclined. The discs around T Tauri stars will become mildly warped and undergo solid body precession around the angular momentum vector of the binary system. It is unclear how planetesimals embedded in such a disc will evolve and affect planet formation. Aims. We investigate the dynamics of planetesimals embedded in gaseous protoplanetary discs that are perturbed by a binary companion on a circular, inclined orbit. We examine the collisional velocities of the planetesimals to determine when planetesimal growth through accretion can occur, instead of disruption. We vary the binary inclination, γF, binary separation, D, disc mass, Md, and planetesimal radius si. Our standard model has D = 60 AU, γF = 45◦, and a disc mass equivalent to that of the minimum mass solar nebula model. Methods. We use a 3D hydrodynamics code to model the disc. Planetesimals are test particles which experience gas drag, the gravitational force of the disc, the companion star gravity. Planetesimal orbital crossing events are detected and used to estimate collisional velocities. Results. For binary systems with modest inclination (γF = 25◦), disc gravity prevents planetesimal orbits from undergoing strong differential nodal precession (which they would do in the absence of the disc), and forces the planetesimals to precess with the disc on average. For planetesimals of different size the orbit planes become modestly mutually inclined, leading to collisional velocities that inhibit planetesimal growth. For larger inclinations (γF = 45◦), the Kozai effect operates, leading to destructively large relative velocities. Conclusions. We conclude that planet formation via planetesimal accretion is difficult in an inclined binary system with parameters similar to those considered in this paper. For highly inclined systems in which the Kozai effect switches on, the prospects for forming planets are very remote indeed. 1. Introduction Of the extrasolar planets detected so far, over 20 percent are found to orbit one component of a multiple/binary star system (Desidera & Barbieri 2007; Eggenberger et al. 2007). Planet formation in binary systems can represent a particular chal- lenge, as each stage of the formation process can be affected by the gravitational perturbation of the binary companion. One crucial stage that is particularly sensitive to the presence of the companion star is the accumulation stage of kilometre-sized planetesimals. The fundamental parameter that controls the efficiency of planetary growth by accretion of planetesimals is the rela- tive velocity between impacting objects. This must be lower than the escape velocity from the accreting objects in order for efficient runaway accretion to occur (Wetherill & Stewart 1989). If the external gravitational perturbation by the bi- nary companion excites relative velocities that exceed the es- cape velocity, runway accretion is prevented and growth re- mains slow. Furthermore, if the relative velocity is excited beyond the threshold velocity at which erosion dominates accretion, planetesimal growth potentially ceases altogether (Benz & Asphaug 1999). The possible effect of a stellar companion on the perturbed velocity distribution of planetesimals has been explored in sev- eral previous publications, where most have considered con- figurations in which the binary orbit is eccentric and coplanar with the disc midplane (Heppenheimer 1978; Marzari & Scholl 2000; Thebault, Marzari & Scholl 2006; Paardekooper et al. 2008). The main conclusion of these studies was that the cou- pled effects of secular perturbations of the binary companion, and friction due to gas drag arising from the protoplanetary gas disc, induce forced eccentricities and a size-dependent phasing of pericentres. This leads to relatively modest collision veloc- ities dominated by the keplerian shear for same-sized bodies, but high relative velocities for different sized planetesimals. As a consequence, binary systems with separations ∼ 50 AU may 2 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems have a strong inhibiting effect on accretion within a swarm of colliding km-size planetesimals. These studies neglected the effect of disc gravity acting on the planetesimals, which can affect the details of the results when the disc becomes eccentric through interaction with a bi- nary system on an eccentric orbit. But, the main conclusion that planetesimals of different sizes experience large collisional ve- locities due to differential phasing of their pericentres remains valid (Kley & Nelson 2008). These previous studies assumed that the orbital plane of the binary companion is coplanar with the disc midplane. According to Hale (1994) the assumption of coplanarity or modest inclination may be reasonable for binary separations below ∼ 40 AU, but systems with larger separations appear to have their orbital planes randomly distributed. The examination of planetesimal dynamics in non-coplanar configurations has received attention only recently (Marzari, Thebault & Scholl 2009; Xie & Zhou 2009). The former authors show that, due to the semi-major axis dependence of the nodal precession rate, the nodal lines of the planetesimals become progressively ran- domized. This may lead to the dispersion of the planetesimal disk that expands into a cloud of bodies surrounding the star. The latter authors considered the effect of gas drag in an sys- tem with a modestly inclined binary, and showed that in addi- tion to inducing size-dependent phasing of pericentres, gas drag also introduces phasing of nodal lines. This leads to a situation is which planetesimals of different size occupy different or- bital planes. Xie & Zhou (2009) suggest that this may provide a favourable channel for planetesimal growth as low velocity collisions between similar sized objects become more frequent than high velocity collisions between different sized objects. It should be noted, however, that these authors neglected the effect of disc gravity on the planetesimals. Although the inclu- sion of disc gravity modifies the dynamics of planetesimals in fully coplanar systems, the changes are not dramatic. In non- coplanar systems, however, where the disc and planetesimals may precess at different rates around the binary angular mo- mentum vector, the deviation of the planetesimals from the disc plane results in the disc gravity providing a strong restoring force that can modify the dynamics in an important qualitative sense. The recent simulations by Marzari, Thebault & Scholl (2009) and Xie & Zhou (2009) either ignored the gaseous pro- toplanetary disc, or treated it as a static object which did not evolve in the gravitational field of the inclined companion. It is known, however, that a gaseous disc orbiting under the in- fluence of a close binary companion on an inclined orbit will develop a mild warp and precess as a solid body around the or- bital angular momentum vector of the binary system if bending waves can propagate efficiently (Papaloizou & Pringle 1983; Papaloizou & Lin 1995; Ogilvie 2000; Papaloizou & Terquem 1995; Lubow& Ogilvie 2000). This process has been studied using numerical simulations by Larwood et al. (1996), who used Smooth Particle Hydrodynamic calculations, and more re- cently by Fragner & Nelson (2010), who performed 3D simu- lations using a grid-based hydrodynamics code . The condition for bending waves to propagate is that α < H/R, where α is the usual viscosity parameter (Shakura & Sunyaev 1973) and H/R is the disc aspect ratio. In protostellar discs, it is estimated H/R ≃ 0.05 and α ≃ 10−3 - 10−2, so we expect such discs to evolve similarly to the models presented in this paper. In addi- tion to generating a mild warp, and causing the disc to precess, the binary can also tidally truncate the disc at its outer edge, where the tidal truncation radius of the disc for a binary system of unit mass ratio is typically ∼ D/3, where D is the binary separation (Artymowicz & Lubow 1994; Larwood et al. 1996; Fragner & Nelson 2010). An observational example of a disc in a misaligned bi- nary system is HK Tau (Stapelfeldt et al. 1998), where the bi- nary system and disc have been imaged directly. More indirect evidence for precessing discs in close binary systems comes from observations of precessing jets in star forming regions (Terquem et al. 1999). In this paper we investigate the dynamics of planetesimals embedded in protoplanetary disc models in inclined binary sys- tems with circular orbits. Although most binaries are on eccen- tric orbits, we chose the case of a circular orbit to allow us to focus on the effects inclination. In contrast to previous work, we solve for the disc evolution in the gravitational field of the inclined binary companion, and include the effects of the disc gravity acting on the planetesimals. We consider two different values of the inclination between the binary orbit plane and disc midplane, γF = 25◦ and γF = 45◦, planetesimals of size 100 m, 1 km and 10 km, and binary separations between 60 and 120 AU. The disc outer radius is 18 AU, consistent with the tidal truncation radius for a binary separation D = 60 AU. Due to the complexity of our model, and the associated computational expense of running the simulations, we are un- able to consider large numbers of planetesimals, and so we are forced to use an approximate method for determining their collisional velocities based on examining the moments when the osculating orbits of the planetesimals intersect. Using this approximation, we find that in general the binary companion causes large planetesimal collision velocities to be generated, largely because the orbit planes of the planetesimals become mutually inclined. For the larger value of the inclination, we find that the Kozai mechanism can switch on, leading to the generation of large orbital eccentricities for the planetesimals, and therefore very large collision velocities. These results in- dicate that planet formation via the accumulation of planetesi- mals will be difficult in binary systems with parameters similar to those we consider in this paper. This paper is organised as follows. In Sect. 2 we present the basic equations and in Sect. 3 we discuss the numerical techniques. In Sect. 4 we investigate numerically the effect of disc gravity on the evolution of inclined planetesimal orbits in the absence of a binary companion. In Sect. 5 we examine the dynamics of planetesimals that are embedded in a disc which is inclined initially with respect to the binary plane, and ex- amine the effect of varying the inclination angle γF. In section Sect. 6 we present calculations for different disc masses and bi- nary separations and examine under which condition the Kozai effect can be suppressed. Finally, we discuss our results and draw conclusions in Sect. 7. M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 3 2. Basic equations The equations of hydrodynamics for a viscous disc that we solve in this paper are given in Fragner & Nelson (2010). The disc evolves under the effects of pressure, viscosity and the gravitational forces due to the binary companion and central star. We employ a large number of variables in this paper to describe the results, and we tabulate these in table 1 for ease of reference. We solve the equations in a frame that precesses around the angular momentum vector of the binary orbit, since it is expected that the disc models we consider in this paper will undergo uniform precession due to interaction with the binary companion. In this frame the disc midplane stays close to the equatorial plane of the spherical polar grid that we adopted in the simulations, provided that the precession frequency of the frame, ΩF, is chosen according to (Fragner & Nelson 2010): ΩF Ωd(R) 3 7 R3 D3 MB M⋆ = − cos (γF), (1) where Ωd(R) is the orbital angular velocity of the disc at its outer radius R, D is the binary separation, M⋆ and MB are the masses of the primary and secondary star, respectively, and γF is the relative inclination between the disc midplane and the binary orbital plane. The binary companion is held on a fixed circular orbit with separation D. Its position vector, D, in the precessing frame is given by (Fragner & Nelson 2010): arise because the coordinate system precesses around a vector ΩF, given by (Fragner & Nelson 2010): . (4) ΩF = ΩF 0 − sin(γF) cos (γF)  The last two terms are indirect terms that account for the accel- eration of the central star by the companion and disc, respec- tively. The gas drag force FD can be written (Weidenschilling 1977): 1 2vi − v(vi − v) FD = −CDπs2 i ρ Here si is the radius of planetesimal i, ρ and v are the gas disc density and velocity. The value of the drag coefficient CD de- pends on the Reynolds number (5) 2sivi − v νmol Re = (6) where the molecular viscosity, νmol = λMcs, λM is the mean free path of the gas molecules, and cs is the sound speed. CD takes values: 24. R−1 CD = 24. R−0.6 e e 0.44 Re < 1 1 < Re ≤ 800 Re > 800 (7) D = D cos ([ωB − ΩF]t) sin ([ωB − ΩF]t) cos(γF) sin ([ωB − ΩF]t)) sin (γF) where ωB is the binary angular frequency measured in the non- precessing binary frame. Thus an observer moving with the precessing frame sees an increased binary frequency ωB − ΩF due to the retrograde precession of the frame (i.e. ΩF is nega- tive). The planetesimals we model do not mutually interact. They feel the gravitational force of the primary star, secondary star and disc, the drag force due to the disc, Coriolis and centrifugal forces due to the precession of the frame and indirect forces that arise because we centre our coordinate system on the primary star. The equations describing the evolution of the planetesi- mals are therefore: ∂2ri ∂t2 dm(r) GM⋆ = − (ri − r) ri − r3 ri − GMB (ri − D) ri − D3 − GZdisc r3 i FD Mi − 2ΩF × vi − ΩF × (ΩF × ri) GMB D3 D − GZdisc dm(r) r r3 + − (3) where G is the gravitational constant, ri, vi and Mi are the po- sition and velocity vector and mass of planetesimal i. The first and second terms are the gravitational acceleration due to the central and companion star, respectively, and the third repre- sents the gravitational acceleration due to the disc. The fifth and sixth terms represent Coriolis and centrifugal forces that (2) 3. Numerical method The hydrodynamic disc equations are integrated using the grid-based hydrodynamics code NIRVANA (Ziegler & Yorke 1997), adapted to solve the equations in a precessing reference frame. This code uses operator-splitting, and the advection rou- tine uses a second-order accurate monotonic transport algo- rithm (van Leer 1977). The planetesimal orbits are integrated using the leap-frog integrator. 3.1. Units The equations are integrated in dimensionless form, where we choose our unit of length to be the radius of the disc inner edge. The unit of mass is that of the central star. The unit of time used in the code corresponds to ΩK(1)−1 (where ΩK(1) is the keple- rian angular frequency at radius R = 1), and the gravitational constant G = 1. When discussing simulation results we will refer to a time unit that corresponds to Pd = 2π/ΩK(10), which is approximately one orbit at the outer edge of the disc. In the sections of this paper which describe the results of simulations we refer to evolution times in units of "orbits", where an orbit should be understood as being a time interval equal to Pd. We scale our unit of length to 2 AU in physical units, and assume that the central star has one solar mass, so that one orbital pe- riod at the outer disc edge corresponds to a time of Pd = 89.44 years. Inclination and precession angles are displayed in units of radians. 4 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 3.2. Initialandboundaryconditions The disc model we use is model 1 described in Fragner & Nelson (2010), and interested readers are re- ferred to that paper for a general description of disc evolution in misaligned binary systems. The disc model extends from 1 − 9 in code units, corresponding to 2 − 18 AU in physical units, has a height to radius ratio h = 0.05, and a dimensionless viscosity parameter α = 0.025 (Shakura & Sunyaev 1973). During its evolution, the disc model develops a very modest warp (the variation in inclination across its radial extent is less than one degree), and it precesses as a rigid body around the angular momentum vector of the binary system with a frequency approximately equal to ΩF given by Eq. (1). The disc exhibits spiral density waves which are excited by the companion, and does not appear to become noticeably eccentric during its evolution -- unlike discs which are per- turbed by lower mass binary companions (Kley et al. 2008), or companions on eccentric orbits (Kley & Nelson 2008). ⊙ In our standard model we normalise the disc mass so that it would contain 0.015 M if extended out to a radius of 40 AU (this being very similar to the mass contained in the minimum mass solar nebula model (Hayashi 1981)), although the model we use nominally only goes out to ∼ 18 AU. The actual disc mass is Md = M = 4.52 × 10−3M⊙ , where M denotes our standard disc mass. In physical units the gas disc density is given by: ρ(r, θ) = 7.1 · 10−11(cid:18) r AU(cid:19)− 3 2 sin (θ)(cid:16) 1 h2 − 5 2(cid:17) g cm−3, where θ is the meridional angle measured relative to the nor- mal to the disc midplane. The binary companion is held on a circular inclined orbit with constant separation D = 30 (≡ 60 AU). We note that most binaries are on eccentric or- bits (Duquennoy & Mayor 1991), but we restrict ourselves to circular inclined orbits in this study so as to isolate effects re- lating to the inclination. Note that we neglect the disc gravity acting on the binary companion, so the binary orbit is fixed. At the beginning of each simulation, the companion mass is in- creased linearly until it reaches its final mass MB = 1M⋆ after 4 orbits. Its angular frequency ωB is increased accordingly to be consistent with a stationary orbit. Periodic boundary conditions were applied in the azimuthal direction. At all other boundaries standard stress-free, outflow conditions were employed. 3.3. Planetesimalset-up The mass of the planetesimals is given by Mi = 4 i ρs, and we choose a solid density of ρs = 2 g cm−3. They are initially embedded within the unperturbed disc model on circular, kep- lerian orbits which are coplanar with the disc midplane. As the disc does not become noticeably warped or eccentric during its evolution, it is not necessary to pre-evolve the disc so that it achieves a steady state structure prior to inserting the planetes- imals. 3 πs3 In the following discussion we will characterise the plan- etesimal evolution using their orbital elements in the fixed bi- nary frame, and these are calculated from the positions and ve- locities that the code outputs in the precessing frame. Hence, we first have to transform the velocities and positions from the precessing frame into the binary frame, in which the angular momentum vector of the binary JB is parallel to the unit vector e3. As we consider two reference frames in this paper, one which precesses with the disc, and the fixed frame based on the binary system, we denote vectors and coordinate values in the precessing frame using the hat-symbol (e.g. r, θ). Vectors and coordinate values in the fixed frame are denoted without the hat-symbol (e.g. r, θ). The transformation of the position and velocity vectors ri and vi defined in the precessing frame into vectors ri and vi defined in the non-precessing binary frame is given by (Fragner & Nelson 2010): ri = R−1 X (γF)R−1 ∂ ri = R−1 ∂t vi = Z (ΩFt)ri X (γF)R−1 Z (ΩFt)vi + ΩF × ri (8) where RZ and RX are rotation matrices around the z and x axes, respectively. The last term accounts for precessional velocities. Note that in a strict sense, we should replace ΩFt by R ΩFdt since the precession frequency is increased during the first 4 orbits of the simulations. For simplicity, however, we keep this notation and understand that this replacement should be made. The velocity and position vectors can be used to calculate or- bital elements in the binary frame. These are denoted ai, ei, Ωi, αi, ωi, fi for the semi-major axis, eccentricity, longitude of as- cending node, inclination with respect to the binary plane, lon- gitude of pericentre and true anomaly of particle i, respectively. Since the planetesimals are initialised to be coplanar with the disc midplane, and the equatorial plane of the spherical com- putation grid, it follows that αi = γF and Ωi = π at t = 0. Additionally, we are interested in the inclination of the planetesimals with respect to the local disc midplane, which we will denote by the symbol δ: cos (δi) = L(ri) . ji L(ri) ji = sin (αi) sin (αd) cos (Ωi − Ωd) + cos (αd) cos (αi) (9) where ji is the specific angular momentum vector of particle i, and L(ri) is the angular momentum vector of the disc an- nulus which has the same distance from the central star, ri, as planetesimal i at the current time. In this way, we measure the inclination of the planetesimal with respect to the local disc, which we will describe as the relative inclination from now on. The symbols αd and Ωd denote the inclination and nodal pre- cession angles, respectively, of the local gas disc annulus with respect to the binary orbital plane. Because the disc is rigidly precessing (and not strongly warped), and because of our ac- curate choice of ΩF, the disc midplane stays very close to the equatorial plane of the computational grid, and we have ap- proximately that αd = γF and Ωd = π + ΩFt during the sim- ulations. If planetesimals stay close to the disc midplane (such that Ωi − Ωd ≪ 1, αi − αd ≪ 1), Eq. (9) can be approximated by: i ≃ (Ωi − Ωd)2 sin2 (αi) + (αi − αd)2. δ2 (10) M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 5 Note that such an approximation is generally valid for two or- bits, whose orbital parameters (Ωi, αi) are very similar. It will be instructive to understand which contributions (gas drag, disc gravity, binary companion) are responsible for changing the orbital elements of the planetesimals. In order to measure this numerically, each force contribution is trans- formed into the binary frame, where we take the sum of the di- rect and indirect parts when considering the gravity of the disc or companion. We then calculate the radial FR, tangential FT and normal FN components with respect to the planetesimal orbit in the binary frame for each of the accelerations. These can be used to calculate the changes of osculating orbital ele- ments (Murray & Dermott 1999). Here we will only state some of them that will be used later for the discussion of results (not- ing that a dotted quantity denotes its time derivative): ri ji ri (11) (12) cos (ωi + fi) sin (ωi + fi). ji − (e3.ji)e3 αi = FN · Ωi = FN · For the change of relative inclination, differentiating Eq. (9) with respect to time gives: − sin (δi) · δi = cos (αi) sin (αd) [ αi cos (Ωi − Ωd) − αD] + sin (αi) cos (αd) [ αD cos (Ωi − Ωd) − αi] + sin (αi) sin (αd) sin (Ωi − Ωd)( Ωd − Ωi). (13) When calculating the change of relative inclination δi we note that changes in the disc inclination or nodal precession can only be caused by the binary companion. Hence when considering the change due to drag force or disc gravity we set αd = 0 and Ωd = 0 in Eq. (13). 3.4. Collisiondetection In order to obtain collisional velocity distributions that are sta- tistically relevant, it is necessary to accumulate data on a large number of direct collisions between planetesimals. Because we include the effect of disc gravity acting on the planetesimals, which incurs a large computational overhead, we are only able to integrate 50 planetesimals for each size that we consider. This means that we need to use an approximate method for es- timating the collision velocities between planetesimals, instead of detecting direct collisions between them. The approximation that we adopt in this work is to treat the planetesimals not as individual particles, but rather as repre- sentatives of their orbits, which we conceive of as being highly populated. These orbits are assumed to have a circular cross section with a radius ∆r = 2 · 10−4 AU in physical units (note that this value has been used for the inflated radii of planetes- imals in Xie & Zhou (2009), where a direct collision detec- tion method was used). In order to estimate impact velocities for colliding planetesimals, we detect the moments when two orbits defined by their osculating elements intersect, and es- timate the velocity of impact that planetesimals at the point of intersection would have. When running simulations which consider the general dynamics of planetesimals in misaligned binary systems, we distribute the planetesimals with a large range of semi-major axes within the protoplanetary disc ini- tially. When running simulations which are designed to cal- culate the collisional velocities, however, we perform separate runs in which the initial planetesimal orbits are distributed with a narrow range of semi-major axes (∆a = 10−3 AU). This is to maximise the number of orbit crossings, and hence improve our collision statistics. The condition for two orbits represented by particles i and j to cross is given by the vector equation: ri(φi, Ωi, αi, ai, ei, ωi) = r j(φ j, Ω j, α j, a j, e j, ω j) (14) where ri is given by: ri = ri(φi, ai, ei, ωi) with cos (Ωi) cos (φi) − sin (Ωi) sin (φi) cos (αi) sin (Ωi) cos (φi) + cos (Ωi) sin (φi) cos (αi) sin (φi) sin (αi) (15) ri(φi, ai, ei, ωi) = ai(1 − e2 i ) 1 + ei cos (φi − ωi) = ai(1 − e2 i ) 1 + ei cos ( fi) , (16) and similarly for the orbit represented by particle j. These vec- tor equations involve orbital elements which are defined in the fixed binary frame. The angles φi and φ j in Eq.(14) are the an- gular distances of the orbit crossing point to the nodal lines of orbits i and j, respectively, where the nodal lines are located in the binary orbit plane. For general crossing orbits, where the semi-major axes and eccentricities of the two orbits differ, it is not possible to solve Eq. (14) directly. For circular orbits with the same semi-major axis(cid:16)ri = r j(cid:17), however, we can solve Eq. (14) for φi and φ j, giving the values of these angles at the two points of intersection for orbits i and j: (φ j1, φi1) and (φ j2, φi2). These angle are defined by the expressions tan (φ j1) = sin (Ωi − Ω j) cos (α j) cos (Ωi − Ω j) − sin (α j)/ tan (αi) φ j2 = φ j1 + π cos (φi1) = cos (φ j1) cos (Ωi − Ω j) + sin (φ j1) cos (α j) sin (Ωi − Ω j) sin (φ j1) sin (α j) sin (φi1) = sin (αi) (17) The solution for φi2 is obtained by using φ j2 instead of φ j1 in the last two equations. The above solutions have been derived under the assump- tion that the orbits are circular with the same semi-major axis. If they have a finite eccentricity and different semi-major axes, we still assume that the point of closest approach is at these longitudes1 (φi1, φ j1) and (φi2, φ j2). Hence we define orbital crossing or collision, if the following condition is fulfilled: (cid:12)(cid:12)(cid:12) ri(φi1, ai, ei, ωi) − r j(φ j1, a j, e j, ω j)(cid:12)(cid:12)(cid:12) ≤ ∆r where ri(φi1, ai, ei, ωi) is given by Eq. (16), and this condition is also checked for the other potential crossing point (φi2, φ j2). (18) 1 Note that this simplification was used to speed up the collision test in the simulations. Full numerical evaluation of the closest approach point of the two orbits gave good agreement with this assumption. 6 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems Variable Variable description h = H/R Disc aspect ratio γF D ωB ΩF Pd Md M α ρ si X X ai ei fi αi ωi Ωi δi αd Ωd αi ) Inclination between disc midplane and binary orbit plane Binary separation Binary orbit frequency Precession frequency of the precessing frame and disc Orbital period at 20 AU The disc mass used in a simulation Our nominal disc mass ( M = 4.52 × 10−3 M ⊙ Viscosity parameter Gas density Planetesimal physical radius Vectors and coordinate quantities in the precessing frame denoted with the 'hat' symbol Vectors and coordinate quantities in the fixed binary frame denoted without the 'hat' symbol Semi-major axis of planetesimal i Eccentricity of planetesimal i True anomaly of planetesimal i Inclination angle between binary orbit plane and orbit plane of planetesimal i Longitude of pericentre of planetesimal i (also referred to as the apsidal precession angle) Longitude of ascending node of planetesimal i (also referred to as the nodal precession angle) Inclination of planetesimal i relative to the local disc Inclination of local disc annulus relative to the binary orbit plane Precession angle of local disc Inclination angle between binary orbit plane and orbit plane of planetesimal i Table 1. Table of variable names. After every one hundred time steps the condition (18) is checked for the 11175 particle pairs that arise from the 50 par- ticles integrated for each size. Since this gives a total on the or- der of 107 pairs for the simulated time intervals we considered, we obtain quite a large number of pairs for which condition (18) is fulfilled. Hence we are able to obtain good statistics on collisional velocities despite the relatively low number of plan- etesimals modelled. For orbit i the velocity at the first orbit crossing location is: vi1 = d dt ri(φi1, Ωi, αi, ai, ei, ωi) = vriri + vφi φi = s GM⋆ ai(1 − e2 i )hei sin (φi1 − ωi)ri + (1 + ei cos (φi1 − ωi)) φii (19) and likewise for orbit j. Here ri is the particle unit vector calcu- lated from Eq. (15) and φi is the unit vector in the φ direction. Likewise vri and vφi are the velocity components in the r and φ directions. Note that this expression does not account for pre- cessional velocities. However, since they do not contribute to the relative velocities at orbital crossing points (where ri = r j) they can be safely omitted. The relative velocity at the first crossing location is then simply ∆v = vi1−v j1, and likewise for the second location (φi2, φ j2). Later we will present averages of collisional velocities calculated in this way. 3.4.1. Validity of the orbit crossing technique An important question to address is under what conditions does the orbit crossing technique that we have described above agree with the direct detection of collisions when calculating average collision velocities within a planetesimal swarm. The first point to note is that the method we use to calcu- late the point of closest approach between two orbits assumes that this occurs along the line of intersection between the two orbit planes. For orbits which are mutually inclined with re- spect to one another, this assumption is valid. But it obviously breaks down for coplanar orbit planes, and our method reports the wrong points of intersection in this case. So our method of detecting orbit crossing is only strictly valid for mutually in- clined orbits. In order to address the more general question of whether the orbit crossing method gives collisional velocities which are similar to the direct collision detection method, we have run a number of pure N-body simulations. The first test we ran used a narrow ring of particles centred around 10 AU with static orbital elements (no binary companion was included) that were very similar to those reported in Fig. 7 of this paper at 60 orbits. The direct detection simulation used 1000 particles and was run for approximately 2000 local orbits. The orbit crossing simula- tion run used 50 particles and was also run for approximately 2000 local orbits. We found that the mean collision velocities reported by each method agreed to within approximately 30 %. The second test was similar, but included a binary compan- ion inclined by 25◦ to the ring of planetesimals, which were initially placed on circular orbits. This was run for 2000 local orbits and again good agreement was found between the two collision detection methods, with the mean collision velocity reported by the orbit crossing method being approximately 40 % larger than that reported by the direct detection method. We conclude that the orbit crossing method gives fairly reliable re- M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 7 sults for the mean collision velocity in a planetesimal swarm provided that the dominant contribution to the relative veloci- ties arises because of differential nodal precession. 3.4.2. Analytical estimate of collisional velocities To gain a better understanding of the contributions that control the collisional velocities, we now derive an analytical estimate. Consider two particle orbits A and B, which have very similar orbital elements (i.e. aA = aB, eB = eA + ∆e, ΩB = ΩA + ∆Ω, αB = αA + ∆α, ωB = ωA + ∆ω), where the ∆-quantities are assumed to be very small. The above orbital elements are mea- sured with respect to the binary orbit plane. Consider now a coordinate system that is coplanar with the orbit of particle A. The velocity of particle A is then given by Eq. (19) -- i.e. vA = vrArA+vφA φA. With respect to the x−y plane of this coordi- nate system, the orbit of particle B will be inclined by an angle δAB, because its plane is slightly different due to the quantities ∆α and ∆Ω. But since these differences are very small by as- sumption, the relative inclination δAB between the two particle orbits is small. Note also that in such a coordinate system we can always choose ΩA such that ΩB = 0. Hence from Eq. (15) it follows that: In the limit e ≪ 1 the relative velocity therefore becomes: ∆v2 = v2 (26) K(cid:16)∆e2 + e2∆ω2 + δ2 AB(cid:17) . The quantity δAB is defined within the coordinate system that is coplanar with orbit A. Since ∆Ω ≪ 1 and ∆α ≪ 1 we can apply the approximation introduced in Eq. (10) to express δAB as: δ2 AB = ∆α2 + sin2 (α)∆Ω2 and the relative velocity becomes: ∆v2 = v2 K(cid:16)∆e2 + e2∆ω2 + ∆α2 + sin2 (α)∆Ω2(cid:17) . This result shows that relative velocities are generated for a variety of reasons. As in the coplanar case, relative velocities can be generated due to different eccentricities ∆e, or phasing of pericentres ∆ω. In three dimensions, relative velocities will also be caused by different inclinations ∆α or phasing of nodal lines ∆Ω. If ∆e = ∆α = 0, ∆ω = ∆Ω ∼ 1 and sin (α) ∼ α we recover the result from Lissauer & Stewart (1993) for ran- domised orbits: ∆v = vK √e2 + α2 (29) (27) (28) rB = rA + dr with , (20) 4. Effect of disc gravity on planetesimal dynamics: single star case dr = 0 0 sin (φB)δAB  where Eq. (20) has been derived using the assumption that the mutual inclination δAB ≪ 1. Hence the orbit crossing point (rA = rB) corresponds to φB = 0, from which it follows that: dr = 0 and . (21) dφ = 0 0 δAB  The velocity vector of orbit B at the orbit crossing point is thus: vB = (vrA + dvr)rA + (vφA + dvφ)( φA + dφ) (22) To first order we neglect the term dvφ dφ, and the relative ve- locity between the two orbits at orbital crossing is: ∆v = vB − vA = dvr rA + dvφ φA + vφA dφ. (23) Because rA. φA = rA. dφ = φA. dφ = 0, the relative velocity becomes: ∆v2 = dv2 r + dv2 φ + v2 φAδ2 AB = dv2 r + dv2 φ + v2 Kδ2 AB. (24) In the second equality we can replace vφA by the keplerian ve- locity vK to this order. From Eq. (19) the radial and azimuthal velocity differences are to first order: dvr = vr(e + ∆e, ω + ∆ω) − vr(e, ω) (∆e sin (φ − ω) − e∆ω cos (φ − ω)) dvφ = vφ(e + ∆e, ω + ∆ω) − vφ(e, ω) ≃ s GM⋆ a(1 − e2) ≃ s GM⋆ a(1 − e2) (∆e cos (φ − ω) + e∆ω sin (φ − ω)) . (25) In this section we present the results of simulations of plan- etesimals that interact gravitationally with the disc and central star only. The planetesimals are on orbits that are inclined with respect to the disc midplane, and we switch off the drag force. The knowledge gained in this section will help to understand the results in later sections when the binary companion and the gas drag force are included in the model, and allow us to iso- late the effect that the disc gravity alone has on the dynamics of planetesimals on inclined orbits. Throughout this paper we use the figure convention that the top left panel is referred to as panel 1, with the remaining panels being labelled as 2, 3, 4... when moving from left to right and from top to bottom. A simulation result for a reference particle is presented in Fig.1 (solid lines in panels 1-3 ). This planetes- imal has an initial semi-major axis of 10 AU, eccentricity of ei = 0.1 and relative inclination (inclination relative to the disc midplane) of δi = 0.1. The quantities displayed are the nodal precession angle Ωi (panel 1), inclination αi (panel 2) and apsi- dal precession angle ωi (panel 3). The quantities are calculated with respect to a reference plane that is coplanar with the disc midplane (αi = δi). We observe that the disc gravity causes retrograde nodal precession ( Ωi < 0) about the disc angular momentum vector, and prograde apsidal precession ( ωi > 0), while the inclination αi remains unaffected. We also studied planetesimals with different initial semi- major axes, eccentricities and relative inclinations. The results are summarised in panels 4-6 of Fig.1, which show the apsidal (solid) and nodal (dashed) precession rates ( ωi and Ωi) as a function of: semi-major axis ai (panel 4); eccentricity ei (panel 5); relative inclination δi (panel 6). The black lines show results for a disc model with the reference disc mass, and the red lines 8 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems are for a model with twice the reference mass. As expected, the precession frequencies (nodal and apsidal) scale roughly with the mass of the disc. Furthermore, the precession rates are larger in magnitude for particles that have smaller semi- major axes, as can be seen in panel 4. Panel 5 shows that the precession rates (nodal and apsidal) are weakly dependent on the eccentricity for the range in ei shown. The dependence on relative inclination, however, is strong (see panel 6). So far we have discussed the evolution of orbital elements measured with respect to a reference plane that is coplanar with the disc midplane. Later on, however, we will consider plan- etesimal orbital elements measured with respect to the fixed binary plane, which will be inclined with respect to the disc midplane. If the inclination of the disc with respect to the bi- nary plane is smaller than the relative inclination of the parti- cles (αd ≤ δi), the nodal and apsidal precession rates caused by the disc remain unchanged, and panels 4-6 of Fig. 1 apply. If the inclination of the disc with respect to the binary plane becomes larger than the relative inclination of the particles (αd ≥ δi), however, the nodal and apsidal precession rates will be differ- ent. In order to understand the effect of the gravity of a disc that is highly inclined with respect to our reference plane, we transform the planetesimal position and velocity vectors via the transformation given by Eq. (8), using a representative inclina- tion of the disc of αd = γF = 0.78 (45◦) > δi = 0.1 (5.7◦) and assume that the disc is non-precessing with Ωd = ΩFt = 0 (inclusion of a small precession frequency ΩF , 0 does not change the result.) The outcome of this transformation is shown by the dashed line in Fig.1 (panels 1-3) for the same reference particle as described earlier. We observe that the nodal preces- sion Ωi angle (panel 1) and inclination angle αi (panel 2) are oscillating around a fixed value. This can be understood as fol- lows. The angular momentum vector of the particle still pre- cesses around the angular momentum vector of the disc as be- fore. Unlike before, however, the disc has an inclination with respect to the reference plane of αd = 0.78 (45◦) > δi. Hence the precession of the particle angular momentum vector causes the measured inclination and nodal precession angles to oscil- late around those of the disc, with an amplitude given by the particle's relative inclination δi. We verify that for the refer- ence particle the nodal precession rate measured with respect to the disc midplane is Ω ∼ −0.23 (−13.2◦) per orbit, as seen from panel 4 (black dashed line). This corresponds to a preces- sion period of 27 orbits, coinciding with the observed period of oscillation of the inclination and nodal precession angles measured with respect to the reference plane (panels 1 and 2, dashed line). Additionally, the particle appears to precess apsi- dally in a retrograde sense (panel 3, dashed line). The precession rates (nodal and apsidal) are displayed as functions of semi-major axis, eccentricity and relative incli- nation in panels 7-9, respectively, for the highly inclined disc case, with the same line style and colour convention as used be- fore. As can be seen from these panels, the net nodal precession rate is zero, since the disc causes oscillation but no net change of the nodal precession angle. From panels 8-9 we observe that the apsidal precession rate is roughly independent of eccentricity ei and relative inclina- tion δi. However it depends on the semi-major axis, ai, and also scales roughly with the disc mass, as shown in panel 7. For fu- ture purposes we fit this apsidal frequency by a second order polynomial: ω = (30) Md M "C0 + C1(cid:18) a AU(cid:19) + C2(cid:18) a AU(cid:19)2# 1 Pd with C0 = −4.187 · 10−2, C1 = 5.118 · 10−3 and C2 = −4.242 · 10−4. Pd is the orbital period at the disc outer edge (which is nominally located at r = 10 ≡ 20 AU), Md is the disc mass and M is the nominal disc mass for the reference model introduced in Sect. 3.3. The fit is displayed in panel 7 (dotted lines) and matches the numerical data (solid line) well. To summarise, the measured effect of the gravity of an in- clined disc on the planetesimal orbit depends on the ratio of the inclination of the disc with respect to the binary plane, αd, and the inclination of the particle with respect to the disc, δi. If δi ≥ αd the resulting nodal and apsidal precession rates caused by the disc are displayed in Fig.1 (panels 4-6). If δi ≤ αd the precession around the disc angular momentum vector causes oscillations in the nodal precession and inclination angles with an amplitude that is given by δi. The apsidal precession in this case is displayed in Fig.1 (panels 7-9) and can be approximated by Eq. (30). 5. Planetesimal dynamics in inclined binary systems Model label γF [degrees] 1 2 3 4 5 6 7 8 9 0 25 45 45 45 45 45 45 45 25 25 25 45 45 45 Md [ M] 1 1 1 1 1 3 6 1 1 1 1 1 1 1 1 D [AU] 60 60 60 60 60 60 60 90 120 60 60 60 60 60 60 si [m] 10000 10000 10000 1000 100 10000 10000 10000 10000 100 1000 10000 100 1000 10000 Table 2. Table of runs: the first column gives the run label; the second column gives the inclination of the binary companion to the gas-plus-planetesimal disc, γF; the third column gives the disc mass in units of the reference mass; the fourth column gives the binary separation, D; the fifth column lists the plan- etesimal radii, si, included in the models. We will now describe the dynamics of planetesimals that are orbiting in the full binary plus disc system. The model pa- rameters are summarised in table 2. The planetesimals are ini- tially set-up on circular, keplerian orbits that are coplanar with M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 9 Fig. 1. Upper panels: nodal (panel 1) and apsidal precession (panel 3) angles as well as inclination angles (panel 2) of a test particle experiencing the gravitational field of a disc and central star only. The solid lines show the results when viewed in a frame in which the disc is assumed to be coplanar with the reference plane. The dashed lines show the same quantities for the same model, but where the disc is now treated as if it was inclined by γF = 0.78 (45◦) to the reference plane, as it will be when the binary companion is included. Middle panels: nodal (dashed lines) and apsidal (solid lines) precession rates when the disc is highly inclined as a function of semi-major axis ai (panel 4), eccentricity ei (panel 5) and relative inclination δi (panel 6) for a standard disc mass (black) and a disc with the double mass (red). Bottom panels: nodal (dashed) and apsidal (solid) precession rates caused by the highly inclined disc as a function of semi-major axis ai (panel 7), eccentricity ei (panel 8) and relative inclination δi (panel 9) for a standard disc mass (black) and a disc with double that mass (red). In panel 7 a polynomial fit to the apsidal precession rates is shown (dotted lines). respect to the disc midplane (i.e. δi = 0). They are distributed radially in the interval ai ∈ [6, 16] AU for models 1-7 (where the disc truncation radius is ≃ 20 AU). In an additional series of runs we confine the planetesimals to a narrow annulus with ∆a = 10−3 AU centred around a = 10 AU (models 8 and 9) to maximize the number of collisions and study collisional veloc- ities in more detail. We treat model 3 as our reference case. The inclination of the binary companion to the disc is initially set to γF = 0.78 (45◦) and its separation from the central object is set to D = 60 AU. The disc mass used in the reference model is Md = 1 M and corresponds to a scaled minimum mass solar nebula as explained in Sect. 3.3. In the other models we varied the initial inclination γF of the binary companion to the gas-planetesimal disc (models 1-3), the disc mass Md (models 4 and 5) and the binary separation D (models 6 and 7). 5.1. Zeroinclinationcase(model1) In this model the binary orbital plane is coplanar with the planetesimal-plus-disc midplane. From time t = 0 the plan- etesimals experience the gravity of the disc and gas drag forces (the planetesimals have size si = 10 km), and the mass of the binary companion is increased to its final value over a time of four orbits. The evolution of the semi-major axes, ai, and eccentricities, ei, are shown in Fig.2. The colours correspond to different initial semi-major axes, with darker blue colours representing the inner planetesimals and the green-red colours representing planetesimals further out in the disc. We will also adopt this colour convention when discussing simulation re- sults later in the paper. As can be seen from Fig.2 (panel 1), the binary companion does not change the semi-major axes of the bodies, as expected from secular theory. Eccentricities are generated, however, that are of the order of (ei ∼ 10−2), as can 10 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems Fig. 2. Orbital elements for the coplanar case. Panel 1 shows the semi-major axes of different planetesimals, where the colour convention introduced here is used in future plots. The eccen- tricity of the various planetesimals is shown in panel 2. The binary separaion D = 60 AU. be seen from Fig.2 (panel 2). Very modest eccentricities are generated because the planetesimals are orbiting in a slightly non-keplerian potential due to the disc gravity. But we also see that the eccentricities grow on a time scale of ∼ 5 orbits due in- teraction with the binary whose mass grows over this time. The values of eccentricity obtained are consistent with those pre- sented by Ciecielag et al. (2007), who considered the evolution of planetesimals in a gas disc with a circular binary companion, but neglected the effects of disc gravity. Because of the closer proximity to the binary companion, the outer planetesimals are more strongly affected by these perturbations and their eccen- tricities are raised to larger values. Note that the data in this and the following figures has been smoothed over a temporal window of 1.8 orbits, but it is clear that after an initial rise, the eccentricities remain essentially constant for 80 orbits (7155 years). Secular perturbations from an eccentric, coplanar binary companion lead to the generation of well-defined forced ec- centricities, and can also lead to strong alignment of perias- tra for same-sized planetesimals in the presence of gas drag (Marzari & Scholl 2000; Thebault, Marzari & Scholl 2006). This causes a dramatic reduction in the impact velocities for these planetesimals. But this effect is absent for a circular com- panion as considered here, and eccentricities are largely gen- erated by high frequency terms in the disturbing function. We find that the periastra are not well aligned near the beginning of the simulation, since a companion on a circular orbit cannot de- fine a direction of preferred alignment (this effect may be seen in Fig. 7 later in the paper where we plot results for a simula- tion with a low inclination (25◦) binary companion). This basic point is in agreement with results presented by Ciecielag et al. (2007), who also considered a circular binary. As such a binary companion on a circular orbit appears to be singular in its effect on the orbits of planetesimals embedded in a gas disc. This has potentially important consequences for the collisions between planetesimals reported later in this paper. Fig. 3. Orbital elements for the low inclination case γF = 25◦. The colours represent planetesimals at different semi-major axes with the convention adopted from Fig.2. The eccentricity is depicted in panel 1. In panels 2 and 3 the nodal precession and inclination angles are shown, where the short dashed line represents the inner and outer edge of the disc. In panel 4 the relative inclination with respect to the disc is shown, where the short dashed line represents one pressure scale height and the dashed-dotted line indicates where δi = αd. The binary separa- tion D = 60 AU. 5.2. Lowinclinationcase(model2) In this section we describe the simulation results for model 2, for which only planetesimals of size 10 km were considered and the binary inclination γF = 25◦. Fig.3 shows the evolution of eccentricities ei (panel 1), nodal precession angles Ωi (panel 2), inclinations αi (panel 3) and relative inclinations δi (panel 4) for the different planetesimals using the same convention of colours as described in Sect. 5.1. As in the coplanar case, eccentricities are raised by interactions with the binary com- panion, but because the binary orbit is circular the forced ec- centricity predicted by secular theory is negligible, and the ec- centricities are generated by high frequency perturbations. We can observe that these are somewhat lower than in the coplanar case due to a reduced coplanar component of the companion's gravity. In panel 2 of Fig.3 we see that most of the planetesimals precess nodally at a joint rate with the disc, except for the out- ermost particle (red line), which will be discussed later in this section. This joint precession can be explained by the presence of the disc, with its gravitational field playing the major role. To illustrate this we also performed simulations of planetesimals that interact with the binary system only. The corresponding nodal precession and inclination angles of the planetesimals for such a simulation are shown in Fig.4. Due to the secular per- turbation of the secondary star alone, the inclinations are ex- M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 11 pected to stay constant while the orbital planes of the planetes- imals precess at their free particle rate (Papaloizou & Terquem 1995): Ω f ree i 3 8ΩK GMB D3 (3 cos2 αi − 1). = − (31) which is a strong function of semi-major axis, as ΩK ∼ r− 3 2 is the keplerian angular velocity. The left panel of Fig.4 shows that the nodal lines of the planetesimals become progres- sively randomised, leading to the dispersion of the planetesimal disk into a cloud of bodies surrounding the star, as found by Marzari, Thebault & Scholl (2009). Comparing this behaviour Fig. 4. Simulation results of planetesimals interacting with the binary system only (no gas disc). The left panel shows the nodal precession, and the right panel shows the inclination. with the results displayed in Fig.3 (panels 2 and 3) we can see that the presence of the disc and its gravitational field causes fundamentally different behaviour of the planetesimals. The planetesimals remain coupled to the disc, as their nodal pre- cession angles stay close to the disc precession angle (dashed line in panel 2 of Fig. 3). The role of gas drag in determining the relative inclination between the disc midplane and the plan- etesimals is negligible for this run for which the planetesimal size is 10 km. But for smaller planetesimals gas drag does play a role, as we will describe later in the paper. In Fig.5 we show the time evolution of the nodal preces- sion rates for the particle at r = 13.3 AU (panel 1) and r = 6.2 AU (panel 2). The various curves represent contributions due to gas drag (blue line), disc gravity (red line), binary compan- ion (green line) and the total rate (black line) as calculated in Sect. 3.4. Because of its different radial location, the outer par- ticle orbit precesses faster in the retrograde sense than the inner particle if the disc is absent. Confirming this we can observe in Fig.5 (panels 1 and 2) that Ωi due to the companion (green line) is about Ωi = −0.09 (−5.15◦) per orbit for the outer body, and Ωi = −0.03 (−1.7◦) per orbit for the inner body. It is the disc gravity, however, that compensates for this effect and causes the planetesimals to precess at a common net rate. Since the disc precesses rigidly at a rate ΩD ≃ ΩF = −0.06 (−3.4◦) per orbit, which corresponds to the free particle rate at about ai = 11.5 AU, disc gravity (red line) gives a positive contribu- tion to the precession rate for the outer particle ( Ωi > 0) and a net negative contribution for the inner particle ( Ωi < 0). Since the particle-plus-disc system precesses in the retrograde sense the outer particle precession is slowed down, while the inner particle precession is speeded up by the disc, such that both bodies precess together with the disc on average at the rate ΩF. This result illustrates the fundamental importance of including the effects of disc gravity when calculating the dynamics of planetsimals in misaligned binary systems. The nodal precession and inclination angles of the planetes- imals show oscillations, as seen in Fig.3 (panels 2 and 3). These are caused by the disc gravity, as explained in Sect. 4 and illus- trated by the red line in panel 5 of Fig. 5. The planetesimals effectively precess around the disc angular momentum vector at the same time as the disc-plus-planetesimal system precesses around the binary angular momentum vector. Initially the plan- etesimals are coplanar with the disc midplane (δi = 0), but as the system evolves during early times, differential nodal preces- sion between particle i and the disc occurs, leading to a build up of relative inclination δi, since δ2 i ∝ (Ωi − Ωd)2 as shown by Eq. (10). The differential precession is greatest for planetes- imals whose semi-major axes deviate most from a = 11.5 AU (the radius at which planetesimals naturally want to precess at the same rate as the disc), so the relative inclination is largest for particles that are furthest inside (black line) or outside (yel- low line) this value, as demonstrated in panel 4 of Fig. 3. The disc model has an aspect rato H/R = 0.05, so that planetsimals which develop relative inclinations δi > 0.05 will spend large fractions of their orbits away from the disc mid- plane. Planetesimals interior to 8 AU and exterior to 13 AU have relative inclinations δi ≃ 0.1, and so spend large fractions of their orbits essentially outside of the disc. We also observe that the oscillation periods of the inclination, αi, and nodal pre- cession angles, Ωi, varies among the planetesimals. This is ex- pected due to their different semi-major axes and relative incli- nations, and the observed periods are all consistent with Fig.1 (dashed lines in panels 4-6), which we recall shows how the disc alone acts on the planetesimals. We note that the inclination oscillations, seen in Fig.3 (panel 3), begin in opposite senses for planetesimals whose semi-major axes are above or below a = 11.5 AU (i.e. plan- etesimals below a = 11.5 AU are perturbed onto higher incli- nation orbits, while planetesimals exterior to a = 11.5 AU are perturbed onto lower inclination orbits). Panel 3 of Fig.5 plots the inclination angle difference αi − αd versus the precession angle difference Ωi − Ωd for a planetesimal located at 13.3 AU. A similar plot for a planetesimal at 6.2 AU is shown in panel 4. Each plot shows the trajectory of the tip of the planetesimal an- gular momentum vector relative to the disc angular momentum vector (which is located at the origin). For the inner particle the companion induces a differential precession Ωi − Ωd > 0 and the anti-clockwise precession around the disc angular mo- mentum vector causes the planetesimal to approach a higher inclination orbit αi − αd > 0 during the first half of this preces- sion cycle. For the outer planetesimal the induced differential precession is Ωi − Ωd < 0, and the anti-clockwise precession causes perturbation onto a lower inclination orbit αi − αd < 0. We now return to the planetesimal that is orbiting at a = 15 AU. This body shows quite distinct behaviour from all the other bodies, as shown in Fig.3 (red lines). This particle is dominated by the gravity of the companion, and it nodally precesses fast 12 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems sion period. Ignoring the possible growth of planetesimals into planets, or their collisional destruction, those bodies which re- main almost coplanar with the disc should continue to do so over long time scales until the disc mass decreases due to vis- cous evolution and/or photoevaporation. Significant reduction of the disc mass would eventually allow the planetesimal dy- namics to become dominated by the companion star, and they would decouple from the disc. But, we also note that that the disc and binary orbit plane will evolve toward alignment on the viscous evolution time at the outer edge of the of the disc (Terquem et al. 1999; Larwood et al. 1996; Fragner & Nelson 2010). So it is possible that an initially misaligned protoplane- tary disc may align significantly over its lifetime, bringing with it any planetary system which has formed within it. The final degree of misalignment for a planetary system formed in an in- clined protoplanetary disc may therefore be substantially less than the initial misalignment of the protostellar disc. 5.2.1. Collisional velocities in low inclination case (model 7) In this section we present the results of a simulation which ex- amines collisions between planetesimals in a system where the binary companion has an inclination of γF = 25◦ relative to the disc midplane. To increase the collision rate to statistically meaningful values which can be measured in a simulation, we initialise the planetesimals with a narrow range of semi-major axes (∆a = 10−3 AU) centred around a = 10 AU. We consider three different planetesimal sizes (100m, 1km and 10km), and for each size we include 50 particles. We check the condition for orbital crossing given by Eq. (18) every 100 time steps for each of the 11175 particles pairs (100 time steps corresponds to 0.011 orbital periods at 10 AU, the orbital radius of the plan- etesimals). If the orbit crossing condition is satisfied, then we calculate the velocity at the crossing point for both planetesi- mal orbits according to Eq. (19). Before discussing the results of model 7, it is worth recap- ping what we might expect based on previous work in which the gravity of the disc was neglected. Same-sized planetesi- mals being perturbed by an eccentric, coplanar binary com- panion will experience a growth in their forced eccentricity, but gas drag damping will cause strong orbital phasing dramat- ically reducing collisional velocities (Marzari & Scholl 2000; Thebault, Marzari & Scholl 2006). The different phasing of pericentres for planetesimals of different size, however, leads to large collision velocities which are likely to be disruptive. The inclusion of a small inclination (αi ≤ 5◦) causes different sized planetesimals to orbit in different planes, such that collisions between similar sized bodies are more frequent than between different sizes. The fact that the pericentre phasing is main- tained in this scenario means that planetesimal growth may be more likely to occur in inclined systems (Xie & Zhou 2009). The collision velocities we obtain are shown in Fig.6 (solid lines) for collisions between equally sized (panel 1) and differ- ently sized planetesimals (panel 2). As mentioned in Sect. 5.1, an important difference between our set-up and previous work is that we consider a circular, inclined binary companion, Fig. 5. Panels 1 and 2 show the nodal precession rates of a plan- etesimal with semi-major axis at ai = 13.3 AU (panel 1) and ai = 6.2 AU (panel 2). Shown are the contributions of the disc gravity (red line), binary companion (green line), gas drag (blue line) and total rate (black line). In panels 5 and 6 we also show the rates of inclination change with respect to the binary and disc midplane respectively for the inner particle. In panels 3 and 4 the trajectory of the tip of the particle's angular momen- tum vector is shown projected onto the (Ωi − Ωd, αi − αd) plane for the outer (panel 3) and inner particle (panel 4). enough to decouple from the disc. As it precesses away from the disc, the relative inclination δi grows (panel 4 of Fig.3). The precession rate around the disc angular momentum vector is a decreasing function of relative inclination, so the precession period becomes long for this particle. Because Ωi − Ωd < 0, we observe the quasi-monotonic decrease of inclination in panel 3 of Fig.3 until the reversal point at t = 47 orbits, when the differential nodal precession is Ωi − Ωd = π. The planetesi- mal orbital plane then approaches the disc midplane from the other side (Ωi − Ωd > 0), and we observe the quasi-monotonic increase of inclination after t = 47 orbits. Over longer time scales than we have been able to consider because of the computational expense of running the simula- tions, we expect the outermost planetesimals which decouple from the disc to show continued oscillations in their inclina- tions, with an oscillation period equal to the synodic preces- M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 13 Fig. 6. Average collisional velocities in the low inclination γF = 25◦ case between equally sized (left panel) and differently sized planetesimals (right panel) for planetesimals centred around 10 AU from the central star. Left panel: collisions between 10km- 10km bodies (green-solid); 1km-1km (blue-solid); 100m-100m (red-solid). Right panel: collisions between 10km-1km bodies (green-solid); 10km-100m (blue-solid); 1km-100m (red-solid). Threshold velocities for catastrophic disruption corresponding to the different size-combinations for the weak (dash-dotted line) and strong aggregates (dashed line) are also shown. resulting in a broad distribution of planetesimal longitudes of pericentre, ωi, even for planetesimals of the same size. Consequently, we see that the collision velocities for equal sized bodies in panel 1 are quite large, being between 50 - 70 ms−1 at the end of the simulation. Although it appears that the circular binary is largely responsible for the growth of ec- centricity and the misaligned periastra of orbits for same sized planetesimals, it is possible that gravitational perturbations as- sociated with spiral waves in the disc also contribute. The col- lision velocities for differently sized bodies, however, are even larger than for same-sized bodies, and exceed 400 ms−1 after 80 orbits (see panel 2). Fig.7 shows averages of all the quantities that determine the relative velocities according to the analytical estimate given by Eq. (28), and we can use this data to determine the main contributors to the collision velocities shown in Fig.6. Panel 1 shows the difference of longitude of pericentre ∆ωi, panel 2 shows the average eccentricity e, panel 3 shows the difference in eccentricity ∆ei, panel 4 shows the difference in inclination ∆αi, panel 5 shows the angle between the nodal lines ∆Ωi and panel 6 shows the average inclination αi. Note that these aver- ages were obtained only for orbits which mutually cross, and do not represent the distributions for the whole ensemble of particles. Using the numbers extracted from these figures in Eq. (28) we can reproduce the relative velocities shown Fig.6 with good accuracy, indicating that Eq. (28) is a valid approxi- mation and that our collision detection technique is generating collision velocities which are consistent with expectations. For collisions between planetesimals of the same size the dominating contribution to the square of the relative velocity comes from the term which is proportional to e2 i , while all i ∆ω2 other terms are at least an order of magnitude smaller. This im- plies that planetesimals with the same size all orbit more or less in the same plane (as shown by panels 4 and 6), and rela- tive velocities are generated due to misalignment of their orbit pericentres, ∆ωi, a result which is broadly consistent with those obtained by Xie & Zhou (2009). We note that our collision de- tection method is not strictly valid in this regime, because it predicts the locations of the orbit crossing points incorrectly for coplanar orbits. But, because the average pericentre mis- alignment is on the order of unity (measured in radians), the average collision velocities reported by the algorithm are con- sistent with the analytical prediction vcoll ≃ e ∆ω to within a factor of order unity. Under these conditions, incorrect detec- tion of the orbit crossing locations still leads to estimates of the average collision velocities being approximately correct. Although panel 1 of Fig.7 shows that the misalignment of pericentres increases slightly over time when considering only those particles whose orbits intersect, we find that the average pericentre misalignment calculated over all same-sized particle pairs actually decreases during the simulation. The reason for this is not clear, but it is interesting to note that even when the binary orbit is circular, the long term evolution appears to be toward one where the pericentres tend toward alignment, and this effect is most marked for the smallest planetesimals suggesting that it is an effect due to gas drag. As seen in panel 2 of Fig.6, planetesimals of different sizes tend to have much larger collisional velocities. One reason is the increased misalignment of pericentres, as can be ob- served in Fig.7 (panel 1, dashed lines). This may be a di- rect consequence of gas drag, as discussed in previous work (Marzari & Scholl 2000; Thebault, Marzari & Scholl 2006), 14 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems bital planes ∆Ωi as seen in panel 5 of Fig.7 (dashed lines) in addition to the misalignment of their pericentres. The frequency of collision between bodies is an impor- tant factor during the growth of planetesimals, and Xie & Zhou (2009) have suggested that the differential phasing of nodal lines for planetesimals of different sizes may increase the rel- ative importance of collisions between similar sized objects rather than different sized bodies. We have examined the colli- sion frequency reported by our collision detection technique by simply counting the number of orbit crossing events re- ported during the simulation. In basic qualitative agreement with Xie & Zhou (2009), we find that the collision frequency during the simulation between same-sized objects is a factor of 3 - 6 times larger than between different sized objects. Our re- sults thus support the idea that planetesimal growth in inclined binaries is likely to proceed via accretion of similar sized bod- ies. We conclude that relative velocities between differently sized planetesimals tends to be large in binary systems whose orbital plane is misaligned with the planetesimal-plus-disc plane, while relative velocities between same sized planetesi- mals are largely unaffected by this non-coplanarity. But the cir- cular orbit of the binary companion prevents strong pericentre alignment for similar sized bodies, leading to significant col- lisional velocities in this case too, in contrast to the situation observed when the companion is on a coplanar eccentric orbit. The question of what happens in the case of an eccentric, in- clined companion unfortunately goes beyond the scope of this paper but should obviously be the focus of future work. Fig. 7. Average orbital parameters of the colliding planetes- imal pairs in the low inclination γF = 25◦ case. The line colours and styles are as follows: green-solid (10km-10km); blue-solid (1km-1km); red-solid (100m-100m); green-dashed (10km-1km); blue-dashed (10km-100m); red-dashed (1km- 100m). but it may also be due to the fact that particles which expe- rience stronger gas drag orbit closer to the disc midplane than larger planetesimals do. This then affects the gravitational per- turbations experienced by the particles as they travel through the disc, which may increase the pericentre misalignment. But a more important contribution to the square of the relative ve- locities comes from the term ∝ sin2 (αi)∆Ω2 i . We find that due to the different gas drag strengths, planetesimals of different sizes tend to precess nodally at different rates for some initial time until disc gravity causes them to precess together with the disc. While smaller planetesimals tend to precess together with the disc immediately and consequently their relative inclina- tions stay low, larger planetesimals tend to precess at their own free particle rates for a longer time, causing a larger build up of relative inclination. In other words, although the planetesi- mal swarm as a whole is forced on average to precess with the disc by the disc gravity, smaller planetesimals oscillate about the midplane of the disc with a smaller amplitude than larger planetesimals. This induces a large misalignment of their or- We now want to determine whether the collisional veloc- ities seen in Fig.6 will lead to growth or erosion of the plan- etesimals. In principle there are three possible outcomes: accre- tion, in which the largest body involved in the collision contains more mass after the collision; catastrophic disruption, in which more than half of the total mass of the system is dispersed af- ter the collision; erosion (or cratering), where the largest body loses a small amount of mass during the collision. Only accre- tion leads to the growth of a planetesimal. For simplicity in interpreting our results, we adopt from remnant mass the universal Stewart & Leinhardt (2009): law for the largest Mlr Mtot = 1 − 1 2 QR QD , (32) where Mlr is the mass of the largest post-collision remnant, and Mtot = M1 + M2 is the total mass of the two colliding bodies M1 and M2. The quantity QR = 0.5 M1M2 ∆v2 is the reduced mass M2 tot kinetic energy scaled by the total mass of the colliding system. For equally sized bodies accretional collisions require Mlr > Mtot 1 2 , from which it follows that QR < QD. Hence QD is called the catastrophic disruption limit of the collisions (the energy required to disperse half the total mass). If QR > QD collisions between equally sized bodies lead to catastrophic disruption. When considering collisions between differently sized bodies the condition has to be modified. Let M1 ≫ M2 such that M2 = µM1 with µ ≪ 1. The condition for accretion is now Mlr > M1, from which it follows that QR < (cid:16) 2µ 1+µ(cid:17) QD, which implies for M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 15 Benz & Asphaug (1999), and find qS = 7 · 104, qG = 10−4, µM = 0.5 and φM = 8. In the limit of very large bodies the disruption curve can be best fitted by their results of colliding rubble piles. In this regime Stewart & Leinhardt (2009) find 12 (35) QRP = 1.7 · 10−6R2 for equal mass bodies and a factor of about three larger than this if the impactor size is much smaller than the target size. For il- lustrative purposes we present the relative velocity threshold δvD obtained by combining Eq. (33) and Eq. (34) in the case of identical colliding bodies (µ = 1) in Fig.8. Note that in the limit of large bodies at low impact velocities, we set QD = QRP, because the rubble-pile solution for equal mass bodies defines a lower limit for disruption in this regime. The figure shows the limiting velocity below which collisions will lead to net accretion as a function of the planetesimal radius R1 for the weak (dashed-dotted line) and strong aggregates (dashed line) at two different impact velocities VI = 2km/s (blue line) and VI = 50m/s (green line). It is apparent that the threshold ve- locity is increased for larger impact velocities as more energy is partitioned into shock deformation. In contrast, at low im- pact velocities the momentum coupling between the colliding bodies is more efficient and less energy is needed to cause ero- sion of planetesimals. As gravity becomes more important than strength for very large bodies the curves join onto the disrup- tion curve for colliding rubble piles (black solid line). In this regime the threshold velocity is about a factor of 3-5 larger than the escape velocity from the bodies surface (dotted line). To compare the relative velocities obtained from our numerical simulations with the disruption threshold velocity, we evaluate Eq. (33) for equally (µ = 1) and differently sized µ = 10−3 planetesimals using VI = ∆v in Eq. (34) and choosing R1 as the bigger of the two planetesimals involved in the collision. If the rubble-pile disruption limit gives a higher estimate for a given size combination we use this limit instead. We note that Stewart & Leinhardt (2009) considered target- projectile mass ratios only down to 0.03, instead of the val- ues 103 and 106 that apply to collisions of 1km or 100m sized bodies with a 10km sized target. As such we are apply- ing the Stewart & Leinhardt (2009) results in a regime where catastrophic disruption is unlikely to occur, and therefore out- side of the regime of validity of their study, strictly speaking. Nonetheless, it reasonable to assume that even if catastrophic disruption does not occur during high velocity impacts, accre- tional growth is also unlikely due to erosion. The results are plotted in Fig.6 for the strong (dashed line) and weak aggregates (dashed-dotted line). Except for colli- sions between equally sized 10km bodies (green), the colli- sional velocities are always substantially larger than the ero- sion/disruption threshold. Hence we conclude that growth of planetesimals probably can not occur for planetesimal sizes < 10km. For the equally (10km) sized bodies on the other hand the situation is not as clear. Although the collisional ve- locities lie below the strong aggregate threshold (green dashed line) they are also slightly above the weak aggregate thresh- old (dashed-dotted line). This suggests that collisions might also lead to erosion in this size regime. But it should be noted (33) (34) Fig. 8. Threshold velocities for collisions between equally sized planetesimals as a function of their spherical radius R1. The disruption curves for the weak (dashed-dotted line) and strong rock (dashed line) are shown for two different impact velocities VI = 50m/s (green) and VI = 2km/s (blue) along with the solution for rubble piles (black-solid line). The bodies escape velocity (black-dotted line) is also shown. the relative velocity threshold: δvD = 2p1 + µpQD The disruption limit QD is sensitive to factors that influence the energy and momentum coupling between the colliding bodies (i.e. impact velocity, mass ratio and material properties such as strength and porosity). Stewart & Leinhardt (2009) show that the catastrophic disruption curve QD can be fit to an analytical formula (Houssen & Holsapple 1990,1999): QD =hqS R9µM /(3−2φM) 12 + qGR3µM 12 i V(2−3µM) I 3 (1 + µ) 1 where µM and φM are material properties. In this expression R12 is the spherical radius of the combined mass Mtot assuming a density of ρs = 1 g cm−3. Since we use a density of ρs = 2 g cm−3 this radius is given by R12 = 2 1 3 R1, where R1 is the spherical radius of the larger mass M1 involved in the collision. The first term on the right hand side of Eq. (34) rep- resents the strength regime, while the second term represents the gravity regime. Very small bodies are supported by mate- rial strength (first term), which decreases as the planetesimal size increases. As gravity becomes more important for larger bodies the second term becomes dominant and increases the disruption limit again. Stewart & Leinhardt (2009) derive the constants qS , qG, µM and φM for weak aggregates, such as weak rock and porous glass, by fitting the disruption curve to their numerical results. For weak aggregates they find (in cgs units) qS = 500, qG = 10−4, µM = 0.4 and φM = 7. For strong rocks they use the basalt laboratory data and modelling results from 16 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems that we have not included a realistic planetesimal size distri- bution in these simulations, and it is possible that collisions between planetesimals clustered in size around 10km, whose orbit planes are very modestly inclined with respect to one another, may allow accretion to occur for strong aggregates. Examination of this possibility will require future simulations that adopt a more realistic and continuous size distribution. If collisions between 10km sized planetesimals do lead to growth then this will most definitely not occur in the standard runaway regime, because the escape velocity is ∼ 10m/s. It is possible, however, that type II runaway growth may occur (Kortenkamp et al. 2001) in which the large velocity dispersion causes growth to be orderly while the bodies remain small, but enters a runaway phase when large bodies form whose escape velocities are larger that the velocity dispersion. We conclude that in an inclined binary system on a circular orbit, relative velocities are excited that will lead to erosion of planetesimals for sizes ≤ 10km which we have considered here, and growth is uncertain for 10km sized bodies. 5.3. HighinclinationcasesandtheKozaieffect For large inclinations between the binary and planetesimal or- bital planes, it is possible that the Kozai effect will switch on, causing cyclic variations of planetesimal eccentricities and in- clinations (Kozai 1962). In this section we describe this effect in more detail in order to simplify the understanding of results which are described later in this paper. The equations describ- ing the secular evolution of the inclination αi, eccentricity, ei, and longitude of pericentre, ωi, can be written as (Innanen et al. 1997): (36) (37) ∂αi ∂τ 15 8 = − e2 i sin (2ωi) sin (αi) cos (αi) i q1 − e2 i sin (2ωi) sin2 (αi) = = 15 8 3 i n2(1 − e2 i ) + 5 sin2 (ωi)he2 eiq1 − e2 4q1 − e2 ∂ei ∂τ ∂ωi ∂τ where for simplicity of notation we have introduced the time variable: τ = [D3ΩKi/GMB]−1t. Hence the time scale on which any of the given quantities change is ∝ D3. It can be seen from Eq. (38) that there is a value of ωi for which ωi = 0. To lowest order in ei this occurs when i − sin2 (αi)io (38) (39) sin2 (ωi) = 2 5 sin2 (αi0) (40) where αi0 is the initial inclination of planetesimal i. Having ob- tained this value, the apsidal precession is halted, as ωi = 0. This leads to the exponential growth of eccentricity. To illus- trate this we can substitute Eq. (40) into Eq. (37) to find: ∂ei ∂τ sin2 (αi0) sin (ωi) cos (ωi)ei 15 4 = = 15 4 eis 2 5"sin2 (αi0) − 2 5#. (41) sin (αi0) ≥ q 2 Hence the Kozai effect is only active for initial inclinations 5 or written differently αi0 ≥ 0.68 radians. The critical angle for which the Kozai effect switches on is thus αK = 39.2◦. Additionally it can be shown that the Delaunay variable, Di, (which is equivalent to the component of the plan- etesimal angular momentum which is parallel to the binary an- gular momentum vector) is a constant of the motion, where Di is defined by (42) Di = qai(1 − e2 i ) cos (αi). As secular variations do not change the semi-major axes, this implies that an increase in eccentricity is coupled to a decrease in inclination, and vice versa. To study the Kozai effect in the absence of the disc, we have performed N-body simulations of planetesimals inter- acting with the binary system only. The results are depicted in Fig.9 for planetesimals with various initial inclinations. It can be seen that the eccentricities and inclinations undergo cyclic variations for planetesimals whose initial inclination are αi ≥ αK (yellow, green and light blue lines in Fig.9). This can be explained as follows. As the eccentricity grows and the in- clination decreases (panels 1 and 3), the latter eventually hits the Kozai threshold, αK, at which point there is no value for ωi for which expression (40) holds. Consequently ωi , 0 and ωi evolves to values for which ei ≤ 0 and αi ≥ 0. Hence, even- tually ωi = 0 can be obtained again, but now for a value of ωi which is causing the opposite evolution of eccentricities and in- clinations until the original configuration is recovered and the Kozai-cycle is completed. We can clearly see in Fig.9 how the Fig. 9. Orbital elements of planetesimals with different initial inclinations interacting with the binary system only. Panel 1 shows the eccentricity and panel 2 shows the nodal precession angle. The inclination with respect to the binary plane and the longitude of pericentre (apsidal precession) are depicted in pan- els 3 and 4, respectively. M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 17 increase or decrease of eccentricities and inclinations is con- nected to the halting of apsidal precession during each half- cycle of the Kozai mechanism (panel 4). Additionally, we ob- serve that the time scale on which the eccentricities and incli- nations vary are increased as the initial inclination αi0 is de- creased. Once the initial inclination is αi0 < αK (Fig.9, blue and black lines) the condition ωi = 0 can never be obtained. Consequently there are no net changes in eccentricities and in- clinations and the Kozai effect is switched off. 5.4. Highinclinationcase(model3) In model 3 we increased the inclination between the disc- planetesimal system and the binary plane to γF = 0.78 (45◦). As described in the previous section, once the inclination ex- ceeds a critical value αK = 0.68 (39.2◦), the Kozai effect may lead to large eccentricity and inclination variations of the plan- etesimals, provided the disc mass is not too large (see later for a discussion about the conditions under which the Kozai ef- fect operates). The presence of our scaled minimum mass solar nebula disc does not prevent the onset of the Kozai effect for planetesimals with semi-major axes a < 12 AU, as seen from Fig.10 (panels 1 and 3), where we use the same colour con- vention for depicting the planetesimals as a function of initial semi-major axis as in models 1 and 2. Fig. 10. Orbital elements for the highly inclined case γF = 0.78 (45◦) (model 3). Panel 1 shows the eccentricity grows due to the Kozai effect for all but the outermost planetesimals . The nodal precession and inclination angles are depicted in panel 2 and 3, respectively, where the short dashed lines represent the inner and outer disc edge. The long dashed line in panel 3 rep- resents the threshold inclination of αK = 39.2◦ above which the Kozai mechanism is expected to operate. The apsidal pre- cession angles are depicted in panel 4. The eccentricity grows earliest for planetesimals with larger semi-major axes (panel 1, green lines). When the Kozai mechanism starts operating, the inclination starts to decrease. In the absence of the disc, this decrease would continue until the threshold of αK = 0.68 (39.2◦) is approached, after which the inclination would increase and the eccentricity would de- crease again. In the presence of the disc, however, the situation is different. As the Kozai effect reduces the inclination, αi the nodal precession induced by the binary accelerates according to Eq. (31). This causes the planetesimal orbits to become sig- nificantly inclined relative to the disc midplane (see panel 3 of Fig.10). Concurrent precession about the disc angular momen- tum vector due to the disc gravity causes a quasi-monotonic decrease of inclination relative to the binary. The planetesi- mals are thus perturbed by the disc onto orbits with inclination αi < αK, causing the Kozai effect to switch off. At this stage the planetesimal has only completed a portion of its Kozai cy- cle and is left with high eccentricity (panel 1) and a reduced inclination (panel 3), at least for the duration of the simulation. The apsidal precession angles are depicted in panel 4 of Fig.10. During the time when the Kozai effect is operating (αi > αK), the apsidal precession is halted such that ωi = 0. After the Kozai effect has switched off (αi < αK) we observe prograde apsidal precession ( ωi > 0). Although this precession rate is dominated by the perturber the disc enhances the prograde pre- cession rate, since δi > αd at this stage. The total rate is thus given by the sum of contributions by the binary companion and disc. Planetesimals with semi-major axes ai > 12 AU decou- ple from the disc due to the differential precession induced by the binary companion. Consequently their inclinations get per- turbed below αK before the Kozai effect can start to operate, and their eccentricities stay low, at least for the duration of the simulations that we present here. The simulation presented in Fig.10 was only run for 110 orbits measured at the disc outer edge (∼ 104 years). Those planetesimals which have experienced the Kozai effect have only undergone half a Kozai cycle, which stalls because the inclination falls below αK. But we see in panel 3 that the in- clination relative to the binary is increasing toward αK at the point when the simulation ends, because the planetesimals have precessed nodally by more than 180◦ with respect to the disc. We would therefore expect that the Kozai effect will switch on again when αi > αK, allowing the previous Kozai cycle to complete. Similarly, we see that the inclinations of the plan- etesimals located beyond 12 AU are also increasing toward αK, such that the Kozai effect will switch on for them eventually. A long term effect of the disc thus appears to be to prolong the period associated with the Kozai cycles because of its effect on the planetesimal inclinations. An important consequence of this is that it increases the dephasing of the Kozai cycles at dif- ferent locations in the disc, ensuring that over long times plan- etesimals with different semi-major axes are at very different phases of their Kozai cycles. There will therefore be very large variations in both eccentricity and inclination within the plan- etesimal swarm, leading to very large collisional velocities be- tween planetesimals that are well separated in their semi-major axes. 18 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 5.4.1. Varying planetesimal sizes (model 3) 5.4.2. Collisional velocities for the highly inclined case (model 9) In this section we investigate the relative velocities for the high inclination (γF = 0.78) (45◦) case. The procedure to detect In the previous discussion we focused on planetesimals with size si = 10km, for which the gas drag forces are very weak. As we decrease the size of the bodies in the system, the gas drag becomes more important since the stopping time is ∝ si. Results for the 1km and 100m sized planetesimals are shown in the left and right panels of Fig.11 respectively. The Kozai effect leads to the growth of eccentricities (panels 3 and 4) and rela- tive inclinations (panels 7 and 8) for both particle sizes shown. This is not surprising, since the time scale on which we expect substantial damping of eccentricity and inclination to occur is of the order of the gas drag stopping time. For our disc model this is given by: Pd, (43) AU(cid:19)2 m(cid:19)(cid:18) a τS = 0.81(cid:18) si where si is expressed in metres. For the 100m sized planetes- imals at a = 6 − 15 AU the stopping time is of the order of 103 − 104 orbits, respectively, and is a factor of 10 longer for the 1km sized planetesimals. This is much longer than the time scale on which the Kozai effect operates (∼ 102 orbits), and hence we would not expect the gas drag to prevent growth of eccentricity and relative inclination in this size range. We com- ment here (without showing results), that we also found the Kozai effect to operate for 10m sized bodies, for which the drag and Kozai time scales are similar. The increased effect of gas drag causes the semi-major axes to decay (see panels 1 and 2 in Fig.11). As planetesimals ex- perience the Kozai effect, and develop large eccentricities and relative inclinations, their relative velocities with respect to the gas disc becomes very large (since vi − v2 ≃ v2 expression [29]). As they travel through the disc they lose ki- netic energy rapidly to the gas, and their semi-major axes de- cay. We note that those 100m sized bodies lying beyond 12 AU, which do not experience the Kozai effect during the simulation, also undergo rapid decay in their semi-major axes. These par- ticles develop large inclinations relative to the disc due to the rapid nodal precession induced by the companion, and the re- sulting gas drag as they pass through the disc leads to their orbital decay. i(cid:17) from K(cid:16)e2 i + δ2 Before the Kozai effect starts to operate we can see that the increased effect of gas drag for the 100m sized bodies (right panels of Fig. 11) causes damping of the relative inclination (panel 8) and therefore a decrease of the amplitude of the oscil- lation about the disc midplane caused by disc gravity (panel 6). In other words, the increased gas drag causes the 100m sized bodies to remain closer to the disc midplane, with the oscilla- tions in the amplitude of relative inclination being damped to a larger degree than occurs for larger bodies. This effect will become important when discussing relative velocities between the differently sized planetesimals, as the size dependence of the inclination perturbations will have an effect on the Kozai effect operating on the planetesimals, causing planetesimals of different sizes to orbit in planes which are significantly inclined with respect to one another. We note that the relative inclina- tions of both the 1km and 10km sized bodies are found to be similar, since the gas drag is weak in both of these cases. Fig. 11. Orbital elements for the 1km (left panels) and 100m sized planetesimals (right panels). The semi-major axes are shown in panels 1 and 2. In panels 3 and 4 eccentricity growth due to the Kozai effect is still apparent for both planetesi- mal sizes. The inclination is shown in panels 5 and 6, where the short dashed lines represent the disc inner and outer edge, and the long dashed line marks the threshold inclination above which the Kozai effect operates. In panels 7 and 8 the relative inclination with respect to the disc is shown, where the short dashed line indicates the one pressure scale height limit for the gas disc. The dash-dotted line represents the inclination of the binary orbit plane relative to the disc midplane. M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 19 collisions and measure relative velocities is identical to the one applied to the low inclination case (model 8). In Fig.12 (panel 1 and 2, solid lines) we present collisional velocities be- tween equally and differently sized planetesimals, respectively. Collisional velocities are even higher in this case (∼ a few km/s) than for the low inclination case for all size combinations con- sidered. The orbital elements are displayed in Fig.13. We observe that the Kozai effect causes an increase of eccentricity up to e = 0.4-0.6, such that the contribution to the relative velocity between planetesimals due to the term ∝ ei∆ωi becomes large, and this is the dominant cause of high velocity collisions be- tween equal sized bodies. The situation is different for planetes- imals of different sizes. The different gas drag strengths change the relative inclination between the planetesimals and the disc, resulting in a variation of the inclination perturbations caused by disc gravity, which on the one hand causes the value of ωi that is approached during the first half of the Kozai cycle to be different, and on the other hand causes the Kozai mechanism Fig. 13. Average orbital parameters of the colliding planetes- imal pairs in the high inclination γF = 45◦ case. The line colours and styles are as follows: green-solid (10km-10km); blue-solid (1km-1km); red-solid (100m-100m); green-dashed (10km-1km); blue-dashed (10km-100m); red-dashed (1km- 100m). to operate on different time scales. As a result planetesimals of different sizes decouple from the disc at different times. As can be seen from panels 4 and 5 of Fig.13, the orbital planes and pericentres of different sized planetesimals become randomly distributed with ∆Ωi ∼ 1 and ∆ωi ∼ 1, and the relative ve- locities are increased substantially due to the terms which are proportional to ei∆ωi, ∆αi, and sin (αi)∆Ωi in Eq. (28), with the latter term being the dominant one. As found in the low incli- nation case, the relative velocities of different sized planetesi- mals are generated due to misalignment of their pericentres ∆ωi and their orbital planes ∆Ωi. In contrast to the low inclination model, however, the eccentricities, ei, inclinations αi, and the differential nodal precession angle, ∆Ωi, are much larger due to the Kozai effect, leading to very large collisional velocities. In Fig.12 (panel 3) we display the number of collisions de- tected for the different size combinations, and it may be ob- served that the number of collisions drops to ∼ 100 for colli- sions between differently sized planetesimals. As the bodies be- gin to orbit in different planes their encounter probability is re- duced, as discussed in Sect. 5.2.1 and reported by Xie & Zhou Fig. 12. Average collisional velocities in the high inclina- tion γF = 0.78(45◦) case between equally (panel 1) and differently sized planetesimals (panel 2). Left panel: 10km- 10km collisions (green-solid); 1km-1km collisions (blue- solid); 100m-100m collisions (red-solid). Right panel: 10km- 1km (green-solid); 10km-100m (blue-solid); 1km-100m (red- solid). Threshold velocities for catastrophic disruption corre- sponding to the different size-combinations for the strong ag- gregates (dashed line) are also shown. Relative velocities have also been calculated using a larger orbit width ∆a = 2·10−3 AU (dotted lines). Panels 3 and 4: collision count for orbital width ∆a = 2 · 10−4 AU and ∆a = 2 · 10−3 AU, respectively. The line colours and styles in panel 3 and 4 are as follows: green- solid (10km-10km); blue-solid (1km-1km); red-solid (100m- 100m); green-dashed (10km-1km); blue-dashed (10km-100m); red-dashed (1km-100m). 20 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems (2009). This implies that the averaged relative velocities (pan- els 1 and 2) are obtained from only ∼ 102−103 data points, and the results are in danger of becoming statistically unreliable. In order to compare these collision velocities with more statisti- cally significant data, we also calculated the relative velocities while adopting a larger cross sectional area for the intersecting orbits corresponding to ∆r = 2 · 10−3 AU. The collision prob- ability is increased by a factor of ∼ 10, as observed in panel 4 of Fig.12. These relative velocities are plotted as dotted lines in panels 1 and 2 of Fig. 12, and are almost indistinguishable from the data obtained with the smaller orbit width, suggesting that the results are reliable. As was discussed for the low inclination model, the reduction in the collision frequency can have poten- tially important consequences for planetesimal accretion, since it may favour collisions between similar sized bodies which or- bit in similar planes. But, it is clear from Fig.12 that when the Kozai effect is active, large collision velocities are obtained for all size combinations. The disruption/erosion threshold velocity for the strong ag- gregates is shown using the dashed lines in panels 1 and 2. The collisional velocities are substantially larger than those re- quired for erosion or catastrophic disruption for all size combi- nations in this case, leading to the unsurprising conclusion that planetesimal accretion is not possible if the Kozai mechanism operates during the epoch of planet formation. For the relative velocities observed in Fig.12 of ∼ 2km/s for equally sized plan- etesimals, we estimate that collisions will always lead to frag- mentation unless bodies of size ∼ 103 km have already formed. Fig. 14. Orbital elements for model 4 (left panels) and model 5 (right panels). Only planetesimals with semi-major axes a ≤ 13 AU are included. In model 4 the Kozai effect is still operating, while in model 5 it is not. Panels 1 and 2 show the eccentricity. Panels 3 and 4 display the inclination, where the short dashed lines represent the inner and outer edge of the disc and the long dashed line indicates the threshold inclination of 0.68 (39.2◦) above which the Kozai mechanism can operate. 6. Suppressing the Kozai effect 6.1. Increasingthediscmass(models4and5) The Kozai effect generates relative velocities which are always too large for collisional growth of planetesimals, at least for sizes in the range (100m-10km). An interesting question is un- der what conditions does the Kozai mechanism cease to oper- ate. As discussed in Sect. 5.3 the Kozai mechanism relies on the apsidal precession induced by the binary companion halting, such that ωi = 0, which then allows for a secular change of the eccentricity and inclination values. If apsidal precession can be induced by another source, however, such as the disk for exam- ple, then we would expect the Kozai mechanism to be ineffec- tive once this induced precession is fast enough (Wu & Murray 2003). Since the apsidal precession rate caused by the disc scales with disc mass (see Eq.30), we have run simulations with an increased disc mass to see if the Kozai mechanism can be suppressed in this way. The results are shown in Fig.14 for model 4 (left panels) and 5 (right panels) with disc masses of Md = 3.3 M and 6.6 M, respectively, where M is the nominal disc mass used in model 3. As can be seen from the figure, the Kozai mechanism still switches on for the Md = 3.3 M case but does not operate in the Md = 6.6 M model. Consequently the eccentricity grows to large values in model 4, as seen in panel 1 of Fig.14, while it remains low in model 5 (panel 2). 6.2. Increasingbinaryseparation(models6and7) Another way to increase the relative importance of the disc- induced apsidal precession is to decrease the gravitational in- fluence of the binary companion. In this section we present simulations in which the separation of the binary system was increased to examine when the Kozai effect is suppressed. We consider separations of D = 90 AU and D = 120 AU in models 6 and 7, respectively. Because the Kozai time scale is ∝ D3, extremely long simulation times are required to prove the exis- tence (or non existence) of the Kozai effect. The required run times would be prohibitive for a full 3D hydrodynamic simula- tion, so we adopted the compromise of switching off the hydro- dynamic evolution of the disc in order to speed up the run times for these models. The disc in these simulations therefore re- mains axisymmetric and static in the precessing frame, but our adoption of a precessing reference frame causes it to precess around the binary angular momentum vector at the prescribed rate. We tested the accuracy of this approach by re-running model 3 using a static, precessing disc model, and found sim- ilar results when compared with the run in which the disc was allowed to evolve self-consistently. Differences between the static, precessing disc model and the full hydrodynamic sim- ulation arise mainly because the disc experiences a low am- plitude nodding motion (oscillation in its inclination) with a period equal to half the binary orbit period in the full simula- tion (Larwood et al. 1996; Fragner & Nelson 2010), where this M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 21 induced apsidal precession. Since we only consider cases for which δi ≤ αd, the latter is given by Eq. (30). Hence the to- tal apsidal precession rate experienced by a planetesimal on a circular orbit (ei = 0) is: = (44) 3π 2D3 (20ai) 3 2 (cid:16)2 − 5 sin2 (ωi) sin2 (αi)(cid:17) − ωD, ∂ωi ∂t where a and D are expressed in units of AU. The first term on the right hand side of Eq. (44) accounts for the binary-induced precession, and ωD is the disc-induced precession rate given by Eq. (30). Both of these rates are expressed in time units of orbits at a = 20 AU (the time unit used throughout the paper). In order for the Kozai mechanism to operate we require that ∂ωi ∂t = 0. This can only be true if the prograde term due to the binary companion is larger than the retrograde term of the disk. For the Kozai effect to operate, we therefore require that: 2 ≥ ωD. 3π D3 (20ai) 3 Using Eq. (30), we can express this a condition on the total disc mass in terms of the planetesimal semi-major axis and binary orbit separation: (45) AU(cid:19)2#−1 . (46) 3π M D3 (20ai) 3 2 "C0 + C1(cid:18) ai AU(cid:19) + C2(cid:18) a Md ≤ If this inequality is fulfilled we expect the Kozai mechanism to operate. In Fig.16 we show how the upper limit on the disc mass varies as a function of binary separation in order for the Kozai mechanism to just switch on/off for a planetesimal or- biting at a representative value of the semi-major axis ai = 11 AU, which is approximately half of the disc tidal truncation ra- dius. The red area corresponds to the regime where the inequal- ity is fulfilled and the Kozai-mechanism should operate. The cyan area marks the parameter space where the Kozai mecha- nism should not operate (large disc mass/large binary separa- tion). The symbols represent the simulation results, where open squares indicate that the Kozai mechanism was found to oper- ate (models 3, 4 and 6), while crosses represent the models in which the Kozai effect was ineffective (models 5 and 7). We see that the numerical results agree with our predictions. We also plot the boundary lines that correspond to semi- major axes of ai = 6 AU (dashed line) and ai = 15 AU (dot- ted line). Between ai = 6 AU and ai = 11 AU the precession frequency due to the disc is nearly independent of semi-major axis, as can be seen from the solid line in panel 7 of Fig.1. Yet the binary precession still increases as ∝ a 3 2 . Hence the param- eter space for which the Kozai mechanism operates becomes larger as the semi-major axis increases. Between ai = 11 AU and ai = 15 AU the disc precession rate increases in magni- tude approximately as ∝ a3/2 . Thus the boundary separating the Kozai-active versus Kozai-inactive regions does not change very much beyond ai = 11 AU. i 7. Discussion and conclusions In this paper we have investigated the dynamics of planetesi- mals that are embedded in a gaseous disc that is perturbed by a binary companion on a circular, inclined orbit. In contrast to Fig. 15. Orbital elements for model 6 (left panels) and model 7 (right panels). The Kozai effect is operating in model 6, but is not in model 7. The eccentricities are shown in panels 1 and 2. The inclination is shown in panels 3 and 4. The long dashed lines in panels 3 and 4 represents the 39.2◦ inclination thresh- old, above which the Kozai effect can operate. is driven by the time-varying gravitational field of the binary. Although the effect is relatively small, it is likely to reduce the accuracy with which a rigidly precessing disc model can be used to determine planetesimal collision velocities in detail. But, it is a useful set-up for determining the disc mass for which the Kozai mechanism operates. The results are shown in Fig.15 for the D = 90 AU (left panels) and D = 120 AU model (right panels). We can observe that the Kozai effect is operating in model 6 while it is not in model 7. The time scale for the Kozai effect to increase the eccentricities is larger by a factor 3.3, as expected, for model 6 compared to the reference model 3 due to the Kozai time scale increasing as ∼ D3. If the Kozai effect was operating in model 7, we would expect to see substantial eccentricity growth after 500 orbits (a factor 8 longer than in our reference model). This is not seen, however, and we conclude that the Kozai effect cannot operate in model 7. 6.3. Atheoreticalargument We have seen in the previous two sections that the Kozai mech- anism can be suppressed provided the disc mass or the binary separation are large enough. Now we want to understand the numerical results by means of a theoretical argument. As has already been discussed, the Kozai mechanism relies on halting the apsidal precession of the planetesimals induced by the bi- nary companion. If this can be prevented then there should be no secular net change of eccentricities or inclinations. The to- tal net rate of apsidal precession is given by the sum of the contributions by the binary companion (Eq.38) and the disc 22 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems of the precessing disc should always be included in future studies of planet formation in inclined binary systems. -- Previous studies that focused on the influence of a coplanar, eccentric binary companion have shown that the eccentric binary generates a forced eccentricity in the planetesimal swarm. Gas drag acts to align the pericentres of planetes- imals which are of the same size, such that collisions be- tween same-sized bodies are dominated by keplerian shear. Collisions between different sized bodies occur with higher velocity due to the misaligned orbits (Marzari & Scholl 2000; Thebault, Marzari & Scholl 2006). We find that for a circular binary, where the eccentricity of planetesimals is largely generated by high frequency terms in the disturbing function, strong alignment of pericentres is not observed. -- Previous work which has examined planetesimal dynamics in modestly inclined and eccentric binary systems suggests that gas drag causes size-dependent phasing of both peri- centres and the lines of nodes of perturbed planetesimals (Xie & Zhou 2009). This has the effect of favouring colli- sions between same sized objects, for which the collision velocity is dominated by the keplerian shear, and as such it has been suggested that this may provide an effective chan- nel for planetesimal growth in inclined binary systems. Our results obtained for a binary with inclination γF = 25◦ are in basic agreement with this, as we find that same-sized planetesimals do indeed occupy very similar orbital planes, whereas the orbits of differently sized bodies are mutually inclined. This arises in our case because of the different amplitudes of oscillation about the disc midplane observed for planetesimals of different size, with smaller bodies re- maining closer to the midplane. We thus find that the col- lision velocities between differently sized bodies are sig- nificantly larger than between same-sized bodies (typically vcoll ≃ 200 ms−1 for different sized objects and vcoll ≃ 50 - 70 ms−1 for same-sized objects). We note that the lack of pericentre alignment observed in our simulations leads to the substantial collision velocities between same size bod- ies. Collisions with the velocities described above are likely to be erosive or disruptive, depending on the mass ratio of the colliding bodies. -- For highly inclined systems with γF = 45◦, we find that the Kozai mechanism can operate, causing large changes in the eccentricities and inclinations of the planetesimals, and leading to collisional velocities that are much too large to allow for planetesimal accretion. The long term influence of the disc in cases where the Kozai mechanism operates is to modify the period associated with the Kozai cycle by periodically forcing the planetesimal inclinations to fall be- low, or rise above, the critical value for the Kozai effect to operate, αK = 39.2◦. We find that planetesimals of size 10m - 10km all experience the Kozai effect. -- Increasing the disc mass or binary separation can suppress the Kozai effect, as disc gravity can induce apsidal preces- sion which is fast enough to render the Kozai mechanism ineffective. For a binary separation of D = 60 AU, and disc truncation radius of 20 AU, we find that we need to in- crease the disc mass by a factor of ∼ 6 above the minimum mass solar nebula (MMSN) value in order to switch off the Fig. 16. Plot of the parameter regime explored in the simu- lations. The diagram shows disc mass against binary separa- tion, with the red area denoting the region where the Kozai mechanism should operate, and the cyan area showing the re- gion where it should not for a planetesimal located at 11 AU. The symbols denote the outcome of the simulations, where a cross implies that the Kozai effect was switched off, and an open square indicates that the Kozai mechanism operates. The dashed and dotted line represents the same boundary for a body at ai = 6 AU and ai = 15 AU respectively. previous work the planetesimals are allowed to interact with the gaseous disc via gas drag and disc gravity. The disc evolves hydrodynamically because of gravitational perturbations due to the binary companion, and the disc model we consider under- goes solid body precession around the orbital angular momen- tum vector of the binary system, as expected when the warp propagation time via bending waves is shorter than the differ- ential precession time. Discs around young stars are expected to show similar behaviour since it is estimated that α < H/R in these systems, where α is the usual viscosity parameter. The key results of our study can be summarized as follows. -- In the absence of disc gravity, planetesimals undergo strong differential nodal precession, such that they eventually or- bit in different planes (Marzari, Thebault & Scholl 2009; Xie & Zhou 2009). They also precess relative to the disc, so their orbits become highly inclined relative to it. We find that disc gravity acts to prevent this differential precession, and forces the planetesimals to precess with the disc on av- erage. Viewed locally, the forcing of nodal precession by the binary companion causes the planetesimals to oscillate about the midplane of the disc, with an amplitude that de- pends on the size of the planetesimals because of the influ- ence of gas drag in damping this relative motion. This key result suggests that the influence of the gravitational field M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems 23 Kozai effect at a representative orbital radius of ai = 11 AU. Increasing the binary separation to D = 120 AU, and main- taining the mass of our disc at its nominal value (equivalent to the MMSN) also rendered the Kozai mechanism inoper- ative. In the light of the above findings, we can conclude that highly inclined, distant binary companions probably will not induce the Kozai effect during planet formation while the disc is present, although it may do so once the disc has dissipated, such that the final planetary system is strongly perturbed. But we note that the peak in the distribution of binary orbit sep- arations occurs at D ≃ 30 AU (Duquennoy & Mayor 1991; Ghez et al. 1993; Leinert et al. 1993), which is small enough for the Kozai mechanism to operate during the planet formation epoch when the gas disc is present. There are number of extra- solar planetary systems, however, which are observed to be in relatively close binary systems, γ Cephei being a notable exam- ple (see Kley & Nelson 2008, and references therein). Here the binary separation is only a few tens of AU, such that the Kozai mechanism could possibly have operated when planetesimals were accreting. The fact that a planet is observed in this system with a modest orbital eccentricity is probably an indication that the binary inclination is too small to allow the Kozai mecha- nism to operate. This is consistent with the findings of Hale (1994) that binary orbits with D < 40 AU tend to be reason- ably well aligned with the stellar spin axes. A key issue that needs to be addressed in the future is the evolution of a planetesimal swarm in a binary system where the orbit is both inclined and eccentric. Our results suggest that a circular binary system does not allow for the pericentres of per- turbed planetesimals to be well aligned, since a circular orbit is unable to impose a preferred direction on the system, and this is in clear contrast to studies which have considered an eccentric companion. It should be noted, however, that the inclusion of the disc gravity will be of crucial importance in such a study, since it will significantly perturb the orbits of the planetesi- mals. This is particularly the case when the planetesimal orbits are inclined relative to the disc midplane, and the presence of an eccentric binary companion may also cause the disc itself to become eccentric (Kley & Nelson 2008; Kley et al. 2008) (where the disc eccentricity may be dependent on the disc self- gravity (Marzari et al. 2009)), further complicating matters. We further note that the inner edge of our computational domain was located at R = 2 AU, so our present study examines planet formation in the region normally associated with giant planet formation. Studies of this type which examine terrestrial planet formation closer to the central star will require deployment of substantially more powerful computational resources than were used here, in order to simulate the larger range of time scales in the problem. Such a study should also include a more realistic planetesimal size distribution so that the possibilty of accretion occuring via collisions between similarly sized bodies can be explored. In this paper we have only considered planet formation via collisions between planetesimals, whereas dust accumu- lation onto individual planetesimals could still lead to plane- tary growth as long as collisions between planetesimals can be avoided. This issue has been recently been addressed by Paardekooper & Leinhardt (2010); Xie et al. (2010), where they show that km-sized planetesimals may grow in size by two orders of magnitude due to accretion of collisional debris if the efficiency of planetesimal formation stays low, i.e. only a few planetesimals form. This study involved two-dimensional simulations, and it will be interesting to examine how the re- sults change for a misaligned system where planetesimals may orbit out of the disc midplane where the majority of the dust and collisional debris will reside. As we have seen in Sect. 5.2, 10 km sized planetesimals tend to have orbits which are in- clined relative to the disc by more that the disc scale height, once their free particle rates are substantially different from the disc precession rate. Planetesimal formation via dust accretion is therefore only expected in radial regions of the disc where the precession rate of the planetesimals and the disc are well- matched. Closer to the inner or outer edge of the disc, efficient dust accretion may not be possible due to the large relative in- clinations of the planetesimals. This, and other issues discussed in this paper, will be the subject of future publications. Acknowledgements. The simulations presented in this paper were per- formed on the QMUL HPC facility purchased under the SRIF ini- tiative. We gratefully acknowledge detailed and insightful comments provided by the referee, Phillipe Th´ebault, which led to significant im- provements in this paper. MMF gratefully acknowledges the support of Stephan Rosswog and staff at the University of Bremen, where part of this work was carried out. References Artymowicz, P., & Lubow, S. H. 1994, ApJ, 421, 651 Benz, W., & Asphaug, E. 1999, Icarus, 142, 5 Ciecielag, P., Ida, S., Gawryszczak, A., & Burkert, A. 2007, A&A, 470, 367 Desidera, S. & Barbieri, M., 2007, A&A, 462, 345 Duquennoy, A., & Mayor, M. 1991, A&A, 248, 485 Eggenberger, A., Udry, S., Chauvin, G., Beuzit, J.-L., Lagrange, A.- M., S´egransan, D., & Mayor, M. 2007, A&A, 474, 273 Fragner, M. M., & Nelson, R. P. 2010, A&A, 511, A77 Ghez, A. M., Neugebauer, G., & Matthews, K. 1993, AJ, 106, 2005 Hale, A., 1994, AJ, 107, 306 Hayashi, C. 1981, Progress of Theoretical Physics Supplement, 70, 35 Heppenheimer, T.A., 1978, A&A, 65, 421 Housen, K.R., Holsapple K.A., 1990, Icarus, 84, 226 Housen, K.R., Holsapple K.A., 1999, Icarus, 142, 21 Innanen, K.A., Zheng, J.Q., Mikkola, S., Valtonen, M.J., 1997, AJ, 113, 1915I Kley, W. & Nelson, R.P., (2008), A& A, 486, 617 Kley, W., Papaloizou, J. C. B., & Ogilvie, G. I. 2008, A&A, 487, 671 Kortenkamp, S. J., Wetherill, G. W., & Inaba, S. 2001, Science, 293, 1127 Kozai, Y. 1962, AJ, 67, 591 Larwood, J.D., Nelson, R.P., Papaloizou, J.C.B., Terquem, C., 1996, MNRAS, 282,597 Leinert, C., Zinnecker, H., Weitzel, N., Christou, J., Ridgway, S. T., Jameson, R., Haas, M., & Lenzen, R. 1993, A&A, 278, 129 Lissauer, J., Stewart, G., 1993, Univ.of Arizona Press, Tucson, pp. 1061-1088 Lubow, S. H., & Ogilvie, G. I. 2000, ApJ, 538, 326 Marzari, F. & Scholl, H., 2000, ApJ, 543, 328 24 M.M. Fragner, R.P. Nelson & W. Kley: On collisional growth of planetesimals in misaligned binary systems Marzari, F., Scholl, H., Th´ebault, P., & Baruteau, C. 2009, A&A, 508, 1493 Marzari, F., Thebault, P. & Scholl, H., 2009, A& A, 507, 505 Murray, C.D., Dermott, S.F., 1999, Solar System Dynamics, Cambridge University Press Ogilvie, G. I. 2000, MNRAS, 317, 607 Paardekooper, S.J., Thebault, P. & Melemma, G., 2008, MNRAS, 386, 973 Paardekooper, S.J., Leinhardt, Z. M., 2010, arXiv1001.3025 Papaloizou, J.C.B. & Lin, D.N.C., 1995, ApJ, 438,841 Papaloizou, J.C.B. & Pringle, J.E., 1983, MNRAS, 202,1181 Papaloizou, J.C.B. & Terquem, C., 1995, MNRAS, 274,987 Shakura, N.I. & Sunyaev, R.A., 1973 A&A, 24 337 Stapelfeldt, K. R., Krist, J. E., Menard, F., Bouvier, J., Padgett, D. L., & Burrows, C. J. 1998, ApJ, 502, L65 Stewart, S.T., Leinhardt, Z.M., 2009, ApJ, 691, 133 Terquem, C., Eisloffel, J., Papaloizou, J. C. B., & Nelson, R. P. 1999, ApJ, 512, L131 Thebault, P., Marzari, F. & Scholl, H., 2006, Icarus, 183, 193 van Leer B., 1977, J. Comput.Phys., 23, 276 Wetherill, G. W., & Stewart, G. R. 1989, Icarus, 77, 330 Weidenschilling, S.J., 1977, MNRAS, 180, 57 Wu, Y., Murray, N., 2003, ApJ, 589,605 Xie, J.W., Zhou J.L., 2009, ApJ, 698, 2066 Xie, J.-W., Payne, M. J., Th´ebault, P., Zhou, J.-L., & Ge, J. 2010, ApJ, 724, 1153 Ziegler, U., Yorke H.W., 1997, Comput.Phys.Commun., 101, 54
1508.04769	1	1508	2015-08-19T20:05:43	Precise radial velocities of giant stars VIII. Testing for the presence of planets with CRIRES Infrared Radial Velocities	[ "astro-ph.EP", "astro-ph.SR" ]	We have been monitoring 373 very bright (V < 6 mag) G and K giants with high precision optical Doppler spectroscopy for more than a decade at Lick Observatory. Our goal was to discover planetary companions around those stars and to better understand planet formation and evolution around intermediate-mass stars. However, in principle, long-term, g-mode nonradial stellar pulsations or rotating stellar features, such as spots, could effectively mimic a planetary signal in the radial velocity data. Our goal is to compare optical and infrared radial velocities for those stars with periodic radial velocity patterns and to test for consistency of their fitted radial velocity semiamplitudes. Thereby, we distinguish processes intrinsic to the star from orbiting companions as reason for the radial velocity periodicity observed in the optical. Stellar spectra with high spectral resolution have been taken in the H-band with the CRIRES near-infrared spectrograph at ESO's VLT for 20 stars of our Lick survey. Radial velocities are derived using many deep and stable telluric CO2 lines for precise wavelength calibration. We find that the optical and near-infrared radial velocities of the giant stars in our sample are consistent. We present detailed results for eight stars in our sample previously reported to have planets or brown dwarf companions. All eight stars passed the infrared test. We conclude that the planet hypothesis provides the best explanation for the periodic radial velocity patterns observed for these giant stars.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. Trifonov_2015 October 10, 2018 c(cid:13)ESO 2018 5 1 0 2 g u A 9 1 . ] P E h p - o r t s a [ 1 v 9 6 7 4 0 . 8 0 5 1 : v i X r a Precise radial velocities of giant stars VIII. Testing for the presence of planets with CRIRES Infrared Radial Velocities (cid:63) Trifon Trifonov1,2, Sabine Reffert1, Mathias Zechmeister3, Ansgar Reiners3, and Andreas Quirrenbach1 1 Landessternwarte, Zentrum für Astronomie der Universität Heidelberg, Königstuhl 12, 69117 Heidelberg, Germany 2 Department of Earth Sciences, The University of Hong Kong, Pokfulam Road, Hong Kong 3 Institut für Astrophysik, Georg-August-Universität, Friedrich-Hund-Platz 1, 37077 Göttingen, Germany Received March 27, 2015; accepted August 4, 2015 ABSTRACT Context. We have been monitoring 373 very bright (V ≤ 6 mag) G and K giants with high precision optical Doppler spectroscopy for more than a decade at Lick Observatory. Our goal was to discover planetary companions around those stars and to better understand planet formation and evolution around intermediate-mass stars. However, in principle, long-term, g-mode nonradial stellar pulsations or rotating stellar features, such as spots, could effectively mimic a planetary signal in the radial velocity data. Aims. Our goal is to compare optical and infrared radial velocities for those stars with periodic radial velocity patterns and to test for consistency of their fitted radial velocity semiamplitudes. Thereby, we distinguish processes intrinsic to the star from orbiting companions as reason for the radial velocity periodicity observed in the optical. Methods. Stellar spectra with high spectral resolution have been taken in the H-band with the CRIRES near-infrared spectrograph at ESO's VLT for 20 stars of our Lick survey. Radial velocities are derived using many deep and stable telluric CO2 lines for precise wavelength calibration. Results. We find that the optical and near-infrared radial velocities of the giant stars in our sample are consistent. We present detailed results for eight stars in our sample previously reported to have planets or brown dwarf companions. All eight stars passed the infrared test. Conclusions. We conclude that the planet hypothesis provides the best explanation for the periodic radial velocity patterns observed for these giant stars. Key words. Planetary systems, Infrared: stars, Instrumentation: spectrographs, Methods: observational, Techniques: radial velocities 1. Introduction By June 2015, the number of confirmed substellar companions discovered with the Doppler technique has reached ∼ 6001. This number constitutes ∼ 30% of the total number of planets dis- covered using all planet search methods. In addition, many more extrasolar planets, in particular those discovered with the tran- siting method, have been confirmed with Doppler spectroscopy. This shows that the precise radial velocity (RV) method still re- mains one of the most valuable extrasolar planet search tools to date. Ultra stable échelle spectrographs, such as HARPS (Mayor et al. 2003), have already reached sub-m s−1 precision and poten- tially allow astronomers to search for Earth mass planets. The Doppler technique, however, can suffer from false positive de- tections due to radial velocity variations intrinsic to a star. In addition to short period p-mode pulsations (Barban et al. 2004; Zechmeister et al. 2008), which are excited by convection and seen as stellar radial velocity scatter, stellar surface spots or even nonradial gravitational g-mode pulsations can potentially cause (cid:63) Based on observations collected at the European Southern Observa- tory, Chile, under program IDs 088.D-0132, 089.D-0186, 090.D-0155 and 091.D-0365. 1 http://exoplanet.eu line shape deformations, which can be misinterpreted as velocity shifts. So far, more than 60 substellar companions have been de- tected around evolved G and K giant stars using the Doppler method. Some of these stars have RV signals clearly consistent with highly eccentric substellar companions that cannot be mis- taken for stellar activity (e.g., Frink et al. 2002; Moutou et al. 2011; Sato et al. 2013), but for others an alternative explanation of the data cannot be excluded despite their periodic RV sig- nals. While long-term g-mode pulsations are rather unlikely to be excited in the large convective layers in these evolved stars, temperature spots on the other hand could effectively mimic a planet (e.g., Hatzes & Cochran 2000). As K giants are inflated stars compared to their main-sequence progenitors, their rota- tional period can be of the order of hundreds of days. Therefore, in case of spot(s) on the stellar surface, the line profile variations can lead to RV variations with long periods and semiamplitudes significantly surpassing the intrinsic stellar jitter that can be eas- ily mistaken for a planet. Detailed studies of spectral line profile bisectors have often been used to support the companion hypothesis. This method requires very high resolution spectra, however, and a stable in- strumental profile, and the absence of variations in the spectral line shape is a necessary condition, but is not sufficient to prove the existence of a substellar companion definitively. Article number, page 1 of 14 A&A proofs: manuscript no. Trifonov_2015 Measuring RV signals in the near-infrared (near-IR) is an- other promising test that can be performed. The contrast ratio be- tween the star's surface blackbody intensity and that of a cooler spot is much larger in the optical than in the IR domain. In case of spots, line shape deformations in the IR data must have a much smaller amplitude (Desort et al. 2007; Reiners et al. 2010). We can also expect different RV amplitudes in the two domains in case of nonradial g-mode pulsations, while this test should yield no difference in the optical and near-IR radial velocities in case of a planet. A prominent example for the discussion of spot-induced ra- dial velocities is the young star TW Hya; Setiawan et al. (2008) found periodic radial velocity variations, but no indication of line bisector variations and concluded that the star is orbited by a planet. Later, Huélamo et al. (2008) observed TW Hya at infrared wavelengths but could not find RV variations con- sistent with the Keplerian solution from optical measurements. Huélamo et al. (2008) concluded that no planet orbits TW Hya. High precision RVs derived from the infrared wavelength regime can form a rather critical test for planet confirmation. This test is valid mostly for active stars for which other viable explanations for periodic radial velocities, such as spots or pul- sations, cannot be excluded by any other means, such as giant stars in particular. An obvious choice for this test is the ESO pre-dispersed CRyogenic InfraRed Echelle Spectrograph (CRIRES), mounted at the Nasmyth focus B of the 8m VLT UT1 (Kaeufl et al. 2004). Several studies demonstrated that radial velocity mea- surements with precision between 10 and 35 m s−1 are possible with CRIRES. Seifahrt & Käufl (2008a) reached a precision of ≈ 35 m s−1 for CRIRES radial velocities, using the N2O gas cell for calibration. Huélamo et al. (2008) and Figueira et al. (2010) showed that a Doppler precision of ≈ 10 m s−1 can be achieved when using telluric CO2 lines in the H−band as wavelength ref- erence. We present our CRIRES near-infrared Doppler results for 20 evolved G8-K4 III giant stars. All selected targets in this study have periodic radial velocities as derived from optical spectra, and are thus potential planet hosts. They constitute a small sub- sample of our K giant planet search sample in the optical (Frink et al. 2001; Reffert et al. 2015). In Section 2 we briefly introduce the background of this pro- gram. Section 3 describes our observational setup with CRIRES. In Section 4 we explain in detail the data reduction and analysis process with our CRIRES pipeline. Our methods of extracting the precise RVs are given in Section 5, and in Section 6 we dis- cuss the consistency between optical and IR data. In Section 7 we discuss our results, and we provide a summary in Section 8. 2. The K giant sample Our Doppler survey started in 1999 at Lick Observatory using the Hamilton spectrograph in conjunction with an iodine cell (Marcy & Butler 1992; Butler et al. 1996). The initial goal of our program and the star selection criteria are described in Frink et al. (2001). Briefly, we regularly observed 373 very bright (V ≤ 6 mag) and photometrically constant G and K giants se- lected from the Hipparcos Catalog. The objective of our program is to investigate and understand giant planet occurrence and evo- lution around intermediate-mass stars. The first planet from our program was discovered around the K giant star ι Dra (Frink et al. 2002). The highly eccentric Keplerian reflex motion of ι Dra seen in the Lick data leaves no doubt on the planet's existence. Therefore, ι Dra b became Article number, page 2 of 14 the first confirmed extrasolar planet around a giant star and en- couraged more scientists to search for extrasolar planets around evolved stars with the Doppler method. As a result, to date more than 60 planets2 have been discovered around giants, and their number is constantly growing. Planet occurrence statistics and more results from our Lick survey are given in Reffert et al. (2015). We distinguish between planets and planet candidates among the companions, depend- ing on how persistent the RV signal is over many cycles, and how large the RV amplitude is compared to the intrinsic RV jit- ter caused by short-term radial pulsations. A subsample of 20 stars with planets and planet candidates from our survey is of particular interest, and are studied further here. All these stars display either one or two clear periodicities in their RVs, consistent with one or two orbiting substellar com- panions with minimum masses ranging between a few Jupiter masses and a low-mass brown dwarf. If stellar spots were re- sponsible for the obtained Doppler velocities, they would corre- spond to photometric variability (Hatzes 2002). Precise photom- etry from Hipparcos shows that our sample is photometrically stable down to 3 mmag, and thus there are no indications for any intrinsic features to be related to a Doppler signal (Reffert et al. 2015). Nevertheless, since we cannot fully exclude long-period pulsations (g−mode or of unknown nature), we have observed our subsample of 20 stars with CRIRES over four semesters in an attempt to provide more evidence in favor of the companion hypothesis. Stars are listed in Table 1, including their Hippar- cos and HD number, apparent magnitude in V and H bands, as well as basic physical parameters, such as stellar mass M, ra- dius R, and luminosity L as given in Reffert et al. (2015). Some planetary systems from the subsample have already been pub- lished: HIP 37826 (Reffert et al. 2006), HIP 88048 (Quirrenbach et al. 2011), HIP 34693 and HIP 114855 (Mitchell et al. 2013), and HIP 5364 (Trifonov et al. 2014). Others have been discov- ered independently: HIP 37826 (Hatzes et al. 2006), HIP 20889 (Sato et al. 2007), HIP 60202 (Liu et al. 2008), and HIP 31592 (Wittenmyer et al. 2011). Publication of the remaining stars with confirmed planets is in preparation. 3. Observational setup The ESO's CRIRES instrument is a temperature stabilized near- infrared spectrograph with a maximum resolution of R ≈ 100 000 when used with a 0.2(cid:48)(cid:48) slit. Despite a broad accessible wave- length range from 950 to 5200 nm (J, H, K, L, and M infrared bands), the available wavelength coverage per single observation in this predispersed échelle spectrograph is much smaller than in optical cross-dispersed échelle spectrographs and currently lim- ited to one spectral order3. A particular wavelength setting can be selected for each observation, with a free spectral range on the order of 20-200 nm, depending on the spectral order and the dispersion for that setting. The spectrum is imaged by a mosaic of four Aladdin III detectors, forming an effective 4096 x 512 pixel array (with a gap of ∼ 280 px between the detectors). This arrangement is somewhat problematic because a precise wave- length solution must be obtained separately for each detector. The standard CRIRES pipeline recipes offer in principle this 2 http://www.lsw.uni-heidelberg.de/users/sreffert/giantplanets.html 3 Currently, CRIRES is removed from UT1 for an upgrade and is ex- pected to be operational again in 2017 under the name CRIRES+. The new instrument is expected to cover simultaneously a wavelength range about ten times larger than the original (Dorn et al. 2014). Table 1. List of observed stars. Trifon Trifonov et al.: Precise radial velocities of giant stars Rβ [R(cid:12)] HIP number 14.3 ± 0.2 5364 13.9 ± 0.6 19011 13.1 ± 0.2 20889 126.3 ± 12.6 23015 5.1 ± 0.3 31592 36.8 ± 0.7 34693 13.2 ± 0.3 36616 8.7 ± 0.3 37826 69.4 ± 11.3 38253 20.4 ± 1.4 39177 15.6 ± 0.4 60202 44.0 ± 2.1 73133 26.5 ± 1.1 74732 26.8 ± 0.8 79540 29.5 ± 1.6 80693 53.4 ± 2.7 84671 14.6 ± 0.3 88048 91004 23.6 ± 0.5 100587 11.0 ± 0.1 114855 α - Hipparcos Catalog, β - Reffert et al. (2015), γ - 2MASS Catalog Sp.typeα Mβ [M(cid:12)] 1.7 ± 0.1 K2 III 2.1 ± 0.3 K1 III 2.4 ± 0.2 K0 III 4.5 ± 0.6 K3 II 1.4 ± 0.2 K1 III 2.3 ± 0.3 K2 III 1.9 ± 0.2 K2 III 2.3 ± 0.2 K0 III 2.0 ± 0.5 K3 III 2.7 ± 0.6 K3 III 2.2 ± 0.3 G8 III 1.2 ± 0.1 K4 III 1.5 ± 0.2 K2 III 1.1 ± 0.2 K3 III 1.2 ± 0.1 K4 III 1.2 ± 0.1 K4 II 2.7 ± 0.2 K0 III 6.4 ± 2.0 K3 III 1.4 ± 0.2 K3 III 1.4 ± 0.1 K0 III HD number 6805 25723 28305 31398 47205 54719 59686 62509 63752 65759 107383 131918 135534 145897 148513 156681 163917 171115 194317 219449 Lβ [L(cid:12)] 71.1 ± 1.1 89.5 ± 7.2 85.3 ± 2.7 3752.8 ± 675.1 12.0 ± 0.6 238.4 ± 11.4 73.3 ± 3.4 39.8 ± 0.6 1169.2 ± 379.8 473.2 ± 134.2 112.0 ± 4.6 408.2 ± 39.8 200.1 ± 16.0 190.9 ± 10.5 206.4 ± 21.5 554.0 ± 54.4 109.3 ± 3.1 159.3 ± 5.4 51.5 ± 1.1 . . . V α [mag] Hγ [mag] N obs. 3.46 5.62 3.53 2.69 3.95 4.41 5.45 1.16 5.60 5.60 4.72 5.48 5.52 5.24 5.41 5.03 3.32 5.49 4.43 4.24 1.0 3.1 1.3 −0.7 1.7 1.8 3.1 −0.8 2.5 2.9 2.5 2.3 2.7 2.3 2.3 1.5 1.3 1.3 1.4 1.7 9 6 8 6 9 8 8 8 8 7 7 6 7 8 8 5 10 7 7 8 . . . kind of calibration, but at a level of radial velocity precision far below that required for the Doppler detection of planets. Thus, we had two options when calibrating our spectra: to use a similar approach to the I2 cell method using N2O or NH3 gas cells (Seifahrt & Käufl 2008a; Bean & Seifahrt 2009), or to use atmospheric telluric lines (Huélamo et al. 2008; Figueira et al. 2010). We chose to follow the telluric method. 3.1. Spectral window Our wavelength setting was selected by inspecting the Arcturus near-IR spectral atlas (Hinkle et al. 1995). We searched for dense stellar line regions that also contained deep and sharp telluric lines to be used for wavelength calibration. We chose a wavelength setting in the H-band (36/1/n in the CRIRES user manual), with a reference wavelength of λ = 1594.5 nm (wavelength in the middle of detector 3). In fact, the selected region is very close to the wavelength setup successfully used by Figueira et al. (2010). Our setup is characterized by the presence of many sharp atmospheric CO2 lines, which we used as wavelength calibrators. Unfortunately, the CO2 lines only fall on detector 1 and 4. Although detectors 2 and 3 contain many stellar lines, we were not able to construct a wavelength solu- tion for these spectra on these detectors, and thus we excluded detectors 2 and 3 from our analysis. 3.2. Observations We performed the observations with a standard ABBA nodding cycle, including three jitter observations per nod to subtract the sky background emission. The total exposure time required for each individual target to reach a S/N ≥ 300 at the reference wave- length has been estimated with the ESO Exposure Time Calcu- lator4 (ETC) to be between 18 seconds for the brightest stars and 4 http://etimecalret.eso.org/observing/etc/bin/gen/form?INS.NAME= CRIRES+INS.MODE=swspectr 3 minutes for the faintest stars. To achieve the highest possible precision, the spectrograph is used with the 0.2(cid:48)(cid:48) slit, resulting in a resolution of R = 100 000. To minimize RV errors related to the imperfect stability of the slit illumination, the observations were done in NoAO mode (without adaptive optics), and nights with poor seeing conditions were requested if possible. Since the periods derived from optical RV data are typically one to two years long, we took at least four RV measurements per year, if possible. During the four semesters of observations some targets were given higher priority than others; the number of data points thus ranges from six to ten. 4. Calibration and data reduction Dark, flat, and nonlinearity corrections, as well as the combina- tion of raw jittered frames in each nodding position, were per- formed using the standard ESO CRIRES pipeline recipes. The final output from the CRIRES pipeline is an extracted, one- dimensional coadded spectrum, but we also obtained spectra from the individual A and B nodding frames separately. The CRIRES pipeline wavelength calibration is based on a robust cross-correlation technique between observed spectra and ThAr lamp lines or, alternatively, from a physical model employing ray tracing. However, the precision of both wavelength solutions was insufficient for obtaining precise radial velocities. We found the wavelength solution from the physical model to be more pre- cise, and thus we used it as an initial guess for the calibration of each spectral frame with telluric lines. For further data analysis, we developed a semiautomatic pipeline based on a sequence of χ2 minimization algorithms. Fig. 1 and 2 illustrate our data reduction steps on detector 1 and 4 for the first observation of the K giant HIP 60202. 4.1. Normalization Continuum normalization of G and K giant spectra is compli- cated by the high density of spectral lines, which leave almost Article number, page 3 of 14 A&A proofs: manuscript no. Trifonov_2015 Fig. 1. Top: model spectra of the telluric transmission for detectors one (left) and four (right). Bottom: calibrated science spectra for both de- tectors (black) showing the stellar and telluric absorption lines. Only nonblended telluric lines were modeled (red) and used for constructing the wavelength solution. no continuum level points to work with. To perform a proper continuum normalization, we thus developed an automated al- gorithm, which identifies areas of the raw spectrum that are free of absorption lines. To find the real continuum points on the raw spectrum, the same algorithm was applied on a noise-free spec- trum constructed by combining a theoretical telluric spectrum and a synthetic stellar spectrum, which was shifted to correct for the barycentric velocity valid for the given epoch of observations (see Section 4.2). The matching continuum points between both spectra were selected for the construction of a low-order polyno- mial fit that represents the estimated continuum level. The raw individual spectrum was then divided by the continuum solu- tion, yielding a normalized spectrum. Extensive testing showed that this algorithm performs a much better continuum normal- ization in the case of high spectral line densities than the usual algorithm, which assumes that all points below a given threshold belong to the continuum. 4.2. Wavelength calibration For precise Doppler measurements with CRIRES, a precise wavelength calibration for each epoch and detector must be ob- tained. We applied several automated steps to first identify all CO2 lines used for wavelength calibration in the spectra and then to derive their line centroids in pixel and wavelength space. First we identified all significant absorption lines in the nor- malized spectrum with relative depths below 0.85 of the con- tinuum level (Pepe et al. 2002). Telluric lines from CO2 are nearly static, with relatively equidistant wavelengths, and thus were easily identified in the spectra as such. Nevertheless, for precise wavelength calibration on the basis of telluric CO2 lines, one should take into account that the line centroids may vary de- pending on the ambient climate above the observatory at the time when the spectrum is obtained. Thus, to assure precise calibra- tion, we adopted methods for telluric spectral synthesis similar Article number, page 4 of 14 to those used in Seifahrt & Käufl (2008b), Seifahrt et al. (2010) and Lebzelter et al. (2012). We used the physical wavelength solution from the CRIRES recipes as initial guess to roughly identify the wavelength re- gion of each observational epoch and for each detector. Next, we constructed a high resolution telluric transmission spectrum using the Line By Line Radiative Transfer Model5 (LBLRTM) code for the wavelength region of interest. The precise molec- ular line positions needed for the model were taken from the HITRAN database (Rothman et al. 1998). For the atmospheric profile above VLT, we adopted the mid-latitude summer/winter meteorological model, which is implemented in the LBLRTM. Additional input parameters for the atmospheric model in- cluded the target's zenith angle as well as ambient atmospheric pressure and temperature at the time of observation. For lack of anything else we used ground-layer pressure and tempera- ture, which might not adequately reflect the conditions higher up in the atmosphere. We were only interested in the CO2 ab- sorption spectra, but to avoid confusion resulting from any unre- solved lines that might appear on science spectra we used all the available atmospheric molecules from HITRAN and obtained the telluric mask. Many other very weak lines caused by other molecules appeared in the theoretical telluric spectra, but they could not be identified separately in the science spectra. Weak molecular lines affect the total continuum level, but apart from that we consider their contribution negligible. The resulting spectra for detectors one and four were always dominated by many deep and sharp CO2 lines in absorption. De- tector two also showed a few CO2 absorption lines, but their in- tensity declined fast toward longer wavelengths and, in general, the lines were heavily blended with stellar lines from the science spectrum. No deep telluric lines could be seen on detector three. We did not use all of the identified telluric and stellar lines in the following steps. The implemented line identification algo- 5 LBLRTM − part of FASCODE (Clough et al. 1981, 1992), available at http://rtweb.aer.com/ Trifon Trifonov et al.: Precise radial velocities of giant stars rithm effectively excludes lines that are either blended or in close proximity to each other. The selected lines depend on the stellar spectrum shift, which is caused mostly by Earth's barycentric movement. For this reason, a different set of telluric lines was used for wavelength calibration in each observational epoch. In the next step, the unblended observed telluric lines in the science spectrum are interpolated and oversampled with a spline function. Line centers are obtained in pixel coordinates by fitting a Gaussian, Lorentzian, or Pseudo-Voigt (weighted sum between a Gaussian and a Lorentzian) profile via χ2 minimization. The same has been done for the synthetic telluric lines to identify the line centers in wavelength space. Synthetic spectra, by de- fault, are constructed with Voigt profiles, and thus the Voigt pro- file (approximated as a Pseudo-Voigt profile) is the best choice for fitting. Observed telluric lines are also best represented by Voigt profiles. To avoid possible weak line contamination near the line wings, however, we decided to fit the wavelength range spanning from the line centroid to the line full width at half max- imum (FWHM) depth. Many tests showed that at that level the lines are best fit by a Lorentzian profile, and thus we selected the Lorentzian model for telluric fitting instead of the Pseudo-Voigt model. Finally, the wavelength solution is obtained by constructing a third order polynomial χ2 fit between the precise CO2 pixel co- ordinates of observed telluric lines and their wavelength centers from the synthetic spectra. The wavelength solutions obtained in this way are more precise than those from the CRIRES pipeline and can be used for obtaining precise RVs. Our RV precision, however, will always be limited by the individual telluric lines precision from the HITRAN catalog (5−50 m s−1) as well as the telluric variability. The HITRAN catalog precision can in prin- ciple be overcome using many lines so that the random errors average out, but the telluric variability is a systematic error that affects all lines in the same way. We tried to limit the random part using as many lines as possible, but blends between atmo- spheric and stellar lines sometimes did not leave many lines to work with. 4.3. Spectra interpolation and telluric removal Each wavelength solution for each chip and nodding position was interpolated and resampled onto 105 regularly spaced pix- els. Based on this oversampled solution, we interpolated the observed and synthetic spectra from their original, irregularly- spaced wavelength solutions onto the regularly spaced pixel grid. The high resolution LBLRTM telluric spectra was interpolated with a sampling factor L ∼ three times higher than the original, while the observed spectra have L ∼ 100 times the number of the original 1024 pixels. Having both the observed and synthetic telluric spectra on the same wavelength scale is a great advantage. In this way, the telluric lines in both spectra match in wavelength space, which allows us to remove their contribution from each CRIRES spectrum. Synthetic spectra, however, have deeper and narrower lines, and must therefore be smoothed with the CRIRES instru- ment profile (IP) to match the observed line widths and depths. Instead of measuring each telluric line to identify the IP, we applied a more general strategy. By using a least squares algo- rithm, we compared the observed spectra, on the one hand, and the synthetic spectra convolved with a Gaussian kernel, on the other hand. An initial guess of the width of the Gaussian ker- nel is estimated from the spectrograph resolution (∼ 0.016 nm at λ = 1594.5 nm); this is the only free parameter in the fit. With this technique, the IP is obtained quickly and independently for each nod and detector. In our test we assumed that the IP is a sim- ple Gaussian, although in reality the IP is a more complicated sum of several Gaussian profiles coming from different optical parts of the spectrograph (Butler et al. 1996; Bean et al. 2010). Telluric lines may be removed in our pipeline by dividing the observed spectra by the synthetic telluric template convolved with the IP, leaving only the stellar lines. The automated line algorithm was then once again applied to the remaining stellar lines. The laboratory wavelengths of the identified stellar spec- tral lines were taken from the Vienna Atomic Line Database (VALD, Kupka et al. 1999) and obtained based on the target's Teff and log g. Fig. 2 (top panel) illustrates the final output from the spectral division. The figure shows many blended and un- blended stellar lines remaining in the spectrum, but there are also many weak stellar lines for which we do not have any informa- tion from VALD. We believe that dividing the spectra in this way provides us with the best stellar template that can be extracted from the CRIRES spectra. This method gives much better results than simply masking the telluric lines from the spectrum. With this approach, we remove the total atmospheric contribution from the science spectra. 5. Obtaining the radial velocities To derive precise radial velocities, we adopted a method that combines steps from the least squares modeling as in the iodine method (Butler et al. 1996) and the cross-correlation method (Baranne et al. 1996). Indeed, our wavelength calibrators are atmospheric CO2 lines, which are always superimposed on the stellar spectra similar to the gas cell method. As we described in Section 4, we were able to model the telluric lines and eliminate their contribution from the science spectra. The ideal case would be to use the iodine method to model the telluric and the stel- lar spectrum simultaneously, where one of the free parameters would be the Doppler shift. In this kind of approach, the blends between telluric and stellar lines would be less problematic. This approach, however, was initially not possible, because we did not have a proper stellar template spectrum in the near-IR region studied. Also, a simultaneous fit would be more complicated in our case because the telluric lines are variable in contrast to the iodine lines, which are stable. Therefore, despite the fact that the four CRIRES detectors cover relatively small spectral regions, our choice was to use a more simplified approach and we obtain radial velocities by cross-correlating the calibrated spectra with a proper stellar mask. Initially, when we only had a few CRIRES observations for our targets, we obtained radial velocities by cross-correlating the stellar spectra with a weighted binary mask (Pepe et al. 2002). Our mask was consistent with noise-free continuum level val- ues at those wavelengths where we did not identify stellar lines and at a noncontinuum level for the stellar wavelength positions. The mask also had adjustable aperture widths to take the stel- lar line's FWHM into account or to select a pre-defined width, thus making the apertures box-shaped. Even though we achieved relatively good results, we realized that a cross-correlation with weighted binary masks for only a few spectral lines available on each detector might not be optimal. Some near-IR RVs de- viated considerably from the Keplerian model prediction, so it became clear that we might be far from the precision goal of 10−25 m s−1. The main problem was that for some epochs one or even both detectors had only two unblended stellar lines (min- imum for our pipeline), which could be used to construct a cross- correlation function (CCF). If one of the two lines is slightly de- Article number, page 5 of 14 A&A proofs: manuscript no. Trifonov_2015 Fig. 2. Top: spectra for detector one (left) and four (right) are divided by the synthetic telluric spectra (blue), removing the atmospheric contribution, and thus only leaving the stellar lines (black). Many of these spectra were automatically identified in the VALD line catalog. Middle: only the nonblended lines with well-defined profiles are used for obtaining the radial velocities via cross-correlation with a synthetic stellar template modeled from the VALD line catalog. Bottom: the telluric-free spectra from all observational epochs are later shifted and median-combined in one very high S/N stellar template, which is then used for cross-correlation. formed by a nearby line, a delta function or box-like aperture mask also leads to a biased CCF, and consequently to spurious velocity shifts. More lines or an accurate stellar template would mitigate those problems. The line position precision, as calculated by state-of-the-art stellar atmosphere models such as PHOENIX6, is usually worse than that obtained from the high S/N near-IR spectra themselves (Figueira et al. 2010), so we decided to construct an accurate stellar template from our CRIRES observations. The two differ- ent ways of constructing the stellar masks are explained below. 5.1. CCF with noise-free stellar mask To construct a noise-free stellar mask from our observations we chose only single deep and sharp stellar lines (with relative depths below 0.85 of the continuum level, as explained in Sec- tion 4.2) and modeled these using a Gaussian profile. In fact, initially we did the modeling with Lorentzian and Pseudo-Voigt profiles, but we found that the better fit and lack of complex- ity of the Gaussian profile suited us well. From this line mod- eling, we obtained the line FWHM and their individual spectral depths. In the next step, we shifted the identified reference stellar wavelengths by the target's absolute radial velocity7 and Earth's barycentric motion (accuracy better than ∼ 1 m s−1, Roytman 6 http://www.hs.uni-hamburg.de/EN/For/ThA/phoenix/index.html 7 Either taken from SIMBAD (http://simbad.u-strasbg.fr/simbad/), or estimated with our pipeline. Article number, page 6 of 14 2013). With this approach, we took all the necessary line shifts into account, so that the synthetic stellar spectrum resulted in line positions very similar to those on the observed stellar spec- trum. We thus constructed a synthetic noise-free stellar spectrum in the same oversampled wavelength space as the observed stel- lar spectrum. On Fig. 2 (middle panel) are shown the final stellar masks for detectors one and four. To save computing time, the cross- correlation is calculated for a range of about ± 5 km s−1 (500 oversampled pixels) around the stellar line centers. The maxi- mum of the CCF can be obtained down to subpixel level using χ2 minimization; we tried fitting with a Gaussian, a Lorentzian, or a low-order polynomial, which fits a parabola in the area of the maximum of the CCF. Gaussian and Lorentzian models fit the CCF very well, but in contrast to the polynomial model they perform poorly around the CCF peak. The most likely reason for this is that these two models were applied over a wider region of the CCF when compared to the polynomial, which onl fits around the maximum (Allende Prieto 2007). Since we were in- terested in precisely modeling the peak of the CCF, we selected the polynomial model for deriving precise velocities. This method of radial velocity computation was applied to the coadded spectra and to the spectra for nodding positions A and B at detectors one and four separately, resulting in six differ- ent radial velocities for a given epoch. The final radial velocity was obtained by calculating a weighted mean of all radial veloc- ities, where the weight was given by the median S/N of the ex- Trifon Trifonov et al.: Precise radial velocities of giant stars tracted spectrum, which was taken from the FITS header. Using the S/N for weighting corresponds to the approach used by But- ler et al. (1996) for the combination of Doppler information from a large number of short spectral chunks. Usually, the highest S/N ratio was achieved for the coadded spectrum from detector one, and the lowest S/N belonged to one of the nodding positions at detector four. 5.2. CCF with median combined stellar mask As can be seen from Fig. 2, there are many weak stellar lines on detectors one and four. On top of that there are a few very deep stellar lines that according to the VALD line list are actually spectral doublets. These lines are difficult to model and cannot be used for precise RVs in the context defined in Section 5.1. After the telluric removal on all available spectra, we median combine all frames into another stellar template to be used for cross-correlation. This method has several major advantages: 1. Median combination removes the telluric artifacts still present in the individual frames after the division by the the- oretical telluric spectra. 2. All available spectral lines can be used, even unknown lines and unresolved doublet lines. This is not possible by cross- correlating with the noise-free template. 3. Cross-correlation between two identical functions leads to a significant CCF maximum, so that one can very precisely obtain the Doppler shift. This method, however, requires several challenging steps. In- dividual spectra have different wavelength shifts and sampling, and thus need to be shifted carefully before median combina- tion. Resampling is done by interpolating each spectrum to the most precise wavelength solution achieved so far (i.e., most tel- luric references have been used). Each spectrum is then shifted exactly by the difference to the reference spectrum and, finally, a median combination is applied. The result is a high S/N stel- lar mask with the most precise wavelength solution for a given target. This method was only applicable after we had acquired 5−8 epochs for each target. The final median combined stellar tem- plate spectrum for each target on detector one and four is shown in Fig. A1 and Fig. A2. The RVs seem to be more precise from detector one than from detector four, where the lower S/N and the abundant and faint stellar lines influence the CCF. The resulting radial velocity for each epoch is obtained with the same cross-correlation steps and weighted combination of in- dividual frame RVs as explained in Section 5.1 for the noise-free stellar template. The measured near-infrared radial velocities for all stars using this method are listed in Table A1. 5.3. Error estimation The RV uncertainties depend on the number of telluric lines used for the calibration (i.e., the reproducibility of the wavelength so- lution), the lines used for cross-correlation, and the stellar flux from each individual star. We did not assess these systematic er- rors quantitatively nor did we consider the HITRAN and VALD line errors individually. Instead, for each observation we used one combined error, which is based on the RV dispersion from all nodding and coadded positions. The RV error computed in this way varies greatly from epoch to epoch, with a mean value of about ∼ 25 m s−1. The individual error for each observation is listed in Table A1. Based on our results, we estimated the total error of our CRIRES measurements to be not more than 40 m s−1. This es- timation was further confirmed by the total r.m.s. dispersion of the near-IR velocities around the best-fit model obtained for each target from the Lick velocities. 6. IR radial velocity analysis 6.1. Consistency between optical and IR data Eight stars in our sample have published planetary companions, including two two-planet systems. For those ten planets8 the five spectroscopic orbital parameters characterizing a Keplerian orbit are known rather accurately. To test whether the CRIRES radial velocities support the planet hypothesis, we fitted Keplerian or- bits to the CRIRES radial velocities for each star as well. How- ever, since we have far fewer data points from CRIRES than in the optical, sometimes with incomplete phase coverage, we only fit for the RV semiamplitude as well as the RV zero point; we also keep the other four spectroscopic parameters (period, pe- riastron time, eccentricity, and longitude of periastron) fixed at the values derived from the optical radial velocities. We note that our observations with CRIRES were scheduled to sample the RV signal as well as possible in phase over the orbital period. With this approach, we tried to avoid uneven sampling of the data be- cause it can lead to possible zero velocity crossings and a poorly resolved periastron passage (e.g., Cumming 2004), which can lead to highly uncertain amplitude ratios. For the two stars with two planets, the fitting was done one planet at a time, while the RV signal of the other planet as de- termined from the optical data was subtracted. This approach assumes that the RV signal of the subtracted planet is consistent in the optical and the IR; when this is not the case then this in- consistency would show up in the comparison of the RV signals for the other planet in the system. As in the optical, we quadratically added an RV jitter term to the RV measurement errors to account for stellar noise (pul- sations). We did not make any attempt to derive a different value for the jitter term from the CRIRES data, but used the jitter term derived from the optical data. For the eight stars investigated here, the RV jitter term varies from about a third of the value of the IR RV measurement error up to about twice its value. The resulting RV semiamplitudes in the optical (Kopt) and in the IR (KIR) are given in Table 2 together with their errors derived from χ2 fitting. Table 2 also shows the r.m.s. values of the optical and IR data from the best optical and IR models, re- spectively. The last column in the table gives the proportion κ between those two RV amplitudes: κ = KIR/Kopt, along with its error derived following simple error propagation. The RV semi- amplitudes Kopt and KIR are also shown in Fig. 3. The solid line corresponds to κ = 1, i.e., the same RV semiamplitude in both wavelength regimes. In Fig. 4, the resulting Keplerian radial ve- locity curves for the optical and the IR wavelength regime are shown together with the CRIRES measurements. The RV semiamplitude of the outer planet in the HIP 88048 system is not constrained by the CRIRES data, which leaves nine planets in eight systems for our analysis. As one can see, the optical and the IR data are consistent (i.e., κ ≈ 1) at the 1 -- 1.5σ level for all nine investigated planets. In the case of a radial velocity signal, which is at least partly due to stellar noise (starspots, pulsations, etc.), one might ex- pect different RV amplitudes or phases in the optical and in the 8 For simplicity we also use the term planet for deuterium burning mass objects (brown dwarf) as defined in Reffert et al. (2015). Article number, page 7 of 14 Table 2. Comparison of best optical (Kopt) and IR (KIR) RV semiamplitudes and companion between the optical and IR data r.m.s. values from these models. The sixth column gives the proportion between the two semiamplitudes. A&A proofs: manuscript no. Trifonov_2015 HIP 5364 b 5364 c 20889 b 31592 b 34693 b 37826 b 60202 b 88048 b 88048 c 114855 b Kopt [m s−1] 51.1 ± 2.5 52.9 ± 2.6 93.2 ± 2.1 45.2 ± 4.7 350.5 ± 3.4 46.0 ± 1.6 296.7 ± 5.6 288.4 ± 1.2 175.2 ± 1.4 91.0 ± 2.3 r.m.s.opt [m s−1] 16.3 16.3 10.7 7.3 21.1 10.9 30.0 9.0 9.0 18.9 KIR [m s−1] 65.7 ± 14.6 71.5 ± 17.9 89.5 ± 8.3 47.0 ± 10.8 326.2 ± 13.6 70.1 ± 12.6 277.8 ± 26.1 253.4 ± 23.1 120.7 ± 24.3 . . . r.m.s.IR [m s−1] 23.3 26.7 44.1 31.8 39.9 15.4 55.9 20.9 . . . 39.8 Trifonov et al. (2014) Trifonov et al. (2014) Sato et al. (2007) κ = KIR/Kopt Reference 1.29 ± 0.29 1.35 ± 0.35 0.96 ± 0.09 1.04 ± 0.24 Wittenmyer et al. (2011) 0.93 ± 0.04 Mitchell et al. (2013) 1.52 ± 0.27 Reffert et al. (2006) 0.94 ± 0.09 Liu et al. (2008) 0.88 ± 0.08 Quirrenbach et al. (2011) Quirrenbach et al. (2011) 1.33 ± 0.27 Mitchell et al. (2013) . . . infrared since the contrast between the various regions on the stellar surface is wavelength dependent. The agreement between optical and infrared amplitude further supports the planetary hy- pothesis in all eight systems, as expected further tests might be required to undoubtedly confirm the planet interpretation. We will discuss the other systems individually in forthcoming pa- pers. 6.2. IR data phase coverage Given the relatively sparse IR data sampling for our targets as well as the moderately large IR RV errors, we investigated whether this might lead to systematic errors in the fitted RV semiamplitudes. We simulated 10 000 alternative IR data sets for each of the eight stars. We assumed the optical orbit to be the cor- rect data set and we randomly simulated IR data drawn from a normal distribution around the best optical fit, but at the same ob- servational times and with errors as in our original CRIRES mea- surements. For each simulated data set, we fitted the RV semi- amplitude as described in Section 6.1, keeping the rest of the orbital parameters fixed at the optical solution. The resulting dis- tribution of RV semiamplitudes was fitted with a Gaussian, from which we obtained the mean RV semiamplitude and its standard deviation. This test should tell us whether the incomplete phase coverage coupled with the relatively large IR RV errors would lead to any systematics in the recovered RV semiamplitudes. We found that for all targets listed in Table 2 the RV semi- amplitude could be recovered from the simulated data without any systematic offsets. The biggest difference we found between the optical and the simulated RV semiamplitude was ∼ 0.3 m s−1, which is completely insignificant. We conclude that there are no systematic differences in κ due to the sparse sampling and large error of the IR data set. were able to use a maximum number of telluric lines for precise wavelength calibration and almost in any epoch we had enough unblended stellar lines to obtain radial velocities. We achieved similar results for these stars when we cross-correlate the telluric free spectra with the median-combined stellar mask (see §5.2). For G8 III -- K2 III giants we achieved an overall mean RV preci- sion of about 25 m s−1. Also, it is worth mentioning that G8 III -- K2 III giants have in general lower levels of stellar RV jitter than later K giants, and the velocity precision achieved was adequate to test the near-IR velocities for agreement with optical data, and hence, to test for the presence of a planetary companion. Although both methods worked well for G8 III -- K2 III gi- ants, the noise-free mask failed to give reasonable velocities for K3 III -- K4 III giants. The reason for that is the abundance of stellar lines typical for late K stars (see Fig. A1 and Fig. A2). Heavy line blending mutually excluded large number of lines us- able for the wavelength solution and cross-correlation. We find that for many cases in given epochs the RV dispersion between the individual detector nods is too large to derive an adequate RV precision. Unlike the noise-free stellar mask, the median mask is less sensitive to blended stellar lines (since we use all available stellar lines to construct the CCF) and generally gave better re- sults for these stars. Although in some cases the low number of unblended telluric lines was still a problem for constructing a precise wavelength solution, the full set of stellar lines used in this method led to a significant CCF peak from which we could obtain the RVs with a decent precision. The only way to obtain reasonable RVs for K3 III -- K4 III giants was with the median stellar template. Using this method, we achieved mean precision of 20 m s−1, which is even slightly better than that for G8 III -- K2 III. 7.2. Substellar companions 7. Discussion 7.1. RV precision As discussed in Section 5.3, the RV precision achieved with CRIRES spectra depends strongly on the number of telluric and stellar lines used for wavelength calibration and cross- correlation, respectively. In this context we find that giants of spectral type G8 III -- K2 III are the best targets from our CRIRES survey in terms of radial velocity precision, when we cross-correlate with the noise-free stellar mask (see §5.1). These stars have a small number of deep stellar lines, and hence, the problematic line contamination with CO2 lines is minimized. We Lick and CRIRES data for HIP 114855 and HIP 34693 have al- ready been published in Mitchell et al. (2013), while the mul- tiple planetary system around HIP 5364 was extensively dis- cussed in Trifonov et al. (2014). Here we present improved CRIRES velocities for these targets based on cross-correlation with the median-combined mask, superceding earlier results. Results from our near-IR study based on the new CRIRES ve- locities for these stars are given in Table 2 and Fig. 4. All three stars have consistent optical and near-IR semiamplitudes at the 1−1.5σ level. The relatively sparse near-IR sample for HIP 114855 and HIP 34693 has excellent phase coverage show- ing consistency with the best optical Keplerian model. HIP 5364 Article number, page 8 of 14 Trifon Trifonov et al.: Precise radial velocities of giant stars has a rather good near-IR phase coverage with data mostly sam- pled around the two-planet signal extrema, but more near-IR RVs for later epochs would be highly desirable. As was demonstrated in Trifonov et al. (2014), Lick and CRIRES data are more con- sistent with a two-planet dynamical model rather than a simple double Keplerian model, and thus strongly argue for the pres- ence of gravitationally interacting planets. However, for longer time spans the difference between the Keplerian and the dynam- ical model is much larger, and thus more data certainly helps to reveal HIP 5364 system's architecture in more detail. We con- clude that HIP 114855, HIP 34693, and HIP 5364 have secure planets. HIP 60202 has a substellar companion with minimum mass in the brown dwarf regime (mb sin i ≈ 17 MJup), first reported in Liu et al. (2008). More optical velocities from Lick for this star will be published in Mitchell et al. (in prep.). Our CRIRES velocities for this star are consistent with the Keplerian model based on the combined optical velocities from Lick and those published in Liu et al. (2008). We found that compared to the optical model the near-IR data favor a slightly smaller RV semi- amplitude (κ = 0.94 ± 0.09), but both data sets are clearly consis- tent with each other. As can be seen in Fig. 4, the sparse near-IR data set does not cover the time of the predicted RV minimum, which is most likely the reason for the lower κ. We conclude that the IR data support an orbiting companion around HIP 60202. Fig. 3. Comparison of the optical (Kopt) and infrared (KIR) RV semi- amplitudes. The straight line corresponds to Kopt = KIR. Overall we observe a good correspondence between Kopt and KIR; the largest devi- ations are at the 1.5σ level. ∼45 m s−1 and it is on the same order with some of the obtained errors from CRIRES (see Fig. 4). Despite the large errors, how- ever, the near-IR data for HIP 31592 are consistent and are best fitted with an amplitude that is about the same as in the optical (κ = 1.04 ± 0.24). These results imply that the planet announced in Wittenmyer et al. (2011) is very likely real, although this star will benefit from more precise RV data with better phase cover- age. 8. Summary The main goal of our study was to confirm or disprove the plan- etary origin of radial-velocity variations observed in G and K giants. For this purpose, we compared a set of high precision op- tical and near-IR radial velocities and searched for consistency between the two wavelength domains. For our test we selected 20 G and K giants that we exten- sively observed for more than a decade at Lick observatory and that all show periodicity in the optical RV data. For some of the stars orbiting planets have been announced or will be published in the future, and for some there are indications (from the optical data alone) that the RV periodicity might also be caused by stel- lar activity. For all stars however the radial velocity signal is very persistent, so that we were able to fit Keplerian orbits consistent with one or two orbiting substellar companions. We obtained precise near-IR radial velocities with ESO's CRIRES spectrograph for these stars and in this paper we pre- sented our observational setup, data reduction strategy, and re- sults. We selected a spectral region in the H-band, which is known to have abundant atmospheric CO2 lines that we used for wavelength calibration. Only detectors one and four were used for our analysis. To achieve a precise wavelength solution, we adopted the CO2 line centroids from a synthetic telluric spectrum constructed from a line-by-line radiative transfer code (LBLRTM), which uses as an input the HITRAN atmospheric line catalog and the ambient conditions at the time of the observations. By perform- ing a division between the synthetic atmospheric spectrum and Article number, page 9 of 14 The optical Doppler velocities for the K0 III giant HIP 88048 are clearly consistent with two massive companions on noncir- cular orbits. Based on our Lick data, two brown dwarf compan- ions were first discovered by Quirrenbach et al. (2011) and later confirmed by Sato et al. (2012). The Keplerian nature of the opti- cal velocity data was generally accepted, since other phenomena intrinsic to the star are very unlikely to be responsible for the Doppler signal. The period of the outer companion exceeds sev- eral times the time span of the CRIRES observations, and thus at this stage we cannot derive any κ value for the second RV signal. The near-IR data cover only about one full period of the inner planet and clearly follow the optical model, but with poor phase coverage at the predicted RV maximum (see Fig. 4). Our semi- amplitude analysis resulted in κ = 0.88 ± 0.08, which is about 1.5σ from the best optical fit. We also confirm the presence of planetary companions around HIP 20889 (Sato et al. 2007) and HIP 37826 (Reffert et al. 2006; Hatzes et al. 2006). For HIP 20889, the near-IR data cover one full period and the velocity amplitude fully agrees with the amplitude from the optical resulting in κ = 0.96 ± 0.09. For HIP 37826, the near-IR velocity amplitude is larger than the pre- dicted amplitude with κ = 1.52 ± 0.27, but the near-IR value is still within 2σ from the best optical semi-amplitude. The reason for the larger κ is most likely the relatively large near-IR RV er- rors compared to the optical semi-amplitude for HIP 37826. An- other complication might be the phase coverage; there are sev- eral measurements around the minimum and maximum RV, but in between there is a gap. The CRIRES data follow the Keple- rian predictions for these stars; the near-IR RV signal has slightly larger velocity dispersion than the optical data, as expected. The CRIRES data are best suited for testing RV models that assume brown dwarf mass objects in orbit (HIP 34693, HIP 60202 and HIP 88048), since these stars have large RV semiamplitudes and the CRIRES radial velocity uncertainties are still adequate to detect the periodic RV signal. It was more challenging for planetary mass objects, since their RV semi- amplitude was usually only between two and four times of the achieved CRIRES measurement precision. This was the case for HIP 31592, where the optical semiamplitude is on the order of 0100200300400Kopt [m/s]0100200300400KIR [m/s]A&A proofs: manuscript no. Trifonov_2015 Fig. 4. CRIRES near-IR radial velocities for the targets known to have planetary companions. Two best fits to the CRIRES data are overplotted: the model for the best-fit value of κ, where KIR is a free parameter is plotted with a solid black line, while the gray dashed line is for κ = 1, which is the best Keplerian model obtained from the literature data (optical). Bottom panels show the residuals around the best-fit value of κ. The dashed line in the residual panel illustrates the difference between the two models. the science spectrum, we precisely removed the telluric line con- tamination, leaving only the stellar lines. Finally, we derived the RVs by cross-correlating the ob- served spectra with noise-free stellar masks constructed from the VALD theoretical wavelengths and the modeled nonblended stellar line profiles from the observed spectra. This method worked well for G8 III -- K1 III and some K2 III giants, but failed to give reasonable velocities for later K giants. There- fore, once we had enough observational epochs for each target we combined all available target spectra into a master template. We cross-correlated this high S/N stellar template with each ob- served spectrum to extract the maximum Doppler information from all stellar lines in the spectra. This improved strategy gave excellent results for almost all spectral types in our sample, in- cluding the late K giants. The RV precision varies considerably from epoch to epoch and from star to star; the mean RV error is about 25 m s−1. The achieved RV precision was adequate to address our scientific question only for the stars with larger RV amplitudes. Article number, page 10 of 14 HIP 5364 - K2III, 1.7 Msol-150-100-50050100150Vrad [m/s]460048005000520054005600[JD - 2451000.0] -80-60-40-200204060o-c [m/s]HIP 20889 - K0III, 2.4 Msol-150-100-50050100150Vrad [m/s]4600480050005200540056005800[JD - 2451000.0] -100-50050o-c [m/s]HIP 31592 - K1III, 1.4 Msol-100-50050100Vrad [m/s]460048005000520054005600[JD - 2451000.0] -50050100o-c [m/s]HIP 34693 - K2III, 2.3 Msol-400-2000200400Vrad [m/s]460048005000520054005600[JD - 2451000.0] -100-50050100o-c [m/s]HIP 37826 - K0III, 2.3 Msol-100-50050Vrad [m/s]460048005000520054005600[JD - 2451000.0] -60-40-2002040o-c [m/s]HIP 60202 - G8III, 2.2 Msol-400-2000200400Vrad [m/s]48005000520054005600[JD - 2451000.0] -50050100o-c [m/s]HIP 88048 - K0III, 2.7 Msol-400-2000200400600Vrad [m/s]48005000520054005600[JD - 2451000.0] -60-40-200204060o-c [m/s]HIP 114855 - K0III, 1.4 Msol-200-1000100200Vrad [m/s]5000520054005600[JD - 2451000.0] -50050100150o-c [m/s]Trifon Trifonov et al.: Precise radial velocities of giant stars The near-IR radial velocities are in general in excellent agreement with the optical data; we did not find a single sys- tem where the two data sets are inconsistent with each other. For a small number of stars the interpretation is unclear, since we would need more precise data to perform the comparison. For all eight stars that we investigated in detail, the derived near-IR radial velocities agree well with the Lick data. They follow the best Keplerian model predictions and in particular they are con- sistent in amplitude, so that we confirm the planets in the follow- ing systems (cf. Table 2): HIP 5364 (η Cet), HIP 20889 ( Tau), HIP 31592 (7 CMa), HIP 34693 (τ Gem), HIP 37826 (β Gem), HIP 60202 (11 Com), HIP 88048 (ν Oph) and HIP 114855 (91 Aqr). We conclude that the vast majority of the planetary systems known or suspected to exist around the massive evolved stars from our sample is most likely real. Acknowledgements. T.T. was supported by the International Max Plank Re- search School of Astronomy and Astrophysics - Heidelberg (IMPRS-HD) and the Heidelberg Graduate School of Fundamental Physics (HGSFP). M.Z. ac- knowledges support by the European Research Council under the FP7 Starting Grant agreement number 279347. A.R. has received financial support from the DFG under RE 1664/9-2. This research has made use of the SIMBAD data base operated at CDS, Strasbourg, France, the Vienna Atomic Line Database (VALD) at Institut für Astronomie, Wien, Austria, and the HITRAN database at Harvard- Smithsonian Center for Astrophysics (CFA), Cambridge, MA, USA. We thank the anonymous referee for the excellent comments that helped to improve this paper. References Allende Prieto, C. 2007, AJ, 134, 1843 Baranne, A., Queloz, D., Mayor, M., et al. 1996, A&AS, 119, 373 Barban, C., De Ridder, J., Mazumdar, A., et al. 2004, in ESA Special Publication, Vol. 559, SOHO 14 Helio- and Asteroseismology: Towards a Golden Future, ed. D. Danesy, 113 Bean, J. L. & Seifahrt, A. 2009, A&A, 496, 249 Bean, J. L., Seifahrt, A., Hartman, H., et al. 2010, ApJ, 713, 410 Butler, R. P., Marcy, G. W., Williams, E., et al. 1996, PASP, 108, 500 Clough, S. A., Iacono, M. J., & Moncet, J.-L. 1992, J. Geophys. Res., 97, 15761 Clough, S. A., Kneizys, F. X., Rothman, L. S., & Gallery, W. O. 1981, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 277, Atmospheric transmission, 152 -- 166 Cumming, A. 2004, MNRAS, 354, 1165 Desort, M., Lagrange, A.-M., Galland, F., Udry, S., & Mayor, M. 2007, A&A, Dorn, R. J., Anglada-Escude, G., Baade, D., et al. 2014, The Messenger, 156, 7 Figueira, P., Pepe, F., Melo, C. H. F., et al. 2010, A&A, 511, A55 Frink, S., Mitchell, D. S., Quirrenbach, A., et al. 2002, ApJ, 576, 478 Frink, S., Quirrenbach, A., Fischer, D., Röser, S., & Schilbach, E. 2001, PASP, 473, 983 113, 173 Hatzes, A. P. 2002, Astronomische Nachrichten, 323, 392 Hatzes, A. P. & Cochran, W. D. 2000, AJ, 120, 979 Hatzes, A. P., Cochran, W. D., Endl, M., et al. 2006, A&A, 457, 335 Hinkle, K., Wallace, L., & Livingston, W. 1995, PASP, 107, 1042 Huélamo, N., Figueira, P., Bonfils, X., et al. 2008, A&A, 489, L9 Kaeufl, H.-U., Ballester, P., Biereichel, P., et al. 2004, in Society of Photo- Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 5492, So- ciety of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. A. F. M. Moorwood & M. Iye, 1218 -- 1227 Kupka, F., Piskunov, N., Ryabchikova, T. A., Stempels, H. C., & Weiss, W. W. 1999, A&AS, 138, 119 Lebzelter, T., Seifahrt, A., Uttenthaler, S., et al. 2012, A&A, 539, A109 Liu, Y.-J., Sato, B., Zhao, G., et al. 2008, ApJ, 672, 553 Marcy, G. W. & Butler, R. P. 1992, PASP, 104, 270 Mayor, M., Pepe, F., Queloz, D., et al. 2003, The Messenger, 114, 20 Mitchell, D. S., Reffert, S., Trifonov, T., Quirrenbach, A., & Fischer, D. A. 2013, A&A, 555, A87 Moutou, C., Mayor, M., Lo Curto, G., et al. 2011, A&A, 527, A63 Pepe, F., Mayor, M., Galland, F., et al. 2002, A&A, 388, 632 Quirrenbach, A., Reffert, S., & Bergmann, C. 2011, in American Institute of Physics Conference Series, Vol. 1331, American Institute of Physics Confer- ence Series, ed. S. Schuh, H. Drechsel, & U. Heber, 102 -- 109 Reffert, S., Bergmann, C., Quirrenbach, A., Trifonov, T., & Künstler, A. 2015, A&A, 574, A116 Reffert, S., Quirrenbach, A., Mitchell, D. S., et al. 2006, ApJ, 652, 661 Reiners, A., Bean, J. L., Huber, K. F., et al. 2010, ApJ, 710, 432 Rothman, L. S., Rinsland, C. P., Goldman, A., et al. 1998, J. Quant. Spec. Ra- diat. Transf., 60, 665 Roytman, N. 2013, Doppler Analysis of K Giant Echelle Spetra from the Lick Exoplanet Search, Diplomarbeit, Landessternwarte, Heidelberg University, Germany Sato, B., Izumiura, H., Toyota, E., et al. 2007, ApJ, 661, 527 Sato, B., Omiya, M., Harakawa, H., et al. 2012, PASJ, 64, 135 Sato, B., Omiya, M., Harakawa, H., et al. 2013, PASJ, 65, 85 Seifahrt, A. & Käufl, H. U. 2008a, A&A, 491, 929 Seifahrt, A. & Käufl, H. U. 2008b, A&A, 491, 929 Seifahrt, A., Käufl, H. U., Zängl, G., et al. 2010, A&A, 524, A11 Setiawan, J., Henning, T., Launhardt, R., et al. 2008, Nature, 451, 38 Trifonov, T., Reffert, S., Tan, X., Lee, M. H., & Quirrenbach, A. 2014, A&A, Wittenmyer, R. A., Endl, M., Wang, L., et al. 2011, ApJ, 743, 184 Zechmeister, M., Reffert, S., Hatzes, A. P., Endl, M., & Quirrenbach, A. 2008, 568, A64 A&A, 491, 531 Article number, page 11 of 14 Table A1. Radial velocities measured with CRIRES A&A proofs: manuscript no. Trifonov_2015 Target HIP 5364 HIP 19011 HIP 20889 HIP 23015 HIP 31592 HIP 34693 HIP 36616 epoch [JD] 2455853.844 2455854.616 2456113.863 2456121.811 2456139.747 2456239.563 2456441.910 2456469.917 2456494.892 2455854.736 2455876.638 2456141.856 2456147.907 2456244.602 2456287.561 2455854.744 2455900.760 2456147.921 2456168.832 2456244.649 2456288.569 2456495.931 2456552.779 2455854.777 2455900.755 2456168.898 2456169.900 2456244.738 2456288.574 2455854.862 2455876.701 2455918.706 2456021.552 2456039.484 2456244.653 2456288.579 2456326.544 2456406.497 2455859.855 2455895.863 2455918.699 2455994.537 2456039.510 2456244.866 2456288.696 2456326.553 2455859.849 2455896.842 2455918.683 2455994.545 2456021.576 RV [km s−1] σ [m s−1] Target 11.623 11.588 11.697 11.717 11.662 11.623 11.774 11.801 11.800 26.864 26.854 27.007 27.020 26.923 26.917 38.407 38.392 38.620 38.604 38.618 38.571 38.438 38.499 16.997 16.728 17.143 17.222 16.933 16.320 2.606 2.564 2.575 2.576 2.601 2.652 2.630 2.702 2.695 22.027 22.222 22.270 21.845 21.592 22.114 21.846 21.644 −35.057 −34.848 −34.780 −34.800 −34.910 19 40 20 22 12 41 33 26 26 6 8 5 7 12 23 14 24 15 14 27 20 13 9 9 7 17 31 16 17 29 18 53 19 22 59 18 22 24 11 23 6 14 25 13 14 12 17 72 33 18 49 HIP 37826 HIP 38253 HIP 39177 HIP 60202 HIP 73133 HIP 74732 HIP 79540 epoch [JD] 2456288.592 2456326.559 2456406.504 2455866.855 2455896.833 2455918.718 2455994.554 2456021.584 2456288.709 2456326.567 2456406.514 2455859.837 2455896.774 2455918.723 2455973.736 2456021.597 2456244.742 2456288.586 2456326.571 2455859.843 2455896.768 2455918.731 2455973.728 2456021.589 2456288.701 2456326.578 2455932.826 2455973.744 2456024.594 2456067.482 2456293.849 2456407.498 2456458.561 2455937.846 2455994.777 2456022.713 2456067.488 2456406.698 2456458.572 2455939.840 2455994.784 2456024.647 2456067.495 2456406.729 2456453.796 2456494.649 2456015.780 2456105.644 2456113.594 2456121.644 2456406.706 RV [km s−1] σ [m s−1] −34.193 −34.333 −34.328 3.706 3.728 3.712 3.759 3.762 3.623 3.647 3.651 −3.777 −3.822 −3.840 −4.039 −3.936 −4.311 −3.759 −3.851 39.473 39.511 39.502 39.390 39.500 39.254 39.264 43.196 43.302 43.431 43.650 43.255 43.586 43.356 17.637 17.749 17.744 17.808 18.099 18.196 −9.514 −9.493 −9.526 −9.601 −9.466 −9.350 −9.323 −23.336 −23.399 −23.406 −23.459 −23.391 29 55 47 17 15 50 19 31 20 15 32 9 20 24 17 21 12 18 8 13 5 7 9 13 8 4 10 10 6 37 13 15 37 6 12 15 11 34 13 10 13 6 9 10 13 18 14 22 18 25 22 Target HIP 80693 HIP 84671 HIP 88048 HIP 91004 HIP 100587 HIP 114855 epoch [JD] 2456453.802 2456496.664 2456503.652 2455990.888 2456015.773 2456025.680 2456025.676 2456121.559 2456406.714 2456459.675 2456511.625 2455994.793 2456018.858 2456105.639 2456121.670 2456148.674 2455997.887 2456018.852 2456105.635 2456113.599 2456121.780 2456148.679 2456453.870 2456496.673 2456497.759 2456497.767 2455855.571 2455997.891 2456114.775 2456119.832 2456119.846 2456141.792 2456239.543 2455860.541 2456113.859 2456119.838 2456144.750 2456405.903 2456440.905 2456466.747 2455854.611 2456112.928 2456119.880 2456144.774 2456239.555 2456453.898 2456476.910 2456511.637 RV [km s−1] σ [m s−1] −23.342 −23.374 −23.376 7.561 7.956 7.820 7.832 7.813 7.826 7.956 8.063 38.930 39.056 39.164 39.315 39.137 12.901 12.912 13.048 13.068 13.093 13.206 12.966 12.882 12.925 12.920 −3.084 −3.574 −3.654 −3.580 −3.538 −3.416 −2.432 −15.468 −15.268 −15.278 −15.207 −14.705 −14.680 −14.607 −25.721 −25.563 −25.576 −25.691 −25.773 −25.617 −25.545 −25.691 15 17 9 13 11 21 24 50 13 19 9 8 10 50 145 21 23 6 49 28 29 26 11 20 16 19 28 9 3 41 23 25 4 4 6 11 50 11 20 37 43 17 43 26 33 86 73 55 Article number, page 12 of 14 Trifon Trifonov et al.: Precise radial velocities of giant stars Fig. A1. Median-combined spectra from detector 1 for all observed targets, shifted by +0.5 in ascending order starting from K4 III giants at the bottom to earlier spectral type stars toward the top. Telluric lines were removed. Obviously late G and early K giants have a smaller number of deep stellar lines in this wavelength region than later K giants. Article number, page 13 of 14 A&A proofs: manuscript no. Trifonov_2015 Fig. A2. Same as Fig. A1, except for detector 4. Detector 4 has in general lower S/N when compared to detector 1. Article number, page 14 of 14
1602.01851	1	1602	2016-02-04T21:05:18	The K2-ESPRINT Project II: Spectroscopic follow-up of three exoplanet systems from Campaign 1 of K2	[ "astro-ph.EP", "astro-ph.SR" ]	We report on Doppler observations of three transiting planet candidates that were detected during Campaign 1 of the K2 mission. The Doppler observations were conducted with FIES, HARPS-N and HARPS. We measure the mass of K2-27b (EPIC 201546283b), and provide constraints and upper limits for EPIC 201295312b and EPIC 201577035b. K2-27b is a warm Neptune orbiting its host star in 6.77 days and has a radius of $4.45^{+0.33}_{-0.33}~\mathrm{R_\oplus}$ and a mass of $29.1^{+7.5}_{-7.4}~\mathrm{M_\oplus}$, which leads to a mean density of $1.80^{+0.70}_{-0.55}~\mathrm{g~cm^{-3}}$. EPIC 201295312b is smaller than Neptune with an orbital period of 5.66 days, radius $2.75^{+0.24}_{-0.22}~\mathrm{R_\oplus}$ and we constrain the mass to be below $12~\mathrm{M_\oplus}$ at 95% confidence. We also find a long-term trend indicative of another body in the system. EPIC 201577035b, previously confirmed as the planet K2-10b, is smaller than Neptune orbiting its host star in 19.3 days, with radius $3.84^{+0.35}_{-0.34}~\mathrm{R_\oplus}$. We determine its mass to be $27^{+17}_{-16}~\mathrm{M_\oplus}$, with a 95% confidence uppler limit at $57~\mathrm{M_\oplus}$, and mean density $2.6^{+2.1}_{-1.6}~{\rm g~cm}^{-3}$. These measurements join the relatively small collection of planets smaller than Neptune with measurements or constraints of the mean density. Our code for performing K2 photometry and detecting planetary transits is now publicly available.	astro-ph.EP	astro-ph	Draft version September 24, 2018 Preprint typeset using LATEX style emulateapj v. 08/22/09 THE K2-ESPRINT PROJECT II: SPECTROSCOPIC FOLLOW-UP OF THREE EXOPLANET SYSTEMS FROM CAMPAIGN 1 OF K2(cid:63) Vincent Van Eylen1,2, Grzegorz Nowak3,4, Simon Albrecht1, Enric Palle3,4, Ignasi Ribas5, Hans Bruntt1, Manuel Perger5, Davide Gandolfi6,7, Teriyuki Hirano8, Roberto Sanchis-Ojeda9,10, Amanda Kiilerich1, Jorge P. Arranz3,4, Mariona Badenas11, Fei Dai2, Hans J. Deeg3,4, Eike W. Guenther12, Pilar Montan´es-Rodr´ıguez3,4, Norio Narita13,14,15, Leslie A. Rogers16, V´ıctor J. S. B´ejar3,4, Tushar S. 1 Stellar Astrophysics Centre, Department of Physics and Astronomy, Aarhus University, Ny Munkegade 120, Shrotriya1, Joshua N. Winn2, Daniel Sebastian12 DK-8000 Aarhus C, Denmark 2 Department of Physics, and Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 3 Instituto de Astrof´ısica de Canarias (IAC), 38205 La Laguna, Tenerife, Spain 4 Departamento de Astrof´ısica, Universidad de La Laguna (ULL), 38206 La Laguna, Tenerife, Spain 5 Institut de Ci´encies de l'Espai (CSIC-IEEC), Carrer de Can Magrans, Campus UAB, 08193 Bellaterra, Spain 6 Dipartimento di Fisica, Universit´a di Torino, via P. Giuria 1, I-10125, Torino, Italy 7 Landessternwarte Konigstuhl, Zentrum fur Astronomie der Universitat Heidelberg, Konigstuhl 12, D-69117 Heidelberg, Germany 8 Department of Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan 9 Department of Astronomy, University of California, Berkeley, CA 94720 10 NASA Sagan Fellow 11 Department of Astronomy, Yale University, New Haven, CT 06511, USA 12 Thuringer Landessternwarte Tautenburg, 07778 Tautenburg, Germany 13 National Astronomical Observatory of Japan, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan 14 SOKENDAI (The Graduate University for Advanced Studies), 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan 15 Astrobiology Center, National Institutes of Natural Sciences, 2-21-1 Osawa, Mitaka, Tokyo 181-8588, Japan and 16 Department of Astronomy and Division of Geological and Planetary Sciences, California Institute of Technology, MC249-17, 1200 East California Boulevard, Pasadena, CA 91125, USA (Received receipt date; Revised revision date; Accepted acceptance date) Draft version September 24, 2018 ABSTRACT We report on Doppler observations of three transiting planet candidates that were detected during Campaign 1 of the K2 mission. The Doppler observations were conducted with FIES, HARPS-N and HARPS. We measure the mass of K2-27b (EPIC 201546283b), and provide constraints and upper limits for EPIC 201295312b and EPIC 201577035b. K2-27b is a warm Neptune orbiting its host star in 6.77 days and has a radius of 4.45+0.33−0.33 R⊕ and a mass of 29.1+7.5−7.4 M⊕, which leads to a mean density of 1.80+0.70−0.55 g cm−3. EPIC 201295312b is smaller than Neptune with an orbital period of 5.66 days, radius 2.75+0.24−0.22 R⊕ and we constrain the mass to be below 12 M⊕ at 95% confidence. We also find a long-term trend indicative of another body in the system. EPIC 201577035b, previously confirmed as the planet K2-10b, is smaller than Neptune orbiting its host star in 19.3 days, with radius 3.84+0.35−0.34 R⊕. We determine its mass to be 27+17−16 M⊕, with a 95% confidence uppler limit at 57 M⊕, and mean density 2.6+2.1−1.6 g cm−3. These measurements join the relatively small collection of planets smaller than Neptune with measurements or constraints of the mean density. Our code for performing K2 photometry and detecting planetary transits is now publicly available. Subject headings: planetary systems -- stars: fundamental parameters -- stars: individual (EPIC 201295312, EPIC 201546283, EPIC 201577035, K2-10, K2-27) 1. INTRODUCTION Although data from the K2 mission (Howell et al. 2014) has only been available for six months, it has already led to several notable exoplanet discoveries. For example, a sub-Neptune orbiting a bright star (using only the 9 days of Engineering Test Data Vanderburg et al. 2015), three super-Earths orbiting a bright M dwarf star (Crossfield et al. 2015), a disintegrating rocky planet with a cometary head and tail (Sanchis-Ojeda et al. 2015), two super-Earth planets orbiting a nearby cool star (Petigura et al. 2015) and two additional planets orbiting the known hot Jupiter host star WASP-47 (Becker et al. 2015). Based on the Campaign 1 photometry, the first lists of plane- tary candidates have been produced (Foreman-Mackey et al. Electronic address: [email protected] (cid:63) Based on observations made with the NOT telescope under programme ID. 50-022/51-503 and 50-213(CAT), the TNG telescope under programme ID. AOT30.13, OPT15A 33, and CAT14B 121 and ESOs 3.6 m telescope at the La Silla Paranal Observatory under programme ID 095.C-0718(A). 2015; Montet et al. 2015). As part of the Equipo de Seguimiento de Planetas Rocosos INterpretando sus Tr´ansitos (ESPRINT) project (see Sanchis- Ojeda et al. 2015), we present our radial velocity follow- up measurements of three Campaign 1 planet candidates (EPIC 201295312, EPIC 201546283, and EPIC 201577035), making use of the FIES (Telting et al. 2014), HARPS-N (Cosentino et al. 2012) and HARPS (Mayor et al. 2003) spec- trographs. Measurements of masses for planets smaller than 4 − 5 R⊕ are notoriously difficult (Marcy et al. 2014), but are of importance to constrain interior models for sub-Neptune planets (e.g. Rogers 2015). In the next section we present our analysis pipeline for K2 photometry, including aperture photometry, light curve de- trending and planet search algorithms (the Python code used for the analysis is publicly available on GitHub2). We de- scribe the planet characterization via spectroscopic observa- 2 https://github.com/vincentvaneylen 2 Van Eylen et al. tions in Section 3, and discuss the results in Section 5. 2. PHOTOMETRY Unlike for the primary Kepler mission, K2 photometry is primarily delivered in the form of pixel files without mission- defined aperture masks, and the task of finding planet can- didates rests upon the community rather than upon the mis- sion team. Because the mission operates with only two func- tioning reaction wheels (Howell et al. 2014) the pointing sta- bility is more limited, which affects the photometric preci- sion. Correction methods make use of the center-of-light off- set (Vanderburg & Johnson 2014; Lund et al. 2015) or use trends which are common to many stars (Foreman-Mackey et al. 2015; Angus et al. 2015). We have developed a photometry pipeline consisting of the following independent modules: (1) Extract aperture photom- etry; (2) Perform light curve detrending; and (3) Search for transits and perform time-domain transit modeling. We de- scribe these steps in the next sections. Our analysis starts from K2 pixel mask files which can be downloaded from the MAST archive3. We perform simple aperture photometry on these pixel masks. First we sum the flux per pixel over the full time series of the K2 campaign. Subsequently, the median flux over the different pixels is cal- culated. As long as the field is not too crowded the median flux is a fairly good estimate of the background flux. The stellar flux can then be identified as the flux that exceeds the background flux by some predetermined threshold (typically 1.05×median). We include the pixels exceeding the threshold and group them according to whether they are spatially adja- cent. If two or more spatially disjoint groups are detected, the largest pixel group is selected (the other, smaller groups are assumed to be caused by other stars and are ignored). The results are shown in Figure 1 for the three stars discussed in this work. Once the aperture is selected in this way, for each time step the total flux is calculated by summing the flux of all pixels in the aperture, and the flux centroid position is cal- culated based on the flux-weighted mean X and Y coordinates of the group of pixels. We also subtract the background flux, which is estimated as the median of the pixels after iteratively clipping all 3σ outliers. For the light curve detrending, we use a linear fit to the cen- troid positions, a technique first used successfully for Spitzer transit observations (e.g. D´esert et al. 2009; Van Eylen et al. 2014; D´esert et al. 2015), and which is similar to techniques developed for K2 photometry by others (Vanderburg & John- son 2014; Lund et al. 2015; Sanchis-Ojeda et al. 2015). In summary, the light curve is divided into chunks of specified length (typically 300 data points), and a polynomial function of centroid position and time is fitted to the flux in each chunk. More precisely, for centroid coordinates Xc and Yc (calculated relative to the mean centroid position), time T, and flux F, we fit the model M: M = t0 + t1T + t2T 2 + t3T 3 + x1Xc + x2X2 +y1Yc + y2Y2 c + y3Y3 c + x3X3 c c + z1XcYc, where ti, xi, yi and z1 are fitting parameters. For each chunk, the light curve flux is then divided by the model to remove variability caused by spacecraft pointing variations (which cause flux variations due to different pixel sensitivi- ties) as well as long-term astrophysical variations. We have 3 See https://archive.stsci.edu/k2/data search/search.php compared this technique with the ones employed by Sanchis- Ojeda et al. (2015) and Vanderburg & Johnson (2014), and found the light curve quality and the transit parameters to be similar. To search for transit events we subsequently run a "box least square" search on the light curves (Kov´acs et al. 2002), which detects periodic transit-like events4. These are then visually inspected in order to check if they are indeed transit-shaped. Based on an initial analysis of the light curves and ground based imaging, interesting planets were selected for spectro- scopic follow-up. Bright planet candidates were observed using the FastCam (FC) lucky imaging camera at the 1.5- m Carlos Sanchez Telescope (TCS) in Tenerife. All images were bias subtracted and then shifted and co-added using FC specific software to produce a final, high SNR, high resolu- tion image. Objects that appeared to be isolated, were then moved forward in the confirmation process to be observed with FIES. We obtained observations (45-60 minute expo- sures) with the FIES spectrograph for a first detailed stel- lar characterization and a small number of Radial Velocities (RVs). Systems which show RV scatter less than 20 m s−1 are selected for further observations. For Campaign 1, these ef- forts were focused on EPIC 201295312, EPIC 201546283, and EPIC 201577035. 3. GROUND BASED FOLLOW-UP OBSERVATIONS These systems were recently discussed by Montet et al. (2015). EPIC 201577035 was validated as a gen- uine planet (also called K2-10b in the NASA Exoplanet Archive5), but high-resolution Adaptive Optics images con- ducted by these authors revealed faint stellar companions for EPIC 201295312 and EPIC 201546283, at distances of 3 and 8 arcsec respectively. This complicates the planet validation because K2 apertures span many pixels (see Figure 1; each pixel measures 3.98 × 3.98 arcsec). Therefore it can be diffi- cult to be certain that the target star is truly the host of the tran- siting planet candidate. Consequentially, Montet et al. (2015) were unable to validate the planetary candidates (despite as- signing false positive probabilities below 10−4 in both cases). We measured RVs using the standard data reduction pipelines for the HARPS and HARPS-N spectrographs. For the case of FIES we used the approach described in Gandolfi et al. (2015). To derive stellar parameters, to measure stellar reflex mo- tion and finally to determine the planetary mass we observed these three systems throughout Spring 2015 with the HARPS- N spectrograph mounted on the TNG on La Palma and the HARPS spectrograph on ESO's 3.6m telescope at La Silla. The exposure times varied between 15 min and 45 min for the different systems and instruments, and we used standard se- tups. All RVs for the three systems are available in electronic form from the ApJ webpage. 4. PARAMETER ESTIMATION 4.1. Estimation of stellar parameters Before modeling the RVs we co-added the available spectra for each system to derive the stellar atmospheric parameters using the VWA software6 developed by Bruntt et al. (2012). 4 We used an implementation of this algorithm in Python by Ruth Angus and Dan Foreman-Mackey; see https://github.com/dfm/python-bls. 5 http://exoplanetarchive.ipac.caltech.edu/ 6 https://sites.google.com/site/vikingpowersoftware/home Spectroscopic follow-up of K2 Campaign 1 3 Fig. 1. -- Pixel masks for EPIC 201295312 (left), EPIC 201546283 (middle) and EPIC 201577035 (right). The colors indicate the electron count, going from red (high) to low (blue). Pixel masks above a threshold electron count are encircled. We use red for those pixels included in the light curve and green for those assumed to be caused by other stars. Fig. 2. -- K2 reduced photometry folded by the best period for EPIC 201295312 (left), EPIC 201546283 (middle) and EPIC 201577035 (right). The best fitted model is shown with a solid line as well as the residuals after subtracting the model. This approach includes systematic errors in the uncertainty es- timate. For the signal-to-noise ratio in our combined spectra, the uncertainty in the stellar parameters is dominated by this systematic noise floor rather than by photon noise, so that the uncertainties in parameters for different stars are sometimes very similar. With obtained values of the effective stellar tem- perature (Teff), the stellar surface gravity (log g), and metal- licity ([Fe/H]) as input, we then used BaSTI evolution tracks7 to infer the stellar mass, radius, and obtain an age estimate. Here we used the SHOTGUN method (Stello et al. 2009). In parallel we also obtained spectra with the High Disper- sion spectrograph (HDS) mounted to the Subaru telescope, one spectrum for each system. These spectra have been analyzed following Takeda et al. (2002) (see also Hirano et al. 2012). There is agreement between the set of param- eters obtained with the two different data sets and methods, with one important exception, i.e. the stellar radius of EPIC 201295312 (the HDS radius is 1.91±0.11 R(cid:12), the VWA radius is 1.52 ± 0.10 R(cid:12)). This is an important parameter, because any uncertainty or error in this parameter translates directly into the planetary radius and the planetary density. However, we have not been able to track down the cause for this dis- agreement. Here we just note that the two methods agree well on all other stellar parameters and for the other systems and that for evolved stars (such as EPIC 201295312), different isochrones can lay close to each other, complicating the stel- lar characterization. Because the HDS spectrum has a lower signal-to-noise ratio than the combined HARPS-N spectrum, we adopt the VWA values in the analysis. 4.2. Estimation of orbital and planetary parameters 7 http://albione.oa-teramo.inaf.it/ Photometric model -- We modeled the K2 light curves to- gether with the RVs obtained from the FIES, HARPS-N, and HARPS spectrographs. We use the transit model by Mandel & Agol (2002) and a simple Keplerian RV model. The transit model used was binned to 30 minutes, to match the integra- tion time of the Kepler observations. The model parameters mainly constrained by photometry (see Figure 2) include the orbital Period (P), a particular time of mid-transit (Tmin), the stellar radius in units of the orbital semi-major axis (R(cid:63)/a), the planetary radius in units of the stellar radius (Rp/R(cid:63)), and the cosine of the orbital inclination (cos io). We further as- sumed a quadratic limb-darkening law (with two parameters, u1 and u2). RV model -- The Keplerian RV model introduces additional parameters, i.e. the semi-amplitude of the projected stellar re- flex motion (K(cid:63)), systemic velocities for each spectrograph (γspec), the orbital eccentricity (e), and the argument of peri- e sin ω to astron (ω). For our analysis, we use avoid boundary issues (see e.g. Lucy & Sweeney 1971). For one system (EPIC 201295312) we found evidence for a long term drift which we model with a second order polynomial function of time. e cos ω and √ √ Prior information -- For all parameters in all three systems we use a flat prior if not mentioned otherwise in this para- graph. We use priors on the quadratic limb darkening param- eters u1 and u2, selected from the tables provided by Claret et al. (2013) appropriate for the Kepler bandpass. In particu- lar we placed a Gaussian prior on u1 + u2 with a width of 0.1. The difference u1 − u2 was held fixed at the tabulated value, since this combination is weakly constrained by the data. The stellar density obtained from the analysis of the stellar spectra 051015X051015Y051015X051015Y051015X051015Y−0.02940.02940.99900.99951.00001.0005relative flux−0.02940.0294−0.00050.00000.0005O − C−202time (hr)−0.02460.02460.9970.9980.9991.000relative flux−0.02460.0246−0.0010−0.00050.00000.00050.0010O − C−202time (hr)−0.01080.01080.99850.99900.99951.00001.0005relative flux−0.01080.0108−0.00050.00000.0005O − C−4−2024time (hr)4 Van Eylen et al. de ∝ 1 (Section 4.1) is used as a Gaussian prior in our fit to the pho- tometric and RV data. This impacts the confidence intervals we obtain for e and ω (e.g. Van Eylen & Albrecht 2015). We further assume an eccentricity prior dN 24 as ob- tained by Shen & Turner (2008), and require that the planet and star do not have crossing orbits. (1+e)4 − e For EPIC 201295312 we found indications of a long term drift, using the FIES and HARPS-N data sets. Unfortu- nately the HARPS data points have been taken after the FIES and HARPS-N campaigns were finished. This compli- cates the characterization of the long term trend, as poten- tial offsets in the RV zero points of the spectrographs could lead to biases. However, previous studies, such as L´opez- Morales et al. (2014) (55 Cnc) and Desidera et al. (2013) (HIP 11952) found that the RVs of HARPS and HARPS-N agree within their uncertainties. In this study, we find the same for EPIC 201546283 (see below). Therefore, we pro- ceed and impose a Gaussian prior with zero mean and a σ of 5 m s−1 on the difference in the systemic velocity of these two spectrographs for EPIC 201295312. We also assumed the pe- riod to be constant in all three systems as we could detect no sign of significant Transit Timing Variations. Parameter estimation -- To estimate the uncertainty intervals for the parameters we use a Markov Chain Monte Carlo (MCMC) approach (see, e.g., Tegmark et al. 2004). Be- fore starting the MCMC chain we added "stellar jitter" to the internally estimated uncertainties for FIES, HARPS-N, and HARPS observations, so that the minimum reduced χ2 for each dataset alone is close to unity. For each system we run three simple chains with 106 steps each, using the Metropolis-Hastings sampling algorithm. The step size was adjusted to obtain a success rate of ≈ 0.25. We removed the first 104 points from each chain and checked for convergence via visual inspection of trace plots and em- ploying the Gelman and Rubin Diagnostic (Gelman & Rubin 1992). Here we find that for all parameters for all systems R < 1.01. The uncertainty intervals presented below have been obtained from the merged chains. In Table 1 we report the results derived from the posterior, quoting uncertainties excluding 15.85 % of all values at both extremes, encompass- ing 68.3 % of the total probability. The key result is K(cid:63), which together with the orbital period, inclination, and stellar mass determines planetary mass and bulk density. 5. RESULTS 5.1. EPIC 201295312 EPIC 201295312b has an orbital period of 5.66 days. We do not clearly detect the RV amplitude caused by the planet, but do find a longer-period trend. Using the procedure described in Section 4.2, we find the long term drift to be adequately described by a second order polynomial, with a linear term of 3.3 cm s−1 d−1 and a quadratic term of 1.83 cm s−1 d−1, while using 2457099.654 HJD as the constant time of reference. We speculate that this trend, which is shown in Figure 3, could be caused by the gravitational influence of an additional body in the system, such that additional monitoring might re- veal the orbital period and mass function of this presumed companion. Our currently data constrains it poorly, but as one example of what may be causing it, we find that a circu- lar orbit with a period of 365 days leads to a good fit with a K(cid:63) amplitude of 155 m s−1. Assuming an inclination of 90 degrees, this implies a mass of 5.9 MJup. However, we cau- tion against overinterpreting these values, as only a small part of such an orbit is covered with the current data, and longer orbital periods cannot be excluded. We furthermore tested if the bisector measurements give any indication that the observed RV trend is caused by a stel- lar companion. The results are inconclusive: the HARPS and HARPS-N CCFs appear to show a small difference, but it is possible this is caused by the atmospheric conditions under which the stars were observed, or by small differences be- tween the two instruments. The data are shown in Figure 4. Fig. 3. -- RV observations of EPIC 201295312 as function of time. The best fitted model using a quadratic long term trend is shown along with the data. Assuming good agreement between the velocity offsets of HARPS and HARPS-N (see also Section 4.2) the data does require a quadratic term to be adequately fitted. Fig. 4. -- BIS from HARPS and HARPS-N CCFs for EPIC 2012995312 plotted versus the stellar RVs. The color code indicates the signal-to-noise ratio in the stellar spectra obtained around a wavelength range of 5560 Å. The BIS values for the low RV points appear slightly shifted from the RV points with higher values, which would indicate the presence of a self-luminous companion, but because those data are taken with different telescopes the results are inconclusive. The uncertainty in the bisector values is taken to be twice the uncertainty in the RV values. We place an upper limit on the planetary mass of 12 M⊕, −150−100−500radial velocity (m s−1)HARPS−NHARPSFIESFIES−30−20−100102030O − C (m s−1)050100150200Time of observation [BJD − 2457024]−20020406080100BIS (m s−1)−100−50050RV − mean RV (m s−1)203040SNRSpectroscopic follow-up of K2 Campaign 1 5 TABLE 1 System parameters Parameter 2MASS ID Right Ascension Declination Magnitude (Kepler) Basic properties EPIC 201295312 EPIC 201546283 EPIC 201577035 11360278-0231150 11260363+0113505 11282927+0141264 11 36 02.790 -02 31 15.17 12.13 11 26 03.638 +01 13 50.66 12.43 11 28 29.269 +01 41 26.29 12.30 Stellar parameters from spectroscopy Effective Temperature, Teff (K) Surface gravity, log g (cgs) Metallicity, [Fe/H] Microturbulence (km s−1) Projected rotation speed, v sin i(cid:63) (km s−1) Assumed Macroturbulence, ζ (km s−1) Stellar Mass, Mp (M(cid:12)) Stellar Radius, Rp (R(cid:12)) Stellar Density, ρ(cid:63) (g cm−3)a Limb darkening prior u1 + u2 Stellar jitter term HARPS (m s−1) Stellar jitter term HARPS-N (m s−1) Stellar jitter term FIES (m s−1) 5830±70 4.04±0.08 0.13±0.07 1.2±0.07 5±1 1.13±0.07 1.52±0.10 0.45±0.14 0.6752±0.1 10.5 6.5 20 5320±70 4.60±0.08 0.14±0.07 0.8±0.07 1±1 2 0.89±0.05 0.85±0.06 2.04±0.38 5620±70 4.50±0.08 −0.07±0.07 0.9±0.07 3±1 0.92±0.05 0.98±0.08 1.38±0.34 0.7009±0.1 0.6876±0.1 6 6 30 -- 7 5 Fitting (prior) parameters Adjusted Parameters from RV and transit fit Orbital Period, P (days) Time of mid-transit, Tmin (BJD−2450000) Orbital Eccentricity, e Cosine orbital inclination, cos io Scaled Stellar Radius, R(cid:63)/a Fractional Planetary Radius, Rp/R(cid:63) Linear combination limb darkening parameters (prior & transit fit), u1 + u2, Stellar Density (prior & transit fit), ρ(cid:63) (g cm−3) Stellar radial velocity amplitude, K(cid:63) (m s−1) Linear RV term, φ1 (m s−1/day) Quadratic RV term, φ2 (m s−1/day) Systemic velocity HARPS-N, γHARPS−N (km s−1) Systemic velocity HARPS, γHARPS (km s−1) Systemic velocity FIES, γFIES (km s−1) Indirectly Derived Parameters Impact parameter, b Planetary Mass, Mp (M⊕)b Mass upper limit (95% confidence), Mp (M⊕)b Planetary Radius, Rp (R⊕)b Planetary Density, ρp (g cm−3) Notes -- a This value is used as a prior during the fitting procedure. b Adopting an Earth radius of 6371 km and mass of 5.9736 · 1024 kg. 5.65639±0.00075 6811.7191±0.0049 6.77145±0.00013 6812.8451±0.0010 19.3044±0.0012 6819.5814±0.0021 0.12+0.22−0.09 0.052+0.039 −0.032 0.115+0.022 −0.011 0.01654+0.00093 −0.00071 0.668±0.098 0.39+0.14−0.16 −0.18+2.6−2.6 −0.03±0.12 0.0183±0.0016 44.561±0.002 44.560±0.006 44.509±0.010 0.16+0.10−0.09 0.020+0.013 −0.013 0.0605+0.0051 −0.0038 0.0478+0.0013 −0.0008 0.676±0.082 1.80+0.70−0.55 10.8+2.7−2.7 -- -- −37.772±0.002 −37.776±0.003 −37.979±0.014 0.31+0.16−0.18 0.018+0.01−0.012 0.0349+0.0042 −0.0029 0.03570+0.0017 −0.0009 0.622±0.092 1.19+0.35−0.34 7.3+4.6−4.2 -- -- 8.203±0.003 8.062±0.005 -- 0.43+0.25−0.26 12.0 2.75+0.24−0.22 ≤ 3.3 (95% confidence) 0.30+0.21−0.20 29.1+7.5−7.4 41.5 4.45+0.33−0.33 1.80+0.70−0.55 0.40+0.27−0.27 27+17−16 57 3.84+0.35−0.34 2.6+2.1−1.6 6 Van Eylen et al. which together with a measured radius of 2.75+0.24−0.22 R⊕, re- sults in a planet density upper limit of 3.3 g cm−3. These limit are one-sided 95% confidence intervals. We note that our best measured mass, −0.5+7.6−7.5M⊕, the median of the distribution, is negative (see Figure 5). We allow for negative (unphysi- cal) mass to avoid positively biasing mass measurements for small planets. While this can easily be avoided using a prior that prohibits the unphysical mass regime, we prefer not to do so because the negative masses are a statistically impor- tant measurement of the planetary mass (see e.g. Marcy et al. 2014, for a detailed discussion). Allowing negative masses accounts naturally for the uncertainty in planet mass due to RV errors, and is key to allow unbiased constraints the interior structure based on a sample of small planets (see e.g. Rogers 2015; Wolfgang et al. 2015). All parameters are available in Table 1. 5.2. K2-27 (EPIC 201546283) For EPIC 201546283b, which has an orbital period P of 6.77 days, we obtain a 3σ mass detection of 29.1+7.5−7.4 M⊕. This confirms the planetary nature of this candidate, which we fur- ther refer to as K2-27b. Earlier work detected the transits of this candidate but was unable to confirm the planetary nature on statistical grounds (Montet et al. 2015). We find a plane- tary radius of 4.45+0.33−0.33 R⊕, which taken together with the mass measurement leads to a planet density of 1.80+0.70−0.55 g cm−3. This makes this planet rather similar to Neptune (which has a density of 1.64 g cm−3). As for EPIC 201295312 we al- lowed for the presence of a linear drift but found that it did not significantly change the results, and therefore we set it to zero. All parameters for this star and its planet are available in Table 1. The planet is plotted on a mass-radius diagram in Figure 6. We now check if the RV signal could be caused by stellar activity. For this we calculate the BIS as defined by Queloz et al. (2001) from the HARPS and HARPS-N CCFs and searched for a correlation with the measured RVs. If such a correlation does exist, then this is a sign of a deformation of the stellar absorption lines by stellar activity instead of a Doppler shift of the CCF caused by the gravitational pull of an unseen companion. However, no such correlation can be found. We calculate the Pearson correlation coefficient, which is 0.105. With 23 degrees of freedom, we also find a t-statistic of 0.505, and a two-tailored test leads to a p-value of 0.61. All these tests indicate there is no significant evidence for a cor- relation between the BIS and RVs. The values are shown in Figure 7. 5.3. K2-10 (EPIC 201577035) K2-10b (EPIC 201577035b) was previously validated to be a true planet by Montet et al. (2015). Here we refine the stellar and planetary parameters and constrain the planet's mass. The planet is smaller than Neptune with an orbital period of 19.3 days and a radius of 3.84+0.35−0.34 R⊕. We measure its mass to be 27+17−16 M⊕, resulting in a planetary density of 2.6+2.1−1.6 g cm−3. Within 95% confidence, the planetary mass is below 57 M⊕. We found no evidence for a linear drift and set it to zero in the final fit. We note that, due to uncooperative weather, the coverage of this system shows a significant phase gap (see Figure 5). All parameters are listed in Table 1, and the system is indicated on a mass-radius diagram in Figure 6. 6. DISCUSSION Fig. 5. -- RV observations over orbital phase for EPIC 201295312 (top), K2-27 (EPIC 201546283, middle) and K2-10 (EPIC 201577035, bottom). The best fitted model is shown with a solid line as well as the residuals after subtracting the model. The internal RV uncertainties are indicated by the black error bars, while the gray error bars include an additional "stellar jitter" term as explained in the text. Note that for K2-27 the residual plot do not show all FIES RVs due to the small RV interval displayed here. For this system the FIES data does not carry a lot of weight for the final solution and the zoom allows for a better inspection of the HARPS and HARPS-N residual which do determine our final solution. −0.051.05−30−1501530radial velocity (m s−1)HARPS−NHARPSFIESFIES0.00.20.40.60.81.0orbital phase−30−20−100102030O − C (m s−1)−0.051.05−60−40−200204060radial velocity (m s−1)HARPS−NHARPSFIESFIES0.00.20.40.60.81.0orbital phase−30−20−100102030O − C (m s−1)−0.051.05−30−1501530radial velocity (m s−1)HARPS−NFIESFIES0.00.20.40.60.81.0orbital phase−20−1001020O − C (m s−1)Spectroscopic follow-up of K2 Campaign 1 7 Neptune. For EPIC 201295312, we also discover a long- period trend indicative of an additional body. Relatively few mass measurements are available for sub-Neptunes due to the small RV amplitudes these planets cause. For example, an ex- tensive Keck campaign following up on 22 Kepler stars with known transiting planets recently lead to 16 secure mass de- tections (Marcy et al. 2014) and 26 more marginal measure- ments or upper limits. Despite thousands of planetary can- didates found by the primary Kepler mission, many of those are too faint for follow-up measurements. Here, K2 has the potential to make a significant contribution. Finally, the significant effort required to measure masses for the planets highlights the need for future missions such as TESS (Ricker et al. 2014) and PLATO (Rauer et al. 2014) which will observe even brighter stars. We appreciate the quick and thoughtful comments and sugges- tions by the referee, which significantly improved the manuscript. We are grateful for the Python implementation of a BLS algorithm by Ruth Angus and Dan Foreman-Mackey. We are thankful to the GAPS consortium, which kindly agreed to exchange time with us, and the KEST team which shared observations. Based on obser- vations made with the NOT and TNG telescopes operated on the island of La Palma in the Observatorio del Roque de Los Mucha- chos of the Instituto de Astrof´ısica de Canarias, as well as obser- vations with the HARPS spectrograph at ESO's La Silla observa- tory (095.C-0718(A)). Funding for the Stellar Astrophysics Centre is provided by The Danish National Research Foundation (Grant agreement no.: DNRF106). The research is supported by the AS- TERISK project (ASTERoseismic Investigations with SONG and Kepler) funded by the European Research Council (Grant agreement no.: 267864). We acknowledge ASK for covering travels in relation to this publication. This research has made use of the Exoplanet Or- bit Database and the Exoplanet Data Explorer at exoplanets.org. This work was performed [in part] under contract with the Jet Propulsion Laboratory (JPL) funded by NASA through the Sagan Fellowship Program executed by the NASA Exoplanet Science Institute. I.R. and M.P. acknowledge support from the Spanish Ministry of Econ- omy and Competitiveness (MINECO) and the Fondo Europeo de Desarrollo Regional (FEDER) through grants ESP2013-48391-C4- 1-R and ESP2014-57495-C2-2-R. N.N. acknowledges supports by the NAOJ Fellowship, Inoue Science Research Award, and Grant- in-Aid for Scientific Research (A) (JSPS KAKENHI Grant Number 25247026). T.H. is supported by Japan Society for Promotion of Sci- ence (JSPS) Fellowship for Research (No.25-3183). L.A.R. grate- fully acknowledges support provided by NASA through Hubble Fel- lowship grant #HF-51313 awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Re- search in Astronomy, Inc., for NASA, under contract NAS 5-26555. HD acknowledges support by grant AYA2012-39346-C02-02 of the Spanish Secretary of State for R&D&i (MICINN). Fig. 6. -- Mass-radius diagram for transiting exoplanets smaller than 10 R⊕ and less massive than 50 M⊕ with masses measured from RV observations (data from exoplanets.org) and TTVs. Our new measurements are indicated. Fig. 7. -- BIS from HARPS and HARPS-N CCFs for K2-27 plotted versus the stellar RVs. The color code indicates the signal-to-noise ratio in the stellar spectra obtained around a wavelength range of 5560 Å. There is no evidence for correlation between the BIS and RV. As expected lower signal-to-noise spectra do have a larger scatter in both BIS and RV. The uncertainty in the bisector values is taken to be twice the uncertainty in the RV values. We have reported our results for three planet candidates observed with K2 during Campaign 1. These planets have been found in the K2 data using two different algorithms and were also discussed by Montet et al. (2015). We measured the mass for the largest of these planets (EPIC 201546283b) and obtain lower significance measurements or upper limits of the masses and densities of the other two systems (EPIC 201295312b and EPIC 201577035b). The first of these plan- ets is similar to Neptune, while the other two are smaller than REFERENCES Angus, R., Foreman-Mackey, D., & Johnson, J. A. 2015, ArXiv e-prints Becker, J. C., Vanderburg, A., Adams, F. C., Rappaport, S. A., & Schwengeler, H. M. 2015, ArXiv e-prints Bruntt, H., Basu, S., Smalley, B., et al. 2012, MNRAS, 423, 122 Claret, A., Hauschildt, P. H., & Witte, S. 2013, A&A, 552, A16 Cosentino, R., Lovis, C., Pepe, F., et al. 2012, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 8446, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, 1 478 e-prints Crossfield, I. J. M., Petigura, E., Schlieder, J. E., et al. 2015, ApJ, 804, 10 D´esert, J.-M., Charbonneau, D., Torres, G., et al. 2015, ApJ, 804, 59 D´esert, J.-M., Lecavelier des Etangs, A., H´ebrard, G., et al. 2009, ApJ, 699, Desidera, S., Sozzetti, A., Bonomo, A. S., et al. 2013, A&A, 554, A29 Foreman-Mackey, D., Montet, B. T., Hogg, D. W., et al. 2015, ArXiv Gandolfi, D., Parviainen, H., Deeg, H. J., et al. 2015, A&A, 576, A11 01020304050Mass [M⊕]0246810Radius [R⊕]EPIC 201295312bEPIC 201546283bEPIC 201577035b−200204060BIS (m s−1)−30−20−100102030RV − mean RV (m s−1)101520253035SNR8 Van Eylen et al. Gelman, A. & Rubin, D. B. 1992, Statistical Science, 7, 457 Hirano, T., Sanchis-Ojeda, R., Takeda, Y., et al. 2012, ApJ, 756, 66 Howell, S. B., Sobeck, C., Haas, M., et al. 2014, PASP, 126, 398 Kov´acs, G., Zucker, S., & Mazeh, T. 2002, A&A, 391, 369 L´opez-Morales, M., Triaud, A. H. M. J., Rodler, F., et al. 2014, ApJ, 792, Lucy, L. B. & Sweeney, M. A. 1971, AJ, 76, 544 Lund, M. N., Handberg, R., Davies, G. R., Chaplin, W. J., & Jones, C. D. 2015, ApJ, 806, 30 Mandel, K. & Agol, E. 2002, ApJ, 580, L171 Marcy, G. W., Isaacson, H., Howard, A. W., et al. 2014, ApJS, 210, 20 Mayor, M., Pepe, F., Queloz, D., et al. 2003, The Messenger, 114, 20 Montet, B. T., Morton, T. D., Foreman-Mackey, D., et al. 2015, ArXiv Petigura, E. A., Schlieder, J. E., Crossfield, I. J. M., et al. 2015, ArXiv L31 e-prints e-prints Queloz, D., Henry, G. W., Sivan, J. P., et al. 2001, A&A, 379, 279 Rauer, H., Catala, C., Aerts, C., et al. 2014, Experimental Astronomy Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2014, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 9143, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, 20 Rogers, L. A. 2015, ApJ, 801, 41 Sanchis-Ojeda, R., Rappaport, S., Pall´e, E., et al. 2015, ArXiv e-prints Shen, Y. & Turner, E. L. 2008, ApJ, 685, 553 Stello, D., Chaplin, W. J., Bruntt, H., et al. 2009, ApJ, 700, 1589 Takeda, Y., Ohkubo, M., & Sadakane, K. 2002, PASJ, 54, 451 Tegmark, M., Strauss, M. A., Blanton, M. R., et al. 2004, Phys. Rev. D, 69, Telting, J. H., Avila, G., Buchhave, L., et al. 2014, Astronomische 103501 Nachrichten, 335, 41 Van Eylen, V. & Albrecht, S. 2015, ApJ, 808, 126 Van Eylen, V., Lund, M. N., Silva Aguirre, V., et al. 2014, ApJ, 782, 14 Vanderburg, A. & Johnson, J. A. 2014, PASP, 126, 948 Vanderburg, A., Montet, B. T., Johnson, J. A., et al. 2015, ApJ, 800, 59 Wolfgang, A., Rogers, L. A., & Ford, E. B. 2015, ArXiv e-prints
1905.12639	1	1905	2019-05-29T18:00:05	Planet-forming material in a protoplanetary disc: the interplay between chemical evolution and pebble drift	[ "astro-ph.EP", "astro-ph.SR" ]	The composition of gas and solids in protoplanetary discs sets the composition of planets that form out of them. Recent chemical models have shown that the composition of gas and dust in discs evolves on Myr time-scales, with volatile species disappearing from the gas phase. However, discs evolve due to gas accretion and radial drift of dust on time-scales similar to these chemical time-scales. Here we present the first model coupling the chemical evolution in the disc mid-planes with the evolution of discs due to accretion and radial drift of dust. Our models show that transport will always overcome the depletion of CO$_2$ from the gas phase, and can also overcome the depletion of CO and CH$_4$ unless both transport is slow (viscous $\alpha \lesssim 10^{-3}$) and the ionization rate is high ($\zeta \approx 10^{-17}$). Including radial drift further enhances the abundances of volatile species because they are carried in on the surface of grains before evaporating left at their ice lines. Due to large differences in the abundances within 10 au for models with and without efficient radial drift, we argue that composition can be used to constrain models of planet formation via pebble accretion.	astro-ph.EP	astro-ph	MNRAS 000, 1 -- 14 (2019) Preprint 31 May 2019 Compiled using MNRAS LATEX style file v3.0 Planet-forming material in a protoplanetary disc: the interplay between chemical evolution and pebble drift R. A. Booth1(cid:63) and J. D. Ilee2† 1Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK 2School of Physics & Astronomy, University of Leeds, Leeds LS2 9JT, UK Accepted 2019 May 27. Received 2019 May 17; in original form 2019 April 10 ABSTRACT The composition of gas and solids in protoplanetary discs sets the composition of plan- ets that form out of them. Recent chemical models have shown that the composition of gas and dust in discs evolves on Myr time-scales, with volatile species disappearing from the gas phase. However, discs evolve due to gas accretion and radial drift of dust on time-scales similar to these chemical time-scales. Here we present the first model coupling the chemical evolution in the disc mid-planes with the evolution of discs due to accretion and radial drift of dust. Our models show that transport will always over- come the depletion of CO2 from the gas phase, and can also overcome the depletion of CO and CH4 unless both transport is slow (viscous α (cid:46) 10−3) and the ionization rate is high (ζ ≈ 10−17). Including radial drift further enhances the abundances of volatile species because they are carried in on the surface of grains before evaporating left at their ice lines. Due to large differences in the abundances within 10 au for models with and without efficient radial drift, we argue that composition can be used to constrain models of planet formation via pebble accretion. Key words: astrochemistry -- protoplanetary discs -- planets and satellites: forma- tion -- planets and satellites: composition 1 INTRODUCTION The structure and composition of protoplanetary discs are fundamental pieces in the puzzle of how planets form and evolve. In the most direct sense, planets form via the ac- cretion of gas and solids -- in the form of planetesimals, pebbles, or dust, which together with the gas are the main components of protoplanetary discs. This provides the op- portunity to learn about planet formation by connecting the composition of planets to the discs in which they form. A challenge for the protoplanetary disc community is thus to determine the composition of the planetary building blocks, a task complicated by uncertainties in the processes govern- ing disc evolution and the associated difficulties with the in- terpretation observational constraints. Here, we explore one aspect of this problem -- namely how differences in the trans- port of gas and solids through the disc compete with their chemical evolution to control the composition of discs. We frame our investigation in terms of a recent idea, which is to use the atmospheric C/O ratio in hot Jupiters to determine their building blocks. The observational utility (cid:63) E-mail: [email protected] (RAB) † E-mail: [email protected] (JDI) © 2019 The Authors of the C/O ratio comes from the relative ease with which it can be constrained in hot Jupiter atmospheres. This is because the abundance of the observed molecules can vary by orders of magnitude for only small changes in the C/O ratio (see Lodders 2010; Madhusudhan 2012). Oberg et al. (2011) realised that the C/O ratio could be used to con- strain the radial location within a disc from which a planet formed. This is possible because the C/O ratio of the gas and ices changes across the ice lines (places where molecular species transition from being predominantly in the gas phase to ices in the solid phase) of the dominant carbon and oxy- gen bearing species, such as CO, CO2 and H2O. Numerous studies have since explored how planet formation and migra- tion affect the C/O ratio of the planet (Madhusudhan et al. 2014; Mordasini et al. 2016; Cridland et al. 2016, 2017; Ali- Dib 2017; Booth et al. 2017; Madhusudhan et al. 2017). Al- though there are differences in the composition of the model planets in these studies, there is general agreement that the partitioning of the carbon and oxygen between the gas and solid phases, along with the amount these species accreted, is important in determining the planet's composition. However, the composition of discs is not static in time, which needs to be taken into account in planet formation models. Although Cridland et al. (2016, 2017) recently cou- 2 Booth & Ilee pled a planet formation model to a disc model including chemical kinetics, they focused on planets that accreted their envelopes inside the water ice line, where the majority of volatile carbon and oxygen bearing species are in the gas phase. Further out in the disc, in regions where the tem- peratures are lower and molecular species freeze out, chem- ical kinetics can have a greater impact on the gas and solid phase composition by exchanging carbon and oxygen be- tween species that are in ices on grains or in the gas phase. Using an extensive gas-grain network Eistrup et al. (2016) showed that CH4 and CO can be removed from the gas phase, with the net effect of lowering the gas-phase C/O ratio inside of around 10 au. However, this process is slow, taking several million years (Eistrup et al. 2018; Bosman et al. 2018b), by which time gas giant must be well under- way (Greaves & Rice 2010; Najita & Kenyon 2014; Manara et al. 2018). By comparison, it is becoming clear that the transport of molecular species is important on time-scales comparable to chemical processes. Bosman et al. (2018a) showed that even conservatively slow models of transport were enough to raise the CO2 abundance above levels that has previously been assumed. Furthermore, transport is an essential com- ponent of a popular new model for planet formation -- peb- ble accretion (Ormel & Klahr 2010; Lambrechts & Johansen 2012) -- which relies on dust grains large enough that they decouple from the gas (e.g. Lambrechts & Johansen 2014; Morbidelli et al. 2015; Bitsch et al. 2015. Johansen et al. 2018 invoke smaller pebbles, although they are still large enough to migrate). These pebbles thus migrate rapidly to- wards the star (Weidenschilling 1977) carrying the volatile molecular species frozen out onto their surfaces with them as they migrate. These volatile species then enter the gas phase when the pebbles cross their respective ice lines. Under the conditions often assumed in pebble accretion models, radial drift enhances the gas phase abundances by a factor of a few within a Myr (Booth et al. 2017; Krijt et al. 2018). Previous studies investigating the interplay between chemical kinetics and transport typically assume that dust and gas are coupled (e.g. Aikawa et al. 1999; Tscharnuter & Gail 2007; Nomura et al. 2009; Semenov & Wiebe 2011; Heinzeller et al. 2011; Walsh et al. 2014; Gail & Trieloff 2017; see Henning & Semenov 2013, for a review). These stud- ies suggest that the transport has two predominant effects: 1) diffusion weakens concentration gradients and 2) trans- port limits the amount of time molecular species spend at a given location, reducing the influence of chemical processes that act on time scales longer than the transport time-scale. While a number of studies have treated the transport of gas and solids separately (e.g. in the context of CO sequestra- tion in disc mid-planes; Piso et al. 2015, 2016; Bergin et al. 2016; Kama et al. 2016; Booth et al. 2017; Krijt et al. 2018; or the destruction of carbon grains in the terrestrial planet- forming region; Klarmann et al. 2018), to our knowledge this is the first study to treat gas and dust transport indepen- dently and gas-phase kinetics in detail. We emphasise that differences in the transport of gas and dust is the only way that the bulk atomic abundances (i.e. gas+dust abundance of carbon, oxygen, and nitrogen, etc) can vary in the disc; the partitioning of different molecular species between the gas and ice phase changes the atomic abundance of the each phase separately, but not the total abundance. In this work we investigate the competition between chemical kinetics and transport in the mid-plane of proto- planetary discs. To this end, we couple a chemical kinetics network (Ilee et al. 2011) with a 1D disc evolution model. Our disc evolution model includes gas evolution (which is treated as viscous), grain growth and radial drift (Booth et al. 2017)1. We allow for differences in the transport of gas and ice phase molecular species, tracking their motion with the gas and dust, respectively. We examine how trans- port and chemical kinetics compete, focusing on the classical giant planet forming region (1 -- 10 au). 2 METHODS 2.1 Physical evolution We consider the physical evolution of gas and dust follow- ing Booth et al. (2017). The gas evolution is treated as- suming a viscous disc, with a constant α = 10−3 or 10−2. The justification of these choices is given in subsection 2.2. Grain growth and radial drift are treated based upon Birn- stiel et al. (2012). The mid-plane temperature is computed taking into account viscous heating and irradiation from the central star, along with external irradiation with a temper- ature of 10 K. We self-consistently take into account the motion of each of the chemical species included in the model, treat- ing the motion of gas and ice phase species independently. The gas phase species are advected at the gas velocity, which is determined by viscous evolution. Ice phase species are ad- vected along with the dust. The dust radial velocity is set by a combination of both the gas velocity due to viscosity and radial drift (e.g. Takeuchi & Lin 2002). The grain growth model assumes that there are two populations of grains, small and large, here we assume that the ice species are partitioned across the two populations in the same way as the dust. This results in the advection of ice phase species being dominated by the large grain popu- lation, in good agreement with calculations treating the full distribution of sizes (Stammler et al. 2017). In addition to advection, we include turbulent diffusion. The diffusion co- efficient D of the gas is given by D = ν/Sc, where ν is the viscosity and the Sc the Schmidt number, for which we take Sc = 1. For the dust and ice phases the diffusion coefficient is scaled to the gas coefficient following Youdin & Lithwick (2007) and assuming the eddy turnover time-scale equals the dynamical time-scale, Ω−1. For the dust sizes considered here, the diffusion coefficients of the dust and gas are the same to within one per cent. The opacity used in the mid-plane temperature calcu- lation has been updated. In Booth et al. (2017) we used the Rosseland mean opacity tables similar to Bell & Lin (1994), as computed by Zhu et al. (2012). Since these opacities assume a grain size distribution appropriate for the inter- stellar medium they can considerably overestimate the dust opacity once the grains have grown and begin to migrate. 1 https://github.com/rbooth200/DiscEvolution MNRAS 000, 1 -- 14 (2019) Here we instead use Rosseland mean opacity computed self- consistently for grain size distributions with number density n(a) ∝ a−3.5 with the maximum grain size taken from the grain growth model. The dust properties follow Tazzari et al. (2016), based on the composition of Pollack et al. (1994) and assuming a porosity of 30 per cent. In addition the models that include the effects of trans- port processes: viscous evolution, radial drift and diffusion; we also consider models that switch off some or all of these processes. This allows us to separate out the effects of phys- ical and chemical processes on the composition. 2.2 Physical parameters We explore two models for the evolution of the disc, with the same initial surface density profile, but choices of the turbulent α parameter -- and correspondingly different ac- cretion rates. Unless otherwise stated, in each model we set the initial disc mass to 0.01 M(cid:12) and stellar mass to 1 M(cid:12). The surface density profile is initialised to a Lynden-Bell & Pringle (1974) profile, − R Σ(R) = Rd exp (cid:19) (cid:18) (1) . Md 2πR2 d We assume that the initial ratio of refractory dust-to-gas surface density is 0.01 everywhere. The total dust-to-gas ra- tio is higher than this by approximately factor of 2 due to the presence of ices frozen onto grain surfaces. The initial grain size is also taken to be 0.1µm. For the turbulent α parameter we use α = 10−3 and α = 10−2. While we note that recent observations suggest turbulence in the outer regions of discs is weaker than this (Pinte et al. 2016; Flaherty et al. 2017, 2018; Teague et al. 2018), there is evidence that the accretion rate through discs is more compatible with higher values of α. In the absence of direct measurements of the radial flow of gas through discs, the accretion rate onto the star combined with the disc mass provide the most reasonable estimates. For example, Boneberg et al. (2016) found that an effective α ∼ 0.1 was required to match the accretion rate of HD 163296, while the turbulent mixing in the vertical direction was lower. Further- more, based on the surface density profile and accretion rate Clarke et al. (2018) derive α ∼ 0.01 for CI Tau. Even TW Hya, which is an old system and has a low accretion rate, has an accretion rate compatible with α ∼ 10−3 (Calvet et al. 2002; Bergin et al. 2013; Ercolano et al. 2017). Based on dust disc sizes and ages, Andrews & Williams (2007) and Guil- loteau et al. (2011) suggest α in the range 10−3 -- 10−2. Sim- ilarly, for most discs Rafikov (2017) inferred α in the range 10−3 -- 10−2 based on the disc radii and masses inferred from dust emission in Lupus and their associated accretion rates. High dust-to-gas ratios in Lupus, as suggested by Miotello et al. (2017), would result in higher estimates of α for Lupus; however, Manara et al. (2016) suggest that the dust mass is well matched to the accretion rate (but see Mulders et al. 2017 and Lodato et al. 2017 for a discussion). Differences be- tween the accretion rate and inferred turbulence levels may point to MHD winds as the source of angular momentum transport instead of viscosity (Suzuki & Inutsuka 2009; Fro- mang et al. 2013; Bai & Stone 2013). However, since we are mainly concerned with the speed of mass transport through MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 3 the disc, this distinction is not critical. Thus we model the discs as viscous, considering α in the range 10−3 -- 10−2 to be typical for nearby, low mass protoplanetary discs. For the dust evolution, we use the same parameters as Booth et al. (2017), for which the standard parameters are based on fits to more detailed models (Birnstiel et al. 2012), except that we use the ice fraction to specify the location where the fragmentation velocity transitions from water ice- like (10 m s−1) to silicate-like (1 m s−1). Rather than varying these parameters, we choose instead to simply run models with radial drift included or neglected (in which case the dust moves with the gas). Combined with the two choices of α, this spans the commonly studied parameter space. For example α = 10−3 is commonly considered (e.g. Birnstiel et al. 2012), for which dust grows large enough to be in the radial drift dominated regime at large radii, leading to rapid evolution of the dust-to-gas ratio. This model is also comparable to the parameters assumed in pebble accretion models of Lambrechts & Johansen (2014); Morbidelli et al. (2015) and Bitsch et al. (2015). In the α = 10−2 model, dust growth is initially limited by fragmentation everywhere, re- sulting in smaller grains and slower radial drift. Slower radial drift combined with a faster gas radial velocity results in the dust-to-gas ratio decreasing more slowly. Such a model may be in better agreement with observed disc properties, where low dust-to-gas ratios are typically not inferred (Boneberg et al. 2016; Ansdell et al. 2016; although note that masses determined from CO observations are uncertain), and is also comparable to the parameters assumed in the pebble accre- tion model of Johansen et al. (2018). The evolution of the surface density, dust-to-gas ratio, temperature, and maximum grain size for our two canonical models are shown in Figure 1 (α = 10−2) and Figure 2 (α = 10−3). The evolution of combined dust-gas models has been described at length in Birnstiel et al. (2012), and Booth et al. (2017) in the case where adsorption and desorption at ice lines is included. Thus we only describe the salient differences here and refer the readers to the above references for further details. The main features of note are: the slower evolution of dust-to-gas ratio at α = 10−2, in which the dust- to-gas ratio is enhanced by a factor ∼ 2 inside 10 au, and only decreases outside the scaling radius (100 au). In comparison the dust-to-gas ratio falls to 10−4 on Myr time-scales in the α = 10−3 model, a consequence of larger grain maximum grain size. Note also that gas surface density decreases in the α = 10−2 model, but barely changes after 1 Myr at α = 10−3. Another consequence of the higher accretion rate at α = 10−2 is the warmer temperature in the disc inside of ∼ 5 au, due to both the larger contribution of viscous heating and also a higher optical depth because of smaller grains. The time-scale for the evolution of the gas and dust sur- face densities are given by the time-scale for the accretion of gas and dust onto the star. These time-scales will be useful in the following sections for interpreting the competition be- tween chemical evolution and transport. In the case of the gas, the accretion time-scale is controlled by the viscous ve- locity, vν, from which we can estimate a viscous time-scale, tν = R/vν, (cid:18) α 10−3 (cid:19)−1(cid:18) R (cid:19) au tν ≈ 2 × 105 yr. (2) 4 Booth & Ilee Figure 1. Evolution of the physical parameters for the model with α = 10−2. Top left: gas (solid) and dust (dashed) surface density. Top right: dust-to-gas-ratio. Bottom left: Temperature. Bottom Right: maximum grain size. Figure 2. Same as Figure 1, but for α = 10−3. The long viscous time-scale at 100 au explains the slow change in gas surface density in α = 10−3 model. However, at 1 au the viscous transport time is always much less than 1 Myr, and for α = 10−2 the viscous transport time is less than 1 Myr everywhere inside of approximately 50 au. In addition to transport with the gas, the dust surface density also evolves due to radial drift, with the associated radial drift time-scale, tdrift = R/vdrift. In this case, the evolu- tion of the dust surface density is always on the shorter of ra- dial drift and viscous time-scales. The radial drift time-scale depends on both the dust grain size and gas surface density profile. Taking the gas surface density to be Σ = Σ0(R/au)−1, we estimate the radial drift time-scale as tdrift ≈ 1 × 105 yr. Σ0 mm 102 g cm−2 (3) Typical grain sizes are 0.1 -- 1 mm for α = 10−2 and 1 -- 10 mm for α = 10−3. Hence the dust evolves on approxi- mately the viscous time-scale for α = 10−2, which explains the slow evolution of dust-to-gas ratio. However, for α = 10−3 (cid:18) (cid:19)(cid:16) a (cid:17)−1 the dust evolves faster than the gas, and the dust-to-gas ra- tio decreases. 2.3 Chemical evolution Within the physical model discussed above, we embed a time-dependent gas-grain chemical evolution model that self-consistently calculates the chemical evolution of each grid cell. For this, we utilise the krome chemical evolution package2 (Grassi et al. 2014). For the chemical reaction net- work, we adopt the network previously used in Ilee et al. 2011; Evans et al. 2015 and Ilee et al. 2017, with modifica- tions described here to provide consistency with more recent disc evolution models. The network consists of 1485 reac- tions involving 136 species containing the elements H, He, C, N, O, Na and S. These reactions were originally selected from a subset of the UMIST Rate 95 database (Millar et al. 2 http://www.kromepackage.org/ MNRAS 000, 1 -- 14 (2019) 10-310-210-1100101102103§G;D10-410-310-210-1100§D=§G10-1100101102103R[au]101102103T[K]0.01Myr0.1Myr0.5Myr1.0Myr10-1100101102103R[au]10-510-410-310-210-1100a[cm]10-310-210-1100101102103§G;D10-410-310-210-1100§D=§G10-1100101102103R[au]101102103T[K]0.01 Myr0.1 Myr0.5 Myr1.0 Myr10-1100101102103R[au]10-510-410-310-210-1100a[cm]1997). Data from the Kinetic Database For Astrochemistry, KIDA3, were used to update some of the rates and rate co- efficients. In addition to the chemical reactions, adsorption, and desorption processes; we self-consistently follow the ad- vection and diffusion of each of the gas and ice phase species following Booth et al. (2017). We overview the key aspects of our chemical model below. The fundamental variables integrated in the disc evolu- tion and transport code are the total surface density, dust fraction, and mass fraction of each chemical species. These are used to compute the mid-plane number density, n(i), and dust-to-gas ratio, , assuming a Gaussian vertical structure using the scale-height computed from the mid-plane temper- ature. The fractional abundance relative to the total number of hydrogen nuclei, n, is X(i) ≡ n(i)/n. The rate equation for the gas-phase fractional abundance of species i is therefore = d dt X(i) ≡ S(i) − − 2 j(cid:48),m j,l,m j(cid:48)(cid:48) k(j)X(l)X(m)n k(j(cid:48))X(i)X(m)n k(j(cid:48)(cid:48))X(i)2n − Srad(i)X(i) + S3(i) + Sd(i) − Sa(i), (cid:19) (cid:18) T (cid:19) β(j) (cid:18)−γ(j) where k(j) is the rate coefficient of the jth reaction, and the summations are restricted so that only reactions involved in the formation or removal of the ith species are included. We denote this total rate S(i). These rate coefficients are of the standard Arrhenius form k(j) = α(j) (5) where α(j) is the room temperature rate coefficient of the reaction (at 300 K), β(j) describes the temperature depen- dence and γ(j) is the activation energy of the reaction. The terms Srad(i)X(i), S3, Sa and Sd represent radiative destruc- tion processes, three-body gas phase reactions, adsorption on to dust grains, and desorption from dust grains, respec- tively. 300 exp T , For three body reactions, we assume that the only third body of importance is H2, and therefore S3(i) = − − 2 j(cid:48),m j,l,m j(cid:48)(cid:48) k(j)X(l)X(m)X(H2)n2 k(j(cid:48))X(i)X(m)X(H2)n2 k(j(cid:48)(cid:48))X(i)2X(H2)n2. The rate of destruction of species i due to radiative pro- cesses is given by Srad = α(i) exp(−γAV) for photoreactions (where α is a proportionality constant containing the dust grain albedo, which we take to be 0.5). (7) 3 http://kida.obs.u-bordeaux1.fr MNRAS 000, 1 -- 14 (2019) (cid:115) Chemistry & transport in discs 5 For cosmic ray ionisation, we take Srad = ζ = 10−17 s−1 every- where, unless otherwise specified. Though photoabsorption of radiation from external sources will affect the chemistry in the upper layers of a disc, we focus on the bulk of the planet forming material toward the disc mid-plane. We therefore as- sume that all material is well shielded from external sources of photons, setting AV = 30 mag everywhere. However, we do include photoreactions due to secondary photons generated by cosmic rays, using the rates from Heays et al. (2017). For the adsorption of species from the gas phase onto the surfaces of icy grain mantles, we assume 8kT πm(i) (cid:104)πa2(cid:105) Sa = Xdust S(i) (8) where Xdust is the dust number density, S(i) is the sticking coefficient (taken to be 0.3), m(i) is the atomic mass of the adsorbed species, and (cid:104)πa2(cid:105) is the average dust grain area. The average grain radius is computed using the maximum grain size from the dust evolution code, assuming a grain size distribution n(a)da ∝ a−3.5da with the minimum grain- size equal to 0.1µm. However, we note that our results are only very weakly dependent on this choice (as shown in sec- tion A). For unsaturated C, N and O species on the surfaces of dust grains, we assume hydrogenation occurs via the Eley-Rideal mechanism, at the rate of adsorption of H from the gas multiplied by the probability of encounter with the species on the grain surface, e.g., gC → gCH → . . .→ gCH4 (where g denotes a species on a grain surface, see Visser et al. 2011). (cid:18)−Eb(i) (cid:19) For the thermal desorption of species from dust grain surfaces into the gas phase, we assume , kT Sd(i) = 1.26 × 10−21 σ ν0(i) exp (9) where σ = 1.5 × 1015 cm−2 is the surface density of bind- ing sites, Eb(i) is the binding energy of the ith species on the surface of the dust grain, and T is the dust temperature (which we assume to be in equilibrium with the gas temper- ature), and ν0 is the characteristic vibrational frequency of the species attached to the grain. Integration of the rate equations in each cell is per- formed independently, using the dlsodes package (Hind- marsh 1983) within the krome package (Grassi et al. 2014) to yield the time-dependent evolution of fractional abun- dance for each species throughout the disc. 2.4 Combined chemical evolution and transport The coupling to the disc evolution model is achieved in a a first-order operator-split fashion, which we briefly describe here. The full equation for the surface density of a given species, Σ(i), may be written as ∂Σ(i) (10) ∂t where vr(i) is the radial velocity of the species (either the gas velocity or dust velocity, depending on whether the species is in the gas or ice phase), and S(i) is the source terms due to chemical reactions (Equation 4). We discretize this equation [RΣ(i)vr(i)] = Σ(i)S(i), ∂ ∂R 1 R + (4) (6) 6 Booth & Ilee Table 1. Initial abundances relative to the number of hydrogen nuclei, X(i) ≡ n(i)/n. Based on Eistrup et al. (2016). Species H H2 He H2O CO CO2 X(i) 0.098 9.11 × 10−5 0.4999545 3.0 × 10−4 6.0 × 10−5 6.0 × 10−5 Species CH4 N2 NH3 H2S Na X(i) 1.8 × 10−5 2.1 × 10−5 2.1 × 10−5 6.0 × 10−6 3.5 × 10−5 on a fixed Eulerian grid, as described in Booth et al. (2017). We then proceed by splitting these equations into two steps, first a transport step without the chemical reactions: ∂Σ(i) ∂t [RΣ(i)vr(i)] = 0, (11) + 1 R ∂ ∂R where the amount of each species being transported between the different radial cells between times t and t + dt is com- puted as described in Booth et al. (2017). This produces a new estimate for the surface density in each cell, Σ∗(i), at time t + dt. The second step in the computation is to integrate = Σ(i)S(i) ∂Σ(i) (12) ∂t over a time dt starting from Σ∗(i). To do this, we first ρ(i) = Σ∗(i)/(√ compute mid-plane density of each chemical species via 2πH), where H is the disc-scale height. We then pass n(i) = ρ(i)/m(i), where m(i) is the mass of the species, into the chemical solver, krome, which integrates the number density from t to t + dt. The new n(i) can then be converted back to Σ(i). Finally, the grain size is updated as described in Booth et al. (2017) and the temperature in the disc updated. 2.5 Chemical initial conditions For the initial abundances, we assume the gas is molecu- lar with the abundance of key molecular species given in Table 1, which follow Eistrup et al. (2016), which are com- parable to the abundances in comets (Mumma & Charn- ley 2011). These abundances correspond to C/H, N/H, and O/H ratios of 0.47, 0.89, and 0.85 times solar, respectively (Asplund et al. 2009). The C/O ratio is 0.29, which is be- low the protosolar value of approximately 0.54. These lower abundances are consistent with the remaining species being locked up in refractory solids (Pollack et al. 1994). Where relevant, we thus assume the remaining C, N, and O atoms are locked up in the cores of dust grains, which for the pur- pose of this study are considered inert. 3 CHEMICAL EVOLUTION 3.1 Chemical evolution without transport We first begin with a simple test case in which all trans- port terms are turned off in the disc evolution model, i.e. the chemical reaction network for each cell is integrated in- dependently with the density fixed at the initial value. This allows a direct comparison with recent works exploring the impact of chemistry on the composition of planets (Eistrup et al. 2016; Cridland et al. 2016; Yu et al. 2016; Cridland et al. 2017; Eistrup et al. 2018), with which we can bench- mark the smaller chemical network used in this study. For this test we present results from the standard disc model with a mass of 0.01M(cid:12) and α = 10−3. The evolution in these models is almost entirely driven by ionization due to cosmic rays (see Appendix A or Eistrup et al. 2016). The abun- dances of the dominant carbon and oxygen carriers at 0 and 106 yr are shown in Figure 3, which can be compared with Figures 2 and 3 of Eistrup et al. (2016). The time evolution can also be seen in Figure A2. 2 and C2H+ Our model without transport is in good qualitative agreement with the results from the more extensive networks used by Eistrup et al. (2016, 2018), and Yu et al. (2016). The most significant effect is the depletion of CH4 inside of 10 au, where CH4 is in the gas phase. The primary path- way for the removal of CH4 are reactions with C+, which in turn is produced via reactions between CO and He+ that is generated by cosmic ray ionization. The reactions between CH4 and C+ produce C2H+ 3 , which can recombine with electrons to form C2H and C2H2 (Walsh et al. 2015; Yu et al. 2016). C2H2 then freezes out, acting as the main reservoir in our network. In reality, C2H2 is likely further hydrogenated on the grain surfaces -- Eistrup et al. (2018) find that the main reservoir in this region is the saturated hydrocarbon C2H6, which is also frozen out. We therefore argue that the conversion of CH4 to larger molecules is ad- equately captured in our model because the time-scale for depletion (a few 105 yr) is captured properly. Furthermore, since the binding energies of these two species are similar (Collings et al. 2004; Oberg et al. 2009, see also Penteado et al. 2017) they will be in the ice phase for similar ranges of parameter space, resulting in similar transport efficiency too. We thus argue that the differences are not significant for our goal, i.e. to asses how the carbon and oxygen abundance of the gas and solids during planet-formation is affected by the competition between transport and chemistry. The most significant differences come from the fact that we do not include detailed grain surface chemistry. However, besides the aforementioned hydrogenation of small hydro- carbons, these differences are most significant on time-scales of a few Myr. E.g. Eistrup et al. (2018) show that this leads to the conversion of CO into CO2 on time-scales of a few Myr in the outer disc where CO is frozen out (see also Bosman et al. 2018b). Without grain surface chemistry CO can only be destroyed in the gas phase, which takes several to 10 Myr. For this reason we focus on the first 1 Myr, where the differ- ences in the CO abundance are within a factor of 2. 3.2 Viscously evolving discs We now turn our attention to models involving transport, first considering models with viscous evolution but no radial drift. In this model, the dust and gas move together as they move towards the star, similar to the models by Aikawa et al. (1999). The differences in the chemical evolution between the models with and without transport are thus due to the gas at a given location having come from larger radii in the disc, where the temperatures and densities are lower. Since MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 7 Figure 3. Abundances relative to hydrogen, X(i), of the major carbon and oxygen carriers after 106 yr in models without radial drift. In the left panel transport is neglected entirely, while the other two panels show models in which the dust moves at the same speed as the gas. Solid lines denote gas phase species and dashed lines denote ice phase species and the dotted lines show the initial abundance of each species. The vertical bars show the approximate ice line locations. Note that C2H2 should be considered a proxy for hydrocarbons more generally. Figure 4. Same as Figure 3, but for models including radial drift. The left hand panel shows the α = 10−3 case in including only adsorption and desorption. each parcel of gas now spends a limited amount of time in the regions where a given chemical process is happening (e.g. the depletion of CH4), this naturally leads to competition between the chemical time-scales and transport time-scales (which are given in subsection 2.2). Transport thus reduces the impact of chemical reactions on the composition in the inner regions. Even in our low ac- cretion rate model with α = 10−3, viscous transport carries gas from 2 au onto the star more rapidly than CO2 depletes ((cid:38) 106 yr), as discussed in detail by Bosman et al. (2018a). However, at α = 10−3, viscous transport is not able to over- come the depletion of CH4, due to a combination of a shorter chemical time-scale (a few 105 yr) and the longer time that material spends at 10 au than at 2 au. At α = 10−2, viscous transport becomes efficient enough that gas-phase chemistry no longer has a significant effect on MNRAS 000, 1 -- 14 (2019) the abundance of the major carbon and oxygen carriers in- side 10 au. While there is still some conversion of CH4 to larger hydrocarbons, this only reduces the abundance of CH4 by a factor of ∼ 2. Thus the composition of the mid-plane inside the CO ice line (∼ 30 au) is largely set by the compo- sition of the material being carried in from larger radii. 3.3 Influence of radial drift The primary effect of radial drift is to enhance the transport of molecular ices from the outer disc (Booth et al. 2017). The molecular species frozen out enter the gas phase when the grains pass each species' respective ice lines, enhancing the local volatile abundance inside the ice line. The strength of this enhancement is controlled by the relative time-scales on which the gas and dust evolve: rapid radial drift leads 10-1100101102R [au]10-810-710-610-510-410-3abundanceno transport100101102R [au]®=10¡3100101102R [au]®=10¡2COCO2CH4H2OC2H210-1100101102R [au]10-810-710-610-510-410-3abundanceno chem100101102R [au]®=10¡3100101102R [au]®=10¡2COCO2CH4H2OC2H28 Booth & Ilee to a rapid increase of volatile abundance inside the ice line. In the opposite limit, when radial drift is slow, there is no enhancement of the abundances because the gas and dust move together. Thus the volatile species are transported away from the ice lines in the gas phase at the same rate they are brought there in ices on the grains. The influence of radial drift can be most clearly seen in the left hand panel of Figure 4, which shows the α = 10−3 model including radial drift and adsorption/desorption, but neglecting other chemical processes. The rapid dust trans- port in this model means that 90 per cent of the volatiles are delivered into the inner disc within 0.5 Myr. Combined with the low gas accretion rate, this results in high gas phase molecular abundances in the inner disc, along with a reduced dust-to-gas ratio (Figure 2). For CO, which enters the gas phase at around 40 au where viscous transport is slow, this results in a spike in the gas phase CO abundance. Inside 5 au the CO abundance is dominated by gas that originated inside the ice line and has been carried inwards by the gas; thus, the CO abundance returns to its initial value. For the species that have their ice lines closer to the star, such as CH4, CO2 and H2O, the abundance inside the ice lines has become close to constant because the transport time-scale is less than 1 Myr. Out- side their ice lines, the gas phase abundance of all molecular species remains low because any molecules that diffuse be- yond the ice line freeze out and are carried back on the grain surfaces. This results in the ice-to-dust abundance of indi- vidual species being enhanced by as much as a factor of ∼ 10 near their ice lines. The interplay between gas-phase chemistry, radial drift and viscous evolution can be seen in the middle panel of Fig- ure 4, for the model with α = 10−3. Here we see again that the conversion of CH4 to larger hydrocarbons is efficient in- side of 10 au, removing the spike in CH4 at the ice line. However, radial drift is efficient enough that the the CH4 abundance is still high after 1 Myr. The high CH4 abundance at ∼ 10 au also translates into a higher rate of hydrocarbon formation, which can be seen through the higher abundance of C2H2 ice relative to water ice in the middle panel of Fig- ure 4 compared to Figure 3. However, efficient radial drift prevents the newly formed ice reaching high abundances, with the ice instead being transported in to its own ice line, as suggested in Booth et al. (2017). A similar process is re- sponsible for the higher CO abundance in the inner region. In this case it is the formation of H2CO from CO near the ice line that is responsible. Once formed, the H2CO freezes out and is transported on the grains to around 5 au, where it enters the gas phase and is converted back to CO. The enhancement of CO in the inner regions thus relies on the formation of H2CO. We note that when grain surface reac- tions are included Eistrup et al. (2016) find CO2 is formed instead. While CO2 would freeze out also and be transported inwards, the conversion of CO2 into CO is slow compared to the transport. Thus the effect of CO conversion to CO2 and transport on the grain surfaces would be to further enhance the CO2 abundance in the inner regions, rather than CO. The evolution of species such as CO, CH4, CO2 and H2O after 1 Myr would be characterized with an inside-out depletion of the species in order of their snow line locations, because species closest to the star accreted onto the star first (Booth et al. 2017). Including chemical reactions only slightly modifies this picture, with the depletion of CO and CH4 being slightly accelerated with respect to the results without chemical reactions through their processing to CO2 and larger hydrocarbon ices, respectively. In the α = 10−2 model the evolution is similar both with and without radial drift. This is due to the viscous time-scale (tν) and radial drift time-scale (tdrift) being com- parable, so radial drift only modestly affects the transport of dust. Furthermore, both of the transport time-scales are shorter than the chemical time-scales for the depletion of the major carbon and oxygen carriers. At α = 10−2 the impact of radial drift is to increase the abundances by a factor ∼ 3 due to the radial drift of ices. The enhancement in molecu- lar abundances is only slightly higher than the enhancement in dust-to-gas ratio because the dust drifts inwards at only slightly higher speeds than the gas. However, suppressing the influence of radial drift entirely would require tν (cid:28) tdrift. Similar to the α = 10−2 models without radial drift, we again see that gas-phase reactions only have a modest effect on the abundances, although the abundance of hydrocarbons pro- duced from CH4 still reaches 20 per cent of CH4 abundance. In Figure 5 we show how transport influences the evo- lution of the main nitrogen reservoirs, N2 and NH3. In the outer disc we find that the evolution is slow(time-scales (cid:29) 1 Myr), even in the absence of transport. We find that N2 converts to NH3, as found by Schwarz & Bergin (2014) in models with a high initial N2 abundance. Conversely, Eistrup et al. (2016) report the conversion of NH3 to N2 in the outer disc, again on time-scales much longer than 1 Myr. The difference is due to the destruction of NH3 ice by cosmic-ray induced photons, which is neglected in this work. Nevertheless, the evolution of the nitrogen abundance in the outer disc is expected to be slow. In the inner disc NH3 is depleted and the N2 abundance increases. The remaining NH3 abundance inside the NH3 ice line is sensitive to the rate of NH3 production from N2. We find that this is fastest via the pathway N2 γCR→ N+ H2→ NH+ H2→ NH+ 2 H2→ NH+ 3 H2O→ NH+ 4 e−→ NH3. (13) Outside the water ice line where water is not present in the gas phase the NH+ 4 step occurs more slowly via reactions with H2, resulting in lower abundances. In the colder regions further out NH3 formation can also occur on grain surfaces. 3 → NH+ As in the case of carbon and oxygen bearing species, we see that even low levels of transport prevent the depletion of NH3, largely because the depletion time-scale, ∼ 0.5 M yr, is long compared to the transport time-scales in the inner disc. We note that high NH3 abundances inside the NH3 ice line is in conflict with recent measurements of the NH3 abun- dance in the inner regions of discs (< 10−7; Pontoppidan et al. 2019), even for α = 10−3. However, it may be possible that the NH3 abundance observed in the upper layers is not representative of the bulk abundance on the disc. A simi- lar conflict between observed CO2 abundances and models including transport was reported by Bosman et al. (2018a). Neither the conversion of N2 to NH3 nor the reverse pro- cess change the total gas phase nitrogen abundance signifi- cantly -- in the inner disc both species are in the gas phase, MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 9 Figure 5. Abundances relative to hydrogen of the key nitrogen bearing molecules. The models are the same as Figure 3 and Figure 4. Solid lines denote gas phase species, dashed lines denote ice phase species. Top: models with viscous transport only. Bottom: models including radial drift. while in the outer disc the destruction of N2 is slow. How- ever, due to the long viscous time in the outer disc (1 Myr at 50 au for α = 10−2), transport only has a modest effect on the evolution of nitrogen in the outer disc. Conversely, in the inner disc where the viscous time is much shorter, transport reduces the conversion of NH3 to N2. Models with radial drift also have a similar effect on the nitrogen reservoirs as the carbon and oxygen reservoirs. Radial drift causes enhancements in the gas phase N2 abun- dance inside the N2 ice line by a factor 2 -- 3. The NH3 abundance increases by factors of 3 -- 10 due to radial drift and is further enhanced by the formation of extra NH3 in the outer disc, which freezes out and is carried in by the drifting grains. We briefly consider how sensitive the results presented here are to assumptions about the disc model. Notably, the radial drift and viscous time-scales are not very sensitive to model assumptions, apart from the dependence on α. For this reason, we investigated how the chemical time-scales vary in models without transport, varying the disc mass, cosmic-ray ionization rate and average grain size used in the chemical models. The results are presented in detail in Ap- MNRAS 000, 1 -- 14 (2019) pendix A, which shows that the time-scale for the depletion of CH4 and CO are only sensitive to the cosmic-ray ioniza- tion rate. Since we adopt the canonical value of cosmic-ray ionization rate (10−17 s−1), which assumes no shielding of the interstellar cosmic ray flux via stellar winds or magnetic fields (Cleeves et al. 2013a), our results likely represent a lower limit of the importance of transport relative to mid- plane chemical kinetics. 3.4 Abundance ratios The bulk abundances of the main atomic species, e.g. car- bon, oxygen, and nitrogen, are of particular interest in the context of planet formation since the composition of exo- planet atmospheres must be connected to the competition of gas and solids in discs (e.g. Oberg et al. 2011; Madhusudhan et al. 2014). We first explore the evolution of the C/O ratio in models without radial drift (Figure 6). In these models, the total abundance (gas and ice phase) of carbon and oxy- gen bearing species does not change, only the partitioning of these species between the gas and ice phases. The changes in composition driven by chemical pro- 10-1100101102R [au]10-810-710-610-510-410-3abundanceno transport100101102R [au]®=10¡3100101102R [au]®=10¡2N2NH310-1100101102R [au]10-810-710-610-510-410-3abundanceno chem100101102R [au]®=10¡3100101102R [au]®=10¡2N2NH310 Booth & Ilee Figure 6. Evolution of the C/O ratio for models with viscous transport, but without radial drift or diffusion. For comparison the left hand panel shows the evolution without transport. The solid lines denote the gas phase abundances, while the dotted lines show the ice abundances. Figure 7. C/O ratio and carbon abundance evolution for models with radial drift. The carbon abundance shown is relative to the initial total carbon abundance (1.38 × 10−4, 0.3 × solar). Note the different scales for the C/O ratio between the left and right panels. MNRAS 000, 1 -- 14 (2019) 10-1100101102R[au]0.20.40.60.81.01.21.4[C=O]solarno transport10-1100101102R[au]solar®=10¡310-1100101102R[au]solar®=10¡20.05 Myr0.1 Myr0.5 Myr1.0 Myr10-1100101102R[au]10-1100101Relative carbon abundanceno chem10-1100101102R[au]10-1100101®=10¡310-1100101102R[au]10-1100101®=10¡20.05 Myr0.1 Myr0.5 Myr1.0 Myr10-1100101102R[au]01234567[C=O]solarno chem10-1100101102R[au]01234567solar®=10¡310-1100101102R[au]0.20.40.60.81.01.21.41.6solar®=10¡20.05 Myr0.1 Myr0.5 Myr1.0 Myrcesses are best seen in the model without transport (and α = 10−3). Inside the water ice line (0.5 au), all of the major carbon and oxygen are in the gas phase and thus chemical processing does not change the gas phase carbon or oxygen abundance. Between the water and CO2 ice lines (1.5 au) the gas phase carbon and oxygen abundance is dominated by CO and CO2, with a small contribution from hydrocarbons (CH4 and C2H2 -- a proxy for C2H6; see subsection 3.1). Here, the C/O ratio increases as CO2 is destroyed, producing CO and O, with O eventually forming H2O, which freezes out. Be- tween the CO2 and CH4 ice lines (10 au), CO and CH4 are initially the main gas phase carbon and oxygen carriers. CH4 gets converted to larger hydrocarbons (C2H2), which freeze out causing the C/O ratio to reduce to 1. The spike in C/O ratio near the CO2 ice line is due to the slightly lower bind- ing energy of C2H2 than CO2 (note that C2H6 also has a lower binding energy than CO2, so Eistrup et al. 2018 see a similar spike in C/O). Outside of the CH4 ice line CO is al- ways the dominant the gas phase carbon and oxygen carrier, so the C/O ratio does not evolve. The nitrogen abundance evolution can be derived from Figure 5 and is is slow, thus the C/N of N/O ratio evolution will be driven primarly by the evolution of carbon and oxygen, respectively. As the speed of the viscous transport increases, the changes in composition driven by chemical reactions become restricted to regions of the disc at larger and larger radii. Al- ready at α = 10−3 the effect of transport is first noticeable between the H2O and CO2 ice lines, since CO2 no longer de- pletes (see also Bosman et al. 2018a), while at α = 10−2 the carbon and oxygen abundance hardly evolve. At α = 10−2 the inward movement of the ice lines as the disc cools gen- erates the only significant changes in abundance ratio. The influence of radial drift on the bulk abundances is sensitive to whether radial drift or viscous evolution is faster. When radial drift is included we see spikes in the gas phase C/H ratio at α = 10−3 (Figure 7), as reported by Booth et al. (2017). These spikes are driven by volatile species being left at their snow lines when grains drift past them. We also see a peak in the C/O ratio inside of 10 au, which is driven by CH4 entering the gas phase. Since radial drift depletes the disc of dust (and ice phase species) within a few 0.1 Myr the C/O ratio at 10 au begins to decline. This decline is due to a combination of the conversion of CH4 to C2H2 and the viscous transport of CH4 inwards from the ice line. After ∼ 0.5 Myr, conversion of CH4 to C2H2 generates a second spike in C/O ratio at a few au (where C2H2 enters the gas phase). At α = 10−2 the viscous and radial drift times are comparable, reducing the influence of radial drift. Nevertheless, we still find an enhancement in the gas phase C/H ratio by a factor 2 -- 3, since radial drift is still able to bring extra material into the inner regions. We again find C/O > 1 between the CH4 and CO2 snow lines since CH4 is transported into this region faster than it is converted to C2H2. 4 DISCUSSION 4.1 Impact of transport on disc composition In this paper we have investigated how the abundances of the main carbon, oxygen, and nitrogen carriers in the mid- MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 11 plane of protoplanetary discs evolve under the competition between chemical kinetics and radial transport. The effects of transport differ between gas dominated transport (viscous evolution, and the transport of small grains at the viscous speed) and radial drift dominated transport. The main effect of gas dominated transport is to reduce the amount of time that a parcel of gas spends in regions where the processing occurs. Thus gas transport weakens the depletion of CO2, CH4 and CO from the gas phase that otherwise occurs inside their ice lines. Radial drift acts differently, enhancing the transport of molecular species inwards on grain surfaces. The molecules are then left at in the gas phase as they sublimate at their ice lines. Thus the main effect of radial drift is to enhance the abundances of molecular species inside their ice lines. Their subsequent evolution is controlled by transport further inward in the gas phase, along with processing by chemical reactions. A secondary effect of radial drift is that new species produced by chemical processing (e.g. C2H2 or H2CO) can freeze out and be carried further in before re- entering the gas phase. The results presented here are relatively insensitive to the parameters in the model, except those that govern the rate of transport. In Appendix A we explore how the chemi- cal depletion time-scales depend on model parameters in the absence of transport, showing that the results are only sen- sitive to the cosmic-ray ionization rate, ζ, which facilitates the removal of CH4 and CO from the gas phase. If ζ is lower than the canonical value of 10−17 (as suggested by Cleeves et al. 2013a), then transport is expected to dominate for α (cid:38) 10−3, which is likely the case for most discs. Our results are otherwise insensitive to the disc mass or assumptions about the grain size distribution. This means that α and ζ essentially control the relative importance of transport and chemical reactions. There will also be a weak dependence on the temperature of the disc as the ice lines move to larger radius where the viscous time is longer. The models presented here support the ideas put for- ward in Booth et al. (2017), using similar assumptions for the transport of chemical species but simpler assump- tions about their abundances. I.e. in models with efficient grain growth and radial drift the abundances inside ice lines are mostly controlled by the inward transport of molecu- lar species from larger distances on grain surfaces, together with transport in the gas phase inside the ice line. Reduc- ing the grain sizes weakens radial drift and results in the compositions being set largely by the composition at larger radii. The exception to this is when transport times are long ((cid:29) 1Myr at 10 au) and ionization rates are high (ζ (cid:38) 10−17), in which case chemical processing in the mid-plane may be able to remove CH4 (and eventually CO) from the gas phase. We also verify the suggestion that, by processing molecules to less volatile species (such as CH4 to C2H2), chemical re- actions act to aid the transport of volatile species to smaller radii (Figure 7). The models presented here assume a turbulent viscos- ity, but it is possible that accretion in protoplanetary discs is primarily laminar, as discussed in subsection 2.2. However, a low level of radial diffusion associated with weak turbulence would not affect the results presented here because the ef- fects of radial diffusion are already localised to regions close to the snow lines. The primary effect of weak turbulence on 12 Booth & Ilee the models is thus on grain growth and radial drift, which is limited by turbulent fragmentation in our α = 0.01 model. Lowering the strength of turbulence in that model by a fac- tor 10 would result in efficient grain growth and radial drift, as in the α = 10−3 model. Reducing the strength of turbu- lence further would have little effect because grain growth becomes limited by radial drift instead. A potentially more important effect associated with laminar accretion in discs is the possibility that the accretion occurs in a narrow layer far away from the mid-plane (e.g. Gressel et al. 2015). In this case, vertical mixing is needed to couple the effects of accretion to the composition of the mid-plane. Two-dimensional models are needed to explore this in detail, but it is unlikely that the accretion flow can be completely decoupled from the mid-plane because verti- cal mixing should occur a factor of (R/H)2 ∼ 102 -- 103 times faster than the radial mixing for the same α. Thus an ex- traordinarily low α would be required to prevent vertical mixing altogether. Dust traps, if widely present in protoplanetary discs, could have a significant impact on the abundance evolution. Such traps have typically been invoked to explain why dust grains can survive in discs for several Myr (e.g. Pinilla et al. 2012). The impact of traps on the composition will depend on the evolution of the traps themselves. If the traps were to move with the gas, then the net effect would be similar to our models neglecting radial drift. However, if the traps remain stationary, preventing the inflow of dust entirely, the chem- ical evolution could be different as volatiles remain trapped at their original location. This would lead to the depletion of volatile species from the gas phase as the gas in the inner disc is accreted onto the star. This effect has been invoked to explain the depletion of refractory material accreting onto Herbig stars with transition discs (Kama et al. 2015). Although the high CO abundance in our models ap- pears in conflict with the observed depletion of CO in TW Hydra (Favre et al. 2013; Zhang et al. 2017), this can poten- tially be reconciled due to TW Hydra's age (8 Myr) and low accretion rate (Calvet et al. 2002; Donaldson et al. 2016), as long as the ionization rate is not too low (ζ ∼ 10−17). In this case the viscous time-scale near the CO ice line is ∼ 5 Myr, long enough that grain surface processes could de- plete the CO abundance (requiring ∼ 3 Myr, Bosman et al. 2018b). The low CO line fluxes observed for a range of discs (e.g. those in Lupus; Miotello et al. 2017) point to a more general depletion of CO within a few Myr. However, for sys- tems requiring a higher α (e.g. HD 163296; Boneberg et al. 2016) or low ionization rate (Cleeves et al. 2013a), grain sur- face processes are unlikely to be able to deplete CO rapidly enough. bidelli et al. 2015; Bitsch et al. 2015, 2019; Ida et al. 2016). Planets forming in these models will have super-solar abun- dances unless they accrete their envelopes beyond the CO ice line (Booth et al. 2017). Our results suggest that these planets will likely also have C/O > 1 and possibly C/O (cid:29) 1 if they accrete their envelopes inside the CH4 ice line (which is the case in the models of Bitsch et al. 2019, for exam- ple). Recently, Johansen et al. (2018) suggested that giant planet formation via pebble accretion can occur for more modest pebble sizes, as long as the pebbles have settled to the mid-plane. Such models will be associated with more modest enhancements in the abundances, more similar to our α = 10−2 model (including radial drift). 5 CONCLUSIONS We have explored the competition between chemical reac- tions in the mid-plane and transport due to viscous evolu- tion and radial drift on the composition of protoplanetary discs. Our models show that the radial transport associated with accretion is able to transport material faster than it can be depleted due to chemical processing if the viscous α parameter is high (α ∼ 10−2, Figure 3) or the ionization rate is low (ζ ∼ 10−18, Figure A2). For α = 10−3 and ζ = 10−17, transport will still overcome the depletion of CO2 from the gas phase inside its ice line, but CO and CH4 will remain depleted. Including radial drift enhances the gas phase abun- dances of the dominant molecular species inside their ice lines (Figure 4). The strength of enhancement depends on how much faster radial drift is than transport in the gas. For commonly assumed parameters in models of dust evolu- tion and planet formation pebble accretion (Birnstiel et al. 2012; Lambrechts & Johansen 2014; Bitsch et al. 2015, e.g.), this leads to factor ∼ 10 enhancements in the abundances. In models with more modest radial drift (grain sizes in the range 0.1 -- 1 mm), radial drift can still lead to a factor 2 -- 3 enhancement in the abundances inside 10 au. In the coming years, the joint operation of the James Webb Space Telescope (JWST) and the Atacama Large Millimetre/submillimetre Array (ALMA) will enable robust measurements of the elemental composition of hot Jupiter atmospheres (such as C/O) alongside measurements of the chemical composition of planet forming discs. Our results demonstrate the need to consider the interplay between both physical and chemical processes during protoplanetary disc evolution when interpretting such observations, with impli- cations for the connection of planet composition to the com- position of the disc from which they formed. 4.2 Implications for planet formation via pebble ACKNOWLEDGEMENTS accretion The large difference in the gas-phase abundances in mod- els with and without efficient radial drift (e.g. models with α = 10−3 and 10−2, respectively) suggests that exoplanet composition will be a powerful way to constrain planet for- mation by pebble accretion. Most models of giant planet formation by via pebble accretion assume large pebble sizes that drift efficiently (Lambrechts & Johansen 2014; Mor- We would like to thank Catherine Walsh, Cathie Clarke, and Mihkel Kama for useful discussions and careful read- ing of the manuscript. We additionally thank the referee, Alexander Cridland, for helpful suggestions which have im- proved this work. RAB and JDI gratefully acknowledge sup- port from the DISCSIM project, grant agreement 341137 under ERC-2013-ADG. JDI also acknowledges support from the STFC under ST/R000549/1. This project has made use MNRAS 000, 1 -- 14 (2019) of the SciPy stack (Jones et al. 2001), including NumPy (Oliphant 2007) and Matplotlib (Hunter 2007). REFERENCES Aikawa Y., Umebayashi T., Nakano T., Miyama S. M., 1999, ApJ, 519, 705 Ali-Dib M., 2017, MNRAS, 464, 4282 Andrews S. M., Williams J. P., 2007, ApJ, 659, 705 Ansdell M., et al., 2016, ApJ, 828, 46 Asplund M., Grevesse N., Sauval A. J., Scott P., 2009, Annual Review of Astronomy and Astrophysics, 47, 481 Bai X.-N., Stone J. M., 2013, ApJ, 769, 76 Bell K. R., Lin D. N. C., 1994, ApJ, 427, 987 Bergin E. A., et al., 2013, Nature, 493, 644 Bergin E. A., Du F., Cleeves L. I., Blake G. A., Schwarz K., Visser R., Zhang K., 2016, ApJ, 831, 101 Birnstiel T., Klahr H., Ercolano B., 2012, A&A, 539, A148 Bitsch B., Lambrechts M., Johansen A., 2015, A&A, 582, A112 Bitsch B., Izidoro A., Johansen A., Raymond S. N., Morbidelli A., Lambrechts M., Jacobson S. A., 2019, A&A, 623, A88 Boneberg D. M., Pani´c O., Haworth T. J., Clarke C. J., Min M., 2016, MNRAS, 461, 385 Booth R. A., Clarke C. J., Madhusudhan N., Ilee J. D., 2017, MNRAS, 469, 3994 Bosman A. D., Tielens A. G. G. M., van Dishoeck E. F., 2018a, A&A, 611, A80 Bosman A. D., Walsh C., van Dishoeck E. F., 2018b, A&A, 618, A182 Calvet N., D'Alessio P., Hartmann L., Wilner D., Walsh A., Sitko M., 2002, ApJ, 568, 1008 Clarke C. J., et al., 2018, ApJ, 866, L6 Cleeves L. I., Adams F. C., Bergin E. A., 2013a, ApJ, 772, 5 Cleeves L. I., Adams F. C., Bergin E. A., Visser R., 2013b, ApJ, 777, 28 Collings M. P., Anderson M. A., Chen R., Dever J. W., Viti S., Williams D. A., McCoustra M. R. S., 2004, MNRAS, 354, 1133 Cridland A. J., Pudritz R. E., Alessi M., 2016, MNRAS, 461, 3274 Cridland A. J., Pudritz R. E., Birnstiel T., Cleeves L. I., Bergin E. A., 2017, MNRAS, 469, 3910 Donaldson J. K., Weinberger A. J., Gagn´e J., Faherty J. K., Boss A. P., Keiser S. A., 2016, ApJ, 833, 95 Eistrup C., Walsh C., van Dishoeck E. F., 2016, A&A, 595, A83 Eistrup C., Walsh C., van Dishoeck E. F., 2018, A&A, 613, A14 Ercolano B., Rosotti G. P., Picogna G., Testi L., 2017, MNRAS, 464, L95 Evans M. G., Ilee J. D., Boley A. C., Caselli P., Durisen R. H., Hartquist T. W., Rawlings J. M. C., 2015, MNRAS, 453, 1147 Favre C., Cleeves L. I., Bergin E. A., Qi C., Blake G. A., 2013, ApJ, 776, L38 Flaherty K. M., et al., 2017, ApJ, 843, 150 Flaherty K. M., Hughes A. M., Teague R., Simon J. B., Andrews S. M., Wilner D. J., 2018, ApJ, 856, 117 Fromang S., Latter H., Lesur G., Ogilvie G. I., 2013, A&A, 552, A71 Gail H.-P., Trieloff M., 2017, A&A, 606, A16 Grassi T., Bovino S., Schleicher D. R. G., Prieto J., Seifried D., Simoncini E., Gianturco F. A., 2014, MNRAS, 439, 2386 Greaves J. S., Rice W. K. M., 2010, MNRAS, 407, 1981 Gressel O., Turner N. J., Nelson R. P., McNally C. P., 2015, ApJ, 801, 84 Guilloteau S., Dutrey A., Pi´etu V., Boehler Y., 2011, A&A, 529, A105 Heays A. N., Bosman A. D., van Dishoeck E. F., 2017, A&A, 602, A105 MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 13 Heinzeller D., Nomura H., Walsh C., Millar T. J., 2011, ApJ, 731, 115 Henning T., Semenov D., 2013, Chemical Reviews, 113, 9016 Hindmarsh A. C., 1983, IMACS Transactions on Scientific Com- putation, 1, 55 Hunter J. D., 2007, Computing in Science Engineering, 9, 90 Ida S., Guillot T., Morbidelli A., 2016, A&A, 591, A72 Ilee J. D., Boley A. C., Caselli P., Durisen R. H., Hartquist T. W., Rawlings J. M. C., 2011, MNRAS, 417, 2950 Ilee J. D., et al., 2017, MNRAS, 472, 189 Johansen A., Ida S., Brasser R., 2018, arXiv e-prints, p. arXiv:1811.00523 Jones E., Oliphant T., Peterson P., et al., 2001, SciPy: Open source scientific tools for Python, http://www.scipy.org/ Kama M., Folsom C. P., Pinilla P., 2015, A&A, 582, L10 Kama M., et al., 2016, A&A, 592, A83 Klarmann L., Ormel C. W., Dominik C., 2018, A&A, 618, L1 Krijt S., Schwarz K. R., Bergin E. A., Ciesla F. J., 2018, ApJ, 864, 78 Lambrechts M., Johansen A., 2012, A&A, 544, A32 Lambrechts M., Johansen A., 2014, A&A, 572, A107 Lodato G., Scardoni C. E., Manara C. F., Testi L., 2017, MNRAS, 472, 4700 Lodders K., 2010, Exoplanet Chemistry. Wiley, p. 157, doi:10.1002/9783527629763.ch8 Lynden-Bell D., Pringle J. E., 1974, MNRAS, 168, 603 Madhusudhan N., 2012, ApJ, 758, 36 Madhusudhan N., Amin M. A., Kennedy G. M., 2014, ApJ, 794, L12 Madhusudhan N., Bitsch B., Johansen A., Eriksson L., 2017, MN- RAS, 469, 4102 Manara C. F., et al., 2016, A&A, 591, L3 Manara C. F., Morbidelli A., Guillot T., 2018, A&A, 618, L3 Millar T. J., Farquhar P. R. A., Willacy K., 1997, A&AS, 121, 139 Miotello A., et al., 2017, A&A, 599, A113 Morbidelli A., Lambrechts M., Jacobson S., Bitsch B., 2015, Icarus, 258, 418 Mordasini C., van Boekel R., Molli`ere P., Henning T., Benneke B., 2016, ApJ, 832, 41 Mulders G. D., Pascucci I., Manara C. F., Testi L., Herczeg G. J., Henning T., Mohanty S., Lodato G., 2017, ApJ, 847, 31 Mumma M. J., Charnley S. B., 2011, Annual Review of Astron- omy and Astrophysics, 49, 471 Najita J. R., Kenyon S. J., 2014, MNRAS, 445, 3315 Nomura H., Aikawa Y., Nakagawa Y., Millar T. J., 2009, A&A, 495, 183 Oberg K. I., Garrod R. T., van Dishoeck E. F., Linnartz H., 2009, A&A, 504, 891 Oberg K. I., Murray-Clay R., Bergin E. A., 2011, ApJ, 743, L16 Oliphant T. E., 2007, Computing in Science Engineering, 9, 10 Ormel C. W., Klahr H. H., 2010, A&A, 520, A43 Penteado E. M., Walsh C., Cuppen H. M., 2017, ApJ, 844, 71 Pinilla P., Birnstiel T., Ricci L., Dullemond C. P., Uribe A. L., Testi L., Natta A., 2012, A&A, 538, A114 Pinte C., Dent W. R. F., M´enard F., Hales A., Hill T., Cortes P., de Gregorio-Monsalvo I., 2016, ApJ, 816, 25 Piso A.-M. A., Oberg K. I., Birnstiel T., Murray-Clay R. A., 2015, ApJ, 815, 109 Piso A.-M. A., Pegues J., Oberg K. I., 2016, ApJ, 833, 203 Pollack J. B., Hollenbach D., Beckwith S., Simonelli D. P., Roush T., Fong W., 1994, ApJ, 421, 615 Pontoppidan K. M., Salyk C., Banzatti A., Blake G. A., Walsh C., Lacy J. H., Richter M. J., 2019, arXiv e-prints, p. arXiv:1902.03647 Rafikov R. R., 2017, ApJ, 837, 163 Schwarz K. R., Bergin E. A., 2014, ApJ, 797, 113 14 Booth & Ilee Semenov D., Wiebe D., 2011, The Astrophysical Journal Supple- ment Series, 196, 25 Stammler S. M., Birnstiel T., Pani´c O., Dullemond C. P., Dominik C., 2017, A&A, 600, A140 Suzuki T. K., Inutsuka S.-i., 2009, ApJ, 691, L49 Takeuchi T., Lin D. N. C., 2002, ApJ, 581, 1344 Tazzari M., et al., 2016, A&A, 588, A53 Teague R., et al., 2018, ApJ, 864, 133 Tscharnuter W. M., Gail H. P., 2007, A&A, 463, 369 Visser R., Doty S. D., van Dishoeck E. F., 2011, A&A, 534, A132 Walsh C., Herbst E., Nomura H., Millar T. J., Weaver S. W., 2014, Faraday Discussions, 168, 389 Walsh C., Nomura H., van Dishoeck E., 2015, A&A, 582, A88 Weidenschilling S. J., 1977, MNRAS, 180, 57 Youdin A. N., Lithwick Y., 2007, Icarus, 192, 588 Yu M., Willacy K., Dodson-Robinson S. E., Turner N. J., Evans Neal J. I., 2016, ApJ, 822, 53 Zhang K., Bergin E. A., Blake G. A., Cleeves L. I., Schwarz K. R., 2017, Nature Astronomy, 1, 0130 Zhu Z., Hartmann L., Nelson R. P., Gammie C. F., 2012, ApJ, 746, 110 APPENDIX A: DEPENDENCE ON MODEL PARAMETERS Here we investigate how the conclusions of our study might depend on the assumed parameters of the model. In partic- ular we vary the cosmic-ray ionization rate, the disc mass, and the average grain size used in the chemical models. For brevity, we report only the results of models neglect- ing transport. The cosmic-ray ionization rate, ζ, is the most impor- tant parameter. Figure A1 shows that the depletion of CH4 and CO from the gas phase are sensitive to this parameter. Reducing the cosmic-ray ionization rate to ζ = 10−18 s−1 is sufficient to significantly reduce the depletion of CH4. Nev- ertheless, CH4 is still depleted by a factor ∼ 10 within a Myr, which is fast enough to compete with viscous trans- port at α = 10−3, but not α = 10−2. Thus our suggestion that mid-plane chemical kinetics will have a significant im- pact at low accretions remains true. Even lower ionization rates would slow the depletion of CH4 further. However, short-lived radioactive nuclei are expected to prevent ion- ization rates falling much below 10−18 s−1 (e.g. Cleeves et al. 2013b). Higher cosmic-ray ionization rates would lead to qual- itatively different conclusions. At α = 10−3 and below, the depletion of CO from the gas phase would now become sig- nificant within 1 Myr. Furthermore, the time-scale for the depletion of CH4 reduces to ∼ 0.1 Myr (Figure A2), which is now comparable with the viscous transport time-scale at α = 10−2. Under such conditions transport is unlikely to strongly suppress the depletion of CO and CH4. However, we note that such high ionization rates are not typically expected for T Tauri star discs, instead the cosmic-ray ion- ization rates are mpre likely lower than the canonical value of 10−17 s−1 (Cleeves et al. 2013a). In addition to varying the ionization rate, we conducted models in which the disc mass has been varied. We find that the surface density in the disc does not greatly affect the composition, as can be seen by comparing the models with Md = 0.001 M(cid:12) and 0.01M(cid:12). At Md = 0.1 M(cid:12) we see differ- ences, but these are due to the higher temperature in the disc, rather than higher density. Since these differences are limited to within a few au from the star where the transport time-scales are short, we conclude that disc mass does not much affect our conclusions. Finally, we tested the effects of our assumptions about the grain size distribution. In the main text we compute the average grain area assuming a grain size distribution with n(a)da ∝ a−3.5da between 0.1µm and amax, where amax is set by the dust evolution model. To test this assumption we compare our results with a model where the grain size used in the gas-grain reactions is fixed at 0.1µm. Figure A1 shows that the evolution is largely insensitive to the average grain area. This paper has been typeset from a TEX/LATEX file prepared by the author. MNRAS 000, 1 -- 14 (2019) Chemistry & transport in discs 15 Figure A1. Dependence of the abundance of the main carbon and oxygen carriers on the disc properties in models without transport. Top: Models with varying cosmic-ray ionization rate, but fixed disc mass, Md = 0.01 M(cid:12): Bottom Left: A model where the grain size has been fixed at 0.1 µm in the chemical model. Bottom Middle and Right: Models with varying disc mass. Figure A2. Time evolution of the CO, CH4 and N2 abundance at 5 au in models with varying cosmic-ray ionization rates. MNRAS 000, 1 -- 14 (2019) 10-1100101102R [au]10-810-710-610-510-410-3abundance³=10¡18COCO2CH4H2OC2H2100101102R [au]³=10¡17100101102R [au]³=10¡1610-1100101102R [au]10-810-710-610-510-410-3abundance-a2®1=2=0:1¹m100101102R [au]0:001M¯100101102R [au]0:1M¯0.00.20.40.60.81.0time [Myr]10-910-810-710-610-510-4abundanceCO³=10¡18CH4³=10¡17N2³=10¡16
1508.06215	1	1508	2015-08-25T16:33:49	Ground-based transit observations of the HAT-P-18, HAT-P-19, HAT-P-27/WASP-40 and WASP-21 systems	[ "astro-ph.EP", "astro-ph.SR" ]	As part of our ongoing effort to investigate transit timing variations (TTVs) of known exoplanets, we monitored transits of the four exoplanets HAT-P-18b, HAT-P-19b, HAT-P-27b/WASP-40b and WASP-21b. All of them are suspected to show TTVs due to the known properties of their host systems based on the respective discovery papers. During the past three years 46 transit observations were carried out, mostly using telescopes of the Young Exoplanet Transit Initiative. The analyses are used to refine the systems orbital parameters. In all cases we found no hints for significant TTVs, or changes in the system parameters inclination, fractional stellar radius and planet to star radius ratio. However, comparing our results with those available in the literature shows that we can confirm the already published values.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 13 (2013) Printed 6th October 2018 (MN LATEX style file v2.2) Ground-based transit observations of the HAT-P-18, HAT-P-19, HAT-P-27/WASP40 and WASP-21 systems 5 1 0 2 g u A 5 2 . ] P E h p - o r t s a [ 1 v 5 1 2 6 0 . 8 0 5 1 : v i X r a M. Seeliger,1⋆ M. Kitze,1 R. Errmann,1,2 S. Richter,1 J. M. Ohlert,3,4 W. P. Chen,5 J. K. Guo,5 E. Gogu¸s,6 T. Guver,7 B. Aydın,6 S. Mottola,8 S. Hellmich,8 M. Fernandez,9 F. J. Aceituno,9 D. Dimitrov,10 D. Kjurkchieva,11 E. Jensen,12 D. Cohen,12 E. Kundra,13 T. Pribulla,13 M. Vanko,13 J. Budaj,14,13 M. Mallonn,15 Z.-Y. Wu,16 X. Zhou,16 St. Raetz,1,17 C. Adam,1 T. O. B. Schmidt,1 A. Ide,1 M. Mugrauer,1 L. Marschall,19 M. Hackstein,20 R. Chini,20,21 M. Haas,20 T. Ak,7 E. Guzel,22 A. Ozdonmez,23 C. Ginski,1,24 C. Marka,1 J. G. Schmidt,1 B. Dincel,1 K. Werner,1 A. Dathe,1 J. Greif,1 V. Wolf,1 S. Buder,1 A. Pannicke,1 D. Puchalski25 and R. Neuhauser1 1Astrophysical Institute and University Observatory Jena, Schillergasschen 2-3, D-07745 Jena, Germany 2Abbe Center of Photonics, Friedrich Schiller Universitat, Max-Wien-Platz 1, D-07743 Jena, Germany 3Astronomie Stiftung Trebur, Michael Adrian Observatorium, Fichtenstrasse 7, D-65468 Trebur, Germany 4University of Applied Sciences, Technische Hochschule Mittelhessen, D-61169 Friedberg, Germany 5Graduate Institute of Astronomy, National Central University, Jhongli City, Taoyuan County 32001, Taiwan (R.O.C.) 6Sabanci University, Orhanli-Tuzla 34956, Istanbul, Turkey 7Faculty of Sciences, Department of Astronomy and Space Sciences, Istanbul University, 34119 Istanbul, Turkey 8Deutsches Zentrum fur Luft- und Raumfahrt e.V., Institut fur Planetenforschung, Rutherfordstr. 2, D-12489 Berlin, Germany 9Instituto de Astrofisica de Andalucia, CSIC, Apdo. 3004, E-18080 Granada, Spain 10Institute of Astronomy and NAO, Bulgarian Academy of Sciences, 72 Tsarigradsko Chaussee Blvd., 1784 Sofia, Bulgaria 11Shumen University, 115 Universitetska str., 9700 Shumen, Bulgaria 12Department of Physics and Astronomy, Swarthmore College, Swarthmore, PA 19081, USA 13Astronomical Institute, Slovak Academy of Sciences, 059 60, Tatransk´a Lomnica, Slovakia 14Research School of Astronomy and Astrophysics, Australian National University, Canberra, ACT 2611, Australia 15Leibnitz-Institut fur Astrophysik Potsdam, An der Sternwarte 16, D-14482 Potsdam, Germany 16Key Laboratory of Optical Astronomy, NAO, Chinese Academy of Sciences, 20A Datun Road, Beijing 100012, China 17European Space Agency, ESTEC, SRE-S, Keplerlaan 1, 2201AZ Noordwijk, The Netherlands 19Gettysburg College Observatory, Department of Physics, 300 North Washington St., Gettysburg, PA 17325, USA 20Astronomisches Institut, Ruhr-Universitat Bochum, Universitatsstrasse 150, D-44780 Bochum, Germany 21Instituto de Astronom´ıa, Universidad Cat´olica del Norte, Avenida Angamos 0610, Casilla 1280 Antofagasta, Chile 22Faculty of Science, Department of Astronomy and Space Sciences, Univserity of Ege, Bornova, 35100 Izmir, Turkey 23Graduate School of Science and Engineering, Department of Astronomy and Space Sciences, Istanbul University, 34116 Istanbul, Turkey 24Sterrewacht Leiden, P.O. Box 9513, Niels Bohrweg 2, 2300RA Leiden, The Netherlands 25Centre for Astronomy, Faculty of Physics, Astronomy and Informatics, N. Copernicus University, Grudziadzka 5, PL-87-100 Toru´n, Poland ABSTRACT As part of our ongoing effort to investigate transit timing variations (TTVs) of known exoplanets, we monitored transits of the four exoplanets HAT-P-18b, HAT-P-19b, HAT-P-27b/WASP-40b and WASP-21b. All of them are suspected to show TTVs due to the known properties of their host systems based on the respective discovery papers. During the past three years 42 transit observations were carried out, mostly using telescopes of the Young Exoplanet Transit Initiative. The analyses are used to refine the systems orbital parameters. In all cases we found no hints for significant TTVs, or changes in the system parameters inclination, fractional stellar radius and planet to star radius ratio and thus could confirm the already published results. Key words: planets and satellites: individual: HAT-P-18b, HAT-P-19b, HAT-P- 27b/WASP-40b, WASP-21b 2 M. Seeliger et al. 1 INTRODUCTION Observing extra solar planets transiting their host stars has become an important tool for planet detection and is used to obtain and constrain fundamental system parameters: The inclination has to be close to 90◦, while the planet to star radius ratio is constrained mainly by the transit depth. Also, in combination with spectroscopy, the semimajor axis and the absolute planet and star radii can be obtained as well. Several years ago, when the first results of the Kepler mission were published (see Borucki et al. 2011 for first sci- entific results, Koch et al. 2010 for an instrument descrip- tion), studying the transit timing became one of the stand- ard techniques in the analysis of transit observations. Since space based missions are able to observe many consecut- ive transit events with a high precision, one can detect even small variations of the transit intervals indicating deviations from a strictly Keplerian motion and thus yet hidden planets in the observed system. Furthermore, with the discovery of multi-planetary systems, transit timing variations (TTVs) are used to find the masses of the companions without the need of radial velocity measurements due to the influence of planetary interaction on TTVs. Since many planet can- didates found in photometric surveys are too faint for radial velocity follow-up even with bigger telescopes, TTV analyses can be considered as a photometric work-around to estimate masses. Although the existence of TTVs can be shown in already known exoplanetary systems, only a few additional planet candidates have been found using TTVs so far (for recent examples of a proposed additional body indicated by TTV analyses see e.g. Maciejewski et al. 2011a and Van Eylen et al. 2014). This is not suprising, since large bodies often can be found using radial velocity measure- ments or direct transit detections, while small (e.g. Earth- like) objects result in small TTV amplitudes and there- fore high precision timing measurements are needed. These measurements can already be acquired with medium size ground-based telescopes. Commonly the transit mid-time of each observation is plotted into an observed minus calculated (O -- C) diagram (Ford & Holman 2007), where the difference between the observed transit mid-time and the mid-time obtained using the initial ephemeris is shown versus the observing epoch. In such a diagram remaining slopes indicate a wrong orbital period, while e.g. periodic deviations from a linear trend indicate perturbing forces. Besides the discovery of small planets, the amount of known massive planets on close-in orbits increased as well. First studies on a larger sample of planet candidates de- tected with Kepler suggest that hot giant planets exist in single planet systems only (Steffen et al. 2012). However, Szab´o et al. (2013) analysed a larger sample of Kepler hot Jupiters and found a few cases where TTVs can not be ex- plained by other causes (e.g. artificial sampling effects due to the observing cadence) but the existence of perturbers -- additional planets or even exo-moons -- in the respective sys- tem. In addition Szab´o et al. (2013) point toward the planet candidates KOI-338, KOI-94 and KOI-1241 who are all hot Jupiters in multi-planetary systems, as well as the HAT-P-13 system with a hot Jupiter accompanied by a massive planet on an eccentric outer orbit (see also Szab´o et al. 2010) and the WASP-12 system with a proposed companion candidate found by TTV analysis (Maciejewski et al. 2011a). The origin of those planets is yet not fully understood. One possible formation scenario shows that close-in giant planets could have migrated inwards after their creation further out (Steffen et al. 2012). In that case, inner and close outer planets would have either been thrown out of the system, or caught in resonance. In the latter case, even small perturbing masses, e.g. Earth-mass objects, can res- ult in TTV amplitudes in the order of several minutes (see Ford & Holman 2007 or Seeliger et al. 2014). Though Kepler is surveying many of those systems, it is necessary to look at the most promising candidates among all close-in giant plan- ets discovered so far, including stars outside the field of view of Kepler. In our ongoing study1 of TTVs in exoplanetary systems we perform photometric follow-up observations of specific promising transiting planets where additional bod- ies are expected. The targets are selected by the following criteria: (i) The orbital solution of the known transiting planet shows non-zero eccentricity (though the circularization time- scale is much shorter than the system age) and/or deviant radial velocity (RV) data points -- both possibly indicating a perturber. (ii) The brightness of the host star is V 6 13 mag and the transit depth is at least 10 mmag to ensure sufficient pho- tometric and timing precision at 1-2m class ground-based telescopes. (iii) The target is visible from the Northern hemisphere. (iv) The target has not been studied for TTV signals before. In the past the (Maciejewski et al. (Maciejewski et al. transiting 2011a, exoplanets WASP- 2013b), WASP- 12b 3b 2013a), WASP-10b (Maciejewski et al. 2011b, Maciejewski et al. 2014 in prep.), WASP-14b (Raetz 2012), TrES-2 (Raetz et al. 2014), and HAT-P-32b (Seeliger et al. 2014) have been studied by our group in detail. In most cases, except for WASP-12b, no TTVs could be confirmed. 2010, Here, we extend our investigations to search for TTVs in the HAT-P-18, HAT-P-19, HAT-P-27/WASP-40 and WASP-21 planetary systems. In Section 2 we give a short description of the targets analysed within this project. Sec- tion 3 explains the principles of data acquisition and reduc- tion and gives an overview of the telescopes used for obser- vation. The modeling procedures are described in Section 4, followed by the results in Section 5. Finally, Section 6 gives a summary of our project. 2 TARGETS 2.1 HAT-P-18b and HAT-P-19b Hartman et al. (2011) reported on the discovery of the exo- planets HAT-P-18b and HAT-P-19b. The two Saturn-mass planets orbit their early K type stars with periods of 5.51d and 4.01d, respectively. In case of HAT-P-18b Hartman et al. (2011) found the 1 see http://ttv.astri.umk.pl/doku.php for a project overview Table 1. The observing telescopes that gathered data within the TTV project for HAT-P-18b, HAT-P-19b, HAT-P-27b/WASP-40b and WASP-21b in order of the number of observed transit events of the Observatory. The table lists the telescopes and corresponding observatories, as well as the telescope diameters ⊘ and number of observed transit events per telescope in this project Ntr. Ground-based transit observations 3 # Observatory Telescope (abbreviation) 3 University Observatory Jena (Germany) 4 T UBITAK National Observatory (Turkey) 5 Calar Alto Astronomical Observatory (Spain) 6 7 8 National Astronomical Observatory Rozhen (Bulgaria) Sierra Nevada Observatory (Spain) Peter van de Kamp Observatory Swarthmore (USA) 1 Michael Adrian Observatory Trebur (Germany) 2 Graduate Institute of Astronomy Lulin (Taiwan & USA) T1T (Trebur 1.2m) Tenagra II (Tenagra 0.8m) RCOS16 (Lulin 0.4m) 90/60 Schmidt (Jena 0.6m) Cassegrain (Jena 0.25m) T100 (Antalya 1.0m) 1.23m Telescope (CA-DLR 1.2m) Ritchey-Chr´etien (OSN 1.5m) RCOS (Swarthmore 0.6m) Ritchey-Chr´etien-Coud´e (Rozhen 2.0m) Cassegrain (Rozhen 0.6m) 9 Teide Observatory, Canarian Islands (Spain) STELLA-I (Stella 1.2m) 10 University Observatory Bochum (Cerro Armazones, Chile) VYSOS6 (Chile 0.15m) 11 Xinglong Observing Station (China) 12 Gettysburg College Observatory (USA) 13 Star´a Lesn´a Observatory (Slovak Rep.) 14 Istanbul University Telescope at C¸ anakkale (Turkey) 15 Toru´n Centre for Astronomy (Poland) ⊘ (m) Ntr 8 1.2 5 0.8 0.4 2 4 0.9/0.6 3 0.25 5 1.0 4 1.23 1.5 2 2 0.6 1 2.0 1 0.6 2 1.2 0.15 1 1 0.9/0.6 90/60 Schmidt (Xinglong 0.6m) 1 0.4 Cassegrain (Gettysburg 0.4m) 1 0.5 0.5m Reflector (StaraLesna 0.5m) 1 0.6m Telescope (C¸ anakkale 0.6m) 0.6 0.6m Cassegrain Telescope (Tor´un 0.6m) 0.6 1 eccentricity to be slightly non-zero (e = 0.084 ± 0.048). Re- cent studies of Esposito et al. (2014) found the eccentri- city to be consistent with a non-eccentric retrograde orbit by analysing the Rossiter-McLaughlin effect. Knutson et al. (2013) also analysed the RV signal and found a jitter in the order of 17.5 m s−1 that remains unexplained. Interest- ingly, the transit data listed in the exoplanet transit data- base (ETD; Poddan´y, Br´at, & Pejcha 2010) shows a huge spread in the transit depth in the order of several tens of milli-magnitudes. Thus we included HAT-P-18b in our list of follow-up objects to also confirm or refute these transit depth variations. In addition, we performed a monitoring project for HAT-P-18 over a longer time span to possibly find overall brightness variations. In Ginski et al. (2012) we used lucky imaging with Astralux at the Calar Alto 2.2m Telescope to search for additional low-mass stellar compan- ions in the system. With the data gathered in this previous study we could already exclude objects down to a mass of 0.140±0.022 M⊙ at angular separations as small as 0.5 arcsec and objects down to 0.099±0.008 M⊙ outside of 2 arcsec. For HAT-P-19b a small eccentricity of e = 0.067 ±0.042 was determined by Hartman et al. (2011). They also found a linear trend in the RV residuals pointing towards the ex- istence of a long period perturber in the system. Within this project we want to address the problem of the proposed perturber using photometric methods, i.e. follow-up transit events to find planetary induced TTV signals. 2.2 HAT-P-27b/WASP-40b HAT-P-27b (B´eky et al. 2011), independently discovered as WASP-40b by Anderson et al. (2011) within the WASP- survey (Pollacco et al. 2006), is a typical hot Jupiter with a period of 3.04d. While the eccentricity was found to be e = 0.078 ± 0.047 by B´eky et al. (2011), Anderson et al. (2011) adopted a non-eccentric orbit. However, the latter au- thor found a huge spread in the RV data with up to 40m s−1 deviation from the circular single planet solution. According to Anderson et al. (2011) one possible explanation, despite a changing activity of the K-type host star, is the existence of a perturber that might not be seen in the B´eky et al. (2011) data due to the limited data set. However, the authors sug- gest further monitoring to clearify the nature of the system. One possibility is to study the companion hypothesis from the TTV point of view. Another interesting aspect of HAT-P-27b is the transit shape which B´eky et al. (2011) fitted using a flat bottom model. Anderson et al. (2011) and Sada et al. (2012) found the transit rather to have a roundish shape. From the graz- ing criterion (Smalley et al. 2011) they concluded that the system is probably grazing, which would explain the unusual shape of the transit. However it is not clear why this is not seen in the B´eky et al. (2011) data, hence it is still not clear which shape is real. 2.3 WASP-21b The planetary host star WASP-21, with its Saturn-mass planet on a 4.32d orbit discovered by Bouchy et al. (2010), is one the most metal-poor planet hosts accompanied by one of the least dense planets discovered by ground-based transit searches to date. Bouchy et al. (2010) found that including a small non-zero eccentricity to the fit does not improve the results. Hence, they concluded that the eccentricity is con- sistent with zero. However, in a later study Barros et al. (2011a) found the G3V star to be in the process of moving off the main sequence. Thus, we included further observations of WASP- 21b planetary transits to improve the knowledge on this sys- tem. 3 DATA ACQUISITION AND REDUCTION Our observations make use of YETI network telescopes (Young Exoplanet Transit Initiative; Neuhauser et al. 2011), a worldwide network of small to medium sized telescopes 4 M. Seeliger et al. Table 2. The list of all transit observations gathered within the TTV project sorted by object and date. Though no preselections for quality or completeness have been applied to this list, transits used for further analysis have been marked by an asterisk. The filter subscripts B, C and J denote the photometric systems of Bessel, Cousins and Johnson, respectively. The last column lists the number of exposures and the exposure time of each observa- tion. # Date Telescope Filter Exposures HAT-P-18b 2011-04-21 Trebur 1.2m 2011-05-02 Trebur 1.2m 2011-05-24 Trebur 1.2m 2011-06-04 Rozhen 2.0m 2012-05-05 Rozhen 0.6m 2012-06-07 CA DLR 1.23m 2013-04-28 Antalya 1.0m 2014-03-30 Toru´n 0.6m HAT-P-19b 2011-11-23 Jena 0.6m 2011-11-23 Jena 0.25m 2011-11-23 Trebur 1.2m 2011-12-05 Jena 0.6m 2011-12-05 Jena 0.25m 2011-12-09 Jena 0.6m 2011-12-09 Jena 0.25m 2011-12-09 Trebur 1.2m 2011-12-17 CA DLR 1.23m 2014-08-01 Antalya 1.0m 2014-08-05 Antalya 1.0m 2014-08-21 2014-10-04 Jena 0.6m Jena 0.6m HAT-P-27b StaraLesna 0.5m Stella 1.2m Lulin 0.4m Lulin 0.4m Stella 1.2m 2011-04-05 2011-04-08 2011-05-03 2011-05-05 Trebur 1.2m 2011-05-08 2011-05-21 Tenagra 0.8m 2012-03-07 2012-03-29 Tenagra 0.8m 2012-04-01 Tenagra 0.8m 2012-04-04 Tenagra 0.8m 2012-04-25 Xinglong 0.6m 2012-05-16 Trebur 1.2m 2012-05-25 Chile 0.15m 2012-06-13 Tenagra 0.8m 2013-06-03 Antalya 1.0m 2013-06-03 OSN 1.5m 2013-06-06 CA DLR 1.23m 2014-06-18 Antalya 1.0m WASP-21b RB RB RB RC IC BJ R clear RB RB RB RB VB RB VB RB RJ R R RB RB RB RB Hα RB Hα R R R R R R RB IJ /RJ R R R RJ R Swarthmore 0.6m RB 2011-08-24 2011-08-24 Gettysburg 0.4m 2012-08-16 Trebur 1.2m 2012-10-20 Antalya 1.0m 2013-09-18 CA DLR 1.23m 2013-09-22 Antalya 1.0m 2013-09-22 Ulupinar 0.6m R RB R RJ R RB 1* 2 3* 4 5* 6 7 8 9* 10 11 12 13 14* 15 16 17* 18 19 20 21* 22* 23* 24 25* 26 27 28 29 30 31 32 33 34 35 36* 37* 38 39* 40* 41 42* 43 44* 45 46 189 x 90 s 123 x 45 s 323 x 60 s 1000 x 10 s 219 x 90 s 250 x 60 s 214 x 50 s 297 x 40 s 246 x 50 s 320 x 50 s 461 x 30 s 129 x 60 s 28 x 300 s 290 x 50 s 118 x 150 s 380 x 35 s 273 x 60 s 148 x 60 s 196 x 40 s 152 x 50 s 280 x 50 s 166 x 40 s 250 x 40 s 180 x 100 s 162 x 70 s 190 x 100 s 141 x 40 s 361 x 30 s 240 x 30 s 329 x 20 s 333 x 20 s 154 x 40 s 231 x 70 s 220 x 80 s 223 x 15 s 156 x 60 s 435 x 30 s 172 x 60 s 146 x 50 s 545 x 45 s 230 x 60 s 365 x 40 s 242 x 40 s 584 x 30 s 208 x 50 s 163 x 110 s mostly on the Northern hemisphere established to explore transiting planets in young open clusters. A summary of all participating telescopes and the num- ber of performed observations can be found in Table 1. Most of the observing telescopes are part of the YETI network. This includes telescopes at Cerro Armazones (Chile, op- erated by the University of Bochum), Gettysburg (USA), Jena (Germany), Lulin (Taiwan), Rozhen (Bulgaria), Sierra Nevada (Spain), Star´a Lesn´a (Slovak Republic), Swarthmore (USA), Tenagra (USA, operated by the National Central University of Taiwan) and Xinglong (China). For details about location, mirror and chip see Neuhauser et al. (2011). In addition to the contribution of the YETI telescopes, we obtained data using the following telescopes: • the 1.2m telescope of the German-Spanish Astronom- ical Center on Calar Alto (Spain), which is operated by Ger- man Aerospace Center (DLR). • the 1.2m robotic telescope STELLA-I, situated at Teide Observatory on Tenerife (Spain) and operated by the Leibnitz-Institut fur Astrophysik Potsdam (AIP). • the Trebur 1Meter Telescope operated at the Michael Adrian Observatory Trebur (Germany) • the T100 telescope of the T UBITAK National Obser- vatory (Turkey) • the 0.6m telescope (CIST60) at Ulupınar Observatory operated by Istanbul University (Turkey) • the 0.6m Cassegrain telescope of the Toru´n Centre for Astronomy (Poland) the transit Besides observations, Jena 0.6m telescope with its Schmidt Teleskop Kamera (Mugrauer & Berthold 2010) was used to perform a long term monitoring of HAT-P-18 as described in Sections 2.1 and 5.1. the Between 2011 April and 2013 June our group observed 45 transit events (see Table 2) using 18 different telescopes (see Table 1). 15 observations could be used for further ana- lysis, while 30 observations had to be rejected due to sev- eral reasons, e.g. no full transit event has been observed or bad weather conditions and hence low signal to noise. E.g. Southworth et al. (2009a,b) showed that defocusing the tele- scope allows to reduce atmospheric and flat fielding effects. Since a defocused image spreads the light over several CCD pixel, one can increase the exposure time and hence the ef- fective duty cycle of the CCD assuming a constant read out time (as mentioned also in the conclusions of Barros et al. 2011a). Thus we tried to defocuse the telescope and increase the exposure time during all our observations. Table 3 lists the ingress/egress durations τ derived using the formula (18) and (19) given in Winn (2010). With our strategy we obtain at least one data point within 90s. This ensures to have at least 10 data points during ingress/egress phase which is re- quired to fit the transit model to the data and get precise transit mid-times. All data has been reduced in a standard way by ap- plying dark/bias and flat field corrections using iraf2. The 2 iraf is distrubuted by the National Optical Astronomy Obser- vatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under coorporative agreement with the National Science Foundation. Table 3. The input parameters for the JKTEBOP & TAP runs for all objects listed in Section 2. All values have been obtained from the original discovery papers. LD coefficients are taken from Claret & Bloemen (2011) linear interpolated in terms of Teff, log g and [Fe/H] using the EXOFAST/QUADLD code (Eastman et al. 2013). Free parameters are marked by an asterisk. At the bottom the duration of ingress and egress according to Winn (2010) has been added. Ground-based transit observations 5 Object rp + rs* Rp/Rs* i (◦)* a/Rs* Mp/Ms e P (d) HAT-P-15b HAT-P-18b HAT-P-19b HAT-P-27b WASP-21b WASP-38b 0.0575(19) 0.1019(09) 89.1(2) 19.16(62) 0.0018(01) 0.190(19) 10.863502(27) 0.0575(19) 0.1365(15) 88.8(3) 16.04(75) 0.0709(33) 0.1418(20) 88.2(4) 12.24(67) 0.000243(26) 0.084(48) 5.5080023(06) 0.000329(37) 0.067(42) 4.008778(06) 0.1159(65) 0.1186(31) 84.7(7) 9.65(54) 0.000663(58) 0.078(47) 3.039586(12) 0.0959(44) 0.1040(35) 88.75(84) 10.54(48) 0.0829(27) 0.0844(21) 89.69(30) 12.15(39) 0.000282(19) 0 4.322482(24) 0.00213(11) 0.0314(46) 6.871815(45) R (mag) Teff (K) log g (cgs) [Fe/H] (dex) v sin i (km s−1) LD law of the star R band linear* R band non-linear* V band linear* V band non-linear* 11.81 12.61 5568(90) 4803(80) 4.38(03) +0.22(08) 2.0(5) 4.57(04) +0.10(08) 0.5(5) 12.82 4990(130) 4.54(05) +0.23(08) 0.7(5) 11.98 5300(90) 4.51(04) +0.29(10) 0.4(4) 11.52 5800(100) 4.2(1) -0.46(11) 1.5(6) 9.22 6150(80) 4.3(1) -0.12(07) 8.6(4) quadratic quadratic quadratic quadratic quadratic quadratic 0.4200 0.2525 0.5274 0.2164 0.5736 0.1474 0.7180 0.0697 0.5433 0.1710 0.6783 0.1039 0.4808 0.2128 0.6002 0.1643 0.3228 0.2982 0.4055 0.2892 0.2998 0.3095 0.3834 0.2992 τegress/ingress (min) 27.8 22.8 23.1 37.8 20.1 21.9 respective calibration images have been obtained in the same night and with the same focus as the scientific observations. This is necessary especially if the pointing of the telescope is not stable. When using calibration images obtained with different foci, patterns remain in the images that lead to distortions in the light curve. Besides our own observations, we also use literature data. This involves data from the respective discovery papers mentioned in Section 2, as well as data from Esposito et al. (2014) in case of HAT-P-18b, Sada et al. (2012) in case of HAT-P-27b, Barros et al. (2011a) and Ciceri et al. (2013) for WASP-21b, and Simpson et al. (2010) for WASP-38b. 4 ANALYSES The light curve extraction and modelling is performed ana- logous to the procedure described in detail in Seeliger et al. (2014). 4.1 Light curve extraction The Julian date of each image is calculated from the header information of the start of the exposure and the exposure time. To precisely determine the mid-time of the transit event these informations have to be stored most accurate. The reliability of the final light curve model thus also de- pends on a precise time synchronization of the telescope computer system. Observing transits with multiple tele- scopes at the same time enables to look for synchroniza- tion errors which would otherwise lead to artifacts in the O -- C diagram (as shown for HAT-P-27b in epoch 415, see Section 5.3). We use differential aperture photometry to extract the light curve from the reduced images by measuring the bright- ness of all bright stars in the field with routines provided by iraf. The typical aperture radius is ≈ 1.5 times the mean full width half maximum of all stars in the field of view (FoV). The best fitting aperture is found by manually varying the aperture radius by a few pixels to minimize the photometric scatter. The final light curve is created by com- paring the star of interest against an artificial standard star composed of the (typically 15-30) brightest stars in the FoV weighted by their respective constantness as introduced by Broeg, Fern´andez, & Neuhauser (2005). The final photometric errors are based on the instru- mental iraf measurement errors. The error of the constant comparison stars are rescaled by their photometric scat- ter using shared scaling factors in order to achieve a mean χ2 ≈ 1 for all comparison stars. The error bars of the transit star are rescaled afterwards using the same scaling factors (for further details on the procedure see Broeg et al. 2005). red Due to atmospheric and air-mass effects transit light curves show trends. They can impact the determination of transit parameters. To eliminate such effects we start the observation about 1 hour before and finish about 1 hour after the transit itself. Thus we can detrend the observations by fitting a second order polynomial to the out-of-transit data. 4.2 Modelling with TAP and JKTEBOP To model the light curves we used the Transit Ana- lysis Package (tap; Gazak et al. 2012). The modelling of the transit light curve is done by using the exofast routines (Eastman, Gaudi, & Agol 2013) with the light curve model of Mandel & Agol (2002). For error estimation tap uses Markov Chain Monte Carlo simulations (in our case 10 times 105 MCMC chains) together with wavelet- based likelihood functions (Carter & Winn 2009). The coef- ficients for the quadratic limb darkening (LD) law used by tap are taken from the exofast/quadld-routine of 6 M. Seeliger et al. Eastman et al. (2013)3 that linearly interpolates the LD tables of Claret & Bloemen (2011). For comparison we also use jktebop (see Southworth 2008, and references therein) which is based on the ebop code for eclipsing binaries (Etzel 1981; Popper & Etzel 1981). To compare the results with those obtained with tap, we only use a quadratic LD law which is sufficient for ground-based data. For error estimation we used Monte Carlo simulations (104 runs), bootstrapping (104 data sets), and a residual shift method as provided by jktebop. As input values we use the system properties presented in the respective discovery papers (see Table 3 for a sum- mary). For both light curve fitting procedures we took the original light curve as well as a threefold binned one. Though the binned light curves result in a lower rms of the fit, no better timing result can be achieved due to a longer cadence. As free parameters we use the mid-transit time Tmid, inclination i, and planet to star radius ratio k = rp/rs (with rp and rs being the planet and stellar radius scaled by the semimajor axis, respectively). In case of tap the inverse frac- tional stellar radius a/Rs = 1/rs, in case of jktebop the sum of the fractional radii (cid:0)rp + rs(cid:1) is fitted as well. Both quantities are an expression of the transit duration and can be transformed into each other according to the following equation: a/Rs = (cid:0)1 + rp/rs(cid:1) / (cid:0)rp + rs(cid:1) The fitting procedure is applied two times. First keeping the limb darkening coefficients fixed at their theoretical values, and afterwards letting them vary. For tap we set the fitting interval to ±0.2. In case of jktebop we use the option to set the LD coefficients fixed for the initial model, but let them being perturbed for the error estimation. Thus the fitted model does not change, but the error bars are increased. The eccentricity was fixed to zero for all our analyses. Since all data are obtained using JDUTC as time base, we transform the fitted mid-transit times to BJDTDB after- wards using the online converter4 provided by Jason East- man (for a detailed description of the barycentric dynamical time see Eastman, Siverd, & Gaudi 2010). Finally, we derive the photometric noise rate (pnr, Fulton et al. 2011) as a quality marker for all light curves, which is defined as the ratio between the root mean square of the model fit and the number of data points per minute. For further analysis we took data with pnr . 4.5 into account. 5 RESULTS For every light curve we get six different models, four from tap (for the binned and unbinned data with the LD coeffi- cients fixed and free) and two from jktebop (for the binned and unbinned data, LD coefficients set free for error estim- ation only). To get one final result we averaged those six results. As for the errors, we got four different estimations from tap and 12 from jktebop. As final error value, we took the maximum of either the largest of the error estimates, or limb darkening 3 the http://astroutils.astronomy.ohio-state.edu/exofast/limbdark.shtml 4 http://astroutils.astronomy.ohio-state.edu/time/utc2bjd.html calculator at is available online Figure 1. Relative R band brightness of the star HAT-P-18 over a time span of 12 months. The dotted line represents the rms of a constant fit. the spread of the model fit results to use a conservative er- ror estimate. It has to be noted, though, that the spread between the different models has always been below size of the error bars. Binning the light curve threefold using an error weighted mean in principle still leaves enough data points during ingress and egress to be able to fit the transit model to the data while reducing the error bar of an in- dividual measurement. However, comparing the results of the threefold binned and unbinned data we do not see sig- nificant differences, neither in the fitted values, nor in the error bars. The same counts for the differences between the tap models obtained with fixed LD values and those obtained with the LD coefficients set free to fit. For a detailed discus- sion of the influence of the LD model on transit light curves see e.g. Raetz et al. (2014). 5.1 HAT-P-18b For HAT-P-18b we obtained five transit light curves (see Fig 2). All light curves show some features which could be caused by stellar activity, e.g. spots. However, the qual- ity of the data does not allow to draw further conclusion. Moreover, there is only a small number of suitable compar- ison stars available in the respective FoV of each observation, hence this could also be an artificial effect. Except for one -- but not significant -- outlier the differ- ences in the O -- C diagram (see Fig. 2) can be explained by redetermining the published period by (0.53 ± 0.36)s. Hence we find a slightly larger period compared to the originally proposed period of Hartman et al. (2011). Regarding the transit depth, and thus the planet to star radius ratio, we do not see any changes. However, having a look at the data provided in the exoplanet transit database (ETD; Poddan´y, Br´at, & Pejcha 2010) one can see that the values for the transit depth reported there vary by several tens of mmag. Such transit depth variations can be caused by close variable stellar companions placed within the aper- ture due to the pixel scale of our detectors and the defocus- sing of the telescopes. Thus we took a more detailed look at Ground-based transit observations 7 Figure 2. top: The threefold binned transit light curves of the three complete transit observations of HAT-P-18b. The upper panels show the light curve, the lower panels show the residuals. The rms of the fit of the threefold binned light curves (dotted lines) are shown as well. bottom: The present result for the HAT-P-18b observing campaign, including the O -- C diagram, as well as the results for the reverse fractional stellar radius a/RS , the inclination i, and the planet to star radius ratio k. The open circle denotes literature data from Hartman et al. (2011), the open triagles denotes data from Esposito et al. (2014). Filled triangles denote our data (from Trebur and Rozhen. The dotted line shows the 1σ error bar of the constant fit. the long term variability of the parent star. Over a timespan of 12 months we obtained 3 images in four bands (B, V, R, I) in each clear night using the Jena 0.6m telescope. As shown in Fig. 1, the mean variation of the R band brightness is ≈ 3.8 mmag taking the individual error bar of the meas- urements into account. However, this variation is too small to be responsible for the seen transit depth variations. The monitoring of the remaining bands is not shown but leads to a similar result. Having a closer look at the light curves listed for HAT-P-18b in ETD, one can see that a large number is of lower quality. This is especially true for those light curves responsible for the spread in the tabulated transit depth. Taking only the higher quality data into account the spread is much smaller. Depending on the quality cut the variation can even reach the order of the error bars of the measure- ments. Thus we believe that the transit depth variation is negligible. Despite a spread in the data, which can be explained by the quality of the light curves, we do not see any significant difference for k, i and a/Rs between the respective observa- tions. A summary of all obtained parameters can be found in Table 4 at the end of the paper, as well as a comparison with literature values. 5.2 HAT-P-19b For HAT-P-19b we got two light curves using the Jena 0.6m and one light curve from the CAHA 1.2m telescope (Fig. 3). In all three cases we obtained high precision data. The light curves show no artifacts that could be ascribed to e.g. spots on the stellar surface. Plotting the 8 M. Seeliger et al. mid-transit times into the O -- C diagram, we can redeter- mine the period by (0.53 ± 0.06) s. As for the inclination and the reverse fractional stellar radius we can confirm the values reported in Hartman et al. (2011). The radius ratio of k = 0.1378 ± 0.0014, however, seems to be smaller than assumed by Hartman et al. (2011) (k = 0.1418 ± 0.0020). 5.3 HAT-P-27b HAT-P-27b planetary transits were observed six times. An advantage of a network such as YETI lies within the possib- ility of simultaneous observations using different telescopes. This enables us to independently check whether the data is reliable. For HAT-P-27b simultaneous observations could be achieved at epoch 415 using two different telescopes (Anta- lya 1.0m and OSN 1.5m). As one can see in Fig. 4 the best-fitting transit shapes differ. Since some data has been acquired using small tele- scopes under unfavorable conditions, this might be an ar- tificial effect. However, these shape variations can also be seen in the literature data. While B´eky et al. (2011) fitted a flat bottom to the transit they observed, the transit of Anderson et al. (2011) shows a rather roundish bottom and so does the Sada et al. (2012) transit. Both Anderson et al. (2011) and Sada et al. (2012) claim, a roundish bottom is in good agreement with a grazing transit. Our most precise transit light curve of HAT-P-27b at epoch 415 the OSN 1.5m telescope also shows no flat bottom. Thus, we would agree with the previous authors that the planet is grazing. The epoch 540 observations also shows a V-shape. However, due to a connection problem parts of the ingress data is missing, thus precise fits of the transit parameters are not possible. In addition to our own data we added data from Sada et al. (2012) and Brown et al. (2012). The latter one only lists system parameters without giving an epoch of ob- servation, thus we artificially put them to epoch 200. Since we do not see any systematic trend in the remaining data, this does not effect the conclusion on the system parameters but improves the precision of the constant fit. The system parameters i, a/RS and k can be determ- ined more precisely than before taking the errors of the in- dividual measurements into account. All three parameters are in good agreement with the results of previous authors. Furthermore, we do not see any significant variation. The larger k-value of the epoch 540 observations are due to the quality of the corresponding light curve. Looking at the mid-transit time we see that a period change of (−0.51 ± 0.12) s explains the data quite well. The mid-transit time of one of the epoch 415 observations was found to be ≈ 4.5 min ahead of time, while the other one is as predicted. This way we could identify a synchronization error during one of the observations. This example shows the importance of simultaneous transit observations. Unfor- tunately this was the only sucessful observation of that kind within this project (for a larger set of double and threefold observations see e.g. Seeliger et al. 2014). 5.4 WASP-21b Four transit light curves of WASP-21b are available, in- cluding one light curve from Barros et al. (2011a) (see Fig.5). In addition, the results of the analysis of two transit events of Ciceri et al. (2013) and one transit observation of Southworth (2012) are also taken into account. Concern- ing the O -- C diagram, we found that a period change of (2.63 ± 0.17) s removes the linear trend which is present in the data fitted with the initial ephemeris. As in the previ- ous analyses no trend or sinusoidal variation in the system parameters can be seen. However, regarding inclination and reverse fractional stellar radius we do see a significant difference between our results and the initial values published by Bouchy et al. (2010). This was also found by other authors before. As dis- cussed in Barros et al. (2011a) this result is a consequence of the assumption of Bouchy et al. (2010) that the planet host star is a main sequence star, while Barros et al. (2011a) found that the star starts evolving off the main sequence and thus its radius increases. This in turn leads to corrections of the stellar and hence planetary properties. 6 SUMMARY We presented the results of the transit observations of the extra solar planets HAT-P-18b, HAT-P-19b, HAT-P- 27b/WASP-40b and WASP-21b which are part of our on- going project on gound-based follow-up observations of exo- planetary transits using small to medium sized telescopes with the help of YETI network telescopes. During the past three years we followed these well chosen objects to refine their orbital parameters as well as to find transit timing variations indicating yet unknown planetary companions. Table 4 contains an overview of the redetermined proper- ties, as well as the available literature values, while Table 5 lists the results of the individual light curve fits. In all cases we could redetermine the orbital paramet- ers. Especially the period could be determined more precise than before. So far, we can not rule out the existence of TTV signals for the planets investigated within this study due to the limited number of available high quality data. Also the parameters a/Rs, rp/rs and inclination have been obtained and compared to the available literature data. Despite some corrections to the literature data, we found no significant variations within these parameters. To distinguish between a real astrophysical source of the remaining scatter and ran- dom noise as a result of the quality of our data more high precision transit observations would be needed. HAT-P-18b was also part of an out-of-transit monitor- ing for a spread in the transit depth was reported in the literature that could be due to a significant variability of the transit host star. Regarding our transit data we can not confirm the spread in transit depth. Looking at the quality of the literature data showing the transit depth variation, it is very likely that this spread is of artificial nature. Thus it is not suprising that we did not find stellar variability larger than ≈ 3.8 mmag. However we do see some structures in the light curves that could be caused by spot activity on the stellar surface. Ground-based transit observations 9 Figure 3. top: The transit light curves obtained for HAT-P-19b. bottom: The present result for the HAT-P-19b observing campaign. All explanations are equal to Fig. 2. The open circle denotes literature data from Hartman et al. (2011), filled triangles denote our data (from Jena and Calar Alto). ACKNOWLEDGMENTS All the participating observatories appreciate the logistic and financial support of their institutions and in particu- lar their technical workshops. MS would like to thank all participating YETI telescopes for their observations, as well as G. Maciejewski for helpful comments on this work. JGS, AP, and RN would like to thank the Deutsche Forschungs- gemeinschaft (DFG) for support in the Collaborative Re- search Center Sonderforschungsbereich SFB TR 7 "Gravit- ationswellenastronomie". RE, MK, SR, and RN would like to thank the DFG for support in the Priority Programme SPP 1385 on the First ten Million years of the Solar System in projects NE 515/34-1 & -2. RN would like to acknowledge financial support from the Thuringian government (B 515- 07010) for the STK CCD camera (Jena 0.6m) used in this project. MM and CG thank DFG in project MU 2695/13-1. 10 M. Seeliger et al. Figure 4. top: The transit light curves obtained for HAT-P-27b. bottom: The present result for the HAT-P-27b observing campaign. All explanations are equal to Fig. 2. The open circles denotes data from the discovery papers of B´eky et al. (2011) and Anderson et al. (2011), open triangles denote literature data from Sada et al. (2012) and Brown et al. (2012) (the latter one set to epoch 200 artificially), filled triangles denote our data (from Lulin, Trebur, Xinglong and Antalya). The research of DD and DK was supported partly by funds of projects DO 02-362, DO 02-85 and DDVU 02/40-2010 of the Bulgarian Scientific Foundation, as well as project RD-08-261 of Shumen University. Wu,Z.Y. was supported by the Chinese National Natural Science Foundation grant Nos. 11373033. The research of RC, MH and MH is sup- ported as a project of the Nordrhein-Westfalische Akademie der Wissenschaften und Kunste in the framework of the academy programme by the Federal Republic of Germany and the state Nordrhein-Westfalen. MF acknowledges fin- ancial support from grants AYA2011-30147-C03-01 of the Spanish Ministry of Economy and Competivity (MINECO), co-funded with EU FEDER funds, and 2011 FQM 7363 of the Consejer´ıa de Econom´ıa, Innovaci´on, Ciencia y Em- Ground-based transit observations 11 Figure 5. top: The transit light curves obtained for WASP-21b. bottom: The present result for the WASP-21b observing campaign. All explanations are equal to Fig. 2. The open circle denotes data from the discovery paper of Bouchy et al. (2010), open triangles denote literature data from Barros et al. (2011a), Ciceri et al. (2013) and Southworth (2012) (the latter one artificially set to epoch 200), filled triangles denote our data (from Swarthmore, Trebur and Calar Alto). pleo (Junta de Andaluc´ıa, Spain) We also wish to thank the T UBITAK National Observatory (TUG) for support- ing this work through project number 12BT100-324-0 an 12CT100-388 using the T100 telescope. MS thanks D. Kee- ley, M. M. Hohle and H. Gilbert for supporting the obser- vations at the University Observatory Jena. This research has made use of NASA's Astrophysics Data System. This research is based on observations obtained with telescopes of the University Observatory Jena, which is operated by the Astrophysical Institute of the Friedrich-Schiller-University. This work has been supported in part by Istanbul University under project number 39742, by a VEGA Grant 2/0143/14 of the Slovak Academy of Sciences and by the joint fund of 12 M. Seeliger et al. Table 4. A comparison between the results obtained in Our analysis and the literature data. All epochs T0 are converted to BJDT DB . T0 (d) P (d) HAT-P-18b a/Rs k = Rp/Rs i (◦) our analysis Hartman et al. (2011) Esposito et al. (2014) 2 454 715.022 54 ± 0.000 39 2 454 715.022 51 ± 0.000 20 2 455 706.7 ± 0.7 5.508 029 1 ± 0.000 004 2 5.508 023 ± 0.000 006 5.507 978 ± 0.000 043 17.09 ± 0.71 16.04 ± 0.75 16.76 ± 0.82 0.136 2 ± 0.001 1 0.136 5 ± 0.001 5 0.136 ± 0.011 88.79 ± 0.21 88.3 ± 0.3 88.79 ± 0.25 HAT-P-19b our analysis Hartman et al. (2011) 2 455 091.535 00 ± 0.000 15 2 455 091.534 94 ± 0.000 34 4.008 784 2 ± 0.000 000 7 4.008 778 ± 0.000 006 12.36 ± 0.09 12.24 ± 0.67 0.137 8 ± 0.001 4 0.141 8 ± 0.002 0 88.51 ± 0.22 88.2 ± 0.4 HAT-P-27b our analysis B´eky et al. (2011) Anderson et al. (2011) Sada et al. (2012) Brown et al. (2012) 2 455 186.019 91 ± 0.000 44 2 455 186.019 55 ± 0.000 54 2 455 368.394 76 ± 0.000 18 2 455 186.198 22 ± 0.000 32 -- 3.039 580 3 ± 0.000 001 5 3.039 486 ± 0.000 012 3.039 572 1 ± 0.000 007 8 3.039 582 4 ± 0.000 003 5 3.039 577 ± 0.000 006 10.01 ± 0.13 9.65 +0.54 −0.40 9.88 ± 0.39 9.11 +0.71 −1.01 9.80 +0.38 −0.29 0.119 2 ± 0.001 5 0.118 6 ± 0.003 1 0.125 0 ± 0.001 5 0.134 4 +0.017 4 −0.038 9 0.120 +0.009 −0.007 85.08 ± 0.07 84.7 +0.7 −0.4 84.98 +0.20 −0.14 84.23 ± 0.88 85.0 ± 0.2 WASP-21b our analysis Bouchy et al. (2010) Barros et al. (2011a) Ciceri et al. (2013) Southworth (2012) 2 454 743.042 17 ± 0.000 65 2 454 743.042 6 ± 0.002 2 2 455 084.520 48 ± 0.000 20 2 454 743.040 54 ± 0.000 71 2 455 084.520 40 ± 0.000 16 4.322 512 6 ± 0.000 002 2 4.322 482 +0.000 024 −0.000 019 4.322 506 0 ± 0.000 003 1 4.322 518 6 ± 0.000 003 0 4.322 506 0 ± 0.000 003 1 9.62 ± 0.17 6.05 +0.03 −0.04 9.68 +0.30 −0.19 9.46 ± 0.27 9.35 ± 0.34 0.103 0 ± 0.000 8 0.104 0 +0.001 7 −0.001 8 0.107 1 +0.000 9 −0.000 8 0.1055 ± 0.0023 0.1095 ± 0.0013 87.12 ± 0.24 88.75 +0.70 −0.84 87.34 ± 0.29 86.97 ± 0.33 86.77 ± 0.45 Table 5. The results of the induvidual fits of the observed complete transit event. The rms of the fit and the resultant pnr are given in the last column. The table also shows the result for the transits with pnr > 4.5 that are not used for redetermining the system properties. date epoch telescope Tmid − 2 450 000 d a/Rs k = Rp/Rs i (◦) rms/pnr (mmag) HAT-P-18b 2011/04/21 2011/05/24 2012/05/05 2012/06/07 2013/04/28 174 Trebur 1.2m 180 Trebur 1.2m 243 Rozhen 0.6m 249 CA-DLR 1.2m 308 Antalya 1.0m 5 673.419 67 ± 0.001 24 16.4 ± 1.4 0.139 9 ± 0.007 2 88.52 ± 0.84 5 706.469 93 ± 0.000 80 18.28 ± 0.83 0.134 3 ± 0.003 9 89.52 ± 0.58 6 053.472 76 ± 0.000 84 16.04 ± 1.36 0.137 3 ± 0.004 7 88.55 ± 0.79 6 086.518 56 ± 0.001 25 6 411.496 38 ± 0.000 84 15.22 ± 1.52 0.146 4 ± 0.006 8 87.89 ± 0.75 -- -- -- 2011/11/23 2011/12/09 2011/12/17 2014/10/04 2011/04/05 2011/04/08 2011/05/05 2012/04/01 2012/04/25 2013/06/03 2013/06/03 2013/06/18 Jena 0.6m Jena 0.6m 199 203 205 CA-DLR 1.2m 460 Jena 0.6m HAT-P-19b 5 899.283 45 ± 0.000 49 12.56 ± 0.34 0.136 9 ± 0.002 6 88.20 ± 0.64 5 905.318 10 ± 0.000 44 12.29 ± 0.35 0.136 9 ± 0.002 3 89.05 ± 0.67 5 913.335 71 ± 0.000 34 11.96 ± 0.53 0.136 8 ± 0.002 7 88.38 ± 0.80 6 935.575 59 ± 0.000 55 12.43 ± 0.36 0.134 0 ± 0.002 6 89.25 ± 0.67 HAT-P-27b Lulin 0.4m Lulin 0.4m 155 156 165 Trebur 1.2m 274 Tenagra 0.8m 282 Xinglong 0.6m 415 Antalya 1.0m 415 OSN 1.5m 540 Antalya 1.0m 5 657.153 33 ± 0.001 07 10.72 ± 1.67 0.123 3 ± 0.008 1 85.53 ± 0.93 5 660.194 81 ± 0.001 16 9.43 ± 1.01 0.122 8 ± 0.014 9 84.69 ± 0.81 5 687.551 22 ± 0.000 51 9.83 ± 0.56 0.115 3 ± 0.002 9 85.07 ± 0.40 6 018.864 57 ± 0.002 32 9.65 ± 1.63 0.119 9 ± 0.012 6 84.13 ± 1.63 6 043.180 95 ± 0.001 35 9.89 ± 1.67 0.118 6 ± 0.006 7 84.83 ± 1.24 6 447.442 68 ± 0.001 66 10.64 ± 1.30 0.118 4 ± 0.008 1 85.51 ± 0.94 6 447.445 71 ± 0.000 30 10.18 ± 0.29 0.122 4 ± 0.003 7 85.23 ± 0.21 6 827.395 45 ± 0.002 20 10.77 ± 1.01 0.146 2 ± 0.014 1 85.26 ± 0.55 WASP-21b 2011/08/24 2012/08/16 2013/09/18 244 327 Trebur 1.2m 420 CA-DLR 1.2m Swarthmore 0.6m 5 797.734 00 ± 0.001 12 9.94 ± 0.93 0.101 4 ± 0.003 2 87.74 ± 1.41 6 156.502 60 ± 0.001 15 9.97 ± 0.92 0.101 7 ± 0.003 2 87.78 ± 1.46 6 558.496 48 ± 0.000 73 9.38 ± 0.69 0.106 4 ± 0.002 7 86.91 ± 0.96 3.0 / 3.3 3.7 / 4.0 3.5 / 4.4 4.1 / 4.5 3.9 / 5.0 2.1 / 2.2 2.3 / 2.4 1.2 / 1.3 2.8 / 3.0 3.4 / 3.8 3.4 / 3.2 1.6 / 1.8 5.7 / 5.8 4.3 / 5.1 2.6 / 3.5 1.2 / 0.9 3.1 / 4.0 3.3 / 3.1 2.9 / 2.6 1.6 / 1.4 Astronomy of the National Science Foundation of China and the Chinese Academy of Science under Grants U1231113. References Adams F. C., Laughlin G., 2006, ApJ, 649, 992 Anderson D. R., et al., 2011, PASP, 123, 555 Bakos G., Noyes R. W., Kov´acs G., Stanek K. Z., Sasselov D. D., Domsa I., 2004, PASP, 116, 266 Ground-based transit observations 13 Winn J. N., 2010, arXiv, arXiv:1001.2010 Wu Z.-Y., Zhou X., Ma J., Jiang Z.-J., Chen J.-S., Wu J.-H., 2007, AJ, 133, 2061 Barros S. C. C., Pollacco D. L., Gibson N. P., Howarth I. D., Keenan F. P., Simpson E. K., Skillen I., Steele I. A., 2011b, MNRAS, 416, 2593 Barros S. C. C., et al., 2011a, A&A, 525, A54 B´eky B., et al., 2011, ApJ, 734, 109 Borucki W. J., et al., 2011, ApJ, 728, 117 Bouchy F., et al., 2010, A&A, 519, A98 Bodenheimer P., Laughlin G., Lin D. N. C., 2003, ApJ, 592, 555 Broeg C., Fern´andez M., Neuhauser R., 2005, AN, 326, 134 Brown D. J. A., et al., 2012, ApJ, 760, 139 Carter J. A., Winn J. N., 2009, ApJ, 704, 51 Ciceri S., et al., 2013, A&A, 557, A30 Claret A., Bloemen S., 2011, A&A, 529, A75 Eastman J., Gaudi B. S., Agol E., 2013, PASP, 125, 83 Eastman J., Siverd R., Gaudi B. S., 2010, PASP, 122, 935 Esposito M., et al., 2014, A&A, 564, L13 Etzel P. B., 1981, in Carling E. B., Kopal Z., eds, Proc. NATO Adv. Study Inst., Photometric and Spectroscopic Binary Systems, Reidel, Dordrecht, p. 111 Ford E. B., Holman M. J., 2007, ApJ, 664, L51 Fulton B. J., Shporer A., Winn J. N., Holman M. J., P´al A., Gazak J. Z., 2011, AJ, 142, 84 Gazak J. Z., Johnson J. A., Tonry J., Dragomir D., East- man J., Mann A. W., Agol E., 2012, AdAst, 2012, 30 Ginski C., Mugrauer M., Seeliger M., Eisenbeiss T., 2012, MNRAS, 421, 2498 Hartman J. D., et al., 2011, ApJ, 726, 52 Knutson H. A., et al., 2013, arXiv, arXiv:1312.2954 Koch D. G., et al., 2010, ApJ, 713, L79 Kov´acs G., et al., 2010, ApJ, 724, 866 Maciejewski G., et al., 2010, MNRAS, 407, 2625 Maciejewski G., et al., 2011a, MNRAS, 411, 1204 Maciejewski G., Errmann R., Raetz St., Seeliger M., Spaleniak I., Neuhauser R., 2011, A&A, 528, A65 Maciejewski G., et al., 2013b, A&A, 551, A108 Maciejewski G., et al., 2013a, AJ, 146, 147 Mandel K., Agol E., 2002, ApJ, 580, L171 Mugrauer M., Berthold T., 2010, AN, 331, 449 Neuhauser R., et al., 2011, AN, 332, 547 Poddan´y S., Br´at L., Pejcha O., 2010, NewA, 15, 297 Pollacco D. L., et al., 2006, PASP, 118, 1407 Popper D. M., Etzel P. B., 1981, AJ, 86, 102 Raetz St., 2012, PhD thesis University Jena Raetz et al., 2014, MNRAS, 444, 1351 Sada P. V., et al., 2012, PASP, 124, 212 Seeliger M., et al., 2014, MNRAS, 441, 304 Simpson E. K., et al., 2010, MNRAS, 405, 1867 Smalley B., et al., 2011, A&A, 526, A130 Southworth J., et al., 2009a, MNRAS, 396, 1023 Southworth J., et al., 2009b, MNRAS, 399, 287 Southworth J., 2008, MNRAS, 386, 1644 Southworth J., 2012, MNRAS, 426, 1291 Steffen J. H., et al., 2012, PNAS, 109, 7982 Szab´o G. M., et al., 2010, A&A, 523, A84 Szab´o R., Szab´o G. M., D´alya G., Simon A. E., Hodos´an G., Kiss L. L., 2013, A&A, 553, A17 Van Eylen V., et al., 2014, ApJ, 782, 14 von Essen C., Schroter S., Agol E., Schmitt J. H. M. M., 2013, A&A, 555, A92 Weber M., Granzer T., Strassmeier K. G., 2012, SPIE, 8451,
1704.06482	1	1704	2017-04-21T11:10:33	Spectra and physical properties of Taurid meteoroids	[ "astro-ph.EP" ]	Taurids are an extensive stream of particles produced by comet 2P/Encke, which can be observed mainly in October and November as a series of meteor showers rich in bright fireballs. Several near-Earth asteroids have also been linked with the meteoroid complex, and recently the orbits of two carbonaceous meteorites were proposed to be related to the stream, raising interesting questions about the origin of the complex and the composition of 2P/Encke. Our aim is to investigate the nature and diversity of Taurid meteoroids by studying their spectral, orbital, and physical properties determined from video meteor observations. Here we analyze 33 Taurid meteor spectra captured during the predicted outburst in November 2015 by stations in Slovakia and Chile, including 14 multi-station observations for which the orbital elements, material strength parameters, dynamic pressures, and mineralogical densities were determined. It was found that while orbits of the 2015 Taurids show similarities with several associated asteroids, the obtained spectral and physical characteristics point towards cometary origin with highly heterogeneous content. Observed spectra exhibited large dispersion of iron content and significant Na intensity in all cases. The determined material strengths are typically cometary in the $K_B$ classification, while $P_E$ criterion is on average close to values characteristic for carbonaceous bodies. The studied meteoroids were found to break up under low dynamic pressures of 0.02 - 0.10 MPa, and were characterized by low mineralogical densities of 1.3 - 2.5 g cm$^{-3}$. The widest spectral classification of Taurid meteors to date is presented.	astro-ph.EP	astro-ph	Spectra and physical properties of Taurid meteoroids Pavol Matlovica, Juraj T´otha, Regina Rudawskab, Leonard Kornosa aFaculty of Mathematics, Physics and Informatics, Comenius University, Bratislava, bESA European Space Research and Technology Centre, Noordwijk, Netherlands Slovakia Abstract Taurids are an extensive stream of particles produced by comet 2P/Encke, which can be observed mainly in October and November as a series of me- teor showers rich in bright fireballs. Several near-Earth asteroids have also been linked with the meteoroid complex, and recently the orbits of two car- bonaceous meteorites were proposed to be related to the stream, raising interesting questions about the origin of the complex and the composition of 2P/Encke. Our aim is to investigate the nature and diversity of Taurid mete- oroids by studying their spectral, orbital, and physical properties determined from video meteor observations. Here we analyze 33 Taurid meteor spectra captured during the predicted outburst in November 2015 by stations in Slo- vakia and Chile, including 14 multi-station observations for which the orbital elements, material strength parameters, dynamic pressures, and mineralog- ical densities were determined. It was found that while orbits of the 2015 Taurids show similarities with several associated asteroids, the obtained spec- tral and physical characteristics point towards cometary origin with highly heterogeneous content. Observed spectra exhibited large dispersion of iron content and significant Na intensity in all cases. The determined material strengths are typically cometary in the KB classification, while PE criterion is on average close to values characteristic for carbonaceous bodies. The studied meteoroids were found to break up under low dynamic pressures of 0.02 - 0.10 MPa, and were characterized by low mineralogical densities of 1.3 - 2.5 g cm-3. The widest spectral classification of Taurid meteors to date is presented. Keywords: meteor shower, meteoroid stream, comet, asteroid, Taurids, Email address: [email protected] (Pavol Matlovic) Preprint submitted to Planetary and Space Science April 24, 2017 2P/Encke 1. Introduction The Taurid complex is an extensive population of bodies associated with comet 2P/Encke and certainly one of the most interesting structures in the Solar System. It is the widest known meteoroid stream in the Solar System, and has been linked with several catastrophic incidents including the Palae- olithic extinctions (Napier, 2010) and the Tunguska event of 1908 (Kres´ak, 1978). The low inclination of the orbit of the stream causes gravitational perturbations of inner solar system planets (Levison et al., 2006), resulting in the diffuse structure of the Taurid complex. The Earth encounters differ- ent parts of the stream annually from September to December, giving rise to several meteor showers, most significantly the Northern and the South- ern Taurids peaking in late October and early November. Owing to the wide spread of the Taurid stream, the zenithal hourly rates of the resulting meteor showers usually do not exceed 5. However, the observed activity of the shower is known to be rich in bright fireballs, which induced discussions about the potential of Taurids in producing meteorites (Brown et al., 2013; Madiedo et al., 2014; Brown et al., 2016). While the orbits of most meteoroids are quite dispersed, it is still likely that the Taurid stream has a narrow and dense core consisting of particles concentrated near the orbit of the stream's parent ob- ject. It is often inferred that strong bombardment episodes have resulted at epochs when the material of the stream's core reached Earth intersection. Although comet 2P/Encke is generally considered as the parent object of the Taurids (Whipple, 1940), various small near-Earth objects (NEOs) and recently even two instrumentally observed carbonaceous meteorite falls have been linked with the orbit of the stream (Haack et al., 2011; Haack et al., 2012). 2P/Encke is a short-period comet moving on an orbit dynamically decoupled from Jupiter with an orbital period of 3.3 years. The peculiar orbit of the 2P/Encke raises interesting questions relating its origin and the origin of the Taurid complex. Today, the most supported hypothesis claims that comet 2P/Encke and several other bodies including large number of meteoroids were formed by a fragmentation of an earlier giant comet. The theory was first suggested by Whipple (1940), and later elaborated by Asher et al. (1993) who argued that the Taurid meteoroid streams, 2P/Encke, and the associated Apollo asteroids were all formed by major comet fragmentation 2 20 to 30 ky ago. Napier (2010) suggested that the debris of this fragmentation event could have caused the Palaeolithic extinctions followed by the return to ice age conditions. The broad structure of the Taurid complex was studied by several au- thors with over 100 of NEO candidates associated with the Taurid complex. Porubcan et al. (2006) identified 15 different sub-streams of the complex and by applying stricter criteria for generic relations found associations with 9 NEOs. The presence of these bodies in the Taurid complex implies the possibility that part of the Taurid meteoroid population might be produced as the decay or impact products of the associated asteroids. The connection to carbonaceous chondrites represented by Maribo and Sutter's Mill mete- orites also needs to be further investigated. We expect significantly different spectral and structural properties of such bodies in comparison to fragile cometary meteoroids originating in 2P/Encke. So far, the efforts in find- ing traces of the common origin of the largest Taurid complex bodies gave rather sceptical results. Popescu et al. (2014) studied the spectral proper- ties of the largest asteroids associated with the Taurid complex, but found no evidence supporting mutual relation. Similarly, the spectroscopic and photometric measurements of Tubiana et al. (2015) found no apparent link between comet 2P/Encke, the Taurid complex NEOs, and the Maribo and Sutter's Mill meteorites. The Taurid meteor shower occasionally exhibits enhanced activity due to a swarm of meteoroids being ejected by the 7:2 resonance with Jupiter (Asher, 1991). The outburst of meteors observed in 2015 was anticipated by the model of Asher & Izumi (1998), which previously predicted enhanced activities of Taurids in 1998, 2005 and 2008. Some features of the Taurid me- teor shower activity suggest that the Taurid swarm exist only in the southern branch (Southern Taurids) and not in the northern branch (Shiba, 2016). The two main branches of the Taurid complex, the Northern and the Southern Taurids are well observed meteor showers with established orbital characteristics, which clearly trace them to comet 2P/Encke. The majority of Taurid shower analyses have been focused on orbital properties of the stream, with only several Taurid meteor spectra observed before the outburst of 2015 (e.g. Srirama Rao & Ramesh, 1965; Madiedo et al., 2014). Borovicka et al. (2005) analyzed six Taurid emission spectra observed by low-resolution video spectrograph, identifying three Na-enhanced meteoroids, while another three Taurids have been classified as normal type. Using the same instrumentation, Voj´acek et al. (2015) observed three more normal type Taurids including one 3 meteoroid with lower iron content. The determined spectral characteristics of normal to enhanced sodium content and lower iron content would suggest cometary parent body. Asteroidal meteoroids are expected to be depleted in volatile sodium by the processes of space weathering (e.g. Borovicka et al., 2005; Trigo-Rodr´ıguez & Llorca, 2007). Determined spectral properties give us valuable input into the studies of the origin of meteoroids; however, for doubtless differentiation between cometary and asteroidal particles, precise orbital and ideally physical characteristics must be obtained. Besides the two major Taurid meteor showers, there are several smaller meteor streams associated with the Taurid complex. Most notably, this in- cludes the Piscids, Arietids, chi Orionids, and the daytime showers of beta Taurids and zeta Perseids, encountered by the Earth in June and July. The beta Taurids and zeta Perseids have been shown to be the daytime twin branches to the Southern and the Northern Taurids respectively. The possi- bility of Taurid sub-streams being produced by a different parent object was discussed by Babadzhanov (2001), who found shower associations to each of the Taurid complex asteroids and interpreted them as the evidence for the cometary origin of these asteroids. Recently, Olech et al. (2016) found very close resemblance between the orbits of two 2015 Southern Taurid fireballs and the orbits of 2005 UR and 2005 TF50 asteroids. Precise observations of Taurid fireballs observed during the 2015 outburst were examined by Spurn´y et al. (2016), who emphasized the orbital resemblance of the 2015 Taurids to the orbits of asteroids 2015 TX24 and 2005 UR. They argue that the outburst may have been caused by Taurid filaments associated with these asteroids. All of the previous associations between different sub-streams of the Tau- rids and the Taurid complex asteroids were based on the orbital similarities. Certainly, studying the spectral and physical properties of the Taurid me- teoroids could extend our understanding of the origin and evolution of the Taurid complex and its individual meteoroid streams. 2. Observations and data reduction The outburst of Taurid meteors in 2015 was observed globally by nu- merous meteor networks. Here we present the detailed analysis of 33 Tau- rid meteor spectra captured by the spectral All-sky Meteor Orbits System (AMOS-Spec) (Rudawska et al., 2016). Particular focus will be placed on 16 of these spectra, which were observed in Modra Observatory, Slovakia. Multi-station observations for 14 of these meteors were provided by four 4 AMOS stations comprising the Slovak Video Meteor Network, which carries out routine meteor observations on every clear night (T´oth et al., 2011; T´oth at al., 2015), supplemented by individual observations provided by Pavel Spurn´y (European Fireball Network) and Jakub Koukal (Central European MEteor NeTwork). Additional 17 single-station Taurid spectra were observed by AMOS sys- tem in Chile during the testing for two new, now already established, south- ern AMOS stations. The system made observations during the activity peak between November 5 and November 10 at San Pedro de Atacama in the Ata- cama Desert. The observational conditions here enable much higher efficiency in capturing meteor events. These single-station meteors were identified as members of the Taurid stream based on their estimated radiant position and geocentric velocity. The AMOS-Spec is a semi-automatic remotely controlled video system for the detection of meteor spectra. The main display components consist of 30mm f/3.5 fish-eye lens, image intensifier (Mullard XX1332), projection lens (Opticon 1.4/19 mm), and digital camera (Imaging Source DMK 51AU02). This setup yields a 100◦ circular field of view with a resolution of 1600 x 1200 pixels and frame rate of 12/s. The incoming light is diffracted by a holographic 1000 grooves/mm grating placed above the lens. The spectral resolution of the system varies due to the geometry of the all-sky lens with a mean value of 1.3 nm/px. A 500 grooves/mm grating was used for the observations in Chile, providing a mean spectral resolution of 2.5 nm/px. The system covers the whole visual spectrum range from app. 370 nm to 900 nm. The spectral response curve of the AMOS-Spec system (camera, image intensifier, and lens) is given in Figure 1. It was determined by measuring the known spectrum of Jupiter and is normalized to unity at 480 nm. The typical limiting magnitude of the system for meteors is approx. +4 mag., while only meteors brighter than approx. 0 mag. can be captured along with its spectrum. More details about the properties and capabilities of the AMOS systems can be found in T´oth et al. (2015). The main disadvantage of the wide-field camera is the interference of the moonlight and bright Moon spectrum causing occasional difficulties in meteor detection. Fortunately, the Moon illumination percentage was descending from 45% to 10% during the activity peak of the 2015 Taurids (November 4 - November 8) and caused no significant problems in spectra analysis. 5 Figure 1: The spectral sensitivity of the AMOS-Spec system. The spectral response curve was determined by measuring the spectrum of Jupiter, and is normalized to unity at 480 nm. 2.1. Spectra reduction The all-sky geometry of the lenses causes slight curvature of the meteor spectra captured near the edge of the FOV. For this reason, each spectrum was scanned manually on individual video frames using ImageJ1 program. Before scanning the spectra, all of the analyzed spectral events were corrected for dark frame, flat-fielded, and had the star background image subtracted. We are particularly interested in the relative intensities of the Na I - 1, Mg I - 2, and Fe I - 15 multiplets, which form the basis of the spectral classifica- tion of meteors established by Borovicka et al. (2005). Each spectrum was fitted with a simple model accounting for all significant contributions to the observed spectral profile. The continuum level was fitted by Planck curve at given temperature. For most meteors, the continuum was well fitted by Planck curve at 4500 K, but in some cases, particularly for fainter meteors, this was lowered down. It should be noted that the fitting procedure does not serve as physical interpretation of the spectrum, but is only used for the reduction of contamination caused by the mixture of continuum radiation and weak unrecognizable lines, in order to achieve the best possible fit of each spectrum. The spectral lines (low temperature, high temperature, at- mospheric, and wake lines) were fitted using the positions and intensities of 1https://imagej.nih.gov/ij/ 6 the most important atomic lines in meteor spectra, as given in Borovicka et al. (2005). The simulated spectrum was fitted by instrumental Gaussian line profiles with appropriate full width of typically 5 nm at 1/e of the peak inten- sity. Furthermore, the most important N2 bands of the first positive system present in the meteor spectra were fitted using the positions and intensities taken from Borovicka et al. (2005), and adjusted for our spectra. The Gaus- sian width of these bands was assumed to be 10 nm. The intensities of all contributions were adjusted in each spectrum to provide to best possible fit. The intensities of the main meteor multiplets (Mg I - 2, Na I - 1, and Fe I - 15) subtracted of the fitted continuous radiation and atmospheric emissions (most prominent lines of O, N, and N2 bands) in the modeled spectrum were then measured and used in the spectral classification. The scale of each spec- trum was determined using known spectral lines in the calibration spectrum and polynomial fitting of 2nd or higher degree. For most cases the lines of Mg I - 2 (518.2 nm), Na I - 1 (589.2 nm), O I - 1 (777.4 nm), and Fe I - 41 (438.4 nm) were used for the scaling of the spectrum, which provided optimal fit for the identification of other lines. Additionally, other prominent spectral lines were employed (e.g., Ca II - 1 at 393.4 nm, O I - 615.3 at nm, or N I - 3 at 744.2 nm.) in cases where other lines were missing in the FOV. The line identification, correction for spectral sensitivity, reduction of continuum radiation and atmospheric emission, and calculation of relative intensity ra- tios of studied spectral lines are all performed by our own developed Matlab code. Each spectrum was scanned and scaled on individual frames and the final profile was obtained by summing measured intensities on each frame. Only bright enough Taurid spectra with signal-to-noise ratio higher than 5 were used for our analysis. The low resolution of the video meteor spectra allows us only to iden- tify limited number of meteoroid spectral lines compared to high dispersion photographic spectrographs used mainly in the past (e.g. Ceplecha, 1964; Nagasawa, 1978; Borovicka, 1993). It has been shown that meteor spec- tra consist of two spectral components (Borovicka, 1994). The majority of the observed meteoric spectral lines belong to the main, low-temperature (≈ 4500 K) spectral component, which comprises mostly of the neutral lines of Na, Mg, Fe, Cr, Ca, and Mn. The high-temperature (≈ 10000 K) spectral It consists of component is mostly significant in bright and fast meteors. atmospheric O and N lines, and meteoric high-excitation lines of Mg II, Si II, and Ca II. Only three meteoric species (Na I, Mg I, and Fe I) can be reliably examined in spectra captured by video spectrographs with resolution similar 7 to our system. Even though other meteoric species are also often observed in meteor spectra (such as the lines of Ca I, Ca II, Si II, Cr I, or Mn I), only the multiplets of these three species have sufficient intensity in almost every meteor spectra to be accurately measured and compared. Nevertheless, the variations of relative intensities of these lines can reveal the different com- positions of studied meteoroids. The full lists of the most important atomic lines identified in typical low resolution video meteor spectra can be found in Borovicka et al. (2005) or Voj´acek et al. (2015). 2.2. Photometry and astrometry Previous measurements and orbital analyses of known meteor streams have demonstrated the astrometric precision of the AMOS system (T´oth et al., 2015). The system achieves higher orbital accuracy compared to most all-sky video systems due to significantly higher pixel resolution. Multiple station observations of the analyzed spectral events were used to determine the heliocentric orbital parameters and geocentric trajectory parameters of the Taurid meteors. The AMOS systems comprising the SVMN yield a stan- dard astrometric error of 0.03 - 0.05◦ that translates to the accuracy of tens to hundreds of meters for atmospheric meteor trajectory. The detailed all-sky reduction described by Ceplecha (1987) and Borovicka et al. (1995) was used. Each meteor was measured individually using UFOCapture software for me- teor detection and UFOAnalyzer for astrometric data reduction (SonotaCo, 2009). Magnitudes of meteors were determined by visual calibration based on comparison with bright stars, planets, and Moon phases in the FOV. These apparent magnitudes were then corrected to standard altitude of 100 km at the observation zenith, and for atmospheric extinction to obtain estimated absolute magnitudes. Using this method, magnitude of each studied me- teor was determined from every available observing station, and resulting value (Table 5) was obtained as an average from individual station measure- ments. Absolute magnitudes of meteors captured in Chile (Table 2) were determined from single-station observations, which is why no error is given. Errors of these measurements from visual calibration are estimated to be ±1 magnitude for meteors of -5 magnitude and fainter, and ±2 magnitudes for brighter meteors. The errors of the estimated photometric masses are influ- enced besides the errors of absolute magnitudes and meteor velocities also by the effective meteoroid fragmentation model, which was not examined in our work. Therefore, the uncertainties of photometric masses is only estimated in 8 given order of magnitude. Few exceptionally bright fireballs among the 2015 Taurids caused significant saturation, delivering higher uncertainty in the determination of the absolute magnitudes and consequently the photomet- ric masses of the meteoroids.One of our multi-station meteors was observed under very low convergence angle, which translated into the low accuracy of the obtained orbital elements. These parameters were therefore omitted in our results (Table 5). In future, we plan to apply a new method based on detailed comparisons to the brightness of the Moon in various phases to gain higher accuracy of the estimated magnitudes. Furthermore, to gain the highest possible accuracy of the determined photometric masses, effective fragmentation model should be taken into account. 3. Orbital classification The orbital properties of the 14 Taurid meteoroids observed by multiple stations are in Table 1. The Tisserand's parameter with respect to Jupiter can be used to distinguish between the different orbit types (cometary, as- teroidal) related to their source regions. It is defined as (cid:114)(cid:16) a (cid:17) TJ = aJ a + 2 aJ (1 − e2) cos i, (1) where aJ is the semi-major axis of Jupiter, and a, e, i are the semi-major axis, eccentricity and inclination of the meteoroid orbit respectively. The Tisserand's parameter places our sample of meteoroids on the borderline be- tween asteroidal orbits (TJ > 3) and Jupiter-family type cometary orbits (2 < TJ < 3), as would be expected for a stream originating in the short- period comet 2P/Encke (TJ = 3.026). We identified four Jupiter-family type cometary orbits and ten asteroidal orbits among the studied meteoroids. Our sample includes particles from the anticipated outburst caused by the 7:2 resonance with Jupiter, along with the background activity of the Taurid stream. Based on the later discussed meteoroid properties (Table 5), we do not assume that found differences between meteoroids defined on asteroidal and Jupiter-family type orbits directly reflect different origin of these parti- cles, but rather demonstrate the broad spatial structure of the stream, which is defined close to the borderline between these orbit classes. The determined Taurid orbits are displayed in Figure 2. For further information about the orbital origin of our sample, we focused on using the Southworth-Hawkins DSH orbital similarity criterion (South- 9 Table 1: Orbital properties of the multi-station Taurid meteor spectra observed during the outburst in November 2015. Each Taurid is designated with meteor number, meteor ID based on the observational date and time, shower assignment, corresponding Southworth- Hawkins criterion (DSH ) with respect to the assigned shower, geocentric velocity (vg), semi-major axis (a), perihelion distance (q), eccentricity (e), inclination (i), argument of perihelion (ω), longitude of the ascending node (Ω), Tisserand's parameter with respect to Jupiter (TJ ), and corresponding orbit type based on TJ (AST for asteroidal orbits, JUP for Jupiter-family type orbits). Meteor no. 7 was omitted due to high uncertainty of the determined orbital elements, caused by the low convergence angle between the two observational stations. Meteor no. 15, and meteors no. 17 - 33 captured in Chile were only observed by one station, orbital properties are not available. No. Meteor ID Shower DSH vg a q e i ω Ω TJ type 1 M20151102 020949 STA 0.08 2 M20151102 024553 NTA 0.14 3 M20151103 212219 STA 0.04 4 M20151103 212454 STA 0.10 5 M20151105 205304 NTA 0.05 6 M20151105 215813 STA 0.03 8 M20151105 220251 STA 0.03 9 M20151105 231200 STA 0.03 10 M20151105 235959 STA 0.17 11 M20151108 234416 STA 0.03 12 M20151109 041942 NTA 0.05 13 M20151111 220940 STA 0.04 14 M20151116 193459 STA 0.10 16 M20151123 012004 NTA 0.10 28.74 ± 0.73 27.62 ± 0.11 27.98 ± 0.35 29.27 ± 0.06 30.12 ± 0.32 28.98 ± 0.12 28.34 ± 0.17 27.94 ± 0.38 26.59 ± 0.20 27.27 ± 0.17 29.30 ± 0.14 27.71 ± 0.15 25.51 ± 0.04 26.06 ± 0.03 0.316 0.008 0.365 0.002 0.340 0.004 0.305 0.001 0.314 0.003 0.354 0.003 0.356 0.005 0.356 0.004 0.438 0.002 0.385 0.007 0.330 0.002 0.400 0.002 0.441 0.000 0.422 0.001 0.831 0.015 0.821 0.003 0.820 0.008 0.839 0.001 0.866 0.007 0.852 0.003 0.836 0.004 0.824 0.009 0.828 0.005 0.816 0.005 0.851 0.003 0.837 0.004 0.788 0.001 0.797 0.001 5.5 0.1 3.3 0.1 4.7 0.1 6.6 0.1 1.8 0.1 4.3 0.1 3.7 0.7 4.9 0.1 0.8 0.1 5.5 1.0 1.0 0.1 5.1 0.1 4.9 0.1 3.0 0.1 120.19 0.25 293.69 0.31 117.32 0.03 121.13 0.01 298.08 0.03 113.45 0.35 114.04 0.72 114.78 0.06 103.60 0.04 111.17 1.02 296.74 0.06 108.00 0.01 104.98 0.01 286.88 0.20 39.10 219.16 40.90 40.90 222.95 42.92 42.92 42.98 42.92 46.01 226.32 48.96 53.88 240.24 1.87 0.12 2.04 0.03 1.89 0.06 1.90 0.01 2.35 0.09 2.39 0.05 2.17 0.07 2.02 0.08 2.55 0.07 2.09 0.07 2.21 0.03 2.45 0.05 2.08 0.01 2.08 0.01 10 JUP JUP 3.45 AST 0.19 3.27 AST 0.04 3.45 AST 0.09 3.40 AST 0.02 2.89 0.09 2.89 0.04 3.11 AST 0.07 3.28 AST 0.10 2.83 0.05 3.22 AST 0.08 3.04 AST 0.04 2.88 0.04 3.28 AST 0.01 3.27 AST 0.02 JUP JUP Figure 2: Heliocentric orbits of 14 multi-station Taurid meteors based on orbital elements given in Table 1. Portrayed 4 Northern Taurid orbits (black), 10 Southern Taurids (grey), complemented with the orbit of Earth and Jupiter, and the direction towards the vernal equinox. worth & Hawkins, 1963) to look for associations of individual meteoroids with different filaments of the broad Taurid stream. Ten of the studied me- teoroids were assigned to the southern branch of the stream, while four were identified as the Northern Taurids. Corresponding DSH values with respect to the assigned Taurid filaments (defined by orbital elements from Porubcan et al., 2006) are given in Table 1. Moreover, we also examined the orbital similarity between the studied meteoroids and the Taurid complex associated NEOs. Applying similarity threshold value of DSH ≤ 0.10, we found possible relation to 10 asteroids: 2005 UR (associated with meteoroids no. 1, 4), 2015 TX24 (no. 1, 3, 4, 6, 8, 9), 2003 UV11 (no. 3, 6, 8, 9, 11, 13), 2007 UL12 (no. 3, 8, 9, 11, 13, 14), 2010 TU149 (no. 2, 6, 8, 10), 2003 WP21 (no. 14), 2010 VN139 (no. 14), 2004 TG10 (no. 5, 12), 2012 UR158 (no. 5, 12), and 2014 NK52 (no. 5, 12, 16). Compared orbital elements of the associated NEOs were taken from the Asteroid Orbital Elements Database currated by Ted Bowell and Bruce Koehn. The most associations in the southern branch of our sample was found with asteroids 2015 TX24, 2003 UV11, and 2007 UL12. Meteoroids from the northern branch showed most markable similar- 11 ities with asteroid 2014 NK 52. These results support the measurements of Spurn´y et al. (2016), who suggested that the 2015 outburst may have been caused by Taurid filaments associated with asteroids 2015 TX24 and 2005 UR. Two of the southern Taurid meteoroids (no. 1 and no. 4) originated on orbits closely resembling the orbit of asteroid 2005 UR (DSH = 0.06 and 0.07). The orbital similarity of 2005 UR with two bright Southern Taurid fireballs observed over Poland was also noted by Olech et al. (2016). The obtained orbital similarities with aforementioned asteroids are however still rather indefinite and could also be, given the broad structure of the Taurid complex, only incidental. In this work, we will focus on using the spectral and physical properties of Taurid meteoroids to look for features suggesting generic associations with asteroidal origin. 4. Spectral classification As mentioned earlier, the low resolution of video spectrographs only al- lows us to study limited number of spectral lines in meteor spectra. However, the relative intensities of the three main meteoric species (Na, Mg, Fe) ob- served in the visual spectrum can reveal the rough composition of meteoroids. This is the basis of the spectral classification established by Borovicka et al. (2005), which distinguishes different types of bodies based on the intensities of spectral multiplets Na I - 1 (589.2 nm), Mg I - 2 (518.2 nm), and Fe I - 15 (526.0 - 545.0 nm). The modeled contributions of all recognized lines of the Fe I - 15 multiplet were summed. These multiplets were chosen as they can be well measured in most meteor spectra (see Section 2.1), and all lay in the region of high instrumental spectral sensitivity. Nevertheless, the measurement of these lines must be careful, because the Na multiplet overlaps with a sequence of N2 bands, while the wake lines of Fe can influ- ence the intensity of Mg line. All these effects were accounted for in the reduction process described in Section 2.1. As the spectral sensitivity of the AMOS-Spec is in this region similar to the sensitivity of the system used by Borovicka et al. (2005), we are able to apply the same classification also for our observations. It must be noted that the classification of Borovicka et al. (2005) was originally developed for fainter meteors in the magnitude range +3 to -1, corresponding to meteoroid sizes 1 - 10 mm. Our system observes meteor spectra of -1 to -10 magnitude, corresponding to meteoroid sizes of app. few mm to tens of cm. However, we expect that the same physical conditions fitted by the thermal equilibrium model applied for bright photo- 12 Figure 3: The spectral classification of 33 Taurid meteors observed in Slovakia and Chile in November 2015. Diagram shows the measured relative intensities of multiplets Mg I - 2, Na I - 1, and Fe I - 15. Meteoroids are marked with numbers ordered according to the observational date. Meteors 1 - 16 were observed in Slovakia, while meteors 17 - 33 were captured in lower spectral resolution in Chile. graphic meteors (Borovicka, 1993) as well as for fainter meteors (Borovicka et al., 2005), can be also assumed for the population observed by our system. The effect of self-absorption in brighter meteor spectra was examined in in- dividual cases. The instrumental pixel saturation in spectra only occurred in individual meteors during bright flares and posed no problem for most me- teors in the resolution provided by the 1000 grooves/mm grating. Individual frames with saturated meteor spectra were excluded from the summation for final spectral profiles and calculation of relative line intensities. The spectral classification of Taurid meteoroids observed in Slovakia and Chile is on Figure 3. The majority (24) of Taurid meteoroids represent normal type spectral class with similar ratios of Na I, Mg I, and Fe I intensities. 13 However, we observe apparent variations in Fe and Na content among the meteoroids in our sample. Particularly, the studied spectra revealed large dispersion of iron content in individual cases, ranging from Fe-poor bodies to almost chondritic Fe/Mg ratios (Figure 4). Normal type meteoroids are positioned in the middle of the ternary diagram, with content close to the ratios characteristic for chondritic composition. It was noted by Borovicka et al. (2005) and later confirmed by Voj´acek et al. (2015) that the majority of observed meteoroids fall in the ternary diagram below the so called chondritic curve, which represents the expected range for chondritic composition, due to the lower content of iron. This is also the case for our sample of Taurid meteoroids, though two of the measured spectra showed quite significant Fe intensities (meteoroids no. 5 and 19). Furthermore, we also identified 8 Taurids depleted in iron, representing Fe-poor class, and two Na-enhanced Taurid meteoroids (no. 13 and 16). As can be clearly seen in Figure 3, all of the Taurids are positioned right from the center of the ternary diagram, exposing slight Na line intensity enhancement over Mg in all of the studied meteoroids. This supports the results of Borovicka et al. (2005), who found Na-enhancement in three of the 6 observed Taurid spectra. The Na line intensity is certainly one of the most interesting features in meteor spectra, revealing the thermal history of meteoroids. The presence of sodium in meteor spectra can be used as a tracer of volatile phases associated with cometary origin. During the formation of the solar system, volatile ele- ments were depleted from the inner protoplanetary disk due to intense solar radiation (Despois, 1992). We expect higher Na abundance in unprocessed bodies such as comets. Borovicka et al. (2005) pointed out three popula- tions of Na-free meteoroids: iron bodies on asteroidal orbits; meteoroids with small perihelia (q ≤ 0.2 AU), where Na was lost by thermal desorption; and Na-free meteoroids on Halley type orbits, where Na loss was possibly caused by irradiation of cometary surfaces by cosmic rays in the Oort cloud. The depletion in Na for small meteoroids is observed as a function of the time exposure to solar radiation, which predicts that asteroidal particles will be Na-poor materials (Trigo-Rodr´ıguez & Llorca, 2007). Even though all of our Taurid spectra have shown unambiguous sodium content, we observe ap- parent effect of Na depletion as a function of perihelion distance (Figure 5). The observed dependence of Na/Mg ratio on perihelion distance however can be partially influenced by the effect of saturation/self-absorption in bright (larger) meteors, while the measurement of line intensity in fainter spectra (typical for smaller particles) is a subject of higher uncertainty. The errors 14 Figure 4: Spectral profiles of Taurid meteoroids no. 5, 7, and 11. The spectra show similar features with significant dispersion of the Fe I - 15 multiplet intensity from the highest (no. 5) to the lowest (no. 11). Displayed spectral profiles are not corrected for the spectral sensitivity of the system (Fig. 1). 15 Figure 5: The dependence of the Na/Mg line intensity ratio on perihelion distance. The varying sizes of the marks display photometric masses of individual meteoroids in loga- rithmic scale. of determined Na/Mg and Fe/Mg intensity ratios from signal-to-noise ratio in individual observations are given in Table 5. As can be seen in Figure 5, the Na enhancement was preferred in smaller Taurid meteoroids, in which the Na/Mg ratio was determined with higher uncertainty. Furthermore, me- teoroids with high surface-area/volume ratios are expected to suffer higher depletion of Na and other volatiles through solar radiation (Trigo-Rodr´ıguez & Llorca, 2007). The increased values of Na/Mg ratios in Taurid spectra are not too sur- prising. It was pointed out that the Na/Mg line intensity ratio varies widely in cometary meteoroids (e.g. Borovicka et al., 2005; Rudawska et al., 2016), and is observed as a function of meteor velocity. Sodium is characterized by a low ionization potential, and the Na line is of low excitation (2.1 eV) compared to the Mg line (5.1 eV) (Borovicka et al., 2005). Consequently, we expect Na line intensity to be dominant at lowest velocities (temperatures) and lower at highest velocities (temperatures). This effect is only observed for meteor velocities below 40 km/s and increases for velocities below 20 km/s. Typical Taurid velocities of approx. 28 km/s would therefore slightly favor Na intensity over Mg, particularly for fainter meteors. Bright fireballs caused by more massive meteoroids can reach very high temperatures, at which both Na and Mg become fully ionized, as observed in Figure 5. Moreover, also the intensities of atmospheric emission lines of O I and N I, as well as the N2 bands increase with the meteor velocity. As a result of the moderate velocities of the Taurid meteors, the atmospheric emission 16 lines were clearly present in the spectra (Figure 4), but did not achieve high intensities, allowing us to readily subtract them from the spectral profiles before determining the intensities of the main meteoroid multiplets. The brightest multi-station Taurid in our sample was observed on Novem- ber 5 at 23:21:00 UT. This fireball (marked no. 9 in Figure 3) of estimated -8.4 absolute magnitude exhibited bright flare, which enabled us to capture two orders of its spectrum. The 1st order and the 2nd order spectral profiles are on Figure 6. While the 1st order spectrum unveils more detailed features with emission lines of the high-temperature spectral component, the satu- ration in several frames caused by the flare restricts us from analyzing the relative intensities of the main spectral multiplets throughout the entire me- teor flight. For this reason, we used the higher dispersion 2nd order spectrum to determine the spectral classification of this Taurid. Clearly, the relative intensities of Mg I - 2, Na I - 1, and Fe I - 15 differ in the two spectral profiles. We can observe the lines of Fe I - 15 blending into the Mg I - I radiation and increasing its intensity. This demonstrates the uncertainty of determining relative intensities in bright spectra of meteor flares. Example of a typical lower-resolution Taurid spectrum captured by a sys- tem in Chile using the 500 grooves/mm grating is on Figure 7. Even though we are not able to study fainter spectral features in these profiles, the intensi- ties of the main spectral multiplets are well measurable. The blending of Mg I - 2 and Fe I - 15 multiplets is more apparent in low-resolution spectra (Fig- ure 7), but it was resolved for the relative intensity measurement by fitting the resulting profile shape with intensities of all expected spectral lines in given area. The general results are in good agreement with the spectral clas- sification of Taurids based on observations in Slovakia. Estimated absolute magnitudes of studied single-station Taurid meteors captured with spectra in San Pedro de Atacama Chile are in Table 2. Generally, the spectral characteristics of the measured Taurid spectra, which are defined by mostly normal-type composition with Fe content below the chondritic values and increased Na content in most meteors point towards the cometary origin of all of the studied Taurid meteoroids. Nevertheless, we observe large variations of Fe and notable Na dispersion in Taurid spectra, suggesting Taurids are very heterogeneous population of meteoroids. Two of the Taurid spectra exhibited slightly enhanced Fe intensities (no. 5 and no. 19). The spectrum of meteor no. 19 was captured from a short, bright flare (estimated -9.4 absolute magnitude) and in lower spectral resolution, which brings higher uncertainty of measured relative intensities related to 17 Figure 6: Brightest multi-station Taurid meteor (no. 9) in our sample of estimated - 8.4 magnitude captured on November 5 at 23:21:00 UT, and the spectral profiles of the first order spectrum (upper picture) and second order spectrum (lower picture) showing different relative intensities of the Mg I - I, Na I - 2, and Fe - 15. multiplets. Displayed spectral profiles are not corrected for the spectral sensitivity of the system (Fig. 1). 18 Figure 7: Example of a low-resolution Taurid spectrum (no. 29) captured on November 8 at 07:00:43 UT in Chile. Displayed spectral profile is not corrected for the spectral sensitivity of the system (Fig. 1). possible effects of saturation/self-absorption. The observed heterogeneity is not unexpected for a stream of such remarkable spatial dispersion. In ad- dition to the original heterogeneity of the parent body of the stream, the diverse thermal evolution in different parts of the stream would result in the observed compositional variations. We did not find any spectral features, which would clearly indicate asteroidal origin in any of the meteoroids in our sample, though the relative intensities Na I, Mg I, and Fe I in several mete- oroids were not too distant from the values expected for chondritic bodies. For more information, we focused on using the multi-station observations of 14 Taurid meteors to determine the physical characteristics of the studied meteoroids. 5. Physical properties of Taurid meteoroids The primary focus of this section will be placed on the material strengths of Taurid meteoroids, which could indicate additional asteroidal source of meteoroids in the stream, and also tell us about the potential of this stream in producing meteorites. The first physical parameter we need to determine is the meteoroid mass, which is also later used in the calculation of other physical parameters. The activity of the Taurid meteor shower is characteristically rich in bright fire- balls indicating larger particles than what we observe in most cometary show- 19 Table 2: Estimated absolute magnitudes (M ag) of single-station Taurid meteors captured with spectra in San Pedro de Atacama, Chile. Each meteor is designated with meteor number and meteor ID based on the observational date and time. Errors of absolute magnitudes are estimated to be ±1 magnitude for meteors of -5 magnitude and fainter, and ±2 magnitudes for brighter meteors. No. Meteor ID M ag 17 M20151106 042700 18 M20151106 052449 19 M20151106 062854 20 M20151106 063233 21 M20151106 070343 22 M20151107 031208 23 M20151107 061350 24 M20151107 063741 25 M20151107 072801 26 M20151107 073808 27 M20151108 020459 28 M20151108 025443 29 M20151108 070043 30 M20151108 071904 31 M20151108 083951 32 M20151110 004036 33 M20151110 054715 -5.5 -4.5 -9.4 -3.5 -8.6 -3.9 -7.7 -7.8 -5.5 -5.2 -2.2 -11.6 -10.4 -2.8 -4.9 -6.8 -4.1 20 ers (Wetherill, 1974). It was suggested that Taurids contain very large me- teoroids at least in the hundreds of kilograms range, which is more than an order of magnitude larger than other showers (Brown et al., 2013; Ceplecha et al., 1998). Our sample included Taurid meteors ranging from -1.3 to - 8.4 magnitude, only confirming these characteristics. The meteoroid masses were determined photometrically, i.e. by integrating along the light curves of meteors. The standard luminosity equation (e.g. Bronshten, 1983) can be rewritten in the form: (cid:90) tE tB Idt τlv2 . m∞ = 2 (2) (3) Here, m∞ is the original mass of the meteoroid, v is the meteor velocity, and I is the light intensity of the meteoroid, integrated in the time interval from the beginning of the light curve (tB) to the end (tE). The luminosity efficiency factor, τl, represents the fraction of the meteoroid's kinetic energy, which has been converted into luminosity in the visual range. For the calculation of τl, we applied the approach described by Hill et al. (2005). The method uses a relation between the luminous efficiency factor and universal excitation coefficient ζ defined by Jones & Halliday (2001): (cid:16) (cid:17) ζ v2 , µ τl = 2 where the values of ζ are split into several regimes based on the meteor velocity and the ratio of the mean excitation energy to the atomic mass (/µ) is computed for meteoric atoms and ions producing the most radiation (e.g. Ceplecha et al., 1998), assuming the majority of the observed radiation comes from the meteoroid atoms (Bronshten, 1983). The estimated masses of the meteoroids in our sample range from 1 to 1500 grams (Table 5). 5.1. Material strength Meteor observations can be used to characterize the material strength of meteoroids by determining empirical parameters KB and PE, which are functions of the beginning and the terminal height of the meteor luminous path. Fragile particles tend to crumble under lower pressures resulting in higher beginning height of the observed luminous trajectory. On the contrary, only stronger materials can withstand the pressure and reach lower parts of our atmosphere, and only the most robust particles will eventually fall on the Earth's surface. It is observed that depending on the entry velocity and 21 meteoroid mass, the ablation begins at higher heights for cometary bodies than asteroidal ones (Ceplecha & McCrosky, 1976; Koten et al., 2004). The KB and PE parameters defined by Ceplecha (1968), and Ceplecha & McCrosky (1976), and later revised by Ceplecha (1988) can be expressed as: KB = log ρB + 2.5 log v∞ − 0.5 log zR, PE = log ρE − 0.42 log m∞ + 1.49 log v∞ − 1.29 log cos zR. (4) (5) Here, ρB and ρE are the atmospheric densities at beginning and end height of the luminous trajectory, v∞ is the pre-atmospheric meteor velocity, zr is the zenith distance of the radiant, and m∞ is the original photometric mass cal- culated using luminous efficiency factor in accordance with Ceplecha & Mc- Crosky (1976). The values of atmospheric densities at different heights were obtained using the MSIS-E-90 Atmosphere Model of Hedin (1991). These parameters form a basis for the empirical classification of meteoroid mate- rial strength (Table 3). The KB classification differentiates between fragile cometary bodies (group D) characteristic e.g. for Draconid meteors, typical cometary bodies (group C), dense cometary bodies (group B), carbonaceous- type bodies (group A), and strong meteoroids of asteroidal origin (group ast). Similarly, the classification based on the PE parameter distinguishes between cometary bodies (group III), carbonaceous-type bodies (group II), and as- teroidal bodies (group I). We clearly observe a large scatter of material strengths among Taurid meteoroids in our sample (Table 5). Following the KB classification (Fig- ure 8), the majority of our Taurids fall into the C group characteristic for standard cometary bodies, with only two meteoroids appearing to be more fragile (D group) and two meteoroids of slightly higher strength (B group). It should be noted that the KB parameter (Eq. 4) assumes that beginning height does not depend on meteoroid mass (meteor magnitude). While this is true for a narrow range of meteor magnitudes, Koten et al. (2004) have shown that the beginning height increases with mass for most meteor show- ers. The classification of the two brightest Taurids in our sample as type D is probably consequence of this fact. The obtained KB values are nevertheless what we would expect from cometary bodies originating in 2P/Encke. How- ever, we see significantly higher dispersion of Taurid material strengths in the PE classification (Figure 8). Here, only three meteoroids have PE values characteristic for cometary bodies, while the majority is defined as group II, similar to carbonaceous bodies, and four Taurids even exposed end-height 22 Table 3: Meteoroid material types based on material strength parameters KB and PE, as defined by Ceplecha (1988). The differences between groups C3 and C2, IIIAi and IIIA are based on orbital properties (for further information, see Ceplecha, 1988) Material type KB group KB PE group PE Fragile cometary Regular cometary long-period Regular cometary long-period Regular cometary short-period Dense cometary Carbonaceous chondrites Ordinary chondrites Asteroids D C3 C2 C1 B A ast KB < 6.6 6.6 ≤ KB < 7.1 6.6 ≤ KB < 7.1 6.6 ≤ KB < 7.1 7.1 ≤ KB < 7.3 7.3 ≤ KB < 8.0 8 ≤ KB IIIB - IIIAi IIIA - II I PE ≤ -5.70 - -5.70 < PE ≤ -5.25 -5.70 < PE ≤ -5.25 - -5.25 < PE ≤ -4.60 -4.60 < PE Figure 8: Classification of meteoroid material strength based on parameters KB and PE compared to the orbital classification of meteoroids based on Tisserands parameter. Taurid meteoroids (filled marks) are compared to diverse population of 22 meteoroids taken from Rudawska et al. (2016). Blue marks represent meteoroids on Halley-type cometary orbits, yellow marks designate Jupiter-family cometary orbits, and black represent asteroidal orbits. The varying sizes of the marks display photometric masses of individual meteoroids in logarithmic scale. 23 strengths typical for rocky asteroidal bodies (group I). Similarly, Brown et al. (2013) found that most Taurids appear to be type II or type III fire- balls. Great variations of material strength in Taurid meteoroids including very strong objects among fireballs were previously indicated by Borovicka (2006). It seems that while the beginning heights of Taurids are typically cometary, the end heights are usually lower and more dispersed (Figure 9). Ceplecha (1988) noted that parent bodies of group II and group A meteoroids are perhaps partly asteroids and partly comets (Wetherill & Revelle, 1981). However, the discovered Taurid features more likely suggest heterogeneous structure of its meteoroids, which are of cometary nature but contain solid, possibly carbonaceous inclusions. The diversity of material types observed on Taurid orbits could be explained by the inhomogeneous interior of the parent comet (2P/Encke or earlier larger body). The existence of solid in- clusions, which relate to CI and CM carbonaceous chondrites has also been indicated in a small population of Perseid meteoroids with high mineralogical densities (Babadzhanov & Kokhirova, 2009), and confirmed in the material of Leonid meteoroids (Borovicka, 2006). Surprisingly low material strength is observed in meteoroid no. 5, which exhibited the highest content of iron in our sample. While this is the second brightest Taurid (estimated -8.1 mag- nitude), no signs of significant effects of saturation/self-absorption, which could cause over-estimation of iron intensity, were observed in the spectrum. There is no clear evidence supporting asteroidal origin in any of the Taurid meteoroids, nevertheless, the connection to carbonaceous chondrites remains unrefuted. Furthermore, the diversity of material strengths among Taurids could also relate to the thermal history of these meteoroids. It is usually observed that meteoroids depleted in sodium are stronger than other parti- cles (Borovicka et al., 2005). However, according to the PE classification, the two Na-enhanced meteoroids in our sample (no. 13 and no. 16) were both identified as group I meteoroids (Table 5). Also the KB classification of Tau- rid meteoroids suggests that particles with slightly enhanced Na/Mg ratios were generally characterized by higher material strength. One explanation of this effect could be the exposure to stronger solar radiation in lower per- ihelia orbits, which may have caused slight structural disruptions, resulting in lower material strengths observed in KB classification. The dependence of material strength parameter KB on sodium content (Figure 10) is ap- parent in the majority of the sample, but was not manifested in a branch of four meteoroids, suggesting this effect has only partial influence on the observed material strength. The observed effect could be however also in- 24 Figure 9: Comparison of the beginning and terminal heights of meteors. Taurid meteoroids (filled marks) are compared to diverse population of 22 meteoroids taken from Rudawska et al. (2016). The marking of meteoroids is the same as in Figure 8. fluenced by the depletion of Na/Mg in brighter meteors, possibly affected by saturation/self-absorption, as was mentioned earlier. The overall physical properties displayed in Table 5 indicate four meteoroids of group I, also char- acterized with higher mineralogical densities, as the best possible candidates for non-cometary origin, however, the asteroidal source cannot be confirmed in any of the studied meteoroids. 5.2. Mineralogical density Mineralogical density is certainly one of the most important physical properties defining the nature meteoroids. There are numerous methods for the determination of meteoroid densities. Bulk densities of meteoroids can be calculated from meteor observations by simultaneously fitting the decel- eration and the light curves of meteors (e.g. Borovicka et al., 2007; Kikwaya et al., 2011). The effective fragmentation model must be taken into account. Mineralogical density represents the density of the substance, irrespective of the structure and shape of the body (Bronshten, 1983). The actual (min- eralogical, or grain) density of the meteoroid material differs from its bulk (volume) density due to the porous structure, presence of voids and porosi- ties, which are characteristic for most meteoroids. We applied the method 25 Figure 10: The dependence of the Na/Mg line intensity ratio on material strength pa- rameter KB. The varying sizes of the marks display photometric masses of individual meteoroids in logarithmic scale. suggested by Ceplecha (1958) and developed by Benyukh (1968), which uses the heat conductivity equation to obtain the mineralogical density of mete- oroids. The equation of heat conductivity can be written in the form: 2TB(λδmc)1/2 Λ = v5/2∞ ρ (b cos zr)1/2 , (6) where TB is the temperature of the frontal surface of a meteoroid; λ is the meteoroid heat conductivity; δm is the mineralogical density of the meteoroid; c is the specific heat of the meteoroid; Λ is the heat transfer coefficient in the beginning of evaporation; v∞ is the pre-atmospheric velocity of the meteoroid; ρ is the atmospheric density; zR is the zenith distance of the radiant; and b = 1/H is the air density gradient, where H is the scale height. The atmospheric parameters (densities, temperatures) were determined from the MSIS-E-90 Atmosphere Model of Hedin (1991). Values of parameters in Eq. 6 are applied at the meteor beginning height. The laboratory data of λ, δm, and c are available for various types of rocks, minerals, and metals (Berch et al., 1949), applied values are given in Table 4. The values of TB and Λ can be assumed for meteoric stone and iron particles. According to Levin (1956), the temperature of a meteoroid frontal surface at the beginning height equals 1600 K for friable stone particles, 2400 K for dense particles, and 2800 K for iron particles. The heat transfer coefficient Λ is assumed to be 1 for stone meteoric particles and 0.75 for iron bodies (Levin, 1956). By applying these values and defining the decimal logarithm 26 Table 4: Laboratory data of mineralogical density (δm), specific heat (c), heat conductivity (λ), and corresponding function f (δm) (Eq. 7) for different materials (Berch et al., 1949). Material δm (g cm-3) c x 107 λ (erg g-1 deg-1) (erg s-1 cm-1 deg-1) f (δm) 9.3 x 103 2.7 x 104 1.7 x 105 1.5 x 106 2.5 x 105 2.0 x 105 1.0 x 106 2.0 x 106 2.4 x 105 3.8 x 106 7.3 x 106 8.72 9.39 9.93 10.13 10.15 10.20 10.25 10.52 10.66 10.82 11.00 Brick (Kizel'gur) Sand (river) Sandstone Schist Granite Basalt Fluorit Fyalit Gematit Arsenopirit Iron 0.40 1.52 2.45 2.67 2.70 2.90 3.20 4.40 5.26 6.07 7.60 0.75 0.70 0.93 0.71 0.65 0.85 0.85 0.55 0.61 0.43 0.44 27 Figure 11: The dependence of the function f (δm) ) (Eq. 7) on mineralogical density δm. The points represent values for different materials assuming physical parameters, which were used by Benyukh (1968). of the left-hand side of Eq. 6 as the function (cid:104)2TB(λδmc)1/2 (cid:105) , (7) f (δm) = log Λ we are able to construct the dependence of f (δm) on δm (Figure 11). The decimal logarithm of the right-hand side of Eq. 6, (cid:104) v5/2∞ ρ (b cos zr)1/2 (cid:105) f (δm) = log , (8) can be calculated from precise video observations and the applied atmosphere model. Using these values of f (δm) in the diagram on Figure 11, we are able to determine the mineralogical density δm of meteoroids. This technique was used by Benyukh (1974) to estimate the mineralogi- cal densities of over 3000 meteoroids from a database of meteors captured by Super-Schmidt cameras. Recently, Babadzhanov & Kokhirova (2009) applied the aforementioned method to determine mineralogical densities of 501 me- teoroids observed by photographic cameras, delivering results in satisfactory 28 agreement with the results of Benyukh (1974). Considering the sensitivity of the AMOS system, which is similar to the sensitivity of Super-Schmidt cameras, we expect to observe comparable beginning heights of meteors and thus comparable results for mineralogical densities. We determined high spread of mineralogical densities in our sample of Taurids ranging from as low as 1.3 g cm-3 to more expected values of approx- imately 2.5 g cm-3 (Table 5). While over half of the meteoroids are positioned close the upper limit of this scale, we also observe a non-negligible popula- tion of low density Taurids. It should be taken into account that the applied technique is highly dependent on the observed beginning height of meteors (Figure 12). Figure 12 reveals that the two most massive meteoroids in our sample are the ones with the lowest mineralogical density. Koten et al. (2004) shown that the beginning heights are functions of meteoroid mass and that this dependence can exhibit different behavior for different meteor showers. The works of Benyukh (1974) and Babadzhanov & Kokhirova (2009) provided mineralogical densities for a sample of photographic Taurids with the mean value of 2.7 ± 0.2 g cm-3. While the heterogeneity of the Taurid stream could provide such observed material variations, the results could also be affected by the slightly higher sensitivity of AMOS system, and the differences in the applied atmospheric model. Nevertheless, we identified compact population of Taurids positioned close to the modus of the mineralogical density values in our sample at around 2.5 g cm-3. Meteoroids with higher mineralogical density expectedly also show higher values of material strength parameters KB and PE. Whereas the relation between mineralogical density and KB is particularly apparent due to the common dependence on meteor beginning height, the PE - density functionality is more complex but still evident. 5.3. Dynamic pressure The bulk strengths of meteoroids can be also inferred from the loading ram pressure causing their fragmentation. During the aerodynamic loading, the internal stresses in the meteoroid are proportional to the exerted ram pressure (Fadeenko, 1967), and so the stagnation pressure at breakup can be used as a measure of meteoroid strength in the disruption (Popova et al., 2011). The fragmentation strength of a meteoroid can be simply taken as the dynamic pressure acting at the front surface of the meteoroid: p = ρhv2 h, 29 (9) Figure 12: The dependence between the mineralogical density and beginning height of Taurid meteoroids. The varying sizes of the marks display photometric masses of individual meteoroids in logarithmic scale. where ρh and vh are the atmospheric density and meteor velocity at the height of the fragmentation. Both of these quantities can be easily measured from the observational data and the atmospheric model, but the determination of the fragmentation height is often difficult. Various methods can be applied for this purpose, depending on the type of available data and the type of meteoroid fragmentation (Popova et al., 2011). The fragmentation heights of the Taurid meteoroids presented here were determined from meteor light curves. The meteor flares (sudden transient increases of brightness) observed mainly in bright fireballs are caused by the sudden release of dust or small fragments from the meteoroid. The position of the flares in Taurid light curves was used to infer the fragmentation height. However, not all of the analyzed Taurid meteors exhibited recognizable flares, and therefore we were not able to determine reliable values of dynamic pres- sures for these, characteristically fainter cases. The inferred fragmentation strengths at first breakup of 8 Taurids for which we observed recognizable flares are in Table 5. Taurid fireballs were previously observed to break-up under 0.05 - 0.18 MPa (Konovalova, 2003). The inferred dynamic pressures in our sample of Taurid meteoroids ranges from 0.02 - 0.10 MPa (Table 5), with the average fragmentation strength of 0.05 MPa placed on the lower limit of the results of Konovalova (2003). Expectedly, meteoroids of higher mineralogical density 30 fragmented under higher pressures than the low-density bodies. The major- ity of C type Taurids were characterized by low fragmentation strengths with very low values observed in the two D type meteors (0.024 and 0.034 MPa). While there are some discrepancies observed between the KB and PE classi- fication with respect to determined dynamic pressures, the obtained values are in general characteristic for fragile cometary bodies. Tensile strengths of average small shower meteoroids were examined by Trigo-Rodr´ıguez & Llorca (2006), who found that next to Geminids, the Taurids contained the strongest shower meteoroids, with bulk densities in some fireballs compa- rable to the Tagish Lake meteorite (Hildebrand et al., 2006). In a survey of instrumentally observed meteorite falls, Popova et al. (2011) found that these bodies show very low bulk strengths of 0.1 to 1 MPa on first breakup. The lower limits of this range are characteristic for the weakest observed meteorite falls - carbonaceous chondrites Tagish Lake, Maribo and Sutter's Mill (Borovicka et al., 2015). Based on the inferred dynamic pressures in our sample, we can assume that Taurid meteoroids are on average weaker compared carbonaceous chondrites. Still, the stream could include particles of comparable mechanical strengths, as the upper limits of the determined fragmentation pressures reach 0.1 MPa. The three instrumentally observed carbonaceous meteorite falls have shown several common characteristics. All objects were initially very large in size and showed features indicating very fragile structure (early fragmentation, high end heights). The very similar orbits of Maribo and Sutter's Mill suggest common and relatively recent ori- gin (Jenniskens et al., 2012). The entry velocities of these meteorite falls are around 28 km/s, which is also the average speed of Taurid meteors, and very close to the cut-off velocity suggested by modeling for meteorite production (ReVelle, 1979). While the Taurids remain the most promising major meteor shower candidate for meteorite recovery (Brown et al., 2013; Madiedo et al., 2014; Brown et al., 2016), the low strengths of its meteoroids, though in the upper limits comparable to the carbonaceous chondrites, would require significant initial mass of the impacting body. 6. Conclusion We presented an analysis of spectral and physical properties of Taurid meteoroids observed during the outburst in November 2015. In a sample of 14 multi-station Taurid meteors with spectra, we identified 10 Southern Taurids and 4 meteors belonging to the Northern Taurids shower. The Tis- 31 Table 5: Physical properties of the studied multi-station Taurid meteoroids. Each Taurid is designated with meteor number, absolute magnitude (M ag), beginning height (HB), terminal height (HE), Na/Mg intensity ratio, Fe/Mg intensity ratio, spectral class, ma- terial strength parameters (KB, PE) and corresponding material type, dynamic pressure (p) and mineralogical density (δm). The uncertainties of photometeric masses are in the given order of magnitude; the accuracy of the beginning and terminal heights is on the order of the last digit. Meteor IDs are the same as in Table 1. No. M ag Mp(g) HB(km) HE(km) Na/Mg Fe/Mg Class KB PE p(MPa) δm(g cm−3) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 16 -4.7 ± 0.7 -4.5 ± 0.7 -5.1 ± 0.8 -6.0 ± 1.2 -8.1 ± 2.0 -2.1 ± 0.7 -5.6 ± 2.0 -1.8 ± 0.7 -8.4 ± 1.4 -1.9 ± 0.6 -5.0 ± 0.9 -5.5 ± 0.6 -1.3 ± 0.9 -3.9 ± 0.9 -2.9 ± 0.7 40 60 50 150 104.0 99.2 99.1 99.7 1000 108.8 1 40 1 99.3 - 99.2 1500 106.0 1 70 99.3 104.9 150 104.4 1 50 10 103.0 99.8 98.1 70.1 52.7 69.3 60.7 56.6 72.2 - 73.4 60.5 68.5 63.7 68.1 62.8 54.5 57.0 1.74 0.14 1.59 0.16 1.30 0.14 1.17 0.09 1.23 0.05 2.61 0.50 1.48 0.07 1.73 0.35 1.59 0.07 2.39 0.22 1.49 0.06 1.57 0.09 2.90 0.38 2.12 0.09 2.44 0.23 0.65 0.08 1.18 0.13 0.77 0.10 1.05 0.09 1.73 0.06 1.70 0.37 1.15 0.06 0.72 0.22 1.10 0.06 0.61 0.10 0.61 0.04 0.80 0.06 0.60 0.16 0.64 0.05 0.39 0.09 Fe poor normal normal normal normal normal normal -5.32 6.76 C1 0.04 0.15 7.11 B -4.36 0.13 0.01 -5.35 7.05 C1 0.02 0.15 -4.98 7.05 C1 0.01 0.21 6.34 D -5.13 0.36 0.02 -5.00 7.09 C1 0.01 0.11 - - - normal normal -5.04 7.07 C1 0.02 0.13 6.53 D -5.51 0.03 0.35 -4.90 7.04 C1 0.11 0.02 -5.14 6.61 C1 0.16 0.02 -5.17 6.74 C1 0.01 0.11 -4.46 0.16 -4.42 0.14 Na enhanced 7.15 B -4.49 0.11 0.02 6.93 C1 0.02 Na enhanced 6.74 C1 Fe poor Fe poor Fe poor normal 0.01 IIIA I IIIA II II II - II IIIA II II II I I I - - 0.063 0.004 0.099 0.003 0.034 0.002 - - - 0.024 0.002 - 0.018 0.002 0.027 0.002 - 0.054 0.003 0.081 0.002 1.94 0.18 2.50 0.05 2.43 0.15 2.44 0.10 1.32 0.15 2.48 0.07 - 2.47 0.13 1.58 0.15 2.41 0.13 1.70 0.08 1.93 0.13 1.92 0.08 2.23 0.02 2.54 0.03 32 serand's parameter defined the orbital origin of the sample on the borderline between asteroidal orbits (10 meteors) and Jupiter-family type cometary or- bits (4 meteors), as would be expected for a stream originating in the short- period comet 2P/Encke. Furthermore, we used a simple method based on the Southworth-Hawkins DSH orbital similarity criterion to look for associ- ations with related Taurid complex NEOs. We found possible association to 10 asteroids satisfying the similarity threshold value of DSH ≤ 0.10. The most meteor associations in the southern branch of our sample was found with asteroids 2015 TX24, 2003 UV11, and 2007 UL12. Meteoroids from the northern branch showed most markable similarities with asteroid 2014 NK52. However, we argue that based on the determined spectral and physical prop- erties of Taurid meteoroids, these associations may be purely incidental. The spectral analysis revealed large dispersion of iron content in Taurid meteoroids, ranging from Fe-poor bodies to meteoroids with almost chon- dritic Fe/Mg ratios. Na content also varied slightly and was found to be a function of the meteoroid perihelion distance, assigning Na loss to space weathering processes, particularly the thermal desorption. It was also found that the Na enrichment is preferred in smaller particles. Nevertheless, the increased Na line intensity over Mg was found in all of the studied spectra, indicating significant volatile content in Taurid meteoroids. The overall spec- tral classification of Taurid meteors observed in Slovakia and Chile showed corresponding results dominated by normal-type spectral classes and several cases of Fe-poor and Na-enhanced meteoroids. Besides the almost chondritic content of iron in few of the studied cases, there are no spectral features clearly suggesting asteroidal origin in any of the analyzed meteoroids. Overall, the determined material strengths showed typically cometary characteristics in the KB classification, dominated by C group meteoroids, and much more dispersed values in the PE classification, in which the major- ity of Taurid meteoroids fall in the group II, characteristic for carbonaceous bodies. The discovered features suggest that Taurids are heterogeneous pop- ulation of meteoroids, which are cometary in nature but contain solid, possi- bly carbonaceous inclusions. The diversity of the observed material may be caused by the reprocessing of meteoroid surfaces, but possibly also reflects the authentic inhomogeneity of 2P/Encke or earlier parent comet of the stream. Four meteoroids of group I characterized with higher mineralogical densities were identified as the best possible candidates for non-cometary origin, how- ever, the asteroidal source cannot be confirmed in any of the meteoroids in our sample. Moreover, we also found high dispersion of mineralogical densi- 33 ties among the Taurid meteoroids. Obtained very low densities in individual cases were probably influenced by the high beginning heights of these mete- ors, which relate to the sensitivity of the observational system. Nevertheless, the largest population was identified with mineralogical densities around 2.5 g cm-3, which is comparable to the mean values of 2.7 g cm-3 determined by Benyukh (1974) and Babadzhanov & Kokhirova (2009). Mechanical strength of several Taurids, which exhibited bright flares, was also estimated from the dynamic pressure causing their fragmentation. It was found that studied me- teoroids started fragmenting under low pressures of 0.02 - 0.10 MPa. While these values suggest that Taurid meteoroids are on average weaker compared to the instrumentally observed carbonaceous meteorite falls (first breakup at 0.1 - 1 MPa), the stream could include bodies of strength comparable to carbonaceous chondrites, as the upper limits of dynamic pressures in our sample reach 0.1 MPa. Acknowledgements. This work was supported by the Slovak Research and Development Agency under the contract No. APVV-0516-10, No. APVV- 0517-12, and by the Slovak Grant Agency for Science, grants No. VEGA 1/0225/14. References Asher D. J., 1991, The Taurid meteoroid complex, PhD thesis, New College, Oxford Asher D. J., Clube S. V. M., Steel D. I., 1993, MNRAS, 264, 93 Asher D. J., Izumi K., 1998, MNRAS, 297, 23 Babadzhanov P. B., 2001, A&A, 373, 329 Babadzhanov P. B., Kokhirova G. I., 2009, A&A, 495, 1, 353 Benyukh V. V., 1968, Vestnik KGU, Seriya astronomii, 10, 51 Benyukh V. V., 1974, PhD. Dissertation Fotometricheskoe issledovanie me- teorov i plotnost meteornych tel, Kiev, 148 Berch et al., 1949, Spravochnik dlya geologov po fizicheskim constantam, Moscow, Inostr. literatura 34 Borovicka J., 1993, A&A, 279, 627 Borovicka J., 1994, Planet. Space Sci., 42, 2, 145 Borovicka J., Spurn´y P., Kecl´ıkov´a J., 1995, Astron. Astrophy. Suppl., 112, 173 Borovicka J., Koten P., Spurn´y P., Bocek J., Stork R., 2005, Icarus, 194, 15 Borovicka J., 2006, In Asteroids, Comets, Meteors 2005, ed. D. Lazzaro, S. Ferraz-Mello, J. A. Fernandez, Proc. IAU Symp., 229, 249 Borovicka J., Spurn´y P., Koten P., 2007, A&A, 473, 661 Borovicka J., Spurn´y P., Brown P., 2015, Small near-Earth asteroids as a source of meteorites. In Asteroids IV, ed. Michel P., DeMeo F. E., and Bottke W. F. Jr., The University of Arizona Press, arXiv:1502.03307 Bronshten V. A., 1983, The Physics of Meteoritic Phenomena, Dordrecht, D. Reidel Publishing Co. Brown P., Marchenko V., Moser D. E., Weryk R., Cooke W., 2013, Meteorit. Planet. Sci., 48, 270 Brown, P. Wiegert P., Clark D., Tagliaferri E., 2016, Icarus, 266, 96 Ceplecha Z., 1958, Bull. Astron. Inst. of Czech. Ac. of Sci., 9, 4, 154 Ceplecha Z., 1964, Bull. Astron. Inst. of Czech., 15, 102 Ceplecha Z., 1968, SAO Special report, 279 Ceplecha Z., McCrosky R. E., 1976, J. Geophys. Res., 81, 6257 Ceplecha Z., 1987, Bull. Astron. Inst. of Czech., 38, 222 Ceplecha Z., 1988, Bull. Astron. Inst. of Czech., 39, 221 Ceplecha Z., Borovicka J., Elford W., ReVelle D., Hawkes R.L., Porubcan V., Simek M., 1998, Space Sci. Rev., 84, 327 Despois D., 1992, Proc. 150th Symp. Int. Astron. Union, Kluwer, Dordrecht, 451 35 Fadeenko Y., 1967, Fizika Gorenija i Vzryva, 3, 276 Haack H., Michelsen R., Stober G., Keuer D., Singer W., Williams I., 2011, Meteorit. Planet. Sci. Suppl., 74, 5271 Haack H., Grau T., Bischoff A., Horstmann M., Wasson J., Sorensen A., Laubenstein M., Ott U., Palme H., Gellissen M., Greenwood R. C., Pearson V. K., Franchi I. A., Gabelica Z., Schmitt-Kopplin P., 2012, Meteorit. Planet. Sci., 47, 30 Hedin A. E., 1991, J. Geophys. Res., 96, 1159 Hildebrand A. R., McCausland P. J. A., Brown P. G., Longstaffe F. J., Russell S. D. J., Tagliaferri E., Wacker J. F., Mazur M. J., 2006, Meteorit. Planet. Sci., 41, 3 Hill K. A., Rogers L. A., Hawkes R. L., 2005, A&A, 444, 615 Jenniskens P., Fries M.D., Yin Q.-Z., et al., 2012, Science, 338, 1583 Jones W., Halliday I., 2001, MNRAS, 320, 4, 417 Kikwaya J.-B., Campbell-Brown M., Brown P.G., 2011, A&A, 530, A113, 1 Konovalova N. A., 2003, A&A, 404, 1145 Koten P., Borovicka J., Spurn´y P., Betlem H., Evans S., 2004, A&A, 428, 683 Kres´ak L., 1978, Bull. Astron. Inst. Czech., 29, 129 Levin B. Y., 1956, Physical Theory of Meteors and Meteor Matter in the Solar System, USSR Academy of Sciences, Moscow Levison H. F., Terrell D., Wiegert P. A., Dones L., Duncan M. J., 2006, Icarus, 225, 475 Madiedo J. M., Ortiz J. L., Trigo-Rodr´ıguez J. M., Dergham J., Castro- Tirado A. J., Cabrera-Cano, J., Pujols P., 2014, Icarus, 231, 356 Napier W. M., 2010, MNRAS, 405, 3, 1901 Nagasawa K., 1978, Annals Tokyo Astron. Obs., Second Series, 16, 4 36 Olech A., Zo(cid:32)l¸adek P., Wisniewski M., Rudawska R., Beben M., Krzyzanowski T., Myszkiewicz M., Stolarz M., Gawronski M., Gozdalski M., Suchodolski T., Wegrzyk W., Tyminski Z., 2016, MNRAS, 461, 1, 674 Popescu M., Birlan M., Nedelcu D. A., Vaubaillon J., Cristescu C. P., 2014, A&A, 572, A106 Popova O., Borovicka J., Hartmann W. K., Spurn´y P., Gnos E., Nemtchinov I., Trigo-Rodr´ıguez J. M., 2011, Meteorit. Planet. Sci., 46, 10, 1525 Porubcan V., Kornos L., Williams I. P., 2006, Contrib. Astron. Obs. Skalnat´e Pleso, 36, 103 ReVelle D. O., 1979, JASTP, 41, 453 Rudawska R., T´oth J., Kalmancok D., Zigo P., Matlovic P., 2016, Planet. Space Sci., 123, 25 Shiba Y., 2016, WGN, Journal of the IMO, 44, 3 SonotaCo, 2009, WGN, Journal of the IMO, 37, 55 Southworth R. B., Hawkins G. S., 1963, Smithson. Contrib. Astrophys., 7, 261 Spurn´y P., Borovicka J., Svoren J., Mucke H., 2016, A&A, in preparation Srirama Rao M., Ramesh P., 1965, Nature, 205, 164 T´oth J., Kornos L., Veres P., Silha J., Kalmancok D., Zigo P., Vil´agi J., 2011, Publ. Astron. Soc. Japan, 63, 331 T´oth J., Kornos L., Zigo P., Gajdos S., Kalmancok D., Vil´agi J., Simon J., Veres P., Silha J., Bucek M., Gal´ad A., Rusn´ak P., Hr´abek P., Duris F., Rudawska R., 2015, Planet. Space Sci., 118, 102 Trigo-Rodr´ıguez J. M., Llorca J., 2006, MNRAS, 372, 2, 655 Trigo-Rodr´ıguez J. M., Llorca J., 2007, Adv. Space Res., 39, 4 Tubiana C., Snodgrass C., Michelsen R., Haack H., Bohnhardt H., Fitzsim- mons A., Williams I. P., 2015, A&A, 584, A97 37 Voj´acek V., Borovicka J., Koten P., Spurn´y P., Stork R, 2015, A&A, 580, A67, 31 Wetherill G. W., 1974, Annu. Rev. Earth Planet. Sci., 2, 303 Wetherill G. W., Revelle D. O., 1981, Icarus, 48, 308 Whipple F., 1940, Proc. Am. Phil. Soc., 83, 711 38
1503.03516	2	1503	2016-10-22T17:40:09	Understanding The Effects Of Stellar Multiplicity On The Derived Planet Radii From Transit Surveys: Implications for Kepler, K2, and TESS	[ "astro-ph.EP" ]	We present a study on the effect of undetected stellar companions on the derived planetary radii for the Kepler Objects of Interest (KOIs). The current production of the KOI list assumes that the each KOI is a single star. Not accounting for stellar multiplicity statistically biases the planets towards smaller radii. The bias towards smaller radii depends on the properties of the companion stars and whether the planets orbit the primary or the companion stars. Defining a planetary radius correction factor $X_R$, we find that if the KOIs are assumed to be single, then, {\it on average}, the planetary radii may be underestimated by a factor of $\langle X_R \rangle \approx 1.5$. If typical radial velocity and high resolution imaging observations are performed and no companions are detected, this factor reduces to $\langle X_R \rangle \approx 1.2$. The correction factor $\langle X_R \rangle$ is dependent upon the primary star properties and ranges from $\langle X_R \rangle \approx 1.6$ for A and F stars to $\langle X_R \rangle \approx 1.2$ for K and M stars. For missions like K2 and TESS where the stars may be closer than the stars in the Kepler target sample, observational vetting (primary imaging) reduces the radius correction factor to $\langle X_R \rangle \approx 1.1$. Finally, we show that if the stellar multiplicity rates are not accounted for correctly, occurrence rate calculations for Earth-sized planets may overestimate the frequency of small planets by as much as $15-20$\%.	astro-ph.EP	astro-ph	Accepted for publication in The Astrophysical Journal Preprint typeset using LATEX style emulateapj v. 5/2/11 6 1 0 2 t c O 2 2 . ] P E h p - o r t s a [ 2 v 6 1 5 3 0 . 3 0 5 1 : v i X r a UNDERSTANDING THE EFFECTS OF STELLAR MULTIPLICITY ON THE DERIVED PLANET RADII FROM TRANSIT SURVEYS: IMPLICATIONS FOR KEPLER, K2, AND TESS David R. Ciardi1, Charles A. Beichman1, Elliott P. Horch2, Steve B. Howell3 Accepted for publication in The Astrophysical Journal ABSTRACT We present a study on the effect of undetected stellar companions on the derived planetary radii for the Kepler Objects of Interest (KOIs). The current production of the KOI list assumes that the each KOI is a single star. Not accounting for stellar multiplicity statistically biases the planets towards smaller radii. The bias towards smaller radii depends on the properties of the companion stars and whether the planets orbit the primary or the companion stars. Defining a planetary radius correction factor XR, we find that if the KOIs are assumed to be single, then, on average, the planetary radii may be underestimated by a factor of hXRi ≈ 1.5. If typical radial velocity and high resolution imaging observations are performed and no companions are detected, this factor reduces to hXRi ≈ 1.2. The correction factor hXRi is dependent upon the primary star properties and ranges from hXRi ≈ 1.6 for A and F stars to hXRi ≈ 1.2 for K and M stars. For missions like K2 and TESS where the stars may be closer than the stars in the Kepler target sample, observational vetting (primary imaging) reduces the radius correction factor to hXRi ≈ 1.1. Finally, we show that if the stellar multiplicity rates are not accounted for correctly, occurrence rate calculations for Earth-sized planets may overestimate the frequency of small planets by as much as 15 − 20%. Subject headings: (stars:) planetary systems, (stars:) binaries: general 1. INTRODUCTION The Kepler Mission Borucki et al. (2010), with the dis- covery of over 4100 planetary candidates in 3200 sys- tems, has spawned a revolution in our understanding of planet occurrence rates around stars of all types. One of Kepler's profound discoveries is that small planets (Rp . 3R⊕) are nearly ubiquitous (e.g., Howard et al. 2012; Dressing & Charbonneau 2013; Fressin et al. 2013; Petigura et al. 2013; Batalha 2014) and, in particular, some of the most common planets have sizes between Earth-sized and Neptune-sized -- a planet type not found in our own solar system. Indeed, it is within this group of super-Earths to mini-Neptunes that there is a tran- sition from "rocky" planets to "non-rocky planets"; the transition is near a planet radius of 1.6R⊕ and is very sharp -- occurring within ≈ 0.2R⊕ of this transition ra- dius (Marcy et al. 2014; Rogers 2014). Unless an intra-system comparison of planetary radii is performed where only the relative planetary sizes are important (Ciardi et al. 2013), having accurate (as well as precise) planetary radii is crucial to our comprehen- sion of the distribution of planetary structures. In par- ticular, understanding the radii of the planets to within ∼ 20% is necessary if we are to understand the relative occurrence rates of "rocky" to "non-rocky" planets, and the relationship between radius, mass, and bulk density.. While there has been a systematic follow-up observa- tion program to obtain spectroscopy and high resolution imaging, only approximately half of the Kepler candi- date stars have been observed (mostly as a result of the [email protected] 1 NASA Exoplanet Science Institute/Caltech Pasadena, CA USA 2 Department of Physics, Southern Connecticut State Univer- brightness distribution of the candidate stars). Those stars that have been observed have been done mostly to eliminate false positives, to determine the stellar param- eters of host stars, and to search for nearby stars that may be blended in the Kepler photometric apertures. Stars that are identified as possible binary or triple stars are noted on the Kepler Community Follow-Up Observation Program website4, and are often handled in individual papers (e.g., Star et al. 2014; Everett et al. 2014). The false positive assessment of an KOI (or all of the KOIs) can take into account the likelihood of stel- lar companions (e.g., Morton & Johnson 2011; Morton 2012), and a false positive probability will likely be in- cluded in future KOI lists. But presently, the current production of the planetary candidate KOI list and the associated parameters are derived assuming that all of the KOI host stars are single. That is, the Kepler pipeline treats each Kepler candidate host star as a single star (e.g., Batalha et al. 2011; Burke et al. 2014; Mullally et al. 2014). Thus, statistical studies based upon the Kepler candidate lists are also assuming that all the stars in the sample set are single stars. The exact fraction of multiple stars in the Kepler can- didate list is not yet determined, but it is certainly not zero. Recent work suggests that a non-negligible fraction (∼ 30 − 40%) of the Kepler host stars may be multiple stars (Adams et al. 2012, 2013; Law et al. 2014; Dressing et al. 2014; Horch et al. 2014), although other work may indicate that (giant) planet formation may be suppressed in multiple star systems (Wang et al. 2014a,b). The presence of a stellar companion does not necessarily invalidate a planetary candidate, but it does change the observed transit depths and, as a result, the planetary radii. Thus, assuming all of the stars in the Ke- sity, New Haven, CT, USA 3 NASA Ames Research Center, Mountain View, CA, USA 4 https://cfop.ipac.caltech.edu/ 2 pler candidate list are single can introduce a systematic uncertainty into the planetary radii and occurrence rate distributions. This has already been discussed for the oc- currence rate of hot Jupiters in the Kepler sample where it was found that ∼ 13% of hot Jupiters were classi- fied as smaller planets because of the unaccounted effects of transit dilution from stellar companions (Wang et al. 2014c). In this paper, we explore the effects of undetected stel- lar gravitationally bound companions on the observed transit depths and the resulting derived planetary radii for the entire Kepler candidate sample. We do not consider the dilution effects of line-of-sight background stars, rather only potential bound companions, as com- panions within 1′′ are most likely bound companions (e.g., Horch et al. 2014; Gilliland et al. 2014), and most stars beyond 1′′ are either in the Kepler Input Catalog (Brown et al. 2011) or in the UKIRT survey of the Ke- pler field and, thus, are already accounted for with re- gards to flux dilution in the Kepler project transit fitting pipeline. Within 1′′, the density of blended background stars is fairly low, ranging between 0.001−0.007 stars/(cid:3)′′ (Lillo-Box et al. 2014). Thus, within a radius of 1′′, we expect to find a blended background (line-of-sight) star only 0.3−2% of the time. Therefore, the primary contam- inant within 1′′ of the host stars are bound companions. We present here probabilistic uncertainties of the plan- etary radii based upon expected stellar multiplicity rates and stellar companion sizes. We show that, in the ab- sence of any spectroscopic or high resolution imaging observations to vet companions, the observed planetary radii will be systematically too small. However, if a can- didate host star is observed with high resolution imaging (HRI) or with radial velocity (RV) spectroscopy to screen the star for companions, the underestimate of the true planet radius is significantly reduced. While imaging and radial velocity vetting is effective for the Kepler candi- date host stars, it will be even more effective for the K2 and TESS candidates which will be, on average, 10 times closer than the Kepler candidate host stars. 2. EFFECTS OF COMPANIONS ON PLANET RADII The planetary radii are not directly observed; rather, the transit depth is the observable which is then related to the planet size. The observed depth (δo) of a plane- tary transit is defined as the fractional difference in the measured out-of-transit flux (Ftotal) and the measured in-transit flux (Ftransit): δo = Ftotal − Ftransit Ftotal . (1) If there are N stars within a system, then the total out- of-transit flux in the system is given by Ftotal = N Xi=1 Fi, (2) and if the planet transits the tth star in the system, then the in -- transit flux can be defined as Ftransit = Ftotal − Ft (Rp/Rt⋆)2 (3) where Ft is the flux of the star with the transiting planet, Rp is the radius of the planet, and Rt⋆ is the radius of the star being transited. Substituting into equation (1), the generalized transit depth equation (in the absence of limb darkening or star spots) becomes δo = (cid:18) Ft Ftotal(cid:19)(cid:18) Rp Rt⋆(cid:19)2 . (4) For a single star, Ftotal = Ft and the transit depth expression simplifies to just the square of the size ratio between the planet and the star. However, for a multiple star system, the relationship between the observed tran- sit depth and the true planetary radius depends upon the brightness ratio of the transited star to the total bright- ness of the system and on the stellar radius which changes depending on which star the planet is transiting: Rp = Rt⋆rδo Ftotal Ft . (5) The Kepler planetary candidates parameters are esti- mated assuming the star is a single star (Batalha et al. 2011; Burke et al. 2014; Mullally et al. 2014), and, there- fore, may incorrectly report the planet radius if the stel- lar host is really a multiple star system. The extra flux contributed by the companion stars will dilute the ob- served transit depth, and the derived planet radius de- pends on the size of star presumed to be transited. The ratio of the true planet radius, Rp(true), to the observed planet radius assuming a single star with no companions, Rp(observed), can be described as: XR ≡ Rp(true) Rp(observed) = (cid:18) Rt⋆ R1⋆(cid:19)r Ftotal Ft , (6) where R1⋆ is the radius of the (assumed single) primary star, and Ft and Rt⋆ are the brightness and the radius, respectively, of the star being transited by the planet. This ratio reduces to unity in the case of a single star (Rt⋆ ≡ R1⋆ and Ftotal ≡ Ft). For a multiple star system where the planet orbits the primary star (Rt⋆ ≡ R1⋆), the planet size is underestimated only by the flux dilution factor: XR ≡ Rp(true) Rp(observed) = r Ftotal F1 . (7) However, if the planet orbits one of the companion stars and not the primary star, then the ratio of the primary star radius (R1⋆) to the radius of the companion star be- ing transited (Rt⋆) affects the observed planetary radius, in addition to the flux dilution factor. 3. POSSIBLE COMPANIONS FROM ISOCHRONES To explore the possible effects of the undetected stel- lar companions on the derived planetary parameters, we first assess what companions are possible for each KOI. For this work, we have downloaded the Cumulative Ke- pler candidate list and stellar parameters table from the NASA Exoplanet Archive. The cumulative list is up- dated with each new release of the KOI lists5; as a result, the details of any one star and planet may have changed since the analysis for this paper was done. However, the 5 The tive KOI http://exoplanetarchive.ipac.caltech.edu/ 2014 October cumula- table was used in the work presented here: update the 23 to 3 Figure 1. Radii corrections factors (XR) are plotted as a function of companion-to-primary brightness ratios (bottom axis) and mass ratios (top axis) for possible binary systems (top plot) or triple (bottom plot) systems. This figure is an example for the G-dwarf KOI-299; similar calculations have been made for every KOI. In each plot, the dark blue stars represent the correction factors if the planet orbits the primary star (equation 7); the red circles represent the correction factors if the planet orbits the secondary star, and the light blue triangles represent the correction factors if the planet orbits the tertiary star (equation 6). The lines are third order polynomials fit to the distributions. Unity is marked with a horizontal dashed line. overall results of the paper presented here should remain largely unchanged. For the KOI lists, the stellar parameters for each KOI were determined by fitting photometric colors and spectroscopically derived parameters (where available) to the Dartmouth Stellar Evolution Database (Dotter et al. 2008; Huber et al. 2014). The planet parameters were then derived based upon the transit curve fitting and the associated stellar parameters. Other stars listed in the Kepler Input Catalog or UKIRT imaging that may be blended with the KOI host stars were accounted for in the transit fitting, but, in general, as mentioned above, each planetary host star was assumed to be a single star. We have restricted the range of possible bound stellar companions to each KOI host star by utilizing the same Dartmouth isochrones used to determine the stellar pa- rameters. Possible gravitationally bound companions are assumed to lie along on the same isochrone as the pri- mary star. For each KOI host star, we found the single best fit isochrone (characterized by mass, metallicity, and age) by minimizing the chi-square fit to the stellar pa- rameters (effective temperature, surface gravity, radius, and metallicity) listed in the KOI table. Figure 2. This figure is the same as Figure 1, but for the M-dwarf KOI-1085, demonstrating that the details of the derived correction factors are dependent upon the KOI properties. We did not try to re-derive stellar parameters or in- dependently find the best isochrone fit for the star; we simply identified the appropriate Dartmouth isochrone as used in the determination of the stellar parameters (Huber et al. 2014). We note that there exists an ad- ditional uncertainty based upon the isochrone finding. In this work, we did not try to re-derive the stellar pa- rameters of the host stars, but rather, we simply find the appropriate isochrone that matches the KOI stellar parameters. Thus, any errors in the stellar parameters derivations in the KOI list are propagated here. This is likely only a significant source of uncertainty for nearly equal brightness companions. Once an isochrone was identified for a given star, all stars along an isochrone with (absolute) Kepler magni- tudes fainter than the (absolute) Kepler magnitude of the host star were considered to be viable companions; i.e., the primary host star was assumed to be the brightest star in the system. The fainter companions listed within that particular isochrone were then used to establish the range of possible planetary radii corrections (equation 6) assuming the host star is actually a binary or triple star. Higher order (e.g., quadruple) stellar multiples are not considered here as they represent only ∼ 3% of the stellar population (Raghavan et al. 2010). We have considered six specific multiplicity scenarios: 1. single star (Xr ≡ 1.0) 2. binary star -- planet orbits primary star 4 Figure 3. Top: The mean correction factor hXRi for each KOI is displayed as a function of the effective temperature of the primary star (see section 4). The black curves are 3rd-order polynomials fitted to the distributions (equation 8). Bottom: Histograms of the correction factors displayed in the top panels. The vertical dashed lines mark the medians of the distributions. Left: The correction factors are computed for the Kepler Cumulative Kepler Objects of Interest list. Right: The corrections are computed for the KOIs but assuming the KOIs are 10 times closer as may be the case for K2 and TESS. The red points and histograms assume each KOI is single as is the case for the published KOI list; the blue points and histograms assume that each KOI has been vetted with radial velocity (RV) and high resolution imaging (HRI), and all stellar companions with orbital periods of 2 years or shorter and all stellar companions located at angular distances of ≥ 0.1′′ have been detected and accounted for in the planetary radii determinations (see Section 4.2). For the vetted stars, the correction factors are only for undetected stellar companions; detected companions have been assumed to be accounted for in the planet radii determinations. 3. binary star -- planet orbits secondary star 4. triple star -- planet orbits primary star 5. triple star -- planet orbits secondary star 6. triple star -- planet orbits tertiary star. Based upon the brightness and size differences between the primary star and the putative secondary or tertiary companions, we have calculated for each KOI the pos- sible factor by which the planetary radii are underesti- mated (XR). If the star is single, the correction factor is unity, and if, in a multiple star system, the planet orbits the primary star, only flux dilution affects the observed transit depth and the derived planetary radius (eq. 7). For the scenarios where the planets orbits the sec- ondary or tertiary star, the planet size correction fac- tors (eq. 6) were determined only for stellar companions where the stellar companion could physically account for the observed transit depth. If more than 100% of the stel- lar companion light had to be eclipsed in order to produce the observed transit in the presence of the flux dilution, then that star (and all subsequent stars on the isochrone with lower mass) was not considered viable as a poten- tial source of the transit. For example, for an observed 1% transit, no binary companions can be fainter than the primary star by 5 magnitudes or more; an eclipse of such a secondary star would need to be more than 100% deep. The stellar brightness limits were calculated inde- pendently for each planet within a KOI system so as to not assume that all planets within a system necessarily orbited the same star. Figures 1 and 2 show representative correction factors (XR) for KOI-299 (a G-dwarf with a super-Earth sized Rp = 1.8 ± 0.24R⊕ planet) and for KOI-1085 (an M- dwarf with an Earth-sized Rp = 0.92 ± 0.13R⊕ planet). The planet radius correction factors (XR) are shown as a function of the companion -- to -- primary brightness ratio (bottom x-axis of plots) and the companion -- to -- primary mass ratio (top x-axis of plots) and are determined for the KOI assuming it is a binary-star system (top plot) or a triple-star system (bottom plot). The amplitude of the correction factor (XR) varies strongly depending on the particular system and which star the planet may orbit. If the planet orbits the pri- mary star, then the largest the correction factors are for equal brightness companions (√2 ∼ 1.4 for a binary sys- tem and √3 ∼ 1.7 for a triple system) with an asymp- totic approach to unity as the companion stars become fainter and fainter. If the planet orbits the secondary or tertiary star, the planet radius correction factor can be significantly larger -- ranging from XR & 2− 5 for binary systems and XR & 2 − 20 or more for triple systems -- depending on the size and brightness of the secondary or tertiary star. 4. MEAN RADII CORRECTION FACTORS (XR) It is important to recognize the full range of the possi- ble correction factors, but in order to have a better under- standing of the statistical correction any given KOI (or the KOI list as a whole) may need, we must understand the mean correction for any one multiplicity scenario and convert these into a single mean correction factor for each star. To do this, we must take into account the proba- bility the star may be a multiple star, the distribution of mass ratios if the star is a multiple, the probability that the planet orbits any one star if the stellar system has multiple stars, and whether or not the star has been vetted (and how well it is has been vetted) for stellar companions. In order to calculate an average correction factor for each multiplicity scenario, we have fitted the individual scenario correction factors as a function of mass ratio with a 3rd-order polynomial (see Fig. 1 and 2). Because the isochrones are not evenly sampled in mass, taking a mean straight from the isochrone points would skew the results; the polynomial parameterization of the correc- tion factor as a function of the mass ratio enables a more robust determination of the mean correction factor for each multiplicity scenario. If the companion -- to -- primary mass ratio distribution was uniform across all mass ratios, then a straight mean of the correction values determined from each polynomial curve would yield the average correction factor for each multiplicity scenario. However, the mass ratio distribu- tion is likely not uniform, and we have adopted the form displayed in Figure 16 of Raghavan et al. (2010). That distribution is a nearly-flat frequency distribution across all mass ratios with a ∼ 2.5× enhancement for nearly equal mass companion stars (q & 0.95). This distribu- tion is in contrast to the Gaussian distribution shown in Duquennoy & Mayor (1991); however, the more recent results of Raghavan et al. (2010) incorporate more stars, a broader breadth of stellar properties, and multiple com- panion detection techniques. The mass ratio distribution is convolved with the poly- nomial curves fitted for each multiplicity scenario, and a weighted mean for each multiplicity scenario was calcu- lated for every KOI. For example, in the case of KOI-299 (Fig. 1), the single star mean correction factor is 1.0 (by definition). For the binary star cases, the average sce- nario correction factors are 1.14 (planet orbits primary) 5 and 2.28 (planet orbits secondary); for the triple stars cases, the correction factors are 1.16 (planet orbits pri- mary), 2.75 (planet orbits secondary), and 4.61 (planet orbits tertiary). For KOI-1085 (Fig. 2), the weighted mean correction factors are 1.18, 1.56, 1.24, 1.61, and 2.29, respectively. To turn these individual scenario correction factors into an overall single mean correction factor hXRi per KOI, the six scenario corrections are convolved with the probability that a KOI will be a single star, a binary star, or a triple star. The multiplicity rate of the Kepler stars is still unclear (Wang et al. 2014b), and, indeed, there may be some contradictory evidence for the the exact value for the multiplicity rates of the KOI host stars (e.g., Horch et al. 2014; Wang et al. 2014b), but the multiplic- ity rates appear to be near 40%, similar to the general field population. In the absence of a more definitive esti- mate, we have chosen to utilize the multiplicity fractions from Raghavan et al. (2010): a 54% single star fraction, a 34% binary star fraction, and a 12% triple star fraction (Raghavan et al. 2010). We have grouped all higher or- der multiples (3+) into the single category of "triples", given the relatively rarity of the quadruple and higher order stellar systems. For the scenarios where there are multiple stars in a system, we have assumed that the planets are equally likely to orbit any one of the stars (50% for binaries, 33.3% for triples). The final mean correction factors hXRi per KOI are displayed in Figure 3; the median value of the correc- tion factor and the dispersion around that median is hXRi = 1.49 ± 0.12. This median correction factor im- plies that assuming a star in the KOI list is single, in the absence of any (observational) companion vetting, yields a statistical bias on the derived planetary radii where the radii are underestimated, on average, by a factor ∼ 1.5, and the mass density of the planets are overestimated by a factor of ∼ 1.53 ∼ 3. From Figure 3, it is clear that the mean correction factor hXRi depends upon the stellar temperature of the host star. As most of the stars in the KOI list are dwarfs, the lower temperature stars are typically lower mass stars and, thus, have a smaller range of possible stellar com- panions. Thus, an average value for the correction factor 1.5 represents the sample as a whole, but a more accurate value for the correction factor can be derived for a given star, with a temperature between 3500 . Tef f . 7500K, using the fitted 3rd-order polynomial: hXRi = a3(Tef f )3 + a2(Tef f )2 + a1(Tef f ) + a0 (8) where a3 = −1.19118× 10−11, a2 = 1.61749× 10−7, a1 = −0.000560, and a0 = 1.64668. In the absence of any specific knowledge of the stel- lar properties (other than the effective temperature) and in the absence of any radial velocity or high resolution imaging to assess the specific companion properties of a given KOI, (see section 4.2), the above parameterization (equation 8) can be used to derive a mean radii correc- tion factor hXRi for a given star. For G-dwarfs and hot- ter stars, the correction factor is near hXRi ∼ 1.6. As the stellar temperature (mass) of the primary decreases to the range of M-dwarfs, the correction factor can be as low as hXRi ∼ 1.2. 6 Figure 4. The distribution of the ratio of the total planetary radii uncertainties (σRp ) to the quoted radii uncertainties (σ′ )from the cumulative KOI list (see equation 10). For the red histogram, it is assumed that the KOIs are single as is the case in the published KOI list; for the blue histogram, it is assumed that each KOI has been vetted with radial velocity (RV) and high resolution imaging (see Section 4.2). The vertical dashed lines represent the median values of the distributions: hσRp /σ′ i = 1.70 for the unvetted KOIs and hσRp /σ′ i = 1.15 for the vetted KOIs (see section 4.2). Rp Rp Rp 4.1. Planet Radius Uncertainty Term from hXRi The mean correction factor is useful for understand- ing how strongly the planetary radii may be underesti- mated, but an additional uncertainty term derived from the mean radius correction factor is potentially more useful as it can be added in quadrature to the formal planetary radii uncertainties. The formal uncertainties, presented in the KOI list, are derived from the uncer- tainties in the transit fitting and the uncertainty in the knowledge of the stellar radius, and they are calculated assuming the KOIs are single stars. We can estimate an additional planet radius uncertainty term based upon the mean radii correction factor as σXR = hXRiRp − Rp = hXRi − 1.0Rp (9) where Rp is the observed radius of the planet. Adding in quadrature to the reported uncertainty, a more complete uncertainty on the planetary radius can be reported as σRp = (cid:16)(σ′ Rp )2 + (σXR )2(cid:17)1/2 (10) where σ′ presented in the KOI list. Rp is the uncertainty of the planetary radius as Rp The distribution of the ratio of the more complete KOI radius uncertainties (σRp ) to the reported KOI radius uncertainties (σ′ ) is shown in Figure 4. Including the possibility that a KOI may be a multiple star increases the planetary radii uncertainties. While the distribution has a long tail dependent upon the specific system, the planetary radii uncertainties are underestimated as re- ported in the KOI list, on average, by a factor of 1.7. 4.2. Effectiveness of Companion Vetting The above analysis has assumed that the KOIs have undergone no companion vetting, as is the assumption in the current KOI list. In reality, the Kepler Project has funded a substantial ground-based follow-up obser- vation program which includes radial velocity vetting and high resolution imaging. In this section, we explore the effectiveness of the observational vetting. The observational vetting reduces the fraction of unde- tected companions. If there is no vetting or all stars are assumed to be single, as is the case for the published KOI list, then the fraction of undetected companions is 100% and the mean correction factors hXRi are as presented above. If every stellar companion is detected and ac- counted for in the planetary parameter derivations, then the fraction of undetected companions is 0%, and the mean correction factors are unity. Reality is somewhere in between these two extremes. To explore the effectiveness of the observational vetting on reducing the radii corrections factors (and the asso- ciated radii uncertainties), we have assumed that every KOI has been vetted equally, and all companions within the reach of the observations have been detected and ac- counted. Thus, the corrections factors depend only on the fraction of companions stars that remain out of the reach of vetting and undetected. In this simulation, we have assumed that all compan- ions with orbital periods of 2 years or less and all compan- ions with angular separations of 0.1′′ or greater have been detected. This, of course, will not quite be true as ran- dom orbital phase effects, inclination effects, companion mass distribution, stellar rotation effects, etc. will dimin- ish the efficiency of the observations to detect compan- ions. We recognize the simplicity of these assumptions; however, the purpose of this section is to assess the use- fulness of observational vetting on reducing the uncer- tainties of the planetary radii estimates, not to explore fully the sensitivities and completeness of the vetting. Typical follow-up observations include stellar spec- troscopy, a few radial velocity measurements, and high resolution imaging. The radial velocity observations usu- ally include 2 − 3 measurements over the span of 6 − 9 months and are typically sufficient to identify potential stellar companions with orbital periods of . 1−2 years or less. While determining full orbits and stellar masses for any stellar companions detected typically requires more intensive observing, we have estimated that 3 measure- ments spanning 6 − 9 months is sufficient to enable the detection of an RV trend for orbital periods of ∼ 2 years or less and mark the star as needing more detailed obser- vations. The amplitude of the RV signature, and hence the ability to detect companions, does depend upon the masses of the primary and companion stars; massive stars with low mass companions will display relatively low RV signatures. However, RV vetting for the Kepler program has been done at a level of . 100 − 200 m/s, which is sufficient to detect (at & 4 − 5σ) a late-type M-dwarf companion in a two-year orbit around a mid B-dwarf primary. Indeed, the RV vetting is made even more effective by searching for companions via spectral signatures (Kolbl et al. 2015). The high resolution imaging via adaptive optics, "lucky imaging", and/or speckle observations typically has resolutions of 0.02′′ − 0.1′′ (e.g., Howell et al. 2011; Horch et al. 2012; Lillo-Box et al. 2012; Adams et al. 2012, 2013; Dressing et al. 2014; Law et al. 2014; Lillo-Box et al. 2014; Wang et al. 2014a; Everett et al. 2014; Horch et al. 2014; Gilliland et al. 2014; Star et al. 7 Figure 5. Top Left: Example of the stellar companion period distribution that is vetted by radial velocity (RV) monitoring and high resolution imaging (HRI) for KOI-299. The Gaussian curve, normalized to unity, is the log-normal orbital period distribution of stellar companions (Raghavan et al. 2010), and the hatched regions mark where the potential observational vetting is assumed to have detected all companions in that period range. The solid (red) region of the Gaussian corresponds to the fraction of companions (31% for KOI-299) that would remain undetected by the RV and HRI observations (see section 4.2). Bottom Left: The distribution of the fraction of companions across all KOIs that remain undetected by observational radial velocity and high resolution imaging vetting (i.e., the red area in the top figure but for all KOIs). The vertical dashed line represents the median fraction (41%) of companions left undetected by the observational vetting of all KOIs. Top Right: The same as the top left figure but assuming that KOI-299 is 10 times closer as may be the case for K2/TESS targets. At such a close distance for a hypothetical KOI-299, RV and HRI vetting leaves only about 7% of the possible stellar companions undetected. Bottom Right: As the bottom left figure but assuming all of the KOI host stars are 10 times closer as may be the case for K2/TESS. The vertical dashed line represents the median fraction (16%) of companions left undetected if the KOIs were at these closer distances. 2014). Based upon Monte Carlo simulations in which we have averaged over random orbital inclinations and eccentricities, we have calculated the fraction of time within its orbit a companion will be detectable via high resolution imaging. With typical high resolution imag- ing of 0.05′′, we have estimated that & 50% of the stel- lar companions will be detected at one full-width half- maximum (FWHM=0.05′′) of the image resolution and beyond and > 90% at & 2 FWHM (0.1′′) of the image resolution and beyond. Kepler magnitude associated with the fitted isochrone. The median distance to the KOIs was found to be ∼ 900 pc, corresponding to ∼ 90 AU for 0.1′′ imaging. Using the isochrone stellar mass, the semi-major axis detection limits were converted to orbital period limits (assuming circular orbits). To determine what fraction of possible stellar compan- ions would be detected in such a scenario, we have used the nearly log-normal orbital period distribution from Raghavan et al. (2010). To convert the high resolution imaging limits into period-limits, we have estimated the distance to each KOI by determining a distance modulus from the observed Kepler magnitude and the absolute Combining the 2-year radial velocity limit and the 0.1′′ imaging limit, we were able to estimate the fraction of undetected companions for each individual KOI (see Fig- ure 5). The distribution of the fraction of undetected companions ranges from ∼ 20 − 60% and, on average, the ground-based observations leave ≈ 41% of the possi- ble companions undetected for the KOIs (see Figure 5). The mean correction factors hXRi are only applicable to the undetected companions. For the stars that are vetted with radial velocity and/or high resolution imag- ing, the intrinsic stellar companion rate for the KOIs of 8 46% (Raghavan et al. 2010) is reduced by the unvetted companion fraction for each KOI. That is, we assume that companion stars detected in the vetting have been accounted for in the planetary radii determinations, and the unvetted companion fraction is the relevant compan- ion rate for determining the correction factors. In the KOI-299 example (Fig. 5), the undetected companion rate used to calculate the mean radii correction factor is 0.46 × 0.31 = 0.1426. This lower fraction of undetected companions in turn reduces the mean correction factors for the vetted stars which are displayed in Figure 3 (blue points). Instead of a mean correction factor of hXRi ∼ 1.5, the average correction factor is hXRi = 1.20 ± 0.06 if the stars are vetted with radial velocity and high resolution imaging. The mean correction factor still changes as a function of the primary star effective temperature but the dependence is much more shallow with coefficients for equation 8 of a3 = −6.73847× 10−12, a2 = 9.38966 × 10−8, a1 = −0.000352, and a0 = 1.40391 (see Figure 3). 4.3. K2 and TESS The above analysis has concentrated on the Kepler Mission and the associated KOI list, but the same effects will apply to all transit surveys including K2 (Howell et al. 2014) and TESS (Ricker et al. 2014). If the planetary host stars from K2 and TESS are also assumed to be single with no observational vetting, the planetary radii will be underestimated by the same amount as the Kepler KOIs (Fig. 3 and eq. 8). Many K2 targets and nearly all of TESS targets will be stars that are typically 4−5 magnitudes brighter than the stars observed by Kepler, and therefore, K2 and TESS targets will be ∼ 10 times closer than the Kepler targets. The effectiveness of the radial velocity vetting will re- main mostly unaffected by the brighter and closer stars, but the effectiveness of the high resolution imaging will be significantly enhanced. Instead of probing the stars to within ∼ 100AU, the imaging will be able to detect companion stars within ∼ 10 AU of the stars. As a result, the fraction of undetected companions will decrease significantly. Even for the Kepler stars that undergo vetting via radial velocity and high resolution imaging, ∼ 40% of the companions remain undetected. But for the stars that are 10 times closer that fraction decreases to ∼ 16% (see Figure 5). This has the strong benefit of greatly reducing the mean correction factors for the stars that are observed by K2 and TESS and are vetted for companions with radial velocity and high resolution imaging. The mean correction factor for vetted K2/TESS-like stars is only hXRi = 1.07 ± 0.03. The correction fac- tor has a much flatter dependence on the primary star effective temperature, because the majority of the possi- ble stellar companions are detected by the vetting. The coefficients for equation 8 become a3 = −4.12309 × 10−12, a2 = 5.89709 × 10−8, a1 = −0.000242, and a0 = 1.30060. The mean radii correction factors for vetted K2/TESS planetary host stars correspond to a correc- tion to the planetary radii uncertainties of only ∼ 2%, in comparison to a correction of ∼ 70% if the K2/TESS stars remain unvetted. For K2 and TESS, where the number of candidate planetary systems may outnumber the KOIs by an or- der of magnitude (or more), single epoch high resolution imaging may prove to be the most important observa- tional vetting performed. While the imaging will not reach the innermost stellar companions, radial velocity observations require multiple visits over a baseline com- parable to the orbital periods an observer is trying to sample. In contrast, the high resolution imaging requires a single visit (or perhaps one per filter on a single night) and will sample the majority of the expected stellar com- panion period distribution. 5. EFFECT OF UNDETECTED COMPANIONS ON THE DERIVED OCCURRENCE RATES Understanding the occurrence rates of the Earth-sized planets is one of the primary goals of the Kepler mission and one of the uses of the KOI list (Borucki et al. 2010). It has been shown that the transition from "rocky" to "non-rocky" planets occurs near a radius of Rp = 1.6 R⊕ and the transition is very sharp (Rogers 2014). However, the amplitude of the uncertainties resulting from unde- tected companions may be large enough to push plan- ets across this boundary and affect our knowledge of the fraction of Earth-sized planets. We have explored the possible effects of undetected companions on the derived occurrence rates. The plan- etary radii can not simply be multiplied by a mean cor- rection factor XR, as that factor is only a measure of the statistical uncertainty of the planetary radius resulting from assuming the stars are single and only a fraction of the stars are truly multiples. Instead a Monte Carlo simulation has been performed to assign randomly the effect of unseen companions on the KOIs. The simulation was performed 10,000 times for each KOI. For each realization of the simulation, we have randomly assigned the star to be single, binary, or triple star via the 54%, the 34% and the 12% fractions (Raghavan et al. 2010). If the KOI is assigned to be a single star, the mean correction factors for the planets If the KOI star in that system are unity: hXRi = 1. is a multiple star system, we have randomly assigned the stellar companion masses according to the masses available from the fitted isochrones and using the mass ratio distribution of (Raghavan et al. 2010). Finally, the planets are randomly assigned to the primary or to the companion stars (i.e., 50% fractions for binary stars and 33.3% fractions for triple stars). Once the details for the system are set for a particular realization, the final cor- rection factor for the planets are determined from the polynomial fits for the individual multiplicity scenarios (e.g., Fig. 1 and 2). For each set of the simulations, we compiled the frac- tion of planets within the following planet-radii bins: Rp ≤ 1.6 R⊕; 1.6 < Rp ≤ 3.0 Rp; 3.0 < Rp ≤ 10 R⊕ cor- responding to Earth-sized, super-Earth/mini-Neptune- sized, and Neptune-to-Jupiter-sized planets. The raw fractions directly from the KOI-list, for these three cat- egories of planets, are 33.3%, 46.0%, and 20.7%. Note that these are the raw fractions and are not corrected for completeness or detectability as must be done for a true occurrence rate calculation; these fractions are necessary for comparing how unseen companions affect the deter- mination of fractions. Finally, we repeated the simula- tions, but using the undetected multiple star fractions af- ter vetting with radial velocity and high resolution imag- 9 Figure 6. The percent change of the measured occurrence rates from the raw KOI list (i.e., uncorrected for completeness) caused by undetected stellar companions and assuming all of the stars in the published KOI list are single. The fractional changes are computed for three different planet size bins as labeled in the plots. The results of assuming all the stars are single (i.e., unvetted) are in the red histogram and the results of vetting the all of the stars with radial velocity and high resolution imaging (section 4.2) are shown in the blue histogram. The histograms are based upon Monte Carlo simulations described in section 5 ing had been performed, thus, effectively increasing the fraction of stars with correction factors of unity. The distributions of the change in the fractions of plan- ets in each planet category, compared to the raw KOI fractions, are shown in Figure 6. If the occurrence rates utilize the assumed-single KOI list (i.e., unvetted), then the Earth-sized planet fraction may be overestimated by as much as 15 − 20% and the giant-planet fraction may be underestimated by as much as 30%. Interestingly, the fraction of super-Earth/mini-Neptune planets does not change substantially; this is a result of smaller planets moving into this bin, and larger planets moving out of the bin. In contrast, if all of the KOIs undergo vetting via radial velocity and high resolution imaging, the frac- tional changes to these bin fractions are much smaller: 5 − 7% for the Earth-sized planets and 10 − 12% for the Neptune/Jupiter-sizes planets. 6. SUMMARY We present an exploration of the effect of undetected companions on the measured radii of planets in the Ke- pler sample. We find that if stars are assumed to be single (as they are in the current Kepler Objects of In- terest list) and no companion vetting with radial velocity and/or high resolution imaging is performed, the plane- tary radii are underestimated, on average, by a factor of hXRi = 1.5, corresponding to an overestimation of the planet bulk density by a factor of ∼ 3. Because lower mass stars will have a smaller range of stellar companion masses than higher mass stars, the planet radius mean correction factor has been quantified as a function of stel- lar effective temperature. If the KOIs are vetted with radial velocity observations and high resolution imaging, the planetary radius mean correction necessary to account for undetected compan- ions is reduced significantly to a factor of hXRi = 1.2. The benefit of radial velocity and imaging vetting is even more powerful for missions like K2 and TESS, where the targets are, on average, ten times closer than the Kepler Objects of Interest. With vetting, the planetary radii for K2 and TESS targets will only be underestimated, on average, by 10%. Given the large number of candidates expected to be produced by K2 and TESS, single epoch high resolution imaging may be the most effective and efficient means of reducing the mean planetary radius correction factor. Finally, we explored the effects of undetected compan- ions on the occurrence rate calculations for Earth-sized, super-Earth/mini-Neptune-sized, and Neptune-sized and larger planets. We find that if the Kepler Objects of In- terest are all assumed to be single (as they currently are in the KOI list), then the fraction of Earth-sized plan- ets may be overestimated by as much as 15-20% and the fraction of large planets may be underestimated by as much as 30% The particular radial velocity observations or high res- olution imaging vetting that any one KOI may (or may not) have undergone differs from star to star. Companion vetting simulations presented here show that a full un- derstanding and characterization of the planetary com- panions is dependent upon also understanding the pres- ence of stellar companions, but is also dependent upon understanding the limits of those observations. For a final occurrence rate determination of Earth-sized plan- ets and, more importantly, an uncertainty on that oc- currence rate, the stellar companion detections (or lack thereof) must be taken into account. The authors would like to thank Ji Wang, Tim Mor- ton, and Gerard van Belle for useful discussions during the writing of this paper. This research has made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, under contract with the National Aeronautics and Space Administration un- der the Exoplanet Exploration Program. Portions of this work were performed at the California Institute of Tech- nology under contract with the National Aeronautics and 10 Space Administration. REFERENCES Howell, S. B., et al., 2014, PASP, 126, 398 Huber, D., Silva Aguirre, V., Matthews, J. M., et al. 2014, ApJS, 211, 2 Howard, A. W., Marcy, G. W., Bryson, S. T., et al. 2012, ApJS, 201, 15 Adams, E. R., Ciardi, D. R., Dupree, A. K., et al. 2012, AJ, 144, Kolbl, R., Marcy, G. W., Isaacson, H., & Howard, A. W. 2015, 42 Adams, E. R., Dupree, A. K., Kulesa, C., & McCarthy, D. 2013, AJ, 146, 9 Batalha, N. M. et al. 2011, ApJ, 729, 27 Batalha, N. M. 2014, Proceedings of the National Academy of Science, 111, 12647 Borucki, W. J., et al. 2010, Science, 327, 977 Brown, T. M., Latham, D. W., Everett, M. E., & Esquerdo, G. A. 2011, AJ, 142, 112 Burke, C. J., Bryson, S. T., Mullally, F., et al. 2014, ApJS, 210, 19 Ciardi, D. R., Fabrycky, D. C., Ford, E. B., et al. 2013, ApJ, 763, 41 AJ, 149, 18 Law, N. M., Morton, T., Baranec, C., et al. 2014, ApJ, 791, 35 Lillo-Box, J., Barrado, D., & Bouy, H. 2012, A&A, 546, AA10 Lillo-Box, J., Barrado, D., & Bouy, H. 2014, A&A, 566, AA103 Marcy, G. W., Isaacson, H., Howard, A. W., et al. 2014, ApJS, 210, 20 Morton, T. D., & Johnson, J. A. 2011, ApJ, 738, 170 Morton, T. D. 2012, ApJ, 761, 6 Mullally, F. et al. 2015, ApJ, submitted Petigura, E. A., Howard, A. W., & Marcy, G. W. 2013, Proceedings of the National Academy of Science, 110, 19273 Raghavan, D., McAlister, H. A., Henry, T. J., et al. 2010, ApJS, Dotter, A., Chaboyer, B., Jevremovi´c, D., et al. 2008, ApJS, 178, 190, 1 89 Dressing, C. D., & Charbonneau, D. 2013, ApJ, 767, 95 Dressing, C. D., Adams, E. R., Dupree, A. K., Kulesa, C., & McCarthy, D. 2014, AJ, 148, 78 Duquennoy, A., & Mayor, M. 1991, A&A, 248, 485 Everett, M. E., Barclay, T., Ciardi, D. R., et al. 2014, arXiv:1411.3621 Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81 Gilliland, R. L., Cartier, K. M. S., Adams, E. R., et al. 2014, arXiv:1407.1009 Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2014, Proc. SPIE, 9143, 914320 Rogers, L. A. 2014, arXiv:1407.4457 Star, K. M., Gilliland, R. L., Wright, J. T., & Ciardi, D. R. 2014, arXiv:1407.1057 Wang, J., Xie, J.-W., Barclay, T., & Fischer, D. A. 2014, ApJ, 783, 4 Wang, J., Fischer, D. A., Xie, J.-W., & Ciardi, D. R. 2014, ApJ, 791, 111 Wang, J., Fischer, D. A., Horch, E. P., & Huang, X. 2014, Horch, E. P., Howell, S. B., Everett, M. E., & Ciardi, D. R. 2012, arXiv:1412.1731 AJ, 144, 165 Horch, E. P., Howell, S. B., Everett, M. E., & Ciardi, D. R. 2014, ApJ, 795, 60 Howell, S. B., Everett, M. E., Sherry, W., Horch, E., & Ciardi, D. R. 2011, AJ, 142, 19
1507.05468	1	1507	2015-07-20T12:28:13	Physical properties of the extreme centaur and super-comet candidate 2013 AZ60	[ "astro-ph.EP" ]	We present estimates of the basic physical properties -- including size and albedo -- of the extreme Centaur 2013 AZ60. These properties have been derived from optical and thermal infrared measurements. Our optical measurements revealed a likely full period of ~9.4 h with a shallow amplitude of 4.5%. By combining optical brightness information and thermal emission data, we are able to derive a diameter of 62.3 +/- 5.3 km and a geometric albedo of 2.9% -- corresponding to an extremely dark surface. Additionally, our finding of ~> 50 Jm^{-2}K^{-1}s^{-1/2} for the thermal inertia is also noticeably for objects in such a distance. The results of dynamical simulations yield an unstable orbit, with a 50% probability that the target will be ejected from the Solar System within 700,000 years. The current orbit of this object as well as its instability could imply a pristine cometary surface. This possibility is in agreement with the observed low geometric albedo and red photometric colour indices for the object, which are a good match for the surface of a dormant comet -- as would be expected for a long-period cometary body approaching perihelion. Despite the fact it was approaching ever closer to the Sun, however, the object exhibited star-like profiles in each of our observations, lacking any sign of cometary activity. By the albedo, 2013 AZ60 is a candidate for the darkest body among the known TNOs.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. ms July 10, 2018 c(cid:13) ESO 2018 Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 A. P´al1,2, Cs. Kiss1, J. Horner3,4, R. Szak´ats1, E. Vilenius5,6, Th. G. Muller5, J. Acosta-Pulido7,8, J. Licandro7,8, A. Cabrera-Lavers7,8, K. S´arneczky1, Gy. M. Szab´o9,1, A. Thirouin10, B. Sipocz11, ´A. D´ozsa9, and R. Duffard12 5 1 0 2 l u J 0 2 . ] P E h p - o r t s a [ 1 v 8 6 4 5 0 . 7 0 5 1 : v i X r a 1 Konkoly Observatory, MTA Research Centre for Astronomy and Earth Sciences, Konkoly-Thege Mikl´os ´ut 15-17, 1121 Budapest, Hungary; e-mail: [email protected] 2 Department of Astronomy, Lor´and Eotvos University, P´azm´any P´eter s´et´any 1/A, 1117 Budapest, Hungary 3 Computational Engineering and Science Research Centre, University of Southern Queensland, Toowoomba, Queensland 4350, Australia 4 Australian Centre for Astrobiology, UNSW Australia, Sydney, New South Wales 2052, Australia 5 Max-Planck-Institut fur extraterrestrische Physik, Postfach 1312, Giessenbachstr., 85741 Garching, Germany 6 Max-Planck-Institut fur Sonnensystemforschung, Justus-von-Liebig-Weg 3, 37077 Gottingen, Germany 7 Instituto de Astrof´ısica de Canarias, 38205 La Laguna, Tenerife, Spain 8 Departamento de Astrosf´ısica, Universidad de La Laguna, 38206 La Laguna, Tenerife, Spain 9 Gothard Astrophysical Observatory, Lor´and Eotvos University, 9700 Szombathely, Hungary 10 Lowell Observatory, 1400 W Mars Hill Rd, 86001, Arizona, USA 11 Centre for Astrophysics Research, University of Hertfordshire, College Lane, Hatfield AL10 9AB, UK 12 Instituto de Astrof´ısica de Andaluc´ıa - CSIC, Apt 3004, 18008 Granada, Spain Received . . . ; accepted . . . ABSTRACT We present estimates of the basic physical properties -- including size and albedo -- of the extreme Centaur 2013 AZ60. These properties have been derived from optical and thermal infrared measurements. Our optical measurements revealed a likely full period of ≈ 9.4 h with a shallow amplitude of 4.5%. By combining optical brightness information and thermal emission data, we are able to derive a diameter of 62.3 ± 5.3 km and a geometric albedo of 2.9% -- corresponding to an extremely dark surface. Additionally, our finding of & 50 Jm−2K−1s−1/2 for the thermal inertia is also noticeably for objects in such a distance. The results of dynamical simulations yield an unstable orbit, with a 50% probability that the target will be ejected from the Solar System within 700,000 years. The current orbit of this object as well as its instability could imply a pristine cometary surface. This possibility is in agreement with the observed low geometric albedo and red photometric colour indices for the object, which are a good match for the surface of a dormant comet -- as would be expected for a long-period cometary body approaching perihelion. Despite the fact it was approaching ever closer to the Sun, however, the object exhibited star-like profiles in each of our observations, lacking any sign of cometary activity. By the albedo, 2013 AZ60 is a candidate for the darkest body among the known TNOs. Key words. Kuiper belt objects: 2013 AZ60 -- Radiation mechanisms: thermal -- Techniques: photometric 1. Introduction The object 2013 AZ60 is a recently discovered extreme Centaur, moving on an eccentric orbit with e ≈ 0.992 and a perihelion distance of q ≈ 7.9 AU. As a result, 2013 AZ60 is among the TNOs with the largest known aphelion distance at ≈ 1950 AU. 2013 AZ60 may be classified as a Centaur, based on its perihelion distance (Horner et al., 2003). However, due to its large semimajor axis, it could equally be considered to be a scattered disk object (Gladman et al., 2008). Its Tisserand parameter (Duncan, Levison & Dones, 2004) w.r.t. Jupiter is TJ = 3.47 which is typical for Centaurs (Horner et al., 2004a,b) and differs from that of Jupiter family comets (2 < TJ < 3) and especially for from that of Damocolids and Halley-type comets (TJ < 2, see Jewitt, 2005) that exhibit cometary dynamics. In order to recover the basic physical and surface char- acteristics of this object, we need measurements both in the visual and in the thermal infrared range. Optical data can yield information about the intrinsic colours, the absolute brightness, rotational period, shape and surface homogene- ity of the object, while thermal observations aid us to decide whether we see a "large but dim" or a "small but bright" surface. For this latter purpose, Herschel Space Observatory (Pilbratt et al., 2010) is an ideal instrument since the ex- pected peak of the thermal emission is close to the shortest wavelengths of its PACS detector (Poglitsch et al., 2010). In our current analysis, we follow the same method- ology as presented in our previous study of the Centaur 2012 DR30 (Kiss et al., 2013), another object moving on a similar orbit to 2013 AZ60. The structure of this pa- per is as follows. In Sec. 2, we describe our observa- tions, including the detection of thermal emission by Herschel/PACS, optical photometry by the IAC-80 tele- scope (Teide Observatory, Tenerife, Spain), optical re- flectance by the Gran Telescopio Canarias (GTC, Roque 1 A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 de los Muchachos Observatory, La Palma, Spain) and near- infrared photometry by the William Herschel Telescope (WHT, Roque de los Muchachos Observatory, La Palma, Spain). In Sec. 3, we derive the basic physical properties of the object by applying well understood thermophysical models. The dynamics of 2013 AZ60 are then discussed in Sec. 4. Finally, our results are summarized in Sec. 5. 2. Observations and data reduction 2.1. Thermal observations and flux estimations Thermal infrared images have been acquired with the Photoconductor Array Camera and Spectrometer (PACS Poglitsch et al., 2010) camera of the Herschel Space Observatory (Pilbratt et al., 2010) in two series, each of 1.3 hour duration. As we summarize in Table. 1, these two measurement cycles were separated by more than 4 hours, allowing the target object 2013 AZ60 to move, but still be in the same field of view. This type of data collection has almost exclusively been employed in the "TNOs are Cool!" Open Time Key Programme of Herschel (Muller et al., 2009, 2010; Vilenius et al., 2012). For both series of measurements, we used both the blue/red (70/160 µm) and green/red (100/160 µm) channel combi- nations. This scheme allowed us to use the second series of images as a background for the first series of images (and vice versa) in order to eliminate the systematic effects of the strong thermal background. This type of data acqui- sition and the respective reduction scheme were described in our former works related to both the "TNOs are Cool!" project (see e.g. Vilenius et al., 2012; P´al et al., 2012) and subsequent measurements (see e.g. Kiss et al., 2013). uncertainties of 2013 AZ60 were relatively large at the time of Herschel observations, due to its rather recent discovery. Thus, the apparent position of the object was slightly (≈ 29′′) off from the image center, which also implied that the double-differential photometric method (Kiss et al., 2014) yielded larger photometric uncertainties. In addition, shortly before the Herschel observations, on February 16, 2013 (at OD-13751), one half of the red (160 µm) channel pixel array became faulty. Hence, only the images from the first visit were sufficient to obtain fluxes at 160 µm and it was not possible to create double-differential maps in this channel. astrometric Unfortunately, the Raw Herschel/PACS data has been processed in the HIPE environment (Ott, 2010) with custom scripts de- scribed in Kiss et al. (2014). The double-differential maps were created and analyzed using the FITSH package (P´al, 2012). The resulting images are displayed in Fig. 1. Photometry on the individual as well as on the combined double-differential images were performed by using aper- ture photometry where the fluxes were corrected by the respective growth curve functions. Photometric uncertain- ties were estimated by involving artificial source implan- tation in a Monte-Carlo fashion (Santos-Sanz et al., 2012; Mommert et al., 2012; Kiss et al., 2014). This method works both for the double-differential images (blue and green channels) as well as on the individual maps (here, the red channel). 1 http://herschel.esac.esa.int/Docs/Herschel/ /html/ch03s02.html#sec3:DeadMat 2 Table 2. Thermal fluxes of 2013 AZ60 derived from our Herschel measurements. Band B G R Flux λ 70 µm 32.5 ± 2.2 mJy 100 µm 23.0 ± 2.8 mJy 160 µm 15.9 ± 4.5 mJy 70 µm 100 µm 160 µm l a u d i v i d n I l a i t n e r e ff i d e l b u o D Fig. 1. Image stamps showing the Herschel/PACS maps of 2013 AZ60 in the 70 µm (blue), 100 µm (green), and 160 µm (red) channels. Each stamp covers an area of 64′′ × 64′′, while the tick marks on the axes show the relative positions in pixels. The effective beam size (i.e. the circle with a di- ameter corresponding to the full width at half maximum) is also displayed in the lower-left corners of the stamps. The upper panels show the stamps directly combined from the individual frames where the lower stamps are obtained us- ing the double-differential method. Due to the failure of the half of the red channel and the astrometric uncertainties, the second visit is nearly unusable in red, hence double- differential maps cannot be created. Based on the individual images, we obtained thermal fluxes of 36.6 ± 2.9, 25.2 ± 3.7 and 15.9 ± 4.5 mJy in the blue, green and red wavelengths, respectively. By involv- ing the double-differential maps, we derived 32.5 ± 2.2 and 23.0 ± 2.8 mJy in the blue and green regimes. Due to the lower level of confusion noise (see Fig. 1), the accuracy of the latter series of fluxes is better. Therefore, for fur- ther modelling we adopt the double-differential fluxes for blue and green. Thermal fluxes should undergo colour cor- rection according to the temperatures of the bodies (see Poglitsch et al., 2010, for the respective coefficients). Since the subsolar temperature of 2013 AZ60 is around 110 K, the colour correction is negligible (less than a percent) in blue and green while it is +4% in red. Our reported fluxes con- sider the respective colour correction factors. In addition, in the error estimation of the aforementioned fluxes, we included the 5% systematic error for the absolute flux cali- bration as well (Balog et al., 2014). The summary of these thermal fluxes are reported in Table 2. A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 Table 1. Summary of Herschel observations of 2013 AZ60, obtained in the DDT program DDT ckiss 3. The columns are: i) visit; ii) observation identifier; iii) date and time; iv) duration; v) filters configuration; vi) scan angle direction with respect to the detector array. Filters (µm/µm) 70/160 70/160 100/160 100/160 70/160 70/160 100/160 100/160 Scan angle (deg) 70 110 70 110 110 70 110 70 Visit OBSID Date & Time Duration Visit-1 Visit-2 1342268974 1342268975 1342268976 1342268977 1342268990 1342268991 1342268992 1342268993 (UT) 2013-03-31 18:10:51 2013-03-31 18:30:46 2013-03-31 18:50:41 2013-03-31 19:10:36 2013-03-31 23:46:55 2013-04-01 00:06:50 2013-04-01 00:26:45 2013-04-01 00:46:40 (s) 1132 1132 1132 1132 1132 1132 1132 1132 1.75 1.5 1.25 1 0.75 e c n a t c e l f e r e v i t a e R l s e d u t i n g a m ) 3 . 0 - ( i d n a r , ) 2 . 0 - ( g n a o S l 19.0 19.5 20.0 20.5 5000 5500 6000 6500 7000 7500 8000 8500 9000 Wavelength (A) Fig. 3. Reflectance spectrum of 2013 AZ60, taken with the OSIRIS spectrometer on the GTC in January, 2014. This spectrum is normalized to unity at λ = 5500 A. 0 0.5 1 1.5 2 Rotational phase Fig. 2. Folded optical light curves of 2013 AZ60 using photometric data taken on 6 subsequent nights of 2013 November 4 -- 9. Note that the folding frequency is related to the preferred double-peaked solution, n = (5.11/2) d−1. See text for further details. 2.2. Optical photometry Since 2013 AZ60 has recently been discovered, one of our goals was to obtain precise photometric time series for this object in order to estimate both the absolute magnitudes (in various passbands) and the rotational period based on light curve variability. For these purposes, we used the IAC-80 telescope located at Teide Observatory, Tenerife. During our observing runs, we used the CAMELOT cam- era, equipped with a CCD-E2V detector of 2k × 2k, with a pixel scale of ≈ 0.3′′ providing a field of view of roughly 10′ × 10′. Time series were gathered for several hours on the nights of 2013 November 4 -- 9 using Sloan g′, r′ and i′ filter sets. Since 2013 AZ60 can currently be found at the edge of the spring Sloan field, several dozen stars with accurate ref- erence magnitudes were available on each image. The night conditions were photometric on 5th, 7th and 9th of 2013 November where the individual photometric uncertainties were nearly constant and varied between 0.04 and 0.05 mags in Sloan r′ band. The conditions were worse on the three another nights (4th, 6th and 8th) when the formal uncer- tainties scattered in the range of 0.04 − 0.08 indicating the variable transparency of the sky (which was also notable during the observations). The worst conditions were on the first night where some of the measurements had a formal uncertainty of 0.09. The scientific images were analyzed with the standard calibration, source extraction, astrometry, cross matching and photometry tasks of the FITSH package (P´al, 2012). As a hint, we used the MPC predictions for the target co- ordinates and then we performed individual centroid fit on each image and smoothed with polynomial regression for a better (much more precise and accurate) input for aperture photometry. Instrumental magnitudes were then extracted for both the reference stars and the target itself and then after applying the standard photometric transformations, we obtained the intrinsic g′, r′ and i′ magnitudes. We searched for possible light curve variations using the most frequently sampled Sloan r' band data (every second image was taken in Sloan r′ while every fourth image was in g′ and i′). In order to search for periodic variations in our data set, we fitted a function in a form of f (t) = a + b · sin(2πnt) + c · cos(2πnt) to the Sloan r′ photometric series where t indicates the independent value (the time in this case). If the value of n is scanned in the appropriate domain 3 A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 (n = 0.01 . . . 15) with a proper stepsize (n = 0.01, that is ∼ 8 times smaller than the stepsize implied by the Nyquist criterion), then the parameters a, b and c can be obtained via a simple weighted linear least squares fit procedure. The unbiased χ2 values can then be compared with the reference value of χ2 0. This reference value is obtained when n is set to 0 and the error bars are scaled by a factor of 1.51 to yield a χ2 0 equivalent to the degrees of freedom. The difference between the χ2 0 and the χ2 related to the adopted period tells the significance of the detection while various additional possible periods can also be checked and/or ruled out according to the difference between the respective χ2 values. We found a significant variation (χ2 0 − χ2 = 24.2) with a corresponding amplitude of ∆r′ = 0.045 ± 0.007 that has a frequency of n = 5.11 ± 0.12 d−1. The folded light curves are displayed in Fig. 2. The mean magnitudes of these observations were g′ 1 = 19.519 ± 0.009 and i′ 1 = 20.274 ± 0.013, r′ 1 = 19.316 ± 0.013. We have to note here that due to the daily aliases, the peaks around n ± 1 d−1 are also remarkable and there is a non-negligible chance that one of these frequencies belong to the intrinsic rotation of the object. The peak at n = 6.11 has a χ2 value which is only larger than that of the main peak by 3.5. In general, minor bodies in the Solar System feature double-peaked light curves. Hence, the rotational frequency of 2013 AZ60 is more likely nrot = n/2 d−1, equivalent to a period of Prot = 9.39 ± 0.22 h. In order to test the signifi- cance of a double-peaked light curve solution, we coadded a sinusoidal component with half of the frequency to the pri- mary variations. The amplitude of this component is found to be 0.013±0.008 mag. This is only a 1.7-σ detection, how- ever, a good argument for confirming the assumption for an intrinsic rotation period of Prot ≈ 9.4 h. 2 = 19.71 ± 0.04 and r′ In addition, we repeated the photometric observations for 2013 AZ60 in 2014 January 28 in g′ and r′ bands. The results of these photometric measurements yielded the Sloan magnitudes of g′ 2 = 18.99 ± 0.03. During the first series of measurements (in 2013 November), the geocentric and heliocentric distance of 2013 AZ60 were ∆1 = 8.176 AU and r1 = 8.244 AU, re- spectively, while in 2014 January 28, these distances were ∆2 = 7.148 AU and r2 = 8.114 AU. Based on these dis- tances, the expected change in the apparent brightness was 5[log10(r2∆2) − log10(r1∆1)] = −0.326, however, the actual brightness changes were ∆g′ = −0.56 ± 0.04 and ∆r′ = −0.53 ± 0.03, whose mean is ∆m = −0.54 ± 0.03. Since the phase angle of 2013 AZ60 was α1 = 6.9◦ in 2013 November 5 and α2 = 1.5◦ in 2014 January 28, these values imply a phase correction factor of β = [(0.54 ± 0.03) − 0.326]/(6.9 − 1.5) = 0.039 ± 0.006 mag/deg. This is is rather good agreement with MPC observations. Based on the MPC observation database, the best-fit phase correc- tion parameter can also be derived, however, with a larger uncertainty: βMPC = 0.040 ± 0.025 mag/deg. These parameters allowed us to derive the absolute brightness of the object 2013 AZ60 in a manner described in the following. First, we employed simple Monte-Carlo run whose input were the observed Sloan brightnesses, the derived phase correction factor as well as the parameters and the respective uncertainties of the corresponding Sloan- UBVRI transformation equation (for converting g′ and r′ brightnesses to V , see Jester et al., 2005). This Monte- Carlo run yielded a value of HV = 10.42 ± 0.07. Next, 4 we checked the available photometric data series presented in the MPC database which yielded slightly fainter val- ues, namely HV,MPC = 10.60 ± 0.15. In order to reflect MPC photometry in our derived absolute brightness value, we adopted the weighted mean value of these two values, namely HV = 10.45 with a conservative uncertainty of ±0.10 in the subsequent modelling. 2.3. Reflectance spectrum In order to accurately compare the surface colour charac- teristics of 2013 AZ60 with other TNOs (see Lacerda et al., 2014), we obtained a low resolution spectrum using the Optical System for Imaging and Low Resolution Integrated Spectroscopy (OSIRIS) camera spectrograph (Cepa et al., 2000; Cepa, 2010) at the 10.4m Gran Telescopio Canarias (GTC), located at the El Roque de los Muchachos Observatory (ORM) in La Palma, Canary Islands, Spain. The OSIRIS instrument consists of a mosaic of two Marconi CCD detectors, each with 2048 × 4096 pixels and a total unvignetted field of view of 7.8′ × 7.8′, giving a plate scale of 0.127′′/pixel. However, to increase the signal to noise for our observations we selected the 2 × 2 binning mode with a readout speed of 200 kHz (that has a gain of 0.95 e−/ADU and a readout noise of 4.5 e−), as corresponds with the standard operation mode of the instrument. A 300 s expo- sure time spectrum was obtained on January 28.17 (UTC), 2014 at an airmass of X = 1.14 using the OSIRIS R300R grism that produces a dispersion of 7.74 A/pixel, covering the 4800−9000 A spectral range. A 1.5′′ slit width was used oriented at the parallactic angle. Spectroscopic reduction has been done using the stan- dard IRAF tasks. Images were initially bias and flat- field corrected, using lamp flats from the GTC Instrument Calibration Module. The two-dimensional spectra were then wavelength calibrated using Xe+Ne+HgAr lamps. After the wavelength calibration, sky background was sub- tracted and a one dimensional spectrum was extracted. To correct for telluric absorption and to obtain the relative reflectance, G2V star Land102 1081 (Landolt, 1992) was observed using the same spectral configuration and at a similar airmass immediately after the Centaur observation. The spectrum of the 2013 AZ60 was then divided by that of Land102 1081, and then normalized to unity at 0.55 µm. The derived spectrum is displayed in Fig. 3. Based on this spectrum, the slope parameter of this object is found to be S ′ = 13.4 ± 3.0 %(1000A)−1 by a linear fit across the interval 5000 − 9000 A. The measured photometric colours (g′ − r′ = 0.755 ± 0.018 and 0.72 ± 0.05 on 2013.11.04 and 2014.01.28, respec- tively) are in complete accordance with the derived spec- tral slope. The spectrum was normalized at 5500 A, just between the g′ and r′ bands. We can therefore write S ′ to equation (2) of Jewitt (2002), if we write SDSS colours in- stead of Bessel ones, and set ∆λ = 1480 A. This results in a synthetic colour index from spectral slope (g − r)synth = 0.76 ± 0.04, in a perfect agreement with our photometry within the errors. 2.4. Near-infrared photometry CCD observations of 2013 AZ60 were obtained on 24 September 2013 with the 4.2-m William Herschel Telescope A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 ) y J m ( x u F l 50 40 30 20 10 0 0 50 100 150 200 250 300 Wavelength (microns) Fig. 4. Spectral energy distribution of 2013 AZ60 in the far-infrared region, based on Herschel/PACS measurements. Left panel: far-infrared measurements superimposed are the best-fit NEATM curves with their respective uncertainty. Middle panel: TPM model curve for thermal inertia of 100 Jm−2K−1s−1/2, rotation period of 9.39 h and equator-on geometry. Right panel: the value of χ2 as the function of thermal inertia for pole-on and equator-on geometries. See text for further details. Table 3. Orbital and optical data for 2013 AZ60 at the time of the Herschel observations. 3. Thermal emission modelling 3.1. Near-Earth Asteroid Thermal Model Quantity Heliocentric distance Distance from Herschel Phase angle Absolute visual magnitude HV Symbol r ∆ α Value 8.702 AU 8.560 AU 6.◦6 10.45 ± 0.10 at La Palma Observatory, equipped with the LIRIS in- strument. LIRIS is a near-IR imager/spectrograph, which uses a 1k × 1k HAWAII detector with a field of view of 4.27′ × 4.27′. The number of exposures taken in different filters are: 5 × 30 s in Y and J, 15 × 20 s in H, 15 × 13 s in CH4, and 180 × 20 s exposures in Ks. Local comparison stars were selected from the 2MASS catalogue and magni- tude transformation were applied following Hodgkin et al. (2009). The result of the photometry is Y = 18.66 ± 0.08, J = 18.34 ± 0.05, H = 18.00 ± 0.06 and Ks = 17.72 ± 0.10 where Y refers to the UKIDDS system, while JHKs are 2MASS magnitudes. Thus, 2013 AZ60 exhibits almost exactly so- lar colour indices, with a slightly redder slope than solar. Namely Y − J = 0.32, J − H = 0.34 ± 0.07 and H − Ks = 0.28 ± 0.11 while according to Casagrande et al. (2012) and estimating solar Y − J according to Hodgkin et al. (2009), the respective solar colours are (Y − J)⊙ = 0.235 ± 0.018 (J − H)⊙ = 0.286 ± 0.018 and (H − Ks)⊙ = 0.076 ± 0.018. We note here that LIRIS is equipped with Mauna Kea Observatories (MKO) system of J, H and K filters. According to Hodgkin et al. (2009)2, the expected system- atic differences between 2MASS and LIRIS/MKO colours of 2013 AZ60 is in the range of −0.010 . . . + 0.015, which is definitely smaller than the photometric uncertainties. This observation indicates a flat and featureless spectrum of 2013 AZ60: the slope is equivalent in the infrared and in the optical, being quite similar to dormant cometary nuclei. During the observation the heliocentric and geocentric distance of 2013 AZ60 were 8.87 and 8.32 AU, respectively, indicating an absolute mid-IR brightness of J = 9.00 ± 0.06 without correcting for the solar phase angle. 2 See their equations (6), (7) and (8) The basic physical properties such as albedo and diameter can be obtained by combining optical brightness data with thermal emission. Assuming an absolute optical brightness for a certain object, the higher the thermal emission, the smaller the actual albedo and hence the larger the diam- eter. Due to the observing strategies constrained by the spatial attitude of the Herschel, objects close to the eclip- tic are likely observable by Herschel during quadratures. Nevertheless, in quadratures these minor objects exhibit a large phase angle, hence Standard Thermal Model (STM Lebofsky et al., 1986) might not be as accurate as it is de- sired. 2013 AZ60 had a phase angle of α = 6.6◦ at the time of our Herschel/PACS observations. In order to have accurate estimates for larger phase angles, we employed the Near- Earth Asteroid Thermal Model (NEATM Harris, 1998): this model integrates the thermal emission for arbitrary viewing angles. Throughout the modelling we use the he- liocentric and geocentric distances at the instance of the Herschel/PACS measurements, namely rhelio = 8.702 AU and rgeo = 8.560 AU. The diameter and albedo can then be derived in a sim- ilar manner like in our earlier works (see e.g. Kiss et al., 2013; P´al et al., 2012). As an input for the fit procedures, we used the previously obtained thermal fluxes (see Table 2) and the absolute brightness HV = 10.45±0.10 (derived ear- lier, see above). The absolute physical parameters (diameter, albedo and beaming parameter) have been obtained in a Monte-Carlo fashion. In each step, a Gaussian value were drawn for the four input values (three thermal fluxes and the absolute brightness HV ) and the model parameters were adjusted via a nonlinear Levenberg-Marquardt fit. A sufficiently long series of such steps yields the best fit values as well as the respective uncertainties and correlations. This procedure were performed in two iterations. First, we let the value for the beaming parameter η to be floated. This run re- sulted relatively high correlations between the parameters and a highly long-tailed distribution for η: we found that the mode for η was 0.8 while the median is 2.6 and the uncertainties yielded by the lower and upper quartiles are 2.6+2.9 −1.1. This skewed distribution is due to the fact that beaming parameters cannot really be constrained if thermal 5 A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 fluxes are not known for shorter wavelengths (i.e. shorter than the peak of the spectral energy distribution). Hence, in the next run we used η as an input (instead of an adjusted variable) while its value was drawn uniformly between 0.8 and 2.6. This domain is also in accordance with the possible physical domain of the beaming parameter (see also Fig. 4 of Lellouch et al., 2013). The results of this second run were d = 62.3 ± 5.3 km, pV = 0.029 ± 0.006 while the beaming parameter can be written as η = 1.7 ± 0.9. The resulting albedo refers to a remarkably dark surface. The fluxes along with the best-fit NEATM model curve are shown in the left panel of Fig. 4. 3.2. Thermophysical Model In addition to the derivation of the NEATM parameters, we conducted an analysis of thermal emission based on the as- teroid thermophysical model (TPM, see Muller & Lagerros, 1998, 2002). The observational constraints employed by this model was the thermal fluxes (see Table. 2), the absolute magnitude of HV = 10.45±0.10 (see earlier), the rotational period of 9.39 h as well as the actual geometry at the time of Herschel observations (see the values for phase angle, heliocentric and geocentric distances above). Our procedures have shown that the best- fit model occurs at high thermal inertia values. Assuming an equator-on geometry, a value for re- duced χ2 . 1 corresponds to Γ & 10 Jm−2K−1s−1/2 (see also the right panel of Fig. 4), however, the gradually decreasing form of the function χ2(Γ) im- plies a lower limit of Γ & 50 Jm−2K−1s−1/2. The cor- responding values at Γ = 50 Jm−2K−1s−1/2 for ge- ometric albedo and diameter are pV = 0.028 and d = 64.9 km, respectively. These values are also com- patible within uncertainties with the ones derived from NEATM analysis (see above). The spectral energy distri- bution provided by these TPM values are shown in the middle panel of Fig. 4. This value for the thermal iner- tia is close to the values of 30 − 300 Jm−2K−1s−1/2 re- ported for comets (Julian, Samarasinha & Belton, 2000; Campins & Fern´andez, 2000; Davidsson et al., 2013, see e.g.) as well as the value of 10 − 50 Jm−2K−1s−1/2 for 67P/Churyumov-Gerasimenko (Gulkis et al., 2015). We note here that models either with thermal inertia values smaller than 50 Jm−2K−1s−1/2 or having an assumption for pole-on view underestimate the observed flux at 160 µm. Our findings for large preferred values of the beaming parameter η as well as for the thermal in- ertia Γ (even Γ & 10 Jm−2K−1s−1/2) can be compared with the statistical expectations of Lellouch et al. (2013). By considering the small heliocentric distance of this object, both of these values are expected to be smaller (see Figs. 6 and 11 in Lellouch et al., 2013). 4. The dynamics of 2013 AZ60 2013 AZ60 moves on a highly eccentric orbit, with perihe- lion between the orbits of Jupiter and Saturn. The best fit solution for the epoch of March 15, 2015 is shown in Table 4. In order to assess the dynamical history, and po- tential future behaviour of 2013 AZ60, we follow a well- established route (see e.g. Horner et al., 2004a,b, 2010, 6 Table 4. The best-fit orbital solution (semi-major axis a, perihelion distance q, eccentricity e, inclination i, longi- tude of ascending node Ω, argument of perihelion ω, mean anomaly M and perihelion date Tperi), and associated un- certainties, for 2013 AZ60, taken from the Minor Planet Center (http://www.minorplanetcenter.net) on March 15, 2015. Note that the uncertainties of the orbital elements involved throughout the planning of the observations sig- nificantly larger than these due to the shorter arcs available at that time. a (AU) q (AU) e i (deg) Ω (deg) ω (deg) M (deg) Tperi 829.7 7.908098 ± 0.000014 0.990468 ± 0.000010 16.535760 ± 0.000011 349.21122 ± 0.00002 158.14327 ± 0.00021 0.00876 2456988.0641 ± 0.0032 Epoch (JD) 2457200.5 2012; Kiss et al., 2013), and used the Hybrid integra- tor within the n-body dynamics package MERCURY (Chambers, 1999), to follow the evolution of a swarm of test particles centered on the best-fit orbit for the object, in or- der to get a statistical overview of the object's behaviour. A total of 91,125 test particles were created, distributed uniformly across the region of orbital element phase space within ±3σof the best-fit perihelion distance, q, eccentric- ity, e, and inclination, i. In this manner, we created a grid of 45 × 45 × 45 test particles distributed in even steps across the ±3σ error ranges about the nominal best fit orbit in each of the three orbital elements studied. Each of these test particles was then followed in our integrations, with its orbit evolving under the gravitational influence of the giant planets Jupiter, Saturn, Uranus and Neptune, for a period of four billion years. Test particles were considered to have been ejected from the Solar System (and were therefore re- moved from the integrations) if they reached a barycentric distance of 10,000 AU. Similarly, any test particles that col- lided with one of the giant planets, or with the Sun, were removed from the simulations. Each time a test particle was removed in either of these manners, the time at which the removal occurred was recorded, allowing us to track the number of surviving test particles as a function of time. The results of our simulations are shown below, in Figs 5 and 6. It is immediately apparent that the population of clones of 2013 AZ60 is highly dynamically unstable, with 63.9% of the particles (58191 of 91125) being removed from the simulations within the first million years of the integrations, as a result of either ejection or collision with one of the giant planets or the Sun. Half of the test particles are ejected within the first 682 kyr of the integrations, revealing that the orbit of 2013 AZ60 is more than two orders of magnitude more unstable than that of the similar object 2012 DR30 (Kiss et al., 2013). The orbit of 2013 AZ60 proves to be highly dynami- cally unstable on timescales of just a few hundred thousand years. Fully half of the test particles in our simulations were removed from the simulations within just 682 kyr, and al- most two-thirds were removed within the first million years. This extreme level of instability is not, however, that sur- A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 0 0 Fig. 5. The decay of our population of 91,125 clones of 2013 AZ60 as a function of the time elapsed in our integra- tions. The plots on the right show the same data as those on the left, but are plotted on a log/log scale. The upper panels show the decay of the population over the first bil- lion years of the four billion year integrations, whilst the lower panels show the decay over the first million years. prising -- 2013 AZ60 passes through the descending node of its orbit3 at essentially the same time it passes through per- ihelion, maximizing the likelihood that it will be perturbed by either Jupiter or Saturn. This extreme level of instabil- ity is typical of objects moving on Centaur like orbits (e.g. Horner et al., 2004a,b) and suggests that 2013 AZ60 may only recently have been captured to its current orbit. This argument is supported by the fact that, averaged over our entire population of 91,125 test particles, the mean lifetime of 2013 AZ60 is just 1.56 Myr. Given that 2013 AZ60 exhibits such extreme instability, and may well be relatively pristine object, it is interest- ing to consider whether it will have experienced significant solar heating, and cometary activity, over its past history. As a result of our large dynamical dataset on the evolu- tion of 2013 AZ60, it is possible to determine the fraction of the population of clones that may one day evolve onto Earth-crossing orbits, and the fraction of the population that approach the Sun to within a given heliocentric dis- tance at some point in their lifetime. Since dynamical evolu- tion under the influence of gravity alone is a time-reversible process, we can use these values to estimate the probability that 2013 AZ60 has moved on orbits that bring it within those heliocentric distances at some point in the past, be- fore being ejected to its current orbit. Due to the extreme instability exhibited by 2013 AZ60, we found that a rela- tively small number of the total population of clones were captured to Earth-crossing orbits through their lifetimes. can be seen in the 3 as elegant Java visualization the object's orbit at http://ssd.jpl.nasa.gov/sbdb.cgi? of sstr=2013%20AZ60;orb=1;cov=0;log=0;cad=0#orb Fig. 6. The mean (upper) and median (lower) lifetimes of 2013 AZ60, as a function of the initial perihelion distance, q, and eccentricity, e, of the orbit tested. The location of the best-fit orbital solution for 2013 AZ60, as detailed in Table 4, is shown by the hollow square at the center of the figure, with the ±1σ uncertainties on the perihelion distance and eccentricity denoted by the solid black lines that radiate from that box. Each coloured square in the fig- ures shows the mean (or median) lifetime of the 45 individ- ual runs carried out at that particular a-e location. Each of those 45 runs tested a different orbital inclination for 2013 AZ60, evenly distributed across the ±3σ uncertainty range on the best-fit orbital solution. As was the case with the high-eccentricity Centaur 2012 DR30 (Kiss et al., 2013), the stability of the orbit of 2013 AZ60 does not vary signif- icantly across the range of perihelion distance and eccen- tricities tested in this work -- a reflection of the relatively high precision with which the object's orbit is known. Indeed, just 3805 of the 91125 test particles we studied (just 4.2% of the population) became Earth-crossing at any point in our integrations, and the total fraction of the object's lifetime spent as an Earth-crossing object (averaged across all 91125 test particles) was 0.12%. Our results for a variety of other perihelion distances are displayed in Table 5, to- gether with estimates of the mean amount of time for which clones of 2013 AZ60 exhibited perihelion distances smaller than the specified value. 7 A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 Number of clones Percentage of total integration time Earth-crossing (q < 1.0616 AU) q < 2 AU q < 4 AU q < 6 AU 3805 6005 12272 27150 0.118 0.291 0.329 2.06 Table 5. The number of the 91,125 clones of 2013 AZ60 simulated in this work that evolved to orbits with perihe- lion distances smaller than 2, 4, and 6 AU, and the number that evolved onto Earth-crossing orbits (following Horner et al., 2003). For each of these values, we also give the frac- tion of the total integration time, across all 91,125 clones, for which clones have perihelion distances within these lim- its. We note that this is the fraction of the time for which the perihelion distance was less than the stated amount and not the fraction of time the clones spend within that heliocentric distance. Even when moving on an orbit with perihelion within that of the Earth, a given clone will spend the vast majority of its time beyond that distance, and only a tiny fraction within it. 60 50 40 30 20 10 0 r e t e m a r a p e p o S l 0.02 0.05 0.10 0.20 0.50 1.00 Albedo Fig. 7. Slope parameter vs. albedo relations for 111 TNOs, including 2013 AZ60 and 2012 DR30. Data (except for these two latter objects) have been taken from Lacerda et al. (2014). The blue and red dots indicate the two major groups identified by Lacerda et al. (2014), black points represent ambiguous objects (due to their large respective uncertain- ties) while green and yellow dots show the large bodies and Haumea-type surfaces, respectively. The isolated pur- ple square shows the place of 2013 AZ60 at the very left side of the diagram. The other purple square indicates 2012 DR30, just in between of the dark neutral (blue) and bright red (red) object groups. 5. Discussion Since the orbit of 2013 AZ60 is highly eccentric, and takes the object out to approximately 1950 AU, it is clear that it spends the vast majority of its orbit at large heliocen- tric distance. It is quite plausible that 2013 AZ60 is a rel- 8 atively recent entrant to the inner Solar System. Hence, it is interesting to consider how much time, cumulative over its entire history since it was first emplaced on a planet crossing orbit, has it spent at a heliocentric distance of less than 1, 10 or 100 AU. Again, we can take advantage of the large dynamical dataset available to us from our integra- tions to get a feel for the amount of time the object will have spent within these distances. Clearly, this is only an estimate (and implicitly assumes that, prior to its injection to a planet-crossing orbit, the object was well beyond the 100 AU boundary -- i.e. that it was injected from the inner or outer Oort cloud, rather than the trans-Neptunian region). Given that implicit assumption, we find that, on average, clones of 2013 AZ60 spend just 6.68 years within 1 AU of the Sun, 4620 years within 10 AU of the Sun, and 273,000 years within 100 AU of the Sun. The time spent within 10 and 100 AU is strongly biased by a few particularly long-lived clones, especially those that are captured onto Centaur-like orbits. We note that more than two-thirds of the clones (64904 objects) spent less than a thousand years within 100 AU of the Sun, and 37493 (41.1%) spent less than one hundred years within 100 AU. Taken into considerations, our dynamical results suggest that 2013 AZ60 has only re- cently been captured to its current planet-crossing orbit, and that it is quite likely that it is a relatively pristine object. Indeed, it seems highly probable that the surface of 2013 AZ60 has experienced only minimal outgassing and loss of volatiles since being captured to a planet crossing orbit, and so it represents a particularly interesting object to target with further observations as it pulls away from the Sun following its recent perihelion passage. Outer Solar System objects can also be characterized in a way recently put forward by Lacerda et al. (2014), us- ing their visual range colours and albedos. In this frame, Centaurs and trans-Neptunian objects form typically two clusters, a dark-neutral and a bright-red one (see Fig. 2 in Lacerda et al., 2014). In this scheme, 2013 AZ60 is located at the dark (very low albedo) edge of the dark-neutral clus- ter, see Fig. 7. 2013 AZ60 is even darker than the object 2002 GZ32, the object with the lowest albedo in the sample of Duffard et al. (2014). Objects with characteristics simi- lar to our target belong rather to "dead comets" or Jupiter family comets which are the end states of Centaurs and Oort cloud comets (Fig. 4 in Lacerda et al., 2014); in this sense 2013 AZ60 is more similar to objects in the inner Solar System than those in the trans-Neptunian population. We also checked the distribution of the slope parameters of var- ious Centaurs based on Fornasier et al. (2009). Although in that work, a correlation between the slope parameters and orbital eccentricity were suspected (the higher the ec- centricity, the redder the object), the large eccentricity of 2013 AZ60 do not fit in this model since it has definitely lower slope parameter than the mean of that sample of Centaurs. While the dynamical analysis indicate that 2013 AZ60 has recently been pulled from the Oort cloud, in the case of this object there is a much higher likelihood that it has spent a considerable time in the inner Solar System then e.g. in the case of 2012 DR30, which might just be in a tran- sitional phase between the two main albedo-colour clusters (Kiss et al., 2013). Acknowledgements. We thank the comments and the thoughtful review of the anonymous referee. The work of A. P., Cs. K. and R. Sz. has been supported by the grant LP2012-31 of the A. P´al et al.: Physical properties of the extreme centaur and super-comet candidate 2013 AZ60 Hungarian Academy of Sciences as well as the ESA PECS grant No. 4000109997/13/NL/KML of the Hungarian Space Office and the European Space Agency, and the K-104607 and K-109276 grants of the Hungarian Research Fund (OTKA). The work of Gy. M. Sz. has also been supported by the Bolyai Research Fellowship of the Hungarian Academy of Sciences. Additionally, Gy. M. Sz. and K. S. has been supported by ESA PECS No. 4000110889/14/NL/NDe and the City of Szombathely under agreements No. S-11-1027 and 61.360-22/2013. K. S. has also been supported by the "Lendulet" 2009 program of the Hungarian Academy of Sciences. J. L. acknowledge support from the project AYA2012-39115-C03-03 (MINECO, Spanish Ministry of Economy and Competitiveness). Part of this work was supported by the German DLR project number 50 OR 1108. Based on observa- tions made with the Gran Telescopio Canarias (GTC), instaled in the Spanish Observatorio del Roque de los Muchachos (ORM) of the Instituto de Astrof´ısica de Canarias (IAC), in the island of La Palma, the William Herschel telescopes (WHT) operated in the ORM by the Isaac Newton Group and on observations made with the IAC-80 tele- scope operated on the island of Tenerife by the IAC in the Spanish Observatorio del Teide. WHT/LIRIS observations were carried out under the proposal SW2013a15. References Balog, Z. et al. 2014, Exp. Astron., 37, 129 Campins, H. & Fern´andez, Y., 2000, EM&P, 89, 117 Casagrande, L.; Ramrez, I.; Melndez, J. & Asplund, M., 2012, ApJ, 761, 16 Cepa, J. et al. 2000, Proc. SPIE Vol. 4008, p. 623-631, Optical and IR Telescope Instrumentation and Detectors, Eds.: Masanori Iye; Alan F. Moorwood Cepa, J. 2010 Highlights of Spanish Astrophysics V.. Astrophysics and Space Science Proceedings, p. 15 Chambers, J., 1999, MNRAS, 304, 793 Davidsson, B. J. R. et al., 2013, Icarus, 224, 154 Duffard, R. et al. 2014, A&A, 564, A92 Duncan, M.; Levison, H. & Dones, L., 2004, "Dynamical evolution of ecliptic comets" in "Comets II", eds. M. C. Festou, H. U. Keller, & H. A. Weaver, University of Arizona Press, Tucson Fornasier, S. et al., 2009, A&A, 508, 457 Gladman, B., Marsden, B. G., & Vanlaerhoven, C. 2008, The Solar System Beyond Neptune, 43 Gulkis, S. et al. 2015, Science, 347, 0709. Harris, A. W. 1998 Icarus, 131, 291 Hodgkin, S. T. et al. 2009, MNRAS, 394, 675 Horner, J., Evans, N. W., Bailey, M. E. & Asher, D. J., 2003, MNRAS, 343, 1057 Horner, J., Evans, N. W. & Bailey, M. E., 2004a, MNRAS, 354, 798 Horner, J., Evans, N. W. & Bailey, M. E., 2004b, MNRAS, 355, 321 Horner, J. & Lykawka, P. S., 2010, MNRAS, 405, 49 Horner, J., Lykawka, P. S., Bannister, M. T. & Francis, P., 2012, MNRAS, 422, 2145 Jester, S. et al., 2005, AJ, 130, 873 Jewitt, D. C., 2002, AJ, 123, 1039 Jewitt, D. C., 2005, AJ, 129, 530 Julian, W. H.; Samarasinha, N. H. & Belton, M. J. S, 2000, Icarus, 144, 160 Kiss, Cs., et al., 2013, A&A, 555, A3 Kiss, Cs. et al., 2014, ExA, 37, 161 Lacerda, P. et al., 2014, ApJ, 793, L2 Landolt, A. U. 1992, AJ, 104, 340 Lebofsky, L. A., Sykes, M. V., Tedesco, E. F. et al. 1986, Icarus, 68, 239 Lellouch, E., et al. 2013, A&A, 557, A60 Mommert, M., Harris, A. W., Kiss, C. et al. 2012, A&A, 541, A93 Muller, T. G. & Lagerros, J. S. V. 1998, A&A, 338, 340 Muller, T. G. & Lagerros, J. S. V. 2002, A&A, 381, 324 Muller, T. G. et al. 2009, EM&P, 105, 209 Muller, T. G. et al. 2010, A&A, 518, 146 Ott, S. 2010, in Astronomical Data Analysis Software and Systems XIX. eds. Y. Mizumoto, K.-I. Morita & M. Ohishi, ASP Conf. Ser., 434, 139 P´al, A. 2012, MNRAS, 421, 1825 P´al, A. et al. 2012, A&A, 541, A6 Pilbratt, G. L. et al., 2010 A&A, 518, 1 Poglitsch, A. et al., 2010, A&A, 518, 2 Ram´ırez, I. et al., 2012, ApJ, 752, 5 Santos-Sanz, P., Lellouch, E., Fornasier, S. et al. 2012, A&A, 541, A92 Sparke, L. S. & Gallagher, J. S., 2007, Galaxies in the Universe, Cambridge University Press, Cambridge, UK Vilenius, E. et al. 2012, A&A, 541, A94 9
1304.5157	1	1304	2013-04-18T15:07:00	Mass-Radius Relationships for Very Low Mass Gaseous Planets	[ "astro-ph.EP" ]	Recently, the Kepler spacecraft has detected a sizable aggregate of objects, characterized by giant-planet-like radii and modest levels of stellar irradiation. With the exception of a handful of objects, the physical nature, and specifically the average densities, of these bodies remain unknown. Here, we propose that the detected giant planet radii may partially belong to planets somewhat less massive than Uranus and Neptune. Accordingly, in this work, we seek to identify a physically sound upper limit to planetary radii at low masses and moderate equilibrium temperatures. As a guiding example, we analyze the interior structure of the Neptune-mass planet Kepler-30d and show that it is acutely deficient in heavy elements, especially compared with its solar system counterparts. Subsequently, we perform numerical simulations of planetary thermal evolution and in agreement with previous studies, show that generally, 10 - 20 Earth mass, multi-billion year old planets, composed of high density cores and extended H/He envelopes can have radii that firmly reside in the giant planet range. We subject our results to stability criteria based on extreme ultraviolet radiation, as well as Roche-lobe overflow driven mass-loss and construct mass-radius relationships for the considered objects. We conclude by discussing observational avenues that may be used to confirm or repudiate the existence of putative low mass, gas-dominated planets.	astro-ph.EP	astro-ph	Draft version August 30, 2018 Preprint typeset using LATEX style emulateapj v. 5/2/11 3 1 0 2 r p A 8 1 . ] P E h p - o r t s a [ 1 v 7 5 1 5 . 4 0 3 1 : v i X r a MASS-RADIUS RELATIONSHIPS FOR VERY LOW MASS GASEOUS PLANETS Konstantin Batygin1 & David J. Stevenson2 1Institute for Theory and Computation, Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cambridge, MA 02138 and 2Division of Geological and Planetary Sciences, California Institute of Technology, 1200 E. California Blvd., Pasadena, CA 91125 Draft version August 30, 2018 ABSTRACT Recently, the Kepler spacecraft has detected a sizable aggregate of objects, characterized by giant- planet-like radii and modest levels of stellar irradiation. With the exception of a handful of objects, the physical nature, and specifically the average densities, of these bodies remain unknown. Here, we propose that the detected giant planet radii may partially belong to planets somewhat less massive than Uranus and Neptune. Accordingly, in this work, we seek to identify a physically sound upper limit to planetary radii at low masses and moderate equilibrium temperatures. As a guiding example, we analyze the interior structure of the Neptune-mass planet Kepler -30d and show that it is acutely deficient in heavy elements, especially compared with its solar system counterparts. Subsequently, we perform numerical simulations of planetary thermal evolution and in agreement with previous studies, show that generally, 10 − 20M⊕, multi-billion year old planets, composed of high density cores and extended H/He envelopes can have radii that firmly reside in the giant planet range. We subject our results to stability criteria based on extreme ultraviolet radiation, as well as Roche-lobe overflow driven mass-loss and construct mass-radius relationships for the considered objects. We conclude by discussing observational avenues that may be used to confirm or repudiate the existence of putative low mass, gas-dominated planets. Subject headings: planets and satellites: interiors, planets and satellites: physical evolution 1. INTRODUCTION The ever-growing transit data set collected by the Kepler spacecraft has proven to be instrumental to the advancement of our understanding of the properties of planetary systems. Thanks to the sheer size of the data set (∼ 2500 planetary candidates as of Quarter 6) and the associated statistical ability to determine the charac- teristics of typical planetary systems (Howard et al. 2010; Youdin 2011), as well as highlight some unexptected ex- amples (Doyle et al. 2011; Welsh et al. 2012), important insights into planet formation have already been gleaned from the analysis (Wolfgang & Laughlin 2011). It is interesting to note that the Kepler data set con- tains objects whose radii are similar to that of Jupiter (and in a few cases even exceed it substantially), in the moderate irradiation range (200K (cid:46) Teq (cid:46) 800K) (see Figure 1). Specifically, the latest application of the pipeline to the sample suggests that of 1333 total planetary candidates in this Teq range, 68 have radii in the (RSAT (cid:46) R (cid:46) 2RJUP) range and 25 have radii that exceed RJ by more than a factor of 2 (Batalha et al. 2012). Although the analysis of Demory & Seager (2011) suggests that a dominant portion of the exces- sively large objects in the Kepler inventory are false pos- itives, the physical nature of objects characterized by Jupiter-like radii is of considerable interest. Neverthe- less, even basic information such as the average density is difficult to acquire since the overwhelming majority of the stars in the Kepler field are rather faint, making radial-velocity follow up observationally expensive. Bar- ring (near-)resonant systems, where transit timing vari- ations can be significant (Holman & Murray 2005), this means that the masses of the planets within the Kepler [email protected] sample will remain observationally unconstrained and theoretical inquiries are desirable. It is well known that giant planets comprising hun- dreds of Earth masses can have large radii, that exhibit only weak dependent on mass, and are instead primarily controlled by their chemical composition and the inte- rior thermal state (Zapolsky & Salpeter 1969; Stevenson 1982a). Furthermore, it is firmly established that radii of gaseous planets can increase with decreasing mass, thanks to the associated softening of the equation of state (Stevenson 1982a). Although, as illustrated by the wide- ranging numerical calculations of Fortney et al. (2007), whether the radius increases or decreases with mass and the extent to which it does so, are rather sensitive to the amount of irradiation received by the planet as well as its chemical composition. In extreme proximity to the host star, the upturn in ra- dius is well pronounced. For example, an evolved 20M⊕ planet irradiated at Teq (cid:39) 1300K is roughly twice as large as its isolated counterpart (Baraffe et al. 2008). Depend- ing on the planetary age, at even higher temperatures (e.g. Teq = 2000K), the discrepancy may be as large as a factor of a few (Guillot 2005). On the other hand, plan- etary radii at Teq (cid:46) 100K, do not differ from those of isolated objects much (Fortney et al. 2007). The current observational frontier lies in between these extremes, and to date, with the exception of only a handful of studies (e.g. Rogers et al. (2011)), this parameter regime re- mains largely unexplored. In this study, we shall perform calculations that will place meaningful constraints on the mass-radius relationship of sub-Saturnian objects in the moderate irradiation regime. Specifically, the identifica- tion of a physically sound upper limit to the planetary radii at low masses and moderate equilibrium tempera- tures is the primary aim of this study. 2 Batygin & Stevenson Fig. 1. -- Planetary radii as a function of planetary equilibrium ir- radiation temperature in the Kepler sample. Note the considerable presence of giant-planet-like radii in this irradiation regime. The data was obtained from http://planetquest.jpl.nasa.gov/kepler. The possible range of chemical compositions of plan- ets is generally not well known. However, the relatively low densities exhibited by some members of the well- characterized subset of the Kepler catalog suggest that low overall metallicities cannot be ruled out. A par- ticularly important example is the planet Kepler -30d (Sanchis-Ojeda et al. 2012) which as we show below, has an envelope whose density does not exceed that of a cosmic H/He mixture substantially and cannot pos- sess a core as massive as that typically invoked in the core-accretion model of planet formation (Pollack et al. 1996). Thus, motivated by the inferred structure of Ke- pler -30d, we shall limit ourselves to a consideration of the most favorable planetary compositions for the fabri- cation of large radii. That is, for definiteness and sim- plicity, here we focus on planets with well-defined cores and H/He gaseous envelopes, though it is possible that in real objects the core material is partially mixed into the envelope1 (Leconte & Chabrier 2012). The paper is organized as follows. In section 2, we de- scribe the setup of our numerical experiments and per- form simulations of planetary thermal evolution to ex- plore the interior structure of Kepler -30d. In section 3, we extend our calculations to lower masses and con- struct generic mass radius relationships, constrained by the hydrodynamical stability of the considered planets. We conclude and discuss our results in section 4. 2. THE STRUCTURE OF KEPLER-30D Following initial detection (Batalha et al. 2012), the Kepler -30 system was studied in greater detail by Sanchis-Ojeda et al. (2012), who determined the mass, radius, equilibrium temperature and age of Kepler -30d to be M = 23.1 ± 2.7M⊕, R = 8.8 ± 0.5R⊕, Teq = 364K (assuming zero albedo) and 2.0 ± 0.8 Gyr respectively. Here, we shall adopt the observed best fit parameters at face value for the generation of interior models. Naturally, any model we consider is subject to hydro- static equilibrium. With the knowledge of the equation 1 Mixing of core and envelope may not change the radius much for Jupiter mass planets when they are compared at similar tem- peratures. However, the consequences of this mixing are in general not simple for the radius-mass relationship because it affects the cooling history of the planet as well as the density distribution for a given temperature. Fig. 2. -- The radius-core mass relationship for Kepler -30d, as- suming various envelope compositions. Blue curves correspond to envelopes with Z = 0.02 while red curves correspond to Z = 0.04. The cartoon in the left-bottom corner of the figure represents the considered two-layer interior models (here drawn to scale with a 5M⊕ core). of state and an assumed radiative structure of the at- mosphere, the construction of a static interior model is relatively straight forward. This is however not enough, since the thermal state of the planet changes in time due to radiative losses of the interior entropy (Guillot 1999). By extension, the planetary radius also contracts. Thus, in order to obtain definitive results that are characteris- tic of multi-Gyr old planets, evolutionary calculations of planetary structure are required. For our numerical experiments, we utilized the MESA stellar and planetary evolution software package (Paxton et al. 2011, 2013). Following Bodenheimer et al. (2001), all of our models comprised constant density (ρcore = 5 g/cc) solid cores embedded in gaseous H/He envelopes. The baseline heat-flux arising from radioactive decay within the cores was taken to be 10−7 ergs/s/g, similar to that of the Earth. The envelope metallicity was varied between Z = 0.02 and Z = 0.04, while solar Y = 0.27 and slightly super-solar Y = 0.35 values of the He mass fraction were explored. We note that if hydrodynamic mass-loss played a significant role in shaping the plane- tary structure (Owen & Wu 2013), a super-solar value of Y can in principle originate from a preferential blow-off of Hydrogen. The analytical radiative equilibrium model of the outer atmosphere was adopted from the work of (Guillot & Havel 2011). In the radiative portion of the atmosphere, following Guillot (2010) we take the constant visible and infrared opacities to be κV = 10−2 cm2 g−1 and κIR = 4 × 10−3 cm2 g−1 respectively. These choices yield the closest agreement between the analytical radia- tive model used here and the state of the art numeri- cal models of Fortney et al. (2008). At optical depths much greater than unity, the tabulated Rosseland mean opacities of Freedman et al. (2008) were used and the radiative convective boundary was computed as dictated by the Schwartzchild criterion. The reported radius of a given planet was taken to be the value corresponding to a chord optical depth of unity in visible light. The search for admissible models of Kepler -30d was performed in the following way. For a given choice of Y and Z, the core-mass was varied between Mcore = 1M⊕ 30040050060070025101RNEPRSATPlanet Temperature (K)Planet Radius (REARTH)RJUP2030040050060070025101RNEPRSATPlanet Temperature (K)Planet Radius (REARTH)RJUP20051015204610Mcore (MEARTH)Planet Radius (REARTH)1282Best-Fit RadiusKepler-30dM = 23.1 Mearth Teq = 364 KAge = 2 Gyr H/He evnvelope5 g/cc high-Z coreρ=Y=0.27 Z=0.02Y=0.27 Z=0.04Y=0.35 Z=0.02Y=0.35 Z=0.04Mcore/M 1/3 ruled out>Big Little Planets 3 and Mcore = 23M⊕. The resulting initial conditions were integrated forward in time, yielding a sequence of model radii that decrease monotonically with Mcore. Impor- tantly, radii also decrease monotonically with enhanced mean molecular weight, which means that there exists a maximum value of Mcore above which the planetary radius cannot be matched. There also exists a maximal extent to which the mean molecular weight of the enve- lope can exceed that of a cosmic H/He mixture. However, such core-less solutions are strongly disfavored because the mass of Kepler-30d is too low for formation by grav- itational instability to be plausible. The R − Mcore sequences for various compositions are shown in Figure (2). Adopting a solar composi- tion envelope yields an upper bound on the core-mass of Mcore (cid:39) 7M⊕. Meanwhile, the corresponding value for a Y = 0.35, Z = 0.04 envelope is a mere Mcore (cid:39) 3.5M⊕. Unfortunately, because the planet's gravitational har- monics are not known, no useful lower bound on Mcore can be formulated. It is noteworthy however, that the upper bound on Mcore is surprisingly low. The dominantly gaseous interior structure we obtained for Kepler -30d is in sharp contrast with the inferred heavy element-dominated interior structures of Neptune and Uranus (Fortney et al. 2011). This suggests that the diversity in composition and overall interior struc- ture of low-mass planets is generally much more exten- sive than what is captured within modern state-of-the-art core accretion models. More specifically, this implies that the nucleated instability mechanism can operate even for comparatively small cores. 3. GENERIC MASS-RADIUS RELATIONSHIPS Motivated by the results attained above, in this sec- tion we construct generic mass-radius relationships for evolved planets with a specified composition, extending down to minimum feasible masses. The radius-mass se- quences were generated in a similar manner to the numer- ical experiments reported in the previous section. How- ever, with the aim to constrain the planetary radii from above, the compositions of the envelopes were kept solar (X = 0.71, Y = 0.27, Z = 0.02) across the models. The core mass was varied between 1, 3 and 5 M⊕, while the total planetary mass range of up to 0.1 MJUP (∼ 33M⊕) was explored. Conventional generation of initial conditions within the framework of thermal evolution calculations is known to encounter numerical instabilities at sufficiently low masses. Consequently, here the initial conditions were constructed by imposing a slow mass-loss on a M = 0.1MJUP model. After the desired mass was attained, we imposed energy dissipation to the core and re-heated the gaseous envelope to the point where the thermal and gravitational energies of the body are comparable. The duration of the evolutionary sequences was formally taken to be 5 Gyr. However, it should be noted that the changes in planetary structure were relatively small after the first ∼Gyr of integration. Likewise, we found the evolved radii to be largely independent of the de- tailed state of the initial condition, in agreement with published literature (Bodenheimer et al. 2001; Hubickyj et al. 2005). Not all generated planetary models are guaranteed to be long-term stable. Indeed, some of the models we con- structed were characterized by radii, exceeding that of Jupiter by as much as a factor of a few, rendering their stability against evaporation questionable. Accordingly, we formulated a criterion for model rejection in terms of the mass-loss rate due to atmospheric escape. Irradiated extrasolar planets can be susceptible to mass-loss due hydrodynamic winds originating in the upper atmosphere (Murray-Clay et al. 2009; Valencia et al. 2010; Lopez et al. 2012). Such winds are gener- ated through the photoionization of H (and the associ- ated heating) by extreme ultraviolet radiation. Provided that downward conductive heatflux or radiative cooling by H+ 3 is not overwhelming (Murray-Clay et al. 2009), the characteristic timescale for energy-limited evapora- tion is given to an order of magnitude by (Watson et al. 1981; Yelle et al. 2008) τe−lim ∼ GM 2Ktide πFEUVR3 EUV , (1) where G is the gravitational constant, FEUV = 4.1(a/1AU)−2 erg s−1 cm−2 is the typical extreme ultra- violet flux of a 5 Gyr old Sun-like star (Ribas et al. 2005; Sanz-Forcada et al. 2010), Ktide = 1−3(REUV/RHill)/2+ (REUV/RHill)3/2 is a geometrical factor that accounts for the fact any given parcel of gas only needs to reach the Hill radius to escape (Erkaev et al. 2007), and REUV is a radius at which the atmosphere becomes optically thick to extreme ultraviolet radiation i.e. nHσXUV ∼ 1 where n is the atmospheric number density, H is the scale-height, and σXUV (cid:39) 10−18 cm2 is the photoion- ization cross-section for Hydrogen (Murray-Clay et al. 2009). Meanwhile, (cid:39) 0.25 is a factor that parameter- izes the efficiency of atmospheric escape. The EUV flux is considered to be constant here since we are not seeking to model loss and stability during early epochs of evolution. However, any model that we deem stable at t = 5 Gyr will also likely be sta- ble at any time greatly exceeding the T-Tauri phase of the evolutionary sequence (e.g. t (cid:38) 100 Myr), because we generally find characteristic loss timescales of order τe−lim ∼ 100 Gyr or greater. In all our models, REUV never exceeded the exobase (a radius at which the molecular mean free path be- comes comparable to H), meaning that the atmospheres were never truncated by Jean's escape. However, for cer- tain models, REUV exceeded RHill, implying mass-loss by Roche-lobe overflow. In such cases the characteristic evaporation timescale is given by (Lubow & Shu 1975; Lai et al. 2010) τRoche ∼ GM 2 πρRHillc2a3 , (2) is the atmospheric density at the Hill ra- where ρRHill dius and c is the speed of sound. Generally, mass-loss by Roche-lobe overflow is orders of magnitude faster than that by extreme ultraviolet radiation-driven winds. Al- though any criterion based on the above estimates is only accurate to within an order of magnitude or so, we find this to be sufficient for our purposes, as we typically find a rapid transition from τ (cid:29) Gyr to τ (cid:28) Gyr across two models that neighbor each-other in mass. The mass-radius relationships for planets with core masses of 1, 3 and 5M⊕ are presented in panels A, B, 4 Batygin & Stevenson and C of Figure (3) respectively. Black dots represent the radii obtained through numerical experiments while the curves depict interpolation functions that run through the data. Thick curves imply models that are secure against evaporation while thin lines depict unstable mod- els. In addition to the irradiated models (shown with blue lines), isolated (i.e. no irradiation) models are also presented and are shown with black lines. For reference, Jupiter's, Saturn's and Neptune's radii are also marked. The results highlight the fact that accounting for stel- lar irradiation, giant planetary radii can persist to sur- prisingly low masses (that is, M (cid:46) 10M⊕). Figure (3) further affirms that the behavior of planetary structure is largely dictated by the associated core mass. Note that all models with a 5M⊕ core are stable against evap- oration and roughly follow the cold (i.e. isolated) mass- radius relationship. On the contrary, 1M⊕ core mod- els are largely unstable below M (cid:46) 15M⊕ but can have radii comparable to that of Jupiter prior to the onset of evaporation. A similar scenario is observed for the 3M⊕, Teq = 500K set of models. Indeed, these models are essentially always characterized by R (cid:39) RJUP above M (cid:38) 8M⊕. It is interesting to note that some of our models (e.g. those corresponding to 1 and 3M⊕ and Teq = 700K) have radii that are bigger than that of Jupiter. As already dis- cussed above, this upturn in radii is a direct consequence of the softening of the equation of state at lower pres- sures (an ideal gas has a softer equation of state than the deep interior of Jupiter). While reminiscent of the inflated Hot Jupiter radii (Guillot 2005; Fortney & Net- telmann 2010), these objects are fundamentally different, since they require no additional heat sources or mecha- nisms for stalling gravitational contraction. That said, it is unclear if such objects are particularly significant within the context of the observational sample, since the models that show such an excess are close to the evapo- ration boundaries of the mass-radius diagrams. In fact, accounting for coupled evolution of gravitational contrac- tion and mass-loss in a more self-consistent matter will likely yield an exclusion region that is a bit larger than what is shown in Figure 3. 4. DISCUSSION In this letter, we have examined the structure of mod- erately irradiated low-mass low-density extrasolar plan- ets. We began by analyzing the interior of a compara- tively well characterized planet Kepler -30d, and showed that the planet is likely composed of an extensive gaseous H/He envelope, surrounding a core that makes up less than a third of its total mass. Although qualitatively this object resembles a scaled down version of Saturn, it is important to recall that the mass of Kepler -30d is typical of much more metal-rich objects such as Uranus or Neptune. The existence of Kepler -30d immediately suggests that range of planetary interior configurations that occur in nature is much wider than that available for detailed study within the realm of the solar system. Prompted by this notion, we extended our calculations to quantify planetary mass-radius relationships for cored low-mass gaseous planetary objects of solar composition. Our calculations underline the importance of stellar ir- radiation on the evolutionary tracks of low-mass objects. In particular, the constructed mass-radius relationships Fig. 3. -- Mass-Radius relationships of low-mass, gas-dominated planetary models. Panels A, B, and C correspond to planets with core-masses of Mcore = 1, 3 and 5M⊕ respectively. On each panel, mass-radius relationships corresponding to equilibrium irradiation temperatures of Teq = 300, 500 and 700K are shown as blue lines. Additionally, isolated mass-radius relationships are shown as black lines. Solid lines run through models that are stable against evap- oration while the converse is true for thin lines. Note that radii characteristic of giant planets are readily attainable for mildly ir- radiated M ∼ 10M⊕, Mcore = 1, 3M⊕ planets. suggest that the radius of an irradiated body may exceed that of its isolated counterpart by as much as a factor of ∼ 2 (e.g. the case of M (cid:39) 10M⊕, Mcore = 3M⊕, Teq = 500), bringing the radius well into the character- istic giant planet range. Collectively, our results suggest that extreme care must be taken in the interpretation of giant transit radii from the Kepler sample, since the mass range corresponding to such radii can be quite ex- tensive (i.e. spanning almost two orders of magnitude). One may wish to argue against a significant popula- tion of bodies like those considered in this work based on 35917305813203047101314710131471013158132030RNEPRSATRJUPRNEPRSATRJUPRNEPRSATRJUPno irradiationno irradiationTeq = 700KTeq = 700KTeq = 700KTeq = 300KTeq = 500KTeq = 300KTeq = 500KTeq = 300KTeq = 500K1 MEARTH core3 MEARTH coreno irradiation5 MEARTH coreABCPlanet Mass (MEARTH)Planet Radius (REARTH)Planet Radius (REARTH)Planet Radius (REARTH)Big Little Planets 5 the (im)probability of their formation, since the gaseous component of our models is much enhanced over the stan- dard models of typical objects in the considered mass range. Indeed, it is often said that one must have a "critical" core mass of Mcore (cid:38) 10M⊕ in order to trigger gas accretion. However, this claim is ill-founded and is not actually relevant since a hydrostatically supported atmosphere around a core can be more massive than that envisioned within the context of the standard mod- els (Pollack et al. 1996) if either accretion is slower, the molecular weight of the envelope is larger, or the opacity is increased. This is evident for example in the simple analytical models of Stevenson (1982b) (see also Ikoma & Genda (2006); Broeg (2009)). Protoplanets may also have circumplanetary disks that qualitatively change the characteristic accretion pattern and affect the planetary energy loss. Furthermore, alternative formation scenar- ios could likely be envisioned, a speculative example be- ing one where objects of this type are sculpted out of more massive planets by intense ultraviolet-driven mass- loss during the first ∼ 100Myr of the stellar lifetime. In- deed, such scenarios have already been proposed in the exo-planetary context (Baraffe et al. 2006; Lopez et al. 2012). formation scenario for Ultimately, our aim here is not to argue for or against any particular sub-Neptune mass gas-dominated planets. Rather, similarly to what has been done for Kepler -30d, we propose that their existence can be validated or ruled out observationally. Beyond standard methods like transit timing variations, the most obvious approach to this is through radial- velocity monitoring of transiting planets. That is, if a giant-plenet-like radius is firmly established for a given object through transit observations but a commensurate radial velocity signal is not observed in the host star, such an object is likely characterized by a very low mass. Another approach to mass discrimination is exclusively photometric, and takes advantage of dependence of the transit radius on spectral frequency. Although the atmospheric scale-heights of Hot Jupiters comprise ∼ 1% of their radii at most, for ∼ 10M⊕ (albeit a factor of ∼ 3 cooler) planets with similar radii, the scale height is increased by about an order of magnitude. As a result, the chord optical depth of unity may correspond to substantially different radii in visible and infrared light. Both of these observational avenues should become readily available as the radial velocity precision continues to improve and future space-based missions such as JWST commence. Acknowledgments We thank Tristan Guillot, Geoff Blake, Ruth Murray-Clay, Adam Burrows and David Kipping for numerous useful conversations. We are grateful to the referee for a careful and insightful report that has greatly increased the quality of the manuscript. K.B. acknowledges the generous support from the ITC Prize Postdoctoral Fellowship at the Institute for The- ory and Computation, Harvard-Smithsonian Center for Astrophysics. REFERENCES Baraffe, I., Alibert, Y., Chabrier, G., & Benz, W. 2006, A&A, 450, 1221 Baraffe, I., Chabrier, G., & Barman, T. 2008, A&A, 482, 315 Batalha, N. M., Rowe, J. F., Bryson, S. T., et al. 2012, arXiv:1202.5852 Bodenheimer, P., Lin, D. N. C., & Mardling, R. A. 2001, ApJ, 548, 466 Broeg, C. H. 2009, Icarus, 204, 15 Demory, B.-O., & Seager, S. 2011, ApJS, 197, 12 Doyle, L. R., Carter, J. A., Fabrycky, D. C., et al. 2011, Science, 333, 1602 Erkaev, N. V., Kulikov, Y. N., Lammer, H., et al. 2007, A&A, 472, 329 Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 668, 1267 Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, 1419 Fortney, J. J., & Nettelmann, N. 2010, Space Sci. Rev., 152, 423 Fortney, J. J., Ikoma, M., Nettelmann, N., Guillot, T., & Marley, M. S. 2011, ApJ, 729, 32 Freedman, R. S., Marley, M. S., & Lodders, K. 2008, ApJS, 174, 504 Guillot, T. 1999, Science, 296, 72 Guillot, T. 2005, Annual Review of Earth and Planetary Sciences, 33, 493 Guillot, T. 2010, A&A, 520, A27 Guillot, T., & Havel, M. 2011, A&A, 527, A20 Holman, M. J., & Murray, N. W. 2005, Science, 307, 1288 Howard, A. W., Marcy, G. W., Johnson, J. A., et al. 2010, Science, 330, 653 Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2005, Icarus, 179, 415 Ikoma, M., & Genda, H. 2006, ApJ, 648, 696 Lai, D., Helling, C., & van den Heuvel, E. P. J. 2010, ApJ, 721, 923 Laughlin, G., Crismani, M., & Adams, F. C. 2011, ApJ, 729, L7 Leconte, J., & Chabrier, G. 2012, A&A, 540, A20 Lopez, E. D., Fortney, J. J., & Miller, N. 2012, ApJ, 761, 59 Lubow, S. H., & Shu, F. H. 1975, ApJ, 198, 383 Murray-Clay, R. A., Chiang, E. I., & Murray, N. 2009, ApJ, 693, 23 Owen, J. E., & Wu, Y. 2013, arXiv:1303.3899 Paxton, B., Bildsten, L., Dotter, A., et al. 2011, ApJS, 192, 3 Paxton, B., Cantiello, M., Arras, P., et al. 2013, arXiv:1301.0319 Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62 Ribas, I., Guinan, E. F., Gudel, M., & Audard, M. 2005, ApJ, 622, 680 Rogers, L. A., Bodenheimer, P., Lissauer, J. J., & Seager, S. 2011, ApJ, 738, 59 Sanchis-Ojeda, R., Fabrycky, D. C., Winn, J. N., et al. 2012, Nature, 487, 449 Sanz-Forcada, J., Ribas, I., Micela, G., et al. 2010, A&A, 511, L8 Saumon, D., Chabrier, G., & van Horn, H. M. 1995, ApJS, 99, 713 Stevenson, D. J. 1982a, Annual Review of Earth and Planetary Sciences, 10, 257 Stevenson, D. J. 1982b, Planet. Space Sci., 30, 755 Valencia, D., Ikoma, M., Guillot, T., & Nettelmann, N. 2010, A&A, 516, A20 Watson, A. J., Donahue, T. M., & Walker, J. C. G. 1981, Icarus, 48, 150 Welsh, W. F., Orosz, J. A., Carter, J. A., et al. 2012, Nature, 481, 475 Wolfgang, A., & Laughlin, G. 2011, arXiv:1108.5842 Yelle, R., Lammer, H., & Ip, W.-H. 2008, Space Sci. Rev., 139, 437 Youdin, A. N. 2011, ApJ, 742, 38 Zapolsky, H. S., & Salpeter, E. E. 1969, ApJ, 158, 809
1409.5525	2	1409	2015-06-22T19:05:14	A Variable Polytrope Index Applied to Planet and Material Models	[ "astro-ph.EP", "astro-ph.SR", "cond-mat.mtrl-sci" ]	We introduce a new approach to a century old assumption which enhances not only planetary interior calculations but also high pressure material physics. We show that the polytropic index is the derivative of the bulk modulus with respect to pressure. We then augment the traditional polytrope theory by including a variable polytrope index within the confines of the Lane-Emden differential equation. To investigate the possibilities of this method we create a high quality universal equation of state, transforming the traditional polytrope method to a tool with the potential for excellent predictive power. The theoretical foundation of our equation of state is the same elastic observable which we found equivalent to the polytrope index, the derivative of the bulk modulus with respect to pressure. We calculate the density-pressure of six common materials up to 10$^{18}$ Pa, mass-radius relationships for the same materials, and produce plausible density-radius models for the rocky planets of our solar system. We argue that the bulk modulus and its derivatives have been under utilized in previous planet formation methods. We constrain the material surface observables for the inner core, outer core, and mantle of planet Earth in a systematic way including pressure, bulk modulus, and the polytrope index in the analysis. We believe this variable polytrope method has the necessary apparatus to be extended further to gas giants and stars. As supplemental material we provide computer code to calculate multi-layered planets.	astro-ph.EP	astro-ph	MNRAS 000, 000–000 (0000) Preprint 23 June 2015 Compiled using MNRAS LATEX style file v3.0 A Variable Polytrope Index Applied to Planet and Material Models S. P. Weppner,1⋆ J. P. McKelvey,2† K. D. Thielen 1 A. K. Zielinski 1 1Eckerd College; St. Petersburg, FL. 33711, USA 2Prof. Emeritus, Clemson University; Clemson, SC. 29634, USA 23 June 2015 ABSTRACT We introduce a new approach to a century old assumption which enhances not only planetary interior calculations but also high pressure material physics. We show that the polytropic index is the derivative of the bulk modulus with respect to pressure. We then augment the traditional polytrope theory by including a variable polytrope index within the confines of the Lane-Emden differential equation. To investigate the possibilities of this method we create a high quality universal equation of state, trans- forming the traditional polytrope method to a tool with the potential for excellent predictive power. The theoretical foundation of our equation of state is the same elas- tic observable which we found equivalent to the polytrope index, the derivative of the bulk modulus with respect to pressure. We calculate the density-pressure of six com- mon materials up to 1018 Pa, mass-radius relationships for the same materials, and produce plausible density-radius models for the rocky planets of our solar system. We argue that the bulk modulus and its derivatives have been under utilized in previous planet formation methods. We constrain the material surface observables for the inner core, outer core, and mantle of planet Earth in a systematic way including pressure, bulk modulus, and the polytrope index in the analysis. We believe this variable poly- trope method has the necessary apparatus to be extended further to gas giants and stars. As supplemental material we provide computer code to calculate multi-layered planets. Key words: planets and satellites:formation - stars:formation - methods:analytical - methods:numerical 1 INTRODUCTION As more exoplanets are discovered (Fran¸coise Roques 2015), the need for a systematic method for planetary models becomes more pronounced (Zapolsky & Salpeter 1969; Stacey & Davis 2004; Valencia et al. 2006, 2007a; Sotin et al. 2007; Seager et al. 2007; Grasset et al. 2009; Leconte & Chabrier 2012; Swift et al. 2012; Wagner et al. 2012; Zeng & Sasselov 2013; Alibert 2014). Our planet formation approach begins with the well known polytrope assumption at its foundation. A polytrope is a simple structural assumption between a material’s pressure and volume, P V n = C, or similarly it’s pressure and density, P = Cρn, where the C is a constant and n is the polytrope index. There have been some attempts to quantify what the polytrope index represents in the limited context of the ideal gas law (Chandrasekhar 1939; Horedt 2004). There have been other recent endeavors to import physical meaning to the polytrope index (de Sousa & de Araujo 2011; Christians 2012) but their results were limited in scope. To our knowledge, this work exhumes for the first time the physical definition of the polytrope index in the general case for all materials. In the body of this article we will first derive the physical interpretation of the polytrope index as the pressure derivative of the bulk modulus and show its kinship to the Murnaghan equation of state (EOS) (Murnaghan 1944). We then define a new variation of the traditional Lane-Emden equation in Sect. 2 and apply it in Sect. 3 by fitting the density-pressure ⋆ email address: [email protected] † deceased July 2014 c(cid:13) 0000 The Authors 5 1 0 2 n u J 2 2 . ] P E h p - o r t s a [ 2 v 5 2 5 5 . 9 0 4 1 : v i X r a 2 profile of the Earth constructed from seismic data. Finding this technique limiting we then permit the polytrope index to vary, modify the Lane-Emden equation which incorporates it, and apply this variable index technique to experimental high pressure-density curves for six common planetary materials up to 1018P a in Sect. 4. We then, using the same theory, develop a systematic approach for planetary models and calculate Mercury, Venus, Earth, and Mars (Sect. 5). Analyzing our EOS relative to previous universal EOSs, we spotlight the functional form of the polytrope index. This survey is done to elucidate the important utility this observable has in the future development of more sophisticated formation models. We finally examine the power of this technique by scrutinizing the interior of the Earth in Sect. 7 where we constrain, with mixed success, elastic material observables extrapolated to the surface of our planet. We conclude reviewing the potential of this variable polytrope approach. 2 THE POLYTROPE INDEX IS A DERIVATIVE OF THE BULK MODULUS Starting with the definition of the bulk modulus (B), the inverse of the more intuitive compressibility (K), we manipulate: from B = K −1 = ρ dP dρ , then then K −1 dρ = ρ, dP K −1 dρ dK dP dP = dK dP ρ, and by using the relationship, dK dP K −1 = − dK−1 dP K = − dB dP K we are able to progress to dρ dP dK dP dB dP = − K ρ and multiplying both sides by the awkward ρ dB dP dP is dimensionless) dB dP ρ −1 ( dB −1 dB dP −1 dB dP dB dP K then ρ dρ dP = − K dρ dP + dK dP dK dP dB dP ρ dB dP = 0. ρ (1) (2) (3) (4) (5) (6) (7) Though this looks complicated we are able to notice a derivative power relationship of the form d dP X n and so we have the intermediate result dY d dP where C ′ is a constant. From the P = 0 boundary condition we have dB dP K) = 0 or ρ dP K = C ′, (ρ dB dP (X nY ) = X n−1nY dX dP + C ′ = ρ dP )0 ( dB 0 K0 = ρn 0 K0 = ρ dB dP K, where the naught subscript represents the values of these variables at zero pressure. This derivation only requires that the exponent, the pressure derivative of the bulk modulus, remains constant because a simple derivative power law was assumed. This relationship, between density and the compressibility (or bulk modulus), has been shown previously by Stacey & Davis (2004, 2008a) in the context of the Murnaghan equation of state (EOS) (Murnaghan 1944). This EOS also assumes that the bulk modulus pressure derivative is constant and thus it reproduces Eq. (5). The derivation using the Murnaghan EOS is reproduced in Appendix A. This result, ρnK = C ′ = ρn 0 K0, will now be demonstrated as equivalent to a polytrope when n = dB dP is a constant. 0. The first step in producing ≡ B ′ Because this derivative is unchanging, it is equivalent to the value at zero pressure, ( dB the polytrope is replacing the compressibility with its definition ρn K = ρn 1 ρ dρ dP = C ′. dP )0 This equation is separable with ρ and P so multiplying both sides by dP and integrating Z ρn−1 dρ =Z C ′ dP, = C ′ P + D, we find ρn n which is in a modified polytrope form allowing for an additional intercept (D = ρn 0 /n) which can be non-zero (solids and liquids described by the Murnaghan EOS or self-bound neutron stars as in Lattimer & Prakash (2001)). In the special case when D = 0 (if ρ ≈ 0 when P = 0) we recover the normal form of Eddington (1916): (8) P = ρn n C ′ = Cρn, (9) MNRAS 000, 000–000 (0000) thus using Eq. (5) 3 nρn 0 B0 . (10) 0 K0 = = n C ′ = nρn 1 C We have successfully recovered a mathematically equivalent polytrope statement in Eq. (5) using the bulk modulus and com- pressibility instead of pressure. Our modified polytrope form of Eq. (8) is also equivalent to the Murnaghan EOS (Murnaghan 1944), see Appendix A for a derivation. More importantly we have shown that the polytrope index is equivalent to the deriva- tive with respect to pressure of the bulk modulus. The physical insight realized by the recognition of this elastic observable is significant. When the traditional polytrope has been successful (P = Cρn), most notably in the interior of stars, than this derivative in that star interior is close to constant. So most of the interior of a main sequence star, which fits closely to a 4/3 index Eddington (1916), implies that the derivative with respect to pressure of the bulk modulus is roughly 4/3. There has also been polytrope modeling success with white dwarfs (Chandrasekhar 1931), brown dwarfs (Stevenson 1991), red giants (Eggleton & Cannon 1991), and neutron stars (Lattimer & Prakash 2001). Thus the same physical interpretation of the index holds. Also the polytrope constant,C of Eq. (10), is now known in the interior of the star which puts powerful constraints on the formation constituents. It requires that Bi ρn i where Bi and ρi are the bulk modulus and density, at any spot in the interior of the star when n ≡ dB = nC = constant, (11) dP = constant. 3 DEVELOPING THE FIXED POLYTROPE TECHNIQUE Now on to deriving the gravitational differential equation for large spheres. This derivation is similar as to those traditional derivations found in Chandrasekhar (1939) and Horedt (2004) except now we use the modified form of our polytrope, ρnK = ρn 0 K0, from Eq. (5), instead of the traditional Eddington form, P = Cρn. One advantage being that Eq. (4) is more general and holds even when the density does not approach zero in the limit of zero pressure, another is that we replace the pressure as our independent variable with elastic constants. If P (pressure) and g (gravity) are functions of the radial distance r, as measured from the core, one may start by considering the equilibrium of radial force components on a concentric spherical shell element of interior radius r and exterior radius r +dr, within which differential changes dr, dP , dρ and dg alter their respective variables. Writing the forces as pressure times area and mass times gravitational acceleration and equating radial force components to zero, with the neglect of second order differentials, one obtains a differential equation for the pressure of the form dP dr This equation expresses the Newtonian equilibrium of hydrostatic forces. The density is affected by pressure through a compressibility relation of the form = −ρ(r)g(r) (12) K(ρ) = − 1 V dV dP = 1 ρ dρ dP (13) In writing these expressions it is to be noted that in a volume element whose mass is conserved, dm = 0, and thus 0 = V = − dρ d(ρV ) = ρdV + dρ and thus dV ρ . From Eq. (13) it is easily seen that ρK(ρ) = dρ/dP , while dρ/dr = (dρ/dP )(dP/dr). Therefore, Eq. (12) can be written as dρ dr This expression depends upon an understanding of how the compressibility K varies as a function of density. First one must observe from Eq. (14) that = −ρ2(r)K(ρ)g(r). (14) g(r) = − 1 ρ2K(ρ) dρ dr . (15) 0 4πr2ρ(r)dr. It is now clear that g(r) can also be written Also, the mass m(r) within a sphere of radius r is given by m(r) =R r as g(r) = Gm(r) r2 = 4πr2ρ(r)dr, where G is the Newtonian universal gravitation constant. Differentiating this expression, one now obtains (16) (17) (18) 0 G r2 Z r r3 Z r 2 0 Using this result along with Eq. (15), we may now write dg dr = G(cid:20)4πρ(r) − dr (cid:20)− ρ2K(ρ) dρ d 1 4πr2ρ(r)dr(cid:21) . r3 Z r 2 0 dr(cid:21) = G(cid:20)4πρ(r) − 4πr2ρ(r)dr(cid:21) . MNRAS 000, 000–000 (0000) 4 14000 12000 4 ) 0 K n 2 0 10000 ) 3 m / g k ( y t i s n e D 8000 6000 4000 ρ ( g o l 0 -2 2.5 3.0 3.5 4.0 4.5 n 2000 0x100 1x106 2x106 3x106 Radius (m) 4x106 5x106 6x106 Figure 1. (color online) In black, significantly buried under the theoretical calculations, is the Preliminary Reference Earth Model (Dziewonski & Anderson 1981) (PREM) and the newer Reference Earth Model (Kustowski et al. 2008) (REF). The colored lines are a sample calculation of ours for inner core (blue), outer core (orange), lower mantle (red), transition (light green) and upper mantle (yellow). There is excellent (less than 1%) disagreement between theory and experiment. The inset depicts the linear range of the polytrope index that share excellent agreement with experimental data using as a legend the same colors as the main plot. = 2g(r)/r, which with the further aid of Eq. (15), leads finally to In this expression, the second term on the right side can be identified with the help of Eq. (16), as 2G/r3R r dr (cid:20)− dr(cid:21) − − 4πGρ(r) = 0. rρ2K(ρ) ρ2K(ρ) dρ dr dρ d 1 2 0 4πr2ρ(r)dr (19) This can be put in a more explicit form by performing the indicated differentiation with respect to r in the leading term. In so doing it is convenient to make a substitution of the form ν(ρ) = ρ2K(ρ), and using this relationship: dν dr . The differentiation with respect to r will clearly yield a nonlinear term containing a factor of the squared derivative (dρ/dr)2. The algebra is straightforward though tedious, but finally leads to a nonlinear differential equation of the form d2ρ dr2 + 4πGρ3K(ρ) = 0, dr = dν dρ . dρ dρ dr (20) K ′ 2 r + + ρ −(cid:20) 2 dr(cid:19)2 K (cid:21)(cid:18) dρ where K ′ is the derivative of the compressibility with respect to density. One can also alternatively use Eqs. (13,19) to remove compressibility for pressure and get a version equivalent to the original Lane-Emden form used by Eddington, d2P dρ dr2 +(cid:20) 2 r − 1 ρ dr(cid:21) dP dr + 4πGρ2 = 0. This traditional version is mathematically simpler than the version derived above but it cannot use our modified polytrope assumption, ρn 0 K0 = ρnK, so we keep the more complicated Eq. (20). Now assuming the density-compressibility relationship of our modified polytrope ρn 0 K0 = ρnK or K = ρn 0 K0 ρn , Where ρ0 and K0 are the density and compressibility of a material at vacuum pressure, and then differentiating to get K ′ = 0 K0 −nρn ρn+1 (21) (22) (23) (24) We can reconstruct Eq. (20) as d2ρ dr2 dr(cid:19)2 ρ (cid:21)(cid:18) dρ −(cid:20) 2 − n dρ dr 2 r + + 4πGρn 0 K0ρ3−n = 0. A difference between our version of the Lane-Emden equation, Eq. (24), and the traditional equation, Eq. (21), is that ours already has assumed the polytrope assumption, Eq. (22), while Eq. (21) has not. If the polytrope assumption of Eq. (8) is used to remove pressure dependence in Eq. (21) than it will be mathematically congruent with our Eq. (24). This equivalence will be true for both zero and non-zero D. In Fig. 1 we show a density profile of planet Earth from the seismic analysis of the Preliminary Reference Earth MNRAS 000, 000–000 (0000) 5 dP ≡ B ′ Model (Dziewonski & Anderson 1981) (PREM) and the newer Reference Earth Model (Kustowski et al. 2008) (REF) giv- ing a profile of Earth’s inner core, outer core, and three stages of the mantle. Most of the seismic data is obscured by our calculation of the solution to Eq. (24) using our fixed polytrope assumption. Our calculation returned excellent results for the density profile and the gravitational and pressure profiles as well (not shown) using a variety of different vacuum densities, ρ0, bulk moduli, B0, and polytrope indices which are equal to the bulk modulus pressure derivative ( dB0 0 ≡ n). The inner most boundary had its density set at the value from the experimental data set, the first derivative initial value was constrained using Eq. (14) and Eq. (22). The poorest fit was for the liquid outer core but results were still well under 1% error. However the predictive power of this fixed index analysis is weak. With an ideal theory, one would assume that by choosing vacuum values which best fit the experimental seismic data one could elicit more detail about the composite materials comprising the interior Earth. Unfortunately a plethora of values for ρ0, B0, and B ′ 0 give excellent simultaneous fits to both the density-radius and the density-pressure profile. The inset of Fig. 1 details this lack of clarity. Each theoretical result which provided a good agreement to the subset seismic data of the interior Earth is provided by a point on a line of the form log10(ρn 0 K0) = m n + b. The polytropic index n worked well over a large range as long as ρ0 and K0 were constrained to the displayed line. By using an EOS which fixes this polytrope observable means that the best fits will reflect only the best average values of ( dB dP ) for that span of the interior of the Earth. For example the inner core is best reproduced with a fixed index range of 3.35 < n < 3.55 but it is know that at the surface the same materials, rich in iron, will have n > 4.5 (Seager et al. 2007; Sha & Cohen 2010; Garai et al. 2011; Yamazaki et al. 2012). Having a fixed polytropic index for planets does not represent the physical reality adequately, despite the fact that the pressure-density and density-radius profiles were described satisfactorily. Unlike the solar models which do have a near constant pressure derivative of the bulk modulus, the planet density-radius models, due to their relatively high compressibility, are rather insensitive to the pressure derivative of the bulk modulus (our newly discovered polytrope index). We will delve further into the analysis of the seismic data for Earth in Sect. 7, but next we advance a variable polytrope procedure which will give our model more physical insight. 4 MAKING THE POLYTROPIC INDEX A VARIABLE The ansatz that the polytropic index is constant is constraining and unrealistic. Eddington inherently recognized this with star formation and attempted a variable index analysis (1938). It is well known that for all materials the derivative of the bulk modulus with respect to pressure does change ( ∂ 2B ∂P 2 6= 0) (Zhang et al. 2007; Garai 2007; Singh 2010). We show later that the bulk modulus pressure derivative can best be approximated as a constant only at pressures below 109 Pa (small rocky planets or asteroids) or above 1014Pa (interiors of stars and very massive planets). Most planets have a pressure at their core which is between these extremes, including our Earth and all the planets in our solar system. Likewise the Murnaghan EOS was quickly deemed unsatisfactorily because of its constant derivative of the bulk modulus with respect to pressure so Birch proposed an extension (Birch 1947). A more realistic approach would allow the index to vary while at the same time retaining the definition for the polytrope index, n = dB dP . 4.1 Modifying the Polytrope So first assuming a modified polytrope with a variable index we must allow for the contingent that the polytrope expression, derived in Sect. 2, is no longer constant ρn(ρ)K = ρn(ρ) (25) 0 K0h(ρ), where we have introduced a new function, h(ρ) which is a measure of the non-conservation of the fixed expression or a weighting function. In the limit of the fixed case h(ρ) = 1. By taking the natural log of this equation and then taking the pressure derivative d dP (cid:18)n(ρ) ln ρ ρ0 + ln K K0(cid:19) = d dP ln h(ρ), and progressing by realizing that B = K −1, d dn(ρ) dB/dP n(ρ) ln + − B B = d dP ln h(ρ), dP , thus two terms cancel and we have dP ln ρ = K, and thus d dP ln K = − dB/dP B dP dP We chose to maintain n = dB dn(ρ) ln = ln h(ρ). d dP ρ ρ0 ρ ρ0 (26) (27) (28) This is a simple prescription on how use a modified polytrope with a variable index equivalent to the pressure derivative of the bulk modulus. We now have a new replacement for compressibility, using Eq. (25) we have K = ρn(ρ) 0 K0h(ρ) ρn(ρ) , MNRAS 000, 000–000 (0000) (29) 6 which is analogous to the fixed n version, Eq. (22), with the addition of a weighting function, h(ρ). 4.2 Changes to the Lane-Emden Equation K ′ −(cid:20) 2 The new modified Lane-Emden equation, which now includes a variable n. Starting with d2ρ dr2 it is trivial to remove K in terms of ρ using Eq. (29). We also need to find K ′ = dK no longer constant and needs to be treated with care. Starting again with Eq. (25) we take the logarithm of both sides dr(cid:19)2 K (cid:21)(cid:18) dρ + 4πGρ3K(ρ) = 0, dρ , which is a larger undertaking since n is dρ dr 2 r + + ρ d dρ recognizing that we can use earlier relationships if we assume a natural log log ρ − n log h − dn dρ = log ρ + dn dρ log ρ0, (30) d dρ log K = (log h + log ρn 0 K0 − n log ρ) d dρ d dρ n − + ln dρ ρ0 ρ dn dρ ρ(cid:19) = K(cid:18) d ln h dK dρ and then using Eq. (28) and the definition K = 1 ρ n dK dρ except now n is no longer a constant but a density dependent function. Thus the only symbolic change to the Lane-Emden equation is in the final term d2ρ dr2 dρ dP a cancellation occurs and we are left with the same result as with a fixed 0 K0ρ3−n(ρ) = 0, + 4πG h(ρ) ρn(ρ) = K ′ = K(− dρ dr (33) (31) (32) n ρ 2 r + ), −(cid:20) 2 − n(ρ) ρ dr(cid:19)2 (cid:21)(cid:18) dρ where the function h(ρ) is introduced and n, the polytrope index and equivalent to the derivative of the bulk modulus with respect to pressure, is also assumed a function of density as expected. This derivation is independent of the EOS function chosen for n(ρ) which, by Eq. (28), is also intrinsically linked to the weighting function h(ρ). 4.3 Our Equation of State To have a working hypothesis for a dynamic polytrope index EOS we examined previous equations of state (Roy & Roy 2005, 2006) and pressure-density relationships (Seager et al. 2007; Swift et al. 2012) for an empirical function form for the changing polytrope index. To be consistent we developed our own EOS which uses the informed polytrope index as the function of reference in place of the more common pressure EOS like the Birch-Murnaghan (Birch 1947; Poirier 2000) or Vinet (Vinet et al. 1987, 1989; Poirier 2000). We found that + A2 (34) n(ρ) = dB dP = A0(cid:18) ρ0 ρ (cid:19)A1 works well (more detail on how this empirical equation was developed is proffered in Sect. 5). It is a respectable EOS for pressures up to the 1013 Pa range. The resulting functional form for the polytrope index is also exceedingly simple. All other major relations to this polytrope index can be found analytically by integration or differentiation, using fundamental relations found in Appendix B. Specifically applying these relations to our EOS as expressed in Eq. (34) we find for pressure P = A0 A1 B0e A1 "(cid:18) ρ ρ0(cid:19)A2 En A1 + A2 A1 , A0 A1 (cid:18) ρ0 ρ (cid:19)A1! − En(cid:18) A1 + A2 A1 , A0 A1(cid:19)# . This pressure equation contains a special function, the generalized exponential integral (En) which is defined as (35) (36) En(n, x) =Z ∞ 1 e−xt tn dt. The parameters A0, A1, A2 are dimensionless, connected to experiment, and well behaved, always greater than zero and less than ten. They are connected to the extremes of the polytrope index: the infinite pressure derivative of the bulk modulus with respect to pressure,B ′ 0 = n0. They were fit as follows: A2 = B ′ ∞, A0 = B ′ . In this work we used the parameter settings in a predictive method, setting B0, and B ′ 0 close to experimental results, choosing a near fixed, but yet to be determined, infinite pressure derivative (1.75 < B ′ ∞ < 2.1), and assuming a universal ratio for B ′′ B ′ 0 0 − A2 ∞ = n∞ and the zero pressure asymptote of the same observable, B ′ 0 − A2, and A1 = −B0B′′ 0 by setting B ′ 0 A0 A1 = 0 −A2 (37) B ′ B′ 0 ≡ , MNRAS 000, 000–000 (0000) 7 which we borrowed from Roy & Roy (2006). Thus for any given material there are only three parameters (ρ0, B0, and B ′ 0). We will show our EOS compared to experiment in subsection 4.5 but first we require that it be continuous with a higher pressure theory to extend its validity. 4.4 Pressures above Ten Tera-Pascal Inspired by Seager et al. (2007) we went higher in pressure, to 1018 Pa, by following their technique to match at some critical density our empirical EOS to the Thomas-Fermi-Dirac theory (Salpeter & Zapolsky 1967) which treats at extreme pressures the quantum mechanic Fermi-Dirac gas whose pressure we approximated as P (ρ) = P0 + F3ρ + F2ρ4/3 + F1ρ5/3 (38) where F1 = F2 = F3 = 5.16 × 1012P a (2690 A Z )5/3 5.16 × 1012P a (2690 A Z )4/3 5.16 × 1012P a (2690 A Z )3/3 (.40726Z2/3 + .20732) (.01407), (39) the variable P0 of Eq. (38) is a small, relatively low pressure, addition to keep pressure continuity across the matching boundary and will be determined last. If the material is not monatomic we use a weighted average for A and Z (atomic and proton number respectively). Following the common definition used between the pressure and the bulk modulus, B = ρ dP dρ , we have for this high pressure theory B = (5/3)F1ρ5/3 − (4/3)F2ρ4/3 − (3/3)F3ρ3/3. Likewise n, our polytrope index equals dB dP thus dB dρ = n = ρ B , n = (5/3)2F1 − (4/3)2F2ρ−1/3 − F3ρ−2/3 (5/3)F1 − (4/3)F2ρ−1/3 − F3ρ−2/3 . (40) (41) (42) A key aspect of this method is to choose the critical density in which to switch from n(ρ) = A0( ρ0 ρ )A1 + A2, to the n of Eq. (42). We found that by requiring at the boundary that the bulk modulus, B, was continuous and using the secant method to find this critical density was all that was required. We also found that at this critical density n, the derivative of the bulk modulus, also was continuous. By adding a relatively small constant to pressure (P0 of Eq. (38)) we can also set the pressures equivalent at the critical density boundary. The reason we get a match between both B and n at the boundary is because these two variables are not independent. Recalling Eq. (41), dB dB dρ , we see that if ρ and B are attuned we get the two derivatives, n = dB dρ as bonus. With this extension to a high pressure EOS, we now feel comfortable in fitting up to 1018 Pa which are core pressures typical of small stars. dp = n = ρ dP and dB B Pragmatically we assume informed values for a material’s ρ0, B0, and B ′ 0. We then make initial approximations for A2 ≈ 2. We then solve for A0, and A1 of Eq. (34). Then a secant algorithm adjusts A2 and then determines A0, and A1 so at the critical density, ρc, the low pressure EOS becomes continuous with the higher pressure Thomas-Fermi-Dirac EOS (TFD) of Eq. (40) by finding at what critical density Bours − BT F D = 0. We then ensure that nours − nT F D = 0 and then we lastly solve for P0 of Eq. (38) to establish continuity for pressure, bulk modulus, and the pressure derivative of the bulk modulus across the critical boundary. As supplemental material we provide an example code to build a multi-layered planet which has this methodology within (see Sect. 5). 4.5 Comparison with experimental results In Fig. 2 we plot density versus pressure for six common materials that make up planet interiors in our solar system ranging from molecular hydrogen to atomic iron. The materials are either solids or liquids throughout the calculation and are also low temperature isothermal calculations (cold curves, 0K - 300K). We choose to examine an intermediate pressure range, 109 Pa to 1012 Pa, because it is challenging; it is at the extreme of our experimental capabilities and the interiors of our solar system planets are in this range. The solid black line of the inset is our EOS which was input into the Lane-Emden equation of Eq. (33) through the polytropic index (as a check of our work we used Eq. (35)). At these lower pressures we used the EOS described with Eq. (34), the values of the parameters can be also found in Appendix C. The filled triangles of various colors are MNRAS 000, 000–000 (0000) 8 6 5 n 4 3 2 6 5 4 n 3 2 1.8x104 1.6x104 1.4x104 1.2x104 1.0x104 8.0x103 6.0x103 4.0x103 7.0x103 6.0x103 5.0x103 4.0x103 3.0x103 2.0x103 1.0x103 ) 3 m / g k ( y t i s n e D ) 3 m / g k ( y t i s n e D 9 9.5 10 10.5 11 11.5 12 Log Pressure (Pa) MgO (+2500 kg/m3) SiO2 9 9.5 10 10.5 11 11.5 12 Log Pressure (Pa) 9 9.5 10 10.5 11 Log Pressure (Pa) Fe H2O He H2 11.5 12 Figure 2. (color online) We describe six materials with our EOS with a solid black line. In the bottom panel an older EOS, the Vinet (Vinet et al. 1987, 1989; Poirier 2000), is represented by a green dashed line (parameters for H2 (Loubeyre et al. 1996),parameters for He (Loubeyre et al. 1993),parameters for H2O (Tkachev & Manghnani 2000), we modified B′ = 7.25, well with in range to give a better fit). In the top panel an older EOS, the Birch-Murnaghan (Birch 1947; Poirier 2000), is also represented by a green dashed line (parameters for MgO (Duffy et al. 1995), parameters for SiO2 (Driver et al. 2010; Ping et al. 2013)), we took an aver- age value between the two references for B = 305 GPa and B′ = 5.0). The iron green dashed line is represented by the Vinet with parameters given by Anderson et al. (2001). More sophisticated first principle EOS (density functional theories) are a red or orange dashed line (H2-DFT (Geng et al. 2012),He-GGA (Khairallah & Militzer 2008), H2O-QMD (red) (French et al. 2009) and H2O- (orange) (Hermann et al. 2011), MgO-GGA-B1 (red) and qMS-Q ff (orange) (Strachan et al. 1999),SiO2-DFT (Driver et al. 2010),Fe- EXAFS (Ping et al. 2013)). Experimental data is expressed in purple, blue, and turquoise triangles (H2-purple (Zha et al. 1993), blue (Geng et al. 2012),He-purple (Cazorla & Boronat 2008),H2O-purple (Sugimura et al. 2008), MgO-purple,blue,turquoise (Duffy et al. 1995) SiO2-purple (Panero et al. 2003b),blue (Xun et al. 2010),Fe-purple, blue, turquoise (Sha & Cohen 2010)). The MgO curve is offset for aesthetic reasons, so that the calculation does not run into the calculation for SiO2. The inset shows the changing value of the polytrope index as the pressure changes in our EOS. All the parameters used for our calculation in this plot can be found in Appendix C. experimental data, the dashed green line is a common empirical theory from the twentieth century (Birch-Murnaghan (Birch 1947) or Vinet (Vinet et al. 1987, 1989; Poirier 2000) which begins to fail in the 1011 Pascal range, the dotted red and orange lines are density functional theories and/or quantum Monte Carlo theories from the twenty-first century. Our theory has the correct general trends, it fits theories and experimental data well considering that it has no adjustable parameters. Our method was that we first chose average experimental values for our material vacuum values of ρ0, B0, and n0. In many cases the literature values varied widely and so some prudence was used to determine the selected values. Our parameter A2 was chosen as the critical boundary value for n = dB dP when the EOS transitioned from our own to the Thomas-Fermi-Dirac MNRAS 000, 000–000 (0000) 1x107 1x106 1x105 1x104 1x103 1x102 ) 3 m / g k ( y t i s n e D 9 7 6 5 n 4 3 2 6 8 10 12 14 16 18 Log Pressure (Pa) Fe SiO2 MgO H2O He H2 6 8 10 Log Pressure (Pa) 12 14 16 18 Figure 3. (color online) Our EOS calculation shown over a large pressure range similar to Fig. 2. The inset also has the same form as the inset of Fig. 2, the change in the polytrope index as a function of the same pressure range as the main plot. EOS (Salpeter & Zapolsky 1967) (1.75 < A2 < 2.1) and we fixed out remaining parameters by constraining A0 + A2 = n0 and A0A1 = n0. All the parameters used to calculate the pressure-density profiles of Fig. 2 can be found in Appendix C. This success is proof of concept, that it is possible to develop an experimentally informed EOS which uses a polytrope function form, equivalent to dB dP , which can be naturally and consistently input into the dynamic index Lane-Emden equa- tion. It rivals the popular universal Birch-Murnaghan (Birch 1947; Poirier 2000) and Vinet (Vinet et al. 1987, 1989; Poirier 2000) in predictive potential as it mimics well the high pressure first-principle theories (French et al. 2009; Geng et al. 2012; Khairallah & Militzer 2008; Hermann et al. 2011; Driver et al. 2010; Ping et al. 2013). The inset of Fig. 2 plots the polytrope index as a function of the same pressure range for the six materials. The polytrope index, n = dB dP does change significantly over this pressure range, thus to keep the polytrope index fixed for planets in our solar system would be worse than approximate. Figure 3 shows our calculation over a wider pressure range, with a similar inset as Fig. 2. Note the general trends: up to 107 Pa the pressure density profile is a nearly incompressible horizontal line, the polytrope index, equivalent to the pressure derivative of the bulk modulus, wanders little from from its zero pressure value. The most dynamic range of change for the polytrope index is from 108 Pa to 1014 Pa as was illustrated in Fig. 2. As the pressure increases, our theory must switch from the low pressure EOS of Eq. (34) to the technique of Thomas-Fermi-Dirac of Sect. 4.4. The polytrope index again becomes nearly constant after 1014 Pa as each material begins its slow asymptotic drive towards the Thomas-Fermi-Dirac value of 5/3 (Salpeter & Zapolsky 1967). The parameters used for the materials in this figure can be found in Appendix C. A word of caution; since this is a simple universal EOS it will not contain any insight about quantum phase transitions which will take place in many of these materials above 1011 Pascals. As one approaches solar pressure range, as we do in Fig. 3 (> 1015P a), it is interesting to note that the polytrope index is again near constant. Was Eddington’s choice of the polytrope index set to 4/3 for main sequence stars (Eddington 1916) thus fortuitous? Assume, in a main sequence star, the most significant interactions under extreme pressure and temperature are gravity, electron repulsion, and heat flow. Eddington assumed only heat flow and gravity but completely neglected the significant electron repulsion in a dense solar gas yet still calculated satisfactory results. If we examine the derivative of the bulk modulus with respect to pressure, our newly revealed polytrope index, we discover why. For large massive objects like suns their partial bulk moduli with respect to pressure have similar values, greater than one and less than two (≈ 4/3 for heat flow (Eddington) and ≈ 5/3 for electron repulsion if the Thomas-Fermi-Dirac scheme of Salpeter & Zapolsky (1967) is followed). The thermal and electric contribution to the change in the bulk modulus is nearly the same as one follows the pressure gradient. Over ninety percent of a stars interior (all but the near surface) can be described well by a constant polytrope index that averages somewhere between 4/3 (thermal) and 5/3 (electrical). As seen in this work the sensitivity of the polytrope index is small enough that this 1/3 difference has little effect on the final results. The EOS developed is an unsophisticated example of the power of this variable polytrope method. At its foundation is the equivalency between the polytrope index and the important elastic variable, the pressure derivative of the bulk modulus. MNRAS 000, 000–000 (0000) 10 1x108 ) m 1x107 ( s u i d a R 1x106 X Y Z[ K10c ♀♁ ♂ ' 1x1023 1x1024 1x1025 1x1026 1x1027 1x1028 1x1029 Mass (kg) Figure 4. (color online) The legend of the six lines are the same as for Fig. 3, six materials that are common in planetary interiors. The symbols run vertically from top to bottom as Jupiter, Saturn, Uranus and Neptune (on top of each other), Earth and Venus (on top of each other), Mars, and Mercury. The symbol k10c stands for exoplanet Kepler 10c which is a super-earth (Dumusque et al. 2014). Other planet formation methods include a universal electric EOS in interplay with thermal effects (Stacey & Davis 2004; Valencia et al. 2006; Sotin et al. 2007; Grasset et al. 2009; Leconte & Chabrier 2012; Swift et al. 2012; Wagner et al. 2012; Zeng & Sasselov 2013) but they vary in their approach. This technique has as its centerpiece one elastic variable from which all dynamic interactions (electric, thermal, magnetic, nuclear) can contribute and thus one prescription to control input into the differential equations. This EOS has the ability to mature alongside its elastic variable polytrope index as we will discuss further in Sect. 7. 5 APPLICATION TO PLANETS We now apply our method to planets of our solar system. For planets the changing polytrope index is a must for all but the smallest planets. One looming approximation that needs to be addressed is the use of cold isothermal material curves to describe planets with hot interiors. We will address this in Sect. 7, but for now let us proceed assuming that this simplification is feasible. In Fig. 4 we solve the modified Lane-Emden equation with a variable index for the same six materials of the two previous figures except now we plot mass versus radius. This is a popular figure in planet modeling papers dating back to at least Zapolsky & Salpeter (1969). For comparison we designate the positions on the figure for the planets of the solar system. Described from vertical top to bottom: Jupiter, Saturn, Uranus and Neptune (almost on top of each other), Earth and Venus (almost on top of each other), Mars and Mercury. They are all placed close to their line of significant composition. Jupiter is close to hydrogen, Saturn has slightly more helium than Jupiter. Uranus and Neptune are between water and helium, Earth, Venus, Mars are between iron and the silicates. Mercury, the most metallic is closest to iron. These logical results give confidence to the validity of the method. Examining the previous results of Zapolsky and Salpeter (1969) and Seager et. al (2007) we have very similar results where the radius hits a maximum for all materials at a mass somewhere between one thousandth and one hundredth of a solar mass. This is not entirely surprising since all three studies used at extreme high densities the theory found in Salpeter & Zapolsky (1967). Our variable polytropic model does add additional physical insight into this maximum. It is well known that the traditional polytrope has a fixed radius for the n = 2 case (Chandrasekhar 1939), the radius grows with mass for fixed n > 2 and the radius shrinks with mass for fixed n < 2. This maximum of Fig. 4 represents the variable polytrope analog to this fixed radius case. All maximum of Fig. 4 occur within a small range, 1.78 < n(ρ) < 1.92. At the same time this result shows little correlation with the vacuum density (ρ0), bulk modulus (B0), and pressure derivative of the bulk modulus (n0) for all six materials. The importance of the n ≈ 2 range is also evident by examining the change of sign in the second term of Eq. (33) which is our final version of the variable polytrope Lane-Emden equation. The general MNRAS 000, 000–000 (0000) 11 Earth and data 1.2x104 Venus Mercury Mars ) 3 m / g k ( y t i s n e D 1.0x104 8.0x103 6.0x103 4.0x103 2.0x103 0x100 1x106 2x106 4x106 3x106 Radius (m) 5x106 6x106 Figure 5. (color online) Mercury, Venus, Earth, and Mars density-radius profile predictions using our EOS. The newer Reference Earth Model (Kustowski et al. 2008) seismic data is plotted as a black line. The details of construction are in the text. trends of Fig. 4 also validate the conclusions of Valencia et al. (2007b) which put maximum radial limits on rocky planets thus making it possible to separate the ice giants from the rocky giants. For an exoplanet super-earth example we put Kepler 10c also on Fig. 4 (labeled ’k10c’). This planet is the mass of Neptune, extremely hot at the core, and very dense (Dumusque et al. 2014), our temperature independent theory still places the planet in the correct bulk composition range. Turning now to specific models for the four rocky planets in our solar system. We constrain the boundary interfaces of these multi-layered planets by substituting for the compressibility, K, in Eq. (29) our variable index assumption K = ρn(ρ) 0 K0h(ρ) ρn(ρ) . Using Eq. (14), we are then left with a constraint on the derivative at each boundary = − ρn(ρ) 0 K0 h(ρ) ρ(r)n(ρ)−2 g(r). dρ dr The method of construction is illustrative. We assume the material makeup of the layers and the size of each layer, we then also assume the core pressure. There are no further adjustable parameters. The materials dictate the parameters in the equations of state, the core pressure determines the starting core density (by using Eq. (35)), and the interface constraint defines the density values and their first derivatives at the boundaries uniquely, the program runs from the core (r = 0) to the surface solving the differential equation of Eq. (33) and using the EOS as summarized in Eq. (34). Numerically this differential equation is very stable and solvable with relative ease, there are no variables that gave us numerical difficulty. Table 1 contains our pertinent values for these multi-layer constructions. The materials at each layer are picked by consulting the best ’general knowledge’ about each of the layers and estimating their average density, bulk modulus, etc. Since the rocky planets are relatively small the significance of Amean and Zmean is minimal since their dependence stems from the extremely high pressure theory (see subsection 4.4 for more detail). The method is a standardized way to put a multi-layered planet together. As supplemental material we include our code for the calculation of Earth (Variable PolyPy), written in Python. (43) In Fig. 5 we plot these hypothetical density-radius profiles for the four rocky planets. In Table 2 we have displayed the results for mass, radius, mean density, and gravity at the surface. For all three models the theoretical results are within a percent of experimental values. This technique provides a systematic method to build mult-layered planets. One of the benefits of using this method is that the EOS has its foundation in the elastic bulk modulus and its pressure derivative. We believe this is an asset that can be utilized to develop EOSs of greater sophistication, we study this question in the next section. 6 USING THE PRESSURE DERIVATIVE OF THE BULK MODULUS AS THE EQUATION OF STATE FOUNDATION The role of the pressure derivative of the bulk modulus (B ′ ≡ dB dP ), which we have shown to be the polytrope index, n, has grown in importance in material science science (Birch 1947; Keane 1954; Vinet et al. 1987, 1989; Poirier 2000; Roy & Roy MNRAS 000, 000–000 (0000) 12 ρ0 B0 kg/m3 GP a n0 Amean Zmean thickness m PC GP a Mercury core 1 core 2 core 3 mantle crust Venus core mantle crust Earth core 1 core 2 mantle 1 mantle 2 mantle 3 crust Mars core 1 core 2 mantle crust 7700 6410 4800 3320 1800 7475 3800 1800 7550 6830 3950 3650 3270 1800 7500 6100 3530 1850 180 132 220 200 160 160 190 170 171 140 190 205 135 160 180 130 200 160 5.0 4.9 4.8 4.1 4.0 4.95 4.0 4.2 5.0 5.0 4.4 4.1 3.8 4.0 5.0 4.9 4.2 4.1 55 55 44 36 30 47 36 30 55 55 36 36 36 30 55 47 36 30 26 26 21 18 15 22 18 15 26 26 18 18 18 15 26 22 18 15 1.12E6 8.80E5 1.00E5 2.00E5 1.36E5 4.04E1 2.67E1 5.63E0 3.63E0 0.94E0 3.00E6 2.98E6 7.11E4 2.95E2 1.22E2 1.14E0 1.22E6 2.26E6 2.23E6 2.51E5 3.84E5 3.72E4 3.64E2 3.29E2 1.36E2 2.36E1 1.37E1 0.66E0 6.00E5 9.20E5 1.82E6 5.10E4 4.04E1 3.66E1 2.20E1 0.35E0 Table 1. Mars in two steps, Jupiter and Uranus in three steps. The variables ρ0, B0, and n0 would be the values of those observables at zero vacuum. Thickness is the radial size of the core or mantle part, PC is the assumed pressure at the largest depth for each layer. mass kg radius mean density surf. gravity m kg/m3 m/s2 Mercury calculation experiment Venus calculation experiment Earth calculation experiment Mars calculation experiment 3.29E23 3.30E23 2.44E6 2.44E6 5.43E3 5.43E3 4.83E24 4.87E24 6.05E6 6.05E6 5.20E3 5.24E3 5.98E24 5.97E24 6.38E6 6.37E6 5.49E3 5.51E3 6.38E23 6.42E23 3.39E6 3.39E6 3.91E3 3.93E3 3.70 3.70 8.80 8.87 9.80 9.80 3.71 3.71 Table 2. The rocky planets of our solar system observable predictions using our EOS. The experimental results are taken from the National Aeronautics and Space Administration website. 2005, 2006). As strain theory continued to progress this elastic observable assumed a more prominent role. It is, frankly, a wonderful observable. It is dimensionless, of order one (for all materials and pressures), and because it is proportional to the second derivative of pressure-density it has tremendous sensitivity. This work has developed a robust technique where the variable polytrope index is equivalent to this elastic observable, B ′. MNRAS 000, 000–000 (0000) n 7 6 5 4 3 2 1 Present work Vinet Birch-Murnaghan Roy-Roy Stacey DFT Exp. Analysis 7 8 13 ) 3 m / g k ( y t i s n e D 800 700 600 500 400 300 200 9 10 10.5 9.5 Log Pressure (Pa) 11 9 Log Pressure (Pa) 10 11 12 13 Figure 6. (color online) A plot of n ≡ dB dB for cold-curve molecular hydrogen using a variety of EOSs. The solid black line is the present calculation, the green dotted line is the EOS of Vinet (Vinet et al. 1987, 1989), the pink dashed line is the third order EOS of Birch-Murnaghan Birch (1947). We also present the EOS of Stacey 2000 (blue dash - dot fixing the infinite dB dP at 2) and the EOS of Roy-Roy (Roy & Roy 2001, 1999) is depicted by a purple long dash - dot. A recent density functional theory (DFT) developed for molecular hydrogen (Geng et al. 2012) in red. All EOS (except the DFT) have fixed ρ0 = 79.43kg/m3, B0 = 0.162GP a. The n0 values are within 3% of each other (ours, Birch-Murnaghan is n = 6.7, Vinet is n = 6.813 and we have set Roy-Roy and Stacey to n = 6.9). Note the dramatic differences in the function n(P ) and its effects on the density-pressure relationship at extreme pressures. The experimental analysis, in blue triangles, was done by taking 2nd derivatives of experiments done by Loubeyre et al. (1996) 6.1 The importance of dB dP in material research Stacey (2000) developed an EOS that used B ′ ≡ dB dP = n as the starting point. This work gives a thorough history of this technique and rightly recognizes Murnaghan (1944) and Keane (1954) as instrumental in the development of the Stacey EOS. Stacey and Davis strongly believe that B ′ should be the center-piece of a universal EOS (Stacey & Davis 2004, 2008a; Lal et al. 2009; Singh & Dwivedi 2012). Calculating from first principles the electronic contribution to the bulk modulus is well established in the literature (Ravindran et al. 1998). The ability for modern material theorists to develop excellent B ′ theories from density functional theory is not in doubt (Zhang et al. (2010) for example). There have also been some recent high pressure models which include temperature as a direct extension of the pressure equation (Sotin et al. 2007; Wagner et al. 2011, 2012). Our calculation technique would use alternative methods which show the thermodynamic connection between B ′, the Gruneisen parameter (Stacey & Davis 2004, 2008a; Shanker et al. 2009) and temperature (Stacey & Davis 2004, 2008b), Sotin et al. (2007) gives a good overview of both techniques. The elastic constants, such as B ′, are a natural environment to develop temperature dependence as shown in a variety of research at low pressures (Zhang et al. 2007), first principle calculations of copper (Narasimhan & de Gironcoli 2002), magnesium oxide (Li et al. 2005) or high pressure iron (Sha & Cohen 2010) and high pressure salt (Singh 2010). We will examine this potential by adding a primitive temperature dependence in Sect. 7. Phase changes would also have a natural place in a foundational B ′ EOS. The discontinuities created by phase changes would be more significant in the bulk modulus and its derivative than in a pressure-density curve. The importance of magnetic effects has also recently been studied (Pourovskii et al. 2013) and similarly the B ′ equation has been shown to be a natural place to add this interaction (Zhang et al. 2010). Classical solid state theory may also be of use. The intermediate polytrope expression we derived in the context of the fixed polytrope index, ρnK = h(ρ)ρn 0 K0 (Eq. (25)) in Sect. 4, has some remarkable similarities to the functional form of the repulsive coulomb potential informed by Born-Madelung theory (Kittel 2005). 6.2 Comparisons of dB dP As stated in Sect. 4 we found that n(ρ) = dB dP = B ′ = A0(cid:18) ρ0 ρ (cid:19)A1 + A2 MNRAS 000, 000–000 (0000) 14 works well. It is a respectable EOS for pressures up to 1013 Pa and it can be extended to higher pressures by attaching it to quantum mechanical models as also discussed in Sect. 4. This EOS was cultivated by examining the functional form of dB dP for the Vinet and Birch-Murnaghan EOS (Birch 1947; Poirier 2000) and also analyzing pressure-density relationships found in previous work (Roy & Roy 2005, 2006; Seager et al. 2007; Swift et al. 2012) including first principle calculations (French et al. 2009; Hermann et al. 2011; Geng et al. 2012) and acquired numerical second derivatives. This was instrumental in developing the intermediate pressure ranges (1011 − 1013P a) which older EOS often do not fit well. Our result is entirely empirical but it meets the criteria that it is using B ′ as its foundation, that it is analytically simple, and that it strives to go at least an order of magnitude higher than the earlier EOSs. The analysis included taking analytical second derivatives of EOSs using B = ρ dP dρ dB dP n = = B ′ = d dP (cid:18)ρ dP dρ(cid:19) so dB B = n ρ dρ. (44) The analytical results for the Vinet, Birch-Murnaghan (Birch 1947; Poirier 2000), Stacey (2000), Roy & Roy (1999), and an EOS for hydrogen derived from density functional theory (Geng et al. 2012) are shown for completeness in Appendix D, these results are non-trivial fractions. One of the goals of this work was to develop a analytical formula for dB dP which is simpler than the two most used universal equations, the Vinet and Birch-Murnaghan, yet at the same time mimic their behavior. We portray dB dP in Fig. 6 for molecular hydrogen as a function of pressure for a variety of EOSs and in the inset we show the pressure-density result. This figure is similar to Figs. 2 and 3 but now the focus is on the behavior of the polytrope index so the plot proper and inset have been switched. We choose hydrogen because it offers challenges at relatively low pressures because of its high compressibility. It was seen in Fig. 2 that a standard Vinet EOS does not do well as one approaches pressures of 1013P a, and Fig. 6 gives the probable cause, the Vinet becomes soft as it heads towards a lower asymptote for n than any of the other EOSs do. In contrast, the complete failure of the Birch-Murnaghan EOS with these chosen parameters is because the asymptote for n is higher than the rest of the EOSs. All the EOSs (except the DFT by Geng et al. (2012) which is valid in a limited range only) approach a constant at extreme pressures but they disagree at what the infinite pressure constant (B ′ ∞) should be. Analysis of the importance of this asymptote is examined in good detail in the research of Stacey and Davis (Stacey 2000; Stacey & Davis 2004, 2008a). The Birch-Murnaghan, in Fig. 6, has the highest B ′ ∞ of 3. The Vinet has the lowest, B ′ ∞ ≤ 2. Our theory, because it connects to the Thomas-Fermi-Dirac high pressure theory has as its asymptote B ′ ∞ = 2 (they treat it as an adjustable parameter). Note the peculiarity of the density function theory (Geng et al. 2012). The authors choose a pressure EOS (see Ap- pendix D) which fit only the pressures applicable to their theory (1 × 1010 Pa to 4 × 1012 Pa). When derivatives are taken to produce dB dP the results are complicated and quickly unstable at the extremes (the ends of the plotted curve head towards infinity). This insensitivity provides an additional lesson on how it is easy to fit the density-pressure results and yet have faulty derivatives, an authoritative EOS will fit the pressure-density, bulk modulus-density and bulk modulus pressure derivative and density profiles. ∞ = 5/3, we set Stacey’s EOS to B ′ ∞ = 2/3, the rest are 5/3 ≤ B ′ The experimental density-pressure profile comes from Loubeyre et al. (1996). The technique we used to analyze this data to produce the polytrope index, as depicted in Fig. 6, is detailed in Appendix D. To summarize, we fit a density-pressure curve to a high order polynomial and then take two derivatives, using Eqs. 44, to achieve a good analysis independently, directly from the data, of n ≡ dB dP . We urge that this technique be used by experimentalists to further analyze their high precision data. With a good experimental set of density-pressure data one can also determine the bulk modulus and the pressure derivative of the bulk modulus over the same range. If one does want to extrapolate to vacuum pressure using a universal EOS, this information will help produce an authoritative EOS and further constrain the parameters chosen. A similar technique and conclusion was proffered by Ziambaras & Schroder (2003). ∞ > 5/3 but not necessarily B ′ It has been argued by Stacey & Davis (2004) that B ′ ∞ = 5/3. The argument is that phase changes and proton number dependence make it difficult to rectify that all materials will approach the same asymp- tote. We also believe this is an open question but still found it convenient to set it to the Thomas-Fermi-Dirac value of 5/3 (Salpeter & Zapolsky 1967). Yet we recognize that the present work and others (Seager et al. 2007) have found that this asymptote is not reached quickly. With the additional insight that the polytrope index is the derivative with respect to pressure of the bulk modulus we look at the interior of stars for guidance. All stars, at the core, have an index < 2. All stars are under extreme electronic, nuclear, and thermal pressure, thus indicating that the contributions to B ′ from these sources are all under 2 in the extreme pressure limit. Too often the universal EOSs are used to put limits on the bulk modulus and its derivative at zero pressure (B0, B ′ 0). All reasonable EOSs will fit any pressure-density curve if the parameters are adjusted enough as discussed in Stacey & Davis (2004), what Fig. 6 shows is the disparity that exists between the EOSs on the functional form and the asymptote of MNRAS 000, 000–000 (0000) dB dP . This observable needs to be measured or analyzed experimentally as well as theoretical approaches from first prin- ciples (Ziambaras & Schroder 2003). The observable should be ascertained independent of any universal EOS so the best functional forms can be developed. Molecular hydrogen, helium, and water, because of their higher compressibilities, are a good place to start. These informed results would catalyze the development of EOSs. With a trusted functional form, which mimics experiment for pressure-density, bulk modulus-density, and especially B ′-density (which is sensitive even at low pres- sures), there will be more faith in the physical interpretation of the parameters being adjusted. To constrain the universal functional form beyond the pressure-density profile would have significant impact on calculations of larger planets. In the next section we use these objectives on fitting Earth using our EOS. 15 7 ANALYSIS OF PLANET EARTH WITH OUR EOS The Earth is our best pseudo-static laboratory for studying materials under high pressures. In this section we analyze the Earth using different constraints on our EOS to better elucidate its strengths and weaknesses. By using the seismic data of planet Earth we now constrain, with our EOS, the material of the inner core, outer core and mantle using the pressure-density profile, and the density-radius profile. The best results still achieve an excellent fit and reveal a linear analysis similar to the inset of Fig. 1, (ρn0 0 K0 vs. n0) An example calculation was shown in Fig. 5 and the code which produced this calculation is included as supplementary material. The linear equations for the inner core, outer core, and lower mantle of the Earth are as follows: inner core: log10(ρn0 0 K0) = 3.994n0 − 11.813 outer core: log10(ρn0 0 K0) = 3.977n0 − 11.836 lower mantle: log10(ρn0 A reminder that the linear analysis constrains ρn0 0 K0 but does not guarantee that every point on the line will be a good fit, but all good fits are very near that line. We offer no explanation for this relationship but that its success does speak to the validity of universal EOSs and that the three variables (ρ0, K0, n0) are not independent. An oversimplification in our technique is that the experimental material fits were all to cold curves, we will address this assumption in the next subsection. 0 K0) = 3.702n0 − 11.717 7.1 Temperature Analysis Seager et al. (2007) have argued that temperature dependencies are significant but manageable (less than 10% in a pressure- density profile). We agree and the reasonable physical observables predicted in Fig. 4 and Fig. 5 for the rocky planets exemplify the power of using approximate cold curves. By using the seismic data of planet Earth we may now, with our equation of state, constrain further the material of the inner core, outer core and mantle and examine the role of temperature dependence for our EOS. As one moves from the surface of a planet to its core the polytropic index, dB dP , moves from close to its vacuum value to a lower value. The effect of temperature is to slow this descent as one immerses into the hotter interior in an approximately adiabatic path. For all of our calculations above, which were cold curves, we fixed the product and sum of our parameters to A0 + A2 = n0 and A0A1 = n0. The sum is required for all curves, the product was a fortuitous choice suggested by Roy-Roy (2006) which worked well for cold curves, it controls the relative size of the second derivative dn dρ = −A0A1 ρA1 0 ρA1+1 (45) (see appendix B for more detail). The cold curves A2 was set by a numerical secant search that connected our cold curve EOS to the Thomas-Fermi-Dirac EOS, the values for A2 was always between 1.75 and 2.1. So by increasing the value of A2 we decrease the magnitude of this derivative thus providing a simple method to include temperature. The procedure is simple: assume a reasonable material with a given vacuum density (ρ0), bulk modulus (B0), and first pressure derivative of the bulk modulus(n0). The constraints on A0, A1, and A2 are that n0 = A0 + A2 and n0 = A0A1. We treat as free parameters ρ0, B0, n0 and A2 and adjust them using a Powell method (constraining A0 and A1 by A0 + A2 = A0A1 = n0). So we are assuming that at the surface of the Earth (vacuum pressure) the density, bulk modulus, first derivative with respect to pressure, and all higher order derivatives have the same value irregardless of whether the isothermal or adiabatic path is chosen. But as one moves into the interior Badiab > Bisotherm thus the elastic derivatives are also disparate. In Stacey & Davis (2004, 2008b) one finds a thermodynamic derivation that provide a mapping Badiab = Bisothermal(1 + γα∆T ), (46) where γ is the Gruneisen parameter and α is the volume expansion coefficient (which are related to the bulk modulus and its pressure derivative). The reference temperature for ∆T is low, the temperature at the surface of the Earth. This ratio of Badiab over Bisothermal remains close to normal, in Stacey & Davis (2004) it never goes above 1.10 for planet Earth. This is a substantial piece of evidence that the isothermal cold curves are a good approximation for the interior of the Earth. MNRAS 000, 000–000 (0000) 16 13800 13600 13400 13200 13000 12800 ) 3 m / g k ( y t i s n e D n 3.5 3.0 2.5 2.0 3.2e+11 Inner Core 3.5e+11 Pressure (Pa) 3.8e+11 3.2e+11 3.3e+11 3.4e+11 3.5e+11 3.6e+11 3.7e+11 12000 ) 3 m / g k ( y t i s n e D 11500 11000 10500 10000 9500 Outer Core n 4.0 3.5 3.0 2.5 1e+11 2e+11 Pressure (Pa) 3e+11 1e+11 1.5e+11 2e+11 2.5e+11 3e+11 5500 5000 ) 3 m / g k ( y t i s n e D 4500 4000 Mantle 2e+10 4e+10 n 4.0 3.5 3.0 2.5 5e+10 1e+11 Pressure (Pa) 6e+10 8e+10 1e+11 1.2e+11 1.4e+11 Pressure (Pa) Figure 7. (color online) This figure is in three panels: the top referencing the inner core, the middle the outer core, and the bottom is the inner mantle. In black, significantly buried under the theoretical calculations, is the newer Reference Earth Model (Kustowski et al. 2008) (REF). The colored lines are all calculations of ours with different sophistication, all parameters are given in Table 3. The three models are named from the section from which they were introduced. Model (VII – two light blue short dashed) includes an adjustable second derivative ( d2B dP 2 ) to simulate the effect of temperature in the interior of the Earth. Model (IV - short dark blue dashed-dot) is the same material as (1) but the second derivative functional form has now been fixed to the cold curve value effectively removing the temperature dependence. Model (III – purple long dash-dot) is the least sophisticated fixed polytrope index (( d2B dP 2 = 0). For reference the cold curve for hcp iron is included for the cores (dashed red) and magnesium oxide (dashed pink) and silicon dioxide (dashed brown) is included for the mantle. The inset contains a depiction of the function n(P ) ≡ dB dP (P ) for our three models along with experimental analysis direct from the seismic wave data of the newer Reference Earth Model (Kustowski et al. 2008) (REF) depicted by the thick black line. dP dBisotherm As Badiab increases faster than Bisothermal as the pressure increases, this implies directly that nadiab ≡ dBadiab > ≡ nisotherm Pragmatically this is not too difficult, we must have an A2 greater than the cold isothermal curves (> 2.2) and so we vary ρ0, B0, n0 and A2 to fit the interior of the Earth. This fitting has only one more adjustable parameter,A2, than the cold isothermal curves. We will label this primitive temperature dependent method ’model VII’, after the section from which it was introduced. dP In Fig. 7 we examine the inner core, outer core and mantle of the Earth in more detail than Fig. 1 and Fig. 5. We begin by plotting the seismic results, as we did in Fig. 1, except now we plot the pressure-density profile (sold black line). Since the data sets are large and precise they are informative. Not only do they contain the pressure-density but they also implicitly contain the derivatives. Here we calculate the bulk modulus directly from the s and p wave velocities (see Stacey & Davis MNRAS 000, 000–000 (0000) 17 Model from Sect. ρ0 [kg/m3] B0 GP a n0 A0 A1 A2 Earth Inner Core Sect. VII Sect. IV Sect. III Fe Earth Outer Core Sect. VII Sect. IV Sect. III Earth Mantle Sect. VII Sect. IV Sect. III MgO SiO2 7563.0 7563.0 7372.6 8300.0 196.39 196.39 194.74 165.0 4.486 4.486 3.525 5.150 1.736 2.369 0.0 2.584 1.893 - 3.080 1.672 2.750 2.117 3.525 2.070 6654.2 6654.2 6596.7 131.32 131.32 153.33 4.748 4.748 3.500 2.096 2.652 0.0 2.266 1.791 - 2.652 2.096 3.500 3988.4 3988.4 3942.9 3580.0 4287.0 207.30 207.30 216.85 157.0 305.0 4.104 4.104 3.200 4.371 4.750 1.804 2.199 0.0 2.465 2.983 2.275 1.867 - 1.772 1.592 2.300 1.905 3.200 1.904 1.767 Table 3. ρ0, B0, n0 are chosen to be a good fit to the models of the Earth based on experiment. Model (VII - two dashed light blue on Fig. 7) is the most realistic allowing the parameters A0 and A1 to vary to simulate temperature dependence. Model (IV - dark blue short dash-dot on Fig. 7) is the same material with the temperature dependence removed and thus not as good fit. The simplest model (III - purple long dash-dot on Fig. 7) is a fixed polytrope index. The fixed polytrope does not require the sophistication of the earlier models but it can still be reduced to a specific case of the variable model by setting A0 = 0, A2 = n and noting that it is independent of A1. For reference we depict the cold curve prediction for ǫ-iron in the core panels, we add magnesium oxide and silicon dioxide for the mantle. (2004) for example) and we then calculate the pressure derivative of the bulk modulus numerically which is represented by the black line on the inset. We also plot three models from this work (named after the section number from which they were introduced): a good fit using the fixed polytrope index from Sect. 3 (model III, d2B dP 2 ≡ 0, purple horizontal line), a good fit using a variable polytrope index from this section (temperature dependent, model VII, light blue dashed line), and the same material as model VII with the temperature dependence removed, thus a isothermal cold curve as described in Sect. 4 (model IV, dark blue dashed line). For reference we also plot, from Sect. 4, the cold curves for iron, silicon dioxide, and magnesium oxide at those pressures. Model VII was fit simultaneously to the density-pressure curve (main graph), the bulk modulus, the density-radius curve (Fig. 1) and the pressure derivative of the bulk modulus-pressure curve (inset). This last constraint, depicted in the inset, shows the wonderful dynamics that are exhibited within the interior of the Earth. The outer core and mantle do not adhere to the monotonically decreasing pressure derivative of the bulk modulus of our cold curve or primitive temperature models but as the layer boundary approaches the second pressure derivative of the bulk modulus changes sign (the material’s rate of hardness growth increases!) as surmised by the analysis of the seismic data. The mantle and outer core have shown to be wonderful examples of phase-changes, non-adiabatic regions, and convection (Alboussi`ere et al. 2010; Katsura et al. 2010; Matyska et al. 2011) as the seismic waves slow as they near these boundary layers. The simple form of our EOS cannot describe this dynamic and so we hope only to stay close. Luckily the sensitivity to the pressure derivative is manageable. This polytrope method forces the user to contemplate beyond the density-pressure profiles to constraining the derivatives thus making it an ideal tool for bulk composition studies in the interior of a rich dynamic planet. Our simple temperature dependence does an adequate job describing the changing polytrope index for the mantle and the outer core, especially away from the inner boundary layer. The direction is correct, the magnitude is of the right order for the thermal pressure and the bulk modulus differential of Eq. (46) (Badiab/Bisothemal ≈ 1.01 for the mantle and ≈ 1.04 for the outer core). Most importantly we can describe the functional form of the polytrope index adequately. The inner core has the right direction and magnitude also (Badiab/Bisothemal ≈ 1.04 for the inner core) but we fail to find an adequate function to match the low experimental analysis of the interior polytrope index(n ≈ 2.3!). The most obvious potential source of error is that the functional form of our temperature dependence does not have the complexity to adequately describe the adiabatic curve relative to the cold temperature curve at inner core pressures. Recent experimental research (Tateno et al. 2010; Anzellini et al. 2013) also indicates the presence of a phase change near the geothermal which a universal EOS cannot describe. We also could easily be missing important magnetic (Pourovskii et al. 2013) or rotational effects. Again this method naturally extends the variables open to sensitivity examination thus providing a systematic tool to examine the successes and failures of any chosen EOS. We do have some faith in our estimation for n0 of the core because we do match the bulk modulus number at the core to a good precision. The results of the three models are systematically logical. For the best fit with realistic values for the vacuum density, MNRAS 000, 000–000 (0000) 18 bulk modulus, and pressure derivative of the bulk modulus the adiabatic curve version with an unsophisticated temperature dependence, model VII, had good success as the listings show in Table 3 and depicted in Fig. 7. This result implies that this technique has the potential to add temperature and any other dynamic variables if the effect of these variables on dB dP are known (profiles). Even with our simple adiabatic temperature model we were able to fit the density-radius profile to < 2%, the density-pressure profile to < 1%, the bulk modulus of the core to < 3%, and the pressure derivative of the bulk modulus for the outer core and mantle to < 10%, the inner core we fit to 30%. Recognizing that choosing our EOS as authoritative (the best functional forms for the cold curves) is a source of possible systematic error (when our EOS errs it is likely to error on the hard side as depicted by Fig. 6) as is our simple model for adiabatic temperature dependence (the slope of the polytrope index-pressure curve is low for mantle and outer core), we choose to constrain the interior of the Earth materials with conservative, fair errors: inner core: ρ0 = 7600 ± 300kg/m3, B0 = 190 ± 40GP a, and B ′ outer core: ρ0 = 6800 ± 300kg/m3, B0 = 130 ± 30GP a, and B ′ mantle: ρ0 = 4100 ± 200kg/m3, B0 = 200 ± 30GP a, and B ′ The constant ρn0 0 K0, which plays a pivotal role in this research, explains why the bulk modulus is difficult to constrain. This constant is extremely sensitive to the compressibility and its inverse, the bulk modulus. A small change in ρ or n will cause a much larger change in the bulk modulus if this term is to stay near constant. 0 = 4.7 ± 0.5 0 = 4.8 ± 0.3 0 = 4.2 ± 0.3 . If the effects of temperature and other dynamic variables are not specified explicitly then a reasonable material cold curve can be used instead and the error will be very manageable, less than 5% for a rocky planet like Earth as detailed in the analysis of the ratio of Eq. (46) which we always found to be under 1.05 using our primitive temperature dependence model. This trend is further bolstered by examining the difference between the density-pressure curves of Fig. 7. So we recommend the use of cold curves, with reasonable material observables, as a very good approximation for the actual geothermal path that describes the rocky planet under study. However we do not recommend the reverse analysis of trying to fit only the pressure-density experimental data with a cold curve to elucidate the the vacuum elastic variables the results will also be suspect and highly dependent on the EOS used. By numerically analyzing the experimental seismic data we have constrained our EOS further than most (fitting to pressure-density, density-radius, bulk modulus-pressure and dB dP vs. pressure) and can thus eliminate many previous parameter choices. We have also added a primitive temperature dependence which has the correct direction and the right order of magnitude. So we felt confident, with our model VII, to constrain the materials in the interior of the Earth with moderation. For better analysis we would need a reasonable temperature, magnetic, phase-change and convection profile, prepared from first principles, which would more tightly constrain the parameters. 7.2 Summary of Technique The method presented here is systematic, builds on history, and is able to grow in complexity as the EOS becomes more complex. To summarize this theoretical apparatus to solve the radius-density profile for a multi-layered planet: • Develop a functional form or differential equation(s) for the polytrope index as a function of density. The dynamic variables that describe this polytrope index function can be sophisticated and include a full thermal, electromagnetic, and nuclear profile: n(ρ) ≡ dB(ρ) for an appropriate core material with a given ρ0, B0, B ′ 0 ≡ n0 and core pressure. dP • Solve analytically or numerically for the weighting function:ln h(ρ) =R ρ ρ0 dn dρ ln( ρ rho0 ) dρ • Starting at the inner core (r = 0) find the core density using the weighting function h(ρ) and core pressure relationship, also set dρ dr = 0. • Solve the Lane-Emden differential equation (Eq. (33)) to a chosen radius (or if the surface, when ρ(r) = ρ0). • If another layer is desired choose an appropriate material with a given ρ0, B0, B ′ 0 then calculating n(ρ) and h(ρ) as shown above. • Choose an initial density for the next layer and a starting radius (r = rf inal of the previous layer). • Find the initial derivative, dρ • Solve the differential equation to a chosen radius (or if the surface, when ρ(r) = ρ0). Repeat the last 4 steps if additional dr by using Eq. (43) and Eq. (33). layers are desired. If one uses a series of differential equations and profiles which produce numerical functions only for the polytrope index, n(ρ), it would be preferred to start at the surface of the planet and work inward since the calculus relations will also not be analytical. This would involve some more computational challenges, but they are not insurmountable as Zeng & Seager (2008) have demonstrated. MNRAS 000, 000–000 (0000) 19 8 SUMMARY AND CONCLUSION In summary this work derived a variable polytrope approach. To investigate the method we also developed a new universal EOS which is distinctive because it is a function of the polytrope index. This work first exhumed, in Sect. 2, that the historic polytrope index is for all cases and materials equivalent to the derivative with respect to pressure of the bulk modulus (n = dB dP ). This result is the foundation of the article, the further development that follows is inspired and motivated by recognizing the physical implications of the polytropic index as an elastic constant thus giving it a life beyond merely index status. We solve the Lane-Emden differential equation for plausible solutions to the density profile of the five layer interior of the Earth. We find that there is a wide range of constant polytropic indexes (n = dB dP ) that agree with the seismic data of the interior of the Earth. In Sect. 4 we expand our technique by making the polytropic index a variable. Since the index has been shown to be equivalent to the derivative with respect to pressure of the bulk modulus (n = dB dP ) this is a natural progression since all materials have a decreasing first derivative along the isotherm when the pressures become very large. We re-derive the Lane-Emden differential equation for a variable polytropic index in a consistent manner. We then introduce an empirically developed functional form for our variable polytropic index, n(ρ) = A0( ρ0 ρ )A1 + A2, which is our submission for a isothermal universal equation of state. With this creation we can create analytical formulas for pressure and bulk modulus as a function of density. We, more importantly, use our variable polytropic index as input in the Lane-Emden differential equation, we are able to make pressure-density profiles for six materials which are common in the interior of planets which compare favorably to experiment. By connecting our results to high pressure theory (Salpeter & Zapolsky 1967) we are able to make theoretical calculations up to 1018P a. We turn to the planets of our solar system in Sect. 5 where we show that their masses and radii compare favorably with mass-radius curves for the six common materials. We then show that this technique can be applied to a multi-layered body as we fit our technique to the rocky planets. In Sect. 6 and Sect. 7 we discuss the importance of developing a universal EOS from dB dP , we then compare our functional form for dB dP to previous work and to experimental analysis. The importance of constraining a EOS fit beyond the mundane pressure-density profile is demonstrated. We show that our EOS has the advantage of being simple and thus easy to grow in sophistication. One addition that is needed is the adding of temperature dependence, we do this in primitive form and show a variety of results using the interior of the Earth as our laboratory. We also verify earlier work that shows that temperature modification is a relatively small effect, under 5% or the Earth’s density-pressure profile. Finally using a variety of experimental analysis, we are able to narrow the constituent vacuum densities of the Earth to 5%, the bulk moduli to 20%, and the pressure derivative of the bulk moduli to 10%. Having revealed the polytrope index as the pressure derivative of the bulk modulus is fortuitous for future research advancement. There are many first principle EOS theories (density functional, quantum Monte-Carlo) which calculate the pressure derivative of the bulk modulus (Ravindran et al. 1998; Sha & Cohen 2010; Xun et al. 2010; Li et al. 2005). The prominence of the bulk modulus in this procedure may also lends itself to solid state Born-Madelung calculations (Kittel 2005). By using the modified polytrope assumption of this work one also has a straight forward procedure to solve the Lane-Emden equation for planets and stars using more sophisticated equations of state. These methods may include dynamic convective interactions, heat flow, rotational dynamics, magnetic kinetics, and nuclear reactions which would all contribute to the overall pressure derivative of the bulk modulus, the variable polytrope index, complementing a new generation of realistic planetary and solar models. We hope that this technique strengthens the relevance of the polytrope method for planet and solar research and continues what Lane, Eddington, and Chandrasekhar began a century ago. ACKNOWLEDGEMENTS S. P. Weppner and J. P. McKelvey would like to thank Don Vroon and Ray Hassard for bringing them together and Rohan Arthur for comments. K. D. Thielen and A. K. Zielinski would like to thank Eckerd College for an internal grant which funded them during the summer of 2013. It is with great sadness that we report the passing of J. P. McKelvey in July of 2014. REFERENCES Alboussi`ere T., Deguen R., Melzani M., 2010, Nature, 466, 744 Alibert Y., 2014, Astron. Astrophys., 561, A41 Anderson O. L., Dubrovinsky L., Saxena S. K., LeBihan T., 2001, Geophys. Res. Lett., 28, 399 Anzellini S., Dewaele A., Mezouar M., Loubeyre P., Morard G., 2013, Science, 340, 464 Birch F., 1947, Physical Review, 71, 809 Cazorla C., Boronat J., 2008, J. Phys.: Condensed Matter, 20, 015223 Chandrasekhar S., 1931, ApJ, 74, 81 Chandrasekhar S., 1939, An Introduction To The Study of Stellar Structure. Dover Publications Christians J., 2012, International Journal of Mechanical Engineering Education, 40, 53 MNRAS 000, 000–000 (0000) 20 Driver K. P., Cohen R. E., Wu Z., Militzer B., R´ıos P. L., Towler M. D., Needs R. J., Wilkins J. W., 2010, Proceedings of the National Academy of Sciences, 107, 9519 Duffy T. S., Hemley R. J., Mao H.-k., 1995, Phys. Rev. Lett., 74, 1371 Dumusque X., et al., 2014, The Astrophysical Journal, 789, 154 Dziewonski A., Anderson D., 1981, Phys.Earth Planet.Interiors, 25, 297 Eddington A. S., 1916, Mon. Not. R. Astron. Soc., 77, 16 Eddington A. S., 1938, Mon. Not. R. Astron. Soc., 99, 4 Eggleton P. P., Cannon R. C., 1991, ApJ, 383, 757 Fran¸coise Roques O. d. P., 2015, ExoSolar Planets Encyclopedia, http://exoplanet.eu French M., Mattsson T. R., Nettelmann N., Redmer R., 2009, Phys. Rev. B, 79, 054107 Garai J., 2007, Journal of Applied Physics, 102, Garai J., Chen J., Telekes G., 2011, American Mineralogist, 96, 828 Geng H. Y., Hong S. X., Li J. F., Wu Q., 2012, J. Appl. Phys., p. 063510 Grasset O., Schneider J., Sotin C., 2009, ApJ, 693, 722 Hermann A., Ashcroft N. W., Hoffmann R., 2011, Proceedings of the National Academy of Sciences Horedt G., 2004, Polytropes Applications in Astrophysics and Related Fields. Kluwer Acasemic Publishers Katsura T., Yoneda A., Yamazaki D., Yoshino T., Ito E., 2010, Physics of the Earth and Planetary Interiors, 183, 212 Keane A., 1954, Aust. J. Phys., 7, 322 Khairallah S. A., Militzer B., 2008, Phys. Rev. Lett., 101, 106407 Kittel C., 2005, Introduction to Solid State Physics, 8 edn. Wiley, Hoboken, NJ, pp 47–87 Kustowski B., Ekstrom G., Dziewo´nski A. M., 2008, J. Geophys. Res., 113 Lal K., Singh C. P., Chauhan R. S., 2009, Indian J. Pure & Ap. Phys., 47, 28 Lattimer J., Prakash M., 2001, ApJ, 550, 426 Leconte J., Chabrier G., 2012, Astron. Astrophys., 540, A20 Li H., Xu L., Liu C.-Q., 2005, Journal of Geophysical Research (Solid Earth), 110, 5203 Loubeyre P., LeToullec R., Pinceaux J. P., Mao H. K., Hu J., Hemley R. J., 1993, Phys. Rev. Lett., 71, 2272 Loubeyre P., Letoullec R., Hausermann D., Hanfland M., Hemley R. J., Mao H. K., Finger L. W., 1996, Nature, 383, 702 Matyska C., Yuen D. A., Wentzcovitch R. M., C´ızkov´a H., 2011, Physics of the Earth and Planetary Interiors, 188, 1 Murnaghan F. D., 1944, Proceedings of the National Academy of Sciences, 30, 244 Narasimhan S., de Gironcoli S., 2002, Phys. Rev. B, 65, 064302 Panero W. R., Benedetti L. R., Jeanloz R., 2003a, J. Geophys. Res.: Solid Earth, 108, ECV 10 Panero W. R., Benedetti L. R., Jeanloz R., 2003b, J. Geophys. Res.: Solid Earth, 108, ECV 5 Ping Y., et al., 2013, Phys. Rev. Lett., 111, 065501 Poirier J.-P., 2000, Introduction to the physics of the Earth’s Interior. Cambridge University Press Pourovskii L. V., Miyake T., Simak S. I., Ruban A. V., Dubrovinsky L., Abrikosov I. A., 2013, Phys. Rev. B, 87, 115130 Ravindran P., Fast L., Korzhavyi P., Johansson B., Wills J., Eriksson O., 1998, Journal of Applied Physics, 84, 4891 Roy S. B., Roy P. B., 1999, J. Phys.: Condensed Matter, 11, 10375 Roy B., Roy B., 2001, physica status solidi (b), 226, 125 Roy P. B., Roy S. B., 2005, J. Phys.: Condensed Matter, 17, 6193 Roy P. B., Roy S. B., 2006, J. Phys.: Condensed Matter, 18, 10481 Salpeter E., Zapolsky H., 1967, Phys.Rev., 158, 876 Seager S., Kuchner M., Hier-Majumder C., Militzer B., 2007, ApJ, 669, 1279 Sha X., Cohen R. E., 2010, Phys. Rev. B, 81, 094105 Shanker J., Singh B. P., Jitendra K., 2009, Condensed Matter Phys., 12, 205 Singh P. K., 2010, Indian J. Pure & Ap. Phys., 48, 403 Singh P. K., Dwivedi A., 2012, Indian J. Pure & Ap. Phys., 50, 734 Sotin C., Grasset O., Mocquet A., 2007, Icarus, 191, 337 Stacey F. D., 2000, Geophys. J. Int., 143, 621 Stacey F. D., Davis P., 2004, Phys. Earth Planet Inter., 142, 137 Stacey F. D., Davis P., 2008a, Physics of the Earth, 4 edn. Cambridge University Press, New York, NY, pp 294–313 Stacey F. D., Davis P., 2008b, Physics of the Earth, 4 edn. Cambridge University Press, New York, NY, pp 314–336 Stevenson D. J., 1991, Ann. Rev. of Astron. Astrophys., 29, 163 Strachan A., Cagin T., Goddard W. A., 1999, Phys. Rev. B, 60, 15084 Sugimura E., Iitaka T., Hirose K., Kawamura K., Sata N., Ohishi Y., 2008, Phys. Rev. B, 77, 214103 Swift D. C., et al., 2012, ApJ, 744, 59 Tateno S., Hirose K., Ohishi Y., Tatsumi Y., 2010, Science, 330, 359 Tkachev S. N., Manghnani M. H., 2000, in Manghnani M. H., Nellis W. J., Nicoles M. F., eds, Vol. 1, Science and Technology of High Pressure:Proceedings of the International Conference on High Pressure Science and Technology. Universities Press, pp 137–140 Valencia D., O’Connell R. J., Sasselov D., 2006, Icarus, 181, 545 Valencia D., Sasselov D. D., O’Connell R. J., 2007a, The Astrophysical Journal, 656, 545 Valencia D., Sasselov D. D., O’Connell R. J., 2007b, The Astrophysical Journal, 665, 1413 Vinet P., Ferrante J., Rose J. H., Smith J. R., 1987, J. Geophys. Res.: Solid Earth, 92, 9319 Vinet P., Rose J. H., Ferrante J., Smith J. R., 1989, J. Phys.: Condensed Matter, 1, 1941 Wagner F. W., Sohl F., Hussmann H., Grott M., Rauer H., 2011, Icarus, 214, 366 Wagner F. W., Tosi, N. Sohl, F. Rauer, H. Spohn, T. 2012, Astron. Astrophys., 541, A103 Xun L., Xian-Ming Z., Zhao-Yi Z., 2010, Chinese Physics B, 19, 127103 Yamazaki D., et al., 2012, Geophys. Res. Lett., 39, n/a Zapolsky H., Salpeter E., 1969, ApJ, 158, 809 MNRAS 000, 000–000 (0000) Zeng L., Sasselov D., 2013, Publ. Astron. Soc. Pac., 125, pp. 227 Zeng L., Seager S., 2008, Publ. Astron. Soc. Pac., 120, 983 Zha C.-s., Duffy T. S., Mao H.-k., Hemley R. J., 1993, Phys. Rev. B, 48, 9246 Zhang Y., Zhao D., Matsui M., Guo G., 2007, J. Geophys. Res.: Solid Earth, 112, 2156 Zhang P., Wang B.-T., Zhao X.-G., 2010, Phys. Rev. B, 82, 144110 Ziambaras E., Schroder E., 2003, Phys. Rev. B, 68, 064112 de Sousa C. M. G., de Araujo E. A., 2011, Mon. Not. R. Astron. Soc., 415, 918 APPENDIX A: RELATIONS BETWEEN DENSITY AND COMPRESSIBILITY FROM THE MURNAGHAN EOS 21 = ( ρ ρ0 )B′ 0 , It was Stacey and Davis 2004; 2008a who we have seen state that in the context of the Murnaghan EOS the relationship B B0 which is equivalent to our own intermediate results for the fixed polytrope index (Eqs.(4,5)). The derivation of Eq. (A1) is straightforward but we include it in some detail because the relation is used extensively in this work. (A1) Assuming only a linear change in the bulk modulus B = B0 + B ′ 0P, where B ′ 0 is the fixed derivative with respect to pressure. One can then insert the definition of the bulk modulus ρ dP dρ = B0 + B ′ 0P, and integrating and describing the zero pressure boundary conditions by using a naught subscript, one gets ln(B = B0 + B ′ 0P ) B ′ 0 = ln + ρ ρ0 ln B0 B ′ 0 which reduces to ρ ρ0 B ′ 0P B0 ) = ( (1 + )B′ 0 . Finally using Eq.(A2) to substitute B ′ 0P = B − B0 we have the desired relation, Eq. (A1). The Murnaghan EOS is found by solving Eq. (A5) for pressure P = B0 B ′ 0 (cid:18)( ρ ρ0 )B′ 0 − 1(cid:19) . The similarity of this equation to the polytrope is undeniable (A2) (A3) (A4) (A5) (A6) APPENDIX B: OUR EMPIRICAL EQUATION OF STATE We developed an equation of state from analysis of earlier empirical equations of state. The major difference is ours uses the polytrope index, the pressure derivative of the bulk modulus, as its place of development. It was first stated in the text proper as Eq. (34), again n(ρ) = dB dP = A0(cid:18) ρ0 ρ (cid:19)A1 + A2. The constants are chosen as follows: A2 → B ′ (1967)), A0 = B ′ details In this work we set B0, and B ′ 0 − A2, and A1 = −B0B′′ B′ 0 ∞ (approximately 1.95 or connecting it to the results of Salpeter & Zapolsky ∞ is the asymptotic value at very high densities, see Appendix C for more where B ′ 0 −A2 0 to experimental results and assume a universal ratio for B ′′ 0 by setting A1 = B ′ 0 0 − A2 B ′ = B ′ 0 A0 , which we borrowed from Roy-Roy (2006). Now using these fundamental relations B = ρ n = d dP (cid:18)ρ dP dρ(cid:19) dP dρ dB dP = dB B ρ ρ0 so = dρ n ρ + ln B0 B ln h = n ln MNRAS 000, 000–000 (0000) (B1) (B2) (B3) (B4) (B5) , A0 A1(cid:19)# (B6) (B7) (B8) (B9) (B10) (B11) (B12) En A1 + A2 A1 , A0 A1 (cid:18) ρ0 ρ (cid:19)A1! − En(cid:18) A1 + A2 A1 ρ ρ0(cid:19) − A0 A1 22 in conjunction with Eq. (28) dn(ρ) dP ln ρ ρ0 = d dP ln h(ρ). It can thus be shown dh h Specifically applying these relations to our EOS as expressed in Eq. (34) we find dn dρ ρ ρ0 = ln dρ. + A2 A0 A1 n = P = dB dP B0e A1 ρ (cid:19)A1 = A0(cid:18) ρ0 ρ0(cid:19)A2 B = B0eM (ρ)(cid:18) ρ "(cid:18) ρ ρ0(cid:19)A2 d2B dP 2 = −A1 A1 (cid:18) ρ0 A0 h = e−M (ρ)(cid:18) ρ ρ (cid:19)A1(cid:18)1 + A1 ln ρ0(cid:19)A0( ( dB dP − A2) ρ0 ρ )A1 = B , dn dP ln h = noting that only n(ρ) and h(ρ) are needed to solve the Lane-Emden differential equation, Eq. (33). The functions in the exponents are given by M (ρ) = A0 A1 (1 − ( ρ0 ρ )A1 ) The only equation which is non-analytical is pressure which contains a special integral function, the generalized exponential integral (En) which is defined as En(n, x) =Z ∞ 1 e−xt tn dt. By using the assumption of Ref. (Roy & Roy 2006), we are equating the second derivative as B ′′ 0 = − B ′ 0 B0 . (B13) (B14) This is true at the surface but we can find the general relation also for our EOS. First taking the derivative of the polytrope index with respect to density, = −A0A1 dn dρ then switching variables to a pressure derivative we have ρA1 0 ρA1+1 = −A0A1 dn dP By doing a substitution for A1 in terms of A2 (using Eq. (B1)) in Eq. (B16) we have finally = −A0A1 = −A1 dρ dP B . ρA1 0 ρA1+1 ρA1 0 BρA1 ( dB dP − A2) dn dP = d2B dP 2 = −( B ′ 0 0 − A2 B ′ ) ( dB dP − A2) B , (B15) (B16) (B17) which has an interesting symmetry. This relationship for the second derivative has been studied in some detail recently in Singh & Dwivedi (2012). APPENDIX C: PARAMETERS CHOSEN FOR MATERIALS In the text proper we used six materials commonly found in the interior of planets in our solar system. For completeness and reproducibility we include our parameters for the lower pressures (< 1013 Pa) in Table C1. The parameters, ρ0, B0, and n0 = dB dP , were not adjusted and were set to values which best reflect the experimental results (Zha et al. 1993; Khairallah & Militzer 2008; Sha & Cohen 2010; Duffy et al. 1995; Panero et al. 2003a; Driver et al. 2010) the rest of these parameters are derived from these three experimental values. To first approximation these parameters are A2 = 1.95,A1 = n0 n0−1.95 (or adjustable for best fit) and A0 = n0 − A2 if one needs pressures only up to the Tera-Pascal range. MNRAS 000, 000–000 (0000) 23 material ρ0 B0 [kg/m3] GP a n0 A0 A1 A2 H2 He H2O MgO SiO2 Fe 79.43 291.73 998.0 3580.0 4287.0 8300.0 0.162 0.224 2.20 157.0 305.0 165.0 6.70 7.15 7.13 4.37 4.75 5.15 4.837 5.141 5.248 2.465 2.983 3.080 1.385 1.391 1.359 1.772 1.592 1.672 1.863 2.009 1.882 1.904 1.767 2.070 Table C1. ρ0, B0, n0 are chosen to be a rough consensus of the experimental values at vacuum pressure. The values of A0, A1, A2 follow the prescription of Eq. (C1) which is slightly adjusted on each iteration of the secant method searching for the critical density. First order approximate values are A2 = 1.95,A1 = n0 n0−1.95 and A0 = n0 − A2 where n0 = dB dP 0 = B′ 0. material H2 He H2O MgO SiO2 Fe ρc kg/m3 1.690E4 1.229E4 3.758E5 1.262E6 1.577E7 9.736E5 Bc GP a 1.154E5 1.623E4 7.360E6 4.478E7 3.967E9 1.998E7 nc 1.866 2.037 1.883 1.905 1.767 2.071 Pc GP a P0 GP a A Z 6.165E4 7.801E3 3.900E6 2.351E7 2.245E9 9.631E6 2.782E3 7.806E2 2.027E5 1.434E6 4.396E7 1.2806E6 2 4 18 40 60 55.85 2 2 10 20 30 26 Table C2. The critical values are found by solving for the critical density, ρc, using the secant method when the bulk modulus of our EOS, Eq. (B11) matches the bulk modulus of the degenerative Fermi gas bulk modulus of Thomas-Fermi-Dirac, Eq. (40). The parameters Pc and P0 are found by forcing the pressure formula of Eq. (40) to equal the pressure formula listed in Eq. (B14) by choosing an appropriate P0 At higher pressures we switch to a Thomas-Fermi-Dirac scheme of Ref. (Salpeter & Zapolsky 1967). To make this trans- formation we find a critical density where we match B and dB dP = n, we then create a match with pressure by adding a constant P0 to the Thomas-Fermi-Dirac scheme. We first assume an initial guess using a secant method to find the critical density. The density is than iterated until it meets the requirement BT F D(ρc) − Bclassical(ρc) = 0. At this critical density the derivative of the bulk modulus with respect to pressure also matches thanks to the relationship in Eq. (41). At each iteration of the search we adjust A2 and then we are able to set A0 and A1: A2 = nT F D − (n0 − nT F D) ∗ (ρ0/ρc) n0 A1 = n0 − A2 A0 = n0 − A2, n0 n0−nT F D (C1) where nT F D is the past iterative guess using the present iteration of ρc in Eq. (42) while n0 is set to the rough experimental consensus value. The results of the secant method with additional useful parameters (atomic number, A, and proton number, Z) are in Table C2. Again no adjustable parameters are in this table, they were found using the secant method by requiring that the bulk modulus matches at the boundary between the low pressure and high pressure theories. APPENDIX D: THE POLYTROPE INDEX FOR VARIOUS EQUATIONS OF STATE Our Equation of state is developed from a functional form of dB dP which is detailed in Appendix B. We repeat them here: B ′ ≡ n = dB dP + A2 (1−( = A0(cid:18) ρ0 ρ0 ρ (cid:19)A1 ρ )A1 )(cid:18) ρ "(cid:18) ρ ρ0(cid:19)A2 ρ0(cid:19)A2 En A1 + A2 A1 B = B0e A0 A1 P = A0 A1 B0e A1 The equation of state of Stacey also starts with B ′ = dB dP , 1 B ′ = 1 B ′ 0 +(cid:18)1 − ∞ B ′ B ′ 0 (cid:19) P B MNRAS 000, 000–000 (0000) , A0 A1 (cid:18) ρ0 ρ (cid:19)A1! − En(cid:18) A1 + A2 A1 , A0 A1(cid:19)# (D1) (D2) (D3) (D4) 24 which then one can derive a formula for the bulk modulus and the density ratio B Most equations of state start from a definition of pressure, one can calculate n = dB ∞ ∞ P B(cid:19) −(cid:18) B ′ 0 B ′ − 1(cid:19) P B′ 0 B′ ∞ B = B0(cid:18)1 − B ′ ln ρ ρ0 = − B ′ 0 B ′ ∞ ∞ P B(cid:19)− 2 ln(cid:18)1 − B ′ dP from P (ρ) analytically dB dP = d dP (ρ dP dρ ) = ρ dB/dρ B Many of the next equations use η = ρ ρ0 = V0 V . The Vinet EOS (Vinet et al. 1987, 1989; Poirier 2000) is P = 3B0η B = 3B0(cid:18) 2 3 2 3 (cid:16)1 − η− 2 1 3(cid:17) exp( 1 η 3 − η 3 + 1 2 1 3 3 )2η 3 )2η ( 2 2 3 − ( 1 B ′ = (B ′ 0 − 1)(1 − η− 1 3 )) 3 2 (B ′ 0 − 1)(η 1 3 − 1)(cid:19) exp( 1 3 − 1 3 ) + 1 0 − 1)(1 − η− (B ′ 3 2 0 − 1)2(1 − η− 1 3 ) 1 3 )) 1 3 + 1 2 3 η 2 (B ′ 3 − 1 3 η 2 0 − 1)(η 3 + 1 1 2 (B ′ 0 − 1)(η 4 (B ′ 3 − 1) 1 The Birch-Murnaghan third order EOS is (Birch 1947; Poirier 2000) P = B = 3 2 3 2 3 )2η ( 7 B ′ = B0(η 7 3 − η 5 3 )(cid:18)1 + 3 4 (B ′ 0 − 4)(η 2 3 − 1)(cid:19) B0(cid:18) 7 3 η 5 3 + η (B ′ 3 4 7 3 − 5 3 3 )2η 3 − 5 7 7 3 − ( 5 5 3 + 3 4 (B ′ 7 3 η 3 η 5 3 + 3 4 (B ′ 3 3 )2η 0 − 4)(cid:18) 9 0 − 4)(cid:16)( 9 0 − 4)(cid:16) 9 3 η 9 η 3 − 2( 7 )η 3 + 7 3 3 )2η 3 + 5 7 7 5 3 η 5 3(cid:19)(cid:19) 3(cid:17) 3 )2η 5 9 3 − 2( 7 3 + ( 5 9 3 − 2( 7 3 )η 3 η 5 3(cid:17) (D5) (D6) (D7) (D8) (D9) (D10) (D11) (D12) (D13) One must be careful with the Roy-Roy equation of state, they have two different forms. The one they consider supe- rior (Roy & Roy 1999, 2006, 2005), which has a logarithmic form, does not have a physical B ′ ∞ asymptote (it is designed to go no higher than 1012 Pa). The one we depict in Fig. 6 and Appendix D has a physical asymptotic form and has B ′ ∞ = 5/3 (Roy & Roy 2001, 1999). The Roy & Roy (1999) EOS is η = 1 + aP (1 + bP )c B = B ′ = aP + abP 2 + (1 + bP )1−c a + ab(1 + c)P abc(1 + bP )−c(−b(1 + c)P − 2) + a2(1 + bP (b(1 + c)P + 2)) (a + ab(1 + c)P )2 where a = b = c = 1 B0 1 + 2B ′ 0)2 − 1 0 − 5(B ′ 6B0(1 − B ′ 0) 0)2) 0)2 − 1 (3(1 − B ′ 0 − 5(B ′ 2B ′ The DFT EOS equation of state (Geng et al. 2012) is P = 10N(ρ) B = − 10N(ρ)(a1ρ4/3 + 2a2ρ + 3a3ρ2/3 + 4a4ρ1/3 + 5a5) ln(10) 3ρ5/3 B ′ = −25a2 5 ln(10) − 25a5ρ5/3 + 40a4a5ρ1/3 ln(10) + 30a3a5ρ2/3 ln(10) 15a5ρ5/3 + 12a4ρ6/3 + 9a3ρ7/3 + 6a2ρ8/3 + 3a1ρ9/3 + − − 20a2a5ρ ln(10) + 5a1a5ρ4/3 ln(100) 15a5ρ5/3 + 12a4ρ6/3 + 9a3ρ7/3 + 6a2ρ8/3 + 3a1ρ9/3 16a2 4ρ2/3 ln(10) + 9a3ρ7/3 + 4a2ρ8/3 + a1ρ9/3 + ρ4/3(3a3 + 2a2ρ1/3 + a1ρ2/3)2 ln(10) 15a5ρ5/3 + 12a4ρ6/3 + 9a3ρ7/3 + 6a2ρ8/3 + 3a1ρ9/3 16a4ρ6/3 + 24a3a4ρ3/3 ln(10) + 8a4a1ρ5/3 ln(10) + 8a4a2ρ4/3 ln(100) 15a5ρ5/3 + 12a4ρ6/3 + 9a3ρ7/3 + 6a2ρ8/3 + 3a1ρ9/3 (D14) (D15) (D16) (D17) (D18) (D19) (D20) (D21) (D22) MNRAS 000, 000–000 (0000) 25 with a0 = 1.0683, a1 = 19.1824(26961/3 ), a2 = −36.3776(26962/3 ), a3 = 28.5165(26963/3 ), a4 = −10.6068(26964/3 ), a5 = 1.5224(26965/3 ). These complicated functions were calculated using Mathematica. The power functions are N (ρ) = 9 + a0 + a1 ρ1/3 + a2 ρ2/3 + a3 ρ3/3 + a4 ρ4/3 + a5 ρ5/3 (D23) In Fig. 6 we also attempted to analyze experimental data independent of a universal EOS. We take high precision pressure- density data that enters the 109Pa range for light elements and the 1010Pa range for heavier elements. It is also advantageous to have at least two points of low pressure data (< 108Pa) that anchors the fit towards the correct origin. In the case of the hydrogen data we found that a 5th order polynomial gave a very low chi-squared. Once we have a polynomial for P , calculating the bulk modulus and the polytrope index are trivial. P (GP a) = a0 + a1ρ + a2ρ2 + a3ρ3 + a4ρ4 + a5ρ5 B(GP a) = a1ρ + 2a2ρ2 + 3a3ρ3 + 4a4ρ4 + 5a5ρ5 n = a1ρ + 4a2ρ2 + 9a3ρ3 + 16a4ρ4 + 25a5ρ5 a1ρ + 2a2ρ2 + 3a3ρ3 + 4a4ρ4 + 5a5ρ5 (D24) (D25) (D26) with a0 = 2.26603GP a, a1 = −4.52911 × 10−2GP a/(kg/m3)1, a2 = 2.16832 × 10−4GP a/(kg/m3)2, a3 = 1.77254 × 10−8GP a/(kg/m3)3, a4 = 1.64780 × 10−10GP a/(kg/m3)4, a5 = −1.48451 × 10−13GP a/(kg/m3)5. It also helps to nor- malize the fitting routine, have as your dependent fitting variable ρ/(2ρmax) so that your domain is much less than one, the range should also be small, here we fit it to Giga Pascals. MNRAS 000, 000–000 (0000)
1803.11307	1	1803	2018-03-30T01:37:23	Kepler's Earth-like Planets Should Not Be Confirmed Without Independent Detection: The Case of Kepler-452b	[ "astro-ph.EP" ]	We show that the claimed confirmed planet Kepler-452b (a.k.a. K07016.01, KIC 8311864) can not be confirmed using a purely statistical validation approach. Kepler detects many more periodic signals from instrumental effects than it does from transits, and it is likely impossible to confidently distinguish the two types of event at low signal-to-noise. As a result, the scenario that the observed signal is due to an instrumental artifact can't be ruled out with 99\% confidence, and the system must still be considered a candidate planet. We discuss the implications for other confirmed planets in or near the habitable zone.	astro-ph.EP	astro-ph	Draft version April 2, 2018 Typeset using LATEX twocolumn style in AASTeX61 KEPLER'S EARTH-LIKE PLANETS SHOULD NOT BE CONFIRMED WITHOUT INDEPENDENT DETECTION: THE CASE OF KEPLER-452b Fergal Mullally,1 Susan E. Thompson,1, 2, 3 Jeffrey L. Coughlin,1, 2 Christopher J. Burke,1, 2, 4 and Jason F. Rowe5 1SETI Institute, 189 Bernardo Ave, Suite 200, Mountain View, CA 94043, USA∗ 2NASA Ames Research Center, Moffett Field, CA 94035, USA 3Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 4MIT Kavli Institute for Astrophysics and Space Research, 77 Massachusetts Avenue, 37-241, Cambridge, MA 02139 5Dept. of Physics and Astronomy, Bishop's University, 2600 College St., Sherbrooke, QC, J1M 1Z7, Canada ABSTRACT We show that the claimed confirmed planet Kepler-452b (a.k.a. K07016.01, KIC 8311864) can not be confirmed using a purely statistical validation approach. Kepler detects many more periodic signals from instrumental effects than it does from transits, and it is likely impossible to confidently distinguish the two types of event at low signal- to-noise. As a result, the scenario that the observed signal is due to an instrumental artifact can't be ruled out with 99% confidence, and the system must still be considered a candidate planet. We discuss the implications for other confirmed planets in or near the habitable zone. Keywords: stars: individual(Kepler-452) - (stars:) planetary systems - stars: statistics 8 1 0 2 r a M 0 3 . ] P E h p - o r t s a [ 1 v 7 0 3 1 1 . 3 0 8 1 : v i X r a ∗ email: [email protected] 2 Mullally et al. 1. INTRODUCTION Kepler's results will cast a long shadow on the field of exoplanets. The abundance of exo-Earths derived from Kepler data will dictate the design of future direct detec- tion missions. An oft neglected component of occurrence rate calculations is an estimate of the reliability of the underlying catalog. As frequently mentioned in Kepler catalog papers (Batalha et al. 2013; Burke et al. 2014; Rowe et al. 2015; Mullally et al. 2015; Coughlin et al. 2016; Thompson et al. 2017), not every listed candidate is actually a planet. Assuming they are all planets leads to an over-estimated planet occurrence rate (see, e.g., § 8 of Burke et al. 2015). This problem is especially pro- nounced for small ((cid:46) 2 R⊕), long-period (> 200 days) candidates, where the signals due to instrumental effects dominate over astrophysical signals. External confirmation of Kepler planets as an inde- pendent measure of reliability is therefore an impor- tant part of Kepler's science. While progressively more sophisticated approaches to false positive identification have been employed by successive Kepler catalogs, it is not possible to identify every false positive with Kepler data alone. Independent, external measures of catalog reliability are a necessary step towards obtaining the most accurate occurrence rates, as well as defining a set of targets for which follow-up observations can be planned. Also, the intangible benefit of being able to point to specific systems that host planets, not just can- didates, should not be understated. Kepler-452b is espe- cially interesting in this regard. The target star is sim- ilar to the sun, the orbital period is close to one Earth year, and its measured radius (1.6± 0.2 R⊕) admits the possibility of a predominantly rocky composition (Wolf- gang et al. 2015; Rogers 2015). While small, short-period planets can be confirmed with radial velocities (see, e.g., Marcy et al. 2014), longer period planets are hard to detect in this manner. Tor- res et al. (2011), building on work by Brown (2003) and Mandushev et al. (2005), introduced the technique of statistical confirmation, which used all available obser- vational evidence (including high resolution imaging to identify possible background eclipsing binaries) to rule out various false positive scenarios involving eclipsing bi- naries to the level that the probability the observed sig- nal was a planet exceeds the probability of an eclipsing binary by a factor of 100 or more. Morton et al. (2016) used a simplified technique which could be tractably run on the entire candidate catalog, finding and confirming 1284 new systems. Lissauer et al. (2012) deduced that the presence of two or more planet candidates around a single star meant that neither was likely to be a false positive and first suggested that a false alarm probability of <1% was sufficient to claim confirmation of a candi- date. Rowe et al. (2014) used Lissauer's argument to claim statistical confirmation of 851 new planets. Rowe et al. and Morton et al. together confirmed the vast majority of confirmed Kepler planets. However, in this paper we argue that statistical con- firmation of long-period, low SNR planets is consider- ably more challenging than previously believed. We use Kepler-452b (Jenkins et al. 2015) as a concrete example of a planet that should not be considered confirmed to il- lustrate our argument, but our argument may also apply to other statistically confirmed small planets (or candi- dates) in or near the habitable zone of solar-type stars.. Our argument relies on two lines of evidence. First, the number of instrumental false positive (hereafter called false alarm) signals seen in Kepler data before vetting dominates over the number of true planet signals at long period and low signal-to-noise (SNR). Second, it is dif- ficult (if not impossible) to adequately filter out enough of those false alarms while preserving the real planets, even with visual inspection. The sample of planet can- didates in this region of parameter space is sufficiently diluted by undetected false alarms that it may be im- possible to conclude that any single, small, long period planet candidate is not an instrumental feature with a confidence greater than 99% without independent ob- servational confirmation that the transit is real. §7.3 of the final Kepler planet candidate catalog (Thompson et al. 2017) discusses the tools and techniques necessary to study catalog reliability, and we draw heavily from that analysis. 2. THE LIMITS OF STATISTICAL CONFIRMATION The basic approach for statistical validation is to com- pute the odds-ratio between the posterior likelihood that an observed signal is due to a planet, to the probabil- ity of some other source. The Blender approach (Torres et al. 2011), used by Jenkins et al. (2015) to confirm Kepler-452b, computes P (planet) P (planet) + P (EB) (1) where P (EB) is the posterior probability the signal is due to an eclipsing binary, and the denominator sums to unity. Morton et al. (2016) expanded on this approach by fitting models of two types of non-transit signals to their data, effectively computing P (planet) P (planet) + P (EB) + P (FA) (2) where FA indicates a false alarm due to a non- astrophysical signal. Notably, they assume the priors on those false alarm models to be low, reflecting their Kepler-452b 3 Figure 1. Each panel shows a single transit from a TCE discovered in one of our data sets. The time scale in each panel is hours from mid-transit, with the points the Kepler pipeline considers "in-transit" highlighted with the vertical blue bar. The panel at far right combines all four events in a folded lightcurve (the red square symbols show the binned folded transit). One row shows transits from a known false alarm TCE found in the inversion run, the other from Kepler-452b. It is not clear from visual examination which one is a claimed planet, and which one is a known systematic signal. (We provide the answer at the end of the paper.) belief that only a small fraction of their sample were non-astronomical signals. For high SNR detections, or for candidates with peri- ods between 50–200 days, P (FA) is indeed small, and can essentially be ignored. However, Mullally et al. (2015) first noted the same is not true for candidates with periods longer than ∼ 300 days and clustered close to the SNR threshold for detection (defined as Multiple Event Statistic, MES, > 7.1). They urged caution in interpreting those candidates as evidence that there was an abundance of planets between 300 and 500 days. The difficulty in distinguishing low signal-to-noise planets from false alarms is illustrated in Figure 1. One of the panels shows individual transits from Kepler- 452b, and the other from a simulated dataset mimick- ing the noise properties of Kepler data (Coughlin 2017). (We summarize how simulated data were created in §3.) Both are detected as TCEs (Threshold Crossing Events, or periodic dips in a lightcurve) by the Kepler pipeline (Twicken et al. 2016). It is not obvious, even to the eye, which is the real planet, and which is the artifact. When planets and false alarms look so similar there is no definitive test to distinguish the two. Any vetting process must trade-off between completeness, the fraction of bona-fide planets correctly identified (or the true positive rate) and effec- tiveness, the fraction of false alarms correctly identified (or the true negative rate). Most of these false alarms are at long period and low SNR. At long periods, a few systematic "kinks" in a lightcurve can often line up, sometimes with low- amplitude stellar variability, to produce a signal that looks plausibly like a transit. These chance alignments are common for 3-4 transit TCEs, but their frequency declines with 5 or more transits. Candidates with 5 or more transits are less likely to be false alarms because the probability of getting 5 events to line up periodically is smaller (see Figure 9 of Thompson et al. 2017, for more details). The combination of less than ideal effec- tiveness at low SNR (because transits and false alarms are difficult to tell apart), and an over-abundance of false alarms at long period, leads to a surprisingly low catalog reliability, or the fraction of claimed planet candidates actually due to transits, and not due to instrumental effects. 3. THE RELIABILITY OF EARTH-LIKE PLANET CANDIDATES The catalog of Thompson et al. (2017) uses a num- ber of tests, collectively called the Robovetter (Coughlin et al. 2016; Thompson et al. 2015; Mullally et al. 2016), to identify false positive and false alarm TCEs. Each test must be tuned to balance the competing demands of completeness (that as many true planets as possible are passed by each test) and effectiveness (that as many false positives of the targeted type as possible are rejected by 4 Mullally et al. the test). The large number of tests applied means that each test must be tuned for high completeness, or the number of bona-fide planets incorrectly rejected quickly becomes unacceptably large. To test the completeness, effectiveness, and reliability of the planet candidate catalog, the Kepler mission pro- duced three synthetic data sets. Christiansen (2017) measured completeness by injecting artificial transits into Kepler lightcurves and seeing how many were re- covered by the Kepler pipeline. Producing a sample of transit free lightcurves with realistic noise and system- atic properties was more difficult. The mission tried two approaches, each with their own strengths and weak- nesses, as discussed in more detail in Thompson et al. (2017). The first approach (called "inversion") was to invert the lightcurves (so that a flux decrement be- comes an excess) The second approach was to shuffle the lightcurves in time, or "scramble" them, to remove the phase coherence of the real transits. All simulated data sets are available for download at the NASA Exo- planet Archive1. In the discussion that follows, we will consider the per- formance of the Robovetter for TCEs with periods in the range 200-500 days and MES (Multiple Event Statistic, or the SNR of detection of the TCE in TPS, Christiansen et al. 2012) less than 10. This parameter space comfort- ably brackets Kepler-452b and contains sufficient TCEs to perform a statistical analysis. For brevity, we will refer to this sample as the Earth-like sample because it encompasses the rocky, habitable zone planets Ke- pler was designed to detect (Koch et al. 2010). We assume for simplicity that completeness and effective- ness is constant across this box, although in reality the Robovetter performance declines with increasing period and decreasing MES. Thompson et al. (2017) ran the Robovetter on a union of TCEs produced by inversion and the first scrambling run (Coughlin 2017)2. They measure effectiveness as the fraction of these simulated TCEs correctly identified as false alarms in the given parameter space. The Robovet- ter performs extremely well on these simulated TCEs, with an effectiveness of 98.3%. For every thousand false alarms, the Robovetter correctly identifies 983 of them and mis-classifies only 17 as planet candidates. The original observations (i.e. the real data) produced a huge number of (mostly false alarm) Earth-like TCEs, as shown in Figure 2. The Robovetter examined 3341 TCEs with periods between 200 and 500 days and MES < 10 in the original data for DR25. Of these, 3274 1 Simulated data available at https://exoplanetarchive.ipac. caltech.edu/docs/KeplerSimulated.html 2 Three scrambling runs were created, but we only use the first one for consistency with Thompson et al. (2017) are identified as not-transit-like (i.e. unlikely to be ei- ther due to a transit or eclipse; not-transit-like is the label given by the Robovetter to TCEs without a realis- tic transit shape). The measured effectiveness suggests that an additional 56 false alarm TCEs were mistakenly labeled as planet candidates. There are only 67 planet candidates in this region of parameter space, but if 56 are likely uncaught false alarms, then only 11 are bona-fide transit signals. The reliability of the sample (the fraction of planet candi- dates that are actually due to an astrophysical event such as a transit or eclipse) is then 11 / 67 = 16.0%. This low reliability means that there is only a 16.0% chance that any one planet candidate is actually due to a planet transit, and not a systematic event, far lower than the 99% threshold for statistical confirmation suggested by Lissauer et al. (2012), or the 99.76% level confidence claimed by Jenkins et al. (2015) for Kepler-452b. The astrophysical false positive rate is clearly only part of the puzzle. For Earth-like candidates, the instrumental false alarm rate must be included in order to statistically confirm planet candidates. Worse, our ability to estimate reliability is limited by the small numbers of simulations in the parameter space of interest. If we assume the number of uncaught false alarms is drawn from a Poisson distribution, our un- certainty in the number of false alarms is 56±7.5. The uncertainty in the reliability is then 16±11%. Not only do we fail to measure a false positive probability of <1%, our current data prevents us from measuring the false positive probability with a precision of < 1%. Our analysis is slightly over-simplified for clarity. A more detailed calculation would correct the number of measured false alarms for the small number of bona-fide transits mistakenly rejected as false alarms. It should also consider the reasons the Robovetter failed simu- lated transits, as some rejected TCEs were rejected as variable stars, eclipsing binaries etc. However, none of these improvements can hope to raise the measured reli- ability to > 99%. The error in the computed reliability due to these simplifications is likely small compared to the systematic error in measured effectiveness due to the different abundances of false alarms in the original and simulated data. In particular, assuming constant relia- bility across the sample slightly over-estimates the reli- ability of the Kepler-452b, as the effectiveness decreases with decreasing numbers of transits. The fidelity of the simulated false alarm populations is discussed in some detail in § 4.2 of Thompson et al. (2017). Combining the TCEs from the inversion run and one scrambling run produces a distribution of sim- ulated TCEs as a function of period that matches the observed distribution quite well, but not exactly. Be- cause the sources of the observed false alarm TCEs is not precisely understood, we can not rule out the possibility Kepler-452b 5 Figure 2. Distribution of planet candidates (large blue circles) and non-candidate TCEs (small grey circles) from the DR25 catalog. The marginal histograms in period (and MES) are shown in the top (and right) panels. The rejected population is dominated by TCEs caused by instrumental systematics in the lightcurves. Because there are so many of these instrumental false alarms, the small number that slip through vetting are a significant fraction of the planet candidate catalog in this region of parameter space. The reliability of any individual planet candidate (or the probability it is due to an astrophysical event and not an instrumental feature) is too low to allow any single candidate to be confirmed without additional observational evidence. The location of Kepler-452b is indicated by the red star. that the simulated datasets are over-producing a certain kind of systematic signal that the robovetter performs well at detecting while under-producing a different kind of signal the robovetter struggles with, thereby overesti- mating our measured reliability. The converse may also be true. The similarity in the period distributions im- plies that the simulated populations are well matched to the observed ones, but there is still a small systematic error in our measured reliability that we don't yet know how to measure. Our argument does not directly apply to the major- ity of statistically confirmed Kepler planets. Transits with high SNR are much less likely to be systematics, and much easier to identify as such if they are. The two largest contributions to the set of statistical confir- mations, Rowe et al. (2014) and Morton et al. (2016), both restrict their analysis to candidates detected with a SNR > 10. Kepler-452b has the lowest SNR of all the long-period planets with claimed confirmations, and thus has the most tenuous claim to confirmation. How- ever, 99% reliability is a high threshold to reach at low SNR, and the confidence with which many of the other small, habitable-zone planets were claimed to be con- firmed is likely over-stated. We leave the analysis of these systems as future work. 3.1. Raising Reliability In this section we discuss some refinements to our analysis that increase the estimated reliability of some of the candidates in the Earth-like sample. However none of these refinements increase the reliability to the desired 99% level. 3.1.1. Improved Parameter Space Selection Following Thompson et al. (2017), if we restrict our analysis to TCEs around main-sequence stars (4000 K < Teff < 7000 K and log g > 4.0) we are looking at signals observed around photometrically quieter stars. The Robovetter performs much better for these stars, and the measured reliability increases to ≈ 50%. Fur- ther restricting the sample to exclude the period range with the highest false alarm rate (360 to 380 days, as shown in Figure 2) increases the reliability to only 54%. 6 Mullally et al. This is close to the ceiling for reliability that can be achieved by analyzing smaller parameter spaces. Choos- ing narrower slices of parameter space can produce small gains in effectiveness, but comparatively larger changes in the numbers of candidates. The measured reliability increases slowly (if at all) and becomes noisier, as the effects of small number statistics begin to dominate. 3.1.2. High score TCEs Thompson et al. (2017) include a score metric, which attempts to quantify the confidence the Robovetter places in its classification of a TCE as a planet can- didate or a false alarm. The score is not the probability that a TCE is a planet. High scores (> 0.8) indicate strong confidence that a TCE is a candidate, while low scores (< 0.2) indicate high confidence that a TCE is either a false positive or false alarm. An intermediate score indicates that the TCE was close to the thresh- old for failure in one or more of the Robovetter metrics. Kepler-452b has a score of 0.77. If we repeat our analysis on the small, long-period TCEs around FGK dwarves, but require a score (cid:62) 0.77 to consider a TCE a candidate, we find an effectiveness in simulated data of 99.97%, and a reliability of the cat- alog (between periods of 200-500 days and MES < 10) of 92% based on 9 planet candidates. While this reliabil- ity is much stronger than before, it is still some distance from the 99% reliability required to claim statistical con- firmation. The effectiveness estimate is also based on a small number of candidates and simulated TCEs passing the Robovetter. If the number of incorrectly identified simulated TCEs changes by just one, the measured reli- ability changes by 8%. Even if the measured reliability for high score TCEs was > 99%, the uncertainty in the measured reliability casts sufficient doubt to counter any claimed confirmation. 3.1.3. Multiple planet systems Lissauer et al. (2012) first noted that the presence of multiple planet candidates around a single target star was strong evidence that all candidates in that system were planets, and not eclipsing binaries. The argument is essentially that while planet detections and eclips- ing binary detections are both rare, multiple detectable planets in a single system are much more likely than an accidental geometric alignment between a planet host system and a background eclipsing binary. As recog- nized by Rowe et al. (2014), a similar argument does not automatically apply to false positives. Lissauer as- sumed that planetary systems and eclipsing binaries are distributed uniformly across all targets. Systematic sig- nals come from a variety of sources, some of which are concentrated in specific regions of the focal plane, or types of stars. Some targets are very much more likely to see multiple false alarm TCEs than a uniform distri- bution would predict. It may be possible to apply the Lissauer analysis once a richer model of the false alarm distribution is known. 3.1.4. Relying on earlier catalogs The reliability of a candidate may be higher if it ap- pears in multiple Kepler catalogs. The TPS algorithm underwent extensive development between the Q1-Q16 catalog of Mullally et al. (2015), the DR24 catalog of Coughlin et al. (2016) and the DR25 catalog (the three catalogs to include 4 years of Kepler data). The dif- ferent catalogs could, in principle, be used as quasi- independent detections at the low SNR limit. Unfortu- nately, due to the lack of diagnostic runs on the earlier catalogs (reliability is unmeasured for DR24, and neither reliability nor completeness for Q1-Q16), it is impossi- ble to quantify the improved reliability of a candidate in this manner. For the specific case of Kepler-452b, it was detected in both the DR24 and DR25 catalogs, but was not detected in Q1-Q16, even though all 4 transits were observed. 4. DISCUSSION & CONCLUSION We show statistical validation is insufficient to con- firm Kepler-452b at the 99% level, and that Kepler-452b should no longer be considered a confirmed planet. Al- though use of this threshold to claim confirmation is somewhat arbitrary, it has seen widespread adoption in the community. To our knowledge, there are no claims in the literature of the claimed confirmation of a tran- siting planet with confidence less than this threshold. Our simplest calculation argues that there is only a 16% chance that the detected signal on the target star is due to a transit, while our most optimistic calcula- tion sets the probability at 92%. Even this most op- timistic assessment is still an order of magnitude less confident than our threshold, and nearly 2 orders less than that claimed in the discovery paper (99.76%, Jenk- ins et al. 2015). While the exact choice of threshold to claim confirmation may be a matter of taste and conven- tion, setting the threshold low enough to admit Kepler- 452b would result in significant contamination of the confirmed planet catalog with false alarms, and negate the goals of creating a confirmed planet list. It may be possible to devise a new statistical approach that validates Kepler-452b. However, given the difficul- ties of creating simulated data that reproduces the time- varying, non-Gaussian noise properties of Kepler data, and the considerable development effort already invested in the Robovetter, we are doubtful such a method will be devised in the near future. Instead, we advocate that Kepler-452b 7 the transit events should be directly confirmed by inde- pendent observations, such as the Hubble follow-up of Kepler-62 by Burke (2017). The un-confirmation of Kepler-452b is interesting in its own right, given the similarity between its mea- sured parameters and those of the Earth, but our result has implications for other long-period confirmed Kepler planets. Torres et al. (2017) claimed to statistically con- firm 6 habitable-zone candidates originally detected at long-period and MES < 12. These planets were all de- tected at higher SNR than Kepler-452b, and are there- fore less likely to be caused by an instrumental artifact. However, the lack of precision for our reliability esti- mates in this work does raise a concern for these other systems. The burden of proof for claiming that a planet exists lies in demonstrating that the probability of all other scenarios is definitely less than the chosen thresh- old. If the likelihood of a transit signal is drawn from a distribution (i.e a posterior) that overlaps significantly with the threshold for acceptance, then that burden of proof is not met. For Kepler-452b, the uncertainty in our measurement of reliability is an order of magnitude higher that what is required for confirmation. The inability to confirm individual planet candidates by statistical techniques should not be considered a fail- ure of the Kepler mission (or of the statistical techniques themselves, which are applicable when their underlying assumptions remain valid). Kepler's goal was to estab- lish the frequency of Earth-like planets in the Galaxy, and the conclusion that at least 2% (and possibly as many as 25%) of stars host an Earth-size planet in the shorter-period range of 50–300 days (Burke et al. 2015) is not strongly affected by this result. The reliability of the Kepler sample in this regime is much higher, and sensitivity of occurrence rate calculations is less sensi- tive to catalog reliability. It will be necessary to account for catalog reliability to properly extend occurrence rate calculations out toward the longer periods that encom- pass the habitable zone around G type stars. We thank the referee, Tim Brown, for his construc- tive comments. This work is supported by NASA un- der grant number NNX16AJ19G. Some of the data pre- sented in this paper were obtained from the Mikulski Archive for Space Telescopes (MAST). STScI is oper- ated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS5-26555. Support for MAST for non-HST data is provided by the NASA Office of Space Science via grant NNX09AF08G and by other grants and contracts. This research has also made use of the NASA Exoplanet Archive, which is operated by the California Institute of Technology, un- der contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Pro- gram. Figure 1 shows Kepler-452b in the top panel, and the inverted TCE 11961208-01 (period 425 days, best fit radius of 2 R⊕) in the lower panel. Facilities: Kepler REFERENCES Batalha, N. M., Rowe, J. F., Bryson, S. T., et al. 2013, Lissauer, J. J., Marcy, G. W., Rowe, J. F., et al. 2012, ApJ, ApJS, 204, 24 750, 112 Brown, T. M. 2003, ApJL, 593, L125 Burke, C. 2017, "Completing Kepler's Mission to Determine the Frequency of Earth-like Planets", HST Proposal 15129 Mandushev, G., Torres, G., Latham, D. W., et al. 2005, ApJ, 621, 1061 Marcy, G. W., Isaacson, H., Howard, A. W., et al. 2014, ApJS, 210, 20 Burke, C. J., Bryson, S. T., Mullally, F., et al. 2014, ApJS, Morton, T. D., Bryson, S. T., Coughlin, J. L., et al. 2016, 210, 19 ApJ, 822, 86 Burke, C. J., Christiansen, J. L., Mullally, F., et al. 2015, Mullally, F., Coughlin, J. L., Thompson, S. E., et al. 2015, ApJ, 809, 8 ApJS, 217, 31 Christiansen, J. L. 2017, Kepler Science Document, KSCI-19110-001 Christiansen, J. L., Jenkins, J. M., Caldwell, D. A., et al. --. 2016, PASP, 128, 074502 Rogers, L. A. 2015, ApJ, 801, 41 Rowe, J. F., Bryson, S. T., Marcy, G. W., et al. 2014, ApJ, 2012, PASP, 124, 1279 784, 45 Coughlin, J. L. 2017, Kepler Science Document, Rowe, J. F., Coughlin, J. L., Antoci, V., et al. 2015, ApJS, KSCI-19114-002 217, 16 Coughlin, J. L., Mullally, F., Thompson, S. E., et al. 2016, Thompson, S. E., Coughlin, J. L., Hoffman, K., et al. 2017, ApJS, 224, 12 ArXiv:1710.06758 Jenkins, J. M., Twicken, J. D., Batalha, N. M., et al. 2015, Thompson, S. E., Mullally, F., Coughlin, J., et al. 2015, AJ, 150, 56 ApJ, 812, 46 Koch, D. G., Borucki, W. J., Basri, G., et al. 2010, ApJL, Torres, G., Fressin, F., Batalha, N. M., et al. 2011, ApJ, 713, L79 727, 24 8 Mullally et al. Torres, G., Kane, S. R., Rowe, J. F., et al. 2017, AJ, 154, Wolfgang, A., & Lopez, E. 2015, ApJ, 806, 183 264 Twicken, J. D., Jenkins, J. M., Seader, S. E., et al. 2016, AJ, 152, 158
1605.05315	1	1605	2016-05-17T19:59:57	Direct Imaging discovery of a second planet candidate around the possibly transiting planet host CVSO 30	[ "astro-ph.EP" ]	We surveyed the 25 Ori association for direct-imaging companions. This association has an age of only few million years. Among other targets, we observed CVSO 30, which has recently been identified as the first T Tauri star found to host a transiting planet candidate. We report on photometric and spectroscopic high-contrast observations with the Very Large Telescope, the Keck telescopes, and the Calar Alto observatory. They reveal a directly imaged planet candidate close to the young M3 star CVSO 30. The JHK-band photometry of the newly identified candidate is at better than 1 sigma consistent with late-type giants, early-T and early-M dwarfs, and free-floating planets. Other hypotheses such as galaxies can be excluded at more than 3.5 sigma. A lucky imaging z' photometric detection limit z'= 20.5 mag excludes early-M dwarfs and results in less than 10 MJup for CVSO 30 c if bound. We present spectroscopic observations of the wide companion that imply that the only remaining explanation for the object is that it is the first very young (< 10 Myr) L-T-type planet bound to a star, meaning that it appears bluer than expected as a result of a decreasing cloud opacity at low effective temperatures. Only a planetary spectral model is consistent with the spectroscopy, and we deduce a best-fit mass of 4-5 Jupiter masses (total range 0.6-10.2 Jupiter masses). This means that CVSO 30 is the first system in which both a close-in and a wide planet candidate are found to have a common host star. The orbits of the two possible planets could not be more different: they have orbital periods of 10.76 hours and about 27000 years. The two orbits may have formed during a mutual catastrophic event of planet-planet scattering.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. CVSO September 11, 2018 c(cid:13)ESO 2018 6 1 0 2 y a M 7 1 . ] P E h p - o r t s a [ 1 v 5 1 3 5 0 . 5 0 6 1 : v i X r a Direct Imaging discovery of a second planet candidate around the possibly transiting planet host CVSO 30 ⋆ T. O. B. Schmidt1, 2, R. Neuhäuser2, C. Briceño3, N. Vogt4, St. Raetz5, A. Seifahrt6, C. Ginski7, M. Mugrauer2, S. Buder2, 8, C. Adam2, P. Hauschildt1, S. Witte1, Ch. Helling9, and J. H. M. M. Schmitt1 1 Hamburger Sternwarte, Gojenbergsweg 112, 21029 Hamburg, Germany, e-mail: [email protected] 2 Astrophysikalisches Institut und Universitäts-Sternwarte, Universität Jena, Schillergässchen 2-3, 07745 Jena, Germany 3 Cerro Tololo Inter-American Observatory CTIO/AURA/NOAO, Colina El Pino s/n. Casilla 603, 1700000 La Serena, Chile 4 Instituto de Física y Astronomía, Universidad de Valparaíso, Avenida Gran Bretaña 1111, 2340000 Valparaíso, Chile 5 European Space Agency ESA, ESTEC, SRE-S, Keplerlaan 1, NL-2201 AZ Noordwijk, the Netherlands 6 Department of Astronomy and Astrophysics, University of Chicago, 5640 S. Ellis Ave., Chicago, IL 60637, USA 7 Sterrewacht Leiden, PO Box 9513, Niels Bohrweg 2, NL-2300RA Leiden, the Netherlands 8 Max-Planck-Institute for Astronomy, Königstuhl 17, 69117 Heidelberg, Germany 9 School of Physics and Astronomy SUPA, University of St. Andrews, North Haugh, St. Andrews, KY16 9SS, UK Received 2015; accepted ABSTRACT Context. Direct imaging has developed into a very successful technique for the detection of exoplanets in wide orbits, especially around young stars. Directly imaged planets can be both followed astrometrically on their orbits and observed spectroscopically and thus provide an essential tool for our understanding of the early solar system. Aims. We surveyed the 25 Ori association for direct-imaging companions. This association has an age of only few million years. Among other targets, we observed CVSO 30, which has recently been identified as the first T Tauri star found to host a transiting planet candidate. Methods.We report on photometric and spectroscopic high-contrast observations with the Very Large Telescope, the Keck telescopes, and the Calar Alto observatory. They reveal a directly imaged planet candidate close to the young M3 star CVSO 30. Results.The JHK-band photometry of the newly identified candidate is at better than 1 σ consistent with late-type giants, early-T and early-M dwarfs, and free-floating planets. Other hypotheses such as galaxies can be excluded at more than 3.5 σ. A lucky imaging z′ photometric detection limit z′= 20.5 mag excludes early-M dwarfs and results in less than 10 MJup for CVSO 30 c if bound. We present spectroscopic observations of the wide companion that imply that the only remaining explanation for the object is that it is the first very young (< 10 Myr) L -- T-type planet bound to a star, meaning that it appears bluer than expected as a result of a decreasing cloud opacity at low effective temperatures. Only a planetary spectral model is consistent with the spectroscopy, and we deduce a best-fit mass of 4 - 5 Jupiter masses (total range 0.6 -- 10.2 Jupiter masses). Conclusions. This means that CVSO 30 is the first system in which both a close-in and a wide planet candidate are found to have a common host star. The orbits of the two possible planets could not be more different: they have orbital periods of 10.76 hours and about 27000 years. The two orbits may have formed during a mutual catastrophic event of planet-planet scattering. Key words. stars: pre-main sequence, low-mass, planetary systems - planets: detection, atmospheres, formation 1. Introduction Since the first definite detection of a planet around another main- sequence star, 51 Peg (Mayor & Queloz 1995) made with high- precision radial velocity measurements, various detection tech- niques have been applied to find a diverse population of exo- planets. The transit method, which was first used for HD 209458 (Charbonneau et al. 2000), later allowed for a boost of exoplanet discoveries after the successful launch of two dedicated satellite missions, CoRoT (Baglin et al. 2007) and Kepler (Koch et al. 2010; Borucki et al. 2010). These two methods indirectly dis- cern the presence of a planet by the influence the planet has on its host star. The methods are most sensitive to small and moderate planet-star-separations around old main-sequence stars that are ⋆ Based on observations made with ESO Telescopes at the La Silla Paranal Observatory under programme IDs 090.C-0448(A), 290.C-5018(B), 092.C-0488(A) and at the Centro Astronómico Hispano-Alemán in programme H15-2.2-002. fairly inactive because of their age. The sensitivity diminishes fast for separations beyond 5 au because transits become less likely and the radial velocity amplitude declines with increas- ing orbital period. In contrast, direct imaging allows discovering planets on wide orbits around nearby pre-main-sequence stars because such young planets are still bright at infrared wave- lengths as a result of the gravitational contraction during their still-ongoing formation process. The total number of imaged planet candidates has increased to about 50-60 objects today. The first detections were made in 2005, when the first four co-moving planetary candidates were found around the solar-like stars DH Tau (Itoh et al. 2005), GQ Lup (Neuhäuser et al. 2005), and AB Pic (Chauvin et al. 2005c), all with masses near the threshold of 13 MJup that di- vides brown dwarfs from planets according to the current IAU working definition, together with the planet candidate around the brown dwarf 2M1207 (Chauvin et al. 2005a). A summary can be found in Neuhäuser & Schmidt (2012), and the current Article number, page 1 of 15 A&Aproofs: manuscript no. CVSO Fig. 1. Direct images of CVSO 30 c. Left: Keck image of data by van Eyken et al. (2012), re-reduced. We note that the companion is north-east and not a contaminant south-east, as given in van Eyken et al. (2012). Right: Our new VLT epoch, clearly showing the planetary companion, which has similar colour as its host star (Fig. 2). This excludes that it is a false positive for the inner planet candidate CVSO 30 b. status is always available in several online encyclopaediae, such as the Extrasolar Planets Encyclopaedia at www.exoplanet.eu (Schneider et al. 2011). As in situ formation at ∼100 au to a few hundreds of au separation seems unlikely according to models, Boss (2006) argued that a third body must exist that tossed these planets outward to their present distance from their young host stars. An alternative explanation might be a stellar encounter (Adams & Laughlin 2001). While early-type stars have less favourable planet-to-star contrast ratios, increasing evidence was found by millimeter- continuum measurements for larger and more massive proto- planetary disks that are available for planet formation around these stars (Mannings & Sargent 1997; Andrews et al. 2013). These conclusions were further strengthened when three of the most prominent planet candidates were found in 2008 and 2009 around the early F-type star HR 8799 (Marois et al. 2008), which is the first system with multiple planets imaged around a star. The other two were around the A-type stars Fomalhaut (Kalas et al. 2008), which is the first planet candidate discovered in the optical regime using the Hubble Space Telescope (HST), and β Pic (Lagrange et al. 2009, 2010), a planet within the large edge-on disk at only about twice the separation of Jupiter from the Sun. This had previously been predicted by Freistetter et al. (2007), for instance, from the structural gaps in the disk. Most of the direct-imaging surveys conducted so far have concentrated on AFGK stars. In 2012 a (proto-)planet candidate was discovered around the ∼2 Myr young solar-like star LkCa 15 (Kraus & Ireland 2012). This is a close (∼11 AU) object found by single-dish interferometry, which is a technique that is also referred to as sparse aperture masking. Two companions of 4-5 MJup were recently discovered around GJ 504 (Kuzuhara et al. 2013), which is a 160 Myr old solar-like star, and around HD 95086 (Rameau et al. 2013), an A-type star at about 10-17 Myr. Additionally, first results from imaging surveys around M dwarfs were published during the past two years, which increased our understanding of planetary systems around the most numerous stars in the Milky Way (Delorme et al. 2013; Bowler et al. 2015). In this article we describe for the first time the direct de- tection of a wide-separation (1.85′′ or 662 au, see Fig. 1) di- rectly imaged planet candidate around a star (CVSO 30) that also hosts a short-period transiting planet candidate; we refer to a more detailed discussion of this object in van Eyken et al. (2012), Barnes et al. (2013), and Yu et al. (2015). A system that harbours two planets with such extreme orbits gives us for the first time the opportunity to study by observations the possible Article number, page 2 of 15 Table 1. Previously known CVSO 30 system data Altern. designations Location RA, Dec Spectral type Mass Luminosity Radius Temperature Opt. extinction Distance Age Hα equivalent width LiI equivalent width v sin(i)∗ Proper motion [E,N] B, V, R photometry J, H, K photometry (Projected) separation Period (circular) Orbit. inclination Orbit. misalignment 2MASS J05250755+0134243, PTF1 J052507.55+013424.3 CVSO 30 25 Ori / Orion OB 1a [1,2] 05h 25m 07.57s, +01◦ 34′ 24.5′′ [2] M3 (weak-line T-Tauri, WTTS) [2] 0.34 / 0.44 M⊙ [2] 0.25 L⊙ [2] 1.39 R⊙ / 1.07 ± 0.10 R⊙ / [1.03 / 1.04 ± 0.01 R⊙] [2,3,4] [323+233 −96 , 322+504 2.39+3.41 3470 K [2] 0.12 mag [2] −122] pc / 357 ± 52 pc [2,5] −2.05 Myr [2,here] -11.40 Å [2] 0.40 Å [2] 80.6 ± 8.1 km s−1 [3] [-0.1 ± 5.3, 0.9 ± 5.5] mas/yr [6] [18.35, 16.26, 15.19] mag [7,2,3] [12.232 ± 0.028, 11.559 ± 0.026, 11.357 ± 0.021] mag [8] CVSO 30 b / PTFO 8-8695 b 0.00838 ± 0.00072 au [3] 0.448413 ± 0.000040 d [3] 61.8 ± 3.7 ◦ [3] 69 ± 2 ◦ / 73.1 ± 0.5 ◦ [4] References: [1] Briceño et al. (2007a), [2] Briceño et al. (2005), [3] van Eyken et al. (2012), [4] Barnes et al. (2013), [5] Downes et al. (2014) [6] Zacharias et al. (2013), [7] Zacharias et al. (2004), [8] Cutri et al. (2003); Skrutskie et al. (2006) Table 2. CVSO 30 astrometry and photometry CVSO 30 b / PTFO 8-8695 b Separation with respect to the host star [E,N] CVSO 30 c 2010 September 25 2012 December 3 (Projected) separation Period (circular) Orbit. inclination Orbit. misalignment z′ band (differential) J band (differential) H band (differential) Ks band (differential) J band (differential) 0.00838 ± 0.00072 au [1] 0.448413 ± 0.000040 d [1] 61.8 ± 3.7 ◦ [1] 69 ± 2 ◦ / 73.1 ± 0.5 ◦ [2] [175.453, 63.395] pixel [1.746 ± 0.006, 0.621 ± 0.010] ′′ 662 ± 96 au ∼ 27250 years > 6.8 mag 7.385 ± 0.045 mag 7.243 ± 0.014 mag 7.351 ± 0.022 mag 7.183 ± 0.035 mag References: [1] van Eyken et al. (2012), [2] Barnes et al. (2013) outcome of planet-planet scattering theories, which have been used to explain the existence of close-in hot Jupiters in 1996 (Rasio & Ford 1996). T. O. B. Schmidt et al.: Direct Imaging of a second planet candidate in the transiting CVSO 30 system. Table 3. VLT/NACO, VLT/SINFONI, archival KeckII/NIRC2, and Calar Alto/2.2m/AstraLux observation log Instrument JD-2455000 [days] NACO J 1264.69416 NACO H 1264.70764 NACO Ks 1264.72079 NACO J 1266.72899 SINFONI H+K 1592.82609 465.05374 NIRC2 H AstraLux z′ 2260.6696 Date of observation 03 Dec 2012 03 Dec 2012 03 Dec 2012 05 Dec 2012 27 Oct 2013 25 Sep 2010 26 Aug 2015 DIT [s] 15 15 15 30 300 3 0.02945 NDIT # images Airmass DIMMa Seeing 4 4 4 2 2 10 1 15 15 15 15 3 12 70000 1.13 1.12 1.11 1.12 1.12 1.25 1.73 0.8 0.6 0.7 1.3 0.5 0.4 1.1 τb 0 [ms] 3.7 4.6 4.6 2.8 5.0 Strehl [%] 3.2 11.2 23.7 2.0 7.0 no AO S/N (brightest pixel) 5.9 24.6 11.1 6.6 /15 7.8 non-detection Remarks: (a) Differential image motion monitor (DIMM) seeing average of all images (b) coherence time of atmospheric fluctuations. Table 4. CVSO 30 deduced planetary properties Opt. extinction Luminosity (vs. ⊙) Eff. temperature Teff Surface gravity log g Radius Mass CVSO 30 b / PTFO 8-8695 b 1.91 ± 0.21 RJup [1] 1.64 / 1.68 ± 0.07 RJup [2] < 5.5 ± 1.4 MJup [1] 3.0 ± 0.2 MJup [2] 3.6 ± 0.3 MJup [2] CVSO 30 c 0.19+2.51 -3.78+0.33 1600+120 3.6+1.4 −0.19 mag −0.13 dex −300 K −0.6 dex 1.63+0.87 −0.34 RJup 4.3+4.9 −2.0 MJup (L, age) −3.7 MJup (log g & Roche) 4.7+5.5 4.7+3.6 −2.0 MJup (L, Teff, age) < 10 MJup (z′ imaging limit) References: [1] van Eyken et al. (2012), [2] Barnes et al. (2013) 2. 25 Ori group and the CVSO 30 system properties Despite their importance for the evolution of protoplanetary disks and the early phases in the planet formation process, suf- ficiently large samples of 10 Myr old stars have been difficult to identify, mainly because the parent molecular clouds dissipate after a few Myr and no longer serve as markers of these popula- tions (see Briceño et al. (2007b) and references therein). The 25 Ori cluster (25 Ori, Briceño et al. 2007a) contains > 200 PMS stars in the mass range 0.1 < M/M⊙ < 3. The Hipparcos OB and earlier A-type stars in 25 Ori are on the zero-age main se- quence (ZAMS, Hernández et al. 2005), implying a distance of ∼330 pc, with some of the A-type stars harbouring debris disks (Hernández et al. 2006). Isochrone fitting of the low-mass stars yields an age of 7-10 Myr (Briceño et al. 2007b). This is the most populous 10 Myr old sample within 500 pc, which we conse- quently chose for a direct-imaging survey with ESO's VLT, the Very Large Telescope of the European Southern Observatory, to find young planetary and sub-stellar companions at or shortly af- ter their formation. For this same reason, the 25 Ori cluster was also targeted in searches for transiting planets, like the Young Exoplanet Transit Initiative (YETI, Neuhäuser et al. 2011) and the Palomar Transient Factory (PTF, van Eyken et al. 2012). CVSO 30 (also 2MASS J05250755+0134243 and PTFO 8- 8695) is a weak-line T Tauri star of spectral type M3 in 25 Ori at an average distance of 357 ± 52 pc (Downes et al. 2014). It was confirmed as a T Tauri member of the 25 Ori cluster by the CIDA Variability Survey of Orion (CVSO), with properties shown in Table 1. As shown in Fig. 1 in van Eyken et al. (2012), CVSO 30 is one of the youngest objects within 25 Ori, its position in the colour-magnitude diagram corresponds to 2.39+3.41 −2.05 Myr (if compared to evolutionary models of Siess et al. (2000)). The ob- ject is highly variable, rotates fast, and has a mass of 0.34 -- 0.44 M⊙ (depending on the evolutionary model), and an effective tem- perature of ∼3470 K. The rotation period of CVSO 30, which is possibly synchronised with the CVSO 30 b orbital period, is still debated (van Eyken et al. 2012; Koen 2015). Kamiaka et al. (2015) concluded that the stellar spin period is shorter than 0.671 d. In 2012 the PTF team (van Eyken et al. 2012) reported a young transiting planet candidate around CVSO 30, named PTFO 8-8695 b, with a fast co-rotating or near co-rotating 0.448413 day orbit. The very same object, henceforth CVSO 30 b for simplicity, was independently detected with smaller tele- scopes within the YETI (Neuhäuser et al. 2013; Errmann et al. 2014), confirming the presence of the transit events by quasi- simultaneous observations. Keck and Hobby-Eberly Telescope (HET) spectra (van Eyken et al. 2012) set an upper limit to the mass of the transiting companion of 5.5 ± 1.4 MJup from the radial velocity variation, which exhibits a phase offset that is likely caused by spots on the surface of the star. This RV limit was already corrected for the derived orbital inclination 61.8 ± 3.7 ◦ of the system. With an orbital radius of only about twice the stellar radius and a planetary radius of 1.91 ± 0.21 RJup, the object appears to be at or within its Roche-limiting orbit, raising the possibility of past or ongoing mass loss. A false positive by a blended eclipsing binary is unlikely because the only present contaminant in Keck near-IR images (see Fig. 1) with 6.96 mag of contrast to the star would have to be very blue to be bright enough in the optical to mimic a transit; it is unlikely to be a star in that case. In 2013 Barnes et al. (2013) fit the two separate light curves observed in 2009 and 2010, which exhibited unusual differ- ing shapes, simultaneously and self-consistently with planetary masses of the companion of 3.0 -- 3.6 MJup. They assumed tran- sits across an oblate, gravity-darkened stellar disk and preces- sion of the planetary orbit ascending node. The fits show a high degree of spin-orbit misalignment of about 70◦, which leads to the prediction that transits disappear for months at a time dur- ing the precession period of this system. The lower planet ra- dius result of ∼1.65 RJup is consistent with a young, hydrogen- dominated planet that results from hot-start formation mecha- nisms (Barnes et al. 2013). 3. Astrometric and photometric analysis After the discovery of the transiting planet candidate by van Eyken et al. (2012) and our independent detection of the transit signals with YETI, we included the system in our 25 Ori VLT/NACO direct-imaging survey with the intent to prove that the object labelled as a contaminant by van Eyken et al. (2012) Article number, page 3 of 15 A&Aproofs: manuscript no. CVSO Quasars S Ori 64 K7III M7III Galaxies M9III M9III HD 204585/M4.5III 0.4 0.4 H−Ks [mag] H−Ks [mag] 0.6 0.6 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1.0 1.0 ] ] g g a a m m [ [ H H − − J J A−K dwarfs White dwarfs PZ Tel B HD 35367 cc3 M dwarfs CVSO 30 L dwarfs CVSO 30 c SDSS 0909 T dwarfs 1RXS 1609 b 0.2 0.2 0.4 0.4 0.6 0.6 0.8 0.8 1.0 1.0 ] ] g g a a m m [ [ H H − − J J F8III G6III K0III CVSO 30 CVSO 30 c HD 175865/ M5III 0.0 0.0 0.2 0.2 0.4 0.4 H−Ks [mag] H−Ks [mag] 0.6 0.6 0.0 0.0 0.2 0.2 Fig. 2. CVSO 30, CVSO 30 c, and comparison objects, superimposed onto the colour data from Hewett et al. (2006). CVSO 30 c clearly stands out in the lower left corner, approximately consistent with colours of giants, early-M and early-T dwarfs, and free-floating planetary mass objects (Zapatero Osorio et al. 2000; Peña Ramírez et al. 2012), e.g. consistent with absolute magnitude and J-Ks colour of S Ori 64. Its unusual blue colour can most likely be attributed to the youth of these objects (Saumon & Marley 2008), leading to L -- T transition opacity drop at high bright- nesses (see Fig. 11). See Fig. A.5 for details. For CVSO 30 c we give the colours before (grey) and after (red) correcting for the NACO to the 2MASS filter set and we give the maximum possible systematic photometric offsets caused by the variability of the primary star that is used as reference (black). Table 5. Astrometric calibration of VLT/NACO Object 47 Tuc JD - 2456000 [days] 264.62525 Pixel scale [mas/pixel] PAa [deg] 13.265 ± 0.041 +0.60 ± 0.31 Remarks: All data from Ks-band images. (a) PA is measured from N over E to S. is not able to produce the detected transiting signal and to con- firm it as second planet. We performed our first high-resolution direct observations in December 2012 and obtained JHK-band photometry (Tables 2 and 3 and Fig. 1). During the course of their study of the transiting planet CVSO 30 b, the PTF team used Keck II/NIRC2 H-band images obtained in 2010 to identify contaminants capable of creating a false-positive signal mimicking a planet. We re-reduced these data and found that they already contain the planetary compan- ion CVSO 30 c that we report here. In Fig. 1 we show the com- panion, erroneously given to lie south-east in van Eyken et al. (2012); it is located north-east of the host star CVSO 30. We astrometrically calibrated the VLT/NACO detector epoch using a sub-field of 47 Tuc (Table 5) to determine pixel scale and detector orientation in order to find precise values for the sepa- ration of CVSO 30 c with respect to CVSO 30 in right ascension and declination. From this, we find the object to be ∼1.85′′ NE of CVSO 30 at a position angle of ∼70◦ from north towards east, corresponding to a projected separation of 662 ± 96 au at the distance of the star. No astrometric calibrator could be found in the night of the Keck observations (hence the position of the ob- ject is given in pixels in Table 2), but we note that the position is consistent with the VLT data. We used the nominal pixel scale of Article number, page 4 of 15 NIRC2 of 0.009942 ′′/pixel (± 0.00005′′) and assumed 0◦ detec- tor orientation for the Keck epoch, which results in 1.744 arcsec right ascension and 0.630 arcsec declination separation in the relative position of CVSO 30 c with respect to its host star. The consistency was expected for a companion because the proper motion of CVSO 30 is too small to distinguish a background source from a sub-stellar companion based on common proper motion (Table 1). CVSO 30 is in general currently not suitable for a com- mon proper motion analysis because the errors in proper mo- tion exceed the proper motion values (Table 1). As orbital mo- tion around the host star might be detectable, we performed a dedicated orbit estimation for the wide companion. The analy- sis shows that even after two to three years of epoch difference, no significant orbital motion is expected for the wide companion (Fig. A.1). Using the Two Micron All Sky Survey (2MASS) (Cutri et al. 2003; Skrutskie et al. 2006) photometry for the primary and our NACO images for differential brightness measurements, we find CVSO 30 c to exhibit an unusually blue H-Ks colour, while its J-H colour indicates the companion candidate to be redder than the primary. This implies that the companion is too red to be an eclipsing background binary mimicking the transiting signal of CVSO 30 as a false-positive signal, which is further indication for the planetary nature of CVSO 30 b. The differential photometry (Table 2) of CVSO 30 c was achieved using psf fitting with the Starfinder package of IDL (Diolaiti et al. 2000) using the primary star CVSO 30 as psf ref- erence. First the noise of the final jittered image was computed, taking the photon noise, the gain, RON, and the number of com- bined images into account, and then it was transferred to the T. O. B. Schmidt et al.: Direct Imaging of a second planet candidate in the transiting CVSO 30 system. Table 6. Photometric rejection significance, spectroscopic reduced χ2 results, and corresponding formal significance without systematics for different comparison objects. Object SpT HD 237903 Gl 846 Gl 229 Gl 806 Gl 388 Gl 213 Gl 51 Gl 406 Gl 644C Gl 752B LHS 2065 LHS 2924 2MUCD 20581 Kelu-1AB 2MUCD 11291 2MUCD 12128 2MUCD 11296 2MUCD 11314 2MUCD 10721 2MUCD 10158 SDSS 1520+354 SDSS 0909+652 SDSS 1254-012 2MASS 055-140 HD 204585 HD 175865 BK Vir HY Aqr Galaxies Quasars White Dwarfs CVSO 30 PZ Tel B CT Cha b 2M0441 Bb 1RXS 1609 b β Pic b 2M1207 b S Ori 64 DP (Fig. 4) K7V M0V M1V M2V M3V M4V M5V M6V M7V M8V M9V L0 L1 L2+L3.5 L3 L4.5 L5.5 L6 L7.5 L8.5 T0 T1.5 T2 T4 M4.5III M5III M7III M8-9III various -- various M3 M7 M9 L1 L4 L4 L7 L/T -- Photometry H-Ks J-H [σ] [σ] 0.5 3.4 2.8 1.8 0.3 0.6 2.5 4.5 2.8 3.7 2.6 5.5 3.5 3.8 3.4 4.6 5.1 4.1 6.4 2.6 7.5 1.7 6.5 1.4 7.5 2.2 9.8 2.2 1.8 >10 >12 5.5 >10 1.3 8.4 2.0 >11 5.8 9.8 2.5 1.0 5.4 0.4 0.3 2.0 0.8 0.1 9.4 0.8 0.4 0.5 0.1 0.4 0.9 8.5 0.6 3.0 4.2 3.9 4.4 3.9 6.4 1.7 2.7 2.1 1.4 1.3 0.4 2.4 0.5 3.0 1.1 3.5 1.0 4.4 3.5 0.7 0.9 -- -- add. ref. [1] [1] [1] [1] [2],[1] [2],[1] [2],[1] [2],[1] [2],[1] [2],[1] [1] [2],[1] [2] [2] [2] [2] [2] [2] [2] [2] [3] [4] [2] [2] [1] [1] [1] [1] [5],[6] [5] [5] [7] [8],[9] [10] [11] [12] [13] [14] [15] Spectroscopy H-band [σ / χ2 r ] >6 / 2.66 >6 / 2.38 >6 / 2.37 >6 / 2.73 >6 / 2.57 >6 / 2.80 >6 / 2.47 >6 / 2.50 >6 / 2.87 >6 / 2.76 >6 / 2.45 >6 / 2.77 >6 / 3.96 >6 / 3.68 >6 / 3.66 >6 / 3.09 >6 / 4.60 >6 / 3.64 >6 / 3.49 >6 / 4.87 >6 / 4.63 >6 / 8.04 >6 / 7.97 >6 / 16.2 >6 / 1.86 >6 / 1.91 >6 / 2.47 >6 / 5.78 >6 / 2.28 -- -- >6 / 3.31 >6 / 3.29 >6 / 2.27 >6 / 3.13 >6 / 2.70 >6 / 2.06 >6 / 2.66 -- K-band [σ / χ2 r ] >6 / 1.60 5.4 / 1.51 5.3 / 1.50 4.3 / 1.40 3.7 / 1.33 2.5 / 1.21 2.6 / 1.21 2.5 / 1.20 2.2 / 1.17 2.3 / 1.18 2.2 / 1.17 2.1 / 1.16 3.7 / 1.33 3.6 / 1.32 3.8 / 1.34 3.4 / 1.29 5.5 / 1.52 >6 / 1.66 3.4 / 1.29 5.0 / 1.47 >6 / 2.15 >6 / 3.64 >6 / 2.90 >6 / 19.1 >6 / 1.88 >6 / 1.78 >6 / 1.72 >6 / 1.61 >6 / 1.61 -- -- 6.0 / 1.57 3.0 / 1.25 1.8 / 1.13 1.9 / 1.13 2.3 / 1.18 -- 2.5 / 1.20 -- 2.2 / 1.16 2.0 / 1.14 References: [1] Rayner et al. (2009), [2] Cushing et al. (2005), [3] Burgasser et al. (2010a), [4] Chiu et al. (2006), [5] Hewett et al. (2006), [6] Mannucci et al. (2001), [7] Schmidt et al. (2014), [8] Schmidt et al. (2009), [9] Schmidt et al. (2008), [10] Bowler & Hillenbrand (2015), [11] Lafrenière et al. (2008), [12] Chilcote et al. (2015), [13] Patience et al. (2010), [14] Peña Ramírez et al. (2012), from VISTA to 2MASS magnitudes using colour equations from http://casu.ast.cam.ac.uk/surveys-projects/vista/technical/photometric-properties, [15] Helling et al. (2008) starfinder routine for psf fitting. This resulted in the values given in Table 2. The values were checked with aperture photometry. As given in van Eyken et al. (2012), our psf reference CVSO 30 varies by 0.17 mag (min to max) in the R band, consistent with our estimates within YETI. The present steep wavelength dependence of the variability amplitudes is best described by hot star-spots (Koen 2015), therefore we can extrapolate from measurements of the very similar T Tauri GQ Lup (Broeg et al. 2007). According to this, 0.17 mag in R correspond to about 0.1 mag and 0.055 mag variability in J and Ks band, respectively. As the hot spots change the bands simultaneously, this gives rise to a maximum systematic offset of 0.045 mag in J-Ks colour. We give an estimate of this variability as black error bars for a possible additional systematic offset of CVSO 30 c in Fig. 2. The colours of CVSO 30 and CVSO 30 c are very similar (Table 2 and Fig. 2). We currently lack a spectrum of CVSO 30 c in J band, therefore we used the M3V star Gl 388 (Cushing et al. 2005; Rayner et al. 2009) and the L3/L4 brown dwarf 2MASS J11463449+2230527 (Cushing et al. 2005) to derive a prelimi- nary filter correction between 2MASS and NACO for CVSO 30 Fig. 3. Median in wavelength direction of the reduced VLT/SINFONI integral field cubes. Left: Cube after reduction. Centre: Cube after re- moving the primary halo, assumed to be centred at the separation of 1.85", as measured in the VLT/NACO images. North is about 70◦ from the right-hand side towards the bottom of the plots. Right: Cube after removing the primary halo, spectral deconvolution, and polynomial flat- tening of the resulting background, used for the extraction of the final spectrum. and CVSO 30 c. The colours of CVSO 30 are well known from 2MASS (Table 1), the differential brightnesses to CVSO 30 c vary from NACO to 2MASS by 28 mmag in J, -21 mmag in H, and -38 mmag in Ks. Thus CVSO 30 c is 49 mmag redder in J-H and 17 mmag redder in H-Ks in 2MASS (red in Fig. 2) than in the NACO results (grey in Fig. 2). In Fig. 2 and Table 6 we compare CVSO 30 c to the colours of several possible sources. We find that background stars of spectral types OBAFGK are too blue in J-H, late-M dwarfs are too blue in J-H and too red in H-K, while foreground L- and late T-dwarfs are either too red in H-K or too blue in J-H. In addition, background galaxies, quasars, and H/He white dwarfs are also inconsistent with the values of CVSO 30 c. Only late-type giants, early-M and early-T dwarfs, and planetary mass free-floating ob- jects such as are found in the σ Orionis star cluster have compa- rable colours (Zapatero Osorio et al. 2000; Peña Ramírez et al. 2012). 4. CVSO 30 c spectroscopic analysis A common proper motion analysis is not feasible because of the low proper motion of the host star (Table 1), therefore we car- ried out spectroscopic follow-up observations at the end of 2013, using the ESO VLT integral field unit SINFONI. The observa- tions were made in H+K band with 100 mas/spaxel scale (FoV: 3 arcsec x 3 arcsec). The instrument provides information in the two spatial directions of the sky in addition to the simultaneous H- and K-band spectra. An unfortunate timing of the observa- tions led to a parallactic angle at which a spike, probably of the telescope secondary mounting, was superimposed onto the well- separated spectrum of the companion candidate CVSO 30 c. After correcting, the resulting spectrum can be compared to model atmospheres to determine its basic properties and to other sub-stellar companions to assess its youth and the reliabil- ity of the models at this low age, surface gravity, and temperature regime. In an attempt to optimally subtract the spike of the host star, we performed several standard and customised reduction steps. After dark subtraction, flat-fielding, wavelength calibration, and cube reconstruction, we found that the spike was superimposed onto the companion in every one of the three individual ex- posures, but at slightly different orientation angles (Fig. 3, left panel). As a first step, we used the NACO astrometry to deter- mine the central position of the primary, which is itself outside the observed field of view of the integral field observations. The orientation of the SINFONI observations was intentionally cho- sen to leave the connection line of primary and companion ex- Article number, page 5 of 15 A&Aproofs: manuscript no. CVSO 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.6 0.3 0.0 ) x u l f d e z i l a m r o n ( λ F s l a u d i s e R 22 1.6 1.8 2.0 Wavelength [µm] 2.2 2.4 Fig. 4. Spectrum of CVSO 30 c as extracted from the spectral deconvolution-corrected cube in the right panel of Fig. 3. Top: The spectrum in resolution 700 (black) is shown after binning of the original extracted spectrum in resolution 1500 (green). The best-fitting Drift-Phoenix model of Helling et al. (2008) is shown in red, fitting both the individually normalised H and K spectra. This type of normalisation was necessary because the redder colour of the models, in comparison to the unusually blue nature of CVSO 30 c, would steer the best-fitting model to higher temperatures, which would prevent fitting the individual features in H and K band. The best-fitting model (red) corresponds to 1600 K, surface gravity log g 3.6 dex, metallicity [M/H] 0.3 dex, and 0.19 mag of visual extinction. Bottom: Absolute value of the difference between spectrum and model from the top panel (black) versus noise floor at the corresponding position (green). actly in x direction. The primary is about 1.85 arcsec exactly to the left of CVSO 30 c in the data because the x direction offers a twice as good sampling regarding the number of pixels for the separation. We were thus able to subtract the radial symmetric halo of the host star from the data cube (Fig. 3, central panel) us- ing the nominal spatial scale. This is necessary because the halo of the primary star is determined by the AO performance at the different wavelength. At this stage, we extracted a first spectrum by subtracting an average spectrum of the spike, left and right of the companion psf from the superposition of companion and spike. We find the results in Fig. A.2 before (red spectrum) and after (blue spectrum) spike subtraction, which also removes the still-present OH lines. The horizontal spike in Fig. 3 appears too narrow to the right. This is a projection effect because the rota- tion of the spike within the three median-combined cubes leads to less overlap on the right-hand side of the cube than on the left-hand side. For this reason, the continuum in Fig. A.2 is not trustworthy because the flux of the spike below the companion candidate is not the average of the spike flux to the left and right of the object. We tried several methods to remove the spike and decided to follow the spectral deconvolution technique (Sparks & Ford 2002; Thatte et al. 2007). This method is able to distinguish both the wavelength-dependent airy rings and speckles and the spike from the light of the wavelength-independent companion posi- tion by using the long wavelength coverage of the observed data cube. As given in Thatte et al. (2007) for the same instrument, the bifurcation radius for SINFONI H+K is for ǫ=1.1 r=246 mas, and for ǫ=1.2 r=268 mas, which means that parts of the data without contamination of the companion could be found at the much higher separation of about 1.85 arcsec. The reduction was then completed by applying a polynomial background correc- tion around CVSO 30 c because the previous reduction steps left a low spatial frequency remnant around it (Fig. 3, right panel). Finally, the optimal extraction algorithm (Horne 1986) was per- formed around the companion and subtracted by the correspond- ing background flux from the close, well-corrected vicinity, and the telluric atmosphere correction using HD 61957, a B3V spec- troscopic standard observed in the same night. Article number, page 6 of 15 We first compare the spectrum of CVSO 30 c to spectra derived from Drift-Phoenix atmosphere simulations. These are dedicated radiative transfer models that take the strong contin- uum altering influence of dust cloud formation in the detectable parts of planetary atmospheres into account (Helling et al. 2008). From a χ2 comparison of the H- and K-band spectra to the model grid, we find an effective temperature of about 1800 - 1900 K, while the individual fit of the H- and K-band spectrum give a lower Te f f of about 1600 K. In addition, the slope of the blue part of the triangular H band is too steep in the atmosphere mod- els of about 1800 K and does not fit the continuum well. The higher Te f f is only needed to fit the unusually blue H-Ks colour of the object, as already discussed in the previous photometry section and visible in Fig. 2, since the models do yet not include a good description of the dust opacity drop at the L-T transition. We thus decided to fit the H and K band simultaneously, but normalising them individually, to cope with the unusual colours, while using all the present information for the fit. In this way, we −300 K, an extinction AV = 0.19+2.51 find a best-fitting Teff= 1600+120 −0.19 mag, a surface gravity log g [cm/s2]= 3.6+1.4 −0.6 dex, and a metallic- ity log[(M/H)/(M/H)⊙]= 0.3−0.9 dex at the upper supersolar edge of the grid. The 1 σ fitting contours are shown in Fig. 5, where they delineate the full regime for the error bars, and the best fit itself is shown in Fig. 4. In Fig. 6 we compare the spectrum of CVSO 30 c to the tri- angular shaped H-band spectrum of the β Pic b planet that was obtained with the Gemini Planet Imager (GPI, Chilcote et al. 2015). We also compare this to other planetary mass objects. β Pic b is particularly suited as comparison object because it is young (10 -- 20 Myr) and has about the same luminosity and ef- fective temperature (1600 -- 1700 K) while being of higher mass (10 -- 12 MJup). We show linear fits to the blue and red part of the H band and the triangular shape of the chosen Drift-Phoenix models. In contrast to M5 -- L5 companions, for which the H2O index in Allers et al. (2007) shows an increase in water absorp- tion, the absorption becomes shallower for later spectral types. This means that even though the formal χ2 fit finds a best tem- perature of 1600 K for CVSO 30 c, the temperature is likely to be lower than for β Pic b, exhibiting a steeper H-band spec- trum. The object's spectrum is not consistent with a giant of any T. O. B. Schmidt et al.: Direct Imaging of a second planet candidate in the transiting CVSO 30 system. HD 35367 cc3 M3 Galaxy CVSO 30 A A SDSS 0909 Drift−Phoenix 1RXS 1609 b T1.5 2M1207 b Beta Pic b CT Cha b CVSO 30 c 1800 K 1700 K 1650 K 1550 K 1.50 1.55 1.60 1.65 1.70 1.75 1.80 wavelength [µm] 1.5 1.6 1.7 wavelength [µm] 3 2 1 0 B C 1.8 D CT Cha b E CVSO 30 c HD 204585 / HD 175865 M4.5III / M5III 2M 0441 Bb Gl 229 M1V Gl 213 M4V Gl 644C M7V LHS 2924 L0 1.50 1.55 1.60 1.65 1.70 1.75 wavelength [µm] 2.0 2.1 2.2 2.3 wavelength [µm] 2.4 t e s f f o + λ F x u l f d e z i l a m r o n Fig. 6. H-band spectrum of CVSO 30 c (lower left) compared to several known planetary candidates and background objects (subplots A, D, E). The triangular shape of the H band (A), with red linear fits guiding the eye, indicates that it is not a background galaxy, but a sub-stellar companion. Beta Pic b has approximately the same luminosity and temperature (Chilcote et al. 2015) but a different surface gravity, hence about twice the mass of CVSO 30 c. As shown (C), the Drift-Phoenix models indicate that the H band becomes less steep with temperature. This means that CVSO 30 c is even slightly lower in temperature than β Pic b. In the upper left panel another candidate is shown, detected at 4.3" from the A1 star HD 35367, which is about 0.5 mag brighter in the K band than CVSO 30 c, but is obviously located in the background. In addition, the H band (D) and K band (E) of CT Cha b and 2M 0441 Bb, the best-fitting comparison objects, are given in K band. These two and CVSO 30 c are given in (D, E) with identical offsets in H band and K band. Additionally, the best-fitting giants and a sample of late-type dwarfs is shown for comparison. References and individual reduced χ2 r comparison values are given in Table 6. Low-resolution spectra of free-floating planetary candidates are not shown, but can be found in Martín et al. (2001). spectral type. The best-fitting giants with consistent photome- try (Fig. 2 and Table 6) are shown as comparison in Fig. 6 and would be at a distance of about 200 Mpc. To improve the fit in the K band, the spectral type would have to be later than M7III, while the H band does not fit for these objects. Finally, CVSO 30 c, which is comparable but younger, must have a lower surface gravity than β Pic b, which is determined to have a 1σ upper limit of log g [cm/s2]= 4.3 dex according to the linear prior orbit fit in Bonnefoy et al. (2014b). This corrects the surface gravity of CVSO 30 c to log g [cm/s2]= 3.6+0.7 −0.6 dex. 5. AstraLux lucky imaging follow-up observations We performed a follow-up of CVSO 30 with 2000 s of AstraLux integration in z′. The individual AstraLux images were com- bined using our own pipeline for the reduction of lucky imaging data. The fully reduced AstraLux image is shown in Fig. 7. z′ photometry of CVSO 30 was not measured so far, but can be de- rived from its magnitudes in other photometric bands using the colour transformation equations1 from Jordi et al. (2006). The V- and R-band photometry of CVSO 30, as given by Briceño et al. (2005) and van Eyken et al. (2012) (V = 16.26 ± 0.19 mag, and R = 15.19 ± 0.085 mag), and the I-band photometry of the star, listed in the 2005 DENIS database (I = 13.695 ± 0.030 mag), yield z′ = 13.66 mag. The (S/N= 5) detection limit reached in the AstraLux ob- servation is given in Fig. 8. At an angular separation of about 1.8 arcsec from CVSO 30 (or ∼640 au of projected separation), companions that are ∆z′= 6.8 mag fainter than the star are still 1 r − R = 0.77 · (V − R) − 0.37 and r − z′ = 1.584 · (R − I) − 0.386 detectable at S/N = 5. The reached detection limit at this angular separation is z′= 20.5 mag, which is just a tenth of magnitude above the limiting magnitude in the background-noise-limited region around the star at angular separations larger than 2 arcsec. This results in a limiting absolute magnitude of Mz′ = 12.7 mag, allowing the detection of sub-stellar companions of the star with masses down to 10 MJup according to the evolutionary models of Baraffe et al. (2015). Furthermore, the AstraLux observations also exclude all young (3 Myr) stellar objects (mass higher than 75 MJup) that are unrelated to CVSO 30, which are located in the AstraLux field of view at distances closer than about 3410 pc. All young M dwarfs with an age of 3 Myr and masses above 15 MJup (Teff > 2400 K) can be ruled out up to 530 pc. All old stellar objects (mass higher than 75 MJup) with an age of 5 Gyr can be excluded when they are located closer than about 130 pc. The AstraLux upper limit results in z′ - Ks & 1.75 mag, which corresponds to excluding & 0.2 M⊙ or & 3300 K (Baraffe et al. 2015) as possible sources or about earlier than M4.5V in spectal type (Kenyon & Hartmann 1995). Because any object later than ∼ M2V / M3V can be excluded by & 4 σ from JHKs photometry (Table 6), no M dwarf can be a false positive of the new com- panion candidate CVSO 30 c. 6. Mass determination and conclusions With the object brightness determined from the direct near-IR imaging and the information provided by the spectroscopic anal- ysis, we can directly estimate the basic parameters of CVSO 30 c. To determine the luminosity, we considered the extinction law Article number, page 7 of 15 A&Aproofs: manuscript no. CVSO ] g a m [ n o i t c n i t x e l a c i t p O ] x e d [ g g o l y t i v a r g e c a f r u S 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1000 1500 2000 2500 Effective temperature [K] 5.0 4.5 4.0 3.5 3.0 1000 1500 2000 2500 Effective temperature [K] ] x e d [ ] H / M [ y t i c i l l a t e M 0.2 −0.0 −0.2 −0.4 −0.6 1000 1500 2000 2500 Effective temperature [K] Fig. 5. 1 σ contour plots of the χ2 Drift-Phoenix model fit to the spec- trum shown in Fig. 4. Contour plot in extinction vs. effective temper- ature (top), surface gravity log g vs. effective temperature (centre) and metallicity [M/H] vs. effective temperature (bottom). The fit shows a best fit at 1600 K, low extinction of 0.19 mag, higher values becom- ing increasingly less likely, and a best fit at log g 3.6. While all surface gravities seem to be almost of equal probability, a high surface gravity foreground brown dwarf can be excluded from the shape of the H band in Fig. 6. Although the young planetary models differ in photometric colours, this could be because of a not yet fully understood change in the cloud properties at the L-T transition that is indicated by the change in brightness of the L-T transition with age of the system, which we show in Fig. 11. Fig. 7. AstraLux z′ band image of CVSO 30, taken on Aug 27, 2015. The dotted circle indicates an angular separation of 1.8 arcsec to CVSO 30 (see Fig. 8). No other objects are detected except for the star, which is located in the centre of the AstraLux image. by Rieke & Lebofsky (1985), a bolometric correction of B.C.K = 3.3+0.0 −0.7 mag for spectral type L5-T4 (Golimowski et al. 2004), and a distance of 357 ± 52 pc to the 25 Orionis cluster. From the 2MASS brightness of the primary and the differential bright- ness measured in our VLT NACO data (Table 2) and the extinc- tion value towards the companion derived from spectroscopy, we Article number, page 8 of 15 Fig. 8. S/N= 5 detection limit of our AstraLux observation of CVSO 30 (Fig. 7). The reached magnitude difference that is dependent on the angular separation to the star is shown. The horizontal dashed lines in- dicate the expected magnitude differences of sub-stellar companions of the star at an age of 3 Myr. Beyond about 1.8 arcsec (or ∼640 au of projected separation), all companions with masses down to 10 MJup can be excluded around CVSO 30. find log Lbol/L⊙ = −3.78+0.33 −0.13 dex. From the luminosity and ef- fective temperature, we calculate the radius to be R= 1.63+0.87 −0.34 RJup. In combination with the derived surface gravity, this would correspond to a mass of M= 4.3 MJup, dominated in its errors by high distance and surface gravity uncertainties. While the lat- ter value and the photometry (Fig. 2) would be consistent with a high surface gravity, thus old foreground T-type brown dwarf, but inconsistent with an L-type brown dwarf, the available spec- troscopy excludes an old T-type brown dwarf (Fig. 6 and Table 6). While the photometry is also consistent with early-M dwarfs, the K-band spectroscopy and z′ upper limit show the opposite behaviour, being only consistent with late-M dwarfs, excluding all types with high significance. Similarly, the remaining H-band spectroscopy excludes all comparison objects. Only the best- fitting Drift-Phoenix model (Fig. 4) shows low deviation in H band, consistent with the fact that the only available very young directly imaged planet candidates exhibit higher temperatures, thus a steeper H band (Fig. 6). Although recent observations by Yu et al. (2015) cast doubts on the existence of the inner transiting planet candidate CVSO 30 b or PTFO 8-8695 b, we assume its existence throughout the remaining discussion because all five hypotheses have dif- ficulties in reproducing the observations presented in Yu et al. (2015), including for example different types of starspots. The inner planet hypothesis gives another constraint, namely that the system has to be stable with both its planets. As described in van Eyken et al. (2012), CVSO 30 b is very close to its Roche radius, the radius of stability. Assuming the values for mass of CVSO 30 b, its radius and orbital period (Tables 2 and 4), we find from the Roche limit an upper limit for the mass of CVSO 30 of ≤ 0.92 M⊙ for a stable inner system comprised of CVSO 30 A and b. This mass limit for CVSO 30 is fulfilled at 1 Myr for masses of CVSO 30 c of ≤ 6.9 MJup at ≤ 760 pc up to 5.8 Myr with masses of CVSO 30 c of ≤ 9.2 MJup at ≤ 455 pc, accord- ing to BT-Settl evolutionary models (Allard 2014; Baraffe et al. 2015). Higher ages are not consistent with the age estimate of the primary, but even at 20 Myr we find a mass of CVSO 30 c of ≤ 12.1 MJup at ≤ 340 pc. With the Roche stability criterion for CVSO 30 b, the previous calculations result in a mass estimate of M= 4.3+4.9 −3.7 MJup for CVSO 30 c. For the approximate age of CVSO 30, 2 -- 3 Myr BT-Settl evo- lutionary models (Allard 2014; Baraffe et al. 2015) predict an T. O. B. Schmidt et al.: Direct Imaging of a second planet candidate in the transiting CVSO 30 system. Brown dwarfs Stars 78 M Jup ] g a m [ y r a m i r p o t e v i t a l e r d n u o r g k c a b e v o b a a m g i s Planets 13.6 M Jup 3 h t p e d d e h c a e r 0 2 4 6 8 Hydrogen Burning Limit Deuterium Burning Limit CVSO 30 c CVSO 30 c GJ 504 b HD 95086 b 2M1207 b HR 8799 b HR 8799 c & d HR 8799 e Beta Pic b 1RXS 1609 b CT Cha b & CXHR 73 b 2M044144 b GQ Lup b HD 100546 b LkCa15 b & c & ROXs 42B b DH Tau b & SR 12 AB c 10 M Jup 5 M Jup Jup 4 M 3 M Jup AB Pic b 2M0103 AB b HD 106906 b GU Psc b 51 Eri b UScoCTIO 108 b & GSC 06214 b 2M0219 b PZ Tel B -1 -2 -3 -4 -5 -6 -7 ) O • L / L ( g o l 0 2 4 separation [arcsec] 6 8 Fig. 10. Dynamic range per pixel achieved in our VLT / NACO Ks-band observations, given as 3 σ contrast to the primary star. The companion would have been detectable until 0.48 arc seconds or 171 au separation. A depth of 20.2 mag was reached at maximum, corresponding to 2.8 MJup. be found from 30 au outwards, planets from 79 au outwards, and CVSO 30 c could have been found from 171 au outwards. The core-accretion model (Safronov & Zvjagina 1969; Goldreich & Ward 1973; Pollack et al. 1996), one of the much debated planet formation scenarios, is unlikely to form an object in situ at ≥662 au because the timescale would be prohibitively long at such separations. In principle, the object could also have formed in a star-like fashion by turbulent core fragmentation as in the case of a binary star system, since the opacity limit for fragmentation is a few Jupiter masses (Bate 2009), but the large separation and high mass ratio argue against this hypothesis. The even more obvious possibility would be planet-planet scattering because an inner planet candidate CVSO 30 b of comparable mass is present that could have been scattered in- ward at the very same scattering event. Several authors sim- ulated such events and found mostly highly eccentric orbits for the outer scattered planets of up to 100s or 1000s of au (Stamatellos & Whitworth 2009; Nagasawa & Ida 2011), which is similar to the minimum separation of our outer planet can- didate of 662 au. The closest match to CVSO 30 bc of a model simulation was presented by Nagasawa & Ida (2011) with an ob- ject at ∼300 au, which has an inner hot planet with which it was scattered. Scattering or gravitational interaction might not be that uncommon as 72% ± 16% of hot Jupiters are part of multi-planet and/or multi-star systems (Ngo et al. 2015). The luminosity of CVSO 30 c is only consistent with hot- start models that usually represent the objects formed by grav- itational disk-instability, not with cold-start models that are at- tributed to core-accretion-formed planets (Marley et al. 2007). However, as stated in Spiegel & Burrows (2012), first-principle calculations cannot yet specify the initial (post-formation) en- tropies of objects with certainty in the different formation sce- narios, hence CVSO 30 c could have formed through a gravi- tational disk-instability or core accretion and might have been scattered with CVSO 30 b afterwards. In this context, it would also be important to clarify the na- ture of the unusually blue H-Ks colour of CVSO 30 c. It is con- Article number, page 9 of 15 6.0 6.5 7.0 7.5 log (age) [yr] 8.0 Fig. 9. Evolution of young stars, brown dwarfs, and planets with BT- Settl evolutionary tracks (Allard 2014; Baraffe et al. 2015). Shown are a few of the planet candidates known so far in comparison to the new sub-stellar companion candidate CVSO 30 c (see Table A.1). apparent brightness of mK ∼ 18.5 mag (assuming the distance to 25 Ori), effective temperature ∼1575 K, mass 4 -- 5 MJup, and log Lbol/L⊙ ∼ -3.8 dex. These expected values are very close to the best-fit atmospheric model spectra fits above, and even the derived visual extinction of about 0.19 mag is very close to the value of the primary ∼0.12 mag (Briceño et al. 2005). Of course, these evolutionary models can also be used to de- termine the resulting mass from the luminosity and age of the companion candidate and system, respectively. To put CVSO 30 c into context, we show the models and several of the cur- rently known directly imaged planet candidates in Fig. 9. The new companion is one of the youngest and lowest mass compan- ions, and we find a mass of 4.7+5.5 −2.0 MJup because the luminosity is not very precise as a result of the rather scarce knowledge of the distance of the system. However, if we take temperature addi- tionally into account, we find a more precise mass determination of 4.7+3.6 −2.0 MJup, which places CVSO 30 c well within the plane- tary regime and would mean that it is very close in mass to the probable inner companion of the system CVSO 30 b with about 2.8 -- 6.9 MJup (van Eyken et al. 2012; Barnes et al. 2013). In Fig. 10 we show the depth reached per pixel in the Ks- band epoch of 20.2 mag, corresponding to 2.8 MJup at the age of CVSO 30, using the same models as above. Brown dwarfs could A&Aproofs: manuscript no. CVSO sistent with colours of free-floating planets (Fig. 2) and might be caused by its youth, allowing the companion to be very bright, still already being at the L-T transition, which would be con- sistent with simulations of cluster brown dwarfs at very young ages and their colours in Saumon & Marley (2008) (Fig. 11). This would imply a temperature at the lower end of the 1σ errors found for CVSO 30 c, ≤ 1400 K, which is consistent with the less steep H band in comparison to β Pic b of about 1600 -- 1700K (Chilcote et al. 2015), however, as shown in Fig. 6. For old brown dwarfs the L-T transition occurs at Te f f 1200 -- 1400 K, when methane absorption bands start to be ubiquitously seen. However, in the ∼30 Myr old planet candidates around HR 8799 no strong methane is found, while the spectrum of the ∼90 Myr old object around GU Psc shows strong methane ab- sorption (Naud et al. 2014) all at temperatures of about 1000 -- 1100 K. Thus the L-T transition might be gravity dependent (Marley et al. 2012). Binarity of CVSO 30 c cannot be excluded either, which would also explain the unusual blue H-Ks colour. Since we cannot confirm that CVSO 30 c is co-moving with its host star from our proper motion analysis, we cannot exclude the possibility that CVSO 30 c is a free-floating young planet be- longing to the 25 Ori cluster, which is not gravitationally bound to CVSO 30. However, such a coincidence is highly improba- ble. Zapatero Osorio et al. (2000) searched 847 arcmin2 of the σ Orionis star cluster for free-floating planets and found only six candidates in the survey with similar colours as CVSO 30 c. This means that the probability to find a free-floating planet by chance within a radius of 1.85" around the transiting planet host star CVSO 30 is about 2·10−5. With a mass ratio of planet candidate to star q= 0.0115 ± 0.0015, CVSO 30 c (and CVSO 30 b) is among the imaged plan- ets with the lowest mass ratio (see e.g. De Rosa et al. 2014). In summary, CVSO 30 b and c for the first time allow a com- prehensive study of both a transiting and a directly imaged planet candidate within the same system, hence at the same age and even similar masses, using RV, transit photometry, direct imag- ing, and spectroscopy. Within a few years, the GAIA satellite mission (Perryman 2005) will provide the distance to the system to a precision of about 10 pc, which will additionally restrict the mass of CVSO 30 c. Simulations of a possible scattering event will profit from the current (end) conditions found for the sys- tem. Considering that the inner planet is very close to the Roche stability limit and the outer planet is far away from its host star, the future evolution and stability of the system is also very inter- esting for dedicated modelling. To investigate how often these scattering events occur, inner planets need also be searched for around other stars with directly imaged wide planets. Acknowledgements. We thank the ESO and CAHA staff for support, espe- cially during service-mode observations. Moreover, we would like to thank Jeff Chilcote and David Lafrenière for kindly providing electronic versions of com- parison spectra from their publications and the anonymous referee, the editor and our language editor for helpful comments that improved this manuscript. TOBS and JHMMS acknowledge support by the DFG Graduiertenkolleg 1351 "Extrasolar Planets and their Host Stars". RN and SR would like to thank the DFG for support in the Priority Programme SPP 1385 on the "First Ten Million Years of the Solar system" in project NE 515/33-1. SR is currently a Research Fellow at ESA/ESTEC. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Mas- sachusetts and the Infrared Processing and Analysis Center/California Institute of Technology, funded by the National Aeronautics and Space Administration and the National Science Foundation. This research has made use of the VizieR catalog access tool and the Simbad database, both operated at the Observatoire Strasbourg. This research has made use of NASA's Astrophysics Data System. Article number, page 10 of 15 8 8 9 9 10 10 11 11 ] ] g g a a m m [ [ K K M M 12 12 13 13 14 14 2M1207 A 2M0441 Bb DH Tau b CT Cha b < 5 Myr 10 Myr 50 Myr 100 Myr 200 Myr 500 Myr L1 1RXS 1609 b CVSO 30 c AB Pic b Beta Pic b L3 L7.5 L5.5 T2 HR 8799 c HR 8799 d T4.5 GJ 504 b 2M1207 b HR 8799 b 0.0 0.0 0.5 0.5 1.0 1.0 1.5 1.5 2.0 2.0 2.5 2.5 3.0 3.0 J−K [mag] J−K [mag] Fig. 11. Colour-magnitude diagram of a simulated cluster brown dwarf population from Saumon & Marley (2008). Each sequence corresponds to a different age as given in the legend. Superimposed we show the positions of several planet candidates and CVSO 30 c. Its unusual blue colour can most likely be attributed to its youth; it is about 2.4 Myr old. The younger the objects, the brighter they are because of not-yet- occurred contraction. Hence they reach the L- and T-dwarf regime at higher brightnesses. If this extrapolation is correct, CVSO 30 c is at the L-T transition, which is roughly consistent with its low effective temperature results. See the discussion and Table A.2 and Fig. A.5 for details. List of Objects 'CVSO 30' on page 2 References Adams, F. C. & Laughlin, G. 2001, Icarus, 150, 151 Allard, F. 2014, in IAU Symposium, Vol. 299, IAU Symposium, ed. M. Booth, B. C. Matthews, & J. R. Graham, 271 -- 272 Aller, K. M., Kraus, A. L., Liu, M. C., et al. 2013, ApJ, 773, 63 Allers, K. N., Jaffe, D. T., Luhman, K. L., et al. 2007, ApJ, 657, 511 Andrews, S. M., Rosenfeld, K. A., Kraus, A. L., & Wilner, D. J. 2013, ApJ, 771, 2013, ApJ, 774, 53 Bate, M. R. 2009, MNRAS, 392, 590 Béjar, V. J. S., Zapatero Osorio, M. R., Pérez-Garrido, A., et al. 2008, ApJ, 673, L185 Biller, B. A., Liu, M. C., Wahhaj, Z., et al. 2010, ApJ, 720, L82 Binks, A. S. & Jeffries, R. D. 2014, MNRAS, 438, L11 Bonavita, M., Daemgen, S., Desidera, S., et al. 2014, ApJ, 791, L40 129 Artigau, É., Gagné, J., Faherty, J., et al. 2015, ApJ, 806, 254 Baglin, A., Auvergne, M., Barge, P., et al. 2007, in American Institute of Physics Conference Series, Vol. 895, Fifty Years of Romanian Astrophysics, ed. C. Dumitrache, N. A. Popescu, M. D. Suran, & V. Mioc, 201 -- 209 Bailey, V., Meshkat, T., Reiter, M., et al. 2014, ApJ, 780, L4 Baraffe, I., Homeier, D., Allard, F., & Chabrier, G. 2015, A&A, 577, A42 Barnes, J. W., van Eyken, J. C., Jackson, B. K., Ciardi, D. R., & Fortney, J. J. T. O. B. Schmidt et al.: Direct Imaging of a second planet candidate in the transiting CVSO 30 system. Bonnefoy, M., Boccaletti, A., Lagrange, A.-M., et al. 2013, A&A, 555, A107 Bonnefoy, M., Chauvin, G., Lagrange, A.-M., et al. 2014a, A&A, 562, A127 Bonnefoy, M., Marleau, G.-D., Galicher, R., et al. 2014b, A&A, 567, L9 Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977 Boss, A. P. 1997, Science, 276, 1836 Boss, A. P. 2006, ApJ, 637, L137 Bowler, B. P. & Hillenbrand, L. A. 2015, ApJ, 811, L30 Bowler, B. P., Liu, M. C., Shkolnik, E. L., & Dupuy, T. J. 2013, ApJ, 774, 55 Bowler, B. P., Liu, M. C., Shkolnik, E. L., & Tamura, M. 2015, ApJS, 216, 7 Briceño, C., Calvet, N., Hernández, J., et al. 2005, AJ, 129, 907 Briceño, C., Hartmann, L., Hernández, J., et al. 2007a, ApJ, 661, 1119 Briceño, C., Preibisch, T., Sherry, W. H., et al. 2007b, Protostars and Planets V, Broeg, C., Schmidt, T. O. B., Guenther, E., et al. 2007, A&A, 468, 1039 Burgasser, A. J., Cruz, K. L., Cushing, M., et al. 2010a, ApJ, 710, 1142 Burgasser, A. J., Simcoe, R. A., Bochanski, J. J., et al. 2010b, ApJ, 725, 1405 Burningham, B., Leggett, S. K., Homeier, D., et al. 2011, MNRAS, 414, 3590 Cameron, A. G. W. 1978, Moon and Planets, 18, 5 Carson, J., Thalmann, C., Janson, M., et al. 2013, ApJ, 763, L32 Charbonneau, D., Brown, T. M., Latham, D. W., & Mayor, M. 2000, ApJ, 529, 345 L45 Chauvin, G., Lagrange, A.-M., Dumas, C., et al. 2004, A&A, 425, L29 Chauvin, G., Lagrange, A.-M., Dumas, C., et al. 2005a, A&A, 438, L25 Chauvin, G., Lagrange, A.-M., Lacombe, F., et al. 2005b, A&A, 430, 1027 Chauvin, G., Lagrange, A.-M., Zuckerman, B., et al. 2005c, A&A, 438, L29 Chilcote, J., Barman, T., Fitzgerald, M. P., et al. 2015, ApJ, 798, L3 Chiu, K., Fan, X., Leggett, S. K., et al. 2006, AJ, 131, 2722 Close, L. 2010, Nature, 468, 1048 Close, L. M., Siegler, N., Freed, M., & Biller, B. 2003, ApJ, 587, 407 Close, L. M., Zuckerman, B., Song, I., et al. 2007, ApJ, 660, 1492 Currie, T., Burrows, A., & Daemgen, S. 2014, ApJ, 787, 104 Cushing, M. C., Rayner, J. T., & Vacca, W. D. 2005, ApJ, 623, 1115 Cutri, R. M., Skrutskie, M. F., van Dyk, S., et al. 2003, 2MASS All Sky Catalog of point sources. De Rosa, R. J., Patience, J., Ward-Duong, K., et al. 2014, MNRAS, 445, 3694 Delorme, P., Gagné, J., Girard, J. H., et al. 2013, A&A, 553, L5 Diolaiti, E., Bendinelli, O., Bonaccini, D., et al. 2000, in Society of Photo- Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 4007, So- ciety of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. P. L. Wizinowich, 879 -- 888 Downes, J. J., Briceño, C., Mateu, C., et al. 2014, MNRAS, 444, 1793 Ducourant, C., Teixeira, R., Chauvin, G., et al. 2008, A&A, 477, L1 Dupuy, T. J., Liu, M. C., & Ireland, M. J. 2014, ApJ, 790, 133 Errmann, R., Raetz, S., Kitze, M., Neuhäuser, R., & YETI Team. 2014, Contri- butions of the Astronomical Observatory Skalnate Pleso, 43, 513 Faherty, J. K., Rice, E. L., Cruz, K. L., Mamajek, E. E., & Núñez, A. 2013, AJ, 145, 2 Ford, E. B. & Rasio, F. A. 2008, ApJ, 686, 621 Freistetter, F., Krivov, A. V., & Löhne, T. 2007, A&A, 466, 389 Galicher, R., Rameau, J., Bonnefoy, M., et al. 2014, A&A, 565, L4 Gauza, B., Béjar, V. J. S., Pérez-Garrido, A., et al. 2015, ApJ, 804, 96 Gizis, J. E., Allers, K. N., Liu, M. C., et al. 2015, ApJ, 799, 203 Goldreich, P. & Ward, W. R. 1973, ApJ, 183, 1051 Golimowski, D. A., Leggett, S. K., Marley, M. S., et al. 2004, AJ, 127, 3516 Helling, C., Dehn, M., Woitke, P., & Hauschildt, P. H. 2008, ApJ, 675, L105 Hernández, J., Briceño, C., Calvet, N., et al. 2006, ApJ, 652, 472 Hernández, J., Calvet, N., Hartmann, L., et al. 2005, AJ, 129, 856 Hewett, P. C., Warren, S. J., Leggett, S. K., & Hodgkin, S. T. 2006, MNRAS, 367, 454 Hinkley, S., Pueyo, L., Faherty, J. K., et al. 2013, ApJ, 779, 153 Horne, K. 1986, PASP, 98, 609 Ireland, M. J., Kraus, A., Martinache, F., Law, N., & Hillenbrand, L. A. 2011, ApJ, 726, 113 Itoh, Y., Hayashi, M., Tamura, M., et al. 2005, ApJ, 620, 984 Jayawardhana, R. & Ivanov, V. D. 2006, Science, 313, 1279 Jenkins, J. S., Pavlenko, Y. V., Ivanyuk, O., et al. 2012, MNRAS, 420, 3587 Jordi, K., Grebel, E. K., & Ammon, K. 2006, A&A, 460, 339 Kalas, P., Graham, J. R., Chiang, E., et al. 2008, Science, 322, 1345 Kamiaka, S., Masuda, K., Xue, Y., et al. 2015, PASJ, 67, 94 Kenyon, S. J. & Hartmann, L. 1995, ApJS, 101, 117 Kirkpatrick, J. D., Dahn, C. C., Monet, D. G., et al. 2001, AJ, 121, 3235 Kirkpatrick, J. D., Reid, I. N., Liebert, J., et al. 2000, AJ, 120, 447 Koch, D. G., Borucki, W. J., Basri, G., et al. 2010, ApJ, 713, L79 Koen, C. 2015, MNRAS, 450, 3991 Konopacky, Q. M., Barman, T. S., Macintosh, B. A., & Marois, C. 2013, Science, 339, 1398 Kraus, A. L. & Ireland, M. J. 2012, ApJ, 745, 5 Kraus, A. L., Ireland, M. J., Cieza, L. A., et al. 2014, ApJ, 781, 20 Kuzuhara, M., Tamura, M., Ishii, M., et al. 2011, AJ, 141, 119 Kuzuhara, M., Tamura, M., Kudo, T., et al. 2013, ApJ, 774, 11 Lafrenière, D., Jayawardhana, R., Janson, M., et al. 2011, ApJ, 730, 42 Lafrenière, D., Jayawardhana, R., & van Kerkwijk, M. H. 2008, ApJ, 689, L153 Lagrange, A.-M., Bonnefoy, M., Chauvin, G., et al. 2010, Science, 329, 57 Lagrange, A.-M., Gratadour, D., Chauvin, G., et al. 2009, A&A, 493, L21 Latham, D. W., Rowe, J. F., Quinn, S. N., et al. 2011, ApJ, 732, L24 Lissauer, J. J., Jontof-Hutter, D., Rowe, J. F., et al. 2013, ApJ, 770, 131 Liu, M. C., Magnier, E. A., Deacon, N. R., et al. 2013, ApJ, 777, L20 Luhman, K. L., Adame, L., D'Alessio, P., et al. 2005a, ApJ, 635, L93 Luhman, K. L., D'Alessio, P., Calvet, N., et al. 2005b, ApJ, 620, L51 Luhman, K. L., Patten, B. M., Marengo, M., et al. 2007, ApJ, 654, 570 Luhman, K. L., Wilson, J. C., Brandner, W., et al. 2006, ApJ, 649, 894 Macintosh, B., Graham, J. R., Barman, T., et al. 2015, Science, 350, 64 Mamajek, E. E. & Bell, C. P. M. 2014, MNRAS, 445, 2169 Mannings, V. & Sargent, A. I. 1997, ApJ, 490, 792 Mannucci, F., Basile, F., Poggianti, B. M., et al. 2001, MNRAS, 326, 745 Marley, M. S. 2013, Science, 339, 1393 Marley, M. S., Fortney, J. J., Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2007, ApJ, 655, 541 Marley, M. S., Saumon, D., Cushing, M., et al. 2012, ApJ, 754, 135 Marois, C., Macintosh, B., Barman, T., et al. 2008, Science, 322, 1348 Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., & Barman, T. 2010, Nature, 468, 1080 Martín, E. L., Zapatero Osorio, M. R., Barrado y Navascués, D., Béjar, V. J. S., & Rebolo, R. 2001, ApJ, 558, L117 Mayor, M. & Queloz, D. 1995, Nature, 378, 355 Metchev, S. A. & Hillenbrand, L. A. 2006, ApJ, 651, 1166 Mohanty, S., Jayawardhana, R., Huélamo, N., & Mamajek, E. 2007, ApJ, 657, 1064 Montet, B. T., Bowler, B. P., Shkolnik, E. L., et al. 2015, ApJ, 813, L11 Moya, A., Amado, P. J., Barrado, D., et al. 2010, MNRAS, 405, L81 Muñoz, D. J., Kratter, K., Vogelsberger, M., Hernquist, L., & Springel, V. 2015, MNRAS, 446, 2010 Mugrauer, M. & Neuhäuser, R. 2005, Astronomische Nachrichten, 326, 701 Mugrauer, M., Vogt, N., Neuhäuser, R., & Schmidt, T. O. B. 2010, A&A, 523, L1 Nagasawa, M. & Ida, S. 2011, ApJ, 742, 72 Naud, M.-E., Artigau, É., Malo, L., et al. 2014, ApJ, 787, 5 Neuhäuser, R., Errmann, R., Berndt, A., et al. 2011, Astronomische Nachrichten, 332, 547 Posters, 47 Neuhäuser, R., Errmann, R., Raetz, S., et al. 2013, in Protostars and Planets VI Neuhäuser, R., Guenther, E. W., Wuchterl, G., et al. 2005, A&A, 435, L13 Neuhäuser, R. & Schmidt, T. O. B. 2012, Topics in Adaptive Optics (InTech) Ngo, H., Knutson, H. A., Hinkley, S., et al. 2015, ApJ, 800, 138 Pani´c, O., Hogerheijde, M. R., Wilner, D., & Qi, C. 2009, A&A, 501, 269 Patience, J., King, R. R., de Rosa, R. J., & Marois, C. 2010, A&A, 517, A76 Peña Ramírez, K., Béjar, V. J. S., Zapatero Osorio, M. R., Petr-Gotzens, M. G., & Martín, E. L. 2012, ApJ, 754, 30 Pecaut, M. J., Mamajek, E. E., & Bubar, E. J. 2012, ApJ, 746, 154 Perryman, M. A. C. 2005, in Astronomical Society of the Pacific Conference Series, Vol. 338, Astrometry in the Age of the Next Generation of Large Tele- scopes, ed. P. K. Seidelmann & A. K. B. Monet, 3 Pollack, J. B., Hubickyj, O., Bodenheimer, P., et al. 1996, Icarus, 124, 62 Potter, D., Martín, E. L., Cushing, M. C., et al. 2002, ApJ, 567, L133 Preibisch, T., Brown, A. G. A., Bridges, T., Guenther, E., & Zinnecker, H. 2002, AJ, 124, 404 Quanz, S. P., Amara, A., Meyer, M. R., et al. 2015, ApJ, 807, 64 Quanz, S. P., Amara, A., Meyer, M. R., et al. 2013, ApJ, 766, L1 Rameau, J., Chauvin, G., Lagrange, A.-M., et al. 2013, ApJ, 772, L15 Rasio, F. A. & Ford, E. B. 1996, Science, 274, 954 Rayner, J. T., Cushing, M. C., & Vacca, W. D. 2009, ApJS, 185, 289 Rebolo, R., Zapatero Osorio, M. R., Madruga, S., et al. 1998, Science, 282, 1309 Reid, I. N. & Walkowicz, L. M. 2006, PASP, 118, 671 Reipurth, B. & Clarke, C. 2001, AJ, 122, 432 Rieke, G. H. & Lebofsky, M. J. 1985, ApJ, 288, 618 Safronov, V. S. & Zvjagina, E. V. 1969, Icarus, 10, 109 Sallum, S., Follette, K. B., Eisner, J. A., et al. 2015, Nature, 527, 342 Saumon, D. & Marley, M. S. 2008, ApJ, 689, 1327 Schmidt, T. O. B., Mugrauer, M., Neuhäuser, R., et al. 2014, A&A, 566, A85 Schmidt, T. O. B., Neuhäuser, R., Mugrauer, M., Bedalov, A., & Vogt, N. 2009, in American Institute of Physics Conference Series, Vol. 1094, 15th Cam- bridge Workshop on Cool Stars, Stellar Systems, and the Sun, ed. E. Stem- pels, 852 -- 855 Schmidt, T. O. B., Neuhäuser, R., Seifahrt, A., et al. 2008, A&A, 491, 311 Schneider, J., Dedieu, C., Le Sidaner, P., Savalle, R., & Zolotukhin, I. 2011, A&A, 532, A79 Siess, L., Dufour, E., & Forestini, M. 2000, A&A, 358, 593 Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163 Sparks, W. B. & Ford, H. C. 2002, ApJ, 578, 543 Spiegel, D. S. & Burrows, A. 2012, ApJ, 745, 174 Article number, page 11 of 15 A&Aproofs: manuscript no. CVSO Stamatellos, D. & Whitworth, A. P. 2009, MNRAS, 392, 413 Thatte, N., Abuter, R., Tecza, M., et al. 2007, MNRAS, 378, 1229 Todorov, K., Luhman, K. L., & McLeod, K. K. 2010, ApJ, 714, L84 van Eyken, J. C., Ciardi, D. R., von Braun, K., et al. 2012, ApJ, 755, 42 Vorobyov, E. I. 2013, A&A, 552, A129 Wahhaj, Z., Liu, M. C., Biller, B. A., et al. 2011, ApJ, 729, 139 Weinberg, M. D., Shapiro, S. L., & Wasserman, I. 1987, ApJ, 312, 367 Wright, J. T., Upadhyay, S., Marcy, G. W., et al. 2009, ApJ, 693, 1084 Yu, L., Winn, J. N., Gillon, M., et al. 2015, ApJ, 812, 48 Zacharias, N., Finch, C. T., Girard, T. M., et al. 2013, AJ, 145, 44 Zacharias, N., Monet, D. G., Levine, S. E., et al. 2004, in Bulletin of the Ameri- can Astronomical Society, Vol. 36, American Astronomical Society Meeting Abstracts, 1418 Zapatero Osorio, M. R., Béjar, V. J. S., Martín, E. L., et al. 2000, Science, 290, 103 Zuckerman, B., Rhee, J. H., Song, I., & Bessell, M. S. 2011, ApJ, 732, 61 Article number, page 12 of 15 A&A -- CVSO, Online Material p 13 Appendix A: Supplementary material CVSO 30 is currently not suitable for a common proper mo- tion analysis (Table 1). Because orbital motion around the host star might be detectable, we simulated the expected maximum separation (top) and position angle (bottom) change in Fig. A.1, dependent on inclination and eccentricity of the companion for an epoch difference of three years. This corresponds to our first astrometrically calibrated epoch from 2012 to a tentative new observation early in 2016. The dedicated orbital analysis shows that even after two to three years of epoch difference, no signifi- cant orbital motion is expected for the wide companion. A first spectrum of CVSO 30 c at an intermediate reduction step, shown in Fig. 3 (central panel), by subtracting an average spectrum of the spike, left and right of the companions psf from the superposition of companion and spike is given in Fig. A.2. We find the results before (red spectrum) and after (blue spec- trum) spike subtraction, which also removes the still-present OH lines. In addition, the spectrum of the host star CVSO 30 is shown in black for comparison. In Fig. A.3 we show the expected signal-to-noise ratio (S/N) for the given conditions and integration times (Tables 3 and 1) using ESO's exposure time calculator for SINFONI and the lat- est available Pickles template spectrum M6 (blue). We derive an almost identical S/N using the flux of the companion after spike removal (Fig. 3) compared to the noise of the background next to the spike (black). However, these S/N estimates are not achieved for our final extracted spectrum and its noise estimate (Fig. 4) because the spike itself adds slight additional noise, and more importantly, because of the imperfect removal of the spike that dominates the final S/N (red). To take this effect that is most likely caused by imperfect primary star positioning into account, we derived our final noise estimate, given as noise floor in Fig. 4, as the standard deviation of the neighbouring spectral channels after removing the continuum at the spectral position of interest. This noise was also used for the spectral model fit- ting (Figs. 4 and 5) and the reduced χ2 estimation for several comparison objects (Table 6). We show the colour-magnitude diagram given in Fig. 11 in the main document with the identification of all the unlabelled objects in a full version in Fig. A.5 with the corresponding refer- ences in Table A.2 . The objects seem to follow the prediction of Saumon & Marley (2008) quite well, especially around 10 Myr. Only 2M1207 b seems to be far off, possibly because of an edge- on disk that heavily reddens the object (Mohanty et al. 2007). Whether HR 8799 c and d are unusual can hardly be judged be- cause no similar object with a very low luminosity is known at this age. HR 8799 b is very low in luminosity (Fig. 9), however. The younger the objects, the higher in luminosity they are at sim- ilar spectral type because of their larger radius because they are still experiencing gravitational contraction. The plot (Fig. A.5) implies that CVSO 30 c is the first very young (< 10 Myr) L-T transition object. The core-accretion model (Safronov & Zvjagina 1969; Goldreich & Ward 1973; Pollack et al. 1996),was also discussed in models for HR 8799 bcde by Close (2010), who argued that the inner planet was most likely formed by core accretion, while for the outer planets the gravitational instability of the disk (Cameron 1978; Boss 1997) is the more probable formation sce- nario. However, HR 8799 is an A- or F-star, and recent numer- ical simulations (Vorobyov 2013) showed that disk fragmenta- tion fails to produce wide-orbit companions around stars with mass < 0.7 M⊙, hence this is unfeasible for the ∼0.34 -- 0.44 M⊙ M3 star CVSO 30. In addition, the disk would have to be large i = 0 deg i = 52 deg i = 89 deg 0.6 0.5 0.4 0.3 0.2 0.1 ] s a m [ p e s ∆ 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 e 0.07 0.06 0.05 ] g e d [ A P ∆ 0.04 0.03 0.02 0.01 0.00 0.0 i = 0 deg i = 52 deg i = 89 deg 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 e Fig. A.1. Expected maximum separation (top) and position angle (bot- tom) change, dependent on inclination and eccentricity of the compan- ion for an epoch difference of three years (early in 2016, since the first calibrated epoch was made at the end of 2012). enough for in situ formation. The most massive disks around M stars (e.g. IM Lupi) might be large enough, but in this case, it was shown to possess almost all of its dust within 400 au sepa- ration (Pani´c et al. 2009), which is still too small for formation at 662 au. If the object has not formed in situ, a very obvious solution would be scattering induced by a stellar flyby close to the sys- tem (Adams & Laughlin 2001; Muñoz et al. 2015) or with an- other object of the system. While Reipurth & Clarke (2001) de- scribed this possibility for the formation of brown dwarfs by disintegration of a small multiple system and possibly a cut- off from the formation material reservoir, which might have oc- curred for example for directly imaged circumbinary planet can- didates such as ROSS 458(AB) c (Burgasser et al. 2010b) or SR 12 AB c (Kuzuhara et al. 2011), the even more obvious possi- bility would be planet-planet scattering because an inner planet candidate CVSO 30 b of similar mass is present that might have been scattered inward at the very same scattering event. A way to distinguish between the formation scenarios would be by higher S/N ratio spectroscopy, as has been done for HR 8799 c (Konopacky et al. 2013). With both H2O and CO de- tected, it is possible to estimate the bulk atmospheric carbon-to- oxygen ratio and whether it differs from that of the primary star, which led Marley (2013) to speculate that HR 8799 c formed by core accretion and not by gas instability. We can place CVSO 30 c best into context by comparing it with the recent M-dwarf survey of Bowler et al. (2015), who A&A -- CVSO, Online Material p 14 Fig. A.4. Direct images of CVSO 30 c. Top row, left to right: Quasi-simultaneous VLT NACO J-, H-, and Ks-band data, taken in a sequence and shown in the same percentage of upper cut-off and lower cut-off value 0. Lower row, left to right: VLT NACO J band with double exposure time per single image, the same in total, Keck image of data by van Eyken et al. (2012), re-reduced. We note that the companion is north-east, not a contaminant south-east, as given in van Eyken et al. (2012), and a JHKs colour composite, showing that CVSO 30 c has similar colours as its host star (Fig. 2). 50 40 30 20 10 0 t e s f f o + λ F x u l F d e z i l a m r o n 1.4 1.6 1.8 2.0 wavelength [micron] 2.2 2.4 Fig. A.2. Spectrum of the primary (black) and the companion at the best illuminated pixel as given in the central panel of Fig. 3 (red; with OH lines), together with the spectrum after subtracting the average spike east and west of the companion (blue), which contributes about 30 % of the light (beforehand). While the H-band spectrum presents a trian- gular shape and bluer colour, indicating a young sub-stellar companion, the full continuum of the companion is not reliable because different amounts of flux are superimposed by the rotating primary spike, which changes the overall continuum shape because of the different spike re- moval quality. found that fewer than 6% of M dwarfs harbour massive giant planets of 5 -- 13 MJup at 10 -- 100 au and that there is currently no statistical evidence for a trend of giant planet frequency with stellar host mass at large separations. We note, however, that CVSO 30 c would probably not have been found at the distance of their targets because it would not have been in the field of view as a result of its large separation of about 662 au. About 20 of the 49 directly imaged planet candidates at www.exoplanet.eu have an M dwarf as host star. 25 20 15 10 5 o i t a r e s i o n o t l a n g i s 0 1.4 1.6 1.8 2.0 wavelength [µm] 2.2 2.4 Fig. A.3. Signal-to-noise ratio (S/N) achieved for the brightest pixel vs. the background noise in the combined cube (black). For compari- son the expected and almost identical S/N is shown, simulated using the exposure time calculator (ETC) of ESO/SINFONI (blue). In red we present the final achieved S/N of the extracted companion spectrum af- ter removing a superimposed spike (Fig. 3), as shown in Fig. 4. At a projected separation of ∼662 au, the system is above the long-term stability limit of ∼390 au for an M3 primary star of 0.34 -- 0.44 M⊙ (Table 1), following the argumentation of Weinberg et al. (1987) and Close et al. (2003). However, as shown in Mugrauer & Neuhäuser (2005), 2M1207 and its com- panion (Chauvin et al. 2005a) are also exceeding this long-term stability limit at about three times the age of CVSO 30. The currently acquired data are consistent with planet-planet scattering simulations in Ford & Rasio (2008), showing that massive planets are more likely to eject one another, whereas smaller planets are more likely to collide, resulting in stabilised systems, as supported by Kepler satellite and Doppler sur- vey results that find predominantly smaller (Wright et al. 2009; A&A -- CVSO, Online Material p 15 Table A.2. Colour-magnitude plot (Figs. 11 and A.5) references reference Chauvin et al. (2004) Mohanty et al. (2007) Ducourant et al. (2008) Lafrenière et al. (2008) Itoh et al. (2005) Luhman et al. (2006) Chauvin et al. (2005c) Burningham et al. (2011) Ireland et al. (2011) Naud et al. (2014) Metchev & Hillenbrand (2006) Chauvin et al. (2005b) Bonnefoy et al. (2014a) Luhman et al. (2007) Carson et al. (2013) Hinkley et al. (2013) Lafrenière et al. (2011) Jayawardhana & Ivanov (2006) 284149 Close et al. (2007) Reid & Walkowicz (2006) Kirkpatrick et al. (2000) Kirkpatrick et al. (2001) Dupuy et al. (2014) Luhman et al. (2006) Schmidt et al. (2008) Gauza et al. (2015) Bowler & Hillenbrand (2015) b HIP 77900 b G196-3 b HD 106906 b W0047 +68 2M0219 b PSO 318 Object HR 8799 b, c, d β Pic b ROXs 42B b SR 12 AB c reference Marois et al. (2008) Bonnefoy et al. (2013) Kraus et al. (2014) Kuzuhara et al. (2011) 2M0103 Delorme et al. (2013) AB b USco CTIO 108 b PZ Tel b GJ 504 b Béjar et al. (2008) Mugrauer et al. (2010) Kuzuhara et al. (2013) 2M0122 Bowler et al. (2013) b HD 130948 B & C 2M0355 CD-35 2722 b OTS 44 Cha 1109 HD Potter et al. (2002) Faherty et al. (2013) Wahhaj et al. (2011) Luhman et al. (2005b) Luhman et al. (2005a) Bonavita et al. (2014) Aller et al. (2013) Rebolo et al. (1998) Bailey et al. (2014) Gizis et al. (2015) Artigau et al. (2015) Liu et al. (2013) Object 2M1207 A & b 1RXS 1609 b DH Tau b AB Pic b Ross 458 AB c GSC 06214 b GU Psc b HD 203030 b GSC 08047 b HN Peg b κ And b HIP 78530 b Oph 11 A & b LP 261-75 b GJ 417 B & C CHXR 73 b CT Cha b VHS 1256 b 2M0441 Bb HIP 77900 b DH Tau b OTS 44 CT Cha b/ CHXR 73 b 2M0441 Bb SR 12 AB c Cha 1109 ROXs 42 B b 1RXS 1609 b 2M0219 b 2M0103 AB b GSC 08047 b L1 < 5 Myr 10 Myr 50 Myr 100 Myr 200 Myr 500 Myr 8 8 9 9 10 10 HIP 78530 b PZ Tel b 2M1207 A Oph 11 A / b GSC 06214 b UScoCTIO 108 b VHS 1256 HD 284149 b CD−35 2722 b Kappa And b CVSO 30 c Beta Pic b L3 11 11 ] ] g g a a m m [ [ K K M M 12 12 HD 106906 b AB Pic b G196−3 b LP 261−75 b HD 130948 B/C 2M0122 b 2M0355 GJ 417 B/C W0047+68 T4.5 13 13 L7.5 L5.5 PSO 318 T2 HN Peg b HR 8799 c HR 8799 d HD 203030 b 2M1207 b 14 14 ROSS 458 AB c GU Psc b GJ 504 b VHS 1256 b HR 8799 b 0.0 0.0 0.5 0.5 1.0 1.0 1.5 1.5 2.0 2.0 2.5 2.5 3.0 3.0 J−K [mag] J−K [mag] Fig. A.5. Colour-magnitude diagram of the simulated cluster brown dwarf population from Saumon & Marley (2008). Each sequence cor- responds to a different age as given in the legend. Superimposed we show the positions of several planet candidates with full identification and CVSO 30 c. See Fig. 11 and Table A.2 for further details. Latham et al. 2011) low-density (e.g. Lissauer et al. 2013) plan- ets in compact close multi-planet systems. Table A.1. Evolutionary plot (Fig. 9) references Object GJ 504 b 2M1207 b HR 8799 e 1RXS 1609 b CT Cha b 2M044 144 b HD 100546b ROXs 42B b DH Tau b AB Pic b 51 Eri b USco CTIO 108 b 2M0219 b reference Kuzuhara et al. (2013) Chauvin et al. (2004) Mohanty et al. (2007) Marois et al. (2010) Zuckerman et al. (2011) Moya et al. (2010) Lafrenière et al. (2008) Neuhäuser & Schmidt (2012) Pecaut et al. (2012) Schmidt et al. (2008) Todorov et al. (2010) Quanz et al. (2013) Quanz et al. (2015) Currie et al. (2014) Itoh et al. (2005) Neuhäuser & Schmidt (2012) Chauvin et al. (2005c) Neuhäuser & Schmidt (2012) Macintosh et al. (2015) Montet et al. (2015) Béjar et al. (2008) Preibisch et al. (2002) Pecaut et al. (2012) Artigau et al. (2015) Object HD 95086 b HR 8799 b, c, d β Pic b CHXR 73 b GQ Lup b LkCA15 b, c SR 12 AB c 2M0103 AB b HD 106906 b GU Psc b GSC 06214 b PZ Tel B reference Rameau et al. (2013) Galicher et al. (2014) Marois et al. (2008) Zuckerman et al. (2011) Moya et al. (2010) Lagrange et al. (2009) Bonnefoy et al. (2014b) Binks & Jeffries (2014) Mamajek & Bell (2014) Luhman et al. (2006) Neuhäuser et al. (2005) Kraus & Ireland (2012) Sallum et al. (2015) Kuzuhara et al. (2011) Delorme et al. (2013) Bailey et al. (2014) Naud et al. (2014) Ireland et al. (2011) Preibisch et al. (2002) Mugrauer et al. (2010) Biller et al. (2010) Jenkins et al. (2012)
1703.09519	2	1703	2017-06-08T20:06:43	Measuring the radius and mass of Planet Nine	[ "astro-ph.EP" ]	Batygin and Brown (2016) have suggested the existence of a new Solar System planet supposed to be responsible for the perturbation of eccentric orbits of small outer bodies. The main challenge is now to detect and characterize this putative body. Here we investigate the principles of the determination of its physical parameters, mainly its mass and radius. For that purpose we concentrate on two methods, stellar occultations and gravitational microlensing effects (amplification, deflection and time delay). We estimate the main characteristics of a possible occultation or gravitational effects: flux variation of a background star, duration and probability of occurence. We investigate also additional benefits of direct imaging and of an occultation.	astro-ph.EP	astro-ph	Revised version 5 June 2017 Measuring the radius and mass of Planet Nine J. Schneider Observatoire de Paris, LUTh-CNRS, UMR 8102, 92190, Meudon, France [email protected] Abstract Batygin and Brown (2016) have suggested the existence of a new Solar System planet supposed to be responsible for the perturbation of eccentric orbits of small outer bodies. The main challenge is now to detect and characterize this putative body. Here we investigate the principles of the determination of its physical parameters, mainly its mass and radius. For that purpose we concentrate on two methods, stellar occultations and gravitational microlensing effects (amplification, deflection and time delay). We estimate the main characteristics of a possible occultation or gravitational effects: flux variation of a background star, duration and probability of occurence. We investigate also additional benefits of direct imaging and of an occultation. keywords Planet P9 - stellar occultations – gravitational lensing 1. Introduction Batygin and Brown (2016) have suggested the existence of a new planet in the outer Solar System (called hereafter P9), supposed to be responsible for the perturbation of eccentric orbits of small bodies. From the characteristics of the observed perturbations of orbits, its mass M9 is estimated to be around 10-30 M ⊕ and its distance is estimated to be around 700 AU (Brown & Batygin 2016). More recent estipmate derived the the Cassini data analyis give rather a distance of 1000 AU . For the numerical estimates below, we take the new distance value. The main challenge is now to detect this putative body and, if it exists, to measure its physical characteristics. Several authors have investigated its search by direct imaging. This search presents three issues: constrain the 1 region on the sky plane where the putative planet is localized, estimate its brightness and choose the best possible imaging facilities and search strategies to detect the object. The success of this search depends on the brightness of the object. For a given distance the brightness depends on the planet albedo and radius. The estimates of the brightness run from V = 19.9 to V = 22.7 (Linder & Mordasini 2016) 1. The advantage of the direct imaging search is that it can be performed any time along the orbit and would be the first step to take spectra of the object. But it cannot provide the size and mass of the planet. Without a value for the radius one cannot infer the planet albedo from its brightness, while this parameter is essential to constrain physical characteristics of the planet surface. A value of the mass would constrain the planet internal structure, its formation scenarii and the mechanism of perturbation of small bodies orbits. Its predicted angular diameter (depending on the exact distance and size), being at most a few tens mas, only an interferometer with a baseline of the order of a kilometer could measure this radius at 1 micron with a 3% precision, a presently difficult task for a V > 20 object. That is why we propose to search for stellar occultations by P9 to measure its radius. For the mass determination, we investigate further the microlensing by P9 of a background star, as already proposed by Philippoc and Chobanu (2016). 2. Stellar occultations by P9 In this section we estimate the parameters of occultations and microlensing of background stars by P9. 2.1 Occultation depth Assuming a mass of 30 terrestrial mass and a standard rocky composition, the planet radius R9 is, according to Linder & Mordasini (2016), 2.9 to 6.3 R⊕ , depending on its evolution track. At a distance of 1000 AU, the angular diameter δ 9 is 40 mas for a 3 R⊕ planet. The angular diameter of the potentially eclipsed background stars are all less than 57 mas (for R Doradus). Therefore, neglecting grazing occultations, during an occultation the star will completely disappear behind P9 and its flux will vanish. More exactly, the star+P9 total flux will drop to the P9 flux, which is of the same order of magnitude, or higher, for faint stars. For a star with stellar flux F* required to detect an occultation with and SNR of 3 must be better than F/3 . fainter than P9, the photometric accuracy Δ F9 of the P9 flux 1 We have renormalized to a 1000 AU distance the Linder \& Mordasini (2016) estimate of V = 20.6 to V = 23.2 for a distance of 700 AU. 2 and its orbital motion. For a 1000 is about 1 km/sec, giving an apparent motion of 70 2.2 Occultation duration The P9 apparent motion is a composition of its parallax π9 AU distance of P9, its orbital velocity v 9 arcsec/yr on the sky plane. The latter dominates the proper motion of bakground stars, being all less than 10 arcsec/yr (for Barnard's star). The Earth orbital velocity, 30 km/sec, being much larger than the P9 orbital velocity, it dominates the apparent motion (in the Earth frame) of the observed occultation shadow. If the star path lies on the equator of the planet, the occultation duration is then Docc=2R9/(30km/ sec)/cosα⊕ where α⊕ is the angle between the P9 line of sight and the Earth orbital velocity vector at the time of occultation. assumes implicitly an occultation of the background star at the 2.3 Occultation geometry and kinematics The above estimate of Docc planet equator. Since the true impact parameter of the occultation, the distance between the occultation path and the planet center, is not known, one cannot a priori infer the planet radius from the relation 2R9=30 Docccosα⊕ km/sec This problem occurs for any stellar occultation by Solar System bodies, but then it is generally solved by the accumulation of several occultations by the same object. For instance, for the minor planet Ixion, up to 425 stellar occultations are predicted (Assafin et al. 2012). Since in our case occultations are rare (see the section "Occultation probability"), one cannot use this approach. We thus now show that the measurement of the duration of the ingess and egress of the occultation can solve the problem. The occultation duration Docc depends on the geometry of the occultation and on the kinematics of P9 and of the occulted star. The geometry of the occultation is represented on Figure 1. The paths of the planet and of the star are generally not parallel. Consequently, as shown on Figure 1, the impact parameter, i.e. the distance between the star path and the planet center, are different at ingress and egress. We designate their angular sizes by δbin and δbout . The angular size of the occultation path ("chord") is given by 3 δocc=δ1+ δ2=√δ9 2−δbin 2 + √δ9 2 2−δbout (1) where the different parameters of equation (1) are represented on Figure 1. We assume that during the ingress and the egress the impact parameter is approximately constant. Therefore the angular sizes of the ingress and egress pathes are given by δin=√(δ9+ δ)2−δbin 2 −√(δ9−δ)2−δbin 2 (2) δout=√(δ9+ δ)2−δbout 2 −√(δ9−δ)2−δbout 2 (3) b b in in P9 path stellar path b out δin δ1 > < δ out δ2 Figure 1 Kinematics and geometry of a stellar occultation by P9 From equations (1), (2) and (3) one has 2= δ9 2 (δ δocc 2δin 2 + δ* 2 δout 2 −2)+ 2δocc 2 )+ 1−(δ* 2/δin 2 + δ* 2/δout 2 ) 2 √4 δ* 2(1/δin δ* 2 δout 4/(δin 2 −1/δout 2 ) The geometrical parameters δocc , δin and δout are not primary observables. But they can be 4 derived from their corresponding durations Docc by , Din and Dout which are related to, them Docc=δocc/(ω9−ω) (4) Din=δin /(ω9−ω) (5) and Dout=δout/(ω9−ω) (6) . ω9 is the angular speed of P9, projected onto the stellar path and ω where ω9 is the angular star speed, i.e. its proper motion μ* is the composition of the parallactic speed ωπ 9 and the of P9: ω9=(ωπ 9+ v9 sin i)cosβ , where i is the inclination of the orbital angular velocity v 9 P9 orbit relative to the ecliptic plane. β is the angle between the stellar and P9 paths on the sky plane. The later angle is inferred from direct images of the P9 and star trajectories on the sky, supposed to be taken in addition to the occultation observations. The orbital inclination i of P9 is inferred from the best fit of the perturbations of the Solar System small bodies by P9 (Brown & Batygin 2016). The angular size δ9 Dout P9 parallax π9 , through equations (4) to (6). The radius R9 itself can be derived from δ9 and from the can finally be derived from observables Docc , ω9 , ω* , Din , and . 3. Gravitational lensing Recently Philippov and Chobanu (2016) have proposed to detect P9 by gravitational deflection and lensing of the background star. Here we develop further numerical estimates and probability of occurence. 3.1 Gravitational amplification For a star at an apparent distance r from P9 at the planet distance, the essential parameter to evaluate the amplification is the Einstein radius RE9 RE9=√R S9 D9 of P9. It is given by 5 its Scharzschild radius GM9/c2 Here D9 is the P9 distance and RS9 of a point-like stellar source, at a distance r from the P9 center, the gravitational amplification AG is given by (Schneider 1989) AG(r)= s4 4 s4−R E . In the approximation where s=(r+ √r 2+ RE9 2 )/2 For a 30 M ⊕ mass and a 1000 AU distance, RS9=3.3 108 cm, while R E9 = 20 108 cm, or about R9 for a 3 R⊕ radius. Therefore, a large part of the gravitational amplification phase will be occulted by the planet. Nevertheless, when the star emerges from the planet disc at a distance r larger than R9, the flux increase is still not negligible: AG(r)−1~ 4RE9/r with a value AG about 4 at r=R9 The flux increase drops down to 10% at a distance r=r10=40R E9 Note that the gravitational amplification takes place at a detectable level larger than 10% even when the star is not occulted, as long as its line of sight passes at a distance less than about 40 R E9 times the occultation probability. from P9, i.e. about 40 R9. The corresponding probability of occurence is then 40 3.2 Deflection The gravitational deflection, given by α(r)=RS9/r for a line of sight at a distance r from P9, is for a M9=30 M ⊕ planet. For a background star at 1 arcsec of P9 at 1 arc sec for r=r10 1000 AU α is 110 µas. Even for very faint stars this kind of deflection will be detectable with large telescopes. Combined with gravitational lensing, the measurement of the deflection would constrain further the planet mass. 3.3 Time delay A time delay Δ T for the time of arrival of pulses from a background source has two origins : the « Shapiro » time delay Δ T E due to the Einstein effect of retardation of clocks and a geometrical effect Δ T Geom due to the bending of the light rays passing close to P9. They are given by Δ T E= 2GM9 c3 (1+ ln(rs D9 2 )) rs 6 and Δ T Geom= r SGM9 2r c3 where r is the impact parameter of the line of sight of a source at a distance rS . They are typically of the order of 10-10 sec. There is no known astrophysical source with pulses sufficiently short to make this kind of time delay measurable. Nevertheless, we present in the conclusion a configuration where these time delays will be measurable, namely by clocks on-bard an in-situ mission.. 3.4 Microlensing duration Reminding that b is the impact parameter of the stellar path (minimum distance of the star to P9 on the sky plane). For a velocity v of the apparent position of P9 on the sky plane (due to the Earth orbital motion), the time variation of the amplification AG(t) is still given by equation (1), where now s=s(t)=(√v2t 2+ b2+ √v2 t2+ b2+ R E9 2 )/2 4. Numerical estimates 4.1 Duration of the occultation and ingress/egress As we are essentially interested by orders of magnitudes, and since the P9 parallactic motion dominates over its orbital motion and the star's proper motion, we neglect the two latter. Then, for a R⊙ star at 100 pc with an angular diameter δ* = 0.5 mas, the ingress/egress duration Din/out The occultation duration, depending on the position of the Earth on its orbit at the time of occultaion will be 2R9/(30km/sec )/cos α⊕ , i.e. at least 1300 sec (for cosα⊕=1 ). is about 12.5 sec for an equatorial occultation. of an occultation by P9 depends on the duration of the survey and 4.2 Probability of occultations The a priori probability Pocc the surface density number of the sample of surveyed stars. The later depends on their limiting magnitudes and on the involved sky region. Pocc Pocc=N * A9 (5) where N * the area of the sky band swept by P9 during the survey. Since the parallactic motion δπ 9 dominates over the P9 orbital motion and the is given by is the number of stars per arcsec2 and A9 7 is about 40 mas and δπ9 is about 200 arc sec, so that A9 . For a 3 R⊕ planet is about 24 stellar proper motions, it is given, at first approximation, by A9=2δπ9 δ9 at 1000 AU, δ9 arcsec2. To refine our estimate of N * orbit. The analysis of the perturbations of Solar System small bodies can only give the orbital elements planet 9. According to Fieng ω = 150 deg and Ω = 113 deg. This dynamical analysis cannot give the present position of P9 on its orbit. Nevertheless a recent estimate by Millholland & Laughlin (2017) is RA = 40 ±10 deg and DEC = 0 ±20 deg. In this region, according to Allen (1976), the number of star brighter than V = 21 is 104 per square degree. , it is suitable to have an idea of the current localization of P9 on its The planet spans its sky band A9 in 6 months. It results from equation (5) that the probability of occultation of a V = 21 star in a 6 months survey is about 0.5 10-5. This very low probability forces to search for occultations of fainter stars. Applying the rule N (m)/N (n)=2m−n 10\% probability it is necessary to monitor V = 32 stars. For a subsequent 6 months survey, the background stars and P9 will have moved and the sky configuration refreshed, so that the probability of an occultation during a second 6 months survey is to be added to the previous one. More generally, for a N times 6 months survey, the occultation probability is 0.005 N , i.e. about 20% in a 10 year survey of V= 32 stars. Such faint stars will be detectable with the coming 30m- class telescopes. , to have a 5. Detection strategies One could a priori search "ab initio" for occultations and gravitational events (i.e. Amplification and deflection), prior to the detection of P9 by direct imaging. But, given that these events are short, with no prediction of occurence, hidden in a large portion of the sky and require large telescopes to achieve the required photometric precision, it is better to search for a direct image of P9 before onje of these events. Indeed, if a direct image is detected somewhere in the sky, it will be easy to extrapolate its apparent trajectory in the sky due to the Earth orbital motion. From there, given an appropriate knowledge of the field of background stars in a band A9=2δπ9∗δ10 δ10=40 δ9 is the angular distance from P9 where the flux excess due to microlensing exceeds 10%), it will be easy to predict the time of occurences of occultation, microlensing and and gravitational deflection. (where 8 6. Conclusion Direct detection of P9 and the detection of stellar occultations and gravitational effects by P9 are complementary. Their combination would allow to infer both the planet radius, mass and its albedo. In the future, once the planet P9 has been detected, if it exists, further observations will be interesting. The detection of a stellar occultation by P9 would have additional benefits. - It would be possible to investigate a possible atmosphere by spectroscopy of occultation with 30 meter class telescopes. - It would allow to detect rings (Barnes & Fortney 2004) and satellites (Sartoretti & Schneider 1999), thanks to the shape of the occultation light curve, and possibly surface irregularities ("mountains"), inferred from an asymmetry between the ingress and egress shapes.. High resolution spectroscopy of the planet would allow to determine its radial velocity and thus the eccentricity of its orbit. Indeed,for an eccentricity e, tthe radial velocity as seen from the barycenter of the solar system has amximum given by VR=e√GM ⊙/√(a(1−e2)) For a = 1000 AU and an eccentricy 0.6 it is about 600 m/sec detectable in the spectrum of the Sun reflected by P9 with moderate resolution spectrographs connected to large telescopes. Finally, needless to say, planet P9 would be an exquisite target for future in situ missions. One would not be able to explore in deep all the P9 characteristics with ground and space telescopes, but an in situ mission would investigate characteristics invisible from Earth, like the composition of its potentially rocky surface and its relief. It would also allow to measure the mass of P9 by the perturbation of the spacecraft trajectory and the measurement of the gravitational time delay of ultra-short on board pulses. And, given its large distance, it would be a useful precursor to consolidate the technology of future interstellar missions. References Allen, C. 1976 Astrophysical quantities. The Athlone Press. Section 117, p. 244. Assafin M., Camargo J., Vieira Martins R. et al. 2012. A&A 541, A142 Barnes, J. & Fortney, J. 2004. ApJ., 616, 1193 Batygin, K. & Brown, M. \2016 Astron. J., 151, 22 Brown, M. & Batygin, K. 2016. ApJ Let. 824, 2 Folkner W., Jacobsen R., Park R. & Williams J., 2016.. DPS Meeting #48 id.120.07 Linder, E. & Mordasini, C., 2016. A&A, submitted arxiv:1602.07465 9 Mallama, A., Krobusek, B. & Pavlov H. 2016. Icarus. accepted arxiv:1609.05048 Philippov, J. & Chobanu, M., 2016. Publ. Astron. Soc. Australia. 33, 1 Sartoretti, P. & Schneider, J., 1999. A&A Suppl. 134, 553 Schneider, J., 1989. A&A 214, 1 10
1809.10309	1	1809	2018-09-27T01:54:43	The 2016 Reactivations of Main-Belt Comets 238P/Read and 288P/(300163) 2006 VW139	[ "astro-ph.EP" ]	We report observations of the reactivations of main-belt comets 238P/Read and 288P/(300163) 2006 VW139, that also track the evolution of each object's activity over several months in 2016 and 2017. We additionally identify and analyze archival SDSS data showing 288P to be active in 2000, meaning that both 238P and 288P have now each been confirmed to be active near perihelion on three separate occasions. From data obtained of 288P from 2012-2015 when it appeared inactive, we find best-fit R-band H,G phase function parameters of H_R=16.80+/-0.12 mag and G_R=0.18+/-0.11, corresponding to effective component radii of r_c=0.80+/-0.04 km, assuming a binary system with equally-sized components. Fitting linear functions to ejected dust masses inferred for 238P and 288P soon after their observed reactivations in 2016, we find an initial average net dust production rate of 0.7+/-0.3 kg/s and a best-fit start date of 2016 March 11 (when the object was at a true anomaly of -63 deg) for 238P, and an initial average net dust production rate of 5.6+/-0.7 kg/s and a best-fit start date of 2016 August 5 (when the object was at a true anomaly of -27 deg) for 288P. Applying similar analyses to archival data, we find similar start points for previous active episodes for both objects, suggesting that minimal mantle growth or ice recession occurred between the active episodes in question. Some changes in dust production rates between active episodes are detected, however. More detailed dust modeling is suggested to further clarify the process of activity evolution in main-belt comets.	astro-ph.EP	astro-ph	Draft version 2021-07-20 Typeset using LATEX twocolumn style in AASTeX62 The 2016 Reactivations of Main-Belt Comets 238P/Read and 288P/(300163) 2006 VW139 ∗ Henry H. Hsieh,1, 2 Masateru Ishiguro,3 Yoonyoung Kim,3, 4 Matthew M. Knight,5 Zhong-Yi Lin,6 Marco Micheli,7, 8 Nicholas A. Moskovitz,9 Scott S. Sheppard,10 Audrey Thirouin,9 and Chadwick A. Trujillo11 1Planetary Science Institute, 1700 East Fort Lowell Rd., Suite 106, Tucson, AZ 85719, USA 2Institute of Astronomy and Astrophysics, Academia Sinica, P.O. Box 23-141, Taipei 10617, Taiwan 3Department of Physics and Astronomy, Seoul National University, Gwanak, Seoul 151-742, Korea 4Max Planck Institute for Solar System Research, Justus-von-Liebig-Weg 3, D-37077 Gottingen, Germany 5Department of Astronomy, University of Maryland, 1113 Physical Sciences Complex, Building 415, College Park, MD 20742, USA 6Institute of Astronomy, National Central University, Jhongli, Taiwan 7ESA SSA-NEO Coordination Centre, Largo Galileo Galilei, 1, 00044 Frascati (RM), Italy 8INAF - Osservatorio Astronomico di Roma, Via Frascati, 33, 00040 Monte Porzio Catone (RM), Italy 9Lowell Observatory, 1400 W. Mars Hill Rd, Flagstaff, AZ 86011, USA 10Department of Terrestrial Magnetism, Carnegie Institution for Science, 5241 Broad Branch Road NW, Washington, DC 20015, USA 11Department of Physics and Astronomy, Northern Arizona University, Flagstaff, AZ 86011, USA (Received 2018 August 27; Accepted 2018 September 26) Submitted to AJ ABSTRACT We report observations of the reactivations of main-belt comets 238P/Read and 288P/(300163) 2006 VW139, that also track the evolution of each object's activity over several months in 2016 and 2017. We additionally identify and analyze archival SDSS data showing 288P to be active in 2000, meaning that both 238P and 288P have now each been confirmed to be active near perihelion on three separate occasions. From data obtained of 288P from 2012-2015 when it appeared inactive, we find best-fit R-band H, G phase function parameters of HR = 16.80 ± 0.12 mag and GR = 0.18 ± 0.11, corresponding to effective component radii of rc = 0.80 ± 0.04 km, assuming a binary system with equally-sized components. Fitting linear functions to ejected dust masses inferred for 238P and 288P soon after their observed reactivations in 2016, we find an initial average net dust production rate Md = 0.7 ± 0.3 kg s−1 and a best-fit start date of 2016 March 11 (when the object was at a true of Md = 5.6±0.7 kg s−1 anomaly of ν = −63◦) for 238P, and an initial average net dust production rate of and a best-fit start date of 2016 August 5 (when the object was at ν = −27◦) for 288P. Applying similar analyses to archival data, we find similar start points for previous active episodes for both objects, suggesting that minimal mantle growth or ice recession occurred between the active episodes in question. Some changes in dust production rates between active episodes are detected, however. More detailed dust modeling is suggested to further clarify the process of activity evolution in main-belt comets. 8 1 0 2 p e S 7 2 . ] P E h p - o r t s a [ 1 v 9 0 3 0 1 . 9 0 8 1 : v i X r a Corresponding author: Henry H. Hsieh [email protected] ∗ Based on observations obtained with MegaPrime/MegaCam, a joint project of CFHT and CEA/DAPNIA, at the Canada- France-Hawaii Telescope (CFHT) (Programs 12BH43, 15AT05, and 16BT05), which is operated by the National Research Council (NRC) of Canada, the Institut National des Science de l'Univers of the Centre National de la Recherche Scientifique (CNRS) of France, and the University of Hawaii, and at the Gemini Obser- vatory (Programs GN-2011B-Q-17, GN-2012A-Q-68, GN-2012B- Q-106, GN-2016B-LP-11, and GS-2016B-LP-11), which is oper- ated by the Association of Universities for Research in Astron- omy, Inc., under a cooperative agreement with the National Sci- ence Foundation (NSF) on behalf of the Gemini partnership: the NSF (United States), the National Research Council (Canada), CONICYT (Chile), Ministerio de Ciencia, Tecnolog´ıa e Innovaci´on Productiva (Argentina), and Minist´erio da Ciencia, Tecnologia e Inova¸c¯ao (Brazil). 2 Hsieh et al. Keywords: asteroids -- comets, nucleus -- comets, dust 1. INTRODUCTION 1.1. Background Main-belt comets (MBCs; Hsieh & Jewitt 2006) are objects that orbit in the main asteroid belt but also ex- hibit comet-like activity that has been determined to at least be partially due to the sublimation of volatile ices, typically assumed to be water ice. Exposed water ice is thermally unstable against sublimation at the temper- atures found in the main asteroid belt, and thus is not expected to remain very long on the surface of main- belt asteroids. Thermal modeling has shown, however, that buried water ice can survive within short distances of asteroid surfaces over the age of the solar system (Schorghofer 2008, 2016; Prialnik & Rosenberg 2009), where it might then be susceptible to excavation and exposure to solar heating by small impacts (e.g., Hsieh et al. 2004; Hsieh 2009; Capria et al. 2012; Haghighipour et al. 2016). While mantling and depletion of volatiles are expected to eventually extinguish activity triggered by an individual impact, an active MBC may nonethe- less be expected to remain active over several orbit pas- sages after its initial activation (cf. Hsieh et al. 2015a). MBCs are a subset of the active asteroids (cf. Jewitt et al. 2015), which include all small bodies that ex- hibit comet-like activity yet occupy asteroid-like orbits. Small solar system bodies are commonly considered dy- namically asteroidal if they have Tisserand parameter values (with respect to Jupiter) of TJ > 3 (Kres´ak 1979), although in practice, the dynamical transition zone between asteroids and comets actually appears to lie roughly between TJ = 3.05 and TJ = 3.10 (Tancredi 2014; Jewitt et al. 2015; Hsieh & Haghighipour 2016). Besides MBCs, the other major type of active asteroids are disrupted asteroids (cf. Hsieh et al. 2012a), which ex- hibit comet-like activity due to effects other than subli- mation, such as impact-driven dust ejection or rotational destabilization. Although there is a non-zero chance that some MBCs (particularly those with both large eccentricities and large inclinations) may be implanted Jupiter-family comets (JFCs) originally from the outer solar system, most MBCs appear to be native to the main asteroid belt (Hsieh & Haghighipour 2016). Thus, it may be possible to use MBCs to set temperature and composi- tional constraints on this region of the early solar system (Hsieh 2016). MBCs are intriguing because they offer new opportunities to constrain models of our solar sys- tem's formation and evolution (cf. Hsieh 2014a), and also test hypotheses that icy objects from the region of the solar system currently occupied by the present-day main asteroid belt may have played a significant role in the primordial delivery of water and other volatiles to the early Earth (Morbidelli et al. 2000; Raymond et al. 2004; Raymond & Izidoro 2017; O'Brien et al. 2006, 2018, and references within). 1.2. Recurrent Activity in MBCs Recurrent activity, especially occurring near perihe- lion with intervening periods of inactivity, is a strong indication that an active asteroid's activity is driven by sublimation, making it key to differentiating between MBCs and disrupted asteroids. Repeated impacts or rotational destabilization events are not expected to oc- cur on the same asteroid on timescales as short as a single main-belt asteroid orbit (i.e., 5-6 years), nor are they expected to exhibit periodic behavior correlated with an object's orbital position. On the other hand, re- peated activity near perihelion is naturally explained by sublimation-driven active behavior where such activity is correlated with the higher temperatures experienced by an object during perihelion passages (Hsieh et al. 2012a). This conclusion is further strengthened by the trend noted by Hsieh & Sheppard (2015) that the active episodes of all nine active asteroids identified as MBCs thus far via other analyses (e.g., dust modeling or photo- metric monitoring) all occur near perihelion (i.e., within true anomaly ranges of −60◦ (cid:46) ν (cid:46) 120◦, where ν = 0◦ at perihelion). This finding supports the idea that their active episodes are correlated with tempera- ture, as would be expected if they were due to subli- mation, and do not occur at arbitrary orbit positions, as would be expected if they were due to impact or ro- tational disruption. To date, seven MBCs have been confirmed to be recurrently active: 133P/Elst-Pizarro (Elst et al. 1996; Hsieh et al. 2004), 238P/Read (Hsieh et al. 2009, 2011), 288P (Hsieh et al. 2012b; Agarwal et al. 2016a), 259P/Garradd (Jewitt et al. 2009; Hsieh & Chavez 2017), 313P/Gibbs (Hsieh et al. 2015b; Hui & Jewitt 2015), 324P/La Sagra (Hsieh et al. 2012c; Hsieh & Sheppard 2015), and 358P/PANSTARRS (Hsieh et al. 2018a). Searching for recurrent activity is also important be- cause most MBCs are discovered while they are already active, and even those that were already known as inac- tive asteroids before being discovered to be active (e.g., 133P, 176P, 288P) were not being regularly monitored Reactivations of 238P and 288P 3 leading up to the discovery of their activity. As such, the onset times of MBC active periods are generally poorly constrained. By performing an observational monitor- ing campaign starting well in advance of the expected return of activity in a MBC candidate, more explicit constraints can be placed on the onset time of that ac- tivity. Coupled with analogous observations of the de- cline of activity, constraints on the onset time of activity can in turn allow stronger constraints to be placed on physical quantities of interest such as the total dura- tion of activity, total mass loss per active episode, and the depth of the buried ice presumed to drive the ob- served activity. Characterizing the evolution of activ- ity strength over multiple active episodes may also help constrain total active lifetimes following each triggering event on a MBC, which in turn may give greater context to discovery statistics (cf. Section 4; Hsieh 2009; Hsieh et al. 2015a). Given the above considerations, we have conducted observational campaigns to search for and characterize the return of activity for MBCs 238P and 288P, and report the results in this work. 1.3. 238P/Read Comet 238P/Read (formerly designated P/2005 U1) was discovered to be an active comet on UT 2005 Oc- tober 24 (Read et al. 2005), shortly after perihelion. It was one of the first three objects that led to the recogni- tion of MBCs as a new class of comets (Hsieh & Jewitt 2006). The absolute magnitude of the object's nucleus has been measured to be HR = 19.05±0.05 mag, corre- sponding to an effective nucleus radius of rn ∼ 0.4 km, assuming an albedo of pR = 0.05 (Hsieh et al. 2011). An initial analysis of 238P's activity employing Finson-Probstein-style dust modeling (Finson & Prob- stein 1968) showed that it was most likely driven by ice sublimation (Hsieh et al. 2009). This conclusion was strongly supported by the 2010 appearance of recurrent activity leading up to the object's next perihelion pas- sage in 2011 (Hsieh et al. 2011). Dust modeling of the 2005 active episode indicated that the coma and tail were optically dominated by dust particles >10 µm in radius with terminal ejection velocities of 0.2-3 m s−1, and that the object's average mass loss rate during the period of observation was approximately 0.2 kg s−1 (Hsieh et al. 2009). A photometric analysis showed indi- cations that the object's activity in 2010-2011 may have been comparable in strength to the activity observed in 2005 (Hsieh et al. 2011). However, the observing pe- riods being compared in that analysis did not actually overlap, and as such, no definitive conclusions could be drawn at the time from the available data about the change (or lack thereof) in activity strength from the first epoch to the next. Notably, 238P was also the primary target of the pro- posed NASA Discovery mission, Proteus, to visit and physically characterize a MBC and its activity (Meech & Castillo-Rogez 2015). 1.4. 288P/(300163) 2006 VW139 Comet 288P is also known by its asteroid designa- tion, (300163) 2006 VW139, by which it was already known at the time of the discovery of its activity in 2011 (Hsieh et al. 2012b). It was the first of now nu- merous active asteroids to be discovered to date by the Pan-STARRS1 (PS1) survey telescope (Chambers et al. 2016). Following its discovery, deep follow-up observa- tions showed a short antisolar dust tail and a longer dust trail aligned with the object's orbit plane, indica- tive of the simultaneous presence of recent dust emission (in the antisolar dust tail) and older dust emission (in the orbit-plane-aligned dust trail), and therefore of a prolonged dust emission event. A photometric analy- sis showed the object maintaining a roughly constant coma brightness for about a month, also suggesting the action of a sustained dust emission event, characteristic of sublimation-driven activity, and not, for instance, of an impulsive driver such as an impact. Spectro- scopic searches for CN emission were unsuccessful, al- though did provide upper limit production rates. Hsieh et al. (2012b) found QCN < 1.3×1024 molecules s−1, roughly equivalent to an upper limit water produc- tion rate of QH2O < 1026 molecules s−1, assuming Jupiter-family comet-like chemical composition ratios, which we note is not necessarily a valid assumption for MBCs. Meanwhile, (Licandro et al. 2013) found QCN < 3.8×1023 molecules s−1. Dust modeling analyses of the object's 2011 activity found an average dust production rate of ∼ 0.2 kg s−1 and typical ejection velocities of ∼ 0.1-0.3 m s−1 (Lican- dro et al. 2013; Agarwal et al. 2016b). At the time, the best estimate of the absolute magnitude of the object's nucleus was HR = 16.4 mag, corresponding to an effec- tive nucleus radius of rn ∼ 1.4 km, assuming an albedo of pR = 0.05 (Hsieh et al. 2012b). A later analysis of Hub- ble Space Telescope data indicated that the nucleus had an absolute magnitude of HV = 17.0±0.1 mag (Agarwal et al. 2016b), equivalent to HR ∼ 16.6 mag, assuming so- lar colors. The nucleus has been spectroscopically clas- sified as a C-type asteroid (Licandro et al. 2013), and has been identified as a binary system (Agarwal et al. 2017). Interestingly, 288P was found to potentially belong to a small, young 11-member cluster of asteroids just Hsieh et al. 4 7.5±0.3 Myr old (Novakovi´c et al. 2012). Together with the discovery that fellow MBC 133P belonged to the <10 Myr-old Beagle family (Nesvorn´y et al. 2008), this finding suggests that the younger (and therefore poten- tially more volatile-rich) surfaces of members of these families could be more susceptible to activation by small impactors excavating shallow buried ice and thus be more likely to develop observable activity (e.g., Hsieh et al. 2018b). As such, Novakovi´c et al. (2012) hypoth- esized both that currently known young families could be productive regions in which to search for new MBCs, and that currently known MBCs might be found in the future to be part of as-yet undiscovered young families. Observations obtained with the Hubble Space Tele- scope on 2016 August 22 revealed that 288P had again become active following its observed 2011-2012 active episode (Agarwal et al. 2016a). With these observations, 288P became the fifth MBC to be confirmed to exhibit recurrent activity, after 133P, 238P, 313P, and 324P. Since these observations, recurrent activity in 259P and 358P has also been confirmed, bringing the current to- tal of MBCs confirmed to exhibit recurrent activity to seven. telescope on Maunakea in Hawaii, the 8.1 m Gemini South (Gemini-S) telescope at Cerro Pachon in Chile, the 6.5 m Baade Magellan telescope at Las Campanas in Chile, Lowell Observatory's 4.3 m Discovery Chan- nel Telescope (DCT) at Happy Jack, Arizona, the 2.5 m Sloan Digital Sky Survey (SDSS) telescope at Apache Point Observatory in New Mexico, and the Lulin One- meter Telescope (LOT) at Lulin Observatory in Taiwan. We employed the Gemini Multi-Object Spectrographs (GMOS; Hook et al. 2004; Gimeno et al. 2016) and Sloan r(cid:48)-band filters for Gemini-N and Gemini-S observations, MegaCam (Boulade et al. 2003) and a Sloan r(cid:48)-band filter for CFHT observations, a 2048×2048 pixel Tex- tronix CCD and a Kron-Cousins R-band filter for UH 2.2 m observations, the Inamori Magellan Areal Cam- era and Spectrograph (IMACS; Dressler et al. 2011) and a Sloan r(cid:48)-band filter for Baade observations, the Large Monolithic Imager (Bida et al. 2014) and a Kron- Cousins R-band filter for DCT observations, and a Ver- sArray:1300B CCD (Kinoshita et al. 2005) and a Bessell- like R-band filter for LOT observations. SDSS data pre- sented here were obtained using a large-format mosaic CCD camera designed for the SDSS survey (Gunn et al. 1998) and a Sloan r(cid:48)-band filter. Non-sidereal track- ing was used for all targeted observations, while sidereal tracking was used for SDSS survey observations. Observations of 238P while it was active in 2011 were obtained using Gemini-N (Program GN-2011B- Q-17), and represent a continuation of the observing campaign previously described in Hsieh et al. (2011). Observations of 238P over several months while it was most recently active in 2016 and 2017 were obtained by Gemini-N (Program GN-2016B-LP-11), CFHT (Pro- gram 16BT05), LOT, and DCT. Details of these obser- vations are listed in Table 1. Orbit positions of both the observations reported here and previously reported ob- servations (Table 2; Hsieh et al. 2009, 2011) are marked in Figure 1. Composite images of the object during each night of newly reported observations are shown in Fig- ure 2. Meanwhile, observations of 288P while it was active in 2012 were obtained by Gemini-N (Program GN-2012A- Q-68), and represent a continuation of the observing campaign previously described in Hsieh et al. (2012b). Since then, we also obtained observations of 288P while it was inactive in 2012 with Gemini-N (Program GN- 2012B-Q-106), in 2012 and 2013 with the UH 2.2 m telescope, and in 2013 and 2015 with CFHT (Programs 12BH43 and 15AT05) as part of a campaign to char- acterize its nucleus, and also obtained observations of the object while it was most recently active in 2016 and 2017 using Gemini-N and Gemini-S (Programs GN- 2. OBSERVATIONS Figure 1. Orbit position plot with the Sun (black dot) at the center, the orbits of Mercury, Venus, Earth, Mars, 238P, and Jupiter marked by thin black lines, and the orbit of 238P marked by a thick black line. Perihelion (P) and aphelion (A) are marked with crosses. Green diamonds mark positions of observations when 238P was active in 2005-2007, open squares mark positions of observations when 238P was apparently inactive in 2010, yellow squares mark positions of observations when 238P was active in 2010-2011, and blue circles mark positions of observations when 238P was active in 2016-2017. Observations of 238P and 288P presented here were obtained with the 8.1 m Gemini North (Gemini-N) telescope, the 3.54 m Canada-France-Hawaii Telescope (CFHT), and the University of Hawaii (UH) 2.2 m Reactivations of 238P and 288P 5 Table 1. Observation Log: 238P - Active αg 8.1 CFHT LOT DCT j HR,t -- k Md -- tc Filter Re ∆f i mR,t -- LOT CFHT h mR,n -- CFHT LOT CFHT CFHT UT Date Tel.a N b 360 900 1620 900 900 900 900 1500 900 900 875 14 4200 7 Af ρl νd -- 0.0 2.361 3.276 4.2±0.6 49.8 2.543 2.751 21.5 21.97±0.05 21.9±0.1 16.4±0.1 (1.8±0.2)×107 r(cid:48) 2.9±0.4 56.7 2.598 2.471 22.7 22.25±0.05 21.9±0.1 16.5±0.1 (1.6±0.2)×107 r(cid:48) 1.1±0.2 1.8 21.54±0.05 21.2±0.1 17.3±0.1 (0.6±0.1)×107 r(cid:48) 79.2 2.825 1.844 2.0±0.3 r(cid:48) −31.5 2.439 2.095 24.4 22.32±0.05 22.3±0.1 17.3±0.1 (0.7±0.1)×107 3.0±0.4 r(cid:48) −23.1 2.405 1.742 21.6 21.60±0.05 21.5±0.1 17.1±0.1 (0.9±0.1)×107 4.1±0.5 r(cid:48) −14.2 2.381 1.467 13.1 20.69±0.05 20.3±0.1 16.6±0.1 (1.4±0.2)×107 3.7±0.5 r(cid:48) −13.9 2.380 1.461 12.7 20.76±0.05 20.3±0.1 16.7±0.1 (1.4±0.2)×107 4.3±0.5 5.1 20.14±0.05 20.0±0.1 16.9±0.1 (1.1±0.1)×107 −8.8 2.372 1.382 r(cid:48) 4.4±0.5 4.3 20.06±0.05 19.7±0.1 16.7±0.1 (1.4±0.1)×107 −8.2 2.371 1.377 r(cid:48) 4.3±0.5 3.8 20.05±0.05 19.7±0.1 16.7±0.1 (1.3±0.1)×107 −7.9 2.371 1.375 r(cid:48) 4.2±0.5 3.3 20.03±0.05 19.7±0.1 16.7±0.1 (1.2±0.1)×107 −7.6 2.370 1.374 r(cid:48) 4.0±0.4 0.6 19.79±0.05 19.7±0.1 17.0±0.1 (0.9±0.1)×107 −5.7 2.369 1.368 r(cid:48) 5.0±0.5 2.8 19.83±0.05 19.4±0.1 16.5±0.1 (1.6±0.2)×107 −4.0 2.367 1.372 1500 R 9.1 0.0 2.366 1.415 8.4±0.9 0.9 2.366 1.428 10.3 19.83±0.05 19.7±0.1 16.2±0.1 (2.1±0.2)×107 r(cid:48) 9.5±1.1 4.2 2.367 1.499 14.6 19.97±0.05 19.9±0.1 16.1±0.1 (2.3±0.3)×107 r(cid:48) 18.3 2.390 2.009 23.9 20.70±0.05 20.5±0.1 15.7±0.1 (3.6±0.5)×107 9.6±1.3 r(cid:48) 9.7±1.3 19.5 2.394 2.061 24.0 20.73±0.05 20.5±0.1 15.6±0.1 (3.8±0.5)×107 r(cid:48) 20.3 2.396 2.100 24.1 20.64±0.05 20.4±0.1 15.5±0.1 (4.4±0.6)×107 10.9±1.4 r(cid:48) 20.3 2.396 2.100 24.1 20.48±0.05 20.4±0.1 15.5±0.1 (4.4±0.6)×107 12.7±1.7 r(cid:48) 26.2 2.416 2.367 23.7 20.70±0.05 20.6±0.1 15.4±0.1 (4.6±0.6)×107 11.7±1.5 r(cid:48) 28.5 2.425 2.473 23.2 20.99±0.05 20.7±0.1 15.4±0.1 (4.6±0.6)×107 9.2±1.2 r(cid:48) 0.0 2.369 3.024 16.7 2011 Mar 10 Perihelion ................. 2011 Aug 29 Gemini-N 2 2011 Sep 25 Gemini-N 5 2011 Dec 31 Gemini-N 9 2016 Jul 8 5 2016 Aug 6 Gemini-N 5 2016 Sep 5 Gemini-N 5 5 2016 Sep 6 5 2016 Sep 23 2016 Sep 25 5 2016 Sep 26 5 2016 Sep 27 Gemini-N 5 2016 Oct 3 2016 Oct 9 2016 Oct 22 Perihelion ................. 5 2016 Oct 25 2016 Nov 5 3 2016 Dec 22 Gemini-N 3 2016 Dec 26 7 2016 Dec 29 Gemini-N 5 2016 Dec 29 4 2017 Jan 18 Gemini-N 1 2017 Jan 26 Gemini-N 5 2022 Jun 5 Perihelion ................. a Telescope used. b Number of exposures. c Total integration time, in seconds. d True anomaly, in degrees. e Heliocentric distance, in au. f Geocentric distance, in au. g Solar phase angle (Sun-object-Earth), in degrees. h Equivalent mean apparent R-band nucleus magnitude, measured within photometry apertures with radii of 4.(cid:48)(cid:48)0. i Equivalent total mean apparent R-band magnitude, including the entire coma and tail, if present. j Total absolute R-band magnitude, using H, G phase function where G = −0.03. k Estimated total dust mass, in kg, assuming ρd ∼ 2500 kg m3. l Af ρ values computed using photometry apertures with radii of 4.(cid:48)(cid:48)0, in cm, where uncertainties are estimated to be ∼10%. 1500 540 540 1260 900 720 180 900 -- -- -- -- -- -- CFHT CFHT -- -- -- -- 2016B-LP-11 and GS-2016B-LP-11), CFHT (Program 16BT05), Magellan, and LOT. In addition, using the Solar System Object Image Search tool1 (SSOIS; Gwyn et al. 2012), provided by the Canadian Astronomical Data Centre, we also identified precovery observations obtained in 2000 by the SDSS survey (Abazajian et al. 2009; York et al. 2000; Fukugita et al. 1996; Gunn et al. 1998, 2006; Aihara et al. 2011) in which the object ap- peared to be active. Details of these observations of 288P are listed in Tables 3 and 4. Orbit positions of both the observations reported here and previously re- ported observations (Table 5; Hsieh et al. 2012b) are marked in Figure 3. Composite images of the object during each night of observations when it was active are shown in Figure 4, while composite images of the object during each night of observations when it was inactive are shown in Figure 5. 1 http://www.cadc-ccda.hia-iha.nrc-cnrc.gc.ca/en/ssois/ We performed standard bias subtraction and flat-field reduction (using dithered images of the twilight sky) for all data from targeted observations, except those from CFHT, using IRAF software (Tody 1986, 1993). Reduction of CFHT data was performed by the Elixir pipeline (Magnier & Cuillandre 2004). Photometric cal- ibration of UH 2.2 m data was accomplished using Lan- dolt (1992) standard stars and field stars, for which net fluxes were measured within circular apertures, with background sampled from surrounding circular annuli. For Gemini-N, Gemini-S, CFHT, Magellan, DCT, and LOT data, absolute photometric calibration was accom- plished using field star magnitudes from SDSS or Pan- STARRS1 field star catalogs (Aihara et al. 2011; Schlafly et al. 2012; Tonry et al. 2012; Magnier et al. 2013, 2016; Flewelling et al. 2016). Conversion of r(cid:48)-band Gemini and PS1 photometry to R-band was accomplished using transformations derived by Tonry et al. (2012) and by R. Lupton (http://www.sdss.org/). Comet photometry was performed using circular apertures with 4.(cid:48)(cid:48)0 radii, where background statistics were measured in nearby, 6 Hsieh et al. Table 2. Previously reported observations of 238P when activea Gemini-N j Md -- Perihelion ... Tel.b UH2.2 UH2.2 UH2.2 UH2.2 UH2.2 UH2.2 UH2.2 UH2.2 NTT NTT Keck UH2.2 UH2.2 g mR,n -- mR,t -- h HR,t -- i αf 25.2 0.5 3.8 4.3 7.3 7.8 9.5 17.1 19.9 10.7 11.0 11.4 20.3 23.5 22.5 8.1 Rd 2.365 2.437 2.449 2.450 2.452 2.453 2.459 2.504 2.506 2.576 2.574 2.572 2.514 2.433 2.416 2.361 ∆e 2.276 1.447 1.469 1.473 1.477 1.481 1.501 1.743 1.754 1.643 1.647 1.651 1.869 2.414 2.566 3.276 νc 0.0 31.4 33.9 34.2 34.5 34.8 35.9 43.6 43.9 −54.1 −53.9 −53.6 −45.7 −31.5 −27.5 0.0 19.28±0.05 <19.3 <16.4 >(1.7±0.2)×107 19.34±0.05 <19.3 <16.1 >(2.4±0.2)×107 19.46±0.05 <19.5 <16.0 >(2.4±0.3)×107 19.37±0.05 <19.4 <16.1 >(2.4±0.2)×107 19.28±0.05 <19.3 <15.9 >(2.7±0.3)×107 19.72±0.05 <19.7 <16.2 >(2.1±0.2)×107 20.12±0.05 <20.1 <15.8 >(3.3±0.4)×107 20.16±0.05 <20.2 <15.8 >(3.1±0.4)×107 22.0±0.1 <22.0 <18.0 >(0.3±0.1)×107 <22.3 <18.3 >(0.2±0.1)×107 22.3±0.1 <22.3 <18.3 >(0.2±0.1)×107 22.3±0.1 <22.3 <17.7 >(0.4±0.1)×107 22.3±0.1 21.8±0.1 <21.8 <16.6 >(1.5±0.2)×107 21.9±0.1 <21.9 <16.6 >(1.5±0.2)×107 -- UT Date 2005 Jul 27 2005 Nov 10 2005 Nov 19 2005 Nov 20 2005 Nov 21 2005 Nov 22 2005 Nov 26 2005 Dec 24 2005 Dec 25 2010 Sep 3 2010 Sep 4 2010 Sep 5 2010 Oct 5 2010 Nov 25 2010 Dec 9 2011 Mar 10 Perihelion ... a All 2005 data from Hsieh et al. (2009), and all 2010 data from Hsieh et al. (2011). b Telescope (UH2.2: UH 2.2 m telescope; NTT: New Technology Telescope; Keck: Keck I Observatory). c True anomaly, in degrees. d Heliocentric distance, in au. e Geocentric distance, in au. f Solar phase angle (Sun-object-Earth), in degrees. g Reported mean apparent R-band nucleus magnitude. h Equivalent total apparent R-band magnitude, including the entire coma and tail, if present. i Total absolute R-band magnitude, using H, G phase function, where G = −0.03. j Estimated total dust mass, in kg, assuming ρd ∼ 2500 kg m3. -- -- -- but non-adjacent, regions of blank sky to avoid dust contamination from the comet. To maximize signal-to- noise ratios of cometary features, we constructed com- posite images of the object for each night of data by shifting and aligning individual images on the object's photocenter using linear interpolation and then adding them together. In addition to performing nucleus photometry, we also measured the total flux from each object in our compos- ite images from each night when they were observed to be active. We did so by using rectangular photometry apertures enclosing the entire visible dust cloud and ori- ented to avoid field star contamination. Background sky levels were then measured from nearby areas of blank sky and subtracted to obtain net fluxes. These net fluxes were then calibrated using standard stars or field stars from SDSS or PS1 to obtain absolute photometry. 3. RESULTS AND ANALYSIS 3.1. 288P Phase Function Analysis To enable detailed analysis of 288P's activity, we first need to characterize the properties of its nucleus, specif- ically its phase function, from which we can derive its size and expected brightnesses at given viewing geome- tries (where 238P's nucleus phase function has already been characterized by Hsieh et al. 2011). In order to determine the phase function of 288P's nucleus, we only consider photometric data obtained for the object when no visible activity is detected in stacked composite im- ages (Figure 5) and the orbit position of the object is also sufficiently far from perihelion that significant activity is not expected, i.e., data from late 2012 to 2015. We nor- malize apparent magnitudes, m(R, ∆, α), meeting these criteria to unit heliocentric and geocentric distances, R and ∆, respectively, where α is the solar phase angle, to obtain reduced magnitudes, m(1, 1, α), using m(1, 1, α) = m(R, ∆, α) − 5 log(R∆) (1) The majority of our observations were short-duration "snapshot" observations at unknown rotational phases. For this analysis, we assume that the sparsely sampled nature of our data means that various deviations of our photometric data from the rotationally averaged bright- ness of the nucleus (i.e., the "mid-point" of its rota- tional lightcurve) at the times of our observations aver- age to zero, allowing us to fit a phase function solution to our data that reflects that rotationally averaged nucleus brightness. To account for nights when at least partial lightcurves were obtained (i.e., where some photometric variation is clearly present), we compute the uncertainty, σm, for the average magnitude of each night's observa- tions using ∆mexp − ∆mobs σm = 2 (2) where ∆mexp is the expected or assumed total photo- metric range and ∆mobs is the observed photometric range. Waniak & Drahus (2016) have measured the lightcurve of 288P to have an amplitude (i.e., peak to Reactivations of 238P and 288P 7 Figure 2. Composite R-band or r(cid:48)-band images of 238P (at the center of each panel) constructed from data listed in Table 1. All panels are 30(cid:48)(cid:48) × 30(cid:48)(cid:48) in size, with north (N), east (E), the antisolar direction (−(cid:12)), and the negative heliocentric velocity vector (−v), as projected on the sky, marked. Panels are also labeled with dates of observations in YYYY-MM-DD format, as well as the telescope used to obtain each observation (CFHT: Canada-France-Hawaii Telescope; DCT: Discovery Channel Telescope; GN: Gemini-N; LOT: Lulin One-meter Telescope). midpoint) of 0.4 mag, and as such, we use ∆mexp = 0.8 mag as the expected total photometric range for this analysis. We use these magnitude uncertainties to compute weighted average magnitudes (mR,mid(1, 1, α) in Ta- ble 4) over the time periods in which we are interested (where in all cases, this assumed rotational uncertainty dominates the photometric uncertainties for all of our data), and then perform a least-squares fit to this data to obtain the best-fit phase function. We find best-fit phase function parameters for 288P's nucleus of HR = 16.80±0.12 mag and GR = 0.18±0.11. Assuming solar colors (i.e., V − R = 0.36), these re- sults correspond to a V -band absolute magnitude of HV ∼ 17.16 ± 0.12 mag, consistent with the results of Agarwal et al. (2016b) (who assumed G = 0.15 and did not perform a full phase function fit) within uncer- tainties. Assuming a R-band albedo of pR = 0.05, this absolute magnitude corresponds to an effective nucleus radius of rN = 1.13 ± 0.06 km or, assuming 288P's nu- cleus to be a binary system with approximately equally Figure 3. Orbit position plot with the Sun (black dot) at the center, the orbits of Mercury, Venus, Earth, Mars, 238P, and Jupiter marked by thin black lines, and the orbit of 288P marked by a thick black line. Perihelion (P) and aphelion (A) are marked with crosses. Green diamonds mark positions of observations when 288P was active in 2000, yellow squares mark positions of observations when 288P was active in 2011- 2012, open circles mark positions of observations when 288P was apparently inactive in 2012-2015, and blue circles mark positions of observations when 288P was active in 2016-2017. 8 Hsieh et al. Figure 4. Composite R-band or r(cid:48)-band images of 288P (at the center of each panel) constructed from data listed in Table 3. All panels are 30(cid:48)(cid:48) × 30(cid:48)(cid:48) in size, with north (N), east (E), the antisolar direction (−(cid:12)), and the negative heliocentric velocity vector (−v), as projected on the sky, marked. Panels are also labeled with dates of observations in YYYY-MM-DD format, as well as the telescope used to obtain each observation (CFHT: Canada-France-Hawaii Telescope; GN: Gemini-N; GS: Gemini-S; LOT: Lulin One-meter Telescope; Mgln: Magellan Baade telescope; SDSS: Sloan Digital Sky Survey telescope). sized components (Agarwal et al. 2017), effective com- ponent radii of rc = 0.80 ± 0.04 km each. 3.2. Overview of Photometric Activity Characterization We compute absolute total R-band magnitudes, HR,t, for 238P and 288P (i.e., at R = ∆ = 1 au and α = 0◦) from measured apparent total magnitudes, mR,t, for each set of observations (Tables 1, 4, and 3), assum- ing inverse-square law fading and H, G phase functions with G =−0.03 for 238P (Hsieh et al. 2011) and G = 0.18 for 288P (Section 3.1) as computed for each object's nucleus. Emitted dust may not have the same photo- metric behavior as the nucleus though, making these assumptions a source of uncertainty. From the observed photometric excesses above the expected brightness of each object's inactive nucleus (using HR = 19.05 mag for 238P and HR = 16.80 mag for 288P; Section 3.1 of this work; Hsieh et al. 2011), we then estimate the total mass, Md, of visible ejected dust using (cid:18) 1 − 100.4(HR,t−HR) (cid:19) (cf. Hsieh 2014b), where rN = 0.4 km is the estimated effective nucleus radius for 238P (Hsieh et al. 2011) and rc = 0.8 km is the estimated effective radius for each component of 288P's binary nucleus (Section 3.1), assuming approximately equally sized components (cf. Agarwal et al. 2017). We assume dust grain densities of ρd = 2500 kg m−3, consistent with CI and CM carbonaceous chondrites, which are associated with primitive C-type objects like the MBCs (Britt et al. 2002), and effective mean dust grain radii of ¯a = 1 mm (assuming power-law particle size distributions from µm- to cm-sized particles deter- mined from dust modeling of other MBCs; Moreno et al. 2011; Hsieh 2014b). For reference, we also compute and report Af ρ values (A'Hearn et al. 1984), although we note that this parameter is not always a reliable mea- surement of the dust contribution to comet photometry in cases of non-spherically symmetric comae (e.g., Fink & Rubin 2012). The results of these calculations are shown in Tables 1 and 3. Md = 4 3 πr2 N ¯aρd 100.4(HR,t−HR) (3) 3.3. 238P Activity Characterization Reactivations of 238P and 288P 9 Figure 5. Composite R-band or r(cid:48)-band images of 288P (at the center of each panel) constructed from data listed in Table 4. All panels are 30(cid:48)(cid:48) × 30(cid:48)(cid:48) in size, with north (N), east (E), the antisolar direction (−(cid:12)), and the negative heliocentric velocity vector (−v), as projected on the sky, marked. Panels are also labeled with dates of observations in YYYY-MM-DD format, as well as the telescope used to obtain each observation (CFHT: Canada-France-Hawaii Telescope; GN: Gemini-N; UH88: UH 2.2 m telescope). To compare 238P's activity in 2016-2017 to previous active epochs, we compute Md and Af ρ for previously reported observations of 238P from Hsieh et al. (2009) and Hsieh et al. (2011). Total magnitudes of 238P (i.e., including the entire coma and tail) were not measured for these data in the same way as we have measured recent data, and as such, we report the equivalent mea- sured dust masses as lower limits in Table 2. We plot total absolute R-band magnitudes and equivalent esti- mated total dust masses for 238P in Figure 7. We fit a linear function to a portion of the data ob- tained during the object's 2016-2017 active period, aim- Md, ing to estimate the initial net dust production rate, over this period as well as the approximate onset time of the observed activity. Specifically, we fit data obtained from 2016 July 8 to 2016 October 9 (where equivalent R and ν ranges are listed in Table 6), when both a rea- sonable number of data points is available for fitting purposes and measured excess dust masses appear to increase approximately linearly. Following the calcu- lations detailed by Hsieh et al. (2015a), we find that the heliocentric distance change over this period corre- sponds to a ∼9% -- 28% increase in the water sublimation rate on the object's surface, depending on whether the subsolar or isothermal approximation is assumed. The resulting dust production rate and corresponding activ- ity start date we find are of course subject to numerous sources of uncertainty including the nonlinearity of the actual dust production rate as a function of heliocen- tric distance, ordinary photometric calibration uncer- tainties, uncertainties specifically associated with mea- suring extended objects (e.g., selection of optimal pho- tometry apertures), and the unknown rotational phases of the object at the times when each photometric point was obtained. We find a best-fit initial average net dust produc- Md = 0.7±0.3 kg s−1 over the time period tion rate of specified above, and a best-fit start date for activity of ∼ 225 ± 85 days prior to perihelion (corresponding to a best-fit start date of 2016 March 11, and an uncertainty range of 2015 December 17 to 2016 June 4), when the ob- ject was at R = 2.66∓ 0.19 au and ν = −63± 21◦. Fulle et al. (2016) found dust-to-gas ratios (by mass), fdg, be- tween 5 and 10 for 67P/Churyumov-Gerasimenko, and 10 Hsieh et al. Table 3. Observation Log: 288P - Active -- -- -- -- -- -- LOT -- -- -- -- -- -- -- -- -- UT Date 1 1 1 54 54 54 r(cid:48) r(cid:48) r(cid:48) k Md -- j HR,t -- i mR,t -- CFHT LOT Tel.a N b tc h mR,n -- SDSS SDSS SDSS CFHT CFHT CFHT CFHT r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) Af ρl -- 3.2±1.1 4.0±1.2 9.3±2.3 αg νd Re ∆f 0.0 2.469 2.409 24.0 −21.1 2.491 1.488 3.1 19.48±0.05 19.5±0.1 16.4±0.1 (0.7±0.3)×107 −20.8 2.490 1.486 2.8 19.36±0.05 19.4±0.1 16.3±0.1 (0.8±0.3)×107 −0.4 2.464 1.980 22.5 20.05±0.05 20.0±0.1 15.5±0.2 (3.0±0.7)×107 0.0 2.464 1.997 22.6 0.0 2.445 3.437 1.7 0.0 2.438 2.292 24.6 3.2±1.2 52.6 2.609 2.453 22.2 21.02±0.05 20.3±0.1 15.2±0.2 (4.3±0.9)×107 5.8±1.5 56.5 2.635 2.687 21.3 20.82±0.05 20.5±0.1 15.2±0.2 (4.2±0.9)×107 5.4±1.4 57.0 2.639 2.716 21.2 20.90±0.05 20.4±0.1 15.1±0.2 (4.9±1.0)×107 4.6±1.3 58.5 2.649 2.802 20.7 21.03±0.05 20.8±0.1 15.5±0.2 (3.2±0.7)×107 1.4±0.1 −42.1 2.546 2.294 23.5 21.36±0.05 21.4±0.1 16.5±0.2 (0.4±0.3)×107 1.0±0.1 −34.2 2.509 1.905 21.6 21.00±0.05 21.0±0.1 16.6±0.2 (0.3±0.3)×107 5.4±1.3 −17.8 2.456 1.452 2.8 19.15±0.05 19.0±0.1 16.0±0.1 (1.6±0.3)×107 5.7±1.3 −17.2 2.454 1.450 2.4 19.09±0.05 19.0±0.1 16.0±0.1 (1.5±0.3)×107 −12.5 2.446 1.474 8.9±1.8 7.5 19.11±0.05 19.1±0.1 15.8±0.1 (2.0±0.4)×107 −12.4 2.446 1.475 9.4±1.9 7.6 19.08±0.05 19.0±0.1 15.7±0.1 (2.4±0.5)×107 −10.1 2.442 1.512 11.0 19.10±0.05 19.0±0.1 15.5±0.1 (3.1±0.6)×107 11.7±2.3 −8.8 2.441 1.541 12.8 19.19±0.05 19.1±0.1 15.5±0.1 (3.1±0.6)×107 11.7±2.4 −3.9 2.437 1.680 18.3 19.53±0.05 19.3±0.1 15.3±0.2 (3.8±0.8)×107 11.1±2.5 −1.7 2.436 1.756 20.1 19.51±0.05 19.2±0.1 15.1±0.2 (5.2±1.0)×107 13.3±2.9 −0.9 2.436 1.788 20.7 19.68±0.05 19.3±0.1 15.1±0.2 (4.9±1.0)×107 11.3±2.6 0.0 2.436 1.823 21.3 5.6 2.438 2.059 23.5 19.87±0.05 19.4±0.1 14.8±0.2 (6.8±1.4)×107 12.7±2.8 6.0 2.438 2.074 23.5 19.85±0.05 19.4±0.1 14.8±0.2 (6.9±1.4)×107 13.2±2.9 13.6 2.448 2.421 23.3 20.10±0.05 19.6±0.1 14.7±0.2 (8.1±1.6)×107 12.5±2.7 14.4 2.449 2.459 23.1 20.04±0.05 19.6±0.1 14.6±0.2 (8.3±1.6)×107 13.7±2.9 14.4 2.449 2.459 23.1 20.10±0.05 19.7±0.1 14.7±0.2 (7.5±1.5)×107 12.8±2.7 19.2 2.459 2.673 21.6 20.05±0.05 19.7±0.1 14.6±0.2 (8.7±1.6)×107 14.7±2.9 0.0 2.437 3.374 Filter 1995 Jul 13 Perihelion ....................... 2000 Sep 3 2000 Sep 4 2000 Nov 17 2000 Nov 18 Perihelion ....................... 2006 Mar 21 Perihelion ....................... 2011 Jul 18 Perihelion ....................... 2012 Jan 28 Gemini-N 7 1260 2012 Feb 13 Gemini-N 2 360 2012 Feb 15 Gemini-N 9 1620 360 2012 Feb 21 Gemini-N 2 4 2016 Jun 8 720 9 1620 2016 Jul 8 900 5 2016 Sep 6 2016 Sep 8 5 900 3 1050 2016 Sep 25 Magellan 5 2016 Sep 25 900 5 1500 2016 Oct 3 5 2016 Oct 8 Gemini-S 750 2016 Oct 25 5 1500 400 2 2016 Nov 2 Magellan 2016 Nov 5 Gemini-S 5 900 2016 Nov 8 Perihelion ....................... 900 2016 Nov 28 Gemini-S 5 900 2016 Nov 29 Gemini-N 5 8 1440 2016 Dec 26 900 2016 Dec 29 5 2016 Dec 29 Gemini-N 4 720 2017 Jan 15 Gemini-N 6 1080 2022 Mar 2 Perihelion ....................... a Telescope used. b Number of exposures. c Total integration time, in seconds. d True anomaly, in degrees. e Heliocentric distance, in au. f Geocentric distance, in au. g Solar phase angle (Sun-object-Earth), in degrees. h Equivalent mean apparent R-band nucleus magnitude, measured within photometry apertures with radii of 4.(cid:48)(cid:48)0. i Equivalent total apparent R-band magnitude, including the entire coma and tail, if present. j Total absolute R-band magnitude, using H, G phase function where G = 0.18. k Estimated total dust mass, in kg, assuming ρd ∼ 2500 kg m3. l Af ρ values computed using photometry apertures with radii of 4.(cid:48)(cid:48)0, in cm, where uncertainties are estimated to be ∼10%. CFHT CFHT -- -- -- 6.5 -- -- -- -- -- -- -- so assuming a conservative value of fdg = 5 (cf. Je- witt et al. 2016), this computed best-fit dust production rate for 238P corresponds to a water production rate of QH2O ∼ 5 × 1024 molecules s−1 (assuming water to be the dominant volatile material). At the midpoint of the time period covered by this fitting analysis, water production rates are expected to range from mw ∼ 3.0 × 10−6 kg s−1 m−2 in the isother- mal (or "fast-rotator") approximation to mw ∼ 5.6 × 10−5 molecules s−1 m−2 in the subsolar (or "flat slab") approximation for a sublimating graybody in thermal equilibrium (see Table 6). Using Aact = Md fdg mw (4) to determine the effective active area, Aact, of a sub- limating object, assuming fdg = 5, we find effective active area estimates ranging from ∼ 3 × 103 m2 (in the subsolar approximation) to ∼ 5 × 104 m2 (in the isothermal approximation). Assuming the nucleus to be a spherical body with an effective radius of rN and us- ing rN = 400 m as determined by Hsieh et al. (2011), we then find effective active fraction estimates ranging from fact ∼ 1 × 10−3 to fact ∼ 2 × 10−2 (Table 6). The best-fit pre-perihelion dust production rate that we find for 238P soon after it was confirmed to be active again in 2016 is a factor of a few larger than the aver- Md = 0.2 kg s−1 determined for age production rate of the object by Hsieh et al. (2009) from post-perihelion observations obtained in 2005-2007. Hsieh et al. (2009) assumed a grain density of ρ = 1000 kg m−3, though, whereas we assume ρ = 2500 kg m−3 here, which could account for some of the difference in inferred dust pro- duction rates between the two epochs. The available Reactivations of 238P and 288P 11 Table 4. Observation Log: 288P - Inactive UT Date Tel.a Filter mR,n h mR,mid(1, 1, α)i r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) R R r(cid:48) r(cid:48) r(cid:48) r(cid:48) R R r(cid:48) r(cid:48) r(cid:48) tc 1800 2340 1800 4320 1260 1800 540 2520 6300 7200 6000 750 900 900 150 4800 4200 360 360 900 Re 3.118 3.160 3.162 3.168 3.170 3.172 3.181 3.208 3.225 3.232 3.234 3.262 3.282 3.284 3.286 3.457 3.458 3.307 3.252 3.250 ∆f 3.226 2.953 2.940 2.902 2.890 2.877 2.815 2.637 2.541 2.503 2.494 2.374 2.325 2.323 2.321 3.551 3.566 2.438 2.239 2.237 νd 108.0 111.7 111.9 112.4 112.6 112.7 113.6 116.1 117.5 118.2 118.3 120.9 122.8 123.0 123.1 140.0 140.2 −125.0 −120.0 −119.9 N b 2012 Oct 18 Gemini-N 10 2012 Nov 9 Gemini-N 13 2012 Nov 10 Gemini-N 10 2012 Nov 13 Gemini-N 24 2012 Nov 14 Gemini-N 7 2012 Nov 15 Gemini-N 10 2012 Nov 20 Gemini-N 3 2012 Dec 5 Gemini-N 14 2012 Dec 14 Gemini-N 35 2012 Dec 18 12 10 2012 Dec 19 5 2013 Jan 4 6 2013 Jan 16 6 2013 Jan 17 2013 Jan 18 1 8 2013 May 12 7 2013 May 13 2 2015 Apr 24 2 2015 May 26 2015 May 27 6 a Telescope used (UH2.2: UH 2.2 m telescope). b Number of exposures. c Total integration time, in seconds. d True anomaly, in degrees. e Heliocentric distance, in au. f Geocentric distance, in au. g Solar phase angle (Sun-object-Earth), in degrees. h Equivalent mean apparent R-band nucleus magnitude. i Estimated reduced R-band magnitude at midpoint of full photometric range (assumed to be 0.80 mag) 22.42±0.05 22.90±0.05 22.73±0.05 22.66±0.05 22.85±0.05 22.72±0.05 22.80±0.05 22.51±0.05 21.77±0.05 22.21±0.05 21.85±0.05 21.66±0.05 21.48±0.05 21.64±0.05 21.55±0.05 23.2±0.1 23.0±0.1 21.61±0.05 21.25±0.05 21.13±0.05 17.39±0.30 18.05±0.35 17.91±0.35 17.68±0.05 18.09±0.30 17.90±0.30 18.04±0.40 17.86±0.40 17.23±0.35 17.66±0.40 17.37±0.40 17.36±0.40 17.09±0.40 17.24±0.15 17.14±0.40 17.8±0.4 17.5±0.4 17.07±0.40 16.94±0.40 16.84±0.35 αg 18.0 18.2 18.2 18.1 18.0 18.0 17.6 15.9 14.2 13.4 13.1 8.7 4.7 4.4 4.0 16.5 16.4 10.3 0.5 0.7 UH2.2 UH2.2 CFHT CFHT CFHT CFHT UH2.2 UH2.2 CFHT CFHT CFHT of rotational light curve. data also do not cover similar orbit arcs, with the data used for the analysis by Hsieh et al. (2009) having been obtained post-perihelion and at larger heliocentric dis- tances, where temperatures and therefore sublimation rates would have been lower, than the pre-perihelion data analyzed here, making it difficult to definitively determine whether there have been changes in activity strength for 238P between 2005-2007 and 2016-2017. On the other hand, data is available for the start of 238P's 2010-2011 active apparition (Table 2; Hsieh et al. 2011), providing an opportunity to perform a more di- rect comparison of activity strength for 238P during dif- ferent active apparitions. We follow the analysis applied above to photometric data from 238P's 2016 reactiva- tion, fitting a linear function to estimated ejected dust masses for 238P computed from data obtained from 2010 September 3 to 2010 December 9, where the object's he- liocentric distance change over this period corresponds to a ∼20% (using the subsolar approximation) to 83% (using the isothermal approximation) increase in the wa- ter sublimation rate on the object's surface. Treating the lower limit excess mass measurements reported by Hsieh et al. (2011) as exact values (likely a reasonable approximation for this early period before dust in the comet's tail expands significantly beyond near-nucleus photometry apertures), we find a best-fit initial net dust Md ∼ 1.4±0.3 kg s−1 (corresponding production rate of to QH2O ∼ 9 × 1024 molecules s−1, assuming fdg = 5), about twice that found in our analysis of the object's 2016 activity. We also find a best-fit start date for activity of ∼205±50 days prior to perihelion (corresponding to a best-fit start date of 2010 August 17 and an uncer- tainty range of 2010 June 28 to 2010 October 6), when the object was at a very similar heliocentric distance (R = 2.61∓ 0.11 au) and true anomaly (ν = −58± 12◦) as it was on the estimated start date of its 2016 activ- ity. We also find effective active area estimates ranging from ∼ 6×103 m2 (using the subsolar approximation) to ∼ 1× 105 m2 (using the isothermal approximation), cor- responding to effective active fraction estimates ranging from ∼ 3 × 10−3 (using the subsolar approximation) to ∼ 7 × 10−2 (using the isothermal approximation). Additional details of our analysis of these data are sum- marized in Table 6. Examining the true anomaly range over which we have overlapping data from 238P's 2010-2011 and 2016-2017 active periods (−35◦ < ν <−20◦), we find a excess dust mass of Md > (1.5±0.5)×107 kg on 2010 December 9 (when 238P was at ν =−27.5◦; Table 2) and an aver- age excess dust mass of Md = (0.8±0.2)×107 kg for the period between 2016 July 8 and 2016 August 6 (when 12 Hsieh et al. Table 5. Previously reported observations of 288P when active Tel.a Perihelion ... i Md -- f mR,n -- PS1 PS1 FTN UH2.2 FTN Perkins mR,t -- g HR,t -- h UT Date 2011 Jul 18 2011 Aug 30 2011 Nov 5 2011 Nov 12 2011 Nov 14 2011 Nov 14 2011 Nov 18 2011 Nov 19 2011 Nov 22 2011 Nov 30 2011 Dec 4 2011 Dec 16 2011 Dec 19 2012 Jan 7 2016 Nov 8 a Telescope (PS1: Pan-STARRS1; FTN: Faulkes Telescope North; UH2.2: UH 2.2 m telescope; Perkins: Lowell 20.16±0.14 <20.2 <15.9 >(1.6±0.8)×107 19.00±0.09 <19.0 <15.7 >(2.2±1.0)×107 18.69±0.09 <18.7 <15.2 >(4.5±1.6)×107 18.62±0.05 <18.6 <15.1 >(5.0±1.8)×107 18.64±0.05 <18.6 <15.1 >(4.9±1.8)×107 18.60±0.10 <18.6 <15.0 >(5.8±2.0)×107 18.64±0.02 <18.6 <15.0 >(5.8±2.0)×107 18.89±0.02 <18.9 <15.1 >(4.8±1.7)×107 19.04±0.02 <19.0 <15.1 >(4.9±1.8)×107 19.12±0.03 <19.1 <15.1 >(5.0±1.8)×107 19.70±0.09 <19.7 <15.4 >(3.5±1.4)×107 19.68±0.03 <19.7 <15.3 >(3.8±1.5)×107 20.43±0.10 <20.4 <15.7 >(2.3±1.0)×107 ∆d 2.292 1.806 1.517 1.555 1.561 1.561 1.586 1.596 1.621 1.685 1.724 1.861 1.895 2.152 1.823 Rc 2.438 2.447 2.496 2.505 2.506 2.506 2.510 2.512 2.516 2.525 2.530 2.546 2.549 2.577 2.436 νb 0.0 12.2 30.7 32.9 33.2 33.2 34.3 34.6 35.5 37.4 38.5 41.7 42.4 47.4 0.0 αe 24.6 21.4 4.6 8.0 8.4 8.4 10.0 10.6 11.9 14.4 15.6 18.7 19.2 21.7 21.3 HCT WHT UH2.2 NTT FTS UH2.2 LOT -- -- -- -- Perihelion ... Ref.j -- [1] [1] [1] [1] [1] [1] [1] [1] [1] [1] [1] [1] [1] -- Observatory Perkins Telescope; HCT: Himalayan Chandra Telescope; WHT: William Herschel Telescope; NTT: New Technology Telescope; FTS: Faulkes Telescope South; LOT: Lulin One-meter Telescope). b True anomaly, in degrees. c Heliocentric distance, in au. d Geocentric distance, in au. e Solar phase angle (Sun-object-Earth), in degrees. f Reported mean apparent R-band nucleus magnitude. g Equivalent mean apparent R-band total magnitude, including the entire coma and tail, if present. h Absolute R-band total magnitude (at R = ∆ = 1 au and α = 0◦), i Estimated total dust mass, in kg, assuming ρd ∼ 2500 kg m3. j [1] Hsieh et al. (2012b) using IAU H, G phase-darkening where G = 0.18. 238P was at an average true anomaly of ν =−27.3◦; Table 1). As such, we find that the excess dust mass present during the early portion of 238P's 2010-2011 ac- tive period may have been approximately twice that (or more) present during the same portion of 238P's 2016- 2017 active period. 3.4. 288P Activity Characterization To compare 288P's activity in 2016-2017 to its pre- viously studied active apparition, we compute Md and Af ρ for observations of 288P reported in Hsieh et al. (2012b). Total magnitudes of 288P (i.e., including the entire coma and tail) were not measured for these data in the same way as we have measured recent data, and as such, we report estimated excess dust masses as lower limits in Table 5. We plot total absolute R-band mag- nitudes and estimated ejected dust masses for 288P in Figure 8. Following the analysis performed above for 238P (Sec- tion 3.3), we fit a linear function to estimated ejected dust masses for 288P computed from data obtained be- tween 2016 September 6 and 2016 October 25, where the object's heliocentric distance change over this pe- riod corresponds to a ∼13% (using the subsolar approx- imation) to 50% (using the isothermal approximation) increase in the water sublimation rate on the object's surface. We find a best-fit initial net dust production rate for 288P early in its 2016-2017 active period of Md = 5.6±0.7 kg s−1 (corresponding to QH2O ∼ 4× 1025 molecules s−1, assuming fdg = 5), and a best-fit start date for activity of ∼ 95 ± 15 days prior to perihelion (corresponding to a best-fit start date of 2016 August 5 and an uncertainty range of 2016 July 21 to 2016 Au- gust 20), when the object was at R = 2.48 ∓ 0.02 au and ν = −27 ± 4◦. We also find effective active area es- timates ranging from ∼ 2 × 104 m2 (using the subsolar approximation) to ∼ 5 × 105 m2 (using the isothermal approximation), corresponding to effective active frac- tion estimates ranging from ∼ 1×10−3 (using the subso- lar approximation) to ∼ 3 × 10−2 (using the isothermal approximation). These active fractions are calculated assuming that 288P's nucleus is a binary system with equally-sized spherical components, each with radii of rc ∼ 800 m (Section 3.1). Additional details of our analysis of these data are summarized in Table 6. While photometry of the object shows it to be slightly brighter than the expected magnitude of its inactive nu- cleus on 2016 June 8 and 2016 July 8, we choose to omit these data from the fitting analysis above as the photometric enhancements measured for the object on these dates are within the range of possible fluctuations in the object's brightness due to rotation, and as such, Reactivations of 238P and 288P Table 6. Estimated Activity Start Times and Strength Parameters Object 238P 238P 288P 288P Data rangea 2010 Sep 3 − 2010 Dec 9 R: 2.576 au → 2.416 au ν: −54.2◦ → −27.6◦ 2016 Jul 8 − 2016 Oct 9 R: 2.439 au → 2.367 au ν: −31.5◦ → −4.0◦ 2000 Sep 3 − 2000 Nov 17 R: 2.491 au → 2.464 au ν: −21.1◦ → −0.4◦ 2016 Sep 6 − 2016 Oct 25 R: 2.546 au → 2.437 au ν: −42.3◦ → −4.0◦ Md = 1.4 ± 0.3 kg s−1 2.1 × 10−6 (IT) 1 × 105 (IT) t0 = 2010 Aug 17 ± 50 days R = 2.49 au 5.0 × 10−5 (SS) 6 × 103 (SS) mH2O d Aact e Midpointc 2010 Oct 21 ν = −41◦ Fit resultsb R0 = 2.61 ∓ 0.11 au ν0 = −58 ± 12◦ Md = 0.7 ± 0.3 kg s−1 2016 Aug 24 3.0 × 10−6 (IT) 5 × 104 (IT) t0 = 2016 Mar 11 ± 85 days R = 2.39 au 5.6 × 10−5 (SS) 3 × 103 (SS) R0 = 2.66 ∓ 0.19 au ν0 = −63 ± 21◦ ν = −18◦ Md = 3.5 ± 0.4 kg s−1 2.3 × 10−6 (IT) 3 × 105 (IT) t0 = 2000 Aug 9 ± 15 days R = 2.47 au 5.1 × 10−5 (SS) 1 × 104 (SS) 2000 Oct 11 R0 = 2.51 ∓ 0.02 au ν0 = −28 ± 4◦ ν = −11◦ Md = 5.6 ± 0.7 kg s−1 2.5 × 10−6 (IT) 4 × 105 (IT) t0 = 2016 Aug 5 ± 15 days R = 2.44 au 5.3 × 10−5 (SS) 2 × 104 (SS) 2016 Sep 30 R0 = 2.48 ∓ 0.02 au ν0 = −27 ± 4◦ ν = −11◦ 13 f fact 7 × 10−2 (IT) 3 × 10−3 (SS) 2 × 10−2 (IT) 1 × 10−3 (SS) 2 × 10−2 (IT) 9 × 10−4 (SS) 3 × 10−2 (IT) 1 × 10−3 (SS) a Range of data used for fitting analysis, including dates and corresponding ranges of heliocentric distances, R, and true anomalies, ν. b Best-fit results from fitting analysis for initial net dust production rate, Md, start date, t0, heliocentric distance at start date, R0, and true anomaly at start date, ν0. c Date at midpoint of range of data used for fitting analysis, including heliocentric distance, R, and true anomaly, ν, on that date. d Expected sublimation rates of water for a sublimating graybody in thermal equilibrium, in kg s−1 m−2, at the heliocentric distance of the object at the indicated midpoint date, using isothermal (IT) and subsolar (SS) approximations. e Estimated effective active areas, in m2, assuming fdg = 5, implied by best-fit initial net dust production rates and expected sublimation rates of water computed using isothermal (IT) and subsolar (SS) approximations. f Estimated effective active fractions, assuming fdg = 5, and rN = 400 m for 238P's nucleus and rc = 800 m for each component of 288P's binary nucleus, for expected sublimation rates of water computed using isothermal (IT) and subsolar (SS) approximations. we cannot be certain that the object is in fact active on those dates. As such, we omit these data points from this fitting analysis, focusing instead on data that we are certain was obtained when the object was active. In their analysis of 288P's 2011-2012 active episode, Licandro et al. (2013) found an average dust produc- tion rate of 0.2 kg s−1 over a 100-day period starting shortly after perihelion (i.e., 0◦ < ν < 30◦; 2.43 au < R < 2.50 au), and a peak mass loss rate of 0.5 kg s−1 about 60 days after perihelion (when the object was at ν ∼ 17◦ and R ∼ 2.46 au). These conclusions were based on analysis of one night of data obtained on 2011 Novem- ber 29 when the object was at ν = 37◦ and R = 2.52 au. Licandro et al. (2013) assumed a grain density of ρ = 1000 kg m−3, whereas we assume ρ = 2500 kg m−3 here, however, and also analyzed observations obtained at a larger heliocentric distance than the observations we dis- cuss here. Given the extremely large uncertainty on our computed mass loss rate and difference in initial assump- tions, we regard our result as approximately consistent with that of Licandro et al. (2013). Although very little data is available for 288P from 2000 (Table 3), we are interested in making at least a rough assessment of its activity strength at the time. Following the analysis performed above, we fit a linear function to estimated ejected dust masses for 288P com- puted from data obtained between 2000 September 3 to 2000 November 17 (cf. Table 5), where the object's helio- centric distance change over this period corresponds to a ∼3% (using the subsolar approximation) to 10% (using the isothermal approximation) increase in the water sub- limation rate on the object's surface. We find a best-fit Md = 3.5±0.4 kg s−1 initial net dust production rate of (corresponding to QH2O ∼ 2 × 1025 molecules s−1, as- suming fdg = 5), and a best-fit start date for activity of ∼ 100 ± 15 days prior to perihelion (corresponding to a best-fit start date of 2000 August 9 and an uncer- tainty range of 2000 July 25 to 2000 August 24), when the object was at R = 2.51 ∓ 0.02 au and ν = −28 ± 4◦. We note that these uncertainties, which are generated from the same fitting routines used to analyze all of the other analogous data sets discussed in this work, imply a potentially unrealistically precise best-fit solution, given the small number of data points used to derive it, and in practice, should be regarded as being somewhat larger. We also find effective active area estimates ranging from ∼ 1 × 104 m2 (using the subsolar approximation) to ∼ 3× 105 m2 (using the isothermal approximation), cor- responding to effective active fraction estimates ranging from ∼ 9 × 10−4 (using the subsolar approximation) to 14 Hsieh et al. 3.5. Comparison to Other Active Asteroids We plot the confirmed active ranges (where visible dust emission or photometric enhancement has been re- ported) of all likely MBCs identified to date in Fig- ure 9, updating the similar figure originally shown by Hsieh & Sheppard (2015). While extending the active ranges confirmed for 238P and 288P based on obser- vations presented here, we also add active ranges re- ported for recently discovered MBC candidates P/2015 X6 (PANSTARRS) and P/2016 J1-A/B (PANSTARRS) (cf. Moreno et al. 2016; Hui et al. 2017) to this figure, and extend the active range for 259P based on the re- cently reported confirmation of its reactivation (Hsieh & Chavez 2017). As noted by Hsieh & Sheppard (2015), the active regions marked in this plot should be con- sidered lower limits to the full ranges over which ac- tivity may be present for each object. This is because good constraints are not always available for the onset or termination of activity for an object given that obser- vational circumstances may prevent direct observations during the onset or termination of activity for objects already known to be active, or the fact that new ac- tive objects are by definition discovered while already exhibiting activity, and as such, must complete at least another full orbit before attempts can be made to di- rectly observationally constrain the onset times of those objects' activity. 324P remains the MBC with the largest observed ac- tive range of all likely MBCs in terms of orbit position, with both the earliest and latest observations of activ- ity in terms of true anomaly, and also the most distant confirmed activity in terms of heliocentric distance on both the inbound and outbound portions of its orbit of all of the MBCs. However, the onset point of 238P's ac- tivity found by both Hsieh et al. (2011) and this work is similar to that of 324P in terms of true anomaly, while 133P has been observed to exhibit residual activity at a similarly large true anomaly and similarly distant helio- centric distance as 324P on the outbound portion of its orbit (cf. Figure 9). 4. DISCUSSION With reported observations of activity in 2016 and 2017 for both 238P and 288P and in 2000 for 288P (this work; Agarwal et al. 2016b; Hsieh et al. 2016), both ob- jects have now been reported to be active near perihelion on three separate occasions. This further solidifies the conclusion that their activity is likely to be due to sub- limation of volatile material, and not due to disruptive events like rotational destabilization events or impacts, which would not be expected to repeat so regularly or so frequently, nor specifically occur near perihelion. Figure 6. Best-fit IAU phase function (solid line) for 288P where estimated reduced R-band magnitudes at the mid- point of the full photometric range (assumed to be 0.8 mag) of the object's rotational light curve for data obtained be- tween 2012 October to 2015 May when no activity was de- tected for 288P are plotted with open circles. Dotted lines indicate the range of uncertainty due to estimated phase function parameter uncertainties, while dashed lines indicate the possible photometric range due to rotational brightness variations, assuming a peak-to-trough photometric range of ∆m = 0.8 mag. ∼ 2 × 10−2 (using the isothermal approximation). Ad- ditional details of our analysis of these data are sum- marized in Table 6. The estimated start time of 288P's activity in 2000 is quite close to its estimated start time in 2016, although the initial net dust production rate for the object appears to have actually increased from 2000 to 2016. Examining the true anomaly range over which we have overlapping data from 288P's 2000 and 2016-2017 ac- tive periods, we find an average excess dust mass of Md = (0.8±0.2)×107 kg for 2000 September 3-4 when 288P had an average true anomaly of ν =−21.0◦, and an average excess dust mass of Md = (1.6±0.2)×107 kg for 2016 September 6-8 when 288P had a similar aver- age true anomaly of ν =−17.5◦. We also find an excess dust mass of Md = (3.0±0.7)×107 kg on 2000 November 17 when 288P was at ν =−0.4◦, and an average excess dust mass of Md = (5.1±0.7)×107 kg for 2016 Novem- ber 2-5 when 288P had a similar average true anomaly of ν =−1.3◦. As such, we find that excess dust masses measured for 288P in 2016 are larger than excess dust masses measured in 2000 for the object when it was at similar points in its orbit, corroborating our earlier con- clusion that the average dust production rate for the object appears to have actually increased over time. Reactivations of 238P and 288P 15 Figure 7. (a) Total absolute R-band magnitude of 238P during its 2005-2007 active period (green diamonds), 2010 inactive period (open squares), 2010-2011 active period (yellow squares), and 2016-2017 active period (blue circles) plotted as a function of true anomaly. The expected magnitude of the inactive nucleus is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (b) Total estimated dust masses measured for 238P during the same periods of observations as in (a) plotted as a function of true anomaly. The excess dust mass expected for the inactive nucleus (i.e., zero) is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (c) Estimated total dust masses measured for 238P during just its 2016-2017 active period plotted as a function of time from perihelion (where negative values denote time before perihelion and positive values denote time after perihelion). A diagonal solid blue line shows a linear fit to data obtained between 2016 July 8 and 2016 November 5 (−31.5◦ < ν < 4.2◦), reflecting the average net dust production rate over this period (over which dust production appears to be roughly linear) and allowing us to estimate the onset time of activity, while diagonal dotted blue lines show the range of uncertainty of the linear fit. A diagonal solid orange line shows a linear fit to data obtained between 2010 September 3 and 2010 December 9 (−54.1◦ < ν <−27.5◦), reflecting the average net dust production rate over this period (over which dust production appears to be roughly linear), while diagonal dotted orange lines show the range of uncertainty of the linear fit. As discussed in Section 1.2, our observations of mul- tiple active episodes are also useful for investigating the evolution of activity between those different active pe- riods. Interestingly, while activity strength (as parame- terized by initial dust production rate) appears to have declined by about a factor of 2 for 238P between 2010- 2011 and 2016-2017, activity strength for 288P appears to have increased between 2000 and 2016-2017. Esti- mates of the evolution of water sublimation rates in the presence of a growing rubble mantle (cf. Jewitt 1996) suggest that activity should decline relatively rapidly soon after activity is initially triggered but then should decline more slowly after a mantle of sufficient thickness has developed (e.g., Hsieh et al. 2015a). This behav- ior can be partly attributed to the dependence of man- tle growth rates on sublimation rates (cf. Hsieh et al. 2015a): as sublimation rates decrease from mantling, mantle growth slows, slowing the change in sublimation rates from one orbit passage to the next. A more de- tailed analysis by Kossacki & Szutowicz (2012) found that the evolution of activity for a MBC from one per- ihelion passage to the next also depend on additional physical properties and circumstances such as spin axis orientation, latitude of specific active sites, grain sizes, bulk density, and porosity. Increases in activity strength over time could conceiv- ably occur if other processes besides mantling also have significant modulating effects on dust production rates. These processes may or may not be associated with the object's ongoing activity, and could include sink- hole collapses (e.g., Vincent et al. 2015) or rotation- or impact-induced landslide activity (e.g., Steckloff et al. 2016; Hofmann et al. 2017) that uncover fresh volatile material. Perihelion distance reduction could also be a potential explanation for increased activity strength (Licandro et al. 2000). In the case of 288P, further anal- ysis (e.g., using detailed dust modeling) must first be done to confirm whether the apparent increase in activ- ity strength between 2000 and 2016-2017 is real before speculating on possible or likely causes of such an in- crease. Ultimately, determining whether physical pro- cesses that may actually increase activity strength over time are in operation on MBCs will likely require more occurrences of increasing activity strength to be identi- 16 Hsieh et al. Figure 8. (a) Total absolute R-band magnitude of 288P during its 2000 active period (green diamonds), 2011-2012 active period (yellow squares), 2012-2015 inactive period (open circles), and 2016-2017 active period (blue circles) plotted as a function of true anomaly. The expected magnitude of the inactive nucleus is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (b) Total estimated dust masses measured for 288P during the same periods of observations as in (a) plotted as a function of true anomaly. The excess dust mass expected for the inactive nucleus (i.e., zero) is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (c) Estimated total dust masses measured for 238P during just its 2016-2017 active period plotted as a function of time from perihelion (where negative values denote time before perihelion and positive values denote time after perihelion). A diagonal solid blue line shows a linear fit to data obtained between 2016 June 8 and 2016 October 8 (−42.1◦ < ν < −8.8◦), reflecting the average net dust production rate over this period (over which dust production appears to be roughly linear) and allowing us to estimate the onset time of activity, while diagonal dotted blue lines show the range of uncertainty of the linear fit. A diagonal solid green line shows a linear fit to data obtained between 2000 September 3 and 2000 November 17 (−21.1◦ < ν < −0.4◦), reflecting the average net dust production rate over this period (over which dust production appears to be roughly linear), while diagonal dotted green lines show the range of uncertainty of the linear fit. A vertical arrow indicates the time of the HST observations of 288P by Agarwal et al. (2016a) when the object was seen to be active. fied and characterized. To achieve this, detailed observa- tional characterization of more repeated active episodes for MBCs will be required, further motivating continued monitoring of known active MBCs. Meanwhile, we find similar start times for different active periods for both 238P (in 2010-2011 and 2016- 2017) and 288P (in 2000 and 2016-2017) (Sections 3.3 and 3.4). Assuming that each object's activity is driven by the sublimation of subsurface ice, the depth of that ice should be a significant controlling factor of when ac- tivity starts, given the finite time needed for solar inso- lation to propagate through surface layers to buried ice reservoirs (cf. Hsieh et al. 2011). As such, the consistent start times for activity for each object for different active periods appear to suggest that ice depths remained rela- tively consistent between the active episodes in question, implying that minimal mantle growth or ice recession occurred during these periods. In terms of follow-up opportunities, 238P was observ- able again from September 2017 to June 2018, during which it covered a true anomaly range of 80◦ < ν < 130◦, while 288P was observable again from August 2017 to May 2018, during which it covered a true anomaly range of 70◦ < ν < 125◦, offering opportunities to monitor both objects for activity past the largest true anomaly at which either 133P or 324P has been seen to exhibit ac- tivity. The 2017-2018 observability window for 238P just barely overlapped the last observation of 238P's 2010- 2011 active period that we report here, and so presented a potential opportunity for another direct comparison of activity strength for the object at the same point in its orbit during two different orbit passages in addition to the ones we present here. The observing window for 288P did not overlap any previous active observations, however, and so no similar opportunity to directly com- pare activity strengths from two different orbit passages was available for this object. We have acquired data for both objects during their respective follow-up periods and will report results based on analyses of those data in a future paper. As indicated above in Section 1.2 and in Figure 9, there are now seven MBCs which have been confirmed Reactivations of 238P and 288P 17 Figure 9. Active ranges, extending between the earliest and latest observations for which activity has been reported (Hsieh et al. 2015a; Moreno et al. 2016; Hui et al. 2017; Hsieh & Chavez 2017, and references within), in terms of true anomaly (left) and heliocentric distance (right) for likely MBCs. Solid blue line segments indicate the inbound (pre-perihelion) portion of each object's orbit while outlined blue line segments indicate the outbound (post-perihelion) portion of each object's orbit. In the left-hand panel, perihelion is marked with a dashed vertical line, while the earliest and latest orbit positions at which activity has been observed for 324P (the MBC with the earliest and latest observed activity in terms of true anomaly) are marked with dotted vertical lines. In the right-hand panel, horizontal black line segments indicate the heliocentric distance range covered by the orbit of each object, and the most distant positions at which activity has been observed for 324P (the MBC for which the most distant activity has been observed inbound to perihelion, as well as outbound away from perihelion) during the inbound and outbound portions of its orbit are marked with a vertical dashed line and a vertical dotted line, respectively. Numbers in parentheses to the far right of the left panel indicate the number of confirmed active apparitions that have been observed for each object. After Hsieh & Sheppard (2015). to exhibit activity on two or more separate occasions (always near perihelion, and often with intervening ob- servational confirmation of inactivity away from perihe- lion). Of those, 238P and 288P have been seen to be active on three separate occasions and 133P has been seen to be active on four separate occasions. Observa- tions obtained during each of these active apparitions will be useful for systematic dust modeling studies, in which mass loss rates are independently determined for each active apparition in a consistent manner. Results from these studies will allow us to more quantitatively compare changes in activity strength of MBCs from one active apparition to the next, giving us insights into the process of activity evolution over time for MBCs. We plan to conduct such a study in the future, incorporating both data reported in the literature to date (including in this paper) and unpublished observations currently in hand or still being obtained of the recent reactivations of 259P, 324P, and 358P (cf. Hsieh & Sheppard 2015; Hsieh & Chavez 2017; Hsieh et al. 2018a). The number of currently active MBCs in the asteroid belt is related to the rate of triggering events (e.g., surface disruption events such as impacts or landslides) and the duration of activity following such triggering events (Hsieh 2009). As such, given improved estimates of the size of the cur- rently active MBC population provided by current and future survey data, better constraints on active lifetimes following triggering events enabled by an improved un- derstanding of activity evolution could provide indepen- dent constraints on the rate of triggering events, provid- ing a means for evaluating the plausibility of proposed triggering mechanisms. 5. SUMMARY In this work, we present the following key findings: 1. We confirm the reactivations of main-belt comets 238P/Read and 288P/(300163) 2006 VW139, pre- viously reported by Hsieh et al. (2016) and Agar- wal et al. (2016a), and have obtained data follow- ing the evolution of each object's activity over sev- eral months in 2016 and 2017. Additionally, we re- port the identification of archival SDSS data from 2000 of 288P in which the object is seen to be active. With these observations, both 238P and 288P have now each been confirmed to be active near perihelion on three separate occasions. 2. A photometric analysis of observations obtained of 288P while the object appeared inactive from 2012 18 Hsieh et al. to 2015 yields best-fit IAU phase function parame- ters of HR = 16.80±0.12 mag and GR = 0.18±0.11, corresponding to an effective nucleus radius of rN = 1.13 ± 0.06 km (assuming a R-band albedo of pR = 0.05) or, assuming 288P's nucleus to be a binary system with approximately equally sized components, effective component radii of rc = 0.80 ± 0.04 km each. 3. In an analysis of our observations of 238P's reacti- vation in 2016, we find a best-fit initial average net Md = 0.7±0.3 kg s−1 and a dust production rate of best-fit start date of activity of ∼225 days prior to perihelion, corresponding to 2016 March 11 when 238P was at R = 2.66 au and ν = −63◦. In an analogous analysis of observations of 238P's activ- ity in 2010-2011, we find a best-fit initial average Md ∼ 1.4±0.3 kg s−1, net dust production rate of i.e., about twice that estimated for the object's 2016-2017 active period, and a best-fit start date of activity of ∼ 205 days prior to perihelion, cor- responding to 2010 August 17 when 238P was at R = 2.61 au and ν = −58◦, i.e., similar to the or- bit position of the start of activity inferred for the object's 2016-2017 active period. Comparing esti- mated dust masses from overlapping true anomaly ranges from 238P's 2010-2011 and 2016-2017 ac- tive periods, we find that the dust mass present in 2010-2011 may have been approximately twice that present over the same orbit arc in 2016-2017. 4. In an analysis of our observations of 288P's reacti- vation in 2016, we find a best-fit initial average net Md = 5.6± 0.7 kg s−1 and dust production rate of a best-fit start date of activity of ∼ 95 days prior to perihelion, corresponding to 2016 August 5 when 288P was at R = 2.48 au and ν = −27◦. In an analogous analysis of our observations of 288P's re- activation in 2016, we find a best-fit initial average Md = 3.5±0.4 kg s−1, net dust production rate of suggesting that the dust production rate actually increased between 2000 and 2016-2017, and a best- fit start date of activity of ∼ 100 days prior to perihelion, corresponding to 2000 August 9 when 288P was at R = 2.51 au and ν = −28◦, i.e., similar to the orbit position of the start of activ- ity inferred for the object's 2016-2017 active pe- riod. Comparing estimated dust masses from sim- ilar true anomaly positions during 288P's 2000 and 2016-2017 active periods, we find that excess dust masses estimated during 2016-2017 to be larger than those measured at similar orbit positions in 2000. More detailed dust modeling and analysis will be required to determine whether the appar- ent increase in 288P's activity strength between 2000 and 2016-2017 is real, and if it is, what mech- anisms could be responsible for such evolution of the object's activity strength. 5. We find similar start times for different active pe- riods for both 238P (in 2010 and 2016) and 288P (in 2000 and 2016). The consistent start times for activity for each object for different active periods suggest that minimal mantle growth or ice recession occurred during the periods in ques- tion, leaving delays in the start of activity caused by the time needed for solar insolation to propa- gate through surface layers to buried ice reservoirs largely unchanged from one episode to the next. We expect that future systematic dust modeling studies of the active apparitions of these objects and other MBCs will provide additional insights into the process of activity evolution for MBCs, with implications for constraining total activity lifetimes and the rate of MBC triggering events from discovery statistics. HHH, MMK, NAM, and SSS acknowledge support from the NASA Solar System Observations program (Grant NNX16AD68G). CAT was funded in part through the State of Arizona Technology and Research Initiative Program. HHH and MMK also thank the International Space Science Institute (ISSI) in Bern, Switzerland, for the hosting and provision of financial support for an international team to discuss the science of MBCs that facilitated discussion related to this work. We are grateful to S. Arnouts, T. Burdullis, A. Draginda, N. Flagey, P. Forshay, A. Petric, L. Wells, and C. Wipper at CFHT, J. Ball, P. Candia, J. Chavez, D. Coulson, L. Fuhrman, M. Gomez, P. Hirst, M. Hoenig, J. Kemp, H. Lee, S. Leggett, A. Lopez, T. Matulonis, R. McDermid, J. Miller, C. Morley, J. O'Donoghue, S. Pakzad, M. Pohlen, J. Rhee, R. Salinas, D. Sanmartim, M. Schwamb, O. Smirnova, A. Smith, A. Stephens, S. Stewart, B. Walp, E. Wenderoth, and S. Yang at Gem- ini, and J. Dvorak, E. Moore, and M. Willman at the UH 2.2 m telescope for assistance in obtaining obser- vations. This work made use of the Discovery Channel Telescope operated by Lowell Observatory. Lowell is a private, non-profit institution dedicated to astrophysical research and public appreciation of astronomy and op- erates the DCT in partnership with Boston University, the University of Maryland, the University of Toledo, Northern Arizona University and Yale University. The Large Monolithic Imager was built by Lowell Obser- Reactivations of 238P and 288P 19 vatory using funds provided by the National Science Foundation (AST-1005313). Funding for the SDSS and SDSS-II has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, the U.S. Department of Energy, the National Aeronautics and Space Administration, the Japanese Monbukagakusho, the Max Planck Society, and the Higher Education Funding Council for England. The SDSS Web Site is http://www.sdss.org/. The SDSS is managed by the Astrophysical Research Consortium for the Participat- ing Institutions. The Participating Institutions are the American Museum of Natural History, Astrophysical In- stitute Potsdam, University of Basel, University of Cam- bridge, Case Western Reserve University, University of Chicago, Drexel University, Fermilab, the Institute for Advanced Study, the Japan Participation Group, Johns Hopkins University, the Joint Institute for Nuclear As- trophysics, the Kavli Institute for Particle Astrophysics and Cosmology, the Korean Scientist Group, the Chi- nese Academy of Sciences (LAMOST), Los Alamos Na- tional Laboratory, the Max-Planck-Institute for As- tronomy (MPIA), the Max-Planck-Institute for Astro- physics (MPA), New Mexico State University, Ohio State University, University of Pittsburgh, University of Portsmouth, Princeton University, the United States Naval Observatory, and the University of Washington. The Pan-STARRS1 Surveys (PS1) and the PS1 public science archive have been made possible through contri- butions by the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max- Planck Society and its participating institutes, the Max Planck Institute for Astronomy, Heidelberg and the Max Planck Institute for Extraterrestrial Physics, Garching, The Johns Hopkins University, Durham University, the University of Edinburgh, the Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, the Las Cumbres Observatory Global Telescope Network Incorporated, the National Central University of Tai- wan, the Space Telescope Science Institute, the National Aeronautics and Space Administration under Grant No. NNX08AR22G issued through the Planetary Science Di- vision of the NASA Science Mission Directorate, the National Science Foundation Grant No. AST-1238877, the University of Maryland, Eotvos Lorand University (ELTE), the Los Alamos National Laboratory, and the Gordon and Betty Moore Foundation. We wish to recognize and acknowledge the very sig- nificant cultural role and reverence that the summit of Maunakea has always had within the indigenous Hawai- ian community. We are fortunate to have the opportu- nity to conduct observations from this mountain. Facilities: CFHT (MegaCam), Discovery Channel Telescope (LMI), Gemini North (GMOS-N), Gemini South (GMOS-S), Lulin One-meter Telescope, Magel- lan Baade (IMACS), University of Hawaii 2.2m, Sloan Digital Sky Survey (SDSS) REFERENCES Abazajian, K. N., Adelman-McCarthy, J. K., Agueros, Boulade, O., Charlot, X., Abbon, P., et al. 2003, in M. A., et al. 2009, ApJS, 182, 543 Agarwal, A., Jewitt, D., Weaver, H., Mutchler, M., & Larson, S. 2016a, Central Bureau Electronic Telegrams, 4306 Proc. SPIE, Vol. 4841, Instrument Design and Performance for Optical/Infrared Ground-based Telescopes, 72 -- 81 Britt, D. T., Yeomans, D., Housen, K., & Consolmagno, G. 2002, Asteroids III (Tucson, University of Arizona Press), Agarwal, J., Jewitt, D., Mutchler, M., Weaver, H., & 485 -- 500 Larson, S. 2017, Nature, 549, 357 Capria, M. T., Marchi, S., de Sanctis, M. C., Coradini, A., Agarwal, J., Jewitt, D., Weaver, H., Mutchler, M., & & Ammannito, E. 2012, A&A, 537, A71 Larson, S. 2016b, AJ, 151, 12 Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016, A'Hearn, M. F., Schleicher, D. G., Millis, R. L., Feldman, P. D., & Thompson, D. T. 1984, AJ, 89, 579 Aihara, H., Allende Prieto, C., An, D., et al. 2011, ApJS, 193, 29 Bida, T. A., Dunham, E. W., Massey, P., & Roe, H. G. 2014, in Proc. SPIE, Vol. 9147, Ground-based and ArXiv e-prints, arXiv:1612.05560 Dressler, A., Bigelow, B., Hare, T., et al. 2011, PASP, 123, 288 Elst, E. W., Pizarro, O., Pollas, C., et al. 1996, IAUC, 6456 Fink, U., & Rubin, M. 2012, Icarus, 221, 721 Finson, M. J., & Probstein, R. F. 1968, ApJ, 154, 327 Flewelling, H. A., Magnier, E. A., Chambers, K. C., et al. Airborne Instrumentation for Astronomy V, 91472N 2016, ArXiv e-prints, arXiv:1612.05243 20 Hsieh et al. Fukugita, M., Ichikawa, T., Gunn, J. E., et al. 1996, AJ, Hsieh, H. H., Hainaut, O., Novakovi´c, B., et al. 2015b, 111, 1748 ApJL, 800, L16 Fulle, M., Marzari, F., Della Corte, V., et al. 2016, ApJ, 821, 19 Gimeno, G., Roth, K., Chiboucas, K., et al. 2016, in Proc. SPIE, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, 99082S Hui, M.-T., & Jewitt, D. 2015, AJ, 149, 134 Hui, M.-T., Jewitt, D., & Du, X. 2017, AJ, 153, 141 Jewitt, D. 1996, Earth Moon and Planets, 72, 185 Jewitt, D., Agarwal, J., Weaver, H., et al. 2016, AJ, 152, 77 Jewitt, D., Hsieh, H., & Agarwal, J. 2015, Asteroids IV Gunn, J. E., Carr, M., Rockosi, C., et al. 1998, AJ, 116, (Tucson, University of Arizona Press), 221 -- 241 3040 Jewitt, D., Yang, B., & Haghighipour, N. 2009, AJ, 137, Gunn, J. E., Siegmund, W. A., Mannery, E. J., et al. 2006, 4313 AJ, 131, 2332 Kinoshita, D., Chen, C.-W., Lin, H.-C., et al. 2005, Gwyn, S. D. J., Hill, N., & Kavelaars, J. J. 2012, PASP, ChJA&A, 5, 315 124, 579 Haghighipour, N., Maindl, T. I., Schafer, C., Speith, R., & Kossacki, K. J., & Szutowicz, S. 2012, Icarus, 217, 66 Kres´ak, L. 1979, Asteroids (Tucson, University of Arizona Dvorak, R. 2016, ApJ, 830, 22 Press), 289 -- 309 Hofmann, M., Sierks, H., & Blum, J. 2017, MNRAS, 469, S73 Landolt, A. U. 1992, AJ, 104, 340 Licandro, J., Moreno, F., de Le´on, J., et al. 2013, A&A, Hook, I. M., Jørgensen, I., Allington-Smith, J. R., et al. 550, A17 2004, PASP, 116, 425 Licandro, J., Tancredi, G., Lindgren, M., Rickman, H., & Hsieh, H. H. 2009, A&A, 505, 1297 Hsieh, H. H. 2014a, in IAU Symposium, Vol. 293, IAU Symposium: Formation, Detection, and Characterization of Extrasolar Habitable Planets, 212 -- 218 -- . 2014b, Icarus, 243, 16 Hsieh, H. H. 2016, in IAU Symposium, Vol. 318, IAU Symposium: Asteroids: New Observations, New Models, 99 -- 110 Hutton, R. G. 2000, Icarus, 147, 161 Magnier, E. A., & Cuillandre, J.-C. 2004, PASP, 116, 449 Magnier, E. A., Schlafly, E., Finkbeiner, D., et al. 2013, ApJS, 205, 20 Magnier, E. A., Schlafly, E. F., Finkbeiner, D. P., et al. 2016, ArXiv e-prints, arXiv:1612.05242 Meech, K. J., & Castillo-Rogez, J. C. 2015, IAU General Assembly, 22, 2257859 Hsieh, H. H., & Chavez, J. 2017, Central Bureau Electronic Morbidelli, A., Chambers, J., Lunine, J. I., et al. 2000, Telegrams, 4388 Meteoritics and Planetary Science, 35, 1309 Hsieh, H. H., Forshay, P., & Schwamb, M. 2016, Central Moreno, F., Lara, L. M., Licandro, J., et al. 2011, ApJL, Bureau Electronic Telegrams, 4307 738, L16 Hsieh, H. H., & Haghighipour, N. 2016, Icarus, 277, 19 Hsieh, H. H., Ishiguro, M., Knight, M. M., et al. 2018a, AJ, Moreno, F., Licandro, J., Cabrera-Lavers, A., & Pozuelos, F. J. 2016, ApJ, 826, 137 156, 39 Nesvorn´y, D., Bottke, W. F., Vokrouhlick´y, D., et al. 2008, Hsieh, H. H., & Jewitt, D. 2006, Science, 312, 561 Hsieh, H. H., Jewitt, D., & Ishiguro, M. 2009, AJ, 137, 157 Hsieh, H. H., Jewitt, D. C., & Fern´andez, Y. R. 2004, AJ, ApJL, 679, L143 Novakovi´c, B., Hsieh, H. H., & Cellino, A. 2012, MNRAS, 424, 1432 127, 2997 O'Brien, D. P., Izidoro, A., Jacobson, S. A., Raymond, Hsieh, H. H., Meech, K. J., & Pittichov´a, J. 2011, ApJL, S. N., & Rubie, D. C. 2018, SSRv, 214, 47 736, L18 O'Brien, D. P., Morbidelli, A., & Levison, H. F. 2006, Hsieh, H. H., Novakovi´c, B., Kim, Y., & Brasser, R. 2018b, Icarus, 184, 39 AJ, 155, 96 Hsieh, H. H., & Sheppard, S. S. 2015, MNRAS, 454, L81 Hsieh, H. H., Yang, B., & Haghighipour, N. 2012a, ApJ, Prialnik, D., & Rosenberg, E. D. 2009, MNRAS, 399, L79 Raymond, S. N., & Izidoro, A. 2017, Icarus, 297, 134 Raymond, S. N., Quinn, T., & Lunine, J. I. 2004, Icarus, 744, 9 168, 1 Hsieh, H. H., Yang, B., Haghighipour, N., et al. 2012b, Read, M. T., Bressi, T. H., Gehrels, T., Scotti, J. V., & ApJL, 748, L15 -- . 2012c, AJ, 143, 104 Hsieh, H. H., Denneau, L., Wainscoat, R. J., et al. 2015a, Icarus, 248, 289 Christensen, E. J. 2005, IAUC, 8624 Schlafly, E. F., Finkbeiner, D. P., Juri´c, M., et al. 2012, ApJ, 756, 158 Schorghofer, N. 2008, ApJ, 682, 697 Reactivations of 238P and 288P 21 -- . 2016, Icarus, 276, 88 Tonry, J. L., Stubbs, C. W., Lykke, K. R., et al. 2012, ApJ, Steckloff, J. K., Graves, K., Hirabayashi, M., Melosh, H. J., 750, 99 & Richardson, J. E. 2016, Icarus, 272, 60 Tancredi, G. 2014, Icarus, 234, 66 Tody, D. 1986, in Proc. SPIE, Vol. 627, Instrumentation in Astronomy VI, 733 Tody, D. 1993, in Astronomical Society of the Pacific Conference Series, Vol. 52, Astronomical Data Analysis Software and Systems II, 173 Vincent, J.-B., Bodewits, D., Besse, S., et al. 2015, Nature, 523, 63 Waniak, W., & Drahus, M. 2016, in AAS/Division for Planetary Sciences Meeting Abstracts, Vol. 48, AAS/Division for Planetary Sciences Meeting Abstracts, 504.01 York, D. G., Adelman, J., Anderson, Jr., J. E., et al. 2000, AJ, 120, 1579
1009.3558	1	1009	2010-09-18T14:46:21	Chirikov Diffusion in the Asteroidal Three-Body Resonance (5,-2,-2)	[ "astro-ph.EP", "nlin.CD" ]	The theory of diffusion in many-dimensional Hamiltonian system is applied to asteroidal dynamics. The general formulations developed by Chirikov is applied to the Nesvorn\'{y}-Morbidelli analytic model of three-body (three-orbit) mean-motion resonances (Jupiter-Saturn-asteroid system). In particular, we investigate the diffusion \emph{along} and \emph{across} the separatrices of the (5,-2,-2) resonance of the (490) Veritas asteroidal family and their relationship to diffusion in semi-major axis and eccentricity. The estimations of diffusion were obtained using the Melnikov integral, a Hadjidemetriou-type sympletic map and numerical integrations for times up to $10^{8}$ years.	astro-ph.EP	astro-ph	Celest Mech Dyn Astron manuscript No. (will be inserted by the editor) Chirikov Diffusion in the Asteroidal Three-Body Resonance (5, −2) −2, F. Cachucho · P. M. Cincotta · S. Ferraz-Mello Received: date / Accepted: date Abstract The theory of diffusion in many-dimensional Hamiltonian system is ap- plied to asteroidal dynamics. The general formulations developed by Chirikov is ap- plied to the Nesvorn´y-Morbidelli analytic model of three-body (three-orbit) mean- motion resonances (Jupiter-Saturn-asteroid system). In particular, we investigate the diffusion along and across the separatrices of the (5,−2,−2) resonance of the (490) Veritas asteroidal family and their relationship to diffusion in semi-major axis and eccentricity. The estimations of diffusion were obtained using the Mel- nikov integral, a Hadjidemetriou-type sympletic map and numerical integrations for times up to 108 years. Keywords Chaotic motion, Chirikov theory, asteroid belt, Nesvorn´y-Morbidelli model, three-body resonances 1 Introduction The application of chaotic dynamics concepts to asteroidal dynamics led to the un- derstanding of the main structural characteristics of asteroids distribution within the solar system. It was verified that chaotic region are generally devoid of larger asteroids while, in contrast, regular regions exhibit a great number of them (see for instance, Berry 1978, Wisdom 1982, Dermott and Murray 1983, Hadjidemetriou and Ichtiaroglou 1984, Ferraz-Mello et al. 1997, Tsiganis et al. 2002b, Knezevi´c 2004, Varvoglis 2004) It was soon accepted that chaos was related inevitably to F. Cachucho E-mail: [email protected] P.M. Cincotta Facultad de Ciencias Astron´omicas y Geof´ısicas Universitas National de La Plata, La Plata, Argentina E-mail: [email protected] S. Ferraz-Mello Instituto de Astronomia, Geof´ısica e Ciencias Atmosf´ericas Universidade de Sao Paulo, Sao Paulo, Brazil E-mail: [email protected] 0 1 0 2 p e S 8 1 . ] P E h p - o r t s a [ 1 v 8 5 5 3 . 9 0 0 1 : v i X r a 2 F. Cachucho et al. instability, which may be local or global. Subsequent investigations searched for initial conditions leading to instabilities in relatively short time. In many appli- cations, the determination of Lyapunov exponent on a grid of initial conditions was used to get quantitative informations on stability. The inverse of the largest Lyapunov exponent, called Lyapunov time, should be in some way linked to the characteristic time for the onset of chaos (Morbidelli and Froeschl´e, 1996). However, some investigations have shown that many asteroids exhibit inter- mediary behavior between chaos and regularity. The first registered case was the asteroid (522) Helga (Milani and Nobili, 1992). This asteroid was in chaotic or- bit with a Lyapunov time inch shorter than the age of the solar system, but it exhibited a long period stability. No significant evolution was observed in the or- bital elements of (522) Helga for times up to one thousand times its Lyapunov time. Since then, other asteroids have been shown to have Lyapunov times much shorter than the stability times unraveled by simulations (e.g., Trojans, cf. Milani 1993). Currently, this behavior is known in literature as stable chaos (Milani et al. 1997, Tsiganis et al. 2002a, Tsiganis et al. 2002b). Indeed, there is strong evidence that local instability does not mean chaotic diffusion, in the sense that nothing can be said about how much global or local integrals (or orbital elements) could change in a chaotic domain, even when a linear stability analysis shows rather short Lyapunov times (see Giordano and Cincotta, 2004, Cincotta and Giordano 2008). Nesvorn´y and Morbidelli (1998, 1999) demonstrated that one source of stable chaos is related with three-body (three-orbit) mean-motion resonances (Jupiter- Saturn-asteroid system). They observed that asteroids in these resonances exhibit a slow diffusion in eccentricity and inclination, but no diffusion in the semi-major axis. According to the estimates of Nesvorn´y and Morbidelli (1998), about 1500 among the first numbered asteroids are affected by three-body mean-motion res- onances. The three-body mean-motion resonances are very narrow since they appear at second order in planetary masses, their typical width being ∼ 10−3 AU, but they are much more dense (in phase space) than standard two-body mean-motion reso- nances of similar size. Nesvorn´y and Morbidelli (1999) developed a detailed model for the three-body mean-motion resonance and presented analytical and numerical evidence that most of them exhibit a highly chaotic dynamics (at moderate-to-low- eccentricities) which may be explained in terms of an overlap of their associated multiplets. By multiplet, we refer to all resonances for which the time-derivative of the resonant angle, σp,pJ ,pS , satisfies σp,pJ ,pS = mJ λJ + mS λS + m λ + p + pJ J + pS S ≃ 0, (1) for given (mJ , mS, m) ∈ Z3/{0}. In (1) the λ's and 's denote, as usual, the mean longitudes and perihelion longitudes, respectively; (p, pJ , pS) ∈ Z3 are integers the asteroid). such that Pi (mi + pi) = 0 for i ranging over three bodies (Jupiter, Saturn and We will be dealing in this paper with the case (mJ , mS, m) = (5,−2,−2). This three-body resonance seems to dominate the dynamics of, for instance, the aster- oids (3460) Ashkova, (2039) Payne-Gaposchkin and (490) Veritas (see Nesvorn´y and Morbidelli 1999). In the case of the first two of those asteroids (with relatively large eccentricity, ∼ 0.15 − 0.20), their behavior looks regular over comparatively Chirikov-Arnold in the (5, −2, −2) resonance 3 long time-scales (typically ∼ 1 − 10 × 103 years) while in case of (490) Veritas (with eccentricity, ∼ 0.06) its dynamics looks rather chaotic over similar time- scales. The determination of the age of (490) Veritas family has been the concern of some authors who studied stable chaos (Milani and Farinella 1994, Knezevi´c 1999, Knezevi´c et al. 2002, Knezevi´c 2003, Knezevi´c et al. 2004, Tsiganis et al. 2007, Knezevi´c 2007, Novakovi´c et al. 2009). Herein, we investigate chaotic diffusion along (and also across) the above men- tioned three-body mean-motion resonance by means of a classical diffusion ap- proach. We use (partially) the formulations given by Chirikov (1979). That for- mulation are developed to study specifically Arnold diffusion or some kind of diffusion that geometrically resembles it initially called Fast-Arnold diffusion by Chirikov and Vecheslavov (1989, 1993), as well as the so called modulational diffu- sion (Chirikov et. al, 1985). However, although from the purely mathematical point of view several restrictions should be imposed, there are many unsolved aspects regarding general phase space diffusion (see for instance Lochak 1999, Cincotta 2002, Cincotta and Giordano 2008). The structure of the Hamiltonian used by Chirikov in his first formulation is similar to the Hamiltonians obtained with the perturbations theories of Celes- tial Mechanics. In particular, the Hamiltonians of analytic models of the three- body mean-motion resonances are directly adaptable, with some restrictions, to Chirikov's formulations. Let us mention that some progress has been done in the study of Arnold diffu- sion, particularly when applied to simple dynamical systems, like maps, the latest ones are for instance, the works of Guzzo et. al (2009a),(2009b), Lega (2009). However the link between strictly Arnold diffusion and general diffusion in phase space is still an open matter. Indeed, Arnold diffusion requires a rather small per- turbation, when the measure of the regular component of phase space is close to one. Thus, as far as we know, almost all investigations regarding Arnold diffu- sion involves relatively simple dynamical systems like quasi -- integrable maps. In more real systems, like the one investigated in this paper, the scenario is much more complex in the sense that the domain of the three body resonance is almost completely chaotic. Finally, this work is justified by the fact that an application of all those theories to real astronomical models is still needed. In Sect. 2, we summarize the general problem of computation of the diffusion rate along the resonance and we discuss the limitations and difficulties to follow Chirikov approach in case of this particular three-body mean-motion resonance. Section 3 is devoted to the resonant Hamiltonian (given in Nesvorn´y and Morbidelli (1999)) and its application to the (5,−2,−2) resonance. In Sect. 4, we construct the simplified (or two-resonance) and complete (or three-resonance) numerical models used in our investigations. Moreover, informations about the algorithms and initial conditions used for numerical integrations and the procedure for estimation of the diffusion are also considered in this section. In Sect. 5, we discuss the numerical results on diffusion in the (5,−2,−2) resonance. In this application of the Chirikov theory, we are concerned with the role of the perturbing resonances in the diffusion across and along the (5,−2,−2) resonance and their relationship to diffusion in semi-major axis and eccentricity. Finally, in Sect. 6, we investigate the behavior of the asymptotic diffusion decreasing the intensity of the perturbations in the (5,−2,−2) resonance. In this case, we are interested in the study of the diffusion 4 F. Cachucho et al. under the action of an arbitrarily weak perturbation considering scenarios close to that of the Arnold diffusion. 2 Chirikov's Diffusion Theory In this section we give Chirikov's (1979) as well as Cincotta's (2002) description of diffusion theory in phase space in order to provide a self-consistent presentation of the subject. Since most of the results and discussions given here are included in at least these two reviews, we just address the basic theoretical aspects. Let us consider a Hamiltonian system having several periodic perturbations that can create resonances. The initial conditions are chosen such that the system is in the domain of a main resonance, called guiding resonance. The term of pertur- bation corresponding to the guiding resonance is separated from the others, which will be called of perturbing resonances. The Hamiltonian has the following form with H = H0 (I) + ǫVG (I) cos (mG · θ) + ǫV (I, θ) , ǫV = ǫ Xm6=mG Vm (I) cos (m · θ) , (2) (3) where VG and mG are, respectively, the amplitude and resonant vector of the guiding resonance, Vm and m are, respectively, the amplitude and resonant vectors of perturbing resonances. Here (I, θ) are the usual N -dimensional action-angle coordinates for the unperturbed Hamiltonian H0 (N ≥ 3), the vectors mG, m ∈ ZN /{0} and VG, Vm are real functions. The small parameter perturbation, ǫ, is a real number such that ǫ ≪ 1. The resonance condition is fixed by S (Ir) =mG · ω (Ir) = 0. (4) The surface S (Ir) = 0 in the action space, is called resonant surface. 2.1 Dynamics of the guiding resonance in the actions space Let us first consider the simple case of one single resonance, that is, let us assume that all Vm = 0 for m 6= mG, and we chose initial conditions close to the separatrix of the guiding resonance. In the ω-space, the resonance condition mG·ωr = 0 has a very simple structure, just a (N − 1)-dimensional plane the normal of which is the resonant vector mG. In the I-space, mG · ωr = 0 leads to the (N − 1)-dimensional resonant surface S (Ir) = 0, whose local normal at the point I = Ir is nr =(cid:18) ∂ ∂I [mG · ω (I)](cid:19)I=Ir (5) In addition, we consider the (N − 1)-dimensional surface H0 (I) = E (in I- space) and, if we suppose that ω (Ir) is an one-to-one application, we can also write eH0(ω) = H0 (I (ω)) = E (in ω-space). The manifolds defined by the intersection of both resonant and energy surfaces has, in general, dimension N −2. By definition, the frequency vector ω is normal to the energy surface in I-space, since it is the I-gradient of H0. The latter condition, Chirikov-Arnold in the (5, −2, −2) resonance 5 together with the resonance condition Eqn. (4), shows that the resonant vector mG lies on a plane tangent to the energy surface at I = Ir. Furthermore, the equations of motion (only with H0 and the guiding resonant term) show that I is parallel to the constant vector mG. Thus the motion under a single resonant perturbation lies on the tangent plane to the energy surface at the point I = Ir in the direction of the resonant vector. 2.2 Local change of basis Now, let us introduce a canonical transformation (I, θ) → (p, ψ) by means of a generating function F (p, θ) = NXi=1 I r i + NXk=1 pkµki! θi, (6) where µik is a N × N matrix with µ1i = (mG)i. The transform action equations are Ii = I r i + NXk=1 pkµki; ψk = µkℓθℓ. NXℓ=1 (7) The phases ψk, k = 1, . . . , N are supposed to be non degenerate, i.e., ∂H0 ∂Ik 6= 0. As Cincotta (2002) has shown, this transformation should better be thought as a local change of basis rather than as a local change of coordinates. The action vector whose components are(cid:0)Ij − I r j(cid:1) in the original basis(cid:8)uj , j = 1, . . . , N(cid:9), has components pj in the new basis (cid:8)µj , j = 1, . . . , N(cid:9) constructed taking advantage of the particular geometry of resonances in action space. We choose, µ1 = m1 ≡ mG and since the vector mG is orthogonal to the frequency vector ωr (due to the resonance condition), it seems natural to take µ2 = ωr/ωr . The remaining vectors of the basis are µk = ek, k = 3, . . . , N, the vectors ek are orthonormal to each other and to µ2. Let us define one of the ek, say es, orthogonal to the normal nr to the guiding resonance surface. In general, all the vectors ek will be orthogonal also to µ1, except es. In general, es will not be orthogonal to mG. Then, considering N = 3 and since p = piµi, i = 1, . . . , 3, we can say that p1 measures the deviations of the actual motion from the resonant point across the guiding resonance surface, p3 measures the deviation from the resonant value along the guiding resonance, while p2 measures the variations in the unperturbed energy. For N ≥ 3 degrees of freedom the subspace of intersection of the two surfaces leads to a manifold of N−2 dimensions. Following Chirikov (1979), this subspace is called diffusion manifold. The N − 2 vectors ek locally span (at the resonant value) a tangent plane to the diffusion manifold called the diffusion plane. Then, in the new basis, the action vector may be written as: p = p1mG + p2ωr/ωr + q, where We write now the Hamiltonian (2) in terms of the new components of the action. Expanding up to second order in pk, using the orthogonal properties of the new basis, recalling that ψ1 is the resonant phase and neglecting the constant q is confined to the diffusion plane q =Pk qkek with qk = pk for k = 3, . . . , N. 6 F. Cachucho et al. terms, we obtain for (k, ℓ) 6= (1, 1) H (p, ψ) ≈ p2 1 2MG + ǫVG cos ψ1 +(cid:12)(cid:12)ωr(cid:12)(cid:12) p2 + NXk=1 NXℓ=1 pkpℓ 2Mkℓ + ǫV (ψ) , (8) with 1 Mkℓ = NXi=1 NXj=1 µki 1 MG = 1 M11 = mGi ∂ωr i ∂Ij ∂ωr i ∂Ij µℓj , mGj; (9) (10) where we have written VG, V (ψ) instead of VG (p) , V (p, ψ). These functions are evaluated at the point I = Ir or p = 0. In absence of perturbation (V = 0), the components pk, k = 2, . . . , N are inte- grals of motion, which we set equal to zero so that Ir is a point of the orbit. Then the Hamiltonian (8) reduces to where H (p,ψ) ≈ H1 (p1, ψ1) + ǫV (ψ) , H1 = p2 1 2MG + ǫVG cos ψ1 (11) (12) is the resonant Hamiltonian associated to the guiding resonance. It is a simple pendulum. Note that the stable equilibrium point of the pendulum is ψ1 = π if MGVG > 0, or ψ1 = 0 if MGVG < 0. To transform the phase variables, we take into account that the dot product is invariant under a change of basis. Recalling that ψk =Pℓ µkℓθℓ, then if ν denotes the vector m in the new basis, we have: ϕm ≡ m·θ = ν · ψ, where mk =Pℓ νℓµℓk. As we can see, while the mk are integers, the quantities νk are, in general, non- integer numbers, due to the scaling of the phase variables. 2.3 Changes due to perturbation As mentioned above, for V = 0 the pk are integrals of motion and since H1 is also an integral, we have the full set of N unperturbed integrals: H1, p2, qk, k = 3, . . . , N. But if we switch on the perturbation, these quantities will change with time. This can be seen using the equations of motion for the Hamiltonian (8), where ψj = ∂H/∂pj, j = 1, . . . , N. Performing derivatives and integrating, considering pℓ, we obtain that for V = 0, pℓ (ℓ 6= 2) are constants and p1 = MG ψ1 −PN MG M1ℓ ℓ=2 ψ1 (t) + ψk0, k > 1 (13) ϕm = m · θ = ν · ψ = ξmψ1 (t) + ωmt + βm + Km, (14) ψk (t) = ωr tδ2k + NPℓ=2(cid:16) 1 Mkℓ − MG MkℓM1ℓ(cid:17) pℓt + MG Mk1 where δij is the Kronecker's delta and ψj0 is a constant. To get ϕm (t), we evaluate the dot productPi νiψi Chirikov-Arnold in the (5, −2, −2) resonance where ξm = and βm is a constant and νk (m) MG Mk1 , ωm = m · ωr, NXk=1 The second relation of (12) is obtained taking into account that Km = MG Mkℓ − νk (m)(cid:18) 1 NXℓ=2 m · ωr = Xi MkℓM1ℓ(cid:19) pℓt. νi (m) µi! ·(cid:0)µ2(cid:12)(cid:12)ωr(cid:12)(cid:12)(cid:1) ωm = m · ωr = ν2 (m)(cid:12)(cid:12)ωr(cid:12)(cid:12) . pk (t) ≈ ǫ Xm6=m G 7 (15) (16) (17) (18) (19) and the fact that, since µ2 is orthogonal to all µi, i 6= 2 and µ2 · µ2 = 1, the dot product only contributes to i = 2. Then, We are now ready to compute the time variation of the unperturbed integrals. From (11) and (3), for pk = −∂H/∂ψk, k 6= 1, we easily find m sin ϕm (t) . νk (m) V r m = Vm (Ir) . This equation holds for every component of the momentum where V r p, except for p1. Since p1 is not an integral, we use H1, instead of p1. Chirikov (1979) calculated the total variation of H1 with the aim of construct- ing a whisker map to describe the Arnold diffusion. However, instead of it, we prefer, in the study of three-body resonances, to compute the evolution of the components of the momentum p by means of numerical integrations or, alterna- tively, by mean of a Hadjidemetriou-type sympletic map (see Sect. 4). However, we use a variation of Chirikov's construction to obtain a theoretical estimate of the slow diffusion. We then proceed and compute the total variation of pk. For details about the construction of the whisker map we refer to Chirikov (1979, Sect. 7.3) (see also Cincotta 2002 and the Appendix B of Ferraz-Mello 2007). If ǫ is small enough, the phase space domains associated with all resonances present in (3) do not overlap. Then a standard procedure is to replace ψ1 (t) and ψ1 (t) by the values on the unperturbed separatrix and to solve analytically (19). We make first the integration of (19) over a complete trajectory inside the stochastic layer assuming that ψ1 = ψsx and Km = 0. Indeed, as mentioned previously for V = 0 the pℓ (ℓ 6= 1) are integrals of motion and the phases ϕm can be estimated considering pℓ (ℓ 6= 1) = 0, such that Km = 0. Then, the total variations of the pk's are given by 1 ∆pk (t) ≈ ǫ Xm6=m G νk (m) V r mZ +∞ −∞ sin ϕsx m (t) dt, (20) m (t) = ξmψsx where ϕsx considering the known solutions for the phase ψsx 1 (t) + ωmt + βm. The estimate of integral into (20) is done 1 (t) obtained near both branches 8 F. Cachucho et al. of unperturbed separatrix of the pendulum H1. More details about these calcu- lations are given in the appendix of this paper. Here we only described the main steps and the final result for ∆pk (t). Chirikov shown that the contributions of integral in (20) in both branches of separatrix are described in terms of the Melnikov integral with arguments ± λm = ±(cid:12)(cid:12)(cid:12)(cid:12) ωm ΩG(cid:12)(cid:12)(cid:12)(cid:12) , where the double sign indicates the both separatrix branches and ΩG is the proper frequency of the pendulum Hamiltonian H1. In order to simplify the calculations, Chirikov considered only even perturbing resonances and the contribution of Mel- nikov integral with negative argument was neglected under the condition λm ≫ 1. In contrast, the perturbations in the three-body mean-motion resonance model are non even and the arguments are small. Moreover, the asymmetry in the Nesvorn´y- Morbidelli model implies that the time of permanence of the motion near each separatrix is different. Thus, we introduce the factor RT which takes into account the difference in the time of permanence of the motion in each separatrix branch. Hence, after some algebraic manipulations the Eqn. (20) is rewritten as ∆pk ≈ ǫ ΩG Xm6=mG νk (m) Qm sin ϕ0 m, (21) (22) (23) Qm = V r m = ϕsx with m(cid:2)RT A2ξm (λm) + (1 − RT ) A2ξm (−λm)(cid:3) , m(cid:0)t = t0(cid:1) with ψsx where ϕ0 estimate for the total variation of the momenta pk's inside the stochastic layer around of separatrix of the pendulum Hamiltonian H1, and it is valid for non-even perturbation and for small λm. Estimations of the Melnikov integral, A2ξm (λm), in terms of ordinary function can be obtained from the values of λm and ξm. On the other hand, the factor RT can be estimated from numerical experiments. 1 (cid:0)t = t0(cid:1) = π. Equation (22) is a theoretical 2.4 The diffusion rate In Chirikov's theory of slow diffusion, each resonance has a role in the dynamics of system. The main resonance, that is the guiding resonance, defines the domain where diffusion occurs. The stronger perturbing resonance is called layer reso- nance. That resonance perturbs the guiding resonance separatrix and it generates the stochastic layer and its properties (width, KS-entropy, etc.). Thus, the layer resonance controls the dynamics across the stochastic layer. The weaker perturb- ing resonances are called driving resonances. They perturb the stochastic layer and control the dynamics along the stochastic layer. Then, the driving resonances are responsible for the drift along the stochastic layer, i.e., the slow diffusion. We are interested in obtaining an analytical estimate for the slow diffusion. To fulfill this task, we will estimate the diffusion in the actions whose direction is given along the stochastic layer. We introduced the slow diffusion tensor Dij = ∆pi (t) ∆pj (t) Ta i, j = 3, . . . N, (24) Chirikov-Arnold in the (5, −2, −2) resonance 9 where Ta = ln (32e/ws) /ΩG is the characteristic time of the motion within the stochastic layer of the guiding resonance (equal to half the period of libration or to one period of circulation of ψ1 near the separatrix) and the average in the nu- merator is done over successive values of ϕ0 m. Here ws is the width of the stochastic layer given by ws = −ωr Ω2 G ωmL ν (mL) ν2 (mL) ξmL QmL > 0. (25) (see Sects. 6.2 and 7.3 of Chirikov 1979). In the last equation, the subscript L indicate the layer resonance. The components of the diffusion tensor (24) are es- timated using the Eqn. (22). Hence, because of dependence with the phase ϕ0 mD , the average in (24) depends: (1) of the correlation between successive values ϕ0 mD when the system approaches the edges of the layer; (2) of the possible interferences of several driving resonances. However, the analysis done by Chirikov shown that the terms that contribute to the diffusion must have the same phase ϕ0 mD (see Sect. 7.5 of Chirikov 1979 and Cincotta 2002 for more details). Hence, using (22) the diffusion tensor components in (24) are described as Dij = ǫ2 TaΩ2 GXmD νi (mD) νj (mD) Q2 mD sin2 ϕ0 mD . (26) Terms with different mD are averaged out. Now, there still remains the problem of estimating sin2 ϕ0 mD . To solve this problem we need to consider that the structure of the stochastic layer affects the motion of the system. In fact, studies of the slow diffusion theories have shown that the stochastic layer is formed by two different regions. The first, more central, near the unperturbed separatrix, is totally chaotic. The second, more external, near the edge of the stochastic layer, includes domains of regular motion forming stability islands. When the solution approaches the edge of the stochastic layer, it could remain rather close to the neighborhood of those stability islands for long times. This phenomenon, called stickiness, leads to a reduction in the diffusion rate (for more details about the stickiness phenomenon see the recent work of Sun and Zhou 2009 and references therein). Thus, near stability islands some correlations in the phases arise, which dominate the motion across and along the stochastic layer. In this case, the evolution of phases ϕ0 mD cannot be random simultaneously, and their correlation decreases the diffusion rate (see Chirikov 1979, Cincotta 2002). mL and ϕ0 In order to estimate the correlation between sin2 ϕ0 mL , we use the so called reduced stochasticity approximation, introduced by Chirikov (1979) like an additional hypothesis. Hence, the theoretical rate of diffusion given by (26) may be now evaluated and has the form mD and sin2 ϕ0 Dij = ǫ2 2Ω2 GTaXmD RmD νi (mD) νj (mD) Q2 mD i, j = 3, . . . , N. (27) The Eqn. (27) is an estimate for the theoretical diffusion inside the stochastic layer. The diffusion coefficient includes two parameters that reduce the diffusion rate: RT due to non-even perturbations and RmD due to the reduced stochasticity approximation. The expression given here for the diffusion tensor is different of that given by Chirikov because of the introduction of the parameter RT and by the 10 F. Cachucho et al. possibility of having a small argument in the Melnikov integral. Moreover, we have considered that the reduction factor due the reduced stochasticity approximation is different for each driving resonance, while Chirikov considers the same value for all of them. 3 Application to 3-body mean-motion resonance The Hamiltonian, in the extended phase space, associated to a given (mJ , mS, m) resonance, in Delaunay action-angles variables, is H = − 1 2L2 + nJ ΛJ + nS ΛS + vJ ΠJ + vSΠS + Psec + Pres, where λ, , λJ , J , λS, S (28) (29) are the mean longitudes and longitudes of the perihelions of the asteroid, Jupiter and Saturn, respectively, and L = √a; Π = √a(cid:16)p1 − e2 − 1(cid:17) , ΛJ , ΠJ , ΛS , ΠS (30) are the actions conjugated to them. The frequencies nJ , vJ , nS , vS are the mean- motion and perihelion motions of Jupiter and Saturn, respectively. The first term in (28) describes the Keplerian motion of the asteroid and the terms proportional to the planetary actions extend the phase space to incorporate the motion of the angles λ, , λJ , J , λS, S in the unperturbed Hamiltonian. Details concerning the derivation of this Hamiltonian are given by Nesvorn´y and Morbidelli (1999), whose main results and formula are used in this paper. Note that this Hamiltonian does not satisfy the convexity condition, however, this fact should not be a restriction for the application of Chirikov's diffusion theory. The perturbing function, following Nesvorn´y and Morbidelli (1999), has been splitted into its secular and resonant parts Psec = µJ aJ XkJ ,kS ,k,iJ ,iS ,i Psec (αres) ekekJ J ekS S cos (iJ J + iSS + i) Pres = µJ aJ XkJ ,kS ,k,pJ ,pS ,p Pres (αres) ekekJ J ekS S cos (σp,pJ ,pS ) (31) (32) where, αres = ares/aJ is the semi-major axis corresponding to the exact resonance, σp,pJ ,pS = mJ λJ + mSλS + mλ + p + pJJ + pSS, µJ is Jupiter's mass, e, eJ , eS are the asteroid, Jupiter and Saturn's eccentricities, respectively, and Psec (αres), Pres (αres) are given functions that are linear in Saturn's mass (see bellow). The harmonic coefficients satisfy d'Alembert rules, iJ +iS +i = 0, mJ +mS +m+p+pJ + pS = 0 and the series are truncated at some order in kJ +kS +k, iJ +iS +i and mJ+mS+m+p+pJ+pS. Next, we reduce the secular part (31) to the quadratic term in asteroid's eccentricity in order to break the degeneracy of the unperturbed Hamiltonian, and introduce in (28) the new action-angle variables, (cid:0)I′, θ′(cid:1): Chirikov-Arnold in the (5, −2, −2) resonance I′ = (N, NJ , NS, Π, ΠJ , ΠS) (actions) θ′ = (ν, νJ , νS, , J , S) (angles) defined by and ν = mJ λJ + mSλS + mλ, νJ = λJ , νS = λS, L = mN, ΛJ = mJ N + NJ , ΛS = mSN + NS . 11 (33) (34) (35) (36) The variables (ΠJ , J , ΠS, S) remain unchanged. We recall that the resonant perturbation (32) does not depend on νJ and νS (so that NJ , NS are constant that we can take as equal to zero). Let us write I ≡ (N, Π, ΠJ , ΠS) , θ ≡ (ν, , J , S) . Eliminating the constant terms, the Hamiltonian (28) may be written H (I,θ) = H0 (I) + V (I,θ) , (37) (38) where H0 (I) = − 1 2m2N 2 − β0(cid:18)1 + Π mN(cid:19)2 + (mJ nJ + mSnS) N + νJ ΠJ + νSΠS, (39) is the unperturbed Hamiltonian and the perturbation is described by with and V (I,θ) =Xm βm (I) cos (m·θ) , m = (1, p, pJ , pS) , βm (I) = µJ aJ XkJ ,kS ,k Pres (αres) ekekJ J ekS S . (40) (41) (42) Nesvorn´y developed a procedure allowing to obtain the coefficients (42) in terms of power series of the asteroid eccentricity only. In the last column of Table 1 are the coefficients calculated by Nesvorn´y for the guiding (G), layer (L) and driving (D) resonances used in our numerical experiments. We have considered the guiding resonance, defined by the vector mG = (1,−1, 0, 0), the layer resonance, defined by the vector mL = (1, 0,−1, 0) and the driving reso- nance defined by vector mD = (1, 0, 0,−1). The unperturbed separatrices of those resonances in the plane a − e are shown in Fig. (1). The next step is to introduce the Chirikov variables (p,ψ) allowing to have a separate representation of the actions across and along the resonance within the stochastic domain of the guiding resonance. The canonical transformation, is performed by the generating function (6), with a transformation matrix, µ, given by 12 F. Cachucho et al. Table 1 Old and new resonant vectors, and coefficients of the guiding (G), layer (L) and driving (D) resonances (The coefficients were taken from Nesvorn´y and Morbidelli, 1999). vectors m vectors ν G L D (1, −1, 0, 0) (1, 0, −1, 0) (1, 0, 0, −1) (1, 0, 0, 0) (0.55, 0.66, 0.76, 0.70) (0.68, 0.47, 0.92, −0.70) coefficients eβm(cid:0)×10−8(cid:1) 45.59e − 32.24e3 −2.76 + 0.93e2 1.18 − 0.38e2  2ωr 2 2 1 ωr ωr 2vSnr qr √2vJ 2 v 2 −1 ωr ωr 2vSnr 1ωr qr √2vJ 2 v 2 − 0 νJ ωr νJ vS (nr 2 − nr 1) qr √2ωr v − 2 − 0 νS ωr v2 (nr 2 − nr 1) qr 0  (43) µ = where J v2 S + v4(cid:1) (nr 1)2 + (nr 2)2, (cid:12)(cid:12)qr(cid:12)(cid:12) =q(cid:0)v2 (cid:12)(cid:12)nr(cid:12)(cid:12) =q(nr 2 − nr 1)2 + 4ν 2 Sωr2 2 nr2, 2)2. J + 2 (ωr v =qν 2 Once the matrix of the transformation is defined, the new variables (p,ψ) can be rapidly obtained using the relations (7). In the new basis the arguments of the periodic terms change. The new vectors ν defined by m · θ = ν · ψ are shown in Table 1, in addition to the resonant vectors m and their respective coefficients. The procedure of previous section was applied in the Nesvorn´y-Morbidelli model, and leads to the Hamiltonian (8) with N = 4. The three perturbation coefficients βG, βL and βD of the guiding, layer and driving resonances, respec- tively, are calculated at the resonant values Ir, which satisfies (4). In the plane N Π the resonant condition (4) leads to a curve satisfying to Π r2 + C1Π r + C2 = 0, (44) where C1 (N r) and C2 (N r) are given in terms of N r. Then, the solutions of (44) can be obtained analytically for a fixed value of N r. However, in Nesvorn´y- Morbidelli model the coefficients βm's are given as functions of the asteroid ec- centricity (see Table 1). Therefore, we must use the definitions of Delaunay vari- ables (30) to determinate (ar, er). The resonant eccentricity is determined through er =q1 − (1 + Π r/N r)2, where (N r, Π r) satisfies (4). The resonant semi-major axis is determinate using ar = (N r/2)2 . 4 Numerical Experiments In this section, we describe the numerical experiments done to investigate the diffusion across and along the stochastic layer of the three-body mean-motion resonance (mJ , mS, m) = (5,−2,−2) and its relations with the diffusion in semi- major axis and eccentricity. In these investigations the diffusion across will be Chirikov-Arnold in the (5, −2, −2) resonance 13 = (1,−1,0,0) = (1,0,−1,0) = (1,0,0,−1) mG mL mD D D L G L G 0.25 0.20 e 0.15 0.10 0.05 3.170 3.172 3.174 3.176 3.178 3.180 a Fig. 1 Unperturbed separatrices of guiding, layer and driving resonances in the plane (a, e). (Nesvorn´y and Morbidelli, 1999) described by the actions (p1, p2) and the diffusion along by the actions (p3, p4). In order to determine the time evolution of each action p, we use the equations of motion obtained from Hamiltonian (8) with N = 4. Then, for each value pk(t), we use the equations of transformation (7) to obtain the respective values of N(t) and Π(t) and the definition of the Delaunay actions in (30) to obtain a(t) and e(t). Two main models were considered in the numerical experiments: (i) simplified (or two-resonance model) and (ii) complete (or three-resonance model). In the first one, only one term of the perturbation - the layer resonance - is considered. In the complete model, two terms are considered: the layer and one driving resonance. In both cases the guiding resonance is given by mG = (1,−1, 0, 0). Two different techniques were used to construct the solutions. In a first set of experiments, the equations of motion of the Hamiltonian (8) were numerically integrated using the Burlish-Stoer method, for times in the interval 102 ≤ tint ≤ 108 years. The results of these simulations were sampled with an output time step of 10 years. The simulations were done for the eccentricities 0.05 and 0.25 with the initial conditions given on the separatrix of the guiding resonance (p10 = 14 F. Cachucho et al. 2q(cid:12)(cid:12)MGβr G(cid:12)(cid:12), p20 = 0, p30 = 0, p40 = 0, ψ = 0). The main goal in these experiments was the study of the variation of the rate of diffusion across and along as a function of the total time of the simulations. Moreover, we investigate the correlations between the diffusion in the Chirikov actions p and the diffusion in semi-axis major and eccentricity. In the other set of experiments, the simulations were done using an Hadjidemetriou- type sympletic mapping (Hadjidemetriou 1986, 1988, 1991, 1993; Ferraz-Mello 1997; Roig and Ferraz-Mello 1999, Lhotka 2009) defined by the canonical trans- formation (pn, ψn) →(cid:0)pn+1, ψn+1(cid:1) , whose generating function is given by S(cid:16)pn+1, ψn(cid:17) = 3Xi=1 ψn i pn+1 i + ηH(cid:16)pn+1, ψn(cid:17) , (45) (46) (47) where η is the mapping step and the Hamiltonian is given by (8). The mapping equations are pn+1 i = pn i − η ∂H(cid:0)pn+1, ψn(cid:1) ∂ψn i ψn+1 i = ψn i + η ∂H(cid:0)pn+1, ψn(cid:1) ∂pn+1 i i = 1, 2, 3. i i , ψn+1 The procedure to determinate the semi-major axis and eccentricity for each point (pn+1 ) of the trajectory is analogous to that discussed above. The goal of these experiments is to obtain the diffusion contour plots in the region of the 5,-2,-2 resonance in the plane (a, e) (that plane is shown in Fig. 1) for the two models considered. In this case, the total time of integration used is the 108 years with η = 10 years. The initial conditions are defined by the knots of a grid in the plane (a, e), on the rectangle (3.17 ≤ a0 ≤ 3.18)U.A., (0.01 ≤ e0 ≤ 0.30). The initial condition of the state vector p0, for each point of the grid, was obtained using the transformation equations (7) and the definitions of Delaunay variables. The initial condition for the phases is ψk0 = 0, k = 1, . . . 4. The use of the Hadjidemetriou map was instrumental allowing the computation of the solutions starting on each point of the grid which, otherwise, would demand an excessively large amount of CPU-time. The comparison of results provided by the map with those obtained by integrating the Hamiltonian flow, do not show significant differences in the numerical computation of the diffusion coefficient (see below), at least for the two values of the eccentricity used (0.05 and 0.25). Finally, we need a numerical procedure to estimate the diffusion coefficient of each element of the set (p1, p2, p3, p4, a, e). In his investigations, Chirikov (1979, et al. 1979, 1985) used a particular method to determine the diffusion coefficient of the total energy H of the system. After Chirikov (1979), this procedure allows the pro- cesses that are really stochastic to be separated from those associated to bounded oscillations of periodic nature. Chirikov's procedure for experimental determina- tion of the diffusion coefficient consist in dividing the total time of simulation tint in Nk sub-intervals of length (∆t)k and the calculation of the mean value, ¯pi, for every sub-interval. The contribution to the diffusion rate for a given pair ¯pim , sep- arated by interval of time (m − ℓ) (∆t)k, is given by (¯pim − ¯piℓ )2 /m − ℓ (∆t)k. To Chirikov-Arnold in the (5, −2, −2) resonance 15 obtain the rate of diffusion, the contributions of the considered pairs are averaged over all the combinations m 6= ℓ. That is, Dk i = 2 Nk (Nk − 1) Xm>ℓ (¯pim − ¯piℓ )2 (∆t)k (m − ℓ) . (48) The sub-intervals, used to estimate the mean values of quantities ¯pi, were obtained with k = 10 and the length (∆t)10 = tint/10. a, and of the eccentricity, Dk The same procedure was used to determine the diffusion of the semi-major axis, Dk e . We have also estimated the eccentricity variation in these experiments using a definition of diffusion rate of the random walking type (see for example Eqn. (24)): δe ∼pDk e tint. (49) 5 Results and discussion In this sections, we discuss the results obtained in the numerical experiments described above. In the discussion we will call action across to (p1, p2), and across diffusion to (D1, D2), where we suppressed the superscript k. In the same way we call action along to (p3, p4) and along diffusion to (D3, D4). 5.1 The role of number of the perturbing resonances in the diffusion In his theory, Chirikov showed that the number of perturbing resonances is im- portant for the dynamics of systems with many-dimensional Hamiltonians. The results, in this case, repeat what is know from the general theory of Hamiltonian systems. In a system with two degrees of freedom, the resonances may be iso- lated by KAM tori, but for (N > 3) the dimensionality may allows, in principle, a solution to visit the whole phase space when t → ∞. Several experiments, using the Burlish-Stoer integrator, were done see the way in which the number of perturbing resonances in the diffusion behavior. Figure 2 shows the results for the diffusion coefficients Di, i = 1, 2, 3, 4 in the simplified and complete models as function of total integration time for eccentricities equal to 0.05 and 0.25, . In the plots of Fig. 2, we see that the estimated diffusion increases in the low eccentricities up to a maximum reached for 103 − 104 years. This behavior is explained by the fact that the solution needs to fill the stochastic domain in the direction across to it. After that maximum, in the simplified model the diffusion coefficients for all actions decrease continuously. This decrease indicates that the variation of the momenta in both directions, across and along the stochastic layer are bounded (as the total time increase, only the denominator of (48) grows making the result to decrease). As predicted by Chirikov's theory of slow diffusion, the actions p3 and p4 along the resonance do not evolve, notwithstanding the absence of topological barriers for its evolution. Without a driving resonance, there is no long- period evolution of the solution along the stochastic domain. In our experiments, a very distinctive reduction in the diffusion is observed in the case e = 0.25 after tint ∼ 107 years. This behavior is likely due to a sticking of the solution to some regular domain. 16 Di Di 10-24 10-14 10-18 10-22 10-26 D1 D2 . . Simplified Model Complete Model D3 D4 102 103 104 e = 0.05 e = 0.25 F. Cachucho et al. 10-12 10-16 10-20 Simplified Model Complete Model . . 105 tint 106 107 108 103 104 106 107 108 105 tint Fig. 2 Diffusion coefficients of the actions associated with motion across and along the guiding resonance, in experiments over times from 102 −108 years for two different initial eccentricities. The behavior of the diffusion in the complete model is more complicated. In the experiment with e = 0.05, the diffusion coefficients for the actions across the stochastic domain after 108 years are smaller than for the actions along it. This difference reaches approximately four orders of magnitude in this case and is due, probably, to the limitation of the motion across the stochastic layer imposed by its width. For e = 0.25 (right plot of Fig. 2), the diffusion coefficients in the two models present almost the same characteristics observed for e = 0.05, except by the fact that, now, the diffusion in the actions along the stochastic layer, present a slow reduction with the integration time after 104 years. This behavior is likely due to the absence of overlapping of resonance at high eccentricities, in contrast with the case of low eccentricities, where the three resonances overlap (see Fig. 1). 5.2 Diffusion in semi-major axis and eccentricity in the complete model The study of the previous section was completed with the computation of the diffusion coefficients for the orbital elements: semi-major axis and eccentricity in the complete model. Figure 3 presents the results. The results for the actions shown in this figure are the same shown in Fig. 2, but with a magnified scale. We see that, for large total times, there exist a correspondence between the diffusion coefficients of the actions across (p1, p2) and of the semi-major axis, and between the diffusion coefficients of the actions along the resonance (p3, p4) and of the eccentricity. This Chirikov-Arnold in the (5, −2, −2) resonance 17 behavior can be understood observing the geometry of resonance (5,−2,−2) shown in the Fig. 1. The separatrices of resonances are straight lines and the motion, along one of these separatrices, has constant semi-major axis and variable eccentricity. Following the discussion presented in Sect. 5.1, and the comparison done in the previous section for the simplified and completed models, we know that the drift along the separatrices only occurs if there is at least one driving resonance. Hence, the eccentricity diffusion is due to the driving resonance. As a complement to the previous discussion, we note that the variations in semi-major axis occurs in the horizontal direction, the same direction of the actions (p1, p2). The behavior of the diffusion in semi-major axis is similar to the diffusion of the actions across the resonance (p1, p2) and is bounded by the width of the stochastic domain. A consequence of this fact is that the diffusion coefficient in the semi-major axis is smaller than that for the eccentricity (in the complete model, the diffusion along is not bounded). The Fig. 4 shows the variation of the eccentricity calculated using initial con- ditions forming a grid in the plane (a, e) (the same grid of Fig. 5) in tint = 108 years. The results for the simplified model are shown in Fig. 4(a). In this case, the larger variations in eccentricity occur for small values of the eccentricity. Two shallow maximums are formed, which are likely related with the eccentricity value at the intersection of the separatrices of the guiding and layer resonances (the only secondary resonance considered in the simplified model). The results for the com- plete model are shown in Fig. 4(b). In this case, the eccentricity variation reaches high values in the domain of low eccentricities - between 0.01and 0.125 - with a maximum for < e >∼ 0.05. This maximum is certainly a result of the overlapping of the resonances in low eccentricities, forcing the actions along the resonance. In this model for mean eccentricities between 0.125 and 0.20 the variations are of the same order. The distributions observed in the Fig. 4(b) is in agreement with Nesvorn´y's unpublished data for 45 numbered asteroids of the (5,−2,−2) resonance (see the Table 2 in Nesvorn´y and Morbidelli 1998). The use of the models with only one perturbing resonance does not allow to get the distribution of the eccentricity variation observed in Fig. 4(b). 5.3 The stochastic domain in the plane (a,e). Dependence on the initial conditions The diffusion coefficients were calculated on a large set of initial conditions to assess the domain where the solutions present stochastic behavior. The analysis was done using simulations over tint ∼ 108 years, on the points of a grid of initial conditions in plane (a, e). A Hadjidemetriou-type sympletic mapping was used instead of expensive numerical integration to allow a large number of simulations. Figure 5 shows the contour plots of the diffusion coefficients of p4. It shows the stochastic domain of the guiding resonance (the light gray areas in Fig. 5). Note that the stochastic domain follows the geometry of the unperturbed separatrix of Fig. 1. Also note that the results for the complete model show a stochastic domain (a, e) larger than that observed for the simplified model. These differences are easily understood if we note the overlapping of the three resonances in the considered range of eccentricities. Figure 1 shows that the sepa- ratrices of the layer and driving resonances are, for almost all eccentricities, interior 18 10-10 10-12 Di 10-14 10-16 10-10 10-12 Di 10-14 10-16 102 103 e = 0.05 e = 0.25 F. Cachucho et al. Da Dp1 Dp2 De Dp3 Dp4 104 105 tint 106 107 108 103 104 106 107 108 105 tint Fig. 3 Diffusion Coefficient for actions across and along, semi-major axis and eccentricity in experiments, obtained for complete model, for times from 102 up to 108 years. Each point is one experiment with initial conditions upon the unperturbed separatrix of the guiding resonance. to the domain of the guiding resonance. At low eccentricities, however, the sep- aratrices cross one another. Thus, in low eccentricities, one solution crossing the chaotic neighborhood of the separatrix of the guiding resonance, also cross the sep- aratrices of the layer and driving resonances. The driving resonance acts pushing the actions along the guiding resonance. The magnitude of the push is determined by the phase ϕmD and amplitude βmD . At variance with the complete model, the simplified model presents very low diffusion, in low eccentricities, as seen in Fig. 2. In this case, the absence of the driving resonance (only the guiding and layer are considered in the simplified model) implies in the absence of evolution along the guiding resonance. A remarkable feature in both results is the formation of a wide region, in the central part of the domain of the guiding resonance, where the diffusion is negligible. The motion appears regular for initial conditions inside that region even when considering very long time spans. This result confirms what Nesvorn´y and Morbidelli (1999) observed in surface of sections for eccentricity 0.20 using this same analytic model reduced to two degrees of freedom and two resonances. This is different from the situation observed in low eccentricities, where the separatrices of the resonances overlap. (a) 19 (b) Chirikov-Arnold in the (5, −2, −2) resonance 0.10 0.09 0.08 0.07 0.06 0.05 0.04 0.03 0.02 0.01 δe 0.00 0.05 0.10 0.15 <e> 0.20 0.25 0.00 0.05 0.10 0.20 0.25 0.15 <e> Fig. 4 Variation of the eccentricity versus mean eccentricity for (a) simplified model and (b) complete model on a net of points in the plane (a, e). 6 Asymptotic behavior Chirikov theory of slow diffusion was constructed to study the diffusion under the action of an arbitrarily weak perturbations, and the diffusion coefficient was com- puted there using the asymptotic estimate of Melnikov's Integral. The asymptotic behavior of the three-body resonance model of Nesvorn´y and Morbidelli was stud- ied using the same technique devised by Chirikov. Figure 6 shows the variation of the diffusion coefficients along the resonance, for two different initial eccentric- ities (0.05 and 0.2), as functions of the parameter λmD = ωmD /ΩG appearing as argument of the Melnikov integrals in Sect. 2.3 in the case the driving resonance m = (1, 0, 0,−1). Small values of λmD are obtained decreasing the intensity of the guiding and perturbing resonances. The Hadjidemetriou-like mapping was used to allow us to compute the solu- tions over 1010 years for a great deal of different conditions. The diffusion coefficient was calculated for initial conditions over the separatrix of the guiding resonance. A background value Db, to be used as reference, was also obtained with initial conditions in the central part of the guiding resonance, far from the separatrices. For small values of the perturbation, the motion in the central part of the res- onance domain is regular and the background diffusion appear as smaller than the diffusion shown by solution starting on the separatrices. For high values of the perturbation intensity, the motion is chaotic over the whole domain and the 20 F. Cachucho et al. Fig. 5 Diffusion coefficients for action along p4 for initial conditions in the interval 3.17 < a0 < 3.18 U.A. for semi-major axis and 0.01 < e0 < 0.30 for eccentricity for both simplified and complete models. The results were obtained for total integration time equal to 108 years. Chirikov-Arnold in the (5, −2, −2) resonance 21 diffusion coefficients in the central part are not different form those of solutions starting on the separatrices. Figure 6 shows the diffusion coefficient D3 for e = 0.2 and for the low eccentric- ity case e = 0.05. The figures for the coefficient D4 are not shown since they are almost identical to those shown for D3. Figure 6 shows the 3 different possibilities. 1. The first section of the figures, corresponding roughly to λmD . 2, is character- ized by complete chaos. For the smallest λmD , one sees the same phenomenon discussed in Sect. 5.1: the background diffusion appear small for some shorter runs because they do not cover the time necessary to allow the solution to fill the chaotic layer; but when time span grows, the diffusion values increase as expected. In this section, in general the diffusion coefficients for solutions start- ing in the central part or on the separatrices are equal showing that the whole resonance domain is chaotic. A few exceptions appear, as shown in Fig. 6(b). In addition we may see in this figure, for λmD ∼ 1, a sudden decrease of the background diffusion indicating that the corresponding solution stuck to some regularity island during its evolution. However this sticking is not permanent and the background diffusion grows when longer time spans are considered. The background values are shown in Fig. 6(a) only for the time span 1010 years to allow a better comparison of the numerical results with the dashed lines representing results from Chirikov's model, 2. For λmD ∼ 2, the background diffusion shows a discontinuity which, for the longest runs, reaches up to 14 orders of magnitude. This means that the cen- ter of the resonance domain becomes regular and the stochasticity remains confined to layers around the separatrix. This is the domain where Chirikov's slow diffusion theories are valid and where the results may be compared to the theoretical results obtained in Sect. 2.4. The integration time is a crucial factor in the detection of the slow diffusion. For instance, one may see that for simulations over only 105 years, the diffusion near separatrix is equal to the background diffusion for values of λmD close to 1, while for simulations over 1010 years, the equality is reached only for λmD = 9. 3. In the last section of the Figs. 6 the solutions starting close to the separatrices show a diffusion equal to the background diffusion. The interpretations is that the stochastic layer is this case is so thin that the used initial conditions are no longer within them. (For that sake, the locus of the separatrices should be computed with very large precision. See e.g. Froeschl´e et al. 2006). One striking feature in this section is that an increase in the time span by a factor 10 means a decrease in the background diffusion by a factor 103. This is a clue for the fact that the solutions are dominated by periodic terms. Indeed, if we consider one periodic term with amplitude proportional to ǫ and frequency ω, its contribution to the average momentum in an interval [a, b] is proportional to 1 ∆t bZa ǫcosωtdt where ∆t = b − a. This integral is elementary and the integration of the result over all frequencies below a upper limit ωlim, gives ǫ ∆t [si (bωlim) − si (aωlim)] (a) n o s u i f f i d d n u o r g k c a b n o s u i f f i l d d o n r A - v o k i r i h C . . . . . . i n a m o d c i t o a h c y l l a b o g l f o e d s n i i n o s u i f f i d e c n a n o s e r e h t f o bg tint (yrs) sep 105 106 107 108 109 1010 theor. (RmD theor. (RmD theor. (RmD = 0.25, RT = 0.6) = 0.25, RT = 0.9) = 0.01, RT = 0.6) F. Cachucho et al. l d o n r A - v o k i r i h C n o s u i f f i d (b) n o s u i f f i d d n u o r g k c a b i n a m o d c i t o a h c y l l a b o g l f o e d s n i i n o s u i f f i d e c n a n o s e r e h t f o e c n a n o s e r e h t f o e d s n i i i n a m o d l e b a t s 22 D3 10-12 10-15 10-18 10-21 10-24 10-27 10-30 10-33 10-36 10-39 10-1 100 λ mD 101 102 100 101 102 λ mD Fig. 6 Asymptotic behavior of the diffusion coefficient D3 for initial conditions over the separatrix and on the central part of the guiding resonance for (a) e = 0.05 and (b) e = 0.20. The dashed lines show the behavior predicted with Chirikov's theory. where si is the sine-integral function. The diffusion coefficients are given by the square of the average variation of the momentum divided by the total time (see Eqn. 24) and then D ∼ ∆t−3. We also have D ∝ ǫ2 ∼ Ω−2 D . The inclination -4 of the straight lines in the log-log plots can be easily checked. G ∼ λ−4 The diffusion of the solutions in the neighborhood of the separatrix may be deter- mined from Eqn. 27. This equation involves the intensity of the perturbation (re- lated to λmD ) and two unknown parameters: the factor of reduction RmD and the factor of odd perturbations RT . The factor of reduction corresponds to Chirikov's hypothesis of reduced stochasticity (due to holes, the solution does not fill the strip around the separatrix); the other factor comes from the fact that the perturbation is not even and thus the values of the diffusion coefficient are not the same for solutions in both separatrices (the solution may remain circulating near one of the separatrices at time different of the time it remain near the other). The results obtained with Chirikov are shown in Figs. 6 by dashed lines. In Figs. 6(a) tree different solutions are shown (calculated with the reduction factors indicated in the figure). The better agreement is obtained with RmD = 0.25. The two values used for RT (0.6 and 0.9) give almost the same result, showing that the motion near deviation for the weakest perturbation (larger λmD ). In the other two figures, only the two solutions with RmD = 0.25 are shown. Chirikov-Arnold in the (5, −2, −2) resonance 23 7 Conclusion Chirikov's theories provide heuristic tools to understand the diffusion observed in both eccentricity and semi-major axis of asteroids inside the (5,−2,−2) resonance. The multi-dimensional Hamiltonians of the three-body (three orbit) mean-motion resonances may be studied with the theories developed by Chirikov and collabo- rators, mainly because of the particular geometry of those resonances in the plane (a, e). The results obtained in this paper for the (5,−2,−2) three-body mean- motion resonance confirms the role of the resonances in the raising of diffusion across and along the main resonance as foreseen in Chirikov's theories. The diffusion calculations presented in this paper show that diffusion in semi- major axis is related with the diffusion in the across actions (p1, p2) while the diffusion in eccentricity is related with the diffusion in the along actions (p3, p4). The diffusion coefficient for the semi-major axis tends to small values showing that the variation of the semi-major axis remains small. It indicates the existence of barriers on both sides of the stochastic layer limiting the motion across the resonance. The comparison between simplified and complete model results shown that the diffusion in eccentricity is presumably due to the presence of at least one resonance driving the motion along the guiding resonance. This behavior is similar to the expected behavior of the Arnold diffusion, but, differently of it, the diffusion here is well apparent and the diffusion coefficients remain high. For this reason it was sometimes called Fast Arnold Diffusion (Chirikov and Vecheslavov 1989, 1993). The structure of the (5,−2,−2) resonance is formed by several overlapping res- onances, particularly at low eccentricities. Thus, diffusion across (5,−2,−2) reso- nance may be no longer limited to the thin chaotic layers (stochastic layers), but it fills the whole resonance zone. The diffusion along the resonance could be due to a multiplet. In this scenario, we have a random motion across the resonance, due to the overlap of several resonances belonging to a multiplet, and another, likely due to weaker resonances, which drive the diffusion along the guiding reso- nance. Arnold diffusion might occur inside the stochastic layer formed around the separatrix of the guiding resonance under the action of sufficiently weak pertur- bations. At variance, thick layer diffusion can appear for perturbation parameters in a broad interval. Although this mechanism show a similar exponential depen- dence of diffusion rate as a function of some system parameters, the mean rate of thick layer diffusion is generally larger than any theoretical estimation of Arnold diffusion. Therefore, it seems that for the real problem would be more appropriate to call the diffusion with another name - asymptotic diffusion of Chirikov-Arnold - due to the fact that this keeps some features of the Arnold diffusion, but late very well characterized by Chirikov like some distinct. As we mentioned before, we believe that the connection between rigorous in- vestigations concerning strictly Arnold diffusion and that observed in real physi- cal systems like this, is still an open subject. As Lochak (1999) pointed out, the global instability properties of near -- integrable Hamiltonian systems are far from well -- understood. It could almost be said that little progress has been made after pioneering work Arnold, and new ideas are definitely called for. Finally, the good results showed that the Chirikov slow diffusion theory can be used in broader investigations considering more resonances for (5,−2,−2), as well 24 F. Cachucho et al. applied for the others three-body (three orbit) mean motion resonances and also can include the inclination of the asteroid orbit. A Estimate of total variation of the momenta pk's To estimate the integral in (20 we use the approach done by Chirikov (1979). In fact, the unperturbed separatrix is defined by where ψsx 1 (t) = 4arctan(cid:16)e±ΩG(t−t0)(cid:17) , psx 1 = ±2 MG ΩGsin ψsx 1 2 , ΩG =sǫ(cid:12)(cid:12)(cid:12)(cid:12) VG MG(cid:12)(cid:12)(cid:12)(cid:12) (50) (51) (52) (53) (54) (55) is the proper frequency of the pendulum Hamiltonian H1. The double sign indicates the two separatrix branches: The positive sign correspond to the upper separatrix(cid:0)0 ≤ ψsx the negative corresponds to the lower separatrix (cid:0)−2π ≤ ψsx points lie at ψ1 = ±π, respectively. (This non usual separation of the intervals where the two branches are considered allows ψ1 and p1 to have the same signal in each separatrix and simplifies the next calculations. For the usual presentation, the reader is referred to the study of the motions near the separatrix of pendulum in the Appendix B of Ferraz-Mello, 2007.) 1 < 2π(cid:1), and 1 < 0(cid:1). The stable equilibrium ψs 1 > 0 (lower separatrix), we have We introduce a time variable change τ = ΩG(cid:0)t − t0(cid:1), with ψsx 1 (cid:0)t0(cid:1) = ψ0 m(cid:1) , m (t) = sin(cid:0)ξmψsx 1 (τ ) + λmτ + ϕ0 sin ϕsx 1 + ωmt0 + βm, with ψ0 1 = π, and where ϕ0 m = ξmψ0 1 = ±π. Then, to Or, after expansion of the right-hand side, λm = ωm ΩG . sin ϕsx m (t) = sin (ξmψsx 1 (τ ) + λmτ ) cos ϕ0 m + cos (ξmψsx 1 (τ ) + λmτ ) sin ϕ0 m. When (55) is substituted into (20), the first term does not give contribution since, by symmetry, Z +∞ −∞ sin [ξmψsx 1 (τ ) + λmτ ] dτ = 0. (56) The contribution of the second term of (55) is determined by the relative signs of ξm and λm. Using the absolute values to ξm and λm, we can introduce the Melnikov integral in the form Z +∞ −∞ cos (ξm ψsx 1 (τ ) ± λm τ ) dτ = 1 ΩG A2ξm (∓ λm) , (57) where A2ξm is the Melnikov integral with argument ± λm. Then, the integral in (20) is Z +∞ −∞ sin ϕsx m (t) dt = 1 ΩG sinϕ0 mA2ξm (∓ λm) . (58) In the other branch of the separatrix, ψsx symmetry of this equation makes it invariant to the sign change of ψsx the same result (58). Indeed, if ψsx 1 < 0 the parity of cosine makes the integral (57) to be 1 has the signal changed, but the particular 1 and, thus, one obtains Z +∞ −∞ cos (ξm ψsx 1 (τ ) ∓ λm τ ) dτ = 1 ΩG A2ξm (± λm) , (59) Chirikov-Arnold in the (5, −2, −2) resonance 25 where we used ψsx introducing the result (59) into (20): 1 (τ ) = −(cid:12)(cid:12)ψsx ΩG Xm6=mG ǫ 1 (τ )(cid:12)(cid:12). Then, the variations in the actions pk can be obtained m(cid:2)A2ξm (λm) + A2ξm (− λm)(cid:3) . (60) ∆pk (t) ≈ νk (m) V r msinϕ0 Chirikov (1979) estimated the diffusion using the result of the last equation. In order to simplify the theoretical estimate of the diffusion, Chirikov considered only even perturbing resonances and neglected the contribution of the perturbation with negative argument under the condition λm ≫ 1. In the case of the three-body mean-motion resonance model, the perturbation are non even and it is not possible to neglect the contribution of perturbations for which λm is small. Then, each perturbation contributes differently when the motion lies close to a separatrix where λm > 0 or λm < 0. Moreover, the odd perturbations in the Nesvorn´y-Morbidelli model makes necessary to take into account that the times of permanence of the motion near each separatrix are not equal. This is done by considering that the solution lies only a fraction of total time near the separatrix with λm > 0. To take into account this asymmetry we introduce the factor RT RT = , (61) Tλ T where Tλ is the time that the solution stay in the neighborhood of separatrix with λm > 0 and T is the total time. Then, the Eqn. (60) is rewritten as ∆pk ≈ ǫ ΩG Xm6=mG νk (m) Qm sin ϕ0 m, (62) (63) with Qm = V r m(cid:2)RT A2ξm (λm) + (1 − RT ) A2ξm (− λm)(cid:3) . Equation (62) is valid for non-even perturbation and for small λm. In order to obtain estimation of (62) in terms of ordinary functions we must know the values of λm and ξm. In general, the relations to A2ξm depend on the exponential term with argument λm (see Appendix A in Chirikov 1979). Acknowledgments The authors are grateful to an anonymous referee for a careful reading of the manuscript and helpful recommendations. PMC is grateful to FAPESP (Brazil) for supporting his visit to the University of Sao Paulo. References 1. Benettin, G., and Gallavotti, G.: Stability of Motions near Resonances in Quasi-Integrable Hamiltonian Systems, J. Stat. Phys. 44, 293-338 (1986). 2. Berry, M.: in Topics in nonlinear dynamics: A tribute to Sir Edward Bullard New York, American Institute of Physics, 1978, p. 16-120 (1978) 3. Chirikov, B.V.: A universal instability of many-dimensional oscillator system. Phys. Rep. 52, 263-379 (1979) 4. Chirikov, B.V., Ford, J. and Vivaldi, F.: Some numerical studies of Arnold diffusion in simple model. In: M. Month. and J.C. Herrera (eds) A.I.P. Conf. Proc.: Nonlinear Dynamics and the Beam-Beam Interaction, N 57, pp. 323-340 (1979) 5. Chirikov, B.V., Lieberman, M.A., Shepelyansky, D.L. and Vivaldi, F.M.: 1985, A theory of modulational diffusion, Physica 14D, 289-304. 6. Chirikov, B.V. and Vecheslavov, V.V.: How fast is the Arnold diffusion? Preprint INP 89-72, Novosibirsk (1989) 7. Chirikov, B.V. and Vecheslavov, V.V.: Theory of fast Arnold diffusion in many frequency system. J. Stat. Phys. 71, 243 (1993) 26 F. Cachucho et al. 8. Cincotta, P.M.: Arnold diffusion: an overview through dynamical astronomy. New Astron- omy Reviews 46, 13-39 (2002) 9. Cincotta, P.M. and Giordano, C.M: Topics on diffusion in phase space of multidimensional Hamiltonian systmes. In: New Nonlinear Phenomena Research, Nova Science Publishers, Inc., pp. 319-336, (2008) 10. Dermott, S. F.and Murray, C. D.: Nature of the Kirkwood gaps in the asteroid belt. Nat.301, 201-205 (1983) 11. Ferraz-Mello, S., Nesvorn, D.and Michtchenko, T. A. On the Lack of Asteroids in the Hecuba Gap. Celest. Mech. Dynam. Astron. 69, 171-185 (1997) 12. Ferraz-Mello, S.: A sympletic mapping approach to the study of the stochasticity of aster- oidal resonances. Cel. Mech. Dyn. Astron. 65, 421-437 (1997) 13. Ferraz-Mello, S.: Canonical Perturbation Theories - Degenerate Systems and Resonance. Springer, New York (2007) 14. Froeschl´e, C., Lega, E. and Guzzo, M.: Analysis of the chaotic behavior of orbits diffusing along the Arnold Web. Celest. Mech. Dynam. Astron. 95, 141-153 (2006) 15. Giordano, C. M. and Cincotta, P.M.: Chaotic diffusion of orbits in systems with divided phase space. A&A. 423, 745-753 (2004) 16. Guzzo, M., Lega, E. and Froeschl, C.: A Numerical Study of Arnold Diffusion in a Priori Unstable Systems. Comm. in Math. Phys., 290, 557-576 (2009a) 17. Guzzo, M., Lega, E. and Froeschl, C.: A numerical study of the topology of normally hy- perbolic invariant manifolds supporting Arnold diffusion in quasi-integrable systems. PhysD, 238, 1797-1807 (2009b) 18. Hadjidemetriou, J. D. and Ichtiaroglou, S.:A qualitative study of the Kirkwood gaps in the asteroids.A&A. 131, 20-32 (1984) 19. Hadjidemetriou, J.D.: A hyperbolic twist mapping model for the study of asteroid orbits near the 3/1 resonance. J. Appl. Math. Phys. 37, 776-796 (1986) 20. Hadjidemetriou, J.D.: Algebric mappings near the resonance with an application to as- teroid motions. In: A.E. Roy (ed) Long Term Dynamical Behavior of Natural Artificial and N-body Systems. Kluver Academic Publishers, 257-276 (1988) 21. Hadjidemetriou, J.D.: Mapping models for Hamiltonian system with application to res- onant asteroidal motion. In: A.E. Roy (ed) Predictability, Stability and Chaos in N-body Dynamical Systems, Plenum Press, 157-175 (1991) 22. Hadjidemetriou, J.D.: Asteroid motion near the 3/1 resonance. Cel. Mech. Dyn. Astron. 56, 563-599 (1993) 23. Knezevi´c, Z.: Veritas family age revisited. In: IAU Coloquium 173: Evolution and sources regions of asteroids and comets, pp 153-158 (1999) 24. Knezevi´c, Z., Tsiganis, K. and Varvoglis, H.: The dynamical portrait of the Veritas family region. In: Proceedings of Asteroids, Comets, Meteors International Conference, Noordwijk, Netherlands: ESA Publications Division, pp. 335-338 (2002) 25. Knezevi´c, Z.: Chaotic diffusion in the Veritas family region. In: Proceedings of the XIII National Conference of Yugoslav Astronomers, Publications of the Astronomical Observatory of Belgrade 75, pp. 251-254 (2003) 26. Knezevi´c, Z.: New Frontiers in Main Belt Asteroid Dynamics. In: Bulletin of the American Astronomical Society, 36, p. 856 (2004) 27. Knezevi´c, Z., Tsiganis, K. and Varvoglis, H.: Age of the Veritas asteroid family from two independent estimates. Astronomical Observatory of Belgrade, 80, p. 161-166 (2004) 28. Knezevi´c, Z.: Dynamical Methods to Estimate the Age of Asteroid Families. (2007) 29. Lega, E. Guzzo, M. and Froeschl, C.: Measure of the exponential splitting of the homoclinic tangle in four-dimensional symplectic mappings. Celest. Mech. Dyn. Astron. 104, 191-204 (2009) 30. Lochak, P.: Arnold diffusion: A compendium of remarks and question. In: C. Sim´o (ed) NATO ASI: Hamiltonian system with Three or More Degrees of Freedom, pp-168, Kluver, Dordrecht (1999) 31. Lhotka, C.: Dynamic expansion points: an extension to Hadjidemetriou's mapping meth- ods. Celest. Mech. Dyn. Astron. 104, 175-189 (2009) 32. Lichtenberg, A.J. and Lieberman, M.A.: Regular and Stochastic Motion. Springer-Verlag, New York, vol. 38 (1983) 33. Milani, A. and Nobili, A. M.: An example of stable chaos in the solar system. Nature 357, 569-571 (1992) 34. Milani, A.: 1993, The trojan asteroid belt: Proper elements, stability, chaos and families. Celest. Mech. Dyn. Astron. 57, 59-94 (1993) Chirikov-Arnold in the (5, −2, −2) resonance 27 35. Milani, A.; and Farinella, P.: The age of the Veritas asteroid family deduced by chaotic chronology. Nature 370, 40-42 (1994) 36. Milani, A.; Nobili, A. M.; Knezevi´c, Z.: Stable chaos in asteroid belt. Icarus 125, 13-31 (1997) 37. Morbidelli, A. and Froeschl´e, C.: On the relationship between Lyapunov times and macro- scopic instability times. Celest. Mech. Dyn. Astron. 63, 227- 239 (1996) 38. Nesvorn´y, D. and Morbidelli, A.: Three-body mean-motion resonances and the chaotic structure of the asteroid belt. Astron. J. 116, 3029-3037 (1998) 39. Nesvorn´y, D. and Morbidelli, A.: An analytic model of three-body mean-motion reso- nances. Celest. Mech. Dyn. Astron. 71, 243-271 (1999) 40. Novakovi´c, B.; Tsiganis, K.; Knezevi´c, Z.: Chaotic transport and chronology of complex asteroid families. (2009) 41. Roig, F. and Ferraz-Mello.: A sympletic mapping approach of the dynamics of the Hecuba gap. Planetary and Space Science 47, 653-664 (1999) 42. Sun, Y and Zhou, L.: Stickiness in three-dimensional volume preserving mappings. Celest. Mech. Dyn. Astron.103, 119-131 (2009) 43. Tsiganis, K., Varvoglis, H. and Hadjidemetriou, J. D. Stable Chaos in High-Order Jovian Resonances. Icarus 155, 454-474 (2002a) 44. Tsiganis, K., Varvoglis, H. and Hadjidemetriou, J. D. Stable Chaos versus Kirkwood Gaps in the Asteroid Belt: A Comparative Study of Mean Motion Resonances. Icarus 159, 284-299 (2002b) 45. Varvoglis, H: Diffusion in the asteroid belt. In: IAU Coloquium 197: Dynamics of Popula- tions of Planetary Systems, pp 157-170 (2004) 46. Tsiganis, K., Knezevi´c, Z. and Varvoglis, H.: Reconstructing the orbital history of the Veritas family. Icarus 186, 484-497 (2007). 47. Wisdom, J.:The origin of the Kirkwood gaps - A mapping for asteroidal motion near the 3/1 commensurability. A.J. 87, 577-593 (1982)
1106.0014	2	1106	2012-09-07T16:30:48	Angular momentum exchange during secular migration of two-planet systems	[ "astro-ph.EP" ]	We investigate the secular dynamics of two-planet coplanar systems evolving under mutual gravitational interactions and dissipative forces. We consider two mechanisms responsible for the planetary migration: star-planet (or planet-satellite) tidal interactions and interactions of a planet with a gaseous disc. We show that each migration mechanism is characterized by a specific law of orbital angular momentum exchange. Calculating stationary solutions of the conservative secular problem and taking into account the orbital angular momentum leakage, we trace the evolutionary routes followed by the planet pairs during the migration process. This procedure allows us to recover the dynamical history of two-planet systems and constrain parameters of the involved physical processes.	astro-ph.EP	astro-ph	Noname manuscript No. (will be inserted by the editor) Angular momentum exchange during secular migration of two - planet systems Adri´an Rodr´ıguez · Tatiana A. Michtchenko · Octavio Miloni 2 1 0 2 p e S 7 . ] P E h p - o r t s a [ 2 v 4 1 0 0 . 6 0 1 1 : v i X r a the date of receipt and acceptance should be inserted later Abstract We investigate the secular dynamics of two-planet coplanar systems evolving under mutual gravitational interactions and dissipative forces. We con- sider two mechanisms responsible for the planetary migration: star-planet (or planet-satellite) tidal interactions and interactions of a planet with a gaseous disc. We show that each migration mechanism is characterized by a specific law of orbital angular momentum exchange. Calculating stationary solutions of the con- servative secular problem and taking into account the orbital angular momentum leakage, we trace the evolutionary routes followed by the planet pairs during the migration process. This procedure allows us to recover the dynamical history of two-planet systems and constrain parameters of the involved physical processes. Keywords Secular evolution · Migration · Dissipation laws · Tidal force · Drag force 1 Introduction Recently, Michtchenko and Rodr´ıguez (2011) have purposed a new method for a qualitative study of the planet migration originated by a generic dissipative mech- anism. It has been shown that, under assumption that the dissipation processes are sufficiently slow, the evolutionary routes of migrating planets in the phase space are traced by stationary solutions of the conservative secular problem. Therefore, the modeling of planet migration consists in the calculation of families composed by Mode I and Mode II stationary solutions, parameterized by the mass ratio, for all possible values of planetary semi-major axes and orbital angular momentum of the system. A. Rodr´ıguez · T. A. Michtchenko Instituto de Astronomia, Geof´ısica e Ciencias Atmosf´ericas, IAG - USP, Universidade de Sao Paulo, Sao Paulo, SP, Brazil E-mail: [email protected] O. Miloni Facultad de Ciencias Astron´omicas y Geof´ısicas, Universidad Nacional de La Plata, La Plata, Argentina 2 A. Rodr´ıguez et al. It is presently accepted that the main mechanisms responsible for planet migra- tion are the star-planet (and planet-satellite) tidal interactions and gravitational interactions of a planet with a gaseous disc. Dissipative interactions remove (or add) orbital energy from the planetary system, producing changes of semi-major axes. In the case of tidal interactions, orbital energy is transformed into thermal energy, which is dissipated in the interior of the tidally deformed body. Migration of the system can be either toward or outward the central body, depending on the properties of the tidal interactions and some parameters such as Love numbers and quality factors (Darwin 1880; Jeffreys 1961; Goldreich and Soter 1966; Hut 1981; Dobbs-Dixon et al. 2004; Ferraz-Mello et al. 2008). During disc-planet interactions, orbital energy is exchanged with a surrounding gaseous protoplanetary disc and, generally, the planetary orbital decay occurs (for a review, see Armitage 2010 and references therein). In addition, migration can also occur through gravitational scattering between planets and a remnant planetesimal debris (e.g. Fern´andez and Ip 1984; Kirsh 2009). Dissipative forces also alter the orbital angular momentum of the system. In- deed, tidal torques modify the rotational angular momenta of the interacting bod- ies, which are transferred to the orbits of planets or satellites (e.g. Mignard 1979; Correia et al. 2008). During disc-planet interactions, the angular momentum is exchanged through gravitational torques between the planets and the disc (e.g. Lin et al. 1996; Goldreich and Sari 2003; Masset et al. 2006). In Michtchenko and Rodr´ıguez (2011), we have purposed a generic approach to describe the exchange of the orbital angular momentum of the system during migration. We have intro- duced the orbital angular momentum leakage, defined as a portion α of the orbital angular momentum variation produced by migration (i.e. expansion or shrinkage of the planetary orbits), which is extracted (or added to) from the planet system. In this paper, we show that each physical process is characterized by a specific law of orbital angular momentum leakage. We show that α is a function of the eccentricity of the planet affected by the dissipative force. Moreover, we show that α also depends on a set of physical parameters related to planets, star and disc. Finally, owning the characteristic α-function, we construct the evolutionary routes of the system for each specific mechanism driving the planet migration. This approach provides us with a general idea of how the system could evolve under a variety of migration conditions and physical models. In Section 2 we briefly introduce some basic aspects of the secular dynamics of two-planet systems evolving under dissipative forces. Section 3 addresses tidal interactions, where we include the cases of a close-in planet and a satellite interact- ing with their parent star and planet, respectively. Migration driven by disc-planet interaction is discussed in Section 4. The dissipation is modeled through a drag Stokes-like force. Finally, Section 5 is devoted to conclusions. 2 Some aspects of the secular dynamics under dissipation 2.1 Orbital angular momentum exchange We consider a three-body system consisting of a central star with mass m0 and two coplanar planets with masses m1 and m2. We assume that the system is far Secular migration of two - planet systems 3 away from any mean-motion resonance. Hereafter, the indices i = 1, 2 stand for the inner and outer planets, respectively. In the astrocentric reference frame, the orbital angular momentum of the sys- tem is given, up to second order in masses, as Lorb = m′ 1) + m′ 1pa1(1 − e2 2pa2(1 − e2 2), where m′ tricities, and G is the gravitational constant. i ≡ mipG(m0 + mi), ai and ei are planetary semi-major axes and eccen- The migration of planets is originated by dissipative forces which affect the energy and the orbital angular momentum of the planets and results in variations of their semi-major axes and eccentricities. If dissipation is sufficiently slow (its rate is smaller than the proper frequency of the secular motion), the long-term variations of the orbital elements can be separated into two components: one is produced by secular interactions with the companion and the other is due to external interactions. Hence, we can write (1) ai = ai ei = ei sec + ai sec + ei ext, ext, (2) (3) where the indices "sec" and "ext" stand for secular and external contributions of the total variation of each element. The secular theory provides that ai sec = 0 (to first order in masses) and, as a ext. We stress that, since we are studying the secular evolution consequence, ai = ai of the system, short-period variations of the orbital elements (of the order of orbital periods) are omitted in the analysis. Thus, Equations (2)-(3) should be considered as the averaged equations describing the long-term variations of semi-major axes and eccentricities. In addition, the contribution of resonant terms is also neglected. Assuming that the planetary masses are unaltered during migration, the time variations of Lorb, in terms of the time variation of the planetary semi-major axes, ai, and eccentricities, ei, is written as Lorb = m′ 1p1 − e2 2√a1 1 a1 − e1 + m′ 2p1 − e2 2√a2 2 a2 − m′ 1√a1e1 p1 − e2 1 e2. (4) m′ 2√a2e2 p1 − e2 2 The total angular momentum of the whole system (including the three-body system under study and the external component) is conserved during migration. Thus, we can write that Lorb + Lext = const, where the external component of the total angular momentum is denoted as Lext. It should be emphasized that the generic definition "ext" does not necessarily mean that the exchange of the orbital angular momentum occurs with an exterior medium. For instance, in the case of tidal interactions of planets with a central star, the 'external' contribution to the angular momentum comes from the rotation of the star, as described in the next section. On the other hand, in the case of disc-planet interactions, there is a flux of angular momentum between the planet system and an external protoplanetary disc, as described in Section 4. For sake of simplicity, we restrict our investigation to the case in which only one planet is directly affected by dissipative forces and, consequently, its semi-major axis is changed. This implies that the semi-major axis of the other planet remains unchanged during secular evolution of the system. Assuming that the i th- body 4 A. Rodr´ıguez et al. is affected by dissipative forces and combining Equations (2) -- (4), we use the conservation of the total angular momentum to find that ai m′ j m′ ir aj (1 − e2 i ) 2aiei j ej eis 1 − e2 1 − e2 ai + p1 − e2 i√aiei m′ i i sec ej Lext, (5) (6) ei sec = − and ei ext = where i 6= j and ei 6= 0. Equation (5) shows that, under mutual secular perturba- tions, the eccentricities of the planetary orbits oscillate in anti-phase. Equation (5) also shows that the evolution of ej depends implicitly on the migrating evolution of the companion, through the external variations of ai and ei. Equation (6) can be re-written as where ei ext = (1 − α) (1 − e2 i ) 2aiei ai, ei 6= 0, α = − 2ai ai Lext Li , ai 6= 0, (7) (8) with Li being the partial angular momentum of the body affected by dissipation. The function α can thus be used as a convenient measure of the non-conservation of Lorb, since it quantifies the variation of the external component of the orbital angular momentum of the system. Indeed, if α = 0, Lext remains constant and Lorb is conserved during migration. According to Equation (7), the limit case ei = 0 implies that ai = 0 for α = 0, that is, the migration must be stopped when the planet orbit is circularized in the system with no exchange of angular momentum. For α 6= 0, the portion 1 − α of the Lorb- change produced by ai is absorbed by the system through the variation of ei, while the rest (α) is transferred to Lext. We say that there is a leakage (or gain) of the orbital angular momentum in the system. According to Equation (7), for α = 1, the dissipative force produces no change of the eccentricity and the change in Lorb due to ai is totally transferred to Lext. This particular case of the angular momentum exchange was used to modeling of planet-planetesimal interactions (Malhotra 1995). The range of theoretical values of α is large; according to Equation (8), α can be positive or negative, depending on the signs of ai and Lext. Moreover, α is a function of the orbital elements of the migrating planet and, consequently, varies during migration. Knowing that each physical process is characterized by a specific law of angular momentum exchange, in the next sections, we develop the α-function for several kinds of dissipative interactions. Using Equation (7), we write α in general form as α = 1 − F(ei, par) 2e2 i 1 − e2 i , (9) where F(ei, par) is a characteristic function of the migration process, defined through the condition eext i ei = F(ei, par) ai ai , (10) where the vector par is composed of physical parameters of the process under study. Secular migration of two - planet systems 5 2.2 Evolutionary routes in the phase space As shown in Michtchenko and Rodr´ıguez (2011), under assumption that dissipation is weak and slow, the evolutionary routes of the migrating non resonant planets are traced by the Mode I and Mode II stationary solutions of the conservative secular problem (see also Hadjidemetriou and Voyatzis 2010). The ultimate convergence and the evolution of the system along one of these secular modes of motion are determined by the condition that the dissipation rate is smaller than the proper secular frequency of the system. We have shown that, for values of α less than 1, the Mode I secular solution (characterized by aligned orbits) plays a role of an attractive center when planetary orbits diverge and the condition pa1/a2 < m2/m1 is satisfied, or, when planetary orbits converge and pa1/a2 > m2/m1. In opposite cases, the Mode II (characterized by anti-aligned orbits) is the attractive center of the migrating two-planet system. In practice, the calculation of evolutionary tracks is simple. The current loca- tion of the system in the phase space provides the starting values of the orbital angular momentum, Lorb, and the semi-major axis ai of the planet directly affected by dissipation. The set, composed by Lorb, masses and semi-major axes, uniquely defines two stationary solutions (e∗ 2) of the conservative secular problem at the current configuration. In order to model migration process, the semi-major axes of the affected planet is incremented by ∆ai (which can be either positive or nega- tive), while the semi-major axis of the other planet is kept unaltered. The orbital angular momentum of the system is then corrected by the amount 1, e∗ ∆Lorb = α Li 2ai ∆ai, where the function α is defined by Equation (9). For a new set of ai and Lorb values (with fixed masses), we calculate new values of stationary solutions, (e∗ 2). The procedure is repeated until the system reaches domains with no possible stationary solutions. The obtained values of (e∗ 2) are then plotted on the representative planes (n1/n2, ei), where ni (i = 1, 2) are mean motions of the planets. The family of stationary solutions shows two evolutionary routes, one of which the system will follow during migration. 1, e∗ 1, e∗ In the described above process, stationary solutions can be calculated using the precise semi-analytical approach developed in Michtchenko and Malhotra (2004), with no restrictions on the values of planet eccentricities. An alternative is the use of the expansion of the disturbing function in series of a1/a2, e1 and e2, given by R = N l Xl=0 Xm=0 N ′ Xl′ 1,l′ =0 2 Rl,m,l′ 1,l′ a2!l 2 a1 1 l′ 1 e 2 l′ 2 cos(m∆), e 2 1,l′ where Rl,m,l′ are numerical coefficients which do not depend on the physical and orbital parameters of the system and ∆ ≡ 1 − 2 is the difference of planetary longitudes of pericenter. The main difficulty in the calculation of migration routes is that the charac- teristic function F(ei, par), which is needed to obtain α through Equation (9), is unknown a priory. Thus, the next sections of this paper will be devoted to calculate the function F for some specific dissipative processes. 6 A. Rodr´ıguez et al. 3 Tidal interactions We first consider the case of tidal interactions, which affect orbital elements and rotations of the deformed bodies, while energy is dissipated due to internal friction. The tidal force is inversely proportional to the 7 th power of the distance between interacting bodies and is effective in both planetary and satellite systems (e.g. Mignard 1979). In this paper, we assume that only the central and close-in bodies are mutually affected by the raised tides, whereas the outer companion is not directly affected by the tidal force. The orbital planes of planets are assumed to be coincident. 3.1 A system with a close-in planet We consider a short-period planet orbiting a slow-rotating central star (Ω0 ≪ n1), where Ω0 and n1 are the angular velocity of rotation of the star and the mean orbital motion of the planet, respectively. The long-term variations (averaged over the inner planet orbital period) of the planetary elements, due to the combined effects of tides raised on the star and on the inner planet, are written, up to O(e3 n1a−4 1 s[(1 + 23e2 1), as: 1) + 7e2 1D] < a1 > = − 4 3 (11) and where with < e1 ext > = − 2 3 n1e1a−5 1 s[9 + 7D], D ≡ p/2s, p ≡ 9 2 k1 Q1 m0 m1 R5 1 and s ≡ 9 4 k0 Q0 m1 m0 R5 0 (12) (13) (14) being the strengths of stellar and planetary tides, respectively (Dobbs-Dixon et al. 2004; Ferraz-Mello et al. 2008, Rodr´ıguez and Ferraz-Mello 2010). ki and Qi are the Love number and the dissipation function of the deformed body, where the indices i = 0 and 1 stand for the central and inner bodies, respectively. In above equations, it is assumed a quasi-synchronous rotation state of the close-in planet (Ω1 ≃ n1). In addition, is is also assumed a linear dependence between phase lags and the corresponding frequencies appearing in the decomposition of the tidal potential of each body (Mignard 1979; Ferraz-Mello et al. 2008). For sake of simplicity, we consider that Qi are constant; however, it should be kept in mind that their dependence on frequencies can be important in the long-term evolution. The reader is referred to Efroimsky and Williams (2009) for more discussions on the frequency dependence of Qi. Combining Equations (10) and (11) -- (12), we obtain the characteristic function F(e1, D) = 9 + 7D 2 [1 + (23 + 7D)e2 1] , (15) Secular migration of two - planet systems 7 α 1 0.8 0.6 0.4 0.2 0 D=0 D=100 D=1000 0 0.05 0.1 0.15 0.2 e1 Fig. 1 The Lorb -- leakage α as a function of e1, in the case of a close-in planet tidal evolution. The curves are parameterized by several constant values of D. where par is defined by the parameter D. The Lorb -- leakage or the function α is then calculated through Equation (9) to give α = 1 − (9 + 7D)e2 1 (1 − e2 1) [1 + (23 + 7D)e2 1] . (16) Figure 1 shows the variation of α with e1. We observe that, for all values of D, α varies in the range between 0 and 1. The immediate consequence is that the tidal decay of a close-in planet is generally accompanied by the loss of orbital angular momentum of the system. The portion 1 − α is absorbed by the system, damping the eccentricity of the inner planet. The amount α is transferred to the central star and accelerates the stellar rotation (see discussion of the next paragraph). The stellar rotation changes due to tides on the star and, from the above discussion, it is clear that its variation plays a role of external source of the angular momentum. The rate of variation of the rotational angular momentum is given by Lext = Lrot = C0 Ω0, where C0 is the stellar moment of inertia (assuming that the inner planet has reached stationary rotation, we can neglect the spin variation of the planet; e.g. Rodr´ıguez et al. 2011 for more details). Hence, using the definition of α given in Equation (8), we obtain, for a1 6= 0, Ω0. (17) α = − 2C0a1 L1 a1 Since a1 < 0 and α > 0 during orbital decay of the inner planet, above equation results in Ω0 > 0, as previously discussed. On one hand, Equation (17) shows that, for α = 0, Ω0 is a constant and there is no tides on the star (i.e. s = 0). In this case, the orbital angular momentum of the system is conserved. Inversely, knowing that D ≡ p/2s, s = 0 means that D → ∞ and, according to Equation (15), F(e1, D) = 0.5e−2 1 . Hence, to second order in e1 and according to Equation (9), we have α = 0. Note that, for α ≃ 0 (very large D), 8 A. Rodr´ıguez et al. Fig. 2 Evolutionary routes of CoRoT-7 system parameterized by constant values of D. The angular momentum exchange is determined by the α -- function (9) defined through (16). The current position of the system is shown by a star symbol, whereas the locations of some mean- motion resonances are indicated by dashed lines. A schematic view of the migration scenario is shown in the box illustration at the top panel. the migration is halted when the orbit circularizes, that is, a1 becomes constant because there is no orbital angular momentum transfer and e1 = 0. On the other hand, when planetary tides are neglected, we have p = D = 0 and, up to second order in e1, (1 − α) = 9e2 1). This last expression shows that, after circularization of the inner planet orbit (e1 = 0), the orbital angular momentum change due to the orbital decay is totally transferred to the stellar spin (α = 1). Evidence for excess of rotational angular momentum of the parent stars due to the tidal interaction with close-in giant planets have been already addressed in previous studies (e.g., Pont 2009). 1/(1 + 23e2 It is important to note that, during tidal interactions, α is not constant but varies as a function of e1, according to Equation (16). The sign of α is always positive, meaning that the decreasing of the orbital angular momentum due to the orbital decay of the inner planet is compensated by both the damping of e1 and the increase of Ω0. Figure 2 shows the evolutionary tracks of a migrating pair of planets. They were obtained as discussed in Section 2.2, where the orbital angular momentum varia- tion for each curve was defined by the function α, according to Equation (16). An application to the CoRoT-7 planetary system, characterized by very small current eccentricities (Ferraz-Mello et al. 2011) was considered, whose results are plotted in the planes (n1/n2, ei). The adopted migration scenario is schematically shown on the top of the figure. The divergent migration, together with the condition Secular migration of two - planet systems 9 m2/m1 >pa1/a2, results in the Mode I (∆ = 0) of secular motion as an attrac- tive center (Michtchenko and Rodr´ıguez 2011). The current position of the system is marked with a star symbol and the arrows indicate the direction of the evolution (orbital decay of the inner planet). We can note that migration tracks are sensible to the values of the parameter D, and consequently, to the orbital angular momentum exchange. For small D (α ≃ 1), corresponding to the nearly-total loss of the angular momentum variation produced by the orbital decay of the inner planet, both eccentricities vary only slightly during evolution. In this case, when the strength of the stellar tide is dominating, the orbital configuration of the system in the past would be not different from the present one. The situation is very different in the case of large values of D, when the strength planetary tide is dominating. Large D-values imply that the orbital angular mo- mentum of the system is almost conserved (α ≃ 0). In this case, the orbital config- uration of the system in the past should be characterized by high eccentricities, as higher as larger are D-values. Therefore, in the cases with dominating planetary tides, the origin of such high eccentric planetary orbits must be investigated. Note that, during migration, the two-planet system could cross several mean- motion resonances, resulting in the temporary excitation of eccentricities but with- out trapping, since the orbits are divergent. Several numerical simulations per- formed in Rodr´ıguez et al. (2011) and Michtchenko and Rodr´ıguez (2011), show that the system ultimately returns to the stationary secular solution after leaving the mean-motion resonance. 3.2 A system of satellites We consider now the case of a pair of satellites orbiting a rapidly rotating parent planet (Ω0 ≫ n1). We assume that only the inner satellite is affected by tidal interactions with the planet. According to the linear tidal model, the averaged variations of the orbital elements are given by and < a1 > = 4 3 n1a−4 1 s[(1 + (27/2 − 7D)e2 1] < eext 1 > = 2 3 n1e1a−5 1 s(11/2 − 7D) (18) (19) (Goldreich and Soter 1966; Ferraz-Mello et al. 2008). Note that the only difference with respect to the previous case is in the tides raised on the planet (or central body). We still note that the mean variations can be positive or negative, depend- ing on the value of D. Hence, the combined effect of planet and satellite tides can induce either inward or outward migration, as well as either damping or excitation of eccentricity. The sign of eext depends on the value of D in such a way that eext 1 > 0 if D < 11/14. The sign of a1 depends on both D and e1-values. For small eccentricities (e1 ≤ 0.2) and moderate D (in the range from 1 to 5), we would ex- pect outward migration and damping of eccentricity, a typical behavior observed in the tidal evolution of Solar System satellites (e.g. Peale 1999). In this case (i.e., a1 > 0 and eext 1 < 0), the total angular momentum conservation implies that α > 1 Ω0 < 0. The Earth-Moon tidal (see Equation (7)) and, according to Equation (17), 1 10 A. Rodr´ıguez et al. 10 α 1 0.1 D=0 D=3 D=5 0 0.05 0.1 0.15 0.2 e1 Fig. 3 Dependence of α on e1, in logarithmic scale, for the tidal evolution of a close-in satellite. Illustrations are shown for different values of D. Here, α can adopt values larger than unity, but not in the close-in tidal evolution case (see Figure. 1 for comparison). evolution, with the satellite moving away from the Earth and the increasing length of day of the planet, corresponds to this scenario. Combining Equations (9)-(10) and (18)-(19), we have F(e1, D) = 11/2 − 7D 2[1 + (27/2 − 7D)e2 1] , and α = 1 − (11/2 − 7D)e2 1)[1 + (27/2 − 7D)e2 1] 1 (1 − e2 (20) (21) . Figure 3 shows the variation of α with e1, for different values of D. With the exception of the example with D = 0, α > 1 always, indicating that the external source of angular momentum (planet rotation) injects angular momentum in the satellite system, which is used to expand the inner planet orbit and circularize both inner and outer orbits. Figure. 4 shows the migration routes of a hypothetical system in which the central body is an Earth-like planet, the inner satellite has the Moon mass and the outer companion is three times smaller than the Moon. The curves are param- eterized by several values of D. For D in the range from 0 to 5, the inner satellite migrates outward when the eccentricity of its orbit is not too large (e1 < 0.2, according to Equation (18)). Since the semi-major axis of the outer satellite is unaltered during the secular interaction with its companion, we have a convergent migration and, together with the condition m2/m1 <pa1/a2, results in the Mode I (∆ = 0) as the attractive center. The direction of migration along evolutionary routes is shown by arrows in Figure. 4. Secular migration of two - planet systems 11 Fig. 4 The same as in Figure 2, except the angular momentum exchange is determined by the α -- function defined through (21). For D ≥ 1, the satellite eccentricities decrease during migration: larger is the value of D, more rapid is the damping of eccentricities. In this way, the satellite system approaches to main mean-motion resonances with low-eccentricity orbits which allow a smooth capture into one of these resonances (e.g. Tittemore and Wisdom 1988). The behavior of the system described by D = 0 is different, be- cause in this case the eccentricities of the satellites increase during migration (see Equation (19)). As a consequence, the capture inside a low-order mean-motion resonance would be strongly improbable. 4 Disk-planet interactions In this section we study the migration of a two-planet coplanar system assum- ing that dissipative forces affect only the outer planet. We also assume that the two planets are interacting secularly, that is, the planets are far enough from any mean-motion resonance. Disc-planet interactions are the natural mechanism which results in the orbital decay of the outer planet due to energy and angular momentum exchange with an outer gaseous disc (e.g. Kley 2000, 2003). 4.1 Evolutionary tracks under Stokes interactions For sake of simplicity, we suppose that the outer planet is forced to migrate under the action of a dissipative drag force (Stokes-like force) of the type 12 A. Rodr´ıguez et al. f = −10−ν(v2 − γ v2c), (22) where ν > 0 and v2c is the Keplerian circular velocity at the astrocentric distance r2. Due to the negative radial pressure gradient, the gas velocity is a bit less than the circular velocity at the same point. Consequently, the value of the factor γ should be slightly smaller than unity (see Adachi et al. 1976; Patterson 1987). γ is also frequently used as a free parameter of the problem, in order to model different conditions of disc-planet interactions (e.g. Beaug´e et al. 2006). Several previous works have investigated the evolution of a two-planet system evolving under a Stokes drag force (e.g., Beaug´e at al. 2006; Hadjidemetriou and Voyatzis 2010), with the majority devoted to the study of the evolution inside mean-motion resonances. In order to obtain the variations of the semi-major axis and the eccentricity of the outer planet produced by the force f , we use the Euler-Gauss's equations (Brouwer and Clemence 1961): [fr e2 sin u2 + ft(1 + e2 cos u2)] (23) da2 dt de2 dt d2 dt d l0 2 dt 2 = 2 2 ft n2 a2 hfr sin u2 + n2p1 − e2 = p1 − e2 = p1 − e2 n2 a2 e2 (cid:20)−fr cos u2 + ft(cid:18) = p1 − e2 (cid:2)(a2(1 − e2 e2 2 2 r2 e2 (cid:16)1 + e2 cos u2 − a2(1 − e2 2) a2(cid:17)i + 1(cid:19) sin u2(cid:21) r2 2) cos u2 − 2e2r2)fr − (r2 + a2(1 − e2 2)) sin u2 ft(cid:3) , 2 +R n2dt. fr and ft are 2 is related to the mean anomaly, l2, through l2 = l0 where l0 the radial and transverse components of f , whereas u2 is the true anomaly of the outer planet. The first-order averaging over the outer planet orbital period gives, to fourth order in e2 and assuming m2 ≪ m0: < a2 > = −10−νa2[2(1 − γ) + γ(5e2 < e2 > = −10−νγ e2(1 − 13e2 2/32) < 2 > = 0 2/8 + 119e4 2/512)] (24) (25) (26) (for comparison, see Beaug´e and Ferraz-Mello 1993). It follows from above equa- tions that the disc-planet interactions produce orbital decay and circularization. If we set γ = 1 (see Hadjidemetriou and Voyatzis 2010) and e2 = 0 into Equation (24), then we have a2 = 0, indicating that there is no orbital decay of the outer planet in the case of circular orbit. Using Equations (9)-(10) and (24)-(25), we obtain F(e2, γ) = γ (1 − 13e2 2/32) 2(1 − γ) + γ(5e2 2/8 + 119e4 2/512) and α = 1 − 2γ (1 − 13e2 2) [2(1 − γ) + γ(5e2 (1 − e2 2/32)e2 2 2/8 + 119e4 2/512)] where par is given by the parameter γ. (27) , (28) Secular migration of two - planet systems 13 α 1 0.5 0 -0.5 -1 -1.5 -2 -2.5 γ=0.995 γ=0.990 0 0.1 0.2 0.3 e2 Fig. 5 Variation of α with e2, for two values of γ, corresponding to the case of disc-planet interaction simulated with a drag force of the type (22). Note that, depending on e2, α can adopt negative values (see text for details). Figure 5 shows the function α(e2, γ) parameterized by two values of γ. We note that, as in the case of star-planet tidal interactions (see Sect. 3.1), α is always smaller than 1, which means that the orbital decay of the outer planet is always accompanied by damping of the eccentricity of its orbit (see Equation (7)). We also see that, for e2 = 0, α = 1 for any value of γ (we have discussed that γ < 1), indicating that, after circularization of the planet orbit, the change of Lorb due to the orbital decay is totally transferred to Lext, in this case, to the gaseous disc. In contrast with the case of star-planet tidal interactions, the function α is not saturated at 0 (see Figure 1), but decreases monotonously with eccentricities towards negative values. According to Equation (8), negative values of α imply in the loss of angular momentum by the disc during the orbital decay of the outer planet ( Lext < 0). Thus, during migration, the pair of planets can either gain or lose orbital angular momentum, depending on the eccentricity of the migrating planet. 2 ≃ 0.085. For e2 = ecr A transition between leakage or gain of angular momentum occurs at α = 0, when e2 reaches a critical value ecr 2 which is determined from Equation (28), for a given γ. For instance, for γ = 0.995, ecr 2 , the orbital angular momentum of the planet system is conserved (since α = 0), however, this condition is only transitory. For e2 < ecr 2 , we have 0 < α < 1, indicating that a portion α of the orbital angular momentum change due to migration is removed from the system and transferred to the disc ( Lext > 0, see Equation (8)), while the portion 1 − α is absorbed by the system altering the value of e2. When e2 > ecr 2 , we have α < 0 and the system gains angular momentum from the disc. Note that, in the latter case, the planet system gains angular momentum, despite the semi-major axis of the outer planet is decreasing. The evolutionary routes of the planet pair, obtained for several values of the parameter γ and for the angular momentum exchange defined by (28), are shown in Figure 6. The hypothetical system is composed of a Sun-like central body and 14 A. Rodr´ıguez et al. Fig. 6 Evolutionary routes for a disc-planet interaction simulated through the force given in Equation (22), parameterized by constant values of γ. In this case, the angular momentum exchange is determined by the α -- function (28). two planets of equal mass. The migration scenario shown on the top of the figure corresponds to the outer planet moving towards the inner planet. In this case, migration is convergent and the Mode II plays a role of an attractive center, since m2/m1 > pa1/a2 (see Michtchenko and Rodr´ıguez 2011). Planet eccentricities are damped during migration (see Equation (25)), while the system approaches to one of the main mean-motion resonances, whose locations are indicated in Figure 6. Because of small eccentricities, the probability of capture in a mean-motion resonance would be high, mainly for larger γ-values. It is worth noting that, to achieve the current position of the system (marked by a star symbol), a past orbital configuration with higher eccentricities is needed for all considered values of γ. It is important to stress that, during the migration path, the system crosses several mean-motion resonances, as shown in Figure 6. One should keep in mind that, when the rate of dissipation is small, the system may be trapped one of such resonances. Since we restrict our investigations to the secular evolution of the sys- tems, we fix the initial configurations of planet pairs far away from mean-motion resonances and, according to our model, neglect the possibility of a resonant cap- ture. Secular migration of two - planet systems 15 4.2 The factor K Several works have simulated the disc-planet interaction adopting a simplified model of planetary migration (e.g. Lee and Peale 2002; Beaug´e et al. 2006, among others). The method avoids the use of hydrodynamic numerical simulations in- volving an accurate description of the interaction process between the gas and the planet body. The model introduces the relationship between the migration rates of semi-major axis and eccentricity as follows e2 e2 = −K(cid:12)(cid:12)(cid:12) a2 a2(cid:12)(cid:12)(cid:12) , (29) where K is a constant (positive) parameter. Given a value of K, the migrating systems are modeled through numerical integrations of the equations of planetary motion, with constant perturbations in the orbital elements according to Equation (29). As shown in Beaug´e and Ferraz-Mello (1993) and Gomes (1995), the time evolution of orbital elements due to the action of the drag force given by (22) is, to first order in eccentricity: a2(t) = a20 exp(−At) and e2(t) = e20 exp(−Et), where A and E are the inverse of the e-folding times of a2 and e2, respectively. In- troducing above equations into Equation (29) we obtain that K = E/A. Moreover, using Equations (24)-(25) we also obtain that, to first order in e2, K = γ/2(1− γ). In this way, we have shown that the condition (29), with a constant value of K, can be directly obtained from the averaged variations of the orbital elements produced by the Stokes force. It is worth noting that, for γ = 0.995, which is the usually adopted value (Beaug´e and Ferraz-Mello 1993), we obtain K = 100, which matches the value determined empirically by Lee and Peale (2002) for the GJ 876 resonant system (the planets b and c evolve inside the 2/1 mean-motion resonance). The relationship (29) is frequently applied in moderate and even high eccentric- ity domains, specially when the evolution of resonant pairs of planets is modeled. However, comparing Equations (10) and (29) we note that K = F(e2, γ), that is, the assumption of constant K is not appropriate in the case of eccentric orbits and the function F(e2, γ), given by Equation (27), should be used instead K = const (see also Kley et al. 2004). The variation of K as a function of e2 is shown in Figure 7. 4.3 A general drag force The Stokes force (22) belongs to the class of a more general dissipative force given by f = −10−νρ(r2)(v2 − γ v2c), (30) where ρ(r2) is the density profile of the gas in the disc (see Smart 1960). In this work, we assume that ρ(r2) = r−β 2 , with β ≥ 0 (see Kominami and Ida 2002, for an example with β = 2). Some works also adopt a more general dependence on 16 K 100 80 60 40 20 0 A. Rodr´ıguez et al. γ=0.995 γ=0.99 0 0.1 0.2 0.3 e2 Fig. 7 Variation of K with e2, for two values of γ. When e2 = 0 and γ = 0.995, K = 100 is obtained, reproducing the result of previous works (Lee and Peale 2002). Moreover, K strongly varies with e2 and for high eccentric orbits is almost independent on the value of γ. the relative velocity, introducing powers of (v2 − γ v2c) larger than 1. This case is related to high Reynolds numbers and correspond to regions of turbulence in the gas medium. Using the Euler-Gauss equations with the external force (30) and applying the averaged procedure (over orbital periods), we obtain the long-term variations of semi-major axis and eccentricity of the outer planet, up to O(e5 2), as follows < a2 > = −10−νa1−β < e2 > = −10−νa−β [2(1 − γ) + e2 e2 [γ(1 − β) + β + e2 2 G1(γ, β) + e4 2 G3(γ, β)], 2 2 2 G2(γ, β)], (31) (32) where 2G 1 = 3β + β2 + γ(cid:16) 5 4 − 2β − β2(cid:17) , 16G2 = 7β + 23 2 +γ(cid:16) 119 32 − 3β − 4G3 = −3β + β2 + 3 2 1 2 β2 β4, 27 4 β2 + 5β3 + 1 2 − 4β3 β3 + γ(β − 1)(cid:16) 13 1 2 − β4(cid:17) , 8 − β − 1 2 β2(cid:17) . It is clear that the case β = 0 corresponds to the previously studied Stokes force (see Equations (24)-(25)). Interesting features can be highlighted from the above equations. On one hand, we note that fixing a2 and e2, the rate of migration increases when β < 1 and decreases if β > 1. Moreover, for β = 1, < a2 > does not depends on a2. On the other hand, the rate of e2 - damping decreases for β > 0 and it is independent of a2 for the Stokes drag (β = 0). Secular migration of two - planet systems 17 α 1 0.5 0 -0.5 -1 -1.5 -2 γ=0.995 β=0 β=1 β=2 0 0.05 0.1 0.15 0.2 0.25 0.3 e2 Fig. 8 Variation of α with e2 for the case of a general dissipative force given by (30). Three values of the power β are illustrated for γ = 0.995. Through Equations (9)-(10) and (31)-(32), the characteristic function F(e2, γ, β) is given by F(e2, γ, β) = and the function α [γ(1 − β) + β + e2 2 G3(γ, β)] [2(1 − γ) + e2 2 G1(γ, β) + e4 2 G2(γ, β)] α = 1 − 2[γ(1 − β) + β + e2 2e2 2)[2(1 − γ) + e2 2 G1(γ, β) + e4 2 G3(γ, β)] 2 G2(γ, β)] (1 − e2 (33) , (34) where par is given by two parameters of the system, γ and β. Figure 8 shows the variation of α(e2, β, γ) as a function of e2, for γ = 0.995. We show curves parameterized by β = 0, 1, 2. Note that α has a strong dependence on e2 for all β. For high eccentricity (e2 ≥ 0.2), α < 0 for all values of β illustrated. When α < 0, the planetary system gains orbital angular momentum, because a2 < 0 and, looking at Equation (8), we have Lext < 0. Moreover, the gain is larger as smaller is the power β. In the limit of very small eccentricity (e2 ≤ 0.05), α is almost independent on β. The dependence of the planet evolution on the power β is illustrated in Figure 9, where we show the migration paths of the fictitious two-planet system, previously analyzed in Section. 4.1. Several values of β are used, with γ fixed at 0.995. We can note that, at least for the given γ, the different density profile distributions produce only quantitative differences in the damping of eccentricities, which are significant only at high eccentricity domains. The convergent migration of the planets can results in the crossing or capture in a mean-motion resonance, as in some previous discussed cases. 18 A. Rodr´ıguez et al. Fig. 9 Evolutionary routes for a disc-planet interaction simulated through the general force (30), parameterized by constant values of the power β and γ = 0.995. The angular momentum exchange is determined by the corresponding α -- function, given by Equation (34) 5 Conclusions In this work we model analytically the orbital angular momentum exchange of two- planet coplanar systems evolving under dissipative forces. Two migration mech- anisms were considered: tidal interactions in star-planet and planet-satellite sys- tems and gaseous disc-planet interactions. Our approach is based on the model developed in Michtchenko and Rodr´ıguez (2011), which shows that the angular momentum exchange between the orbital and the exterior components of the total angular momentum can be calculated through the α-function, referred to as the orbital angular momentum leakage. For each dissipative mechanism considered, α was calculated as a function of the planet eccentricity of the migrating planet and physical parameters involved in the process. Using the obtained α-function, stationary solutions of the secular conservative problem were calculated for each specific migration scenario and for several val- ues of physical parameters. Stationary solutions provide evolutionary routes that the system would follow in the process of planetary migration and allows us to understand the dynamical history of the system evolving under dissipative forces. The tidal interactions between a close-in planet with its host star results in the orbital decay of the inner planet. The angular momentum exchange between the orbital component and the stellar rotation imposes constraints on the α-function in such a way that 0 ≤ α ≤ 1. During migration, the planet system always loses some part of the orbital angular momentum, which is used to accelerate the rotation of Secular migration of two - planet systems 19 the star. Large (small) values of the parameter D (ratio between the strengths of planetary and stellar tides) are associated to weak (strong) stellar tides, enabling a substantial conservation (dissipation) of the orbital angular momentum of the system. In the planet-satellite tidal interaction, the situation is generally opposite. In this case, the migration of the inner satellite is predominantly outward. For typical values of the parameter D (in the range between 1 and 5) and moderate eccentric- ities (e1 < 0.2), the α-function is always larger than 1. This means that angular momentum is injected into the satellite system, with the planet spin as the supplier source. Disk-planet interactions modeled through a drag Stokes-like force, have shown that α ≤ 1 always. Angular momentum can be removed (0 < α ≤ 1) or injected (α < 0) into the planet system. We have shown that the factor K (the ratio between eccentricity damping and orbital decay), which is frequently adopted as a free parameter of the dissipative problem, is a function of the outer planet eccentricity. In addition, K also depends on the parameter γ, related to the disc properties. Moreover, the assumption of constant K is only valid in the domain of very small orbital eccentricities. Finally, the development of the α-function for each dissipative process, and the consequent calculation of evolutionary routes allow us to reassemble the start- ing configurations and migration history of the planet systems on the basis of their current orbital configurations. In addition, the analysis of the orbital angu- lar momentum evolution during migration of the system allows us to constrain parameters of the involved dissipative physical processes. Acknowledgements This work has been supported by CNPq, FAPESP (2009/16900-5) and INCT-A (Brazil). The authors gratefully acknowledge the support of the Computation Center of the University of Sao Paulo (LCCA-USP). We also thank the two anonymous referees for their stimulating revisions. References 1. Adachi, I., Hayashi, C., Nakazawa, K.: The gas drag effect on the elliptical motion of a solid body in the primordial solar nebula. Progress of Theoretical Physics 56, 1756-1771 (1976) 2. Armitage, P.J.: Astrophysics of Planet Formation. Cambridge (2010) 3. Beauge, C., Ferraz-Mello, S.: Resonance trapping in the primordial solar nebula - The case of a Stokes drag dissipation. Icarus 103, 301-318 (1993) 4. Beaug´e, C., Michtchenko, T.A., Ferraz-Mello, S.: Planetary migration and extrasolar planets in the 2/1 mean-motion resonance. Mon. Not. R. Astron. Soc. 365, 1160-1170 (2006) 5. Brouwer, D., Clemence, G.M.: Methods of celestial mechanics. New York (1961) 6. Correia, A.C.M., Levrard, B., Laskar, J.: On the equilibrium rotation of Earth-like extra- solar planets. Astron. Astrophys. 488, L63-L66 (2008) 7. Darwin, G.H.: On the secular change in the elements of the orbit of a satellite revolv- ing about a tidally distorted planet, Philos. Trans. 171, 713-891 (repr. Scientific Papers, Cambridge, Vol. II, 1908) (1880) 8. Dobbs-Dixon, I., Lin, D.N.C., Mardling, R.A.: Spin-orbit evolution of short-period Planets. Astrophys. J. 610, 464-476 (2004) 9. Efroimsky, M., Williams, J.G.: Tidal torques: a critical review of some techniques. Celest. Mech. Dyn. Astr. 104, 257-289 (2009) 10. Fernandez, J.A., Ip, W.-H.: Some dynamical aspects of the accretion of Uranus and Nep- tune - The exchange of orbital angular momentum with planetesimals. Icarus 58, 109-120 (1984) 20 A. Rodr´ıguez et al. 11. Ferraz-Mello, S., Rodr´ıguez, A., Hussmann, H.: Tidal friction in close-in satellites and exoplanets: The Darwin theory re-visited. Celest. Mech. Dyn. Astr. 101, 171-201 (2008). Erratum: 104, 319-320 (2009) 12. Ferraz-Mello, S., Tadeu dos Santos, M., Beaug´e, C., Michtchenko, T.A., Rodr´ıguez, A.: On planetary mass determination in the case of super-Earths orbiting active stars. The case of the CoRoT-7 system. (Astron. Astrophys., in press, preprint: arXiv:1011.2144) (2011) 13. Goldreich, P., Soter, S.: Q in the Solar System. Icarus 5 375-389 (1966) 14. Goldreich, P., Sari, R.: Eccentricity Evolution for Planets in Gaseous Disks. Astrophys. J. 585, 1024-1037 (2003) 15. Gomes, R.S.: The effect of nonconservative forces on resonance lock: Stability and insta- bility. Icarus 115, 47-59 (1995) 16. Hadjidemetriou, J.D., Voyatzis, G.: On the dynamics of extrasolar planetary systems under dissipation: Migration of planets. Celest. Mech. Dyn. Astr. 107, 3-19 (2010) 17. Hut, P.: Tidal evolution in close binary systems. Astron. Astrophys. 99, 126-140 (1981) 18. Jeffreys, H.: The effect of tidal friction on eccentricity and inclination. Mon. Not. R. Astron. Soc. 122, 339-343 (1961) 19. Kirsh, D.R., Duncan, M., Brasser, R., Levison, H.F.: Simulations of planet migration driven by planetesimal scattering. Icarus 199, 197-209 (2009) 20. Kley, W.: On the migration of a system of protoplanets. Mon. Not. R. Astron. Soc. 313, L47-L51 (2000) 21. Kley, W.: Dynamical Evolution of Planets in Disks. Celest. Mech. Dyn. Astr. 87, 85-97 (2003) 22. Kley, W., Peitz, J., Bryden, G.: Evolution of planetary systems in resonance. Astron. Astrophys. 414, 735-747 (2004) 23. Kominami, J., Ida, S.: The Effect of Tidal Interaction with a Gas Disk on Formation of Terrestrial Planets. Icarus 157, 43-56 (2002) 24. Lee, M.H., Peale, S.J.: Dynamics and Origin of the 2:1 Orbital Resonances of the GJ 876 Planets. Astrophys. J. 567, 596-609 (2002) 25. Lin, D.N.C., Bodenheimer, P., Richardson, D.C.: Orbital migration of the planetary com- panion of 51 Pegasi to its present location. Nature 380, 606-607 (1996) 26. Malhotra, R.: The Origin of Pluto's Orbit: Implications for the Solar System Beyond Neptune. Astron. J. 110, 420-429 (1995) 27. Masset, F.S., Morbidelli, A., Crida, A., Ferreira, J.: Disk Surface Density Transitions as Protoplanet Traps. Astrophys. J. 642, 478-487 (2006) 28. Michtchenko, T.A., Malhotra, R.: Secular dynamics of the three-body problem: application to the υ Andromedae planetary system. Icarus 168, 237-248 (2004) 29. Michtchenko, T.A., Rodr´ıguez, A.: Modeling the secular evolution of migrating planet pairs. (Mon. Not. R. Astron. Soc., in press, preprint: arXiv:1103.5485) (2011) 30. Mignard, F.: The evolution of the lunar orbit revisited - I. Moon and Planets 20, 301-315 (1979) 31. Patterson, C.W.: Resonance capture and the evolution of the planets. Icarus 70, 319-333 (1987) 32. Peale, S.J.: Origin and Evolution of the Natural Satellites. Ann. Rev. Astron. Astrophys. 37, 533-602 (1999) 33. Pont, F.: Empirical evidence for tidal evolution in transiting planetary systems. Mon. Not. R. Astron. Soc. 396, 1789-1796 (2009) 34. Rodr´ıguez, A., Ferraz-Mello, S.: Tidal decay and circularization of the orbits of short- period planets. EAS Publications Series, 42, 411-418 (2010) 35. Rodr´ıguez, A., Ferraz-Mello, S., Michtchenko, T.A., Beaug´e, C., Miloni, O.: Tidal decay and orbital circularization in close-in two-planet systems. (Mon. Not. R. Astron. Soc., in press, preprint: arXiv:1104.0964) (2011) 36. Tittemore, W.C., Wisdom, J.: Tidal evolution of the Uranian satellites. I - Passage of Ariel and Umbriel through the 5:3 mean-motion commensurability. Icarus 74, 172-230 (1988) 37. Smart, W.M.: Celestial mechanics. London (1960)
1311.3498	2	1311	2013-11-18T20:33:18	A quantification of hydrodynamical effects on protoplanetary dust growth	[ "astro-ph.EP" ]	Context. The growth process of dust particles in protoplanetary disks can be modeled via numerical dust coagulation codes. In this approach, physical effects that dominate the dust growth process often must be implemented in a parameterized form. Due to a lack of these parameterizations, existing studies of dust coagulation have ignored the effects a hydrodynamical gas flow can have on grain growth, even though it is often argued that the flow could significantly contribute either positively or negatively to the growth process. Aims. We intend to provide a quantification of hydrodynamical effects on the growth of dust particles, such that these effects can be parameterized and implemented in a dust coagulation code. Methods. We numerically integrate the trajectories of small dust particles in the flow of disk gas around a proto-planetesimal, sampling a large parameter space in proto-planetesimal radii, headwind velocities, and dust stopping times. Results. The gas flow deflects most particles away from the proto-planetesimal, such that its effective collisional cross section, and therefore the mass accretion rate, is reduced. The gas flow however also reduces the impact velocity of small dust particles onto a proto-planetesimal. This can be beneficial for its growth, since large impact velocities are known to lead to erosion. We also demonstrate why such a gas flow does not return collisional debris to the surface of a proto-planetesimal. Conclusions. We predict that a laminar hydrodynamical flow around a proto-planetesimal will have a significant effect on its growth. However, we cannot easily predict which result, the reduction of the impact velocity or the sweep-up cross section, will be more important. Therefore, we provide parameterizations ready for implementation into a dust coagulation code.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. sellentinetal_v8.1 June 21, 2021 c(cid:13)ESO 2021 A quantification of hydrodynamical effects on protoplanetary dust growth E. Sellentin, J. P. Ramsey, F. Windmark(cid:63), C. P. Dullemond Universität Heidelberg, Zentrum für Astronomie, Institut für Theoretische Astrophysik, Albert-Überle-Str. 2, D-69120 Heidelberg, Germany e-mail: [email protected] June 21, 2021 ABSTRACT Context. The growth process of dust particles in protoplanetary disks can be modeled via numerical dust coagulation codes. In this approach, physical effects that dominate the dust growth process often must be implemented in a parameterized form. Due to a lack of these parameterizations, existing studies of dust coagulation have ignored the effects a hydrodynamical gas flow can have on grain growth, even though it is often argued that the flow could significantly contribute either positively or negatively to the growth process. Aims. We intend to qualitatively describe the factors affecting small particle sweep-up under hydrodynamical effects, followed by a quantification of these effects on the growth of dust particles, such that they can be parameterized and implemented in a dust coagulation code. Methods. Using a simple model for the flow, we numerically integrate the trajectories of small dust particles in disk gas around a proto-planetesimal, sampling a large parameter space in proto-planetesimal radii, headwind velocities, and dust stopping times. Results. The gas flow deflects most particles away from the proto-planetesimal, such that its effective collisional cross section, and therefore the mass accretion rate, is reduced. The gas flow however also reduces the impact velocity of small dust particles onto a proto-planetesimal. This can be beneficial for its growth, since large impact velocities are known to lead to erosion. We also demonstrate why such a gas flow does not return collisional debris to the surface of a proto-planetesimal. Conclusions. We predict that a laminar hydrodynamical flow around a proto-planetesimal will have a significant effect on its growth. However, we cannot easily predict which result, the reduction of the impact velocity or the sweep-up cross section, will be more important. Therefore, we provide parameterizations ready for implementation into a dust coagulation code. Key words. accretion, accretion disks -- protoplanetary disks -- stars: circumstellar matter -- planets and satellites: formation In the classical incremental growth scenario, planetesimals are formed by the coagulation of dust across several orders of mag- nitude in mass. During this growth phase, many different phys- ical processes are important, but not all are beneficial. Indeed, several barriers have been found that stop dust particles from growing to sizes where gravity can aid in coagulation. For exam- ple, the charge barrier (Okuzumi et al. 2009), the radial drift and fragmentation barriers (Nakagawa et al. 1986; Weidenschilling et al. 1997; Brauer et al. 2008; Birnstiel et al. 2010), or the bouncing barrier (Zsom et al. 2010; Windmark et al. 2012b). In order to overcome these barriers, it is important to exam- ine what effects neglected physical processes might have. One such effect is the hydrodynamical flow of disk gas around larger dust particles such as pebbles or small planetesimals, to which we will collectively refer to as proto-planetesimals. In this paper, we focus on the question of whether such a flow pattern is bene- ficial for the growth of protoplanetary dust. Although this ques- tion has already been partially addressed in Sekiya & Takeda (2003, hereafter ST03), a quantification of these effects over a large parameter space such that the results can be implemented in a numerical dust growth code still lacks in the literature. Whether colliding dust particles stick to each other and grow, bounce off each other, or disrupt one or both of the collision part- ners depends on a series of parameters, out of which the impact velocity is one of the most important. Experimental work has (cid:63) Member of IMPRS for Astronomy & Cosmic Physics at the Uni- versity of Heidelberg shown (for a summary, see Blum & Wurm 2008; Güttler et al. 2010) that particles preferentially stick to each other if they col- lide at low velocities (Wurm & Blum 1998), and bounce off or disrupt at high collision velocities (e.g. Wurm et al. 2005; Kothe et al. 2013). It was also found that larger dust aggregates are eroded due to high velocity impacts of dust monomers, a pro- cess that has become known as monomeric erosion (Schräpler & Blum 2011). Many of these experimental findings are in- cluded in the numerical simulations that model dust growth in protoplanetary disks (e.g. Weidenschilling 1997; Ormel & Cuzzi 2007; Brauer et al. 2008; Birnstiel et al. 2010; Zsom et al. 2010; Okuzumi et al. 2012). Numerical studies also show that the growth of dust does not always proceed by a hierarchical coagu- lation of equal sized particles, but instead that collisions between particles of large size ratios can lead to the formation of planetes- imals (Xie et al. 2010; Windmark et al. 2012a). In this scenario, proto-planetesimals can grow by sweeping up a secondary pop- ulation of particles kept small by the collisional growth barriers. To date, collision velocities in dust growth codes have been calculated from Brownian motion, radial and azimuthal drift, vertical settling, and turbulence induced motions of the dust par- ticles. The collisional cross section of the particles is further- more assumed to be equal to the geometrical cross section. This neglects potential effects of hydrodynamical gas flow past dust particles on the growth process. It is however expected that these effects can be quite important for collisions between particles of a large size ratio, i.e., in a sweep-up growth scenario. Geomet- Article number, page 1 of 9 3 1 0 2 v o N 8 1 . ] P E h p - o r t s a [ 2 v 8 9 4 3 . 1 1 3 1 : v i X r a rical cross sections are only applicable if the trajectories of the colliding particles follow straight lines. This is not necessar- ily the case for a small dust particle passing near a large proto- planetesimal surrounded by a hydrodynamical flow pattern: The drag of the gas flow around the proto-planetesimal can deflect the small dust particle, resulting in a curved trajectory. In the most extreme case, this deflection can prevent a collision, preventing the sweep-up of the dust particle. The collisional cross section must therefore be modified to take into account this deflection. The drag force of the gas flow on a small particle can also af- fect its speed. This then modifies the collision velocity between the dust particle and proto-planetesimal, which is decisive for determining whether a collision leads to sticking, bouncing, or erosion. In protoplanetary disks, we expect hydrodynamical flow pat- terns to form around dust particles much larger than the mean free path of the gas because the gas orbits at sub-Keplerian speed while the dust orbits at Keplerian speed. This gives rise to a rel- ative motion between disk gas and dust, appearing in the rest frame of the dust as a headwind. In this work, we aim to use numerical simulations to study the effect that such a flow pattern will have on the dust coagulation. We assume that our proto-planetesimals are small enough so that gravity is negligible, and constraints on this assump- tion are presented below. For studies which include the grav- ity of larger planetesimals, see, for e.g., Lambrechts & Johansen (2012); Morbidelli & Nesvorny (2012); Ormel (2013). This paper is organized as follows: In Sect. 1, we present our assumptions and general properties of the flow pattern around a proto-planetesimal. In Sect. 2, we describe the changes to the collisional cross section and the impact velocity which we ob- serve from integrating trajectories of small dust particles in the flow around a proto-planetesimal. In Sect. 3, we turn to the ques- tion of whether gas flow around proto-planetesimals can return debris from a disruptive collision to the proto-planetesimal sur- face. In Sect. 4, we summarize our results. 1. Validity range of the model In the limit of high viscosity (i.e. small Reynolds numbers), the steady-state (∂/∂t = 0) velocity field vg of a gas flow around a sphere of radius R in spherical polar coordinates can be derived from the Navier-Stokes equations (Greiner & Stock 1991; origi- nally due to Stokes): (cid:33) (cid:32) R3 (cid:16) + −1 3R 4rs 4r3 s cos θ r − sin θ θ + 3R 4r3 s (cid:17) (cid:32) (cid:33) (v∞(θ)·rs)rs 1− R2 r2 s , (1) vg (rs, θ) = v∞(θ) where v∞(θ) = v∞ is the upstream velocity, v∞ is the headwind velocity and constant, and rs denotes the radial distance measured from the center of the sphere. As rs → ∞, vg reduces to −v∞. Eq. (1) describes the flow around a spherical proto-planetesimal if a) the proto-planetesimal radius R is much larger than the mean free path λ in the disk, b) the flow is in- compressible, c) laminar and unseparated, d) the gravity of the proto-planetesimal is negligible, and e) the proto-planetesimal does not rotate. In the following, we assume (e) to be true and calculate when (a) -- (d) are valid for a protoplanetary disk in both the minimum mass solar nebula model (MMSN) (Weidenschilling 1977b; Hayashi et al. 1985) and the model of Desch (2007). Al- though the two models provide similar constraints, we quote the results for both. Article number, page 2 of 9 Assumption a) Hydrodynamics is suitable to describe the macroscopic properties of a flow around a proto-planetesimal if R (cid:38) 100λ. The mean free path λ is related to the number density n of gas molecules by λ = 1/(nσg), where σg is the collisional cross section of the gas molecules. For n, we use the number density n0 at the mid-plane of a protoplanetary disk, which is related to the surface density Σ(r) by: (2) n0(r) = √ Σ(r) 2π hµmp , mass µmp, where mp is the mass of a proton, cs = (cid:112)kBT(r)/µmp with the disk scale height h = cs/ΩK, and average molecular is the isothermal sound speed, kB is the Boltzmann constant, and ΩK is the Keplerian angular frequency. For the disk model, we assume µ = 2.3 and σg = 2·10−19 m2 (ST03) for the collisional cross section of Hydrogen, and a stellar mass of 1 M(cid:12). We adopt the disk temperature profile of Alexan- der et al. (2004), which is consistent with observations: T(r) = 100 (r/1 AU)−1/2 K, where r is the heliocentric distance, expressed in astronomical units. The gas surface densities for the Desch (2007) and MMSN (Hayashi et al. 1985) models are, respectively: ΣDesch(r) = 5 · 105 (r/1 AU)−2.17 kg m−2; ΣMMSN(r) = 1.7 · 104 (r/1 AU)−3/2 kg m−2. Substituting the surface densities into Eq. (2), the minimal radius Rmin = 100λ that a proto-planetesimal must have for assumption (a) to hold is then: Rmin,Desch(r) = 3 · 10−2 (r/1 AU)3.42 m; Rmin,MMSN(r) = 8 · 10−1 (r/1 AU)2.75 m, which increases with heliocentric distance for both disk models. (6) (7) (4) (5) (3) Incompressibility approximately holds true for Assumption b) small Mach numbers, M (cid:46) 0.1. From the temperature profile (3), the isothermal sound speed can be written as: cs(r) = 600 (r/1 AU)−1/4 m s−1. (8) The headwind velocity v∞ = vK − vg originates from the sub- Keplerian rotation of the disk gas, and can be written as v∞ = ηvK with (ST03): η = − (9) 1 Kµmpn0 ∂P ∂r , 2rΩ2 where P is the gas pressure as given by the ideal gas law. Substituting the surface density profiles Eqs. (4) & (5), and the temperature profile Eq. (3) into the expression for η, we find the headwind velocity for the Desch and MMSN models to be v∞,Desch ≈ 24 m s−1; v∞,MMSN ≈ 20 m s−1, and independent of heliocentric distance. From Eq. (8), the Mach number of a gas flow around a proto- planetesimal is thus on the order of 10−2 to 10−1 in both models for reasonable values of r, and assumption (b) is valid. (10) (11) Sellentin et al.: Hydrodynamical effects on dust growth Assumption c) To determine whether the flow around a proto- planetesimal can be described as laminar, we first estimate the molecular viscosity of gas in a protoplanetary disk, which can be determined from: 1 2 ν = (12) ¯uthλ. √ Using Eq. (8), the mean thermal speed of the disk gas is given by ¯uth(r) = 8/π cs(r). The mean free path can be calculated from Eq. (2), and thus we have, νDesch(r) = 0.13 (r/1 AU)3.17 m2s−1; νMMSN(r) = 3.6 (r/1 AU)2.5 m2s−1. Based on experimental results, viscous hydrodynamic flow past a sphere will remain perfectly laminar and unseparated if the Reynolds number of the flow is Recrit (cid:46) 22 (e.g. Taneda 1956), where the Reynolds number of flow around a spherical proto- planetesimal is given by: (13) (14) Note also that the gravity of a proto-planetesimal with R (cid:46) Rgrav does not affect the disk gas because the escape velocity from its surface is much smaller than the local thermal speed of the gas molecules. In assumption (a), we found that the minimally required ra- dius of a proto-planetesimal which induces a hydrodynamical flow grows with heliocentric distance. There must consequently be a distance when the radius exceeds the above specified Rgrav. Setting Rmin = Rgrav in Eqs. (6) & (7), this occurs at rgrav,Desch = 14 AU; rgrav,MMSN = 8 AU, (20) (21) for the Desch and MMSN models, respectively. Beyond these heliocentric distances, a proto-planetesimal cannot fulfill both prerequisites of being large enough to induce a hydrodynami- cal flow and small enough for gravity to be negligible, and our model is no longer valid. Re = 2Rv∞ ν . (15) 2. Hydrodynamical effects on collisions between proto-planetesimals and dust grains Other than the proto-planetesimal radius, all variables in Eq. (15) are now fixed by the choice of disk model. The statement that a flow will remain perfectly laminar below Recrit ≈ 22 can then be used to solve for the maximal radius of a proto-planetesimal: Rcrit,Desch = 0.059 (r/1 AU)3.17 m; Rcrit,MMSN = 1.98 (r/1 AU)2.5 m. Laminarity therefore holds if the proto-planetesimal radius does not exceed Rcrit(r). (16) (17) Assumption d) We assume that the gravity of the proto- planetesimal is negligible if the drag force due to the friction of the fluid, FD, is at least FD > 102FG, where FG is the gravita- tional force of the proto-planetesimal. The drag force can be written as FD = mv/ts where m, v, and ts are a dust particle's mass, speed, and stopping time. The largest stopping time that leads to a deflection in the flow pattern is approximately equal to the time tp that a particle needs to be advected by the gas flow through the perturbed region around the proto-planetesimal. This timescale is thus tp ≈ L/v∞, where L is roughly the extent of the perturbed region. At a distance of rs = 100 R, the flow pattern Eq. (1) virtually equals the upstream velocity v∞. We therefore set the extent of the perturbed region to L = 100 R and thus FD = mv2∞/100 R. We evaluate the gravitational force at the surface of the proto-planetesimal by assuming early proto-planetesimals have a volume filling factor of 0.6 (Geretshauser et al. 2011). We also assume that early proto-planetesimals consist of the same mate- rial as asteroids, which have typical densities of 3 · 103 kg m−3 (Carry 2012). From this, we estimate the density of the proto- planetesimal to be ρp = 0.6 × 3 · 103 kg m−3. The statement FD > 102FG then becomes: 100 R > 102 Gm 4 mv2∞ 3 πR3ρp R2 , (18) where G is the gravitational constant. The maximally allowed proto-planetesimal radius before the gravitational force on the dust particles becomes comparable to the drag force is then Rgrav ≈ 250 m. (19) In order to investigate the change of the effective sweep-up cross section σeff and the impact velocity vimp due to the drag in a hy- drodynamical flow pattern, we calculate the trajectories of dust particles from the equation of motion, , ts (22) v = −(v − vg) where the dot denotes a time derivative, v is the dust particle's velocity, vg is the velocity of the gas flow (Eq. 1), ts = p/FD is the stopping time of the dust particle, and p = mv is the dust particle momentum. For example, in the Epstein regime, when the particle radius a < 9λ/4, the stopping time is given by ts,Ep = ρpa/ρg ¯uth. This quantity can be regarded as a parti- cle property parameterizing the effects which contribute to the drag of this particle in a specified medium. Hence, the follow- ing work can be applied to real protoplanetary disks if stopping times of dust particles are known. Theoretical stopping times for particles in different drag regimes are summarized in Wei- denschilling (1977a). 2.1. Qualitative aspects of the flow pattern and sweep-up cross section Far upstream and downstream of a spherical proto-planetesimal, the streamlines of the chosen flow pattern Eq. (1) are parallel. Meanwhile, near the proto-planetesimal, the stream lines diverge upstream and converge again downstream, creating a flow pat- tern that is rotationally symmetric about the flow axis. If dust particles approach the proto-planetesimal in this flow pattern from all directions -- which could be possible if they are stirred up by nearby turbulent eddies and acquire randomly orientated velocities, or if the headwind velocity is modified by a large- scale turbulent eddy -- then they should experience the following scenarios, all of which are depicted in Fig. 1: arrow a) Dust particles that approach the proto-planetesimal from upstream and parallel to the flow axis will be de- flected away from the proto-planetesimal due to the diverging flow pattern. In particular, particles with a large impact parameter that would collide with the proto-planetesimal in the absence of the flow pattern Article number, page 3 of 9 can be deflected such that they pass over the rim of the proto-planetesimal and miss it. The collisional cross section of the proto-planetesimal is then reduced rela- tive to its geometrical cross section. Furthermore, the magnitude of the deflection will depend on the dust particle stopping time: The larger the stopping time, the smaller the deflection. Lastly, due to the rotational symmetry of the flow pattern, the collisional cross sec- tion will be circular. arrow b) Particles that approach the proto-planetesimal from downstream and parallel to the flow axis do so under a converging gas flow. Even particles that lie outside the geometrical cross section of the proto-planetesimal can then be pushed onto a collision course by the gas flow. A collision, however, only occurs if the particles have sufficient inertia such that the gas flow cannot re- verse their direction of motion (i.e. a large stopping time). For particles that approach from downstream and collide with the proto-planetesimal, the collisional cross section must then be larger than the geometrical cross section, but will remain circular due to rotational symmetry. arrow c) Particles that approach the proto-planetesimal at an angle θ to the flow axis experience the flow pattern as a cross wind. Particles not initially on a collision course can be deflected such that they now collide with the proto-planetesimal. In this scenario, the cross wind introduces an additional dependence on sin θ to the collisional cross section. For a spherical proto- planetesimal, this causes a small stretching of the col- lisional cross section in the upstream direction, result- ing in a top-down symmetric oval. The slight breaking of left-right symmetry occurs because the flow pattern also depends on θ (Eq. 1). arrow d) In contrast, particles that approach the proto- planetesimal under an angle to the flow axis, but that were originally on a collision course, can be deflected such that they now miss the proto-planetesimal. To test for the above qualitative features of the dust trajectories, we ran a set of example simulations (see Sect. 2.2), and found that all cases occur as described. The reduction of the proto- planetesimal collisional cross section for particles that approach from upstream (arrow (a) in Fig. 1) was found to be the most significant, while the enhancement (or distortion) of the cross section for particles that approach from downstream (or under an angle) is effectively negligible. We expect particles that approach from upstream to be the most frequently occurring situation in a sweep-up scenario, and thus in the following we focus only on the case of upstream par- ticles travelling parallel to the flow axis. 2.2. Description of simulations and initial conditions In total, we performed 16000 simulations that numerically inte- grate Eq. (22) for the trajectories of dust particles in the flow pat- tern around a proto-planetesimal using a 4th-order Runge-Kutta based, backwards Euler method with adaptive stepping. Each simulation uses a different combination of proto-planetesimal radii, headwind velocities, and particle stopping times, con- strained in Sect. 1 and summarized in Table 1. The calculation of the headwind velocity v∞ in Sect. 1 only accounts for the sub-Keplerian gas velocity as the origin of a relative motion between disk gas and proto-planetesimals. The Article number, page 4 of 9 Fig. 1. Trajectories of dust particles resulting from their interaction with the gas flow are indicated by the solid arrows. The gray lines represents stream lines of the gas flow around the proto-planetesimal, as calculated from Eq. (1). The feathered arrows indicate the direction of gas flow. See the text for a discussion of scenarios (a) -- (d). total headwind velocity should also include, for example, disk turbulence, and for the sake of generality we vary the headwind velocities between v∞ ∈ [15, 60] m s−1 (c.f. the headwind veloc- ities calculated in Weidenschilling 1977a). In our simulations of particle trajectories, we assume the par- ticles are initially at rest with respect to the gas, and thus we set the initial velocity of the dust particles equal to the headwind velocity. Although dust particles that react significantly to the disturbed gas flow around the proto-planetesimal consequently must have a relatively small stopping time, and are thus essen- tially comoving with the gas, only dust particles with ts = 0 will always remain perfectly at rest with respect to the gas. The dust particles start at a distance H · R upstream from the proto-planetesimal, where the parameter 1 < H ≤ 150 is a multiple of the proto-planetesimal radius. The larger the value of H, the closer the gas is to a uniform flow pattern, and the longer the particles will remain at rest with respect to the gas. To investigate which impact parameters lead to collisions, we vary the initial impact parameter p in steps of 10−4R between zero and 0.01R, and then by steps of 0.01R up to the planetesimal radius R. Fig. 2 shows a representative example of trajectories for dust particles with different impact parameters. While particles with small impact parameters collide with the proto-planetesimal, beyond a certain value of p = pmax (< R) the particle will be deflected around the proto-planetesimal, and impacts cease to occur. The maximum impact parameter, pmax, that leads to a collision is related to the effective sweep-up cross section σeff by: σeff = πp2 max. (23) acbdTable 1. Simulation parameter space. The range of values for each parameter is determined with regards to Sect. 1. Sellentin et al.: Hydrodynamical effects on dust growth Parameter R v∞ ts Definition proto-planetesimal radius headwind velocity dust particle stopping time Range 1 − 150 m 15 − 40 m s−1 10−1 − 104 s As restricted by gas mean free path & neglect of gravity protoplanetary disk model size of perturbed gas region slightly less than one stopping time is required for the flow to deflect a particle around the proto-planetesimal. Fig. 3 demonstrates how the effective sweep-up cross section varies with the parameter x. The open points are taken from our simulations, while the dashed line plots the function which best- fits the data points. This function is given by = exp (−Dσ(H) x) , σeff σgeom (25) where σgeom = πR2 is the geometrical cross section. The func- tion Dσ(H) accounts for the dependence of the effective cross section (i.e. the amount of deflection) on the starting distance H from the surface of the proto-planetesimal. Although Eq. (25) typically fits the simulation results to within 10% difference, it does not reproduce the effect that par- ticles no longer impact the proto-planetesimal for x > 0.8 := xcutoff. We therefore suggest the following form for the best-fit function: x < xcutoff; x ≥ xcutoff. (26) (cid:40) exp (−Dσ(H) x) σeff σgeom = 0 Fig. 2. The numerically calculated trajectories of dust particles with different impact parameters in the viscous laminar flow around a spheri- cal proto-planetesimal are indicated by solid lines. In this representative example, the proto-planetesimal radius is 10 m, the headwind velocity of the gas is 20 m s−1, and the particle stopping time is 1 s. The dust particles enter the plot from the right, and the axes have units of meters. 2.3. The influence of the flow pattern on the sweep-up cross section For the entire set of 16000 simulations, we find the outcome depends only on the dimensionless parameter: x := R v∞ts . (24) This parameter x is the ratio of the hydrodynamical time scale for flow past the proto-planetesimal, tp = R/v∞, and the parti- cle stopping time, ts. Note that x is equal to the reciprocal of the dimensionless stopping time (c.f. ST03). We prefer to use x here because it characterises the strength of the deflection by the flow for a particle with a given ts. As x increases, so does the particle deflection, eventually reaching a point where the particle no longer collides with the proto-planetesimal and the effective sweep-up cross section goes to zero. Conversely, as x → 0, the influence of the gas flow on the particle becomes insignificant, and the resulting effective cross section reduces to the geometri- cal cross section. When x > 0.8, we observe that σeff/σgeom < 10−4, and we therefore consider it to be zero. As such, for x > 0.8, the deflec- tion of dust particles is substantial enough to entirely prevent impacts onto the proto-planetesimal, and the particles are in- stead advected around it. That collisions cease for x < 1 means Fig. 3. Decreasing effective sweep-up cross section as a function of the parameter x for H = 10. The open circles represent simulation data, and the dashed line corresponds to the best-fitting function Eq. (25). Above x > xcutoff, the simulations show that particles no longer impact the proto-planetesimal. In this regime, the best-fitting function overes- timates the sweep-up cross section. If a dust particle starts very close to the proto-planetesimal surface, i.e. H (cid:38) 1, the flow does not have a significant opportu- nity to deflect the particle and the sweep-up cross section is then very close to the geometrical cross section. Comparing simula- tions with different values of H, we find: Dσ(H) = 4.00 − 4.75 · exp (cid:18)− H (27) (cid:19) . 5.38 Article number, page 5 of 9 -15-10-505101520-15-10-505101520y[m]x[m] 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1σeff / σgeomxThe maximum deviation of Eq. (27) from the simulation results is 10%, although the typical deviation is only ∼ 2%. For H > 100, the function Dσ(H) (cid:39) 4. This behaviour is physical: In Sect. 1 we calculated that the viscous, laminar flow pattern reaches a gas velocity of 98% the upstream velocity v∞ at a distance of 100 R. That the function Dσ(H) is constant be- yond H = 100 therefore originates from the gas velocity being effectively constant in this regime. In a protoplanetary disk, it is reasonable to assume that the collision timescale for small dust is long enough that the parti- cles have spent several stopping times in the gas flow, and they are at rest with respect to the gas as they approach the proto- planetesimal. Thus, one can safely choose the asymptotic value Dσ(H) = 4 when applying Eq. (25) to a sweep-up growth sce- nario. If one wants to consider the influence of turbulence on parti- cle velocities, then one might need to consider small values of of H and Dσ(H) < 4. However, this means the particle is not ini- tially at rest with respect to the gas flow, nor travelling parallel to the flow axis, and this is beyond the scope of the current study. 2.4. The influence of the flow pattern on the impact velocity Here, we summarize the effects that we observe for the flow pat- tern on the measured impact velocity vimp = vimp of dust par- ticles onto the proto-planetesimal. From Eq. (1), the gas veloc- ity in the perturbed region steadily decreases as a particle ap- proaches the proto-planetesimal surface. Directly at the surface, the gas velocity is zero. The reduced gas velocity exerts a drag force on approaching dust particles, thereby reducing their im- pact velocity. In our simulations, we find the impact velocity to be virtually independent of the impact parameter -- under the condition that the chosen impact parameter leads to a collision at all. In the following, we therefore take the average of the impact velocities, weighted by the impact parameter p, over all impact parameters that lead to collisions and refer to it as the impact velocity vimp. Fig. 4. Decreasing impact velocity of dust particles onto a proto- planetesimal as a function of the parameter x. The open circles repre- sent simulation data, while the solid line corresponds to the best-fitting function Eq. (28). Fig. (4) demonstrates how the impact velocity of dust parti- cles varies with x. The open circles are taken from our simula- Article number, page 6 of 9 (28) = exp (−Dvx) ; Dv = 3.3. tions and the solid line depicts the best fitting function: vimp v∞ Eq. (28) describes that particles with small x will impact the proto-planetesimal with velocity equal to v∞. The impact ve- locity decreases with increasing x because the drag that brakes the particles also increases with x. Eq. (28) fits the simulation results to better than 10% and, as with Eq. (25), Eq. (28) is only applicable for x < xcutoff because an impact velocity is ill-defined for particles that do not impact the proto-planetesimal. Schräpler & Blum (2011) find that large dust particles are prone to erosion by the impact of smaller particles in a process that has become known as monomeric erosion. More specifi- cally, they find the mass loss of the target scales linearly with the impact velocity of the monomers, and that the process can result in losses of up to ∼ 10× the mass of the impactor. We find that a flow around proto-planetesimals reduces the impact velocity of small dust particles (i.e. large x, all else being equal), and thus will have the consequence of reducing the effi- ciency of monomeric erosion. In the extreme limit of x > xcutoff, the flow entirely prevents impacts and thereby also monomeric erosion. Thus, if a proto-planetesimal grows large enough that a hydrodynamical gas flow develops around it, it can become partially shielded against erosive high velocity impacts. 2.5. Dependence of the results on the Reynolds number From Eqs. (13), (14), and the range of values for R and v∞ in Table 1, one can calculate the minimal and maximal values of the Reynolds numbers for our parameter space: Remax = 2Rmaxv∞,max/ν(r); Remin = 2Rminv∞,min/ν(r). The resulting range of values, constrained by the assumptions of Sect. 1, are shown in Fig. 5. For example, at r = 5 AU, in the Desch (MMSN) model with a headwind velocity of 24 m s−1 (20 m s−1), requiring Re ≤ 22 restricts our results to proto- planetesimal radii R ≤ 9.8 m (≤ 110.7 m). Eq. 1 assumes Re (cid:28) 1, and is therefore strictly valid only in this regime. Meanwhile, experimental results affirm that the flow streamlines upwind of a sphere remain remarkably similar to Eq. (1) for Reynolds numbers Re (cid:46) 22 (e.g. Taneda 1956; Van Dyke 1982 and references therein). At larger Reynolds numbers however, the flow pattern downstream of the sphere qualitatively changes as the flow separates and vortices begin to form. As we are only concerned with the sweep-up of dust particles on the upstream side of the proto-planetesimal, the application of Eq. (1) is thus expected to accurately describe the deflection of particles even when Re (cid:54)(cid:28) 1. In an attempt to quantify this expectation, we examined two additional analytical approximations for viscous flow past a sphere, described in detail in Van Dyke (1964), and which ex- plicitly depend on the Reynolds number. The first approxima- tion, due to Oseen, accounts for the convective terms of the full Navier-Stokes equations (which Eq. 1 ignores) via linearisation. However, this approximation suffers from the issue that the ve- locity at the surface of the sphere depends on a term O(Re). In contrast, Eq. (1) correctly predicts zero velocity at the surface of the sphere. The second, which we refer to here as "Proudman and Pear- son's two-term approximation" (Proudman & Pearson 1957; 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1Vimp/V∞xSellentin et al.: Hydrodynamical effects on dust growth Fig. 5. Reynolds number as a function of heliocentric distance r for the Desch and MMSN disk models. The Remin (dashed) line corresponds to a proto-planetesimal radius and headwind velocity of (R, v∞) = (1 m, 15 m s−1), while the Remax (dotted) line corresponds to (150 m, 40 m s−1). The solid horizontal lines denote Reynolds numbers of 2 and 22, while the solid vertical line indicates the critical gravitational heliocentric distance for each disk model (Eqs. 20 & 21). The successively darker shading distinguishes the regions of our parameter space that fall below Reynolds numbers of 22 and 2, respectively. "PP's approximation" for short), is derived with the region near the sphere in mind, and preserves a zero velocity at the surface. Indeed, in the immediate vicinity of the sphere, this approxi- mation does a remarkable job of matching experiment, even for Re > 22 (Van Dyke 1964, p. 160). The trade-off, however, is that the velocity far from the sphere can significantly exceed v∞ when Re > 1 and, in general, does not return to a uniform flow (as do Eq. 1 and Oseen's approximation). Consequently, particles can be accelerated beyond v∞, impacting the proto-planetesimal with vimp/v∞ > 1. Although both approximations are strictly valid only for small Reynolds numbers, because of their explicit dependence on Re, we apply them here to provide some intuition on the behaviour of particle accretion when Re > 1. Furthermore, as our results strongly depend on the conditions in the immedi- ate vicinity of the proto-planetesimal, and Oseen's approxima- tion does not recover a zero velocity at the surface of the proto- planetesimal, we choose to henceforth discuss only PP's approx- imation. Relative to PP's approximation, we expect simulations ap- plying Eq. (1) to progressively overestimate the deflection of dust particles as Re increases. Indeed, additional simulations run using PP's approximation demonstrate that xcutoff increases linearly with Re. In other words, as the relative importance of in- ertial forces increase, all else being equal, a successively smaller stopping time is required for a particle to be deflected around the proto-planetesimal. Similarly, the additional simulations show that Dσ(H) and Dv decrease with increasing Re before asymptoting to small but non-zero [O(10−1)] values for Re (cid:38) 15. This manifests as an upward shift of the curves in Figs. 3 & 4 and, as with xcutoff, this illustrates that the amount of particle deflection decreases with increasing Re. In regards to the accuracy of using Eq. (1) when Re (cid:54)(cid:28) 1, we observe that xcutoff, Dσ(H), and Dv match the results from PP's approximation for Re (cid:46) 0.1. Moreover, these variables agree with PP's approximation to within 50% for Re ≤ 1, and within a factor of two for Re ≤ 2. Beyond Re = 2, notwithstanding the variation of xcutoff, Dσ(H), and Dv, the results applying Eq. (1) remain qualitatively correct with respect to PP's approximation. Thus, Fig. 5 depicts where in our parameter space the results of Stokes' and PP's approximations agree to within a factor of 2 (dark grey), and where these approximations still qualitatively agree (light gray). That said, keep in mind that the additional approximations considered here are derived under the assumption of small Re, and thus the above trends are at best estimates. It should also be noted that the effects of Re on xcutoff, Dσ(H), and Dv can at least be partially explained by the locally enhanced particle velocities (relative to v∞) observed when using PP's approximation. Direct numerical simulations of subsonic, laminar flow past a proto- planetesimal are needed to verify these results, and this is left to future work.1. 3. Hydrodynamical effects on the reaccretion of collisional debris If a dust aggregate collides with a proto-planetesimal and the collision velocity is large enough, the aggregate will be dis- rupted into fragments. Debris of the particle can then re-enter the gas flow around the proto-planetesimal. In the literature, the question has been raised whether the gas flow can return such fragments to the surface of the proto-planetesimal, and thereby increase the efficiency of sweep-up growth (Wurm et al. 2001, 2004; Sekiya & Takeda 2003, 2005). In this section, we attempt to address the escape of collisional debris from an impact site by launching dust particles directly from the surface of the proto-planetesimal into the flow pattern with random directions. 3.1. On the free molecular flow and straight trajectories Sekiya & Takeda (2003) find that the gas can return collisional debris to the proto-planetesimal surface in the free molecular flow. The term free molecular flow describes the situation where 1 For direct numerical simulations of particle accretion in fully turbu- lent flow see, for e.g., Mitra et al. 2013. Article number, page 7 of 9 02468101214r[AU]10−210−1100101102103ReDeschReminRemax02468101214r[AU]10−210−1100101102103MMSNReminRemaxthe proto-planetesimal radius is comparable to or smaller than the mean free path of the gas. The possibility of averaging over the random thermal motion of the gas particles is then no longer available, and the fluid has instead to be described by the motion of individual particles. Even in free molecular flow, the transformation into the rest frame of the proto-planetesimal induces an apparent macro- scopic motion of the gas, namely the headwind. ST03 assume the gas motion can then be described by straight, parallel stream- lines that intersect with the proto-planetesimal. However, the thermal motion of the gas will be superimposed on the ordered headwind. From the temperature profile Eq. (3), the mean thermal speed of gas molecules in the disk is then: ¯uth(r) (cid:39) 960 (r/1 AU)−1/4 m s−1. (29) At 1 AU, the random thermal motion outweighs the headwind velocity of v∞ ≈ 24 m s−1 (20 m s−1) by a factor of 40 (48), and by a factor of (cid:38) 20 (28) within the inner 14 AU (8 AU) of the disk for the Desch (MMSN) model. Clearly, the ordered macroscopic motion of the headwind will disappear under the random thermal motion of the free molecular flow, and thus parallel stream lines are not physically applicable in this context. However, if we launch dust particles from the surface of the proto-planetesimal into a flow pattern of parallel stream lines, then we do reproduce the result of ST03, namely that parallel streamlines lead to reaccretion of collisional debris on the up- stream side of the proto-planetesimal. 3.2. Reaccretion in a viscous laminar flow We also launched collisional debris from the surface of the proto- planetesimal into the laminar flow pattern of Eq. (1) with multi- ple ejection velocities and angles in the range 0 ≤ α ≤ π. Independent of the ratio between headwind velocity and the ejection velocity of the collisional debris, we find virtually no reaccretion, in agreement with the findings of ST03. In our simu- lations however, we observe a maximal fraction of ∼10−3 of par- ticles return to the proto-planetesimal surface. These re-impacts only occur for angles nearly tangential to the proto-planetesimal surface, and it is our opinion that they are the result of limited numerical precision and thus, physically, no reaccretion occurs. Fig. 6 shows a representative example of the possible trajec- tories for collisional debris. Plotted are the trajectories of four dust particles that leave the proto-planetesimal surface with the same initial location, direction α, and ejection velocity (2 m s−1), but with different stopping times. Particles with small stopping times only escape into the fluid layers just above the proto-planetesimal surface before being de- flected (trajectories 1 and 2). Since the gas flows around the proto-planetesimal, and the particles are dragged with it, they are not returned to the surface. Particles with a larger stopping time that are ejected against the headwind can reach the region upstream of the proto- planetesimal where the gas velocity points towards the proto- planetesimal. In this region, the particle direction is reversed by the gas flow and it then approaches the planetesimal (trajectory 3). However, because the direction has been reversed, the parti- cle stopping time must consequently still be relatively small. In our simulations, stopping times that lead to a reversal of direc- tion in front of the planetesimal also lead to a complete deflection of the particle and thus no reaccretion. For particles with even larger stopping times (trajectory 4), the dust particles leave the collisional cross section of the proto- planetesimal before their motion is reversed. Article number, page 8 of 9 In all four cases above, the ejected particles do not collide with the proto-planetesimal, and reaccretion does not occur. Fig. 6. Trajectories of collisional debris in a laminar flow. In this representative example, the headwind velocity is 20 m s−1, and the axes have units of meters. All dust particles begin with the same location, direction α, and velocity (20 m s−1), indicated by a dot on the proto- planetesimal surface. The stopping times of the dust particles are, for trajectories 1 -- 4, ts = 0.1, 1, 8, and 40 s, respectively. See the text for a discussion of the different outcomes. 4. Discussion & conclusions We have examined a number of issues related to the effects of hy- drodynamical gas flow around proto-planetesimals, with an in- terest in the consequences for coagulation efficiency, expanding upon the work of ST03. By numerically integrating the trajec- tories of dust particles in the gas flow described by Eq. (1), we have quantified how these particles are deflected and their impact velocities affected. We have also studied whether reaccretion of collisional debris remains possible in the presence of a laminar hydrodynamical flow. We have found that small particles that would impact the proto-planetesimal in the absence of a flow pattern can instead be deflected by the streaming gas and pass over the rim of the proto-planetesimal, avoiding a collision. Even if an impact oc- curs, the gas flow can decrease the relative velocity between the small particle and the proto-planetesimal, leading to generally less disruptive collisions. These effects are mostly important in a sweep-up scenario, occurring in the regime between small dust aggregates and larger proto-planetesimals, as studied by, e.g., Xie et al. (2010); Windmark et al. (2012a). Our model is valid for spherical, non-rotating proto- planetesimals with radii (cid:46) 250 m, and within 8 AU of the central star in a MMSN disk, 14 AU in a Desch (2007) type disk. The results of our model are summarized as follows: -- Sect. 2: The amount of deflection or deceleration experi- enced by a dust particle as a result of the flow pattern around -30-20-100102030-30-20-1001020304312Sellentin et al.: Hydrodynamical effects on dust growth a proto-planetesimal is purely a function of the parameter x = R/(v∞ts), where R is the radius of the proto-planetesimal, ts is the dust particle stopping time, and v∞ is the headwind velocity of the disk gas. -- Sect. 2.1 -- 2.3: The flow of disk gas around a proto- planetesimal reduces the effective cross section with which it can sweep up dust particles from the surrounding gas. For a spherical proto-planetesimal, the effective cross section is easily parameterized (Eq. 25). When x > xcutoff = 0.8, dust particles no longer impact the proto-planetesimal, and sweep-up is impossible. This corresponds to a dimension- less stopping time of 1.25, in rough agreement with ST03. For example, a proto-planetesimal with a radius R = 100 m at 5 AU would, in a MMSN (Desch) type disk with a headwind velocity of v∞ = 20 m s−1 (24 m s−1), have its effective cross section reduced by half (i.e. x (cid:39) 0.17 when D(H) = 4) for collisions with particles of stopping time ts ∼ 30 s (25 s), corresponding to a particle size of 0.5 µm (4 µm). Particles of stopping time ts (cid:46) 6.2 s (5.2 s) would not collide with the proto-planetesimal at all (x ≥ xcutoff), corresponding to particle sizes of (cid:46) 0.1 µm (0.9 µm). -- Sect. 2.4: The flow pattern reduces the impact velocity of small dust particles onto the proto-planetesimal, and this ef- fect can be parameterized with Eq. 28. The reduced impact velocity decreases the erosion efficiency, and could result in enhanced sticking. -- Sect. 3.1: A flow pattern of parallel stream lines in free molecular flow leads to enhanced reaccretion, in agreement with ST03. However, because the random thermal motion of gas molecules in the disk will dominate over the headwind, parallel streamlines are not physically justifiable for a proto- planetesimal with radius comparable to the mean free path of the gas. -- Sect. 3.2: A laminar flow pattern around a spherical proto- planetesimal does not result in an enhanced reaccretion rate, in agreement with ST03. In the cases studied here, col- lisional debris of different stopping times is either suffi- ciently deflected by the flow to be advected past the proto- planetesimal, or moves beyond the effective cross section be- fore it can be reaccreted. The adopted flow pattern, Eq. 1, is strictly valid only for Re (cid:28) 1, but should provide accurate quantitative results for the sweep-up of dust particles by a spherical proto-planetesimal if Re (cid:46) 2 (Sect. 2.5). A relatively simple flow pattern was chosen here to provide a straightforward approximation of the effects of laminar hydrodynamic flow on the accretion of dust particles. Of course, the dependence of the flow pattern on the Reynolds num- ber influences the results, and while we have explored three an- alytical approximations (Stokes, Oseen, and Proudman & Pear- son), only full hydrodynamical simulations of the Navier-Stokes equations over the broad parameter space presented here can definitively determine the effects. These simulations will be the subject of future work. Additional and obvious refinements to our model would en- compass surface irregularities, velocity shear of the disk gas, and a non-spherical or rotating proto-planetesimal. Although a non-spherical, or rotating proto-planetesimal, or a shear ve- locity could produce significant differences in the hydrodynami- cal flow pattern (e.g. Kurose & Komori 1999; Ormel 2013), we expect that surface irregularities of planetesimals will introduce only minor differences. For proto-planetesimals of size R ∼ 100 m at 5 AU, our results predict that hydrodynamical deflection and deceleration will considerably reduce the effect of monomeric erosion (Schrä- pler & Blum 2011). Meanwhile, the detailed consequences of including the reduced sweep-up cross section and impact veloc- ities into a dust growth code are not easily predicted. The rela- tively low Reynolds numbers (≤ 22) used in this study prevent us from generalising to larger proto-planetesimal radii at low he- liocentric distances, and therefore to larger dust particle stopping times. Even when considering the results of ST03 (Re = 50 and cut-off point x ∼ 3), the high Reynolds numbers at low r in the Desch & MMSN disk models (> 3000; Fig. 5) make it difficult to predict what effect a ∼ 100 m size body might have on the deflection and deceleration of dust particles at r ∼ 1 AU. This further underscores the need for direct numerical simulation of dust particle sweep-up at high Reynolds numbers. This study has sampled a broad parameter space and demon- strated that a hydrodynamical gas flow can significantly affect the ability of a proto-planetesimal to sweep-up smaller particles. To further quantify these effects, Eq. (25) for the effective sweep- up cross section and Eq. (28) for the impact velocities must be implemented into a dust coagulation code which treats sweep- up growth. In this regard, the appropriate value of D(H) to use in coagulation models is the asymptotic limit of 4 (i.e. the dust particles are already well-coupled to the gas flow). Acknowledgements. We thank the anonymous referee for a careful and thorough review which improved the manuscript. JPR is supported by DFG grant DU 414/9-1. FW is funded by the DFG within Forschergruppe 759 "The Formation of Planets: The Critical First Growth Phase". References Alexander, R. D., Clarke, C. J., & Pringle, J. E. 2004, MNRAS, 354, 71 Birnstiel, T., Dullemond, C. P., & Brauer, F. 2010, A&A, 513, A79 Blum, J. & Wurm, G. 2008, ARA&A, 46, 21 Brauer, F., Dullemond, C. P., & Henning, T. 2008, A&A, 480, 859 Carry, B. 2012, Planet. Space Sci., 73, 98 Desch, S. J. 2007, ApJ, 671, 878 Geretshauser, R. J., Speith, R., & Kley, W. 2011, A&A, 536, A104 Greiner, W. & Stock, H. 1991, Hydrodynamik (Verlag Harri Deutsch, Frankfurt am Main) 513, A56 Güttler, C., Blum, J., Zsom, A., Ormel, C. W., & Dullemond, C. P. 2010, A&A, Hayashi, C., Nakazawa, K., & Nakagawa, Y. 1985, in Protostars and Planets II, ed. D. C. Black & M. S. Matthews, 1100 -- 1153 Kothe, S., Blum, J., Weidling, R., & Güttler, C. 2013, astro-ph:1302.5532 Kurose, R. & Komori, S. 1999, Journal of Fluid Mechanics, 384, 183 Lambrechts, M. & Johansen, A. 2012, A&A, 544, A32 Mitra, D., Wettlaufer, J. S., & Brandenburg, A. 2013, ApJ, 773, 120 Morbidelli, A. & Nesvorny, D. 2012, A&A, 546, A18 Nakagawa, Y., Sekiya, M., & Hayashi, C. 1986, Icarus, 67, 375 Okuzumi, S., Tanaka, H., Kobayashi, H., & Wada, K. 2012, ApJ, 752, 106 Okuzumi, S., Tanaka, H., & Sakagami, M.-a. 2009, ApJ, 707, 1247 Ormel, C. W. 2013, MNRAS, 428, 3526 Ormel, C. W. & Cuzzi, J. N. 2007, A&A, 466, 413 Proudman, I. & Pearson, J. R. A. 1957, Journal of Fluid Mechanics, 2, 237 Schräpler, R. & Blum, J. 2011, ApJ, 734, 108 Sekiya, M. & Takeda, H. 2003, Earth, Planets, and Space, 55, 263 Sekiya, M. & Takeda, H. 2005, Icarus, 176, 220 Taneda, S. 1956, Journal of the Physical Society of Japan, 11, 1104 Van Dyke, M. 1964, Perturbation methods in fluid mechanics (Academic Press, New York) Van Dyke, M. 1982, An Album of Fluid Motion (Parabolic Press, Stanford) Weidenschilling, S. J. 1977a, MNRAS, 180, 57 Weidenschilling, S. J. 1977b, A&SS, 51, 153 Weidenschilling, S. J. 1997, Icarus, 127, 290 Weidenschilling, S. J., Spaute, D., Davis, D. R., Marzari, F., & Ohtsuki, K. 1997, Windmark, F., Birnstiel, T., Güttler, C., et al. 2012a, A&A, 540, A73 Windmark, F., Birnstiel, T., Ormel, C. W., & Dullemond, C. P. 2012b, A&A, Icarus, 128, 429 544, L16 Wurm, G. & Blum, J. 1998, Icarus, 132, 125 Wurm, G., Blum, J., & Colwell, J. E. 2001, Icarus, 151, 318 Wurm, G., Paraskov, G., & Krauss, O. 2004, ApJ, 606, 983 Wurm, G., Paraskov, G., & Krauss, O. 2005, Icarus, 178, 253 Xie, J.-W., Payne, M. J., Thébault, P., Zhou, J.-L., & Ge, J. 2010, ApJ, 724, 1153 Zsom, A., Ormel, C. W., Güttler, C., Blum, J., & Dullemond, C. P. 2010, A&A, 513, A57 Article number, page 9 of 9
1103.0035	1	1103	2011-02-28T21:57:18	The GROUSE project II: Detection of the Ks-band secondary eclipse of exoplanet HAT-P-1b	[ "astro-ph.EP" ]	Context: Only recently it has become possible to measure the thermal emission from hot-Jupiters at near-Infrared wavelengths using ground-based telescopes, by secondary eclipse observations. This allows the planet flux to be probed around the peak of its spectral energy distribution, which is vital for the understanding of its energy budget. Aims: The aim of the reported work is to measure the eclipse depth of the planet HAT-P-1b at 2.2micron. This planet is an interesting case, since the amount of stellar irradiation it receives falls in between that of the two best studied systems (HD209458 and HD189733), and it has been suggested to have a weak thermal inversion layer. Methods: We have used the LIRIS instrument on the William Herschel Telescope (WHT) to observe the secondary eclipse of HATP-1b in the Ks-band, as part of our Ground-based secondary eclipse (GROUSE) project. The observations were done in staring mode, while significantly defocusing the telescope to avoid saturation on the K=8.4 star. With an average cadence of 2.5 seconds, we collected 6520 frames during one night. Results: The eclipse is detected at the 4sigma level, the measured depth being 0.109+/-0.025%. The uncertainties are dominated by residual systematic effects, as estimated from different reduction/analysis procedures. The measured depth corresponds to a brightness temperature of 2136+150-170K. This brightness temperature is significantly higher than those derived from longer wavelengths, making it difficult to fit all available data points with a plausible atmospheric model. However, it may be that we underestimate the true uncertainties of our measurements, since it is notoriously difficult to assign precise statistical significance to a result when systematic effects are important.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. demooijhatp1 April 2, 2018 c(cid:13) ESO 2018 1 1 0 2 b e F 8 2 . ] P E h p - o r t s a [ 1 v 5 3 0 0 . 3 0 1 1 : v i X r a The GROUSE project II: Detection of the Ks-band secondary eclipse of exoplanet HAT-P-1b⋆ E.J.W. de Mooij1, R.J. de Kok2 S.V. Nefs1 and I.A.G. Snellen1 1 Leiden Observatory, Leiden University, Postbus 9513, 2300 RA, Leiden, The Netherlands; e-mail: [email protected] 2 SRON Netherlands Institute for Space Research, Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands; ABSTRACT Context. Only recently it has become possible to measure the thermal emission from hot-Jupiters at near-Infrared wavelengths using ground-based telescopes, by secondary eclipse observations. This allows the planet flux to be probed around the peak of its spectral energy distribution, which is vital for the understanding of its energy budget. Aims.The aim of the reported work is to measure the eclipse depth of the planet HAT-P-1b at 2.2µm. This planet is an interesting case, since the amount of stellar irradiation it receives falls in between that of the two best studied systems (HD209458 and HD189733), and it has been suggested to have a weak thermal inversion layer. Methods. We have used the LIRIS instrument on the William Herschel Telescope (WHT) to observe the secondary eclipse of HAT- P-1b in the Ks-band, as part of our Ground-based secondary eclipse (GROUSE) project. The observations were done in staring mode, while significantly defocusing the telescope to avoid saturation on the K=8.4 star. With an average cadence of 2.5 seconds, we collected 6520 frames during one night. Results. The eclipse is detected at the 4-σ level, the measured depth being 0.109±0.025%. The uncertainties are dominated by residual systematic effects, as estimated from different reduction/analysis procedures. The measured depth corresponds to a brightness temperature of 2136+150 −170K. This brightness temperature is significantly higher than those derived from longer wavelengths, making it difficult to fit all available data points with a plausible atmospheric model. However, it may be that we underestimate the true uncertainties of our measurements, since it is notoriously difficult to assign precise statistical significance to a result when systematic effects are important. Key words. techniques: photometric -- stars: individual: HAT-P-1 -- planetary systems 1. Introduction Measurements of the secondary eclipse of an exoplanet, the mo- ment it passes behind its host star, allow us to probe the prop- erties of the atmosphere on the day-side of the planet. The first successful secondary eclipse measurements have been obtained with the Spitzer Space Telescope (Charbonneau et al. 2005; Deming et al. 2005), followed by many more secondary eclipse measurements from 3.6µm to 24µm (e.g. Knutson et al. 2008; Machalek et al. 2008), see also the review by Deming (2009). These Spitzer observations indicate that some planets ex- hibit a thermal inversion in their atmospheres (e.g. Knutson et al. 2008). This inversion is apparent when molecular bands in the infrared change from absorption to emission, resulting in a different shape of the planet's spectral energy distribution. Fortney et al. (2008) and Burrows et al. (2008) have proposed that such an inversion layer could be due to the presence of a strong optical absorber high in the planet's atmosphere, de- pending on the amount of stellar radiation the planet receives. Only for the planets receiving very high levels of irradiation, the hypothetical absorbing compound can stay in the gas-phase high up in the atmosphere, absorbing the stellar light very ef- ficiently and causing the thermal inversion. As a possible ab- sorber, Hubeny et al. (2003) suggested TiO and VO, although ⋆ Photometric time-series are only available in electronic form at the CDS via anonymous ftp to cdsarc.u-strasbg.fr (130.79.128.5) or via http://cdsweb.u-strasbg.fr/cgi-bin/qcat?J/A+A/ recent work (Spiegel et al. 2009) show that it might be difficult to keep these molecules in the gas phase. At lower levels of stel- lar irradiation, the compound possibly condenses out and is sub- sequently removed from the higher layers of the atmosphere. Fortney et al. (2008) dub these two classes pM and pL respec- tively, in analogy with L and M stellar dwarfs. With the increase in the number of exoplanets studied with the Spitzer Space Telescope, it became clear that the above scheme cannot be solely dependent on the level of stellar ir- radiation. There are planets apparently exhibiting an inversion layer which receive less light from their host-star than required to keep the proposed absorber in the gas-phase (e.g. XO-1b, Machalek et al. 2008), while there are also planets that receive very high levels of irradiation, which do not apparently ex- hibit an inversion layer (e.g. TrES-3b, Fressin et al. 2010). Recently, Knutson et al. (2010), showed that there appears to be a trend between the presence of an inversion layer and the stellar activity (as determined from the calcium lines), with only plan- ets around stars with low activity exhibiting an inversion layer. Madhusudhan & Seager (2010) showed that the inference of a thermal inversion is not robust, and can also depend on the chem- ical composition of the atmosphere. Recently, several groups have presented measurements for a number of exoplanets in the near-infrared using ground-based telescopes (de Mooij & Snellen 2009; Sing & L´opez-Morales 2009; Gillon et al. 2009; Rogers et al. 2009; Anderson et al. 2010; Alonso et al. 2010; Gibson et al. 2010; Croll et al. 2010a; 1 De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b L´opez-Morales et al. 2010; Croll et al. 2010b, 2011). Data in this wavelength region are very interesting because they probe the thermal emission at, or close to, the peak of the spectral en- ergy distribution of hot-Jupiters, which is vital for the under- standing of their energy budgets. Here we present the second result from the GROUnd-based Secondary Eclipse project (GROUSE), which aims to use tele- scopes at a variety of observatories for exoplanet secondary eclipse observations. As part of this survey we have already published our detection of the secondary eclipse of TrES- 3b (de Mooij & Snellen 2009), which we retrospectively include as paper I. In this paper we present observations of the secondary eclipse of the hot-Jupiter HAT-P-1b (Bakos et al. 2007) in Ks- band as obtained with the Long-slit Intermediate Resolution Infrared Spectrograph (LIRIS; Acosta-Pulido et al. 2002) in- strument on the William Herschel Telescope (WHT). HAT-P- 1b, with a mass of 0.52M jup and a radius of 1.2R jup, orbits its G0V stellar host with a period of P=4.5 days at a dis- tance of 0.055 AU. This large orbital separation places HAT- P-1 on the lower edge of the proposed pM/pL transition bound- ary of Fortney et al. (2008). Recent observations of this planet by Todorov et al. (2010), who used the IRAC instrument on the Spitzer Space Telescope to determine the infrared brightness of HAT-P-1, showed evidence for a weak inversion layer. Its host star appears not to be very active (Knutson et al. 2010), which provides an alternative explanation for the planet's inversion layer, in the case that stellar variability is one of the driving fac- tors for the structure of the atmosphere. In Sec. 2 we present our observations and data-reduction. In Sec. 3 we will present and discuss the results, and in Sec. 4 we will give our conclusions. Fig. 1. The flux in the highest pixel of the PSF of the brightest star as a function of time. occurs. The stars also slowly drifted by 2 pixels in declination during the night. The exposure times were varied between 1.3 and 2 seconds, in order to keep the flux-levels of the star in the linear regime of the detector. To allow these relatively long exposure times for these rather bright stars, we strongly defocused the telescope. This resulted in a donut-shaped PSF with a diameter of 15 pix- els (3.75 arcsec). Despite the defocus, the PSF was still small enough to avoid overlap between the target and the reference star (which are separated by ∼11.2 arcsec), although the diffrac- tion spikes originating from the support of the secondary mirror did overlap for certain periods (see section 2.4). 2. Observations, data reduction and analysis 2.1. Crosstalk,non-linearitycorrectionsandflat-fielding The secondary eclipse of HAT-P-1 was observed on October 2, 2009 using the imaging mode of the Long-slit Intermediate Resolution Infrared Spectrograph (LIRIS; Acosta-Pulido et al. 2002) instrument on the William Herschel Telescope (WHT) on La Palma. The expected time of mid-eclipse was 21h:55mUT. The observations started at 19h:59mUT and lasted for 6.5 hours, although ∼15 minutes were lost because the observer got locked up in the bathroom due to a broken door handle. Since this caused a gap in the photometry, we excluded the final 630 frames from further analysis. The exposures were taken in sequences of 50 or 100 frames. The first two frames of every sequence are known to suffer from a reset-anomaly1, visible by a very strong gradient in the background. We excluded the frames affected by the anomaly from further analysis. The LIRIS detector was windowed to one sub-array of 512 by 512 pixels in order to increase the observing cadence. The pixel-scale of LIRIS is 0.25 arcsec per pixel, hence the field of view for these windowed observations was 128 by 128 arcsec- onds. Since HAT-P-1b has a slightly brighter companion star nearby, the field of view was enough to allow us to observe both HAT-P-1 (K=8.86) and the reference star, ADS 16402 A (K=8.41), simultaneously. The observations were performed in staring mode, keeping the stars as much as possible at the same position on the chip. Despite guiding, a slow drift of 3 pixels over 4 hours is visible, and subsequently a sudden jump of 5 pixels in right ascension 1 http://www.ing.iac.es/astronomy/instruments/liris/liris cookbook.pdf As a first step in the data reduction we removed the intra- quadrant crosstalk, which is present as an additional flux in each pixel at the level of 10−5 of the total flux along its row. We subse- quently performed a non-linearity correction, for which we ob- tained a new set of measurements in August 2010. A plot of the value of the highest pixel in the PSF of the brightest star is shown in Fig. 1. For the non-linearity correction we observed several sequences of Ks-band dome-flats with varying exposure times to which we subsequently fitted a second order polynomial to the non-linearity for each pixel of the detector. Our current re- sults show that the detector is non-linear for all fluxes, reaching ∼8.5% at 30000 ADU, above this flux level the non-linearity in- creases rapidly. We used this non-linearity solution to correct all our frames. We excluded all frames with the peak flux of the brightest star above 30000 ADU. All the frames were flatfielded using a flatfield created from a set of dome flats, both with the dome-lights on and off to remove both the effects of emission from the telescope structure, as well as structure due to the dark-current from the detector. The flat- field images were corrected for the intra-quadrant crosstalk and non-linearity in the same way as the science images. 2.2. Removalofbad-pixels During most of the observations, one hot-pixel is located within the PSF of the target. This hot-pixel was corrected for by replac- ing it with the flux of the corresponding pixel in the reference star, scaled by the flux ratio of the stars. Since the position dif- 2 De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b Fig. 2. The variation of the sky background during the night. ference between the two stars is not an integer, and the precise flux in inner edge of the donut is a strong function of position, we interpolated the PSF of the reference star to the same grid as HAT-P-1. before measuring the flux in the corresponding pixel. 2.3. Backgroundsubtraction The background level during the observations varied, and the variation of the background per pixel is shown in Fig. 2. A back- ground map was constructed from a separate set of observations of a different field taken on the same night. After subtracting this background map from the images, a small gradient was still ap- parent along the y-axis, which we believe is due to different ex- posure times (and dark currents) for the images used for the cre- ation of the background map and the science images. To remove this gradient, we first masked out all the stars and bad-pixels and subsequently determined the mean along a row of the detector, rejecting outliers deviating by more than 3 times the standard de- viation from the median of the row. The resultant profile along the y-axis was then fitted with a polynomial, in order to create a smooth background map, which was subsequently subtracted from all the columns of the image. The overall impact of the subtraction of a background image on the results is small, but not subtracting the background image results in an increased noise-level. 2.4. Diffractionspokesfromthesecondarymirrorsupport The close proximity of the reference star causes the diffraction pattern from the support structure for the secondary mirror to overlap with the other star's PSF. Since the WHT is on an alt-az mount, the instrument is rotated to maintain the same position angle on the sky. This causes the diffraction spikes at certain times to overlap, depending on the position-angle of the mount. In Fig. 3 two images are shown, on the left side with a rotator angle of -79.2 degrees, with the diffraction spikes not overlap- ing, while in the right side the rotator angle was +55 degrees, where the diffraction spikes do overlap. To account for this, the amount of flux in a region offset by the same amount as the sep- aration between the two stars was measured, and subsequently subtracted from the measured flux values of the stars, after ac- counting for the difference in flux levels. Fig. 3. Sample of PSFs for our observations at high and low count-levels for different mount rotator angles. Left panels: mount rotator angle of -79.2 degrees. Right panels: +55.0 de- grees. The spokes caused by the support structure of the sec- ondary mirror can be clearly seen to rotate due to the alt-az mounting of the telescope. Fig. 4. Data points show the uncorrected lightcurve of HAT-P- 1b with the WHT on La Palma. The solid line indicates the best fitting model where we fit simultaneous for both the systematic effects as well as the eclipse depth. 2.5. AperturePhotometry Subsequently, we performed aperture photometry on the two stars, with a radius of 14 pixels (3.5 arcsec), chosen to maximise the flux, but minimise the influence of the background. Any residual background was determined by measuring the mean flux in two boxes located 60 pixels (15 arcsec) above and below the midpoint between the two stars, with a box-size of is 131 by 41 pixels (32.75 by 10.25 arcsec). In these boxes, we excluded bad- pixels from the background-determination, as well as masked the areas that can be affected by the diffraction spikes from the secondary mirror support. 3 De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b Fig. 5. Different components fitted to the lightcurve due to the eclipse and systematic effects. Shown are, from top to bottom the eclipse model, the airmass, the exposure time and the x position on the detector. The different parameters are offset for clarity. . Fig. 6. The lightcurve for the secondary eclipse of HAT-P-1b corrected for systematic effects. Overplotted is our best fitting eclipse model. Table 1. Parameters for HAT-P-1. References: (1) Winn et al. (2007); this work; (4) Todorov et al. (2010) Johnson et al. (2) (2008); (3) parameter Value Reference Semi-major axis a (au) R∗ (R⊙) Rp (R jup) i (deg) P (days) b Rp/R∗ Eclipse depth at 2.2µm Eclipse depth at 3.6µm Eclipse depth at 4.5µm Eclipse depth at 5.8µm Eclipse depth at 8.0µm 0.0551±0.0015 1.115±0.043 1.204±0.051 86.22±0.24 4.4652934±0.00072 0.701±0.023 0.11295±0.00073 0.109±0.025% 0.080±0.008% 0.135±0.022% 0.203±0.031% 0.238±0.040% (1) (1) (1) (1) (2) (1) (2) (3) (4) (4) (4) (4) Fig. 7. The same as Fig. 6, but binned by 41 points. 2.6. Correctionforsystematiceffects The lightcurve of HAT-P-1b, corrected for the hot-pixel, diffrac- tion spikes, and normalised with that of the reference star, is shown in Fig. 4. Although the eclipse is already visible, the lightcurve is still clearly influenced by systematic effects, which we found to be correlated with the x-position on the detector, the exposure time and the geometric airmass. We fitted for these effects simultaneously with the eclipse depth. Other parameters, such as the background level, were tried, but resulted in no dif- ference in the final lightcurve. For the model of the secondary eclipse, we used the formalism from Mandel & Agol (2002), where all the orbital parameters were kept fixed to the litera- ture values (given in Table 1), with the limbdarkening set to 0, and the timing of the secondary eclipse fixed. We did not fit for a time offset, because Todorov et al. (2010) already showed that the timing of the secondary eclipses measured using the Spitzer Space Telescope shows no indication of any offset, using higher signal-to-noise data. We performed the fit both using the standard multi-linear re- gression algorithm in IDL (regress.pro) as well as using a Monte- Carlo Markov Chain (MCMC) analysis. After an initial fit, we performed a clipping of the residuals by excluding all points within bins of 38 pixels, for which the binned value deviate by more that 1.5·10−3. In this way we removed 228 unbinned points. After this clipping we performed an MCMC analysis. For the MCMC analysis, we created 5 chains with a length of 2 million steps each, using different starting positions. We trimmed away the first 200.000 points of each chain, to make sure that the initial starting point no longer influence the measured parameters. The best fitting model, overplotted on the data, is shown in Fig. 4, while the contributions for the individual components are shown in Fig. 5. 3. Results The final, corrected lightcurve is shown in Fig. 6, with its binned (by 41 points) version in Fig. 7. Assuming pure, white, Gaussian noise, we find an eclipse depth of 0.1089±0.01 % (11- σ). However, clear residual systematic effects are visible in the binned lightcurve. The level of this red noise was estimated by comparing the standard deviation in the unbinned data with that of gradually increased binned lightcurves. From this we es- timate a red noise component at the level of 0.023%, which we added in quadrature to our eclipse depth uncertainty to be- come 0.109±0.025 %. We also performed the residual permua- tion ("Prayer Bead") method (e.g. Gillon et al. 2007), where we add the best fit model to the residuals, after shifting the resid- 4 De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b 3.1. Atmosphericmodels We modelled the thermal emission spectra for a few ad hoc cases to try to understand what type of atmosphere would be needed to fit the measured eclipse depths, including our new measure- ment in Ks-band. We used a radiative transfer model that can cal- culate multiple scattering of the thermal radiation, since we in- cluded a case with scattering clouds (see below). The scattering is calculated using the doubling-adding method, which makes use of the fact that one can calculate the reflection and trans- mission properties of a combination of two atmospheric layers from the properties of the individual layers. For each layer in the model atmosphere we started with an optically thin layer with the same scattering properties as the model layer. For the opti- cal thin case analytical expressions are available to calculate the layer transmission and reflection. The thin layer was then dou- bled several times to match the real optical depth of the model layer. Subsequently, the different model layers were added to calculate the spectra for the entire inhomogeneous atmosphere. For more explanations we refer to Wauben et al. (1994), who describe the model in great detail. Spectra were calculated at 10 different emission angles, from which a disc-averaged spectrum was calculated. We included absorption of water and CO, whose ab- sorption properties were taken from the (old) HITEMP database (Rothman et al. 1995). Although the water data in this database is not the most up-to-date, a line-by-line comparison in K-band with the new HITEMP database (Rothman et al. 2010) yielded only differences many times smaller than the errors on the measured eclipse depths, which we neglect for this qualita- tive assessment. For speedy calculations, we pre-computed ab- sorption properties at a range of temperatures and pressure using the correlated-k method (Lacis & Oinas 1991) with a spectral resolution of 0.01 µm between 1.8 µm and 5 µm and 0.1 µm be- tween 5 µm and 10µm, and a Voigt line shape. Assumed mixing ratios of water and CO were 2e-4 and 5e-4 respectively for all cases. The models extend from 1.8µm to 10µm, and were made for different temperature-pressure (T-P) profiles and composi- tions. We have made models for the T-P profile for this planet from Fortney et al. (2008), which does not include an inversion layer, as well as for models with an inversion layer. We also var- ied the composition of the atmosphere and made a model with Venus-like clouds. In Fig. 8 we show the temperature-pressure profiles for the different models, and in Fig. 9 we show the ex- pected eclipse depths within the observed bands for the different models, binned to a resolution of 0.1µm. Fig. 9 shows that an atmosphere calculated from first prin- ciples (Fortney et al. 2008) cannot reproduce the Spitzer mea- surements, and that a weak inversion is needed (Todorov et al. 2010). However, a weak inversion throughout the atmosphere produces a very low planet signal in Ks-band, unlike what we derive from our measurements. The reason for this is that Ks- band is a region of very low absorption and hence we see deep in the planet atmosphere, where temperatures are relatively low in the inversion-only case (green lines). Contribution functions for the different wavelengths are shown in Fig. 10. To increase the eclipse depth in Ks-band the lower atmosphere needs to be made hotter. However, this also increases the eclipse depth in L-band, another spectral region of low absorption (red cureve in Fig. 9). L-band emission can be suppressed somewhat by in- cluding large amounts of CO2 and CH4 (calculated here from HITEMP and HITRAN 2008 (Rothman et al. 2009) data respec- tively), but this also reduces the signal in Ks-band (blue curve in 5 Fig. 8. The temperature-pressure (T-P) profile for the our at- mospheric models used for modelling the secondary eclipse depth. The black (solid) line shows the uninverted model from Fortney et al. (2008), the green (dashed-triple dotted) line is for a model with an ad-hoc temperature inversion as in Todorov et al. (2010), the red (dashed) line shows the tem- perature pressure profile for an atmosphere with a normal tropo- sphere and an inversion, and the orange (dash-dotted) line is for an atmosphere with Venus-like clouds. The temperatures for the latter profiles are chosen to provide a reasonable fit to the IRAC bands longward of 4µm uals by N-points. Points are wrapped around when performing the shift. By refitting the data for all possible shifts between 0 and the number of data points, and measuring the best-fit eclipse depth, we can get an estimate for the true uncertainties in the data. We find a depth of 0.1089+0.024 −0.025 %, which is fully consis- tent with the uncertainties estimated from the red-noise. This results overall in an eclipse depth of 0.109±0.025 %, which corresponds to a brightness temperature in the Ks-band of Tb=2136+150 −170K. This brightness temperature is significantly higher than the expected equilibrium temperature of this planet, which lies between Teq=1540 K for an albedo of 0 and inefficient energy redistribution from the dayside to the nightside (Pn=0), and Teq=1190 K for an albedo of 0.3 and efficient redistribu- tion of energy from the dayside to the nightside (Pn=0.5). The higher brightness temperature could be due to the fact that in the Ks-band we are looking deeper into the atmosphere, where the layers are warmer. Todorov et al. (2010) observed the secondary eclipse of HAT-P-1b with the IRAC instrument on board the Spitzer Space Telescope, at 3.6, 4.5, 5.8 and 8.0 µm, and find eclipse depths of 0.080±0.008%, 0.135±0.022%, 0.203±0.031% and 0.238±0.040% in the different bands respectively. Combining our own measurement of the secondary eclipse in Ks-band with the IRAC measurements, we can construct a spectral energy distribution (SED) from the near-infrared to the mid-infrared, which is shown in Fig. 9, and compare it to atmospheric models. De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b Fig. 9. Models for the emission from the atmosphere of HAT-P-1b, for the T-P profiles of Fig. 8, using the same colour scheme. The diamonds indicate the expected flux within the observed bands for the different models. The points with errorbars indicate the observational data. The dotted lines indicate the transmission curves of the different bands. In addition, a blue (long dashed) curve is shown for an atmosphere with the same T-P profile as the red (short dashed) curve, but with a high concentration of methane and carbon-dioxide (1·10−4 and 2·10−5 respectively). Fig. 10. Normalised contribution functions for the atmospheric emission of HAT-P-1b in the Ks-band and the different IRAC chan- nels, using the same colour scheme as in Figs. 8 and 9, although the different linestyles are now used for different bands. The distribution functions are normalised by their sum. The left panel shows the contribution functions for the non-inverted model from Fortney et al. (2008), the middle panel shows the contribution functions for the models with both a troposphere and an inver- sion layer, and the right panel shows the same for the ad-hoc temperature inversion used in Todorov et al. (2010) 6 De Mooij et al.: The GROUSE project II: The secondary eclipse of HAT-P-1b Fig. 9). Reducing the water or CO abundances also do not pro- duce a better fit. Hence, we cannot fit all data points well using any clear atmosphere with gases that are predicted to be abun- dant on hot exoplanets. One remedy to fit both the Spitzer data and our Ks-band eclipse depth is to include a Venus-like cloud. The clouds on Venus are made of concentrated sulphuric acid droplets, which have the property that they are strongly scatter- ing below wavelengths of ∼3µm and strongly absorbing above 3 µm, with only little variation of extinction with wavelength (e.g. Grinspoon et al. 1993). Because of the scattering nature of the clouds, radiation from Venus' hot lower atmosphere still reaches space at the night side below 3µm, despite an optically thick cloud layer surrounding the planet. Above 3 µm the night side emission originates from the clouds. To show this potential effect for HAT-P-1b, we inserted a cloud layer with an optical depth of 1.5 at the tropopause with properties identical to those of Venus' 1-µm sized cloud-particles. Indeed, brightness temperatures at Ks-band are higher than anywhere else in Fig. 9 and all data points could potentially be fitted well if the lower atmosphere is made even hotter. However, we could not find a physically plausible candidate for cloud materials that could mimic Venus' clouds on a hot exoplanet. So, at present we do not find any suitable atmosphere that could explain the high Ks-band eclipse depth. 4. Conclusion Using the LIRIS infrared camera on the WHT, we determine the eclipse depth of the extrasolar planet HAT-P-1b in Ks-band to be 0.109±0.025% (∼4σ), with the uncertainties in the eclipse depth dominated by residual systematic effects, as estimated from dif- ferent reduction/analysis procedures. The measured depth cor- responds to a brightness temperature of 2136+150 −170K. This bright- ness temperature is significantly higher than those derived from longer wavelengths, making it difficult to fit all available data points with a plausible atmospheric model. It may be that we underestimate the true uncertainties of our measurements, since it is notoriously difficult to assign precise statistical significance to a result when systematic effects are important. Acknowledgements. We are grateful to the staff WHT telescope for there assis- tance with these observations. The William Herschell Telescope is operated on the island of La Palma by the Isaac Newton Group in the Spanish Observatorio del Roque de los Muchachos of the Instituto de Astrof´ısica de Canarias. References Acosta-Pulido, J., Ballesteros, E., Barreto, M., et al. 2002, The Newsletter of the Isaac Newton Group of Telescopes, 6, 22 Alonso, R., Deeg, H. J., Kabath, P., & Rabus, M. 2010, AJ, 139, 1481 Anderson, D. R., Gillon, M., Maxted, P. F. L., et al. 2010, A&A, 513, L3+ Bakos, G. ´A., Noyes, R. W., Kov´acs, G., et al. 2007, ApJ, 656, 552 Burrows, A., Budaj, J., & Hubeny, I. 2008, ApJ, 678, 1436 Charbonneau, D., Allen, L. E., Megeath, S. T., et al. 2005, ApJ, 626, 523 Croll, B., Albert, L., Lafreniere, D., Jayawardhana, R., & Fortney, J. J. 2010a, ApJ, 717, 1084 Croll, B., Jayawardhana, R., Fortney, J. J., Lafreni`ere, D., & Albert, L. 2010b, ApJ, 718, 920 Croll, B., Lafreniere, D., Albert, L., et al. 2011, AJ, 141, 30 de Mooij, E. J. W. & Snellen, I. A. G. 2009, A&A, 493, L35 Deming, D. 2009, in IAU Symposium, Vol. 253, IAU Symposium, 197 -- 207 Deming, D., Seager, S., Richardson, L. J., & Harrington, J. 2005, Nature, 434, 740 Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, 1419 Fressin, F., Knutson, H. A., Charbonneau, D., et al. 2010, ApJ, 711, 374 Gibson, N. P., Aigrain, S., Pollacco, D. L., et al. 2010, MNRAS, 404, L114 Gillon, M., Demory, B., Barman, T., et al. 2007, A&A, 471, L51 Gillon, M., Demory, B., Triaud, A. H. M. J., et al. 2009, A&A, 506, 359 Grinspoon, D. H., Pollack, J. B., Sitton, B. R., et al. 1993, Planet. Space Sci., 41, 515 Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011 Johnson, J. A., Winn, J. N., Narita, N., et al. 2008, ApJ, 686, 649 Knutson, H. A., Charbonneau, D., Allen, L. E., Burrows, A., & Megeath, S. T. 2008, ApJ, 673, 526 Knutson, H. A., Howard, A. W., & Isaacson, H. 2010, ApJ, 720, 1569 Lacis, A. A. & Oinas, V. 1991, J. Geophys. Res., 96, 9027 L´opez-Morales, M., Coughlin, J. L., Sing, D. K., et al. 2010, ApJ, 716, L36 Machalek, P., McCullough, P. R., Burke, C. J., et al. 2008, ApJ, 684, 1427 Madhusudhan, N. & Seager, S. 2010, ApJ, 725, 261 Mandel, K. & Agol, E. 2002, ApJ, 580, L171 Rogers, J. C., Apai, D., L´opez-Morales, M., Sing, D. K., & Burrows, A. 2009, ApJ, 707, 1707 Rothman, L. S., Gordon, I. E., Barbe, A., et al. 2009, J. Quant. Spec. Radiat. Transf., 110, 533 Rothman, L. S., Gordon, I. E., Barber, R. J., et al. 2010, J. Quant. Spec. Radiat. Transf., 111, 2139 Rothman, L. S., Wattson, R. B., Gamache, R., Schroeder, J. W., & McCann, A. 1995, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 2471, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, ed. J. C. Dainty, 105 -- 111 Sing, D. K. & L´opez-Morales, M. 2009, A&A, 493, L31 Spiegel, D. S., Silverio, K., & Burrows, A. 2009, ApJ, 699, 1487 Todorov, K., Deming, D., Harrington, J., et al. 2010, ApJ, 708, 498 Wauben, W. M. F., de Haan, J. F., & Hovenier, J. W. 1994, A&A, 282, 277 Winn, J. N., Holman, M. J., Bakos, G. ´A., et al. 2007, AJ, 134, 1707 7
1801.04652	1	1801	2018-01-15T03:15:16	Production and fate of the G ring arc particles due to Aegaeon (Saturn LIII)	[ "astro-ph.EP" ]	The G ring arc hosts the smallest satellite of Saturn, Aegaeon, observed with a set of images sent by Cassini spacecraft. Along with Aegaeon, the arc particles are trapped in a 7:6 corotation eccentric resonance with the satellite Mimas. Due to this resonance, both Aegaeon and the arc material are confined to within sixty degrees of corotating longitudes. The arc particles are dust grains which can have their orbital motions severely disturbed by the solar radiation force. Our numerical simulations showed that Aegaeon is responsible for depleting the arc dust population by removing them through collisions. The solar radiation force hastens these collisions by removing most of the 10$~\mu$m sized grains in less than 40 years. Some debris released from Aegaeon's surface by meteoroid impacts can populate the arc. However, it would take 30,000 years for Aegaeon to supply the observed amount of arc material, and so it is unlikely that Aegaeon alone is the source of dust in the arc.	astro-ph.EP	astro-ph	MNRAS 000, 1 -- 7 (0000) Preprint 11 October 2018 Compiled using MNRAS LATEX style file v3.0 Production and fate of the G ring arc particles due to Aegaeon (Saturn LIII) Gustavo Madeira⋆, R. Sfair, D.C. Mourao and S.M. Giuliatti Winter Univ. Estadual Paulista -UNESP, Grupo de Dinamica Orbital e Planetologia, Guaratinguet´a, CEP 12516-410, Brazil 8 1 0 2 n a J 5 1 . ] P E h p - o r t s a [ 1 v 2 5 6 4 0 . 1 0 8 1 : v i X r a ABSTRACT The G ring arc hosts the smallest satellite of Saturn, Aegaeon, observed with a set of images sent by Cassini spacecraft. Along with Aegaeon, the arc particles are trapped in a 7:6 corotation eccentric resonance with the satellite Mimas. Due to this resonance, both Aegaeon and the arc material are confined to within sixty degrees of corotating longitudes. The arc particles are dust grains which can have their orbital motions severely disturbed by the solar radiation force. Our numerical simulations showed that Aegaeon is responsible for depleting the arc dust population by removing them through collisions. The solar radiation force hastens these collisions by removing most of the 10 µm sized grains in less than 40 years. Some debris released from Aegaeon's surface by meteoroid impacts can populate the arc. However, it would take 30,000 years for Aegaeon to supply the observed amount of arc material, and so it is unlikely that Aegaeon alone is the source of dust in the arc. Key words: planets and satellites: rings 1 INTRODUCTION Before the discovery of the small satellite Aegaeon, a bright arc close to the inner edge of Saturn's G ring was imaged by the cameras onboard the Cassini spacecraft. Located at about 167500 km from Saturn's centre, this arc extends over ∼ 60◦ in longitude and has a radial width of 250 km, while the rest of the G ring is 6000 km wide (Hedman et al. 2007). Cassini data showed that most of the arc is populated by µm sized dust grains, although larger bodies (cm to meters in size) can also be present. Hedman et al. (2007) argued that a decrease in the flux of energetic electrons, observed by the Cassini instruments, could be caused by a population of cm- m sized bodies. They proposed that these large bodies could be the source of the arc and also the G ring. The mean motion of the arc is close to the 7:6 corota- tion eccentric resonance (CER) with Mimas (Hedman et al. 2007). The resonant argument φ = 7λM − 6λ − ϖM is equal to 180◦, where λM and λ are the mean longitudes of Mimas and the particle, respectively, and ϖM is the longitude of Mimas' pericenter. Hedman et al. (2007) numerically simu- lated a sample of particles initially located in this arc and verified that they stay confined for at least 80 years. Several Cassini images taken between 2007 and 2009 showed a small satellite, named Aegaeon, embedded in the G ring arc. With a diameter about 500 m, Ae- gaeon is trapped in the same 7:6 corotation eccentric res- ⋆ E-mail: [email protected] c(cid:13) 0000 The Authors onance with Mimas with a libration amplitude of about 10◦ (Hedman et al. 2010). In this work we analyze the influence of Aegaeon on the small particles located in the G ring arc, after their possi- ble ejection from the surface of Aegaeon, as well as their orbital evolution due to gravitational and dissipative forces. In section 2 we analyze the orbital evolution of a set of µm sized particles under the effects of the solar radiation force and the gravitational effects of the planet and the saturnian satellites, Mimas, Tethys and Aegaeon. In section 3 we ana- lyze the time evolution of those dust particles ejected from the surface of the small satellite. Section 4 compares these estimates of dust lifetimes with the production rate due to impacts onto Aegaeon's surface in order to analyze the role of the satellite on the maintenance of the arc population. Our results are discussed in the last section. 2 ORBITAL EVOLUTION OF THE G RING ARC PARTICLES First of all we analyze the gravitational effects of Mimas on the particles located in the G ring arc. The dynam- ical system is formed by Saturn, including the gravita- tional coefficients J2, J4 and J6, and the satellites Mimas, Aegaeon and Tethys. The numerical simulations were per- formed using the Mercury integrator package (Chambers 1999) with the Burlish Stoer algorithm. We also used the algorithm described in Renner & Sicardy (2006) to convert the state vector into the geometric orbital elements that ac- 2 Madeira et al. Table 1. Mass (m), density (d) and osculating elements of the satellites (2454700.5 JD): a is the semi-major axis, e is the ec- centricity, I is the inclination, and the angles ϖ, Ω and λ are the argument of pericentre, longitude of node and mean anomaly, re- spectively. These values were extract from JPL-Horizons System. Aegaeon Mimas Tethys a (×105km) e (×10−2) I(◦) ϖ (◦) Ω (◦) λ (◦) 1.6803396819 0.3121285727 0.0014761087 145.76280592 233.02211168 5.3594249045 1.8600466879 1.7245219209 1.5687571620 163.18023984 259.15258436 197.73278953 2.9497426488 0.0828116581 1.0915162973 15.906312542 355.42083194 4.5221318570 m (kg) d (g/cm3) 5.997 × 1010 0.500 3.754 × 1019 1.152 61.760 × 1019 0.956 Table 2. Physical parameters of Saturn. Radius (km) Mass (×1026kg) J2 (×10−6) J4 (×10−6) J6 (×10−6) 60330 5.68683765495 16290.543820 −936.700366 86.623065 ) g e d ( φ 210 200 190 180 170 160 with Tethys without Tethys 0 5 10 15 20 time (years) Figure 1. (Colour online) Resonant argument of Aegaeon as a function of time with (dashed line) and without the gravitational effects of Tethys. Tethys disturbs the resonant angle of Aegaeon. count for the orbital precession caused by the gravity co- efficients of Saturn. Table 1 shows the initial osculating el- ements (2454700.5 JD), mass and density of the satellites, while Table 2 presents the parameters of Saturn: radius (in km), mass (in kg) (Thomas et al. 2013), J2, J4 and J6 [con- sistent with Hedman et al. (2010)]. The resonant argument of Aegaeon as a function of time is showed in Figure 1. Gravitational effects of the satellite Tethys induce small variations in the resonant argument which can be seen in Figure 1 (dashed line). Although these small variations did not alter the lifetime of the particles, the effects of Tethys were added to the system just for com- pleteness. A sample of 6000 test particles was randomly dis- tributed, with uniform probability, in a resonant arc con- fined 60 km in radius and 60◦ azimuthally. The initial or- bital elements, e, I,ω and Ω, of the particles have the same values of the initial orbital elements of Aegaeon (Table 1). This sample of particles was used in all numerical simula- tions described in this section. When the distance between ) m k ( a 167535 167520 167505 167490 167475 167460 167445 0 0.6 1.2 1.8 2.4 3 3.6 time (years) Figure 2. (Colour online) This figure shows the geometric semi- major axis (in km) as a function of time (in years) of Aegaeon and 8 particles near the outer (full line) and inner (dashed line) edges of the arc. the particle and Aegaeon is less than the radius of the small satellite (r = 240 m, Hedman et al. 2010), a collision is de- tected. Our numerical simulations for a timespan of 500 years showed that the particles, due to Mimas 7:6 CER, are az- imuthally confined in the arc with an amplitude of 60◦. Fig- ure 2 presents the geometric semi-major axis as a function of time of Aegaeon and eight arc particles. These particles are initially displaced by 5, 15, 25 and 35 km from the resonant semi-major axis (167493.73 km). Although the particles dis- placed by 5-25 km from Aegaeon's semi-major axis show some variation in the semi-major axis, they remain located in the arc region. The 7:6 CER with Mimas has a width of about 60 km (El Moutamid et al. 2014), therefore those particles displaced by more than 30 km from the resonant semi-major axis are not azimuthally confined. Besides the gravitational effects of Mimas on the arc particles, the small satellite Aegaeon also causes small vari- ations in their orbital elements. Variations in the geometric eccentricity and inclination of the particles are of order 10−6, and these effects are not strong enough to remove these par- ticles from the resonance. Our numerical simulations showed that about 75% of the initial set of arc particles, under the gravitational effects of Saturn, Mimas and Aegaeon, collide with Aegaeon in 500 years. Thus, Aegaeon acts as a sink for these arc particles. The Cassini cameras observed a population of µm sized particles located in the G ring arc. These tiny particles, 1- 10µm in radius (r), can be strongly influenced by the effects of the solar radiation force. Considering that the planet has its heliocentric position vector as r sp (rsp = r sp) and veloc- ity as VP, the solar radiation force (SRF) experienced by a circumplanetary particle is (Mignard 1984) F = ΦA c Qpr(cid:26)(cid:20)1 − r sp rsp ·(cid:18) VP c + V c (cid:19)(cid:21) r sp rsp − V P + V c (cid:27) (1) where c is the speed of light and V is the velocity vector of the particle with respect to the planet. The particles' cross section is A, and we considered that they are made of ideal material which implies Qpr = 1. In our model the planet is in a circular orbit, hence rsp, the magnitude of VP and the solar flux Φ are constants. Furthermore, we assumed that the Sun lies in the equatorial plane of the planet (i.e. the obliquity of the planet was neglected), which is a simplification that does not change significantly the magnitude of the solar radiation MNRAS 000, 1 -- 7 (0000) Production and fate of small particles due to Aegaeon 3 y t i c i r t n e c c e 0.030 0.024 0.018 0.012 0.006 0.000 1µm 3µm 5µm 10µm 0 30 60 90 120 150 180 210 240 time (days) Figure 3. (Colour online) Time evolution of the geometric ec- centricity of particles with different sizes due to the radiation pressure component. As the size of the particle increases, its ∆e decreases. force and saves computational time. We also disregarded secondary and, at least an order of magnitude, weaker effects such as the planetary light reflection and shadow, and the Yarkovsky effect (Hamilton 1996). We do not include the effects of the plasma drag and the electromagnetic force. These forces are responsible to cause an outward drift of the particle and the precession of or- bital pericenter, respectively (Sun et al. 2015; Burns et al. 2001). As discussed in more detail in section 5, including these forces will probably further reduce the lifetimes of these particles. The solar radiation force equation was decomposed and its components were included in the Mercury package (Sfair & Giuliatti Winter 2009) in order to analyze the or- bital evolution of particles with sizes of 1, 3, 5 and 10µm in radius. These particles are also perturbed by the gravity of Mimas, Tethys and Aegaeon, and the gravity coefficients of Saturn. The radiation pressure component (RP, the term that does not depend on the velocities in Eq. (1)) mainly causes a variation in the eccentricity of the particles which can be seen in Figure 3. Each curve shows the time variation of the eccentricity of four particles, initially located at 10 km from the CER semi-major axis, with sizes of 1, 3, 5 and 10 µm in radius. The smaller particle (1 µm) has the larger variation in the eccentricity, from 0 to 10−2. The 10 µm sized particle has the smaller variation in the eccentricity, from 0 to 10−3. The radiation pressure component provokes, besides a variation in the eccentricity, short-period oscillations (about 40 days) in the semi-major axis of the particles, while the Poynting-Robertson component (PR, those terms velocity- dependent in Eq. 1) causes a slow decay of the semi-major axis, but in a timescale much longer than the effects of the RP component. Figure 4 shows the variation of the semi- major axis (∆a), and the resonant argument (φ) for a 1 µm in radius particle initially with ∆a=15km, without the SRF and when considering each component separately. We can see that the effects of the Poynting-Robertson are negligible in this timescale, while the radiation pressure component causes kilometer variations in the semi-major axis. For this particle, the variation in a is enough to remove the particle from the resonance in less than 2 years. Whether or not the particle remains in resonance may change its lifetime. Figure 5 shows the difference between the semi-major axis of the particle and Aegaeon, and the resonant argument as a function of time for a 10µm sized MNRAS 000, 1 -- 7 (0000) ) m k ( a ∆ 80 60 40 20 0 -20 without SRF RP PR 0 2 4 6 time (years) (a) 8 10 ) g e d ( φ 360 300 240 180 120 60 0 0 2 4 6 time (years) (b) 8 10 Figure 4. (Colour online) Time variation of (a) the difference be- tween the semi-major axis of the particle and Aegaeon and (b) the resonant argument without the solar radiation force (SRF, dot- ted line), with the Poynting-Robertson component (dashed line) and with the radiation pressure (full line). The 1µm sized parti- cle is initially with ∆a=15km and has the same mean anomaly of Aegaeon. particle, initially with the same semi-major axis (∆a0 = 0km) of Aegaeon and mean anomaly displacement by ∆λ0 = 20◦, with (dashed line) and without the effects of the solar radi- ation force. When no solar radiation force is acting in the system, the particle is trapped in the 7:6 CER with Mimas until colliding with Aegaeon in less than 10 years. The ef- fects of the solar radiation force remove the particle from the resonance after about 10 years, when the resonant argument starts to circulate, and the lifetime of the particle increases by a factor 2. In some cases a particle remains trapped in the res- onance, despite of the perturbation of the SRF. Figure 6 shows the difference between the geometric semi-major axis of the particle and Aegaeon and the resonant argument of a representative particle with 10µm in radius, initially with ∆a0 = −5km (a=167488km) and ∆λ0 = 0◦ from Aegaeon. This particle remains trapped in the 7:6 CER with Mimas for almost 35 years, until colliding with the satellite. We also found some particles that leave the arc, as it is shown in Figure 7 for a 10 µm grain in radius, initially with ∆a0 = 25 km (a=167522 km) and ∆λ0 = 0◦. This par- ticle leaves the 7:6 CER with Mimas after about 20 years. After this, the semi-major axis starts to decrease due to the Poynting-Robertson component, and the particle moves away from the arc region to the inner edge of the G ring. We then computed the percentage of particles which re- main in resonance, and those that the resonance argument circulates. The first corresponds to the column 'arc' in Ta- ble 3, while the column 'ring' corresponds to the percentage of particles that leave the CER. It is also presented the time 4 Madeira et al. ) m k ( a ∆ 30 20 10 0 -10 -20 -30 without SRF with SRF 0 5 10 15 20 time (years) (a) ) g e d ( φ 360 300 240 180 120 60 0 0 5 10 15 20 time (years) (b) Figure 5. Time variation of (a) the difference between the semi- major axis of a 10µm sized particle and Aegaeon, and (b) the resonant argument with (dashed line) and without the SRF. In both cases the particle is located at the same semi-major axis of Aegaeon and displaced by ∆λ0 = 20◦. ) m k ( a ∆ ) g e d ( φ 45 30 15 0 -15 -30 -45 0 5 10 15 20 25 30 35 40 time (years) (a) 360 300 240 180 120 60 0 0 5 10 15 20 25 30 35 40 time (years) (b) Figure 6. (a) The difference between the geometric semi-major axis of the particle and Aegaeon and (b) the resonant argument as a function of time in years for a 10 µm sized particle initially at ∆a0 = −5km and ∆λ0 = 0◦. The particle collides with the satellite in less than 35 years. ) m k ( a ∆ ) g e d ( φ 45 30 15 0 -15 -30 -45 0 20 40 60 80 100 time (years) (a) 360 300 240 180 120 60 0 0 20 40 60 time (years) (b) 80 100 Figure 7. The same as Figure 6 for a particle initially with ∆a0 = 25 km (a=167522km) and ∆λ0 = 0◦. Table 3. Percentage of particles of different sizes that remain in the arc (in resonance) and those particles that leave the arc but remain in the G ring. The time (in years) necessary to 90% of the total ensemble of particles, trapped in the arc, to be removed by collisions is shown in the last column. r (µm) arc (%) ring (%) t90 (years) 1 3 5 10 55 58 60 64 45 42 40 36 3 9 16 26 necessary to remove through collisions 90% of the "arc" par- ticles (t90). Even though Aegaeon's and the particles' semi-major axes periodically cross due to the resonance, the excitation of the eccentricity caused by the SRF increases the number of times that the particles' orbital radius intersect the or- bit of the satellite. Since the amplitude of the eccentricity varies inversely with the particles' size, it is expected that the smaller grains have a larger probability to leave the arc and also have a shorter lifetime. After 3 years, more than 90% of the 1 µm particles collide with the satellite, while 10µm sized grains have longer lifetimes. 3 PARTICLES EJECTED FROM AEGAEON'S SURFACE Throughout the Solar System there is a flux of interplan- etary dust particles (IDPs), which can be focused by the presence of a planet. While moving towards the planet, these IDPs can collide with a satellite at a speed of O(10 km/s), and if the satellite is small (radius up to a few tens of kilome- MNRAS 000, 1 -- 7 (0000) Production and fate of small particles due to Aegaeon 5 Table 4. Summary of the numerical simulations considering par- ticles of different sizes and ejection velocities. The particles were classified as arc when they remain the entire simulation in reso- nance and as ring otherwise. t90 corresponds to the time in years necessary to 90% of the ensemble to be removed by collision. since they stay closer to Aegaeon's orbit. The longer lived ensemble is the one formed by the faster and larger grains, but even in this case the particles do not survive more than 300 years. 1µm 3µm 5µm 10µm 4 MASS PRODUCTION RATE 1vesc 5vesc 10vesc arc t90 ring t90 arc t90 ring t90 arc t90 ring t90 100% 2.5 0% - 92% 3.0 8% 7.1 100% 4.6 0% - 100% 12.6 0% - 100% 14.4 0% - 100% 20.6 0% - 100% 38.2 0% - 100% 36.7 0% - 66.3% 68.7% 70.6% 71.3% 3.4 33.7% 31.3% 29.4% 28.7% 35.0 278.6 136.3 22.5 41.9 13.9 94.9 ters), the outcome of these hypervelocity impacts is the ejec- tion of micrometric particles. Before calculating the amount of material produced by these collisions on Aegaeon surface (section 4) we will analyze the orbital evolution of a set of particles ejected from its surface. We considered particles with sizes of 1, 3, 5 and 10µm in radius, leaving the surface of the satellite with initial ve- locities equal to 1, 5 and 10 times the escape velocity of the satellite (vesc). These parameters were chosen in a man- ner to cover the most likely distribution expected for the dust production mechanism (Krivov et al. 2003). Similarly to the previous section, the dust grains evolved under the influence of the planet and its gravity coefficients (J2, J4 and J6), the gravitational effects of Mimas, Tethys and Aegaeon, and also the solar radiation force. For each combination of particle size and ejection velocity, an ensemble of 1,000 par- ticles, launched at the same time, was analyzed. All particles were launched radially away from Aegaeon and in the equa- torial plane of Saturn. The angular position, related to the surface of the satellite, was randomly chosen from 0 to 360◦ with uniform probability. Table 4 summarizes the outcome of the numerical sim- ulations. We divided each sample into arc particles, corre- sponding to those remaining all the integration time in res- onance (thus azimuthally confined), and the ring particles the ones that make excursions through the ring. Since all particles hit the surface of Aegaeon we computed the time necessary to remove 90% of each set (t90). The amount of particles that remain in the arc de- creases slightly for smaller grains, but we note a substan- tial change when the ejection velocity increases. Smaller and faster grains are more likely to leave the arc and therefore they can survive longer. It is mainly caused by those par- ticles that experience stronger variations in the semi-major axis due to the SRF and, as a consequence, they leave the resonance. At this point those particles are classified as 'ring' and they present larger survival time, since their orbital paths go further from the satellite. When launched at lower speeds, the particles remain entirely confined in the arc. Their survival time is shorter MNRAS 000, 1 -- 7 (0000) Besides acting as a sink for the particles of the G ring/arc, Aegaeon can produce dust due to the impacts of interplan- etary grains. Through the simplified model presented in Sfair & Giuliatti Winter (2012) we compute the mass pro- duction rate (M+) due to the impacts of projectiles directly onto the surface of the satellite as M+ = FimpY S (2) where Fimp is the mass flux of impactors that reaches the satellite, Y is the ejecta yield and S is the satellite cross section. The characterization of the interplanetary dust grain environment is a difficult task, specially for the outer part of the Solar System due to the small number of direct measure- ments. Combining data from several missions, Poppe (2016) estimates that the mass flux at Saturn's heliocentric dis- tance is 1 × 10−17 kg/(m2·s). This value is enhanced by the gravitational focusing due to the planet, so that the effective flux at Aegaeon's orbit is Fimp ∼ 5.5 × 10−17 kg/m2/s. It was assumed that the radius of Aegaeon is 240 m, so its cross section area is ∼ 1.8 × 105 m2. The yield measures efficiency of the ejection process, and for a satellite with a pure ice surface (no silicates) it can be written as (Koschny & Grun 2011) Y = 2.64 × 10−5 m0.23 (3) imp v2.46 imp . For a typical impactor of 10−8 kg (r ∼ 100µm) with velocity of 23 km/s (after the gravitational focusing), the yield is Y ∼ 2 × 104. By this mechanism, impacts with Aegaeon produce dust grains at a rate of M+ ∼ 2 × 10−7 kg/s. (4) In order to determine if ejecta from Aegaeon can be the source of visible dust in the arc, it is necessary to estimate the mass of the arc. First we assume a power law distribution for the particle size distribution as dN = Cr−q dr (5) where dN is the number of particles and q = 3.5 is assumed as a typical value. The optical depth can be written as τ = Z r2 r1 dτ = Z r2 r1 πr2 dN (6) the dust and the mass of (Sfair & Giuliatti Winter 2012) can be calculated by m =Aarc(cid:18) 4 3 πρ(cid:19)Z r2 r1 r3 dN (7) Aarc is the surface area of the arc. We assumed a simplified model considering the arc as a 60◦ circular sector with radius 167493 ± 125km. If we consider an uniform optical depth for the arc as τ = 10−5 (Hedman et al. 2007) and dominated by small ice particles (r = [1 − 10]µm), which is expected by the 6 Madeira et al. impact process, it gives m ∼ 2 × 106 kg. This value is close to the high end of the dust mass estimates from Hedman et al. (2007). Neglecting any loss mechanism, the total amount of dust in the arc would need about 30,000 years to accumulate. Since this time is at least three orders of magnitude larger than those presented in Table 4, it is unlikely that Aegaeon alone could keep the arc dust in a steady state. 5 DISCUSSION The G ring arc region is a dynamic environment composed by dust particles, probably cm-m sized bodies, and a small satellite Aegaeon. The satellite and the particles are both trapped in a 7:6 CER with the satellite Mimas, responsible for keeping the particles and Aegaeon azimuthally confined in 60◦ of longitude. Besides the gravitational effects, the µm sized dust particles are also strongly affected by the solar radiation force, which can lead them to collisions or ejection from the arc. In this work we analyze Aegaeon's effects on the G-ring arc. In the numerical simulations the set of µm sized parti- cles is under the gravitational effects of the massive bodies Saturn, Mimas, Tethys and Aegaeon. Our results showed that Aegaeon acts as a sink for the particles, removing them by collisions. About 75% of the confined particles are re- moved from the arc in less than 500 years. The solar radiation component induces short period variations in the semi-major axis of the particles, which changes the resonant argument, and as a result most of the particles are removed from the resonance. This force also changes the eccentricities of the particles, leading to orbital crossing and eventually collisions between these particles and the satellite Aegaeon. The lifetime of 90% of the ini- tial set of 1 µm sized particles is about 3 years, and 26 years for particles 10 µm in radius. Therefore, the presence of Ae- gaeon reduces the lifetime of the particles leading to the extinction of the arc. This lifetime may be even shorter, since in our numerical simulations the plasma drag and the electromagnetic force were not taken into account. The for- mer perturbation causes an outward drift (Sun et al. 2015), while the latter may increase even more the eccentricity of the particles (Hamilton 1993), thus both perturbations will probably cause particles to be removed from the arc even faster. Hedman et al. (2009) argued that debris ejected from the surface of small satellites can stay confined in the same resonance of these satellites. Our results showed it is true for the particles ejected at slower velocities. However, the confined particles have a very short lifetime, 90% of the en- tire population leave the arc in less than 20 years. Only 8% of the smaller particles goes to the G ring, the majority of them collide with the satellite Aegaeon. The longer lived ensemble is composed by larger particles leaving the surface of the satellite with 10vesc, these particles can last up to ∼ 300 years. Although Aegaeon is below the optimum size to gen- erate dust by impacts of interplanetary dust particles (Burns et al. 2001) we calculated the amount of dust this small satellite can contribute to the arc population. Our re- sult shows that Aegaeon can produce particles at a rate of 2 × 10−7 kg/s, and if the mass of the arc is about 2 × 106 kg, neglecting any loss mechanism, it would take more than 300,000 years to Aegaeon populated the arc. Our simula- tions therefore show that Aegaeon is probably a net sink for arc particles. Our simplified model takes into account only the direct process of dust. The assumed yield and interplanetary dust flux must be taken at least with one order of uncertainty. However, even considering a lower estimate for the arc mass and the upper limits for the flux and yield, the time neces- sary to accumulate such amount of dust is several orders of magnitude larger than the survival time of the particles. In order to maintain the dust population in a steady state additional processes must be invoked. For instance, there is evidence that the ring may be populated by multi- ple objects O(cm-m) across, which are below the threshold level of the cameras to be detected. A more complex process involving secondary impacts of IDPs with these objects, or even impacts among themselves, could produce more dust particles than the primary impacts. 6 ACKNOWLEDGEMENTS We would like to thank the anonymous reviewer who greatly improved the final text. The authors thank Fapesp (Proc. 2016/2488-0 and Proc. 2011/08171-3) and CNPq (Proc. 309714/2016-8 and Proc. 305737/2015-5) for the fi- nancial support. REFERENCES Burns J. A., Lamy P.L., Soter S., 1979, Icarus, 40, 48 Burns J.A., Hamilton D.P., Showalter M.S., 2001. Dusty rings and circumplanetary dust. In Interplanetary Dust, Springer Verlag, Berlin Chambers J.E., 1999, Monthly Notices of the Royal Astronomical Society, 304, 793 El Moutamid M., Sicardy B., Renner S., 2014, Celestial Mechanics and Dynamical Astronomy, 118, 235 Hamilton D. P., 1993, Icarus, 101, 244 Hamilton D. P., 1996, Icarus, 123, 503 Hedman M.M., Burns J.A., Tiscareno M.S., Porco C.C., Jones G.H., Roussis E., Krupp N., Paranicas C., Kempf S., 2007, Science, 317, 653 Hedman M.M., Murray C.D., Cooper N.J., Tiscareno M.S., Beurle K., Evans M.W., Burns J.A., 2009, Icarus, 199, 378 Hedman M.M., Cooper N.J., Murray C.D., Beurle K., Evans M.W., Tiscareno M.S., Burns J.A., 2010, Icarus, 207, 433 Koschny D., Grun E., 2011, Icarus, 154, 402 Krivov A. V., Sremcevic M., Spahn F., Dikarev V. V., Khol- shevnikov K. V., 2003, Planet. Space Sci., 51, 251 Mignard F., 1984. Planetary rings, University of Arizona Press, Tucson Poppe A. R., 2016, Icarus, 264, 369 Renner S., Sicardy B., 2006, Celestial Mechanics and Dynamical Astronomy, 94, 237 Sfair R., Giuliatti Winter S. M., 2009, Astronomy and Astro- physics, 505, 845 Sfair R., Giuliatti Winter S. M., 2012, Astronomy and Astro- physics, 543, 17 Sun K. L., Seiss M., Spahn F., 2015, arXiv, 1510.07722 Thomas P. C., Burns J. A., Hedman M., Helfenstein P., Morrison S., Tiscareno M. S., Veverka, J., 2013, Icarus, 226, 999 MNRAS 000, 1 -- 7 (0000) This paper has been typeset from a TEX/LATEX file prepared by the author. Production and fate of small particles due to Aegaeon 7 MNRAS 000, 1 -- 7 (0000)
1007.4497	1	1007	2010-07-26T16:18:11	Dipolar Magnetic Moment of the Bodies of the Solar System and the Hot Jupiters	[ "astro-ph.EP", "physics.space-ph" ]	The planets magnetic field has been explained based on the dynamo theory, which presents as many difficulties in mathematical terms as well as in predictions. It proves to be extremely difficult to calculate the dipolar magnetic moment of the extrasolar planets using the dynamo theory. The aim is to find an empirical relationship (justifying using first principles) between the planetary magnetic moment, the mass of the planet, its rotation period and the electrical conductivity of its most conductive layer. Then this is applied to Hot Jupiters. Using all the magnetic planetary bodies of the solar system and tracing a graph of the dipolar magnetic moment versus body mass parameter, the rotation period and electrical conductivity of the internal conductive layer is obtained. An empirical, functional relation was constructed, which was adjusted to a power law curve in order to fit the data. Once this empirical relation has been defined, it is theoretically justified and applied to the calculation of the dipolar magnetic moment of the extra solar planets known as Hot Jupiters. Almost all data calculated is interpolated, bestowing confidence in terms of their validity. The value for the dipolar magnetic moment, obtained for the exoplanet Osiris (HD209458b), helps understand the way in which the atmosphere of a planet with an intense magnetic field can be eroded by stellar wind. The relationship observed also helps understand why Venus and Mars do not present any magnetic field.	astro-ph.EP	astro-ph	Published in Planetary and Space Science 57:1405-1411 (2009) Dipolar Magnetic Moment of the Bodies of the Solar System and the Hot Jupiters Hector Javier Durand-Manterola Departamento de Ciencias Espaciales, Instituto de Geofísica, Universidad Nacional Autónoma de México [email protected] Abstract The planets magnetic field has been explained based on the dynamo theory, which presents as many difficulties in mathematical terms as well as in predictions. It proves to be extremely difficult to calculate the dipolar magnetic moment of the extrasolar planets using the dynamo theory. The aim is to find an empirical relationship (justifying using first principles) between the planetary magnetic moment, the mass of the planet, its rotation period and the electrical conductivity of its most conductive layer. Then this is applied to Hot Jupiters. Using all the magnetic planetary bodies of the solar system and tracing a graph of the dipolar magnetic moment versus body mass parameter, the rotation period and electrical conductivity of the internal conductive layer is obtained. An empirical, functional relation was constructed, which was adjusted to a power law curve in order to fit the data. Once this empirical relation has been defined, it is theoretically justified and applied to the calculation of the dipolar magnetic moment of the extra solar planets known as Hot Jupiters. Almost all data calculated is interpolated, bestowing confidence in terms of their validity. The value for the dipolar magnetic moment, obtained for the exoplanet Osiris (HD209458b), helps understand the way in which the atmosphere of a planet with an intense magnetic field can be eroded by stellar wind. The relationship observed also helps understand why Venus and Mars do not present any magnetic field. 1 Introduction 1 Several credited theories exist, which attempt to provide an explanation for the magnetic field of a planet. The first, proposed by William Gilbert in 1600, in his book De Magnete was that of permanent magnetization; or in other words, a planet body behaves as a permanent magnet. Over time, other explanations have emerged, such as electric charge in rotation in the planetary body (Sutherland, 1900; Parkinson, 1983), the free fall of electric currents (Lamb, 1932; Rikitake, 1966), giro-magnetic effect (Barnett, 1933), thermo- electrical effects (Elsasser, 1939), Hall’s effect (Vestine, 1954), differential rotation (Inglis, 1955), electromagnetic induction caused by magnetic storms (Chatterjee, 1956), etc. Currently, the most accepted explanation is the dynamo model. In spite of the fact that this represents the best candidate for inducing the magnetic field of a planet, it presents some problems. For example, the theory predicts that the axis of the dipole cannot be parallel to the axis of rotation (Ferraro and Plumpton, 1966, pp 2; Alexeff, 1989) and yet Saturn has these two axes parallel to each other (de Pater and Lissauer, 2001, pp 264). In short, the exact mechanism for generating the magnetic field of the planets is not known with certainty. And the various theories which have been developed, based on the dynamo model, have not managed to explain all aspects of this phenomenon. We know from the Ampere-Maxwell law that a magnetic field can be generated in two ways; either by electric currents or by temporal variations in the electric field. In the interior of a planet there may be free charges which generate electrical fields, which, when varied, (because of the separation and recombination of charges) are able to generate magnetic fields. However, because of the neutral, global character of the material that forms planets, we may assume that the separations of charge that occur will not be very great and because of this, the electric and magnetic fields produced will not be very intense. In the case of electric currents, these can produce an intense magnetic field, so that in the case of the magnetic planetary fields it is very probable that these are generated by currents in the interior of the body. These currents may constitute molecular currents, like those that generate permanent magnetism. However, given the temperatures that prevail in the interior of the planets, temperatures exceeding Curie’s temperature point; this magnetism could only exist in the external layers of the planet, where temperatures are lower. We do not know what level of magnetism exists in the outer layer of the Mercury, but in the case of the Earth, the magnetic intensity of the planet by far exceeds the magnetism of the rocks of the outer layer and in the case of the gaseous giants, the upper layers are 2 conformed of gas, thus impeding the presence of any permanent magnetism. All these arguments lead us to reject the idea of permanent magnetism or that molecular currents should be a source of planetary magnetism; and thus the only remaining possibility is that they represent macroscopic currents. For this to be possible, a conducting material must be present within the body of the planet where the currents are produced. If this is the case, the magnetic moment of the planet should therefore depend on its conductivity and on the quantity of conducting material present, which may also depend on the total mass of the planet. If the mechanism that generates the magnetic field is of the dynamo type, then the magnetic moment will also depend on the speed at which the planet rotates, or, in other words, its rotation period. In this study, these three parameters are correlated: mass m, rotation period P, and electric conductivity σ for all the magnetic planets of the solar system, together with their dipolar magnetic moment M, in order to obtain an empirical equation, allowing the calculation of the magnetic moment for the extra solar planets. Section 2 describes the relationship that exists between the magnetic moment and the mass of the planet; section 3 describes the relationship between the magnetic moment and the rotation period of the planet.; section 4 describes the relationship that exists between the magnetic moment and electric conductivity; section 5 describes the relationship of the magnetic moment with a function of the three previous parameters; section 6 provides a rough outline of the theoretical justification for the empirical formula. And in section 7 the empirical formula is employed to calculate the magnetic moment of certain extra solar planets; those denominated as the “Hot Jupiters” 2 The functional relationship between magnetic moment and mass Seven magnetic planets are known to exist in the solar system: Jupiter, Saturn, Uranus, Neptune, Earth, Mercury and Ganymede. From the measurements of the magnetic moment and the mass of these seven planetary objects (see Table I), a curve may be fitted to the data as can be seen in figure 1. The relationship between the magnetic moment and the mass is a power law: (1) Hence, increasing planet mass increases magnetic field. This relationship between the data has a correlation coefficient of 0.9804, reinforcing the idea 10x2M = 21 m .1 − 7365 3 that a physical relationship exists between these two variables. If it is assumed, as with the dynamo theory, that the magnetic field is produced by electric currents in a conducting layer, then relationship (1) indicates that if planet mass increases, the electric current that generates the magnetic field also increases. This is coherent, as mass increases, we also expect the amount of conductor material to increase, so the electric current will be more intense and consequently the magnetic field will be stronger. However, the generation of the magnetic field does not only depend on the mass. If we apply equation (1) to the case of the planet Venus, according to its mass it should have a magnetic moment similar to that of the earth (3.953x1022 Amp-m2; Cox, 2000). However in reality, its magnetic moment is either non-existent or less than 5x1017 Amp-m2 (Cox, 2000). This makes us conclude that even though planetary mass may be significant; it is not the only factor intervening in the generation of a magnetic field. Figure 1 Functional relationship between the planetary dipolar magnetic moment and the mass for the seven magnetic planetary bodies of the solar system. Table I Data of the Solar System Magnetic Planets 4 Planet Mass m (x1024 kg) Mass Error (kg) Period P (s) P. Error (s) Mercury * 0.33022 5.9742 Earth * 1898.7 Jupiter * 568.51 Saturn * 86.849 Uranus * Neptune * 102.44 Ganymede + 0.148186 * Tholen et al., 2000 + Kivelson et al., 2002 ± 5x1018 ± 5x1019 ± 5x1022 ± 5x1021 ± 5x1020 ± 5x1021 ± 5x1017 5067031 ± 5 86164.1003 ± 5e-4 35729.8 ± 0.5 38362.4 ± 0.5 62063.7 ± 0.5 57996 ± 0.5 618153.3757 ± 5x10-4 Dipolar Magnetic Moment M (Amp-m2) 4x1019 7.84x1022 1.55x1027 4.6x1025 3.9x1024 2.2x1024 1.32x1020 M. Error (Amp-m2) ± 5x1018 ± 5x1019 ± 5x1024 ± 5x1023 ± 5x1022 ± 5x1022 ± 5x1017 3 Functional relationship between the magnetic moment and the rotation period Once again, considering the data but now referring to the dipolar magnetic moment and the rotation period for the seven magnetic planets (see Table I), we can observe in figure 2 that in the case of these planets, the functional relationship between the magnetic moment and the rotation period represents an exponential function, meaning, that when the inverse of the rotation period is increased or the rotation period is decreased; the magnetic moment increases and can be expressed as: 10x4M = 19 ⎛ exp ⎜ ⎝ 579696 P (2) ⎞ ⎟ ⎠ This relationship between the data has a correlation coefficient of 0.9742, reinforcing the idea that a physical relationship exists between these two variables. However, the generation of the magnetic field does not only depend on the rotation period. If we were to apply equation (2) to the case of the planet Mars, its rotation should induce a magnetic moment similar to that of the Earth (3.953x1022 Amp-m2; Cox, 2000). However, in reality, its dipolar moment is non-existent or less than 5x1018 Amp-m2.(Cox, 2000) Thus, we can conclude that although the rotation period is important; it is not the only factor which intervenes in the generation of the magnetic field. 5 Figure 2 Functional relationship between the dipolar magnetic moment and the rotation period for the seven magnetic planetary bodies of the solar system. 4 Functional relationship of M with σ Contrary to mass and rotation periods, which are both parameters that can be measured, the value of conductivity in the conducting layer is not easy to calculate. From models of the interior of the planet Earth, iron-nickel (Fe-Ni) alloy is the most abundant component in the Earth’s core (Lin et al., 2002) forming a conductive layer with a mean electric conductivity of 1.2 ± 0.2 x105 S/m, between 1.02±0.005 x105 S/m (Iron conductivity) and 1.43±0.005 x105 S/m (nickel conductivity) (Kittel, 2004). Models for the interior of Mercury also suggest that the interior is made of iron-nickel (Hamblin and Christiansen, 1990) which is responsible for Mercury’s magnetic field. Some contemporary geodynamic models say that shear motions between the core and the base of the Earth's mantle drives the geodynamo (Labrosse et al., 2007, and Labrosse et al., 2003). If this is the case, the composition of this layer is uncertain but most probably enriched by iron. The core conductivity upper limit is probable and so will be used here. In the case of the gaseous giants, it has been suggested that the conductive layer is formed from metallic hydrogen, with a conductivity of 2 ± 0.5 x105 S/m (Shvets, 2007). In the case of Ganymede, Kivelson et al. (2002) suggest that salt water forms the conductive layer. As we do not know the concentration of these salts, it 6 would be risky to calculate the conductivity of the Ganymedean Ocean. The maximum conductivity of the Earth’s seawater is 6.5683 S/m (Kennish, 2001), so this may provide a first approximation for Ganymede. However, the absence of a magnetic field in Europa or Callisto (Gurnett et al., 1997), most of all in Europa, where there are indications that a sub-superficial ocean exists throws doubts on the possibility that an ocean of this type would be capable of generating a magnetic field, such as that which is calculated for Ganymede. For this reason and others mentioned later in this study, a conductivity of 1.2 ± 0.2 x105 S/m was attributed to Ganymede, assuming it has an iron nucleus. With only two conductivity values for the seven planets it would be pointless to try and calculate the functional relationship between σ and M, however, in figure 3, we can see that all the bodies with less conductivity (those of iron- nickel) have magnetic moments which fall below those with layers of metallic hydrogen (greater conductivity). Figure 3 Relationship between the dipolar magnetic moment and electrical conductivity. 5 Functional relationship of M with mσ/P Finally, putting together all three factors under a single parameter mσ/P; the values of the seven planets constitute a power law curve which we can observe in figure 4 7 10x1M = − 5 106.1 m ⎡ σ ⎢⎣ P ⎤ ⎥⎦ (3) In order to determine the variation interval from the inclination of the curve (exponent of equation 3), various fittings for the data were carried out, in each case removing one of the planets and calculating the value for this group of data. In Table II, the values are presented for each case. The average and error were calculated for extreme values. In the past some authors proposed a potential relationship between the magnetic moment (M) and the angular momentum (L), the so-called magnetic Bode’s law (Blackett, 1947, Kennel, 1973; Hill and Michel, 1975; Dessler, 1976; Siscoe, 1978; Russell, 1978). Because no physical justification existed some other authors question it (Cain et al, 1995). In the next section we derive equation (3) from the first principles. Figure 4 Planetary magnetic moment as a function of the parameter σm/P, where σ represents the conductivity of the conductor layer of the planet, m its mass and P the rotation period of the body. The diamonds represent the 7 magnetic bodies of the solar system. The straight line represents the graph for equation (3). The filled in triangle is Ganymede only if its conductivity were the same as that of sea water. The filled in circle and square represent the maximum values for the magnetic moment of Mars and Venus, respectively, graphed against the parameter σm/P, assuming a conductivity of 1 S/m which is the conductivity of silicates (perovskite) The empty triangles and squares 8 represent the predicted values for equation (3), in the case that Venus and Mars have either a conductivity of iron-nickel (triangles) or silicates (squares). All Without Mercury Without Earth Without Jupiter Without Saturn Without Uranus Without Neptune Without Ganymede ------ Table II Variation of the exponent of equation (3) Eliminating one planet of the seven. R2 is the Correlation coeficient of data Exponent Error of the Average 1.1218 1.1685 1.1212 1.0435 1.1388 1.1267 1.1418 1.1143 ------ ------ 1.1685 1.0435 1.10 ± 0.13 Larger Minor Average R2 0.9814 0.975 0.9813 0.9959 0.9805 0.9814 0.9878 0.9739 ------ 6 Theoretical justifications for formula (3) The scalar magnetic potential for a dipole seen from a distance r is (Reitz y Milford, 1981, pp 178): ( ) rU = ( ) cosMr θ 3r4 π (4) Where θ is the angle between the vector M and the vector r. On the other hand, the scalar magnetic potential of any circuit seen from afar is (Reitz and Milford, 1981, pp 179): (5) Where I is the current in the circuit and Ω is the solid angle from which the circuit can be seen from a distance r. −= ( ) rU I Ω 4 π 9 Near to the planet, the magnetic field is seen as a magnetic dipole, and because of this (4) and (5) can be made equal in order to obtain M. We also substitute the electric current for sε, where s represents electric conductance and ε represents the electromotive force. And, if we are placed at the magnetic equator of the planet, where cos (θ) = 1 and r = R, R representing the radius of the planet, then, 4πR3 = V represents the volume of the planet and s/2πR = σ represents electrical conductivity. Thus, the dipolar moment can be expressed as: V (6) (7) M −= Ωσε 2 d BΦ dt −=ε From Faraday’s law: And by the chain rule: =ε d- Φ B dt −= d Φ B d ϕ d ϕ dt −= d Φ B d ϕ ω (8) And from equations (6) and (8) we have: M = d ΦΩ B 2 d ϕ V ωσ (9) But ω = 2π/P. We substitute this and multiply the numerator and the denominator by ρ the average density of the planet. On the other hand ρV = m. Then equation (9) is expressed as: M = Ωπ ρ d m Φ B σ d P ϕ (10) That is to say m M σ ∝ P In equation (3), the exponent is different from unity, but close to it. (11) 10 7 Magnetic fields of the “Hot Jupiters” If relationship (3) represents a general formula for planets, then it will serve for predicting the magnetic field of extrasolar planets known as “Hot Jupiters”. The reason “Hot Jupiters” were chosen from among the other exoplanets is because they are located at short distance from the star, we can assume that their rotation has been trapped and therefore it is the same as the period it takes to travel around the star. Once its rotation period and mass are known, assuming that these are gaseous planets, we can use the conductivity of metallic hydrogen in order to calculate the parameter σm/P. Using formula (3), the magnetic moment can be calculated. Table III shows the calculated values for 31 Hot Jupiters. Inserted into the same table were Rss values of the distance from planet to the sub stellar point of the magnetopause calculated with the formula: Rss = 1 6 M 2 V4 πρ ⎡ ⎢ ⎣ 2 ⎤ ⎥ ⎦ (12) 8 Discussion and Conclusions In the case of Mars and Venus, because they are terrestrial planets, one would expect them to have an iron-nickel nucleus. If we take the conductivity of iron for Venus and Mars and use relationship (3) to calculate M, it gives us a greater magnetic moment than that of Mercury, greater than the upper limits measured for either planet (figure 4, empty triangles). It is known that in the past Mars had a magnetic field (Acuña et al, 2001) and for some reason it disappeared. It is possible that great impacts, such as those which produced Argyre and Hellas could eliminate the dynamo which generated the magnetic field. That impact could have injected material with a low conductivity (i.e. silicates) into the planet’s nucleus. In figure 4 (filled square and circle), we see the maximum estimate for the magnetic moment of both these planets, which has been calculated based on the observations. The arrows indicate that the real value of the magnetic moment must be lower than these values. In the same figure (empty squares) we can observe what the magnetic moment for both planets would be if we reduce the conductivity, assigning them that of silicates (perovskita) and applying equation (3). The idea here is that even though metals exist in the nucleus, these do not necessarily form a continuous mass, but are instead mixed with a material of poor conductivity (silicates) and that these are 11 unable to generate generalized currents for the entire nucleus. In the case of Venus another possibility exists. The Earth’s magnetic field has been inverted countless times, as indicated in studies of oceanic rocks. However, we do not know what happens when these inversions occur. We do not know whether the magnetic field diminishes until it disappears and then grows in an inverted way, or if it simply rotates (Merrill and McFaden, 1999). If it were the former, how long does this period of field zero endure? It is known from the oceanic rocks that this moment is geologically rapid. However, one thousand years is rapid in geological terms. Is it possible that Venus is currently going through one of these moments? Some estimates of the time of an inversion lie between 1000 and 8000 years (Merrill and McFaden, 1999). The slow retrograde rotation of Venus would anyways lead to a theoretical value of 0.33% of the Earths magnetic moment. According to some contemporary geodynamic models (Labrosse et al., 2007, and Labrosse et al., 2003) one may also conceive the absence of a magnetic dynamo as the cessation of intrinsic shear motion between the inner mantle and the nucleus induced by a catastrophic event. In the same figure, it can be seen that the relationship between the parameter mσ/P and M for Ganymede, does not coincide with the relationship found for other planets using the conductivity of sea water (5S/m) (full triangle). On the other hand, using the conductivity of iron-nickel, it would fall within the relationship. This appears to indicate that within the interior of Ganymede, there is a conducting, metallic nucleus. Taking the average density of Ganymede ( ρ =1,406 kg/m3) and its radius (R = 2.634x106 m) and assuming an interior model of 2 layers; a covering of ice (ρm = 1000 kg/m3) and a nucleus of iron-nickel (ρn = 11,000 kg/m3), we can calculate the radius of the nucleus applying the following formula r = ⎡ ⎢ ⎣ ⎤ ρ−ρ m ⎥ ρ−ρ ⎦ n m R (12) This radius would be 107 km; that is to say, a very small nucleus. It may be assumed that among the components of the satellites of the Giant planets, metals are present but in the case of the majority, the metal is mixed with non-conducting compounds, without forming a continuous mass and because of this, they cannot create strong electric currents. This is the reason that bodies such as Titan, Europa and Callisto do not have a magnetic field. However, in the case of Ganymede this being the largest satellite, it is 12 possible that the differentiation of the body has been total and that the metal has been separated, forming an iron nucleus. In Table 1, we can observe an interesting result, when all the magnetic planets are correlated, or if we remove one that is not Jupiter, the exponent will always exceed 1.1. However, in the moment we remove Jupiter, the exponent reduces to almost unity and the correlation coefficient is almost one. This tells us that Jupiter may have an unusually intense magnetic field, either because the planet is anomalous, or because it is passing through a brief interval of greater intensity than usual. The benefit that the values for the magnetic moment can provide for the exoplanets is that with these, together with the radius of the planet (if known); the magnetic intensity of the surface of the planet can be calculated. Similarly, if the value of the magnetic field at the surface is known, then the distance of the sub stellar point of the magnetopause from the body may be found (see Table III). Schneider et al. (1998) predicted that exoplanets of the Hot Jupiter type would have a cometary exosphere. Schneiter et al. (2007) made simulations for the interaction of stellar wind with the expanded atmosphere of the Hot Júpiter HD209458b (“Osiris”), obtaining this type of cometary exosphere. They concluded that the existence of a great stellar cometary wake suggests that Osiris does not possess a global magnetic field. In this work, the magnetic moment calculated for Osiris is equal to 1.26x1026 Amp-m2, which is equivalent to 5.5 times the magnetic moment of Saturn and 1/6 that of Jupiter. Sánchez-La Vega (2004) using a completely different method came to the same conclusion, that Osiris has a magnetic moment that is intermediate between that of Saturn and of Jupiter. This means an intense magnetic field, but does not necessarily contradict the results of Schneiter et al. (2007). Osiris is located at 0.045 U.A. from the star and for this reason its atmosphere is evaporating. The neutral gases escape from the magnetosphere without problem. The particles ionised by the EUV and X-rays radiation out of the plasmasphere are convected to the front of the magnetosphere and expelled to the stellar wind. It is also possible that an interaction between Osiris and its star (Preusse et al., 2006) permit the escape of gas from the cusps of the planet magnetosphere. From this study the following conclusions can be made: • Using known data from the solar system, it is possible to construct a function which permits us to calculate the magnetic moment for the Hot Jupiters. • Theoretical justification of the above function was obtained. 13 • The values calculated for magnetic moment are mostly interpolated. • Once the values for magnetic moments are known; it is possible to calculate the sub stellar point of the magnetopause and subsequently the size of the magnetosphere. • The value for magnetic moment calculated for the planet Osiris permits us to understand how a body with a magnetic field can simultaneously be eroded by solar wind. • The same function permits us to understand the very weak magnetic moment, manifested by Mars and Venus, assuming that the metallic nucleus underwent a mixing of silicates from the mantle, leading to catastrophic events. Table III Magnetic Moment M and sub stellar distance Rss of the magnetopause calculated for 31 Hot Jupiters Orbital period P (s) 1.0471E+05 1.2377E+05 2.2843E+05 1.4600E+05 2.4346E+05 2.2020E+05 2.4278E+05 2.6180E+05 2.5796E+05 2.6719E+05 2.6127E+05 2.6795E+05 2.6758E+05 2.9108E+05 2.9758E+05 3.0454E+05 3.0136E+05 3.0326E+05 2.9462E+05 3.4685E+05 3.4309E+05 2.8629E+05 3.6554E+05 4.2714E+05 3.9892E+05 Mass m Planet (kg) 1 OGLE-TR-56 b 2.7538E+27 2 OGLE-TR-113 b 2.5639E+27 3 GJ 436 b 1.2724E+26 4 OGLE-TR-132 b 2.26E+27 5 7.2169E+26 HD 63454 b 6 3.5515E+27 HD 73256 b 7 55 CnC e 8.5463E+25 8 1.1585E+27 TrES-1 9 7.7866E+26 HD 83443 b 10 HD 179949 b 1.8612E+27 11 HD 46375 b 4.7289E+26 12 OGLE-TR-10 b 1.0825E+27 13 HD 187123 b 9.8757E+26 14 HD 330075 b 1.4434E+27 15 HD 2638 b 9.116E+26 16 HD 209458 b 1.3104E+27 17 BD -10 3166 b 9.116E+26 18 HD 75289 b 7.9765E+26 19 HD 88133 b 4.1782E+26 20 OGLE-TR-111 b 1.0066E+27 21 HD 76700 b 3.7414E+26 22 Tau-Boo 7.8436E+27 23 51 Peg b 8.8881E+26 24 HD 49674 b 2.279E+26 25 ups-And b 1.3104E+27 mσ/P (kg S/ms) 5.2599E+27 4.1431E+27 1.1141E+26 3.0958E+27 5.9286E+26 3.2257E+27 7.0402E+25 8.8503E+26 6.0371E+26 1.3931E+27 3.6199E+26 8.08E+26 7.3815E+26 9.9173E+26 6.1268E+26 8.606E+26 6.0499E+26 5.2605E+26 2.8363E+26 5.804E+26 2.181E+26 5.4795E+27 4.863E+26 1.0671E+26 6.57E+26 M (Amp-m2) 9.3346E+26 7.169E+26 1.314E+25 5.1939E+26 8.3479E+25 5.4354E+26 7.9091E+24 1.3003E+26 8.5171E+25 2.1476E+26 4.8374E+25 1.1757E+26 1.0638E+26 1.4747E+26 8.6571E+25 1.2606E+26 8.537E+25 7.3138E+25 3.6935E+25 8.1542E+25 2.7621E+25 9.7666E+26 6.7052E+25 1.2528E+25 9.3523E+25 Rss (km) 4.08e6 3.76e6 1.06e6 3.72e6 2.13e6 4.02e6 9.89e5 2.54e6 2.22e6 3.02e6 1.85e6 2.51e6 2.43e6 2.73e6 2.31e6 2.63e6 2.33e6 2.21e6 1.77e6 2.31e6 1.63e6 5.39e6 2.24e6 1.32e6 2.61e6 14 4.3681E+26 2.4309E+27 3.6084E+27 2.6114E+28 2.0511E+27 7.9765E+25 5.5322E+05 6.1576E+05 5.4225E+05 7.2820E+05 9.2655E+05 8.2512E+05 1.59e6 2.95e6 3.59e6 6.72e6 2.57e6 8.16e5 1.5792E+26 7.8957E+26 1.3309E+27 7.1721E+27 4.4274E+26 1.9334E+25 1.9326E+25 1.1461E+26 2.0418E+26 1.3153E+27 6.0441E+25 1.894E+24 26 HD 168746 b 27 HD 217107 b 28 HD 68988 b 29 HD 162020 b 30 HD 130322 b 31 HD 160691 d References Acuña, M.H. et al., (2001) Magnetic field of Mars: Summary of results from the aerobraking and mapping orbits. JGR 106(E10): 23403-23417. Alexeff, I. (1989) Disproof of Cowling Theorem. IEEE Transaction on Plasma Science. 17(2): 282-283. Barnett, S.J. (1933) Giromagnetics effects: history, theory and experiments. Physica 13, 241 Blackett, PMS (1947) The Magnetic Field of Massive Rotating Bodies Nature 159:658-666 Cain, J.C., P. Beamont, W. Holter, Z. Wang, and H. Nevanlinna. (1995) The Magnetic Bode Fallacy. JGR 100(E5): 9439-9454 Chatterjee, J.S. (1956) Induction in the core by magnetic storms and Earth’s magnetism. Sci. Cult. 21, 623 De Pater, I. and J.J. Lissauer (2001) Planetary Sciences. Cambridge University Press. Dessler, A.J. (1976) In Solar Wind Interactions with the Planets Mercury, Venus and Mars, 159-166 (NASA SP-397, Washington DC) Elsasser, W.M. (1939) On the origin of the Earth’s magnetic field. Phys. Rev. 55, 489. Ferraro, V.C.A. and C. Plumpton (1966) An Introduction to magneto-fluid Mechanics. Oxford University Press. Gurnett, D.A., W.S. Kurth, A. Roux and S.J. Bolton (1997) Absence of magnetic field signature in plasma-wave observations at Callisto. Nature 387: 261- 262. Hamblin, W.K. and E.H. Christiansen (1990) Exploring the planets. McMillan- Collier Hill, T.W. and F.C. Michel (1975) Rev. geophys. Space Phys. 13: 967-974 Inglis, D.R. (1955) Theories of the Earth’s magnetism. Rev. Mod. Phys. 27, 212. Kennish, M.J. (2001) Practical Handbook of Marine Science, CRC Press, Boca Raton, FL. Kennel, C.F. (1973) Space Sci. Rev. 14:511-533 15 Kivelson, M.G., K.K. Khuruna and M. Volwerk (2002) The Permanent and Inductive Magnetic Moments of Ganymede. Icarus 157: 507-522 Kittel, C. (2004) Introduction to Solid State Physics. Wiley & Sons Labrosse, S., M. Macouin (2003) The inner core and the geodynamo. Geoscience 335: 37-50 Lambrose, S., J.W. Hernlund and N. Coltice (2007) A crystallizing dense magma ocean at the base of the Earth’s mantle. Nature 450: 866-869. doi:10.1038/nature06355 Lamb, H. (1932) Hydrodinamics. 6a edition. Cambridge Univversity Press. London. Lin, J., D.L. Heinz, A.J. Campbell, J.M. Devine, W.L. Mao and G. Shen. (2002) Iron-Nickel alloy in the (cid:31)elás’s core. Geophys. Res. Lett. 29(10): 1471 Merrill, R.T. and P.L. McFadden (1999) Geomagnetic Polarity Transitions. Rev. Geophys. 37,2 / May 1999 Parkinson, W.D. (1983) Introduction to Magnetism. Scottish Academia Press. London Preusse, S., A. Kopp, J. Büchner, and U. Motschmann. (2006) A magnetic communication scenario for Hot Jupiters. A&A 460: 317-322. Reitz, J.R. and F.J. Milford (1981) Fundamentos de la teoría electromagnética. Editorial UTEHA Rikitake, T. (1966) Electromagnetism and the (cid:31)elás’s Interior. Elsevier Pub. Co. (cid:31)elásquez Russell, C.T. (1978) Re-evaluating Bode’s law of planetary magnetism. Nature 272(5649):147-148, March 9 Sánchez-La Vega, A. (2004) The magnetic field in giant extrasolar planets. The Astroph. J. 609: L87-L90 Schneider, J., H. Rauer, J.P. Lasota, S. Bonazzola, and E. Chassefire. (1998) Brown Dwarfs and Extrasolar Planets. In ASP Conf. Ser. 134. Ed. R. Reboldo, E.L. Martin and M.R. Zapatero-Osorio (San Francisco: ASP), 241. Schneiter, E.M., P.F. (cid:31)elásquez, A. Esquivel, A.C. Raga and X. Blanco-Cano. (2007) Three-Dimensional Hydrodynamical Simulation of the Exoplanet HD 209458b. The Astrophys. J. 671: L57-L60 Shvets, V.T. (2007) Electrical Conductivity of Metallic Hydrogen in the Nearly-Free-Electron Model. The Physics of Metals and Metallography 103(4): 330-336 Siscoe, G.L. (1978) In Solar System Plasma Physics-A Twentieth Anniversary Review (eds Kennel, CF, Lanzaroti, L.J. and Parker, E.N.) 16 Sutherland, W. (1900) A possible cause of the Earth’s magnetism and a theory of its variations. Terrest. Magnetism Atmospheric Elec. 5, 73. Tholen, D.J., V.G. Tejfel and A.N. Cox. Chapter 12 Planets and Satellites. In Allen´s Astrophysical Quantities. 4th Edition. Springer-Verlag, Ney York. Vestine, E.H. (1954) The Earth’s core. Trans. Am. Geophys. Union, 35, 63 17
1905.00760	1	1905	2019-05-02T14:13:10	The floatability of aerosols and waves damping on Titan's seas	[ "astro-ph.EP" ]	Titan, Saturn's largest moon, has a dense atmosphere, together with lakes and seas of liquid hydrocarbons. These liquid bodies, which are in polar regions and up to several hundred kilometres in diameter, generally have smooth surfaces despite evidence of near-surface winds. Photochemically generated organic aerosols form a haze that can settle and potentially interact with the liquid surface. Here we investigate the floatability of these aerosols on Titan's seas and their potential to dampen waves. We find that the majority of aerosols are denser than the liquid hydrocarbons, but that some could have liquid-repelling properties. From calculation of the capillary forces, we propose that these 'liquidophobic' aerosols could float and form a persistent film on Titan's seas. We numerically model the wave damping efficiency of such a film under the conditions on Titan, demonstrating that even a film one molecule thick may inhibit formation of waves larger than a few centimetres in wavelength. We conclude that the presence of a floating film of aerosols deposited on Titan's lakes and seas could explain the remarkable smoothness of their surfaces.	astro-ph.EP	astro-ph	The floatability of aerosols and waves damping on Titan's seas Daniel Cordier,1 Nathalie Carrasco2 3 , 1Groupe de Spectrom´etrie Mol´eculaire et Atmosph´erique - UMR CNRS 7331 Campus Moulin de la Housse - BP 1039, Universit´e de Reims Champagne-Ardenne 51687 REIMS -- France. 2LATMOS, UMR CNRS 8190, Universit´e Versailles St Quentin, UPMC Univ. Paris 06, 11 blvd d'Alembert, 78280 Guyancourt, France 3Institut Universitaire de France, 103, bd Saint-Michel, 75005 Paris, France May 3, 2019 Titan, Saturns largest moon, has a dense atmosphere, together with lakes and seas of liquid hydrocarbons. These liquid bodies, which are in polar regions and up to several hundred kilometres in diameter, generally have smooth sur- faces despite evidence of near-surface winds. Photochemically generated or- ganic aerosols form a haze that can settle and potentially interact with the liq- uid surface. Here we investigate the floatability of these aerosols on Titans seas and their potential to dampen waves. We find that the majority of aerosols are denser than the liquid hydrocarbons, but that some could have liquid-repelling 1 properties. From calculation of the capillary forces, we propose that these liq- uidophobic aerosols could float and form a persistent film on Titans seas. We numerically model the wave damping efficiency of such a film under the condi- tions on Titan, demonstrating that even a film one molecule thick may inhibit formation of waves larger than a few centimetres in wavelength. We conclude that the presence of a floating film of aerosols deposited on Titans lakes and seas could explain the remarkable smoothness of their surfaces. Titan, the main satellite of Saturn, is the only satellite of the solar system possessing a dense atmosphere. However, the most striking feature of Titan is perhaps the presence, of a thick layer of haze, source of inspiration of many works focused on photochemical products and aerosols properties1. Besides, the presence of oceans of liquid hydrocarbons was conceived in the early eighties, while Cassini orbiter has revealed a collection of seas and lakes in the polar regions of Titan3. These structures involve diameters up to more than several hundreds kilometers. Air- fall deposits of photochemically produced organics, may coat the surface of the seas. Potential interactions between haze particles and liquid surfaces are therefore likely. To understand the fate of the aerosols at the surface of Titan's lakes, a first parameter to be investigated is their floatability. This issue of flotability is neither anecdotal nor formal. During pre-Cassini era, in the different context of radar volume scattering, the near surface properties of Titan's ocean have already been considered2, in the scenario envisaged, particles or macromolecules were suspended in the liquid, maintained by some vertical currents. In the case where some mate- rial could accumulate at the sea surface over time, and build up a layer of some thickness, the presence of a such surface film can affect the gas, heat and momentum exchange between the sea of the atmosphere. On Earth, it is well known, for centuries, that an oil film damps the sea surface waves (Aristotle, Problematica Physica, 23, no. 38, and historical records mentioned in Ref. 4). At the surface of terrestrial oceans, biogenic surface microlayers appear due to se- 2 cretion of fish and plankton5, and they are detectable by satellite RADAR, through their waves damping effect6. Therefore, discussing the existence of a possible film, made of atmospheric products, covering some areas of Titan's sea surface, appears particularly relevant. During fly- bys other major seas and lakes7 -- 9, Cassini RADAR measurements have suggested mm-level flatness surfaces. Infrared ground imagery has also revealed very smooth surfaces for two of them10, 11. Even if infrared off-specular glints have been observed12, corresponding probably to rougher zones, Cassini observations draw a picture of impressively smooth marine surfaces. If the physical and/or chemical properties of some molecules or particles, may lead to the ap- pearance of a layer, or microlayer, at the surface of liquid formations, then the properties of the seas could be drastically affected. The question touches major properties of this unique case of "exo-oceanography". Material that can Sediment to the Surface of Titan To avoid any ambiguity, we reserve the term "aerosols" to solid particles that make the thick hazy layer of Titan. This photochemical haze extends roughly from the surface up to about 1000 km altitude13 and it is made of aggregates of monomers. Aggregates show a fractal structure (see Supplementary Fig. 1) and count up to several thousands of monomers14; each of them can be approximated by a sphere15. The monomers radius determinations, agree for a value around 50 nm16. Each Titan's haze model, based on a microphysical description, depends on the rate of particle production among several parameters17. The aerosols mass production, derived em- pirically, has its values17 spreading around ∼ 10−13 kg m−2 s−1. With the haze layers in steady state, the "mass production rate" is also the average "mass deposit rate" of aerosols, i.e. the sed- imentation rate over the surface of Titan. The adopted value corresponds, at ground level, to one nanometer per year, if we assume a density around 103 kg m−3. These organic particles should not be surrounded by liquid, in Titan's dry regions, while in the most humid ones, aerosols can 3 play the role of nucleation cores for liquid methane droplets formation18. Even if observational evidences for rainfalls are rare19; at Titan, the polar regions are recognized to be the wettest from climate simulations20. In these regions, the precipitations of liquid methane are governed by the presence of small particles (micronic or submicronic) known as cloud condensation nu- clei, aerosols are very good candidates for this role. In summary, a certain amount of aerosols should reach the seas as dry particles, whereas the remaining could get the sea embedded in liquid droplets. Titan's aerosols are the end-products of a complex chemistry, in which a plethora of small molecules is generated. Some of them are detected by spaceborne instruments21 or by Earth's telescopes22. On the theoretical side, models account for the production of these species23. Due to local thermodynamic conditions, some of these compounds can condensate to form ei- ther liquid droplets or ice crystals. For instance, the VIMS instrument aboard Cassini allowed the detection of micrometre-sized particles of frozen hydrogen cyanide (HCN ice) over Titans southern pole24. The mass flux, to the surface, of these compounds is not negligible. Models indicate a mass flux for HCN of the same order of magnitude as that of aerosols25. Hydro- gen cyanide is not the only species that can produce organic crystals in the atmosphere, among many others, molecules like C2H2 and HC3N have also a similar potential26. Nonetheless, HCN appears to be the most abundant. The key point for our purpose, is the potential propensity of these micron-sized crystals to aggregate with each other into, micron to millimeter-sized parti- cles, analogs of terrestrial snowflakes27. The physical properties of HCN do not preclude this type of process. In this perspective, Titan's troposphere could be the scene of "exotic snowfalls" composed of "HCN-flakes" (or C2H2-flakes, ...). Even if CO2 "snowfalls" has been considered in the case of Mars28, perhaps curiously this possibility has never be investigated from the point of view of microphysics in the context of Titan. 4 Finally, two Cassini instruments detected large molecules in Titan's thermosphere, with charge/mass ratios up to29 ∼ 10, 000. In addition, the presence of polycyclic aromatic hydrocarbons above an altitude of ∼ 900 km has been also suggested30. These facts plead in favor of the presence of large molecules at ground level, analogs of terrestrial surfactants31. In summary, we have determined three sources of material that can end at the surface of Ti- tan's hydrocarbon seas: (1) the haze particles, (2) crystallized organics, (3) large molecules, harboring at least one "liquidophobic" function. Existence and persistence of a floating film Two distinct effects may be invoked when the floatability of an object is questioned: (1) the Archimedes' buoyancy, (2) the effect of capillary processes. In an idealized case, the only relevant parameter is the density of monomers material, com- pared to the density of sea liquid. In the absence of any wetting effect the liquid penetrates in the whole aerosol free volume. In such a situation, the fractal structure of aerosols cannot be invoked to introduce an "effective density", lower than the monomers one. This is why we focus on the density of monomers. These latter are recognized to be formed by molecules harboring a large number of carbon and nitrogen atoms32. As a first guess, we can adopt Earth fossil carbon forms, like oil or bitumen, as analogs for monomers the organic matter. Since petroleum in- dustry products are made of complex mixtures of numerous species, their density is not unique, and for oil33 ranges between 0.8 and 0.95 g cm−3. During its descent to Titan's surface, the Huygens probe made a lot of measurements, and the ACP (Aerosol Collector and Pyrolyser)-GCMS (Gas Chromatograph and Mass Spectrom- eter) experiment analyzed the chemical composition of the collected aerosols14. Their nuclei were found to be made up of N-rich organics, without information about the molecular structure. 5 The best estimations of their composition and density are, to date, provided by Titan's aerosol laboratory analogues, named "tholins"34. The few available measurements may be classified into two categories: the high pressure experiments35 producing relatively light materials with a density around ∼ 0.8 g cm−3, and the low pressure simulations36, 37 leading to heavier products, with a mean density in the range 1.3 − 1.4 g cm−3. For low pressure measurements, individ- ual density determinations can be found down to35 0.4 g cm−3. Concerning "exotic snows", densities of solid organics that could be common at the surface of Titan, may be found in the literature38: for the most abundant, and less soluble, HCN, the value should be ∼ 1.03 g cm−3. Even if the chemical composition of Titan's seas is not known in details, there is a general consensus to consider that the main components are methane, ethane with some amounts of nitrogen39, 40. In Supplementary Table 1 we have gathered the densities of these species in con- ditions relevant for Titan's polar surface. A quick inspection of this table convinces that, as a general tendency, monomers density should remain above the expected value for the liquid. In such circumstances, the majority of aerosols particles, or exotic "snowflakes", should sink to the depths of hydrocarbons seas. This do not exclude the formation of a floating deposit, supported only by Archimedes buoyancy, and formed by the lightest in the mass spectrum. We turn now our attention to capillary processes. It is well known that small bodies heavier than the supporting liquid, including those made of iron, can float under the influence of the so- called capillary force. Even some animals, bugs of the family of the Gerridae (water striders) take advantage of this kind of force to survive at the surface of water41. As recalled in Method, the action of a liquid on a tiny object is a function of two parameters: (1) the surface tension σ (N m−1), an intrinsic property of the interface liquid-air; (2) the contact angle θc, which represents the interaction between the liquid and the material of the considered object. For 0o ≤ θc ≤ 90o the liquid is rather "attracted" by the solid material, thus called "liquidophillic" in this case. 6 When 90o ≤ θc ≤ 180o, the material is liquid-repellant and is named "liquidophobic". Clearly, an aerosol monomer, seen as a small sphere, may be maintained at the surface solely if the monomer is made of "liquidophobic" matter. A simple derivation (see Methods), based on a balance between weight and capillarity forces, leads to the layer thickness e ≃ 3 σ cos θc r gTit ρmono (1) In the case of a perfect non-wetting liquid (i.e. for θc = 180o), a numerical estimate can be obtained for e, assuming a surface tension σ fixed to 2 × 10−2 N m−1 (see Supplementary Table 2), a radius of monomers at 50 nm and taking ρmono = 800 kg m−3, we found e ≃ 500 m. Such an unrealistically large value is the signature of the existence of strong limiting factors. The most obvious limitations are the aerosols sedimentation rate representing a few nanometer per year, and the idealized poor wettability. This, then, leads naturally to a discussion concerning expected contact angles. On Titan, maritime surfaces are not the unique place for possible interaction between liquids and solid particles. This kind of interaction is known to play a crucial role in the formation of cloud particles, which are generated by heterogeneous nucleation42. On the Earth, heteroge- neous nucleation on micronic and sub-micronic aerosols is the dominant mechanism in forming liquid cloud droplets. In the context of Titan, given the large abundance of aerosols, a similar microphysics has been proposed for the nucleation of liquid methane droplets or small ethane crystals43. In these approaches, the contact angle plays a key role that can be easily understood: the more aerosols are wettable, the more the liquid can spread over its surface and favor the formation of a liquid "envelope". Unfortunately, contact angles are very unconstrained param- eters44, what can be found in the literature is either not perfectly relevant15, 45, or comes from informal personal communication43. Except concerning the nucleation of solid butane, we did not find proper peer reviewed publications providing θc values for nucleation onto "tholins". 7 Then, cloud formation models include values close to43 θc ≃ 0o and which obviously favors the formation of droplets onto organic aerosols particles. In other words, microphysics models of clouds assume the existence of "liquidophilic" aerosols to play the role of condensation nuclei, whereas "liquidophobic" particles are required to form a floating layer over Titan's lakes sur- face. Let us now examine if some clues can be found about the wettability of aerosols. The actual chemical composition of Titan's aerosols is not known. Many teams published the global stoi- chiometry CxHyNz of "tholins", which spectral signature is compatible with what is observed at the Saturn moon. Nonetheless, spectroscopy is not sensitive neither to the detailed chemical composition nor to the exact composition and physical state of aerosol surface. These surface properties determine the "liquidophilic" or "liquidophobic" character of aerosols46. During their fall to Titan's ground, particles may also undergo a variety of alterations, due to charging, photolysis or radiolysis26, 47. This "aging", changes the surface properties of aerosols. This way, their wettability may evolve before they get to the sea surface. Laboratory measurements show a very low solubility of tholins in non-polar solvents48. An high solubility is generally recog- nized to be associated with a liquidophilic character. Thus, the low solubility of tholins may be regarded as an indication of a liquidophobia. Similarly, HCN snow may also float due to a strong liquidophobic properties. Considering the very likely existence of a rich variety of aerosol surface properties, we propose the presence of both "liquidophilic" and "liquidophobic" aerosols in Titan's atmosphere. The first family of particles sink when they reach the maritime surface, even if they arrive in a "dry state". The particles belonging to the second category do not participate to the clouds formation and float when touching the surface of the sea. These kind of particles are good candidates for building up a more or less thick layer at the surface of hydrocarbons seas. In a sense, the surface of Titan's lakes/seas could retain liquidophobic material. 8 Since the precipitation rates of atmospheric products are small, one might wonder how an organic microlayer can build up and be maintained, rather than being destroyed by weather- ing. Titan surface is a dynamic environment: wind, rain, fluvial runoff or tides could impede such a formation. It is well known that saltation of particles is much more difficult from a wet substrate rather than from a dry surface49. Thus, if lands surrounding seas are wet, the wind should let organic dusty material lieing down over these terrains, and similarly should not rip marine floating film. On the contrary, if polar lands are dry, the saltation should be easy and the wind could transport material to sea surface, which could behave as a "wet trap", leading to an accumulation process. Methane rain droplets, or nitrogen bubbles coming from seabed39, 50, may locally disrupt the layer. Thanks to basics physics laws, the momenta associated to the impact of such objects can be estimated respectively to 5 × 10−2 kg m s−1 and 1 kg m s−1, revealing that bubbles could be more efficient than rain droplets. Nevertheless, more specific conclusion cannot be drawn since the mechanical properties of films are not known. But, bub- bles and droplets have a significant difference: droplets bring to seas material washed up along their fall. Indeed, droplets transport solid particles on which they have nucleated. If rainfalls are heavy enough, fluvial run-off can also favor the appearance of surface layers, by transporting material from lands to seas. Finally, according to numerical simulations51, Titan's seas undergo a moderate tidal activity. Except along the shores, where material could be periodically de- posited and returned to the liquid, the tides should not alter any large film due to their large "wavelength". Nevertheless, relatively strong tidal currents, through the straits may generate some wave fields52. If lakes behave like a trap, an almost continuous shore-to-shore deposit can be expected. In the contrary, where only a partial coverage is at work, this aspect could introduce some tempo- ral and spatial variabilities in surface properties. Even if it seems difficult to destroy floating 9 layers by wind, rain, run-off or tides, these effects could induce migration or fragmentation of slicks. Intrinsic properties of floating material could also induce some evolution. Floating small objects53 can make large structures by self-assembly processes driven by lateral capillary in- teractions. Finally, we stress that observations of specular reflections54 over lakes is consistent with a partial and evolving film coverage. In the case of a shore-to-shore slick, a large range of refractive index values is compatible with glint observations, for which the photon fluxes are uncertain due to the lack of knowledge about the optical thickness of the hazy cap. Damping of Sea Waves by Surface Films The first fully satisfying theoretical explanation has been published in the sixties55. For a monomolecular film, the damping of a wave of initial amplitude a0, after a propagation along a distance x can be written a(x) = a0 exp −∆ x (2) with ∆ (m−1), the damping coefficient, which depends on the wavelength λ. A "clean surface", i.e. free of slick, has a damping coefficient noted ∆0 (see Method). In order to characterize the damping effect of a supernatant film, it is usual to introduce the relative damping ratio defined as4 y(λ) = ∆/∆0 (3) This ratio depends on the intrinsic mechanical properties of the surface slick, which are rep- resented by E0 (N m−1) the modulus of its coefficient of elasticity, and ωd (s−1) a parameter accounting for the relaxation time of the layer (see Method for details). In Fig. 1, we have reported the variations of y(λ), employing values representative for monomolecular films (e.g for hexadecanoic acid methyl ester E0 = 4.5 × 10−2 N m−1 and ωd = 22 rad s−1, while for oleic acid E0 = 1.4 × 10−2 N m−1 and ωd = 38 rad s−1). Not surprisingly, large viscoelastic 10 modulii E0 produce a strong damping effect; whereas, long relaxation times (i.e. low frequen- cies ωd) lead to efficient damping. If we compare the Earth and Titan, the general tendency is at least a similar damping effect at short wavelengths, and a much stronger effect at longer wavelengths. The properties of the sea liquid have also their influence on the waves formation. Except the surface tension σ (see Supplementary Fig. 2), which has a minor influence, all the other parameters tend to enhance the wave damping at Titan. Undoubtedly, the sea viscosity ν has the strongest effect by, in our example, multiplying the value of y by a factor of ∼ 4; corre- sponding to a factor of ∼ exp 4 ≃ 55 on the wave amplitude. According to this first approach, Titan seems to be more favorable for a wave damping caused by a monomolecular film, than the Earth, because liquid hydrocarbons have a density and a viscosity smaller than that of liquid water. A monomolecular film is the thinnest blanket that one could imagine, but thicker deposits are also conceivable. A formalism, specifically adapted to these finite-thickness layers, have also been developed (see Method). In that more general frame, the relative damping ratio y depends explicitly on the slick thickness d, and it firmly increases when d becomes larger. Essentially, results obtained with monomolecular films remain valid with thicker ones. Common observations, and numerous academic studies, show that winds, blowing over water, are found to result in the birth and growth of waves upon sea surface56. The global picture of waves generation can be divided into three physical processes. First turbulence in the wind pro- duces random stress variations on the surface. These pressure and tangential shear fluctuations give rise to small wavelets, due to resonances in the wind-sea coupling57, 58. Secondly the waves amplitude is reinforced by the air flow, the pressure being maximum on the windward side of the crest and minimum on the leeward side58. Finally the waves start to interact each with other, exciting longer wavelength modes56. Many effects conspire to limit the wave growth in height and wavelength. For instance, the fetch length over which the wind blows and the so-called 11 "whitecapping", affect the final spectrum of waves56. It is worth noting that without the generation of the very first ripples, due to air turbulent eddies near the surface, the large waves cannot be produced, and the surface of the ocean would re- main mirror-smooth. An estimation, for the wavelength λr, of these initial wavelets caused by resonances is given by57 λr = 2πs σ ρg (4) with the notation already adopted in previous paragraphs. In the context of the Earth, this equa- tion leads to λr ≃ 1.7 cm, whereas a transposition to Titan yields to a similar value λr ≃ 3.4 cm. Our discussion about the damping rate of waves, indicates that a very strong damping could occur at Titan mares, with a maximum efficiency around a wavelength of a few centimeters (see Fig. 3), depending on the nature and actual properties of the floating deposit. Therefore, if the surface of a Titan sea was covered, at least partially, by such a film/slick; the onset of wave formation could be impeded, leading to the non-existence of waves at all larger wavelengths, in the corresponding regions. Compatibility of a strong wave damping with observations In this paper, we have considered the massive presence of aerosols, an other organic products (large molecules, HCN crystals/snow flakes, ...), in the atmosphere of Titan; that sediment to the ground where in polar regions hydrocarbon seas and lakes are observed. The formation of a more or less thin deposit at the sea surface appears to be plausible. As already mentioned, the off-specular infrared observations may not be in conflict with this scenario: (1) the deposit may be patchy, letting free liquid being wavy, and/or (2) the floating layer may itself produce these "reflections", if the local deposit has a king of "roughness" at infrared wavelengths. Recently, a mechanism has been proposed to explain the occurrence of efficient RADAR reflectors at Ligeia Mare, one of the main Titan's seas59, 60. These, so-called, "Magic Islands" could be produced 12 by streams of nitrogen bubbles rising from the sea depths39, 50. This scenario is not in conflict with the existence of a thin film at the sea surface: bubbles arriving at the surface could locally break the layer, and, in the same time, the RADAR-waves transparency of that slick could not prevent the observation of bubbles still in the volume of the sea liquid, as it is proposed. This work has strongly highlighted the need for laboratory studies of interactions between cryogenic liquids, relevant for Titan, and tholins. Particularly, reliable contact angles deter- minations are fundamental for the behavior of hydrocarbon seas together with nucleation of liquid droplets within atmospheric microphysics processes. This new class of experimentations includes studies of surfaces states and compositions of tholins particles. As an extension of preliminary works61, wind-tunnels may also be used, at room temperature, with liquids and fine particles or floating films, analog to what is expected on Titan. Given its, potentially crucial, role in the carbon cycle, floating film/slick could be an important target for possible future in situ explorations62. And, much more speculatively, it could harbor an original "exobiological" activity. References 1. S. M. Horst, "Titan's atmosphere and climate," J. Geophys. Res. 122, 432 -- 482 (2017). 2. J. I. Lunine, Symposium on Titan, B. Kaldeich, ed. (1992), vol. 338 of ESA Special Publi- cation, pp. 233 -- 239. 3. E. R. Stofan, C. Elachi, J. I. Lunine, R. D. Lorenz, B. Stiles, K. L. Mitchell, S. Ostro, L. Soderblom, C. Wood, H. Zebker, S. Wall, M. Janssen, R. Kirk, R. Lopes, F. Paganelli, J. Radebaugh, L. Wye, Y. Anderson, M. Allison, R. Boehmer, P. Callahan, P. Encrenaz, 13 E. Flamini, G. Francescetti, Y. Gim, G. Hamilton, S. Hensley, W. T. K. Johnson, K. Kelle- her, D. Muhleman, P. Paillou, G. Picardi, F. Posa, L. Roth, R. Seu, S. Shaffer, S. Vetrella, and R. West, "The lakes of Titan," Nature 445, 61 -- 64 (2007). 4. W. Alpers and Huhnerfuss, "The damping of Ocean Waves by Surface Films: A New Look at an Old Problem," J. Geophys. Res. 94, 6251 -- 6265 (1989). 5. C. Lancelot and S. Mathot, "Dynamics of a Phaeocystis-dominated spring bloom in Belgian coastal waters. I. Phytoplanktonic activities and related parameters," Mar. Ecol. Prog. Ser. 37, 239 -- 248 (1987). 6. I. I. Lin, W. Alpers, and W. T. Liu, "First evidence for the detection of natural surface films by the QuickSCAT scatterometer," Geophys. Res. Lett. 30, 1713 (2003). 7. L. C. Wye, H. A. Zebker, and R. D. Lorenz, "Smoothness of Titan's Ontario Lacus: Con- straints from Cassini RADAR specular reflection data," Geophys. Res. Lett. 36, L16201 (2009). 8. H. Zebker, A. Hayes, M. Janssen, A. Le Gall, R. Lorenz, and L. Wye, "Surface of Ligeia Mare, Titan, from Cassini altimeter and radiometer analysis," Geophys. Res. Lett. 41, 308 -- 313 (2014). 9. C. Grima, M. Mastrogiuseppe, A. G. Hayes, S. D. Wall, R. D. Lorenz, J. D. Hofgartner, B. Stiles, C. Elachi, and Cassini Radar Team, "Surface roughness of Titan's hydrocarbon seas," Earth Planet. Sci. Lett. 474, 20 -- 24 (2017). 10. K. Stephan, R. Jaumann, R. H. Brown, J. M. Soderblom, L. A. Soderblom, J. W. Barnes, C. Sotin, C. A. Griffith, R. L. Kirk, K. H. Baines, B. J. Buratti, R. N. Clark, D. M. Lytle, R. M. Nelson, and P. D. Nicholson, "Specular reflection on Titan: Liquids in Kraken Mare," Geophys. Res. Lett. 37, L07104 (2010). 14 11. J. W. Barnes, J. M. Soderblom, R. H. Brown, L. A. Soderblom, K. Stephan, R. Jaumann, S. L. Mou´elic, S. Rodriguez, C. Sotin, B. J. Buratti, K. H. Baines, R. N. Clark, and P. D. Nicholson, "Wave constraints for Titan's Jingpo Lacus and Kraken Mare from VIMS spec- ular reflection lightcurves," Icarus 211, 722 -- 731 (2011). 12. J. W. Barnes, C. Sotin, J. M. Soderblom, R. H. Brown, A. G. Hayes, M. Donelan, S. Ro- driguez, S. Le Mouelic, K. H. Baines, and T. B. McCord, "Cassini/VIMS Observes Rough Surfaces on Titan's Punga Mare in Specular Reflection," Planet. Sci., vol. 3, p. 3, Aug. 2014. 13. R. A. West, P. Lavvas, C. Anderson, and H. Imanaka, Titans haze, ch. 8, pp. 285 -- 321. New York: Cambridge University Press (2014). 14. M. G. Tomasko, L. Doose, S. Engel, L. E. Dafoe, R. West, M. Lemmon, E. Karkoschka, and C. See, "A model of Titan's aerosols based on measurements made inside the atmosphere," Planet. Space Sci. 56, 669 -- 707 (2008). 15. D. B. Curtis, C. D. Hatch, C. A. Hasenkopf, O. B. Toon, M. A. Tolbert, C. P. McKay, and B. N. Khare, "Laboratory studies of methane and ethane adsorption and nucleation onto organic particles: Application to Titan's clouds," Icarus 195, 792 -- 801 (2008). 16. B. Seignovert, P. Rannou, P. Lavvas, T. Cours, and R. A. West, "Aerosols optical properties in Titan's detached haze layer before the equinox," Icarus 292, 13 -- 21 (2017). 17. P. Rannou, F. Hourdin, C. P. McKay, and D. Luz, "A coupled dynamics-microphysics model of Titan's atmosphere," Icarus 170, 443 -- 462 (2004). 18. P. Rannou, F. Montmessin, F. Hourdin, and S. Lebonnois, "The Latitudinal Distribution of Clouds on Titan," Sci 311, 201 -- 205 (2006). 15 19. E. P. Turtle, J. E. Perry, A. G. Hayes, R. D. Lorenz, J. W. Barnes, A. S. McEwen, R. A. West, A. D. Del Genio, J. M. Barbara, J. I. Lunine, E. L. Schaller, T. L. Ray, R. M. C. Lopes, and E. R. Stofan, "Rapid and Extensive Surface Changes Near Titan's Equator: Evidence of April Showers," Sci 331, 1414 -- 1417 (2011). 20. T. Schneider, S. D. B. Graves, E. L. Schaller, and M. E. Brown, "Polar methane accumu- lation and rainstorms on Titan from simulations of the methane cycle," Nature 481, 58 -- 61 (2012). 21. A. Coustenis, D. E. Jennings, R. K. Achterberg, G. Bampasidis, P. Lavvas, C. A. Nixon, N. A. Teanby, C. M. Anderson, V. Cottini, and F. M. Flasar, "Titan's temporal evolution in stratospheric trace gases near the poles," Icarus 270, 409 -- 420 (2016). 22. E. M. Molter, C. A. Nixon, M. A. Cordiner, J. Serigano, P. G. J. Irwin, N. A. Teanby, S. B. Charnley, and J. E. Lindberg, "ALMA Observations of HCN and Its Isotopologues on Titan," AJ 152, 42 (2016). 23. V. A. Krasnopolsky, "Chemical composition of Titan's atmosphere and ionosphere: Obser- vations and the photochemical model," Icarus 236, 83 -- 91 (2014). 24. R. J. de Kok, N. A. Teanby, L. Maltagliati, P. G. J. Irwin, and S. Vinatier, "HCN ice in Titan's high-altitude southern polar cloud," Nature 514, 65 -- 67 (2014). 25. P. Lavvas, C. A. Griffith, and R. V. Yelle, "Condensation in Titan's atmosphere at the Huy- gens landing site," Icarus 215, 732 -- 750 (2011). 26. I. Couturier-Tamburelli, N. Pi´etri, V. Le Letty, T. Chiavassa, and M. Gudipati, "UV -- Vis Light-induced Aging of Titan's Haze and Ice," ApJ 852, 117 (2018). 16 27. P. V. Hobbs, S. Chang, and J. D. Locatelli, "The Dimensions and Aggregation of the Ice Crystals in Natural Clouds," J. Geophys. Res. 79, 2199 -- 2206 (1974). 28. F. Forget, F. Hourdin, and O. Talagrand, "CO 2 Snowfall on Mars: Simulation with a Gen- eral Circulation Model," Icarus 131, 302 -- 316 (1998). 29. J. H. Waite, D. T. Young, T. E. Cravens, A. J. Coates, F. J. Crary, B. Magee, and J. Westlake, "The Process of Tholin Formation in Titan's Upper Atmosphere," Sci 316, 870 (2007). 30. M. L´opez-Puertas, B. M. Dinelli, A. Adriani, B. Funke, M. Garc´ıa-Comas, M. L. Mori- coni, E. D'Aversa, C. Boersma, and L. J. Allamandola, "Large Abundances of Polycyclic Aromatic Hydrocarbons in Titan's Upper Atmosphere," ApJ 770, 132 (2013). 31. J. Stevenson, J. Lunine, and P. Clancy, "Membrane alternatives in worlds without oxygen: Creation of an azotosome," Sci. Adv. 1, (2015). 32. T. Gautier, N. Carrasco, I. Schmitz-Afonso, D. Touboul, C. Szopa, A. Buch, and P. Pernot, "Nitrogen incorporation in Titan's tholins inferred by high resolution orbitrap mass spec- trometry and gas chromatography-mass spectrometry," Earth Planet. Sci. Lett. 404, 33 -- 42 (2014). 33. J. Ancheyta and J. G. Speight, Hydroprocessing of Heavy Oils and Residua. CRC Press, 1st edition ed. (2007). 34. C. Sagan, W. R. Thomson, and B. Khare, "Titan: A laboratory for pre-biological organic chemistry," Acc. Chem. Res. 25, 286 -- 292 (1992). 35. S. M. Horst and M. A. Tolbert, "In Situ Measurements of the Size and Density of Titan Aerosol Analogs," ApJL 770, L10 (2013). 17 36. H. Imanaka, D. P. Cruikshank, B. N. Khare, and C. P. McKay, "Optical constants of Titan tholins at mid-infrared wavelengths (2.5-25 µm) and the possible chemical nature of Titan's haze particles," Icarus 218, 247 -- 261 (2012). 37. Y. Brouet, A. C. Levasseur-Regourd, P. Sabouroux, L. Neves, P. Encrenaz, O. Poch, A. Pommerol, N. Thomas, W. Kofman, A. Le Gall, V. Ciarletti, A. H´erique, A. Lethuil- lier, N. Carrasco, and C. Szopa, "A porosity gradient in 67P/C-G nucleus suggested from CONSERT and SESAME-PP results: an interpretation based on new laboratory permittiv- ity measurements of porous icy analogues," MNRAS 462, S89 -- S98 (2016). 38. D. Cordier, T. Cornet, J. W. Barnes, S. M. MacKenzie, T. Le Bahers, D. Nna Mvondo, and A. G. Ferreira, "Structure of titan's evaporites," Icarus 270, 41 -- 56 (2016). 39. D. Cordier, F. Garc´ıa-S´anchez, D. N. Justo-Garc´ıa, and G. Liger-Belair, "Bubble streams in Titan's seas as a product of liquid N2 + CH4 + C2H6 cryogenic mixture," Nat. Astron. 1, 0102 (2017). 40. A. Le Gall, M. J. Malaska, R. D. Lorenz, M. A. Janssen, T. Tokano, A. G. Hayes, M. Mas- trogiuseppe, J. I. Lunine, G. Veyssi`ere, P. Encrenaz, and O. Karatekin, "Composition, sea- sonal change, and bathymetry of Ligeia Mare, Titan, derived from its microwave thermal emission," J. Geophys. Res. 121, 233 -- 251 (2016). 41. X. Gao and L. Jiang, "Biophysics: Water-repellent legs of water striders," Nature 432, 36 (2004). 42. A. S´anchez-Lavega, An Introduction to Planetary Atmospheres. CRC Press (2010). 43. E. L. Barth and O. B. Toon, "Methane, ethane, and mixed clouds in Titan's atmosphere: Properties derived from microphysical modeling," Icarus 182, 230 -- 250 (2006). 18 44. S. Rodriguez, P. Paillou, M. Dobrijevic, G. Ruffi´e, P. Coll, J. M. Bernard, and P. Encre- naz, "Impact of aerosols present in Titan's atmosphere on the CASSINI radar experiment," Icarus 164, 213 -- 227 (2003). 45. D. B. Curtis, O. B. Toon, M. A. Tolbert, C. P. McKay, and B. N. Khare, "Laboratory studies of butane nucleation on organic haze particles: Application to Titan's clouds," Icarus 109, 1382 -- 1390 (2005). 46. H. R. Pruppacher and J. D. Klett, Microphysics of Clouds and Precipitation. Dordrecht: Reidel (1978). 47. R. Courtin, S. J. Kim, and A. Bar-Nun, "Three-micron extinction of the Titan haze in the 250700 km altitude range: Possible evidence of a particle-aging process," A&A 573, A21 (2015). 48. N. Carrasco, I. Schmitz-Afonso, J.-Y. Bonnet, E. Quirico, R. Thissen, O. Dutuit, A. Bagag, O. Lapr´evote, A. Buch, A. Giulani, G. Adand´e, F. Ouni, E. Hadamcik, C. Szopa, and G. Cernogora, "Chemical Characterization of Titans Tholins: Solubility, Morphology and Molecular Structure Revisited," J. Phys. Chem. A 113, 11195 -- 11203 (2009). PMID: 19827851. 49. R. D. Lorenz, "Physics of saltation and sand transport on Titan: A brief review," Icarus 230, 162 -- 167 (2014). 50. D. Cordier and G. Liger-Belair, "Bubbles in Titans Seas: Nucleation, Growth, and RADAR Signature," ApJ 859, 26 (2018). 51. D. Vincent, K. Ozgur, V. Vallaeys, A. G. Hayes, M. Mastrogiuseppe, C. Notarnicola, V. De- hant, and E. Deleersnijder, "Numerical study of tides in Ontario Lacus, a hydrocarbon lake on the surface of the Saturnian moon Titan," Ocean Dynamics 66, 461 -- 482 (2016). 19 52. N. Kurata, K. Vella, B. Hamilton, M. Shivji, A. Soloviev, S. Matt, A. Tartar, and W. Perrie Sci. Rep. 6, 19123 (2016). 53. G. M. Whitesides and M. Boncheva, "Beyond molecules: Self-assembly of mesoscopic and macroscopic components," PNAS 99, 4769 -- 4774 (2002). 54. J. M. Soderblom, J. W. Barnes, L. A. Soderblom, R. H. Brown, C. A. Griffith, P. D. Nichol- son, K. Stephan, R. Jaumann, C. Sotin, K. H. Baines, B. J. Buratti, and R. N. Clark, "Mod- eling specular reflections from hydrocarbon lakes on Titan," Icarus 220, 744 -- 751 (2012). 55. M. van den Tempel and R. P. van de Riet J. Chem. Phys. 42, 2769 (1965). 56. G. J. Komen, L. Cavaleri, M. Donelan, K. Hasselmann, S. Hasselmann, and P. A. E. M. Janssen, Dynamics and Modelling of Ocean Waves. Cambridge, UK: Cambridge University Press, 1st ed. (1994). 57. F. T. Phillips, "On the generation of waves by turbulent wind," J. Fluid Mech. 2, 417 -- 445 (1957). 58. J. W. Miles, "On the generation of the surface waves by shear flows," J. Fluid Mech. 3, 185 -- 204 (1957). 59. J. D. Hofgartner, A. G. Hayes, J. I. Lunine, H. Zebker, B. W. Stiles, C. Sotin, J. W. Barnes, E. P. Turtle, K. H. Baines, R. H. Brown, B. J. Buratti, R. N. Clarck, P. Encrenaz, R. D. Kirk, A. Le Gall, R. M. Lopes, R. D. Lorenz, M. J. Malaska, K. L. Mitchell, P. Nicholson, P. D. Paillou, J. Radebaugh, S. D. Wall, and C. Wood, "Transient features in a Titan sea," Nat. Geosci. 7, 493 -- 496 (2014). 20 60. J. D. Hofgartner, A. G. Hayes, J. I. Lunine, H. Zebker, R. D. Lorenz, M. J. Malaska, M. Mastrogiuseppe, C. Notarnicola, and J. M. Soderblom, "Titan's "Magic Islands": Tran- sient features in a hydrocarbon sea," Icarus 271, 338 -- 349 (2016). 61. R. D. Lorenz, E. R. Kraal, E. E. Eddlemon, J. Cheney, and R. Greeley, "Sea-surface wave growth under extraterrestrial atmospheres: Preliminary wind tunnel experiments with ap- plication to Mars and Titan," Icarus 175, 556 -- 560 (2005). 62. J. W. Hartwig, A. Colozza, R. D. Lorenz, S. Oleson, G. Landis, P. Schmitz, M. Paul, and J. Walsh, "Exploring the depths of Kraken Mare - Power, thermal analysis, and ballast control for the Saturn Titan submarine," Cryogenics 74, 31 -- 46 (2016). Additional information Supplementary information is available in the on-line version of the paper. Reprints and per- missions information is available on-line at www.nature.com/reprints. Publishers note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and in- stitutional affiliations. Correspondence and requests for materials should be addressed to D.C., e-mail: [email protected] Acknowledgements The authors thanks the anonymous Reviewers who help them to improve the clarity and scien- tific significance of their work. N. Carrasco thank the European Research Council for funding via the ERC PrimChem project (grant agreement No. 636829). This work was also supported by the Programme National de Plan´etologie (PNP) of CNRS-INSU co-funded by CNES. The authors warmly thank S´ebastien Lebonnois, Jan Vatant d'Ollone, Tetsuya Tokano and Benjamin Charnay for fruitful scientific discussions. 21 Author contributions D. Cordier wrote the paper and performed numerical simulations, N. Carrasco provided exper- tise concerning the properties of Titan's aerosols. Competing interests The authors declare no competing financial interests. 22 Methods Interaction between a monomer and the sea liquid In Fig. 2 we recall some basics about the capillarity forces acting on a tiny object. If the latter is assimilated to a partially immersed sphere, each elementary portion dl (m) of the float line undergoes an elementary force, which vertical component is given by dfcap = σ cos θc dl (5) where σ is the surface tension of the liquid (N m−1), and θc is the contact angle. The surface ten- sion is an intrinsic property of the liquid, it depends on thermodynamic conditions together with the chemical composition of the liquid. The contact angle accounts for the interactions between the liquid and the solid substrate. The range 0o ≤ θc ≤ 90o corresponds to high wettabilities: the solid is "liquidophilic". The liquid has the tendency to spread out over the entire surface of the solid, for a perfect wetting θc = 0o. A "liquidophobic" liquid, i.e. presenting a low wettability, leads to 90o ≤ θc ≤ 180o. The perfectly non-wetting case occurs when θc = 180o. On a flat surface the liquid tends to form spherical droplets. Figure 2 describes the interaction between an idealized spherical monomer and the liquid phase of a Titan's sea. Clearly, when 0o ≤ θc < 90o the monomer is "attracted" by the liquid and consequently sinks into the liquid (see Fig. 2b). In the contrary, if 90o < θc ≤ 180o a vertical force can balance the effect of grav- ity, this force is at maximum when the monomer is perfectly non-wettable (Fig. 2a: θc = 180o). The contact angles are very unconstrained parameters, as discussed in the main text. For a single monomer, the resulting vertical force fcap (N), produced by capillarity, may ap- proximated by fcap ≃ 2πrσ cos θc (6) where r (m) represents the radius of the monomer. This expression assumes that the plane of flotation contains the center of the monomer, which is not necessary the case. Furthermore, 23 real monomers are built from clusters of large organic molecules and their shape are certainly not perfectly spherical ; therefore equation (6) has to be understood as an approximation. Un- ambiguously, the floatability of the considered monomer is governed by the value of the angle of contact: if the wettability of the monomer is high (i.e. for 0o ≤ θc ≤ 90o) the capillarity force is pointing down and the monomer sinks. In the opposite, for low wettability (i.e. for 90o ≤ θc ≤ 180o) the capillarity force has a vertical ascending component, that can counterbal- ance the weight of the object. Maximum thickness of a slick supported by capillarity By definition, the first layer of aerosols deposited at the surface of a Titan's sea is in contact with the liquid by monomers at the lowest positions. We denote N ∗ (N m−2) the number, per unit of surface of these monomers located right at the surface of the liquid. If p is the "porosity" of aerosols, then N ∗ can be estimated by N ∗ ∼ (1 − p) × 1 s (7) where s (m2) is the cross-section of a individual monomer, specifically s ∼ πr2. Thus, the force F ∗ cap (N m−2) per unit of surface, due to the effect of surface tension and acting on the aerosols layer covering the sea, can be derived F ∗ cap ∼ 2(1 − p)σ cos θc r (8) M ∗ aero (kg m−2) is the mass, per unit of surface, of the aerosols deposited at the surface of the liquid. It can be expressed as M ∗ aero ∼ N ∗n∗ 4 3 πr3 ρmono (9) with n∗ (m−1) the mean number of monomers, per unit of length, along the vertical axis. The average density of the monomers is noted ρmono (kg m−3). An estimation of n∗ can be made by 24 adopting n∗ ∼ e/2r with e (m) the total thickness of the aerosol slick. The weight, per unit of surface, supported by the liquid, is then P ∗ aero = M ∗ aerogTit (N m−2), where gTit stands for the gravity. With some algebra, we get P ∗ aero ∼ 2 3 (1 − p) e ρmono gTit By equalizing Eq. (8) and Eq. (10), we can extract the thickness e ∼ 3 σ cos θc r gTit ρmono (10) (11) As it could be expected, high surface tension and small monomers favor thick aerosols deposits, while high gravity together with large densities decrease the value of e. Perhaps more surpris- ingly, this result does not depend on the "porosity" of the aerosols. This behavior is easily explained: both the layer weight, and the surface capillary force, are proportional to (1 − p). The wave relative damping ratio For a surface, free of floating material, the damping coefficient ∆0 is given by the Stokes' equation4 ∆0 = 4k2ηω ρg + 3σk2 (12) where k (m−1) is the wavenumber related to the wavelength λ by the well known equation k = 2π/λ. The liquid density, the dynamic viscosity and surface tension are denoted respectively ρ (kg m−3), η (Pa s) and σ (N m−1). The planet gravity is represented by g (m s−2), whereas ω (s−1) is the angular frequency. In the present context, all these quantities are also constrained by the dispersion relation for gravity-capillary waves ω2 = gk + σ ρ k3 (13) Together with ∆, the actual damping coefficient of the considered sea surface, ∆0 appears in the relative damping ratio defined as4 y(λ) = ∆/∆0 25 (14) The proper mechanical properties of a surface film are a function of its complex viscoelastic modulus E (N m−1). Its real and imaginary part, respectively denoted Er and Ei, as a function of the radial frequency ω (s−1), are given by63, 64 and Er = E0 1 + (ωd/2ω)1/2 1 + ωd/ω + (2ωd/ω)1/2 Ei = E0 (ωd/2ω)1/2 1 + ωd/ω + (2ωd/ω)1/2 (15) (16) where E0 is the coefficient of elasticity in compression of the film and ωd a parameter which accounts for the relaxation time of the layer. The relative damping ratio is finally expressed as64 r + E2 E2 i 4(21/2) a2k3 + Ei 4 ak2 1 + (E2 i )a2k2 − (21/2)(Er − Ei)ak 1 − Er − Ei 21/2 ak + r + E2 (17) y = with a = k3/ρω2. The effect of a finite-thickness, i.e. non-monomolecular, surface deposit has been investigated in the context of heavy fuel slicks and greace ice65, 66. Up to a thickness of ∼ 10 µm, the computed damping rate does not differ essentially from what we get with zero thickness. In the frame of the formalism, relevant for finite-thickness film, the properties of such a film is no longer represented by only two parameters. Instead, more specific quantities are introduced: the kinematic viscosities of the film ν+ and that of the bulk liquid ν− (m2 s−1), their respective densities ρ+ and ρ− (kg m−3); together with surface and interfacial tensions. The thickness d (m) is also taken into account. The model parametrization is summarized in Supplementary Fig. 3. The behavior of an oil film over water at the Earth surface is illustrated by the damping ratio y for d = 0.01 mm (i.e. 10 µm) in Fig 3 . This curves has to be understood as the reference for what would be produced by a quasi-monomolecular film66. At all wavelengths, an increase of d leads to a substantially larger relative damping ratio y; the conclusions of our previous discussion, developed using the simple monomolecular formalism, remain valid and 26 the increase factor of y, found in this frame, must be taken as a minimum value. Although totally unknown, the intrinsic properties of a possible slick deposited at a Titan's sea surface, should play an important role. For instance, a high value for the kinematic viscosity ν+ yields naturally to a much more efficient damping. In fig. 3, the consequences of a ten times more viscous "oil" can be seen for a d = 5 mm thick layer. As it could be expected, the waves damping appears much stronger, particularly at long wavelengths. References 63. J. Lucassen and D. Giles J. Chem. Soc. Faraday Trans. 71, 217 (1975). 64. R. Cini, "Damping Effect of Monolayers on Surface Wave Motion in a Liquid," J. Colloid Interface Sci. 65, 387 -- 389 (1978). 65. J. E. Weber, "Wave Attenuation and Wave Drift in the Marginal Ice Zone," J. Phys. Oceanogr. 7, 2351 (1987). 66. D. Jenkins and S. J. Jacobs, "Wave damping by a thin layer of viscous fluid," Phys. Fluids 9, 1256 (1997). Code and Data availability Computer codes and data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request. 27 120 100 80 60 40 20 y o i t a r g n i p m a D ω d= 0.1 s-1 Titan (a) Earth 120 100 80 60 40 20 ω d= 10 s-1 (b) Titan Earth 120 100 80 60 40 20 ω d= 1000 s-1 (c) Titan Earth 0 0.001 0.01 0.1 λ (m) 1 0 0.001 10 0.01 0.1 λ (m) 1 0 0.001 10 0.01 1 10 0.1 λ (m) Figure 1 Comparison of the wave damping efficiency, due to a floating film, between Titan con- text and under Earth conditions. The wave relative damping ratio y caused by a monomolecular film deposited over the surface of a liquid is a function of the wavelength λ. The three panels correspond to different values of ωd which account for the relaxation time of the material form- ing the slick. In each panel, three values (i.e. 36.5, 21.3 and 7.3 in 10−3 N m−1, respectively in solid, dashed and dot-dashed lines) are considered for the coefficient of elasticity in compres- sion E0 of the film. The parameters concerning the Earth (blue lines) are the gravity g = 9.81 m s−2, the surface tension σ = 73 mN m−1, the viscosity η = 10−3 Pa s, and the density ρ = 103 kg m−3, value relevant for liquid water. In the case of Titan (red lines), we took g = 1.352 m s−2 for the gravity, and values expected for liquid methane: σ = 2 × 10−2 N m−1, η = 2 × 10−4 Pa s and ρ = 452 kg m−3. 28 d fcap o = 180 θ C d fcap (a) (b) Figure 2 A monomer (symbolized by a shaded disk) in contact with a liquid, which is repre- sented in blue. The contact wettability angle between this aerosol monomer and the liquid is noted θc. (a) The ideal situation of perfect non-wettability of the monomer, in this case the cap- illarity force sustaining the monomer is maximum. (b) A wettable monomer: 0o ≤ θc < 90o, in such a case capillarity cannot produce a floatability effect by balancing the gravity. 29 Earth d= 10 mm 300 250 200 150 100 50 ) λ ( y o i t a r e t a r g n i p m a D d= 5 mm ν+= 103 ν− d= 5 mm ν+= 102 ν− d= 0.01 mm 0 0 5 10 λ (cm) 15 20 Figure 3 The relative damping ratio y as a function of the wavelength λ in the case of a thin finite thickness film deposited at the surface of water, i.e. in the context of the Earth. The red curves correspond to three different thicknesses d = 0.01, 5 and 10 mm, using a "standard" ratio for kinematic viscosities: ν+/ν− = 100. The case where d = 0.01 mm is approximately similar to the case of a monomolecular floating layer. The solid blue curve shows the influence of the film kinematic viscosity ν+. 30
1003.4798	1	1003	2010-03-25T04:35:46	Critical Core Masses for Gas Giant Formation with Grain-Free Envelopes	[ "astro-ph.EP" ]	We investigate the critical core mass and the envelope growth timescale, assuming grain-free envelopes, to examine how small cores are allowed to form gas giants in the framework of the core accretion model. This is motivated by a theoretical dilemma concerning Jupiter formation: Modelings of Jupiter's interior suggest that it contains a small core of < 10 Earth mass, while many core accretion models of Jupiter formation require a large core of > 10 Earth mass to finish its formation by the time of disk dissipation. Reduction of opacity in the accreting envelope is known to hasten gas giant formation. Almost all the previous studies assumed grain-dominated opacity in the envelope. Instead, we examine cases of grain-free envelopes in this study. Our numerical simulations show that an isolated core of as small as 1.7 Earth mass is able to capture disk gas to form a gas giant on a timescale of million years, if the accreting envelope is grain-free; that value decreases to 0.75 Earth mass, if the envelope is metal-free, namely, composed purely of hydrogen and helium. It is also shown that alkali atoms, which are known to be one of the dominant opacity sources near 1500 K in the atmospheres of hot Jupiters, have little contribution to determine the critical core mass. Our results confirm that sedimentation and coagulation of grains in the accreting envelope is a key to resolve the dilemma about Jupiter formation.	astro-ph.EP	astro-ph	Critical Core Masses for Gas Giant Formation with Grain-Free Envelopes Yasunori Hori and Masahiro Ikoma Department of Earth and Planetary Sciences, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo 152-8551, Japan [email protected] Received ; accepted 0 1 0 2 r a M 5 2 . ] P E h p - o r t s a [ 1 v 8 9 7 4 . 3 0 0 1 : v i X r a -- 2 -- ABSTRACT We investigate the critical core mass and the envelope growth timescale, as- suming grain-free envelopes, to examine how small cores are allowed to form gas giants in the framework of the core accretion model. This is motivated by a the- oretical dilemma concerning Jupiter formation: Modelings of Jupiter's interior suggest that it contains a small core of < 10M⊕, while many core accretion mod- els of Jupiter formation require a large core of > 10M⊕ to finish its formation by the time of disk dissipation. Reduction of opacity in the accreting envelope is known to hasten gas giant formation. Almost all the previous studies assumed grain-dominated opacity in the envelope. Instead, we examine cases of grain-free envelopes in this study. Our numerical simulations show that an isolated core of as small as 1.7M⊕ is able to capture disk gas to form a gas giant on a timescale of million years, if the accreting envelope is grain-free; that value decreases to 0.75M⊕, if the envelope is metal-free, namely, composed purely of hydrogen and helium. It is also shown that alkali atoms, which are known to be one of the dominant opacity sources near 1500K in the atmospheres of hot Jupiters, have little contribution to determine the critical core mass. Our results confirm that sedimentation and coagulation of grains in the accreting envelope is a key to resolve the dilemma about Jupiter formation. Subject headings: accretion, accretion disks -- planets and satellites: formation -- 3 -- 1. Introduction Core masses provide clues to unveiling the origins of gas giants. In the core accretion scenario for gas giant formation, a forming solid protoplanet (i.e., a proto-core) experiences a rapid gas capture from a protoplanetary disk to form a massive gas envelope, when its mass exceeds a critical mass (Mizuno 1980; Bodenheimer & Pollack 1986). That critical core mass must be reached within the lifetime of the disk gas of several million years (e.g. Haisch et al. 2001), which places a limit on the core mass of a formed gas giant. Because of the slow increase in the critical core mass with core accretion rate (Stevenson 1982; Ikoma et al. 2000), faster formation in general results in larger core mass. Indeed, in many core-accretion models that are successful in forming Jupiter within several Myr (Pollack et al. 1996; Inaba et al. 2003; Alibert et al. 2005, etc.), the resultant core mass is as large as ∼ 10M⊕ or more. In contrast, the mass of Jupiter's present core is inferred to be small. Saumon & Guillot (2004) made an extensive investigation of the interior structure of Jupiter, finding successful models that are consistent with the observed values of its gravitational moments and equatorial radius by using a variety of equations of state (EOSs) for hydrogen and helium. They demonstrated that the possible core mass of Jupiter is smaller than ∼ 10 M⊕. This is also supported by recent calculations with an ab initio EOS derived in the first-principle approach (Nettelmann et al. 2008). While a more massive core of > 10M⊕ is reported by Militzer et al. (2008) who used their own EOS of hydrogen-helium mixtures based on density functional molecular dynamics, Fortney & Nettelmann (2009) pointed out the difference in the mass fraction of helium used by the two groups is responsible for this discrepancy in the derived value of Jupiter's core mass. Although this pending problem about Jupiter's core may arise from the uncertainty of EOS, the core mass suggested by interior modeling is, on an average, smaller than that derived by formation theories. This -- 4 -- fact motivates us to know how small a core can start the rapid gas accretion to form a massive envelope within several Myr. Reduction of opacity in the protoplanet's envelope has the potential to make a small core possible. As the opacity becomes small, the critical core mass decreases and the post-critical-mass gas accretion becomes fast, because low opacity in the envelope makes it difficult to maintain the envelope's hydrostatic structure without gravitational energy released by contraction of the envelope (Mizuno 1980; Stevenson 1982; Ikoma et al. 2000). Since the opacity sources are dust grains and gaseous components in the envelope, a minimum critical core mass is achieved in the case of grain-free envelopes. All the previous studies except one calculation done by Mizuno (1980) (see Section 2) assumed grain-dominated opacity in the outer envelope. Thus, in this paper, we consider grain-free envelopes and make an extensive investigation of the critical core mass and timescale for gas accretion. Results from the recent works by Podolak and his colleagues (Podolak 2003; Movshovitz & Podolak 2008) are encouraging. They have directly simulated the dynamical behavior of dust grains to determine their size distribution, and then calculated grain opacity in the accreting envelope. Their numerical simulations revealed that grain opacities in the envelope can be much lower than those in the protoplanetary disk. The reason is that small grains initially suspended in the outer envelope quickly grow large in size and then settle down into the deep envelope where temperature is high enough that grains evaporate. In the following section, the details of the gas opacity used in this study are described. Our numerical calculations and results for the critical core mass and the timescale of gas accretion are shown in Sections 3 and 4, respectively. We discuss a possibility of Jupiter formation with a small core in Section 5. -- 5 -- 2. Gas Opacity The opacity of the envelope gas is lowest when the envelope contains only hydrogen and helium. In this study, we first consider such a metal-free case1. We compute chemical equilibrium between H2, H, H+, H−, H+ 2 , H+ 3 , and e−, and then calculate the opacity that includes the bound-free and free-free absorptions by H−, Rayleigh scatterings by H2, H, and He, Thomson scattering by e−, and the collision-induced absorptions (CIA) due to H2−H2, H2−H, H2−He, and H−He. Values of quantities relevant to those calculations are given in Lenzuni et al. (1991) and references therein. The CIA opacities are computed by using the latest programs and tables available on Borysow's web page2 (Borysow et al. 1985; Borysow & Frommhold 1990; Borysow 1991, 1992; Zhen & Borysow 1995; Birnbaum et al. 1996; Borysow et al. 1997, 2000; Borysow 2002). Metal-containing envelopes with no grains are also considered in this study. It is uncertain what fraction of metals in protoplanetary disks are incorporated in dust grains and what faction remains in their gaseous forms. For example, the amount of water adsorbed onto grain surfaces depends strongly on the thermal state of the protoplanetary disk. Water is adsorbed onto grain surfaces in a relatively cold disk, while it remains in the gas phase in a relatively hot disk (Markwick et al. 2002; Woods & Willacy 2009). Adsorption rates of water are also sensitive to the abundances and sizes of dust grains in the disk (Aikawa & Nomura 2006), but such dust properties remain poorly known. Alkali atoms may be important: They are known to be one of the dominant gas opacity sources near 1500K in the outer atmospheres of brown dwarfs and giant planets (Guillot et al. 1994; Burrows et al. 2000), though alkali atoms have a tendency to reside onto grain surfaces 1In the community of astrophysics, the word 'metal' represents elements heavier than helium. 2http://www.astro.ku.dk/ aborysow/programs/index.html -- 6 -- in protoplanetary disks (Hasegawa & Herbst 1993). Therefore, we consider two cases in addition to the metal-free case: In one case called the alkali case hereafter, alkali atoms are all in the gas phase; in another case called the no-alkali case, they are absent in the gas. In the alkali case, we use the gas-opacity data provided by Freedman et al. (2008), which incorporate the revised solar abundances and alkali atoms. The solar abundances have been recently revised by re-analyses of two forbidden lines of neutral oxygen and carbon from the solar photosphere under the assumption of the local thermodynamic equilibrium, although their values should still require scrutiny because of uncertainties in 3-D hydrodynamical models for the solar atmosphere (Allende Prieto et al. 2001, 2002; Lodders 2003). Opacity data without alkali atoms are kindly provided by Dr. Freedman (in personal communication). In practice, we use the up-to-date opacity data of Freedman et al. (2008), which cover the wider ranges of temperature and pressure compared to the published ones and include the latest HITRAN spectroscopic data and an additional opacity source, CO2 (in personal communication). 3. Critical Core Mass The spherically-symmetric hydrostatic structure of a protoplanet is simulated in a manner similar to previous studies (e.g. Mizuno 1980). The protoplanet consists of a solid core and a gaseous envelope. The core has a constant density of 3.2g cm−3 and its structure is not computed. The envelope is assumed to be in purely hydrostatic equilibrium and have a uniform chemical composition. While the composition depends on the case, the abundance ratios of elements taken into account are always solar (Lodders 2003). All simulations are computed with spline-interpolated SCVH EOS tables for hydrogen and helium provided by Saumon et al. (1995) , contributions from heavy elements being ignored. The inner boundary conditions are applied at the core surface. At the outer boundary, the -- 7 -- temperature and density are equal to the midplane values of a protoplanetary disk at the protoplanet's orbit, Tdisk and ρdisk, because the envelope is assumed to be in equilibrium with the disk gas at the outer edge. The outer radius of the protoplanet is defined by the smaller of the accretion radius and the tidal radius. Note that our results are insensitive to the outer boundary conditions. We handle the accretion rate of planetesimals, Mc, as a free parameter in this study. We follow the same procedure as Mizuno (1980) to determine the critical core mass, Mcrit (please see Mizuno (1980) in detail). Table 1 summarizes input parameters and their values used in this study. Figure 1 shows Mcrit as a function of Mc for the three cases. For comparison, we plot results for additional two cases in which the envelope contains both grains and gas as opacity sources. We assume that the grain opacity takes f times values of the grain opacity given by Pollack et al. (1985); their calculations assumed a nearly interstellar size distribution of dust grains. Reduced opacity results in small Mcrit, as shown in Figure 1. For instance, we now focus on results for Mc = 1 × 10−6M⊕/yr. In the case of f = 0.01 (double dot-dashed line), which is often adopted in core accretion models (e.g. Hubickyj et al. 2005), Mcrit = 10M⊕. Removal of grains lowersMcrit to a few M⊕; 3.5M⊕ for the alkali case (dashed line) and 1.5M⊕ for the metal-free case (solid line). Compared to the alkali case, the metal-free case always produces smaller Mcrit. This is because molecules composed of oxygen and/or carbon such as H2O and CO2 are effective opacity sources as for determination of Mcrit. In contrast, comparison between the results for the alkali (dashed line) and no-alkali cases (dotted line) demonstrates that alkali atoms have little contribution to determine Mcrit. The reason is that convection governs heat transfer in deep, hot parts of the envelope where alkali atoms have a great contribution to the opacity. -- 8 -- Figure 1 also demonstrates that all the lines except one for the metal-free case are converging, as Mc decreases. In the case of low Mc, namely, low luminosity, the outermost isothermal layer extends deep in the envelope. In that isothermal layer, the density increases rapidly to keep the pressure gradient needed for supporting the core's gravity. Thus, because of high densities, the opacity from H2O and CO2 dominates the grain opacity in the deep radiative envelope in spite of low temperature. Finally it is worth mentioning differences from previous studies. Mizuno (1980) calculated Mcrit in the metal-free case for only one value of core accretion rate, 1 × 10−6M⊕/yr, and derived 1.5M⊕. While the CIA opacities at high temperatures have been revised thanks to progresses of quantum mechanical models and improvement of experimental data in the 1990s, Mizuno's (1980) value of Mcrit is in good agreement with our result for the same parameters. Ikoma et al. (2000) investigated cases with low grain opacity. Our calculations for f = 0.01 and f = 0.001 yield larger Mcrit than those from Ikoma et al. (2000), because they assumed relatively small grain opacities compared to those used in this study. 4. Envelope Growth Timescale The critical core mass decreases as the core accretion rate decreases, as shown in Fig.1. Therefore, the case of Mc → 0 (i.e., isothermal envelope) yields the absolute minimum of Mcrit (Sasaki 1989; Pecnik & Wuchterl 2005). However, for smaller Mcrit, it takes longer for the core to capture disk gas (Ikoma et al. 2000; Ikoma & Genda 2006). Because there are time constraints on gas giant formation such as the disk's lifetime (Haisch et al. 2001), a practical minimum of Mcrit is determined in this respect. It is necessary to estimate how long it takes for a protoplanet with a given core mass, Mcore, to capture disk gas. -- 9 -- We consider an isolated protoplanet, namely, accumulation of the envelope after planetesimal accretion is halted. The isolated protoplanet always experiences contraction of its envelope and captures disk gas, because of no energy supply due to planetesimal accretion. We simulate the quasi-static evolution of the envelope with a given Mcore after planetesimal accretion is halted and evaluate the growth timescale of the envelope. As the growth timescale of the envelope, we present values of the characteristic growth time, τg, that is defined by Ikoma et al. (2000). Figure 2 plots τg as functions of Mcore. For comparison, the results for f = 0.01 and 0.001 are also shown. The growth time τg is found to be much shorter in the case of the grain-free opacities than τg for the grain-dominated opacities. For example, when Mc = 3M⊕, τg = 7 × 102 yr in the metal-free case (solid line) and 7 × 104 yr in the alkali case (dashed line), while 2 × 106 yr for f = 0.01 (double dot-dashed line). The envelope growth is regulated by the Kelvin-Helmholtz contraction of the envelope; that is, τg is proportional to 1/L (Ikoma et al. 2000), where L is the luminosity at the envelope's outer edge. Since a change in opacity compensates that in luminosity, ¯κ being an averaged opacity, τg should be proportional to ¯κ/L. In the case of the grain-dominated opacities, the dependence is simple, namely, τg ∝ f , as shown in Ikoma et al. (2000) and Ikoma & Genda (2006). Figure 2, however, demonstrates that such a simple scaling is inappropriate in low ¯κ cases. The curves for grain-free cases are steeper than those for grain-dominated cases; the reason has been already described in the previous section. In any case, we have found that the envelope growth time is significant shorter in the case of the grain-free opacities. Based on this fact, we discuss the minimum critical core mass from the viewpoint of gas giant formation, and the possibility of Jupiter formation with a small core in the following section. -- 10 -- 5. Discussion There is a theoretical dilemma concerning Jupiter formation, as described in Introduction. Modelings of Jupiter's interior suggest that Jupiter has a small core of < 10M⊕ (e.g., Saumon & Guillot 2004), while many core-accretion models of Jupiter formation require a large core of > 10M⊕ to finish its formation by the time of disk dissipation (Pollack et al. 1996; Alibert et al. 2004, 2005; Fortier et al. 2007, 2009). The disk instability scenario has been revisited as an alternative scenario of their formation (e.g. Boss 2000; Mayer et al. 2002). In this study, we have demonstrated that reduced opacities in the protoplanet's envelope have the potential to resolve this dilemma. From Fig. 2, one finds that Mcore = 0.75M⊕ for τg = 1 Myr in the metal-free case; Mcore = 1.7M⊕ even in the alkali case. Given observed lifetimes of protoplanetary disks of several Myr, the fact above indicates the reduction of opacity allows Jupiter to have a small core that is consistent with interior modelings in principle. The feasibility of such minimum Mcrit depends on opacities in the protoplanet's envelope, while this does not change our conclusion that the minimum Mcrit obtained here provides the lowest limit to core masses of gas giants to which the core accretion model can apply. We need more extensive investigation of gas giant formation into which sedimentation and coagulation of grains in the accreting envelope are incorporated, although reduction of opacity was already pronounced (Podolak 2003; Movshovitz & Podolak 2008). Our results also shed light upon growth of solid cores. Many core accretion models assumed the presence of a single protoplanet (e.g. Pollack et al. 1996). However, a gas giant is in practice thought of as being formed in a system of multiple protoplanets embedded in a protoplanetary disk. Compared to cases of a single protoplanet, the final mass of a core should be small in the case of a multiple-protoplanet system. According to Kokubo & Ida (1998, 2000), the isolation mass is a few M⊕ around 5AU. Even such a small core is enough -- 11 -- for a gas giant to capture disk gas within several Myr, as demonstrated in this study. We need to perform comprehensive simulations on gas giant formation in multiple-protoplanet systems, which incorporate adequate models of planetary accretion such as fragmentation of planetesimals and e-damping due to gas drag (e.g. Duncan et al. 2009). These calculations will be our future work. We are grateful to S. Ida for his continuous encouragement. We thank R.S. Freedman for his kindness of calculating newly Rosseland mean opacities of gas with no alkali atom and providing updated gas opacities. We also thank H. Tanaka and H. Nomura for giving us helpful comments. Y.H. is supported by Grant-in-Aid for JSPS Fellows (No.21009495) from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan. -- 12 -- REFERENCES Aikawa, Y., & Nomura, H. 2006, ApJ, 642, 1152 Alibert, Y., Mordasini, C., & Benz, W. 2004, A&A, 417, L25 Alibert, Y., Mordasini, C., Benz, W., & Winisdoerffer, C. 2005, A&A, 434, 343 Allende Prieto, C., Lambert, D. L., & Asplund, M. 2001, ApJ, 556, L63 Allende Prieto, C., Lambert, D. L., & Asplund, M. 2002, ApJ, 573, L137 Birnbaum, G., Borysow, A., & Orton, G. S. 1996, Icarus, 123, 4 Bodenheimer, P., & Pollack, J. B. 1986, Icarus, 67, 391 Borysow, A. 1991, Icarus, 92, 273 Borysow, A. 1992, Icarus, 96, 169 Borysow, A. 2002, A&A, 390, 779 Borysow, A., Borysow, J., & Fu, Y. 2000, Icarus, 145, 601 Borysow, A., & Frommhold, L. 1990, ApJ, 348, L41 Borysow, A., Jorgensen, U. G., & Zheng, C. 1997, A&A, 324, 185 Borysow, J., Trafton, L., Frommhold, L., & Birnbaum, G. 1985, ApJ, 296, 644 Boss, A. P. 2000, ApJ, 536 L101 Burrows, A., Marley, M. S., & Sharp, C. M. 2000, ApJ, 531, 438 Fortney, J. J. & Nettelmann, N. 2009, Space Sci. Rev., 115 Fortier, A., Benvenuto, O. G., & Brunini, A. 2007, A&A, 473, 311 -- 13 -- Fortier, A., Benvenuto, O. G., & Brunini, A. 2009, A&A, 500, 1249 Duncan, H. F., Levison, E., Thommes, M. J. 2009, arXiv:0912.3144 Freedman, R. S., Marley, M. S., & Lodders, K. 2008, ApJS, 174, 504 Guillot, T., Gautier, D., Chabrier, G., & Mosser, B. 1994, Icarus, 112, 337 Haisch, K. E. Jr., Lada, E. A., & Lada, C. J. 2001, ApJ, 553, L153 Hasegawa, T. I., & Herbst, E. 1993, MNRAS, 263, 589 Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 2005 Icarus, 179, 415 Inaba, S., Wetherill, G. W., & Ikoma, M. 2003, Icarus, 166, 46 Ikoma, M., & Genda, H. 2006, ApJ, 648, 696 Ikoma, M., Nakazawa, K., & Emori, H. 2000, ApJ, 537, 1013 Lodders, K. 2003, ApJ, 591, 1220 Kokubo, E. & Ida, S. 1998, Icarus, 131, 171 Kokubo, E. & Ida, S. 2000, Icarus, 143, 15 Lodders, K. 2003, ApJ, 591, 1220 Lenzuni, P., Chernoff, D. F. & Salpeter, E. E. 1991, ApJ, 76, 759 Markwick, A. J., Ilgner, M., Millar, T. J., & Henning, Th. 2002, A&A, 385, 632 Mayer, L., Quinn, T., Wadsley, J., & Stadel, J. 2002, Science, 298, 1756 Militzer, B., Hubbard, W. B., Vorberger, J., Tamblyn, I., & Bonev, S. A. 2008, ApJ, 688, L45 Mizuno, H. 1980, Prog. Theor. Phys., 64, 544 -- 14 -- Movshovitz, N., & Podolak, M. 2008, Icarus, 194, 368 Nettelmann, N., Holst, B., Kietzmann, A., French, M., Redmer, R., & Blaschke, D. 2008, ApJ, 683, 1217 Pecnik, B., & Wuchterl, G. 2005, A&A, 440, 1183 Podolak, M. 2003, Icarus, 165, 428 Pollack, J. B., McKey, C. P., & Christofferson, B. M. 1985, Icarus, 64, 471 Pollack, J. B., Hubickyj, O., Bodenheimer, P., & Lissauer, J. J. 1996, Icarus, 124, 62 Sasaki, S. 1989, A&A, 215, 177 Saumon, D., Chabrier, G., & Van Horn, H. M. 1995, ApJS, 99, 713 Saumon, D., & Guillot, T. 2004, ApJ, 609, 1170 Stevenson, D. J. 1982, Planet. Space Sci., 30, 755 Woods, P. M., & Willacy, K. 2009, ApJ, 693, 1360 Zheng, C., & Borysow, A. 1995, ApJ, 441, 960 This manuscript was prepared with the AAS LATEX macros v5.2. -- 15 -- 100 ) + M 10 ( t i r c M 1 0.1 -10 -9 -8 -7 -6 -5 -4 -3 log Mc (M+ /yr) Fig. 1. -- Critical core masses, Mcrit, as functions of core accretion rate, Mc. The solid, dashed, and dotted lines represent the results for the metal-free, alkali, and no-alkali cases (see Section 2 for the definitions). For comparison, the results of two cases in which there exist grains in the envelope are also shown with the double dot-dashed and dot-dashed lines: f = 0.01 and 0.001, respectively, where f is the grain depletion factor (see the text for its definition). We have used the grain opacity tables derived from Pollack et al. (1985). [A color version of this plot is available in the electronic edition of The Astrophysical Journal.] -- 16 -- 9 8 7 6 5 4 3 ) r y ( g g o l 2 0.1 1 Mcore (M+ ) 10 30 Fig. 2. -- Typical growth timescale of the envelope, τg (see the text for the definition), as a function of Mcore. As in Fig.1, the solid, dashed, dotted lines represent the results of the metal-free, alkali, and no-alkali cases respectively. The results of f = 0.01 (double dot-dashed line) and 0.001 (dot-dashed line) are also shown. [A color version of this plot is available in the electronic edition of The Astrophysical Journal.] -- 17 -- Table 1: Input Parameters and Their Values. Parameter semimajor axis, a core density disk temperature, Tdisk Value 5.2 AU 3.2 g/cm3 150 K disk density, ρdisk 5.0 × 10−11 g/cm3 planetesimal accretion rate, Mc 1.0 × 10−3 to 1.0 × 10−10 M⊕/yr
1510.07052	1	1510	2015-10-23T20:21:08	A Chemical Kinetics Network for Lightning and Life in Planetary Atmospheres	[ "astro-ph.EP" ]	There are many open questions about prebiotic chemistry in both planetary and exoplanetary environments. The increasing number of known exoplanets and other ultra-cool, substellar objects has propelled the desire to detect life and prebiotic chemistry outside the solar system. We present an ion-neutral chemical network constructed from scratch, Stand2015, that treats hydrogen, nitrogen, carbon and oxygen chemistry accurately within a temperature range between 100 K and 30000 K. Formation pathways for glycine and other organic molecules are included. The network is complete up to H6C2N2O3. Stand2015 is successfully tested against atmospheric chemistry models for HD209458b, Jupiter and the present-day Earth using a simple 1D photochemistry/diffusion code. Our results for the early Earth agree with those of Kasting (1993) for CO2, H2, CO and O2, but do not agree for water and atomic oxygen. We use the network to simulate an experiment where varied chemical initial conditions are irradiated by UV light. The result from our simulation is that more glycine is produced when more ammonia and methane is present. Very little glycine is produced in the absence of any molecular nitrogen and oxygen. This suggests that production of glycine is inhibited if a gas is too strongly reducing. Possible applications and limitations of the chemical kinetics network are also discussed.	astro-ph.EP	astro-ph	A Chemical Kinetics Network for Lightning and Life in Planetary Atmospheres School of Physics and Astronomy, University of St Andrews, St Andrews, KY16 9SS, United Kingdom P. B. Rimmer1 and Ch Helling ABSTRACT There are many open questions about prebiotic chemistry in both planetary and exoplane- tary environments. The increasing number of known exoplanets and other ultra-cool, substellar objects has propelled the desire to detect life and prebiotic chemistry outside the solar system. We present an ion-neutral chemical network constructed from scratch, Stand2015, that treats hydrogen, nitrogen, carbon and oxygen chemistry accurately within a temperature range between 100 K and 30000 K. Formation pathways for glycine and other organic molecules are included. The network is complete up to H6C2N2O3. Stand2015 is successfully tested against atmo- spheric chemistry models for HD209458b, Jupiter and the present-day Earth using a simple 1D photochemistry/diffusion code. Our results for the early Earth agree with those of Kasting (1993) for CO2, H2, CO and O2, but do not agree for water and atomic oxygen. We use the network to simulate an experiment where varied chemical initial conditions are irradiated by UV light. The result from our simulation is that more glycine is produced when more ammonia and methane is present. Very little glycine is produced in the absence of any molecular nitrogen and oxygen. This suggests that production of glycine is inhibited if a gas is too strongly reducing. Possible applications and limitations of the chemical kinetics network are also discussed. Subject headings: astrobiology — atmospheric effects — molecular processes — planetary systems 1. Introduction The potential connection between a focused source of energy and life was first made apparent in the Miller-Urey experiment (Miller 1953), set to test a hypothesis proposed by Haldane (1928). In this experiment, a gas composed of water va- por, ammonia, methane and molecular hydrogen was circulated past an electric discharge. After a week’s time, various biologically relevant chemi- cals had developed, including glycine and alanine, identified with a paper chromatrogram. A follow- up study of Miller’s samples, carried out approx- imately fifty years later, discovered a much richer variety of prebiotic compounds than originally thought (Johnson et al. 2008). Since then, nu- merous related experiments have been carried out under a variety of conditions (see Miller & Urey 1959; Cleaves et al. 2008, and references therein). [email protected] The input energy source and the initial chem- istry have been varied across these different ex- periments. An energy source may have been im- portant for the production of prebiotic species on Earth, because the pathways to formation have considerable activation barriers, often on the order of 0.1-1 eV. Patel et al. (2015) generated prebiotic species by exposing HCN and H2S to ultraviolet light. The experimental results from Powner et al. (2009) suggest that the aqueous synthesis of amino acids, nucleobases and ribose is predisposed, start- ing from glyceraldehyde and glycoaldehyde, which they suggest would most likely form through heat- ing and UV irradiation. Shock synthesis of amino acids due to the atmospheric entry of cometary meteors and micrometeorites or thunder is also sufficient to overcome these barriers and produce amino acids (Bar-Nun et al. 1970). The initial chemical conditions are naturally significant to the formation of prebiotic chemistry. 1 Of course, in an environment where hydrogen or carbon were lacking, there would be no complex hydrocarbons. Nitrogen and phosphorus are also essential to the origins of terrestrial life, although some scientists, such as Benner et al. (2004), have speculated that life could occur under very differ- ent chemistries; presently, we lack the ability to explore this possibility. The initial chemical com- position also has an effect on the production of prebiotic chemical species. For example, hydro- gen can be bound in a reducing species, CH4, in an oxidizing species, H2SO4, or into the neutral species of water (H2O). Both Schlesinger & Miller (1983) and Miyakawa et al. (2002) have found that performing a Miller-Urey like experiment in an ox- idizing environment produces only trace amounts of prebiotic materials, whereas performing the ex- periment in a reducing environment produces a great number of prebiotic materials. The atmosphere of Earth in its present state is oxidizing (≈ 21% O2, 78 % N2). The atmosphere of the Earth during its first billion years (first 1 Gyr) would have had a very different composition, probably oxidizing or at least only weakly reducing (Kasting 1993), although Tian et al. (2005) sug- gest that the Earth’s atmosphere was once highly reducing. Even if the Earth never possessed a strongly reducing atmosphere, other planets and moons are known to have both reducing atmo- spheres and active lighting and UV photochem- istry, Jupiter for example. Extrasolar planets may not simply have diverse compositions, but also widely varied gas-phase C/O ratios, either intrinsi- cally at formation, as may be the case with Wasp- 12b, XO-1b, CoRoT-2b (Madhusudhan et al. 2011; Moses et al. 2013), and possibly the inte- rior of 55 Cancri e (Madhusudhan et al. 2012, but see also Nissen 2013); or alternatively due to oxy- gen depletion into the cloud particles (Bilger et al. 2013; Helling et al. 2014). The question of the C/O ratio is not a settled matter (Benneke 2015). These diverse planetary and exoplanetary en- vironments provide unique “laboratories” within which to explore prebiotic chemistry. There are many potential drivers for prebiotic chemistry in planets and exoplanets, from the steep ther- mal gradients in hot Jupiters and close-in Super- Earths to the thermal production of organics and complex hydrocarbons in Saturn’s storms (Moses 2015) and photochemical production of complex 2 organics in Titan (Yung et al. 1984; Loison et al. 2015). There is some evidence that cosmic rays drive the formation of hydrogen cyanide in Nep- tune (Lellouch et al. 1994). Molina et al. (1999) have proposed pathways to formation of a rich variety of nitriles via cosmic rays in Titan’s atmo- sphere. As mentioned above, electric discharges may also be an important source of energy driv- ing the production of prebiotic species, and are ubiquitous throughout the gas giants. Dis- charges in the form of lightning are known to occur within our solar system, on Earth, Jupiter (Little et al. 1999), Saturn (Dyudina et al. 2007), Uranus (Zarka & Pedersen 1986), and Neptune (Gurnett et al. 1990). There are some indications of lightning discharges on Venus (Taylor et al. 1979), and possibly also in Titan’s nitrogen chem- istry (Borucki et al. 1984), although these traces are still tentative. Lightning is hypothesized to occur on exoplanets (Aplin 2013; Helling et al. 2013) and brown dwarfs (Helling et al. 2013; Bailey et al. 2014). Simulated plasma discharges initiated within Jupiter-like gas compositions sug- gest that lightning on Jupiter may produce a significant amount of trace gases (Borucki et al. 1985). The comparison between experimental rates of production of organic compounds in high-temperature plasmas to chemical equilibrium models is unsurprisingly poor (Scattergood et al. 1989), and indicates that a chemical kinetics ap- proach will be important in explaining the results of these experiments, Chemical kinetics seems to be necessary for exploring any of these pathways to the formation of prebiotic species. Chemical kinetics models have been applied to planetary and exoplanetary atmospheric condi- tions in such a diverse range that it is impractical to provide complete references, so a brief summary of the work will instead be provided. Photochemi- cal models of the modern Earth have been applied in the context of 1D models (Owens et al. 1985), up to fully coupled 3D general circulation models (Roble & Ridley 1994), and even within a flexi- ble modular framework that can be included as a module within other codes (Sander et al. 2005). The Earth’s atmosphere during its first billion years has been extensively modeled (Kasting 1993; Zahnle 1986). Chemical kinetics models have been applied also to Jupiter’s atmosphere, from the deep atmosphere (Fegley & Lodders 1994; Visscher et al. 2010) through the stratosphere (Zahnle et al. 1995; Moses et al. 2005). The at- mosphere of the moon Titan has also been ana- lyzed using ion-neutral chemical kinetics to better explain the abundance of rich hydrocarbons in its atmosphere and its stratospheric haze (Yung et al. 1984; Keller et al. 1998; Lavvas et al. 2008a,b). Chemical kinetics models for exoplanetary at- mospheres have been typically developed for hot Jupiters, especially HD189733b and HD209458b (Moses et al. 2011; Venot et al. 2012; Zahnle et al. 2009). Almost all of the models for hot Jupiters have been applied only in two dimensions, and so have not taken a more complete account of the atmospheric dynamics, instead relying on a pa- rameterization of vertical mixing using the eddy diffusion coefficient, Kzz [cm2 s−1] (see Lee et al. 2015, their Sec. 4.2). Ag´undez et al. (2014) have taken on the ambitious task of coupling a chemical kinetics model to 2D dynamics for both HD189733b and HD209458b. Ion-neutral models have been applied to exoplanets, taking account of photochemistry (Lavvas et al. 2014), and addi- tionally of cosmic ray ionization (Walsh & Millar 2011; Rimmer et al. 2014). Chemical kinetics models have also been applied to the extraso- lar super-earths (Hu et al. 2012; Hu et al. 2013; Hu & Seager 2014), and have been used to ex- plore possible biosignatures on rocky planets (Seager et al. 2013a,b). There has also been some recent investigation into chemistry on helium dom- inated exoplanets (Hu et al. 2015). Lightning chemistry has been explored with some basic chemical kinetics models, e.g. within Earth’s mesosphere (Luque & Ebert 2009 and Parra-Rojas et al. 2013) and Saturn’s lower iono- sphere (Dubrovin et al. 2014). Dubrovin et al. (2014) present interesting results for Saturn’s lower ionosphere, predicting that TLE’s within this region would produce mostly H+ 3 , what they identify as the primary positive charge carrier dur- ing the duration of the TLE and for sometime af- ter. This would mimic the effect of cosmic ray ion- ization. Parra-Rojas et al. (2013) presented simi- lar results involving terrestrial nitrogen chemistry. The products of discharge chemistry in the upper part of both hydrogen-rich and nitrogen-rich at- mospheres seem to be similar to the products of cosmic ray chemistry in these same atmospheres. 3 There are many open questions about prebi- otic chemistry in diverse planetary and exoplan- etary environments, as well as in the lab. In this paper, we present a candidate network for exploring UV photochemistry, cosmic ray chem- istry and lightning-driven chemistry, constructed from scratch. We will explore mostly the photo- chemistry and thermochemistry within this paper, leaving the exploration of lightning-driven chem- istry and cosmic ray chemistry to future work. The largest task in developing this network has been the collation of a full set of chemical re- actions that treat both reducing and oxidizing chemistries at temperatures ranging from 100 K through 30000 K (the approximate peak temper- ature of lightning, see Orville 1968; Price et al. 1997) and the selection of rate constants when more than one is published. Since one interest is the investigation of the formation rate of prebi- otic species in diverse environments, the network is made extensive enough to include the simplest amino acid, glycine. In this paper, we present this chemical network (Stand2015), and test in a di- versity of environments. For these tests, we devel- oped a simple 1D photochemistry/diffusion code (Argo). Argo was developed based on Nahoon (Wakelam et al. 2012) by including wavelength- dependent photochemistry, cosmic ray transport, water condensation and chemical mixing. The Stand2015 network is presented in Sec- tion 2. We compare the predictions of our network using a simplified 1D photochemistry/diffusion code called Argo (Section 3). The model and net- work are then combined and tested against other model results for HD209458b and the early Earth, and compared to observation for Jupiter and the present-day Earth in Section 4. Finally, in Sec- tion 5 we simulate a Miller-Urey type experiment and explore the formation of glycine under various chemical conditions. Section 6 contains a short discussion of the results and possible future appli- cations of this model. 2. The Chemical Network The Stand2015 Atmospheric Chemical Net- work is an H/C/N/O network with reactions in- volving He, Na, Mg, Si, Cl, Ar, K, Ti and Fe, developed from scratch. It contains all known re- actions for species of up to 6 hydrogen, 2 carbon, 2 nitrogen and 3 oxygen atoms, for which a rate con- stant has been published, as well as a less complete network involving species with 3+ carbon atoms, 3 nitrogen atoms and/or 4 oxygen atoms. A chem- ical network is effectively a list of chemical reac- tions and reaction rate constants. Rate constants are used to calculate the rates of production and loss of a particular molecular or ionic species, Pi [cm−3 s−1] and Li [cm−3 s−1] respectively, and i is enumerated over the list of species. Rate constants are of zeroth order (e.g., source terms, Si [cm−3 s−1]), first order (involving interactions with parti- cles not accounted in the network, such as photons or cosmic rays, k1 [s−1]), second order (collisions between particle i and other particles within the network, k2 [cm3 s−1]), or third order (collisions between particle i and other particles, as well as a third body, denoted here as k3 [cm6 s−1]). The rates of production and loss for a given species, i, in terms of rate constants, are generally: Pi = Si +X k1 nj +X k2 njnk +X k3 ngasnjnk, Li =X k1 ni +X k2 njni +X k3 ngasnjni. (2) (1) Summation is over all the relevant reactions, some involving species j and/or k, that result in the production (Eq. (1)) or loss (Eq. (2)) of species i. The symbol ni [cm−3] denotes the number density of species i and ngas [cm−3] denotes the total gas number density. The reaction rate constants have been as- sembled from various databases. With only a couple hundred exceptions, the rate constants for 2-body and 3-body neutral reactions have been assembled from the NIST Chemical Ki- netics Database (Manion et al. 2013). Virtually all of the ion-neutral reactions were taken from Ikezoe et al. (1987). Several rate constants that we have used, relevant for terrestrial atmospheric chemistry, are taken from Sander et al. (2011). The KIDA database provided the rate constants for several dissociative recombination reactions (Wakelam et al. 2012). Coefficients for the cosmic ray ionization rate constant were taken from the OSU chemical network (Harada et al. 2010). Rate constants were compared to the pub- licly available networks of Moses et al. (2011); Venot et al. (2012), and ion-neutral rate coeffi- cients were checked against the KIDA database (Wakelam et al. 2012)1, as well as the OSU 09 2010 high temperature network (Harada et al. 2010)2. Some further ion-neutral reactions in- volving the alkali ion chemistry were appropriated from Lavvas et al. (2014). Finally, ∼ 20 more reactions for suspected formation pathways for glycine have been added to the network, from Blagojevic et al. (2003); Patel et al. (2015). The full network and references are provided in Ap- pendix A. The following subsections contain brief discussions about the different classes of reactions, their rate coefficients and whether reverse reac- tions have been included. 2.1. 2-Body Neutral-Neutral and Ion- Neutral Reactions Two-body neutral-neutral and ion-neutral reac- tions follow the basic scheme: A + B → Y + Z, and A+ + B → Y+ + Z. (3) (4) The rate constants for these reactions are approx- imated by the Kooij equation (Kooij 1893): k2 = α(cid:16) T 300 K(cid:17)β e−γ/T , (5) where T [K] is the gas temperature3, k2 [cm3 s−1] is the rate constant, and α [cm3 s−1], β and γ are constants characterizing the reaction. All of these reactions are reversed in our network and we use the rate coefficients for the best character- ized direction for each reaction, which is typically the exothermic direction. For neutral-neutral re- actions, even when exothermic, there is often a siz- able barrier to reaction, allowing certain elements to be locked into non-equilibrium configurations at low temperatures effectively for eternity, because the barrier to the lower energy state is too large to be overcome in the current environment. Ion-neutral reactions do not typically have bar- riers in the exothermic direction, and in many cases the rate constants are altogether temper- closely approximating the ature independent, 1http://kida.obs.u-bordeaux1.fr/ 2http://faculty.virginia.edu/ericherb/research.html 3Surface chemistry is not considered in this paper, and the temperature of all chemical species including electrons is set equal to the gas-phase temperature. 4 Langevin approximation. A notable exception are charge exchange reactions, A+ + B → B+ + A, (6) which, due to the differences in energy between ionic and neutral ground states, often contains barriers on the order of a few x 100 K. The rate constants for the forward reactions are given in Appendix A with the label ‘2n’, reactions 577-1352. These reactions are reversed following the scheme described in Appendix B. The ion- neutral reactions are also reversed, are listed in Appendix A with ‘2i’, reactions 1353-2569. 2.2. 3-Body Neutral Reactions, Dissocia- tion Reactions, and Radiative Associ- ation Reactions Reactions that involve a third body occur pri- marily in the two forms: A + M → Y + Z + M, A + B + M → Z + M, (7) (8) where M represents any third body. Decompo- sition reactions are well studied at high temper- atures, being important for various combustion processes. Just as in Section 2.1, we choose the reactions best characterized, which in this case of- ten involve endothermic reactions. The rate coef- ficients for the majority of these reactions follow the Lindemann form (Lindemann et al. 1922). In this form, we first determine the rate constants in the low-pressure (k0 [cm6 s−1]) and high-pressure (k∞ [cm3 s−1]) limits: k0 = α0(cid:16) T k∞ = α∞(cid:16) T 300 K(cid:17)β0 300 K(cid:17)β∞ (9) (10) e−γ0/T , e−γ∞/T . These are combined with the number density of the neutral third species, [M] [cm−3] to determine the reduced pressure, pr = k0[M]/k∞, and this can then be utilized to set the pressure-dependent effective “two-body” rate: approximate the transition between the low- pressure and high-pressure limits, and this pro- vides the Troe form (Troe 1983). The coefficients for the Troe form are not explicitly given. We favor using the rate constants for three- body combination reactions, and reversing these reaction to determine the rate of thermal decom- position. In many cases, however, the rate con- stants are unavailable. When we have only the rate coefficients for the decomposition reactions, we add an additional 500 K barrier to both the de- composition and three body combination rate con- stants. This barrier is added in order to limit run- away three body reactions that can result from reversing decomposition reactions at low temper- ature. Additionally, we incorporate a small number of radiative association reactions, of the form: A + B → Z + γ, (12) where γ is the radiated photon that carries the ex- cess energy from the association. We appropriate the Kooij form for this reaction, as with 2-Body Neutral-Neutral reactions, in order to determine the rate constant kra [cm3 s−1]. We then apply this rate constant, along with the rate constant for the corresponding three-body reaction, to the adduct form of the overall rate constant (H´ebrard et al. 2013, their eq. (B.2)): k = (cid:0)k0[M]F + kr(cid:1)k∞ k0[M] + k∞ , (13) where the function F is from the Troe form of the transition from high to low pressure. The rate constants for the forward reactions are given in Appendix A with the labels ‘2d’ for the neutral species and ‘3i’ for ion-neutral species. These reactions are reversed in the manner de- scribed by Appendix B. Reactions 1-420 are reac- tions of this type, for which each odd-numbered reaction gives the low-pressure rate constant k0 [cm6 s−1] and each even-numbered reaction gives the high-pressure rate constant k∞ [cm3 s−1]. Re- actions labeled ’ra’ are radiative association reac- tions, numbered 2974-2980. k2 = k∞pr 1 + pr . (11) 2.3. Thermal Ionization & Recombination Reactions Sometimes this expression is multiplied by a di- mensionless function F (p, T ) to more accurately A special set of three-body reactions are ther- mal ionization and three-body recombination re- 5 actions, which proceed by the pair of equations (analogous to Eq’s (7) and (8)): A + M → Z+ + e− + M, A+ + e− + M → Z + M. (14) (15) For which we again use published rates wherever possible for the ionization reactions, (Eq. (14)), but in many cases here use the simple approxima- tion: k0 = 8πe8 mekBT!1/2 e−I/kB T , (16) where e = 4.9032 × 10−10 esu is the elementary charge, kB = 1.38065 × 10−16 erg/K is the Boltz- mann constant, me = 9.1084× 10−28 g is the mass of the electron, I is the ionization energy (here in units of erg) which we determine from the change in the Gibbs free energy for the reaction. k∞ is then estimated from k0. Three-body recombination and ionization reac- tions have been well studied, and in many cases have well-characterized rate constants. Here we treat the three body recombinations as the reverse reactions for the collisional ionization reactions, but the studied rate coefficients for these reac- tions generally have a temperature dependence of T −4.5, at least for T > 1 K (Hahn 1997). This cre- ates a problem for reversibility. Using these rates will not allow us to reproduce chemical equilibrium for plasmas and this is largely because we are not properly treating the time-dependent plasma con- ditions in which these rates are often measured. Many of these rate constants may accurately de- scribe the time to achieve an equilibrium electron density in a regime where a strong ionizing source has recently been removed from the environment. With this in mind, we instead set the recom- bination rate constants such that, when disso- ciative recombination reactions are disabled, the Saha equation is upheld. These reactions and rate coefficients are also given in Appendix A. The ionization reactions are labeled ‘ti’ and numbered 421-576. As with Sec- tion 2.2, the odd reactions are k0 [cm6 s−1] and the even numbers are k∞ [cm3 s−1]. Finally, we include a series of dissociative re- combination reactions, which take the form: A+ + e− → Y + Z. (17) 6 These have rate constants parameterized in the form of Eq. (5). The reverse reactions can in prin- ciple be calculated, and their rate constants could be calculated straight-forwardly using the same principles used for the three-body reactions. This would effectively be analogous to the rates of three body recombination for any third body, and we do not find that reversing these reactions changes the results much. When we compare with chemical equilibrium, however, we disable these reactions. The dissociative recombination reactions are taken only from the OSU 09 2010 high temperature net- work (Harada et al. 2010), and shown in Appendix A, numbered 2777-2973, and labeled ‘dr’. 2.4. Photochemistry and Cosmic Ray Chem- istry Photochemistry is considered for the species H, H−, He, C, C(1D), C(1S), N, O, O(1D), O(1S), H−, C2, CH, CN, CO, H2, N2, NO, O2, OH, CO2, H2O, HO2, HCN, NH2, NO2, O3, C2H2, H2CO, H2O2, NH3, NO3, CH4, HCOOH, HNO3, N2O3, C2H4, C2H6, CH3CHO, C4H2, C4H4, Na, K, and HCl. The photoionization and pho- todissociation cross-sections are taken almost en- tirely from PhIDRates4 (Huebner & Carpenter 1979; Huebner et al. 1992; Huebner & Mukherjee 2015), with the exception of C4H2, C4H4 and N2O3, the cross-sections of which are taken from the MPI-Mainz UV/VIS Spectral Atlas5 (Keller-Rudek et al. 2013). We divide the cross-sections between 200 bins each ≈ 50 A wide. A comparison between our binned cross-sections and the raw cross-sections from PhIDRates is plotted for an example reac- tion (Figure 1). The cross-sections, both in the database and here, of the form σ(λ) with σ in units cm2 and wavelength in units of A. The reso- lution for the UV cross-sections is fairly low, and cannot encapsulate the fine structure of the UV emission lines or the UV cross-sections. This is especially important when treating ionospheres of gas giants, since, e.g. the fine structure in the H2 bands leave small spectral windows through which photons can penetrate and effectively ion- ize deeper in the atmosphere. Such a low resolu- tion spectrum will effectively close these windows 4phidrates.space.swri.edu 5http://satellite.mpic.de/spectral_atlas/index.html and underestimate the ion production in the iono- sphere (Kim & Fox 1994; Kim et al. 2014). High resolution is also a important for capturing where the UV flux and cross sections both peak; a low resolution cross section can in this case underesti- mate the destruction rate of the species with this resonant photochemical cross-section. As can be seen below, these issues do not significantly af- fect the comparisons of this model for HD209458b, Jupiter or Earth. For photoionization deep in the atmosphere, where high resolution is essential, the network itself need not be modified. The trans- port of UV photons line by line would need to be calculated. The tabulated chemical cross-sections are com- bined with F (λ, z) [photons cm−2 s−1 A−1], the radiant flux density onto a unit sphere (hereafter called the actinic flux) located at atmospheric height, z [cm], to determine the photochemical rate constants, kph,i(z) = τf Z 104A 1A σi F (λ, z) dλ, (18) where i is indexed over the molecules listed above, for which photochemistry is considered. τf is a di- mensionless parameter representing the fraction of time (over a period much longer than the longest characteristic time scale for the atmosphere) the particular atmospheric region is irradiated; for tidally locked planets, τf = 1 (dayside) or 0 (night- side), the diurnal average for a rotating planet is τf = 1/2. The photoionization and photodis- sociation reactions are listed in Appendix A, re- actions numbered 2570-2693, and labeled ‘pi’ for photoionization reactions & ‘pd’ for photodissoci- ation reactions. Cosmic ray ionization and dissociation is pa- rameterized by ζ (Rimmer & Helling 2013), to treat both direct ionization by galactic cosmic rays and ionization by secondary particles produced in air showers. The cosmic ray ionization rate de- pends on the chemical species in question, since different species will have different chemical cross- sections for the photons produced by cosmic rays, and this is accounted for by multiplying ζ(z) by a constant κCR,i such that: kCR,i(z) = κCR,iζ(z). (19) We treat low energy cosmic rays (E < 1 GeV) for these objects as though they have been sig- 7 nificantly shielded by the astrospheres of the host stars, and therefore set the fitting parameters for the incident cosmic ray flux to α = 0.1 and γ = −1.3 in the equation for the flux of cosmic ray particles: , if Ecut < E < E2 if E < Ecut , if E > E2 0, (20) j(E) =   p(E1)(cid:19)γ j(E1)(cid:18) p(E) p(E1)(cid:19)γ (cid:18) p(E) p(E2)(cid:19)α j(E1)(cid:18) p(E2) √E2 + 2EE0, E0 = 9.38 × 108 where p(E) = 1 c eV, E1 = 109 eV, and E2 = 2 × 108 eV, and the flux at E1 is set to j(E1) = 0.22 cm−2 s−1 sr−1 (GeV/nucleon)−1. All of these parame- ters except α are observationally well-constrained (Indriolo et al. 2009). For a demonstration of how α affects the cosmic ray spectrum, and a discus- sion of the Monte Carlo transport we use for cos- mic rays of energy < 1 GeV, see Rimmer et al. (2012); Rimmer & Helling (2013). For ioniza- tion rate by cosmic rays of energy > 1 GeV, QHECR [cm−3 s−1], we use the analytical method of Velinov & Mateev (2008). Cosmic ray reactions are listed in Appendix A, numbered 2694-2776, and labeled ‘cr’. 2.5. Test for Chemical Equilibrium At sufficiently high temperatures and pressures, a gas should rapidly settle into chemical equilib- rium. An important test for a chemical network is that its steady state solution converges to the chemical equilibrium solution. To perform this test of our network, we solve the chemical kinetics at a single (T, p) point, using the rate constants from the Stand2015 network, disabling the cos- mic ray reactions, photochemistry and dissociative recombination. We compute a time-dependent so- lution of the equation dni dt = Pi − Li. (21) We solve this equation for T = 1000 K and p = 1 bar, with solar abundances from Asplund et al. (2009). We compare our results to chemical equi- librium calculations using the Burcat polynomials (Burcat & Ruscic 2005), and plot our comparisons in Figure 2 and find excellent agreement. This agreement is not surprising; we have used the same thermochemical data to reverse our reactions, and only include reversed reactions in this test, so once the system achieves steady state, computationally achievable at this pressure and temperature, the chemistry has effectively settled into equilibrium. We also compare our electron number density to the electron number density achieved using the Saha equation, this time at a pressure of 10−4 bar and over a range of temperatures from 1000 K to 10000 K. This comparison is plotted in Fig- ure 3. The comparison is virtually perfect when T & 2000 K, unsurprising given the way the three- body recombination reactions are calculated (see Section 2.3). At ∼ 1000 K, our results diverge from the Saha equation. This is because the integrator does not reliably calculate mixing ra- tios below ∼ 10−30. Indeed, at this stage, the electron number density achieves ∼ 10−300 cm−3 while the H+ number density rests at ∼ 10−60 cm−3, producing significant charge balance errors. These large errors in the charge balance fluctuate, and only appear when the ionization fraction is . 10−30, at which point ion-neutral chemistry is inconsequential. 3. 1-D Photochemistry/Diffusion Code We have developed a simple 1D photochem- istry/diffusion code (Argo), for the purposes of testing the Stand2015 network. The required in- puts for Argo are: • (p, T ) profile of the atmosphere. • Vertical eddy diffusion (Kzz [cm2 s−1]) pro- file of the atmosphere (see discussion in Lee et al. 2015). • Atmospheric elemental abundances. • Boundary conditions at top and bottom of the p, T profile. • Actinic flux6 at the top of the atmosphere. • Chemical Network (in our case, Stand2015). • Initial chemical composition 6The actinic flux is the radiance integrated over all angles, expressing flow of energy through a unit sphere. There are subtle differences between the actinic flux and the spectral irradiance, see Madronich (1987). All of these inputs except the chemical composi- tion are fixed. With these inputs, Argo solves molecular transport in a fully Lagrangian manner, similar to Alam & Lin (2008) and Zahnle et al. (1995). The model consists of two parts: (1) A chemical trans- port model (Section 3.1) and (2) calculation of the photochemical and cosmic ray chemical rate con- stants from cross-sections and a depth-dependent actinic flux (Section 3.2). A conceptual illustra- tion is shown in Figure 4. 3.1. The Continuity Equations for Chem- ical Species The coupled 1D continuity equations describing the time-dependent vertical atmospheric chem- istry are: , (22) ∂ni ∂t = Pi − Li − ∂Φi ∂z where ni [cm−3] is the number density of species i, and i = 1, ..., Ns, and Ns is the total num- ber of species. Pi [cm−3 s−1] is the rate of pro- duction and Li [cm−3 s−1] is the rate of loss of species i. The right-most term is the vertical change in flux Φi [cm−2 s−1], and represents the flux due to both eddy (K [cm2 s−1]) and molecu- lar diffusion (D [cm2 s−1]) respectively, related as (Banks & Kockarts 1973, their Eq. (15.14)), ∂z Φi = − Kh ∂ni − Dh ∂ni ∂z H0 + ni(cid:16) 1 + ni(cid:16) 1 Hi + + dT 1 T 1 + αT dz(cid:17)i T dT dz(cid:17)i, (23) where H0[cm] is the pressure scale height of the at- mosphere at z [cm], Hi[cm] is the molecular scale height of the atmosphere for species i, and αT is the thermal diffusion factor (Banks & Kockarts 1973; Yung & Demore 1999; Zahnle et al. 2006; Hu et al. 2012). For molecular diffusion coeffi- cients, we adopt Chapman-Enskog theory (Enskog 1917; Chapman & Cowling 1991). Eddy diffusion coefficients are either determined empirically, as with Earth and Jupiter, or are derived from global circulation models, as is the case for HD209458b. In Eq. (23), the terms dealing with eddy dif- fusion and molecular diffusion are separated out, clarifying four regions that Eq’s (22), (23) de- scribe. (1) Deep within the atmosphere, where pressures and temperatures are sufficiently large, 8 the thermochemistry dominates, and the equa- tion simplifies to Equation (21). The atmo- spheric chemical composition converges to chem- ical equilibrium or at least to some stable quasi- equilibrium. (2) Higher in the atmosphere, the eddy diffusion may dominate, and the species are quenched, their abundance mixed evenly over a wide range of the atmosphere at time-scales shorter than the chemical time-scales. (3) Above this region, molecular diffusion may dominate, and at that point, species lighter than the mean molecular mass of the atmospheric gas will rise up, and species heaver than the mean molecular mass will settle down, and the chemistry will largely be determined by the individual scale heights of the atmospheric constituents. (4) Non-equilibrium processes, such as photochemistry or cosmic ray chemistry, may create a fourth region, the com- position of which is determined by irreversible chemical reactions. Since the purpose of this paper is to introduce a new chemical kinetics network for lightning and prebiotic processes, our focus is not on the atmo- spheric dynamics (for this, see Lee et al. 2015). We therefore apply a simple approximation to Eq. (22), inspired by Alam & Lin (2008). We first cast Eq. (22) in a Lagrangian formulation, and con- sider Eddy diffusion to be moving small parcels of the gas vertically. We follow a single parcel as it moves up from the lower boundary of the tem- perature profile, and then returns down again. In reality, the parcel would be jostled in all three di- mensions as it makes a complex journey up to the top of the atmosphere, but 1D transport models are unable to capture this effect in full. The differential diffusion of molecules into and out of the parcel requires a different approach. The discrete formulas used by Hu et al. (2012, their Eq. 9) in the Lagrangian frame are: ∂ni,j ∂t = Pi,j − Li,jni,j − dj+1/2 −(cid:16)dj+1/2 ngas,j+1/2 ngas,j − dj−1/2 ngas,j−1/2 ngas,j−1 ni,j−1. + dj−1/2 ngas,j+1/2 ngas,j+1 ni,j+1 ngas,j−1/2 ngas,j (cid:17)ni,j (24) Here, j represents the parcel being followed, j − 1 the parcel directly beneath j, j+1 the parcel above j, and j±1/2 an arithmetic average between j and j ± 1. n without any i subscript represents ngas at 9 the relevant parcel, and dj±1/2 = Dj±1/2 2(∆z)2h ( ¯m − mi)g∆z kBTj±1/2 − αT Tj±1/2 (Tj±1−Tj)i. (25) ¯m[g] denotes the mean molecular mass of the at- mosphere at z and mi[g] the mass of species i. Both the third and last terms on the R.H.S. of Eq. (24) do not depend on ni and can therefore be treated as source terms, Pi. The fourth term can be treated as a term in Li, such that molecules “destroyed” by this reaction are “banked”, A → BA. The “banked” molecules re-enter the parcel at a rate determined by the third and last terms on the R.H.S. of the equation, thus conserving mass throughout the parcel’s travels. Violations of this conservation do not appear here, but can be accounted for via further reactions, settling, con- densation and evaporation, outgassing and escape, discussed in Appendix C. Although it is straight- forward to handle atmospheric escape with this method, we do not do so for any of the test cases in this paper. Equation (24) is solved within Argo in the same numerical manner as Nahoon (Wakelam et al. 2012), by the implicit time-dependent Gear method as incorporated by the Livermore Solver for Ordinary Differential Equations (DLSODE) (Gear 1971; Brown & Hindmarsh 1989). 3.2. Calculating the XUV and Cosmic Ray Flux Once the fluid parcel has completed the atmo- spheric profile, the solar XUV actinic flux from 1 A to 10000 A as a function of depth, z [cm]7 and wavelength λ [A] is calculated. We consider both the direct and approximate diffusive actinic flux. The local height-dependent actinic flux is calcu- lated without any iteration on the local tempera- ture. The cross-sections for various photochemical reactions (Sect. 2.4), are multiplied by each ver- tical step (∆z)j [cm], where (∆z)j is the size of the step at height zj. The total optical depth as a function of the wavelength takes the form τ (λ, z) = Σj(cid:2)(∆z)jΣiσinij(cid:3) + τs, (26) 7The depth for this model extends from z = 0, the bottom of the temperature profile for the planet in question, to z = ztop, the top of the profile. where i is summed over all species for which pho- toabsorption is considered (see Section 2.4 for a list of these species). τs is the optical depth due to Rayleigh scattering, and the actinic flux as a function of depth is defined as (Hu et al. 2012): F (λ, z) = F (λ, ztop)e−τ (λ,z)/µ0 + Fdiff , (27) where µ0 = cos θ, where θ is the stellar zenith an- gle; we set µ0 = 1/2 for all calculations within this paper (see Hu et al. 2012, their Fig. 7). Fdiff de- notes the actinic flux of the diffusive radiation, de- termined using the δ-Eddington 2-stream method (Toon et al. 1989). Once the actinic flux is cal- culated, the photochemical rates are determined as in Section 2.4. Once the depth dependent flux, F (z, λ) [cm−2 s−1 A−1], is determined for all lay- ers, the parcel’s path through the atmospheric profile is repeated, now accounting for the pho- tochemistry. The cosmic ray ionization rate, ζ(z) [s−1] is likewise calculated in a depth-dependent manner following Rimmer & Helling (2013) and incorporated into the chemistry (Sec. 2.4). A new depth-dependent composition is con- structed, then applied to Eq. (26) to solve again for F (z, λ). The value of ζ(z) does not change significantly between iterations. This process is repeated until the results converge; i.e. until the profile from the previous global calculation (trans- port + depth-dependent flux) agrees to within 1% the profile from the current global calculation. The number of repetitions depends on the param- eters, but is typically between 5 and 12 global it- erations. This iterative process is represented as a flow-chart in Figure 5. This method is both simple and functional, re- quiring relatively little computational resources. It is also straight-forward to adapt to diverse chemical environments, since it does not require the selection of “fast” and “slow” chemistry to ease computational speed. These strengths do not come without a cost: The simplistic dynamics does not transition as smoothly from the eddy dif- fusion regime to the molecular diffusion regime as the Eulerian formulation, and can result in steep changes over a handful of height-steps. 3.3. Testing the Atmospheric Transport Model for Molecular Diffusion In order to benchmark the Stand2015 chemi- cal network in different planetary atmospheres, we 10 test the molecular diffusion within Argo. We con- sider a 1D isothermal gas under a constant surface gravity, g = 103 cm/s2, with temperature T = 300 K, at hydrostatic equilibrium. The gas is initially composed of carbon and hydrogen atoms, each with a mixing ratio of X0(C) = n(C)/ngas = 0.5 and X0(O) = 0.5 throughout. All chemistry is disabled. It is expected that the heavier species, carbon, will settle into the atmosphere, and the lighter species, hydrogen, will rise up, until they stratify. The analytic solution to this system is well-known. The mixing ratio should be de- termined by the scale-heights of the individual species such that, for the carbon abundance: X(C) = X0(C)e−z/HC X0(H)e−z/HH + X0(C)e−z/HC , (28) where X(C) is the final steady state carbon mix- ing ratio, and HH [cm] and HC [cm] are the atmo- spheric scale heights for the hydrogen and carbon. The code is run until steady state is achieved, when the carbon in the very upper atmosphere dif- fuses into the lower atmosphere. The steady-state mixing ratio, as a function of height is compared the analytic mixing ratio, Eq. (28), in Figure 6. The comparison is reasonable through the extent of the atmosphere. 4. Testing the Network for Planetary En- vironments The Stand2015 network contains chemical re- actions for an H/C/N/O gas, and including both highly reducing to highly oxidizing atmospheres, and for a temperature range of 100 K to 30000 K. The network should then be tested for a vari- ety of planetary atmospheres with different chem- ical compositions, from the (probably) oxidizing atmosphere of the early Earth to the highly re- ducing atmosphere of Jupiter. The large range of temperatures is tested for the irradiated exoplanet HD209458b. We also test our model against the height dependent measurements of select trace species within the atmosphere of the present-day Earth. It would be interesting to apply our model to Titan, due to its rich nitrile and organic chem- istry. Titan’s atmosphere is a very rich and com- plex environment, and it is important to account for these complexities when modeling Titan. Ti- tan has upper atmospheric hazes, temperatures low enough to condense several molecular species, and ionization and dissociation by energetic parti- cles including cosmic rays, Saturn magnetospheric particles, solar wind protons and interplanetary electrons. As useful as a study of the atmosphere of Titan would be for exploring Miller-Urey-like chemistry (Waite et al. 2007), such a model is be- yond the scope of this paper. The boundary con- ditions for these various objects are given in Sec- tion 4.1. We then compare our results to the results from other chemical kinetics models and, where possible, with observations, for HD209458b (Section 4.2), Jupiter (Section 4.3), and the Earth (Section 4.4). 4.1. Boundary Conditions for Three Test Cases: HD209458b, Jupiter and the Earth Below, we compare the results of our chemi- cal kinetics to other results for HD209458b and also for Jupiter and the Earth. Each of these ob- jects has different boundary conditions and pa- rameters. These conditions and parameters in- clude the temperature profile of the object’s at- mosphere, the eddy diffusion profile, the elemental abundances, the initial composition at the lower boundary of the atmospheric profile, and the unat- tenuated UV flux. For HD209458b, the conditions at the lower boundary of the atmospheric profile rapidly develop from the prescribed initial condi- tions toward chemical equilibrium. For Jupiter and the early Earth, the composition at the lower boundary is stable over the dynamical time-scale (dni(z = 0)/dt ≈ 0), and so the initial compo- sition effectively acts as a lower boundary con- dition. The assumed elemental abundances and initial conditions at the lower boundary of the at- mospheric profile are given in Table 1. We take HD209458b to have solar elemental abundances throughout its atmosphere, and set the initial conditions at the lower boundary of the atmosphere to be entirely atomic. The ini- tial composition hardly matters here, since the composition quickly settles to chemical equilib- rium at such a high temperature and pressure. The temperature profile and eddy diffusion profile for HD209458b are both taken from Moses et al. (2011), so that we can directly compare results. Since HD209458 is a G0 star, we use the so- lar UV flux. The unattenuated solar UV flux 11 at 1 AU is obtained from the SORCE data (Rottman et al. 2006) for 1 A - 350 A and 1150 A - 10000 A with data from PhIDRates for the 350 - 1150 A range. The binned flux we use is plotted in Figure 7. This flux is adapted to HD209458b by multiplying the solar UV flux by a factor of (d⊕/dp)2, where d⊕ [AU] is the distance from the Earth to the Sun and dp ≈ 0.047 AU is the ap- proximate distance between HD209458b and its host star. This may not be the most accurate approximation to the UV behavior of HD209458, since it might have quite different activity from our sun (Tu et al. 2015). For Jupiter, we use the temperature and eddy profiles from Moses et al. (2005). For consistency, we set the initial conditions at the lower bound- ary of Jupiter’s atmosphere to be the same as Moses et al. (2005); see Table 1. The solar UV spectrum at 1 AU is used for Jupiter, although multiplied by a factor of (d⊕/dJ )−2, where dJ ≈ 4.5 AU is the square of the distance between the sun and Jupiter. For the present-day Earth, we use the measured surface mixing ratios from the US Standard At- mosphere 1976 (see Table1) and the temperature profile from Hedin (1987, 1991), Fig. 13. We use the present day solar flux at 1 AU as our incident UV flux. We use the same chemical lower boundary con- ditions as from Kasting (1993) for the atmosphere of the early Earth (Table 1). The temperature pro- file for the early Earth is assumed to be the same as that of the present Earth (Hedin 1987, 1991), Fig. 13. The UV field used for this model is that of the young Sun calculated using the scaling re- lationships of Ribas et al. (2005) for wavelengths between 1 A – 1200 A and the UV field of the solar analogue κ1 Cet above 1200 A (Ribas et al. 2010). 4.2. HD209458b HD209458b was first observed by Henry et al. (2000), and is one of a growing number of Hot Jupiters to have a measured spectrum, via tran- sit (e.g. Queloz et al. 2000), and also in emis- sion (e.g. Knutson et al. 2008). Various molecu- lar species have been tentatively identified in the spectrum, such as TiO (D´esert et al. 2008), water (Madhusudhan & Seager 2009; Swain et al. 2009; Beaulieu et al. 2010), CO, CO2 and methane fea- tures (Madhusudhan & Seager 2009; Swain et al. 2009). HD209458b has been extensively modeled, with retrieval modeling (Madhusudhan & Seager 2009), and with hydrodynamic global circulation models (Showman et al. 2008). This planet has also been a popular target for non-equilibrium chemistry models such as those of Liang et al. (2003); Zahnle et al. (2009); Moses et al. (2011); Venot et al. (2012); Ag´undez et al. (2014) and Lavvas et al. (2014). We have chosen the atmosphere of HD209458b as one candidate for benchmarking our results be- cause it is well characterized and has been the subject of several non-equilibrium chemistry mod- els, and it has a very high temperature even among Hot Jupiters. An additional benefit to HD209458b is its suspected temperature inversion (Knutson et al. 2008, although this is debated, see also Schwarz et al. 2015), which allows us to test our chemistry at very high temperatures both at both high and low pressures. The thermal profile of HD209458b from Moses et al. (2011) is shown in Figure 8. The local gas-phase temperature T > 2000 K both when p > 100 bar and when the gas-phase pressure, p < 10−4 bar. This is a wide parameter space relevant for ion-neutral chemistry initiated via thermal ionization. We compare our results to the predictions of two different chemical kinetics models. (1) We compare our results to the results of Moses et al. (2011) with the ion-neutral chemistry disabled. (2) We compare the ionic abundances for our most abundant ions to the results of Lavvas et al. (2014). Also in this case, we disable cosmic ray chemistry in order to draw a better comparison to the ion-neutral chemistry. We compare our network and transport model to Moses et al. (2011) by examining the volume mixing ratios of major neutral species: H, H2, He (hydrogen/helium chemistry), OH, H2O, O and O2 (oxygen/water chemistry), N2 and NH3 (ni- trogen chemistry), and CO, CH4 and CO2 (car- bon chemistry). See Figure 9. These species were chosen because they are abundant and, in the case of H2 and N2, play an important role in the non- equilibrium chemistry. N2 provides the reservoir for the transition between N2 ⇋ NH3. Other species were chosen because they contribute to fea- tures observed in transit spectroscopy, (e.g. CO2). The molecules CO and H2O do both. Helium was 12 chosen because its mixing ratio is not significantly affected by the chemistry. It changes with pres- sure due to molecular diffusion, and so it provides a useful comparison between our dynamical calcu- lations and those of Moses et al. (2011). The transition of carbon between CO and CH4, and nitrogen between N2 and NH3 is very sensitive to non-equilibrium chemistry, as CH4 ≈ CO when p ∼ 100 bar and T ∼ 2000 K. As the pressure decreases rapidly while the temperature remains relatively high (T > 1000 K), the thermochemi- cal equilibrium ratio for CH4/CO plummets, ap- proaching 10−7 at 0.1 bar in the HD209458b atmo- sphere. The time it takes the carbon to meander from CH4 to CO, however, becomes significantly longer than the relevant dynamical timescales (for HD209458b, this time-scale is prescribed by the eddy diffusion coefficient, see Bilger et al. 2013), and the CH4 and CO abundances are quenched. The same sort of process governs the transition of nitrogen from N2 to NH3. The pathways for both CH4 ⇋ CO and N2 ⇋ NH3 interconversions are not well understood. In both cases, the paths competing with one an- other are often circuitous, and tend to be regulated by one of several reactions encountered along the journey, a slow rate-limiting step (Moses 2014). The time-scale of the transition between species is almost entirely set by the rate by which that single reaction proceeds. As discussed in Section 2, rate coefficients can be frustratingly uncertain, with different estimations sometimes varying by more than an order of magnitude. For example, com- pare the rate experimental and theoretical rate constants for C2H6 → CH3 + CH3 (Yang et al. 2009 and Kiefer et al. 2005, respectively). The path that one believes regulates these central tran- sitions can be very different depending on what rate coefficients are used. An illustrative example is the reaction CH3 + H2O → CH3OH + H. Hidaka et al. (1989) has determined the rate for CH3 + H2O → Products, Reaction (29), proceeds with a barrier of ≈ 2670 K (see Visscher et al. 2010, for a discussion on this reaction). With reasonable assumptions of the branching ratios for this reaction, namely that the branching ratios do not change much with temperature, one would set the same bar- rier to CH3 + H2O → CH3OH + H, as done by Venot et al. (2012). However, Moses et al. (2011) carried out quantum chemical calculations for this reaction using MOLPRO and estimate a barrier for this particular branch of ≈ 10380 K, much larger than the activation energies of the other branches. With the smaller barrier, the path car- bon takes from CH4 to CO proceeds as: H2 + M → H + H + M CH4 + H → CH3 + H2 CH3 + H2O → CH3OH + H CH3OH + H → CH2OH + H2 CH2OH + M → H2CO + H2 + M H2CO + H → HCO + H2 HCO + H → CO + H2 HCO + M → CO + H + M (29) CH4 + H2O → CO + 3H2. (30) We adopt the rates of Moses et al. (2011) for this pathway, as well as the smaller rate coefficient for the three-body reaction H2O + CH2 + M → CH3OH. An examination of our results would re- veal that, as with Venot et al. (2012), the tran- sition of carbon from CH4 to CO is much more efficient than with Moses et al. (2011). We have examined the rates at which reactions proceed in our network and find another formation pathway: H2 + OH ↔ H2O + H OH + O → O2 + H CH4 + H → CH3 + H2 CH3 + H → CH2 + H2 CH2 + O2 → COOH + H COOH + H2O → CH2O2 + OH CH2O2 + M → CO2 + H2 + M CO2 + H → CO + OH (31) CH4 + O → CO + 2H2, (32) The atomic oxygen arises from thermal dissocia- tion of OH or photodissociation of H2O followed by diffusion downward. This pathway is critically dependent on Reaction (31). To our knowledge, the three-body rate coefficient for this reaction has not been determined. This reaction has in- stead appeared in our network as the reverse reac- tion of CH2O2 + OH → COOH + H2O, for which 13 we use an estimate based on reaction energetics (Mansergas & Anglada 2006). This pathway is highly uncertain, and removing it makes up the majority of the difference between our results and those of Moses et al. (2011) for methane between 1 – 10−4 bar. We suspect further differences owe to our different thermochemical constants and the use of slightly different solar abundances. The path of nitrogen from NH3 to N2 is con- siderably more uncertain. The path is believed to roughly follow from NH3 to NH via hydrogen ab- straction, which will in turn react with another NHX species to form N2HY . This species will be destroyed either by reacting with hydrogen or via thermal decomposition, to form N2. The reac- tions N2HX+2 → NH2 + NHX involve large un- certainties, which result in variations of the NH3 quenched abundance by an order of magnitude. We find, similar to Moses et al. (2011), that: H2 + M → H + H + M NH3 + H → NH2 + H2 NH2 + H → NH + H2 NH2 + NH → N2H2 + H N2H2 + H → NNH + H2 NNH + M → N2 + H + M (33) 2NH3 → N2 + 3H2 (34) with Reaction (33) as the rate limiting step. The profile we have for NH3 deviates considerably from the results of Moses et al. (2011), but this is for large part due to a difference in the nitrogen thermochemistry and initial abundances at high pressures propagating up through the atmosphere. Figure 9 shows that our quenching height is in both cases higher than for Moses et al. (2011), suggesting that the nitrogen in NH3 migrates to N2 more slowly in our network, even overtaking Moses et al. (2011) at ∼ 10−4 bar, but that we start with less NH3 than Moses et al. (2011). The increase in NH3 abundance at ∼ 5 × 10−6 bar is due to a formation path for NH3 in Moses et al. (2011) that is less efficient in our network. We conclude this section with a brief discus- sion of the most neutral ions, in comparison with Lavvas et al. (2014). We have plotted the most abundant ions in Figure 10. Note that, for this paper, ngas is a sum of all neutral gas particles, cations, ions and electrons, so the mixing ratio of ions cannot increase above unity. This plot allows a direct comparison to Lavvas et al. (2014, their Fig’s 5 & 6). In our model, K+ is the most abun- dant ion deep within the atmosphere, followed by Mg+ and Fe+. Lavvas et al. (2014) does not con- sider these species, but they don’t seem to af- fect the abundances of other ions very much deep within the atmosphere. When the pressure delves to 10−2 bar, K+ deviates considerably between our results and those of Lavvas et al. (2014). This is likely due to the inclusion of several other ions in our model that become dominant charge carriers at this height, including several complex hydro- carbon ions, of the form CnH+ m. This indicates that ion-neutral chemistry can be significantly in- fluenced by the variety of ions and neutral species under consideration. This will be especially true for the potassium chemistry. Our network con- tains a small number of potassium-bearing species. Including new species and reactions could signifi- cantly affect the degree of ionization. It will be interesting to discover how an expanded potas- sium and sodium chemistry affects the overall ion- neutral chemistry and the resulting abundances of trace species. Between 10−3 and 10−4 bar, Na+ overtakes K+ as the dominant positive charge carrier, and re- mains so until ∼ 10−7 bar. This transition, the ratios between the ions, and the abundances of the ions, are nearly identical between our model and that of Lavvas et al. (2014). Within the thermo- sphere of HD209458b, there are some small dis- crepancies between our model and Lavvas et al. (2014) for He+, and quite large discrepancies for C+ which we suggest are owing to the non-Alkali photochemistry that Lavvas et al. (2014) include, but that we have not included here. 4.3. Jupiter The atmosphere of Jupiter is divided into three regions: (1) the troposphere, where the gas-phase temperature T decreases with atmospheric height, (2) the stratosphere, where T is roughly con- stant with increasing height, and (3) the ther- mosphere, where T increases with height. In this section, we consider the chemical composi- tion of Jupiter’s stratosphere. The stratosphere of Jupiter is rich in hydrocarbons, owing to its large gas-phase C/O ratio, because the majority of the 14 oxygen is locked in water ice and then gravita- tionally settling to below the tropopause. This is predicted to lead to a C/O ∼ 2× 106 (Moses et al. 2005) in the absence of external sources of H2O and CO2 (Feuchtgruber et al. 1997; Moses et al. 2000a,b) such as Shoemaker-Levy 9 (Cavali´e et al. 2012). Jupiter’s stratosphere provides an ex- treme example of how surface deposition can rad- ically affect the C/O ratio, an effect more re- cently predicted for exoplanets and brown dwarfs (Bilger et al. 2013; Helling et al. 2014). The high C/O ratio, in combination with the large abun- dance of hydrogen (H2 and CH4 are the two most abundant volatiles in the stratosphere and lower thermosphere), means that the stratosphere of Jupiter is strongly reducing (Strobel 1983). Fouchet et al. (2000) have observed ethane and acetylene in Jupiter’s stratosphere. Ethylene has also been observed by B´ezard et al. (2001). The stratospheric chemistry of Jupiter has been mod- eled by several groups, including Gladstone et al. (1996) and Moses et al. (2005). We adopt the lower boundary conditions and temperature pro- file that Moses et al. (2005) used and model the carbon-oxygen chemistry in the stratosphere of Jupiter, ignoring the nitrogen chemistry (most of the nitrogen will be locked in NH3 ice). Boundary conditions are discussed in Section 4.1. Our lower boundary is set to be identical to Moses et al. (2005). These boundary conditions are somewhat artificial; the carbon budget is con- trolled by the photochemistry and the dynamics. There is no effective destruction pathway for the stable hydrocarbons, but the time-scale for their formation is often competing with the dynamical time-scales. In the thermosphere, ∼ 10−7 − 10−8 bar, these hydrocarbons are lost through photodis- sociation and photoionization as well as molecular diffusion. At the base, the chemistry is halted once the dynamical timescale is reached, effectively treating the bottom boundary as an open bound- ary through which the hydrocarbons would con- tinue to diffuse. In reality, the complex hydrocar- bons are carried into Jupiter’s deep atmosphere, where the high temperatures and pressures dis- sociate these hydrocarbons, and force the carbon budget to return to chemical equilibrium values: CH4 with trace amounts of CO and other species. Visscher et al. (2010, their Fig. 6) demonstrate how the carbon budget is set deep within Jupiter’s atmosphere; we do not model this region. With these reactions removed from the net- work, we ran the network using the tempera- ture and Kzz profiles from Moses et al. (2005), shown in Figure 11. Comparisons between our results and a representative set of observations for the depth dependent mixing ratios, for the species CH4, C2H2, C2H4, C2H6 and C4H2, are shown in Figure 12. The observations for CH4 are taken from Drossart et al. (1999) and Yelle et al. (1996), C2H2 observations are from Fouchet et al. (2000), Moses et al. (2005) and Kim et al. (2010), C2H4 observations are from Romani et al. (2008) and Moses et al. (2005), C2H6 observations are from Fouchet et al. (2000), Moses et al. (2005), Yelle et al. (2001) and Kim et al. (2010), and the C4H2 observations are from Fouchet et al. (2000) and Moses et al. (2005). We also incorpo- rate observations for C2H2, C2H4 and C2H6 from Gladstone et al. (1996) and references therein. Many of the published observations do not in- clude error bars in atmospheric pressure. Ad- ditionally, there may seasonal in the pressure- temperature structure and the location of the ho- mopause, which adds uncertainty to our predic- tions as a function of pressure. To account for these sources of uncertainty, we place error bars for the pressure at a factor of two above and below the published observations when errors in pressure were not given. These errors in pressure are of the same order as observations where errors in pres- sure are given. We do not compare our results for oxygen-bearing species, because the abundances of these species are expected to be greatly enhanced in the stratosphere by the addition of an external source of oxygen, such as Shoemaker-Levy 9. The differences between our results and those of other models arise primarily because of differ- ent photochemistries and different rate constants, especially for the re-formation of methane after its photodissociation, CH3 + H2 → CH4 + H, and CH3 + H + M → CH4 + M. (35) (36) Differences between Jovian photochemical mod- els can result in very large discrepancies between stratospheric abundances of complex hydrocar- bons. The differences between Gladstone et al. (1996) and Moses et al. (2005) span several orders 15 of magnitude in some cases (see Moses et al. 2005, their Fig. 14). Both ethane and acetylene agree reasonably well between our model and the observations, and the results for C4H2 lie more than a factor of five below the observational upper limits. Our predictions for the location of the methane ho- mopause do not agree very well with observations. We use the eddy diffusion coefficient from Model C in Moses et al. (2005), and either this or the use of the Chapman-Enskog diffusion coefficient for Methane may be the source of the discrep- ancy. Our results are similar to the Model C re- sults of Moses et al. (2005, their Fig. 14). The molecule with the largest discrepancy between the two models is ethylene (C2H4), with the largest discrepancy between our predictions and the 1 mil- libar observations (ignoring the observation from Gladstone et al. 1996 that predicts a mixing ratio of ∼ 10−8). In our model, the primary path of for- mation for ethylene follows from the photodissoci- ation of ethane (Reaction 2679 in the network), C2H6 + γ → C2H4 + H2, (37) and ethane is formed from CH4 following paths to formation like this one: 2(cid:0)CH4 + γ → 1CH2 + H2(cid:1), 2(cid:0)1CH2 + H2 → CH3 + H(cid:1), CH3 + CH3 + M → C2H6 + M; 2CH4 + 2γ → C2H6 + 2H. (38) These differences may be resolved by a more care- ful accounting of pressure-dependent branching ratios, such as those of: H + C2H5 → CH3 + CH3 (39) from Loison et al. (2015). We use the Kooij form for these reactions (Section 2.1), which does not account for the effect that pressure has on the rate constant. Ion-neutral chemistry also makes a contribu- tion, via the formation of C2H4 from the reaction C2H+ 3 + CH4 CH+ 5 + C2H2 → C2H+ 3 + e− + M → C2H3 + M C2H3 + CH4 → C2H4 + CH3, (40) x+1 + H. and CH+ 5 forms from a series of reactions start- ing with the photoionization of CH3 and then a series of hydrogen abstractions, CH+ x + H2 → CH+ It should be emphasized that this is not the primary formation pathway for ethy- lene, but it is an important path of formation in our chemistry and makes some contribution to the mixing ratios at 1 millibar. Finally, there is a large discrepancy for CO, but this is not due to differences in the chem- istry. Rather, this results from Moses et al. (2005) injecting CO, CO2 and H2O into Jupiter’s strato- sphere. Inclusion of this external source of oxygen- bearing species is justified by a number of data- model comparisons mentioned at the beginning of this section. We neglected to include these exter- nal sources, and therefore oxygen-bearing species, especially H2O and CO2 (not shown) fail to agree with observations. Our results therefore sug- gest that some external source of oxygen-bearing species is necessary to explain the H2O and CO2 observations in Jupiter’s stratosphere. 4.4. The Earth The Earth’s atmosphere is well studied, and the profiles of trace species are well constrained, and the formation and destruction of these species is controlled by photochemistry and deposition. Comparing our results to the present day Earth at- mosphere therefore provides a comprehensive test of our chemical network (Section 4.4.1). Addi- tionally, the connection between lightning-driven and NOx chemistry8 has been extensively stud- ied with experiments, observations and models, and provides a useful regime in which to compare the results of Stand2015 applied to a lightning shock model (Section 4.4.2). It is important to find out what our model predicts in habitable en- vironments before the onset of life, and so we apply our model to the Early Earth (Section 4.4.3). 4.4.1. Present Day Earth Atmosphere The best understood planetary atmosphere, in terms of both models and observations, is the at- mosphere of the present day Earth. Earth’s at- mosphere has been studied in situ, with the use of countless balloon experiments used to mea- 8Referring primarily to NO and NO2 chemistry. sure various trace elements, and remotely, with satellite measurements. Models of Earth’s atmo- sphere range from simple to complex, both dy- namically (1D diffusion to 3D global circulation models) and chemically (from treating only oxygen and hydrogen chemistry to modeling the transport and chemistry of chlorofluorocarbons and biologi- cal aerosols). Seinfeld & Pandis (2006) provide a useful introduction and review to the subject. Our interest is in validating our photochemi- cal network to the present-day Earth, and not in coupling Earth’s geochemistry to its atmospheric chemistry. We therefore make some simplifying as- sumptions when we set our boundary conditions. We compare our model to the contemporary Earth by setting the lower boundary conditions, temper- ature profile and external UV field as given in Sec- tion 4.1. We present these comparisons for O3, CH4 and N2O (Figure 14), NO and NO2 (Figure 15) and OH and H2O (Figure 16). The data for O3, CH4 and N2O is taken from the globally averaged mixing ratios from Massie & Hunten (1981). Following Hu et al. (2012), we apply error bars spanning an order of magnitude in mixing ratio to reflect the tem- poral and spatial variations. Our model fits the measured CH4 to within the error bars throughout the atmosphere. The O3 predicted by the model deviates from the data with errors at 15 km, and the N2O deviates from the data with errors be- tween 40 km – 55 km. This may be due to an over-estimation of the optical depth. If more UV photons in the model penetrated through to ∼ 10 km, the O3 mixing ratios would be enhanced at 15 km, and the N2O mixing ratios would be destroyed more efficiently deeper in the atmosphere. The data for NO and NO2 is taken from bal- loon observations at 35 deg N in 1993 (Sen et al. 1998), and here also we apply error bars span- ning an order of magnitude to reflect spatial and temporal variations. As with Hu et al. (2012), we seem to overpredict the abundance of NO in the upper atmosphere (30-40 km). We find that this overprediction is due to Reaction 1300 in the net- work: N2O + O(1D) → NO + NO; k = 7.25 × 10−11 cm3 s−1. (41) We use the rate suggested by the JPL Chemical 16 Kinetics and Photochemical Data for Use in At- mospheric Studies (Sander et al. 2011). If the rate constant for this reaction is decreased by a factor somewhere between 2 and 10, we come into much better agreement at 30-40 km, and worse agree- ment between 20-30 km (see Figure 15). Finally, the data from OH and H2O was taken from balloon measurements at various latitudes and heights in 2005 (Kovalenko et al. 2007). We plot each individual datapoint without error bars in order to represent the observed variations; changes at other points of the globe or at other times of the year or day may lead to more sig- nificant variations in the abundances. The H2O predictions are within a factor of five of the ob- served water abundance, and our OH predictions lie within the measurements, indicating that the model correctly reproduces the water and OH mixing ratios. 4.4.2. Lightning Shock Model and NOx chem- istry It is also useful to to the model’s NOx lightning- driven chemistry in the present day atmosphere. For this purpose, we apply a simple shock model in order to explore the formation of NOx species due to lightning at a single small region in the atmo- sphere. We employ the temperature and pressure calculations of Orville (1968, his Fig’s 1 and 3) and the time-scaled results of Jebens et al. (1992, their Fig’s 2 and 3), fitting these to an exponen- tial function. We use the following functions of temperature and pressure: T (t) = 300 K + (29800.0 K) e−t/(55.56 µs); P (t) = 1.0 bar + (7.0 bar) e−t/(5.88 µs). (42) (43) a We start with present day atmospheric chemistry at the base of the troposphere, except without the N2O, NO and NO2 species, and with T = 300 K and p = 1 bar. The shock occurs at 1 ns, and is allowed to evolve until 0.1 ms. At this point the calculation is terminated, and another calcu- lation initiated using for its initial conditions the final conditions of the shock model, except with temperature and pressure returned to 300 K and 1 bar, respectively. This model is run until 104 s and results are shown in Figure 17. We find that the NOx species are formed in our model thermally by the Zel’dovich mechanism (Zel’dovich & Raizer 1966): O2 + M ⇋ O + O + M, N2 + M ⇋ N + N + M, O + N2 → NO + N, N + O2 → NO + O, O2 + N2 → 2NO. (44) (45) (46) (47) (48) We compare our NO yield to the lightning dis- charge experiments performed by Navarro-Gonz´alez et al. (2001). We use for our NO mixing ratio the values found before the end of the shock (10−4 s in Fig. to (Navarro-Gonz´alez et al. 2001, their Eq. 4). We find that: 17), between 10−2 and 10−3, P (NO) ≈(cid:0)2.4 × 1022 K/J(cid:1) X(NO) Tf ≈ 2 − 20 × 1016 molecules J−1, (49) where Tf [K] is the “freeze-out” temperature af- ter which the NO mixing ratio does not change appreciably over the time-scale of the experiment, which we set to 1000 K (the approximate temper- ature of our model at t ≈ 10−4 s). This is consis- tent with the production of NO in the “hot core” region of the experiment. This is also roughly con- sistent with the literature values for NO produc- tion of 1017 molecules J−1 (Borucki et al. 1984; Price et al. 1997). This is an order-of-magnitude comparison be- tween the code and lightning experiments and models, and for a more complete comparison will need to be applied to a model atmosphere, where diffusion and photochemistry together will further process the NOx species. We plan to do this in a future paper. 4.4.3. The Early Earth The presence of life and the evolution of the sun both have radically altered Earth’s atmospheric chemistry. Oparin (1957) and Miller & Urey (1959) thought that the atmosphere of the early Earth9 was largely reducing, domi- nated by methane, ammonia and molecular hy- drogen. Kasting (1993) made a compelling case 9“early Earth” in this context means the Earth in its first 1 Gyr 17 that prebiotic formation of hydrogen would be too slow to allow for much molecular hydrogen in the atmosphere of the early Earth. Furthermore, a major constituent in the early Earth atmosphere needs to be a strong greenhouse gas, in order to compensate for the cooler young sun. The atmo- spheric chemistry of the early Earth is difficult to determine, and a severe lack of data results in many possible early Earth chemistries. As an illustrative example, Tian et al. (2005) argue that hydrogen escape was less efficient during the first 1 Gyr as was previously thought10. If Tian et al. (2005) are correct, then Earth’s early atmospheric composition could have been reducing. We present a model of the atmosphere of the early Earth, using the same lower boundary con- ditions as shown in Kasting (1993, his Fig. 1), and a temperature profile for the present Earth (Hedin 1987, 1991)11, shown in Figure 13. The lower boundary conditions used for the early Earth are given in Section 4.1. We treat outgassing using the deposition method (Appendix C). We compare our results to those of Kasting (1993, esp his Fig. 1). Our results are presented in Figure 18. The results compare reasonably well for CO and O2, but not for H2O and O. The CO abun- dance begins to increase at 30 km, 10 km higher than for Kasting (1993), and achieves a mixing ra- tio of ≈ 5×10−3 at 60 km, which is within a factor of 2 of Kasting (1993). The O2 likewise begins to rise above a mixing ratio of 10−6 10 km higher in the atmosphere, and also achieves a mixing ratio of ≈ 2.5× 10−3, again within a factor of 2 of Kasting (1993). The water vapor profile is quite different, however. Instead of falling below a mixing ratio of 10−6 at 10 km, the H2O mixing ratio in our model levels out at 5 × 10−4, increasing slightly at ∼ 50 km before plummeting. Also, the oxygen mixing ratio only reaches ≈ 3 × 10−6, approximately two orders of magnitude below the mixing ratio pre- dicted by Kasting (1993). These differences may be due to the different assumed young solar UV fields between ourselves and Kasting (1993), but we suspect that the differences are more likely due either to differences in the water condensation or the temperature profiles used. This seems espe- cially likely for atomic oxygen, which is primarily 10The debate is ongoing (Claire et al. 2006; Catling 2006). 11http://omniweb.sci.gsfc.nasa.gov/vitmo/ destroyed by the reaction: O + H2O → OH + OH, (50) in spite of the sizeable 7640 K barrier. When the water vapor drops off at ∼ 55 km, this destruction route becomes unviable, and the atomic oxygen mixing ratio rapidly increases. a 5. Glycine Formation in a Laboratory En- vironment The formation of glycine, among several other amino acids, amines and nucleotides, has been investigated for a variety of chemical composi- tions, from reducing (Miller 1953) to oxidizing (Schlesinger & Miller 1983; Miyakawa et al. 2002; Cleaves et al. 2008), and exploring various energy sources (see Miller & Urey 1959, and references therein). In a recent experiment, HCN and H2S were exposed to UV light (peak frequency 2540 A), resulting in the formation of numerous com- plex prebiotic compounds (Patel et al. 2015). The techniques used in this experiment afforded the ex- perimenters to track the pathways of formation for these various species. compounds Prebiotic species are produced in smaller con- centrations within a more oxidizing environ- ments (Miller & Urey 1959). Methane has been found to be important for the formation of pre- biotic (Schlesinger & Miller 1983; Miyakawa et al. 2002). The correlation between reducing chemistry and the efficient production of prebiotic molecules, combined with compelling evidence that the atmosphere of the early Earth was oxidizing (Kasting 1993), would suggest that other processes were responsible for producing the prebiotic chemical inventory on Earth. This pro- cess is hypothesized to have taken place within hy- drothermal vents (e.g. Ferris 1992), on the surfaces of crystals (Vijayan 1980), or possibly within the interstellar medium (e.g. Greenberg et al. 1995). Cleaves et al. (2008) have repeated Miller’s ex- periment in a reducing environment, and dis- covered that amino acids can be efficiently pro- duced in such environments, but that nitrites (e.g. HONO) destroy these species as quickly as they are produced. Adding ferrous iron, in the form of FeO or FeS2 (in the form of pyrite surfaces) effec- tively removes the nitrites and allows the amino acids to survive. 18 We explore the formation of glycine in the con- text of a weak radiating source. An unattenuated monochromatic beam of light at λ0 = 1000 A is applied with an intensity of ≈ 2 × 10−3 erg cm−2 s−1, corresponding to a flux of F0 = 108 photons cm−2 s−1. This flux is applied to Eq. (18) such that: kph,i(z) =Z 104A 1A = F0σi(λ0), σi(λ) F0δ(λ − λ0) dλ; (51) where δ is the Dirac delta function. The formation pathways for glycine have not been rigorously determined, although there are some proposed pathways. We include four possi- ble pathways to glycine formation in our network. First, we include glycine formation via the three body interaction of various species. These reac- tions have significant barriers, and so will only oc- cur efficiently at rather high temperatures. The reactions are: C2H4 + HNO2 → NH2CH2COOH, C2H5 + NO2 → NH2CH2COOH, CH3NO + H2CO → NH2CH2COOH, (52) (53) (54) with rate constants set equal to the three-body formation for analogous chemical species (e.g. CH2COOH). Also included is the ion-neutral pathway proposed for interstellar formation for glycine from Charnley (1997), CH6NO+ + HCOOH → C2H6NO+ 2 + H2O, (55) C2H6NO+ 2 + e− → NH2CH2COOH + H. (56) And finally, the formation of glycine by a possible pathway similar to that suggested by Patel et al. (2015), CH3NO + CH3O → NH2CH2COOH + H (57) is included. Additionally, we include FeO and reactions be- tween FeO and nitrites. We also inject our gas with HCOOH in order to facilitate the ion-neutral formation pathway; it is likely that there are other presently unknown paths of formation for formic acid. We run this network for a set of five differ- ent initial compositions given in Table 2, labeled 19 Model A - E. Model A is a strongly reducing envi- ronment, with only the gases NH3, CH4, H2 and H2O, FeO and HCOOH (Model A). We transition to a more reducing environment in the successive models (Models B, C, D). Finally, for Model E, we run the experiment starting solely from CO2, N2, H2O, FeO and HCOOH. We run all models using the unattenuated UV flux, at 1 bar pressure and 300 K temperature. The model is run to t ≈ 1 week. Our results are plotted in Figure 19. Moving from Model B to E, less and less glycine is formed, falling from a mixing ratio of 10−6 for Model B to 10−8 for Model E. This is what is expected from the Miller-Urey experiments per- formed for various chemical compositions: as the chemistry becomes less reducing, it becomes more difficult to form prebiotic molecules. More interesting is Model A. If all N2 and CO2 are removed, certain formation pathways to NO2, HNO2 and especially H2CO are inhibited. Addi- tionally, HCNO forms more slowly from HCN, and especially the ionic form, CHNO+ (in its various permutations) is difficult to form without some ex- cess unbonded atomic nitrogen or oxygen present in the gas. Model A produces virtually no glycine. We traced this back to the key reactions: 2 + e−, N2 + γ → N+ N+ 2 + e− → N + N, CO2 + γ → CO + O (58) the same formation pathway for amines in the early Earth as suggested by Zahnle (1986). In our case, however, the atomic nitrogen and oxygen are both important in completing the formation of HCNO and its isomers. 6. Conclusion In this paper, we have presented a gas-phase chemical network, Stand2015. The photochem- istry/diffusion code, Argo, was used to test the network. We have shown that the predictions from Stand2015 converge to chemical equilib- rium under the appropriate conditions and also that the molecular diffusion modeled by Argo makes a reasonable approximation to analytical calculations of molecular diffusion for an isother- mal gas in hydrostatic equilibrium. We have com- pared our model results (Stand2015+Argo) to perimental and theoretical determinations of the reaction rates are made available. More accurate determinations, especially of the reaction rates for the nitrogen chemistry, would be extremely helpful. This network and model provide a win- dow into a detailed analysis of prebiotic chem- istry, but much work must still be done in order to accurately predict the full budget of prebiotic molecules in the variety of environments in which they may occur. Both authors gratefully acknowledge the sup- port of the ERC Starting Grant #257431. P. B. R. is grateful to J. I. Moses for several helpful discus- sions about the network and model comparisons, to G. Laibe for his help with the Lagrangian nu- merical methods, and to C. R. Stark for his help understanding prebiotic formation in plasma en- vironments. Both authors are grateful for the anonymous referee whose report has helped sig- nificantly improve this paper. They also express thanks to Ian Taylor at St Andrews for his help with computational resources. Finally, they ac- knowledge the National Institute of Standards and Technology, the databases of which were essential to the completion of this project. chemical kinetics models for HD209458b, Jupiter and the Earth. For Jupiter, we found that ion- neutral chemistry may provide significant alterna- tive pathways to formation of various hydrocar- bons, especially ethylene (C2H4). Finally, we numerically simulate a Urey-Miller- like experiment12 under various initial chemistries. We found that, in an artificial environment, when derivatives of FeO and pyrite (FeS2) can destroy nitrites in the presence of a reservoir of formic acid, the formation of glycine is considerable also in re- ducing environments, approaching a mixing ratio of ∼ 10−6. For an environment more similar to the atmosphere of the early Earth, the mixing ra- tio drops to ∼ 10−8. Surprisingly, for a gas with- out any CO2, O2 or N2, virtually no glycine is formed. If this result is robust for various other en- ergy sources (shocks, thermal energy, etc.) and for other prebiotic species, this would suggest that the early Earth chemistry should not be too strongly reducing, else the formation of glycine and other prebiotic species would be severely inhibited. This network has limitations. It has only been tested for 1D atmosphere models, with non-self- consistent temperature profiles. Using this net- work within a global circulation model is presently unrealistic, but a reduced version of this network, constructed specifically for given atmospheres, could in principle be employed in 2D or 3D at- mosphere simulations. Sulfur chemistry has been shown to play an important role in the formation of prebiotics, and is an essential constituent in volcano plumes. The inclusion of sulfur chemistry will be a natural next step to take the model. Additionally, the models of prebiotic chemistry should consider the formation of species other than glycine. The formation of ribose (C5H10O5) of nucleotides, such as adenine (C5H5N5), and of phosphorus-bearing species should also be in- cluded to more fully encapsulate the formation of the prebiotic chemical reservoir. One serious problem with this network, and in- deed with any chemical kinetics network, is the uncertainty in rate coefficients. The effects of this uncertainty can be estimated using sensitivity analysis (e.g., within Venot et al. 2012), but can ultimately only be resolved slowly as better ex- 12The experiment we simulate is more like that of Patel et al. (2015). 20 A. List of Species, Reactions and Rates The purpose of this appendix is to explicitly lay out the content of the chemical network itself. We list the species considered in the network and the reactions. The species include the elements H/C/N/O, and the network includes a complete chemistry for molecules and ions of up to 2 carbon, 6 hydrogen, 2 nitrogen and 3 oxygen atoms. The different chemical kinetics for various neutral molecular isomers is included as completely as possible, although much about branching ratios for reactions is presently not well understood. A list of all the neutral species is given in Table 4. This table lists the species considered and includes the formula as used in the network, the standard formula, the name of the molecule and the source we used for the thermochemical data. In some cases, the chemical formula in the network is different from the standard chemical formula. This is because we incorporated our own method for distinguishing isomers, in order to make sure that we did not incorporate the same molecule under two different formulas. This list also includes some species with the elements Na, Mg, Si, Cl, K, Ti and Fe. The chemistry attempts to include only the dominant species with these elements, in which they would be present in the gas phase. These species are generally only present in the gas-phase for very high temperatures (generally > 1000 K). For cooler objects, these species are typically ignored. The noble gases He and Ar are included, both for the sake of completeness, and because they can play an important role in organic ion-neutral chemistry through charge-exchange reactions. Ions are also included, and a list of the ionic species is given in Table 5. In this case, the uncertainty in reaction rates and branching ratios is much more severe, and so we made no attempt at present to distinguish isomers of ionic species. It is difficult to determine which rate constants to use for a specific reaction, since there are often many to choose from, and they do not always agree well with each other. We employed the following method for determining which rate constant to include in our network, after plotting all the rate constants versus temperature over a range of 100 K to 30000 K: 1. If there exists only one published rate constant for a given reaction, we use that value. 2. Reject all rate constants that become unrealistically large at extreme temperature. 3. Choose rate constants that agree with each other over the range of validity. 4. If the most recent published rate constant disagrees with (3), and the authors give convincing arguments for why the previous rates were mistaken, we use the most recently published rate. The full list of forward reactions and rate constants determined by this method comprise the Stand2015 network and are given in Table 6. Reverse reactions are not explicitly shown; when reactions are reversible, bidirectional arrows are shown. When they are irreversible, or simply not reversed in the network, only unidirectional arrows are shown. Table 6 additionally includes a full list of the references for the rate constants used for each given reaction. B. Reversing Reactions For reverse reactions, we follow the prescription given by Burcat & Ruscic (2005). For the reaction: there is a rate constant, kf . We resolve to determine the reverse rate constant, kr, for the reaction: A + B → C + D + E, C + D + E → A + B. 21 (B1) (B2) Note that the number of species is different between the r.h.s. and l.h.s. of Eq. (B1). We denote this difference in number of reactants and products (nreact and nprod, respectively) by ∆ν, which in our case = nprod − nreact = 2 − 3 = −1. We then solve for the reaction rate constant as (Burcat & Ruscic 2005, their Eq. (6)), Kc =(RT )−∆ν exp(cid:16)∆a1(cid:0) log T − 1(cid:1) + + ∆a7(cid:17), 20 − ∆a5T 4 ∆a4T 3 ∆a6 12 T + + ∆a2T 2 + ∆a3T 2 6 (B3) where R = 8.314472 J mol−1 K−1 is the gas constant, and ∆ai = ai(C + D + E) − ai(A + B) for 1 ≤ i ≤ 7 are the NASA thermodynamics coefficients, which Burcat & Ruscic (2005) describes and tabulates. It is important to emphasize here, as done by Visscher & Moses (2011), the multiplicative factor (RT )−∆ν, which in our example would be 1.38065 × 10−22 T . The Burcat values for the NASA coefficients have been used for all possible species (see Tab. 4. For some species, however, the coefficients had to be obtained from other sources. For sources with elements Na, Mg, Si, Cl, K, Ti and Fe, the Burcat values were sparse, so we made use instead of the NASA-CEA values (McBride et al. 1993; Gordon & McBride 1999), which use 9-coefficient polynomials, so we fit them to a series of 7-coefficient polynomials for various temperature ranges. We do the same for the monatomic gases and ions at high temperatures 6000 K < T < 20000 K, using fits to the polynomials provided by Gordon & McBride (1999). For some species, the thermodynamic properties have not been determined. In these cases, for neutral species we use Benson’s additivity method as described by Cohen & Benson (1993). Benson’s additivity method can be naturally combined with the NASA and Burcat polynomial coefficients using the experimental values for the small alkanes listed within Cohen & Benson (1993). For the arbitrary alcohol from Cohen & Benson (1993), we use methanol, and for the arbitrary ether, dimethyl ether. The Benson coefficients are: Pi = 1 2 ai(C2H6), Si = ai(C3H8) − ai(C2H6), Di = ai(C2H6O) − ai(C2H6), Fi = ai(CH3OH) − ai(CH4). (B4) (B5) (B6) (B7) Here, ai(X) denotes the seven coefficients, i = 1, ..., 7 for species X. The coefficients for fundamental bonds are calculated using these coefficients as follows: S, 1 2 3 2 1 2 ai([C − H]) = ai([C − C]) = ai([C − O]) = ai([O − H]) = F − 1 P − 4 S − P, 3 D + S − 4 1 D − 2 1 2 1 2 P, S. (B8) (B9) (B10) (B11) The values for these bonds are given in Table 3. The values for [N-H], [N-C] and [N-O] are similarly determined. It has been suggested by Lias (1988) and Cohen & Benson (1993) that using Benson’s additivity method to determine the thermodynamic properties of ions, or at least strongly of strongly polarizing groups, can lead to large errors, because the thermodynamic properties of ions do not depend linearly on their length, although Hammerum & Sølling (1999) have had some success applying Benson’s method to ions. 22 We found, by investigating the thermodynamic properties of ionic species tabulated by Burcat & Ruscic (2005), that the thermodynamic properties of ions do depend nonlinearly but predictably based on size. We therefore placed all the known thermodynamic properties of ions into a database, and have extrapolated to calculate the thermodynamic properties for the undetermined ions. C. Outgassing, Condensation, Evaporation and Escape Boundary conditions play a key role in determining the atmospheric compositions of planets. For rocky planets, these boundary conditions are set by outgassing and escape into the exosphere. At temperatures . 1500 K, metals such as silicates, iron, corundum, begin to condense out of the exoplanet and brown dwarf atmospheric gas phase. At much lower temperatures, when, various other species (e.g. water, ammonia, methane) may also condense out. It is important for comparison to previous models to consider both the atmospheric boundary conditions and atmospheric condensation. As discussed in Section 3.1, there exist, in addition to the Stand2015 reactions, a series of “banking” reactions for all major species, that collect particles and reintroduce them to the fluid parcel at a rate determined by the diffusion time-scales. The very bottom banking reaction can be set to act effectively as an outgassing rate. Imagine a particular reservoir for a species, A. This reservoir is outgassing into the atmosphere with a flux, Φ(A) [cm−2 s−1]. This can be accounted by first taking a reservoir concentration of A, which for a large reservoir will be ≫ [A](z = 0), the bottom of the atmosphere. For a reservoir that will not be appreciably depleted over the chemical-dynamical time-scale of the atmosphere, the rate is simply: Pout(A) = Φ(A) ∆z . (C1) For a finite reservoir, we can place the reservoir concentration into the bottom “bank” for the species in question, and the t = 0 flux, Φ(A, 0), and concentration ([BA]) can be used to determine the rate of outgassing, where Pout(A) = L(BA) [BA], L(BA) = Φ(A, 0) [BA(t = 0)] ∆z . (C2) (C3) These approximations are not used anywhere in this paper. For HD209458b, we simply start with solar elemental abundances, with everything in atomic form at the bottom of the atmosphere. For both Jupiter and Earth, we start with fixed lower boundary conditions. Condensation or evaporation of species A can be treated by the reactions (“JA” represents “A in conden- sate form”): A → JA, JA → A, for condensation; for evaporation. (C4) (C5) This physical process is treated in two ways in this paper for Earth and Jupiter. The first method is by considering supersaturation concentrations, above which the species in question condenses out and below which the species in question will evaporate. This method is given by (Hamill et al. 1977; Toon & Farlow 1981; Hu et al. 2012), and has the form (for species A): P = tc = , [A] tc mAvth 4ρnuc L = [JA] tc ngas − nc(T, p) a 23 (C6) (C7) where mA [g] is the mass of the condensing species, vth is the thermal velocity of the gas, ρnuc [g cm−3] is the material density of the condensation seed, ngas [cm−3] is the density of the gas, and nc [cm−3] is the saturation number density, at the given temperature and pressure, and a [cm] is the average radius of the nucleation site. We consider condensation only for low temperatures, so nc = pv/kBT , where pv [dyn cm−2] is the vapor pressure, and is estimated using the relatively simple Antoine equation: log pv = A − B C + T , (C8) where A, B and C are all parameters taken from the tabulated NIST chemistry webbook13. Alternatively, one can use the method commonly used in the astrochemical context (Hasegawa et al. 1992; Caselli et al. 1998), where and L [s−1] = πa2vthnnuc P = ν0[JA] e−ED/kB T . (C9) (C10) Here, ED is the desorption energy, an empirically determined quantity, taken from Garrod et al. (2008). The frequency, ν0 [Hz] =(cid:16) 2nsED π2mA (cid:17)1/2 , (C11) relation ns = 1.5 × 1015 cm2 (cid:0)a/a0(cid:1)2 is the characteristic frequency of the surface. The number of sites is estimated, also empirically, by the , where a0 = 0.1 µm. The advantage of this approximation is that it is identical to the form generally used for complex surface chemistry in protoplanetary disks. This would allow one to take the results from disk chemistry and utilize them straightforwardly in atmospheric outgassing models. It is worth pointing out that exponentiating Eq. (C8), dividing by kBT , and then placing the resulting form of nc into Eq. (C7), yields a form: P L ∼ Const. vth eTc/T (C12) where Tc/T = B/(C + T ) from Eq. (C8). The two forms are therefore analogous parameterizations, with the same temperature dependence, but the saturation approach is dependent on the parameterized vapor pressure, and the deposition approach is parameterized by the number of nucleation sites and the binding energy of the nucleation particle. Neither the supersaturation method nor the deposition method explain where the condensation seeds first arise. It is assumed that the condensation seeds are already present, and therefore that condensation occurs whenever the supersaturation ratio, S & 1. In some environments like Earth, the condensation seeds come in the form of sand or ash particles, and the supersaturation ratio for water to condense is very small, S ≈ 1.01. If the seed particles are not already present in the atmosphere, they must form within the gas phase by the growth from small to large, complex clusters. This requires a supersaturation ratio S ≫ 100, which only occurs when T ≪ Tc (Helling & Fomins 2013). Zsom et al. (2012) explore the microphysics of water condensation and cloud formation for Earth and Earth-like planets None of these reactions appear in the generic kinetic network, because their inclusion is atmosphere dependent. Condensation is not considered at all for HD209458b because it is too hot, but is considered for Earth and Jupiter for water. Ammonia and methane condensation can also be considered for Jupiter and methane and other condensation should be considered for even colder planets, such as Uranus and Neptune. 13http://webbook.nist.gov/chemistry/ 24 REFERENCES Anglada, J. M. 2004, JAChS, 126, 9809 Ackerman, M. 1971, in Mesospheric Models and Related Experiments (Springer), 149 Anicich, V. G., Huntress, W. T., & Futrell, J. H. 1976, CPL, 40, 233 Adachi, H., Basco, N., & James, D. 1981, Int J Anicich, V. G., Huntress, W. T., & McEwan, M. J. Chem Kin, 13, 1251 1986, JPhCh, 90, 2446 Adam, L., Hack, W., Zhu, H., Qu, Z., & Schinke, R. 2005, JChPh, 122, 114301 Adams, N., Bohme, D., & Ferguson, E. 1970, JChPh, 52, 5101 Adams, N., & Smith, D. 1976a, JPhB, 9, 1439 —. 1976b, IJMIP, 21, 349 —. 1977, CPL, 47, 383 —. 1978, CPL, 54, 530 Adams, N., Smith, D., & Grief, D. 1978a, IJMIP, 26, 405 —. 1978b, IJMIP, 26, 405 Adams, N. G., Smith, D., & Paulson, J. F. 1980, JChPh, 72, 288 Aplin, K. L. 2013, Electrifying Atmospheres: Charging, Ionisation and Lightning in the Solar System and Beyond (Springer Science & Busi- ness Media) Arenas, J. F., Marcos, J. I., L´opez-Toc´on, I., Otero, J. C., & Soto, J. 2000, JChPh, 113, 2282 Armentrout, P., Berman, D., & Beauchamp, J. 1978, CPL, 53, 255 Asplund, M., Grevesse, N., Sauval, A. J., & Scott, P. 2009, ARA&A, 47, 481 Atkinson, R., Baulch, D., Cox, R., et al. 1989, JPCRD, 18, 881 —. 1992, AtmEn A, 26, 1187 —. 1997, JPCRD, 26, 521 Adamson, J., DeSain, J., Curl, R., & Glass, G. 1997, JPCA, 101, 864 —. 2001, Summary of Evaluated Kinetic & Pho- tochemical Data for Atmospheric Chemistry Aders, W.-K., & Wagner, H. G. 1973, Berich Bun- Atkinson, R., Baulch, D., Cox, R., et al. 2004, sen Gesell, 77, 712 ACP, 4, 1461 Ag´undez, M., Parmentier, V., Venot, O., Hersant, F., & Selsis, F. 2014, A&A, 564, A73 Atkinson, R., Finlayson, B., & Pitts, J. 1973, JAChS, 95, 7592 Alam, J. M., & Lin, J. C. 2008, MWRv, 136, 4653 Albritton, D., Dotan, I., Streit, G., et al. 1983, JChPh, 78, 6614 Allison, T. C., Lynch, G. C., Truhlar, D. G., & Gordon, M. S. 1996, JPhCh, 100, 13575 Almatarneh, M., Flinn, C., & Poirier, R. 2005, CaJCh, 83, 2082 Alvarez, R. A., & Moore, C. B. 1994, JPhCh, 98, 174 Atreya, S. K., Mahaffy, P. R., Niemann, H. B., Wong, M. H., & Owen, T. C. 2003, Planet. Space Sci., 51, 105 Ausloos, P., 1975, Interactions between Ions and Molecules, ed. P. Ausloos (Plenum Press), 489 Ausloos, P., & Lias, S. 1981, JAChS, 103, 3641 Avramenko, L., & Krasnen’kov, V. 1966, Russ Chem B, 15, 394 Back, R., & Griffiths, D. 1967, JChPh, 46, 4839 Anastasi, C., & Hancock, D. U. 1988, FaTr II, 84, Badnell, N. R., Bautista, M., Butler, K., et al. 1697 2005, MNRAS, 360, 458 Andersson, S., Markovic, N., & Nyman, G. 2003, Bailey, R. L., Helling, Ch., Hodos´an, G., Bilger, JPCA, 107, 5439 C., & Stark, C. R. 2014, ApJ, 784, 43 25 Baker, R., Baldwin, R., & Walker, R. 1971 in In- ternational Symposium on Combustion (Else- vier), 291 Baker, R., Kerr, J., & Trotman-Dickenson, A. 1969, J Chem Soc A, 390 Baldwin, R. R., Booth, D., & Brattan, D. 1961, CaJCh, 39, 2130 Baldwin, R. R., Keen, A., & Walker, R. W. 1984, FaTr I, 80, 435 Baldwin, R. R., Langford, D., Matchan, M., Walker, R., & Yorke, D. 1971 in International Symposium on Combustion (Elsevier), 251 —. 1994, JPCRD, 23, 847 Baulch, D., Cox, R., Crutzen, P., et al. 1982, JPCRD, 11, 327 Baulch, D., Duxbury, J., Grant, S., & Mon- tague, D. 1981, Evaluated kinetic data for high temperature reactions. Volume 4. Homogeneous gas phase reactions of halogen-and cyanide- containing species, Tech. rep., DTIC Document Beaty, E., & Patterson, P. 1965, PhRv, 137, A346 Beaulieu, J. P., Kipping, D. M., Batista, V., et al. 2010, MNRAS, 409, 963 Becker, E., Rahman, M., & Schindler, R. 1992a, Banks, P. M., & Kockarts, G. 1973, Aeronomy Berich Bunsen Gesell, 96, 776 (Elsevier) Bar-Nun, A., Bar-Nun, N., Bauer, S. H., & Sagan, C. 1970, Science, 168.3930, 470 Becker, K., Konig, R., Meuser, R., Wiesen, P., & Bayes, K. D. 1992b, J Photoch Photobio A, 64, 1 Barassin, J., Barassin, A., & Thomas, R. 1983, Benneke, B. 2015, arXiv:1504.07655 IJMIP, 49, 51 Benner, S. A., Ricardo, A., & Carrigan, M. A. Barfield, W., Koontz, G., & Huebner, W. 1972, 2004, Curr Opin Chem Biol, 8, 672 JQSRT, 12, 1409 Barnett, A. J., Marston, G., & Wayne, R. P. 1987, FaTr II, 83, 1453 Barsuhn, J., & Nesbet, R. 1978, JChPh, 68, 2783 Bass, A. M., Ledford, A. E., & Laufer, A. H. 1976, J Res NBS A Phys Chem, 80A, 143 Batt, L., McCulloch, R., & Milne, R. 1975, Int J Chem Kinet, 7, 1 Benson, S. W. 1994, Int J Chem Kin, 26, 997 Bergeat, A., Calvo, T., Daugey, N., Loison, J.-C., & Dorthe, G. 1998, JCPA, 102, 8124 Bergeat, A., Moisan, S., M´ereau, R., & Loison, J.-C. 2009, CPL, 480, 21 Bethe, H. A., & Salpeter, E. E. 1957, Quan- tum mechanics of one-and two-electron atoms, Vol. 63 (Springer Berlin) Batt, L., Milne, R., & McCulloch, R. 1977, Int J Chem Kin, 9, 567 Betowski, D., Payzant, J., Mackay, G. I., & Bohme, D. 1975, CPL, 31, 321 Batt, L., & Rattray, G. 1979, Int J Chem Kin, 11, 1183 B´ezard, B., Drossart, P., Encrenaz, T., & Feucht- gruber, H. 2001, Icarus, 154, 492 Bauer, W., Becker, K., & Meuser, R. 1985, Berich Bunsen Gesell, 89, 340 Bierbaum, V. M., DePuy, C., Shapiro, R., & Stew- art, J. H. 1976, JAChS, 98, 4229 Bauerle, S., Klatt, M., & Wagner, H. 1995, Berich Bunsen Gesell, 99, 870 Bierbaum, V. M., Grabowski, J. J., & DePuy, C. H. 1984, JPhCh, 88, 1389 Baulch, D., Bowman, C. T., Cobos, C. J., et al. 2005, JPCRD, 34, 757 Baulch, D., Cobos, C., Cox, R., et al. 1992, JPCRD, 21, 411 Biggs, P., Canosa-Mass, C. E., Fracheboud, J.-M., Shallcross, D. E., & Wayne, R. P. 1995, FaTr, 91, 817 26 Bilger, C., Rimmer, P., & Helling, Ch. 2013, MN- Borucki, W. J., McKay, C. P., & Whitten, R. C. RAS, 435, 1888 1984, Icarus, 60, 260 Blagojevic, V., Petrie, S., & Bohme, D. K. 2003, Bose, D., & Candler, G. V. 1996, JChPh, 104, MNRAS, 339, L7 2825 Blake, P., & Jackson, G. 1969, J Chem Soc B, 94 Boughton, J., Kristyan, S., & Lin, M. 1997, CP, Blitz, M., Johnson, D., Pilling, M., et al. 1997, JCS(FaTr), 93, 1473 Bogan, D. J., & Hand, C. W. 1978, JPhCh, 82, 2067 Bogdanchikov, G., Baklanov, A., & Parker, D. 2004, CPL, 385, 486 Bohland, T., Dob, S., Temps, F., & Wagner, H. G. 1985, Berichte Bunsen Gesell, 89, 1110 Bohme, D., Adams, N., Mosesman, M., Dunkin, D., & Ferguson, E. 1970, JChPh, 52, 5094 Bohme, D. K., & Fehsenfeld, F. 1969, CaJCh, 47, 2717 Bohme, D. K., Lee-Ruff, E., & Young, L. B. 1971, JAChS, 93, 4608 214, 219 Bowers, M., Elleman, D., & King, J. 1969, JChPh, 50, 4787 Bozzelli, J. W., Chang, A. Y., & Dean, A. M. 1994 in International Symposium on Combus- tion (Elsevier), 965 Bozzelli, J. W., & Dean, A. M. 1989, JPhCh, 93, 1058 —. 1990, JPhCh, 94, 3313 —. 1995, Int J Chem Kin, 27, 1097 Bravo-P´erez, G., Alvarez-Idaboy, J. R., Cruz- Torres, A., & Ru´ız, M. E. 2002, JCPA, 106, 4645 Breen, J., & Glass, G. 1971, Int J Chem Kin, 3, Bohme, D. K., & Mackay, G. I. 1981, JAChS, 103, 145 2173 Bohme, D. K., Mackay, G.-I., & Schiff, H. 1980, JChPh, 73, 4976 Brion, C., Tan, K., Van der Wiel, M., & Van der Leeuw, P. E. 1979, JESRP, 17, 101 Broad, J. T., & Reinhardt, W. P. 1976, PhRvA, Bohme, D. K., Mackay, G., Schiff, H., & 14, 2159 Hemsworth, R. 1974, JChPh, 61, 2175 Brown, P. N., & Hindmarsh, A. C. 1989, ApMaC, Bohme, D. K., Mackay, G., & Tanner, S. 1979, 31, 40 JAChS, 101, 3724 Bohme, D. K., & Raksit, A. B. 1985, CaJCh, 63, 3007 Bohme, D. K., Raksit, A., & Schiff, H. 1982, CPL, 93, 592 Browning, R., & Fryar, J. 1973, JPhB, 6, 364 Brownsword, R., Gatenby, S., Herbert, L., et al. 1996, FaTr, 92, 723 Brunetti, B., & Liuti, G. 1975, ZPC, 94, 19 Bolden, R., Hemsworth, R., Shaw, M., & Twiddy, N. 1970, JPhB, 3, 45 Bryukov, M. G., Dellinger, B., & Knyazev, V. D. 2006, JPCA, 110, 936 Bolden, R., & Twiddy, N. 1972, FaDi, 53, 192 Borisov, A., Zamanskii, V., Potmishil, K., Skachkov, G., & Foteenkov, V. 1977, Kinet. Catal. (USSR), 18 Borucki, W., Kenzie, R. M., McKay, C., Duong, N., & Boac, D. 1985, Icarus, 64, 221 Bryukov, M. G., Knyazev, V. D., Lomnicki, S. M., McFerrin, C. A., & Dellinger, B. 2004, JCPA, 108, 10464 Bulatov, V., Buloyan, A., Cheskis, S., et al. 1980, CPL, 74, 288 27 Burcat, A., & Ruscic, B. 2005, Third millenium ideal gas and condensed phase thermochemical database for combustion with updates from ac- tive thermochemical tables Burt, J., Dunn, J., McEwan, M., et al. 1970, JChPh, 52, 6062 Buttrill, S., Kim, J., & Huntress, W. 1974, J. Chem. Phys, 61, 2122 Cairns, R. B., & Samson, J. A. R. 1965, J. Geo- phys. Res., 70, 99 Calvert, J., & Pitts, J. 1966, Photochemistry (Wi- ley: New York) Chan, W.-T., Heck, S. M., & Pritchard, H. O. 2001, PCCP, 3, 56 Chang, J.-G., Chen, H.-T., Xu, S., & Lin, M. 2007, JPCA, 111, 6789 Chang, N.-Y., & Yu, C.-H. 1995, CPL, 242, 232 Chapman, S., & Cowling, T. G. 1991, The Math- ematical Theory of Non-uniform Gases (Cam- bridge University Press) Charnley, S. B. 1997, in IAU Colloq. 161: As- tronomical and Biochemical Origins and the Search for Life in the Universe, ed. C. Batalli Cosmovici, S. Bowyer, & D. Werthimer, 89 Campomanes, P., Men´endez, I., & Sordo, T. L. Chau, M., & Bowers, M. T. 1976, CPL, 44, 490 2001, JPCA, 105, 229 Canosa, C., Penzhorn, R.-D., & Von Sonntag, C. 1979, Berich Bunsen Gesell, 83, 217 Canosa, C., Smith, S. J., Toby, S., & Wayne, R. P. 1988, FaTr II, 84, 263 Caridade, P., Rodrigues, S., Sousa, F., & Varan- das, A. 2005, JPCA, 109, 2356 Carl, S., Sun, Q., Teugels, L., & Peeters, J. 2003, PCCP, 5, 5424 Carstensen, H.-H., & Dean, A. M. 2005, P Com- bust Inst, 30, 995 —. 2008, JCPA, 113, 367 Caselli, P., Hasegawa, T. I., & Herbst, E. 1998, ApJ, 495, 309 Catling, D. C. 2006, Science, 311, 38 Cavali´e, T., Biver, N., Hartogh, P., et al. 2012, Planet. Space Sci., 61, 3 Cermak, V., Dalgarno, A., Ferguson, E., Fried- man, L., & McDaniel, E. 1970, Ion Molecule Reactions (New York: Wiley) Ceursters, B., Nguyen, H. M. T., Nguyen, M. T., Peeters, J., & Vereecken, L. 2001 PCCP, 3, 3070 Chakraborty, D., & Lin, M. 1999, JCPA, 103, 601 Chakraborty, D., Park, J., & Lin, M. 1998, CP, 231, 39 Cheng, T., & Lampe, F. 1973, JPhCh, 77, 2841 Cheng, T., Yu, T.-Y., & Lampe, F. 1973, JPhCh, 77, 2587 —. 1974, JPhCh, 78, 1184 Choi, Y., & Lin, M. 2005, Int J Chem Kin, 37, 261 Christie, M. I., & Voisey, M. 1967, TrFa, 63, 2702 Chuchani, G., Martin, I., Rotinov, A., & Dominguez, R. M. 1993, J Phys Org Chem, 6, 54 Cimas, A., & Largo, A. 2006, JCPA, 110, 10912 Claire, M. W., Catling, D. C., & Zahnle, K. J. 2006, Geobiology, 4, 239 Clark, J. H., Moore, C. B., & Nogar, N. S. 1978, JChPh, 68, 1264 Clary, D., Smith, D., & Adams, N. 1985, CPL, 119, 320 Cleaves, H. J., Chalmers, J. H., Lazcano, A., Miller, S. L., & Bada, J. L. 2008, OLEB, 38, 105 Cobos, C., & Troe, J. 1985, JChPh, 83, 1010 Cohen, N. 1991, Int J Chem Kin, 23, 397 Cohen, N., & Benson, S. 1993, ChRv, 93, 2419 Cohen, N., & Westberg, K. 1991, JPCRD, 20, 1211 28 Colberg, M., & Friedrichs, G. 2006, JCPA, 110, Demott, P. J., Cziczo, D. J., Prenni, A. J., et al. 160 2003, PNAS, 100, 14655 Cook, G., & Metzger, P. 1964, JOSA, 54, 968 Cook, R. D., Davidson, D. F., & Hanson, R. K. 2009, JPCA, 113, 9974 Corchado, J. C., & Espinosa-Garcıa, J. 1997, JChPh, 106, 4013 Corchado, J. C., Espinosa-Garcıa, J., Hu, W.-P., Rossi, I., & Truhlar, D. G. 1995, JPhCh, 99, 687 Corchado, J. C., Espinosa-Garc´ıa, J., Roberto- Neto, O., Chuang, Y.-Y., & Truhlar, D. G. 1998, JCPA, 102, 4899 Deppe, J., Friedrichs, G., Ibrahim, A., Romming, H.-J., & Wagner, H. G. 1998, Berich Bunsen Gesell, 102, 1474 DeSain, J. D., Klippenstein, S. J., Miller, J. A., & Taatjes, C. A. 2003, JCPA, 107, 4415 D´esert, J.-M., Vidal-Madjar, A., Lecavelier Des Etangs, A., et al. 2008, A&A, 492, 585 Dheandhanoo, S., Johnsen, R., & Biondi, M. A. 1984, P&SS, 32, 1301 Dibeler, V. H., Walker, J. A., & Rosenstock, H. M. 1966, J Res Nat Bur Stand A, 70, 459 Cox, R., & Derwent, R. 1977, J Photochem, 6, 23 Ditchburn, R. 1955, RSPSA, 229, 44 Cribb, P. H., Dove, J. E., & Yamazaki, S. 1992, Dixon, R., & Kirby, G. 1968, TrFa, 64, 2002 CoFl, 88, 169 Crosley, D. R. 1989, JPhCh, 93, 6273 Curran, H. 2006, Int J Chem Kin, 38, 250 Cvetanovi´c, R. J. 1987, JPCRD, 16, 261 Dombrowsky, C., Hoffmann, A., Klatt, M., et al. 1991, Berich Bunsen Gesell, 95, 1685 Dombrowsky, C., & Wagner, H. G. 1992, Berich Bunsen Gesell, 96, 1048 Dong, H., Ding, Y.-h., & Sun, C.-c. 2005, JChPh, Daele, V., Laverdet, G., Le Bras, G., & Poulet, G. 122, 204321 1995, JPhCh, 99, 1470 Donovan, T., Dorko, W., & Harrison, A. 1971, Dammeier, J., Colberg, M., & Friedrichs, G. 2007, CaJCh, 49, 828 PCCP, 9, 4177 Davidson, D. F., Kohse-Hoinghaus, K., Chang, A. Y., & Hanson, R. K. 1990, Int J Chem Kin, 22, 513 De Cobos, A. E. C., & Troe, J. 1984, Int J Chem Kinet, 16, 1519 Dean, A. J. 1985, JPhCh, 89, 4600 Dean, A. J., Davidson, D., & Hanson, R. 1991, JPhCh, 95, 183 Dean, A. J., & Hanson, R. K. 1992, Int J Chem Kin, 24, 517 Dean, A. J., & Kistiakowsky, G. 1971, JChPh, 54, 1718 Debrou, G. B., Fulford, J. E., Lewars, E. G., & March, R. E. 1978, IJMIP, 26, 345 Dorko, E. A., Pchelkin, N. R., Wert, J. C., & Mueller, G. W. 1979, JPhCh, 83, 297 Dotan, I., & Lindinger, W. 1982, JChPh, 76, 4972 Dotan, I., Lindinger, W., Rowe, B., et al. 1980, CPL, 72, 67 Drossart, P., Fouchet, T., Crovisier, J., et al. 1999, “The Universe as Seen by ISO”, Vol. 427, 169 Duan, X., & Page, M. 1995, JAChS, 117, 5114 Dubrovin, D., Luque, A., Gordillo-Vazquez, F. J., et al. 2014, Icarus, 241, 313 Duff, J., & Sharma, R. 1996, GeoRL, 23, 2777 Dunbar, R. C., Shen, J., & Olah, G. A. 1972, JChPh, 56, 3794 Dunkin, D., Fehsenfeld, F., & Ferguson, E. 1970, JChPh, 53, 987 29 —. 1972, CPL, 15, 257 Fegley, B., & Lodders, K. 1994, Icarus, 110, 117 Dunkin, D., McFarland, M., Fehsenfeld, F., & Fer- Fehsenfeld, F. 1969, CaJCh, 47, 1808 guson, E. 1971, JGR, 76, 3820 Duran, R., Amorebieta, V., & Colussi, A. 1988, JPhCh, 92, 636 Durup-Ferguson, M., Bohringer, H., Fahey, D. W., & Ferguson, E. E. 1984, JChPh, 79, 265 Dyudina, U. A., Ingersoll, A. P., Ewald, S. P., et al. 2007, Icarus, 190, 545 Edelbuttel-Einhaus, J., Hoyermann, K., Rohde, G., & Seeba, J. 1992 in International Sympo- sium on Combustion (Elsevier), 661 Edwards, D., Kerr, J., Lloyd, A., & Trotman- Dickenson, A. 1966, J Chem Soc A, 1500 England, C., & Corcoran, W. H. 1975, Ind Eng Chem, 14, 55 Enskog, D. 1917, Kinetische Theorie der Vor- gaenge in maessig verduennten Gasen. I. All- gemeiner Teil, Dissertation Ercolano, B., & Storey, P. J. 2006, MNRAS, 372, 1875 Eremin, A., Ziborov, V., Shumova, V., Voiki, D., & Roth, P. 1997, Kin Catal, 38, 1 Espinosa-Garcia, J., Corchado, J., & Sana, M. 1993, J Chim Phys, 90, 1181 Fahey, D., Bohringer, H., Fehsenfeld, F., & Fergu- son, E. 1982, JChPh, 76, 1799 Fahey, D., Fehsenfeld, F., Ferguson, E., & Viehland, L. 1981, JChPh, 75, 669 Fahr, A., & Nayak, A. 1994, CP, 189, 725 Fahr, A., & Nayak, A. 1996, CP, 203, 351 Fahr, A., & Stein, S. 1989 in International Sym- posium on Combustion (Elsevier), 1023 Faigle, J. F. G., Isolani, P. C., & Riveros, J. M. 1976, JAChS, 98, 2049 Fairbairn, A. 1969, RSPSA, 312, 207 Faravelli, T., Goldaniga, A., Zappella, L., et al. 2000, P Combust Inst, 28, 2601 —. 1975, JChPh, 63, 1686 —. 1976, ApJ, 209, 638 —. 1977, P&SS, 25, 195 Fehsenfeld, F., Albritton, D., Burt, J., & Schiff, H. 1969, CaJCh, 47, 1793 Fehsenfeld, F., Dotan, I., Albritton, D., Howard, C., & Ferguson, E. 1978, JGR:Oceans, 83, 1333 Fehsenfeld, F., Dunkin, D., & Ferguson, E. 1970, P&SS, 18, 1267 —. 1974, ApJ, 188, 43 Fehsenfeld, F., & Ferguson, E. 1970, JChPh, 53, 2614 —. 1971, JGR, 76, 8453 —. 1972, JChPh, 56, 3066 —. 1974, JChPh, 61, 3181 Fehsenfeld, F., Ferguson, E., & Bohme, D. 1969b, P&SS, 17, 1759 Fehsenfeld, F., Ferguson, E., & Mosesman, M. 1969c, CPL, 4, 73 Fehsenfeld, F., Ferguson, E., & Schmeltekopf, A. 1966a, JChPh, 45, 1844 Fehsenfeld, F., Howard, C. J., & Ferguson, E. 1973, JChPh, 58, 5841 Fehsenfeld, F., Howard, C. J., & Schmeltekopf, A. 1975a, JChPh, 63, 2835 Fehsenfeld, F., Lindinger, W., & Albritton, D. 1975b, JChPh, 63, 443 Fehsenfeld, F., Lindinger, W., Schiff, H., Hemsworth, R., & Bohme, D. 1976, JChPh, 64, 4887 Fehsenfeld, F., Schmeltekopf, A., & Ferguson, E. 1967a, The JChPh, 46, 2802 Fehsenfeld, F., Schmeltekopf, A., Goldan, P., Schiff, H., & Ferguson, E. 1966b, JChPh, 44, 4087 30 Fehsenfeld, F., Schmeltekopf, A., Schiff, H., & Fer- guson, E. 1967b, P&SS, 15, 373 Feng, W., & Hershberger, J. F. 2007, JCPA, 111, 3831 Frank, P., Bhaskaran, K., & Just, T. 1988 in the International Symposium on Combustion (El- sevier), 885 Freeman, C., Harland, P., & McEwan, M. 1978a, Ferguson, E. 1968, Adv Electron Electron Phys, 24, 1 IJMIP, 28, 19 —. 1978b, AJCh, 31, 2157 Ferguson, E., & Fehsenfeld, F. 1968, JGR, 73, 6215 Friedrichs, G., Colberg, M., Dammeier, J., Bentz, T., & Olzmann, M. 2008, PCCP, 10, 6520 Ferguson, E., Fehsenfeld, F., & Schmeltekopf, Friedrichs, G., DAvidson, D. F., & Hanson, R. K. A. L. 1969, Adv Chem Ser, 80, 83 2002, Int J Chem Kinet, 37, 374 Fernandes, R. X., Giri, B. R., Hippler, H., Kachi- ani, C., & Striebel, F. 2005, JCPA, 109, 1063 Fern´andez-Ramos, A., Mart´ınez-N´unez, E., R´ıos, M. A., et al. 1998, JAChS, 120, 7594 Fulle, D., & Hippler, H. 1997, JChPh, 106, 8691 Gannon, K. L., Glowacki, D. R., Blitz, M. A., et al. 2007, JPCA, 111, 6679 Gao, Y., & Macdonald, R. G. 2006, JCPA, 110, Ferradaz, T., Benilan, Y., Fraya, A., et al. 2009, 977 P&SS, 57, 10 Ferris, J. P. 1992, OLEB, 22, 109 Feuchtgruber, H., Lellouch, E., de Graauw, T., et al. 1997, Nature, 389, 159 Field, F., Franklin, J., & Lampe, F. 1957, JAChS, 79, 2419 Fifer, R. 1975, Ber Bunsenges Phys Chem, 10 Fluegge, R. A. 1969a in the Bulletin of the Amer- ican Physical Society, (New York: APS), 14(2), 261 Fluegge, R. A. 1969b, JChPh, 50, 4373 Fontijn, A., Fernandez, A., Ristanovic, A., Ran- dall, M. Y., & Jankowiak, J. T. 2001, JCPA, 105, 3182 Forst, W., Evans, H., & Winkler, C. 1957, JPhCh, 61, 320 Fouchet, T., Lellouch, E., B´ezard, B., et al. 2000, A&A, 355, L13 Frank, P. 1986, in 15th International Symposium on Rarefied Gas Dynamics (Tuebner) Garrod, R. T., Weaver, S. L. W., & Herbst, E. 2008, ApJ, 682, 283 Gear, C. W. 1971, Comm ACM, 14, 185 Gehring, M., Hoyermann, K., Wagner, H. G., & Wolfrum, J. 1969, Berich Bunsen Gesell, 73, 956 Geiger, H., Wiesen, P., & Becker, K. H. 1999, PCCP, 1, 5601 Geltman, S. 1962, The Astrophysical Journal, 136, 935 Gentieu, E., & Mentall, J. 1970, Science, 169, 681 Gill, R., Johnson, W., & Atkinson, G. 1981, CP, 58, 29 Gladstone, G. R., Allen, M., & Yung, Y. L. 1996, Icarus, 119, 1 Glanzer, K., & Troe, J. 1973, AcHCh, 56, 577 —. 1975, Berich Bunsen Gesell, 79, 465 Glarborg, P., Dam-Johansen, K., & Miller, J. A. 1995, Int J Chem Kinet, 27, 1207 Glosik, J., Raksit, A., Twiddy, N., Adams, N., & Frank, P., Bhaskaran, K., & Just, T. 1986, JPhCh, Smith, D. 1978, JPhB, 11, 3365 90, 2226 Gonzalez, C., Theisen, J., Schlegel, H. B., Hase, W. L., & Kaiser, E. 1992, JPhCh, 96, 1767 31 Gorden, R., & Ausloos, P. 1961, JPhCh, 65, 1033 Harada, N., Herbst, E., & Wakelam, V. 2010, ApJ, —. 1967, JChPh, 46, 4823 Gordon, S., & McBride, B. J. 1999, Thermody- namic Data to 20000 K for Monatomic Gases (Cleveland: NASA) Grabowski, J. J. 1983, Doctoral Thesis (University of Colorado) Graham, R. A., & Johnston, H. S. 1978a, JPhCh, 82, 254 —. 1978b, JPhCh, 82, 254 Graham, E., James, D., Keever, W., et al. 1973, JChPh, 59, 4648 Gray, P., & Herod, A. 1968, Transactions of the Faraday Society, 64, 2723 Greenberg, J. M., Kouchi, A., Niessen, W., et al. 1995, J Biol Phys, 20, 61 Griggs, M. 1968, JChPh, 49, 857 Grotheer, H., & Just, T. 1981, CPL, 78, 71 Grussdorf, J., Nolte, J., Temps, F., & Wagner, H. G. 1994, Berich Bunsen Gesell, 98, 546 Gupta, S., Jones, E., Harrison, A. G., & Myher, J. J. 1967, CaJCh, 45, 3107 Gurnett, D. A., Kurth, W. S., Cairns, I. H., & Granroth, L. J. 1990, NASA STI/Recon Tech- nical Report N, 91, 11642 Guyon, P. M., Chupka, W. A., & Berkowitz, J. 1976, JChPh, 64, 1419 Hack, W., Hold, M., Hoyermann, K., Wehmeyer, J., & Zeuch, T. 2005, PCCP, 7, 1977 Hahn, Y. 1997, PhLA, 231, 82 721, 1570 Harding, L. B., Guadagnini, R., & Schatz, G. C. 1993, JPhCh, 97, 5472 Harding, L. B., Klippenstein, S. Georgievskii, Y. 2005, P Combust 30, 985 J., & Inst, Harding, L. B., Klippenstein, S. J., & Miller, J. A. 2008, JPCA, 112, 522 Harding, L. B., & Wagner, A. F. 1989 in Inter- national Symposium on Combustion (Elsevier), 983 Hartmann, D., Karthauser, J., Sawerysyn, J., & Zellner, R. 1990, Berich Bunsen Gesell, 94, 639 Hasegawa, T. I., Herbst, E., & Leung, C. M. 1992, ApJS, 82, 167 Hassinen, E., Kalliorinne, K., & Koskikallio, J. 1990, Int J Chem Kin, 22, 741 Hassinen, E., Riepponen, P., Blomqvist, K., et al. 1985, Int J Chem Kin, 17, 1125 Hastie, D., Freeman, C., McEwan, M., & Schiff, H. 1976, Int J Chem Kin, 8, 307 Haworth, N. L., Mackie, J. C., & Bacskay, G. B. 2003, JPCA, 107, 6792 He, Y., Liu, X., Lin, M., & Melius, C. 1993, Int J Chem Kin, 25, 845 He, Y., Sanders, W., & Lin, M. 1988, JPhCh, 92, 5474 H´ebrard, E., Dobrijevic, M., Loison, J. C., et al. 2013, A&A, 552, A132 Hedin, A. E. 1987, JGR: Space Physics, 92, 4649 Haldane, J. B. S. 1928, Ration Annu, 148, 3 —. 1991, JGR: Space Physics, 96, 1159 Hamill, P., Toon, O. B., & Kiang, C. S. 1977, JAtS, 34, 1104 Hammerum, S., & Sølling, T. I. 1999, JAChS, 121, 6002 Hanson, R. K., & Salimian, S. 1984, in Symposium on Combustion Chemistry (Springer), 361 Heicklen, J., & Johnston, H. S. 1962, JAChS, 84, 4394 Heimerl, J., Johnsen, R., & Biondi, M. A. 1969, JChPh, 51, 5041 Helling, Ch., & Fomins, A. 2013, RSPTA, 371, 20110581 32 Helling, Ch., Jardine, M., Diver, D., & Witte, S. Horne, D., & Norrish, R. 1970, RSPSA, 315, 301 2013, Planet. Space Sci., 77, 152 Helling, Ch., Woitke, P., Rimmer, P. B., et al. 2014, Life, 4, 142 Hemsworth, R., Payzant, J., Schiff, H., & Bohme, D. 1974, CPL, 26, 417 Hemsworth, R., Rundle, H., Bohme, D., et al. 1973, JChPh, 59, 61 Henis, J., Stewart, G., & Gaspar, P. 1973, JChPh, 58, 3639 Hennig, G., & Wagner, H. 1994, Berich Bunsen Gesell, 98, 749 Henry, G. W., Marcy, G. W., Butler, R. P., & Vogt, S. S. 2000, ApJ, 529, L41 Henry, R. J. 1970, ApJ, 161, 1153 Henry, R. J., & McElroy, M. B. 1968, Adv Space Res, 1, 251 Herbst, E., Payzant, J., Schiff, H., & Bohme, D. 1975, ApJ, 201, 603 Herman, J., & Mentall, J. 1982, JGR: Oceans, 87, 8967 Herron, J. T. 1988, JPCRD, 17, 967 —. 1999, JPCRD, 28, 1453 Hidaka, Y., Nakamura, T., Tanaka, H., Inami, K., & Kawano, H. 1990, Int J Chem Kin, 22, 701 Hidaka, Y., Oki, T., Kawano, H., & Higashihara, T. 1989, JPhCh, 93, 7134 Hidaka, Y., Sato, K., & Yamane, M. 2000, CoFl, 123, 1 Hinshelwood, C., & Askey, P. 1927, RSPSA, 215 Howard, C. J. 1979, JChPh, 71, 2352 Howard, C. J., Fehsenfeld, F., & McFarland, M. 1974, JChPh, 60, 5086 Howorka, F., Lindinger, W., & Varney, R. N. 1974, JChPh, 61, 1180 Hoyermann, K., Olzmann, M., Seeba, J., & Viskolcz, B. 1999, JPCA, 103, 5692 Hsu, C.-C., Lin, M., Mebel, A., & Melius, C. 1997, JPCA, 101, 60 Hu, R., & Seager, S. 2014, ApJ, 784, 63 Hu, R., Seager, S., & Bains, W. 2012, ApJ, 761, 166 Hu, R., Seager, S., & Bains, W. 2013, ApJ, 769, 6 Hu, R., Seager, S., & Yung, Y. L. 2015, arXiv:1505.02221 Huebner, W. F., & Carpenter, C. W. 1979, NASA STI/Recon Technical Report N, 80, 24243 Huebner, W. F., Keady, J. J., & Lyon, S. P. 1992, Ap&SS, 195, 1 Huebner, W. F., & Mukherjee, J. 2015, Planet. Space Sci., 106, 11 Huffman, R. E. 1969, CaJCh, 47, 1823 Huffman, R. E., Tanaka, Y., & Larrabee, J. 1963, JChPh, 39, 910 Humpfer, R., Oser, H., & Grotheer, H.-H. 1995, Int J Chem Kin, 27, 577 Hunt, M., Kerr, J., & Trotman-Dickenson, A. 1965, J Chem Soc, 5074 Huntress, W. T., Anicich, V., McEwan, M., & Hiraoka, K., & Kebarle, P. 1980, CaJCh, 58, 2262 Karpas, Z. 1980, ApJS, 44, 481 Hoehlein, G., & Freeman, G. 1970, JAChS, 92, Huntress, W. T., & Baldeschwieler, J. D. 1969, 6118 Nature, 223, 468 Homann, K., & Wellmann, C. 1983, Berich Bunsen Huntress, W. T., & Elleman, D. 1970, JAChS, 92, Gesell, 87, 609 3565 Hopkinson, A., Mackay, G., & Bohme, D. 1979, Huntress, W. T., Mosesman, M. M., & Elleman, CaJCh, 57, 2996 D. D. 1971, JChPh, 54, 843 33 Huntress, W. T., Pinizzotto, R. F., & Lau- Jebens, D. S., Lakkaraju, H. S., McKay, C. P., & denslager, J. B. 1973, JAChS, 95, 4107 Borucki, W. J. 1992, GRL, 19, 273 Husain, D., & Lee, Y. 1988, Int J Chem Kin, 20, Jensen, D. E. 1982, FaTr I, 78, 2835 223 Johnsen, R., Brown, H., & Biondi, M. A. 1970, Husain, D., & Marshall, P. 1986, Int J Chem Kin, JChPh, 52, 5080 18, 83 Johnsen, R., Castell, F., & Biondi, M. A. 1974, Husain, D., & Young, A. N. 1975, FaTr II, 71, 525 JChPh, 61, 5404 Huynh, L. K., & Truong, T. N. 2008, Theoretical Johnson, A. P., Cleaves, H. J., Dworkin, J. P., Chemistry Accounts, 120, 107 et al. 2008, Science, 322, 404 Huynh, L. K., & Violi, A. 2008, J Org Chem, 73, Johnston, H. S. 1951, JAChS, 73, 4542 94 Jones, J., Birkinshaw, K., & Twiddy, N. 1981a, Ibragimova, L. 1986, Kin Catal, 27 CPL, 77, 484 Ikeda, E., & Mackie, J. C. 1996 in International —. 1981b, JPhB, 14, 2705 Symposium on Combustion (Elsevier), 597 Ikezoe, Y., Matsuoka, S., & Takebe, M. 1987, Gas phase ion-molecule reaction rate constants through 1986 (Ion reaction research group of the Mass spectroscopy society of Japan) Imai, N., & Toyama, O. 1962, Bulletin of the Chemical Society of Japan, 35, 860 Indriolo, N., Fields, B. D., & McCall, B. J. 2009, ApJ, 694, 257 Ing, W.-C., Sheng, C. Y., & Bozzelli, J. W. 2003, Fuel Process Technol, 83, 111 Inn, E. C. 1975, JAtS, 32, 2375 Jachimowski, C. J. 1977, CoFl, 29, 55 Jaffe, S., Karpas, Z., & Klein, F. S. 1973, JChPh, 58, 2190 Jaffe, S., & Klein, F. 1974, IJMIP, 14, 459 Jamieson, J., Brown, G., & Tanner, J. 1970, Ca- JCh, 48, 3619 Jarrold, M. F., Bass, L. M., Kemper, P. R., van Koppen, P. A., & Bowers, M. T. 1983, JChPh, 78, 3756 Jasper, A. W., Klippenstein, S. J., Harding, L. B., & Ruscic, B. 2007, JCPA, 111, 3932 Javoy, S., Naudet, V., Abid, S., & Paillard, C. 2003, ExTFS, 27, 371 34 Jones, J., Lister, D., & Twiddy, N. 1979, JPhB, 12, 2723 Joshi, A., You, X., Barckholtz, T. A., & Wang, H. 2005, JPCA, 109, 8016 Ju, L.-P., Han, K.-L., & Varandas, A. J. 2007, Int J Chem Kin, 39, 148 Kappes, M. M., & Staley, R. H. 1981, JAChS, 103, 1286 Karkach, S. P., & Osherov, V. I. 1999, JChPh, 110, 11918 Karpas, Z., Anicich, V., & Huntress, W. 1978, CPL, 59, 84 —. 1979, JChPh, 70, 2877 Karpas, Z., & Klein, F. S. 1975, IJMIP, 16, 289 Kasper, S., & Franklin, J. 1972, JChPh, 56, 1156 Kasting, J. F. 1993, Science, 259, 920 Katayama, D., Huffman, R., & O’Bryan, C. 1973, JChPh, 59, 4309 Kato, A., & Cvetanovic, R. 1967, CaJCh, 45, 1845 Keller, C., Anicich, V., & Cravens, T. 1998, P&SS, 46, 1157 Keller-Rudek, H., Moortgat, G. K., Sander, R., & Sorensen, R. 2013, ESSD, 5, 365 Kelly, N., & Heicklen, J. 1978, J Photochem, 8, 83 Kretschmer, C., & Petersen, H. 1963, JChPh, 39, Kemper, P. R., & Bowers, M. 1984, Int J Chem 1772 Kin, 16, 707 Kronebusch, P., & Berkowitz, J. 1976, IJMIP, 22, Kemper, P. R., Bowers, M. T., Parent, D. C., et al. 283 1983, JChPh, 79, 160 Kruse, T., & Roth, P. 1997, JCPA, 101, 2138 Kern, R., Singh, H., & Wu, C. 1988, Int J Chem —. 1999, Int J Chem Kin, 31, 11 Kin, 20, 731 Kukui, A., Jungkamp, T., & Schindler, R. 1995, Kiefer, J., Mitchell, K., Kern, R., & Yong, J. 1988, Berich Bunsen Gesell, 99, 1565 JPhCh, 92, 677 Kiefer, J., Santhanam, S., Srinivasan, N. K., et al. 2005, Proceedings of the Combustion Institute, 30, 1129 Kim, J., Theard, L., & Huntress, W.-T. 1975, JChPh, 62, 45 Kumakura, M., Arakawa, K., & Sugiura, T. 1978a, B Chem Soc Jpn, 51, 49 —. 1978b, IJMIP, 26, 303 —. 1979, IJMIP, 29, 21 Laidler, K., & McKenney, D. 1964, RSPSA, 278, Kim, S. J., Geballe, T. R., Kim, J., et al. 2010, 517 Icarus, 208, 837 Laidler, K., & Wojciechowski, B. 1961, RSPSA, Kim, Y. H., & Fox, J. L. 1994, Icarus, 112, 310 260, 103 Kim, Y. H., Fox, J. L., Black, J. H., & Moses, J. I. 2014, JGRA, 119, 384 Klatt, M., Spindler, B., & Wagner, H. G. 1995, ZPC, 191, 241 Klemm, R. 1965, CaJCh, 43, 2633 Knutson, H. A., Charbonneau, D., Allen, L. E., Burrows, A., & Megeath, S. T. 2008, ApJ, 673, 526 Knyazev, V. D., Benscura, ´A, Stoliarov, S. I., et al. 1996, JPhCh, 100, 11346 Knyazev, V. D., Stoliarov, S. I., & Slagle, I. R. 1996b in International Symposium on Combus- tion (Elsevier), 513 Lamb, J. J., Mozurkewich, M., & Benson, S. W. 1984, JPhCh, 88, 6441 Lambert, R., Christie, M., Golesworthy, R., & Linnett, J. 1968, RSPSA 302, 167 Lambert, R., Christie, M., & Linnett, J. 1967, Chem Commun (London), 388 Langer, S., & Ljungstrom, E. 1994, Int J Chem Kin, 26, 367 —. 1995, FaTr, 91, 405 Laudenslager, J. B., Huntress, W. T., & Bowers, M. T. 1974, JChPh, 61, 4600 Laufer, A. H., & Fahr, A. 2004, ChRv, 104, 2813 Lavendy, H., Gandara, G., & Robbe, J. 1984, Koch, E.-E., & Skibowski, d. M. 1971, CPL, 9, JMoSp, 106, 395 429 Lavendy, H., Robbe, J., & Gandara, G. 1987, Koike, T., Kudo, M., Maeda, I., & Yamada, H. JPhB, 20, 3067 2000, Int J Chem Kin, 32, 1 Kooij, D. M. 1893, ZPC, 12, 155 Lavvas, P., Coustenis, A., & Vardavas, I. 2008a, P&SS, 56, 27 Korovkina, T. 1976, High Energ Chem, 10, 75 —. 2008b, P&SS, 56, 67 Kovalenko, L. J., Jucks, K. W., Salawitch, R. J., Lavvas, P., Koskinen, T., & Yelle, R. V. 2014, et al. 2007, JGL, 34, 19801 ApJ, 796, 15 35 Lawson, G., Bonner, R. F., Mather, R. E., Todd, Lifshitz, A., Tamburu, C., & Carroll, H. F. 1997, J. F., & March, R. E. 1976, FaTr I, 72, 545 Int J Chem Kin, 29, 839 Lee, G., Helling, Ch., Dobbs-Dixon, I., & Juncher, Lifshitz, A., Tamburu, C., Frank, P., & Just, T. D. 2015, arXiv:1505.06576 1993, JPhCh, 97, 4085 Lee, J., & Bozzelli, J. W. 2003, Int J Chem Kin, Lifshitz, C., & Tassa, R. 1973, IJMIP, 12, 433 35, 20 Lifshitz, C., Wu, R., Haartz, J., & Tiernan, T. Lee, L., Carlson, R., Judge, D., & Ogawa, M. 1973, 1977, JChPh, 67, 2381 JQSRT, 13, 1023 Lifshitz, C., Wu, R., Tiernan, T., & Terwilliger, Lellouch, E., Romani, P. N., & Rosenqvist, J. D. 1978, JChPh, 68, 247 1994, Icarus, 108, 112 Lin, M., He, Y., & Melius, C. 1992, Int J Chem Levush, S., Abadzhev, S., & Shevchuk, V. 1969, Kin, 24, 1103 Neftekhimiya, 9, 215 Levy, J. B. 1956, JAChS, 78, 1780 —. 1993, JPhCh, 97, 9124 Lindemann, F. A., Arrhenius, S., Langmuir, I., Li, Q.-S., & Wang, C. Y. 2004, JCoCh, 25, 251 et al. 1922, TrFa, 17, 598 Li, Q. S., Zhang, Y., & Zhang, S. 2004, JCPA, Linder, D. P., Duan, X., & Page, M. 1996, JChPh, 108, 2014 104, 6298 Li, S., & Williams, F. 1996 in International Sym- Lindinger, W. 1973, PhRvA, 7, 328 posium on Combustion (Elsevier), 1017 Li, S., Zhang, Q., & Wang, W. 2006, CPL, 428, 262 Liang, M.-C., Parkinson, C. D., Lee, A. Y.-T., Yung, Y. L., & Seager, S. 2003, ApJ, 596, L247 Lias, S. 1988, JPCRD, 17, 1 Lias, S., Collin, G., Rebbert, R., & Ausloos, P. 1970, JChPh, 52, 1841 Lichtin, D., Berman, M., & Lin, M. 1984, CPL, 108, 18 Liddy, J., Freeman, C., & McEwan, M. 1977a, MNRAS, 180, 683 —. 1977b, IJMIP, 23, 153 Lifshitz, A., & Ben-Hamou, H. 1983, JPhCh, 87, 1782 Lifshitz, A., & Frenklach, M. 1980, Int J Chem Kin, 12, 159 Lifshitz, A., & Tamburu, C. 1994, JPhCh, 98, 1161 —. 1998, Int J Chem Kin, 30, 341 —. 1976, JChPh, 64, 3720 Lindinger, W., Albritton, D., & Fehsenfeld, F. 1979, JChPh, 70, 2038 Lindinger, W., Albritton, D., Fehsenfeld, F., & Ferguson, E. 1975a, JGR, 80, 3725 —. 1975b, JChPh, 63, 3238 Lindinger, W., Albritton, D., Fehsenfeld, F., Schmeltekopf, A., & Ferguson, E. 1975c, JChPh, 62, 3549 Lindinger, W., Albritton, D., Howard, C. J., Fehsenfeld, F., & Ferguson, E. 1975d, JChPh, 63, 5220 Lindinger, W., Albritton, D., McFarland, M., et al. 1975e, JChPh, 62, 4101 Lindinger, W., Fehsenfeld, F., Schmeltekopf, A., & Ferguson, E. 1974, JGR, 79, 4753 Lindinger, W., Howorka, F., Lukac, P., et al. 1981, PhRvA, 23, 2319 Lindinger, W., McFarland, M., Fehsenfeld, F., et al. 1975f, JChPh, 63, 2175 36 Lindley, C. R., Calvert, J. G., & Shaw, J. H. 1979, Madhusudhan, N., Lee, K. K. M., & Mousis, O. CPL, 67, 57 2012, ApJ, 759, L40 Little, B., Anger, C. D., Ingersoll, A. P., et al. Madhusudhan, N., & Seager, S. 2009, ApJ, 707, 1999, Icarus, 142, 306 24 Liu, G.-x., Ding, Y.-h., Li, Z.-s., et al. 2002, Madronich, S. 1987, JGR: Atmospheres, 92, 9740 PCCP, 4, 1021 Magnotta, F., & Johnston, H. S. 1980, GeoRL, 7, Lloyd, A. C. 1974, Int J Chem Kin, 6, 169 769 Loison, J. C., H´ebrard, E., Dobrijevic, M., Hick- son, K. M., Caralp, F., et al. 2015, Icarus, 247, 218 Lombos, B., Sauvageau, P., & Sandorfy, C. 1967, JMoSp, 24, 253 Louge, M. Y., & Hanson, R. K. 1984, CoFl, 58, 291 Lukirskii, A., Brytov, I., & Zimkina, T. 1964, OptSp, 17, 234 Luque, A., & Ebert, U. 2009, NatGe, 2, 757 Macdonald, R. G. 2007, PCCP, 9, 4301 Mackay, G. I., Betowski, L., Payzant, J., Schiff, Mahmud, K., Marshall, P., & Fontijn, A. 1987, JPhCh, 91, 1568 Manion, J. A., Huie, R. E., Levin, R. D., et al. 2013, NIST Standard Reference Database 17, Version 7.0 (Web Version), Release 1.6.8, Data version 2013.03 Mansergas, A., & Anglada, J. M. 2006, JCPA, 110, 4001 Marinov, N. M., Pitz, W. J., Westbrook, C. K., et al. 1998, CoFl, 114, 192 Mark, T. D., & Oskam, H. 1971, PhRvA, 4, 1445 Marmo, F. 1953, JOSA, 43, 1186 H., & Bohme, D. 1976a, JPhCh, 80, 2919 Marx, R., Mauclaire, G., Fehsenfeld, F., Dunkin, Mackay, G. I., & Bohme, D. K. 1978, IJMIP, 26, D., & Ferguson, E. 1973, JChPh, 58, 3267 327 Massie, S. T., & Hunten, D. M. 1981, JGR, 86, Mackay, G. I., Hemsworth, R. S., & Bohme, D. K. 9859 1976b, CaJCh, 54, 1624 Masuoka, T., & Samson, J. A. 1981, JChPh, 74, Mackay, G. I., Hopkinson, A., & Bohme, D. 1978, 1093 JAChS, 100, 7460 Matsui, Y., & Nomaguchi, T. 1978, CoFl, 32, 205 Mackay, G. I., Rakshit, A. B., & Bohme, D. K. 1982, CaJCh, 60, 2594 Mackay, G. I., Schiff, H., & Bohme, D. 1981, Ca- Matsumoto, A., Okada, S., Misaki, T., Taniguchi, S., & Hayakawa, T. 1975, B Chem Soc Jpn, 48, 794 JCh, 59, 1771 Matsunaga, F., & Watanabe, K. 1967, Sci Light, Mackay, G. I., Tanaka, K., & Bohme, D. K. 1977, 16, 31 IJMIP, 24, 125 Matsuoka, S., & Ikezoe, Y. 1988, JPhCh, 92, 1126 Mackay, G. I., Tanner, S. D., Hopkinson, A. C., & Mauclaire, G., Derai, R., & Marx, R. 1978, IJMIP, Bohme, D. K. 1979, CaJCh, 57, 1518 26, 289 Mackay, G. I., Vlachos, G., Bohme, D.-K., & Mayer, S., & Schieler, L. 1968, JPhCh, 72, 2628 Schiff, H. 1980, IJMIP, 36, 259 Mayer, S., Schieler, L., & Johnston, H. S. 1966, Madhusudhan, N., Harrington, J., Stevenson, JChPh, 45, 385 K. B., et al. 2011, Nature, 469, 64 37 —. 1967 in International Symposium on Combus- Meot-Ner, M., Karpas, Z., & Deakyne, C. A. 1986, tion (Elsevier), 837 JAChS, 108, 3913 Mayer, T., & Lampe, F. 1974a, JPhCh, 78, 2645 —. 1974b, JPhCh, 78, 2433 McAllister, T. 1973, IJMIP, 10, 419 McAllister, T., & Pitman, P. 1976, IJMIP, 19, 423 McBride, B. J., Gordon, S., & Reno, M. A. 1993, Coefficients for calculating thermodynamic and transport properties of individual species, Tech. rep. McElroy, M. B., & McConnell, J. C. 1971, JGR, 76, 6674 McEwan, M., Anicich, V., & Huntress, W. 1981, IJMIP, 37, 273 McEwan, M., Anicich, V., Huntress, W., Kemper, P., & Bowers, M. 1983, IJMIP, 50, 179 McFarland, M., Dunkin, D., Fehsenfeld, F., Schmeltekopf, A., & Ferguson, E. 1972, JChPh, 56, 2358 McKenney, D., Wojciechowski, B., & Laidler, K. 1963, CaJCh, 41, 1993 McKnight, L. 1970, PhRvA, 2, 762 McNesby, J. R., & Okabe, H. 1964, Adv Pho- tochem, 3, 157 McNesby, J. R., Tanaka, I., & Okabe, H. 1962, J. Chem. Phys., 36, 605 Meaburn, G., & Gordon, S. 1968, JPhCh, 72, 1592 Mertens, J. D., & Hanson, R. K. 1996 in Inter- national Symposium on Combustion (Elsevier), 551 Mertens, J. D., Kohse-Hoinghaus, K., Hanson, R. K., & Bowman, C. T. 1991, Int J Chem Kin, 23, 655 Metcalfe, E., Booth, D., McAndrew, H., & Woo- ley, W. 1983, Fire Mater, 7, 185 Metzger, P., & Cook, G. 1964, JChPh, 41, 642 Meyer, E., Olschewski, H., Troe, J., & Wag- ner, H. G. 1969 in International Symposium on Combustion (Elsevier), 345 Meyer, J. P., & Hershberger, J. F. 2005, JCPA, 109, 4772 Michael, J., Kumaran, S., & Su, M.-C. 1999, JPhCh A, 103, 5942 Mick, H.-J., Burmeister, M., & Roth, P. 1993, AIAA, 31, 671 Miller, J. A., & Glarborg, P. 1999, Int J Chem Kin, 31, 757 Miller, J. A., & Melius, C. F. 1988 in International Symposium on Combustion (Elsevier), 919 Miller, J. A., & Melius, C. F. 1989 in International Symposium on Combustion (Elsevier), 1031 Miller, J. A., & Melius, C. F. 1992, Int J Chem Kin, 24, 421 Miller, J. A., Pilling, M. J., & Troe, J. 2005, P Meagher, N. E., & Anderson, W. R. 2000, JCPA, Combust Inst, 30, 43 104, 6013 Mebel, A., & Lin, M. 1997, IRPC, 16, 249 —. 1999, JCPA, 103, 2088 Mebel, A., Lin, M., Morokuma, K., & Melius, C. 1996, Int J Chem Kin, 28, 693 Melton, C., & Rudolph, P. 1960, JChPh, 32, 1128 Mentall, J., Gentieu, E., Krauss, M., & Neumann, D. 1971, JChPh, 55, 5471 Miller, J. L., McCunn, L. R., Krisch, M. J., Butler, L. J., & Shu, J. 2004, JChPh, 121, 1830 Miller, S. L. 1953, Science, 117, 528 Miller, S. L., & Urey, H. C. 1959, Science, 130, 245 Miller, T. M., Wetterskog, R. E., & Paulson, J. F. 1984, JChPh, 80, 4922 Miyakawa, S., Yamanashi, H., Kobayashi, K., Cleaves, H. J., & Miller, S. L. 2002, PNAS, 99, 14628 38 Miyoshi, A., Ohmori, K., Tsuchiya, K., & Matsui, H. 1993, CPL, 204, 241 Molina, L. T., & Molina, M. J. 1981, J Photochem, 15, 97 Mulvihill, J. N., & Phillips, L. F. 1975 in Inter- national Symposium on Combustion (Elsevier), 1113 Munson, M. S., & Field, F. H. 1969, JAChS, 91, Molina-Cuberos, G. J., Lopez-Moreno, J. J., Ro- drigo, R., Lara, L. M., & O’Brien, K. 1999, P&SS, 47, 1347 3413 Munson, M. S., Franklin, J., & Field, F. 1964, JPhCh, 68, 3098 Monks, P., Romani, P., Nesbitt, F., Scanlon, M., Murrell, J., & Rodriguez, J. 1986, JMoSt, 139, 267 & Stief, L. 1993, JGR: Planets, 98, 17115 Moortgat, G., SSlemr, F., & Warneck, P. 1977, Int J Chem Kin, 9, 249 Moortgat, G., & Warneck, P. 1975, ZNatA, 30, 835 Morris, E., & Niki, H. 1973, Int J Chem Kin, 5, 47 Morrissey, R. J., & Schubert, C. 1963, CoFl, 7, 263 Moses, J. I. 2014, RSPTA, 372, 20130073 Moses, J. I., Armstrong, E. S., Fletcher, L. N., et al. 2015, Icarus, in press Moses, J. I., B´ezard, B., Lellouch, E., et al. 2000a, Icarus, 143, 244 Moses, J. I., Fouchet, T., B´ezard, B., et al. 2005, JGR: Planets, 110, 8001 Moses, J. I., Lellouch, E., B´ezard, B., et al. 2000b, Icarus, 145, 166 Moses, J. I., Madhusudhan, N., Visscher, C., & Freedman, R. S. 2013, ApJ, 763, 25 Moses, J. I., Visscher, C., Fortney, J. J., et al. 2011, ApJ, 737, 15 Moshkina, R., Polyak, S., Sokolova, N., Mas- terovoi, I., & Nalbandyan, A. 1980, Int J Chem Kin, 12, 315 Mount, G., & Moos, H. 1978, ApJ, 224, L35 Musin, R., & Lin, M. 1998, JCPA, 102, 1808 Myer, J. A., & Samson, J. A. 1970, JChPh, 52, 266 Myerson, A. L. 1973in , Elsevier, 219–228 Nadtochenko, V., Sarkisov, O., & Vedeneev, V. 1979, Doklady Akademii Nauk SSSR, 244, 152 Nakata, R., Watanabe, K., & Matsunaga, F. 1965, Sci Light, 14, 54V71 Nakayama, T., Kitamura, M. Y., & Watanabe, K. 1959, JChPh, 30, 1180 Nakayama, T., & Watanabe, K. 1964, JChPh, 40, 558 Natarajan, K., & Bhaskaran, K. 1981, Experimen- tal and analytical investigation of high temper- ature ignition of ethanol, Tech rep, Indian Inst of Tech, Madras Dept of Mechanical Engineer- ing Natarajan, K., Thielen, K., Hermanns, H., & Roth, P. 1986, Berich Bunsen Gesell, 90, 533 Navarro-Gonz´alez, R., Villagr´an-Muniz, M., So- bral, H., Molina, L. T., & Molina, M. J. 2001, GRL, 28, 3867 Nee, J. B., & Lee, L. 1984, JChPh, 81, 31 Neilson, P. V., Bowers, M. T., Chau, M., David- son, W. R., & Aue, D. H. 1978, JAChS, 100, 3649 Neiman, M., & Feklisov, G. 1961, Zh Fiz Khim, Mousavipour, S. H., & Saheb, V. 2007, B Chem 35, 1064 Soc Jpn, 80, 1901 Moylan, C. R., Jasinski, J. M., & Brauman, J. I. 1985, JAChS, 107, 1934 Nissen, P. E. 2013, A&A, 552, 10 Nguyen, H. M. T., Zhang, S., Peeters, J., Truong, T. N., & Nguyen, M. T. 2004, CPL, 388, 94 39 Nielsen, O. J., Sidebottom, H. W., Donlon, M., & Park, J., & Hershberger, J. F. 1993, JChPh, 99, Treacy, J. 1991, CPL, 178, 163 3488 Nizamov, B., & Dagdigian, P. J. 2003, JCPA, 107, Park, J., & Lin, M. 1997, JCPA, 101, 5 2256 Oehlschlaeger, M. A., Davidson, D. F., & Hanson, R. K. 2004, JPCA, 108, 4247 Ohmori, K., Miyoshi, A., Matsui, H., & Washida, N. 1990, JPhCh, 94, 3253 Okabe, H. 1970, JChPh, 53, 3507 —. 1980, JChPh, 72, 6642 —. 1981, JChPh, 75, 2772 —. 1983, JChPh, 78, 1312 Okabe, H., & Becker, D. 1963, JChPh, 39, 2549 Okabe, H., et al. 1978, Photochemistry of Small Molecules, Vol. 431 (New York: Wiley) Okada, S., Matsumoto, A., Dohmaru, T., Taniguchi, S., & Hayakawa, T. 1972, Mass Spectroscopy (Japan), 20, 311 O’Neal, E., & Benson, S. W. 1962, JChPh, 36, 2196 Opansky, B. J., & Leone, S. R. 1996a, JPhCh, 100, 4888 Opansky, B. J., & Leone, S. R. 1996b, JPhCh, 100, 19904 Oparin, A. I. 1957, The Origin of Life on the Earth. Orville, R. E. 1968, JAtS, 25, 852 Osborn, D. L. 2003, JCPA, 107, 3728 Owens, A., Hales, C., Filkin, D., et al. 1985, JGR: Atmospheres, 90, 2283 Padial, N., Collins, L., & Schneider, B. 1985, ApJ, 298, 369 Pang, J.-L., Xie, H.-B., Zhang, S.-W., Ding, Y.- H., & Tang, A.-Q. 2008, JCPA, 112, 5251 Paraskevopoulos, G., & Winkler, C. A. 1967, JPhCh, 71, 947 Parkes, D. A. 1972a, FaTr I, 68, 613 —. 1972b, FaTr I, 68, 627 Parra-Rojas, F. C., Luque, A., & Gordillo- V´aZquez, F. J. 2013, JGR:Space Physics, 118, 5190 Patel, B. H., Percivalle, C., Ritson, D. J., Duffy, C. D., & Sutherland, J. D. 2015, NatCh, 7, 301 Patrick, R., & Golden, D. M. 1984, Int J Chem Kin, 16, 1567 Patterson, W., & Greene, E. 1962, JChPh, 36, 1146 Payzant, J., Tanaka, K., Betowski, L., & Bohme, D. 1976, JAChS, 98, 894 Peeters, J., Boullart, W., & Devriendt, K. 1995, JPhCh, 99, 3583 Petrov, Y. P., Turetskii, S., & Bulgakov, A. 2009, Kin Catal, 50, 344 Petty, J. T., Harrison, J. A., & Moore, C. B. 1993, JPhCh, 97, 11194 Phillips, E., Lee, L., & Judge, D. 1977, JQSRT, 18, 309 Pitts, W. M., Pasternack, L. & McDonald, J. R. 1982, CP, 68, 417 Porter, R. P., & Noyes, W. A. 1959, JAChS, 81, 2307 Pouilly, B., Robbe, J., Schamps, J., & Roueff, E. 1983, JPhB, 16, 437 Powner, M. W., Gerland, B, & Sutherland, J. D. 2009, Nature, 459.7244, 239 Prasad, S. S., & Huntress, W. T. 1980, ApJ, 43, 1 Price, C., Penner, J., & Prather, M. 1997, J. Geo- phys. Res., 102, 5929 Pshezhetskii, S. Y., Morozov, N., Kamenetskaya, S., Siryatskaya, V., & Gribova, E. 1959, Russ J Phys Chem, 33, 402 40 Quandt, R. W., & Hershberger, J. F. 1995, Rogers, J. D. 1990, JPhCh, 94, 4011 JPhCh, 99, 16939 Rohrig, M., Romming, H.-J., & Wagner, H. G. Queloz, D., Eggenberger, A., Mayor, M., et al. 1994, Berich Bunsen Gesell, 98, 1332 2000, A&A, 359, L13 Raksit, A. B. 1982, IJMIP, 41, 185 —. 1986, IJMSI, 69, 45 Raksit, A. B., & Bohme, D. K. 1984, IJMSI, 57, 211 —. 1985, IJMSI, 63, 217 Raksit, A. B., Schiff, H., & Bohme, D. 1984, IJMSI, 56, 321 Raksit, A. B., & Warneck, P. 1979, ZNatA, 34, 1410 —. 1980a, JChPh, 73, 2673 —. 1980b, IJMIP, 35, 23 —. 1980c, FaTr II, 76, 1084 —. 1981, JChPh, 74, 2853 Ray, A., Daele, V., Vassalli, I., Poulet, G., & Le Bras, G. 1996, JPhCh, 100, 5737 Reitel’boim, M., Romanovich, L., & Vedeneev, B. 1978, Kin Catal, 19, 1131 Ribas, I., Guinan, E. F., Gudel, M., & Audard, M. 2005, ApJ, 622, 680 Ribas, I., Porto de Mello, G. F., Ferreira, L. D., H´ebrard, E., Selsis, F., et al. 2010, ApJ714, 384 Rim, K. T., & Hershberger, J. F. 1999, JCPA, 103, 3721 Rimmer, P. B., Herbst, E., Morata, O., & Roueff, E. 2012, A&A, 537, Rimmer, P. B., & Helling, Ch. 2013, ApJ, 774, 108 Rimmer, P. B., Helling, Ch., & Bilger, C. 2014, IJAsB, 13, 173 Robertson, R., Hils, D., Chatham, H., & Gal- lagher, A. 1983, ApPhL, 43, 544 Roble, R., & Ridley, E. 1994, GeoRL, 21, 417 Rohrig, M., & Wagner, H. G. 1994 in International Symposium on Combustion (Elsevier), 975 Romani, P. N., Jennings, D. E., Bjoraker, P. V., et al. 2008, Icarus, 198, 420 Roose, T., Hanson, R., & Kruger, C. 1978, in 11th International Symposium on Shock Tubes and Waves, 245 Roscoe, J. M., & Roscoe, S. G. 1973, CaJCh, 51, 3671 Ross, S. K., Sutherland, J. W., Kuo, S.-C., & Klemm, R. B. 1997, JPCA, 101, 1104 Rottman, G. J., Woods, T. N., & McClintock, W. 2006, AdSpR, 37, 201 Rowe, B., Fahey, D., Fehsenfeld, F., & Albritton, D. 1980, JChPh, 73, 194 Rowe, B., Fahey, D., Ferguson, E., & Fehsenfeld, F. 1981, JChPh, 75, 3325 Rustgi, O. P. 1964, JOSA, 54, 464 Saeys, M., Reyniers, M.-F., Van Speybroeck, V., Waroquier, M., & Marin, G. B. 2006, CPPC, 7, 188 Safrany, D. R., & Jaster, W. 1968, JPhCh, 72, 3305 Sahetchian, K., Heiss, A., & Rigny, R. 1987, JPhCh, 91, 2382 Saito, K., Kakumoto, T., & Murakami, I. 1984, CPL, 110, 478 Saito, K., Mochizuki, Y., Yoshinobu, I., & Ima- mura, A. 1990, CPL, 167, 347 Salahub, D., & Sandorfy, C. 1971, CPL, 8, 71 Samson, J. A., & Cairns, R. 1964, JGR, 69, 4583 —. 1965, JOSA, 55, 1035 Sander, R., Kerkweg, A., Jockel, P., & Lelieveld, Roche, A., Sutton, M., Bohme, D., & Schiff, H. J. 2005, ACP, 5, 445 1971, JChPh, 55, 5480 41 Sander, S. P., Friedl, R. R., Barker, J. R., et al. 2011, Chemical Kinetics and Photochemical Data for use in Atmospheric Studies 17, (JPL Publications) Seinfeld, J. H., & Pandis, S. N. 2006, Atmospheric Chemistry and Physics: From Air Pollution to Climate Change (2nd ed.; New Jersey: Wiley) Selwyn, G., Podolske, J., & Johnston, H. S. 1977, Sanders, N., Butler, J., Pasternack, L., & McDon- GeoRL, 4, 427 ald, J. 1980, CP, 48, 203 Sen, B., Toon, G. C., Osterman, G. B. 1998, JGR, Sanders, W., Lin, C., & Lin, M. 1987, CST, 51, 103, 3571 103 Senosiain, J. P., Klippenstein, S. J., & Miller, J. A. Sato, K., & Hidaka, Y. 2000, CoFl, 122, 291 2006, JPCA, 110, 5772 Saxon, R. P., Lengsfield III, B. H., & Liu, B. 1983, Setser, D., & Rabinovitch, B. 1962, CaJCh, 40, JChPh, 78, 312 1425 Sayah, N., Li, X., Caballero, J., & Jackson, W. M. Shardanand, & Rao, A. D. P. 1977, 1988, J Photoch Photobio A, 45, 177 J. Quant. Spec. Radiat. Transf., 17, 433 Scattergood, T. W., McKay, C. P., Borucki, W. J., Shaw, R. 1977, Int J Chem Kin, 9, 929 et al. 1989, Icarus, 81, 413 Schacke, H., Schmatjko, K., & Wolfrum, J. 1974, Arch Procesow Spalania, 5 Scherzer, K., Loser, U., & Stiller, W. 1987, ZCh, 27, 300 Schiff, H., & Bohme, D. K. 1979, ApJ, 232, 740 —. 1978, JPCRD, 7, 1179 Sheng, C. Y., Bozzelli, J. W., Dean, A. M., & Chang, A. Y. 2002, JPCA, 106, 7276 Showman, A. P., Cooper, C. S., Fortney, J. J., & Marley, M. S. 2008, ApJ, 682, 559 Shul, R., Upschulte, B., Passarella, R., Keesee, R., Schildcrout, S. M., & Franklin, J. 1970, JAChS, & Castleman, A. 1987a, JPhCh, 91, 2556 92, 251 Schlesinger, G., & Miller, S. L. 1983, JMolE, 19, 376 Schoen, R. I. 1962, JChPh, 37, 2032 Schulz, G., Klotz, H.-D., & Spangenberg, H.-J. 1985, ZCh, 25, 88 Schurath, U., Tiedemann, P., & Schindler, R. N. 1969, JPhCh, 73, 456 Schwarz, H., Brogi, M., de Kok, R., Birkby, J., & Snellen, I. 2015, arXiv:1502.04713 Seager, S., Bains, W., & Hu, R. 2013a, ApJ, 775, 104 —. 2013b, ApJ, 777, 95 Seery, D. 1969 in the International Symposium on Combustion, 12 —. 1987b, JPhCh, 91, 2556 Sieck, L. W. 1978, Int J Chem Kin, 10, 335 Sieck, L. W., & Futrell, J. 1968, JChPh, 48, 1409 Sieck, L. W., & Searles, S. K. 1970, JChPh, 53, 2601 Sims, I. R., Queffelec, J.-L., Travers, D., et al. 1993, CPL, 211, 461 Sivaramakrishnan, R., Michael, J., & Klippen- stein, S. 2009, JPCA, 114, 755 Skinner, G. B., & Ruehrwein, R. A. 1959, JPhCh, 63, 1736 Slack, M., & Fishburne, E. 1970, JChPh, 52, 5830 Slanger, T. G., & Black, G. 1982, JChPh, 77, 2432 Smith, D. L., & Adams, N. 1977a, ApJ, 217, 741 Seetula, J., Blomqvist, K., Kalliorinne, K., & Koskikallio, J. 1986, Acta Chem Scand, 40, 653 —. 1977b, CPL, 47, 145 42 —. 1978, CPL, 54, 535 —. 1980, CPL, 76, 418 —. 1981, MNRAS, 197, 377 Smith, D. L., Adams, N. G., & Alge, E. 1982, JChPh, 77, 1261 Smith, D. L., Adams, N. G., & Miller, T. 1978, JChPh, 69, 308 Smith, R. D., Smith, D. L., & Futrell, J. H. 1976, IJMIP, 19, 369 Song, S., Golden, D. M., Hanson, R. K., et al. 2003, Int J Chem Kin, 35, 304 Song, X., Hou, H., & Wang, B. 2005, PCCP, 7, 3980 Sumathi, R., & Nguyen, M. T. 1998, JCPA, 102, 8013 Sumathi, R., & Peyerimhoff, S. 1996, CPL, 263, 742 Sun, F., DeSain, J., Scott, G., et al. 2001, JPCA, 105, 6121 Sun, H., He, H.-Q., Hong, B., et al. 2006, IJQC, 106, 894 Sun, H., & Weissler, G. 1955, JChPh, 23, 1160 Sun, Y., Zhang, Q.-y., Ai, X.-c., Zhang, J.-p., & Sun, C.-c. 2004, JMoSt, 686, 123 Suzaki, K., Kanno, N., Tonokura, K., et al. 2006, CPL, 425, 179 Suzaki, K., Tsuchiya, K., Koshi, M., & Tezaki, A. Spokes, G. N., & Benson, S. W. 1967, JAChS, 89, 2007, JPCA, 111, 3776 6030 Swain, M. R., Tinetti, G., Vasisht, G., et al. 2009, Sridharan, U., & Kaufman, F. 1983, CPL, 102, 45 ApJ, 704, 1616 Srinivasan, N., Su, M.-C., & Michael, J. 2007, Szabo, I., & Derrick, P. 1971, IJMIP, 7, 55 JPCA, 111, 3951 Srinivasan, N., Su, M.-C., Sutherland, J., & Michael, J. 2005, JPCA, 109, 1857 Szekely, A., Hanson, R. K., & Bowman, C. T. 1985 in International Symposium on Combus- tion (Elsevier), 647 Stevenson, D. P., & Schissler, D. O. 1955, JChPh, Tabayashi, K., & Bauer, S. 1979, CoFl, 34, 63 23, 1353 Takahashi, K., Yamamoto, O., Inomata, T., & Ko- Stief, L., Donn, B., Glicker, S., Gentieu, E., & goma, M. 2007, Int J Chem Kin, 39, 97 Mentall, J. 1972, ApJ, 171, 21 Takahashi, S. 1972, Mem Def Acad: Phys & Chem Stockwell, W. R., & Calvert, J. G. 1978, J Pho- Eng, 12 tochem, 8, 193 Tanaka, K., Betowksi, L., Mackay, G., & Bohme, Stothard, N., Humpfer, R., & Grotheer, H.-H. D. 1976, JCP, 65, 3203 1995, CPL, 240, 474 Tanaka, Y., Inn, E. C., & Watanabe, K. 1953, Strausz, O. P., Duholke, W., & Gunning, H. E. JChPh, 21, 1651 1970, JAChS, 92, 4128 Tang, Y., Wang, R., & Wang, B. 2008, JCPA, 112, Streit, G. E. 1982, JPhCh, 86, 2321 5295 Striebel, F., Jusinski, L. E., Fahr, A., et al. 2004, PCCP, 6, 2216 Strobel, D. F. 1983, IRPC, 3, 145 Su, M.-C., Kumaran, S., Lim, K., et al. 2002, JPCA, 106, 8261 Tanner, S. D., Mackay, G. I., & Bohme, D. K. 1979b, CaJCh, 57, 2350 —. 1979c, CaJCh, 57, 2350 —. 1981, CaJCh, 59, 1615 Tanner, S. D., Mackay, G. I., Hopkinson, A., & Bohme, D. K. 1979a, IJMIP, 29, 153 43 Tanzawa, T., & Gardiner, W. 1980, JPhCh, 84, Tsang, W., & Hampson, R. 1986, JPCRD, 15, 236 1087 Tao, Y.-g., Ding, Y.-h., Li, Z.-s., Huang, X.-r., & Tsang, W., & Herron, J. T. 1991, JPCRD, 20, 609 Sun, C.-C. 2001, JCPA, 105, 9598 Taylor, W. W. L., Scarf, F. L., Russell, C. T., & Brace, L. H. 1979, Nature, 279, 614 Thaxton, A. G., Hsu, C.-C., & Lin, M. 1997, Int J Chem Kin, 29, 245 Theard, L. P., & Huntress, W. T. 1974, JChPh, 60, 2840 Thielen, K., & Roth, P. 1986, AIAA, 24, 1102 Tsuboi, T., & Hashimoto, K. 1981, CoFl, 42, 61 Tsuboi, T., Katoh, M., Kikuchi, S., & Hashimoto, K. 1981, JaJAP, 20, 985 Tu, L., Johnstone, C. P., Gudel, M., & Lammer, H. 2015, arXiv:1504.04546 Tuazon, E. C., Carter, W. P., Atkinson, R., Winer, A. M., & Pitts, J. N. 1984, EnST, 18, 49 Tzeng, S.-Y., Chen, P.-H., Wang, N. S., et al. Thomas, R., Barassin, A., & Burke, R. 1978, 2009, JPCA, 113, 6314 IJMSI, 28, 275 Thompson, B., Harteck, P., & Reeves, R. 1963, JGR, 68, 6431 Thweatt, W. D., Erickson, M. A., & Hershberger, J. F. 2004, JPCA, 108, 74 Thynne, J., & Gray, P. 1963a, TrFa, 59, 1149 —. 1963b, TrFa, 59, 1149 Tian, F., Toon, O. B., Pavlov, A. A., & De Sterck, H. 2005, Science, 308, 1014 Tich`y, M., Raksit, A., Lister, D., et al. 1979, IJMIP, 29, 231 Tomeczek, J., & Grado´n, B. 2003, CoFl, 133, 311 Tonkyn, R., & Weisshaar, J. C. 1986, JPhCh, 90, 2305 Vaghjiani, G. L. 1995, Int J Chem Kin, 27, 777 Vakhtin, A. B., Hard, D. E., Smith, I. W. M., & Leone, S. R. 2001, CPL, 344, 317 van Dishoeck, E. F. 1984, Dissertation van Dishoeck, E. F. 1987, JChPh, 86, 196 Vandooren, J., Bian, J., & Van Tiggelen, P. 1994, CoFl, 98, 402 Vandooren, J., & Van Tiggelen, P. 1977in , Else- vier, 1133–1144 Vardanyan, I., Sachyan, G., Philiposyan, A., & Nalbandyan, A. 1974, CoFl, 22, 153 Velinov, P. I. Y., & Mateev, L. N. 2008, Journal of Atmospheric and Solar-Terrestrial Physics, 70, 574 Toon, O. B., & Farlow, N. H. 1981, AREPS, 9, 19 Venot, O., H´ebrard, E., Ag´undez, M., et al. 2012, Toon, O. B., McKay, C. P., Ackerman, T. P., & Santhanam, K. 1989, J. Geophys. Res., 94, 16287 A&A, 546, A43 Verner, D., Ferland, G., Korista, K., & Yakovlev, D. 1996, ApJ, 465, 487 Trenwith, A. 1960, J Chem Soc, 3722 Verner, D., & Yakovlev, D. 1995, A&AS, 109, 125 Troe, J. 1983, Berich Bunsen Gesell, 87, 161 Verner, D., Yakovlev, D., Band, I., & —. 2005, JCPA, 109, 8320 Tsang, W. 1987, JPCRD, 16, 471 —. 1992, JPCRD, 21, 753 —. 2004, Int J Chem Kin, 36, 456 Trzhaskovskaya, M. 1993, ADNDT, 55, 233 Veyret, B., Rayez, J. C., & Lesclaux, R. 1982, JPhCh, 86, 3424 Viggiano, A., Albritton, D., Fehsenfeld, F., et al. 1980, ApJ, 236, 492 44 Viggiano, A., & Paulson, J. F. 1983, JChPh, 79, Watanabe, K. 1954, JChPh, 22, 1564 2241 Vijayan, M. 1980, FEBS Lett, 112, 135 Watanabe, K., & Jursa, A. 1964, JChPh, 41, 1650 Watanabe, K., Matsunaga, F. M., & Sakai, H. Villinger, H., Futrell, J., Howorka, F., Duric, N., 1967, ApOpt, 6, 391 & Lindinger, W. 1982, JChPh, 76, 3529 Visscher, C., & Moses, J. I. 2011, ApJ, 738, 72 Visscher, C., Moses, J. I., & Saslow, S. A. 2010, Icarus, 209, 602 Vogt, J., Williamson, A. D., & Beauchamp, J. 1978, JAChS, 100, 3478 Vuitton, V., Yelle, R. V., Lavvas, P., & Klippen- stein, S. J. 2012, ApJ, 744, 11 Wagner, A. F., & Bowman, J. M. 1987, JPhCh, 91, 5314 Wagner, H. G., Warnatz, J., & Zetzsch, C. 1971, An Asoc Quim Argent, 59 Wagner-Redeker, W., Kemper, P. R., Jarrold, M. F., & Bowers, M. T. 1985, JChPh, 83, 1121 Waite, J. H., Young, D. T., Cravens, T. E., Coates, A. J., Crary, F. J., et al. 2007, Science, 316.5826, 870 Wakamatsu, H., & Hidaka, Y. 2008, Int J Chem Kin, 40, 320 Wakelam, V., Herbst, E., Loison, J.-C., et al. 2012, ApJS, 199, 21 Walker, T., & Kelly, H. 1972, CPL, 16, 511 Wallington, T. J., & Japar, S. M. 1989, JAtC, 9, 399 Walsh, C., & Millar, T. J. 2011, in IAU Sympo- sium, Vol. 280, IAU Symposium, ed. J. Cer- nicharo & R. Bachiller, 56 Wang, C. Y., Zhang, S., & Li, Q. S. 2002, Theor Watanabe, K., & Sood, S. 1965, Sci Light, 14, 36 Watanabe, K., & Zelikoff, M. 1953, JOSA, 43, 753 Weaver, J., Meagher, J., & Heicklen, J. 1977, J Photochem, 6, 111 Westenberg, A. A., & De Haas, N. 1969, JPhCh, 73, 1181 Whyte, A., & Phillips, L. 1983, CPL, 98, 590 Wight, C. A., & Beauchamp, J. 1980, JPhCh, 84, 2503 Wijnen, M. 1960, JAChS, 82, 3034 Williamson, A. D., & Beauchamp, J. 1975, JAChS, 97, 5714 Williamson, D. G., & Bayes, K. D. 1967, JAChS, 89, 3390 Wilson, W. E. 1972, JPCRD, 1, 535 Woods, I., & Haynes, B. 1994 in International Symposium on Combustion (Elsevier), 909 Wooldridge, M. S., Hanson, R. K., & Bowman, C. T. 1996, Int J Chem Kin, 28, 361 Wu, C., Wang, H., Lin, M., & Fifer, R. 1990, JPhCh, 94, 3344 Wu, C.-W., Lee, Y.-P., Xu, S., & Lin, M. 2007, JPCA, 111, 6693 Wu, J.-Y., Liu, J.-Y., Li, Z.-S., & Sun, C.-c. 2003, JChPh, 118, 10986 Xu, S., & Lin, M. 2007, JCPA, 111, 6730 Chem Acc, 108, 341 Xu, Z.-F., & Lin, M. 2004, Int J Chem Kin, 36, Wang, S., Cui, J.-P., He, Y.-Z., Fan, B.-C., & 205 Wang, J. 2001, ChPhL, 18, 289 Xu, Z.-F., & Sun, C.-C. 1999, JMoSt, 459, 37 Warnatz, J. 1984, in Combustion Chemistry Xu, Z.-F., & Sun, J.-Z. 1998, JCPA, 102, 1194 (Springer), 197 Yang, J., Li, Q. S., & Zhang, S. 2008, JCoCh, 29, Warneck, P. 1972, Berich Bunsen Gesell, 76, 421 247 45 Yang, X., Goldsmith, C. F., & Tranter, R. S. 2009, Zaslonko, I., Petrov, Y. P., & Smirnov, V. 1997, JPCA, 113, 8307 Kinetics and catalysis, 38, 321 Yang, Y., Zhang, W., Pei, S., et al. 2005, JMoSt, Zaslonko, I., Smirnov, V., & Tereza, A. 1993, Kin 725, 133 Catal, 34, 531 Yasunaga, K., Kubo, S., Hoshikawa, H., Kame- sawa, T., & Hidaka, Y. 2008, Int J Chem Kin, 40, 73 Yee Quee, M., & Thynne, J. 1968, Berich Bunsen Zel’dovich, Y. B., & Raizer, Y. P. 1996, Physics of Shock Waves and High Temperature Hydrody- namic Phenomena, (New York, NY: Academic Press) Gesell, 72, 211 Zelikoff, M., Watanabe, K., & Inn, E. C. 1953, Yelle, R. V., Griffith, C. A. & Young, L. A. 2001, JChPh, 21, 1643 Icarus, 152, 331 Zhang, Y.-X., & Bauer, S. 1997, JPCB, 101, 8717 Yelle, R. V., Young, L. A., Vervack, R. J., et al. Zhang, Y.-X., Zhang, S., & Li, Q. S. 2004, CP, 1996, JGR, 101, 2149 296, 79 You, X., Wang, H., Goos, E., Sung, C.-J., & Klip- —. 2005, CP, 308, 109 penstein, S. J. 2007, JCPA, 111, 4031 Young, J. 1958, J Chem Soc, 2909 Young, L. B., Lee-Ruff, E., & Bohme, D. 1971, CaJCh, 49, 979 Yumura, M., & Asaba, T. 1981 in International Symposium on Combustion (Elsevier), 863 Yung, Y. L., Allen, M., & Pinto, J. P. 1984, ApJS, 55, 465 Yung, Y. L., & Demore, W. B., eds. 1999, Pho- tochemistry of Planetary Atmospheres (New York: OUP) Zabarnick, S., & Heicklen, J. 1985, Int J Chem Kin, 17, 455 Zahnle, K. 1986, JGR: Atmospheres, 91, 2819 Zahnle, K., Claire, M., & Catling, D. 2006, Geo- biology, 4, 271 Zahnle, K., Mac Low, M.-M., Lodders, K., & Fe- gley, B. 1995, GeoRL, 22, 1593 Zahnle, K., Marley, M. S., Freedman, R. S., Lod- ders, K., & Fortney, J. J. 2009, ApJ, 701, L20 Zalotai, L., Hunyadi-zolt´an, Z., B´erces, T., & Marta, F. 1983, Int J Chem Kin, 15, 505 Zarka, P., & Pedersen, B. M. 1986, Nature, 323, 605 Zhu, R., & Lin, M. 2003, JChPh, 119, 10667 —. 2007, JCPA, 111, 6766 —. 2009, CPL, 478, 11 Zhu, R., & Lin, M.-C. 2005, Int J Chem Kin, 37, 593 Zielinska, T. J., & Wincel, H. 1970, Nukleonika, 15, 343 Zsom, A., Kaltenegger, L., & Goldblatt, C. 2012, Icarus, 221, 603 This 2-column preprint was prepared with the AAS LATEX macros v5.2. 46 Table 2: Mixing Ratios for Laboratory Simulations∗ Model H2O CH4 NH3 0.08 A 0.06 B 0.04 C D 0.02 0.00 E 0.80 0.80 0.80 0.80 0.84 0.08 0.06 0.04 0.02 0.00 H2 0.04 0.04 0.04 0.04 0.00 CO2 0.0 0.02 0.04 0.06 0.08 N2 0.0 0.02 0.04 0.06 0.08 ∗ Not including the injected FeO and HCOOH ] 2 m c [ 10−17 10−18 10−19 10−20 10−21 10−22 10−23 10−24 10−25 0 500 1000 1500 2000 2500 3000 [Å] Fig. 1.— The photodissociation cross sections of NH3 → 1NH + H2, σ [cm2], as a function of wave- length, λ (A), from PhIDRates (original data from McNesby et al. 1962; Schurath et al. 1969, red line). The data is compared to our binned fit (blue line). 100 10-2 10-4 10-6 10-8 10-10 10-12 10-14 o i t a r g n i x i m 2000 4000 6000 8000 10000 T (K) Fig. 3.— Mixing ratio as a function of temper- ature. The solid line is from the Saha equation and the dashed line is the result from our model calculation. T4,P4 T3,P3 T3,P3 T2,P2 T2,P2 T1,P1 T1,P1 e r u s s e r P , e r u t a r e p m e T e r u s s e r P , e r u t a r e p m e T F(λ,0) F(λ,Δz1) F(λ,Δz1+Δz2) Fig. 4.— Picture representation of the model. The picture on the left represents the motion of the single parcel from the bottom of the atmosphere, T1, P1, up to the top of the atmosphere, T4, P4, and then back down, see Section 3.1. Once this journey is completed, we irradiate the atmosphere, by stacking up the parcel at different times, when it was located at different parts of the atmosphere. The picture on the right represents the calculation of the depth-dependent actinic flux discussed in Section 3.2. Only photons of wavelength between 1-10000 A are considered Figure 5 gives a flowchart for the calculation. Start Input p,T ,element abundances,Kzz, STAND network Bλ(τ = 0), σph, n(X) from ARGO no First global run? Solve for col dep. rad field, determine photo rate coefficients yes Run ARGO Global Run Update/restart p, T , Kzz, elemen- tal abundances At end of profile? no yes Check for con- vergence between two full p, T ARGO calculations Does it converge? no yes Output: Volume fraction as function of pressure Stop Fig. 5.— A flow-chart representation for the pro- gram. 47 s l o i t a r g n i x i m 100 10-2 10-4 10-6 10-8 10-10 H2 CO H 100 t [s] 10-6 10-4 10-2 10-3 o i t a r g n i x i m 10-4 H2O CH4 HNC (x105) 102 104 106 10-5 10-6 10-4 10-2 102 104 106 100 t [s] Fig. 2.— Mixing ratios as a function of time [s] at 1 bar and 1000 K (dashed lines), compared to chemical equilibrium (solid lines) for H2, H, CO, CH4 and H2O. H, H2 and He Chemistry H2 H He 10-10 10-8 10-6 10-4 10-2 100 102 10-10 10-8 10-6 10-4 10-2 100 102 ] r a b [ p Water and O2 Chemistry O2 O OH H2O 10-6 10-5 10-4 10-3 10-2 mix ratio 10-1 100 101 10-1210-1110-10 10-9 10-8 10-7 10-6 10-5 10-4 10-3 10-10 10-8 10-6 10-4 10-2 100 102 Nitrogen Chemistry NH3 ] r a b [ p N2 10-10 10-8 10-6 10-4 10-2 100 102 mix ratio Carbon Chemistry CO CO2 CH4 ] r a b [ p ] r a b [ p 10-12 10-11 10-10 10-9 10-8 10-7 10-6 10-5 10-4 10-10 10-9 10-8 mix ratio 10-7 10-6 mix ratio 10-5 10-4 10-3 Fig. 9.— Mixing ratios for various chemical species as a function of pressure, p [bar]. A comparison between our model (solid lines) and that of Moses et al. (2011, dashed lines), for H/H2 chemistry, water and O2 chemistry, nitrogen chemistry and carbon chemistry in the atmosphere of HD209458b. 48 Table 1: Initial Conditions for the Chemistry at the Lower Boundary in terms of n(X)/ngas 9.2092(-1) 7.8383(-2) 2.4787(-4) 6.2262(-5) 4.5105(-4) 2.3133(-6) 9.8766(-8) 2.9122(-7) 2.9122(-5) 3.6663(-5) 1.6004(-6) 2.9800(-5) 8.2077(-8) Species HD209458ba H He C N O Ar K Cl Fe Mg Na Si Ti CO H2 N2 NO O2 CO2 H2O N2O NH3 CH4 Earthb Early Earthc Jupiterd 1.3600(-1) 9.1150(-3) 1.1300(-7) 1.0000(-6) 7.9172(-1) 2.4000(-11) 1.9793(-1) 3.5000(-4) 1.0000(-2) 3.0200(-7) 2.4000(-10) 1.9390(-6) 4.9005(-5) 9.8010(-4) 7.8408(-1) 8.0000(-10) 8.6219(-1) 1.9602(-1) 9.8010(-3) 1.8100(-3) a Solar metallicity, from Asplund et al. (2009). b Surface mixing ratios based on the US Standard Atmosphere 1976. c Based on Early Earth models (Kasting 1993). d Moses et al. (2005). Table 3: Bond Constants for Benson Additivity a1 Species C–H 300-1000K 1.1E0 1000-6000K 4.0E-1 C–C 300-1000K -2.4E0 1000-6000K 1.8E0 C–O 300-1000K -6.8E-1 1000-6000K 1.7E0 O–H 5.0E-2 300-1000K 1000-6000K -4.5E-1 a2 a3 a4 a5 a6 a7 -3.0E-3 2.3E-3 1.2E-5 -1.9E-6 5.8E-8 1.4E-10 4.8E-12 -8.3E-15 -2.5E3 -2.0E3 -2.5E-2 2.8E0 1.3E-3 1.5E-3 -1.4E-5 5.7E-7 -2.8E-7 4.0E-11 -5.0E-13 -2.5E-15 3.0E3 2.5E2 3.0E0 -1.8E1 8.3E-3 1.3E-3 -1.0E-5 8.3E-6 -2.0E-7 5.5E-11 -2.3E-12 -3.8E-15 9.0E2 -6.5E2 4.8E-1 -1.1E1 -7.3E-3 -3.1E-3 1.8E-5 -6.7E-6 1.2E-7 -1.6E10 8.5E-12 -5.4E-14 -9.0E3 -7.5E3 -4.2E0 5.2E0 49 Neutral Species included in the Stand2015 Network Table 4 Network Formula Standard Formula Name Thermochem Data H C C(1D) C(1S) N O O(1D) O(1S) He Na Mg Si Cl Ar K Ti Fe H2 C2 N2 O2 O2(a1∆) H C C(1D) C(1S) N O O(1D) O(1S) He Na Mg Si Cl Ar K Ti Fe H2 C2 N2 O2 O2(a1∆) CH HN HN(a1∆) CH NH NH(a1∆) HO CN CO KH NO HCl NaH MgO SiH SiO KCl TiO FeO O3 3CH2 1CH2 OH CN CO KH NO HCl NaH MgO SiH SiO KCl TiO FeO O3 CH2(X3B1) CH2(a1A1) Atomic hydrogen Atomic carbon Singlet D carbon Singlet S carbon Atomic nitrogen Atomic oxygen Singlet D oxygen Singlet S oxygen Helium Atomic sodium Atomic magnesium Atomic silicon Atomic chlorine Argon Atomic potassium Atomic titanium Atomic iron Molecular hydrogen Dicarbon Molecular nitrogen Molecular oxygen Singlet oxygen Methylidyne radical Nitrogen monohydride Singlet nitrogen monohydride Hydroxyl radical Cyano radical Carbon monoxide Potassium hydride Nitric oxide Hydrogen chloride Sodium hydride Magnesium oxide Silylidyne Silicon monoxide Potassium chloride Titanium(II) oxide Iron(II) oxide Ozone Triplet methylene Singlet methylene 50 Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat NASA-CEA NASA-CEA Burcat NASA-CEA Burcat Burcat Burcat Burcat 10-8 10-7 10-6 10-5 10-4 10-3 10-2 10-1 100 ) r a b ( p C2H4 C4H2 C2H2 C2H6 10-14 10-12 10-10 10-8 10-6 mixing ratio 10-4 Drossart+1999 Yelle+1996 Fouchet+2000 Moses+2005 Kim+2010 Gladstone+1996 Romani+2008 ISO Moses+2005 Gladstone+1996 Fouchet+2000 Moses+2005 Yelle+2001 Kim+2010 Gladstone+1996 Fouchet+2000 Moses+2005 CH4 10-2 Fig. 12.— Mixing ratios for various chemical species as a function of pressure, p [bar]. A comparison between our model (solid lines) and that of various observations, for complex hydrocarbons in the stratosphere of Jupiter. 100 10-2 10-4 10-6 10-8 o i t a r g n i x i m n o b r a c 10-10 101 ] 1 − Å 1 − s 2 − m c s n o t o h p [ ) 102 height [km] 103 ( 1014 1013 1012 1011 1010 109 108 107 106 0 2000 4000 6000 8000 10000 [Å] Fig. 6.— The carbon mixing ratio as a function of atmospheric height [km]. We test for diffusion, with chemistry turned off, for carbon atoms and hydrogen atoms in a gas at hydrostatic equilibrium for an isothermal gas (g = 103 cm s−2, T = 300 K). The solid line is the result from Argo and the dashed line is the analytic result (Eq. (28)). Fig. 7.— The solar flux used in our model [pho- tons cm−2 s−1 A−1], as a function of wavelength, λ [A], taken from Huebner & Carpenter (1979); Huebner et al. (1992); Huebner & Mukherjee (2015). Weighted versions of this flux are used for HD209458b and Jupiter. This flux is used, unadjusted, for the early Earth. 51 F l l 10-10 10-8 10-6 10-4 10-2 100 102 ] r a b [ p 10-10 10-8 10-6 10-4 10-2 100 ) r a b ( p 2000 4000 6000 8000 10000 12000 T [K] 100 200 300 400 500 600 700 800 900 1000 T (K) Fig. 8.— Temperature profile for HD209458b, T [K] as a function of p [bar], as used by Moses et al. (2011). Fig. 11.— Temperature profile for Jupiter, T [K] as a function of p [bar] (Moses et al. 2005). 10-10 10-8 10-6 10-4 10-2 100 102 ] r a b [ p Ion Chemistry e- H+ H- K+ Na+ He+ Mg+ Fe+ 10-14 10-12 10-10 10-8 10-6 10-4 10-2 100 mixing ratio Fig. 10.— Mixing ratios for the dominant ionic species as a function of pressure, p [bar] for the atmosphere of HD209458b. 52 300 250 200 150 100 50 ) m k ( h 0 100 200 300 400 500 600 700 800 900 1000 T (K) Fig. 13.— Temperature profile used for the Early Earth chemistry, temperature [K] vs. height [km]. This profile is a synthetic profile for the Earth’s atmosphere generated with the MSIS-E-90 model for the date 2000/1/1 (Hedin 1987, 1991). 70 60 50 40 30 20 10 ) m k ( h N2O ) m k ( h H2O OH (x103) 80 70 60 50 40 30 20 10 0 10-9 10-8 10-7 10-6 10-5 10-4 mixing ratio Fig. 16.— Mixing ratios of OH and H2O as a func- tion of atmospheric height [km] for the atmosphere of the present-day Earth. The lines are produced by our model and the points are taken from bal- loon measurements at various latitudes, heights, and times (Kovalenko et al. 2007). o i t a r g n i x i m 100 10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8 NO NO2 10-6 10-4 10-2 time (s) 100 102 104 O3 CH4 0 10-10 10-9 10-8 10-7 10-6 10-5 10-4 Mixing Ratio Fig. 14.— Mixing ratios of ozone, methane and nitrous oxide as a function of atmospheric height [km] for the atmosphere of the present-day Earth. The lines are produced by our model and the points are taken from globally averaged measure- ments (Massie & Hunten 1981). Errors are set to an order of magnitude to account for diurnal and latitudinal variations. ) m k ( h 80 70 60 50 40 30 20 10 0 10-12 NO NO2 10-11 10-10 10-9 Mixing Ratio 10-8 10-7 Fig. 15.— Mixing ratios of NO and NO2 as a func- tion of atmospheric height [km]for the atmosphere of the present-day Earth. The lines are produced by our model and the points are taken from bal- loon measurements (Sen et al. 1998). Errors are set to an order of magnitude to account for diurnal and latitudinal variations. We also show the re- sults from suppressing the rate constant for Reac- tion 1300 in the network by a factor of 2 (dashed) and a factor of 10 (dotted). Fig. 17.— Mixing ratios of NO (solid) and NO2 (dashed) vs time [s] in a simulation of a lightning shock on a parcel of gas with an Earth-like atmo- spheric composition, initially at 300 K and 1 bar. The temperature and pressure vary as a function of time as described by Orville (1968), until 10−4 s, at which time conditions are returned to 300 K and 1 bar, and the system is allowed to further evolve. 53 ) m k ( h 80 70 60 50 40 30 20 10 0 10-6 O O2 H2 CO2 H 2O CO 10-5 10-4 10-3 10-2 10-1 100 mixing ratio Fig. 18.— Mixing ratios for O, H2, CO, O2, H2O and CO2, as a function of height [km], for early Earth photochemistry. These results can be com- pared to the results of Kasting (1993, his Fig. 1). 10-5 10-10 10-15 10-20 10-25 o i t a r g n i x i m B - D s M o d e l M o d e l E A M o d e l 10-30 100 101 102 104 105 103 t [s] Fig. 19.— Mixing ratio of glycine as a function of time, for five lab simulations, labelled Models A- E, with parameters given in Table 2 and described in Section 5. 54 Network Formula Standard Formula Name Thermochem Data Table 4—Continued 1CH2 C2H H2N HN2 H2O HO2 C2N CNC CN2 C2O CO2 N2O NO2 HCN HNC HNO NCO NaOH MgHO NaCl SiH2 KOH FeO2 CH3 C2H2 H3N H2N2 HNNH H2O2 NCCN H2CN HCCN CHN2 CH2O HOCH HCCO COOH NH2O HNO2 OCCN HCNO CH2(a1A1) CCH NH2 N2H H2O HO2 CCN CNC CNN C2O CO2 N2O NO2 HCN HNC HNO NCO NaOH MgOH NaCl SiH2 KOH FeO2 CH3 C2H2 NH3 N2H2 HNNH H2O2 (CN)2 H2CN HCCN HCNN H2CO HCOH HCCO COOH NHOH HNO2 NCCO HCNO Singlet methylene Ethynyl radical Amidogen Amino radical Water Hydroperoxyl Cyano-methylidyne CNC radical CNN radical Dicarbon monoxide Carbon dioxide Nitrous oxide Nitrogen dioxide Hydrogen cyanide Hydrogen isocyanide Nitroxyl Isocyanato radical Sodium hydroxide Magnesium monohydroxide Sodium chloride Silylene Potassium hydroxide Iron oxide Methyl radical Acetylene Ammonia Diimide (Z)-Diazene Hydrogen peroxide Cyanogen Dihydrogen cyanide HCCN radical HCNN Formaldehyde Hydroxymethylene Ethynyloxy radical Hydrocarboxyl radical NHOH Nitrous acid NCCO Fulminic acid 55 Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat NASA-CEA Burcat NASA-CEA Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Network Formula Standard Formula Name Thermochem Data Table 4—Continued HNCO CHNO HCNO SiH3 CH4 C2H3 NH2NH N2O3 CH3N CH2CN CH2N2 HC3N CH2OH CH3O C2H2O HCOOH CH2O2 HNO3 NH2OH HCOCN MgO2H2 SiH4 C2H4 C3H3 H4N2 H4O2 CH3CN C2H3N CH3N2H CH3OH CH3O2 C2H3O cyc-C2H3O CH2CHO C3H2O C2H2O2 H2NNO2 CH3NO C2H5 C3H4 CH5N Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Benson NASA-CEA Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Benson Burcat Burcat Burcat Burcat Burcat Burcat HNCO CHNO HCNO SiH3 CH4 C2H3 NH2NH N2O3 CH2NH CH2CN CH2N2 · · · CH2OH CH3O H2CCO HCOOH CH2OO HNO3 NH2OH HCOCN H2MgO2 SiH4 C2H4 · · · N2H4 H2O·H2O CH3CN CH2CNH CH3N2H CH3OH CH3O2 CH3CO Oxyranyl CH2CHO · · · (CHO)2 H2N–NO2 HCONH2 C2H5 · · · CH3NH2 Isocyanic acid Cyanic acid HCNO Silyl radical Methane Vinyl radical Hydrazinyl radical Nitrogen trioxide Methanimine Cyanomethyl radical Diazomethane Propiolonitrile∗ Hydroxymethyl radical Methoxy radical Ethenone Formic acid CH2OO Nitric acid Hydroxylamine HCOCN Magnesium hydroxide Silane Ethylene Propargyl radical∗ Hydrazine Water dimer Acetonitrile CH2=C=NH Methyl diazene Methanol CH3O2 Acetyl radical Oxiranyl radical CH2CHO 2-Propynal∗ Glyoxal H2N-NO2 Formamide Ethyl radical Propyne∗ Methylamine 56 Network Formula Standard Formula Name Thermochem Data Table 4—Continued C3H3N C2H4O H2C2HOH Oxirane C3H3O CH4O2 CH3OCO CH3NO2 CH3ONO C2H6 C4H4 CH2NCH3 CH3COOH CH3OCHO CH3CHOH CH3CH2O CH3OCH2 CH3NO3 C2H3NO2 Si2H6 C3H6 C2H6N C2H5OH C2H5OO C2H4O3 (CH3)2O C2H5NO C4H6 (CH3N)2 (CH3O)2 (CH2OH)2 (CH3)2CO C2H6O2 C2H5NO2 aC2H5NO2 bC2H5NO2 C3H8 (CH3)2N2O C4H10 · · · CH3CHO CH2CHOH Oxirane · · · CH3OOH CH3OCO CH3NO2 CH3ONO C2H6 · · · NH2NCH3 CH3COOH CHOOCH3 CH3CHOH CH3CH2O CH3OCH2 CH3NO3 C2H3NO2 Si2H6 · · · (CH3)2N CH3CH2OH C2H5OO HOCH2COOH (CH3)2O C2H5NO · · · (CH3N)2 (CH3O)2 (CH2OH)2 (CH3)2CO HOCH2CH2OH NH2CH2COOH C2H5NO2 C2H5ONO (CH3)2N2O · · · · · · Acrylonitrile∗ Acetaldehyde Vinyl alcohol Oxirane 1-Oxoprop-2-3nyl∗ Methyl peroxide CH3OC(·)(O) Nitromethane Methyl nitrite Ethane 1-Buten-3-yne∗ N-Methyl methanimine Acetic acid Methyl formate 1-hydroxy Ethyl radical Ethoxy radical Methoxymethyl radical Methyl nitrate Nitroethylene Disilane Propene∗ Dimethyl amidogen Ethanol C2H5OO Glycolic acid Dimethyl ether Acetaldoxime 1,3-Butadiene∗ Dimethyl diazene Dimethyl peroxide 1,2-Ethanediol Acetone Ethylene glycol Glycine Nitroethane Ethyl nitrate Propane∗ Dimethylnitrosamine Butane∗ 57 Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Benson Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Benson Burcat Burcat Burcat Burcat Burcat Burcat Burcat Burcat Benson Burcat Benson Burcat 58 e− O+ C− 2 Fe+ Mg+ O+ 2 ArH+ C3O+ CHN+ CO+ 4 H3O+ HO+ 4 NO− 2 ArH+ 3 C2H+ 6 C3H+ 3 C3HN+ C4H+ 7 CH3O+ CH6N+ H3O+ 2 SiH+ 2 C2H2N− C2H5N+ C3H4O+ C4H2N+ CH2OH+ Si2H+ 5 Si4H+ 2 SiCH+ 3 C2H4NO− SiC2H+ 3 O+ C+ O+(2D) C+ 3 H+ 2 Na+ 2 (X2Πg) C2H+ C4H+ CHO+ CO− 4 H4N+ HSi+ NO− 3 C2H+ 2 C2H+ 7 C3H+ 4 C3N+ 2 C4H+ 8 CH3O− CHNO+ HN2O+ SiH+ 3 C2H2O+ C2H5O+ C3H5O+ CH2NO+ CH3O+ 2 Si3H+ 2 Si4H+ 3 SiCH+ 4 C2H5O+ 2 SiC2H+ 4 C+ 4 H+ 3 N+ 2 O− 2 C2H− C5N+ CN+ 2 FeO+ HN+ 2 He+ 2 Si+ 2 C2H− 2 C2HN+ C3H+ 5 C3N+ 3 C4H+ 9 CH4N+ CHO+ 2 HNO+ 2 SiH− 3 H− Ar+ CN+ HO+ N− 3 Si+ C2O+ CH+ 3 CNO− H2N− HNO+ MgO+ SiO+ C2H− 3 C2HO− C3H+ 7 C4H+ 3 CH2N+ CH4O+ FeO+ 2 MgO+ 2 SiH+ 5 H+ O− CH+ HN+ N+ 3 O− 3 C2N+ CH+ 2 CNO+ H2N+ HN+ 3 HeH+ SiH+ C2H+ 3 C2HO+ C3H+ 6 C4H+ 2 C5HN+ CH4N− CHO− 2 MgHO+ K+ Cl+ CN− HO− NO+ Ti+ C3H+ CH+ 4 CO+ 2 H2O+ HO+ 2 N2O+ TiO+ C2H+ 4 C2N+ 2 C3H+ 8 C4H+ 4 CH2O+ CH5N+ H2NO+ Si2H+ SiHO+ C2H4N+ C2H3N+ C2HN+ C2H5O− 2 C3H9O+ C3H6N+ CH3NO+ CH4NO+ CH5NO+ CH6NO+ H2NO+ 2 Si3H+ 3 Si4H+ 4 SiCH+ 5 C4H7O+ 2 SiC2H+ 5 H2NO+ 3 Si3H+ 4 Si4H+ 5 SiC2H+ CH2NO− 2 SiC2H+ 6 Si2H+ 2 Si3H+ 5 Si4H+ 6 SiH3O+ CH3NO+ 2 TiC2H+ 4 Si2H+ 3 Si3H+ 6 Si4H+ 7 SiH3O− CH4NO+ SiH+ 4 C2H3O+ C2H6O+ C3H6O+ N+ C+ 2 CO+ He+ NO− Ar+ 2 C3N+ CH+ 5 CO− 3 H3N+ HO− 2 NO+ 2 Ar2H+ C2H+ 5 C3H+ 2 C3H+ 9 C4H+ 5 CH3N+ CH5O+ H2O+ 2 SiCH+ C2H2N+ C2H4O+ C3H2N+ C3HN+ 2 CH2O+ 2 Si2H+ 4 Si3H+ 7 SiCH+ 2 aCHNO+ 2 MgH2O+ 2 Table 5: List of ions included the Stand2015 Net- work C− O+(3P) C2H3O− C2H7O+ C3H7O+ 59 Table 6 Stand2015 Chemical Kinetics Network # Type Reaction α β γ Ref 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2.49e-08 C2 ⇌ C + C 6.01e+11 C2 ⇌ C + C 3.16e-10 CH ⇌ C + H 7.63e+09 CH ⇌ C + H 1.00e-09 CN ⇌ C + N 2.42e+10 CN ⇌ C + N 1.52e-04 CO ⇌ C + O 3.67e+15 CO ⇌ C + O 9.13e-33 H + H ⇌ H2 1.00e-11 H + H ⇌ H2 2.99e-10 HN ⇌ H + N 7.22e+09 HN ⇌ H + N 4.33e-32 H + O ⇌ HO 1.05e-12 H + O ⇌ HO 8.86e-05 N2 ⇌ N + N 2.14e+15 N2 ⇌ N + N 4.98e-10 NO ⇌ N + O 1.20e+10 NO ⇌ N + O 5.65e-10 O2 ⇌ O + O 1.36e+10 O2 ⇌ O + O 6.00e-34 O2 + O ⇌ O3 2.81e-12 O2 + O ⇌ O3 5.00e-01 C2H ⇌ C2 + H 1.21e+18 C2H ⇌ C2 + H 5.00e-01 CNC ⇌ C2 + N 1.21e+18 CNC ⇌ C2 + N 5.00e-01 C2O ⇌ C2 + O 1.21e+18 C2O ⇌ C2 + O 9.33e-09 CH2 ⇌ CH + H 2.25e+11 CH2 ⇌ CH + H 7.00e-32 C + H2 ⇌ CH2 1.70e-11 C + H2 ⇌ CH2 1.45e-06 HCN ⇌ HNC HCN ⇌ HNC 3.50e+13 HCN ⇌ CN + H 6.14e-06 HCN ⇌ CN + H 2.55e+12 H + CO ⇌ CHO 1.40e-34 H + CO ⇌ CHO 1.96e-13 NCO ⇌ CO + N 3.65e-10 NCO ⇌ CO + N 8.82e+09 CO2 ⇌ CO + O 1.81e-10 0.00 -1.00 0.00 -1.00 0.00 0.00 -3.10 -4.10 -0.60 0.00 0.00 -1.00 -1.00 -2.00 -3.33 -4.33 0.00 -1.00 0.00 -1.00 -2.40 0.00 -5.16 -6.16 -5.16 -6.16 -5.16 -6.16 0.00 -1.00 0.00 0.00 1.00 0.00 -1.58 0.00 0.00 0.00 0.00 -1.00 0.00 71600 71600 33700 33700 71000 71000 129000 129000 0 0 37700 37700 0 0 113000 113000 76600 76600 55700 55700 0 0 57400 57400 57400 57400 57400 57400 45100 45100 0 57 23800 23800 61500 0 100 1370 27200 27200 49000 1 · · · 2 · · · 3 · · · 4 · · · 30 5 6 · · · 7 · · · 8 · · · 9 · · · 10 · · · 169 11 1 · · · Est · · · Est · · · 12 · · · 13 · · · 14 · · · Est · · · 15 · · · 16 · · · 17 60 Table 6—Continued # Type Reaction α β γ Ref 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 8.99e+12 CO2 ⇌ CO + O 1.99e-09 H2N ⇌ HN + H 4.81e+10 H2N ⇌ HN + H 6.78e-31 HO + H ⇌ H2O 2.09e-10 HO + H ⇌ H2O 5.48e-07 HNO ⇌ NO + H 1.00e+13 HNO ⇌ NO + H 2.41e-08 HO2 ⇌ O2 + H 5.82e+11 HO2 ⇌ O2 + H 9.51e-10 N2O ⇌ N2 + O 2.30e+10 N2O ⇌ N2 + O 9.00e-32 NO + O ⇌ NO2 3.00e-11 NO + O ⇌ NO2 9.00e-31 CH + H2 ⇌ CH3 2.01e-10 CH + H2 ⇌ CH3 5.63e-31 H + CH2 ⇌ CH3 2.01e-11 H + CH2 ⇌ CH3 2.96e-08 H3N ⇌ H2N + H 6.60e+17 H3N ⇌ H2N + H 1.05e-09 H3N ⇌ HN + H2 3.62e+10 H3N ⇌ HN + H2 2.51e-14 NO3 ⇌ NO + O2 6.06e+05 NO3 ⇌ NO + O2 8.90e-29 H + CH3 ⇌ CH4 3.20e-10 H + CH3 ⇌ CH4 1.15e-10 HN2 ⇌ N2 + H 6.35e+13 HN2 ⇌ N2 + H 5.00e-01 C2N ⇌ CNC 1.21e+18 C2N ⇌ CNC 3.95e-06 CNO ⇌ CO + N 9.54e+13 CNO ⇌ CO + N 2.63e-26 H + C2H ⇌ C2H2 H + C2H ⇌ C2H2 3.00e-10 HCCN ⇌ CNC + H 5.00e-01 HCCN ⇌ CNC + H 1.21e+18 NCCN ⇌ CN + CN 2.97e-07 NCCN ⇌ CN + CN 7.17e+12 OCCN ⇌ CN + CO 2.97e-07 OCCN ⇌ CN + CO 7.17e+12 CH2O ⇌ CHO + H 6.20e-09 CH2O ⇌ CHO + H 8.00e+15 0.00 0.00 -1.00 -2.00 0.00 -1.24 0.00 -1.18 -2.18 0.00 -1.00 -1.50 0.00 0.00 0.15 0.00 0.15 0.00 0.00 0.00 -1.00 0.00 -1.00 -1.80 0.13 -0.11 -0.53 -5.16 -6.16 -1.90 -2.90 -3.10 0.00 -5.16 -6.16 0.00 -1.00 0.00 -1.00 0.00 0.00 65300 38300 38300 0 0 25300 24200 24400 24400 29000 29000 0 0 550 0 0 0 46300 49000 47000 47000 1230 1230 318 25 2510 3400 57400 57400 30100 30100 740 0 57400 57400 53600 53600 53600 53600 36900 44100 · · · 6 · · · 5 · · · 18 · · · 7 · · · 19 · · · 169 · · · 21 20 21 · · · 22 · · · 23 · · · 24 · · · 31 · · · 25 · · · Est · · · 3 · · · 7 · · · Est · · · 26 · · · Est · · · 27 · · · 61 Table 6—Continued # Type Reaction α β γ Ref 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d CH2O ⇌ CO + H2 9.40e-09 CH2O ⇌ CO + H2 3.70e+13 CHN2 ⇌ CH + N2 2.97e-07 CHN2 ⇌ CH + N2 7.17e+12 HNCO ⇌ NCO + H 1.66e-07 HNCO ⇌ NCO + H 4.01e+12 HNCO ⇌ CO + HN 7.69e-04 HNCO ⇌ CO + HN 6.00e+13 HOCN ⇌ CN + HO 1.66e-09 HOCN ⇌ CN + HO 4.01e+10 HCNO ⇌ CNO + H 1.66e-07 HCNO ⇌ CNO + H 4.01e+12 HCNO ⇌ CN + HO 1.66e-09 HCNO ⇌ CN + HO 4.01e+10 HONC ⇌ CN + HO 1.66e-09 HONC ⇌ CN + HO 4.01e+10 H2N2 ⇌ H2 + N2 1.00e-09 2.42e+10 H2N2 ⇌ H2 + N2 3.01e-08 H2O2 ⇌ HO + HO 3.00e+14 H2O2 ⇌ HO + HO HNO2 ⇌ HO + NO 2.23e-03 1.09e+16 HNO2 ⇌ HO + NO 1.00e-09 N2O2 ⇌ NO + NO 2.42e+10 N2O2 ⇌ NO + NO 4.87e-30 H + C2H2 ⇌ C2H3 H + C2H2 ⇌ C2H3 1.06e-11 5.00e-01 CH3N ⇌ HCN + H2 1.21e+18 CH3N ⇌ HCN + H2 CH3O ⇌ CH2O + H 6.59e-05 CH3O ⇌ CH2O + H 2.87e+13 1.14e-06 NO2 + HO ⇌ HNO3 9.33e+15 NO2 + HO ⇌ HNO3 8.00e-02 HO2 + NO ⇌ HNO3 HO2 + NO ⇌ HNO3 5.56e+17 N2O3 ⇌ NO2 + NO 1.91e-07 N2O3 ⇌ NO2 + NO 4.70e+15 1.49e-29 H + C2H3 ⇌ C2H4 H + C2H3 ⇌ C2H4 2.30e-10 5.80e-08 C2H4 ⇌ C2H2 + H2 9.75e+13 C2H4 ⇌ C2H2 + H2 H + C4H ⇌ C4H2 2.64e-26 0.00 0.00 0.00 -1.00 0.00 0.00 -3.10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -1.00 0.00 0.00 -3.80 -1.23 0.00 -1.00 -1.07 1.64 -5.16 -6.16 -2.70 1.31 0.00 0.00 -6.55 -2.27 -8.70 0.40 -1.00 0.20 0.00 0.44 -3.10 33200 36200 53600 53600 56400 56400 51400 50200 0 0 56400 56400 0 0 0 0 10000 10000 21700 24400 25200 25000 1500 1500 83 1060 57400 57400 15400 15800 23100 24700 26100 26300 4880 4880 0 0 36000 44700 72 27 · · · Est · · · 28 · · · 3 · · · 29 · · · 28 · · · 29 · · · 29 · · · Est · · · Est · · · 18 · · · Est · · · 31 30 Est · · · Est · · · 169 · · · 169 · · · 32 · · · 21 · · · 33 · · · 21 62 Table 6—Continued # Type Reaction α β 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 3.00e-10 H + C4H ⇌ C4H2 5.56e-28 C2H + C2H ⇌ C4H2 2.60e-10 C2H + C2H ⇌ C4H2 5.53e-25 H + C4H2 ⇌ C4H3 2.66e-12 H + C4H2 ⇌ C4H3 5.41e-23 H + C4H3 ⇌ C4H4 1.39e-10 H + C4H3 ⇌ C4H4 6.61e-09 H4N2 ⇌ H2N + H2N 7.94e+13 H4N2 ⇌ H2N + H2N 1.00e-09 H4O2 ⇌ H2O + H2O 2.42e+10 H4O2 ⇌ H2O + H2O 9.23e-29 H + C2H4 ⇌ C2H5 1.08e-12 H + C2H4 ⇌ C2H5 5.25e-11 CH5N ⇌ CH3 + H2N 6.92e+10 CH5N ⇌ CH3 + H2N 2.66e-11 CH5N ⇌ CH4 + HN 6.43e+08 CH5N ⇌ CH4 + HN 3.05e-26 CH3 + CH3 ⇌ C2H6 6.78e-11 CH3 + CH3 ⇌ C2H6 2.00e-28 H + C2H5 ⇌ C2H6 1.70e-10 H + C2H5 ⇌ C2H6 3.80e-07 C2H6 ⇌ C2H4 + H2 9.18e+12 C2H6 ⇌ C2H4 + H2 2.19e-10 C3H4 ⇌ C2H2 + CH2 5.30e+09 C3H4 ⇌ C2H2 + CH2 2.40e-13 C3H4 ⇌ C2H4 + C C3H4 ⇌ C2H4 + C 5.79e+06 C3H4 ⇌ CH4 + C + C 2.82e-11 C3H4 ⇌ CH4 + C + C 6.81e-08 C3H4 ⇌ C3H3 + H 7.80e-06 1.88e+14 C3H4 ⇌ C3H3 + H 1.00e-14 C3H3 ⇌ C2H2 + CH 2.00e+05 C3H3 ⇌ C2H2 + CH C3H5 ⇌ C3H4 + H 4.14e-05 1.00e+15 C3H5 ⇌ C3H4 + H 5.22e-07 C3H5 ⇌ C2H2 + CH3 1.26e+13 C3H5 ⇌ C2H2 + CH3 HOCH ⇌ CO + H2 9.40e-09 3.70e+13 HOCH ⇌ CO + H2 6.20e-09 HOCH ⇌ CHO + H HOCH ⇌ CHO + H 8.00e+15 0.00 -3.00 0.00 -2.93 5.55 -3.97 0.00 0.00 0.00 0.00 -1.00 -1.51 5.31 0.00 0.00 0.00 -1.00 -3.77 -0.36 -1.50 0.00 0.00 -1.00 1.00 0.00 1.00 0.00 1.00 0.00 0.00 -1.00 0.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 γ 0 300 0 176 0 177 0 20600 27700 1500 1500 73 0 17700 24200 24300 24300 62 30 0 0 34000 34000 27900 27900 20100 20100 23500 23500 40300 40300 0 0 32000 32000 16800 16800 33200 36200 36900 44100 Ref · · · 31 · · · 31 · · · 31 · · · 34 · · · Est · · · 31 · · · 35 · · · 36 · · · 31 · · · 31 · · · 37 · · · Est · · · Est · · · Est · · · Est · · · Est · · · 38 · · · 39 · · · 27 · · · 27 · · · 63 Table 6—Continued # Type Reaction α β γ Ref 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 1.00e-08 HC3N ⇌ HCN + C2 2.42e+11 HC3N ⇌ HCN + C2 4.14e-07 NH2O ⇌ H2 + NO 1.00e+13 NH2O ⇌ H2 + NO 1.08e-08 HCCO ⇌ CO + CH 2.61e+11 HCCO ⇌ CO + CH 5.34e-05 C4H3 ⇌ C2H2 + C2H 1.29e+15 C4H3 ⇌ C2H2 + C2H 5.34e-05 C4H4 ⇌ C2H2 + C2H2 1.29e+15 C4H4 ⇌ C2H2 + C2H2 2.90e-05 C4H6 ⇌ C4H4 + H2 7.00e+14 C4H6 ⇌ C4H4 + H2 4.14e-08 C4H6 ⇌ C3H3 + CH3 1.00e+12 C4H6 ⇌ C3H3 + CH3 1.27e-07 C4H6 ⇌ C2H3 + C2H3 3.07e+12 C4H6 ⇌ C2H3 + C2H3 2.90e-07 C4H6 ⇌ C2H4 + C2H2 7.00e+12 C4H6 ⇌ C2H4 + C2H2 1.00e-08 HNNH ⇌ N2 + H2 2.42e+11 HNNH ⇌ N2 + H2 1.00e-08 HNNH ⇌ HN + HN 2.42e+11 HNNH ⇌ HN + HN 1.00e-08 H2CN ⇌ HCN + H 2.42e+11 H2CN ⇌ HCN + H 5.00e-01 CH2CN ⇌ CH2 + CN 1.21e+18 CH2CN ⇌ CH2 + CN 5.98e-09 C2H2O ⇌ CH2 + CO C2H2O ⇌ CH2 + CO 3.00e+14 HCOCN ⇌ HNC + CO 9.23e-06 HCOCN ⇌ HNC + CO 2.23e+14 HCOCN ⇌ HCN + CO 1.00e-05 HCOCN ⇌ HCN + CO 2.42e+14 1.66e-10 CH2OH ⇌ CH2O + H CH2OH ⇌ CH2O + H 5.60e+10 4.97e-08 CH2N2 ⇌ CH2 + N2 1.20e+12 CH2N2 ⇌ CH2 + N2 2.81e-09 CH2O2 ⇌ CO2 + H2 CH2O2 ⇌ CO2 + H2 4.46e+13 2.81e-09 HCOOH ⇌ CO2 + H2 4.46e+13 HCOOH ⇌ CO2 + H2 HCOOH ⇌ H2O + CO 6.73e-09 0.00 -1.00 1.00 0.00 0.00 -1.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 -5.16 -6.16 0.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 81900 81900 6830 6830 29600 29600 41500 41500 41500 41500 47800 47800 30000 30000 33600 33600 33800 33800 10000 10000 70000 70000 27000 27000 57400 57400 29000 35700 32100 32100 33100 33100 12600 14600 17100 17100 25700 34400 25700 34400 26700 Est · · · 40 · · · 41 · · · Est · · · 42 · · · Est · · · 43 · · · 43 · · · 43 · · · Est · · · Est · · · 44 · · · Est · · · 45 · · · 46 · · · 46 · · · 47 · · · 48 · · · Est · · · 49 · · · 49 64 Table 6—Continued # Type Reaction α β γ Ref 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 7.49e+14 HCOOH ⇌ H2O + CO 1.20e-12 HCOOH ⇌ CH2O2 2.90e+07 HCOOH ⇌ CH2O2 4.17e-11 CH3CN ⇌ C2H3N 1.01e+09 CH3CN ⇌ C2H3N CH3CN ⇌ CH2CN + H 4.00e-04 CH3CN ⇌ CH2CN + H 1.00e+16 4.00e-04 C2H3N ⇌ CH2CN + H C2H3N ⇌ CH2CN + H 1.00e+16 1.24e-04 C2H3O ⇌ CH2CHO 3.00e+15 C2H3O ⇌ CH2CHO 5.25e-10 C2H3O ⇌ CH3 + CO C2H3O ⇌ CH3 + CO 2.00e+10 5.00e-30 C2H2 + HO ⇌ C2H3O 9.00e-13 C2H2 + HO ⇌ C2H3O 3.61e-07 C2H3O ⇌ C2H2O + H C2H3O ⇌ C2H2O + H 8.74e+12 1.16e-08 CH3OH ⇌ H2O + CH2 CH3OH ⇌ H2O + CH2 2.80e+11 CH3OH ⇌ CH2OH + H 2.16e-08 CH3OH ⇌ CH2OH + H 2.30e+10 6.40e-29 HO + CH3 ⇌ CH3OH 6.33e-11 HO + CH3 ⇌ CH3OH 4.04e-07 CH3OH ⇌ CH3O + H 9.75e+12 CH3OH ⇌ CH3O + H CH3OH ⇌ HOCH + H2 8.78e-07 2.12e+13 CH3OH ⇌ HOCH + H2 2.48e-05 C2H4O ⇌ CH3 + CHO 6.00e+14 C2H4O ⇌ CH3 + CHO C2H4O ⇌ C2H2O + H2 1.24e-05 3.00e+14 C2H4O ⇌ C2H2O + H2 4.14e-05 C2H4O ⇌ CH4 + CO 1.00e+15 C2H4O ⇌ CH4 + CO CH4O2 ⇌ CH3O + HO 2.48e-05 CH4O2 ⇌ CH3O + HO 6.00e+14 5.25e-11 C2H6N ⇌ CH3N + CH3 6.92e+10 C2H6N ⇌ CH3N + CH3 NH2OH ⇌ H2N + HO 9.30e-11 2.25e+09 NH2OH ⇌ H2N + HO 8.28e-07 C3H3N ⇌ HC3N + H2 C3H3N ⇌ HC3N + H2 2.00e+13 0.00 0.00 -1.00 0.00 -1.00 1.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 -1.50 2.00 2.94 1.94 0.00 -1.00 0.00 5.04 0.00 0.10 3.39 2.39 2.22 1.22 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 0.20 -0.80 1.00 0.00 34500 0 0 19100 19100 48400 48400 48400 48400 14200 14200 6040 7550 0 0 23100 23100 33400 33400 33600 42500 1030 0 50200 50200 43500 43500 39800 39800 42200 42200 42800 42800 21300 21300 17700 24200 45800 45800 38700 38700 · · · 49 · · · 50 · · · 51 · · · Est · · · 52 · · · 53 · · · 54 · · · 55 · · · 56 · · · 57 · · · 21 · · · 58 · · · 58 · · · 59 · · · 59 · · · 59 · · · 33 · · · Est · · · 60 · · · 61 · · · 65 Table 6—Continued # Type Reaction α β γ Ref 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 7.37e-08 C3H3N ⇌ C2H2 + HCN 1.78e+12 C3H3N ⇌ C2H2 + HCN 5.22e-06 C3H3O ⇌ C2H3 + CO 1.26e+14 C3H3O ⇌ C2H3 + CO 5.25e-11 CH3NO ⇌ HCN + H2O 1.27e+09 CH3NO ⇌ HCN + H2O 7.84e-06 C4H10 ⇌ C2H5 + C2H5 2.72e+15 C4H10 ⇌ C2H5 + C2H5 3.52e-05 C3H2O ⇌ C2H2 + CO 8.51e+14 C3H2O ⇌ C2H2 + CO 2.03e+00 CH3O2 ⇌ CH3 + O2 4.90e+19 CH3O2 ⇌ CH3 + O2 1.00e-12 CH2CHO ⇌ C2H2O + H 2.42e+07 CH2CHO ⇌ C2H2O + H 1.00e-09 CH2CHO ⇌ CH3 + CO 2.42e+10 CH2CHO ⇌ CH3 + CO 1.06e+04 C2H2O2 ⇌ CHO + CHO 2.57e+23 C2H2O2 ⇌ CHO + CHO 1.00e-09 H2NNO2 ⇌ H2N + NO2 2.42e+10 H2NNO2 ⇌ H2N + NO2 2.38e-06 Oxyrane ⇌ CH2CHO + H 5.75e+13 Oxyrane ⇌ CH2CHO + H 4.18e-06 Oxyrane ⇌ C2H3O + H 1.01e+14 Oxyrane ⇌ C2H3O + H 2.25e-05 Oxyrane ⇌ CH3 + CHO 5.44e+14 Oxyrane ⇌ CH3 + CHO 4.72e-08 Oxyrane ⇌ C2H2O + H2 1.14e+12 Oxyrane ⇌ C2H2O + H2 1.83e-09 Oxyrane ⇌ C2H4O 4.43e+10 Oxyrane ⇌ C2H4O 4.43e-07 Oxyrane ⇌ C2H2 + H2O 1.07e+13 Oxyrane ⇌ C2H2 + H2O 8.28e-07 Oxyrane ⇌ CH4 + CO 2.00e+13 Oxyrane ⇌ CH4 + CO 2.66e-11 CH3N2H ⇌ HCN + H3N 2.66e-11 CH3N2H ⇌ HCN + H3N 3.32e-07 CH3NO2 ⇌ CH3 + NO2 1.58e+16 CH3NO2 ⇌ CH3 + NO2 CH3ONO ⇌ CH2O + HNO 1.15e-06 CH3ONO ⇌ CH2O + HNO 2.77e+13 CH3ONO ⇌ CH3O + NO 1.32e-06 1.00 0.00 1.63 0.63 0.00 -1.00 0.00 0.00 1.00 0.00 -10.00 -11.00 0.00 -1.00 0.00 -1.00 -9.90 -10.90 0.00 -1.00 1.20 0.20 1.25 0.25 1.40 0.40 0.80 -0.20 0.25 -0.75 1.06 0.06 1.11 0.11 0.00 -1.00 0.00 0.00 1.00 0.00 0.00 34200 34200 13700 13700 25300 25300 24900 38000 35700 35700 16700 16700 20700 20700 20700 20700 41300 41300 5000 5000 36100 36100 32800 32800 31200 31200 31800 31800 23300 23300 35000 35000 32100 32100 24300 24300 21100 30000 19500 19500 17200 61 · · · 55 · · · 62 · · · 63 · · · 64 · · · 7 · · · 65 · · · 66 · · · 67 · · · Est · · · 68 · · · 68 · · · 68 · · · 68 · · · 68 · · · 68 · · · 68 · · · 69 · · · 70 · · · 71 · · · 71 66 Table 6—Continued # Type Reaction α β γ Ref 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 1.06e+15 CH3ONO ⇌ CH3O + NO 4.14e-07 CH3NO3 ⇌ CH3O + NO2 1.00e+13 CH3NO3 ⇌ CH3O + NO2 5.42e+14 C2H5OH ⇌ CH2OH + CH3 2.99e+18 C2H5OH ⇌ CH2OH + CH3 1.14e+16 C2H5OH ⇌ C2H5 + HO 1.37e+17 C2H5OH ⇌ C2H5 + HO 2.05e+12 C2H5OH ⇌ C2H4 + H2O 1.41e+13 C2H5OH ⇌ C2H4 + H2O 1.25e-08 (CH3)2O ⇌ CH3O + CH3 5.71e+17 (CH3)2O ⇌ CH3O + CH3 4.17e-11 (CH3)2O ⇌ CH2O + CH4 1.53e+13 (CH3)2O ⇌ CH2O + CH4 2.07e-05 C2H5NO ⇌ C2H2O + H3N 5.01e+14 C2H5NO ⇌ C2H2O + H3N 7.37e-05 C2H5OO ⇌ CH3CH2O + O 1.78e+15 C2H5OO ⇌ CH3CH2O + O 1.40e+02 C2H5OO ⇌ C2H5 + O2 5.30e+15 C2H5OO ⇌ C2H5 + O2 1.34e-07 C2H5OO ⇌ C2H4O + HO 3.24e+12 C2H5OO ⇌ C2H4O + HO 3.59e-05 C2H5OO ⇌ C2H4 + HO2 C2H5OO ⇌ C2H4 + HO2 8.66e+14 C2H4O3 ⇌ CH2O + H2O + CO 4.43e-06 C2H4O3 ⇌ CH2O + H2O + CO 1.07e+14 (CH3N)2 ⇌ CH3 + CH3 + N2 8.28e-09 2.00e+11 (CH3N)2 ⇌ CH3 + CH3 + N2 1.04e-03 (CH3O)2 ⇌ CH3O + CH3O 2.51e+16 (CH3O)2 ⇌ CH3O + CH3O cyc-C2H3O ⇌ CH2CHO 3.06e-05 7.40e+14 cyc-C2H3O ⇌ CH2CHO 3.03e-07 cyc-C2H3O ⇌ CH3 + CO 7.31e+12 cyc-C2H3O ⇌ CH3 + CO cyc-C2H3O ⇌ C2H2O + H 2.05e-06 4.96e+13 cyc-C2H3O ⇌ C2H2O + H 1.35e-06 CH3OCO ⇌ CH3O + CO 3.27e+13 CH3OCO ⇌ CH3O + CO CH3OCO ⇌ CH3 + CO2 1.14e-06 2.76e+13 CH3OCO ⇌ CH3 + CO2 1.66e-04 C2H6O2 ⇌ CH3CH2O + HO C2H6O2 ⇌ CH3CH2O + HO 4.00e+15 0.00 1.00 0.00 -18.90 -2.16 -19.70 -2.16 -17.90 1.36 0.00 -1.57 0.00 0.00 1.00 0.00 0.91 -0.09 -9.85 -0.83 2.37 1.37 -5.88 -6.88 1.00 0.00 1.00 0.00 1.00 0.00 -5.90 -6.90 1.00 0.00 1.00 0.00 1.65 0.65 2.11 1.11 1.00 0.00 18000 16800 16800 57700 44300 57600 48600 42700 33100 21500 42200 20500 29400 36900 36900 31000 31000 19600 17200 20900 20900 17100 17100 25300 25300 16800 16800 19400 19400 7540 7540 7190 7190 7480 7480 10600 10600 7990 7990 21700 21700 · · · 72 · · · 73 · · · 73 · · · 73 · · · 74 · · · 75 76 77 · · · 78 · · · 78 · · · 78 · · · 79 · · · 80 · · · 81 · · · 82 · · · 68 · · · Est · · · 68 · · · 55 · · · 55 · · · 33 · · · 67 Table 6—Continued # Type Reaction α β γ Ref 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d H2C2HOH ⇌ CH3 + CHO H2C2HOH ⇌ CH3 + CHO CH3COOH ⇌ C2H2O + H2O CH3COOH ⇌ C2H2O + H2O CH3COOH ⇌ CH4 + CO2 CH3COOH ⇌ CH4 + CO2 C2H3NO2 ⇌ CH2CHO + NO C2H3NO2 ⇌ CH2CHO + NO (CH2OH)2 ⇌ CH3CH2O + HO (CH2OH)2 ⇌ CH3CH2O + HO C2H5NO2 ⇌ C2H4 + HNO2 C2H5NO2 ⇌ C2H4 + HNO2 C2H5NO2 ⇌ C2H5 + NO2 C2H5NO2 ⇌ C2H5 + NO2 C2H5NO2 ⇌ CH3NO + CH2O C2H5NO2 ⇌ CH3NO + CH2O aCO2H5N ⇌ C2H4 + HNO2 aCO2H5N ⇌ C2H4 + HNO2 aCO2H5N ⇌ C2H5 + NO2 aCO2H5N ⇌ C2H5 + NO2 bCO2H5N ⇌ CH3CH2O + NO bCO2H5N ⇌ CH3CH2O + NO bCO2H5N ⇌ C2H4O + HNO bCO2H5N ⇌ C2H4O + HNO CH3OCH2 ⇌ CH2O + CH3 CH3OCH2 ⇌ CH2O + CH3 CH3CHOH ⇌ C2H4O + H CH3CHOH ⇌ C2H4O + H CH2NCH3 ⇌ CH3CN + H2 CH2NCH3 ⇌ CH3CN + H2 (CH3)2CO ⇌ CH2O + C2H4 (CH3)2CO ⇌ CH2O + C2H4 (CH3)2CO ⇌ cyc-C2H3O + CH3 (CH3)2CO ⇌ cyc-C2H3O + CH3 (CH3)2CO ⇌ C2H5 + CHO (CH3)2CO ⇌ C2H5 + CHO (CH3)2CO ⇌ C2H3O + CH3 (CH3)2CO ⇌ C2H3O + CH3 CH3CH2O ⇌ CH3CHOH CH3CH2O ⇌ CH3CHOH CH3CH2O ⇌ C2H4O + H 2.48e-05 6.00e+14 5.50e-12 2.82e+12 2.93e-06 7.08e+13 1.32e-06 1.60e+15 1.66e-04 4.00e+15 2.32e-08 5.60e+11 1.66e-06 7.94e+15 2.42e-11 5.85e+08 2.32e-08 5.60e+11 1.66e-06 7.94e+15 2.53e-06 6.10e+13 2.61e-06 6.31e+13 2.09e-10 1.27e+14 8.30e-11 2.00e+09 1.00e-08 2.10e+17 1.09e-04 2.63e+15 1.33e-08 3.21e+11 1.01e-06 2.45e+13 5.80e-06 1.40e+14 3.29e-07 1.87e+00 1.76e-07 1.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 1.00 0.00 1.00 0.00 0.00 0.00 0.50 -0.50 1.00 0.00 0.00 0.00 1.00 0.00 1.00 0.00 0.00 -0.22 0.00 -1.00 1.00 0.00 1.00 0.00 0.00 -1.00 1.00 0.00 1.00 0.00 1.00 12.40 2.42 39800 39800 19700 32700 37500 37500 17200 18000 21700 21700 21600 21600 18000 28800 2340 2340 21600 21600 18000 28800 18900 18900 18900 18900 9560 13700 11000 11000 7950 7950 31200 31200 46300 46300 29400 29400 35700 35700 14800 2130 10300 Est · · · 83 · · · 84 · · · Est · · · 33 · · · 85 · · · 86 · · · Est · · · 85 · · · 86 · · · 87 · · · 88 · · · 89 · · · 90 · · · 91 · · · 92 · · · 93 · · · 93 · · · 94 · · · 95 96 97 68 Table 6—Continued # Type Reaction α β γ Ref 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 2d 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i 2 + H2O 2 + H2O CH3CH2O ⇌ C2H4O + H CH3CH2O ⇌ CH2O + CH3 CH3CH2O ⇌ CH2O + CH3 Na + HO ⇌ NaOH Na + HO ⇌ NaOH Na + H ⇌ NaH Na + H ⇌ NaH K + HO ⇌ KOH K + HO ⇌ KOH K + H ⇌ KH K + H ⇌ KH HCl ⇌ Cl + H HCl ⇌ Cl + H O(1D) + N2 ⇌ N2O O(1D) + N2 ⇌ N2O CH3OH ⇌ CH∗ CH3OH ⇌ CH∗ HN∗ ⇌ HN HN∗ ⇌ HN Ar+ + Ar ⇌ Ar+ 2 Ar+ + Ar ⇌ Ar+ 2 C+ + H2 ⇌ CH+ 2 C+ + H2 ⇌ CH+ 2 CH+ CH+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ Fe+ + O2 ⇌ FeO+ 2 Fe+ + O2 ⇌ FeO+ 2 H+ + H2 ⇌ H+ 3 H+ + H2 ⇌ H+ 3 H+ H+ H+ H+ 3 + H2 ⇌ CH+ 5 3 + H2 ⇌ CH+ 5 2 + C2H2 ⇌ C4H+ 4 2 + C2H2 ⇌ C4H+ 4 2 + H2 ⇌ C2H+ 4 2 + H2 ⇌ C2H+ 4 3 + H2 ⇌ C2H+ 5 3 + H2 ⇌ C2H+ 5 4 + C2H4 ⇌ C4H+ 8 4 + C2H4 ⇌ C4H+ 8 3 + Ar ⇌ ArH+ 3 3 + Ar ⇌ ArH+ 3 3 + O2 ⇌ H3O+ 2 3 + O2 ⇌ H3O+ 2 69 4.24e+12 5.51e-05 1.33e+15 1.90e-25 1.00e-11 1.90e-25 1.00e-11 1.90e-25 1.00e-11 1.90e-25 1.00e-11 7.31e-11 1.81e+09 2.80e-36 6.76e-17 3.94e-04 9.51e+15 1.35e-10 3.29e+09 1.46e-31 3.53e-12 2.10e-29 5.07e-10 1.10e-28 2.66e-09 1.60e-26 3.86e-07 1.20e-27 2.90e-08 1.49e-29 3.60e-10 6.30e-26 1.52e-06 1.00e-30 2.42e-11 3.05e-29 7.37e-10 1.00e-31 2.42e-12 3.00e-29 7.25e-10 1.42 -1.02 -2.02 -2.21 0.00 -2.21 0.00 -2.21 0.00 -2.21 0.00 0.00 -1.00 -0.90 -1.90 -0.02 -1.02 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 10300 10400 10400 41 0 41 41 41 41 41 41 41000 41000 0 0 46200 46200 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 · · · 98 · · · 99 100 100 · · · 99 100 100 · · · 101 · · · 21 · · · 21 · · · 102 · · · 103 · · · 104 · · · 105 · · · 106 · · · 107 · · · 107 · · · 107 · · · 108 · · · 109 · · · 110 · · · 111 · · · Table 6—Continued # Type Reaction α β γ Ref 1.06e-31 2.56e-12 2.50e-30 6.04e-11 2.00e-28 4.83e-09 5.00e-28 1.21e-08 4.60e-30 1.11e-10 1.15e-09 2.77e+10 5.31e-12 1.28e+08 1.83e-09 4.41e+10 2.72e-09 6.57e+10 3.10e-08 7.48e+11 9.18e-09 2.22e+11 1.83e-09 4.41e+10 1.17e-08 2.81e+11 1.45e-08 3.53e+11 9.33e-08 2.25e+12 4.25e-11 1.03e+09 2.72e-09 6.57e+10 3.87e-09 9.35e+10 4.25e-11 1.03e+09 9.18e-09 2.22e+11 1.45e-08 -1.00 -2.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.00 -1.00 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 0 0 0 0 0 0 0 0 0 0 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 112 · · · 108 · · · 113 · · · 114 · · · 115 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 3i 3i 3i 3i 3i 3i 3i 3i 3i 3i ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti 2 + H2O ⇌ H2NO+ 3 2 + H2O ⇌ H2NO+ 3 2 + CO2 ⇌ CO+ 4 2 + CO2 ⇌ CO+ 4 He+ + He ⇌ He+ 2 He+ + He ⇌ He+ 2 Mg+ + O2 ⇌ MgO+ 2 Mg+ + O2 ⇌ MgO+ 2 N+ + N2 ⇌ N+ 3 N+ + N2 ⇌ N+ 3 NO+ NO+ O+ O+ C ⇌ C+ + e− C ⇌ C+ + e− H ⇌ H+ + e− H ⇌ H+ + e− N ⇌ N+ + e− N ⇌ N+ + e− O ⇌ O+ + e− O ⇌ O+ + e− Ar ⇌ Ar+ + e− Ar ⇌ Ar+ + e− C2 ⇌ C+ 2 + e− C2 ⇌ C+ 2 + e− CH ⇌ CH+ + e− CH ⇌ CH+ + e− CN ⇌ CN+ + e− CN ⇌ CN+ + e− CO ⇌ CO+ + e− CO ⇌ CO+ + e− Fe ⇌ Fe+ + e− Fe ⇌ Fe+ + e− H2 ⇌ H+ 2 + e− H2 ⇌ H+ 2 + e− HN ⇌ HN+ + e− HN ⇌ HN+ + e− HO ⇌ HO+ + e− HO ⇌ HO+ + e− He ⇌ He+ + e− He ⇌ He+ + e− Mg ⇌ Mg+ + e− Mg ⇌ Mg+ + e− N2 ⇌ N+ 2 + e− 70 Table 6—Continued # Type Reaction α β γ Ref 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti 2 + e− N2 ⇌ N+ 3.53e+11 NO ⇌ NO+ + e− 1.80e-08 NO ⇌ NO+ + e− 4.32e+11 O2 ⇌ O+ 2 + e− 2.18e-08 O2 ⇌ O+ 2 + e− 5.26e+11 Si ⇌ Si+ + e− 1.45e-08 Si ⇌ Si+ + e− 3.53e+11 Ti ⇌ Ti+ + e− 5.65e-08 Ti ⇌ Ti+ + e− 1.37e+12 C2H ⇌ C2H+ + e− 1.17e-08 C2H ⇌ C2H+ + e− 2.81e+11 C2N ⇌ C2N+ + e− 3.64e-08 C2N ⇌ C2N+ + e− 8.80e+11 C2O ⇌ C2O+ + e− 4.25e-08 C2O ⇌ C2O+ + e− 1.03e+12 CH2 ⇌ CH+ 2 + e− 2.72e-09 CH2 ⇌ CH+ 2 + e− 6.57e+10 CH3 ⇌ CH+ 3 + e− 3.87e-09 CH3 ⇌ CH+ 3 + e− 9.35e+10 CH4 ⇌ CH+ 4 + e− 5.31e-09 CH4 ⇌ CH+ 4 + e− 1.28e+11 HCN ⇌ CHN+ + e− 1.45e-08 HCN ⇌ CHN+ + e− 3.53e+11 CHO ⇌ CHO+ + e− 1.80e-08 CHO ⇌ CHO+ + e− 4.32e+11 CN2 ⇌ CN+ 4.25e-08 CN2 ⇌ CN+ 1.03e+12 CNO ⇌ CNO+ + e− 4.91e-08 CNO ⇌ CNO+ + e− 1.19e+12 CO2 ⇌ CO+ 2 + e− 5.65e-08 CO2 ⇌ CO+ 2 + e− 1.37e+12 FeO ⇌ FeO+ + e− 2.09e-07 FeO ⇌ FeO+ + e− 5.04e+12 H2N ⇌ H2N+ + e− 3.87e-09 H2N ⇌ H2N+ + e− 9.35e+10 H2O ⇌ H2O+ + e− 5.31e-09 H2O ⇌ H2O+ + e− 1.28e+11 H3N ⇌ H3N+ + e− 5.31e-09 H3N ⇌ H3N+ + e− 1.28e+11 HN2 ⇌ HN+ 2 + e− 1.80e-08 HN2 ⇌ HN+ 2 + e− 4.32e+11 2 + e− 2 + e− -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 71 Table 6—Continued # Type Reaction α β γ Ref 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti 2 + e− 2 + e− HNO ⇌ HNO+ + e− 2.18e-08 HNO ⇌ HNO+ + e− 5.26e+11 HO2 ⇌ HO+ 2.61e-08 HO2 ⇌ HO+ 6.30e+11 MgO ⇌ MgO+ + e− 4.25e-08 MgO ⇌ MgO+ + e− 1.03e+12 N2O ⇌ N2O+ + e− 5.65e-08 N2O ⇌ N2O+ + e− 1.37e+12 NO2 ⇌ NO+ 2 + e− 6.46e-08 NO2 ⇌ NO+ 2 + e− 1.56e+12 SiH ⇌ SiH+ + e− 1.80e-08 SiH ⇌ SiH+ + e− 4.32e+11 SiO ⇌ SiO+ + e− 5.65e-08 SiO ⇌ SiO+ + e− 1.37e+12 TiO ⇌ TiO+ + e− 1.43e-07 TiO ⇌ TiO+ + e− 3.46e+12 C2H2 ⇌ C2H+ 2 + e− 1.45e-08 C2H2 ⇌ C2H+ 2 + e− 3.53e+11 C2H3 ⇌ C2H+ 3 + e− 1.80e-08 C2H3 ⇌ C2H+ 3 + e− 4.32e+11 C2H4 ⇌ C2H+ 4 + e− 2.18e-08 C2H4 ⇌ C2H+ 4 + e− 5.26e+11 C2H5 ⇌ C2H+ 5 + e− 2.61e-08 C2H5 ⇌ C2H+ 5 + e− 6.30e+11 C2H6 ⇌ C2H+ 6 + e− 3.10e-08 C2H6 ⇌ C2H+ 6 + e− 7.48e+11 NCCN ⇌ C2N+ 2 + e− 9.33e-08 NCCN ⇌ C2N+ 2 + e− 2.25e+12 CH2O ⇌ CH2O+ + e− 2.18e-08 CH2O ⇌ CH2O+ + e− 5.26e+11 CH3N ⇌ CH3N+ + e− 2.18e-08 CH3N ⇌ CH3N+ + e− 5.26e+11 CH3O ⇌ CH3O+ + e− 2.61e-08 CH3O ⇌ CH3O+ + e− 6.30e+11 CH3OH ⇌ CH4O+ + e− 3.10e-08 CH3OH ⇌ CH4O+ + e− 7.48e+11 CH5N ⇌ CH5N+ + e− 3.10e-08 CH5N ⇌ CH5N+ + e− 7.48e+11 HNCO ⇌ CHNO+ + e− 5.65e-08 HNCO ⇌ CHNO+ + e− 1.37e+12 FeO2 ⇌ FeO+ 3.94e-07 2 + e− 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 72 Table 6—Continued # Type Reaction α β γ Ref 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti ti 2 + e− 2 + e− 2 + e− 2 + e− 3 + e− 3 + e− 4 + e− 4 + e− 2 + e− 2 + e− 2 + e− 2 + e− 2 + e− FeO2 ⇌ FeO+ H2O2 ⇌ H2O+ H2O2 ⇌ H2O+ HNO2 ⇌ HNO+ HNO2 ⇌ HNO+ MgHO ⇌ MgHO+ + e− MgHO ⇌ MgHO+ + e− MgO2 ⇌ MgO+ MgO2 ⇌ MgO+ SiH2 ⇌ SiH+ SiH2 ⇌ SiH+ SiH3 ⇌ SiH+ SiH3 ⇌ SiH+ SiH4 ⇌ SiH+ SiH4 ⇌ SiH+ CH2CN ⇌ C2H2N+ + e− CH2CN ⇌ C2H2N+ + e− C2H2O ⇌ C2H2O+ + e− C2H2O ⇌ C2H2O+ + e− C2H3N ⇌ C2H3N+ + e− C2H3N ⇌ C2H3N+ + e− C2H3O ⇌ C2H3O+ + e− C2H3O ⇌ C2H3O+ + e− C2H4O ⇌ C2H4O+ + e− C2H4O ⇌ C2H4O+ + e− CH2NCH3 ⇌ C2H5N+ + e− CH2NCH3 ⇌ C2H5N+ + e− CH3CH2O ⇌ C2H5O+ + e− CH3CH2O ⇌ C2H5O+ + e− C2H5OH ⇌ C2H6O+ + e− C2H5OH ⇌ C2H6O+ + e− CH2O2 ⇌ CH2O+ 2 + e− CH2O2 ⇌ CH2O+ 2 + e− HCOOH ⇌ CH2O+ 2 + e− HCOOH ⇌ CH2O+ 2 + e− CH2OH ⇌ CH2OH+ + e− CH2OH ⇌ CH2OH+ + e− CH3O2 ⇌ CH3O+ CH3O2 ⇌ CH3O+ Na ⇌ Na+ + e− Na ⇌ Na+ + e− 2 + e− 2 + e− 9.51e+12 3.10e-08 7.48e+11 7.34e-08 1.77e+12 4.91e-08 1.19e+12 1.17e-07 2.81e+12 2.18e-08 5.26e+11 2.61e-08 6.30e+11 3.10e-08 7.48e+11 4.91e-08 1.19e+12 5.65e-08 1.37e+12 5.65e-08 1.37e+12 6.46e-08 1.56e+12 7.34e-08 1.77e+12 7.34e-08 1.77e+12 8.30e-08 2.00e+12 9.33e-08 2.25e+12 7.34e-08 1.77e+12 7.34e-08 1.77e+12 2.61e-08 6.30e+11 8.30e-08 2.00e+12 8.49e+05 2.48e+12 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.50 -0.50 0.77 -0.23 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ 0∗ · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 116 · · · 73 Table 6—Continued # Type Reaction α β 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 ti ti 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n K ⇌ K+ + e− 8.49e+05 K ⇌ K+ + e− 2.48e+12 6.64e-11 C + CN ⇌ C2 + N 1.30e-10 CH + H ⇌ C + H2 8.70e-11 C + N2 ⇌ CN + N 3.49e-11 C + NO ⇌ CO + N 1.99e-10 C + O2 ⇌ CO + O 2.69e-12 C + CH2 ⇌ CH + CH 9.61e-13 C + H2N ⇌ CH + HN C + H2O ⇌ CH + HO 1.30e-12 C + NCCN ⇌ CN + C2N 3.01e-11 4.02e-10 H + HN ⇌ H + H + N H + HN ⇌ H2 + N 2.14e-12 6.86e-14 H + HO ⇌ H2 + O 5.27e-10 H + N2 ⇌ HN + N 9.30e-10 H + NO ⇌ HN + O H + NO ⇌ HO + N 3.60e-10 6.73e-10 H + O2 ⇌ HO + O 5.99e-11 H + C2H ⇌ H2 + C2 2.19e-11 H + C2O ⇌ CO + CH H + CH2 ⇌ H2 + CH 2.60e-10 3.53e-13 H + HCN ⇌ HNC + H 6.19e-10 H + HCN ⇌ CN + H2 6.61e-11 H + CHO ⇌ O + CH2 1.50e-10 H + CHO ⇌ CO + H2 H + NCO ⇌ CO + HN 8.90e-11 1.86e-11 H + NCO ⇌ HCN + O 2.51e-10 H + CO2 ⇌ CO + HO 1.05e-10 H + H2N ⇌ H2 + HN H + H2O ⇌ H2 + HO 6.82e-12 1.05e-09 H + HNO ⇌ H2N + O 2.41e-09 H + HNO ⇌ HO + HN 5.63e-11 H + HNO ⇌ H2 + NO H + HO2 ⇌ H2O + O 6.55e-12 5.50e-11 H + HO2 ⇌ HO + HO 5.60e-12 H + HO2 ⇌ H2 + O2 2.18e-08 H + N2O ⇌ O + HN2 H + N2O ⇌ NO + HN 5.03e-07 1.60e-10 H + N2O ⇌ HO + N2 1.47e-10 H + NO2 ⇌ HO + NO H + O3 ⇌ HO + O2 2.72e-11 0.77 -0.23 0.00 0.00 0.00 -0.02 0.00 0.00 0.00 0.00 0.00 -0.20 1.55 2.80 0.50 -0.10 0.00 -0.59 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.90 0.00 0.00 1.60 -0.30 -0.50 0.62 1.47 0.88 1.72 -1.06 -2.16 0.00 0.00 0.75 γ 0∗ 0∗ 0 80 22600 0 2010 23500 10500 19800 0 27300 120 1950 74500 35200 24900 8150 14200 0 0 0 12500 51600 0 0 2920 13400 4450 9720 14700 9010 179 6990 0 0 23800 18600 7600 0 0 Ref 116 · · · 118 163 33 120 119 121 121 121 122 123 124 7 123 125 33 126 7 127 5 128 18 129 5 3 3 7 130 5 125 125 131 132 132 132 133 133 18 134 135 74 Table 6—Continued # Type Reaction α β γ Ref 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n H + C2H2 ⇌ H2 + C2H H + NCCN ⇌ CN + HCN H + CH3 ⇌ H2 + CH2 H + CH2O ⇌ H2 + CHO H + HNCO ⇌ H2 + NCO H + HNCO ⇌ CO + H2N H + HOCN ⇌ H2 + NCO H + HOCN ⇌ CN + H2O H + H3N ⇌ H2 + H2N H + H2N2 ⇌ HN2 + H2 H + H2O2 ⇌ HO + H2O H + H2O2 ⇌ H2 + HO2 H + HNO2 ⇌ H2O + NO H + HNO2 ⇌ HO + HNO H + HNO2 ⇌ H2 + NO2 H + NO3 ⇌ HO + NO2 H + C2H3 ⇌ C2H2 + H2 H + C2H2O ⇌ CO + CH3 H + CH4 ⇌ H2 + CH3 H + CH3O ⇌ CH2O + H2 H + CH2OH ⇌ CH3 + HO H + CH2OH ⇌ CH2O + H2 H + CH2N2 ⇌ CH3 + N2 H + HNO3 ⇌ H2O + NO2 H + HNO3 ⇌ H2 + NO3 H + C2H4 ⇌ H2 + C2H3 H + CH3CN ⇌ HCN + CH3 H + CH3CN ⇌ CN + CH4 H + C2H3O ⇌ CH3 + CHO H + C2H2O2 ⇌ CO + H2 + CHO H + CH3OH ⇌ CH3 + H2O H + CH3OH ⇌ CH2OH + H2 H + CH3OH ⇌ CH3O + H2 H + H4N2 ⇌ NH2NH + H2 H + H2NNO2 ⇌ HNO + NO + H2 H + C2H5 ⇌ CH3 + CH3 H + C2H5 ⇌ C2H4 + H2 H + C2H4O ⇌ C2H3O + H2 H + C2H4O ⇌ CH2CHO + H2 H + C2H4O ⇌ CH3 + CO + H2 H + C2H4O ⇌ CH4 + CHO 2.49e-10 5.25e-10 1.00e-10 2.14e-12 2.67e-13 5.99e-13 1.66e-12 1.66e-12 6.54e-13 4.53e-13 4.00e-11 8.00e-11 6.39e-13 1.26e-11 2.27e-12 1.10e-10 2.03e-10 4.99e-12 5.94e-13 3.14e-10 1.60e-10 1.00e-11 1.60e-11 1.39e-14 5.65e-12 5.08e-12 3.39e-12 1.66e-13 3.32e-11 8.97e-11 9.41e-09 2.42e-12 6.64e-11 1.17e-11 1.66e-12 2.00e-10 3.01e-12 5.27e-13 2.12e-13 4.88e-13 8.80e-14 1.30 0.00 0.00 1.62 2.50 1.70 0.00 0.00 2.76 2.63 0.00 0.00 1.89 0.86 1.55 0.00 -1.00 1.45 3.00 -0.58 0.00 0.00 0.00 3.29 1.53 1.93 0.00 0.00 0.00 0.00 0.00 2.00 0.00 0.00 0.00 0.00 0.00 2.58 3.10 2.75 0.00 75 5170 0 0 0 0 0 0 0 0 1400 4050 855 1800 4000 1940 2500 3330 15400 4030 7600 1090 6690 1910 373 136 5 33 137 137 138 138 139 140 7 7 141 141 141 142 21 143 5 89 145 145 146 147 147 148 149 149 150 151 12400 Est 152 2270 45 3070 1260 153 14 144 7 154 154 154 155 3160 8250 6520 3950 1530 614 2620 486 0 0 0 0 0 0 Table 6—Continued # Type Reaction α β γ Ref 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 9.22e-11 H + Oxyrane ⇌ C2H4O + H 1.58e-13 H + Oxyrane ⇌ C2H4 + HO 8.30e-15 H + Oxyrane ⇌ C2H3 + H2O 3.32e-11 H + Oxyrane ⇌ cyc-C2H3O + H2 5.28e-09 H + CH3NO2 ⇌ CH3 + HNO2 2.03e-13 H + CH3ONO ⇌ CH3OH + NO 1.22e-11 H + C2H6 ⇌ C2H5 + H2 9.80e-13 H + C2H5OH ⇌ C2H5 + H2O 3.02e-11 N + CH ⇌ C + HN 1.66e-10 N + CH ⇌ CN + H 3.84e-09 N + CO ⇌ CN + O 1.88e-11 N + HO ⇌ HN + O 3.40e-11 N + NO ⇌ N2 + O 3.44e-12 N + O2 ⇌ NO + O 1.00e-10 N + C2N ⇌ CN + CN 5.50e-10 N + C2O ⇌ CN + CO 9.96e-13 N + CH2 ⇌ CH + HN 3.30e-11 N + NCO ⇌ CO + N2 1.66e-12 N + NCO ⇌ CN + NO 3.20e-13 N + CO2 ⇌ CO + NO 2.99e-13 N + H2N ⇌ HN + HN 6.03e-11 N + H2O ⇌ HO + HN 5.80e-12 N + NO2 ⇌ N2O + O 1.50e-11 N + N2O ⇌ N2 + NO 1.00e-16 N + O3 ⇌ O2 + NO 6.00e-15 N + HCCN ⇌ NCCN + H 1.18e-10 N + CH3 ⇌ H2CN + H 3.33e-13 N + CH3 ⇌ HCN + H + H 3.85e-05 N + HNCO ⇌ HN + NCO 7.70e-11 N + C2H3 ⇌ CH2CN + H 2.51e-14 N + CH4 ⇌ HCN + H2 + H N + C2H4 ⇌ HCN + CH3 2.66e-14 N + CH3CN ⇌ HCN + HCN + H 2.27e-15 N + CH3OH ⇌ CH3 + HNO 3.99e-10 1.99e-14 N + C2H4O ⇌ HCN + CHO + H2 3.32e-10 N + C2H5OH ⇌ C2H5 + HNO 5.99e-10 O + C2 ⇌ CO + C O + CH ⇌ HO + C 2.52e-11 1.02e-10 O + CH ⇌ CO + H 5.37e-11 O + CN ⇌ NO + C O + C2H ⇌ CO + CH 1.69e-11 0.00 0.00 0.00 0.00 0.00 0.00 1.50 0.00 0.65 0.00 0.00 0.10 0.00 1.18 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.20 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 5510 2520 2520 4180 7920 665 3730 1740 1210 0 36000 10700 0 4000 0 0 20500 0 0 1710 7600 19200 0 51 0 0 0 0 26500 0 0 352 813 4330 0 4210 0 2380 914 13700 0 68 156 156 156 157 158 144 159 160 161 162 125 164 165 122 166 121 5 167 168 121 125 169 169 170 171 172 173 174 175 176 177 178 179 180 179 181 182 182 120 5 76 Table 6—Continued # Type Reaction α β 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n O + C2N ⇌ CN + CO 5.99e-12 O + C2O ⇌ CO + CO 8.60e-11 O + CH2 ⇌ CO + H + H 1.33e-10 O + CH2 ⇌ CO + H2 6.64e-11 O + HCN ⇌ CO + HN 8.88e-13 O + HCN ⇌ CN + HO 3.65e-11 O + HNC ⇌ CO + HN 7.64e-12 O + CHO ⇌ CO + HO 5.00e-11 O + CHO ⇌ CO2 + H 5.00e-11 O + NCN ⇌ CO + N2 2.22e-16 O + NCN ⇌ NCO + N 2.26e-11 O + NCO ⇌ CO + NO 7.51e-11 O + NCO ⇌ CN + O2 4.05e-10 O + CO2 ⇌ CO + O2 1.41e-11 O + H2N ⇌ HO + HN 1.16e-11 O + H2N ⇌ H2 + NO 8.30e-12 O + H2O ⇌ HO + HO 6.68e-13 O + H2O ⇌ H2 + O2 4.48e-12 O + HNO ⇌ HO + NO 5.99e-11 O + HO2 ⇌ HO + O2 1.36e-11 O + N2O ⇌ NO + NO 1.52e-10 O + N2O ⇌ N2 + O2 6.13e-12 O + NO2 ⇌ NO + O2 6.51e-12 O + O3 ⇌ O2 + O2 2.23e-12 O + C2H2 ⇌ HCCO + H 7.14e-10 O + C2H2 ⇌ CH2 + CO 3.49e-12 O + NCCN ⇌ CN + NCO 4.15e-11 O + CH3 ⇌ CH2O + H 1.40e-10 O + CH3 ⇌ CO + H2 + H 5.72e-11 O + CH2O ⇌ CHO + HO 1.78e-11 O + CH2O ⇌ CO + HO + H 1.00e-10 4.93e-13 O + HNCO ⇌ CO2 + HN 6.61e-13 O + HNCO ⇌ NCO + HO O + H3N ⇌ HO + H2N 1.60e-11 2.98e-13 O + H3N ⇌ H2O + HN 1.42e-12 O + H2O2 ⇌ HO2 + HO 2.01e-11 O + HNO2 ⇌ HO + NO2 O + NO3 ⇌ O2 + NO2 1.70e-11 1.60e-10 O + C2H3 ⇌ C2H2O + H 5.50e-12 O + C2H3 ⇌ C2H2 + HO O + CH4 ⇌ CH3 + HO 5.63e-10 0.00 0.00 0.00 0.00 1.21 1.58 0.00 0.00 0.00 2.32 0.18 0.00 -1.43 0.00 0.00 0.00 2.60 0.97 0.00 0.75 0.00 0.00 0.00 0.75 0.00 1.50 0.00 0.00 0.00 0.57 0.00 1.41 2.11 0.00 0.00 2.00 0.00 0.00 0.00 0.20 0.00 γ 0 0 0 0 3820 13300 1100 0 0 0 0 0 3400 22100 0 0 7640 35300 0 0 14000 8020 0 1580 6100 850 5500 0 0 1390 0 4290 5750 3670 2420 2000 3000 0 0 0 6230 Ref 166 127 183 41 18 18 14 5 5 184 184 3 3 185 125 125 186 187 18 135 188 188 169 135 189 189 101 5 190 5 191 3 3 5 Est 7 18 32 7 192 193 77 Table 6—Continued # Type Reaction α β γ Ref 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 7.51e-12 O + CH4 ⇌ CH2O + H2 1.00e-11 O + CH3O ⇌ CH2O + HO 3.55e-11 O + CH3O ⇌ CH3 + O2 7.01e-11 O + CH2OH ⇌ CH2O + HO 6.24e-13 O + C2H4 ⇌ C2H3O + H 2.31e-11 O + C2H4 ⇌ C2H3 + HO 1.50e-12 O + C2H4 ⇌ CH3 + CHO 3.82e-14 O + C2H4 ⇌ C2H2O + H2 8.30e-12 O + C2H4 ⇌ CH2O + CH2 6.40e-11 O + C2H3O ⇌ C2H2O + HO 1.60e-11 O + C2H3O ⇌ CO2 + CH3 5.71e-11 O + CH3OH ⇌ CH2OH + HO 1.66e-11 O + CH3OH ⇌ CH3O + HO 1.41e-10 O + H4N2 ⇌ H2N2 + H2O 1.33e-10 O + C2H5 ⇌ H2C2HOH + H 6.31e-12 O + C2H5 ⇌ C2H4 + HO 2.67e-11 O + C2H5 ⇌ CH2O + CH3 2.49e-11 O + C2H4O ⇌ CH2CHO + HO O + C2H4O ⇌ C2H3O + HO 2.81e-11 O + Oxyrane ⇌ cyc-C2H3O + HO 3.17e-12 O + C2H6 ⇌ C2H5 + HO 5.11e-12 O + C2H5OH ⇌ CH3CHOH + HO 1.03e-13 O + C2H5OH ⇌ CH3CH2O + HO 1.23e-15 O + (CH3)2O ⇌ CH3OCH2 + HO 4.68e-10 2.66e-09 CN + CN ⇌ C2 + N2 HN + HN ⇌ N2 + H + H 1.16e-09 5.10e-10 NO + NO ⇌ N2 + O2 7.47e-12 O3 + O3 ⇌ O2 + O2 + O2 8.30e-12 C2 + HO ⇌ CO + CH C2 + NO ⇌ C2O + N 3.75e-11 8.75e-11 C2 + NO ⇌ C2N + O 1.10e-11 C2 + O2 ⇌ CO + CO 4.06e-13 CH + N2 ⇌ NCN + H CH + N2 ⇌ HCN + N 6.64e-13 4.40e-11 CH + NO ⇌ NCO + H 1.33e-11 CH + NO ⇌ CHO + N 2.00e-10 CH + NO ⇌ CO + HN CH + NO ⇌ HCN + O 1.37e-10 1.40e-10 CH + NO ⇌ CN + HO 1.66e-11 CH + O2 ⇌ CHO + O CH + O2 ⇌ CO + HO 8.30e-11 0.00 0.00 0.00 0.00 0.00 0.00 1.55 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.03 0.00 0.00 0.00 0.00 2.40 0.00 4.73 0.00 0.00 0.00 0.50 0.00 0.00 0.00 0.00 0.00 1.12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 78 6230 0 0 0 0 3620 0 0 754 0 0 2750 2360 603 0 0 0 0 0 0 Est 7 194 145 195 196 7 33 197 33 7 198 45 199 Est 192 870 2590 21700 2520 1680 2640 2930 7 59 200 201 202 203 204 205 206 207 30600 Est 208 9310 209 210 210 211 212 213 214 215 214 215 214 216 217 4350 4350 381 8820 6840 0 0 0 0 0 0 0 Table 6—Continued # Type Reaction α β γ Ref 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2.82e-11 CH + H2O ⇌ CH2O + H 7.82e-11 CH + N2O ⇌ HCN + NO 3.09e-11 CH + N2O ⇌ CO + N2 + H 1.45e-10 CH + NO2 ⇌ CHO + NO 9.25e-11 CH + CH4 ⇌ C2H4 + H 2.84e-10 CH + C2H4 ⇌ C3H4 + H 7.01e-11 CN + HO ⇌ NCO + H 1.79e-10 CN + NO ⇌ CO + N2 2.42e-12 CN + O2 ⇌ CO + NO 1.00e-10 CN + CHO ⇌ HCN + CO 5.18e-11 CN + NCO ⇌ NCN + CO 3.01e-11 CN + NCO ⇌ CN2 + CO 3.82e-11 CN + H2O ⇌ HCN + HO 3.01e-11 CN + HNO ⇌ HCN + NO 2.01e-12 CN + N2O ⇌ NCO + N2 1.73e-14 CN + N2O ⇌ CNN + NO 4.00e-11 CN + NO2 ⇌ NCO + NO 7.11e-12 CN + NO2 ⇌ N2O + CO 5.20e-12 CN + NO2 ⇌ CO2 + N2 2.43e-10 CN + C2H2 ⇌ HC3N + H 2.19e-10 CN + C2H2 ⇌ HCN + C2H 7.01e-11 CN + CH2O ⇌ HCN + CHO 2.51e-11 CN + HNCO ⇌ HCN + NCO 3.95e-11 CN + HCNO ⇌ HCCN + NO 1.10e-10 CN + HCNO ⇌ HCN + NCO 1.66e-11 CN + H3N ⇌ HCN + H2N CN + HNO2 ⇌ HCN + NO2 2.01e-11 CN + C2H2O ⇌ HNC + HCCO 2.37e-11 5.11e-13 CN + CH4 ⇌ HCN + CH3 CN + C2H4 ⇌ C3H3N + H 3.49e-10 CN + C2H4 ⇌ C2H3 + HCN 2.09e-10 CN + CH3OH ⇌ CH3O + HCN 1.20e-10 2.08e-11 CN + C2H6 ⇌ C2H5 + HCN CO + HNO ⇌ CO2 + HN 3.32e-12 6.46e-14 CO + HO2 ⇌ CO2 + HO 5.30e-13 CO + N2O ⇌ CO2 + N2 1.50e-10 CO + NO2 ⇌ CO2 + NO CO + C2H2 ⇌ C2H + CHO 8.00e-10 6.31e-11 CO + CH3 ⇌ C2H2 + HO 2.61e-11 CO + CH3O ⇌ CH3 + CO2 CO + CH3O ⇌ CH2O + CHO 3.23e-13 -1.22 0.00 0.00 0.00 -0.90 -0.31 0.00 0.00 0.18 0.00 0.16 0.00 0.00 0.00 0.00 2.60 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.64 0.00 0.00 0.00 0.22 0.00 2.18 0.00 0.00 0.00 0.00 0.00 2.28 79 4040 6700 1860 0 0 0 0 0 0 0 0 0 0 0 0 0 218 Est Est 219 220 Est 3 221 222 3 223 3 224 3 171 3 3 225 225 226 227 3 3 228 229 101 3 230 33 Est 227 Est 231 130 6190 232 9030 18 10200 18 17000 7 53600 233 30400 5940 7 10100 Est 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Table 6—Continued # Type Reaction α β γ Ref 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2.51e-10 CO + C2H4 ⇌ C2H3 + CHO 4.15e-11 H2 + O2 ⇌ HO + HO 1.66e-10 H2 + HNO ⇌ H2O + HN 5.73e-12 H2 + N2O ⇌ H2O + N2 7.04e-08 H2 + H2O2 ⇌ HO + HO + H2 3.07e-11 HN + NO ⇌ HN2 + O 5.86e-12 HN + NO ⇌ HO + N2 6.79e-14 HN + O2 ⇌ HNO + O 1.09e-14 HN + O2 ⇌ HO + NO 1.44e-10 HN + H2N ⇌ H2N2 + H 1.81e-12 HN + H2O ⇌ H2N + HO 9.50e-12 HN + NO2 ⇌ HNO + NO 6.60e-12 HN + NO2 ⇌ N2O + HO 1.04e-11 HN + HNCO ⇌ H2N + NCO 5.25e-10 HN + H3N ⇌ H2N + H2N HN + CH4 ⇌ CH3 + H2N 1.49e-10 HN + C2H4O ⇌ C2H3O + H2N 8.30e-11 1.16e-10 HN + C2H6 ⇌ C2H5 + H2N 3.01e-11 HO + C2H ⇌ CH2 + CO 3.01e-11 HO + C2H ⇌ C2H2 + O HO + CH2 ⇌ CH2O + H 3.01e-11 1.07e-13 HO + HCN ⇌ H2N + CO 2.01e-11 HO + HCN ⇌ HOCN + H 2.84e-13 HO + HCN ⇌ HNCO + H 4.65e-11 HO + HNC ⇌ HNCO + H HO + CHO ⇌ H2O + CO 1.69e-10 3.99e-13 HO + NCO ⇌ CHO + NO 2.99e-13 HO + NCO ⇌ CO + NO + H 4.72e-12 HO + HNO ⇌ H2O + NO HO + HO2 ⇌ H2O + O2 7.11e-11 1.01e-17 HO + N2O ⇌ HNO + NO 1.03e-14 HO + N2O ⇌ HO2 + N2 2.25e-11 HO + NO2 ⇌ HO2 + NO HO + O3 ⇌ HO2 + O2 3.76e-13 4.12e-13 HO + C2H2 ⇌ C2H2O + H 1.03e-13 HO + C2H2 ⇌ C2H + H2O 1.91e-13 HO + C2H2 ⇌ HCCO + H2 HO + CH3 ⇌ CH2 + H2O 1.20e-10 6.45e-13 HO + CH3 ⇌ CH3O + H 9.10e-11 HO + CH3 ⇌ CH2O + H2 HO + CHN2 ⇌ NCO + H2N 7.93e-13 0.00 0.44 0.00 0.50 0.00 0.21 -0.50 2.00 1.50 -0.50 1.60 0.00 0.00 1.82 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.94 -0.21 4.33 4.72 0.00 1.99 2.30 2.68 0.00 0.00 1.01 0.00 -0.18 45600 34800 8060 0 22900 5470 0 3270 0 0 14100 0 0 12000 13500 10100 5770 8420 0 0 0 5890 8520 4390 1860 0 0 0 0 0 12600 18400 3830 603 6790 6060 0 1400 6010 1500 4770 7 187 130 234 235 133 133 137 137 236 125 Est 237 238 239 130 130 130 7 7 7 240 240 240 14 5 241 241 131 242 243 243 244 245 246 7 247 33 248 249 250 80 Table 6—Continued # Type Reaction α β γ Ref 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 1.09e-17 HO + CHN2 ⇌ HNC + HNO 1.16e-15 HO + CHN2 ⇌ NCN + H2O 6.21e-14 HO + CHN2 ⇌ HCN + HNO 2.94e-13 HO + HNCO ⇌ NCO + H2O 1.54e-14 HO + HNCO ⇌ CO2 + H2N 2.55e-12 HO + HONC ⇌ NCO + H2O 1.34e-13 HO + HONC ⇌ CO2 + H2N 5.25e-12 HO + H3N ⇌ H2O + H2N 2.54e-14 HO + H2N2 ⇌ H2O + HN2 1.30e-11 HO + H2O2 ⇌ HO2 + H2O 6.24e-12 HO + HNO2 ⇌ H2O + NO2 2.32e-11 HO + NO3 ⇌ HO2 + NO2 5.00e-11 HO + C2H3 ⇌ C2H2 + H2O 1.01e-11 HO + C2H2O ⇌ CH2OH + CO 4.98e-12 HO + C2H2O ⇌ CH3 + CO2 4.65e-11 HO + C2H2O ⇌ CH2O + CHO 2.96e-13 HO + CH4 ⇌ CH3 + H2O 3.01e-11 HO + CH3O ⇌ CH2O + H2O 4.00e-11 HO + CH2OH ⇌ CH2O + H2O 1.41e-11 HO + CH2N2 ⇌ CHN2 + H2O 9.85e-13 HO + CH2O2 ⇌ COOH + H2O HO + HCOOH ⇌ CHO + HO + HO 4.50e-13 4.50e-13 HO + HCOOH ⇌ CH2O + HO2 1.71e-14 HO + HNO3 ⇌ NO3 + H2O 1.37e-12 HO + C2H4 ⇌ C2H3 + H2O HO + CH3CN ⇌ CH2CN + H2O 6.43e-13 HO + C2H3O ⇌ C2H2O + H2O 2.01e-11 HO + C2H2O2 ⇌ H2O + CHO + CO 1.10e-11 2.13e-13 HO + CH3OH ⇌ CH2OH + H2O HO + CH3OH ⇌ CH3O + H2O 1.66e-11 1.10e-12 HO + CH3OH ⇌ CH2O + H2O + H 6.09e-11 HO + H4N2 ⇌ H3N + NH2O 4.00e-11 HO + C2H5 ⇌ C2H4 + H2O HO + C2H4O ⇌ CH2CHO + H2O 2.66e-11 HO + C2H4O ⇌ C2H3O + H2O 1.66e-11 HO + Oxyrane ⇌ cyc-C2H3O + H2O 2.96e-11 1.20e-12 HO + CH4O2 ⇌ CH3O2 + H2O HO + C2H6 ⇌ C2H5 + H2O 1.88e-12 HO + CH3NO3 ⇌ CH3O + HNO3 3.01e-13 HO + C2H5OH ⇌ CH3CH2O + H2O 1.60e-13 HO + (CH3)2O ⇌ CH3OCH2 + H2O 5.28e-13 4.44 2.92 0.74 1.50 1.50 0.00 0.00 0.00 3.40 0.00 1.00 0.00 0.00 0.00 0.00 0.00 2.38 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.53 2.49 0.00 0.00 2.00 0.00 1.44 0.00 0.00 0.00 0.00 0.00 0.00 1.80 0.00 0.00 2.91 4960 0 6750 1810 1810 0 0 1010 0 670 0 0 0 0 0 0 1140 0 0 3280 1040 0 0 0 2190 2120 0 0 0 834 0 0 0 1010 1010 1820 0 570 0 0 410 250 250 250 251 251 Est Est 252 140 5 18 142 7 253 254 255 256 7 145 257 258 Est Est 259 260 261 7 54 152 45 262 Est 7 59 45 263 5 264 265 266 267 81 # Type Reaction α β γ Ref Table 6—Continued 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n HO + bCO2H5N ⇌ C2H4O + H2O + NO 8.21e-13 8.00e-12 N2 + CH2 ⇌ HCN + HN 2.16e-11 NO + C2O ⇌ CN + CO2 4.17e-12 NO + CH2 ⇌ HNCO + H NO + CH2 ⇌ HCN + HO 8.32e-13 1.18e-11 NO + CHO ⇌ HNO + CO 5.15e-11 NO + NCO ⇌ N2O + CO 4.21e-11 NO + NCO ⇌ N2 + CO + O NO + NCO ⇌ N2 + CO2 8.05e-11 1.28e-10 NO + H2N ⇌ N2 + H2O 5.15e-09 NO + H2N ⇌ HN2 + HO 1.49e-12 NO + H2N ⇌ N2 + HO + H NO + HO2 ⇌ HNO + O2 1.17e-14 2.87e-13 NO + N2O ⇌ NO2 + N2 3.00e-12 NO + O3 ⇌ NO2 + O2 8.97e-12 NO + C2H2 ⇌ HCN + CO + H NO + CH3 ⇌ H2CN + HO 8.63e-12 4.00e-12 NO + CH3 ⇌ HCN + H2O 1.69e-11 NO + CH2O ⇌ CHO + HNO 8.28e-13 NO + HNCO ⇌ HN2 + CO2 NO + H3N ⇌ H2N + HNO 3.29e-13 2.60e-11 NO + NO3 ⇌ NO2 + NO2 5.02e-11 NO + C2H3 ⇌ CH2O + HCN 4.00e-12 NO + CH3O ⇌ CH2O + HNO 1.66e-10 NO + C2H6 ⇌ C2H5 + HNO NO + (CH3)2O ⇌ CH3OCH2 + HNO 1.66e-10 8.70e-12 NO + C2H5OO ⇌ CH3CH2O + NO2 4.05e-13 O2 + C2O ⇌ CO2 + CO 4.21e-13 O2 + C2O ⇌ CO + CO + O O2 + CH2 ⇌ COOH + H 5.74e-12 4.00e-13 O2 + CH2 ⇌ H2O + CO 3.74e-11 O2 + CH2 ⇌ CO2 + H + H 2.99e-11 O2 + CH2 ⇌ CO2 + H2 O2 + CH2 ⇌ CH2O + O 3.74e-11 6.14e-11 O2 + CHO ⇌ HO2 + CO 1.17e-11 O2 + CHO ⇌ CO2 + HO 2.19e-14 O2 + NCN ⇌ CNO + NO O2 + NCN ⇌ NCO + NO 1.15e-13 1.32e-12 O2 + NCO ⇌ CO2 + NO 8.93e-10 O2 + H2N ⇌ NH2O + O O2 + H2N ⇌ HNO + HO 2.72e-13 0.00 0.00 0.00 0.00 0.00 0.00 -1.34 -1.98 -1.98 -2.65 -3.02 0.00 0.00 2.23 0.00 0.00 0.00 0.00 0.00 1.00 1.73 0.00 -3.38 -0.70 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -3.30 -3.30 -3.30 0.00 0.00 0.54 0.51 0.00 -1.34 -0.39 960 18000 337 3010 1440 0 360 380 380 632 4830 0 0 23300 1500 19000 12200 7900 20500 0 28500 0 515 0 26200 21800 0 0 0 751 0 1440 1440 1440 1560 1560 12300 12400 0 17000 18200 268 269 270 12 12 271 272 3 3 273 274 275 276 243 169 278 279 279 18 280 281 32 282 54 283 284 54 285 286 287 7 288 288 288 151 289 290 290 291 292 292 82 Table 6—Continued # Type Reaction α β γ Ref 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2.01e-11 O2 + C2H2 ⇌ C2H + HO2 5.17e-13 O2 + OCCN ⇌ CO2 + NCO 2.81e-13 O2 + CH3 ⇌ CH2O + HO 1.66e-12 O2 + CH3 ⇌ CHO + H2O 9.14e-12 O2 + CH2O ⇌ CHO + HO2 9.00e-11 O2 + H2O2 ⇌ HO2 + HO2 1.79e-11 O2 + C2H3 ⇌ CH2CHO + O 6.58e-12 O2 + C2H3 ⇌ C2H2 + HO2 2.03e-08 O2 + C2H3 ⇌ CH2O + CHO 4.01e-12 O2 + CH4 ⇌ CH3O2 + H 1.26e-12 O2 + CH4 ⇌ CH3 + HO2 3.60e-14 O2 + CH3O ⇌ CH2O + HO2 2.01e-12 O2 + CH2OH ⇌ CH2O + HO2 O2 + C2H4 ⇌ C2H3 + HO2 7.01e-11 O2 + C2H3O ⇌ C2H2O + HO + O 4.20e-12 1.10e-12 O2 + CH2CHO ⇌ CH3O + CO2 O2 + CH3OH ⇌ CH2OH + HO2 3.40e-11 6.14e-12 O2 + C2H5 ⇌ CH3CH2O + O 4.32e-15 O2 + C2H5 ⇌ Oxyrane + HO 2.14e-13 O2 + C2H5 ⇌ C2H4O + HO O2 + C2H5 ⇌ C2H4 + HO2 1.61e-12 3.32e-10 O2 + C2H4O ⇌ CH2CHO + HO2 5.00e-11 O2 + C2H4O ⇌ C2H3O + HO2 1.00e-10 O2 + C2H6 ⇌ C2H5 + HO2 1.00e-14 O3 + C2H2 ⇌ C2H2O + O2 O3 + CH3 ⇌ CH3O + O2 5.10e-12 2.66e-13 O3 + CH4 ⇌ HO3 + CH3 1.20e-14 O3 + C2H4 ⇌ C2H3 + HO + O2 3.32e-14 O3 + C2H5 ⇌ CH3CH2O + O2 O3 + C2H4O ⇌ CH2CHO + HO3 7.12e-18 9.37e-18 O3 + C2H4O ⇌ C2H3O + HO3 6.76e-13 O3 + CH3ONO ⇌ CH3NO3 + O2 3.94e-13 O3 + C2H6 ⇌ C2H5OH + O2 HN2 + H ⇌ N2 + H2 1.66e-12 9.72e-19 HN2 + O ⇌ HNO + N 2.56e-11 HN2 + O ⇌ N2 + HO 3.31e-12 HN2 + CH ⇌ HCN + HN HN2 + HO ⇌ H2O + N2 2.97e-09 6.94e-14 HN2 + O2 ⇌ N2O + HO 2.87e-13 HN2 + O2 ⇌ HO2 + N2 HN2 + CH3 ⇌ CH2N2 + H2 4.14e-13 0.00 1.64 0.00 0.00 2.05 0.00 -0.61 -1.26 -5.31 1.96 2.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.20 0.00 -1.03 -1.87 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.94 3.90 0.00 0.00 0.00 4.84 -1.23 1.00 -2.88 -0.34 -0.34 1.00 37500 593 4980 0 19100 20000 2650 1670 3270 43900 26200 880 0 29000 0 0 22600 14100 0 4860 707 24400 19700 26100 4100 0 7700 2630 0 14600 7710 5320 7200 0 0 0 0 1230 0 0 0 7 293 294 295 296 7 297 297 297 298 299 33 145 7 65 300 145 301 302 301 79 59 5 5 Est Est Est 169 303 304 304 305 306 25 307 25 Est 25 25 25 Est 83 # Type Reaction α Table 6—Continued 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 3.01e-12 C2H + C2H ⇌ C2H2 + C2 2.99e-10 CH2 + CH2 ⇌ C2H2 + H + H 1.80e-10 CH2 + CH2 ⇌ C2H2 + H2 3.64e-11 CHO + CHO ⇌ CO + CO + H2 5.00e-11 CHO + CHO ⇌ CH2O + CO 3.01e-11 NCO + NCO ⇌ CO + CO + N2 1.00e-13 H2N + H2N ⇌ NH2NH + H 1.30e-12 H2N + H2N ⇌ HNNH + H2 1.40e-15 HNO + HNO ⇌ H2O + N2O 1.49e-12 HO2 + HO2 ⇌ H2 + O2 + O2 2.71e-12 NO2 + NO2 ⇌ NO + NO + O2 1.66e-10 CH3 + CH3 ⇌ C2H4 + H2 7.14e-12 CH3 + CH3 ⇌ CH4 + CH2 8.50e-13 NO3 + NO3 ⇌ NO2 + NO2 + O2 3.01e-11 C2H + CH2 ⇌ C2H2 + CH 3.01e-11 C2H + HO2 ⇌ HCCO + HO 4.00e-11 C2H + CH3 ⇌ C3H3 + H 4.00e-11 C2H + CH2O ⇌ C3H2O + H 4.00e-11 C2H + CH2O ⇌ C2H2 + CO + H 1.60e-12 C2H + C2H3 ⇌ C2H2 + C2H2 1.20e-11 C2H + CH4 ⇌ C2H2 + CH3 4.00e-11 C2H + CH3O ⇌ CH2O + C2H2 5.99e-11 C2H + CH2OH ⇌ CH2O + C2H2 2.01e-11 C2H + CH2OH ⇌ C3H3 + HO 1.22e-10 C2H + C2H4 ⇌ C4H4 + H 1.00e-11 C2H + CH3OH ⇌ C2H2 + CH2OH 2.01e-12 C2H + CH3OH ⇌ C2H2 + CH3O 3.01e-12 C2H + C2H5 ⇌ C2H4 + C2H2 3.01e-11 C2H + C2H5 ⇌ C3H3 + CH3 4.72e-11 C2H + C2H6 ⇌ C2H5 + C2H2 2.84e-11 C2O + NO2 ⇌ CO + CO + NO 2.84e-11 C2O + NO2 ⇌ CO2 + CNO 2.84e-11 C2O + NO2 ⇌ CO2 + NCO 1.00e-14 C2O + C2H4 ⇌ C3H4 + CO C2O + Oxyrane ⇌ C2H4 + CO + CO 1.60e-12 C2O + Oxyrane ⇌ CH2O + C2H2 + CO 1.00e-14 3.01e-11 CH2 + CHO ⇌ CH3 + CO CH2 + CO2 ⇌ CH2O + CO 3.90e-14 1.00e-14 CH2 + N2O ⇌ HCN + NO + H 1.00e-14 CH2 + N2O ⇌ CH2O + N2 CH2 + C2H2 ⇌ C3H3 + H 1.00e-12 β 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.54 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 γ 0 397 400 0 0 0 0 0 1560 503 13100 16100 5050 2450 0 0 0 0 0 0 491 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 3330 Ref 7 12 5 308 5 3 309 309 18 310 18 311 183 24 7 7 7 312 312 7 277 7 145 Est 277 145 145 7 7 356 Est Est Est Est 313 313 7 7 Est Est 21 84 # Type Reaction α β γ Ref Table 6—Continued 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 1.20e-10 CH2 + CH3 ⇌ C2H4 + H 1.00e-14 CH2 + CH2O ⇌ CH3 + CHO 1.00e-14 CH2 + H2O2 ⇌ CH3 + HO2 8.00e-11 CH2 + C2H3 ⇌ C2H2 + CH3 2.00e-10 CH2 + C2H2O ⇌ C2H4 + CO 3.01e-11 CH2 + CH3O ⇌ CH2O + CH3 4.00e-11 CH2 + CH2OH ⇌ C2H4 + HO 2.01e-12 CH2 + CH2OH ⇌ CH2O + CH3 3.01e-11 CH2 + C2H3O ⇌ C2H2O + CH3 4.38e-15 CH2 + CH3OH ⇌ CH2OH + CH3 1.12e-15 CH2 + CH3OH ⇌ CH3O + CH3 8.00e-11 CH2 + C2H5 ⇌ C2H4 + CH3 1.07e-11 CH2 + C2H6 ⇌ C2H5 + CH3 4.50e-14 HCN + C2H3 ⇌ C3H3N + H 4.50e-14 HNC + C2H3 ⇌ C3H3N + H 5.99e-11 CHO + NCO ⇌ HNCO + CO 8.54e-13 CHO + H2O ⇌ CH2O + HO 8.13e-23 CHO + HNO ⇌ NH2O + CO 1.94e-10 CHO + NO2 ⇌ CO2 + NO + H 8.30e-13 CHO + O3 ⇌ CO2 + O2 + H 2.01e-10 CHO + CH3 ⇌ CH4 + CO 5.00e-12 CHO + HNCO ⇌ CH2O + NCO 1.69e-13 CHO + H2O2 ⇌ CH2O + HO2 3.05e-15 CHO + HNO2 ⇌ CH2O + NO2 1.36e-13 CHO + CH4 ⇌ CH2O + CH3 1.50e-10 CHO + CH3O ⇌ CH3OH + CO 2.01e-10 CHO + CH2OH ⇌ CH3OH + CO 3.01e-10 CHO + CH2OH ⇌ CH2O + CH2O CHO + C2H3O ⇌ C2H4O + CO 1.50e-11 CHO + CH3OH ⇌ CH2OH + CH2O 2.41e-13 2.01e-10 CHO + C2H5 ⇌ C2H6 + CO 6.71e-15 CHO + C2H6 ⇌ C2H5 + CH2O 3.01e-11 NCO + HNO ⇌ HNCO + NO NCO + HO2 ⇌ HNCO + O2 3.32e-11 1.50e-10 NCO + N2O ⇌ CO + N2 + NO 2.79e-11 NCO + NO2 ⇌ CO2 + N2O 2.21e-12 NCO + NO2 ⇌ CO + NO + NO NCO + CH3 ⇌ CH2O + HNC 7.57e-11 1.34e-10 NCO + CH3 ⇌ CH2O + HCN NCO + HCNO ⇌ HCCO + N2O 3.95e-12 NCO + HCNO ⇌ CHO + N2 + CO 3.95e-12 0.00 0.00 0.00 0.00 -2.40 0.00 0.00 0.00 0.00 3.20 3.10 0.00 0.00 0.00 0.00 0.00 1.53 3.27 -0.75 0.00 0.00 0.00 0.00 4.18 2.85 0.00 0.00 0.00 0.00 2.90 0.00 3.74 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 3610 3490 0 3980 0 0 0 13100 882 971 0 0 13100 3490 1060 11300 0 0 0 0 6600 0 8520 0 0 14000 0 0 1840 1840 0 0 30 7 7 30 315 7 145 145 7 145 145 21 316 317 Est 3 7 318 18 303 7 3 7 318 7 7 145 145 7 145 7 319 3 142 3 3 142 320 320 Est Est 85 # Type Reaction α β γ Ref Table 6—Continued 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n NCO + HCNO ⇌ CO2 + CHN2 3.95e-12 NCO + HCNO ⇌ HCN + CO + NO 3.95e-12 5.99e-12 NCO + HNO2 ⇌ HNCO + NO2 4.61e-12 NCO + C2H2O ⇌ HCNO + HCCO NCO + C2H2O ⇌ HOCN + HCCO 4.61e-12 4.61e-12 NCO + C2H2O ⇌ HNCO + HCCO 6.20e-14 NCO + C2H6 ⇌ HNCO + C2H5 2.61e-10 H2N + HO2 ⇌ NH2O + HO H2N + HO2 ⇌ H3N + O2 1.88e-16 1.60e-11 H2N + HO2 ⇌ HNO + H2O 5.81e-12 H2N + NO2 ⇌ NH2O + NO 7.01e-12 H2N + NO2 ⇌ N2O + H2O H2N + O3 ⇌ NH2O + O2 4.10e-12 8.30e-13 H2N + HNCO ⇌ H3N + NCO 7.33e-10 H2N + HNO2 ⇌ H3N + NO2 6.84e-14 H2N + CH4 ⇌ CH3 + H3N H2N + H4N2 ⇌ NH2NH + H3N 6.46e-15 2.74e-14 H2N + C2H6 ⇌ C2H5 + H3N 6.29e-11 H2O + N2O3 ⇌ HNO2 + HNO2 1.00e-12 HNO + NO2 ⇌ HNO2 + NO HNO + O3 ⇌ HNO2 + O2 9.61e-15 1.16e-14 HNO + CH3 ⇌ CH3NO + H 1.85e-11 HNO + CH3 ⇌ CH4 + NO 5.00e-11 HNO + CH3O ⇌ CH3OH + NO 2.31e-13 HO2 + NO2 ⇌ HNO2 + O2 HO2 + O3 ⇌ HO + O2 + O2 1.66e-13 1.00e-14 HO2 + C2H2 ⇌ C2H2O + HO 3.01e-11 HO2 + CH3 ⇌ CH3O + HO 1.00e-13 HO2 + H2O2 ⇌ H2O + HO + O2 HO2 + NO3 ⇌ HNO3 + O2 1.91e-12 2.51e-12 HO2 + NO3 ⇌ NO2 + O2 + HO 1.50e-11 HO2 + CH4 ⇌ CH3 + H2O2 5.00e-13 HO2 + CH3O ⇌ CH2O + H2O2 HO2 + CH2OH ⇌ CH2O + H2O2 2.01e-11 3.70e-12 HO2 + C2H4 ⇌ Oxyrane + HO 1.00e-14 HO2 + C2H4 ⇌ C2H4O + HO 1.66e-12 HO2 + CH3OH ⇌ CH2OH + H2O2 HO2 + C2H5 ⇌ CH3CH2O + HO 4.98e-11 5.00e-13 HO2 + C2H5 ⇌ C2H4 + H2O2 1.66e-12 HO2 + C2H4O ⇌ CH2CHO + H2O2 HO2 + C2H4O ⇌ C2H3O + H2O2 5.00e-12 0.00 0.00 0.00 0.00 0.00 0.00 3.27 -1.32 1.55 -1.12 0.00 -1.44 0.00 0.00 0.00 3.01 3.60 3.46 0.00 0.00 3.59 2.40 0.76 0.00 0.58 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 628 1020 0 0 0 1160 0 0 5000 386 2820 4470 999 0 3100 0 0 720 1410 4000 0 0 0 0 12400 0 0 8650 4000 5050 0 0 7050 6000 Est Est 3 Est Est Est 321 322 322 322 323 274 324 325 326 327 40 328 329 18 250 330 330 331 332 333 7 5 334 335 335 5 7 145 5 7 129 301 7 59 5 86 # Type Reaction α β γ Ref Table 6—Continued 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n HO2 + C2H6 ⇌ C2H5 + H2O2 HO2 + C2H5OO ⇌ C2H6O2 + O2 N2O + CH3 ⇌ CH3O + N2 N2O + C2H4 ⇌ C2H4O + N2 NO2 + O3 ⇌ NO3 + O2 NO2 + C2H2 ⇌ C2H2O + NO NO2 + CH3 ⇌ CH3O + NO NO2 + CH3 ⇌ CH2O + HNO NO2 + NO3 ⇌ NO2 + NO + O2 NO2 + CH4 ⇌ HNO2 + CH3 NO2 + CH3O ⇌ CH2O + HNO2 NO2 + C2H3O ⇌ CH3 + CO2 + NO NO2 + CH3OH ⇌ CH2OH + HNO2 NO2 + C2H5 ⇌ CH3CH2O + NO NO2 + C2H4O ⇌ C2H3O + HNO2 NO2 + C2H6 ⇌ C2H5 + HNO2 NO2 + C2H6N ⇌ CH2NCH3 + HNO2 CH3 + C2H3 ⇌ C3H5 + H CH3 + C2H3 ⇌ CH4 + C2H2 CH3 + C2H2O ⇌ C2H5 + CO CH3 + CH4 ⇌ C2H6 + H CH3 + CH4 ⇌ C2H5 + H2 CH3 + CH3O ⇌ CH2O + CH4 CH3 + CH2OH ⇌ CH2O + CH4 CH3 + C2H4 ⇌ C2H3 + CH4 CH3 + CH3CN ⇌ CH2CN + CH4 CH3 + C2H3O ⇌ C2H6 + CO CH3 + C2H3O ⇌ C2H2O + CH4 CH3 + CH3OH ⇌ CH2OH + CH4 CH3 + CH3OH ⇌ CH3O + CH4 CH3 + H4N2 ⇌ NH2NH + CH4 CH3 + C2H5 ⇌ C2H4 + CH4 CH3 + C2H4O ⇌ C2H3O + CH4 CH3 + C2H4O ⇌ CH2CHO + CH4 CH3 + C2H4O ⇌ (CH3)2CO + H CH3 + Oxyrane ⇌ cyc-C2H3O + CH4 CH3 + C2H6 ⇌ C2H5 + CH4 CH3 + CH3COOH ⇌ CH3OCO + CH4 CH3 + C2H6N ⇌ CH2NCH3 + CH4 CH3 + C2H5OH ⇌ CH3OCH2 + CH4 CH3 + C2H5OH ⇌ CH3CHOH + CH4 6.54e-13 7.63e-12 1.66e-09 1.32e-14 1.40e-13 2.09e-12 2.26e-11 5.39e-12 6.64e-12 1.16e-12 3.01e-13 1.66e-12 6.09e-13 4.34e-12 5.18e-13 2.66e-10 6.97e-14 1.20e-10 3.00e-11 8.30e-12 4.95e-13 1.66e-11 4.00e-11 4.00e-12 6.91e-12 1.66e-14 5.43e-11 1.01e-11 4.38e-15 1.12e-15 1.04e-14 1.50e-12 2.66e-12 9.96e-12 2.76e-14 1.78e-12 2.99e-10 8.32e-13 1.31e-11 1.32e-13 6.61e-13 2.69 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.34 0.00 0.00 0.00 0.00 0.00 0.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.20 3.10 4.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 9510 0 14300 19100 2740 7550 0 0 2520 15200 0 0 10800 0 13600 17000 0 0 0 0 22600 11600 0 0 5600 0 0 0 3630 3490 2040 0 4030 5530 6240 5950 7510 5180 0 4730 4880 336 266 337 338 32 Est 339 340 341 342 343 344 345 340 346 342 347 348 349 350 351 352 7 145 5 353 354 355 145 145 40 30 357 59 Est 263 358 359 360 361 361 87 # Type Reaction α β γ Ref Table 6—Continued 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 1.33e-11 CH3 + (CH3)2O ⇌ CH3OCH2 + CH4 6.03e-12 CH3 + (CH3O)2 ⇌ CH3O + CH2O + CH4 8.32e-14 H3N + CH3O ⇌ CH3OH + H2N 5.18e-16 H3N + HNO3 ⇌ H2NNO2 + H2O 8.12e-11 NO3 + CH4 ⇌ HNO3 + CH3 1.79e-12 NO3 + CH3O ⇌ CH3O2 + NO2 1.51e-12 NO3 + CH3O ⇌ CH2O + HNO3 9.40e-13 NO3 + CH3OH ⇌ CH2OH + HNO3 4.00e-11 NO3 + C2H5 ⇌ C2H4 + HNO3 1.40e-12 NO3 + C2H4O ⇌ C2H3O + HNO3 1.38e-10 NO3 + C2H6 ⇌ C2H5 + HNO3 6.99e-13 NO3 + C2H5OH ⇌ CH3CHOH + HNO3 1.40e-12 NO3 + (CH3)2O ⇌ CH3CHOH + HNO3 2.30e-12 NO3 + C2H5OO ⇌ CH3CH2O + NO2 + O2 1.79e-11 HOCH + HO ⇌ CO2 + H2 + H 1.66e-12 NH2O + H ⇌ HNO + H2 1.80e-11 NH2O + HO ⇌ HNO + H2O 1.54e-05 NH2O + CHO ⇌ NH2OH + CO 1.87e-08 NH2O + CHO ⇌ HNO + CO + H2 1.68e-05 NH2O + CHO ⇌ CH2O + HNO 1.66e-12 NH2O + HNO ⇌ NH2OH + NO NH2O + H2NNO2 ⇌ NH2OH + HNO + NO 1.66e-12 2.49e-10 HCCO + H ⇌ CH2 + CO 1.60e-10 HCCO + O ⇌ CO + CO + H 4.90e-11 HCCO + O ⇌ CO2 + CH HCCO + H2 ⇌ C2H + H2O 2.20e-11 6.38e-11 HCCO + NO ⇌ HCNO + CO 1.16e-11 HCCO + NO ⇌ HCN + CO2 2.44e-13 HCCO + O2 ⇌ CO2 + CO + H HCCO + O2 ⇌ CO + CO + HO 2.71e-13 2.58e-11 HCCO + NO2 ⇌ CHO + CO + NO 1.66e-14 HCCO + C2H2 ⇌ C3H3 + CO 6.67e-13 HNNH + H ⇌ HN2 + H2 HNNH + HO ⇌ HN2 + H2O 2.49e-13 2.27e-14 HNNH + H2N ⇌ H3N + HN2 1.00e-14 COOH + CO ⇌ CO2 + CHO 2.09e-12 COOH + O2 ⇌ CO2 + HO2 COOH + C2H2 ⇌ C2H3 + CO2 3.01e-14 1.00e-14 COOH + C2H4 ⇌ C2H5 + CO2 2.07e-12 H2CN + H ⇌ HCN + H2 H2CN + O ⇌ HCN + HO 2.07e-12 0.00 0.00 0.00 3.47 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -2.15 -1.09 -3.46 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -0.75 0.00 0.00 0.00 0.00 2.63 3.40 4.05 0.00 0.00 0.00 0.00 0.00 0.00 6290 5010 0 21700 7060 0 0 2650 0 1860 4960 1820 2500 0 0 0 0 0 757 0 1500 1500 0 0 560 2000 0 0 431 431 0 0 0 0 0 0 0 0 0 0 0 362 363 364 365 366 367 368 369 370 266 366 369 371 266 249 14 372 318 318 318 14 14 41 5 373 374 374 374 375 375 376 314 140 140 140 377 377 377 377 Est Est 88 # Type Reaction α Table 6—Continued 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 7.70e-12 H2CN + HO ⇌ HCN + H2O 9.93e-12 H2CN + N2O ⇌ CH2N2 + NO 1.00e-11 HOCH + CNO ⇌ HNCO + CHO 3.30e-12 HOCH + CH3O2 ⇌ CH4O2 + CHO 3.32e-12 C2H2 + C2H2 ⇌ C4H3 + H 1.30e-11 C2H3 + C2H3 ⇌ C4H4 + H + H 1.40e-10 C2H3 + C2H3 ⇌ C2H4 + C2H2 1.00e-10 CH3O + CH3O ⇌ CH3OH + CH2O 4.00e-11 CH3O + CH2OH ⇌ CH3OH + CH2O 8.00e-13 C2H5 + C2H3 ⇌ C2H4 + C2H4 4.47e-08 C2H4 + C2H4 ⇌ C4H6 + H2 2.41e-12 C2H5 + C2H5 ⇌ C2H6 + C2H4 4.90e-13 C2H2 + NO3 ⇌ C2H2O + NO2 2.62e-11 C2H2 + C2H3 ⇌ C4H4 + H 2.01e-13 C2H2 + CH3O ⇌ CH2O + C2H3 1.20e-12 C2H2 + CH2OH ⇌ CH2O + C2H3 8.00e-13 C2H5 + C2H3 ⇌ C2H2 + C2H6 1.00e-13 HCCN + CH4 ⇌ CH3N + C2H2 1.00e-13 HCCN + C2H4 ⇌ C2H3N + C2H2 8.07e-14 CH2O + C2H3 ⇌ C2H4 + CHO 1.69e-13 CH2O + CH3O ⇌ CH3OH + CHO 3.01e-13 CH2O + C2H3O ⇌ C2H4O + CHO 1.66e-12 HNCO + NO3 ⇌ HNO + CO2 + NO 5.45e-14 H2O2 + C2H3 ⇌ C2H4 + HO2 2.01e-15 HNO2 + NO3 ⇌ HNO3 + NO2 4.00e-11 C2H3 + CH3O ⇌ CH2O + C2H4 5.00e-11 C2H3 + CH2OH ⇌ CH2O + C2H4 2.01e-11 C2H3 + CH2OH ⇌ C3H5 + HO 8.30e-13 C2H3 + C2H4 ⇌ C4H6 + H 3.01e-11 C2H3 + C2H3O ⇌ C3H3O + CH3 4.38e-15 C2H3 + CH3OH ⇌ CH2OH + C2H4 1.12e-15 C2H3 + CH3OH ⇌ CH3O + C2H4 1.35e-13 C2H3 + C2H4O ⇌ C2H3O + C2H4 1.46e-13 C2H3 + C2H6 ⇌ C2H5 + C2H4 1.00e-11 CH3O + C2H3O ⇌ C2H2O + CH3OH 1.00e-11 CH3O + C2H3O ⇌ CH2O + C2H4O 4.00e-11 CH3O + C2H5 ⇌ CH2O + C2H6 8.30e-15 CH3O + C2H4O ⇌ CH3OH + C2H3O CH3O + C2H6 ⇌ CH3OH + C2H5 4.00e-13 CH3O + CH3COOH ⇌ CH3OH + CH3OCO 1.69e-11 C2H4 + C2H5OO ⇌ Oxyrane + CH3CH2O 3.79e-08 β 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.81 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.20 3.10 0.00 3.30 0.00 0.00 0.00 0.00 0.00 0.00 0.00 γ 0 22000 0 5790 32200 0 0 0 0 0 28800 0 2770 12600 10000 4530 0 0 0 2950 1500 6500 5030 0 0 0 0 0 3860 0 3630 3490 1850 5280 0 0 0 0 3570 4120 11000 Ref 378 379 3 7 380 381 21 7 145 7 382 5 383 384 7 145 7 386 386 7 7 7 142 7 387 7 145 145 388 7 145 145 389 7 7 7 7 390 7 391 392 89 # Type Reaction α β γ Ref Table 6—Continued 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n H4N2 + C2H5 ⇌ C2H6 + NH2NH C2H5 + C2H4O ⇌ C2H6 + C2H3O C2H5 + C2H5NO ⇌ C4H10 + NO C2H6 + C2H5OO ⇌ (CH3O)2 + C2H5 H2CN + H2CN ⇌ CH3N + HCN CH3O2 + H ⇌ CH3O + HO CH3O2 + O ⇌ CH3O + O2 CH3O2 + H2 ⇌ CH4O2 + H CH3O2 + HO ⇌ CH3OH + O2 CH3O2 + NO ⇌ CH3O + NO2 CH3O2 + O2 ⇌ HCOOH + HO2 CH3O2 + C2H ⇌ CH3O + HCCO CH3O2 + CH2 ⇌ CH3O + CH2O CH3O2 + HO2 ⇌ CH4O2 + O2 CH3O2 + CH3 ⇌ CH3O + CH3O CH3O2 + CH2O ⇌ CH4O2 + CHO CH3O2 + H2O2 ⇌ CH4O2 + HO2 CH3O2 + NO3 ⇌ CH3O + NO2 + O2 CH3O2 + C2H3 ⇌ cyc-C2H3O + CH3O CH3O2 + CH4 ⇌ CH4O2 + CH3 CH3O2 + CH3O ⇌ CH4O2 + CH2O CH3O2 + CH2OH ⇌ CH4O2 + CH2O CH3O2 + CH3OH ⇌ CH4O2 + CH2OH CH3O2 + C2H5 ⇌ CH3CH2O + CH3O CH3O2 + C2H6 ⇌ CH4O2 + C2H5 C2H2O + C2H2O ⇌ C3H4 + CO2 CH2OH + CH2OH ⇌ CH3OH + CH2O C2H3O + C2H3O ⇌ C2H2O + C2H4O CH2OH + C2H5 ⇌ CH2O + C2H6 CH2OH + C2H5 ⇌ CH3OH + C2H4 CH2OH + C2H6 ⇌ CH3OH + C2H5 C2H3O + CH3OH ⇌ C2H4O + CH2OH C2H3O + C2H4O ⇌ (CH3)2CO + CHO C2H3O + C2H6N ⇌ C2H4O + CH2NCH3 CH3NO + CH3O ⇌ C2H5NO2 + H NH2OH + HO ⇌ H2O + NH2O CH3O2 + CH3O2 ⇌ CH3O + CH3O + O2 CH3O2 + CH3O2 ⇌ (CH3O)2 + O2 CH3O2 + CH3O2 ⇌ CH3OH + CH2O + O2 CH3OCO + CH3O ⇌ CH3COOH + CH2O C2H6O2 + O ⇌ CH3CH2O + HO2 8.32e-14 2.09e-12 1.66e-10 2.87e-14 7.70e-12 1.60e-10 8.30e-11 5.00e-11 1.00e-10 7.51e-12 3.50e-14 4.00e-11 3.01e-11 1.57e-11 4.00e-11 3.30e-12 4.00e-12 3.11e-12 4.00e-11 3.01e-13 5.00e-13 2.01e-11 3.01e-12 4.00e-11 4.90e-13 1.83e-11 8.00e-12 1.49e-11 4.00e-12 4.00e-12 8.73e-15 2.13e-13 2.84e-13 5.98e-11 1.00e-10 4.13e-11 7.40e-13 3.01e-14 2.22e-13 3.79e-11 8.80e-12 0.00 0.00 0.00 3.76 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 3.00 3.00 0.00 0.00 0.50 0.00 0.00 0.00 0.00 0.00 0.57 2300 4280 0 8650 0 0 0 13100 0 0 0 0 0 0 0 5870 5000 0 0 9300 0 0 6900 0 7520 18000 0 0 0 0 7030 6210 0 0 2340 2140 519 0 0 0 1390 393 394 395 396 397 7 398 7 7 399 400 7 7 7 7 7 7 368 7 7 7 Est 145 7 7 401 145 355 145 145 145 145 402 360 403 257 266 Est Est 404 5 90 # Type Reaction α β γ Ref Table 6—Continued 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 8.80e-12 C2H6O2 + O ⇌ CH3CHOH + HO2 2.13e-13 CH2CHO + CH3OH ⇌ C2H4O + CH2OH 2.84e-13 CH2CHO + C2H4O ⇌ (CH3)2CO + CHO 1.91e-13 CH2CHO + C2H6 ⇌ C2H4O + C2H5 5.98e-11 CH2CHO + C2H6N ⇌ C2H4O + CH2NCH3 3.01e-12 (CH3)2N2O + HO ⇌ C2H6 + HO2 + N2 3.32e-11 CH3OCH2 + H ⇌ C2H4O + H2 3.03e-09 CH3OCH2 + O ⇌ CH3COOH + H 1.50e-10 CH3OCH2 + O ⇌ CH3O + CH2O 3.16e-10 CH3OCH2 + O ⇌ C2H4O + HO 1.30e-11 CH3OCH2 + NO ⇌ C2H4O + HNO 1.11e-13 CH3OCH2 + O2 ⇌ CH2O + CHO + H2O 8.30e-12 CH3OCH2 + O2 ⇌ CH2O + CH2O + HO 6.61e-12 CH3OCH2 + NO2 ⇌ C2H4O + HNO2 3.30e-12 CH3OCH2 + NO3 ⇌ C2H5OO + NO2 6.46e-08 CH3OCH2 + C2H5 ⇌ C2H4O + C2H6 1.14e-07 CH3OCH2 + C2H5 ⇌ C2H5OH + C2H4 3.32e-11 CH3CHOH + H ⇌ C2H4O + H2 3.16e-10 CH3CHOH + O ⇌ C2H4O + HO 2.41e-11 CH3CHOH + NO ⇌ C2H4O + HNO 1.90e-11 CH3CHOH + O2 ⇌ C2H4O + HO2 9.61e-14 CH2NCH3 + CH3 ⇌ CH3N + CH4 + CH 1.30e-11 CH3CH2O + NO ⇌ C2H4O + HNO 7.11e-14 CH3CH2O + O2 ⇌ C2H4O + HO2 6.61e-12 CH3CH2O + NO2 ⇌ C2H4O + HNO2 3.30e-12 CH3CH2O + NO3 ⇌ C2H5OO + NO2 6.46e-08 CH3CH2O + C2H5 ⇌ C2H6 + C2H4O CH3CH2O + C2H5 ⇌ C2H4 + C2H5OH 1.14e-07 CH3CH2O + CH3CH2O ⇌ C2H5OH + C2H4O 6.72e-10 O(1D) + O2 ⇌ O + O2 4.00e-11 O(1D) + O3 ⇌ O2 + O2 1.20e-10 O(1D) + O3 ⇌ O2 + O + O 1.20e-10 O(1D) + H2 ⇌ HO + H 1.20e-10 O(1D) + H2O ⇌ HO + HO 2.00e-10 O(1D) + N2 ⇌ O + N2 3.10e-11 O(1D) + N2O ⇌ N2 + O2 4.95e-11 O(1D) + N2O ⇌ NO + NO 7.25e-11 O(1D) + H3N ⇌ H2N + HO 2.50e-10 O(1D) + CO2 ⇌ O + CO2 1.10e-10 O(1D) + CH4 ⇌ CH3 + HO 1.31e-10 O(1D) + CH4 ⇌ CH3O + H 3.50e-11 0.57 3.00 0.00 2.75 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1.00 1.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 1390 6210 0 8820 0 0 0 0 0 0 0 388 1060 0 0 0 0 0 0 0 2800 2660 0 552 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 5 145 402 7 360 405 406 407 407 406 54 408 409 410 411 Est Est 406 406 410 412 413 54 414 410 411 415 415 416 169 169 169 169 169 169 169 169 169 169 169 169 91 Table 6—Continued # Type Reaction α β γ Ref 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2n 2i 2 + H ⇌ CH + H2 2 + H2 ⇌ CH3 + H 2 + H2 ⇌ CH2 + H2 2 + O2 ⇌ HO + CO + H 2 + O2 ⇌ H2O + CO O(1D) + CH4 ⇌ CH2O + H2 9.00e-12 O(1D) + CH4 ⇌ CH2O + H2 9.00e-12 O(1S) + O2 ⇌ O + O2 2.85e-13 O(1S) + O3 ⇌ O2 + O2 2.30e-10 O(1S) + O3 ⇌ O2 + O + O 2.30e-10 O(1S) + H2O ⇌ HO + HO 6.38e-10 O(1S) + CO2 ⇌ O + CO2 3.09e-13 O(1S) + H2 ⇌ HO + H 3.60e-10 O(1S) + N2 ⇌ O + N2 3.09e-13 C(1D) + H2 ⇌ CH + H 9.50e-11 C(1S) + H2 ⇌ CH + H 2.10e-10 O2(a1∆g) + O3 ⇌ O2 + O2 + O 5.20e-11 O2(a1∆g) + H2O ⇌ O2 + H2O 1.00e-11 CH∗ 3.00e-10 CH∗ 1.26e-11 CH∗ 9.24e-11 CH∗ 4.60e-11 CH∗ 2.00e-11 CH3 + HO ⇌ CH∗ 2 + H2O 1.61e-13 CH∗ 2 + H2O ⇌ CH2 + H2O 5.00e-11 CH∗ 2 + H2O ⇌ CH2 + H2O 5.00e-11 CH∗ 2 + CO ⇌ CH2 + CO 5.00e-11 CH∗ 2 + N2 ⇌ HN + HCN 1.66e-13 CH∗ 2 + NO ⇌ HCNO + H 6.43e-12 CH∗ 8.86e-12 2 + NO ⇌ HCN + HO CH∗ 2 + CH3 ⇌ C2H4 + H 7.00e-11 3.00e-11 CH2 + C3H3 ⇌ C4H4 + H 5.05e-11 C2 + CH4 ⇌ C2H + CH3 1.30e-10 C2H + C2H2 ⇌ C4H2 + H C4H + H2 ⇌ C4H2 + H 2.18e-12 1.20e-11 C4H + CH4 ⇌ C4H2 + CH3 4.72e-11 C4H + C2H6 ⇌ C4H2 + C2H5 2.38e-12 NaH + H ⇌ Na + H2 Na + H2O ⇌ NaOH + H 4.07e-10 2.38e-12 KH + H ⇌ K + H2 5.00e-10 K + H2O ⇌ KOH + H 6.84e-19 H + HCl ⇌ H2 + Cl HCl + HO ⇌ H2O + Cl 5.89e-18 5.60e-10 K + HCl ⇌ KCl + H 4.00e-10 Na + HCl ⇌ NaCl + H Ar+ + CH4 ⇌ CH+ 9.00e-10 2 + H2 + Ar 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 2.57 0.00 0.00 0.00 0.00 -0.40 -0.70 0.00 0.00 0.00 0.00 2.17 0.00 0.54 0.69 0.00 0.69 0.00 0.00 2.12 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 2840 0 0 0 0 0 0 2010 0 0 0 32700 292 382 0 0 297 0 478 491 0 2360 21900 2360 20000 1730 0 4170 4090 0 169 169 169 169 169 169 169 169 169 169 169 169 169 21 21 21 21 21 21 21 21 21 21 21 21 385 21 685 686 385 385 385 160 417 100 417 418 419 420 421 422 92 # Type Reaction α β γ Ref Table 6—Continued 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 5 + H + Ar 4 + H2 + Ar 3 + H2 + H + Ar 2 + H2 + H2 + Ar 2 + Ar 2 + Ar Ar+ + CH4 ⇌ CH+ 3 + H + Ar Ar+ + CH4 ⇌ CH+ 4 + Ar Ar+ + CO ⇌ CO+ + Ar Ar+ + CO2 ⇌ CO+ 2 + Ar Ar+ + C2H6 ⇌ C2H+ Ar+ + C2H6 ⇌ C2H+ Ar+ + C2H6 ⇌ C2H+ Ar+ + C2H6 ⇌ C2H+ Ar+ + H2 ⇌ ArH+ + H Ar+ + H2 ⇌ H+ Ar+ + H2O ⇌ ArH+ + HO Ar+ + H2O ⇌ H2O+ + Ar Ar+ + H3N ⇌ H3N+ + Ar Ar+ + H3N ⇌ H2N+ + H + Ar Ar+ + H3N ⇌ ArH+ + H2N Ar+ + NO ⇌ NO+ + Ar Ar+ + NO2 ⇌ NO+ 2 + Ar Ar+ + NO2 ⇌ NO+ + O + Ar Ar+ + N2 ⇌ N+ Ar+ + N2O ⇌ N2O+ + Ar Ar+ + O2 ⇌ O+ 2 + Ar ArH+ + CH4 ⇌ CH+ 5 + Ar ArH+ + CO ⇌ CHO+ + Ar ArH+ + CO2 ⇌ CHO+ 2 + Ar ArH+ + H2 ⇌ H+ 3 + Ar ArH+ + H2O ⇌ H3O+ + Ar ArH+ + N2 ⇌ HN+ ArH+ + O2 ⇌ HO+ ArH+ ArH+ ArH+ ArH+ ArH+ ArH+ ArH+ Ar+ Ar+ Ar+ Ar+ Ar+ Ar+ 2 + CO ⇌ CO+ + Ar + Ar 2 + CO2 ⇌ CO+ 2 + Ar + Ar 2 + H2 ⇌ Ar2H+ + H 2 + H2 ⇌ ArH+ + H + Ar 2 + H3N ⇌ H3N+ + Ar + Ar 2 + NO ⇌ NO+ + Ar + Ar 2 + Ar 2 + Ar 3 + CO ⇌ CHO+ + H2 + Ar 3 + CO2 ⇌ CHO+ 2 + H2 + Ar 3 + C2H4 ⇌ C2H+ 3 + H2 + H2 + Ar 3 + C2H4 ⇌ C2H+ 5 + H2 + Ar 3 + NO ⇌ HNO+ + H2 + Ar 3 + N2 ⇌ HN+ 3 + N2O ⇌ HN2O+ + H2 + Ar 2 + H2 + Ar 9.00e-10 9.00e-10 5.00e-11 5.00e-10 5.00e-10 5.00e-10 5.00e-10 5.00e-10 2.20e-09 1.00e-09 1.30e-09 1.50e-10 1.69e-09 5.52e-11 9.20e-11 4.00e-10 1.30e-10 1.30e-10 6.50e-10 3.10e-10 6.00e-11 1.00e-9 1.25e-9 1.10e-9 1.50e-9 4.50e-09 8.00e-10 4.10e-10 1.20e-09 1.20e-09 1.75e-09 7.50e-10 1.60e-09 8.50e-10 1.80e-09 8.50e-10 1.10e-09 4.90e-10 4.70e-10 4.50e-10 2.40e-11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 2320 0 0 0 0 0 0 464 464 464 1740 1740 464 580 580 580 0 0 580 0 0 0 0 0 0 0 0 0 0 0 0 0 0 422 422 423 424 425 425 425 425 426 426 427 427 428 428 428 429 430 430 431 426 429 432 432 432 433 434 432 435 436 436 436 436 436 436 436 437 437 438 439 430 437 93 Table 6—Continued # Type Reaction α β γ Ref 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + Ar + Ar 2 + Ar + Ar 3 + Ar 2 + CO 3 + HCN + H 3 + H 2 + H2 2 + NO2 ⇌ NO+ 2 + NO2 ⇌ NO+ + O + Ar + Ar 2 + N2O ⇌ N2O+ + Ar + Ar 2 + O2 ⇌ O+ Ar+ Ar+ Ar+ Ar+ Ar2H+ + H2 ⇌ ArH+ C+ + HCN ⇌ C2N+ + H C+ + CH2O ⇌ CH+ C+ + CH2O ⇌ CHO+ + CH C+ + CH2O ⇌ CH2O+ + C C+ + HCOOH ⇌ CHO+ + CHO C+ + CH4 ⇌ C2H+ C+ + CH4 ⇌ C2H+ C+ + CH3OH ⇌ CH4O+ + C C+ + CH3OH ⇌ CH3O+ + CH C+ + CH3OH ⇌ CH+ 3 + CHO C+ + CH3OH ⇌ CHO+ + CH3 C+ + CH5N ⇌ CH5N+ + C C+ + CH5N ⇌ CH4N+ + CH C+ + CH5N ⇌ CH2N+ + CH3 C+ + CH5N ⇌ CH+ C+ + CO2 ⇌ CO+ + CO C+ + C2H2 ⇌ C3H+ + H C+ + C2H4 ⇌ C3H+ 2 + H2 C+ + C2H4 ⇌ C3H+ 3 + H C+ + C2H4 ⇌ C2H+ 4 + C C+ + C2H4 ⇌ C2H+ 3 + CH C+ + C2H6 ⇌ C3H+ 3 + H2 + H C+ + C2H6 ⇌ C2H+ 3 + CH3 C+ + C2H6 ⇌ C2H+ 4 + CH2 C+ + C2H6 ⇌ C2H+ 5 + CH C+ + NCCN ⇌ C2N+ + CN C+ + H2 ⇌ CH+ + H C+ + H2O ⇌ CHO+ + H C+ + H3N ⇌ H3N+ + C C+ + H3N ⇌ CH2N+ + H C+ + H3N ⇌ CHN+ + H2 C+ + SiH4 ⇌ Si+ + H2 + H2 + C C+ + SiH4 ⇌ SiH+ + H2 + H + C C+ + SiH4 ⇌ SiH+ C+ + SiH4 ⇌ SiH+ C+ + SiH4 ⇌ SiH+ 2 + H2 + C 3 + H + C 4 + C 2.80e-10 2.80e-10 8.20e-10 7.40e-11 1.20e-10 1.30e-09 2.34e-09 7.80e-10 7.80e-10 3.30e-09 1.03e-09 4.21e-10 1.35e-09 1.23e-09 1.19e-09 3.28e-10 3.07e-09 7.98e-10 2.52e-10 1.26e-10 1.60e-09 2.20e-09 5.20e-10 3.90e-10 1.95e-10 1.95e-10 4.80e-10 4.80e-10 3.20e-10 3.20e-10 1.90e-09 2.20e-12 2.70e-09 1.15e-09 1.08e-09 6.90e-11 4.36e-09 4.36e-09 4.36e-09 4.36e-09 4.36e-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4480 0 0 0 0 0 0 0 0 0 430 430 430 430 440 441 442 442 442 443 442 442 442 442 442 442 442 442 442 442 444 445 445 445 445 445 445 445 445 445 446 Est 442 442 442 442 447 447 447 447 447 94 Table 6—Continued # Type Reaction α β γ Ref 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 3 + H2 2 + H2 + H 4 + H 3 + CH2O 2 + H 2 + H2 C+ + SiH4 ⇌ SiCH+ + H2 + H 4.36e-09 C+ + SiH4 ⇌ SiCH+ 4.36e-09 C+ + NO ⇌ NO+ + C 8.50e-10 C+ + N2O ⇌ NO+ + CN 9.10e-10 C+ + O2 ⇌ CO+ + O 7.81e-10 C+ + O2 ⇌ O+ + CO 4.39e-10 CH+ + HCN ⇌ C2N+ + H2 4.29e-10 CH+ + HCN ⇌ C2HN+ + H 2.31e-10 CH+ + HCN ⇌ CH2N+ + C 2.80e-09 CH+ + CH2O ⇌ CHO+ + CH2 9.60e-10 CH+ + CH2O ⇌ CH+ 9.60e-10 3 + CO CH+ + CH2O ⇌ CH3O+ + C 9.60e-10 CH+ + CH2O ⇌ C2H2O+ + H 3.20e-10 CH+ + CH4 ⇌ C2H+ 1.09e-09 CH+ + CH4 ⇌ C2H+ 1.43e-10 CH+ + CH4 ⇌ C2H+ 6.50e-11 CH+ + CH3OH ⇌ CH+ 1.45e-09 CH+ + CH3OH ⇌ CH5O+ + C 1.16e-09 CH+ + CH3OH ⇌ CH3O+ + CH2 2.90e-10 CH+ + CH5N ⇌ CH4N+ + CH2 1.10e-09 CH+ + CH5N ⇌ CH6N+ + C 8.80e-10 CH+ + CH5N ⇌ CH5N+ + CH 2.20e-10 CH+ + CO ⇌ CHO+ + C 7.00e-12 CH+ + CO2 ⇌ CHO+ + CO 1.60e-09 CH+ + C2H2 ⇌ C3H+ 2.40e-09 CH+ + H2 ⇌ CH+ 1.20e-09 2 + H CH+ + H2O ⇌ CHO+ + H2 1.35e-09 CH+ + H2O ⇌ CH2O+ + H 6.75e-10 CH+ + H2O ⇌ H3O+ + C 6.75e-10 CH+ + H3N ⇌ CH2N+ + H2 1.84e-09 CH+ + H3N ⇌ H3N+ + CH 4.59e-10 CH+ + H3N ⇌ H4N+ + C 4.05e-10 CH+ + SiH4 ⇌ Si+ + H2 + H2 + CH 4.56e-09 CH+ + SiH4 ⇌ SiH+ + H2 + CH + H 4.56e-09 CH+ + SiH4 ⇌ SiH+ 4.56e-09 CH+ + SiH4 ⇌ SiH+ 4.56e-09 CH+ + SiH4 ⇌ SiCH+ 4.56e-09 CH+ + SiH4 ⇌ SiCH+ 4.56e-09 CH+ + N ⇌ CN+ + H 1.90e-10 CH+ + N ⇌ H+ + CN 1.90e-10 CH+ + NO ⇌ NO+ + CH 7.60e-10 2 + H2 + CH 3 + CH + H 2 + H2 + H 3 + H2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 447 447 442 448 442 442 441 441 441 449 449 449 449 107 107 107 449 449 449 449 449 449 450 450 451 450 450 450 450 452 452 452 447 447 447 447 447 447 453 453 453 95 # Type Reaction α β γ Ref Table 6—Continued 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H 3 + H2N CH+ + O ⇌ CO+ + H 3.50e-10 CH+ + O ⇌ H+ + CO 3.50e-10 CH+ + O2 ⇌ CHO+ + O 4.85e-10 CH+ + O2 ⇌ CO+ + HO 2.43e-10 CH+ + O2 ⇌ O+ + CHO 2.43e-10 CHN+ + HCN ⇌ CH2N+ + CN 1.20e-09 CHN+ + HCN ⇌ C2HN+ 3.20e-11 CHN+ + CH4 ⇌ CH2N+ + CH3 1.09e-09 CHN+ + CH4 ⇌ C2H+ 2.08e-10 CHN+ + CH5N ⇌ CH6N+ + CN 1.50e-09 CHN+ + CO ⇌ CHO+ + CN 1.40e-10 CHN+ + CO2 ⇌ CHO+ 2.10e-10 2 + CN CHN+ + H2 ⇌ CH2N+ + H 9.81e-10 CHN+ + H2O ⇌ H3O+ + CN 1.80e-09 CHN+ + H2O ⇌ H2O+ + HCN 1.80e-09 CHN+ + H2O ⇌ CH2N+ + HO 1.44e-10 CHN+ + H3N ⇌ H3N+ + HCN 1.89e-09 CHN+ + H3N ⇌ CH2N+ + H2N 8.40e-10 CHN+ + H3N ⇌ H4N+ + CN 1.40e-10 CHN+ + O2 ⇌ O+ 3.20e-10 CHNO+ + HNCO ⇌ CH2NO+ + CNO 8.30e-10 CHO+ + Ar ⇌ ArH+ + CO 3.90e-10 CHO+ + HCN ⇌ CH2N+ + CO 4.00e-09 CHO+ + HNCO ⇌ CH2NO+ + CO 1.34e-09 CHO+ + CH2O ⇌ CH3O+ + CO 6.10e-10 CHO+ + HCOOH ⇌ CH3O+ 1.80e-09 CHO+ + CH3NO2 ⇌ CH4NO+ 3.33e-09 CHO+ + CH3NO2 ⇌ NO+ + CH3OH + CO 3.33e-09 CHO+ + CH3OH ⇌ CH5O+ + CO 2.40e-09 CHO+ + CO2 ⇌ CHO+ 1.25e-09 CHO+ + C2H2 ⇌ C2H+ 1.36e-09 CHO+ + C2H2O ⇌ C2H3O+ + CO 1.80e-09 CHO+ + CH3CN ⇌ C2H4N+ + CO 4.10e-09 CHO+ + C2H4O ⇌ C2H5O+ + CO 3.69e-09 CHO+ + CH3COOH ⇌ C2H5O+ 2.50e-09 CHO+ + C2H6 ⇌ C2H+ 1.20e-10 CHO+ + C2H5OH ⇌ C2H7O+ + CO 2.20e-09 CHO+ + (CH3)2O ⇌ C2H7O+ + CO 2.10e-09 CHO+ + HC3N ⇌ C3H2N+ + CO 4.00e-09 CHO+ + H2O ⇌ H3O+ + CO 2.50e-09 CHO+ + H3N ⇌ H4N+ + CO 1.12e-09 2 + CO 3 + CO 2 + HCN 2 + CO 2 + CO 2 + CO 7 + CO 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 453 453 450 450 450 441 441 454 454 455 456 456 457 456 456 456 456 456 456 456 458 459 457 458 460 461 462 462 463 459 464 461 465 466 461 467 461 468 469 463 470 96 # Type Reaction α β γ Ref Table 6—Continued 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + CO 2 + CO2 5 + CO2 3 + CO2 4 + H2 5 + H CHO+ + N2O ⇌ HN2O+ + CO CHO+ + O2 ⇌ HO+ CHO+ CHO+ CHO+ CHO+ CHO+ CHO+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH+ CH2N+ + CH2O ⇌ CH3O+ + HCN CH2N+ + CH3NO2 ⇌ CH4NO+ 2 + CH3NO2 ⇌ CH4NO+ 2 + CH3NO2 ⇌ NO+ + CH3OH + CO2 2 + CH4 ⇌ CH+ 2 + C2H2 ⇌ C2H+ 2 + CH3CN ⇌ C2H4N+ + CO2 2 + H2O ⇌ H3O+ + CO2 2 + HCN ⇌ C2H2N+ + H 2 + CH2O ⇌ CH+ 3 + CHO 2 + CH2O ⇌ CHO+ + CH3 2 + CH2O ⇌ C2H3O+ + H 2 + CH2O ⇌ C2H2O+ + H2 2 + CH4 ⇌ C2H+ 2 + CH4 ⇌ C2H+ 2 + CH3OH ⇌ CH3O+ + CH3 2 + CH3OH ⇌ CH5O+ + CH 2 + CH5N ⇌ CH4N+ + CH3 2 + CH5N ⇌ CH5N+ + CH2 2 + CH5N ⇌ CH6N+ + CH 2 + CO ⇌ CHO+ + CH 2 + CO2 ⇌ CH2O+ + CO 2 + C2H2 ⇌ C3H+ 2 + C2H2O ⇌ C2H2O+ + CH2 2 + H2 ⇌ CH+ 2 + H2O ⇌ CH3O+ + H 2 + H2O ⇌ H3O+ + CH 2 + H3N ⇌ CH4N+ + H 2 + H3N ⇌ H4N+ + CH 2 + SiH4 ⇌ SiH+ + CH + H2 + H2 2 + SiH4 ⇌ SiH+ 2 + CH + H2 + H 2 + SiH4 ⇌ SiH+ 3 + CH + H2 2 + SiH4 ⇌ SiCH+ 2 + H2 + H2 2 + SiH4 ⇌ SiCH+ 3 + H2 + H 2 + N ⇌ CN+ + H2 2 + N ⇌ CHN+ + H 2 + NO ⇌ NO+ + CH2 2 + O2 ⇌ CHO+ + HO 2 + O2 ⇌ CH2O+ + O 3 + H 3 + H 2 + HCN 97 2.70e-12 4.80e-10 2.81e-09 8.46e-12 1.30e-10 1.37e-09 4.10e-09 3.00e-09 1.80e-09 6.40e-10 2.80e-09 3.30e-10 1.65e-10 8.40e-10 3.60e-10 1.30e-09 1.30e-09 1.16e-09 7.35e-10 2.10e-10 4.99e-12 1.60e-09 2.50e-09 3.80e-10 1.60e-09 2.90e-09 2.90e-09 1.54e-09 1.26e-09 3.49e-09 3.49e-09 3.49e-09 3.49e-09 3.49e-09 2.20e-10 2.20e-10 4.20e-10 9.10e-10 9.10e-10 2.10e-09 3.75e-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -2.50 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 435 459 462 462 471 464 465 472 473 460 449 449 449 107 107 449 449 449 449 449 104 450 473 474 450 450 450 452 452 447 447 447 447 447 453 453 453 450 450 475 462 Table 6—Continued # Type Reaction α β γ Ref 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + CO 5 + CH2 5 + H2 3 + H2 CH2N+ + CH5N ⇌ CH6N+ + HNC 5.00e-10 CH2N+ + H3N ⇌ H4N+ + HCN 2.40e-09 CH2O+ + CH2O ⇌ CH3O+ + CHO 1.65e-09 CH2O+ + CH2O ⇌ CH4O+ + CO 1.24e-10 CH2O+ + CH4 ⇌ CH3O+ + CH3 9.35e-11 CH2O+ + CH4 ⇌ C2H5O+ + H 1.65e-11 CH2O+ + CH3OH ⇌ CH5O+ + CHO 2.16e-09 CH2O+ + CH3OH ⇌ CH3O+ + CH3O 2.40e-10 CH2O+ + H2O ⇌ H3O+ + CHO 2.60e-09 CH2O+ + H3N ⇌ H4N+ + CHO 1.28e-09 CH2O+ + H3N ⇌ H3N+ + CH2O 4.25e-10 CH2O+ + O2 ⇌ CHO+ + HO2 7.70e-11 CH2O+ + O2 ⇌ H2O+ 3.30e-11 CH+ 1.60e-09 CH+ 1.50e-09 CH+ 1.00e-09 CH+ 2.30e-09 CH+ 1.21e-09 CH+ 9.90e-10 CH+ 1.20e-09 CH+ 2.10e-10 CH+ 5.61e-10 CH+ 3.08e-10 CH+ 1.98e-10 CH+ 3.30e-11 CH+ 3.30e-11 CH+ 1.68e-11 CH+ 1.98e-09 CH+ 2.52e-09 CH+ 5.00e-13 CH+ 9.99e-12 CH+ 2.00e-09 CH+ 1.66e-10 CH+ 2.39e-09 CH+ 2.39e-09 CH+ 2.39e-09 CH+ 6.70e-11 CH+ 1.00e-09 CH+ 3.08e-10 CH+ 8.80e-11 CH+ 4.40e-11 3 + CH2O ⇌ CHO+ + CH4 3 + CH4 ⇌ CH+ 3 + CH4 ⇌ C2H+ 3 + CH3OH ⇌ CH3O+ + CH4 3 + CH5N ⇌ CH5N+ + CH3 3 + CH5N ⇌ CH4N+ + CH4 3 + C2H2 ⇌ C3H+ 3 + C2H4O ⇌ C2H3O+ + CH4 3 + C2H4O ⇌ CH3O+ + C2H4 3 + C2H4O ⇌ C2H+ 3 + CH3OH 3 + C2H4O ⇌ C2H+ 4 + CH3O 3 + C2H4O ⇌ C3H5O+ + H2 3 + C2H4O ⇌ C3H5O+ + H2 3 + C2H4O ⇌ C3H6O+ + H 3 + (CH3)2O ⇌ C2H5O+ + CH4 3 + HC3N ⇌ C3H+ 3 + H2 ⇌ CH+ 4 + H 3 + H2O ⇌ CH2OH+ + H2 3 + H3N ⇌ CH4N+ + H2 3 + H3N ⇌ H4N+ + CH2 3 + SiH4 ⇌ SiH+ 3 + CH4 3 + SiH4 ⇌ SiCH+ 3 + SiH4 ⇌ SiCH+ 3 + N ⇌ CH2N+ + H 3 + NO ⇌ NO+ + CH3 3 + O ⇌ CHO+ + H2 3 + O ⇌ H+ 3 + O ⇌ CH2O+ + H 3 + H2 + H2 5 + H2 3 + HCN 3 + CO 98 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 455 457 460 460 463 463 463 463 463 463 463 463 463 449 476 476 449 477 477 477 478 478 478 478 478 478 478 Est 479 480 481 482 482 447 447 447 483 453 483 483 483 # Type Reaction α β γ Ref Table 6—Continued 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 5 + CH3 CH3O+ + HCN ⇌ CH2N+ + CH2O 1.30e-09 CH3O+ + CH2O ⇌ C2H2O+ + H2O + H 1.81e-12 CH3O+ + CH2O ⇌ C2H3O+ + H2O 1.81e-12 CH3O+ + CH2O ⇌ C2H5O+ + O 1.81e-12 CH3O+ + CH2O ⇌ CH5O+ + CO 1.21e-12 CH3O+ + H2O ⇌ H3O+ + CH2O 3.00e-11 CH3O+ + H3N ⇌ H4N+ + CH2O 8.57e-10 CH+ 3.23e-09 CH+ 6.60e-11 CH+ 1.98e-09 CH+ 1.62e-09 CH+ 1.20e-09 CH+ 1.80e-09 CH+ 1.20e-09 CH+ 1.32e-09 CH+ 8.80e-10 CH+ 1.40e-09 CH+ 6.08e-10 CH+ 1.20e-09 CH+ 9.60e-12 CH+ 1.25e-09 CH+ 1.13e-09 CH+ 1.25e-10 CH+ 2.90e-09 CH+ 2.90e-09 CH+ 2.90e-09 CH+ 2.90e-09 CH+ 3.30e-11 CH+ 3.33e-09 CH+ 1.75e-10 CH+ 1.65e-09 CH+ 1.15e-09 CH+ 6.32e-11 CH+ 2.04e-09 CH+ 2.04e-09 CH+ 2.04e-09 CH+ 2.04e-09 CH+ 2.80e-10 CH+ 1.01e-09 CH+ 3.12e-11 CH+ 4.40e-10 4 + HCN ⇌ CH2N+ + CH3 4 + HCN ⇌ C2H4N+ + H 4 + CH2O ⇌ CH3O+ + CH3 4 + CH2O ⇌ CH2O+ + CH4 4 + CH4 ⇌ CH+ 4 + CH3OH ⇌ CH4O+ + CH4 4 + CH3OH ⇌ CH5O+ + CH3 4 + CH5N ⇌ CH5N+ + CH4 4 + CH5N ⇌ CH4N+ + CH4 + H 4 + CO ⇌ CHO+ + CH3 4 + CO ⇌ C2H3O+ + H 4 + CO2 ⇌ CHO+ 2 + CH3 4 + CO2 ⇌ C2H3O+ + HO 4 + C2H2 ⇌ C2H+ 3 + CH3 4 + C2H2 ⇌ C2H+ 2 + CH4 4 + C2H2 ⇌ C3H+ 3 + H2 + H 4 + C2H4 ⇌ C2H+ 5 + CH3 4 + C2H4 ⇌ C2H+ 4 + CH4 4 + C2H4 ⇌ C3H+ 6 + H2 4 + C2H4 ⇌ C3H+ 7 + H 4 + H2 ⇌ CH+ 5 + H 4 + H2O ⇌ H3O+ + CH3 4 + H2O ⇌ CH5O+ + H 4 + H3N ⇌ H3N+ + CH4 4 + H3N ⇌ H4N+ + CH3 4 + H3N ⇌ CH+ 5 + H2N 4 + SiH4 ⇌ SiH+ 2 + CH4 + H2 4 + SiH4 ⇌ SiH+ 3 + CH4 + H 4 + SiH4 ⇌ SiCH+ 4 + H2 + H2 4 + SiH4 ⇌ SiCH+ 5 + H2 + H 4 + NO ⇌ NO+ + CH4 4 + N2O ⇌ HN2O+ + CH3 4 + N2O ⇌ HNO+ + CH3N 4 + O2 ⇌ O+ 2 + CH4 99 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 475 460 460 460 460 463 470 473 473 449 449 484 449 449 449 449 485 485 485 485 473 473 473 486 486 486 486 450 427 427 485 485 485 447 447 447 447 487 488 488 450 # Type Reaction α β γ Ref Table 6—Continued 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + CH4 3 + CH4 + H2 2 + CH4 2 + CH4 2 + HC3N ⇌ C3H2N+ + CH3NO2 CH4N+ + CH5N ⇌ CH6N+ + CH3N CH4N+ + CH5N ⇌ CH6N+ + CH3N CH4NO+ CH4O+ + CH3OH ⇌ CH5O+ + CH3O CH4O+ + H3N ⇌ H4N+ + CH3O CH+ 5 + CH2O ⇌ CH3O+ + CH4 CH+ 5 + HCOOH ⇌ CH3O+ CH+ 5 + CH3NO2 ⇌ CH4NO+ CH+ 5 + CH3NO2 ⇌ NO+ + CH3OH + CH4 CH+ 5 + CO ⇌ CHO+ + CH4 CH+ 5 + CO2 ⇌ CHO+ 2 + CH4 5 + C2H2 ⇌ C2H+ CH+ 3 + CH4 CH+ 5 + C2H4 ⇌ C2H+ 5 + CH4 5 + CH3OCHO ⇌ C2H5O+ CH+ CH+ 5 + C2H6 ⇌ C2H+ 7 + CH4 CH+ 5 + C2H6 ⇌ C2H+ 5 + CH4 + H2 CH+ 5 + H2O ⇌ H3O+ + CH4 CH+ 5 + H3N ⇌ H4N+ + CH4 5 + SiH4 ⇌ SiH+ CH+ CH+ 5 + NO ⇌ HNO+ + CH4 CH+ 5 + N2O ⇌ HN2O+ + CH4 CH+ 5 + O ⇌ H3O+ + CH2 CH+ 5 + O ⇌ CH3O+ + H2 CH5O+ + HCN ⇌ C2H4N+ + H2O CH5O+ + CH2O ⇌ C2H5O+ + H2O CH5O+ + CH3NO2 ⇌ CH4NO+ CH5O+ + CH3OH ⇌ C2H7O+ + H2O CN+ + HCN ⇌ CHN+ + CN CN+ + HCN ⇌ C2N+ CN+ + HCOOH ⇌ CHO+ + HCNO CN+ + CH4 ⇌ CH+ CN+ + CH4 ⇌ CH+ CN+ + CH4 ⇌ CHN+ + CH3 CN+ + CH4 ⇌ CH2N+ + CH2 CN+ + CH4 ⇌ C2H2N+ + H2 CN+ + CH3OH ⇌ CH3O+ + HCN CN+ + CH3OH ⇌ CH4O+ + CN CN+ + CH3OH ⇌ CH2N+ + CH2O CN+ + CO ⇌ CO+ + CN CN+ + CO2 ⇌ CO+ 2 + CN CN+ + CO2 ⇌ CNO+ + CO 4.00e-10 4.00e-10 9.00e-10 2.50e-09 3.40e-09 4.50e-09 2.90e-09 4.13e-09 2.89e-12 9.90e-10 3.30e-11 1.56e-09 1.00e-09 4.10e-09 1.28e-09 2.25e-10 3.70e-09 2.30e-09 1.99e-09 9.99e-12 9.50e-10 2.16e-10 4.40e-12 2.30e-11 2.10e-11 2 + CH3OH 1.32e-09 1.08e-10 1.79e-09 3.15e-10 5.30e-09 4.50e-10 1.35e-10 1.35e-10 9.00e-11 9.00e-11 1.56e-09 7.80e-10 2.60e-10 6.30e-10 3.00e-10 2.25e-10 2 + H 3 + HCN 4 + CN 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 455 455 469 489 470 490 491 462 462 492 493 464 494 495 467 467 472 496 447 433 497 453 453 498 499 462 499 500 500 443 500 500 500 500 500 500 500 500 501 500 500 100 Table 6—Continued # Type Reaction α β γ Ref 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i CN+ + CO2 ⇌ C2O+ + NO CN+ + C2H4 ⇌ C2H+ 4 + CN CN+ + C2H4 ⇌ CHN+ + C2H3 CN+ + C2H4 ⇌ C2H+ 3 + HCN CN+ + C2H4 ⇌ CH2N+ + C2H2 CN+ + C2H6 ⇌ C2H+ 4 + HCN + H CN+ + C2H6 ⇌ C2H+ 5 + HCN CN+ + C2H6 ⇌ C2H+ 3 + HCN + H2 CN+ + NCCN ⇌ C2N+ 2 + CN CN+ + NCCN ⇌ C3N+ + N2 CN+ + NCCN ⇌ C2N+ + CN2 CN+ + HC3N ⇌ C3HN+ + CN CN+ + HC3N ⇌ C3N+ + HCN CN+ + H2 ⇌ CHN+ + H CN+ + H2O ⇌ CHN+ + HO CN+ + H2O ⇌ CHNO+ + H CN+ + H2O ⇌ CH2N+ + O CN+ + H2O ⇌ H2O+ + CN CN+ + H2O ⇌ CHO+ + HN CN+ + H3N ⇌ H3N+ + CN CN+ + H3N ⇌ CHN+ + H2N CN+ + H3N ⇌ CH2N+ + HN CN+ + H3N ⇌ H2N+ + HCN CN+ + NO ⇌ NO+ + CN CN+ + NO ⇌ CNO+ + N CN+ + N2O ⇌ N2O+ + CN CN+ + N2O ⇌ CNO+ + N2 CN+ + N2O ⇌ NO+ + CN2 CN+ + O2 ⇌ O+ CN+ + O2 ⇌ CNO+ + O CN+ + O2 ⇌ NO+ + CO CNO+ + HCNO ⇌ CHNO+ + NCO CO+ + HCN ⇌ CHN+ + CO CO+ + HCN ⇌ CHO+ + CN CO+ + HNCO ⇌ CHO+ + NCO CO+ + CH2O ⇌ CHO+ + CHO CO+ + CH2O ⇌ CH2O+ + CO CO+ + HCOOH ⇌ CHO+ + CH + O2 CO+ + CH4 ⇌ CH+ CO+ + CH4 ⇌ CHO+ + CH3 CO+ + CH4 ⇌ C2H3O+ + H 4 + CO 2 + CN 2.25e-10 8.80e-10 4.00e-10 1.60e-10 1.60e-10 1.24e-09 3.80e-10 2.85e-10 1.19e-09 1.40e-10 7.00e-11 3.12e-09 7.80e-10 1.00e-09 1.60e-09 6.40e-10 4.80e-10 3.20e-10 1.60e-10 1.20e-09 4.00e-10 3.00e-10 1.00e-10 5.70e-10 1.90e-10 4.56e-10 1.52e-10 1.52e-10 2.58e-10 8.60e-11 8.60e-11 1.06e-09 3.06e-09 3.40e-10 2.16e-09 1.65e-09 1.35e-09 4.10e-09 7.93e-10 4.55e-10 5.20e-11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 500 501 501 501 501 501 501 501 500 500 500 479 479 500 501 501 501 501 501 501 501 501 501 500 500 500 500 500 500 500 500 458 456 456 458 463 463 443 443 443 443 101 Table 6—Continued # Type Reaction α β γ Ref 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + CO CO+ + CO2 ⇌ CO+ CO+ + C2H2O ⇌ C2H2O+ + CO CO+ + HC3N ⇌ C3HN+ + CO CO+ + H2 ⇌ CHO+ + H CO+ + H2O ⇌ H2O+ + CO CO+ + H2O ⇌ CHO+ + HO CO+ + H3N ⇌ H3N+ + CO CO+ + H3N ⇌ CHO+ + H2N CO+ + H3N ⇌ CHO+ + H2N CO+ + N ⇌ NO+ + C CO+ + NO ⇌ NO+ + CO CO+ + O ⇌ O+ + CO CO+ + O2 ⇌ O+ 2 + CO CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ CO+ C+ C+ C+ C+ C+ C+ C+ C+ C+ 2 + HCN ⇌ CHN+ + CO2 2 + HCN ⇌ CHO+ 2 + CN 2 + CH4 ⇌ CH+ 2 + CH4 ⇌ CHO+ 2 + C2H2 ⇌ C2H+ 2 + C2H4 ⇌ C2H+ 2 + C2H4 ⇌ C2H+ 2 + C2H4 ⇌ C2H+ 2 + H ⇌ CHO+ + O 2 + H ⇌ H+ + CO2 2 + H2 ⇌ CHO+ 2 + H2O ⇌ H2O+ + CO2 2 + H2O ⇌ CHO+ 2 + HO 2 + H3N ⇌ H3N+ + CO2 2 + NO ⇌ NO+ + CO2 2 + NO2 ⇌ NO+ 2 + CO2 2 + O ⇌ O+ 2 + CO 2 + O ⇌ O+ + CO2 2 + O2 ⇌ O+ 2 + CO2 2 + HCN ⇌ C3N+ + H 2 + HCN ⇌ C3H+ + N 2 + HCN ⇌ C2H+ + CN 2 + CH4 ⇌ C3H+ 2 + H2 2 + CH4 ⇌ C2H+ + CH3 2 + CH4 ⇌ C3H+ 2 + CH4 ⇌ C3H+ + H2 + H 2 + CH4 ⇌ C2H+ 2 + CH2 2 + C2H2 ⇌ C4H+ + H 3 + H 4 + CO2 2 + CH3 2 + CO2 2 + CO2 + H2 3 + CO2 + H 4 + CO2 2 + H 3.70e-09 3.20e-11 3.10e-09 1.80e-09 1.72e-09 8.84e-10 1.20e-09 3.00e-10 3.00e-10 1.99e-11 3.30e-10 1.40e-10 1.20e-10 8.10e-10 9.00e-11 5.50e-10 5.50e-10 7.30e-11 5.04e-10 2.43e-10 1.53e-10 2.41e-10 4.93e-11 1.40e-09 2.04e-09 7.56e-10 1.86e-09 1.20e-10 7.80e-10 1.64e-10 9.62e-11 6.30e-11 1.56e-09 7.80e-10 2.60e-10 5.74e-10 2.38e-10 2.10e-10 1.96e-10 1.82e-10 1.70e-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 -1.06 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 502 474 479 463 427 427 428 428 485 503 503 503 463 456 456 471 471 504 505 505 505 506 506 110 427 427 428 507 508 507 507 509 473 473 473 107 107 107 107 107 473 102 Table 6—Continued # Type Reaction α β γ Ref 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 4 + N2 2 + NCCN ⇌ C2N+ + C2N 2 + NCCN ⇌ C3N+ + CN 2 + NCCN ⇌ C+ 2 + HC3N ⇌ C3N+ + C2H 2 + HC3N ⇌ C5N+ + H 2 + HC3N ⇌ C3H+ + C2N 2 + H2 ⇌ C2H+ + H 2 + N ⇌ C+ + CN 2 + NO ⇌ NO+ + C2 2 + O ⇌ C+ + CO 2 + O ⇌ CO+ + C C+ C+ C+ C+ C+ C+ C+ C+ C+ C+ C+ C2H+ + HCN ⇌ C2H+ 2 + CN C2H+ + HCN ⇌ CH2N+ + C2 C2H+ + CH4 ⇌ C2H+ 2 + CH3 C2H+ + CH4 ⇌ C3H+ 3 + H2 C2H+ + CH4 ⇌ C3H+ 4 + H C2H+ + C2H2 ⇌ C2H+ 3 + C2 C2H+ + C2H2 ⇌ C4H+ 2 + H C2H+ + C2H4 ⇌ C4H+ 3 + H2 C2H+ + HC3N ⇌ C3H2N+ + C2 C2H+ + HC3N ⇌ C4H+ 2 + CN C2H+ + HC3N ⇌ C5HN+ + H C2H+ + HC3N ⇌ C3HN+ + C2H C2H+ + H2 ⇌ C2H+ 2 + H C2H+ + N ⇌ CH+ + CN C2H+ + NO ⇌ NO+ + C2H C2H+ + O ⇌ C+ + CHO C2H+ + O ⇌ CH+ + CO C2H+ + O ⇌ CHO+ + C C2H+ + O ⇌ CO+ + CH C2HN+ + HCN ⇌ C3HN+ C2HN+ C2HN+ C2HN+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ 7.50e-10 4.50e-10 3.00e-10 2.40e-09 9.60e-10 9.60e-10 1.40e-09 3.99e-11 3.40e-10 3.10e-10 3.10e-10 1.40e-09 1.40e-09 3.74e-10 3.74e-10 1.32e-10 2.40e-09 2.40e-09 5.80e-10 1.60e-09 6.40e-10 6.40e-10 3.20e-10 1.70e-09 9.00e-11 1.20e-10 3.30e-10 3.30e-10 3.30e-10 3.30e-10 1.40e-11 2 + CH3OH ⇌ CH5O+ + NCCN 1.50e-09 2 + C2H4 ⇌ C2H+ 3.92e-10 5 + NCCN 2 + H2O ⇌ H3O+ + NCCN 5.10e-10 2.34e-10 1.26e-10 7.80e-10 2.20e-10 1.25e-09 1.25e-09 1.25e-09 2 + H 2 + HCN ⇌ CH2N+ + C2H 2 + HCN ⇌ C3H2N+ + H 2 + CH4 ⇌ C3H+ 2 + CH4 ⇌ C3H+ 2 + C2H2 ⇌ C4H+ 2 + C2H2 ⇌ C4H+ 2 + C2H2 ⇌ C2H+ 5 + H 4 + H2 3 + H 2 + H2 3 + C2H 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 446 446 446 479 479 479 107 453 453 453 453 510 510 107 107 107 451 451 511 479 479 479 479 107 453 453 453 453 453 453 441 469 469 469 510 510 107 107 486 486 486 103 # Type Reaction α β γ Ref Table 6—Continued 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 4 + C2H2 3 + CH3 5 + H 2 + HCN 3 + H 3 + H2 + H 5 + H 2 + C2H4 ⇌ C2H+ 2 + C2H4 ⇌ C3H+ 2 + C2H4 ⇌ C4H+ 2 + HC3N ⇌ C4H+ 2 + H2 ⇌ C2H+ 2 + SiH4 ⇌ Si+ + C2H6 2 + SiH4 ⇌ SiH+ + C2H5 2 + SiH4 ⇌ SiH+ 2 + C2H4 2 + SiH4 ⇌ SiH+ 3 + C2H3 2 + SiH4 ⇌ SiCH+ 3 + CH3 2 + SiH4 ⇌ SiC2H+ + H2 + H2 + H 2 + SiH4 ⇌ SiC2H+ 2 + SiH4 ⇌ SiC2H+ 2 + N ⇌ C2HN+ + H 2 + N ⇌ C2N+ + H2 2 + N ⇌ CH+ + HCN 2 + NO ⇌ NO+ + C2H2 2 + O ⇌ CHO+ + CH 2 + O ⇌ C2HO+ + H C2H+ 8.96e-10 C2H+ 2.80e-10 C2H+ 2.24e-10 C2H+ 1.60e-09 C2H+ 1.00e-11 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.79e-09 C2H+ 1.50e-10 C2H+ 7.50e-11 C2H+ 2.50e-11 C2H+ 1.20e-10 C2H+ 8.50e-11 C2H+ 8.50e-11 C2H2O+ + C2H2O ⇌ C3H4O+ + CO 3.84e-10 C2H2O+ + C2H2O ⇌ C2H+ 1.42e-10 C2H2O+ + C2H2O ⇌ C2H3O+ + CHO + C 6.49e-11 C2H+ 2.90e-09 C2H+ 1.70e-10 C2H+ 2.16e-10 C2H+ 3.80e-10 C2H+ 4.80e-10 C2H+ 4.00e-10 C2H+ 9.99e-13 C2H+ 2.49e-10 C2H+ 3.08e-11 C2H3N+ + CH3CN ⇌ C2H4N+ + CH2CN 1.96e-09 C2H3O+ + C2H4O ⇌ C2H5O+ + C2H2O 7.14e-10 C2H3O+ + H3N ⇌ H4N+ + C2H2O 4.99e-11 C2H+ 6.73e-10 C2H+ 2.37e-10 C2H+ 3.29e-10 C2H+ 1.53e-10 C2H+ 1.53e-10 C2H+ 8.40e-10 C2H+ 1.41e-09 4 + C2H2 ⇌ C3H+ 4 + C2H2 ⇌ C4H+ 4 + C2H4 ⇌ C3H+ 4 + C2H4 ⇌ C4H+ 4 + C2H4 ⇌ C4H+ 4 + C2H6 ⇌ C3H+ 4 + SiH4 ⇌ SiH+ 3 + CH3 5 + H 5 + CH3 7 + H 7 + H 7 + CH3 2 + C2H6 5 + H2 3 + H2 5 + C2H2 5 + C2H4 3 + HCN ⇌ CH2N+ + C2H2 3 + CH4 ⇌ C3H+ 3 + C2H2 ⇌ C4H+ 3 + C2H4 ⇌ C2H+ 3 + C2H6 ⇌ C2H+ 3 + NCCN ⇌ C2HN+ 3 + H2 ⇌ C2H+ 4 + H 3 + SiH4 ⇌ SiH+ 3 + C2H4 3 + SiH4 ⇌ SiH+ 5 + C2H2 2 + C2H2 4 + CO + CO 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1960 0 512 512 512 479 107 513 513 513 513 513 513 513 513 453 453 453 453 453 453 514 514 514 510 107 473 515 516 469 517 518 518 489 519 470 512 512 520 520 520 521 518 104 # Type Reaction α β γ Ref Table 6—Continued 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 3 + C2H5 4 + SiH4 ⇌ SiH+ 4 + SiH4 ⇌ SiC2H+ 4 + H2 + H2 4 + SiH4 ⇌ SiC2H+ 5 + H2 + H 4 + SiH4 ⇌ SiC2H+ 6 + H2 4 + NO ⇌ NO+ + C2H4 5 + HCN ⇌ CH2N+ + C2H4 5 + CH2O ⇌ CH3O+ + C2H4 5 + HCOOH ⇌ CH3O+ 5 + C2H2 ⇌ C4H+ 5 + H2 5 + C2H2 ⇌ C3H+ 3 + CH4 5 + C2H4 ⇌ C3H+ 5 + CH4 5 + C2H6 ⇌ C4H+ 9 + H2 5 + C2H6 ⇌ C3H+ 7 + CH4 5 + H2 ⇌ C2H+ 5 + H3N ⇌ H4N+ + C2H4 C2H+ C2H+ C2H+ C2H+ C2H+ C2H4N+ + CH3OH ⇌ C3H6N+ + H2O C2H4N+ + CH3CN ⇌ C3H6N+ + HCN C2H4O+ + C2H4O ⇌ C2H5O+ + C2H3O C2H4O+ + H3N ⇌ H4N+ + C2H3O C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H5O+ + (CH3)2O ⇌ C3H9O+ + CH2O C2H5O+ + H3N ⇌ H4N+ + C2H4O C2H5O+ C2H5O+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H+ C2H6O+ + (CH3)2O ⇌ C2H7O+ + CH3OCH2 C2H6O+ + H3N ⇌ H4N+ + CH3CH2O C2H+ C2H+ C2H+ C2H7O+ + HCN ⇌ C3H6N+ + H2O C2H7O+ + HCN ⇌ C2H4N+ + CH3OH C2H7O+ + CH2O ⇌ C3H7O+ + H2O C2H7O+ + CH3CN ⇌ C3H6N+ + CH3OH C2H7O+ + (CH3)2O ⇌ C3H9O+ + CH3OH C2N+ + HCN ⇌ C3N+ 7 + HCN ⇌ CH2N+ + C2H6 7 + HCN ⇌ C2H4N+ + CH4 7 + H3N ⇌ H4N+ + C2H6 2 + H2O 2 + C2H4O ⇌ C2H5O+ + CH3COOH 2 + C2H4O ⇌ C4H7O+ 8 + CH4 9 + H2 + H 9 + CH3 7 + CH4 + H 5 + C2H6 + H 6 + C2H6 ⇌ C3H+ 6 + C2H6 ⇌ C4H+ 6 + C2H6 ⇌ C3H+ 6 + C2H6 ⇌ C3H+ 6 + C2H6 ⇌ C2H+ 6 + H2O ⇌ H3O+ + C2H5 6 + O2 ⇌ C2H+ 5 + HO2 1.41e-09 1.41e-09 1.41e-09 1.41e-09 3.60e-10 2.70e-11 6.30e-11 2.95e-09 1.20e-10 2.70e-09 3.10e-09 1.50e-09 1.24e-10 6.65e-11 6.10e-10 3.44e-11 5.60e-12 7.30e-14 2.10e-09 2.60e-11 4.73e-10 2.43e-09 8.42e-10 1.01e-10 1.01e-10 1.01e-10 1.01e-10 1.01e-10 1.20e-09 1.15e-10 1.93e-09 8.43e-10 1.98e-09 2.20e-10 2.00e-09 1.50e-11 1.80e-11 1.70e-11 7.20e-11 1.20e-11 1.50e-11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 589 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 518 518 518 518 487 522 523 466 470 510 524 443 473 473 111 525 525 Est 497 499 470 519 519 526 526 526 526 526 527 528 489 470 510 510 497 522 522 499 522 499 441 2 + C2H4 6 + H 2 + H 105 Table 6—Continued # Type Reaction α β γ Ref 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + C2H2 3 + H 3 + H2 4 + H 3 + HCN + H2 2 + HCN ⇌ C3N+ 2 + HC3N ⇌ C3HN+ + NCCN C2N+ + CH4 ⇌ C2H+ 2.64e-12 3 + HCN C2N+ + CH4 ⇌ C3H2N+ + H2 1.32e-12 C2N+ + CH4 ⇌ CH2N+ + C2H2 4.40e-13 C2N+ + C2H2 ⇌ C3H+ + HCN 8.01e-10 C2N+ + C2H4 ⇌ C2H2N+ + C2H2 5.85e-10 C2N+ + C2H4 ⇌ C3H+ 3.51e-10 3 + HCN C2N+ + C2H4 ⇌ C2H+ 1.17e-10 4 + C2N C2N+ + C2H4 ⇌ C4H2N+ + H2 1.17e-10 C2N+ + C2H6 ⇌ C3H+ 3.60e-10 C2N+ + C2H6 ⇌ C2H2N+ + C2H4 3.00e-10 C2N+ + C2H6 ⇌ C2H+ 1.20e-10 3 + CH3CN C2N+ + C2H6 ⇌ C2H+ 1.20e-10 5 + HCCN C2N+ + HC3N ⇌ C3H+ + NCCN 3.10e-09 C2N+ + H2O ⇌ CHO+ + HCN 2.55e-10 C2N+ + H2O ⇌ C2HN+ + HO 8.50e-11 C2N+ + H3N ⇌ CH2N+ + HCN 1.71e-09 C2N+ + H3N ⇌ HN+ 1.90e-10 C2N+ 1.60e-11 C2N+ 1.60e-09 H+ + HCN ⇌ CHN+ + H 1.10e-09 H+ + CH4 ⇌ CH+ 4.50e-09 H+ + CH4 ⇌ CH+ 4.50e-09 H+ + CO2 ⇌ CHO+ + O 3.00e-09 H+ + C2H6 ⇌ C2H+ 3.90e-09 H+ + C2H6 ⇌ C2H+ 3.90e-09 H+ + C2H6 ⇌ C2H+ 3.90e-09 H+ + NO ⇌ NO+ + H 1.90e-09 H+ + O ⇌ O+ + H 3.75e-10 HN+ + CH2O ⇌ CHO+ + H2N 1.82e-09 HN+ + CH2O ⇌ CH2O+ + HN 9.90e-10 HN+ + CH2O ⇌ CH3O+ + N 4.95e-10 HN+ + CH4 ⇌ CH2N+ + H2 + H 6.72e-10 HN+ + CH4 ⇌ H2N+ + CH3 1.92e-10 HN+ + CH4 ⇌ CH+ 9.60e-11 HN+ + CH3OH ⇌ CH3O+ + H2N 2.10e-09 HN+ + CH3OH ⇌ CHO+ + H3N + H 4.50e-10 HN+ + CH3OH ⇌ CH2O+ + H3N 4.50e-10 HN+ + CH3OH ⇌ CH5O+ + N 2.10e-10 HN+ + CH5N ⇌ CH4N+ + H2N 9.45e-10 HN+ + CH5N ⇌ CH5N+ + HN 4.20e-10 HN+ + CH5N ⇌ CH6N+ + N 4.20e-10 4 + H2 + H 3 + H2 + H2 5 + H2 5 + N 106 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 454 454 454 501 501 501 501 501 501 501 501 501 479 501 501 501 501 441 479 529 525 525 506 467 467 467 530 530 531 531 531 531 531 531 531 531 531 531 531 531 531 # Type Reaction α β γ Ref Table 6—Continued 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 3 + N 2 + H3N HN+ + CH5N ⇌ CH2N+ + H3N + H HN+ + CH5N ⇌ CH3N+ + H3N HN+ + CO ⇌ CNO+ + H HN+ + CO ⇌ CHO+ + N HN+ + CO2 ⇌ CHO+ 2 + N HN+ + CO2 ⇌ HNO+ + CO HN+ + CO2 ⇌ NO+ + CHO HN+ + C2H4 ⇌ C2H+ 3 + H2N HN+ + C2H4 ⇌ C2H+ 4 + HN HN+ + C2H4 ⇌ CH2N+ + CH3 HN+ + C2H4 ⇌ C2H+ HN+ + C2H4 ⇌ CH3N+ + CH2 HN+ + C2H4 ⇌ C2H3N+ + H2 HN+ + H2 ⇌ H2N+ + H HN+ + H2 ⇌ H+ HN+ + H2O ⇌ H3O+ + N HN+ + H2O ⇌ H2O+ + HN HN+ + H2O ⇌ H2N+ + HO HN+ + H2O ⇌ HNO+ + H2 HN+ + H3N ⇌ H3N+ + HN HN+ + H3N ⇌ H4N+ + N HN+ + NO ⇌ NO+ + HN HN+ + NO ⇌ HN+ HN+ + N2 ⇌ HN+ HN+ + O2 ⇌ O+ HN+ + O2 ⇌ NO+ + HO HN+ + O2 ⇌ HO+ HNO+ + CH4 ⇌ CH+ 5 + NO HNO+ + CO ⇌ CHO+ + NO HNO+ + CO2 ⇌ CHO+ 2 + NO HNO+ + H2O ⇌ H3O+ + NO HNO+ + NO ⇌ NO+ + HNO HNO+ + N2 ⇌ HN+ HN+ HN+ HN+ HN+ HN+ HN+ HN+ HN+ 2 + HCN ⇌ CH2N+ + N2 2 + CH2O ⇌ CH3O+ + N2 2 + CH3NO2 ⇌ CH4NO+ 2 + CH3NO2 ⇌ NO+ + CH3OH + N2 2 + CH4 ⇌ CH+ 5 + N2 2 + CO ⇌ CHO+ + N2 2 + CO2 ⇌ CHO+ 2 + N2 2 + C2H2 ⇌ C2H+ 3 + N2 2 + O 2 + N 2 + HN 2 + N 2 + NO 2 + N2 107 4.20e-10 1.05e-10 5.39e-10 4.41e-10 3.85e-10 3.85e-10 3.30e-10 3.75e-10 3.75e-10 3.00e-10 1.50e-10 1.50e-10 1.50e-10 1.28e-09 2.25e-10 1.05e-09 1.05e-09 8.75e-10 3.50e-10 1.80e-09 6.00e-10 7.12e-10 1.78e-10 6.50e-10 4.51e-09 2.05e-09 1.64e-09 1.01e-10 1.01e-10 1.01e-10 2.30e-09 7.00e-10 9.99e-12 3.20e-09 3.30e-09 3.27e-09 1.65e-11 8.90e-10 8.79e-10 9.80e-10 1.41e-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 531 531 531 531 531 531 531 532 532 532 532 532 532 531 531 531 531 531 531 531 531 531 531 531 531 531 531 433 433 433 436 111 433 465 490 462 462 111 533 534 464 # Type Reaction α β γ Ref Table 6—Continued 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + N2O 3 + N2O 5 + H2N2 7 + N2 7 + N2O 5 + N2O + H2 2 + CH3CN ⇌ C2H4N+ + N2 2 + C2H6 ⇌ C2H+ 2 + C2H6 ⇌ C2H+ 2 + NCCN ⇌ C2HN+ 2 + N2 2 + HC3N ⇌ C3H2N+ + N2 2 + H2O ⇌ H3O+ + N2 2 + H3N ⇌ H4N+ + N2 2 + NO ⇌ HNO+ + N2 2 + N2O ⇌ HN2O+ + N2 2 + O2(a1∆g) ⇌ HO+ HN+ 4.10e-09 HN+ 1.13e-09 HN+ 1.69e-10 HN+ 1.20e-09 HN+ 4.20e-09 HN+ 5.50e-09 HN+ 2.30e-09 HN+ 3.40e-10 HN+ 7.90e-10 HN+ 8.00e-11 2 + N2 HN2O+ + CH3NO2 ⇌ CH4NO+ 2.70e-09 HN2O+ + CH3NO2 ⇌ NO+ + CH3OH + N2O 1.62e-12 HN2O+ + C2H2 ⇌ C2H+ 1.21e-09 HN2O+ + CH3CN ⇌ C2H4N+ + N2O 3.80e-09 HN2O+ + C2H6 ⇌ C2H+ 1.05e-09 HN2O+ + C2H6 ⇌ C2H+ 5.50e-11 HN2O+ + H2O ⇌ H3O+ + N2O 2.80e-09 HN2O+ + H3N ⇌ H4N+ + N2O 2.10e-09 HO+ + CH4 ⇌ H3O+ + CH2 1.31e-09 HO+ + CH4 ⇌ CH+ 1.95e-10 5 + O HO+ + CO ⇌ CHO+ + O 8.20e-10 HO+ + CO2 ⇌ CHO+ 1.44e-09 2 + O HO+ + C2H6 ⇌ C2H+ 1.04e-09 4 + H2O + H HO+ + C2H6 ⇌ C2H+ 3.20e-10 5 + H2O HO+ + C2H6 ⇌ H3O+ + C2H4 1.60e-10 HO+ + C2H6 ⇌ C2H+ 4.80e-11 HO+ + C2H6 ⇌ C2H+ 3.20e-11 HO+ + H2 ⇌ H2O+ + H 1.50e-09 HO+ + H2O ⇌ H2O+ + HO 2.87e-09 HO+ + H2O ⇌ H3O+ + O 2.87e-09 HO+ + NO ⇌ HNO+ + O 6.11e-10 HO+ + NO ⇌ NO+ + HO 3.59e-10 HO+ + NO2 ⇌ NO+ 1.30e-09 2 + HO HO+ + N2 ⇌ HN+ 3.60e-10 HO+ + N2O ⇌ HN2O+ + O 1.12e-09 HO+ + N2O ⇌ N2O+ + HO 2.50e-10 HO+ + N2O ⇌ NO+ + HNO 1.72e-10 HO+ + O2 ⇌ O+ 2.00e-10 HO+ 1.90e-11 HO+ 1.10e-09 HO+ 3.00e-10 2 + HO 2 + Ar ⇌ ArH+ + O2 2 + CO2 ⇌ CHO+ 2 + H2 ⇌ H+ 3 + O2 2 + O2 6 + HO 7 + O 2 + O 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 465 467 467 469 469 535 497 436 111 536 462 462 464 465 467 467 472 497 492 492 492 537 467 467 467 467 467 110 538 538 537 537 539 537 537 537 537 110 435 496 540 108 # Type Reaction α β γ Ref Table 6—Continued 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H3N ⇌ H4N+ + O2 2 + NO ⇌ HNO+ + O2 2 + NO ⇌ NO+ + HO2 2 + N2 ⇌ HN+ 2 + O2 2 + H 2 + CO2 ⇌ CHO+ 2 + H ⇌ H+ + H2 2 + H2 ⇌ H+ 3 + H 2 + He ⇌ HeH+ + H 2 + O2 ⇌ HO+ 2 + H HO+ HO+ HO+ HO+ SiH+ + SiH4 ⇌ Si2H+ + H2 + H2 SiH+ + SiH4 ⇌ Si2H+ 2 + H2 + H SiH+ + Si2H6 ⇌ Si3H+ 5 + H2 SiH+ + Si2H6 ⇌ Si2H+ 3 + SiH4 SiH+ + Si2H6 ⇌ Si3H+ 3 + H2 + H2 SiH+ + Si2H6 ⇌ Si2H+ + SiH4 + H2 Si2H+ + Si2H6 ⇌ Si4H+ 3 + H2 + H2 Si2H+ + Si2H6 ⇌ Si4H+ 5 + H2 Si2H+ + Si2H6 ⇌ Si3H+ 3 + SiH4 Si2H+ + Si2H6 ⇌ Si3H+ 3 + SiH4 H+ H+ H+ H+ H+ H2N+ + CH2O ⇌ CH3O+ + HN H2N+ + CH2O ⇌ H3N+ + CHO H2N+ + HCOOH ⇌ H4N+ + CO2 H2N+ + CH4 ⇌ H3N+ + CH3 H2N+ + CH3OH ⇌ CH5O+ + HN H2N+ + CH3OH ⇌ H3N+ + CHO + H2 H2N+ + CH5N ⇌ CH5N+ + H2N H2N+ + CH5N ⇌ CH6N+ + HN H2N+ + CH5N ⇌ CH4N+ + H3N H2N+ + CH5N ⇌ H4N+ + CH3N H2N+ + C2H4 ⇌ C2H+ H2N+ + C2H4 ⇌ CH4N+ + CH2 H2N+ + C2H4 ⇌ C2H+ 5 + HN H2N+ + C2H4 ⇌ C2H5N+ + H H2N+ + H2 ⇌ H3N+ + H H2N+ + H2O ⇌ H3O+ + HN H2N+ + H2O ⇌ H4N+ + O H2N+ + H3N ⇌ H4N+ + HN H2N+ + H3N ⇌ H3N+ + H2N H2N+ + NO ⇌ NO+ + H2N H2N+ + O2 ⇌ H2NO+ + O H2N+ + O2 ⇌ HNO+ + HO 4 + H2N 2.00e-09 7.30e-10 7.30e-10 8.00e-10 1.20e-10 5.20e-11 3.80e-10 3.30e-10 1.80e-10 1.10e-10 3.19e-10 1.73e-10 4.86e-11 4.86e-11 2.35e-09 6.40e-10 2.11e-09 1.40e-10 7.56e-09 2.24e-09 5.60e-10 2.70e-09 9.20e-10 2.64e-09 4.65e-10 9.00e-10 3.60e-10 3.60e-10 1.80e-10 4.50e-10 4.50e-10 3.00e-10 3.00e-10 1.00e-09 2.76e-09 1.45e-10 1.10e-09 1.10e-09 7.00e-10 1.19e-10 2.10e-11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 496 541 541 496 542 542 543 543 543 543 543 543 543 543 544 545 546 547 548 531 531 443 531 531 531 549 549 549 549 532 532 532 532 110 531 531 550 550 531 531 531 109 Table 6—Continued # Type Reaction α β γ Ref 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H2O 4 + H2O 5 + HO 4 + H2O + H2 6 + H2O 5 + H2O + H 2 + HO H2NO+ 3 + H3N ⇌ H4N+ + HNO3 3 + NO ⇌ H2NO+ H2NO+ 2 + NO2 H2O+ + HCN ⇌ CH2N+ + HO H2O+ + HCN ⇌ H3O+ + CN H2O+ + CH4 ⇌ CH+ 4 + H2O H2O+ + CH4 ⇌ H3O+ + CH3 H2O+ + CO ⇌ CHO+ + HO H2O+ + C2H2 ⇌ C2H+ H2O+ + C2H4 ⇌ C2H+ H2O+ + C2H4 ⇌ C2H+ H2O+ + C2H6 ⇌ H3O+ + C2H5 H2O+ + C2H6 ⇌ C2H+ H2O+ + C2H6 ⇌ C2H+ H2O+ + C2H6 ⇌ C2H+ H2O+ + NCCN ⇌ C2HN+ H2O+ + H2 ⇌ H3O+ + H H2O+ + H2O ⇌ H3O+ + HO H2O+ + N ⇌ HNO+ + H H2O+ + N ⇌ NO+ + H2 H2O+ + NO ⇌ NO+ + H2O H2O+ + NO2 ⇌ NO+ 2 + H2O H2O+ + N2O ⇌ HN2O+ + HO H2O+ + O ⇌ O+ H2O+ + O2 ⇌ O+ H2O+ H2O+ H2O+ H2O+ H2O+ SiH+ SiH+ SiH+ SiH+ SiH+ SiH+ SiH+ SiH+ SiH+ Si2H+ Si2H+ Si2H+ 6.40e-10 3.10e-11 2.10e-09 2.10e-09 1.20e-09 1.20e-09 5.30e-10 1.90e-09 1.60e-09 1.60e-09 1.33e-09 1.92e-10 6.40e-11 1.60e-11 1.00e-09 1.40e-09 1.80e-09 1.90e-10 1.90e-10 3.60e-10 1.20e-09 4.80e-12 3.99e-11 2.00e-10 2 + CO ⇌ CHO+ + HO2 5.50e-11 2 + H2O ⇌ H3O+ + HO2 1.70e-09 2 + H2O2 ⇌ H3O+ 2 + HO2 6.00e-10 2 + H3N ⇌ H4N+ + HO2 1.80e-09 2 + NO ⇌ NO+ + H2O2 5.00e-10 2 + SiH4 ⇌ Si2H+ 6.98e-11 2 + SiH4 ⇌ Si2H+ + H2 + H2 + H 1.26e-11 2 + SiH4 ⇌ Si2H+ 1.16e-11 2 + SiH4 ⇌ Si+ 2.91e-12 2 + Si2H6 ⇌ Si2H+ 3.70e-10 2 + Si2H6 ⇌ Si2H+ 3.15e-10 2 + Si2H6 ⇌ Si3H+ 2.33e-10 2 + Si2H6 ⇌ Si3H+ 2.33e-10 2 + Si2H6 ⇌ Si2H+ 1.37e-10 2 + Si2H6 ⇌ Si4H+ 2.26e-10 2 + Si2H6 ⇌ Si3H+ 1.11e-10 2 + Si2H6 ⇌ Si4H+ 7.38e-11 5 + SiH3 4 + SiH4 6 + H2 7 + H 2 + SiH4 + H2 4 + H2 + H2 4 + SiH4 6 + H2 3 + H2 + H 2 + H2 + H2 + H2 2 + H2 2 + H2O 2 + H2 + H2 110 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 114 114 456 456 551 551 492 504 551 551 467 467 467 467 469 110 538 453 453 552 551 553 453 110 554 554 554 554 554 542 542 542 542 543 543 543 543 543 543 543 543 # Type Reaction α β γ Ref Table 6—Continued 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H2 2 + H2 + H 2 + H2 5 + H2 3 + H2 + H2 7 + H2 5 + H2 + H2 3 + Ar ⇌ ArH+ + H2 3 + HCN ⇌ CH2N+ + H2 3 + CH2O ⇌ CH3O+ + H2 3 + HCOOH ⇌ CHO+ + H2O + H2 3 + HCOOH ⇌ H3O+ + CO + H2 3 + CH3NO2 ⇌ CH4NO+ 3 + CH3NO2 ⇌ NO+ + CH3OH + H2 3 + CH3NO2 ⇌ CH3NO+ 3 + CH4 ⇌ CH+ 5 + H2 3 + CO ⇌ CHO+ + H2 3 + CO2 ⇌ CHO+ 2 + H2 3 + C2H2 ⇌ C2H+ 3 + H2 3 + CH3CN ⇌ C2H4N+ + H2 3 + C2H4 ⇌ C2H+ 3 + C2H4 ⇌ C2H+ 3 + CH3COOH ⇌ C2H3O+ + H2O + H2 3 + CH3OCHO ⇌ C2H5O+ 3 + CH3OCHO ⇌ CH5O+ + CH2O 3 + C2H6 ⇌ C2H+ 3 + C2H6 ⇌ C2H+ 3 + NCCN ⇌ C2HN+ 2 + H2 3 + HC3N ⇌ C3H2N+ + H2 3 + H2O ⇌ H3O+ + H2 3 + H3N ⇌ H4N+ + H2 3 + NO ⇌ HNO+ + H2 3 + NO2 ⇌ NO+ + HO + H2 3 + N2 ⇌ HN+ 3 + N2O ⇌ HN2O+ + H2 3 + O2 ⇌ HO+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H+ H3N+ + HCN ⇌ H4N+ + CN H3N+ + CH2O ⇌ H4N+ + CHO H3N+ + HCOOH ⇌ H4N+ + CH + O2 H3N+ + CH4 ⇌ H4N+ + CH3 H3N+ + CH3OH ⇌ H4N+ + CH3O H3N+ + CH5N ⇌ CH5N+ + H3N H3N+ + CH5N ⇌ CH6N+ + H2N H3N+ + CH5N ⇌ H4N+ + CH3 + HN H3N+ + C2H4 ⇌ H4N+ + C2H3 H3N+ + C2H4O ⇌ H4N+ + C2H3O H3N+ + (CH3)2O ⇌ H4N+ + CH3CH2O H3N+ + H2 ⇌ H4N+ + H 2 + H2 2 + H2 9.99e-12 7.00e-09 6.30e-09 4.27e-09 1.83e-09 4.42e-09 3.53e-09 8.03e-11 1.60e-09 1.40e-09 1.90e-09 2.90e-09 7.30e-09 1.90e-09 1.21e-09 6.80e-09 7.30e-09 7.30e-09 2.02e-11 2.40e-09 3.70e-09 1.05e-08 4.30e-09 4.40e-09 1.40e-09 7.00e-10 2.00e-09 1.80e-09 7.00e-10 6.00e-10 1.10e-09 9.00e-10 2.90e-10 9.00e-10 9.00e-10 6.30e-10 2.70e-10 1.40e-09 4.37e-10 9.37e-10 4.89e-12 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 510 433 555 490 491 491 462 462 462 111 111 111 464 555 111 111 491 495 495 111 111 555 469 472 496 111 111 556 111 540 456 531 443 485 470 531 531 531 532 470 470 557 111 # Type Reaction α β γ Ref Table 6—Continued 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 5 + H2O 2 + H2O 2 + H2O 2 + H2O 2 + H2O H3N+ + H2O ⇌ H4N+ + HO 1.63e-10 H3N+ + H2O ⇌ H3O+ + H2N 8.75e-11 H3N+ + H3N ⇌ H4N+ + H2N 1.07e-09 H3N+ + NO ⇌ NO+ + H3N 7.20e-10 H3O+ + HCN ⇌ CH2N+ + H2O 3.50e-09 H3O+ + CH2O ⇌ CH3O+ + H2O 3.00e-09 H3O+ + HCOOH ⇌ CH3O+ 2.70e-09 H3O+ + CH3NO2 ⇌ CH4NO+ 4.11e-09 H3O+ + CH3OH ⇌ CH5O+ + H2O 2.20e-09 H3O+ + C2H2O ⇌ C2H3O+ + H2O 2.00e-09 H3O+ + CH3CN ⇌ C2H4N+ + H2O 4.70e-09 H3O+ + C2H4 ⇌ C2H+ 6.30e-11 H3O+ + C2H4O ⇌ C2H5O+ + H2O 3.60e-09 H3O+ + CH3COOH ⇌ C2H5O+ 2.85e-09 H3O+ + CH3COOH ⇌ C2H3O+ + H2O + H2O 1.50e-10 H3O+ + CH3OCHO ⇌ C2H5O+ 3.00e-09 H3O+ + (CH3)2O ⇌ C2H7O+ + H2O 2.70e-09 H3O+ + C2H5OH ⇌ C2H7O+ + H2O 2.80e-09 H3O+ + HC3N ⇌ C3H2N+ + H2O 3.80e-09 H3O+ + HNO3 ⇌ H2NO+ 1.60e-09 3 + H2O H3O+ + H3N ⇌ H4N+ + H2O 2.50e-09 H3O+ + NO ⇌ NO+ + H2O + H 1.50e-12 H3O+ 1.00e-10 H3O+ 5.00e-10 SiH+ 5.80e-12 SiH+ 1.30e-10 SiH+ 1.20e-10 SiH+ 2.50e-11 SiH+ 5.22e-10 SiH+ 3.48e-10 Si2H+ 3.13e-10 Si2H+ 1.20e-10 Si2H+ 6.77e-11 Si2H+ 2.08e-11 H4N+ + CH5N ⇌ CH6N+ + H3N 2.00e-09 Fe+ + N2O ⇌ FeO+ + N2 7.00e-11 Fe+ + O3 ⇌ FeO+ + O2 1.50e-10 FeO+ + CO ⇌ Fe+ + CO2 9.00e-10 He+ + Ar ⇌ Ar+ + He 5.00e-14 He+ + HCN ⇌ CN+ + H + He 1.95e-09 He+ + HCN ⇌ CHN+ + He 7.80e-10 2 + H2O ⇌ H3O+ + H2O2 2 + O2 ⇌ HO+ 3 + H2O ⇌ SiH3O+ + H2 3 + SiH4 ⇌ Si2H+ + H2 + H2 + H2 3 + SiH4 ⇌ Si2H+ 3 + SiH4 ⇌ Si2H+ 3 + Si2H6 ⇌ Si2H+ 3 + Si2H6 ⇌ Si3H+ 3 + Si2H6 ⇌ Si4H+ 3 + Si2H6 ⇌ Si3H+ 3 + Si2H6 ⇌ Si3H+ 3 + Si2H6 ⇌ Si4H+ 3 + H2 + H2 2 + H2 + H2 + H 5 + SiH4 7 + H2 5 + H2 + H2 5 + SiH4 3 + SiH4 + H2 7 + H2 4 + H2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 558 558 538 531 465 463 491 462 559 560 465 561 560 491 491 562 560 560 469 114 563 564 554 111 565 542 542 542 543 543 543 543 543 543 566 567 108 567 568 457 457 112 # Type Reaction α β γ Ref Table 6—Continued 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 7.80e-10 3.90e-10 2.87e-09 1.23e-09 4.10e-10 9.12e-10 3.04e-10 2.72e-10 9.60e-11 1.60e-11 1.60e-09 1.05e-09 3.50e-10 8.69e-10 1.21e-10 9.90e-11 1.10e-11 2.66e-09 2.66e-09 2.66e-09 2.66e-09 2.66e-09 2.30e-09 2.30e-09 2.30e-09 5.00e-13 5.00e-13 2.00e-12 1.00e-10 4.57e-10 9.35e-11 1.60e-09 2.40e-10 1.60e-10 2.40e-09 2.40e-09 2.40e-09 2.40e-09 1.60e-09 9.38e-10 3.13e-10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 457 457 443 443 443 569 569 569 569 569 423 422 422 569 569 569 569 425 425 425 425 425 479 479 479 547 547 547 570 570 570 428 428 428 425 425 425 425 423 422 422 2 + He 2 + C + He He+ + HCN ⇌ CH+ + He + N He+ + HCN ⇌ C+ + H + He + N He+ + HCOOH ⇌ CHO+ + HO + He He+ + HCOOH ⇌ CHO+ 2 + H + He He+ + HCOOH ⇌ CH2O+ 2 + He He+ + CH4 ⇌ CH+ 2 + H2 + He He+ + CH4 ⇌ H+ + CH3 + He He+ + CH4 ⇌ CH+ + H2 + H + He He+ + CH4 ⇌ CH+ 3 + H + He He+ + CH4 ⇌ CH+ 4 + He He+ + CO ⇌ CO+ + He He+ + CO ⇌ C+ + O + He He+ + CO ⇌ O+ + C + He He+ + CO2 ⇌ CO+ + O + He He+ + CO2 ⇌ CO+ He+ + CO2 ⇌ O+ + CO + He He+ + CO2 ⇌ O+ He+ + C2H6 ⇌ C2H+ 6 + He He+ + C2H6 ⇌ C2H+ 5 + H + He He+ + C2H6 ⇌ H+ + C2H5 + He He+ + C2H6 ⇌ C2H+ 4 + H2 + He He+ + C2H6 ⇌ C2H+ 3 + H2 + H + He He+ + HC3N ⇌ C+ He+ + HC3N ⇌ C+ He+ + HC3N ⇌ C2H+ + CN + He He+ + H2 ⇌ HeH+ + H He+ + H2 ⇌ H+ He+ + H2 ⇌ H+ + H + He He+ + H2O ⇌ HeH+ + HO He+ + H2O ⇌ HO+ + H + He He+ + H2O ⇌ H2O+ + He He+ + H3N ⇌ H2N+ + H + He He+ + H3N ⇌ H3N+ + He He+ + H3N ⇌ HN+ + H2 + He He+ + SiH4 ⇌ Si+ + H2 + H2 + He He+ + SiH4 ⇌ SiH+ + H2 + H + He He+ + SiH4 ⇌ H+ + SiH3 + He He+ + Si2H6 ⇌ Si2H+ He+ + NO ⇌ NO+ + He He+ + NO ⇌ N+ + O + He He+ + NO ⇌ O+ + N + He 2 + HCN + He 3 + HN + He 2 + He 2 + H2 + H2 + He 113 # Type Reaction α β γ Ref Table 6—Continued 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 8.25e-10 6.75e-10 2.30e-09 6.80e-10 1.70e-10 2.10e-09 2.10e-09 1.50e-09 1.70e-09 1.10e-09 2.00e-10 1.40e-09 1.80e-09 1.80e-09 1.80e-09 1.30e-09 1.20e-09 7.88e-10 2.63e-10 2.63e-10 5.80e-11 1.30e-09 7.00e-10 2.30e-10 3.20e-10 4.30e-10 1.00e-10 8.00e-10 4.40e-10 1.00e-09 1.89e-09 7.25e-10 2.90e-10 6.20e-09 4.79e-10 3.76e-10 5.64e-11 2.82e-11 1.40e-09 6.00e-10 3.20e-10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 571 571 423 571 571 467 467 572 556 492 437 437 437 437 437 437 437 437 437 437 573 573 573 108 573 573 108 573 573 573 574 574 574 443 574 574 574 574 574 574 574 5 + H2 + He 3 + H2 + H2 + He 2 + He 2 + He + He 2 + He 2 + He 2 + He + He 2 + He + He He+ + N2 ⇌ N+ + N + He He+ + N2 ⇌ N+ He+ + N2O ⇌ N2O+ + He He+ + O2 ⇌ O+ + O + He He+ + O2 ⇌ O+ 2 + He HeH+ + C2H6 ⇌ C2H+ HeH+ + C2H6 ⇌ C2H+ HeH+ + H2 ⇌ H+ 3 + He HeH+ + N2 ⇌ HN+ HeH+ + O2 ⇌ HO+ He+ 2 + Ar ⇌ Ar+ + He + He He+ 2 + CO ⇌ CO+ + He + He He+ 2 + CO2 ⇌ CO+ + O + He + He He+ 2 + CO2 ⇌ CO+ He+ 2 + CO2 ⇌ O+ + CO + He + He He+ 2 + NO ⇌ NO+ + He + He He+ 2 + N2 ⇌ N+ He+ 2 + O2 ⇌ O+ He+ 2 + O2 ⇌ O+ + O + He + He He+ 2 + O2 ⇌ O+ + O + He + He Mg+ + HNO3 ⇌ MgHO+ + NO2 Mg+ + H2O2 ⇌ MgHO+ + HO Mg+ + N2O ⇌ MgO+ + N2 Mg+ + O3 ⇌ MgO+ + O2 MgO+ + CO ⇌ Mg+ + CO2 MgO+ + NO ⇌ Mg+ + NO2 MgO+ + O ⇌ Mg+ + O2 MgO+ + O3 ⇌ Mg+ + O2 + O2 MgHO+ + HNO3 ⇌ NO+ MgHO+ + H2O2 ⇌ MgH2O+ N+ + CH2O ⇌ CH2O+ + N N+ + CH2O ⇌ CHO+ + HN N+ + CH2O ⇌ NO+ + CH2 N+ + HCOOH ⇌ CHO+ + HNO N+ + CH4 ⇌ CH+ N+ + CH4 ⇌ CH2N+ + H2 N+ + CH4 ⇌ CHN+ + H2 + H N+ + CH4 ⇌ CH+ N+ + CH3OH ⇌ CH4O+ + N N+ + CH3OH ⇌ CH2O+ + H2N N+ + CH3OH ⇌ CH3O+ + HN 2 + MgO2H2 2 + HO 3 + HN 4 + N 114 Table 6—Continued # Type Reaction α β γ Ref 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 3 + H2 + N2 3 + HN 4 + N N+ + CH3OH ⇌ NO+ + CH4 2.00e-10 N+ + CH3OH ⇌ CH+ 8.00e-11 3 + HNO N+ + CH5N ⇌ CH4N+ + HN 1.40e-09 N+ + CH5N ⇌ CH2N+ + HN + H2 2.00e-10 N+ + CH5N ⇌ CH3N+ + H2N 1.40e-10 N+ + CH5N ⇌ CH5N+ + N 1.40e-10 N+ + CH5N ⇌ CH+ 1.20e-10 N+ + CO ⇌ CO+ + N 3.96e-10 N+ + CO ⇌ NO+ + C 5.40e-11 N+ + CO2 ⇌ CO+ 7.50e-10 2 + N N+ + CO2 ⇌ CO+ + NO 2.50e-10 N+ + C2H4 ⇌ C2H+ 4.80e-10 N+ + C2H4 ⇌ C2H+ 4.00e-10 N+ + C2H4 ⇌ CHN+ + CH3 2.40e-10 N+ + C2H4 ⇌ C2H+ 1.60e-10 2 + H2N N+ + C2H4 ⇌ CH2N+ + CH2 1.60e-10 N+ + C2H4 ⇌ C2H2N+ + H2 1.60e-10 N+ + NCCN ⇌ C2N+ + N2 9.80e-10 N+ + NCCN ⇌ C2N+ 4.20e-10 2 + N N+ + H2 ⇌ HN+ + H 4.80e-10 N+ + H2O ⇌ H2O+ + N 2.80e-09 N+ + H3N ⇌ H3N+ + N 1.97e-09 N+ + H3N ⇌ HN+ 2.16e-10 2 + H2 N+ + H3N ⇌ H2N+ + HN 2.16e-10 N+ + NO ⇌ NO+ + N 4.51e-10 N+ + NO ⇌ N+ 7.95e-11 N+ + O2 ⇌ O+ 3.11e-10 N+ + O2 ⇌ NO+ + O 2.62e-10 N+ + O2 ⇌ O+ + NO 3.66e-11 NO+ + CH5N ⇌ CH5N+ + NO 8.20e-10 NO+ + C2H4O ⇌ C2H3O+ + HNO 3.50e-10 NO+ + C2H5OH ⇌ C2H5O+ + HNO 8.00e-10 NO+ 8.90e-10 NO+ 7.99e-12 NO+ 2.90e-10 NO+ 7.99e-12 N+ 2 + Ar ⇌ Ar+ + N2 3.50e-11 N+ 2 + CH2O ⇌ CHO+ + N2 + H 2.52e-09 N+ 2 + CH2O ⇌ CH2O+ + N2 3.77e-10 N+ 2 + HCOOH ⇌ CHO+ + N2 + HO 4.60e-09 2 + CH4 ⇌ CH+ N+ 9.30e-10 2 + C2H6 ⇌ C2H+ 2 + N ⇌ NO+ + NO 2 + NO ⇌ NO+ + NO2 2 + O ⇌ NO+ + O2 5 + NO + HO 2 + O 2 + N 3 + N2 + H 115 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 574 574 574 574 574 574 574 574 574 574 574 532 532 532 532 532 532 446 446 574 574 574 574 574 575 575 574 574 574 574 576 576 528 577 578 577 431 574 574 443 574 Table 6—Continued # Type Reaction α β γ Ref 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306 2307 2308 2309 2310 2311 2312 2313 2314 2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330 2331 2332 2333 2334 2335 2336 2337 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H 3 + N 2 + N2 + H2 3 + HO + N2 2 + N2 3 + C2N N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N2O+ + CH4 ⇌ HN2O+ + CH3 N2O+ + CH4 ⇌ CH+ 4 + N2O N2O+ + CO ⇌ NO+ + NCO N2O+ + CO ⇌ CO+ 2 + N2 N2O+ + H2 ⇌ HN2O+ + H N2O+ + H3N ⇌ H3N+ + N2O N2O+ + NO ⇌ NO+ + N2O N2O+ + NO2 ⇌ NO+ + N2 + O2 N2O+ + NO2 ⇌ NO+ N2O+ + N2O ⇌ NO+ + N2 + NO N2O+ + O2 ⇌ O+ N+ N+ N+ N+ N+ N+ 2 + CH4 ⇌ CH+ 7.00e-11 2 + CH3OH ⇌ CH+ 1.11e-09 2 + CH3OH ⇌ CH3O+ + N2 + H 1.68e-10 2 + CH3OH ⇌ CH4O+ + N2 1.26e-10 2 + CH5N ⇌ CH4N+ + N2 + H 8.76e-10 2 + CH5N ⇌ CH+ 2.52e-10 3 + H2N + N2 2 + CH5N ⇌ CH5N+ + N2 7.20e-11 2 + CO ⇌ CO+ + N2 7.40e-11 2 + CO2 ⇌ CO+ 2 + N2 7.70e-10 2 + NCCN ⇌ C2N+ 8.17e-10 2 + NCCN ⇌ N+ 2.15e-11 2 + NCCN ⇌ CN+ 2.15e-11 2 + CN2 2 + HC3N ⇌ C3HN+ + N2 4.10e-09 2 + H2 ⇌ HN+ 1.41e-09 2 + H2O ⇌ HN+ 9.88e-10 2 + HO 2 + H2O ⇌ H2O+ + N2 9.12e-10 2 + H3N ⇌ H3N+ + N2 1.90e-09 2 + N ⇌ N+ + N2 9.99e-12 2 + NO ⇌ NO+ + N2 3.30e-10 2 + N2 ⇌ N+ 2.76e-12 2 + N2O ⇌ N2O+ + N2 7.00e-10 2 + O ⇌ NO+ + N 1.34e-10 2 + O ⇌ O+ + N2 5.60e-12 2 + O2 ⇌ O+ 5.10e-11 2 + N2 9.74e-10 1.23e-10 3.14e-10 2.57e-10 2.60e-10 1.82e-09 4.30e-10 4.29e-10 2.21e-10 1.20e-11 3.90e-10 9.88e-10 4.94e-10 3.04e-10 1.14e-10 4.56e-11 2.40e-12 3 + CHO 3 + CH2O ⇌ HN+ 3 + CH2O ⇌ CH2O+ + N2 + N 3 + CH2O ⇌ CHO+ + HN + N2 3 + CH2O ⇌ H2N+ + CO + N2 3 + CH4 ⇌ CH2N+ + H2 + N2 3 + CH4 ⇌ CH4N+ + N2 2 + N2O 2 + N2O 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 574 574 574 574 574 574 574 574 574 436 Est Est 479 574 535 535 574 579 507 580 579 507 507 581 582 582 582 582 582 582 582 583 583 583 582 574 574 574 574 574 574 116 Table 6—Continued # Type Reaction α β γ Ref 2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349 2350 2351 2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + HN 2 + N2 3 + CH3OH ⇌ CH4O+ + N2 + N 3 + CH3OH ⇌ CH3O+ + HN + N2 3 + CH3OH ⇌ NO+ + CH4 + N2 3 + CH5N ⇌ CH4N+ + HN + N2 3 + CH5N ⇌ CH5N+ + N2 + N 3 + H2 ⇌ HN+ 3 + H2O ⇌ H2NO+ + N2 3 + H3N ⇌ H3N+ + N2 + N 3 + NO ⇌ NO+ + N2 + N 3 + NO ⇌ N2O+ + N2 3 + O2 ⇌ NO+ + N2O 3 + O2 ⇌ NO+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ N+ O+ + HCN ⇌ CO+ + HN O+ + HCN ⇌ CHO+ + N O+ + HCN ⇌ NO+ + CH O+ + CH2O ⇌ CH2O+ + O O+ + CH2O ⇌ CHO+ + HO O+ + HCOOH ⇌ CHO+ + O2 + H O+ + HCOOH ⇌ CHO+ + HO2 O+ + CH4 ⇌ CH+ O+ + CH4 ⇌ CH+ O+ + CH3OH ⇌ CH3O+ + HO O+ + CH3OH ⇌ CH4O+ + O O+ + CH3OH ⇌ CH2O+ + H2O O+ + CH5N ⇌ CH4N+ + HO O+ + CH5N ⇌ CH3N+ + H2O O+ + CH5N ⇌ CH5N+ + O O+(2D) + CO ⇌ CO+ + O O+(3P) + CO ⇌ CO+ + O O+ + CO2 ⇌ CO+ O+ + CO2 ⇌ O+ O+ + C2H6 ⇌ C2H+ O+ + C2H6 ⇌ C2H+ O+ + H2 ⇌ HO+ + H O+ + H2O ⇌ H2O+ + O O+ + H3N ⇌ H3N+ + O O+(2D) + NO ⇌ NO+ + O O+(3P) + NO ⇌ NO+ + O O+ + NO2 ⇌ NO+ O+(2D) + N2 ⇌ N+ O+(2D) + N2 ⇌ NO+ + N 4 + O 3 + HO 2 + O 2 + CO 4 + H2O 5 + HO 2 + O 2 + O 117 5.00e-10 2.80e-10 2.20e-10 8.69e-10 2.31e-10 2.00e-13 3.30e-10 2.10e-09 1.40e-10 1.40e-10 3.57e-11 1.53e-11 3.50e-09 3.50e-09 3.50e-09 2.10e-09 1.40e-09 3.50e-09 1.50e-09 8.90e-10 1.10e-10 1.33e-09 4.75e-10 9.50e-11 1.66e-09 3.15e-10 1.26e-10 1.30e-09 1.30e-09 9.00e-10 9.00e-10 1.33e-09 5.70e-10 1.70e-09 3.20e-09 1.20e-09 1.20e-09 1.20e-09 1.60e-09 1.35e-10 1.50e-11 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 574 574 574 574 574 574 574 574 584 584 574 574 529 529 529 574 574 443 443 574 574 574 574 574 574 574 574 585 585 586 586 467 467 574 574 574 585 585 587 585 585 Table 6—Continued # Type Reaction α β γ Ref 2379 2380 2381 2382 2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + O 2 + O2 2 + HO2 O+(3P) + N2 ⇌ N+ O+(3P) + N2 ⇌ NO+ + N O+(2D) + O2 ⇌ O+ 2 + O O+(2D) + O2 ⇌ O+ + O2 SiO+ + H2 ⇌ SiHO+ + H SiO+ + N ⇌ Si+ + NO SiO+ + N ⇌ NO+ + Si O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ O+ 1.35e-10 1.50e-11 5.67e-10 2.43e-10 3.20e-10 2.01e-10 9.90e-11 2 (X2Πg) + Ar ⇌ Ar+ + O2 5.00e-10 2 (X2Πg) + Ar ⇌ O+ 5.00e-10 2 + Ar 2 + CH2O ⇌ CH2O+ + O2 2.07e-09 2 + CH2O ⇌ CHO+ + HO2 2.30e-10 2 + HCOOH ⇌ CH2O+ 1.17e-09 2 + HCOOH ⇌ CHO+ 6.30e-10 2 + CH4 ⇌ CH3O+ 4.41e-12 2 + CH4 ⇌ CH2O+ + H2O 9.45e-13 2 + CH4 ⇌ H2O+ + CH2O 9.45e-13 2 (X2Πg) + CH4 ⇌ CH3O+ 7.70e-10 2 (X2Πg) + CH4 ⇌ CH2O+ + H2O 1.65e-10 2 (X2Πg) + CH4 ⇌ H2O+ + CH2O 1.65e-10 2 + CH3OH ⇌ CH3O+ + HO2 5.00e-10 2 + CH3OH ⇌ CH4O+ + O2 5.00e-10 2 + CH5N ⇌ CH5N+ + O2 8.45e-10 2 + CH5N ⇌ CH4N+ + HO2 4.55e-10 2 (X2Πg) + CO ⇌ CO+ + O2 1.49e-10 2 (X2Πg) + CO ⇌ O+ 1.00e-10 2 + CO 2 (X2Πg) + CO ⇌ CO+ 2 + O 3.06e-11 2 (X2Πg) + CO2 ⇌ CO+ 9.00e-10 2 + O2 2 (X2Πg) + CO2 ⇌ O+ 9.00e-10 2 + CO2 2 + C2H4 ⇌ C2H+ 6.80e-10 2 + C2H6 ⇌ C2H+ 1.21e-09 2 (X2Πg) + H2 ⇌ O+ 6.00e-10 2 (X2Πg) + H2 ⇌ HO+ 1.02e-09 2 (X2Πg) + H2 ⇌ H+ 1.80e-10 2 + H2O2 ⇌ H2O+ 1.50e-09 2 + O2 2 + H3N ⇌ H3N+ + O2 1.00e-09 2 + N ⇌ NO+ + O 1.20e-10 2 + NO ⇌ NO+ + O2 3.00e-10 2 (X2Πg) + NO ⇌ NO+ + O2 1.10e-09 2 (X2Πg) + NO ⇌ O+ 1.10e-09 2 + NO 2 + NO2 ⇌ NO+ 8.20e-10 2 + O2 2 (X2Πg) + N2 ⇌ N+ 4.10e-10 4 + O2 6 + O2 2 + H2 2 + H 2 + O2 2 + H 2 + H 2 + O2 118 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 585 585 588 588 589 590 590 591 591 574 574 443 443 574 574 574 592 592 592 574 574 574 574 591 591 585 591 591 593 528 591 585 585 554 574 577 581 541 541 593 591 # Type Reaction α β γ Ref Table 6—Continued 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451 2452 2453 2454 2455 2456 2457 2458 2459 2460 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + H2 + H2 2 + SiH4 4 + H2 2 (X2Πg) + N2 ⇌ O+ 2 (X2Πg) + O2 ⇌ O+ 2 + N2 2 + O2 2 + H2 + H2 3 + H2 + H 4 + H2 2 + SiH4 3 + SiH3 4 + H2 O+ 4.10e-10 O+ 4.60e-10 Si+ + H2O ⇌ SiHO+ + H 2.30e-10 Si+ + SiH4 ⇌ Si2H+ + H2 + H 2.08e-11 Si+ + SiH4 ⇌ Si+ 1.02e-11 Si+ + Si2H6 ⇌ Si2H+ 7.85e-10 Si+ + Si2H6 ⇌ Si3H+ 2.35e-10 Si+ + O2 ⇌ SiO+ + O 9.20e-11 Si+ 2 + Si2H6 ⇌ Si4H+ 3.06e-10 Si+ 2 + Si2H6 ⇌ Si4H+ 1.53e-10 2 + Si2H6 ⇌ Si4H+ Si+ 1.30e-10 Si+ 2 + Si2H6 ⇌ Si3H+ 1.22e-10 Si+ 2 + Si2H6 ⇌ Si3H+ 5.36e-11 Ti+ + C2H6 ⇌ TiC2H+ 7.50e-11 Ti+ + NO ⇌ TiO+ + N 9.50e-11 Ti+ + O2 ⇌ TiO+ + O 5.00e-10 C− + HCN ⇌ CN− + CH 1.10e-09 C− + CO ⇌ C2O + e− 4.10e-10 C− + CO2 ⇌ CO + CO + e− 4.70e-11 C− + H2 ⇌ CH2 + e− 1.00e-13 C− + SiH4 ⇌ SiH− 6.20e-11 C− + N2O ⇌ CO + N2 + e− 9.00e-10 C− + O2 ⇌ O− + CO 3.40e-10 C− + O2 ⇌ CO2 + e− 6.00e-11 CH3O− + HCN ⇌ CN− + CH3OH 3.30e-09 CH3O− + C2H2O ⇌ C2HO− + CH3OH 1.40e-09 CH3O− + CH3CN ⇌ C2H2N− + CH3OH 3.50e-09 CH3O− + C2H4O ⇌ C2H3O− + CH3OH 2.00e-09 CH3O− + C2H5OH ⇌ C2H5O− + CH3OH 3.30e-09 CH4N− + C2H2 ⇌ C2H− + CH5N 1.33e-09 CH4N− + H2 ⇌ H− + CH5N 2.00e-10 CN− + C2H4O ⇌ CNO− + C2H4 9.99e-13 CN− + H ⇌ HCN + e− 1.30e-09 CN− + HNO3 ⇌ NO− 2.00e-09 CO− 1.70e-10 CO− 1.00e-11 CO− 2.00e-10 CO− 1.10e-10 CO− 1.65e-10 CO− 5.50e-11 CO− 4.80e-11 3 + H ⇌ HO− + CO2 3 + NO ⇌ NO− 3 + NO2 ⇌ NO− 3 + O ⇌ O− 4 + H ⇌ CO− 4 + H ⇌ HO− + CO + O2 4 + NO ⇌ NO− 3 + CH 3 + HCN 2 + CO2 3 + CO2 2 + CO2 3 + HO 3 + CO2 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 4000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 591 591 589 542 542 543 543 589 543 543 543 543 543 594 595 595 596 597 597 597 596 597 597 597 598 599 465 598 598 464 600 458 601 602 603 604 605 606 603 603 607 119 Table 6—Continued # Type Reaction α β γ Ref 2461 2462 2463 2464 2465 2466 2467 2468 2469 2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 3 + O2 3 + CH3CN 2 + CH2CN 3 + CO2 3 + CO2 + O2 2 + HCN ⇌ CN− + C2H 2 + O2 ⇌ CO2 + C + e− 4 + O ⇌ CO− 4 + O ⇌ O− + CO2 + O2 4 + O ⇌ O− 4 + O3 ⇌ O− CO− CO− CO− CO− C− C− C2H− + HCN ⇌ CN− + C2H2 C2H− + CH3NO2 ⇌ CH2NO− C2H− + CH3NO2 ⇌ NO− C2H− + CH3OH ⇌ CH3O− + C2H2 C2H− + CH3CN ⇌ C2H2N− + C2H2 C2H− + C2H5OH ⇌ C2H5O− + C2H2 C2H− + O3 ⇌ O− C2H− 2 + O2 ⇌ O− C2H2N− + H ⇌ CH3CN + e− C2H2N− + HNO3 ⇌ NO− C2H2N− + NO2 ⇌ NO− H− + HCN ⇌ CN− + H2 H− + CH3NO2 ⇌ CH2NO− H− + CH3NO2 ⇌ NO− H− + CO ⇌ CHO + e− H− + C2H2 ⇌ C2H− + H2 H− + CH3CN ⇌ C2H2N− + H2 H− + H ⇌ H2 + e− H− + H2O ⇌ HO− + H2 H− + H3N ⇌ H2N− + H2 H− + SiH4 ⇌ SiH− 3 + H2 H− + NO ⇌ HNO + e− H− + NO2 ⇌ NO− 2 + H H− + N2O ⇌ HO− + N2 H− + O2 ⇌ HO2 + e− HO− + HCN ⇌ CN− + H2O HO− + HCOOH ⇌ CHO− HO− + CH3NO2 ⇌ CH2NO− HO− + CH3NO2 ⇌ NO− HO− + CH3OH ⇌ CH3O− + H2O HO− + C2H2 ⇌ C2H− + H2O HO− + C2H2O ⇌ C2HO− + H2O HO− + CH3CN ⇌ C2H2N− + H2O HO− + C2H4 ⇌ C2H− HO− + C2H4O ⇌ C2H3O− + H2O 2 + H2O 2 + H2O 2 + CH3OH 2 + H2 2 + CH4 2 + C2H2 2 + C3H4 3 + C2H 2 + C2H2 3 + H2O 120 7.00e-11 3.50e-11 3.50e-11 1.30e-10 2.00e-09 2.10e-11 3.90e-09 2.38e-09 1.26e-10 5.00e-11 1.50e-09 1.00e-10 2.00e-12 3.00e-10 1.90e-09 1.40e-09 1.00e-09 1.50e-08 1.22e-08 6.45e-10 5.00e-11 4.42e-09 1.30e-08 1.80e-09 3.70e-09 8.80e-13 5.70e-10 4.60e-10 2.90e-09 1.10e-09 1.20e-09 4.10e-09 2.20e-09 3.61e-08 1.90e-09 2.20e-09 2.18e-09 2.20e-09 4.40e-09 3.00e-11 3.10e-09 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 606 606 606 605 536 608 465 462 462 609 465 609 610 611 612 602 602 465 462 462 613 464 465 612 472 600 614 613 615 613 613 465 616 462 462 616 464 616 465 611 617 # Type Reaction α β γ Ref Table 6—Continued 2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2 + HO 2 + H2N 3 + HNO2 2 + H3N 2 + CH5N 2 + CH3OH HO− + CH3OCHO ⇌ CHO− HO− + CH3OCHO ⇌ CH3O− + H2O + CO HO− + C2H5OH ⇌ C2H5O− + H2O HO− + H ⇌ H2O + e− HO− + H3N ⇌ H2N− + H2O HO− + SiH4 ⇌ SiH− 3 + H2O HO− + SiH4 ⇌ SiH3O− + H2 HO− + N ⇌ HNO + e− HO− + NO2 ⇌ NO− HO− + O ⇌ HO2 + e− HO− + O3 ⇌ O− 3 + HO H2N− + HCN ⇌ CN− + H3N H2N− + CH3NO2 ⇌ CH2NO− H2N− + CH3NO2 ⇌ NO− H2N− + CH5N ⇌ CH4N− + H3N H2N− + CO2 ⇌ CNO− + H2O H2N− + C2H2 ⇌ C2H− + H3N H2N− + CH3CN ⇌ C2H2N− + H3N H2N− + C2H4O ⇌ CN− + CH2O + H2 + H2 H2N− + C2H4O ⇌ C2H4NO− + H2 H2N− + NO2 ⇌ NO− H2N− + N2O ⇌ N− 3 + H2O H2N− + O2 ⇌ HO− + HNO NO− + NO2 ⇌ NO− 2 + NO NO− + O2 ⇌ O− NO− NO− NO− NO− 3 + O2 NO− 3 + NO2 O− + HCN ⇌ CN− + HO O− + CH4 ⇌ HO− + CH3 O− + CO ⇌ CO2 + e− O− + C2H2 ⇌ C2H2O + e− O− + C2H2 ⇌ C2H− + HO O− + C2H2 ⇌ C2HO− + H O− + CH3CN ⇌ C2H2N− + HO O− + C2H4 ⇌ C2H4O + e− O− + C2H4 ⇌ C2H− 2 + H2O O− + C2H4 ⇌ C2H3O− + H O− + C2H6 ⇌ HO− + C2H5 2 + NO 2 + H ⇌ HNO2 + e− 2 + H ⇌ HO− + NO 2 + HNO3 ⇌ NO− 2 + O3 ⇌ NO− 2 + O3 ⇌ O− 8.40e-10 6.60e-10 2.70e-09 1.40e-09 5.00e-12 8.71e-10 4.29e-10 9.99e-12 1.10e-09 2.00e-10 9.00e-10 4.80e-09 4.61e-09 2.23e-10 1.01e-10 9.30e-10 1.84e-09 5.10e-09 1.20e-10 3.00e-11 1.00e-09 2.09e-10 4.60e-11 7.40e-10 5.00e-10 1.85e-10 1.85e-10 1.60e-09 1.20e-10 9.00e-11 3.70e-09 1.10e-10 7.30e-10 1.31e-09 8.07e-10 6.54e-11 3.50e-09 4.10e-10 1.89e-10 1.95e-11 7.00e-10 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 618 618 616 619 620 614 614 621 622 621 622 465 462 462 600 463 464 465 617 617 615 624 625 626 626 603 603 114 622 610 536 627 628 627 627 627 602 627 627 627 629 121 Table 6—Continued # Type Reaction α β γ Ref 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i 2i pi pi pi pi pi pi pi pi pi pi pi pd pi pi 3 + O 2 + O2 2 + H ⇌ HO2 + e− 2 + H ⇌ H− + O2 2 + N ⇌ NO2 + e− 2 + NO2 ⇌ NO− 2 + O ⇌ O3 + e− 2 + O ⇌ O− + O2 2 + O3 ⇌ O− 3 + O2 3 + CO2 ⇌ CO− 3 + H ⇌ HO− + O2 3 + NO ⇌ NO− 3 + NO2 ⇌ NO− 3 + O ⇌ O− O− + H2 ⇌ H2O + e− 7.01e-10 O− + H2 ⇌ HO− + H 2.92e-11 O− + H3N ⇌ HO− + H2N 1.50e-09 O− + N ⇌ NO + e− 2.20e-10 O− + NO ⇌ NO2 + e− 3.10e-10 O− + NO2 ⇌ NO− 1.25e-09 2 + O O− + N2O ⇌ NO− + NO 2.20e-10 O− + O ⇌ O2 + e− 1.90e-10 O− + O2 ⇌ O− 7.00e-13 2 + O O− + O2(a1∆g) ⇌ O3 + e− 3.00e-10 O− + O3 ⇌ O2 + O2 + e− 3.01e-10 O− + O3 ⇌ O− 1.99e-10 O− + O3 ⇌ O− 1.02e-11 O− 1.40e-09 O− 1.40e-09 O− 4.00e-10 O− 7.00e-10 O− 1.50e-10 O− 1.50e-10 O− 7.80e-10 O− 5.50e-10 O− 8.40e-10 O− 1.65e-12 O− 2.80e-10 O− 2 + O2 2.50e-10 Na+ + K ⇌ K+ + Na 1.00e-11 Na + H ⇌ Na+ + H− 1.00e-11 C + γ → C+ + e− C(1D) + γ → C+ + e− C(1S) + γ → C+ + e− H + γ → H+ + e− He + γ → He+ + e− N + γ → N+ + e− O + γ → O+ + e− O(1D) + γ → O+ + e− O(1S) + γ → O+ + e− H− + γ → H+ + e− + e− H− + γ → H + e− C2 + γ → C(1D) + C(1D) C2 + γ → C+ 2 + e− CH + γ → CH+ + e− · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 2 + O2 3 + O2 2 + O2 3 + O2 0.00 0.00 0.00 0.00 -0.83 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 50900 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 627 627 611 630 631 632 633 606 634 635 636 636 636 603 603 630 622 603 603 637 628 603 604 638 606 100 100 639 640 640 641 642 643 644 645 645 646 646 647 647 648 122 Table 6—Continued # Type Reaction 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 pd pd pd pd pi pi pi pd pd pd pi pi pd pi pi pi pi pi pd pd pd pi pd pi pd pd pi pd pd pi pi pi pi pd pd pd pd pi pi pi pi 2 + e− 2 + e− CH + γ → C(1S) + H CH + γ → C(1D) + H CH + γ → C + H CN + γ → C + N CO + γ → CO+ + e− CO + γ → C+ + O + e− CO + γ → O+ + C + e− CO + γ → C + O CO + γ → C(1D) + O(1D) H2 + γ → H + H H2 + γ → H+ H2 + γ → H+ + H + e− N2 + γ → N + N N2 + γ → N+ + N + e− N2 + γ → N+ NO + γ → NO+ + e− NO + γ → O+ + N + e− NO + γ → N+ + O + e− NO + γ → N + O O2 + γ → O + O O2 + γ → O + O(1D) O2 + γ → O+ + O + e− O2 + γ → O(1S) + O(1S) O2 + γ → O+ 2 + e− HO + γ → O + H HO + γ → O(1S) + H HO + γ → HO+ + e− HO + γ → O(1D) + H CO2 + γ → CO + O(1D) CO2 + γ → CO+ 2 + e− CO2 + γ → CO+ + O + e− CO2 + γ → O+ + CO + e− CO2 + γ → C+ + O2 + e− CO2 + γ → CO + O H2O + γ → HO + H H2O + γ → H2 + O(1D) H2O + γ → O + H + H H2O + γ → HO+ + H + e− H2O + γ → O+ + H2 + e− H2O + γ → H+ + HO + e− H2O + γ → H2O+ + e− 123 α · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · β · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · γ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · Ref 648 648 648 649 650 651 651 652 652 653 653 653 654 654 654 655 655 655 655 656 656 656 656 656 657 657 657 657 658 658 658 658 658 658 659 659 659 659 659 659 659 Table 6—Continued # Type Reaction 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 pd pi pd pd pd pd pi pd pd pd pd pi pi pd pd pd pd pd pi pi pi pd pd pd pd pd pd pd pi pi pi pd pd pd pd pd pi pi pi pd pd 2 + e− HCN + γ → CN + H HCN + γ → CHN+ + e− HO2 + γ → HO + O N2O + γ → N2 + O(1S) N2O + γ → N2 + O(1D) H2N + γ → HN + H NO2 + γ → NO+ 2 + e− NO2 + γ → NO + O(1D) NO2 + γ → NO + O O3 + γ → O2 + O O3 + γ → O2 + O(1S) C2H2 + γ → C2H+ C2H2 + γ → C2H+ + H + e− C2H2 + γ → C2 + H2 C2H2 + γ → C2H + H CH2O + γ → CHO + H CH2O + γ → CO + H2 CH2O + γ → CO + H + H CH2O + γ → CH2O+ + e− CH2O + γ → CHO+ + H + e− CH2O + γ → CO+ + H2 + e− H2O2 + γ → HO + HO HNCO + γ → HN + CO HNCO + γ → NCO + H HNO2 + γ → HO + NO H3N + γ → H2N + H H3N + γ → HN∗ + H2 H3N + γ → HN + H + H H3N + γ → H3N+ + e− H3N + γ → H2N+ + H + e− H3N + γ → HN+ + H2 + e− NO3 + γ → NO2 + O NO3 + γ → NO + O2 CH4 + γ → CH∗ 2 + H2 CH4 + γ → CH3 + H CH4 + γ → CH2 + H + H CH4 + γ → CH+ 4 + e− CH4 + γ → CH+ 3 + H + e− CH4 + γ → CH+ 2 + H2 + e− CH4 + γ → CH + H2 + H HCOOH + γ → CO2 + H2 124 α · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · β · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · γ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · Ref 660 660 661 662 662 663 664 664 664 665 665 666 666 666 666 667 667 667 667 667 667 667 668 668 669 670 670 670 670 670 670 671 671 672 672 672 672 672 672 672 673 Table 6—Continued # Type Reaction 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 pi pi pd pd pi pi pi pd pd pd pd pd pi pd pd pd pd pi pd pd pd pd pd pd pi pi pd pd cr cr cr cr cr cr cr cr cr cr cr cr cr 2 + e− HCOOH + γ → CH2O+ HCOOH + γ → CHO+ + HO + e− HCOOH + γ → CHO + HO HNO3 + γ → NO2 + HO C2H4 + γ → C2H+ 4 + e− C2H4 + γ → C2H+ 2 + H2 + e− C2H4 + γ → C2H+ 3 + H + e− C2H4 + γ → C2H2 + H + H C2H4 + γ → C2H2 + H2 C2H6 + γ → CH3 + CH3 C2H6 + γ → C2H5 + H C2H6 + γ → CH∗ 2 + CH4 C2H6 + γ → C2H+ 6 + e− C2H6 + γ → C2H4 + H2 C2H4O + γ → CH4 + CO C2H4O + γ → CH3 + CHO CH3OH + γ → CH3 + HO CH3OH + γ → CH4O+ + e− CH3OH + γ → CH2O + H2 C4H2 + γ → C2H2 + C2 C4H2 + γ → C2H + C2H C4H2 + γ → C4H + H C4H4 + γ → C4H2 + H2 C4H4 + γ → C2H2 + C2H2 Na + γ → Na+ + e− K + γ → K+ + e− HCl + γ → H + Cl N2O3 + γ → NO2 + NO C + CR → C+ + e− + CR Fe + CR → Fe+ + e− + CR H + CR → H+ + e− + CR He + CR → He+ + e− + CR Mg + CR → Mg+ + e− + CR N + CR → N+ + e− + CR O + CR → O+ + e− + CR Si + CR → Si+ + e− + CR C2 + CR → C + C + CR CH + CR → C + H + CR CN + CR → C + N + CR CO + CR → C + O + CR CO + CR → CO+ + e− + CR α · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · β · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · γ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · Ref 673 673 673 674 675 675 675 675 675 676 676 676 676 676 677 677 678 678 678 687 687 687 688 688 679 680 681 684 682 682 682 682 682 682 682 682 682 682 682 682 682 125 Table 6—Continued # Type Reaction 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr 2 + e− + CR H2 + CR → H + H + CR H2 + CR → H+ + H + e− + CR H2 + CR → H+ + H− + CR H2 + CR → H+ 2 + e− + CR N2 + CR → N + N + CR HN + CR → N + H + CR NO + CR → N + O + CR NO + CR → NO+ + e− + CR O2 + CR → O + O + CR O2 + CR → O+ 2 + e− + CR HO + CR → O + H + CR SiH + CR → Si + H + CR SiO + CR → Si + O + CR C2H + CR → C2 + H + CR C2N + CR → C + CN + CR C2O + CR → C2 + O + CR C2O + CR → CO + C + CR CH2 + CR → CH+ CO2 + CR → CO + O + CR H2O + CR → HO + H + CR HCN + CR → CN + H + CR CHO + CR → CO + H + CR CHO + CR → CHO+ + e− + CR HNC + CR → CN + H + CR HNO + CR → HNO+ + e− + CR N2O + CR → NO + N + CR H2N + CR → HN + H + CR H2N + CR → H2N+ + e− + CR NO2 + CR → NO + O + CR HO2 + CR → O + HO + CR HO2 + CR → O2 + H + CR CNO + CR → CN + O + CR SiH2 + CR → SiH + H + CR C2H2 + CR → C2H + H + CR C2H2 + CR → C2H+ CH3 + CR → CH2 + H + CR CH3 + CR → CH+ 3 + e− + CR CH2O + CR → CO + H2 + CR H2O2 + CR → HO + HO + CR H3N + CR → HN + H2 + CR H3N + CR → H2N + H + CR 2 + e− + CR 126 α · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · β · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · γ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · Ref 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 # Type Reaction α β Table 6—Continued 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr cr dr dr dr dr dr dr dr dr dr dr dr dr H3N + CR → H3N+ + e− + CR C2H2O + CR → CH2 + CO + CR C2H2O + CR → C2H2O+ + e− + CR C2H3 + CR → C2H2 + H + CR CH2O2 + CR → CHO + HO + CR CH2O2 + CR → CH2O+ 2 + e− + CR CH3N + CR → HCN + H2 + CR CH4 + CR → CH2 + H2 + CR HC3N + CR → C2H + CN + CR SiH4 + CR → SiH2 + H2 + CR C2H3N + CR → CH3 + CN + CR C2H3N + CR → C2H3N+ + e− + CR C2H4 + CR → C2H2 + H2 + CR C2H4 + CR → C2H+ 4 + e− + CR CH3OH + CR → CH3O+ + H + e− + CR CH3OH + CR → CH4O+ + e− + CR C2H4O + CR → CH3 + CHO + CR C2H4O + CR → CH4 + CO + CR C2H4O + CR → C2H4O+ + e− + CR C2H5 + CR → C2H4 + H + CR C3H3N + CR → C2H3 + CN + CR C3H4 + CR → C3H3 + H + CR C3H4 + CR → C3H+ 4 + e− + CR CH5N + CR → HCN + H2 + H + CR CH+ + CR → C + H+ + CR Cl + CR → Cl+ + e− + CR K + CR → K+ + e− + CR Na + CR → Na+ + e− + CR NaH + CR → Na + H + CR C+ 2 + e− → C + C CH+ + e− → C + H CN+ + e− → C + N CO+ + e− → O + C H+ 2 + e− → H + H HeH+ + e− → H + He N+ 2 + e− → N + N HN+ + e− → N + H NO+ + e− → N + O O+ 2 + e− → O + O HO+ + e− → O + H SiH+ + e− → Si + H · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 8.84e-08 7.00e-08 3.38e-07 2.75e-07 2.53e-07 3.00e-08 1.80e-07 1.18e-07 4.10e-07 1.95e-07 6.30e-09 2.00e-07 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · -0.50 -0.50 -0.55 -0.55 -0.50 -0.50 -0.39 -0.50 -1.00 -0.70 -0.48 -0.50 γ · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · 0 0 0 0 0 0 0 0 0 0 0 0 Ref 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 682 683 683 683 683 683 683 683 683 683 683 683 683 127 Table 6—Continued # Type Reaction α β γ Ref 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr SiO+ + e− → Si + O 2.00e-07 C2H+ + e− → C2 + H 1.16e-07 C2H+ + e− → CH + C 1.05e-07 C2H+ + e− → C + C + H 4.80e-08 C2N+ + e− → C + CN 1.50e-07 C2N+ + e− → C2 + N 1.50e-07 C2O+ + e− → CO + C 3.00e-07 CH+ 2 + e− → C + H2 7.70e-08 CH+ 2 + e− → CH + H 1.60e-07 CH+ 2 + e− → C + H + H 4.00e-07 SiCH+ + e− → Si + CH 1.50e-07 CO+ 2 + e− → O + CO 4.20e-07 H2O+ + e− → O + H2 3.90e-08 H2O+ + e− → HO + H 8.60e-08 H2O+ + e− → O + H + H 3.05e-07 H+ 3 + e− → H + H + H 4.36e-08 H+ 3 + e− → H2 + H 2.34e-08 CHN+ + e− → CN + H 2.00e-07 CHO+ + e− → CO + H 2.80e-07 CHN+ + e− → CN + H 2.00e-07 HNO+ + e− → NO + H 3.00e-07 SiHO+ + e− → Si + HO 1.50e-07 SiHO+ + e− → SiO + H 1.50e-07 HN+ 2 + e− → N2 + H 9.00e-08 HN+ 2 + e− → HN + N 1.00e-08 CNO+ + e− → CO + N 3.00e-07 H2N+ + e− → N + H + H 2.00e-07 H2N+ + e− → HN + H 1.00e-07 NO+ 2 + e− → O + NO 3.00e-07 HO+ 2 + e− → O2 + H 3.00e-07 SiH+ 2 + e− → Si + H + H 2.00e-07 SiH+ 2 + e− → Si + H2 1.50e-07 SiH+ 2 + e− → SiH + H 2.00e-07 C2H+ 2 + e− → C2H + H 2.90e-07 C2H+ 2 + e− → C2 + H + H 1.70e-07 C2H+ 2 + e− → CH + CH 7.50e-08 C2H+ 2 + e− → CH2 + C 2.89e-08 C2H+ 2 + e− → C2 + H2 1.15e-08 C2HO+ + e− → CO + C + H 1.50e-07 C2HO+ + e− → CO + CH 1.00e-07 C2HO+ + e− → C2H + O 1.50e-07 -0.50 -0.76 -0.76 -0.76 -0.50 -0.50 -0.50 -0.60 -0.60 -0.60 -0.50 -0.75 -0.50 -0.50 -0.50 -0.52 -0.52 -0.50 -0.69 -0.50 -0.50 -0.50 -0.50 -0.51 -0.51 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 128 Table 6—Continued # Type Reaction α β γ Ref 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr C2HO+ + e− → C2O + H 1.00e-07 C2N+ 2 + e− → CN + CN 1.50e-07 C2N+ 2 + e− → CNC + N 1.50e-07 C2HN+ + e− → CH + CN 1.50e-07 C2HN+ + e− → CNC + H 1.50e-07 C3H+ + e− → C2H + C 1.50e-07 C3N+ + e− → C2 + CN 3.00e-07 SiCH+ 2 + e− → Si + CH2 2.00e-07 CH+ 3 + e− → H2 + C + H 3.00e-07 CH+ 3 + e− → CH + H + H 1.60e-07 CH+ 3 + e− → CH + H2 1.40e-07 CH+ 3 + e− → CH2 + H 4.00e-07 CH2O+ + e− → CO + H + H 5.00e-07 CH2O+ + e− → CHO + H 1.00e-07 H3O+ + e− → HO + H + H 2.60e-07 H3O+ + e− → H2O + H 1.10e-07 H3O+ + e− → HO + H2 6.00e-08 H3O+ + e− → H2 + H + O 5.60e-09 CHO+ 2 + e− → CO + H + O 8.10e-07 CHO+ 2 + e− → HO + CO 3.20e-07 CHO+ 2 + e− → CO2 + H 6.00e-08 CHNO+ + e− → CO + HN 3.00e-07 H3N+ + e− → HN + H + H 1.55e-07 H3N+ + e− → H2N + H 1.55e-07 SiC2H+ + e− → C2H + Si 1.50e-07 SiH+ 3 + e− → SiH + H2 1.50e-07 SiH+ 3 + e− → SiH2 + H 1.50e-07 C2H2N+ + e− → CH + HCN 3.00e-07 C2H2N+ + e− → CN + CH2 3.00e-07 C2H2N+ + e− → C2N + H2 3.00e-07 C2H2O+ + e− → C2 + H2O 2.00e-07 C2H2O+ + e− → CH2 + CO 2.00e-07 C2H2O+ + e− → C2H2 + O 2.00e-07 C2H+ 3 + e− → C2H + H + H 2.95e-07 C2H+ 3 + e− → C2H + H2 3.00e-08 C2H+ 3 + e− → C2H2 + H 1.45e-07 C2H+ 3 + e− → C2 + H + H2 1.50e-08 C2H+ 3 + e− → CH3 + C 3.00e-09 C2H+ 3 + e− → CH2 + CH 1.50e-08 C3H+ 2 + e− → C2H2 + C 3.00e-08 C3HN+ + e− → C2 + HCN 3.00e-07 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.30 -0.30 -0.30 -0.30 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.64 -0.64 -0.64 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.84 -0.84 -0.84 -0.84 -0.84 -0.84 -0.50 -0.50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 129 Table 6—Continued # Type Reaction α β γ Ref 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr 2 + e− → CO2 + H + H C3HN+ + e− → C2H + CN 1.50e-07 CH2O+ 3.00e-07 CH+ 4 + e− → CH2 + H + H 3.00e-07 CH+ 4 + e− → CH3 + H 3.00e-07 CH3O+ + e− → CO + H + H2 2.00e-07 CH3O+ + e− → CHO + H + H 2.00e-07 CH3O+ + e− → CH2O + H 2.00e-07 SiH3O+ + e− → SiO + H2 + H 1.50e-07 H4N+ + e− → H2N + H + H 3.20e-07 H4N+ + e− → H2N + H2 1.50e-07 H4N+ + e− → H3N + H 1.05e-06 SiH+ 4 + e− → SiH2 + H2 1.50e-07 SiH+ 4 + e− → SiH3 + H 1.50e-07 C2H3N+ + e− → C2N + H2 + H 2.00e-07 C2H3N+ + e− → CH2 + HCN 3.00e-07 C2H3N+ + e− → CH3 + CN 2.00e-07 C2H3O+ + e− → CH3 + CO 1.50e-07 C2H3O+ + e− → C2H2O + H 1.50e-07 C2H+ 4 + e− → C2H2 + H + H 3.70e-07 C2H+ 4 + e− → C2H2 + H2 3.36e-08 C2H+ 4 + e− → C2H3 + H 6.16e-08 C2H+ 4 + e− → C2H + H2 + H 5.60e-08 C2H+ 4 + e− → CH4 + C 5.60e-09 C2H+ 4 + e− → CH3 + CH 1.12e-08 C2H+ 4 + e− → CH2 + CH2 2.24e-08 C3H2N+ + e− → C2H + HNC 7.50e-08 C3H2N+ + e− → HC3N + H 1.50e-07 C3H+ 3 + e− → C2H2 + CH 6.99e-08 CH3O+ 2 + e− → CO2 + H2 + H 1.50e-07 CH3O+ 2 + e− → CH2O2 + H 1.50e-07 CH4N+ + e− → CN + H2 + H2 3.00e-08 CH4N+ + e− → CH2 + H2N 1.50e-07 CH4N+ + e− → HCN + H + H2 3.00e-07 CH4N+ + e− → CH3N + H 1.50e-07 CH4O+ + e− → CH3 + HO 3.00e-07 CH4O+ + e− → CH2O + H2 3.00e-07 CH+ 1.40e-08 CH+ 1.40e-08 CH+ 1.95e-07 CH+ 4.80e-08 CH+ 3.00e-09 5 + e− → CH3 + H2 5 + e− → CH4 + H 5 + e− → CH3 + H + H 5 + e− → CH2 + H2 + H 5 + e− → CH + H2 + H2 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.76 -0.76 -0.76 -0.76 -0.76 -0.76 -0.76 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.52 -0.52 -0.52 -0.52 -0.52 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 130 Table 6—Continued # Type Reaction α β γ Ref 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr 5 + e− → SiH3 + H2 5 + e− → SiH4 + H SiH+ SiH+ C2H4N+ + e− → CH2CN + H + H C2H4N+ + e− → C2H3N + H C2H4O+ + e− → CH3 + CHO C2H4O+ + e− → C2H2O + H + H C2H4O+ + e− → C2H2O + H2 C2H+ 5 + e− → C2H + H2 + H2 C2H+ 5 + e− → C2H2 + H2 + H C2H+ 5 + e− → C2H3 + H2 C2H+ 5 + e− → C2H4 + H C3H+ 4 + e− → C3H3 + H C3H+ 4 + e− → C2H3 + CH C3H+ 4 + e− → C2H2 + CH2 C3H+ 4 + e− → C2H + CH3 CH5N+ + e− → CH3 + H2N CH5N+ + e− → CH3N + H2 CH5O+ + e− → CH2O + H2 + H CH5O+ + e− → CH3 + H2O CH5O+ + e− → CH3 + HO + H CH5O+ + e− → CH2 + H2O + H CH5O+ + e− → CH3OH + H C4H2N+ + e− → HC3N + CH C2H5O+ + e− → CH2O + CH3 C2H5O+ + e− → C2H2O + H2 + H C2H5O+ + e− → CH4 + CO + H C2H5O+ + e− → C2H4O + H C2H+ C2H+ C3H+ C3H+ C4H+ CH6N+ + e− → CH3N + H2 + H CH6N+ + e− → CH5N + H C2H6O+ + e− → C2H2O + H2 + H2 C2H6O+ + e− → CH4 + CH2O C2H6O+ + e− → C2H4O + H2 C2H6O+ + e− → C2H5 + HO C2H7O+ + e− → C2H4 + H2O + H C2H7O+ + e− → C2H4O + H2 + H C2H7O+ + e− → C2H5OH + H 6 + e− → C2H4 + H2 6 + e− → C2H5 + H 5 + e− → C3H3 + H2 5 + e− → C3H4 + H 4 + e− → C4H3 + H 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 3.00e-07 1.50e-07 1.50e-07 2.57e-06 2.95e-08 1.77e-07 2.95e-08 1.50e-07 1.50e-07 9.10e-08 8.19e-08 4.64e-07 1.91e-07 2.73e-08 3.00e-07 1.50e-07 1.50e-07 3.00e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 3.30e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 1.50e-07 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.67 -0.67 -0.67 -0.67 -0.50 -0.50 -0.67 -0.67 -0.67 -0.67 -0.67 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 131 Table 6—Continued # Type Reaction α β γ Ref 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr dr ra ra ra ra ra ra ra 2 + e− → C2H5NO2 + H 7 + e− → C3H3 + CH4 7 + e− → C2H3 + C2H4 7 + e− → C2H2 + C2H5 7 + e− → C2H5 + H2 7 + e− → C2H6 + H 6 + e− → C3H3 + H2 + H 6 + e− → C3H4 + H2 7 + e− → C2H2 + CH3 + H2 7 + e− → C2H3 + CH3 + H 7 + e− → C2H4 + CH2 + H 7 + e− → C3H4 + H2 + H 7 + e− → C2H6 + CH 2 + e− → C2H + C2H 2 + e− → C4H + H 8 + e− → C2H6 + CH2 9 + e− → C2H6 + CH3 9 + e− → C2H6 + C2H3 C3H6O+ + e− → CH3 + CH3 + CO 1.50e-07 C3H6O+ + e− → C2H4O + CH2 1.50e-07 C3H7O+ + e− → C2H4O + CH3 1.50e-07 C4H+ 1.95e-07 C4H+ 2.25e-08 C4H+ 2.25e-08 C2H6NO+ 2.34e-08 C2H+ 1.50e-07 C2H+ 1.50e-07 C3H+ 1.50e-07 C3H+ 1.50e-07 C3H+ 2.53e-08 C3H+ 4.37e-08 C3H+ 9.20e-09 C3H+ 6.60e-08 C3H+ 1.50e-07 C4H+ 2.75e-07 C4H+ 8.25e-07 C3H+ 1.50e-07 C3H+ 1.50e-07 C4H+ 1.50e-07 H + CH3 → CH4 + γ 1.31e-16 H + C2H4 → C2H5 + γ 4.46e-17 5.64e-13 H + C4H2 → C4H3 + γ 6.22e-12 H + C4H3 → C4H4 + γ CH3+ CH3 → C2H6 + γ 2.96e-14 H+ + e− → H + γ 3.50e-12 He++ e− → He + γ 5.36e-12 -0.50 -0.50 -0.50 -0.50 -0.50 -0.50 -0.52 -0.50 -0.50 -0.50 -0.50 -0.73 -0.73 -0.73 -0.73 -0.73 -0.79 -0.79 -0.50 -0.50 -0.50 -1.29 -0.53 2.75 -3.01 -3.23 -0.75 -0.50 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 20 19 50 162 75 0 0 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 683 31 31 31 31 31 689 690 ∗Consult Section 2.3 on how the reaction barrier is calculated. (4) Mick et al. 1993; (5) Baulch et al. 1992; (6) Deppe et al. 1998; (2) Dean & Hanson 1992; References. — (1) Kruse & Roth 1997; (3) Tsang (7) 1992; Tsang & Hampson 1986; (8) Thielen & Roth 1986; (9) Myerson 1973; (10) Javoy et al. 2003; (11) De Cobos & Troe 1984; (12) Bauerle et al. 1995; (13) Husain & Young 1975; (14) Lin et al. 1992; (15) Wagner & Bowman 1987; (16) Mertens & Hanson 1996; (19) Ross et al. 1997; (20) Fulle & Hippler 1997; (21) Moses et al. 2005; (22) Yumura & Asaba 1981; (23) Hanson & Salimian 1984; (24) Graham & Johnston 1978a; (25) Bozzelli & Dean 1995; (26) Natarajan et al. 1986; (27) Troe 2005; (28) Wu et al. 1990; (29) Lifshitz et al. 1997; (30) Baulch et al. 2005; (31) Vuitton et al. 2012; (32) Atkinson et al. 2004; (33) (17) Eremin et al. 1997; (18) Tsang & Herron 1991; 132 (41) Frank et al. 1988; (42) Kiefer et al. 1988; (43) Kern et al. 1988; Baulch et al. 1994; (34) Meyer et al. 1969; (35) Dorko et al. 1979; (36) Klatt et al. 1995; (37) Schulz et al. 1985; (38) Fernandes et al. 2005; (39) Dean 1985; (40) Li et al. 2006; (44) Yang et al. 2005; (45) Warnatz 1984; (46) Chang & Yu 1995; (47) Cribb et al. 1992; (48) Setser & Rabinovitch 1962; (49) Chang et al. 2007; (50) Saito et al. 1984; (51) Ikeda & Mackie 1996; (52) Wakamatsu & Hidaka 2008; (53) O’Neal & Benson 1962; (54) Atkinson et al. 1997; (55) Huynh & Violi 2008; (56) Dombrowsky et al. 1991; (57) Koike et al. 2000; (58) Ing et al. 2003; (59) Yasunaga et al. 2008; (60) Herron 1999; (61) Lifshitz & Tamburu 1998; (62) Lifshitz et al. 1993; (63) Oehlschlaeger et al. 2004; (64) Saito et al. 1990; (65) Lee & Bozzelli 2003; (66) Miller et al. 2004; (67) Friedrichs et al. 2008; (68) Joshi et al. 2005; (69) Almatarneh et al. 2005; (70) Zaslonko et al. 1997; (71) Fern´andez-Ramos et al. 1998; (72) Zaslonko et al. 1993; (73) Tsang 2004; (74) Cook et al. 2009; (75) Imai & Toyama 1962; (76) Hinshelwood & Askey 1927; (77) Hunt et al. 1965; (78) Sheng et al. 2002; (79) DeSain et al. 2003; (80) Chuchani et al. 1993; (81) Petrov et al. 2009; (82) Batt & Rattray 1979; (83) Blake & Jackson 1969; (84) Duan & Page 1995; (85) Spokes & Benson 1967; (86) Glanzer & Troe 1973; (87) Levy 1956; (88) Batt et al. 1975; (89) Li et al. 2004; (90) Natarajan & Bhaskaran 1981; (91) Arenas et al. 2000; (92) Zalotai et al. 1983; (93) Lifshitz & Tamburu 1994; (94) Sato & Hidaka 2000; (95) Hoyermann et al. 1999; (96) Zhang et al. 2005; (97) Zhang et al. 2004; (98) Curran 2006; (99) Patrick & Golden 1984; (100) Lavvas et al. 2014; (101) Baulch et al. 1981; (102) Crosley 1989; (103) Kretschmer & Petersen 1963; (104) Fehsenfeld et al. 1974; (105) Smith et al. 1982; (106) Raksit & Warneck 1979; (107) Adams & Smith 1977; (108) Ferguson & Fehsenfeld 1968; (109) Graham et al. 1973; (110) Fehsenfeld et al. 1967a; (111) Burt et al. 1970; (112) Beaty & Patterson 1965;(113) Mark & Oskam 1971; (114) Fehsenfeld et al. 1975a; (115) Sieck 1978; (116) See Section 2.3; (117) Wang et al. 2001; (118) Slack & Fishburne 1970; (119) Dean et al. 1991; (120) Andersson et al. 2003; (121) Mayer et al. 1967; (122) Whyte & Phillips 1983; (123) Caridade et al. 2005; (124) Adam et al. 2005; (125) Cohen & Westberg 1991; (126) Miller et al. 2005; (127) Bauer et al. 1985; (128) Sumathi & Nguyen 1998; (129) Tsuboi & Hashimoto 1981; (130) Rohrig & Wagner 1994; (131) Nguyen et al. 2004; (132) Mousavipour & Saheb 2007; (133) Bozzelli et al. 1994; (134) Su et al. 2002; (135) Shaw 1977; (136) Louge & Hanson 1984; (137) Miller & Melius 1992; (138) Szekely et al. 1985; (139) Corchado & Espinosa-Garcıa 1997; (140) Linder et al. (143) Senosiain et al. 2006; 1996; (144) Loison et al. 2015; (147) Boughton et al. 1997; (148) Knyazev et al. 1996a; (149) Jamieson et al. 1970; (150) Ohmori et al. 1990; (151) Colberg & Friedrichs 2006; (152) Li & Williams 1996; (153) Vaghjiani 1995; (154) Sivaramakrishnan et al. 2009; (155) Lambert et al. 1967; (156) Lifshitz & Ben-Hamou 1983; (157) Zhang & Bauer 1997; (158) Moortgat et al. 1977; (159) Aders & Wagner 1973; (160) Mayer et al. 1966; (161) Brownsword et al. 1996; (162) Ibragimova 1986; (163) Harding, et al. 1993; (164) Duff & Sharma 1996; (165) Bose & Candler 1996; (166) Brunetti & Liuti 1975; (167) Lifshitz & Frenklach 1980; (168) Avramenko & Krasnen’kov 1966; (169) Sander et al. 2011; (170) Barnett et al. 1987; (171) Safrany & Jaster 1968; (172) Cimas & Largo 2006; (173) Wagner et al. 1971; (177) Paraskevopoulos & Winkler 1967; (179) Roscoe & Roscoe 1973; (180) Lambert et al. 1968; (181) Fairbairn 1969; (182) Murrell & Rodriguez 1986; (186) Harding & Wagner 1989; (187) Karkach & Osherov 1999; (188) Meagher & Anderson 2000; (189) Cvetanovi´c 1987; (190) Hack et al. 2005; (191) Dean & Kistiakowsky 1971; (146) Morris & Niki 1973; (178) Forst et al. 1957; (184) Zhu & Lin 2007; (174) Xu & Sun 1999; (175) Sun et al. 2004; (176) Takahashi 1972; (185) Korovkina 1976; (183) Frank 1986; (141) Hsu et al. 1997; (142) He et al. 1993; (145) Tsang 1987; 133 (210) Kruse & Roth 1999; (211) Fontijn et al. 2001; (232) You et al. 2007; (233) Tsuboi et al. 1981; (223) Tzeng et al. 2009; (224) Wang et al. 2002; (249) Humpfer et al. 1995; (252) Corchado et al. 1995; (248) Jasper et al. 2007; (251) Wooldridge et al. 1996; (192) Harding et al. 2005; (193) Corchado et al. 1998; (194) Cobos & Troe 1985; (195) Sridharan & Kaufman 1983; (196) Mahmud et al. 1987; (197) Westenberg & De Haas 1969; (198) Grotheer & Just 1981; (199) Gehring et al. 1969; (200) Mayer & Schieler 1968; (201) Bogan & Hand 1978; (202) Miyoshi et al. 1993; (203) Kato & Cvetanovic 1967; (204) Wu et al. 2007; (205) Takahashi et al. 2007; (206) Patterson & Greene 1962; (207) Meaburn & Gordon 1968; (208) Pshezhetskii et al. 1959; (209) Wilson 1972; (212) Harding et al. 2008; (213) Matsui & Nomaguchi 1978; (214) Geiger et al. 1999; (215) Bergeat et al. 1998; (216) Jachimowski 1977; (217) Lichtin et al. 1984; (218) Bergeat et al. 2009; (219) Tao et al. 2001; (220) Blitz et al. 1997; (221) Mulvihill & Phillips 1975; (222) Rim & Hershberger 1999; (225) Park & Hershberger 1993; (226) Gannon et al. 2007; (227) Sayah et al. 1988; (228) Feng & Hershberger 2007; (229) Pang et al. 2008; (230) Sun et al. 2006; (231) Sims et al. 1993; (234) Roose et al. 1978; (235) Baldwin et al. 1961; (236) Davidson et al. 1990; (237) Quandt & Hershberger 1995; (238) Xu & Sun 1998; (239) Rohrig et al. 1994; (240) Miller & Melius 1988; (241) Campomanes et al. 2001; (242) Gonzalez et al. 1992; (243) Mebel et al. 1996; (244) Srinivasan et al. 2007; (245) Ju et al. 2007; (246) Miller & Melius 1989; (247) Breen & Glass 1971; (250) Xu & Lin 2007; (253) Grussdorf et al. 1994; (255) Vandooren & Van Tiggelen 1977; (256) Bryukov et al. 2004; (257) Espinosa-Garcia et al. 1993; (258) Anglada 2004; (259) Lamb et al. 1984; (260) Liu et al. 2002; (261) Li & Wang 2004; (262) Srinivasan et al. 2007; (263) Baldwin et al. 1984; (264) Cohen 1991; (265) Nielsen et al. 1991; (266) Atkinson et al. 2001; (267) Wu et al. 2003; (268) Zabarnick & Heicklen 1985; (269) Sanders et al. 1987; (270) Thweatt et al. 2004; (271) Dammeier et al. 2007; (272) Lin et al. 1993; (273) Miller & Glarborg 1999; (274) Park & Lin 1997; (275) Vandooren et al. 1994; (276) Howard 1979; (277) Opansky & Leone (1996a,b); (278) Benson 1994; (279) Hennig & Wagner 1994; (280) Tomeczek & Grado´n 2003; (281) Mebel & Lin 1997; (282) Striebel et al. 2004; (283) Laidler & Wojciechowski 1961; (284) McKenney et al. 1963; (285) Becker et al. 1992b; (286) Williamson & Bayes 1967; (287) Alvarez & Moore 1994; (288) Dombrowsky & Wagner 1992; (289) Nadtochenko et al. 1979; (290) Zhu & Lin 2005; (291) Schacke et al. 1974; (292) Bozzelli & Dean 1989; (293) Tang et al. 2008; (294) Reitel’boim et al. 1978; (295) Seery 1969; (296) Michael et al. 1999; (297) Marinov et al. 1998; (298) Bogdanchikov et al. 2004; (299) Shaw 1978; (300) Neiman & Feklisov 1961; (301) Bozzelli & Dean 1990; (302) Baker et al. 1971; (303) Atkinson et al. 1973; (304) Yang et al. 2008; (305) Hastie et al. 1976; (306) Morrissey & Schubert 1963; (308) Yee Quee & Thynne 1968; (309) Stothard et al. 1995; (310) Sahetchian et al. 1987; (311) Hidaka et al. 1990; (312) Dong et al. 2005; (313) Baker et al. 1969; (314) Homann & Wellmann 1983; (315) Frank et al. 1986; (316) Bohland et al. 1985; (317) Monks et al. 1993; (318) Xu & Lin 2004; (319) Huynh & Truong 2008; (320) Gao & Macdonald 2006; (321) Macdonald 2007; (322) Sumathi & Peyerimhoff 1996; (323) Glarborg et al. 1995; (324) Bulatov et al. 1980; (325) Mertens et al. 1991; (326) Thaxton et al. 1997; (327) Song et al. 2003; (328) Mebel & Lin 1999; (329) England & Corcoran 1975; (330) Choi & Lin 2005; (331) He et al. 1988; (332) Glanzer & Troe 1975; (333) Lloyd 1974; (334) Vardanyan et al. 1974; (335) Becker et al. 1992a; (336) Carstensen & Dean 2008; (337) Borisov et al. 1977; (338) Trenwith 1960; (339) Srinivasan et al. 2005; (340) Canosa et al. 1979; (341) Johnston 1951; (342) Chan et al. 2001; (343) Atkinson et al. 1989;a(344) Christie & Voisey 1967; (345) Anastasi & Hancock 1988; (346) Fifer 1975; (254) Faravelli et al. 2000; (307) Haworth et al. 2003; 134 (360) Seetula et al. 1986; (415) Wijnen 1960; (416) Heicklen & Johnston 1962; (353) Metcalfe et al. 1983; (354) Adachi et al. 1981; (363) Thynne & Gray 1963a; (366) Bravo-P´erez et al. 2002; (347) Lindley et al. 1979; (348) H´ebrard et al. 2013; (349) Laufer & Fahr 2004; (350) Woods & Haynes 1994; (351) Skinner & Ruehrwein 1959; (352) Tabayashi & Bauer 1979; (355) Hassinen et al. 1990; (356) Ceursters et al. 2001; (357) Baldwin et al. 1971; (358) Saeys et al. 2006; (361) Gray & Herod 1968; (359) Donovan et al. 1971; (362) Hidaka et al. 2000; (364) Sanders et al. 1980; (365) Musin & Lin 1998; (367) Daele et al. 1995; (368) Kukui et al. 1995; (369) Langer & Ljungstrom 1995; (370) Biggs et al. 1995; (371) Langer & Ljungstrom 1994; (372) Sun et al. 2001; (373) Peeters et al. 1995; (374) Carl et al. 2003; (375) Osborn 2003; (376) Meyer & Hershberger 2005; (377) Petty et al. 1993; (378) Nizamov & Dagdigian 2003; (379) Chakraborty & Lin 1999; (380) Duran et al. 1988; (381) Knyazev et al. 1996b; (382) Levush et al. 1969; (383) Canosa-Mas et al. 1988; (384) Tanzawa & Gardiner 1980; (385) Moses (priv. comm.); (386) Adamson et al. 1997; (387) Wallington & Japar 1989; (388) Fahr & Stein 1989; (389) Scherzer et al. 1987; (390) Kelly & Heicklen 1978; (391) Thynne & Gray 1963b; (392) Moshkina et al. 1980; (393) Edwards et al. 1966; (394) Hoehlein & Freeman 1970; (395) Laidler & McKenney 1964; (396) Carstensen & Dean 2005; (397) Horne & Norrish 1970; (398) Herron 1988; (399) Atkinson et al. 1992; (400) Veyret et al. 1982; (401) Young 1958; (402) Gill et al. 1981; (403) Zhu & Lin 2009; (404) Hassinen et al. 1985; (405) Tuazon et al. 1984; (406) Edelbuttel-Einhaus et al. 1992; (407) Song et al. 2005; (408) Suzaki et al. 2007; (409) Suzaki et al. 2006; (410) Batt et al. 1977; (411) Ray et al. 1996; (412) Natarajan & Bhaskaran 1981; (413) Klemm 1965; (414) Hartmann et al. 1990; (417) Jensen 1982; (418) Allison et al. 1996; (419) Bryukov et al. 2006; (420) Husain & Lee 1988; (421) Husain & Marshall 1986; (422) Bolden et al. 1970; (423) Laudenslager et al. 1974; (424) Raksit & Warneck 1980a; (425) Robertson et al. 1983; (426) Dotan & Lindinger 1982; (429) Thomas et al. 1978; (430) Shul et al. 1987a; (431) Lindinger et al. 1981; (432) Villinger et al. 1982; (433) Roche et al. 1971; (434) Lindinger 1973; (435) Lindinger et al. 1975f; (436) Raksit 1982; (437) Bohme et al. 1970; (438) Adams et al. 1970; (439) Shul et al. 1987b; (440) Raksit & Warneck 1981; (441) Huntress & Baldeschwieler 1969; (442) Anicich et al. 1976; (443) Freeman et al. 1978b; (444) Schildcrout & Franklin 1970; (445) Bohme et al. 1982; (446) Raksit & Bohme 1985; (447) Cheng et al. 1973; (448) Cermak et al. 1970; (449) Adams & Smith 1978; (450) Adams et al. 1978a; (451) Szabo & Derrick 1971; (452) Smith & Adams 1977b; (453) Viggiano et al. 1980; (454) Schiff & Bohme 1979; (455) Zielinska & Wincel 1970; (456) McEwan et al. 1981; (457) Liddy et al. 1977a; (459) Wagner-Redeker et al. 1985; (460) Karpas & Klein 1975; (461) Tanner et al. 1979a; (462) Mackay & Bohme 1978; (463) Adams et al. 1978b; (464) Mackay et al. 1977; (465) Mackay et al. 1976a; (466) Okada et al. 1972; (468) Kumakura et al. 1978a; (469) Raksit & Bohme 1984; (470) Matsumoto et al. 1975; (471) Kasper & Franklin 1972; (472) Betowski et al. 1975; (473) Anicich et al. 1986; (474) Debrou et al. 1978; (475) Freeman et al. 1978a; (476) Fluegge 1969a; (477) Smith & Adams 1978; (478) Kumakura et al. 1978b; (479) Bohme & Raksit 1985; (480) Kim et al. 1975; (481) Huntress et al. 1973; (482) Huntress & Elleman 1970; (483) Fehsenfeld 1976; (484) Fluegge 1969b; (485) Huntress et al. 1980; (486) Warneck 1972; (487) Sieck & Futrell 1968; (490) Tanner et al. 1979b; (493) Hemsworth et al. 1973; (491) Mackay et al. 1978; (494) Munson & Field 1969; (495) Hopkinson et al. 1979; (496) Lindinger et al. 1975c; (497) Hemsworth et al. 1974; (498) Ikezoe et al. 1987; (499) Meot-Ner et al. 1986; (458) Wight & Beauchamp 1980; (488) McAllister 1973; (489) Gupta et al. 1967; (427) Karpas et al. 1978; (428) Chau & Bowers 1976; (467) Mackay et al. 1981; (492) Bohme et al. 1980; 135 (553) Ikezoe et al. 1987; (507) Fehsenfeld et al. 1970; (558) Kemper et al. 1983; (559) Fehsenfeld et al. 1978; (506) Fehsenfeld & Ferguson 1971; (500) Raksit et al. 1984; (501) McEwan et al. 1983; (502) Jaffe & Klein 1974; (503) Fehsenfeld & Ferguson 1972; (504) Raksit & Warneck 1980c; (505) Raksit & Warneck 1980b; (508) Ikezoe et al. 1987; (509) Miller et al. 1984; (510) Mackay et al. 1980; (511) Field et al. 1957; (512) Jarrold et al. 1983; (513) Mayer & Lampe 1974a; (514) Vogt et al. 1978; (515) Melton & Rudolph 1960; (516) Munson et al. 1964; (517) Buttrill et al. 1974; (518) Mayer & Lampe 1974b; (519) Kumakura et al. 1979; (520) Strausz et al. 1970; (521) Hiraoka & Kebarle 1980; (522) Ikezoe et al. 1987; (523) Ausloos & Lias 1981; (524) Tanner et al. 1979c; (525) Dheandhanoo et al. 1984; (526) Dunbar et al. 1972; (527) Sieck & Searles 1970; (528) Matsuoka & Ikezoe 1988; (529) Clary et al. 1985; (530) Armentrout et al. 1978; (531) Adams et al. 1980; (532) Smith & Adams 1980; (533) Herbst et al. 1975; (534) Fehsenfeld et al. 1976; (535) Howorka et al. 1974; (536) Ausloos 1975; (537) Jones et al. 1981a; (538) Lawson et al. 1976; (539) Ikezoe et al. 1987; (540) Fehsenfeld et al. 1975b; (541) Lindinger et al. 1975a; (542) Henis et al. 1973; (543) Cheng et al. 1974; (544) McAllister & Pitman 1976; (545) Karpas et al. 1979; (546) Bowers et al. 1969; (547) Smith et al. 1976; (548) Stevenson & Schissler 1955; (549) Barassin et al. 1983; (550) Huntress et al. 1971; (551) Dotan et al. 1980; (552) Durup-Ferguson et al. 1984; (554) Lindinger et al. 1975d; (555) Liddy et al. 1977b; (556) Adams & Smith 1976b; (557) Smith & Adams 1981; (560) Mackay et al. 1979; (561) Bohme & Mackay 1981; (562) Bohme et al. 1979; (563) Smith & Adams 1977a; (564) Ikezoe et al. 1987; (565) Cheng & Lampe 1973; (566) Neilson et al. 1978; (567) Kappes & Staley 1981; (568) Fehsenfeld et al. 1966b; Jones et al. 1979; (569) Adams & Smith 1976a; (570) Bolden & Twiddy 1972; Mauclaire et al. 1978; (571) Heimerl et al. 1969; (572) Theard & Huntress 1974; (573) Rowe et al. 1981; (574) Smith et al. 1978; (575) Tich`y et al. 1979; (576) Williamson & Beauchamp 1975; (577) Fehsenfeld 1977; (578) Fehsenfeld et al. 1969c; (579) Ferguson 1968; (580) Jaffe et al. 1973; (581) Johnsen et al. 1970; (582) Jones et al. 1981b; (583) Kemper & Bowers 1984; (584) Lindinger 1976; (585) Glosik et al. 1978; (586) Lindinger et al. 1974; (587) Dunkin et al. 1971; (588) Rowe et al. 1980; (589) Fahey et al. 1981; (590) Fehsenfeld 1969; (591) Lindinger et al. 1975e; (592) Lindinger et al. 1979; (593) Raksit 1986; (594) Tonkyn & Weisshaar 1986; (595) Johnsen et al. 1974; (596) Tanaka et al. 1976; (597) Fehsenfeld & Ferguson 1970; (598) Mackay et al. 1982; (599) Moylan et al. 1985; (600) Mackay et al. 1976b; (601) Ausloos 1975; (602) Ikezoe et al. 1987; (603) Fehsenfeld 1975; (604) Albritton et al. 1983; (605) Fehsenfeld & Ferguson 1974; (606) Ikezoe et al. 1987; (609) Bohme et al. 1971; (610) Lifshitz et al. 1978; (611) Lindinger et al. 1975b; (612) Fehsenfeld et al. 1973; (613) Dunkin et al. 1970; (614) Payzant et al. 1976; (615) Ferguson et al. 1969; (616) Tanner et al. 1981; (617) Bierbaum et al. 1976; (618) Faigle et al. 1976; (619) Howard et al. 1974; (622) Ikezoe et al. 1987; (623) Bierbaum et al. 1984; (624) Grabowski 1983; (625) Streit 1982; (626) McFarland et al. 1972; (627) Parkes 1972a; (628) Parkes 1972b; (629) Bohme & Fehsenfeld 1969; (630) Fehsenfeld et al. 1967b; (631) Viggiano & Paulson 1983; (632) Lifshitz & Tassa 1973; (633) Marx et al. 1973; (634) McKnight 1970; (635) Fehsenfeld et al. 1969; (636) Lifshitz et al. 1977; (637) Fahey et al. 1982; (638) Dunkin et al. 1972; (639) 0.1 A – 370 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996, 375.8 A – 1083 A: Badnell et al. 2005; (640) 0 A – 120 A: Barfield et al. 1972, 120 A – 310 A: Henry 1970, 313.3 A – 1226 A: Badnell et al. 2005; (641) Bethe & Salpeter 1957; (642) 0 A – 20 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996, 22.78 A – 504.26 A: Badnell et al. 2005; (643) 0.1 A – 250 (607) Fehsenfeld et al. 1969b; (608) Bohme et al. 1974; (620) Young et al. 1971; (621) Fehsenfeld et al. 1966a; 136 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996, 256.8 A – 851.3 A: Badnell et al. 2005; (644) 0.1 A – 230 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996, 235.9 A – 910.44 A: Badnell et al. 2005; (645) 1 A – 130 A: Barfield et al. 1972, 130 A – 200 A: Henry 1970, 227.3 A – 790.1 A: Badnell et al. 2005; (646) 180 A – 16640 A: Geltman 1962; Broad & Reinhardt 1976; (647) 0.6 A – 918 A: Barfield et al. 1972; Padial et al. 1985, 918 A – 1210 A: Pouilly et al. 1983; Padial et al. 1985; (648) 12 A – 617 A: Walker & Kelly 1972, 827 A – 1170 A: Barsuhn & Nesbet 1978; van Dishoeck 1987, 1200 A – 3589.9 A: van Dishoeck 1987, branching ratio: Barsuhn & Nesbet 1978; (649) 0.61 A – 626.8 A: Barfield et al. 1972, 905.8 A – 1108 A: Lavendy et al. 1984, 1987; (650) 89.6 A – 564.5 A: Masuoka & Samson 1981, 584.3 A – 835.29 A: Cairns & Samson 1965; Kronebusch & Berkowitz 1976; (651) 89.6 A – 584 A: Masuoka & Samson 1981; Kronebusch & Berkowitz 1976; (652) 0 A – 1117.8 A: McElroy & McConnell 1971; (653) 1 A – 200 A: σ(H2) ≈ 2σ(H), 209.3 A – 500 A: Samson & Cairns 1965; Browning & Fryar 1973, 500 A – 844 A: Cook & Metzger 1964; Browning & Fryar 1973; (654) 9.9 A – 247.2 A: Huffman 1969, 303.8 A – 1037 A: Samson & Cairns 1965; Cook & Metzger 1964; Huffman et al. 1963; Kronebusch & Berkowitz 1976; (655) 1 A – 180 A: Huebner & Mukherjee 2015, 180 A – 580 A: Lee et al. 1973; Kronebusch & Berkowitz 1976, 580 A – 1350 A: Watanabe et al. 1967; Kronebusch & Berkowitz 1976, 1350 A – 1910 A: Marmo 1953; (656) 1 A – 1771.2 A: Barfield et al. 1972; Brion et al. 1979, 1771.2 A – 2000 A: Ackerman 1971, 2000 A – 2200 A: Herman & Mentall 1982, 2250 A – 2423.7 A: Shardanand & Rao 1977, branching ratio: Huffman 1969; Samson & Cairns 1964; Matsunaga & Watanabe 1967; Brion et al. 1979; (657) Experimental 0.61 A – 625.8 A: Barfield et al. 1972, 1150 A – 1830 A: Nee & Lee 1984, Theoretical: Van Dishoeck 1984; (658) 2 A – 270 A: Henry & McElroy 1968, 303.78 A – 555.26 A: Cairns & Samson 1965, 580 A – 1670 A: Nakata et al. 1965, branching ratio: Kronebusch & Berkowitz 1976; Nakata et al. 1965; (659) 1 A – 100 A: Barfield et al. 1972, 180 A – 700 A: Phillips et al. 1977; Dibeler et al. 1966, 700 A – 980.8 A: Phillips et al. 1977; Katayama et al. 1973; Watanabe & Jursa 1964, 980.8 A – 1860 A: Watanabe & Jursa 1964; Watanabe & Zelikoff 1953, branching ratio: McNesby et al. 1962; Slanger & Black 1982; Kronebusch & Berkowitz 1976; (660) 1 A – 1950 A: Huebner & Mukherjee 2015; (661) 1 A – 1850 A: Huebner & Mukherjee 2015; (662) 0.61 A – 625 A: Huebner & Mukherjee 2015; Barfield et al. 1972, 1080 A – 1700 A: Zelikoff et al. 1953, 1730 A – 2400 A: Selwyn et al. 1977, branching ratio: Okabe et al. 1978; (663) 1 A – 840 A: σ(H2N) ≈ 2σ(H) + σ(N), 1250 A – 1970 A: Saxon et al. 1983; (664) 0.6 A – 940 A: Huebner & Mukherjee 2015, 1080 A – 1800 A: Nakayama et al. 1959, 1850 A – 3978 A: Bass et al. 1976, branching ratio: Nakayama et al. 1959; (665) 0.6 A – 742 A: σ(O3) ≈ 3σ(O), 1060 A – 1360 A: Tanaka et al. 1953, 1365 A – 2000 A: Ackerman 1971, 2975 A – 3300 A: Moortgat & Warneck 1975, 3300 A – 8500 A: Griggs 1968; (666) 600 A – 1000 A: Metzger & Cook 1964, 1050 A – 2011 A: Nakayama & Watanabe 1964, branching ratio: Schoen 1962; Okabe 1981, 1983; (667) 0 A – 500 A: Barfield et al. 1972, 600 A – 1760 A: Mentall et al. 1971, 1760 A – 1850 A: Gentieu & Mentall 1970, 2000 A – 2634.7 A: Calvert & Pitts 1966, 2635.7 A – 3531.7 A: Rogers 1990, 3531.7 A – 3740 A: Calvert & Pitts 1966, branching ra- tio: Clark et al. 1978; Mentall et al. 1971; Guyon et al. 1976; (668) 0.61 A – 627 A: Barfield et al. 1972, 1200 A – 2000 A: Okabe 1970, 2100 A – 2550 A: Dixon & Kirby 1968; (669) 0.6 A – 940 A: Barfield et al. 1972, 2000 A – 4000 A: Cox & Derwent 1977, 3120 A – 3900 A: Stockwell & Calvert 1978; (670) 1 A – 350 A: Huebner & Mukherjee 2015, 374.1 A – 1650 A: Sun & Weissler 1955; Watanabe & Sood 1965, 1650 A – 2170 A: Watanabe 1954, 2140 A – 2330 A: Thompson et al. 1963, branching ratio: McNesby et al. 1962; Schurath et al. 1969; Kronebusch & Berkowitz 1976; (671) 0.6 A 137 – 630 A: Huebner & Mukherjee 2015, 915 A – 3980 A: σ(NO3) ≈ σ(NO2), 4000 A – 7030 A: Graham & Johnston 1978b, branching ratio: Magnotta & Johnston 1980; (672) 23.6 A – 1370 A: Lukirskii et al. 1964; Rustgi 1964; Ditchburn 1955, 1380 A – 1600 A: Mount & Moos 1978, branching ratio: Gorden & Ausloos 1967; Calvert & Pitts 1966; Stief et al. 1972; Slanger & Black 1982; Kronebusch & Berkowitz 1976; (673) 1 A – 1100 A: Huebner & Mukherjee 2015, branching ratio: Gorden & Ausloos 1961; Huebner & Mukherjee 2015; (674) 0 A – 1100 A: σ(HNO3) ≈ σ(NO3), 1100 A – 1900A: Okabe 1980, 1900 A – 3300A: Molina & Molina 1981; (675) 1 A – 100 A: σ(C2H4) ≈ 2σ(C) + 4σ(H), 180 A – 1065 A: Lee et al. 1973; Schoen 1962, 1065 A – 1960 A: Schoen 1962; Zelikoff et al. 1953, branching ratio: Lee et al. 1973; McNesby & Okabe 1964; Back & Griffiths 1967; (676) 0.61 A – 250 A: σ(C2H4) ≈ 2σ(C) + 6σ(H), 354 A – 1127 A: Koch & Skibowski 1971, 1160 A – 1200 A: Lombos et al. 1967, 1200 A – 1380 A: Okabe & Becker 1963, 1380 A – 1600 A: Mount & Moos 1978, branching ratio: Lias et al. 1970; Huebner & Mukherjee 2015; (677) 2000 A – 3450 A: Baulch et al. 1982, branching ratio: Weaver et al. 1977; (678) 1200 A – 2053 A: Salahub & Sandorfy 1971, branching ratio: Porter & Noyes 1959; Huebner & Mukherjee 2015; (679) 0.1 A – 2412.63 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996; (680) 0.1 A – 2856.34 A: Verner et al. 1993; Verner & Yakovlev 1995; Verner et al. 1996; (681) 0.61 A – 877.46 A: Barfield et al. 1972, 1050 A – 1350 A: Myer & Samson 1970, 1400 A – 2200 A: Inn 1975; (682) Harada et al. 2010; Rimmer & Helling 2013; (683) Harada et al. 2010; (684) 1 A – 3000 A: σ(N2O3) ≈ σ(NO2) + σ(NO), 3000 A – 4000 A: Stockwell & Calvert 1978; (685) Pitts et al. 1982; (686) Vakhtin et al. 2001; (687) Fahr & Nayak 1994; Ferradaz et al. 2009; Friedrichs et al. 2002; Okabe 1981; (688) Fahr & Nayak 1996; (689) Prasad & Huntress 1980; (690) Ercolano & Storey 2006. 138
1002.2861	1	1002	2010-02-15T11:54:25	Exoplanetary Systems with SAFARI: A Far Infrared Imaging Spectrometer for SPICA	[ "astro-ph.EP", "astro-ph.IM" ]	The far-infrared (far-IR) spectral window plays host to a wide range of spectroscopic diagnostics with which to study planetary disk systems and exoplanets at wavelengths completely blocked by the Earth atmosphere. These include the thermal emission of dusty belts in debris disks, the water ice features in the "snow lines" of protoplanetary disks, as well as many key chemical species (O, OH, H2O, NH3, HD, etc). These tracers play a critical diagnostic role in a number of key areas including the early stages of planet formation and potentially, exoplanets. The proposed Japanese-led IR space telescope SPICA, with its 3m-class cooled mirror (~5 K) will be the next step in sensitivity after ESA's Herschel Space Observatory (successfully launched in May 2009). SPICA is a candidate "M-mission" in ESA's Cosmic Vision~2015-2025 process. We summarize the science possibilities of SAFARI: a far-IR imaging-spectrometer (covering the ~34-210um band) that is one of a suite of instruments for SPICA.	astro-ph.EP	astro-ph	Pathways towards Habitable Planets ASP Conference Series, Vol. , 2010 Eds. Dawn Gelino, Ignasi Ribas, Vincent Coude du Foresto Exoplanetary Systems with SAFARI: A Far Infrared Imaging Spectrometer for SPICA J. R. Goicoechea1 and B. Swinyard2 on behalf of the SPICA/SAFARI science teams 1Centro de Astrobiolog´ıa (CSIC-INTA), Madrid, Spain 2Rutherford Appleton Laboratory, Chilton, Didcot, UK The far-infrared (far-IR) spectral window plays host to a wide Abstract. range of spectroscopic diagnostics with which to study planetary disk systems and exoplanets at wavelengths completely blocked by the Earth atmosphere. These include the thermal emission of dusty belts in debris disks, the water ice features in the "snow lines" of protoplanetary disks, as well as many key chemical species (O, OH, H2O, NH3, HD, etc). These tracers play a critical diagnostic role in a number of key areas including the early stages of planet formation and potentially, exoplanets. The proposed Japanese-led IR space telescope SPICA, with its 3m -- class cooled mirror (∼5 K) will be the next step in sensitivity after ESA's Herschel Space Observatory (successfully launched in May 2009). SPICA is a candidate "M-mission" in ESA's Cosmic Vision 2015-2025 process. We summarize the science possibilities of SAFARI: a far-IR imaging-spectrometer (covering the ∼34 -- 210 µm band) that is one of a suite of instruments for SPICA. 1. A New Window in Exoplanet Research The study of exo-planets (EPs) requires many different approaches across the full wavelength spectrum to both discover and characterize the newly discovered objects in order that we might fully understand the prevalence, formation, and evolution of planetary systems. The mid -- IR and far -- IR spectral regions are especially important in the study of planetary atmospheres as it spans both the peak of thermal emission from the majority of EPs thus far discovered (up to ∼1000 K) and is particularly rich in molecular features that can uniquely identify the chemical composition, from protoplanetary disks to planetary atmospheres and trace the fingerprints of primitive biological activity. In the coming decades many space- and ground-based facilities are planned that are designed to search for EPs on all scales from massive, young "hot Jupiters", through large rocky super-Earths down to the detection of exo-Earths. Few of the planned facilities, however, will have the ability to characterize the planetary atmospheres which they discover through the application of mid-IR and far-IR spectroscopy. SPICA will be realized within ∼10 years and has a suite of instruments that can be applied to the detection and characterization of EPs over the ∼5 -- 210 µm spectral range (see e.g. our White Paper ; Goicoechea et al. 2008). The SAFARI instrument (Swinyard et al. 2009) will provide capabilities to complement SPICA studies in the mid-IR (either coronagraphic or transit studies). Indeed, SAFARI could be the only planned instrument able to study 1 2 Goicoechea and Swinyard Figure 1. Left: Fit to HD 209458b "hot Jupiter" (Teff ≃ 1000 K) fluxes inferred from a secondary transit with Spitzer (Swain et al. 2008) around a G0 star (d ∼ 47 pc, in black) and interpolation to d = 10 pc (gray). The emission of a cooler Jupiter-like planet at 5 pc is shown in dashed (reflected emission neglected). Thick horizontal lines are the 5σ-1hr photometric sensitivities of SPICA mid-IR instruments and SAFARI. Dotted lines show sensitivities in spectrophotometric mode (R ≃ 25). SPICA will observe similar transits of inner "hot Jupiter" routinely and will potentially extract their IR spectrum (rich in H2O, O3, CH4, NH3 and HD features as in Solar System planets). Right: Increasing planet-to-star contrast at longer mid -- and far -- infrared wavelengths (Goicoechea et al. 2008). EPs in a completely new wavelength domain (for SAFARI's band 1) not covered by JWST nor by Herschel (Fig. 1). This situation is often associated with unexpected discoveries. Since cool EPs show much higher contrast in the far-IR than in the near -- /mid -- IR (e.g. Jupiter's effective temperature is ∼ 110 K), if such EPs are found in the next 10 years, their transit studies with SPICA will help to constrain their main properties, which are much more difficult to infer at shorter wavelengths. Note that following Infrared Space Observatory (ISO) observations, Jupiter seen at 5 pc will produce a flux of ∼ 35 µJy at 37 µm, but less than ∼ 1 µJy at 15 µm. SAFARI band1 (∼ 34 -- 60 µm) hosts a variety of interesting atmospheric molecular features (e.g. H2O at 39 µm, HD at 37 µm and NH3 at 40 and 42 µm). Strong emission/absorption of these features was first detected by ISO in the atmospheres of Jupiter, Saturn, Titan, Uranus and Neptune (Feuchtgruber et al. 1999). Acknowledgments. We warmly thank the SPICA/SAFARI science teams for fruitful discussions. JRG is supported by a Ram´on y Cajal research contract. References Feuchtgruber, H., Lellouch, E., Bezard, B. et al. 1999, A&A, 341, L17. Goicoechea, J. R., Swinyard, B., Tinetti, G. et al. 2008, A White Paper for ESA's Exo-Planet Roadmap Advisory Team (2008 July 29), astro-ph/0809.0242. Swain, M. R., Bouwman, J., Akeson, R. L. et al. 2008, ApJ, 674, 482. Swinyard, B., Nakagawa, T., et al. 2009, Exp. Astron., 23, 193
1712.08460	1	1712	2017-12-22T14:27:26	Secondary resonances and the boundary of effective stability of Trojan motions	[ "astro-ph.EP" ]	One of the most interesting features in the libration domain of co-orbital motions is the existence of secondary resonances. For some combinations of physical parameters, these resonances occupy a large fraction of the domain of stability and rule the dynamics within the stable tadpole region. In this work, we present an application of a recently introduced `basic Hamiltonian model' Hb for Trojan dynamics, in Paez and Efthymiopoulos (2015), Paez, Locatelli and Efthymiopoulos (2016): we show that the inner border of the secondary resonance of lowermost order, as defined by Hb, provides a good estimation of the region in phase-space for which the orbits remain regular regardless the orbital parameters of the system. The computation of this boundary is straightforward by combining a resonant normal form calculation in conjunction with an `asymmetric expansion' of the Hamiltonian around the libration points, which speeds up convergence. Applications to the determination of the effective stability domain for exoplanetary Trojans (planet-sized objects or asteroids) which may accompany giant exoplanets are discussed.	astro-ph.EP	astro-ph	P´aez & Efthymiopoulos (2017) Submitted to CMDA Secondary resonances and the boundary of effective stability of Trojan motions Roc´ıo Isabel P´aez1 and Christos Efthymiopoulos1 1Research Center for Astronomy and Applied Mathematics, Academy of Athens Abstract: One of the most interesting features in the libration domain of co-orbital motions is the existence of secondary resonances. For some combinations of physical parameters, these resonances occupy a large fraction of the domain of stability and rule the dynamics within the stable tadpole region. In this work, we present an application of a recently introduced 'basic Hamiltonian model' Hb for Trojan dynamics [33], [35]: we show that the inner border of the secondary resonance of lowermost order, as defined by Hb, provides a good estimation of the region in phase-space for which the orbits remain regular regardless the orbital parameters of the system. The computation of this boundary is straightforward by combining a resonant normal form calculation in conjunction with an 'asymmetric expansion' of the Hamiltonian around the libration points, which speeds up convergence. Applications to the determination of the effective stability domain for exoplanetary Trojans (planet-sized objects or asteroids) which may accompany giant exoplanets are discussed. 1 Introduction Despite the theoretical possibility of the existence of Trojan exoplanets ([18], [1], [3]), no such body has been identified so far in exoplanet surveys. This lack of identification may reflect formation constrains, constrains to detectability ([17], [4], [19], [20], [21]), or it may simply be due to stability reasons. In this framework, the question of 'effective stability', i.e. stability of the orbit of a Trojan body for times as long as a considerable fraction of the age of the hosting system, comes to the surface. The question of effective stability has been addressed nearly exhaustively in the case of Trojan asteroids in our Solar System (see, for example, [27], [22], [15], [39], [37], [23], [5], [24]) from both numerical and analytical approaches, but only scarcely in the case of exoplanetary systems (see [30], [12], [38], [9]). One main reason for the scarcity of results in this latter case is the vast volume of parameter space to be investigated, in conjunction with the multi-body nature of the problem: to determine the long-term stability of Trojan motions becomes essentially a problem of secular dynamics with as many degrees of freedom as the number of planets in the system under consideration. Any attempt to face the problem other than numerical simulation clearly requires a simplification of the dynamical model, without this leading to oversimplified conclusions regarding the long-term orbital stability. In the present work, we discuss a key property of the dynamics induced by secondary resonances in the domain of Trojan motions, which in addition to its own proper interest, can serve also the purpose of obtaining a simple analytical estimate of the effective stability boundary of Trojan motions in hypothetical exoplanetary systems. Our analysis of the resonant dynamics stems from a set of considerations or assumptions, whose validity can be most easily judged by comparison with some results and figures of a previous work of ours ([33]) as follows: 1) In [33] we provided a formalism of the problem of the dynamics of Trojan bodies in the Hamiltonian context, which recovers all essential features as discovered in previous literature ([10], [11], [28], [29]); preceding works, however, focus mostly on a direct investigation of the equations of motion, averaged or not with respect to short period terms. In our works we stressed, instead, a main advantage of the new formalism, namely the allowance to recruit the full machinery of Hamiltonian methods in order to better analyze the problem under study. 2) We investigated the features of Trojan dynamics which hold under three physically relevant assumptions: i) that the motions of all planets, including the Trojan body, are close to planar, ii) that the Trojan body is small enough to be considered as test particle, in agreement with formation scenaria which suggest that exo-Trojans should be at most Mars-sized objects ([1]), and iii) that the 1 P´aez & Efthymiopoulos (2017) Submitted to CMDA secular dynamics of the hosting system is such that the eccentricity vector of the primary companion of the Trojan body undergoes circulation with a nearly constant frequency g(cid:48), and has a length which undergoes variations around some non-zero value e0. Under assumptions (i) to (iii), we find that the Hamiltonian of motion of the Trojan body, averaged over short period terms for the motions of the remaining planets, can be decomposed in the form H = Hb + Hsec where: Hb, called the 'basic model', describes short period and synodic motions, and yields a constant proper eccentricity for the Trojan body, and Hsec contains all remaining secular perturbations. Furthermore Hb has a universal form, i.e., it suffices to redefine the physical meaning of the angular canonical variables, to keep its form unaltered in the whole hierarchy of restricted problems (circular, elliptic, secular with more than one perturbing planets). 3) The decomposition H = Hb + Hsec leads to a specific physical understanding of the dynamics when the primary planet has a mass in the giant planet range. In this case, the three timescales related to the short-period, synodic and secular motions have a separation by about one or less order of magnitude from each other. Then, due to the specific features of Hb described above, we arrive at the following key remark: the model Hb produces, in phase space, a set of secondary resonances corresponding to commensurabilities between the frequencies of the short-period and synodic mo- tions ([13]). It is easy to see that these are the only secondary resonances which occupy a non-zero volume in phase space in the whole hierarchy of restricted problems that one could use as dynamical models for the Trojan body. However, there exists a modulation effect ([2]) due to the influence of Hsec on this set of secondary resonances: the separatrices pulsate slowly (with one or more secular frequen- cies) and, as a result, in the 'domain of uncertainty' ([31]) created by such pulsations, the motions become chaotic. Such an effect is possible to visualize already in the Elliptic Restricted Three-Body Problem, namely the simplest model with non trivial Hsec. The reader is refered to Figures 5 to 15 of [33] which show in detail the statements below, by exemplifying the outcome of the modulation effects for the secondary resonances 1:5 up to 1:12, when the modulus of the eccentricity vector e0 of the primary companion varies from e0 = 0 to just a moderate value e0 = 0.1. By inspecting the stability maps in the space of the Trojan body's proper elements, one sees that, for e0 slightly larger than zero, the separatrix pulsation for the secondary resonances becomes large enough so as to wipe out nearly completely the domain of stable motions occupied by such resonances. As a result, the only remaining stable motions are those in a inner (closer to the libration center) domain devoid of secondary resonances. In fact, as found in many works (e.g. [37], [26]) there can still be resonances involving one or more secular frequencies which penetrate this innermost stability domain. However, since these resonances are thin and typically do not overlap, they can only induce a very slow chaotic diffusion of the Arnold type, which can be neglected for all practical purposes. Hence, the innermost domain, devoid of the secondary resonances of Hb, meets all criteria of effective stability, and, indeed, stability maps indicate the robustness of this domain against variations of the orbital parameters of the Trojan body. 1.1 Summary of the method identify the largest in size (typically lowest in order) secondary resonance of Hb for given Stemming from remarks (1) to (3) above, we propose below a practical method to define the effective stability domain of Trojan motions. This is based on the following steps: Step 1: analyze a given system where hypothetical Trojan bodies are sought for and compute the Hamiltonian Hb, Step 2: parameter values, Step 3: compute a resonant normal form and evaluate the theoretical separatrices of the identified secondary resonance, Step 4: assume that all stable domains of resonant motions extending beyond the innermost (closest to the libration center) branch of the theoretical separatrices were wiped out by secular modulation effects. Then, the locus S formed by the intersection of the family of all computed innermost theoretical separatrices, along the dominant secondary resonance, with any chosen (with respect to phases) plane P P of Trojan proper elements, yields the boundary of the effectively stability domain in the plane P P . 2 P´aez & Efthymiopoulos (2017) Submitted to CMDA The above computation is fast and straightforward to perform with modern computer algebra programs, and thus competitive to large grid computations of effective stability maps. In our own implementation we use the normal form method adopted in [34], [35]. In the rest of the paper, we discuss both the dynamical role of the secondary resonances in delimit- ing the main domain of effective stability as well as our particular analytical method of computing the border of this domain. The structure of the paper is as follows: In Section 2, we review the derivation and features of the Hamiltonian Hb, as well as our way to expand Hb in a form form suitable for resonant normal form computations. A novel feature is the adoption of an 'asymmetric expansion' which improves convergence. Section 3 explains in detail the realization of the Steps 1-4, in partic- ular the computation of the theoretical separatrices of the dominant secondary resonance and their superposition to stability maps in the space of proper elements. Section 4 contains the main results: (a) we provide numerical evidence, based on stability maps, of how the separatrices of the secondary resonances of the 'basic model' Hb act as delimiters of the effective stability domain; (b) we use an analytical method to estimate this boundary; (c) we discuss the robustness of the present approach against changing the model's parameters (masses and eccentricities), as well as when considering, in the numerical integrations, the full three-body problem instead of the ERTBP. Section 5 summarizes our main conclusions. 2 Basic Hamiltonian Hb and its asymmetric expansion 2.1 Main features of the basic model Hb In [33], a Hamiltonian formulation was provided for the Trojan motion which applies to the planar Elliptic Restricted Three-Body Problem (ERTBP) with a central mass, a primary perturber or simply 'primary', and the Trojan test particle, or when S additional perturbing bodies are present but far from MMRs, the so-called 'Restricted Multi-Planet Problem' (RMPP)). The Hamiltonian reads H = Hb (Yf , φf , u, v, Yp; µ, e(cid:48) 0) + Hsec (Yf , φf , u, v, Yp, φ, Y1, φ1, . . . , YS, φS) . (1) In Eq. (1), the variables (φf , Yf ), (u, v) and (φ, Yp) are pairs of action-angle variables, whose definition stems from Delaunay-like variables following a sequence of four consecutive canonical transformations (see Appendix A). In particular, (Yf , φf ) are action-angle variables describing the fast degree of freedom, of frequency ωf ≡ φf = 1 − 27 8 µ + g(cid:48) + . . . (2) where g(cid:48) is fundamental frequency of precession of the primary's perihelion. The pair (u, v) describe the particle's synodic librations, u (cid:39) λ − λ(cid:48) − π/3, v (cid:39) √ a − 1, with λ, λ(cid:48) the mean longitudes of the test particle and of the primary, a the particle's major semi-axis, and a(cid:48) = 1. The associated (cid:114) 27µ frequency at the libration center is , ωs ≡ φs = − + . . . . 4 (3) Finally, the secular motion of the test particle's eccentricity vector (e cos(ω − ω(cid:48)), e sin(ω − ω(cid:48))), where e is the eccentricity and ω,ω(cid:48) are the arguments of the perihelion of the test particle and the primary respectively, is described by a circulation around the forced equilibrium point, given in our variables by a set of action angle variables (Yp, φ). The associated secular frequency is g ≡ φ = 27 8 µ − g(cid:48) + . . . . (4) We call the term Hb in the Hamiltonian of Eq. (1) the 'basic Hamiltonian model' for Trojan motions in the 1:1 MMR. Its detailed form is given in the Supplementary Online Material of [33]. We find Hb = − 1 2(1 + v)2 − v + (1 + g(cid:48))Yf − g(cid:48)Yp − µF (0)(u, φf , v, Yf − Yp; e(cid:48) 0) . (5) 3 P´aez & Efthymiopoulos (2017) Submitted to CMDA The physical parameters entering into Hb are i) the mass parameter µ = m(cid:48) m(cid:48)+M , where M is the mass of the central mass and m(cid:48) the mass of the primary, ii) the mean value of the length of the eccentricity vector of the heliocentric orbit of the primary perturber, e(cid:48) 0. In the ERTBP, one has simply e(cid:48) = e(cid:48) 0, g(cid:48) = 0 (implying also ω(cid:48) ≡ const). However, the form of Hb remains the same in both the ERTBP and the RMPP. In particular, the angle φ is defined via a 'shift transformation' depending only on the relative difference ∆ω = ω − ω(cid:48). Physically, the secular dynamics induced under Hb appears the same in the ERTBP and in the RMPP, when, in the latter case, it is viewed in apsidal co-rotation with the primary. Furthermore, since the angle φ is ignorable in Hb, the action variable Yp is an integral of the basic Hamiltonian. Then, the ERTBP and the RMPP are diversified only by their different form of the functions Hsec. In particular, in the RMPP case Hsec contains also pairs of action angle variables associated with the secular precessions of the S additional bodies, while in the ERTBP it contains only the angle φ associated with the secular precession of the test particle. Finally, Hsec disappears all together in the circular RTBP. Thus, Hb becomes the exact Hamiltonian in this case. Note, however, that in the ERTBP Hb is not equal to the ensemble of all terms independent of e(cid:48). The basic model Hb represents a drastic reduction of the number of degrees of freedom with respect to the original problem. In the sequel, we will focus on one particular feature of Hb, namely the presence of secondary resonances, which correspond to commensurability relations between ωf and ωs. In particular, we focus on the role of these resonances in practically determining the boundary of the effective domain of stability for the Trojan motions. 2.2 Asymmetric expansion The resonant normal form computed in Section 3 below provides a model for studying the dynamics within or near a secondary resonance of the form: mf ωf + msωs = 0 . (6) A non-resonant normal form for the model Hb, allows to find the location of secondary resonances in a space of suitably defined proper elements for the Trojan body (see [35]). However, the non-resonant normal form does not allow to compute the local phase portrait, i.e., the separatrices associated with each resonance. Furthermore, all series expansions which are polynomial in the variables u, v exhibit poor convergence, a fact associated with the singularity (collision with the primary) at u = −π/3. In order to deal with this problem, a partially expanded version of the Hb can be used [34], in which all (with τ = u + π/3) are kept unexpanded. This leads to the powers of the quantity β(τ ) = a Hamiltonian of the form 2−2 cos τ 1√ ∞(cid:88) i=0 Hb(v,Y, τ, φf , Yp) = −v + (−1)i−1(i + 1) + Y + Yp vi 2 (cid:88) + µ am1,m2,m3,k1,k2,j e(cid:48)k3 vm1 cosk1 (τ ) sink2 (τ )Y m4 cosm2 φf sinm3 φf βj(τ ) , (7) m1,m2,m3 k1,k2,k3,j where τ = u + π/3, Y = Yf − Yp, and am1,m2,m3,k1,k2,j are rational numbers. The librations in τ (or u) are represented in terms of the synodic angle variable φs, i.e., the phase of the synodic libration. The computation of a resonant normal form requires to explicitly Fourier expand the terms of Hb in both angles φf and φs. Although the Hamiltonian (7) represents a Fourier expansion for the fast d.o.f. (angle φf ), there still remain the powers of β that must be expanded in powers of u in order to obtain a complete Fourier expansion in the angle φs as well. Due to the singularity at τ = 0 (or u = −π/3), any Taylor expansion of the functions β(τ )N = (2−2 cos τ )N/2 , with N ∈ N, around a certain τ0 is convergent only in the domain Dτ0,δ centered at τ0 and of radius δ = Min{τ0, 2π − τ0}. The most common approach consists of Taylor expansions around the libration equilibrium point, located at τ0 = π 3 for L5. The corresponding δ in this case is π 3 . One finds that many Trojan orbits, and important secondary resonances, may cross this domain. In such cases, the resonant normal form construction is obstructed by the poor convergence of the original Hamiltonian expansion. 3 , for L4, or τ0 = 5π 1 4 P´aez & Efthymiopoulos (2017) Submitted to CMDA In order to face this problem, we find a different polynomial representation of the Hamiltonian Hb in the variables (u, v) by performing an asymmetric expansion, i.e. expansion around a non- equilibrium point τ0 (cid:54)= π/3, selected to be further away from the singularity but close enough to the libration point, so that a re-ordering of the expansion in powers of u yields a negligible term linear in u (since u = 0 represents the equilibrium point of Hb). Here we choose τ0 = π 2 . In this case, we obtain a polynomial expansion of the Hamiltonian in powers of the quantity (τ − π/2). Re-ordering the terms, we express it as a polynomial in powers of u. It is immediate to see that any finite truncation of this expression yields a different polynomial than the one obtained by a finite truncation of the direct Taylor expansion around τ = π/3. However, the new expression better represents the quantities β(τ )N in a domain extended up to τ ∼ π. We call the expansion around τ0 = π 2 asymmetric, while the one around τ0 = π 3 symmetric. Figure 1 shows the benefits of the asymmetric expansion when compared to the symmetric one. We consider the functions B1(τ ) = B5(τ ) = cos τ β(τ ) cos τ β5(τ ) = = cos τ (2 − 2 cos τ )1/2 (2 − 2 cos τ )5/2 cos τ , B3(τ ) = , B7(τ ) = cos τ β3(τ ) cos τ β7(τ ) = = cos τ (2 − 2 cos τ )3/2 (2 − 2 cos τ )7/2 cos τ , , (8) which represent the most common terms in powers of β(τ ) appearing in Eq. (7). The symmetric Taylor expansion of B1, B3, B5 and B7 around τ0 = π/3 yield the polynomials BM,π/3(u) = BM (π/3) + B(1) M (π/3) u + 1 2 B(2) M (π/3) u2 + 1 6 B(3) M (π/3) u3 + . . . , (9) where B(n) M (π/3) is the n-th derivative of the function BM , evaluated at u = π/3, M = 1, 3, 5, 7 and u = τ − π/3. On the other hand, the asymmetric Taylor expansions of the same functions around τ0 = π/2 yield the polynomials BM,π/2(u) = BM (π/2) + B(1) M (π/2) (u− π 6 ) + 1 2 M (π/2) (u− π B(2) 6 )2 + 1 6 M (π/2) (u− π B(3) 6 )3 + . . . . (10) Fig. 1 compares the graphs of the original functions BM (u) (pink) with the two corresponding expansions BM,π/3(u) (symmetric, blue) and BM,π/2(u) (asymmetric, green) up to order 10 in u. We see that both expansions provide a good representation of the original function up to a certain extent in u, but for increasing values of u, the asymmetric expansions are more accurate than the symmetric ones in a domain extending to higher values of u. The improvement in accuracy is more notorious as M increases. In fact, we find that the polynomial approximations to Hb found by the asymmetric expansion is accurate up to u ∼ 1 rad, which is enough to cover the effective stability domain in most physically relevant parameter values. Only very close to u ∼ 0, the symmetric expansion is marginally more accurate than the asymmetric one. This yields a slight shift of the equilibrium point of Hb with respect to u = 0, typically of about ∼ 10−8 rad, i.e. practically negligible. Similar results are found for the asymmetric and symmetric expansions of the functions sin τ β(τ ) ap- pearing in Hb. Finally, the asymmetric expansions for both types of functions can be easily performed by a closed set of formulas, given in the Appendix B. 3 Resonant normal form 3.1 Hamiltonian preparation The construction of the asymmetrically expanded Hb consists of two steps: expansions for the variables (u, v) in Eq. (7), and ii) transformation to action-angle variables. i) replacement of the In order to replace the polynomial truncations for the functions of Eq. (8) into Eq. (7), we adopt the asymmetric expansion (10), using the formulae provided in the Appendix B. Regarding v, it is enough to consider the Taylor expansion of Hb with respect to v around zero. The maximum 5 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 1: Comparison of the functions BM (u) (pink), with the expansions BM,π/3(u) (blue) and BM,π/2(u) (green), up to order 10 in u, for M = 1, 3, 5, 7 and u ∈ [0.2, 2]. truncation order is determined in terms of a 'book-keeping parameter' ([8]; see below). After these replacements, the Hamiltonian Hb takes the form Hb(v,Y, u, φf , Yp) = Yp + am1,m2,m3,m4 vm1 um2 ( √Y)m3 cos(m4φf ) , (11) (cid:88) m1,m2, m3,m4 where the real coefficients am1,m2,m3,m4 depend on the parameters µ and e(cid:48) ERTBP). Next, we diagonalize Hb in order to obtain a harmonic oscillator quadratic part for the synodic degree of freedom. Diagonalization is performed by the linear canonical transformation (u, v) → (U, V ) defined by the set of formulas: 0 (or simply e(cid:48) in the From the variables U and V we then pass to the action-angle variables (Y, φf ) with M = (cid:112) 2a(2,0,0,0) −2a(0,2,0,0) −a(1,1,0,0) . (cid:112) U = 2Ys sin φs , V = 2Ys cos φs . (13) (14) The last step corresponds to a re-organization of the terms of the Hamiltonian, according to a book-keeping parameter [8]. This is a parameter with numerical value equal to = 1. To every term in the Hamiltonian (11), we asign a power of indicating the order of the normalization at which the term will be treated. Thus, coefficients with powers of propagate throughout the series at all normalization steps, helping to organize the terms in different orders of smallness. Regarding the original Hamiltonian, we adopt the following book-keeping rule: 6 where E is a 2×2 matrix with columns any two eigenvectors e1,2 associated with the eigenvalues λ1,2 = ±ωs of the matrix M , B = , (12) (cid:18)u (cid:19) v 1(cid:112)Det(E) = (cid:19) (E · B) (cid:18)U (cid:18) a(1,1,0,0) V (cid:32) 1√ 2 −i√ 2 (cid:33) −i√ 2 1√ 2 (cid:19) P´aez & Efthymiopoulos (2017) Submitted to CMDA Rule 3.1 To every monomial of the type c(k1,k2,k3,k4) ( Ys)k1( sin (k3φs + k4 φf ) , √Y)k2 cos assign a book-keeping coefficient r(k1,k2,k4), where the exponent r(k1, k2, k4) is given by r(k1, k2, k4) = Max(0 , k1 + k2 − 2) Max(0 , k1 + k2 − 2) + 1 if k4 = 0 if k4 (cid:54)= 0 . (cid:112) (cid:40) This book-keeping rule ensures also that the terms of zero-th order in are linear in Y and Ys. The Hamiltonian now takes the form: Hb(Ys,Y, φs, φf , Yp) = Yp + ωs Ys + ωf Y rmax(cid:88) (cid:112) √Y)k2 cos + c(k1,k2,k3,k4) r( Ys)k1 ( sin (k3φs + k4 φf ) . (15) r=1 From the canonical transformation in Eq. (14), it is straighforward to check that the harmonics of the angles φf and φs have the same parity as the powers of the corresponding functions in the variables √Y and √ Ys. 3.2 Resonant normalization The resonant normalization of the Hamiltonian (15) consists of a sequence of near-identity canonical transformations, in ascending powers of the book-keeping parameter , aiming to eliminate from the Hamiltonian the trigonometric dependence on the angles in any linear combination other than the one which corresponds to the selected secondary resonance (Eq. 6). The resulting normal form includes, besides terms depending just on the actions, also terms of the form b(p(r)) e i(k·q(r)) . (16) Here, q(0) = (φf , φs), p(0) = (Y, Ys), and the superscript (r) indicates the variables found after r consecutive near-identity normalizing transformations of (q(0), p(0)). Also, k = (k1, k2) belongs to the set M called the resonant module (Eq. 17 below). The terms (16) allow to determine the theoretical separatrices of the secondary resonance via the process described in subsection 3.3 below. The general recursive resonant normalization algorithm is defined as follows: Let m1, m2 be two . The resonant module M is the set of integer ≈ ωf integers marking the secondary resonance m1 m2 vectors defined by ωs M = {k = (k1, k2) : k1 m1 + k2 m2 = 0} , Let us assume that the Hamiltonian is in normal form up to order r in the book-keeping parameter, H = Z0 + Z1 + . . . + rZr + r+1H(r) From the terms of order r+1, in the Fourier expansion, r+1 + r+2H(r) r+2 + . . . . where(cid:80)2 i.e. i=1 mi (cid:54)= 0. where we isolate the terms that we want to eliminate in the present step, denoted by (cid:88) (cid:88) k k /∈M H(r) r+1 = b(p(r)) e i(k·q(r)) , ∗H(r) r+1 = b(p(r)) e i(k · q(r)) . The homological equation r+1 ∗H(r) r+1 + {Z0, χr+1} = 0 7 (17) (18) (19) (20) (21) P´aez & Efthymiopoulos (2017) Submitted to CMDA has the solution with ω = (ωf , ωs). χr+1 = r+1 (cid:88) k /∈M b(p(r)) i (k · ω) e i(k · q(r)) , Having the expression of the generating function, we compute the transformed Hamiltonian where exp H(r+1) = exp(Lχr+1)H(r) , (cid:16)Lχ (cid:17)· = I · +(Lχ · ) + (L2 χ · ) + . . . . 1 2 and the Lie operator Lχ ≡ {·, χ} ({·,·} denotes the Poisson bracket). By construction, the Hamiltonian in Eq. (23) is in normal form up to order r+1, i.e. H = Z0 + Z1 + . . . + rZr + r+1Zr+1 + r+2H(r) r+2 + r+3H(r) r+3 + . . . . (22) (23) (24) (25) 3.3 Computation of theoretical separatrices Let us consider the function Hb given in Eq. (15) as the starting Hamiltonian H (0) of the normalizing scheme. We apply the normalizing scheme presented above, up to a maximum normalization order R in . In the examples that follow, the maximum normalization order examined was R = 22. However, since the resonant normal form series are asymptotic, depending on the parameters and resonance considered, the optimal normalization order (yielding the minimum remainder as computed e.g. in [7]) varies, yielding optimal orders between R = 14 and R = 20. b be the final normalized Hamiltonian. According to Eq. (16), the form of H (R) Let H (R) b by H (R) b = rb(Y (R), Y (R) s ) e i(kf φ(R) f +ksφ(R) s ) . b is given (26) R(cid:88) r=0 (kf ,ks)∈M If we replace the book-keeping parameter for its value equal to 1, we recover the final normal form, depending on the actions and the angles through the combination, (cid:88) H (R) b = (kf ,ks)∈M c(df ,ds,kf ,ks) df(cid:113) (cid:112)Y (R) ds Y (R) s e i(kf φ(R) f +ksφ(R) s ) , (27) where the pairs (df , kf ) and (ds, ks) have the same parity, and the values of the Fourier wavenumbers are bounded by kf ≤ df and ks ≤ ds. The integers (df , ds) are limited by the value of R, through the book-keeping Rule 3.1. We define the quantity Ψ = m1 Y (R) + m2 Y (R) s (28) as a resonant integral of the normal form H (R) resonant module M in Eq. (17). Considering Eq. (16), it is straightforward to prove that , where m1 and m2 are the integers that define the b LH (R) b Ψ = {H (R) b , Ψ} = 0 , i.e. Ψ is a formal integral of H (R) . b By considering the transformation C (R), where C (R) = ϕ(1) ◦ ϕ(2) ◦ . . . ◦ ϕ(R−1) ◦ ϕ(R) , ϕ(r) = exp (Lχr ) (Y (r), Y (r) s , φ(r) f , φ(r) s ) 8 (29) (30) (31) P´aez & Efthymiopoulos (2017) Submitted to CMDA we can represent the resonant integral in terms of the original variables (Y (0), Y (0) , φ(0) f , φ(0) s ), via s Ψ(Y (0), Y (0) s , φ(0) f , φ(0) s ) = Ψ , φ(R) f , φ(R) s ) . (32) (cid:16)C (R)(Y (R), Y (R) s (cid:17) Finally, applying the inverse transformations to those of Eqs. (14) and (12), we are able to express the resonant integral in (32) as function of the variables used in Eq. (11) Ψ ≡ Ψ(v,Y, u, φf ) . (33) Having arrived at a final expression for the resonant integral Ψ in terms of the original canonical variables, we can compute the form of the theoretical separatrices of the corresponding secondary resonance in any suitably defined surface of section of the Hamiltonian Hb. In the numerical results below, we adopt a section of the form φf = φf 0, as well as a constant value of the energy E = Hb, the equation E = Hb(v,Y, u, φf 0) can be solved for Y. Substitution to (33) yields then the resonant integral on the surface of section as a function of u and v only, viz. (cid:16) (cid:17) Ψ ≡ Ψ v,Y(u, v; E, φf 0), u, φf 0 . (34) The theoretical phase portrait is now obtained by the level curves of Eq. (34). Figure 2, left panel, summarizes the main features of the theoretical phase portrait. In particular, the stable periodic orbit of the secondary resonance is represented by the points of extremum of the level set of Ψ, while the unstable periodic orbit corresponds to the minimax (saddle) points of the level set of Ψ. The level curves with Ψ = Ψnmx, where Ψnmx is the value of the resonant integral at the saddle points, are the curves representing the theoretical separatrices of the secondary resonance. 4 Numerical results: boundary of the effective stability do- main 4.1 Analytical vs. numerical stability boundary We present below numerical results based on the computation of stability maps for selected values of the parameters µ and e(cid:48), characterized by the presence of conspicuous secondary resonances of the Hamiltonian Hb. The stability maps are given in color scale of the values of the Fast Lyapunov Indicator ([14]), for orbits with initial conditions labeled in terms of two quantities (∆u, ep0). These quantities also serve as proper elements, i.e. quasi integrals of motion, for the subset of all regular orbits in every stability map. Working on fixed surfaces of section φf = −π/3, the relation between 2 , ∆u = u − u0, where u0 initial conditions (u, v,Y) and (∆u, ep0) is given by the relations Y = is the point of intersection of the short-period orbit around L4 with the surface of section (see [33] for analytical expressions), and v = B∆u, for fixed parameters B (depending on µ) selected in such a way that the straight line v = B(u − u0) in the surface of section passes right through one of the islands of the secondary resonance chain. The half-witdh of the libration in u as a function of ∆u, B, µ, ep and e(cid:48), reads e2 p,0 (cid:34) 3B2/2 + µ(cid:0)9/8 + 63e(cid:48)2/16 + 129e2 p/64(cid:1) µ(cid:0)9/8 + 63e(cid:48)2/16 + 129e2 p/64(cid:1) (cid:35)1/2 Dp = ∆u + O(∆u2) . (35) The values of B used in the various stability maps below are given explicitly in the caption of each figure. We can now superpose the theoretical computation of the phase portrait of the secondary reson- ance to the numerical results found in the stability maps. For given parameters µ, e(cid:48), B, and choosing one value of the energy E, one obtains the resonant integral (34) as a function of u only. An example is shown in the right panel of Fig. 2. The value u = u marks the position of local maximum of the resonant integral Ψ along the line v = B(u − u0). This corresponds a central locus passing approx- imately through the middle of the resonant domain along the corresponding secondary resonance. 9 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 2: Left panel - Schematic representation of the plane (u, v) for a surface of section of the Hb. The central blue dot represents the location of a stable periodic orbit, whose co-ordinate in equal to u = ures. At this point, the resonant integral Ψ presents a global extremum. Additional quasi-periodic orbits inside the island of stability are labeled with the corresponding values of Ψ, i.e. Ψ∗1, Ψ∗2, Ψ∗3, Ψmnx, accomplishing Ψ∗1 > Ψ∗2 > Ψ∗3 > Ψmnx. The value Ψmnx represents a theoretical separatrix of the resonance in the resonant integral approximation (in reality, instead of the separatrix we have a thin separatrix-like chaotic layer). For the initial conditions taken along the line B(u − u0), the orbit for which Ψ is maximum corresponds to a level curve tangent to the line, labeled Ψ∗1. The initial condition for u along this line, u, represents a good approximation to the exact resonant position ures. The two values of u on the line B(u − u0) satisfying Ψ = Ψmnx correspond to the intersection of the separatrix with the line B(u − u0) (∆umin and ∆umin, in blue), and provide an estimation of the width of the resonance. Right panel - Values of the resonant integral Ψ along the line B(u−u0). The position of the maximum of the function corresponds to u (green dot). The value of Ψmnx (black line) defines the position of the two borders of the resonance ∆umin and ∆umax (blue dots). On the other hand, the points of intersection of the line Ψ = Ψmnx with the curve of the resonant integral mark the values u1, u2, and hence ∆umin = u1 − u0, ∆umax = u2 − u0, where the theoretical separatrix intersects the plane of the stability map. The corresponding values of ep0 can be found (cid:17)(cid:105)1/2 (cid:104) − 2Y(cid:16) through ep0,i = ui, vi = B(ui − u0), φf 0 ; E , with i = 1, 2. Repeating, now, the same process for different values of the energy E allows to obtain the whole locus of the theoretical center as well as the theoretical boundary of the secondary resonance on the FLI stability map. Figure 3 shows an example of the location of the center and borders of a secondary resonance, with the method of the resonant integral, for the case of the 1:6 secondary resonance (µ = 0.0041) and e(cid:48) = 0.02. The position of the center of the resonance is denoted by a dashed line, and the inner and outer borders are denoted by thick solid lines. By comparison with the underlaying FLI stability map, we can see that both the center of the resonance and the outer border ∆umax are understimated by this computation, proving that the overall estimation of the resonance width is not accurate. On the other hand, the key remark is that the method turns to be extremely efficient in the location of the inner border. The approximate position of ∆umin is well determined in the whole range of proper eccentricity values considered 0 < ep,0 < 0.1. Figure 4 shows more examples of the method of determination of the effective stability domain through the application of the resonant normal form in the cases of the secondary resonances 1:5 (µ = 0.0056, panel a), 1:6 (µ = 0.0041, panel b), 1:7 (µ = 0.0031, panel b) and 1:8 (µ = 0.0024, panel d), and primary's eccentricity e(cid:48) = 0.02. In all the panels, the location of the inner border ∆umin is 10 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 3: Theoretical location of the center and borders of the 1:6 secondary resonance for µ = 0.0041, B = 0 and e(cid:48) = 0.02. The solid lines correspond to the inner an outer border of the resonance, the dashed line correspond to the estimation of the center of the resonance. The underlying image gives the numerical stability map, using the FLI value in grayscale. shown with a thick black line on top of the corresponding FLI stability map. We observe that this limit divides the space of proper elements in two regions: the inner domain from ∆u = 0 to ∆umin is populated mainly by regular orbits, and exhibits also some isolated resonances of small width, in which the orbits can only be weakly chaotic and remain practically stable. On the contrary, the domain external to ∆umin is dominated by the presence of conspicuous resonances as well as regions of strong chaos. It is remarkable that the analytical determination of the inner border of the resonances, which is based on an integrable approximation to the Hamiltonian (i.e. the resonant normal form), can still provide an accurate limit even in domains of the phase space where the resonant orbits are, in reality, chaotic. It is this robustness of the inner border determination which renders the whole approach useful in practice. Figures 5 and 6 show, now, more examples of the applicability as well as the level of approximation of the method. Figure 5 shows the stability maps for µ = 0.0041 (corresponding to a conspicuous 1:6 secondary resonance) and three different values of the primary's eccentricity, e(cid:48) = 0.02 (panel a), e(cid:48) = 0.06 (panel b), e(cid:48) = 0.1 (panel c). In the same plots we show the effective stability borders from the resonant normal form computation for the 1:6 secondary resonance, but for two values of the primary's eccentricity in each case, namely e(cid:48) = 0 (dotted thin line) and e(cid:48) = 0.02 (thick line) in (a), e(cid:48) = 0 (dotted thin line) and e(cid:48) = 0.06 (thick line) in (b) and e(cid:48) = 0 (dotted thin line) and e(cid:48) = 0.1 (thick line) in (c). We observe that altering the primary's eccentricity from e(cid:48) = 0 to only e(cid:48) = 0.1 suffices to completely wipe out the entire structure of secondary resonances beyond ∆u (cid:39) 0.4. In fact, we observe that, with increasing e(cid:48), so called 'transverse' resonances, i.e. involving also the secular frequency g, i.e. of the form mf ωf + msωs + mgg = 0 with mg (cid:54)= 0, appear near this border. For example, the 1:6:1 resonance at u ∼ 0.25 in panel (a) of Fig. 5 moves towards the border at u ≈ 0.35 in panel C of the same figure. A careful inspection of the stability maps shows that, for small e(cid:48) these resonances have a small width and remain isolated within the inner stability domain, while, as e(cid:48) increases, all resonances (main or transverse) grow in size and move outwards, until they enter to the region of strong chaos. As revealed in the panels of Fig. 5, these two effects (the moving of the resonances outwards and the refilling of the stable region with transverse resonances) counteract each other in such a way that the border separating the inner domain of stability from the outer chaotic 11 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 4: Determination of the effective stability domain (∆umin, thick black line), for the secondary resonances 1:5 (µ = 0.0056, B = 0.03, panel a), 1:6 (µ = 0.0041, B = 0, panel b), 1:7 (µ = 0.0031, B = 0.015, panel c) and 1:8 (µ = 0.0024, B = 0, panel d), and e(cid:48) = 0.02. domain remains practically in the same place. Due to this effect, we can see that even the estimation of the border via the resonant normal form corresponding to the circular case (e(cid:48) = 0, dotted thin line) practically suffices to obtain a good approximation of the border of the effective stability domain. Also, regarding the Trojan's body eccentricity, parameterized by ep,0, one remarks that stable domains of all the secondary resonances, beyond the main stability domain, survive only for small values of ep,0. This is because the amplitude of the separatrix pulsation increases as the eccentricity of the Trojan body increases. As a consequence, we find that the border of the main domain of stability is more sharp, and, thus, in general, better represented by the analytical resonance limit as ep,0 increases. Similar results are found in Fig. 6, showing the stability maps for µ = 0.0056, corresponding to a conspicuous 1:5 secondary resonance, and for the primary's eccentricity values e(cid:48) = 0.02 (panel a), e(cid:48) = 0.08 (panel b). The estimated borders are found by the resonant normal form determination for µ = 0.0056, using the parameters e(cid:48) = 0 (circular case, dotted thin line) and e(cid:48) = 0.02 (thick line) in (a), and e(cid:48) = 0 (dotted thin line) and e = 0.08 (thick line) in (b). The margin between the two theoretical curves is again small (of about 0.02 rad in ∆u), while, again, the determination of the border of the stability domain using the circular model suffices to practically obtain an accurate limit of the domain of stability. In fact, in both Figures 5 and 6 the extent occupied by the stable parts of the corresponding resonances is determined by the separatrix pulsation effect. The amplitude of the pulsation depends on terms absent from the 'basic model', thus this effect cannot be modelled using only the resonant integrals of the basic model. However, as a rule of thumb we find that the border of the domain of stability lies always between two theoretical border determinations by the basic model, i.e., one using the circular model e(cid:48) = 0 and a second using a moderate value of the 12 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 5: FLI stability maps for the 1:6 secondary resonance (µ = 0.0041, B = 0) and three values of the eccentricity e(cid:48) = 0.02 (a), e(cid:48) = 0.06 (b), e(cid:48) = 0.1 (c). The dotted thin line corresponds to the analytical determination of ∆umin for the parameters µ = 0.0041 and e(cid:48) = 0 (circular case) in all three panels, while the thick line corresponds to e(cid:48) = 0.02 in (a), e(cid:48) = 0.06 in (b) and e(cid:48) = 0.1 in (c). primary's eccentricity, e.g. e(cid:48) = 0.1. 4.2 Robustness with respect to parameter values The investigation in the previous subsection focused on particular values of µ selected with the criterion that, for low eccentricities of either the primary perturber or the test body (e(cid:48), ep,0 < 0.1), the phase space of the basic model is dominated by a low-order secondary resonance of the form 1:n with n = 5, 6, .... Repeating a comparison between FLI maps and innermost separatrix borders of secondary resonances, a behavior similar to Figures 6 (resonance 1:5, for µ = 0.0056) and 5 (resonance 1:6, for µ = 0.0041) for low eccentricities is found when one considers the resonances 1:7 for µ = 0.0031, 1:8 for µ = 0.0024, 1:9 for µ = 0.0021, 1:10 for µ = 0.0016, 1:11 for µ = 0.0014, 1:12 for µ = 0.0012. These values of µ are shifted positively with respect to the bifurcation values µ = µ1:n of each corresponding 1:n short period family in the basic model. The shift reflects the fact that, keeping e(cid:48), ep,0 constant, and increasing µ as µ = µ1:n + ∆µ, with ∆µ > 0, the resonance 1:n moves outwards from the libration center, i.e., towards higher libration amplitudes ∆u, as ∆µ increases. In the resonant integral approximation, the outward motion of each resonance is accompanied by an increase of its separatrix width. However, the integrable aproximation fails due to resonance overlap with nearby resonances as ∆µ increases. This antagonism between outward expansion and resonance overlap determines the real limit of the domain of stability (see [40], [6] for a description of this phenomenon in simple dynamical maps). The bifurcation value µm:n for the m:n short-period family of the basic model can be estimated by the root for µ of the equation: m(1 − 27µ/8) = n (cid:115) 6µ (cid:18) 9 8 (cid:19) 63e(cid:48)2 16 + + 129e2 p 64 (36) Applying Eq. (36) to the 1:6 resonance, we find µ1:6 ≈ 0.0040 for e(cid:48) = ep = 0.02, while µ1:6 ≈ 0.0038 for e(cid:48) = ep = 0.1. As evident from Fig. 5, the resonance is clearly dominant at µ = 0.0042. In fact, we find that the 1:6 resonant integral inner separatrix limit applies already when ∆µ ≥ 0.001 with respect to the bifuration value for low eccentricities. On the other hand, as shown in panels (a) and (e) of Fig. 7, the analytical series computation with the 1:6 resonant integral starts collapsing when µ = 0.0044, or ∆µ ≈ 0.004. In practice, the whole separatrix domain around the 1:6 resonance has been transformed into a chaotic domain. Thus, while it remains true that the 1:6 resonance of the basic model delimits the main stability domain, the convergence of the series representing the theoretical computation of the corresponding resonant integral becomes poor. Implementing, now, Eq. (36) to the 1:5 resonance we find µ1:5 ≈ 0.0057 for e(cid:48) = ep,0 = 0.02, while µ1:5 ≈ 0.0054 for e(cid:48) = ep,0 = 0.1. Thus µ1:5 − µ1:6 ≈ 0.0016, which implies that the distance in µ 13 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 6: FLI stability maps for the 1:5 secondary resonance (µ = 0.0056, B = 0.03) and two values of the eccentricity e(cid:48) = 0.02 (a), e(cid:48) = 0.08 (b). The dotted thin line corresponds to the analytical determination of ∆umin for the parameters µ = 0.0056 and e(cid:48) = 0 (circular case) in the two panels, while the thick line corresponds to e(cid:48) = 0.02 in (a) and e(cid:48) = 0.08 in (b). separating the resonances 1:6 and 1:5 is about 3-4 times larger than the interval of values of ∆µ for which the validity of the resonant integral computation using one particular resonance is satisfactory. In principle, in order to bridge the gap between the two resonances, one has to use higher order resonances of the basic model, since the border of the domain of stability is always delimited by one such resonance. In practice, we find that it suffices to consider the basic resonances 1:n and their first Farey tree combination, i.e., the resonances 2:(2n− 1) which bifurcate at intermediate values of µ, i.e. µ1:n < µ2:2n−1 < µ1:n−1 for fixed e(cid:48), ep. Figure 7 exemplifies the transition from the dominance of the 1:6 to the 1:5 resonance via the 2:11 resonance of the basic model, for two values of the primary's eccentricity e(cid:48) = 0.02 (upper row) and e(cid:48) = 0.08 (lower row). The collapse of the inner border calculation for the 1:6 resonances starts near µ = 0.0044. However, the computation using the 2:11 resonant integral restores a correct estimate of the main domain of stability for µ = 0.0048, leaving only secondary resonances outside this domain. The 2:11 resonance remains dominant in this respect up to ≈ µ = 0.0054. At this value of µ the 1:5 secondary resonance of the basic model bifurcates for large enough values of the eccentricities, a fact which implies that the whole domain beyond the innermost separatrix of the 1:5 resonance should now be considered as outside the main stability domain. Indeed, although these secondary resonances are still very stable for very low eccentricities, we see that they essentially disappear for values of the eccentricities near ≈ 0.1 (compare panels (d) and (i) of Fig. 7). This marks the transition from the dominance of the 2:11 to the 1:5 resonance, the latter one being clearly dominant for a somewhat still higher value of µ (µ = 0.0056 in panels e, j). Figure 8 shows in greater detail the transition from the 2:11 to the 1:5 resonance, which, using FLI stability maps of the full problem, is actually seen to involve also some resonances coined transverse in [33], i.e. resonances involving all three short, synodic and secular frequencies. In particular, we see that the border of stability, which for µ = 0.0049 is practically delimited by the 2:11 resonance, starts being gradually penetrated by the transverse resonances 2:11:1, 1:5:2 and 1:5:1. The penetration appears earlier, as µ increases, for higher values of the eccentricities. This effect leaves small windows of values of µ for which, for low eccentricities, the border of stability may appear dominated by the innermost separatrix of some transverse resonance (e.g. the resonance 1:5:1 in panel (c) for µ = 0.0053, e(cid:48) = 0.02). However, for the same value of µ, the innermost separatrix border of the 1:5 resonance appears also in the upper part of the stability map for higher primary's eccentricity, i.e., e(cid:48) = 0.08 (panel g). As a consequence, although a clear dominance of the 1:5 resonance occurs for all eccentricities beyond µ = 0.0055 (panels d, h), the 1:5 resonance practically dominates in a 14 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 7: FLI stability maps for µ = 0.0042 and e(cid:48) = 0.02 (a), µ = 0.0044 and e(cid:48) = 0.02 (b), µ = 0.0048 and e(cid:48) = 0.02 (c), µ = 0.0054 and e(cid:48) = 0.02 (d), µ = 0.0056 and e(cid:48) = 0.02 (e), µ = 0.0042 and e(cid:48) = 0.08 (f), µ = 0.0044 and e(cid:48) = 0.08 (g), µ = 0.0048 and e(cid:48) = 0.08 (h), µ = 0.0054 and e(cid:48) = 0.08 (i), and µ = 0.0056 and e(cid:48) = 0.08 (j). The thick lines yield the analytical estimation of the border of stability for the corresponding value of µ and e(cid:48) = 0.02 for the upper row panels, and e(cid:48) = 0.08 for the lower row panels. The dotted thin line yields the analytical estimation of the border for the corresponding value of µ and e(cid:48) = 0 (circular case) in all the panels. B = 0.03 for all the panels. wide range of eccentricities already at µ = 0.0053. In fact, this dominance can only become more pronounced when additional perturbations are added to the model. In conclusion, except for small transient windows of parameter values, one can practically always find a resonance of the basic model for which the innermost separatrix provides the limit of the main domain of stability. It is to be stressed that this is a physical property induced by resonant dynamics, which holds independently of the efficiency by which the innermost separatrix border of the resonance can be computed analytically using some form of resonant integral series. On the other hand, using the method presented in Section 3.2, we find precise results by limiting the choice of resonance of the basic model among the set 1:n or 2:(2n− 1), with n integer. As a rule of thumb, for given parameters µ, e(cid:48) we choose the limiting resonance as the rational number closer to the frequency ratio: (cid:17) (cid:114) (cid:16) 9 6µ f = 8 + 63e(cid:48)2 16 + (1 − 27µ/8) 129e2 p 64 (37) for values of ep within our domain of interest. 4.3 Robustness with respect to the choice of model As an additional test, we examine the robustness of the above results against changing the dynamical model for Trojan orbits. Several formation scenaria discussed in literature ([1], [3], [16], [25], [32]) have allowed relatively massive Trojan planets (of mass ∼ 1 Earth mass) to exist. Allowing the Trojan body to have considerable mass, we examine whether the stability borders found in the framework of the 'basic model', which is only derived from the ERTBP, are still applicable in the framework of the full planar three body problem. 15 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 8: Details of the transition from the dominance of the 2:11 to the 1:5 resonance as µ varies from µ = 0.0049 to µ = 0.0055. The parameters in each panel are (a). µ = 0.0049, e(cid:48) = 0.02, (b). µ = 0.0051, e(cid:48) = 0.02, (c). µ = 0.0053, e(cid:48) = 0.02, (d). µ = 0.0055, e(cid:48) = 0.02, (e). µ = 0.0049, e(cid:48) = 0.08, (f). µ = 0.0051, e(cid:48) = 0.08, (g). µ = 0.0053, e(cid:48) = 0.08, H. µ = 0.0055, e(cid:48) = 0.08. The domains of various resonances (including transverse ones) are marked in the same plots. As an example, Figure 9 compares the border of stability in the planar ERTBP for µ = 0.0041 with one computed in the full three body problem with the Trojan body having mass equal to 1 or 10 Earth masses. We consider the Hamiltonian in Poincar´e variables: − Gm0m2 − Gm0m1 − Gm1m2 (p1 + p2)2 H = (38) p2 1 2m1 + p2 2 2m2 + 2m0 r1 r2 ∆ where m0, m1, m2 are the masses of the star, perturbing primary and Trojan planet respectively, r1, r2 the heliocentric positions of the primary and Trojan planet, ∆ = r1−r2 and p1, b2 their corresponding barycentric momenta. In order to use same units as in the ERTBP, one notes that the equations of motion in Cartesian heliocentric co-ordinates r1 ≡ (r1x, r1y), r2 ≡ (r2x, r2y), and barycentric velocities p1/m1 ≡ (v1x, v1y), p2/m2 ≡ (v2x, v2y) only depend on the variables rix, riy, vix, viy, i = 1, 2 and on the constants Gm0, Gm1, Gm2. Then, we solve Gm0 + Gm1 = 1, Gm1 = µ and assign a value to Gm2 = Gm0(m2/m0) according to the considered mass ratio µ2 = m2/m0. In order now to obtain comparable FLI maps in the two problems, we proceed as follows. For every point on the plane of the stability map of the ERTBP (such as in Fig. 9a), we compute the corresponding heliocentric positions and velocities of both the primary and the Trojan, i.e. (rix(t = 0), riy(t = 0), rix(t = 0) and riy(t = 0), with i = 1, 2. From Hamilton's equations of (38) one readily sees that the barycentric velocities (vix, viy), i = 1, 2 of both bodies are linear functions of the heliocentric ones. Thus, from every point of the FLI map in the ERTBP, we compute the complete set of corresponding initial conditions rix(t = 0), riy(t = 0), vix(t = 0) and viy(t = 0) needed in order to integrate the full Three Body problem. Via the same process we assign also corresponding initial conditions for the variational equations of motion in the two problems. Figure 9 shows the comparison of the FLI stability maps in the case of dominance of the 1:6 resonance, at µ = 0.0041 as µ2 evolves, i.e., µ2 = 3×10−6 (1 Earth mass, upper row), or µ2 = 3×10−5 (10 Earth masses, lower row). The left, middle and right panels correspond to initial eccentricities of the primary equal to e(cid:48) = 0.02, e(cid:48) = 0.06 and e(cid:48) = 0.1. Thus, these maps are comparable with the ones under the ERTBP (Fig. 5). The main observation is that switching on the mass m2 results in a considerable reduction of the area occupied by the stable domains of the secondary resonances. 16 P´aez & Efthymiopoulos (2017) Submitted to CMDA Figure 9: The FLI stability maps obtained with the same initial conditions as in Fig.5, but running the full planar three body model instead of the ERTBP, with a mass µ2 (cid:54)= 0 assigned to the Trojan body. The mass of the primary is always µ = 0.0041, while the remaining parameters (initial eccentricity e(cid:48) of the primary and mass µ2 of the Trojan) are (a). e(cid:48) = 0.02, µ2 = 3× 10−6, (b). e(cid:48) = 0.06, µ2 = 3× 10−6, (c). e(cid:48) = 0.1, µ2 = 3× 10−6, (d). e(cid:48) = 0.02, µ2 = 3 × 10−5, (e). e(cid:48) = 0.06, µ2 = 3 × 10−5, (f). e(cid:48) = 0.1, µ2 = 3 × 10−5. The analytical curves are those of Fig. 5. This is mostly caused by the secular variations induced on the orbit of the primary planet, which increase the amplitude of modulation of the separatrices of each secondary resonance. However, the main domain of stability remains nearly unaffected by these phenomena, and retains a quite similar width in all simulations with different masses µ2. We only see some transverse resonances penetrating the lowermost (with respect to the eccentricities) part of the stability map for µ2 as large as 10 Earth masses. On the other hand, the analytical determination of the innermost separatrix via the resonant integral of the 'basic model' yields an estimate of the border of the main stability domain which remains robust against the increase of µ2. 5 Conclusions In the present work, we discussed a new application for the 'basic Hamiltonian model' Hb for Trojan motions presented originally in [33]: this is the determination of the border of effective stability, using the theoretical separatrices of the most conspicuous secondary resonances of Hb. In detail: 1) We compute resonant normal forms for various secondary resonances of Hb, using an 'asym- metric expansion' for the Hamiltonian (see Section 2), which allows to speed up the convergence of both the original polynomial representation of the Hamiltonian as well as its normal form. The improvement obtained by the asymmetric expansion is demonstrated with numerical examples. 2) Using the classical normal form construction with Lie series in order to compute a resonant normal form for a specific secondary resonance, one ends with an expression for an invariant of the normal form called the 'resonant integral' Ψ (see Section 3). The level curves of Ψ allow, in turn, to obtain a theoretical phase portrait on a surface of section, and in particular to compute theoretical separatrices as well as the center of the secondary resonance. 3) The method typically yields underestimates of the position of the center and outer separatrix of 17 P´aez & Efthymiopoulos (2017) Submitted to CMDA the resonance, but a very efficient determination of the inner (closer to the libration center) separatrix of the resonance. 4) We argued that the inner limit ∆umin(ep,0) found in this way represents a clear border which exists in numerical stability maps between two well distinct domains in the space of the proper elements (∆u, ep,0) (see Section 4 for definitions). The inner domain is populated by regular orbits and isolated resonances with regular or marginally chaotic orbits, while the outer domain hosts either closely packed secondary resonances or a strongly chaotic domain. In fact, with increasing value of the primary's eccentricity e(cid:48), a modulation mechanism essentially wipes out all the resonances, creating a large outer domain of strong chaos. As a consequence, we argued that the inner domain, delimited by the innermost theoretical separatrix of the most conspicuous secondary resonance of Hb practically coincides with the limit of the effective stability domain for Trojan motions. 5) We demonstrated that the role of the secondary resonances of the basic model Hb, as delimiters of the domain of effective stability, covers most of the values of the parameters entering the problem (primary's mass and eccentricity, Trojan body's eccentricity), while it remains robust even in the full Three Body problem, for Trojan bodies of mass ∼ 1 Earth mass. Acknowledgements: Useful discussions with Prof. U. Locatelli are gratefully acknowledged. R.I.P. was supported by the Research Comittee of the Academy of Athens, under the grant 200/854. Appendix A The variables corresponding to the three degrees of freedom appearing in the expression of the basic Hamiltonian Hb in Eq.(5), (u, v), (Yf , φf ) and (Yp, φp) are given in terms of the orbital elements as follows: u = λ − λ(cid:48) − π 3 a − 1 , v = √ , β = ω − φ(cid:48) , (cid:16)(cid:112) √ (cid:17) , a y = 1 − e2 − 1 V =(cid:112)−2y sin β −(cid:112)−2y0 sin β0 , W =(cid:112)−2y cos β −(cid:112)−2y0 cos β0 , Y = − (cid:19) (cid:18) W 2 + V 2 (cid:19) (cid:18) V 2 φ = arctan W φf = λ(cid:48) − φ , (cid:90) ∂E Yf = ∂λ(cid:48) dt + v , Yp = Y − Yf , (39) (40) (41) (42) (43) (44) where λ, ω, a and e are the mean longitude, the longitude of the perihelion, the major semiaxis and eccentricity of the Trojan body, λ(cid:48) and φ(cid:48) = ω(cid:48) are the mean longitude and longitude of the perihelion 1 − e(cid:48)2 − 1, and E represents the total energy of the Trojan as of the perturber, β0 = π/3, y0 = computed from Eq. (1) (see [33] for further details in the construction). √ 18 P´aez & Efthymiopoulos (2017) Submitted to CMDA Appendix B The asymmetric expansion in terms of u = τ − π/3, up to a generic order K for the functions (2−2 cos τ )N/2 , (2−2 cos τ )N/2 , cosM τ and sinM τ , with N, M ∈ N fixed is given by cos τ sin τ K(cid:88) K(cid:88) k=0 k=0 cos τ (2 − 2 cos τ )N/2 = 1 2N/2 sin τ (2 − 2 cos τ )N/2 = 1 2N/2 K(cid:88) K(cid:88) k=0 cosM τ = sinM τ = and M1(k) uk+O(uK) , where M1(k) = F (i)(π/2) M2(k) uk+O(uK) , where M2(k) = G(i)(π/2) (cid:17)i−k (cid:17)i−k , , K(cid:88) K(cid:88) i=k i=k 1 i! 1 i! K(cid:88) K(cid:88) i=k 1 i! 1 i! B(i) M,M C (i) M,M 6 k (cid:19)(cid:16)− π (cid:18) i (cid:18) i (cid:19)(cid:16)− π (cid:19)(cid:16)− π (cid:17)i−k (cid:19)(cid:16)− π (cid:17)i−k k 6 6 , , 6 (cid:18) i (cid:18) i k k M3(k) uk + O(uK), where M3(k) = M4(k) uk + O(uK), where M4(k) = k=0 i=k [ n−1 2 ](cid:88) 2 ](cid:88) i=1 [ n i=0 F (n)(π/2) = G(n)(π/2) = (n, 2i − 1) (−1)i f (n−(2i−1))(π/2) , (n, 2i) (−1)i f (n−2i)(π/2) , with [ n−1 2 ] the integer part of n−1 2 , and [ n 2 ] the integer part of n 2 ; the derivatives f (n) are given by n(cid:88) m=1 f (n) (π/2) = A(n) m,m ; the coefficients A(n) m,m, B(n) M,M and C (n) M,M are given by (cid:18) 2(m − 1) + N (cid:19) m,m = −A(n−1) A(n) M,M = −B(n−1) B(n) M,M = C (n−1) C (n) m,m−1 − M,M−1 + (M + 1) B(n−1) M,M−1 − (M + 1) C (n−1) M,M +1, 2 M,M +1, A(n−1) m−1,m−1 , 1,1 = − N A(1) , 2 1,1 = −M , B(1) C (1) 1,1 = M . For a proof of these formulae, we refer the reader to [36]. References [1] Beaug´e, C., S´andor, Z., ´Erdi, B., Suli, A., (2007) Co-orbital terrestrial planets in exoplanetary systems: a formation scenario, Astron. Astrophys. 463, p 359. [2] Chirikov, B.V., Lieberman, M.A., Shepelyansky, D.L., Vivaldi, F.M., (1985) A theory of modula- tional diffusion, Physica D 12, p 289. [3] Cresswell, P., Nelson, R.P., (2009) On the growth and stability of Trojan planets, Astron. Astro- phys. 493, p 1141. [4] Dobrovolskis, A., (2013) Effects of Trojan exoplanets on the reflex motions of their parent stars, Icarus 226, p 1635. 19 P´aez & Efthymiopoulos (2017) Submitted to CMDA [5] Dvorak, R., Bazs´o, A., Zhou, L.Y., (2010) Where are the Uranus Trojans?, Celest. Mech. Dyn. Astron. 107, p 51. [6] Efthymiopoulos, C., Contopoulos, G., Voglis, N., (1999) Cantori, islands and asymptotic curves in the stickiness region Celest. Mech. Dyn. Astron. 73, p 221. [7] Efthymiopoulos, C., Giorgilli, A., Contopoulos, G., (2004) Nonconvergence on formal integrals: II. Improved estimates for the optimal order of truncations J. Phys. A 37, p 10831. [8] Efthymiopoulos, C. (2012) Canonical perturbation theory, stability and diffusion in Hamiltonian systems: applications in dynamical astronomy, in Third La Plata International School on Astro- nomy and Geophysics: Chaos, diffusion and non-integrability in Hamiltonian Systems - Applica- tions to Astronomy, Cincotta, P.M., Giordano, C.M., Efthymiopoulos, C., eds, p 3. [9] Efthymiopoulos, C., (2013) High order normal form stability estimates for co-orbital motion, Celest. Mech. Dyn. Astron. 117, p 101. [10] ´Erdi, B., (1988) Long periodic perturbations of Trojan asteroids, Celest. Mech. Dyn. Astron. 43, p 303. [11] ´Erdi, B., (1997) The Trojan Problem, Celest. Mech. Dyn. Astron. 65, p 149. [12] ´Erdi, B., Sandor, Z., (2005) Stability of co-orbital motion in exoplanetary systems, Celest. Mech. Dyn. Astron. 92, p 113. [13] ´Erdi, B., Nagy, I., S´andor, Z., Suli, A., Frohlich, G., (2007) Secondary resonances of co-orbital motions, MNRAS 381, p 33. [14] Froeschl´e, C., Guzzo, M., Lega, E., (2000) Graphical evolution of the Arnold web: from order to chaos, Science 289, p 2108. [15] Giorgilli, A., Skokos, C., (1997) On the stability of the Trojan asteroids, Astron. Astrophys. 317, p 254. [16] Giuppone, C., Ben´ıtez-Llambay, P, Beaug´e, C., (2012) Origin and detectability of co-orbital planets from radial velocity data, MNRAS 421, p 356. [17] Haghighipour, N., Capen, S., Hinse, T., (2013) Detection of Earth-mass and super-Earth Trojan planets using transit timing variation method, Celest. Mech. Dyn. Astron. 117, p 75. [18] Laughlin, G., Chambers, J.E., (2002) Extrasolar Trojans: the viability and detectability of planets in the 1:1 Resonance, Astron.J. 124, p 592. [19] Leleu, A., Robutel, P., Correia, A.C.M. (2015) Detectability of quasi-circular co-orbital planets. Application to the radial velocity technique, Astron.Astrophys. 581, p A128 [20] Leleu, A., (2016) Dynamics of co-orbital exoplanets - Ph.D. Thesis, arXiv:1701.05585 [21] Leleu, A., Robutel, P., Correia, A.C.M., Lillo-Box, J. (2017) Detection of co-orbital planets by combining transit and radial-velocity measurements, Astron.Astrophys. 599, p L7 [22] Levison, H., Shoemaker, E., Shoemaker, C., (1997) Dynamical evolution of Jupiter's Trojan asteroids, Nature 385, p 42. [23] Lhotka, C., Efthymiopoulos, C., Dvorak, R., (2008) Nekhoroshev stability at L4 or L5 in the elliptic restricted three-body problem -- Application to Trojan asteroids, MNRAS 384, p 1165. [24] Lykawka, P.S., Horner, J., Jones, B.W., Mukai, T., (2011) Origin and dynamical evolution of Neptune Trojans - II. Long term evolution, MNRAS 412(1), p 537. [25] Lyra, W., Johansen, A., Klahr, H., Piskunov, N., (2009) Standing on the shoulders of giants: Trojan Earths and vortex trapping in low mass self-gravitating protoplanetary disks of gas and solids, MNRAS 493, p 1125. [26] Marzari, F., Scholl, H., (2007) Dynamics of Jupiter Trojans during the 2:1 mean motion resonance crossing of Jupiter and Saturn, MNRAS 380, p 479. 20 P´aez & Efthymiopoulos (2017) Submitted to CMDA [27] Milani, A., (1993) The Trojan asteroid belt: proper elements, stability, chaos and families, Celest. Mech. Dyn. Astron. 57, p 59. [28] Morais, M.H.M, (1999) A secular theory for Trojan-type motion, Astron. Astrophys. 350, p 318. [29] Morais, M.H.M, (2001) Hamiltonian formulation on the secular theory for a Trojan-type motion, Astron. Astrophys. 369, p 677. [30] Nauenberg, M., (2002) Stability and eccentricity for two planets in a 1:1 resonances, and their possible occurrence in extrasolar planetary systems, Astron. J. 124, p 2332. [31] Neishtadt, A.I., (1987) On the change in the adiabatic invariant on crossing a separatrix in systems with two degrees of freedom, Prikl. Matem. Mekhan. 51(5), p. 750; PMM USSR 51(5), p 586. [32] Pierens, A., Raymond, S.N., (2014) Disruption of co-orbital (1:1) planetary resonances during gas-driven orbital migration, MNRAS, 442(2), p 2296. [33] P´aez, R.I., Efthymiopoulos, C., (2015) Trojan resonant dynamics, stability, and chaotic diffusion, for parameters relevant to exoplanetary systems, Celest. Mech. Dyn. Astron., 121(2), p 139. [34] P´aez, R.I., Locatelli, U., (2015) Trojan dynamics well approximated by a new Hamiltonian normal form, MNRAS 453(2), p 2177. [35] P´aez, R.I., Locatelli, U., Efthymiopoulos, C. (2016) Celest.Mech.Dyn.Astron 126, p 519 [36] P´aez, R.I. (2016) New normal form approaches adapted to the Trojan problem - Ph.D. Thesis, arXiv:1703.08819 [37] Robutel, P., Gabern, F., (2006) The resonant structure of Jupiter's Trojan asteroids - I. Long term stability and diffusion, MNRAS 372, p 1463. [38] Schwarz, R., Suli, ´A., Dvorak, R., Pilat-Lohinger, E., (2009) Stability of Trojan planets in multiplanetary systems, Celest. Mech. Dyn. Astron. 104, p 69. [39] Tsiganis, K., Varvoglis, H., Dvorak, R., (2005) Chaotic diffusion and effective stability of Jupiter Trojans, Celest. Mech. Dyn. Astron. 92, p 71. [40] Voglis, N., Efthymiopoulos, C., (1998) Angular dynamical spectra. A new method for determining frequencies, weak chaos and cantori J. Phys. A 31, p. 2913. 21
1109.4330	1	1109	2011-09-16T23:47:18	Merging Criteria for Giant Impacts of Protoplanets	[ "astro-ph.EP" ]	At the final stage of terrestrial planet formation, known as the giant impact stage, a few tens of Mars-sized protoplanets collide with one another to form terrestrial planets. Almost all previous studies on the orbital and accretional evolution of protoplanets in this stage have been based on the assumption of perfect accretion, where two colliding protoplanets always merge. However, recent impact simulations have shown that collisions among protoplanets are not always merging events, that is, two colliding protoplanets sometimes move apart after the collision (hit-and-run collision). As a first step towards studying the effects of such imperfect accretion of protoplanets on terrestrial planet formation, we investigated the merging criteria for collisions of rocky protoplanets. Using the smoothed particle hydrodynamic (SPH) method, we performed more than 1000 simulations of giant impacts with various parameter sets, such as the mass ratio of protoplanets, $\gamma$, the total mass of two protoplanets, $M_{\rm T}$, the impact angle, $\theta$, and the impact velocity, $v_{\rm imp}$. We investigated the critical impact velocity, $v_{\rm cr}$, at the transition between merging and hit-and-run collisions. We found that the normalized critical impact velocity, $v_{\rm cr}/v_{\rm esc}$, depends on $\gamma$ and $\theta$, but does not depend on $M_{\rm T}$, where $v_{\rm esc}$ is the two-body escape velocity. We derived a simple formula for $v_{\rm cr}/v_{\rm esc}$ as a function of $\gamma$ and $\theta$, and applied it to the giant impact events obtained by \textit{N}-body calculations in the previous studies. We found that 40% of these events should not be merging events.	astro-ph.EP	astro-ph	Accepted for publication in ApJ Merging Criteria for Giant Impacts of Protoplanets H. Genda Department of Earth and Planetary Science, The University of Tokyo, Hongo, Bunkyo-ku, Tokyo 113-0033, Japan [email protected] E. Kokubo Division of Theoretical Astronomy, National Astronomical Observatory of Japan, Osawa, Mitaka, Tokyo 181-8588, Japan and S. Ida Earth and Planetary Sciences, Tokyo Institute of Technology, Ookayama, Meguro-ku, Tokyo 152-8551, Japan ABSTRACT At the final stage of terrestrial planet formation, known as the giant impact stage, a few tens of Mars-sized protoplanets collide with one another to form terrestrial planets. Almost all previous studies on the orbital and accretional evolution of protoplanets in this stage have been based on the assumption of per- fect accretion, where two colliding protoplanets always merge. However, recent impact simulations have shown that collisions among protoplanets are not always merging events, that is, two colliding protoplanets sometimes move apart after the collision (hit-and-run collision). As a first step towards studying the effects of such imperfect accretion of protoplanets on terrestrial planet formation, we investigated the merging criteria for collisions of rocky protoplanets. Using the smoothed particle hydrodynamic (SPH) method, we performed more than 1000 simulations of giant impacts with various parameter sets, such as the mass ratio of protoplanets, γ, the total mass of two protoplanets, MT, the impact angle, θ, and the impact velocity, vimp. We investigated the critical impact velocity, vcr, at the transition between merging and hit-and-run collisions. We found that the – 2 – normalized critical impact velocity, vcr/vesc, depends on γ and θ, but does not depend on MT, where vesc is the two-body escape velocity. We derived a simple formula for vcr/vesc as a function of γ and θ (Eq. (16)), and applied it to the giant impact events obtained by N -body calculations in the previous studies. We found that 40% of these events should not be merging events. Subject headings: accretion, accretion disk - planets and satellites: formation - solar system: formation 1. Introduction Planets are formed in a disk around a star called a protoplanetary disk, which is com- posed of gas and dust. Terrestrial planets are formed mainly from the dust component. Their formation process can be divided into three stages. The first stage is the formation of a large number of kilometer sized bodies called planetesimals by accretion among dust particles (e.g., Goldreich & Ward 1973; Youdin & Shu 2002). In the second stage, these planetesimals collide to produce a few tens of Mars-sized objects called protoplanets (e.g., Wetherill 1985; Kokubo & Ida 1998). The final stage is the formation of terrestrial plan- ets from protoplanets (e.g., Chambers & Wetherill 1998; Agnor et al. 1999). The collisions among protoplanets are referred to as giant impacts, and thus this final stage is known as the giant impact stage. Giant impacts have a large influence on the various features such as the number of terrestrial planets formed, their mass and spin state (e.g., Agnor et al. 1999; Kokubo et al. 2006). Giant impacts are highly energetic events, and are responsible for the creation of large satellites, like the Moon (e.g., Canup 2004a) and planets with extremely large cores such as Mercury (e.g., Benz et al. 2007). Moreover, giant impacts are closely related to the thermal state such as a magma ocean (e.g., Tonks & Melosh 1992), and the origins of the terrestrial planet atmospheres (Genda & Abe 2005). A large number of simulations of giant impacts have been devoted to the specific gi- ant impact events related to the origin of the Moon or Mercury. However, since it is re- cently believed that multiple giant impacts are common during the last stage of terrestrial planet formation, several studies (Agnor & Asphaug 2004a; Asphaug 2009; Marcus et al. 2009, 2010) have investigated the giant impact simulations under various impact parame- ters. Agnor & Asphaug (2004a) were the first to show that collisions of protoplanets during the giant impact stage are not always merging events, that is, two colliding protoplanets sometimes move apart after the collision. They called such a collision a hit-and-run collision. – 3 – Except for Kokubo & Genda (2010), all the previous studies on the orbital and accretional evolution of protoplanets during the giant impact stage have been based on the assumption of perfect accretion, where two colliding protoplanets always merge. However, the hit-and- run collisions demonstrated by Agnor & Asphaug (2004a) may have an important influence on many of the physical characteristics of terrestrial planets. In order to investigate the effects of such imperfect accretion of protoplanets on terres- trial planet formation, the merging criteria during protoplanet collisions must be clarified. Agnor & Asphaug (2004a) performed 48 simulations of the collisions between same-sized protoplanets with masses of 0.1M⊕, where M⊕ is the Earth mass. They found that hit-and- run collisions occurred when vimp ≥ 1.5vesc for θ = 30◦, and vimp ≥ 1.2vesc for θ = 45◦ or 60◦, where vimp is the impact velocity, vesc is the two-body escape velocity (equation [8]), and θ is the impact angle. Since they varied vimp in steps of 0.1vesc for low-velocity col- lisions, the transition between the merging and hit-and-run collisions was estimated to be vimp = 1.4 − 1.5vesc for θ = 30◦, and vimp = 1.1 − 1.2vesc for θ = 45◦ or 60◦. Subsequently, Agnor & Asphaug (2004b) investigated collisions between different-sized protoplanets with mass ratios of 1:2 and 1:10. Their results were presented in Asphaug (2009). The transition was able to be estimated as follows. In the case of a 1:2 mass ratio, the transition occurs at vimp = 1.4 − 1.5vesc for θ = 30◦ and vimp = 1.1 − 1.2vesc for θ = 45◦ or 60◦, whereas in a case of 1:10 mass ratio, vimp = 1.5 − 2.0vesc for θ = 30◦, vimp = 1.2 − 1.3vesc for θ = 45◦, and vimp = 1.0 − 1.1vesc for θ = 60◦. Marcus et al. (2009) and Marcus et al. (2010) investigated collisions of the rocky and icy super-Earths (up to 10M⊕), respectively. They focused on the stripping of the rocky or icy mantle resulting from a high-velocity impact. Although they did not perform the various simulations with low-impact velocities, we were able to find that the transition occurs at roughly vimp = 1.0 − 1.5vesc from their figures. As described above, the merging criteria has already been roughly determined for certain discrete values of the impact parameters. However, in order to carry out N-body orbital calculations with the merging criteria for the giant impact stage, a simple formula describing the dependence of the merging criteria on the impact parameters is required. To achieve this, it is necessary to determine the merging criteria over a wide range of impact parameters. In the present study, we performed more than 1000 simulations of giant impacts for various impact parameter sets using the smoothed particle hydrodynamic (SPH) method in order to formulate the merging criteria. In Section 2, we present the SPH code and initial conditions used in our giant impact In Section 3, we show the collision outcomes, and investigate the transition simulations. – 4 – between merging and hit-and-run collisions. We also perform a resolution test on the simu- lations. In Section 4, we derive the merging criteria as a function of the impact parameters, and compare the results with those of previous studies. Using the derived criteria, we then discuss the merging probability of protoplanets during the giant impact stage in Section 5. 2. Calculation Method 2.1. Numerical Code In order to perform impact simulations for protoplanets, we used the SPH method (e.g., Monaghan 1992), which is a flexible Lagrangian method of solving hydrodynamic equations, and has been widely used in previous giant impact simulations. The SPH method can easily deal with large deformations and shock waves. Our numerical code is based on Canup (2004b); here, we briefly describe its essential points. The equation of the motion for the i -th SPH particle is given by dvi dt = − neighbor Xj Fij − all Xj Gij, (1) where vi is the velocity of the i -th SPH particle, t is the time, and Fij and Gij are the pressure gradient and mutual gravity terms between the i -th and j -th particles, respectively. Several forms have been used for the pressure gradient term, none of which appear to be clearly superior to the others. In this paper, we use the following symmetric expression, Fij = mj (cid:18) Pi ρ2 i + Pj ρ2 j + Πij(cid:19) ∇iW (rij, hij), (2) where mj, Pj, and ρj are the mass, pressure, and density of the j -th particle, respectively, Πij is the artificial viscosity, W is the kernel function, rij is the distance between the i -th and j -th particles, and hij is the average smoothing length of the i -th and j -th particles. For the artificial viscosity Πij, we use a Von Neumann-Richtmyer-type viscosity with parameters of αvis = 1.5 and βvis = 3.0, as described in Monaghan (1992). For the kernel function W , we use the spherically symmetric spline kernel function proposed by Monaghan & Lattanzio (1985): W (r, h) = 2( r 1 − 3 4(2 − r 1 0, 1 πh3   This function satisfies R W (r, h)dr = 1, and has a zero value when r ≥ 2h. In our code, h is variable for each particle and with time, and determined to satisfy the condition that the 4( r h )2 + 3 h )3, h )3, 0 ≤ r h < 1, 1 ≤ r h < 2, 2 ≤ r h . (3) number of neighboring particles (Nnei) within 2h is almost constant, Nnei = 64 ± 2. We used a maximum value of the smoothing length (hmax) to save computational cost when searching for neighboring particles and calculating the pressure gradient term. The value of hmax is determined from m ρmin (2hmax)3, – 5 – = 4π 3 (4) (5) where ρmin is the minimum density, which is set to 5 × 10−3 kg/m3 in our simulations. The mutual gravity term in equation (1) can be written as Gij = GXj mj ri − rj r3 ij , where G is the gravitational constant, and mj is the effective mass of the j -th particle toward the i -th particle defined by mj = Z rij 0 4πr2mjW (r, hj)dr. (6) This equation gives mj = mj when rij ≥ 2hj. The mutual gravity term between all SPH particles was directly computed using a special-purpose computer for gravitational N-body systems named GRAPE-6A (Fukushige et al. 2005). The GRAPE-6A can search for and produce lists of neighboring particles while si- multaneously calculating their mutual gravity. This list of neighbors is used to compute the pressure gradient term in the equation of motion and the time derivative of the internal energy. In a simulation over a period of 105 sec with 20,000 SPH particles, the typical CPU time was about 4 hours. Thus, the GRAPE-6A allowed us to systematically explore a wide range of impact parameters. Time integration was performed using a PEC (predict, evaluate, and correct) scheme with variable time steps (e.g., Serna et al. 1996), which is second-order accurate in time. 2.2. Pre-impact Protoplanets Here we describe the modeling method for the pre-impact protoplanets. All the proto- planets are assumed to be differentiated, with a 30% iron core and 70% silicate mantle by mass. In our SPH simulations, we used the Tillotson equation of state (Tillotson 1962), which has been widely applied to giant impact simulations involving shock waves (e.g., Benz et al. 1987; Canup & Asphaug 2001; Agnor & Asphaug 2004a; Asphaug et al. 2006). The Tillot- son equation of state contains ten material parameters, and the pressure is expressed as a – 6 – function of the density and the specific internal energy, which is convenient for treating fluid dynamics. We used the parameter sets of granite for the silicate mantle and iron for the iron core, which are listed on page 234 of Melosh (1989). All SPH particles in the protoplanets was set to have the same mass, and the total number of particles used for impact simulations was fixed at 20,000. For example, in the case of a collision of protoplanets with a mass ratio of 1:9, the smaller protoplanet consisted of 2,000 particles, and the larger one 18,000 particles. To model the pre-impact protoplanets, we placed the SPH particles in a 3D lattice (face-centered cubic) with iron particles on the inside, and rocky particles outside. The internal energy of the SPH particles was set to 1.0 × 106 J/kg. Beginning with this configuration, we calculated vibrations of the protoplanet until the particle velocities become slower than 100 m/s, which is much less than the impact velocity (order of km/s). After this operation, we used these relaxed objects as the protoplanets for impact simulation. As a first step, the protoplanets were assumed to have no spin. In order to set impact parameters such as the impact velocity (vimp) and impact an- gle (θ), we need to determine the radius of the pre-impact protoplanets. Since the surface boundary of the protoplanets described by the SPH particles is obscure owing to the smooth- ing length (h), we determined the radius of the pre-impact protoplanet (Rp) by the following equation. 4π 3 R3 p = Xi mi ρi . (7) 2.3. Initial Conditions for Collisions We prepared more than 1000 sets of initial conditions for the giant impact simulations. The parameters used were the mass ratio of the protoplanets (γ = Mi/Mt, where Mt and Mi are the mass of the target and impactor, respectively), the total mass of the two protoplanets (MT = Mi + Mt), the impact angle (θ), and the impact velocity (vimp). We systematically varied the mass ratio as γ = 1, 2/3, 1/2, 1/3, 1/4, 1/6, and 1/9. For γ = 1, 1/4, and 1/9, we considered three different values of MT. In total, we used 13 different combinations for the two colliding protoplanets (see table 1). For each mass combination, we varied the impact angle in the range θ = 0◦ − 75◦ in 15◦ steps, and the impact velocity in the range vimp = 1.0 − 3.0vesc in 0.2vesc steps, where vesc is the two-body escape velocity defined as vesc = r 2GMT Rt + Ri , (8) – 7 – where Rt and Ri are the radius of the target (larger protoplanet) and the impactor (smaller protoplanet), respectively. To precisely determine the transition between merging and hit- and-run collisions, we varied vimp with a smaller step size of 0.02 vesc near the transition. In total, we performed more than 1000 runs, consisting of 13 (mass combinations) × 6 (angles) × 16 (11 runs with 0.2vesc steps and ∼ 5 runs with 0.02vesc steps). The impact parameters vimp and θ are defined when the two protoplanets are in contact with each other (see Figure 1). We assumed that the two protoplanets are mass points, and calculated backward the positions of two mass points until their distance was apart at a distance of 3 (Ri+Rt). Then, we performed the giant impact simulations over a period of 105 sec. 3. Collision Outcomes 3.1. Merging and Hit-and-Run Collisions The outcomes of the collisions between the protoplanets are divided into two types: merging and hit-and-run collisions. According to the previous studies (Asphaug 2010; Leinhardt et al. 2010; Leinhardt & Stewart 2011), the collision outcomes are subdivided into several regimes (e.g., partial accretion, fragmentation and so on). However, we classify col- lision outcomes into only two regimes of merging and hit-and-run collisions, because it is the most essential to the evolution of protoplanets during the giant impact stage that only one big body is left after the giant impact (i.e., merging collision) or two big bodies are left (i.e., hit-and-run collision). Additionally, we need to classify collision outcomes as simply as possible, in order to incorporate those into N -body simulation. Figure 2 shows snapshots of two typical collisions. Panels (a) to (h) in Figure 2 show the time sequence for a relatively low-velocity collision (vimp = 1.3vesc) of same-sized protoplanets (Mi = Mt = 0.1M⊕) with θ = 30◦. After the first contact, the protoplanets become separated (see panel (e)), but remain gravitationally bound. Although some amount of mantle material is ejected, almost all parts of the colliding protoplanets finally merge. We refer to this type of collision as "a merging collision". On the other hand, a relatively high-velocity collision leads to a completely different result. Panels (i) to (l) show the time sequence of a collision with vimp = 1.5vesc. The impact angle and protoplanet masses are the same as in panels (a) to (h). After the first contact, the protoplanets escape from each other and are no longer gravitationally bound. We refer to this type of collision as "a hit-and-run collision". By examining snapshots at t = 105sec, we could easily classify almost all the simulations performed in this study as either merging or hit-and-run collisions. However, in some cases, – 8 – the protoplanets were separated but still gravitationally bound at t = 105 sec. Since those protoplanets are expected to eventually merge, we classified such a case as a merging collision. 3.2. Mass of Protoplanet after a Collision Here, we consider the mass of the protoplanet after the collision. We briefly describe the method to determine the mass of the gravitationally bound objects from the SPH particle data after t = 105 sec. In the first step, we roughly identify clumps of SPH particles using a friends-of-friends algorithm (e.g., Huchra & Geller 1982). We then iteratively check whether or not any SPH particles not belonging to clumps are gravitationally bound to clumps. Finally, we iteratively identify pairs of clumps that are gravitationally bound. Such clumps should merge after t = 105 sec, and we can thus regard them as a single object. We define the mass of the largest object as M1 and the second largest object as M2. Figure 3 shows the mass of the largest object normalized by the total mass, M1/MT, as a function of the impact velocity normalized by the escape velocity, vimp/vesc, for collisions of same-sized protoplanets with θ = 30◦. This figure also shows the numerical results of Agnor & Asphaug (2004a) for collisions of same-sized protoplanets with 0.1M⊕, which are in good agreement with our results for total mass MT of 0.2M⊕. We also performed simulations for MT = 0.4M⊕ and 1.0M⊕. As seen in Figure 3, M1/MT does not depend on the total mass, when the normalized impact velocity, vimp/vesc, is considered. The insensitivity to the total mass holds true to collisions between protoplanets, which has been predicted by Asphaug (2010). If the material properties such as strength dominates over gravity (typically collision between smaller bodies with less than 1 km in radius), the collision outcomes would depend on the total mass (Asphaug 2010). 3.3. Transition between Merging and Hit-and-Run Collisions Figure 3 indicates that M1/MT changes sharply around 1.4vesc. Collisions at impact velocities less than this velocity result in almost perfect accretion (i.e., M1/MT ≃ 1), and thus are classified as merging collisions. For the case of a near head-on collision (θ ≤ 15◦), M1/MT gradually decreases with the increase of vimp. Although such a collision should be classified as a partial accretion collision or a fragmentation collision according to Asphaug (2010) and Leinhardt & Stewart (2011), we here classify those as a merging collision because only one large body remains after the collision. Since the probability of a near head-on collision with high velocity is quite low during the giant impact stage, our treatment would not become a – 9 – serious problem. On the other hand, Figure 3 indicates that the collisions at impact velocities higher than 1.4vesc result in M1/MT ≃ 0.5. Some amount of material become stripped from the protoplanets, and the protoplanets escape from each other. These collisions are classified as hit-and-run collisions. We refer to the impact velocity at the transition between merging and hit-and-run collisions as the critical impact velocity, vcr. The normalized critical impact velocity, vcr/vesc, is not strongly dependent on the total mass, MT. We obtain vcr/vesc = 1.39 ± 0.01, 1.37 ± 0.01, and 1.37 ± 0.01 for MT = 0.2, 0.4 and 1.0M⊕, respectively. The critical impact velocity is expected to depend on the impact angle. Figure 4 is similar to Figure 3, but for the collisions with θ = 60◦. As is the same in the case of θ = 30◦, M1/MT for θ = 60◦ does not depend on the total mass. However, collisions with θ = 60◦ result in lower vcr/vesc values than the θ = 30◦ case. The calculated values are vcr/vesc = 1.11±0.01, 1.09±0.01, and 1.09±0.01 for MT = 0.2, 0.4 and 1.0M⊕, respectively. This result implies that collisions at higher impact angles are more like to be hit-and-run collisions. The critical impact velocity is also expected to depend on the mass ratio of the proto- planets (γ). Figure 5 shows the results for γ = 1/4 (mass ratio of 1:4) and θ = 30◦. It can be seen that in this case also, M1/MT does not depend on the total mass. This has also been verified for γ = 1/9 (mass ratio of 1:9), although the results are not shown here. As shown in Figure 5, M1/MT changes sharply around 1.5vesc. The normalized critical impact velocities are vcr/vesc = 1.57 ± 0.01, 1.53 ± 0.01, and 1.53 ± 0.01 for MT = 0.5, 1.0 and 1.5M⊕, respectively, which are larger than for the case of collisions between same-sized protoplanets. 3.4. Dependence on Particle Number and Initial Internal Energy In addition to the impact simulations with 20,000 SPH particles (standard case), we performed simulations with 3,000 (low-resolution case), 60,000 and 100,000 particles (high- resolution cases) for certain impact parameters in order to check the dependence of con- vergence on particle number. Although we found that the critical impact velocity for the low-resolution case was slightly different from that for the standard case, the high-resolution cases yielded the same results. For example, for γ = 1, MT = 0.2M⊕, and θ = 30◦, we obtained vcr/vesc = 1.39 ± 0.01 for both the standard and high-resolution cases. Therefore, using 20,000 SPH particles is enough for determining the critical impact velocity. In subsection 2.2, the initial internal energy of the SPH particles was set to 1.0 × 106 J/kg. To investigate the effect of the initial thermal state of the protoplanets, we prepared pre-impact protoplanets with an internal energies of 1.0 × 104 (cold-state case) and 3.0 × 106 J/kg (hot-state case). We then performed simulations of collisions between same-sized – 10 – protoplanets with impact parameters near the transition between merging and hit-and-run collisions. We found that although the results for the very low-impact angle case (θ = 15◦) showed a very slight dependence on the initial thermal state, for all other cases no such dependence was observed. 4. Merging Criteria for Colliding Protoplanets In the previous section, we determined the critical impact velocities for several impact parameters. In Figure 6, we summarize vcr/vesc for all parameter sets of the mass ratios (γ) and impact angles (θ). For γ = 1, 1/4, and 1/9, we performed simulations for three sets of MT (see Table 1). In Figure 6, we plot the average of these three results. The critical impact velocities estimated by Agnor & Asphaug (2004a) and Agnor & Asphaug (2004b) are also plotted in Figure 6. Agnor & Asphaug (2004a) investigated collisions be- tween same-sized protoplanets with masses of 0.1M⊕, and found that vcr/vesc = 1.4 − 1.5 for θ = 30◦, and vcr/vesc = 1.1 − 1.2 for θ = 45◦ or 60◦. Our results are in good agreement with those of Agnor & Asphaug (2004a), since in our simulations, vcr/vesc = 1.39 ± 0.01 for θ = 30◦, 1.19 ± 0.01 for θ = 45◦, and 1.11 ± 0.01 for θ = 60◦. Agnor & Asphaug (2004b) investigated collisions between different-sized protoplanets with mass ratios of 1:2 and 1:10. Based on Figure 17 in Asphaug (2009), for a mass ratio of 1:2, vcr/vesc = 1.4 − 1.5 for θ = 30◦, and 1.1 − 1.2 for θ = 45◦ or 60◦. In the present study, for a mass ratio of 1:2, we obtained vcr/vesc = 1.43 ± 0.01 for θ = 30◦, 1.19 ± 0.01 for θ = 45◦, and 1.09 ± 0.01 for θ = 60◦, and these values are in good agreement with those of Agnor & Asphaug (2004b). Although we did not perform simulations for a mass ratio of 1:10, our data points for 1:9 (γ = 1/9) seem to fall within the range obtained by Agnor & Asphaug (2004b). As shown in Figure 6, vcr/vesc increases with decreasing impact angle or mass ratio, which means that collisions with low impact angles or low mass ratios tend to be merging events. This can be explained in terms of the size of the overlapping volume of the colliding protoplanets. Since this is geometrically smaller for higher impact angle, the fraction of ki- netic energy converted to thermal energy of protoplanets and kinetic energy of the fragments is small, resulting in a hit-and-run collision. In addition, in the case of a small impactor (i.e., small γ), most of the volume of the impactor tends to overlap with the target. Therefore, the impactor can not easily be ejected, which leads to be a merging collision. In the following, we consider a simple physical model in order to express the critical impact velocity as a function of the impact angle and mass ratio. For two spheres with – 11 – radii Rt and Ri colliding with an impact angle θ (see Figure 7), the mass fractions of the overlapping volumes for the target and impactor (βt and βi, respectively) are geometrically given by (1 − sin θ)2(cid:26) 3 4 (cid:17)2 Rt M ov t Mt = (cid:16) Rt + Ri Ri (cid:17)2 (cid:16) Rt+Ri (1 − sin θ)2(cid:26) 3 1 − Rt 4(cid:16)Rt + Ri Ri (cid:17)(1 − sin θ)(cid:27), 4(cid:16) Rt+Ri 4 − 1 1, (cid:17)(1 − sin θ)(cid:27), (9) if sin θ ≥ 1 − 2Rt Ri+Rt , otherwise, βi = M ov i Mi βt = =   where M ov respectively. A constant density is assumed for simplicity. t and M ov (10) i are the masses of the overlapping volumes for the target and impactor, As illustrated in Figure 7, we divide the spheres into overlapping parts and non- overlapping parts, and consider the momentum exchange between the overlapping parts. Using the parameter of the degree of the momentum exchange (α), the post impact veloci- ties of the overlapping parts (vov i,post) are expressed as t,post and vov (cid:26) vov t,post = (1 − α)(vt − vov vov i,post = (1 − α)(vi − vov COM) + vov COM) + vov COM, COM, (11) where vov COM is the velocity of the center of mass of the overlapping parts, and is written as vov COM = M ov t vt + M ov M ov t + M ov i i vi . (12) For example, when α = 1, the velocities of the overlapping parts (vov vov COM owing to the complete momentum exchange. t,post and vov i,post) become The post impact velocities of the entire target and impactor can be derived based on the conservation of momentum, and written as (cid:26) Mtvt,post = (Mt − M ov Mivi,post = (Mi − M ov t )vt + M ov i )vi + M ov t vov i vov t,post, i,post, (13) where vt,post and vi,post are the velocities for the post-impact target and impactor, respec- tivery. Combining equations. (11) - (13) gives vt,post = (cid:16)1 − αβtβi vi,post = (cid:16)1 − αβtβi   Mt+Mi βtMt+βiMi(cid:17)vt, βtMt+βiMi(cid:17)vi. Mt+Mi (14) – 12 – If the relative velocity of post-impact objects is higher than the two-body escape velocity, a hit-and-run collision should occur. Therefore, to solve vi,post −vt,post = vesc using vcr = vi −vt, the critical impact velocity is obtained as follows: vcr vesc = (cid:20)1 − αβtβi Mt + Mi βtMt + βiMi(cid:21)−1 . (15) The calculated results for α = 0.6 are drawn as gray curves in Figure 8. We could roughly reproduce the dependence of vcr/vesc on θ and γ, but detail features, especially the case for the low-impact angle and low-mass ratio, could not be reproduced. For such a collision, the role of fragmentation that has not been considered here may become important. It should be also noted that α = 0.6 would not be applied to collisions other than giant impacts between rocky protoplanets. In addition to the above physical model, we tried to mathematically fit the numerical data for the critical impact velocity. Of the many possibilities available, we found that the following simple formula with five fitting parameters, c1 to c5, was most effective, vcr vesc = c1Γ Θc + cΓ + cΘc + c, (16) where Γ = (1 − γ)/(1 + γ) = (Mt − Mi)/MT, and Θ = 1 − sin θ. The fitting parameters are c1 = 2.43, c2 = −0.0408, c3 = 1.86, c4 = 1.08, and c5 = 5/2. The fitting curves produced by equation (16) are shown in Figure 8 as thick curves, and are in excellent agreement with numerical results. The value of the fitting parameters derived here is limited to the collision between protoplanets. 5. Merging Probability of Protoplanets Almost all previous N-body simulations of terrestrial planet formation during the giant impact stage have been performed based on the assumption of perfect accretion. In order to investigate the statistical properties of fully formed terrestrial planets, Kokubo et al. (2006) considered 10 sets of protoplanet initial conditions, and performed 20 runs for each set under the assumption of perfect accretion. Subsequently, to investigate the spin state of the formed planets, Kokubo & Ida (2007) performed additional 30 runs each for 7 sets of protoplanet initial conditions. Using the formula for the merging criteria (equation [16]) derived in the present study, we can now determine whether each of giant impacts was a merging or a hit-and-run event. Figure 9 shows the normalized impact velocity as a function of impact angle (left panel) and mass ratio (right panel) for 635 giant impact events during 50 runs under standard initial – 13 – conditions (Model 1 in Kokubo & Ida (2007)). The symbols denoted by crosses represent hit-and-run events, as determined by equation (16). In fact, 40% of all impact events (256 out of 635) are expected to be hit-and-run collisions. This result is consistent with a previous study by Agnor & Asphaug (2004a), who estimated a hit-and-run probability of roughly half. 6. Summary and Discussion During the giant impact stage of terrestrial planet formation in our solar system, a few tens of Mars-sized protoplanets collide with each other to form terrestrial planets. Almost all previous studies on N-body calculations of the giant impact stage have been based on the assumption of perfect accretion. However, recent impact simulations have shown that collisions of protoplanets are not always merging events. As a first step towards studying the effects of such imperfect accretion on terrestrial planet formation, we investigated the merging criteria for a collision of rocky protoplanets. Using the SPH method, we performed more than 1000 simulations of giant impacts for various parameter sets, such as the mass ratio of colliding protoplanets (γ), the total mass of two protoplanets (MT), the impact angle (θ), and the impact velocity (vimp). We investigated the critical impact velocity (vcr) at the transition between merging and hit-and-run collisions. We found that the normalized critical impact velocity, vcr/vesc, depends on γ and θ, but does not depend on MT. We derived a simple formula for vcr/vesc as a function of γ and θ (see equation [16]), and applied it to the giant impact events considered by Kokubo et al. (2006) and Kokubo & Ida (2007). We found that 40% of these events should not be merging events. Kokubo & Genda (2010) was the first to performed N-body simulations of the giant im- pact stage taking into account the merging criteria shown in equation (16), and investigated the effects of imperfect accretion on terrestrial planet formation. They found that some basic properties such as the final number, mass, orbital elements, and growth timescale of planets did not change very much, but the spin angular velocity of the fully formed planets was about 30% smaller than that for the perfect accretion model. They also determined that 49% of all impact events were hit-and-run collisions, which is also consistent with our estimate. In this paper, we focused on the merging criteria for protoplanet collisions. In the future, we plan to investigate additional collisional phenomena such as mantle stripping and ejection of small particles. Using our simulation data for more than 1000 collisional events, we can estimate the change in the core-mantle ratio during the giant impact stage. This is highly relevant to the formation of Mercury, and the formation probability of such planets – 14 – with very large cores will be investigated. The ejection of small particles during each collision in the giant impact stage may influence the orbital evolution of terrestrial planets. The ejected material may have damped the eccentricities of the terrestrial planets to their present low values, although it depends on the total amount of material ejected during the giant impact stage. acknowledgments - We thank Sarah T. Stewart for valuable comments on the manuscript. This research was partially supported by JSPS, the Grant-in-Aid for Young Scientists B (22740291), and MEXT, the Grant-in-Aid for Scientific Research on Priority Areas, and the Special Coordination Fund for Promoting Science and Technology. Agnor, C., & Asphaug, E. 2004a, ApJ, 613, L157 REFERENCES Agnor, C., & Asphaug, E. 2004b, American Geophysical Union, Fall Meeting, abstract #P32A-02 Agnor, C. B., Canup, R. M., & Levison, H. F. 1999, Icarus, 37, 219 Asphaug, E. 2009, Annu. Rev. Earth Planet. Sci., 37, 413 Asphaug, E. 2010, Chemie der Erde, 70, 199 Asphaug, E., Agnor, C., & Williams, Q. 2006, Nature, 439, 155 Benz, W., Slattery, W. L., & Cameron, A. G. W. 1987, Icarus, 71, 30 Benz, W., Anic, A., & Whitby, J. A. 2007, Space Sci. Rev., 132, 189 Canup, R. M., & Asphug, E. 2001, Nature, 412, 708 Canup, R. M. 2004a, ARA&A, 42, 441 Canup, R. M. 2004b, Icarus, 168, 433 Chambers, J. E., & Wetherill, G. W. 1998, Icarus, 136, 304 Fukushige, T., Makino, J., & Kawai, A. 2005, PASJ, 57, 1009 Genda, H., & Abe, Y. 2005, Nature, 433, 842 – 15 – Goldreich, P., & Ward, W. R. 1973, ApJ, 183, 1051 Huchra, J. P., & Geller, M. J. 1982, ApJ, 257, 423 Kokubo, E., & Genda, H. 2010, ApJ, 714, L21 Kokubo, E., & Ida, S. 1998, Icarus, 131, 171 Kokubo, E., & Ida, S. 2007, ApJ, 671, 2082 Kokubo, E., Kominami, J., & Ida, S. 2006, ApJ, 642, 1131 Leinhardt, Z. M., Marcus, R. A., & Stewart, S. T. 2010, ApJ, 714, 1789 Leinhardt, Z. M., & Stewart, S. T. 2011, MNRAS, submitted Marcus, R. A., Stewart, S. T., Sasselov, D., & Hernquist, L. 2009, ApJ, 700, L118 Marcus, R. A., Sasselov, D., Stewart, S. T., & Hernquist, L. 2010, ApJ, 719, L45 Melosh, H. J. 1989, Impact Cratering: A Geologic Process (New York: Oxford Univ. Press) Monaghan, J. J. 1992, ARA&A, 30, 543 Monaghan, J. J., & Lattanzio, J. C. 1985, A&A, 149, 135 Serna, A., Alimi, J.-M., & Chieze, J.-P. 1996, ApJ, 461, 884 Tillotson, J. H. 1962, Report No. GA-3216, July 18 (General Atomic, San Diego, Calfornia, 1962) Tonks, W. B., & Melosh, H. J. 1992, Icarus, 100, 326 Youdin, A. N., & Shu, F. H. 2002, ApJ, 580, 494 Wetherill, G. W. 1985, Science, 228, 877 This preprint was prepared with the AAS LATEX macros v5.2. – 16 – LE θ 4E 4J θ LJ Fig. 1.- Geometry of the collision between a larger (target) and smaller (impactor) pro- toplanets with radii of Rt and Ri, respectively. Since a center of mass coordinate system is used, the impact velocity, vimp, is given by vi − vt, where vt and vi are the velocities of the larger and smaller protoplanets, respectively. The impact angle is θ. A head-on collision corresponds to θ = 0◦. – 17 – Fig. 2.- Snapshots of two typical giant impacts between equal-mass protoplanets with 0.1M⊕. Panels (a) to (h) show the time sequence for a relatively low-velocity collision of protoplanets (vimp = 1.3vesc) with the impact angle 30◦. The colliding protoplanets finally merge so that this type of collision is referred to as a merging collision. Panels (i) to (l) show the time sequence for a relatively high-velocity collision (vimp = 1.5vesc). The impact angle and mass of the protoplanets are the same as in (a) to (h), but the protoplanets do not merge. This type of collision is referred to as a hit-and-run collision. The dark gray circle at the top-right corner in panels (i) to (l) indicates the size of the initial protoplanets. – 18 – 0.1M vs. 0.1M 0.2M vs. 0.2M 0.5M vs. 0.5M Agnor &Asphaug (2004) 1.5 2.0 / vimp vesc 2.5 3.0 1.0 0.8 0.6 0.4 0.2 T M / 1 M 0 1.0 Fig. 3.- Normalized mass of the largest gravitationally bound object, M1/MT, as a function of normalized impact velocity, vimp/vesc. Data for θ = 30◦ impacts between two equal-mass protoplanets with different total masses are plotted. Our collision outcomes are very similar to those obtained by Agnor & Asphaug (2004a) who performed simulations for collisions of same-sized protoplanets with masses of 0.1M⊕. – 19 – θ = 60o 0.1M vs. 0.1M 0.2M vs. 0.2M 0.5M vs. 0.5M Agnor &Asphaug (2004) 1.0 0.8 0.6 0.4 0.2 T M / 1 M 0 1.0 1.5 2.0 / vimp vesc 2.5 3.0 Fig. 4.- The same as Figure 3 but for θ = 60◦. – 20 – 1.0 0.8 0.6 0.4 0.2 T M / 2 M r o T M / 1 M θ = 30o 0.1M vs. 0.4M 0.2M vs. 0.8M 0.3M vs. 1.2M 0.1M vs. 0.4M 0.2M vs. 0.8M 0.3M vs. 1.2M M1 M2 0 1.0 1.5 2.0 / vimp vesc 2.5 3.0 Fig. 5.- The same as Figure 3 but for collisions of different-sized protoplanets with the mass ratio of 1:4 (γ = 1/4). The mass of the second-largest object, M2, is also plotted. – 21 – 30o (AA04b) 30o (AA04b) 45o&60o (AA04b) 45o 60o (AA04b) (AA04b) 0.4 0.6 (Mt (cid:15) Mi)/MT 0.8 1.0 3.0 2.5 2.0 1.5 c s e v / r c v this study 15o 30o 45o 60o 75o 30o (AA04a) 45o&60o (AA04a) 1.0 0 0.2 Fig. 6.- Critical impact velocities for the cases of the various impact angles and mass ratios of protoplanets. Note that (Mt − Mi)/MT is a function of γ, i.e., (1 − γ)/(1 + γ). Data points represent our numerical results for θ = 15◦ (filled circles), 30◦ (filled triangles), 45◦ (filled squares), 60◦ (open circles), and 75◦ (open triangles). Bars labeled with AA04a and AA04b are the results obtained by Agnor & Asphaug (2004a) and Agnor & Asphaug (2004b), respectively. – 22 – vi ov vi,post vi,post ov vt,post vi ov Mt ov Mi vt vt vt,post (a) pre-impact (b) momentum exchange (c) post-impact Fig. 7.- Configuration of a target and an impactor for the simple physical model. (a) Before the impact, the velocities of the target and the impactor are vt and vi, respectively. (b) During the impact, the overlapping parts (shaded areas) exchange momentum, and the post-impact velocity of the overlapping area is reduced to vov i,post for the impactor. (c) After the impact, velocities of the target and impactor are vt,post and vi,post, respectively. t,post for the target and vov – 23 – this study 15o 30o 45o 60o 75o 3.0 2.5 2.0 1.5 c s e v / r c v 1.0 0 0.2 0.4 0.6 (Mt (cid:15) Mi)/MT 0.8 1.0 Fig. 8.- Fits to the normalized critical impact velocity data. Data points are our numerical results and the same as shown in Figure 3. Thin and thick curves represent fits using equation (15) with α = 0.6, and equation (16), respectively. – 24 – c s e v / (a) p m i v (b) 4.0 3.5 3.0 2.5 2.0 1.5 0 10 20 30 40 50 60 70 80 90 1.0 0 0.2 0.4 0.6 (Mt (cid:15) Mi)/MT 0.8 1.0 4.0 3.5 3.0 2.5 2.0 1.5 1.0 c s e v / p m i v Fig. 9.- Normalized impact velocity for 635 giant impacts reported by Kokubo et al. (2006) and Kokubo & Ida (2007) as a function of the impact angle (a) and the mass ratio of proto- planets (b). Based on the critical impact velocity (equation [16]), the giant impact events are distinguished as hit-and-run collisions (cross symbols) or merging collisions (circle symbols). Although Kokubo et al. (2006) and Kokubo & Ida (2007) assumed the perfect accretion of protoplanets in their N-body calculations, the present study reveals that 40% of the impact events (256 out of 635) are hit-and-run collisions. – 25 – Table 1: Parameter sets for the mass ratio of colliding protoplanets γ = 1 2/3 1/2 1/3 1/4 1/6 1/9 0.2M⊕ vs. 0.2M⊕ 0.5M⊕ vs. 0.5M⊕ 0.1M⊕ vs. 0.1M⊕ 0.2M⊕ vs. 0.3M⊕ 0.1M⊕ vs. 0.2M⊕ 0.1M⊕ vs. 0.3M⊕ 0.1M⊕ vs. 0.4M⊕ 0.1M⊕ vs. 0.6M⊕ 0.05M⊕ vs. 0.45M⊕ 0.1M⊕ vs. 0.9M⊕ 0.2M⊕ vs. 1.8M⊕ 0.2M⊕ vs. 0.8M⊕ 0.3M⊕ vs. 1.2M⊕ - - - - - - - -
1302.1620	2	1302	2013-10-24T17:08:40	Secular Orbital Evolution of Compact Planet Systems	[ "astro-ph.EP" ]	Recent observations have shown that at least some close-in exoplanets maintain eccentric orbits despite tidal circularization timescales that are typically shorter than stellar ages. We explore gravitational interactions with a distant planetary companion as a possible cause of these non-zero eccentricities. For simplicity, we focus on the evolution of a planar two-planet system subject to slow eccentricity damping and provide an intuitive interpretation of the resulting long-term orbital evolution. We show that dissipation shifts the two normal eigenmode frequencies and eccentricity ratios of the standard secular theory slightly, and that each mode decays at its own rate. Tidal damping of the eccentricities drives orbits to transition between periods of pericenter circulation and libration, and the planetary system settles into a locked state where the pericenters are nearly aligned or anti-aligned. Once in the locked state, the eccentricities of the two orbits decrease very slowly due to tides rather than at the much more rapid single-planet rate, and thus eccentric orbits, even for close-in planets, can often survive much longer than the age of the system. Assuming that an observed close-in planet on an elliptical orbit is apsidally-locked to a more distant, and perhaps unseen companion, we provide a constraint on the mass, semi-major axis, and eccentricity of the companion. We find the observed two-planet system HAT-P-13 might be in just such an apsidally-locked state, with parameters that obey our constraint well. We also survey close-in single planets, and found that none provide compelling evidence for unseen companions. Instead, we suspect that (1) orbits are circular, (2) tidal damping rates are slower than our assumption, or (3) a recent event has excited these eccentricities. Our method should prove useful for interpreting the results of current and future planet searches.	astro-ph.EP	astro-ph	Draft version October 30, 2018 Preprint typeset using LATEX style emulateapj v. 5/2/11 SECULAR ORBITAL EVOLUTION OF COMPACT PLANET SYSTEMS Ke Zhang1, Douglas P. Hamilton1, and Soko Matsumura1,2 Draft version October 30, 2018 ABSTRACT Recent observations have shown that at least some close-in exoplanets maintain eccentric orbits despite tidal circularization timescales that are typically much shorter than stellar ages. We explore gravitational interactions with a more distant planetary companion as a possible cause of these unex- pected non-zero eccentricities. For simplicity, we focus on the evolution of a planar two-planet system subject to slow eccentricity damping and provide an intuitive interpretation of the resulting long-term orbital evolution. We show that dissipation shifts the two normal eigenmode frequencies and eccen- tricity ratios of the standard secular theory slightly, and we confirm that each mode decays at its own rate. Tidal damping of the eccentricities drives orbits to transition relatively quickly between periods of pericenter circulation and libration, and the planetary system settles into a locked state in which the pericenters are nearly aligned or nearly anti-aligned. Once in the locked state, the eccentricities of the two orbits decrease very slowly because of tides rather than at the much more rapid single-planet rate, and thus eccentric orbits, even for close-in planets, can often survive much longer than the age of the system. Assuming that an observed close- in planet on an elliptical orbit is apsidally-locked to a more distant, and perhaps unseen companion, we provide a constraint on the mass, semi-major axis, and eccentricity of the companion. We find that the observed two-planet system HAT-P-13 might be in just such an apsidally locked state, with parameters that obey our constraint reasonably well. We also survey close-in single planets, some with and some without an indication of an outer companion. None of the dozen systems that we investigate provides compelling evidence for unseen companions. Instead, we suspect that (1) orbits are in fact circular, (2) tidal damping rates are much slower than we have assumed, or (3) a recent event has excited these eccentricities. Our method should prove useful for interpreting the results of both current and future planet searches. Subject headings: planetary systems - planets and satellites: dynamical evolution and stability - planets and satellites: general 1. INTRODUCTION In the past decade, many mechanisms have been pro- posed to explain a wide range of eccentricities seen among exoplanets including planet-planet scattering, planet- disk interaction, mean-motion resonance passage, and Kozai resonances (see, e.g., a review by Namouni (2007) and references there-in). All of these mechanisms can excite orbital eccentricities effectively. While the orbits of long-period planets could stay eccentric for billions of years, those of most close-in planets are likely to be cir- cularized within stellar ages because of tidal interactions between the stars and the planets, except in some inter- esting special cases considered by Correia et al. (2012). In Figure 1, we plot the eccentricities of all 222 known (as of 2012 February) close-in planets with orbital pe- riods of less than 20 days. The average eccentricity for these planets (∼ 0.06) is significantly smaller than that of all of the confirmed planets (∼ 0.18), which indicates that the tidal interaction is an effective ec- centricity damping mechanism and that circularization timescales are typically shorter than stellar ages (about 1-10 Gyr). Nevertheless, nearly half of these close-in planets have non-zero orbital eccentricities. Recent stud- Corresponding Authors: [email protected], [email protected] 1 Department of Astronomy, University of Maryland, College Park, MD 20742, USA 2 School of Engineering, Physics, and Mathematics, University of Dundee, Scotland DD1 4HN 0.7 0.6 0.5 y t i c i r t n e c c E 0.4 0.3 0.2 0.1 0.0 100 Period [days] 101 Fig. 1. -- Eccentricities of close-in planets with orbital periods less than 20 days. The blue circles and orange triangles correspond to single- and multiple-planet systems, respectively. Error bars are plotted for both period and eccentricity. However, errors in period are not apparent because they are very small. Orbital data, cour- tesy of the Exoplanet Orbit Database (http://exoplanets.org/). ies have shown, however, that the orbital fits tend to overestimate eccentricities (e.g., Shen & Turner 2008; Zakamska et al. 2011; Pont et al. 2011), so some mea- sured eccentric orbits for close-in planets might in fact be circular. However, it is unlikely that all close-in plan- ets have perfectly circular orbits and thus at least some of the non-zero eccentricities require a physical explana- tion. The main possibilities include (1) systems have only 2 recently attained their orbital configurations and eccen- tricities are damping quickly, (2) planetary tidal quality factors are much larger than those of the giant planets in our solar system and so eccentricities damp slowly, and (3) eccentricity excitation caused by an exterior planet slows the orbital circularization. In this paper, we ex- plore the third option and investigate the gravitational interactions between a close-in planet and a more distant companion. In systems with more than one planet, the most sig- nificant orbit-orbit interactions are often mean-motion resonances, which have been studied in detail for the satellite systems of the giant planets (see reviews by Greenberg 1977; Peale 1986). Resonances in extra- solar planetary systems have also received increased attention (e.g., Chiang 2003; Beaugé et al. 2003; Lee 2004; Ketchum et al. 2013; Batygin & Morbidelli 2013). Mean-motion resonance passages during planetary mi- gration can be effective in exciting orbital eccentricities. These resonance passages can be divergent, in which case impulsive changes to the orbital elements result, or con- vergent, in which case trapping into resonance usually occurs (Hamilton 1994; Zhang & Hamilton 2007, 2008). Although resonance capture typically leads to excited ec- centricities, we do not consider this process further here. Instead, we focus on the more prosaic secular interactions which are capable of maintaining orbital eccentricities for a greater variety of orbital configurations. systems (Butler et al. 1999), Secular perturbations have been studied for cen- turies in the context of the Solar System (see, e.g., Brouwer et al. 1950). With the discovery of extrasolar multi-planet applica- tions for secular theory have expanded rapidly (e.g., Wu & Goldreich 2002; Barnes & Greenberg 2006b; Adams & Laughlin 2006a; Mardling 2007; Batygin et al. 2009; Greenberg & Van Laerhoven 2011; Laskar et al. 2012). Wu & Goldreich (2002) were the first to use secular interactions to explain the non-zero eccentricity of a hot Jupiter. They showed that a close-in planet with a companion could sustain a substantial orbital eccentricity even though tidal damping was efficient. Zhang & Hamilton (2003) and Zhang (2007) confirmed the results of Wu & Goldreich (2002) and obtained most of the results contained in Section 2 of the present article. Mardling (2007) developed a detailed octopole- order secular theory to study orbital evolution of close-in planets with companions and also confirmed the results of Wu & Goldreich (2002). Laskar et al. (2012) pointed out the importance of precession due to tidal effects. If these well-studied secular interactions play an important role in delaying eccentricity damping, we might expect differences in the eccentricity distributions between single- and multiple-planet systems. There is no obvious difference between these two groups of planets, however, as can be seen in Figure 1. Perhaps future observations might reveal such a difference, but it is also plausible that many single close-in planets are actually accom- panied by unobserved companions that help maintain their eccentricities against tidal dissipation. In this paper, we revisit the problem of secular inter- actions with a distant companion in maintaining the ec- centricities of close-in planets. Our goals are to develop an intuitive interpretation of the secular theory of a two- planet system and to test the model against observed planetary systems. In the next section, we present a lin- ear Laplace-Lagrange model for secular orbital evolution during tidal dissipation, starting with a review of secu- lar orbital interactions in a stable non-dissipative system consisting of a star and two planets. We then add tidal dissipation of eccentricities, and solve the coupled sys- tem to investigate how eccentricity damping affects the apsidal state of the two orbits. We add the precession due to planetary and stellar tidal and rotational bulges as well as general relativity (GR) terms which can be sig- nificant for close-in planets. In Section 3, we apply this model to extrasolar planetary systems, illustrating how it might help to guide planet searches. Last, we discuss and summarize our work in Section 4. 2. MODEL 2.1. Stable Non-dissipative Systems with Two Planets The secular solution of a stable non-dissipative 2- planet system is known as Laplace-Lagrange theory and is discussed in great detail in Murray & Dermott (1999). In this section, we will develop a graphical interpretation of the solution that will help us understand the more complicated systems studied later in this paper. After averaging the disturbing functions of each planet on the other, Rj, over both planets' orbital periods, Lagrange's planetary equations can be linearized for small eccentric- ities and inclinations to (Murray & Dermott 1999, Sec- tion 7.1) aj = 0; ej = − Ij = − 1 nja2 j ej 1 nja2 j Ij ∂Rj ∂j ∂Rj ∂Ωj , , j = + Ωj = + 1 nja2 j ej 1 nja2 j Ij (1) ∂Rj ∂ej ∂Rj ∂Ij ; . Here, nj, aj, ej, Ij , Ωj, and j are the mean motion, semi-major axis, eccentricity, inclination, longitude of as- cending node, and argument of the pericenter of the jth planetary orbit, respectively. Planetary semi-major axes remain constant and hence no long-term energy transfer between the orbits occurs because we have averaged a conservative perturbation force over each planet's orbital period. Note that be- cause of our assumption of small eccentricities and in- clinations, the evolution equations of (ej, j) and (Ij , Ωj) are completely decoupled and can be analyzed sep- arately. In this paper, we neglect the inclination and node pair since they are less easily observable for extra- solar planets. Because the two sets of equations take the same form, however, our eccentricity results below can be easily applied to secular coupling of vertical motions. The disturbing function R is defined for a planet location as at (Murray & Dermott 1999, Section 6). two- planet systems without any external perturbations, it is simply the gravitational potential caused by the other planet. Here, we consider a simple planetary system consisting of a central star and two planets in co-planar orbits. In terms of osculating elements (see, e.g., Murray & Dermott 1999, Section 2.9) and to second order in small eccentricities, Murray & Dermott (1999) show that the orbit-averaged disturbing functions non-Keplerian potential For the its are given by: 3 R1 = n1a2 R2 = n2a2 2 e2 1 σq(cid:20) 1 2 σ√α(cid:20) 1 1 − β e1e2 cos(1 − 2)(cid:21) , 2 − β e1e2 cos(1 − 2)(cid:21) , (3) (2) e2 2 where the subscript "1" refers to the inner planet and "2" to the outer one. We define the mass ratio between the two planets q = m2/m1, and the semi-major ratio of the two orbits α = a1/a2. The remaining parameters are defined as follows: β = b(2) 3/2(α) b(1) 3/2(α) and σ = 1 4 n1 m1 m∗ α2b(1) 3/2(α), 3/2(α) and b(2) where m∗ is the stellar mass and b(1) 3/2(α) are two of the Laplace coefficients (Murray & Dermott 1999). The parameter σ has units of frequency, which we will soon see characterizes the secular precession rates. Both β and σ decrease with increasing planetary separa- tion and, for small α, the Laplace coefficients reduce to 3/2(α) ≈ 3α and b(2) b(1) Following Brouwer et al. (1950), we transform a (ej, j) pair into a complex Poincaré canonical variable hj with the mapping: 3/2(α) ≈ 15α2/4. hj = ej exp(ij), (4) where i = √−1. Substituting Equations (2) and (3) into Equation (1) and rewriting in terms of hj yields a set of linear homogeneous ordinary differential equations similar to those for a double pendulum system: 2 where the coefficient matrix Ajkhk, hj = i Xk=1 −√αβ √α (cid:27) . A = σ(cid:26) q −qβ (5) Two orthogonal special solutions of Equation. (5), or the secular eigenmodes of the system, are given by (cid:18) h1± h2± (cid:19) =(cid:18) 1 ηs ± (cid:19) exp(igs ±t), where the eigen-frequencies gs ters ηs ± can be obtained from the matrix A: ± and eigenvector parame- gs ± = ηs ± = σ 2 (cid:20)q + √α ∓q(q − √α)2 + 4q√αβ2(cid:21) , q − √α ±p(q − √α)2 + 4q√αβ2 2qβ . (6) (7) We use the superscript "s" to indicate that those param- eters are for a "static", or non-dissipative, system. The eigenmode components h1± and h2± depend only on the fixed constants α and q through Equations. (6) and (7), and not on the initial eccentricities and pericenter angles. The physical meaning of the two modes can be elu- cidated by transforming the solution of hj back to the Fig. 2. -- Aligned and anti-aligned secular modes. The planets (solid dots) follow nearly elliptical orbits about the central star. Arrows point from the star to orbital pericenters. 2 1 0 s+ h b = 0.9 0.5 0.2 0.1 s − h 1 − 2 − 0.0 0.5 b = 0.1 1.0 √a /q 0.2 0.5 0.9 1.5 2.0 Fig. 3. -- Eccentricity ratios vs. √α/q in the pure secular modes as given by Equation (7). Different curves represent different β values. Along the dashed line q = √α (or m2 2a2) and orbits have e1 = e2 in either mode. 1a1 = m2 orbital elements (e, ) with Equation (4). If the system is fully in either the "+" or the " -- " mode, we find the following: 1± = 2± = gs ±, (e2/e1)± = ηs ±, ±/ηs cos(∆±) = ηs ±, (8) (9) (10) where ∆ = 2 − 1 is the difference between the two pericenter angles. In either mode, the two orbits precess together at the rate g± (Equation (8)), and their eccen- tricities keep a fixed ratio (Equation (9)). Furthermore, Equation (7) shows that ηs − < 0. Thus, Equation (10) states that the pericenters of the two or- bits are always aligned in the "+" mode (cos(∆) = 1), and anti-aligned in the " -- " mode (cos(∆) = −1). In an eigenmode, the system behaves as a rigid body with the shapes and relative orientation of the elliptical orbits remaining fixed (Figure 2). The frequency of the anti- aligned mode is faster (g− > g+) because of closer ap- proaches between the planetary orbits and thus stronger perturbations (Figure 2). + > 0 while ηs − = σ√α, ηs + = σq(1 − β2), ηs It is instructive to consider the small q limit which In this case, Equa- + = β) corresponds to a tiny outer mass. tions (6) and (7) simplify to (gs − = −√α/qβ). For a true outer test and (gs particle q → 0 and only the first mode is possible. This aligned mode is stationary (g+ = 0) and, since ηs + < 1, the planet's eccentricity exceeds that of the test particle. Similarly, with an inner test particle only the aligned mode survives, it is stationary, and the massive planet again has the higher eccentricity. 4 ± 1a1 = m2 Equation (9) gives the ratio between the eccentricities in a perfect secular eigenmode. In Figure 3, we plot ηs as a function of √α/q for different β values. Although rare in real systems, q = √α (or m2 2a2) makes an interesting special case. When this condition is met, the planets have similar angular momenta and the inner and outer planet orbital precession rates (the diagonal terms of the matrix A) are equal. In addition, ηs ± = ±1 (Equation (7)), and therefore the inner and outer orbits have the same eccentricity (Equation (9)). When q > √α (m2 2a2), the inner planet precesses fastest, has a lower eccentricity in the aligned mode and a higher eccentricity in the anti-aligned mode. The opposite is true for q < √α. 1a1 < m2 In general, a system is in a mixed state composed of a linear combination of the two modes: hj = e+ exp(iϕ+) hj+ + e− exp(iϕ−) hj−. (11) Here the mode amplitudes e± and phases ϕ± are deter- mined by the initial orbits. Plotting the solution Equa- tion (11) on the complex plane yields a phase plot of e cos versus e sin as shown in Figure 4a. The e1 and e2 vectors in the plot represent (e, ) pairs for the two orbits at a given time. The length of a vector is the instantaneous eccentricity, and its polar angle is the instantaneous longitude of pericenter. Each eccentric- ity vector is a vector sum of an aligned component (e+ and ηs +e+ are parallel) and an anti-aligned component (e− and ηs − e− are antiparallel). The lengths of all of these vectors are determined by initial conditions. In a pure aligned eigenmode, e− = e− = 0 and the e1 and e2 vectors are parallel. As time progresses, these ec- centricity vectors rotate together at rate gs + while main- taining their lengths. This solution corresponds to the aligned orbits in Figure 2. For the anti-aligned eigen- mode, e+ = e+ = 0 so that e1 and e2 are anti-parallel and rotate together at rate gs −. This state is depicted by the pair of anti-aligned orbits in Figure 2. In the most general system, both motions occur simultaneously: the parallel eccentricity vectors (e+ and ηs +e+) rotate at rate gs +, while the anti-parallel vectors (e− and ηs − e−) rotate at rate gs −. The resulting lengths of the eccentricities e1 and e2, thus vary periodically as illustrated in Figure 4b. The maximum value of e1 occurs when e+ is parallel to e−. At the same time, however, ηs −e− are anti-parallel to each other, leading to a minimum value for e2. The simultaneous maximum for e1 and minimum for e2 seen in Figure 4b is a general result guaranteed by angular momentum conservation. The two eccentricities, in mathematical form are as follows: − − gs −)2 + 2e+e−ηs − + 2e+e− cos(gs +e+ and ηs +)2 + (e−ηs − cos(gs + + e2 +ηs +)t, e1 =qe2 e2 =q(e+ηs (Murray & Dermott 1999). Both eccentricities oscillate at the same frequency (gs +) as shown in Figure 4b. − − gs +)t − − gs 2.2. Secular Modes with Eccentricity Damping Since a planet close to its host star experiences a drag force caused by planetary tides raised by the star, we seek a way to include tides into the mathematical formalism of the previous section. For small eccentricities for which the secular solution Equation (11) is valid, tidal changes in a are usually negligible compared with the damping in e (Goldreich 1963). Since the damping timescales for close-in planets (in the order of 108 yr) are much longer than the secular timescales (typically ∼ 103 yr), we treat the damping effect as a small perturbation to the secular solution. Stars and planets raise tides on each other, which, in turn, perturb the planet's orbit. For very close-in plan- ets, the combination of stellar tides raised by the planet and planetary tides raised by the star typically act to de- crease the planet's orbital period and eccentricity. Tidal dissipation in the star usually leads to orbital decay and circularization on a timescale much longer than the age of most planetary systems, while that tides in the planet can damp its orbital eccentricity rather quickly (e.g., Goldreich 1963; Burns 1977; Rasio et al. 1996). Thus, we assume that the orbital circularization is largely dom- inated by planetary tides, and approximate the eccen- tricity damping rate as follows (e.g., Murray & Dermott 1999): λ = − e e = 63 4 1 Q′ p m∗ a (cid:19)5 mp (cid:18) Rp n , (12) where Q′ p ≡ 1.5Qp/k2 is the modified tidal quality factor, Qp is the tidal quality factor, and k2 is the Love number of degree 2. The addition of the constant tidal damping (Equa- tion (12)) adds an extra term to the eccentricity equation in Equation (1), which now reads as follows: ej = − 1 nja2 j ej ∂Rj ∂j − λjej, with λj being the eccentricity damping rate. The coeffi- cient matrix of Equation (5) is now the following: A = σ(cid:26) q + i ξ1 −√αβ √α + i ξ2 (cid:27) , −qβ (13) where the dimensionless ξj = λj /σ ≪ 1 parameterizes the damping strength. Eccentricity damping causes the eigen-frequencies of matrix A to have both real and imaginary parts. As in other dynamical systems, the real parts (g±) of the eigen-frequencies still represent the precession rates of the secular modes, while the imaginary parts (γ±) indi- cate that the amplitudes of the modes change over time. This becomes clearer if we rewrite the two orthogonal special solutions of the system as follows: η± (cid:19) exp(−γ±t) exp(ig±t), (cid:18) h1± h2± (cid:19) =(cid:18) 1 For ξj ≪ 1, we solve the new matrix for the complex where η± is the new eccentricity ratio for each mode. frequencies and find the following: (14) q√αβ2 [(q − √α)2 + 4 q√αβ2]3/2 σ (ξ1 − ξ2)2, (15) g± = gs γ± = ± ± 2"λ1 + λ2 ± 1 √α − q p(q − √α)2 + 4 q√αβ2 (λ1 − λ2)# . (16) Eccentricity damping increases the precession rate of 5 e1 e− − g−t+j 1 e e+ g+t+j + e cos 2 e 0.020 0.018 0.016 0.014 0.012 0.010 0.008 0.006 0.004 n i s e (a) e2 − −e− +e+ g−t+j g+t+j + 0 2000 4000 time (yr) (b) 6000 8000 + − e−) rotate at rate gs e+, rotate (precess) at rate gs Fig. 4. -- General solution for a two-planet system. (a) The solution (Equation (11)) on a phase plot. The arrows represent the eccentricity vector e exp(iω). The total eccentricity of each orbit (e1 or e2) is the magnitude of the vector sum of aligned (+) and anti- aligned (−) components. The two aligned components, e+ and ηs +, while the two anti-aligned components (e− and ηs −. (b) Secular evolution of orbital eccentricities from an N-body simulation. Plot shows the eccentricities of two planetary orbits in a computer-simulated system consisting of a 1 Solar-mass star, a 1 Jupiter-mass hot-Jupiter at 0.05 AU, and a 0.8 Jupiter-mass companion at 0.2 AU. The simulation shows an oscillation period of ∼ 1673 yr, in good agreement with the prediction of the secular model: 2π/(g− − g+) = 1670 yr. the aligned mode and decreases that of the anti-aligned mode, but only by an extremely small amount (of or- der ξ2 j ). These tiny frequency changes were neglected by Laskar et al. (2012), but otherwise our damping rates are in perfect agreement. The tiny frequency changes are due to the slightly different orbital configurations in the secular modes as we shall describe below. The new eigen-vectors of the matrix A are as follows: η±, are nearly the same as ηs ±. Thus, we continue to use "aligned" and "anti-aligned" to refer to the two modes. Nevertheless, the small deviation angle is physically im- portant because it is what enables monotonic damping of the outer planet's eccentricity. (17) η± = ηs ±"1 ± i ξ1 − ξ2 p(q − √α)2 + 4 q√αβ2# , where we have neglected terms of second- and higher- order power of ξj. In general, the η±'s are complex with small imaginary If we ignore the imaginary parts for the components. moment, then η± are real and Equation (14) shows that the pericenter angles of the two orbits are the same (pos- itive η+, aligned mode) or 180◦ apart (negative η−, anti- aligned mode). For complex η±, however, the two orbits are not exactly aligned or anti-aligned any longer. In- stead, ∆± shifts from 0◦ and 180◦ by a small angle ǫ = tan−1 ξ1 − ξ2 p(q − √α)2 + 4 q√αβ2! ≈ ξ1 − ξ2 p(q − √α)2 + 4 q√αβ2 In the "aligned" mode, the new pericenter difference is ∆+ = ǫ so that the inner exoplanet's pericenter slightly lags that of the outer exoplanet. The lag is maximized for q = √α, the case with equal eccentricities and equal precession rates for the two planets. Because of this mis- alignment, the minimum distance between the two orbits is slightly less than that in the undamped case (see Fig- ure 2). This causes the average interaction between the two orbits to be stronger, leading to an increase of the precession frequency as indicated by Equation (15). Sim- ilarly, ∆− = 180◦ − ǫ in the "anti-aligned" mode; the slight rotation results in a weaker average interaction and a slightly slower mode-precession rate. The deviation an- gle ǫ is tiny, and the eccentricity ratios in the two modes, In addition to the slight mis-alignment, each mode am- plitude also damps at the rate given by Equation (16). If only planetary tides contribute to eccentricity damp- ing, Equation (12) shows that the damping rate decreases rapidly with the planet's semi-major axis (λ ∝ a−6.5). In the absence of secular interactions between the planets, the outer orbit is hardly affected. With this interaction, however, the damping applied to the eccentricity of the inner orbit is partially transmitted to the outer planet, causing a decrease of its eccentricity as well. The damp- ing rates of the two modes are different, unless q = √α. An interesting result from Equation (16) is that the sum of the two mode-damping rates is equal to the sum of the two individual eccentricity damping rates: γ+ + γ− = λ1 + λ2. . The physical interpretation of this expression is that sec- ular interactions between the planets simply act to redis- tribute where the damping occurs. In Figure 5, we compare our analytical results with nu- merical integration of both the secular equations and the direct N-body equations, with an artificial eccentricity damping added only to the inner planet in all cases. The eccentricity evolution curves of the inner planet are plot- ted for two different cases, and e-folding rates are mea- sured and labeled for all curves. Figure 5a illustrates a system with q > √α. The top panel shows results from secular equations, and the bottom panel shows the cor- responding N-body simulations. Each panel plots three curves: (1) the single planet case, in which the eccen- tricity of the inner planet damps at rate λ1, (2) a two- planet aligned mode with damping rate γ+, and (3) a two-planet anti-aligned mode (damping rate γ−). For q > √α, Figure 5a, eccentricities damp much faster in the anti-aligned mode than in the aligned mode, as pre- v v h h 6 (a) (b) Fig. 5. -- Eccentricity damping of systems in different secular modes found by integration of the secular equations (top panels) and direct N-body simulations (bottom panels). The plots show the eccentricity evolution of a hot-Jupiter (1 Jupiter-mass planet at 0.05 AU from a 1 solar-mass star) with a companion; γ+ and γ− represent the mode damping rates, which are e-folding times measured directly from the curves. Also plotted is a single hot-Jupiter subject to an artificial eccentricity damping with a rate λ1 = 7.90 × 10−7 yr−1. (a) A 0.8 Jupiter-mass companion is located at 0.2 AU (q > √α), with predicted mode damping rates γ+ = 1.5499× 10−7 yr−1, γ− = 6.3501× 10−7 yr−1 (Equation (16)). (b) A 0.3 Jupiter-mass companion is located at 0.2 AU (q < √α), with predicted γ+ = 6.4779 × 10−7 yr−1 and γ− = 1.4221 × 10−7 yr−1. dicted by Equation (16). A comparison between the top and bottom panel shows close agreement (within 2%) be- tween full-scale N-body simulation and integration of the approximate secular equations. Damping rates predicted by Equation (16) match the observed secular decay rates almost perfectly. Figure 5b shows a system with q < √α, for which the aligned mode damps faster than the anti- aligned mode. The faster damping rate of the aligned mode in the N-body simulation is within 0.5% of the pre- diction, but that of the slow anti-aligned mode, however, is ∼ 15% off. This discrepancy might be the result of unmodeled tidal perturbations to the inner body's semi- major axis. The smaller m2 of Figure 5b weakens the secular interaction, thereby emphasizing these tidal ef- fects. Equation (17). Figure 6b shows the corresponding plots for q < √α. The orbital elements undergo similar evo- lution, except that the aligned mode damps quickly and the system ends up in the anti-aligned mode. 2.3. Apsidal Circulation and Libration In Figure 6, the apsidal motion of the two orbits changes from libration of ∆ about 0◦ or 180◦ to circu- lation of ∆ through a full 360◦, and to libration again during the eccentricity damping. In order to understand what determines the apsidal state of the orbits, we plot the aligned and anti-aligned components of the eccen- tricities for Figure 6a in the top two panels of Figure 7. Recall that the total eccentricities of the orbits at any time can be obtained from the components as illustrated in Figure 4. The lower panels in the figure show the phase diagrams of both orbits (e cos(∆) versus e sin(∆)) at five different time indicated in the top two panels. The two orbits move along the phase curves, which them- selves change slowly over time. Two critical instants, labeled s− and s+, are circulation-libration separatrices, which represent the transitions of the apsidal state from anti-aligned libration to circulation (s−), and from circu- lation to aligned libration (s+). These two points divide the evolution curves into three regions. The different damping rates for the two modes are par- ticularly interesting, especially for well-separated nearly- decoupled orbits for which α and β are small. In this case, the 4q√αβ2 term under the square root of Equa- tion (16) is much smaller than the other term. As a result, if the eccentricity damping on one orbit is much faster than on the other (λ1 ≫ λ2), as in the case of tides, one mode damps rapidly at nearly the single-planet tidal rate λ1. The system evolves quickly into a single mode which decays substantially more slowly. Because of the different damping rates for the two modes, a system will evolve into a single mode even if it starts in a mixed state. Figure 6 shows the eccentric- ity and apsidal angle evolution of the systems depicted in Figure 5, but with initial conditions that lead to mixed states. Figure 6a shows the case of q > √α. Before 4 Myr, the system is in a mixed state, so both eccentric- ities, as well as their ratio, oscillate (cf. Figure 4). As the short-lived anti-aligned mode damps away, the orbits begin to librate around ∆ ≈ 0◦, the two eccentricities oscillate less and less, and in the end, the eccentricity ratio settles to the aligned mode ratio η+ predicted by +e+). −e− > ηs In region I (t < 106 yr), the anti-aligned components are stronger than the aligned ones for both orbits (e− > e+ and ηs In the phase plots, both the e1 and the e2 curves are closed and stay on the left side of the vertical e sin(∆) axis, indicating the libration of ∆ about 180◦. With the decrease in the amplitudes of all components, especially the faster damping of the anti- aligned ones, the curves move closer toward the origin, resulting in an increased libration width of ∆. The anti-aligned separatrix s− crossing occurs at t = 106 yr, when the two components for the outer orbit are equal (ηs +e+) and e2 might drop to zero, −e− = ηs 7 (a) (b) Fig. 6. -- Secular evolution of the same systems shown in Figure 5, but with different initial conditions so that the systems begin in mixed states. (a) For q > √α, the anti-aligned mode damps quickly at rate γ− from Figure 5a, and the system evolves to the aligned mode (∆ ≈ 0◦). (b) For q < √α, the aligned mode damps more rapidly and the system evolves to the anti-aligned mode (∆ ≈ 180◦). resulting in a phase curve for the orbit that is tangent to the vertical axis at the origin (s− curve on the inner phase plot of Figure 7). The phase curve for e1 at s− is a half- oval whose straight edge includes the origin. Accordingly, when e2 drops to zero, ∆ jumps from 90◦ to −90◦ for the largest possible full libration amplitude of 180◦. As the system moves past s−, the phase curves for both orbits enclose the origin and circulation results. The cir- culation region (II) is located between the two separa- trices (i.e., when 106 yr < t < 3.95 × 106 yr), where the anti-aligned component of the inner orbit is stronger than the aligned one (e− > e+), while it is weaker for the outer orbit (ηs +e+). With the continuous fast damping of the anti-aligned mode, the system crosses the aligned separatrix s+ at t = 3.95 × 106 yr, when the two components for the inner orbit are equal (e− = e+). The two separatrices occur at those times when each phase curve in Figure 7 touches the origin. in these coordinates can be visualized as circles whose radii and horizontal distances from the origin shrink at the different rates γ+ and γ−. Since the anti-aligned components initially dominate the aligned ones (region I), the e1 vector stays on the right side of the vertical axis, and the e2 vector on the left side in Figure 8. When the aligned components are parallel to the vertical axis, the angle between e1 and e2, ∆ > 90◦, is at minimum; thus the orbits librate about ∆ = 180◦ (anti-aligned li- bration). When the e2 circle moves to enclose the origin and the e1 circle is still confined in the first and fourth quadrants, the system reaches circulation region II. This geometry enables ∆ to cycle through a full 360◦ (Fig- ure 8). Last, when the anti-aligned components are suf- ficiently damped and both circles contain the origin, the system goes to aligned mode libration (region III). Now the maximum value of ∆ < 90◦ occurs when the two aligned components are parallel to the vertical axis, and the orbits librate about ∆ = 0◦. −e− < ηs Figure 9 shows the case of Figure 6b, where the two aligned components are initially stronger and the system starts in the aligned libration region III. The two orbits evolve to cross the aligned separatrix s+ into the circula- tion region II, and then pass the anti-aligned separatrix s− to reach the final anti-aligned libration region I. The equivalent of Figure 8 for this system would show two circles that initially encompass the origin; here the radii of the circles would shrink faster than the distances of their centers from the origin. −e− < ηs After s+, both phase curves are to the right of the vertical axis, indicating libration about the aligned mode (region III). The two anti-aligned components are both significantly damped and the system now has both e− < e+ and ηs The geometry of the orbits can also be illustrated with a component diagram similar to Figure 4, but in a frame rotating at the same rate as the anti-aligned mode (Fig- ure 8). In this rotating frame, the aligned component vectors always rotate clockwise because their precessions are slower than those of the anti-aligned ones. Evolution +e+. 8 Fig. 7. -- Evolution of the apsidal states during eccentricity damping. The top two panels show the time evolution of the eccentricity 2a2 and the anti-aligned mode damps fastest. The inner orbit has an aligned components for the system in Figure 6(a) for which m2 component e+ and an anti-aligned one e−, while ηs −e− are the components for the outer orbit. Two circulation-libration separatrices (s− and s+) divide the evolution curves into three parts: anti-aligned libration (region I), circulation (region II), and aligned libration (region III). The bottom panels show phase diagrams of the inner and outer orbits on the complex e exp(i∆) plane at the corresponding points indicated in the top panels. The shape of the phase curves depends on the relative strength of the two components for each orbit. Banana shapes result from the large difference between the two components: e− ≪ e+ at (a) and ηs 1a1 < m2 +e+ and ηs −e− ≪ ηs +e+ at (c). In conclusion, the apsidal state of a two-planet secular system depends on the sign of the simple product: P = (e+ − e−)(ηs +e+ + ηs −e−). Libration occurs when the same mode components are stronger for both orbits (P > 0), and circulation occurs when one mode is stronger for the inner orbit, but weaker for the outer one (P < 0). This result is in full agree- ment with a slightly more complicated formula given by Barnes & Greenberg (2006b). Eccentricity damping is effective in changing the apsi- dal state of the orbits because the two modes damp at different rates. Eccentricity excitation is equally capable of moving the two orbits across libration-circulation sep- aratrices. This can be easily visualized by running the plots in Figures 7 and 9 backward in time. For systems with m2 2a2 (see Figure 7), eccentricity excitation would eventually bring the orbits into anti-aligned libra- 1a1 < m2 tion (region I), while eccentricity damping brings them into aligned libration (region III). The opposite is true for systems with m2 2a2 (Figure 9). All mecha- nisms that change eccentricities slowly cause planetary systems to move toward apsidal libration. 1a1 > m2 2.4. Additional Apsidal Precessions In Sections 2.2 and 2.3, we have considered tides as an orbital circularization mechanism. Tides, however, also cause apsidal precession (see, e.g., Ragozzine & Wolf 2009): T ,p = T ,∗ = 15 2 15 2 m k2p(cid:18) R k2∗(cid:18) R∗ a(cid:19)5 m∗ a (cid:19)5 m m∗ n, n. (I) n i s e e1 e2 e2 n i s e e1 (II) e cos e cos (III) n i s e e2 e1 e cos Fig. 8. -- Eccentricity component diagrams for different regions in Figure 7. These diagrams are similar to Figure 4, but now shown in a frame rotating at the same rate as the anti-aligned mode so that the horizontal vectors are e− and η− e−. Here, the anti-aligned mode damps faster than the aligned mode (γ− > γ+) so that the circles move horizontally toward the origin faster than their radii shrink. The system starts with ∆ ≈ 0◦ in region I and evolves to ∆ ≈ 180◦ in region III. Here, T ,p and T ,∗ are the tidal precession rates due to planetary and stellar tides, respectively; k2p is the Love number of the planet; and k2∗ is that of the star. These expressions assume that tidal bulges are directly underneath the distant body (non-dissipative tides) but could adjusted to account for slight angular offsets in the tidal bulges (dissipative tides). For close-in exoplanets, orbital precession caused by GR effects is also important. To lowest order in eccen- tricity, the precession rate (e.g., Danby 1988) is the fol- lowing: GR = 3 a2n3 c2 , where c is the speed of light. In addition, the rotational bulges raised on the planet and its star also lead to orbital precession, with respec- tive rates (Ragozzine & Wolf 2009): R,p = R,∗ = Ω2 n2 n, k2p m 2 (cid:18) R 2 (cid:18) R∗ a(cid:19)5 m∗ a (cid:19)5 Ω2 n2 n. ∗ k2∗ Here, Ω and Ω∗ are the spin rates of the planet and the star, respectively. To account for these additional orbital precessions, we define a dimensionless quantity: κ = T ,p 1 + GR 1 + R,p 1 + R,∗ 1 1 + T ,∗ σ , which is the ratio of the sum of all additional precessions to the characteristic secular precession. These additional precessions add extra terms to Equation (1), which now reads as follows: j = + 1 nja2 j ej ∂Rj ∂ej + κσ. Since κ is independent of e for small eccentricities, the ex- tra terms do not change the form of Equation (5), and so all discussion of the general secular modes still holds. In 9 particular, the system still has aligned and anti-aligned modes and the two modes damp separately. The mode frequencies, damping rates, and eccentricity ratios, how- ever, need to be revised. Now the diagonal terms of the coefficient matrix Ajk should be adjusted to the follow- ing: A = σ(cid:26) q + κ + iξ1 −√αβ √α(1 + α2κ) + iξ2 (cid:27) , −qβ gs ± = which gives the new mode frequencies and eccentricity ratios: 1 2 σ(cid:26)(q + κ) + √α(1 + α2κ) ∓q[q + κ − √α(1 + α2κ)]2 + 4 q√αβ2(cid:27) , q + κ − √α(1 + α2κ) ±p[q + κ − √α(1 + α2κ)]2 + 4 q√αβ2 σ (ξ1 − ξ2)2, q√αβ2 g± = gs ηs ± = 2 qβ , (18) (20) (19) {[q + κ − √α(1 + α2κ)]2 + 4q√αβ2}3/2 √α(1 + α2κ) − (q + κ) (λ1 − λ2)# , (21) (22) 1 ± ± 2 "λ1 + λ2 ± ±(1 ± i γ± = η± = ηs ξ1 − ξ2 p[q + κ − √α(1 + α2κ)]2 + 4q√αβ2 p[q + κ − √α(1 + α2κ)]2 + 4 q√αβ2) . stitution (q ± √α) → [q + κ ± √α(1 + α2κ)]. Note that these equations can be obtained from Equa- tions (6), (7), (15), (16), and (17) with the simple sub- These additional apsidal precessions increase both sec- ular rates gs ± (Equation 18) since they cause the orbits to precess in the same direction as the secular interaction does. The mode eccentricity ratio ηs + (Equation 19) in- creases significantly with any source of precession that favors the inner orbit, indicating that it is more difficult to force the eccentricity of a rapidly precessing inner orbit in the aligned mode. The ratio ηs −, however, decreases slightly; increasing the precession of the inner orbit in the anti-aligned mode actually strengthens secular cou- pling. As for the mode damping rates (Equation 21), additional precession decreases the aligned mode damp- ing rate, but increases that of the anti-aligned mode. It also decreases ǫ, the deviation angle of the mode apsidal lines from perfect alignment (Equation 22). 3. APPLICATIONS TO THE OBSERVED SYSTEMS We now apply the aforementioned theory to close-in exoplanets, beginning with known two-planet systems. We then proceed to systems in which there is linear trend in the star's radial velocity (RV) that might signal the presence of an outer companion and finally we consider systems with no hint of a companion. 3.1. Two-planet Systems There are more than 30 multi-planet systems that host one or more close-in planets with the orbital pe- riod Porb . 20 days. Out of these, many systems in- cluding Gliese 876, 55 Cnc, and υ And have three or more planets, which makes apsidal analysis more com- plicated (e.g., Barnes & Greenberg 2006a). Moreover, most Kepler-detected planets have unknown eccentric- ities, and are not the best candidates for our analysis. After removing multiple planet systems and those with poorly determined eccentricities, we are left with eight two-planet systems, with which we can test the linear tidal model. v v v v v v D v D v D v 10 Fig. 9. -- Evolution of the apsidal state during eccentricity damping. Similar to Figure 7, but using data from Figure 6(b) for which 2a2. The aligned mode damps fastest and the system moves from aligned libration to circulation, and finally to anti-aligned 1a1 > m2 m2 libration. First, we compute eccentricity damping timescales for the inner planets (τe), and compare τe with the stellar age (τAge) to determine whether an orbit has had sufficient time to circularize. For all systems, we adopt conventional planetary and stellar tidal quality factors Qp = 105 and Q∗ = 106, respectively (see, e.g., Jackson et al. 2008; Matsumura et al. 2010, and refer- ences therein). We convert these to modified tidal qual- ity factors as Q′ = 1.5Q/k2. For giant planets in our solar system (Jupiter, Saturn, Uranus, and Neptune), the measured gravitational moments agree well with an n ∼ 1 polytrope (Hubbard 1974; Bobrov et al. 1978) which corresponds to the Love number k2 = 0.52 (Motz 1952). For most stars, on the other hand, the n = 3 polytrope is a good approximation (e.g., Horedt 2004), which yields k2 = 0.028 (Motz 1952). We finally ob- ∗ = 5.36 × 107. We esti- tain Q′ mate the eccentricity damping timescale τe by integrat- ing a set of tidal equations based on the equilibrium tide model from the measured planetary eccentricity down to e = 10−4, and further assume that the tidal quality fac- p = 2.88 × 105 and Q′ tors evolve proportional to the inverse of the mean mo- tion Q ∝ 1/n1 (see, e.g., Matsumura et al. 2010). Some of our results appear in the final column of Table 1. Note that these simulation results compare favorably to the simpler approximation of Equation (12) when one prop- erly accounts for the multiple e-folding times needed to damp eccentricities to e = 10−4. Only 2 of the 8 two- planet systems (HD 187123 and HAT-P-13) have short tidal circularization times compared with the stellar ages; physical and orbital parameters of these two systems can be found at the top of Table 1. Since τe << τAge, the systems are likely to have been significantly modified by tides. Next, we check the strength of secular interactions for the two systems. In the HD 187123 system, the inner planet is at ∼ 0.04 AU while the outer one is at ∼ 5 AU. With this configuration, the inner planet is strongly bound to the star and its interaction with the outer planet is weak. The inner and outer planets of HAT- P-13, however, are more closely spaced at ∼ 0.04 and ∼ 1.2 AU, respectively. Furthermore, since m2 >> m1 11 τe (Gyr) ∼ 0.125 ∼ 0.3 −0 0.013 ± 0.0041 0.662 ± 0.0054 0.010 ± 0.00593 0.25 ± 0.0334 0.16 ± 0.019 0.02+0.042 0.12 ± 0.06 0.04 ± 0.0012 0.016 ± 0.01 0.23 ± 0.016 0.11 ± 0.044 0.16 ± 0.061 0.15 ± 0.081 0.13 ± 0.072 ω (◦) 210+27 −36 175.3 ± 0.35 24.5 240 ± 18.6 351 ± 1.2 334 m∗ (m⊙) 1.2+0.05 −0.1 1.2+0.05 −0.1 1.04+0.026 −0.024 1.04+0.026 −0.024 0.45+0.014 −0.012 0.92+0.046 −0.024 0.82 ± 0.033 1.01 ± 0.07 ∼ 1.2 ± 0.1 0.95 ± 0.042 1.13 ± 0.035 1.18+0.043 −0.07 1.40 ± 0.096 1.2 ± 0 Age (Gyr) 5+2.5 −0.8 5+2.5 −0.8 5.33 5.33 6+4 −5 4.18 9+3 −4.9 6.7+6.9 −4.5 7.6 ± 1.2 10.2 ± 2.5 4 ± 1 3.8+1.5 −0.5 2.4 ± 0.4 9.56 Planet HAT-P-13 b HAT-P-13 c HD 187123 b HD 187123 c GJ 436 b BD -10 3166 b HAT-P-26 b WASP-34 b HD 149143 b HAT-P-21 b HAT-P-23 b HAT-P-32 b HAT-P-33 b HD 88133 b Properties of Planets and Stars Discussed in This Paper TABLE 1 a (AU) e mp (mJ ) 0.85 ± 0.0354 14.3 ± 0.691 0.51 ± 0.0173 1.9 ± 0.152 Rp (RJ ) 1.28 ± 0.079 1∗ 0.0427 ± 0.000875 1.23 ± 0.0251 0.0421 ± 0.000702 4.8 ± 0.367 0.3767+0.0082 −0.0092 1∗ 0.565 ± 0.052 1.22+0.11 −0.08 1∗ 0.073 ± 0.00318 0.43 ± 0.0174 0.059 ± 0.00718 0.58 ± 0.0285 1.33 ± 0.0784 4.1 ± 0.173 2.1 ± 0.122 1.0 ± 0.169 0.8 ± 0.117 0.30 ± 0.0270 ∼ 2 ∼ 0.42 ∼ 2.6 ∼ 0.7 ∼ 2.75 ∼ 6 ∼ 0.005 ∼ 0.005 ∼ 0.06 ∼ 0.38 References. -- Data for HAT-P-13 system are from Winn et al. (2010); HD 187123 from Wright et al. (2009); HAT-P-21 and HAT-P-23 from Bakos et al. (2011); HAT-P-26 Hartman et al. (2011a); HAT-P-32 and HAT-P-33 from Hartman et al. (2011b); HD 88133 and BD - 10 3166 from Butler et al. (2006); GJ 436 from Maness et al. (2007); WASP-34 from Smalley et al. (2011); HD 149143 from Fischer et al. (2006). The planetary mass mp assumes the lower limit mp sin i for listed planets without measured planetary radius Rp. The circularization timescale τe is estimated by integrating a set of tidal equations (e.g., Matsumura et al. 2010) from the tabulated eccentricity to e = 10−4. An asterisk signifies an assumed value. 0.0287 ± 0.000479 0.0438 ± 0.000730 0.0479 ± 0.000798 0.052 ± 0.00120 0.053 ± 0.00147 0.0495 ± 0.000825 0.0232 ± 0.000387 0.0344 ± 0.000574 0.050 ± 0.00115 0.0472 ± 0.000786 1.024 ± 0.092 1.368 ± 0.09 2.037 ± 0.099 1.83 ± 0.29 100 ± 165 320 ± 20.9 309 ± 3 120 ± 25 50 ± 29 100 ± 119 349 1∗ 0 (Rauch & Hamilton 2002). Since HAT-P-13c's large ec- centricity of ∼ 0.66 violates the assumption of the linear secular theory, it is understandable that the discrepancy between the two integrations is ∼ 33% (Figure 10). It appears that the system librates with a large amplitude about ∆ = 0◦ so that the system is not far from the aligned separatrix s+ (Figure 7). Here, we do not take ac- count of any additional apsidal precessions. By taking ac- count of the GR effect, we find that the results stay sim- ilar -- the system librates with a large amplitude. How- ever, by considering all of the additional apsidal preces- sions, the system appears to circulate. Thus, HAT-P-13 is likely to be yet another example of a multi-planet sys- tem being near a secular separatrix (Barnes & Greenberg 2006b). The near-separatrix state of this system is some- what surprising given that the tidal decay time is a tiny fraction of the stellar age (Table 1) and so the system should have long ago damped to the apsidally-locked state. Perhaps, given the uncertainty in Q′, our esti- mation of τe is off by a large factor. Alternatively, there might be a third as-yet-undiscovered planet affecting this secular system. We next compare the current state of the system with the expectation from tidal-damping theory as developed in Section 2.2. We assume that the parameters of the inner planet's orbit are known, assume further that one mode has fully damped away (despite the contrary evi- dence of Figure 10), and proceed to predict parameters of the outer planet. We begin by asking which mode is favored, which depends on the damping rates given by Equation (16), or Equation (21) when other apsidal precession effects are important. When a system is locked into one of the secular eigen- modes, the outer and inner orbits have a predictable ec- centricity ratio η± = (e2/e1)± (see Equations (17) and (22)). In Figure 11, we show contour plots of the pre- dicted eccentricity of the outer planet e2 for the more slowly damped eigen-mode in the parameter space (q, α). The dashed curve divides the space into a region in which the slow anti-aligned mode persists (top left, red Fig. 10. -- Current apsidal state of the HAT-P-13 system. We show the orbits of both planets on the e exp(i∆) plane, similar to the lower plots in Figures 7 and 9. The orange curves represent the solution of secular equations, and the blue curves are obtained from an N-body simulation. The orbits librate about ∆ = 0◦, with a libration amplitude ∼ 43◦ predicted by secular theory and ∼ 57◦ measured from the N-body simulation. The narrowness of the outer planet's arc in both cases is due to the fact that m2 >> m1. and e2 = 0.66 is large (Table 1), the secular forcing of the inner planet is substantial. Thus, out of the eight systems, only HAT-P-13 has both strong secular interac- tions and a short tidal damping time; it is, accordingly, the best test case for our theory. We now utilize the linear secular theory without tides developed in Section 2.1, and show that the model pre- dicts the current apsidal state of HAT-P-13 reasonably well. In Figure 10, we compare the apsidal state of HAT- P-13 estimated by secular theory with that obtained from a direct N-body simulation done by the HNBody code 12 0.8 0.6 2 a / 1 a = 0.4 1 . 0 0 0 0 0 . . . 5 0 3 0 2 0 0 . 1 0 5 0 . 0 n o i t a r b i L d e n g i l a - i t n A 0.2 1 . 0 0 0.0 0.01 0 0 0 . . . 5 3 . 0 5 0 . 0 1 0 0 0 2 0 0.10 1.00 q = m2/m1 0.8 0.6 2 a / 1 a = 0.4 HAT-P-13 1 . 0 0 0 . 5 0 0 0 . . 3 0 2 0 5 0 . 0 0 . 1 0 d e n g i l a - i t n A n o i t a r b i L Aligned Libration 0 . 0 5 0.05 0.10 0.20 0.30 0.50 1.00 0.2 0.10 0.20 0.30 0.50 10.00 1.00 100.00 0.0 0.01 (a) without additional precessions (b) with General Relativity 0.10 1.00 q = m2/m1 10.00 100.00 HAT-P-13 Aligned Libration 0.10 0.20 0.30 0.50 1.00 HAT-P-13 Aligned Libration 0.10 0.20 0.30 0.50 1.00 0 3 . 0 0 2 . 0 0 1 . 0 5 0 . 0 0.8 0.6 1 d . 0 e 0 n g i l a - i t n A 0 5 . 0 n o i t a r b i L 2 a / 1 a = 0.4 0.10 0.20 0.30 0.50 1.00 0.05 0.2 0.0 0.01 HAT-P-13 Aligned Libration 0.10 0.20 0.30 0.50 1.00 0 3 . 0 0 2 . 0 0 1 . 0 5 0 . 0 0.8 0.6 0 d 0 e . 1 n g i l a - i t n A 0 5 . 0 n o i t a r b i L 2 a / 1 a = 0.4 0.10 0.20 0.30 0.50 1.00 0.05 0.2 0.0 0.01 0.10 1.00 q = m2/m1 10.00 100.00 0.10 1.00 q = m2/m1 10.00 100.00 (c) with GR and planetary tides (d) with all additional precessions Fig. 11. -- Orbital states of possible companions for HAT-P-13b (a) with no additional apsidal precessions, (b) with GR precession, (c) with GR and planetary tidal precessions, and (d) with different assumptions about the orbital precession terms from Section 2.4. Each point in these q - α plots represents an outer companion with corresponding mass and semi-major axis. The dashed curve divides the plane into regions in which the aligned mode damps faster (top left) and the anti-aligned mode damps faster (bottom right). Furthermore, in the shaded regions, either the aligned mode (blue shading) or the anti-aligned mode (red shading) can survive tidal dissipation and last longer than the age of the system (5 Gyr). Last, the solid contour lines represent the eccentricity of the outer planet assuming that it is in the long-lived mode. The location of HAT-P-13c is marked by a white symbol with error bars. area) and a region in which the slow aligned mode sur- vives (bottom right, blue area). In these shaded areas, the lifetime of the slower mode is longer than the age of the system. Conversely, in the white central area, both modes should have already damped away and both or- bits would be circular by now. Given the inner planet's non-zero e1 = 0.013 (Table 1) and that τe << τAge, we expect the outer planet to be in one of the shaded re- gions. Furthermore, since e2 > 1 correspond to unbound orbits, these parts of the shaded areas in Figure 11 are also off limits. The boundaries of these colored areas are determined by equating the age of the system (5 Gyr for HAT-P-13) to the circularization time τe in Table 1. An older system age τAge and/or faster damping timescale τe would ex- pand the white area outward away from the dashed line. If tides have not been active over the full stellar age, as is possible for a recent resonance crossing or a planet-planet scattering event, the white area would shrink inward to- ward the dashed line. The effect of the additional apsidal precessions is sub- stantial and can be quantified by comparing panels (a)- (d) in Figure 11. The panels (a)-(d) show the estimates from the linear secular theory (a) without additional pre- cessions, (b) with GR precession, (c) with GR and plan- etary tidal precessions, and (d) with all apsidal preces- sions, respectively. By comparing panel (a) with panels (b) and (c), we find that GR and planetary tidal preces- sions significantly change the q−α plot. Conversely, from the comparison of panels (c) and (d), we can tell that the other precessions have negligible effects on HAT-P-13. In Figure 12, we compare the GR precession rate with each of other precession rates for all of the systems listed in Table 1. For these close-in systems, we find that a a a a 13 the eccentricity of the outer planet exceeds one, while the observed value is actually 0.662 ± 0.054 (Table 1). It is clear that e2 ≥ 1 is an unphysical result, which might be attributable to: (1) using linear secular theory despite large eccentricities, and (2) assuming that one mode dominates despite the evidence from Figure 10. Although the quantitative agreement is not very good, the figure does predict apsidal alignment and a large ec- centricity for the outer planet. We accordingly conclude that the secular perturbations between the two planets in the HAT-P-13 system might be responsible for the non-zero eccentricity of the inner planet. This section shows the power and pitfalls of our method. If estimates for the stellar ages and tidal damp- ing timescales are accurate, we can determine whether a given system should currently be near a single eigen- mode. The HAT-P-13 system with a stellar age 40 times longer than the estimated damping time should have had ample time to reach such a state, and yet Figure 10 shows that it has not. Perhaps the time estimates are inaccu- rate or perhaps there was a recent disruptive event in the system. In either case, this suggests that a certain amount of caution is warranted when proceeding to inves- tigate single-planet systems. With this in mind, in the following two sections, we study single-planet systems with and without observed linear trends in the stellar radical velocity that might be indicative of a companion. We investigate whether the observed non-zero eccentrici- ties could be explained by unseen potential companions. 3.2. Single-planet Systems with a Hint of a Companion As shown in Sections 2.2 and 2.3, a two-planet system should have evolved into either an aligned or an anti- aligned apsidally locked state when the tidal dissipation is strong enough. Equations (19) and (21) thus provide a single constraint on the three parameters of the unknown outer companion: the mass ratio q, the semi-major axis ratio α, and the eccentricity ratio of the two planets. Therefore, we can predict a range of possible companions that might force a non-zero eccentricity on an observed close-in planet. We illustrate our method with several examples here and in the next section. There are 16 single, close-in planet systems with Porb ≤ 20 days and non-zero eccentricities that have an observed linear trend in the stellar RV, which indicates the pos- sible existence of a companion on a more distant orbit. For these systems, we can place a unique constraint on the potential companion. For simplicity, we exclude the two systems that have a large projected stellar obliquity (HAT-P-11 and WASP-8), and compare the estimated τe with the stellar age τAge for each of the remaining systems, as described earlier. We find that 5 of 14 sys- tems (GJ 436, BD -10 3166, HAT-P-26, WASP-34, and HD 149143) have τe < τAge. We list parameters for these systems in the middle section of Table 1. Because the properties of the putative companions are unknown, we cannot test for a strong secular interaction as in the pre- vious section and so we investigate all five systems. Figure 13(a) is similar to Figure 11(d), but for the planetary system around GJ 436. The thick, dashed curve is an upper limit to the outer planet's mass estimated from the observed linear trend. Here, we simply assume that the minimum mass of a poten- tial outer planet is expressed as m2 = a2 2 alt/G, where Fig. 12. -- Comparisons of apsidal precession rates for planetary systems listed in Table 1. Blue and orange symbols represent the ratios of rotational and GR precessions (R/GR) and tidal and GR precessions (T /GR), respectively. Circles and triangles cor- respond to precessions due to stellar and planetary deformations, respectively. For all of the close-in systems listed here, either GR or planetary tidal precession dominates the additional apsidal pre- cession. either GR or planetary tidal precession dominates the additional apsidal precession, while the effects of stellar tidal and rotational precessions tend to be much smaller. For HAT-P-13, the precession rate due to tidal deforma- tion of the inner planet is a factor of a few larger than the GR precession rate, while the other precessions are much smaller than GR. As discussed in Section 2.4, adding extra preces- sions diminishes the anti-aligned area (red shading) due to faster damping rates but significantly expands the aligned area (blue shading) in accordance with Equa- tion (21). Notice that, as expected, the lower left quad- rant of the plot (i.e., small q and small α) experiences the greatest changes from panel (a) to panels (b)-(d). Con- versely, the changes to the shading of the other three quadrants are relatively minor. The dashed lines in Figures 11(a) and (b)-(d) are given by q = √α and q + κ = √α(1 + α2κ), respectively, where κ includes only the corresponding terms in (b)-(d). These expres- sions simply compare the real diagonal elements of the respective A matrices. The anti-aligned mode damps most quickly if the inner planet precesses faster while the aligned mode damps first if the outer planet precesses faster. Along this dashed line, the difference between the mode precession rates (gs −) is minimized and the mode damping rates γ± = (λ1 + λ2)/2 are identical. Last, note that including additional apsidal precessions significantly changes the contours for the outer planet's eccentricity. In the region of aligned libration, the ec- centricity contours are moved upward by the inclusion of these precessions, indicating that a more eccentric outer planet is required to maintain the mode, for a given α and q. In the anti-aligned region, the contours move to the left, indicating that a lower eccentricity on the outer planet is needed to preserve the mode. The reasons for these changes were discussed in Section 2.4. + − gs The actual location of the outer planet HAT-P-13c is marked in all of the panels with white error bars. The planet resides well within the more slowly damped aligned mode region, as expected from linear secular the- ory (see also Figure 10). Figures 11(b)-(d) suggest that 14 0.8 0.6 2 a / 1 a = 0.4 0.2 0.0 0.01 R V i L m i t : 1 m / s 0 . 1 1.0 6 . 0 4 . 0 3 . 0 2 . 0 GJ436 0.2 0.8 0.6 1 . 6 . 0 4 . 0 0 3 . 0 2 . 0 1 . 0 WASP-34 R V i L m i t : 1 m s / 0.6 0.3 0.4 2 a / 1 a = 0.4 0.3 0.4 0.6 1.0 0.2 0.1 1.0 Linear Trend 10.00 100.00 0.2 0.0 0.01 0.3 0.4 0.6 1.0 0.2 Linear Trend 0.10 1.00 q = m2/m1 10.00 100.00 (b) 0.10 1.00 q = m2/m1 (a) Fig. 13. -- Eccentricities of possible companions for hot Jupiters (a) GJ 436, and (b) WASP-34 including all apsidal precession effects as in Figure 11(d). In the white area, both modes should damp away within the age of the system of 6 and 6.7 Gyr, respectively. The central stars of these systems each have an observed linear trend in their radial velocities, 1.36 and 55 m s−1 yr−1, respectively, that might be indicative of second planets. The thick, dashed curves represent these constraints. For WASP-34, the region to the right of the solid portion of the curve represents where a potential candidate is expected to exist (Smalley et al. 2011). The dotted curves represent the RV observation limit of 1 m s−1; only planets to the right of the nearly vertical solid part of this curve are detectable with current technology in a 1 yr observational period. alt = 1.36 ± 0.4 m s−1yr−1 is an observed linear trend (Maness et al. 2007).3 The dotted curve indicates an observation limit for the RV method. The solid por- tion of this curve is plotted as a reference, and it shows the limit estimated for a 1 yr observation period. To the right of this solid curve, a full orbit of a hypothet- ical outer planet is observable within a year. To plot this, we express the mass of a potential outer planet as angle, and assume the RV limit of v∗ sin i = 1 m s−1. m2 sin i = m∗ v∗ sin i/pGm∗/a2, where i is the viewing The observed planet GJ 436 b is about a Neptune- mass object (m1 sin i = 0.073 mJ ∼ 1.35 mN ) which is 0.0287 AU away from the central star (orbital period ∼ 2.6 days), and has an orbital eccentricity of 0.16±0.019 (Table 1). If this eccentricity is due to another planet and the system has damped to an eigen-mode, then eccen- tricity contours in Figure 13(a) show that broad aligned and anti-aligned regions are allowed for a potential com- panion, except for small (m2 . 0.15 m1 ∼ 0.2 mN ) and/or distant (a2 & 6.3 a1 ∼ 0.18 AU) planets. Fur- thermore, the plotted RV limit indicates that nearly all hypothetical outer planets which could be responsible for the high eccentricity of GJ 436 b should be observ- able within a year. The curve representing the maximum linear trend, however, is far below the e2 = 1 contour, implying that this potential planet cannot be responsi- ble for the current eccentricity of the observed planet; given its great distance, the secular interactions are sim- ply too weak. This result is consistent with the compari- son of the secular and tidal circularization timescales by Matsumura et al. (2008). Since the system has a tidal dissipation timescale (τe ∼ 2 Gyr) comparable to the stel- 3 This long-term trend has not been confirmed by HARPS (Bonfils et al. 2013). lar age (τAge ∼ 6+4 −5 Gyr), a non-zero eccentricity of this planet might also be explained within the uncertainties of the stellar age. Another example is shown in Figure 13(b) for WASP- 34, which has an observed planet of m1 sin i = 0.58 mJ, a1 = 0.052 AU, and e1 = 0.038 ± 0.012 (see Table 1). Again, most of the parameter space of the q − α plane is available for a possible secular companion, except for small (m2 . 7.4 mE) and/or distant (a2 & 0.74 AU) planets. The solid portion of the RV limit again indi- cates that such a companion should be observable within a year. The system has an observed linear trend of 55 ± 4 m s−1yr−1 (Smalley et al. 2011). Since the long- term trend has not reached its maxima or minima, the orbital period of the outer body has to be greater than twice the RV data baseline. The solid portion of the linear trend corresponds to this limit of a2 & 1.2 AU and m2 & 0.45 mJ (Smalley et al. 2011). A yet-to-be- observed companion should lie in the triangular area right of the solid part of the linear trend and below the observable region of the RV Limit. It is clear that such a region does not have any overlap with the critical e2 = 1 contour. However, they lie relatively close to each other so that the uncertainties in both observations and the secular model could bring them closer to have an over- lap. Our model predicts that the companion planet re- sponsible for both this trend and the eccentricity of the WASP-34 b would have a very high eccentricity. If no companion is present, how do we explain WASP- 34? For this system, although the estimated eccentricity damping time (τe ∼ 700 Myr) is short compared with the stellar age (τAge ∼ 6.7+6.9 −4.5 Gyr), the eccentric orbit model gives only a slightly better fit than the circular one (Smalley et al. 2011). Thus, the inner orbit might well be circular. Alternatively, if the orbit is truly eccentric, we a a would need to assume about an order of magnitude less efficient tidal dissipation to explain this system. Last, the system could also have undergone some dynamical event lately which changed the original eccentricities. The other systems with a linear trend (BD -10 3166, HD 149143, and HAT-P-26) show a similar result to GJ 436 (Figure 13(a)), and thus a potential planet is too far to force the eccentricity of the inner planet to its current value. It is interesting that the observed eccen- tricities are low and consistent with zero for BD -10 3166 and HD 149143 (Butler et al. 2000; Fischer et al. 2006), and poorly constrained for HAT-P-26 (Hartman et al. 2011a). BD -10 3166 has a short tidal dissipation timescale (τe ∼ 420 Myr) compared with the stellar age (τAge ∼ 4.18 Gyr), so the circular orbit assump- tion makes sense. HD 149143 and HAT-P-26 have rel- atively long dissipation timescales (τe ∼ 2.75 Gyr and ∼ 2.6 Gyr, respectively) compared with the stellar ages (τAge ∼ 7.6 ± 1.2 Gyr and ∼ 9+3 −4.9 Gyr, respectively). Thus, uncertainties in the age estimates and/or in the tidal dissipation rates could allow close-in planets to maintain their eccentricities without assistance. In summary, we have found no single-planet systems, where the none-zero eccentricities could be explained by perturbations from hypothetical planets corresponding to the observed linear trends. However, their orbits could be circular (WASP-34, BD -10 3166, HD 149143, and HAT-P-26), or the eccentric orbit could be ex- plained within the uncertainties in the estimated stellar age (GJ 436). 3.3. Single-planet Systems with no Hint of a Companion Given that there are many single, close-in planet sys- tems without linear trends, we focus on planets whose non-zero eccentricities are hardest to explain -- the closest-in exoplanets. There are eight single-planet sys- tems with an orbital period Porb ≤ 5 days, eccentricity ≥ 0.1, and a small or unknown stellar obliquity. Compar- ing the tidal timescale to the stellar age, we find that five of eight such systems have τe < τAge (HAT-P-21, HAT- P-23, HAT-P-32, HAT-P-33, and HD 88133, see Table 1). All of these are hot Jupiter systems. For the remaining three systems, the stellar ages of KOI-254 and Kepler-15 are unknown, while GJ 674 has a very long τe > 10 Gyr compared with the stellar age (∼ 0.55 Gyr). Figure 14(a) is similar to Figure 11(d), but for the plan- etary system around HD 88133. The planet is 0.0472 AU away from the central star with an orbital period of 3.4 days and an eccentricity of 0.13 ± 0.072 (see Ta- ble 1). Since the circularization time is estimated to be very short (τe ∼ 380 Myr) compared with the stel- lar age (τAge ∼ 9.56 Gyr), a moderately high eccentric- ity of this planet is surprising. Figure 14(a) excludes the entire anti-aligned region for a potential compan- ion. The figure also excludes most companions with small mass (m2 . 0.03 mJ) and/or long orbital period (a2 & 0.295 AU). This is understandable because these planets would have weak secular interactions with the inner planet. Thus, a potential companion is expected to be in the aligned region, massive, and close to the star. As the RV observation limit shows, such massive planets in the aligned libration region would be observ- able within a year, although no such planet has been 15 found. HD 88133 has a low stellar jitter (∼ 3.2 m s−1), and the eccentric orbit assumption works only slightly better than the circular one (Fischer et al. 2005). Thus, unless the tidal quality factor for this system is very dif- ferent from what we have assumed here, we argue that the true eccentricity of HD 88133 b is actually near zero. Follow-up observations would better constrain the eccen- tricity and the existence or absence of a potential com- panion for this system. The case for HAT-P-21 is shown in Figure 14(b). The planet is 0.0495 AU away from the central star (orbital period ∼ 4.1 days), and has an orbital eccentricity of 0.23 ± 0.016 (Table 1). Figure 14(b) allows a very broad parameter space for a possible secular companion; the white zone indicating efficient eccentricity damping is nearly absent. However, this undercuts our assumption that the system has had time to damp into a pure eigen- mode and, accordingly, the eccentricity contours are not reliable. If we proceed with the dubious assumption of a single mode, the figure does not allow most companions with m2 . 0.82 mJ and/or a2 & 0.495 AU. Furthermore, the 1 ms−1 RV limit observationally precludes almost any planet that can be significantly coupled to HAT-P-21 b. Accordingly, we seek another explanation for the eccen- tricity of this system; it can be naturally explained if tidal dissipation were just slightly less efficient than we have assumed, since the estimated tidal circularization time is relatively long (τe ∼ 2.6 Gyr) compared with its stellar age (τAge ∼ 9+3 The other systems (HAT-P-23, HAT-P-32, and HAT- P-33) have a similar trend to HD 88133 (Figure 14(a)), with the entire anti-aligned region being excluded for a potential companion. Since all of their circularization times are more than 1-2 orders of magnitude shorter than the estimated stellar ages, the moderately-high ec- centricities (e1 > 0.1) of these planets need to be ex- plained. From figures similar to Figure 14, we find that potential companions for these systems tend to be more massive than the observed planets and thus are likely to be observable by the RV method. However, no com- panions have been found. It is interesting that all of these systems have high stellar jitters (Bakos et al. 2011; Hartman et al. 2011b). Although this could mean that potential companion planets are difficult to observe, high jitters also lead to poorly-constrained orbital eccentric- ities. All of these planets can also be fit well with the circular orbit model. Our model suggests that the circu- lar orbits are probably the most likely solution. Future observations that yield a more accurate solution for the eccentricity are needed. −4.9 Gyr). For our analysis of exoplanetary systems in this sec- tion, we did not explicitly take account of the effects of uncertainties in orbital or stellar parameters. In par- ticular, errors in stellar ages and eccentricities are often large and might change our results significantly. We have tested such effects for all of the systems we discussed in this section, and found that our conclusions will not change within the currently estimated uncertainties in parameters. Also, the assumption of an apsidal lock that we made in Sections 3.2 and 3.3 might be too strong; in- stead, it is possible that not enough time has elapsed for complete damping of one secular mode. In this case, for a close-in planet with known mass, semi-major axis, and 16 0.8 0.6 2 a / 1 a = 0.4 0.2 0.0 0.01 6 . 0 0 . 1 2 . 0 4 . 0 3 . 0 1.0 0.6 0.3 0.4 HD88133 0.2 1.0 0.10 1.00 q = m2/m1 10.00 100.00 0.8 0.6 2 a / 1 a = 0.4 0.2 0.0 0.01 0 . 1 6 . 0 4 . 0 3 . 0 0 . 1 6 . 0 0.3 0.4 HAT-P-21 0.3 0.4 0 . 6 1 . 0 0.10 1.00 q = m2/m1 10.00 100.00 (a) (b) Fig. 14. -- Eccentricities of possible companions for hot Jupiters (a) HD 88133, and (b) HAT-P-21 including all apsidal precession effects as in Figure 11(d). In the white area, both modes should damp away within the age of the system of 9.56 and 10.2 ± 2.5 Gyr, respectively. Because the inner HAT-P-21 planet has a long damping time of about 6 Gyr, all eccentricities can be sustained and almost no white area is visible. The dotted curves represent the RV observation limit of 1 m s−1, and the solid portion indicates a 1 yr observing period. All potential HD 88133 outer planets reside in the aligned zone are detectable. Aligned and anti-aligned solutions exist for HAT-P-21, and nearly all outer planets are detectable. eccentricity, the constraint on m2, a2, and e2 is approxi- mate rather than exact. 4. DISCUSSIONS AND CONCLUSIONS The eccentric orbits of single, close-in planets are gen- erally circularized on timescales shorter than the stellar ages. Close-in planets in multiple-planet systems have much longer tidal circularization timescales and thus are able to maintain eccentric orbits for the stellar ages or longer. Given this difference in tidal circularization times, we might expect a difference in the eccentricity distributions of close-in planets with and without known companions. The eccentricities forced by secular pertur- bations from an outer planet are typically small, however, and so it is perhaps not surprising that no such difference has yet been observed. In this paper, we have explored the possibility that the non-zero eccentricities of close-in planets are due to ob- served or hypothetical planetary companions. We have provided an intuitive interpretation of a simple secular evolution model of a coplanar two-planet system that in- cludes both the effect of the orbital circularization (Sec- tion 2.2) and apsidal precessions due to GR corrections as well as tidal and rotational deformations (Section 2.4). We have tested our model by comparing the evolution of apsidal states and orbital eccentricities with N-body simulations, and found that the agreement between the model and the simulations is very good. We have also ap- plied our model to all of the relevant two-planet systems (Section 3.1), as well as single-planet systems with and without a long-term trend in the RV to indicate a pos- sible second planet (Sections 3.2 and 3.3, respectively). The following is a summary of our main results. 1. In the lowest-order secular theory, the evolution of non-dissipative two-planet systems is described by a linear combination of two modes characterized by pericenter alignment and anti-alignment. Ec- centricity damping slightly shifts the two normal modes from perfect symmetry, which speeds up the precession rate of the aligned mode and slows that of the anti-aligned mode (see Section 2.2). 2. Eccentricity damping affects the two modes at dif- ferent rates. Accordingly, the apsidal state of a two-planet system transitions between libration and circulation, and eventually is locked to either an aligned or anti-aligned state (see Section 2.3). The eccentricity of both planets subsequently de- cays at a very slow rate. 3. GR, tidal, and rotational effects increase the pre- cession rates of both aligned and anti-aligned modes. As a result, they decrease the aligned-mode damping rate, and increase the anti-aligned mode damping rate (see Section 2.4). of confirm previous 2002; Mardling We results studies (e.g., Wu & Goldreich 2007; Greenberg & Van Laerhoven 2011) and show that close-in planets in multiple-planet systems can maintain non-zero orbital eccentricities substantially longer than can single ones. We find, however, that there are currently few two- or one-planet systems that show signs of secular interactions that are strong enough to significantly slow tidal circularization. In Section 3.1, we find that only one out of eight systems (HAT-P-13) shows tentative evidence that secular interactions are slowing orbital circularization. In Section 3.2, we apply our model to 14 single-planet systems with a linear trend and find that 5 of 14 have τe < τAge. Our secular model predicts that none of their eccentricities is likely to be affected by hypothetical planets that could cause the long-term linear trends. We have further studied a a eight very close-in (Porb ≤ 5 days), significantly eccentric (e ≥ 0.1) single-planet systems in Section 3.3. We find five of eight systems that cannot be explained by a single-planet orbital circularization with the conven- tional tidal quality factors. Potential companions for all of these systems are massive planets in the apsidally aligned region and should be observable with current technology. Since all of the host stars have high stellar jitters, it is possible that the planetary eccentricities are systematically overestimated or that outer planets are more difficult to observe than we have assumed here. Our model has some limitations and caveats. We have adopted the leading-order secular theory for two- planet systems, and in principle our model cannot be applied to high-eccentricity or high-inclination systems. However, the model predicts the general trend of ap- sidal states fairly well even for a highly eccentric case (see Section 3.1). Moreover, recent observations indi- cate that multiple-planet systems tend to be well-aligned (e.g., Figueira et al. 2012; Fabrycky et al. 2012). Nev- ertheless, it is useful to extend this kind of a study to higher eccentricities and inclinations and to systems with more than two planets. Also, we have ignored the slow decay in semimajor axis due to tides, which should be weaker than eccentricity damping by a fac- tor of e2 1; this is consistent with the low e assumption made by linear secular theory. Nevertheless, previous studies have shown that the eccentricity of the inner planet does damp faster than that of the outer planet 17 as a result of inward migration (e.g., Wu & Goldreich 2002; Greenberg & Van Laerhoven 2011). Another con- sequence of different damping rates is that the eccen- tricity ratio does not remain constant as the apsidally locked state evolves. Our expression for the mode- damping rates (Equation (21)) is consistent with that of (Greenberg & Van Laerhoven 2011, see their Equa- tion 19), in the limit of no migration. The tidal and rotational deformations of planets and stars also change orbital precession rates, and these effects might become more important than the GR effect for very close-in plan- ets or rapidly spinning stars. These effects can be easily added to our model using the techniques of Section 2.4 (e.g., Laskar et al. 2012). Overall, our study indicates that secular interactions slow down the tidal circularization of the inner planet while speeding up that of the outer planet. Our survey of likely systems available to us in 2012 indicate that secular interactions might not be a dominant cause for the cur- rently observed hot, eccentric planets. The lack of close- in planets with strong secular interactions might be par- tially explained by inward orbital decay that is acceler- ated by non-zero eccentricities (e.g., Adams & Laughlin 2006b). Further research should determine whether the scarcity of compact secular systems will persist. With improved statistics and precision measurements of close- in exoplanet eccentricities, it might become possible to find diagnostic differences in the eccentricity distribu- tions for single- and multiple-planet systems. REFERENCES Adams, F. C., & Laughlin, G. 2006a, ApJ, 649, 992, 992 -- . 2006b, ApJ, 649, 1004, 1004 Bakos, G. Á., Hartman, J., Torres, G., et al. 2011, ApJ, 742, 116, Greenberg, R., & Van Laerhoven, C. 2011, ApJ, 733, 8, 8 Hamilton, D. P. 1994, Icarus, 109, 221, 221 Hartman, J. D., Bakos, G. Á., Kipping, D. M., et al. 2011a, ApJ, Butler, R. P., Wright, J. T., Marcy, G. W., et al. 2006, ApJ, 646, Matsumura, S., Takeda, G., & Rasio, F. A. 2008, ApJ, 686, L29, Figueira, P., Marmier, M., Boué, G., et al. 2012, A&A, 541, A139, (Tuscon, AZ, USA: Univ. of Arizona Press), 159 -- 223 Fischer, D. A., Laughlin, G., Butler, P., et al. 2005, ApJ, 620, 414, 1278, 1278 Fischer, D. A., Laughlin, G., Marcy, G. W., et al. 2006, ApJ, 637, Goldreich, R. 1963, MNRAS, 126, 257, 257 Greenberg, R. 1977, in Planetary Satellites, ed. J. A. Burns (Tuscon, AZ, USA: Univ. of Arizona Press), 157 -- 168 Rauch, K. P., & Hamilton, D. P. 2002, in Bulletin of the American Astronomical Society, Vol. 34, Bull. Am. Astron. Soc., 938 Shen, Y., & Turner, E. L. 2008, ApJ, 685, 553, 553 116 23 Barnes, R., & Greenberg, R. 2006a, ApJ, 652, L53, L53 -- . 2006b, ApJ, 638, 478, 478 Batygin, K., Laughlin, G., Meschiari, S., et al. 2009, ApJ, 699, 23, Batygin, K., & Morbidelli, A. 2013, AJ, 145, 1, 1 Beaugé, C., Ferraz-Mello, S., & Michtchenko, T. A. 2003, ApJ, 593, 1124, 1124 Bobrov, A. M., Vasil'Ev, P. P., Zharkov, V. N., & Trubitsyn, V. P. 1978, Soviet Ast., 22, 489, 489 Bonfils, X., Delfosse, X., Udry, S., et al. 2013, A&A, 549, A109, Brouwer, D., van Woerkom, A., & Jasper, J. 1950 Burns, J. A. 1977, in Planetary Satellites, ed. J. A. Burns (Tuscon, AZ, USA: Univ. of Arizona Press), 113 -- 156 Butler, R. P., Marcy, G. W., Fischer, D. A., et al. 1999, ApJ, 526, 90, 90 Butler, R. P., Vogt, S. S., Marcy, G. W., et al. 2000, ApJ, 545, Chiang, E. I. 2003, ApJ, 584, 465, 465 Correia, A. C. M., Boué, G., & Laskar, J. 2012, ApJ, 744, L23, Danby, J. M. A. 1988 Fabrycky, D. C., Lissauer, J. J., Ragozzine, D., et al. 2012, ArXiv e-prints, arXiv:1202.6328 728, 138, 138 59, 59 Hartman, J. D., Bakos, G. Á., Torres, G., et al. 2011b, ApJ, 742, Horedt, G. P., ed. 2004, Astrophysics and Space Science Library, Vol. 306, Polytropes - Applications in Astrophysics and Related Fields Hubbard, W. B. 1974, Icarus, 23, 42, 42 Jackson, B., Greenberg, R., & Barnes, R. 2008, ApJ, 678, 1396, Ketchum, J. A., Adams, F. C., & Bloch, A. M. 2013, ApJ, 762, Laskar, J., Boué, G., & Correia, A. C. M. 2012, A&A, 538, A105, Lee, M. H. 2004, ApJ, 611, 517, 517 Maness, H. L., Marcy, G. W., Ford, E. B., et al. 2007, PASP, 119, Mardling, R. A. 2007, MNRAS, 382, 1768, 1768 Matsumura, S., Peale, S. J., & Rasio, F. A. 2010, ApJ, 725, 1995, 1396 71, 71 A105 1995 L29 Motz, L. 1952, ApJ, 115, 562, 562 Murray, C. D., & Dermott, S. F. 1999 Namouni, F. 2007, in Lecture Notes in Physics, Berlin Springer Verlag, Vol. 729, Lecture Notes in Physics, Berlin Springer Verlag, ed. D. Benest, C. Froeschle, & E. Lega, 233 Peale, S. J. 1986, in Satellites, ed. J. A. Burns & M. S. Matthews Pont, F., Husnoo, N., Mazeh, T., & Fabrycky, D. 2011, MNRAS, Ragozzine, D., & Wolf, A. S. 2009, ApJ, 698, 1778, 1778 Rasio, F. A., Tout, C. A., Lubow, S. H., & Livio, M. 1996, ApJ, 470, 1187, 1187 A109 916, 916 504, 504 505, 505 L23 A139 481, 481 1094, 1094 18 718, 575, 575 1084, 1084 1895, 1895 Smalley, B., Anderson, D. R., Collier Cameron, A., et al. 2011, A&A, 526, A130, A130 Winn, J. N., Johnson, J. A., Howard, A. W., et al. 2010, ApJ, Wright, J. T., Upadhyay, S., Marcy, G. W., et al. 2009, ApJ, 693, Wu, Y., & Goldreich, P. 2002, ApJ, 564, 1024, 1024 Zakamska, N. L., Pan, M., & Ford, E. B. 2011, MNRAS, 410, Zhang, K. 2007, PhD thesis, Zhang, K., & Hamilton, D. P. 2003, in Bulletin of the American Astronomical Society, Vol. 35, AAS/Division for Planetary Sciences Meeting Abstracts #35, 1485 Zhang, K., & Hamilton, D. P. 2007, Icarus, 188, 386, 386 -- . 2008, Icarus, 193, 267, 267
1605.01892	4	1605	2016-09-22T03:10:34	Response to the Comment by Haack et al. (2015) on the paper by Anfinogenov et al. (2014): John's stone: A possible fragment of the 1908 Tunguska meteorite	[ "astro-ph.EP" ]	The article provides an open discussion and a critical feedback to the comments of Haack et al. (2015) and emphasizes a significance of the first macroscopic evidence for a candidate meteorite of a new type: planetary-origin meteorite composed of silica-rich sedimentary rock. Discussion concerns the arguments for (i) candidate parental bodies including the Earth, Mars and icy moons of Saturn and Jupiter; (ii) PGE anomaly versus glassy silicate microspherules and quartz grains anomaly in the area of the 1908 Tunguska catastrophe; (iii) isotopic heterogeneity of unmixed silicate reservoirs on Mars; (iv) possible terrestrial loss or contamination in the noble gas signatures in meteorites that spent time in the extreme weather conditions; (v) cosmogenic isotopes and shielding; and (vi) pseudo meteorites. We conclude that the list of candidate parental bodies for hypothetical sedimentary-origin meteorites includes, but is not limited by the Earth, Mars, Enceladus, Ganymede, and Europa. A parental body should be identified based on the entire body of evidence which is not limited solely by tests of oxygen and noble gas isotopes whose signatures may undergo terrestrial contamination and may exhibit significant heterogeneity within the parental bodies.	astro-ph.EP	astro-ph	Response to the Comment by Haack et al. (2015) on the paper by Anfinogenov et al. (2014): John's stone: A possible fragment of the 1908 Tunguska meteorite Yana Anfinogenova,1* John Anfinogenov,2 Larisa Budaeva,3 and Dmitry Kuznetsov1 Authors: Affiliations: 1Yana Anfinogenova, Ph.D., Institute of Physics and Technology, National Research Tomsk Polytechnic University 2John Anfinogenov, Tungussky Nature Reserve, Ministry of Natural Resources and Ecology of the Russian Federation 3Larisa Budaeva, National Research Tomsk State University 1Dmitry Kuznetsov, Ph.D., National Research Tomsk Polytechnic University *Corresponding author: Dr. Yana Anfinogenova. Address: TPU, 30 Lenin Ave., Tomsk, 634050, Russia. Tel: +79095390220. E-mail: [email protected] and [email protected] Abstract Here we present our response to the Comment on "John's stone: A possible fragment of the 1908 Tunguska meteorite" (Haack et al., 2015, Icarus 265, 238–240). The article provides an open discussion and a critical feedback to the comments of Haack et al. (2015) and emphasizes a significance of the first macroscopic evidence for a candidate meteorite of a new type: planetary-origin meteorite composed of silica-rich sedimentary rock. Discussion concerns the arguments for (i) candidate parental bodies including the Earth, Mars and icy moons of Saturn and Jupiter; (ii) PGE anomaly versus glassy silicate microspherules/quartz grains anomaly in the area of the 1908 Tunguska catastrophe; (iii) isotopic heterogeneity of unmixed silicate reservoirs on Mars; (iv) possible terrestrial loss or contamination in the noble gas signatures in meteorites that spent time in the extreme weather conditions; (v) cosmogenic isotopes and shielding; and (vi) "pseudo meteorites". We conclude that the list of candidate parental bodies for hypothetical sedimentary-origin meteorites includes, but is not limited by the Earth, Mars, Enceladus, Ganymede, and Europa. A parental body should be identified based on the entire body of evidence which is not limited solely by tests of oxygen and noble gas isotopes whose signatures may undergo terrestrial contamination and may exhibit significant heterogeneity within the parental bodies. Key words: meteorites; 1908 Tunguska event; Earth; Mars; icy moons Introduction Our team (Anfinogenov et al., 2014; Anfinogenov and Budaeva, 1998) is grateful to authors of the comment (Haack et al., 2015) for their interest in the 1908 Tunguska event and for their work on updating Δ17O values and on determining the noble gases isotopic compositions in the samples of John's Stone, an exotic boulder with the numerous signs of high-energy impact in the epicenter of the 1908 Tunguska catastrophe (Anfinogenov et al., 2014). Briefly, the exotic boulder known as John's Stone has been discovered on the Stoykovich Mountain in the epicenter area of the Tunguska catastrophe (Anfinogenov et al., 2014). Pattern of permafrost destruction suggests high-speed entry and lateral ricochet of John's Stone in the ground with further deceleration and breakage. Landing velocity of John's Stone is estimated to be 547 m/s (Anfinogenov et al., 2014). John's Stone is composed of highly silicified gravelite sandstone (~99% SiO2). Outer surface of several splinters of this rock shows continuous glassy coating reminiscent of fresh enamel or fusion crust. There is clear consistency in geometry of Tunguska meteoroid flight trajectory, locations of John's Stone fragments and cleaved pebbles, and directions of impact furrows. John's Stone locates on quaternary deposits at the top of the Stoykovich Mountain and is exotic to the territory within at least hundreds of kilometers around the epicenter (Sapronov, 1986). There are no signs of past glaciation throughout the region (Sapronov, 1986). Decoding of aerial survey photographs covering area within 40 km from the epicenter shows the absence of active diatremes. There is significantly higher (by hundred-fold) content of glassy silicate microspherules precipitated from the atmosphere in the peat layer of 1908 (Dolgov et al., 1973). Macroscopic evidence and data regarding glassy silicate microspherules suggest that John's Stone may be a fragment of the Tunguska meteorite 1908 and may represent a new type of meteorites of planetary origin (Anfinogenov et al., 2014). Carl Sagan said once that "extraordinary claims require extraordinary evidence". We believe that macroscopic signs of impact associated with John's Stone and its nature which is alien to the region of the Tunguska catastrophe represent indeed extraordinary evidence. However, based solely on the studies of Δ17O and noble gases isotopes in the samples of John's Stone, Haack et al. (2015) conclude that "John's Stone is a terrestrial rock" and an "unlikely fragment of the projectile responsible for the Tunguska event". We believe it is necessary to raise comments regarding the method and the reasoning Haack et al. (2015) used for their attempt for identification of the parental body of John's Stone. Along with providing an open discussion and a critical feedback to the comments of Haack et al. (2015), we would like to emphasize the novelty of our idea and significance of the first macroscopic evidence for a candidate meteorite of a new type. We invite our opponents to reconsider our hypothesis based on the discussion we provide below. Planetary- origin meteorites composed of highly metamorphic sedimentary rocks may indeed exist keeping in mind recent discovery of sedimentary pebble-conglomerate rocks on Mars (Daily Mail Reporter. 2013), the presence of liquid water (Heller et al., 2015; Saur et al., 2015; Steinbrügge et al., 2015; Tanigawa et al., 2014; Vance et al., 2014) and hydrothermal activity within the Solar System (Hsu et al., 2012; Postberg et al., 2011), and possibility of sedimentary rock formation in the presence of liquid water flows generated by tidal forces on the satellites of the giant planets. Parental body Haack et al. (2015) argue that "since sandstones can only form on a parental body with liquid water and, by inference also an atmosphere… there are only two possible parental bodies in the Solar System: the Earth and Mars". Evidence does not support this dictum. Many bodies without a significant atmosphere have abundant liquid water (Heller et al., 2015; Saur et al., 2015; Steinbrügge et al., 2015; Tanigawa et al. 2014; Vance et al. 2014). Hydrothermal rocks can form on the bodies other than the Earth and Mars (Hsu et al., 2012; Postberg et al., 2011). Moreover, the presence of powerful tidal currents of water in subsurface oceans on icy satellites of Saturn and Jupiter may provide conditions sufficient for formation of sedimentary rocks including sandstones with a variety of grain sizes. The Saturnian moon Enceladus can be considered a candidate parental body for sedimentary-origin meteorites due to (i) the presence of global hydrothermal activity, (ii) the presence of powerful tidal currents of oceanic water in subsurface oceans potentially resulting in formation of sediments, and (iii) the past history of large-scale impacts explaining ejection of Enceladus' crust fragments into space. Indeed, the plume of Enceladus emits nanometre- sized SiO2 (silica)-containing ice grains (Hsu et al., 2012) formed as frozen droplets from a liquid water reservoir contacting with rock (Postberg et al., 2011). Characteristics of these silica nanoparticles indicate ongoing high temperature (>90 °C) global-scale geothermal and hydrothermal reactions on Enceladus favored by large impacts (Hsu et al., 2015). Enceladus has a differentiated interior consisting of a rocky core, an internal ocean and an icy mantle. Simulation studies suggest that large heterogeneity in the interior, possibly including significant core topography may be due to collisions with large differentiated impactors with radius ranging between 25 and 100 km. Impacts played the crucial role on the evolution of Enceladus (Monteux et al., 2016) and similar effects on evolution are very likely on the other moons of Saturn as well as on other planetary objects, such as Ceres (Davison et al., 2015; Ivanov, 2015). Recent evidence suggests that a putative subsurface water ocean is present on Ganymede. Iron core of Ganymede is surrounded by a silicate rock mantle and by a globe- encircling, briny subsurface water ocean with alternating layers between high pressure ices and salty liquid water (Saur et al., 2015; Steinbrügge et al., 2015; Vance et al., 2014). If Ganymede or Callisto had acquired their H2O from newly accreted planetesimals after the Grand Track (Mosqueira and Estrada 2003), then Io and Europa would be water-rich, too (Heller et al., 2015; Tanigawa et al., 2014). Tidal dissipation and tidal resonance in icy satellites with subsurface oceans are major heat sources for the icy satellites of the giant planets (Kamata et al., 2015). Tidal forces generate heat and currents of liquid water or brine powerful enough to produce sediments that can undergo metamorphic transformations due to hydrothermal activity. Therefore, there are several candidate parental bodies in the Solar System that may be a place of origin for the sedimentary-type meteorites resembling John's Stone. PGE anomaly versus glassy silicate microspherules/quartz grains anomaly Haack et al. (2015) rule out the Earth as a parental body of John's Stone based on data of Rasmussen et al. (1999) and Hou et al. (2004) who reported the presence of PGE anomaly in peat cores at an estimated depth of 1908. We believe that the presence of this anomaly could not rule out the possibility of an independent event or the presence of another anomaly at the same area, essentially, based on the laws of logic. As a matter of fact, PGE anomaly is not the only anomaly known in the region of the 1908 Tunguska catastrophe. In our original article (Anfinogenov et al., 2014) we cited paper reporting the discovery of significantly higher content of glassy silicate microspherules precipitated from the atmosphere in peat layer of 1908 throughout the area of the Tunguska catastrophe (Dolgov et al. 1973). Peat layer of 1908 contained up to hundredfold-higher count of gray and colorless transparent silicate microspherules than the adjacent peat layers. Data of neutron activation analysis showed that chemical composition of microspherules was distinct from that of industrial glass, known stony meteorites, tektites, and Moon rocks (Kolesnikov et al., 1976). It is essential to note that Dr. E.M. Kolesnikov who performed the aforementioned neutron activation analysis (Kolesnikov et al., 1976) of anomalous silicate microspherules from peat layer of 1908 is also a co-author of both papers by Rasmussen et al. (1999) and by Hou et al. (2004) reporting PGE anomaly. Therefore, two anomalies are present in the area of interest and they both are reported by the same Tunguska meteorite explorer, Dr. E.M. Kolesnikov: (i) PGE anomaly and (ii) anomalous abundance of glassy silicate microspherules precipitated from the atmosphere in peat layer of 1908. Even more so, the quartz grains were found in sediment cores collected from Lake Cheko (Gasperini et al., 2009) suggesting that they might have resulted from dust produced by the explosion in the atmosphere of the main body if the Tunguska cosmic body were silica-rich. No macroscopic pieces of chondritic or cometary projectile have been indeed found in the area of the 1908 Tunguska catastrophe. In 1993, Chyba et al. reported in Nature magazine that carbonaceous asteroids and especially comets are unlikely candidates for the Tunguska object. The Tunguska event represents a typical fate for stony asteroids tens of meters in radius entering the Earth's atmosphere at common hypersonic velocities (Chyba et al., 1993). In this regard, John's Stone represents a sound macroscopic candidate for a stony impactor though of a previously unknown type. This hypothesis is consistent with John's Stone phenomenon bearing the numerous signs of high-energy impact and glassy fusion crust-like surface modification on some splinters. It is also consistent with the discovery of the quartz grains in sediment cores collected from Lake Cheko (Gasperini et al., 2009) and the glassy silicate microspherules anomaly associated with the area of the 1908 Tunguska catastrophe (Dolgov et al., 1973; Kolesnikov et al., 1976). Studies by Rasmussen et al. (1999) and by Hou et al. (2004) invoke questions regarding the methodology. These studies are based on data from an insignificant number of peat core samples showing PGE anomaly: only four adjacent peat columns and only one peat column were studied by Rasmussen et al. (1999) and by Hou et al. (2004), respectively. Peat cores from singular topographic locations were tested in these studies. On the contrary, Dolgov et al. (1973) described anomalous content of glassy silicate microspherules in peat layer of 1908 in hundreds of peat samples associated with the entire area of the 1908 Tunguska catastrophe. Two explanations may be proposed for the presence of these two anomalies (PGE and glassy silicate microspherules/quartz grains) in the area of the 1908 Tunguska catastrophe: (i) occurrence of two impact events independently involving silica-rich impactor and chondritic or cometary projectile or (ii) complex conglomerate composition of the impactor consisting of multiple parts that merged due to either collision of parental asteroids in outer space or due to a co-ejection of different adjacent rocks from a parental planetary body due to a high- energy impact. For example, it could be a co-ejection of enclosing bedrocks, intrusive igneous rocks, and impactor material where any of these components could partially melt due to impact. These processes could produce so-called rubble-pile asteroids. Therefore, neither the Earth nor other bodies in the Solar System can be ruled out as candidate parental bodies solely on the ground of PGE anomaly reported by Rasmussen et al. (1999) and Hou et al. (2004). Moreover, the anomaly of glassy silicate microspherules/quartz grains (Dolgov et al., 1973; Gasperini et al., 2009) provides a statistically significant evidence for silica-rich projectile responsible for the 1908 Tunguska event. Isotope tests Haack et al. (2015) rule out Mars as a candidate parental body for John's Stone based on oxygen and noble gas isotope tests. It is necessary to emphasize that isotopic characterizations per se are very essential and further elemental and isotopic characterizations of this exotic Tunguska boulder are highly encouraged. However, numerical results of isotopic characterizations though important on their own do not prove the conclusions of Haack et al. (2015). No rock samples have ever been delivered from Mars to the Earth before. Isotopic compositions of Martian rocks have never been tested by Mars rovers yet. All which was tested up to day were the so called Martian meteorites. However, diversities in the rock- forming processes and in the corresponding rock types within planets are great and insufficiently studied. A significant heterogeneity in Δ17O in rocks of different types has been reported (Wang et al., 2013). Pack and Herwartz (2015; 2014) provide evidence that the concept of a single terrestrial mass fractionation is invalid on small-scale. They conclude that mineral assemblages in rocks fall on individual "rock" mass fractionation lines with individual slopes and intercepts. We believe the same may be true for other bodies of the Solar System such as Mars, large moons of Jupiter and Saturn, and large asteroids. An idea of separate long-lived silicate reservoirs on Mars is supported by radiogenic isotope studies (Borg et al., 1997; 2003). The distinct Δ17O and δ18O values of the silicate fraction of NWA 7034 compared to other SNC meteorites supports the idea of distinct lithospheric reservoirs on Mars that have remained unmixed throughout Martian history (Agee et al., 2013; Ziegler et al., 2013). Isotopic heterogeneity, including that in the noble gases, can be significant in the Martian mantle. Models for accretion and early differentiation of Mars were tested with chronometers several of which provided evidence of very early isotopic heterogeneity preserved within Mars (Halliday et al., 2001). If significant heterogeneities are reported for known Martian meteorites which are all igneous rocks, then even greater isotopic heterogeneities may exist for the rocks of different types within planet. The most essential issue is that the work of Haack et al. (Haack et al., 2015) did not provide any characterizations of any particular terrestrial and extraterrestrial rocks similar to John's Stone (highly metamorphic highly silicified gravelite sandstones with SiO2 content of nearly 99%), which would be helpful for methodologically proper justification of the conclusions they drew. The tests, performed with the samples of John's Stone, should be done with similar extraterrestrial rocks which may be feasible midterm considering that matching sedimentary silica-rich rocks have been found on Mars (Bandfield et al., 2004; 2006; Christensen et al., 2005; Edgett and Malin 2000; Jerolmack, 2013; Kerber and Head, 2012; McLennan, 2003; Michalski, 2013; Smith and Bandfield, 2012; Squyres et al., 2008; Williams et al., 2013). Indeed, recent studies showed that there are sedimentary rocks on Mars including pebble conglomerates and sandstones (Williams et al., 2013; Kerber and Head, 2012). Quartz-bearing deposits are consistently co-located with hydrated silica on Mars (Smith and Bandfield, 2012). There is striking similarity between Martian pebble conglomerates (images are available at Daily Mail Reporter 2013) and John's Stone (Anfinogenov et al., 2014). Notably, the Tunguska boulder, in addition to SiO2, contains traces of a Ti-oxide phase and its bulk composition (Bonatti et al., 2015) is similar to the composition inferred from APXS data (Squyres et al., 2008) for the silica deposits of the Gusev crater on Mars. A paper by Bonatti et al. (2015) contains a section with detailed review of sedimentary rocks and hydrothermal activity on Mars. On the other hand, terrestrial rocks reminiscent of John's Stone would represent necessary terrestrial control required to be tested in the same set of experiments by using the same equipment. All the more so because the difference between estimations of Δ17O by Bonatti et al. (2015) and by Haack et al. (2015) may suggest the presence of equipment- specific errors or human factors. Instead of comparing John's Stone with similar terrestrial and extraterrestrial rocks, Haack et al. (2015) compared isotopic compositions of John's Stone with published data on Chelyabinsk meteorite (Nishiizumi et al., 2013; Buzemann et al., 2014), SNC meteorites (Franchi et al., 1999), NWA 7034 (Agee et al., 2013), CIs (Clayton and Mayeda, 1999), Lunar rocks (no reference was provided by Haack et al., 2015), angrites and HEDs (Greenwood et al., 2005), aubrites and enstatite chondrites (Newton et al., 2000), and averaged characteristics of unidentified quartz-rich terrestrial rocks. This comparison is valuable on its own, but, due to unique occurrence of the Tunguska boulder, it does not prove the conclusions of Haack et al. (2015). Therefore, we strongly suggest more careful consideration of Mars as a candidate parental body of John's Stone. Possible terrestrial loss or contamination in the noble gas signatures Terrestrial loss or contamination in the noble gas signatures should be considered for any meteorite including Martian meteorites that spent time in terrestrial environment (Schwenzer et al., 2013). All the more so for Tunguska meteorite because weather conditions in Tunguska are extreme and temperatures range from –61 °C during winter to +40 °C during summer. If John's Stone is a fragment of the Tunguska impactor, then the repeated freezing of water easily penetrating into tiny pores in its material for decades could significantly accelerate contamination of the rock with terrestrial elements concealing original isotopic signatures. Shielding Haack et al. (2015) state that "The lack of any cosmogenic noble gases (particularly striking in 3He, 21Ne, 38Ar) would be consistent with an extraterrestrial origin under large shielding". We agree with this notion. Indeed, a pre-atmospheric diameter of Tunguska cosmic body is estimated to be tens of meters in diameter (Chyba et al., 1993) suggesting up to 50 m of shielding whereas a core part of the Tunguska impactor would have better chances of surviving passage through the atmosphere. Pseudo meteorites We found a total of 226 records for meteorites with status of "pseudo" in the category of "terrestrial meteorites" in the Meteoritical Bulletin Database of the International Society for Meteorites and Planetary Science (The Meteoritical Society). Except NWA 6944, all other "pseudo" meteorites were found in Antarctica. No characteristics are provided for description of these rocks, but the fact that these specimens were considered candidates for meteorites and were included in the Meteoritical Bulletin Database might suggest that their minerology was exotic to the areas of discovery and/or they presented with a glassy cover or a fusion crust-like surface. It would be helpful to know more about those rocks. Sedimentary-origin meteorites may be present among them. Technical issues with representation of the noble gas signatures reported by Haack et al. (2015) We also would like to point out some inconsistencies in data published by Haack et al. (2015). In particular, Table 1 shows that 36Ar was not detected in four out of five samples of John's Stone. This absence of 36Ar would agree with data generated by Mars rover suggesting a depletion of Martian atmosphere in light Ar isotopes (Atreya et al., 2013). However, for some reason, Haack et al. (2015) state that the ratio of light to heavy argon isotopes in the samples of John's Stone corresponds to the terrestrial one. Table 1 (Haack et al., 2015) does not suggest that. We do not understand how five ratios of 36Ar/40Ar were calculated when 36Ar was not detected in four out of five samples. 38Ar contents cannot be understood from Table 1 for the same reason. Summary The numerical results of the isotopic tests performed by Haack et al. (2015) present valuable phenomenological information regarding the Tunguska boulder historically called John's Stone. However, the conclusions of Haack et al. (2015) regarding the identification of its parental body do not directly follow from the results of their tests. These conclusions represent an opinion rather than an interpretation which is, in our view, insufficiently evidence-based and sounds like "extraterrestrial rocks of this particular isotopic composition do not exist because they could not exist" according to the scientific dictum. Moreover, while drawing the conclusions, Haack et al. (2015) ignored the entire pool of evidence suggesting high-energy impact associated with the John's Stone phenomenon described by Anfinogenov et al. (2014). Authors of the original article (Anfinogenov et al., 2014) invite scientific community to consider the significance of our hypothesis for the extraterrestrial origin of John's Stone found with signs of high-energy impact in the epicenter of the 1908 Tunguska catastrophe. Further in-depth study of this rock should be undertaken including the thermoluminescence analysis, rock age determination, and the comparison of John's Stone signatures with those of similar terrestrial and extraterrestrial rocks. The site of impact associated with John's Stone requires comprehensive interdisciplinary field examination. We conclude that the list of candidate parental bodies for hypothetical sedimentary- origin meteorites includes, but is not limited by the Earth, Mars, Enceladus, Ganymede, and Europa. A parental body should be identified based on consensus of all evidence, not limited solely by tests for oxygen and noble gas isotopes which may undergo terrestrial contamination and exhibit significant heterogeneity within the parental bodies. The study of the Solar System is just in the beginning. Acknowledgments We thank Dr. Haack and two anonymous reviewers who provided critical feedback in a framework of private discussion in Icarus that allowed us to clarify our viewpoints and improve linguistics of the manuscript. References 1. Agee C. B., Wilson N. V., McCubbin F. M., Ziegler K., Polyak V. J., Sharp Z. D., Asmerom Y., Nunn M. H., Shaheen R., Thiemens M. H., Steele A., Fogel M. L., Bowden R., Glamoclija M., Zhang Z, and Elardo S. M. 2013. Unique Meteorite from Early Amazonian Mars: Water-Rich Basaltic Breccia Northwest Africa 7034. Science 339:780- 785, doi: 10.1126/science.1228858. 2. Anfinogenov J. and Budaeva L. I. 1998. Tunguska meteorite problem in light of the so called martian meteorites. International Conference on 90th anniversary of the Tunguska Event, Krasnoyark (Russia) (available at http://www.th.bo.infn.it). 3. Anfinogenov J., Budaeva L., Kuznetsov D., and Anfinogenova Y. 2014. John's Stone: a the 1908 Tunguska meteorite. Icarus 243:139-147. doi: possible fragment of 10.1016/j.icarus.2014.09.006. 4. Atreya S. K., Trainer M. G., Franz H. B., Wong M. H., Manning H. L. K., Malespin C. A., Mahaffy P. R., Conrad P. G., Brunner A. E., Leshin L. A., Jones J. H., Webster C. R., Owen T. C., Pepin R. O., Navarro-Gonzalez R. 2013. Primordial argon isotope fractionation in the atmosphere of Mars measured by the SAM instrument on Curiosity and implications for atmospheric loss. Geophysical Research Letters 40:5605-5609, doi: 10.1002/2013GL057763. 5. Bandfield J. L. 2006. Extended surface exposures of granitoid compositions in Syrtis Major, Mars. Geophys. Res. Lett. 33, L06203, doi:10.1029/2005GL025559. 6. Bandfield J. L., Hamilton V. E., Christensen P. R., and McSween Jr. H. Y. 2004. Identification of quartzofeldspathic materials on Mars, J. Geophys. Res. 109, E10009, doi:10.1029/2004JE002290. 7. Bonatti E., Breger D., Di Rocco T., Franchia F., Gasperini L., Polonia A., Anfinogenov J., and Anfinogenova Y. 2015. Origin of John's Stone: A quartzitic boulder from the site of 258:297-308, Tunguska doi:10.1016/j.icarus.2015.06.018. explosion. (Siberia) the 1908 Icarus 8. Borg L. E., Nyquist L. E., Taylor L. A., Wiesmann H., and Shih C.Y. 1997. Constraints on Martian differentiation processes from Rb-Sr and Sm-Nd isotopic analyses of the basaltic shergottite QUE 94201. Geochimica et Cosmochimica Acta 61:4915-4931. doi: 10.1016/S0016-7037(97)00276-7. 9. Borg L. E., Nyquist L. E., Wiesmann H., Shih C. Y., and Reese Y. 2003. The age of Dar al Gani 476 and the differentiation history of the martian meteorites inferred from their radiogenic isotopic systematics. Geochimica et Cosmochimica Acta 67:3519-3536, doi: 10.1016/S0016-7037(00)00094-2. 10. Busemann H., Toth E. R., Clay P. L., Gilmour J. D., Nottingham M., Strashnov I., Wieler R., Nishiizumi K., and Jones R. H. 2014. Noble gases in the LL5 chondrite Chelyabinsk. Lunar Planet. Sci. Conf. 45, #2805. 11. Christensen P. R., Bandfield J. L., Clark R. N., Edgett K. S., Hamilton V. E., Hoefen T., Kieffer H. H., Kuzmin R. O., Lane M. D., Malin M. C., Morris R. V., Pearl J. C., Pearson R., Roush T. L., Ruff S. W., and Smith M. D. 2000. Detection of crystalline hematite mineralization on Mars by the thermal emission spectrometer: Evidence for near-surface water. J. Geophys. Res. – Planets 105 (E4):9623:9642, doi: 10.1029/1999JE001093. 12. Chyba C. F., Thomas P. J., and Zahnle K. J. 1993. The 1908 Tunguska explosion: atmospheric disruption of a stony asteroid. Nature 361:40-44, doi:10.1038/361040a0. 13. Clayton R. N. and Mayeda T. K. 1999. Oxygen isotope studies of carbonaceous chondrites. Geochim. Cosmochim. Acta 63:2089-2104. 14. Daily Mail Reporter. 2013. Pebbles prove streams once flowed across Mars - and raise hopes of finding life on the red planet. MailOnline (30 May 2013) (available at http://www.dailymail.co.uk/sciencetech/article-2333471/Pebbles-reveal-streams-flowed- Mars--raise-hopes-finding-life-red-planet.html#ixzz2XhImHYT2). 15. Davison T. M., Collins G. S., O'Brien D. P., Ciesla F. J., Bland P. A., and Travis B. J. 2015. Impact bombardment of ceres. Proc. Lunar Sci. Conf. 46, 2116. 16. Dolgov Yu. A., Vasiliev N. V., Shugurova N. A., Lvov Yu. A., Lavrentiev Yu. G., and Grishin Yu. A. 1973. Composition of microspherules and peat of Tunguska meteorite fall region. Meteoritics 32:147-149. 17. Edgett K. S. and Malin M. C. 2000. Examples of martian sandstone: indurated, lithified, and cratered eolian dunes in MGS MOC images. Lunar and Plantery Science Conference 31: Abs. 1071. 18. Franchi I. A., Wright I. P., Sexton A. S., and Pillinger C. T. 1999. The oxygen-isotopic composition of Earth and Mars. Meteorit. Planet. Sci. 34:657-661. 19. Gasperini L., Bonatti E., Albertazzi S., Forlani L., Accorsi C.A., Longo G., Ravaioli M., Alvisi F., Polonia A., and Sacchetti F. 2009. Sediments from Lake Cheko (Siberia), a possible impact crater for the 1908 Tunguska Event. Terra Nova 21: 489-494, doi:10.1111/j.1365-3121.2009.00906.x. 20. Greenwood R. C., Franchi I. A., Jambon A., and Buchanan P. C. 2005. Widespread magma oceans on asteroidal bodies in the early solar system. Nature 435:916-918. 21. Haack H., Greenwood R. C., and Busemann H. 2015. Comment on "John's stone: A possible fragment of the 1908 Tunguska meteorite" (Anfinogenov et al., 2014, Icarus 243, 139-147). Icarus 265:238-240, doi:10.1016/j.icarus.2015.09.018. 22. Halliday A. N., Wanke H., Birck J-L., and Clayton R. N.. The accretion, composition and 96;1-4:197-230, Science Reviews of Mars. differentiation early doi:10.1023/A:1011997206080. Space 23. Heller R., Marleau G. D., and Pudritz R. E. 2015. The formation of the Galilean moons and Titan in the Grand Tack scenario. Astronomy & Astrophysics 579: L4, doi:10.1051/0004-6361/201526348. 24. Hou Q. L., Kolesnikov E. M., Xie L. W., Kolesnikova N. V., Zhou M. F., and Sun M. 2004. Platinum group element abundances in a peat layer associated with the Tunguska event, further evidence for a cosmic origin. Planetary and Space Science 52: 331-340, doi:10.1016/j.pss.2003.08.002. 25. Hsu H. W., Postberg F., Sekine Y., Shibuya T., Kempf S., Horanyi M., Juhasz A., Altobelli N., Suzuki K., Masaki Y., Kuwatani T., Tachibana S., Sirono S. I., Moragas- Klostermeyer G., and Srama R. 2015. Ongoing hydrothermal activities within Enceladus. Nature 519:207-+, doi:10.1038/nature14262. 26. Ivanov B. A. 2015. Ceres: possible records of giant impacts. Proc. Lunar Sci. Conf. 46, 1077. 27. Jerolmack D. J. 2013. Pebbles on Mars. Science 340:1055-1056, doi: 10.1126/science.1239343. 28. Kerber L. and Head J. W. 2012. A progression of induration in Medusae Fossae Formation transverse aeolian ridges: evidence for ancient aeolian bedforms and extensive reworking. Earth Surf. Process. Landforms 37:422-433, doi: 10.1002/esp.2259. 29. Kolesnikov E. M., Lul A. Yu., and Ivanova G. M. 1976. Neutron activation analysis of some elements in silica spherules from peat of Tunguska meteorite fall region, in Cosmic matter on Earth. Novosibirsk: Nauka Publishing House, Siberian Branch. In Russian. 30. McLennan S. M. 2003. Sedimentary silica on Mars. Geology 31:315-318, doi:10.1130/0091-7613(2003)031<0315:SSOM>2.0.CO;2. 31. Michalski J. R., Cuadros J., Niles P. B., Parnell J., Rogers A. D., and Wright S. P. 2013. Groundwater activity on Mars and implications for a deep biosphere. Nature Geoscience 6:133-138, doi:10.1038/ngeo1706. 32. Monteux J., Collins G. S., Tobie G., and Choblet G. 2016. Consequences of large impacts on Enceladus' core shape. Icarus 264:300-310, doi:10.1016/j.icarus.2015.09.034 33. Mosqueira I. and Estrada P. R. 2003. Formation of Large Regular Satellites of Giant Planets in an Extended Gaseous Nebula I: Subnebula Model and Accretion of Satellites. Icarus 163:198-231. 34. Pack A. and Herwartz D. 2014. The triple oxygen isotope composition of the Earth mantle and understanding Delta O-17 variations in terrestrial rocks and minerals. Earth Planet. Sci. Lett. 390: 138-145, doi:10.1016/j.epsl.2014.01.017. 35. Pack A. and Herwartz D. 2015. Observation and interpretation of Delta O-17 variations in terrestrial rocks - Response to the comment by Miller et al. on the paper by Pack & Herwartz (2014). Earth Planet. Sci. Lett. 418:184-186, doi:10.1016/j.epsl.2015.02.044. 36. Postberg F., Schmidt J., Hillier J., Kempf S., and Srama R. 2011. A salt-water reservoir as the source of a compositionally stratified plume on Enceladus. Nature 474:620-622, doi: 10.1038/nature10175. 37. Rasmussen K. L., Olsen H. J. F., Gwozdz R., and Kolesnikov E. M. 1999. Evidence for a very high carbon/iridium ratio in the Tunguska impactor. Meteoritics & Planetary Science 34: 891-895, doi:10.1111/j.1945-5100.1999.tb01407.x. 38. Sapronov N. L. 1986. Ancient volcanic structures in the south of the Tunguska syneclise. 104 pp. In Russian. 39. Saur J., Duling S., Roth L., Jia X., Strobel D. F., Feldman P. D., Christensen U. R., Retherford K. D., McGrath M. A., Musacchio F., Wennmacher A., Neubauer F. M., Simon S., and Hartkorn O. 2015. The search for a subsurface ocean in Ganymede with Hubble Space Telescope observations of its auroral ovals. J. Geophys. Res.: Space Phys. 120:715-1737, doi:10.1002/2014JA020778. 40. Schwenzer S. P., Greenwood R. C., Kelley S. P., Ott U., Tindle A. G., Haubold R., Herrmann S., Gibson J. M., Anand M., Hammond S., and Franchi I. A. 2013. Quantifying noble gas contamination during terrestrial alteration in Martian meteorites from Antarctica. Meteoritics & Planetary Science 48;6: 929-954. 41. Smith M. R. and Bandfield J. L. 2012. Geology of quartz and hydrated silica-bearing deposits near Antoniadi Crater, Mars. J. Geophys. Res. 117, E6 (30 June 2012), in press, doi: 10.1029/2011JE004038. 42. Squyres S. W., Arvidson R. E., Ruff S., Gellert R., Morris R. V., Ming D. W., Crumpler L., Farmer J. D., Des Marais D. J., Yen A., McLennan S. M., Calvin W., Bell J. F. 3rd, Clark B. C., Wang A., Mccoy T. J., Schmidt M. E., de Souza P. A. 2008. Detection of silica-rich deposits on Mars. Science 320:1063-1067, 10.1126/science.1155429. 43. Steinbrügge G., Stark A., Hussmann H., Sohl F., and Oberst J. 2015. Measuring tidal deformations by laser altimetry. A performance model for the Ganymede Laser Altimeter. Planetary and Space Science 117:184-191, doi:10.1016/j.pss.2015.06.013. 44. Tanigawa T., Maruta A., and Machida M. N., 2014. Accretion of Solid Materials onto Circumplanetary Disks from Protoplanetary Disks. The Astrophysical Journal 784:109, doi: 10.1088/0004-637X/784/2/109. 45. The Meteoritical Society, Meteoritical Bulletin Database (available at http://www.lpi.usra.edu/meteor/index.php). 46. Vance S., Bouffard M., Choukroun M., and Sotin C. 2014. Ganymede's internal structure including thermodynamics of magnesium sulfate oceans in contact with ice. Planet. Space Sci. 96:62-70. http://dx.doi.org/10.1016/j.pss.2014.03.011. 47. Wang S. J., Li S. G., and Liu S. A. 2013. The origin and evolution of low-δ18O magma recorded by multi-growth zircons in granite. Earth Planet. Sci. Lett. 373: 233-241, doi:10.1016/j.epsl.2013.05.009. 48. Williams R. M. E., Grotzinger J. P., Dietrich W. E., Gupta S., Sumner D. Y., Sumner D. Y., Wiens R. C., Mangold N., Malin M. C., Edgett K. S., Maurice S., Forni O., Gasnault O., Ollila A., Newsom H. E., Dromart G., Palucis M. C., Yingst R. A., Anderson R. B., Herkenhoff K. E., Le Mouélic S., Goetz W., Madsen M. B., Koefoed A., Jensen J. K., Bridges J. C., Schwenzer S. P., Lewis K. W., Stack K. M., Rubin D., Kah L. C., Bell J. F. III, Farmer J. D., Sullivan R., Van Beek T., Blaney D. L., Pariser O., Deen R. G., and MSL Science Team. 2013. Martian fluvial conglomerates at Gale Crater. Science 340:1068-1072, doi: 10.1126/science.1237317. 49. Ziegler K., Sharp Z. D., and Agee C. B. 2013. The unique NWA 7034 martian meteorite: evidence for multiple oxygen isotope reservoirs. 44th Lunar and Planetary Science Conference. 2639. (available at http://www.lpi.usra.edu/meetings/lpsc2013/pdf/2639.pdf).
1810.12172	1	1810	2018-10-29T14:58:44	The limit of the gyrochronology - A new age determination technique: the tidal-chronology	[ "astro-ph.EP", "astro-ph.SR" ]	While the number of detected planets is continuously increasing since 1995, their impact on their central star still remain poorly understood. Yet, the presence of a massive close-in planet can strongly modify the surface angular velocity evolution of the star. In these circumstances, age estimation techniques based on rotation, such as the gyrochronology, can't be applied to stars that experienced significant star-planet magnetic and tidal interactions during their evolution. With the use of a numerical model that combines an angular momentum evolution and an orbital evolution code, we investigated the evolution of initial distribution of star-planet systems, in which the orbital evolution of the planet is driven by the tidal dissipation formalism. Based on these initial distributions we highlighted the limits of applicability of the gyrochronology analysis and proposed a new age determination technique based on the observation of the couple $\rm P_{rot,\star}$-$a$, where $\rm P_{rot,\star}$ is the stellar rotational period and $a$ the planetary semi-major axis. This technique called tidal-chronology can be very helpful for planetary system composed of a star between 0.3 and 1.2 $\rm M_{\odot}$ and a planet more massive than 1 $\rm M_{jup}$ initially located at few hundredth au of the host star.	astro-ph.EP	astro-ph	SF2A 2018 P. Di Matteo, F. Billebaud, F. Herpin, N. Lagarde, J.-B. Marquette, A. Robin, O. Venot (eds) THE LIMIT OF THE GYROCHRONOLOGY - A NEW AGE DETERMINATION TECHNIQUE: THE TIDAL-CHRONOLOGY F. Gallet 1 and P. Delorme1 Abstract. While the number of detected planets is continuously increasing since 1995, their impact on their central star still remain poorly understood. Yet, the presence of a massive close-in planet can strongly modify the surface angular velocity evolution of the star. In these circumstances, age estimation techniques based on rotation, such as the gyrochronology, can't be applied to stars that experienced significant star-planet magnetic and tidal interactions during their evolution. With the use of a numerical model that combines an angular momentum evolution and an orbital evolution code, we investigated the evolution of initial distribution of star-planet systems, in which the orbital evolution of the planet is driven by the tidal dissipation formalism. Based on these initial distributions we highlighted the limits of applicability of the gyrochronology analysis and proposed a new age determination technique based on the observation of the couple Prot,(cid:63)-a, where Prot,(cid:63) is the stellar rotational period and a the planetary semi-major axis. This technique called tidal- chronology can be very helpful for planetary system composed of a star between 0.3 and 1.2 M(cid:12) and a planet more massive than 1 Mjup initially located at few hundredth au of the host star. Keywords: planet-star: interactions -- stars: evolution -- stars:rotation 1 Introduction Estimating the age of a planetary system is fundamental to constrain: 1) current planetary models (Ida & Lin 2008; Mordasini et al. 2009, 2012; Alibert et al. 2013); 2) star-planet interaction efficiency (Lanza et al. 2011); 3) internal structure and chemical composition of the star (based on stellar model); and 4) the star-planet system previous evolution (Gallet et al. 2017b). Empirical age determination techniques such as the gyrochronology (Barnes 2003) and the magnetochronology (Vi- dotto et al. 2014) are proposed in the literature. Both are based on the observation that during the main-sequence phase (hereafter MS) the evolution of the surface rotation and magnetic field of a star, for a given stellar mass, only depend on age. However, as long as the star departs from an isolated state, i.e. if a massive planet orbits around the star, these tech- niques can no longer be used since star-planet tidal interaction could have modified the evolution of the surface rotation rate along the system's evolution (see Gallet et al. 2018). As a consequence, the age of several planetary systems hosting a hot Jupiter could appear younger than they really are. We aim to provide the community with a new age estimation technique based on the modelling of the evolution of the star-planet tidal interaction and observation of the surface rotation rate of the host star and current location of the massive planet orbiting it. 2 Numerical model 2.1 Description In this work we combined the stellar rotational evolution model described in Gallet & Bouvier (2015) to the modified orbital evolution model used in Bolmont & Mathis (2016). The link between the two is done through the grid of tidal dissipation from Gallet et al. (2017a). The stellar structure is provided by the stellar evolution code STAREVOL (see Amard et al. 2016, and references therein). The aim of this coupling is to follow, through a realistic approach, the evolution of a given star-planet system. This code is specifically designed for stars between 0.3 (fully convective limit) and 1.2 M(cid:12), and for planetary systems composed of only one massive planet. 1 Univ. Grenoble Alpes, CNRS, IPAG, 38000 Grenoble, France c(cid:13) Soci´et´e Francaise d'Astronomie et d'Astrophysique (SF2A) 2018 8 1 0 2 t c O 9 2 . ] P E h p - o r t s a [ 1 v 2 7 1 2 1 . 0 1 8 1 : v i X r a 238 SF2A 2018 (a) Mp = 2 Mjup (b) Mp = 5 Mjup Fig. 1. In green, region in which the gyrochronology analysis can be used. Prot,(cid:63)(t) as a function of SMA(t) and stellar mass for a 2 (left) and 5 (right) Mjup mass planet. The color gradient corresponds to the rotational departure δProt = 1 − Prot,isol./Prot,(cid:63). Only systems for which the planet is still present are plotted. Using this meta-code, we computed a grid composed of 0.5, 0.7, and 1.0 M(cid:12) with initial rotational period between one and 11 days (using a parametrization that follows Gallet & Bouvier 2015; Gallet et al. 2018). We considered a 2 Mjup and a 5 Mjup mass planet initially located between 0.1 and 1.0 Rco, with the corotation radius Rco = (cid:32)GM(cid:63) (cid:33)1/3 Ω2 (cid:63) = GM(cid:63)P2 (2π)2 rot,(cid:63) 1/3 , where Prot,(cid:63) is the surface rotation period of the host star, G the gravitational constant, M(cid:63) the stellar mass, and Ω(cid:63) = 2π/Prot,(cid:63) the surface angular velocity of the host star. 2.2 The gyrochronology and its domain of applicability The gyrochronology technique was initially proposed by Barnes (2003) as a way to estimate the age of isolated stars. It is based on the behaviour of the surface rotation rate of the stars (between 0.3 and 1.0 M(cid:12)) that seems to evolves as t−1/2 during the MS phase (see Skumanich 1972). Figure 1 shows the evolution of Prot,(cid:63)(t) as a function of the semi-major axis (hereafter SMA), stellar mass (0.5, 0.7, and 1.0 M(cid:12)), and planetary mass (2 Mjup and 5 Mjup). The color gradient corresponds to the variation of δProt (±5%) for each couple [Prot,(cid:63)(t)− SMA(t)− M(cid:63)− Mp], where δProt = 1− Prot,isol./Prot,(cid:63) traces the departure of the rotational evolution of star in star-planet system to the rotation evolution of the same but isolated (i.e. without planet) star. Figure 1 displays the domain of validity of the gyrochronology analysis (the green parts). In this figure the reddest parts correspond to the region where the star is 5% faster (δProt ≥ 5%) than an isolated star, which corresponds, following the Skumanich relationship (Ω(cid:63) ∝ t−1/2, Skumanich 1972), to an error of about 10% on the age estimation using the gyrochronology analysis (Delorme et al. 2011). Increasing the planetary mass slightly extends this region toward lower SMA(t). If a massive close-in planet is detected orbiting its host star, then the gyrochronology analysis can't be applied if the planet is more massive than about 1 Mjup and located below 0.1 au. If no planets are detected but that there is a suspicion of planetary engulfment then the gyrochronology analysis can't be used. In some cases, the gyrochronology can't be applied even several Gyr after the engulfment (see Gallet & Delorme in prep.) as the footprint of this interaction survive a large fraction of the stellar rotational evolution history. 3 Tidal-chronology Following the work of Gallet et al. (2018) we developed a new age estimation technique based on the measurement of both the surface rotation rate of the star and location of the planet around it. It relies on the fact that the star-planet interaction produces a rotation cycle that can be used to estimate the age of a given close-in system. Since the rate of the evolution of the SMA strongly depends on its value (see Eqs. 3-6 from Gallet et al. 2018), a given observed couple Prot,(cid:63)(t) − SMA(t) is thus only produced at a small range of possible ages and initial conditions. 239 (a) Mp = 2 Mjup (b) Mp = 5 Mjup Fig. 2. Synthetic Prot,(cid:63)(t) and SMA(t) estimated for a system composed of a 0.5, 0.7, and 1.0 M(cid:12) star around which orbits a 2 and a 5 Mjup planet. Prot,init = 1-11 days (with ∆Prot,init = 0.2 days) and SMAinit = 0.1, 0.3, 0.4, 0.45, 0.5, 0.55, 0.7, 0.8, 0.92, 0.94, 0.96, 0.98, and 1.0 Rco (see Gallet et al. 2018). The color gradient depict the age (in Myr) at which the couple Prot,(cid:63)(t)-SMA(t) is extracted. Only planetary systems in which the planet is still orbiting the star are plotted. To determine the age of the system we computed a grid of star-planet system's evolution composed of stars with Prot,init between one and 11 days and planetary initial SMA between 0.1 and 1.0 Rco for a given stellar and planetary mass that match the observed system's properties. We then explore the grid so as to extract at which age the observed couple Prot,obs − SMAobs is retrieved. We finally estimate the departure of the observed couple Prot,obs − SMAobs to each of the Prot,(cid:63)-SMA(cid:63) couples from the grid using this chosen expression: S 2 = (Prot,(cid:63) − Prot,obs)2 P2 rot,obs + (SMA(cid:63) − SMAobs)2 SMA2 obs . We also applied a 3D interpolation method using the Python-SciPy griddata routine (to interpolate unstructured 3- dimensional data, see Jones et al. 2001). To investigate the degeneracies of this technique we considered a 0.5, 0.7, and 1.0 M(cid:12) mass star and a 2 and 5 Mjup mass planet and ran a grid constituted of Prot,init= 1-11 days (with ∆Prot,init = 0.2 days) and SMAinit between 0.1 and 1.0 Rco (see Gallet et al. 2018). Figure 2 shows Prot,(cid:63)(t) as a function of SMA(t), M(cid:63), and Mp, and displayed the age as a color gradient. It shows that the solutions are degenerated when using only Prot,(cid:63) or SMA, but that these degeneracies are lifted when using both quantities simultaneously. Additionally, the information about the fact that the planet is still (or not) orbiting the star adds another criteria and helps the lift of these degeneracies. In the case of migrating planets, the evolution of star-planet systems in the Prot,(cid:63)(t)-SMA(t) space depends on their initial conditions. As the star evolves, its surface rotation rate is impacted by the inward migration of the planet. The star-planet system thus moves toward smaller SMA and rotation period (if the acceleration torque produced by the planet is stronger than the breaking torque of the stellar winds). We applied our technique on the planetary system WAPS-43, a 0.7 M(cid:12) K7V stars around which orbits a 2 Mjup mass planet (Hellier et al. 2011). The rotation period of the star is estimated at 15.6 ± 0.4 days and the planet is observed at a distance of 0.01526 au from the star. For this system we expect to get a small range of possible age and the fact that the planet is observed indicates that no engulfment has occurred, which discard the models in which the planet is engulfed by the star. 240 SF2A 2018 Using the gyrochronology technique the age of WASP-43 is estimated at 400 Myr (Hellier et al. 2011). The most probable solutions of our technique aim towards a more advanced system with an age between 3 and 6 Gyr. A 3D linear interpolation of the observed couple leads to an age estimation of 4.9+0.3−0.4 Gyr, which is consistent with our S 2 exploration. With this technique we can also constrain the initial condition of the WASP-43 system. According to our simulations the initial rotation rate of the star was between 8 and 11 days (the slow part of the initial observed rotational distribution) and the planet was initially located between 0.025 and 0.03 au. 4 Conclusions In massive close-in planetary systems, the tidal interaction between the central star and the planet is expected to have strongly modified the evolution of the surface rotation rate of the host star. In that case, it is no longer possible to use age determination techniques based on stellar surface rotation and magnetic field. The corresponding forbidden region is located quite close from the stellar surface (few hundredth of au) and tends to expend for increasing stellar and planetary mass. To overcome this issue, we proposed a new age determination techniques that can be applied to such massive close-in planetary systems: the tidal-chronology. This technique is based on the uniqueness of the path followed by a planetary system on the Prot,(cid:63)(t)-SMA(t) plane. However, the numerical and physical description of the tidal dissipation in stellar and planetary interior is currently not good enough to only use this age estimation, which should be considered with caution and in combination with other age determination techniques (as any empirical age determination techniques). Indeed, in this work we do not included the dissipation in the planet (that is still badly theoretically constrained) and the magnetic star-planet interaction (see Strugarek et al. 2017). The dissipation inside of the stellar radiative core is also currently not physically described and hence not numerically included. Nevertheless, we developed a promising technique that will benefit the community when all aspect of tidal and magnetic star-planet interactions will be included in angular momentum evolution models. These dissipations could produce an additional torque that could increase the rate of the planetary migration, and consequently reduce the estimated age of an observed planetary system. The impact of the planetary companion on the rotational evolution of the host stars should increase with the inclusion of the neglected effects mentioned above. Hence, the take-home message of our work is thus to be careful and aware of the limits of the gyrochronology analysis when using it to estimate the age of planetary systems. We thank the SF2A 2018 organizers as well as Roxanne Ligi, Yveline Lebreton, Tristan Guillot, and Magali Deleuil for the organisation of the S17 session. F.G acknowledges financial support from the CNES fellowship. References Alibert, Y., Carron, F., Fortier, A., et al. 2013, A&A, 558, A109 Amard, L., Palacios, A., Charbonnel, C., Gallet, F., & Bouvier, J. 2016, A&A, 587, A105 Barnes, S. A. 2003, ApJ, 586, 464 Bolmont, E. & Mathis, S. 2016, Celestial Mechanics and Dynamical Astronomy, 126, 275 Delorme, P., Collier Cameron, A., Hebb, L., et al. 2011, MNRAS, 413, 2218 Gallet, F., Bolmont, E., Bouvier, J., Mathis, S., & Charbonnel, C. 2018, ArXiv e-prints Gallet, F., Bolmont, E., Mathis, S., Charbonnel, C., & Amard, L. 2017a, A&A, 604, A112 Gallet, F. & Bouvier, J. 2015, A&A, 577, A98 Gallet, F., Charbonnel, C., Amard, L., et al. 2017b, A&A, 597, A14 Hellier, C., Anderson, D. R., Collier Cameron, A., et al. 2011, A&A, 535, L7 Ida, S. & Lin, D. N. C. 2008, ApJ, 685, 584 Jones, E., Oliphant, T., Peterson, P., et al. 2001, SciPy: Open source scientific tools for Python Lanza, A. F., Damiani, C., & Gandolfi, D. 2011, A&A, 529, A50 Mordasini, C., Alibert, Y., Benz, W., Klahr, H., & Henning, T. 2012, A&A, 541, A97 Mordasini, C., Alibert, Y., Benz, W., & Naef, D. 2009, A&A, 501, 1161 Skumanich, A. 1972, ApJ, 171, 565 Strugarek, A., Bolmont, E., Mathis, S., et al. 2017, ApJ, 847, L16 Vidotto, A. A., Gregory, S. G., Jardine, M., et al. 2014, MNRAS, 441, 2361
1511.08640	1	1511	2015-11-27T12:14:23	Asteroids@home - A BOINC distributed computing project for asteroid shape reconstruction	[ "astro-ph.EP", "astro-ph.IM" ]	We present the project Asteroids@home that uses distributed computing to solve the time-consuming inverse problem of shape reconstruction of asteroids. The project uses the Berkeley Open Infrastructure for Network Computing (BOINC) framework to distribute, collect, and validate small computational units that are solved independently at individual computers of volunteers connected to the project. Shapes, rotational periods, and orientations of the spin axes of asteroids are reconstructed from their disk-integrated photometry by the lightcurve inversion method.	astro-ph.EP	astro-ph	Asteroids@home – A BOINC distributed computing project for asteroid shape reconstruction Josef Durecha, Josef Hanusb, Radim Vancoc aAstronomical Institute, Faculty of Mathematics and Physics, Charles University in Prague, V Holesovick´ach 2, 180 00 Prague 8, Czech bUNS-CNRS-Observatoire de la Cote d'Azur, BP 4229, 06304 Nice Cedex 4, France cCzech National Team Republic 5 1 0 2 v o N 7 2 . ] P E h p - o r t s a [ 1 v 0 4 6 8 0 . 1 1 5 1 : v i X r a Abstract We present the project Asteroids@home that uses distributed computing to solve the time-consuming inverse problem of shape reconstruction of asteroids. The project uses the Berkeley Open Infrastructure for Network Computing (BOINC) framework to distribute, collect, and validate small computational units that are solved independently at individual computers of volunteers connected to the project. Shapes, rotational periods, and orientations of the spin axes of asteroids are reconstructed from their disk-integrated photometry by the lightcurve inversion method. Keywords: asteroids, distributed computing, BOINC 1. Introduction 2. Asteroid lightcurve inversion With more than 500,000 discovered objects, asteroids form a large population of small bodies in the solar sys- tem that was affected by all processes that were acting during the formation and evolution of the solar system. By studying asteroids, we can reveal the history and cur- rent state of our cosmic neighborhood. In general, study of asteroids can be done either in situ by spacecrafts or by remote-sensing techniques. In situ measurements are lim- ited to only a very small sample of all asteroids that are direct targets of spacecraft missions or fly-by opportuni- ties. However, remote-sensing techniques are, in principle, feasible for most of the known population. Although the level of information we are able to get from remote sensing is inevitably much lower than from a detailed spacecraft mission, the basic physical properties can be successfully obtained for a substantial part of the asteroid population. One of the main sources of information about aster- oid shapes and spin states (i.e., rotational periods and spin axis directions) is their disk-integrated photometry that is available for all known asteroids. Because as- teroids have irregular shapes and rotate, the amount of light reflected towards the observer at Earth varies with asteroid's rotation – we observe a lightcurve. An ef- fective method to reconstruct asteroid shapes and spin states from their lightcurves (so called lightcurve in- version) has been developed by Kaasalainen and Torppa (2001); Kaasalainen et al. (2001). In the following sections, we describe our project of dis- tributed computing named Asteroids@home that is aimed for solving the lightcurve inversion problem for a substan- tial part of the asteroid population. A method for lightcurve inversion was developed by Kaasalainen et al. (2001). It uses all available photometric data to reconstruct a convex shape model of an asteroid (together with its sidereal rotation period and the spin axis direction) that provides the best fit to the data. The review of the method can be found in Kaasalainen et al. (2002). Its mathematical stability and uniqueness was proved by Lamberg and Kaasalainen (2001) and its re- sults were independently confirmed by disk-resolved im- ages (Marchis et al., 2006; Hanus et al., 2013a), stellar oc- cultation data ( Durech et al., 2011), or spacecraft images (Keller et al., 2010). Although real asteroids have in general complex shapes with (sometimes large) concavities, their lightcurves can be successfully reproduced with convex shapes. Noncon- vex models are an alternative to convex ones, but they lack the mathematical uniqueness and stability and in practice they are needed only when high-quality lightcurves ob- served at high solar phase angles or disk-resolved data are available. In this sense, convex models should be taken as approximations to the real shapes of asteroids. Since the publication of the method, models of about 500 asteroids were derived by this technique ( Durech et al., 2009; Durech et al., 2011; Hanus et al., 2011, 2013b; Marciniak et al., 2011, for example); most of them are publicly available in the Database of Asteroid Models from Inversion Techniques (DAMIT1, Durech et al., 2010). At this site, the source codes for the lightcurve inversion called convexinv can be down- loaded. This code written in the c programming language 1http://astro.troja.mff.cuni.cz/projects/asteroids3D Preprint submitted to Astronomy and Computing July 11, 2018 is a default version of the shape optimization and is widely used by the scientific community. Only minor changes re- lated to input/output file format were necessary to satisfy the BOINC specifications. Recently, the convexinv pro- gram has been also modified to (i) deal with disk-resolved data and nonconvex features of the shape (Carry et al., 2012), (ii) changing rotation rate to model the YORP ef- fect ( Durech et al., 2008), or (iii) excited rotation state (Kaasalainen, 2001; Pravec et al., 2014). The lightcurve inversion code uses the Levenberg- Marquardt algorithm to converge to a local minimum in χ2, where χ2 is a usual measure of difference between the observed and modeled brightness L taking into account the errors σ: χ2 =Xi L(i) obs − L(i) σi model !2 . (1) For each epoch corresponding to the i-th observation, the brightness L(i) model is computed as a sum of contributions from surface elements that are illuminated by Sun and seen by the observer. The orientation of the model in inertial space is determined by the direction of the spin axis given in ecliptic longitude λ and latitude β, and the rotation period P . The shape is represented as convex polyhedron and parametrized by spherical harmonics. The coefficients of spherical harmonics series are optimized together with the spin parameters to get the lowest value of χ2 (for more details see Kaasalainen et al. (2002)). Depending on the resolution of the shape (degree and order of the spherical harmonics series), the number of parameters to be opti- mized is typically between 20 and 90. To find the global minimum, however, we need to start at many initial points in the parameter space to go through all relevant local minima and then select the solution with the lowest χ2. Convergence of the shape towards the lo- cal minimum is robust, so the initial shape can be al- ways a sphere, for example, while the initial spin and pe- riod parameters have to cover the whole parameter sub- space. The number of initial pole directions is usually ten (isotropically distributed in ecliptic coordinates), which is enough for a safe convergence into the global minimum in the spin subspace. What makes the problem time con- suming is a large number of closely packed local minima in the period subspace. The local minima are separated by about 0.5P 2/∆T , where P is the rotation period and ∆T is the length of the time interval covered by observa- tions (Kaasalainen, 2001). For a typical set of lightcurves sufficient for inversion (tens of lightcurves observed dur- ing several apparitions), the rotation period can be es- timated very accurately without any modeling from the period analysis of the signal. Then the interval of periods that has to be searched for the global minimum is nar- row and the process is fast. However, with the data that are sparse in time with respect to the rotation period, the classical Fourier-based or phase dispersion minimization methods cannot be used and the rotation period cannot 3 2 1 s e n t h g i r b e v i t l a e R 0 1990 1995 2000 Date of observation 2005 2010 2015 Figure 1: Sparse-in-time photometry of asteroid (243) Ida down- loaded from AstDyS. There are 700 individual photometric points observed during 17 apparitions. The brightness was reduced to unit distance from Earth and Sun. be easily estimated from the lightcurve data. The sparse- in-time photometry is typically provided by all-sky astro- metric surveys (Catalina, Pan-STARRS, Gaia satellite, for example). With a typical rotation period of several hours and ∼15 years of observations, we have to test hundreds of thousands initial periods to be sure not to miss the global minimum. As an example of sparse-in-time photometry and a corre- sponding model, we present results for asteroid (243) Ida. The photometry obtained by sky surveys and downloaded from AstDyS2 site is shown in Fig. 1. The brightness was reduced to the unit distance from Earth and Sun. The groups of points correspond to individual apparitions. The periodogram is shown in Fig. 2, it consists of about 240,000 points. There are two clear minima with the low- est χ2 at 4.634 and 2.317 h. The best-fit model corre- sponds to the period 4.634 h and is shown in Fig. 3. The model is compared with the real shape of Ida as recon- structed from the fly-by images obtained by Galileo probe (Thomas et al., 1996). The pole direction from sparse data is (λ, β) = (255 ± 4◦, −59 ± 5◦) in J2000.0 ecliptic coordi- nates, which is not far from the value (263◦, −67◦) derived by Davies et al. (1996). The difference between these two poles is 9◦ of arc, which is within a typical uncertainty of the pole direction expected from models based on sparse data. The period P = 4.633631 ± 0.000005h is close to the value 4.633638 ± 0.000002 h of Kaasalainen et al. (2001) based on a set of 40 lightcurves from 1988 to 1993. Fig. 3 also illustrates a typical accuracy of the shapes derived from sparse photometry: they are only a rough approxi- mation of the real (in general unknown) shape and can- not provide any surface details, but rather only global characteristics. They can be further refined with dense lightcurves or other complementary data. The relative ac- curacy of the period determination is however high, de- pending mainly on the length of the interval of observa- tions. A typical uncertainty of the pole direction is 10–20◦, depending mainly on the number of brightness measure- ments, their photometric accuracy, and the number of ap- paritions. 2http://hamilton.dm.unipi.it/astdys 2 The time needed to scan the period interval for one aster- oid with hundreds of measurements covering ten years or more is of the order of weeks at 1 CPU. It means that the processing of data for hundreds of thousands of known as- teroids requires huge computational power. Therefore we set up a distributed computing project Asteroids@home3 that uses idle time of tens of thousands computers of vol- unteers. 4. BOINC The project is built on the Berkeley Open Infrastruc- ture for Network Computing (BOINC) framework, which is major infrastructure for volunteer computing based on the SETI@home project (Korpela, 2012). The BOINC server not only distributes unit tasks, collects and vali- dates results, but also offers community-based tools like forums, credits system, user of the day acknowledgment, etc. The principles of BOINC and the parts of BOINC server and client are described in Korpela (2012) or Vinsen and Thilker (2013). Here we cover only the setup specific for our project. The period search interval for each asteroid is divided into few hundred smaller intervals. The number of inter- vals for a given asteroid depends on the number of data points, the length of observations, and the resolution of the shape model. It is set up such that the CPU time needed to process one interval is constant. Each compu- tational unit contains initial values for the optimization, interval of periods, and the data. The data consist of time- brightness pairs and the geometry of observation: vectors towards Sun and the observer centered to the asteroid. Typically, the size of the input data file is less than 100 kB. At volunteer's side, the lightcurve inversion gradient-based optimization is run. The χ2 values for each trial period are stored and sent back to the server. The typical size of the file sent back to the server is about 1 MB or less. Each unit is sent to at least two volunteers and the results are compared and validated. The limit for delivering the unit is set to 14 days, an average time needed for finishing one unit is about three hours but can vary a lot across devices (hundreds of hours on mobile devices, for example). After finishing all units for a given asteroid, the periodogram is analyzed outside the BOINC platform and the best-fitting model and its uniqueness is checked. Although the best- fitting period and the corresponding shape model always exist, it is only rarely unique/prominent because of the low photometric quality of the currently available data. To date (August 2015) ∼60,000 volunteers with ∼100,000 computers contributed to the project. The number of volunteers steadily increases with hundreds of new users each day. The current computing power is ∼150 TFLOPs. The application is compiled for all main operating systems: 32- or 64-bit Windows and Linux, Mac 3http://asteroidsathome.net 3 Figure 2: Period search results for asteroid (243) Ida for sparse pho- tometry (Fig. 1). Each point corresponds to the local minimum in the parameter space. The lowest χ2 corresponds to the best-fit model for period P = 4.633631 h. Figure 3: Comparison of the convex model of (243) Ida reconstructed from sparse photometry (top) with the detailed shape models re- constructed from fly-by images obtained by Galileo probe (bottom, Thomas et al., 1996). 3. Distributed computing Because the problem of finding the global minimum in the parameter space consists of running the gradient opti- mization many times with different initial values and each run is independent of the others, the task is ideal for par- allelization – it is a so called embarrassingly parallel prob- lem. This approach of scanning a range of periods is safe in the sense that we are sure that the best-fitting model is indeed the global minimum in the parameter space, be- cause the convergence in the pole direction and shape is robust. An example of another approach to finding the global minimum, which will be implemented in the Gaia processing software (Cellino et al., 2009), is the use of ge- netic algorithms. The default interval of periods we scan is usually 2– 100 hours. The lower limit roughly corresponds to the spin barrier of asteroids larger than ∼200 m (Pravec et al., 2002), the upper limit was chosen arbitrarily – the fraction of asteroids rotating so slowly is small and they are often in the excited rotation state that cannot be described by our simple model. These limits can be changed if needed. OS, and FreeBSD. It can run also on mobile devices with Android and Raspberry Pi. A version for GPUs can run on CUDA 5.5 or higher. The code was optimized for speed, to further improve its performance, versions using SSE2/3 and AVX instructions sets are available. 5. Data and Results As the source of photometric data, we currently use the so called Lowell Photometric Database (Bowell et al., 2014). It corresponds to the photometry reported to the Minor Planet Center (MPC), but the systematic shifts be- tween the observatories were removed. Typically, the data set for one asteroid consists of hundreds of brightness mea- surements in V band covering several years. The accuracy of the photometry is about 0.15-0.20 mag. It means that in most cases, the signal is drowned by the level of random noise and systematic errors and there are many indistin- guishable best-fitting solutions of the inverse problem. So far, we have processed about 100,000 asteroids out of 330,000 in the Lowell database. The preliminary results are posted on the web page of the project,4 where the de- rived rotation period, the pole orientation, and an image of the shape model are shown. An important feedback for the volunteers is also the information about who was "lucky enough" to process the period interval that turned out to contain the global minimum. Because each unit is com- puted at least two times and then validated, there are two or more users assigned to each result. After further vali- dation of reliability of solutions and their comparison with independent data – periods from the Asteroid Lightcurve Database Warner et al. (2009), for example – the models will be published in a peer-review journal and stored in the DAMIT database together with the data, shape files, and all other relevant information. Although we have already processed tens of thousands of asteroids, the number of successfully derived models is low. This is because the quality of the data and their number is low, thus no unique period could be derived. A typical periodogram then does not have one or two clear minima like in the case of Ida in Fig. 2, but has many periods and corresponding models that fit the data to the same χ2 level. In such cases, only additional photometric data can help to derive a unique model. 6. Future Processing all data from the Lowell Photometric Database is the first step on the way to reconstruct the distribution of physical properties of the asteroid popula- tion. Because the success rate is very low with these data, a crucial part of interpreting the results will be the proper de-biasing. For this, we will run an extensive test with synthetic data produced by realistic asteroid shapes and scattering and random distribution of spins and periods. We also plan to combine Lowell photometry with dense lightcurves available at the Minor Planet Center Aster- oid Light Curve Database,5 with future Gaia photometry, and with any photometry from other future surveys. Cur- rently we are also computing shape models from all avail- able photometry for several tens of asteroids with future mass estimates based on Gaia astrometry. Another chal- lenge will be the combined inversion of photometric and thermal infrared data ( Durech et al., 2012). Acknowledgments The success of Asteroids@home project would not be possible without tens of thousand of volunteers who pro- vided computing resources of their computers. We greatly appreciate their contribution and patience. We also ap- preciate the help and support of enthusiastic members of the Czech National Team, who tested the application and significantly contributed to the optimization of the code. The work of J D was supported by the research grant 15- 04816S of the Czech Science Foundation. JH is supported by the Centre national d'´etudes spatiales (CNES) post- doctoral fellowship. References Bowell, E., Oszkiewicz, D.A., Wasserman, L.H., Muinonen, K., Pent- tila, A., Trilling, D.E., 2014. Asteroid spin-axis longitudes from the Lowell Observatory database. Meteoritics and Planetary Sci- ence 49, 95–102. doi:10.1111/maps.12230, arXiv:1310.3617. Carry, B., Kaasalainen, M., Merline, W.J., Muller, T.G., Jorda, L., Drummond, J.D., Berthier, J., O'Rourke, L., Durech, J., Kuppers, M., Conrad, A., Tamblyn, P., Dumas, C., Sierks, H., Osiris Team, 2012. Shape modeling technique KOALA validated by ESA Rosetta at (21) Lutetia. Planetary and Space Science 66, 200–212. doi:10.1016/j.pss.2011.12.018 , arXiv:1112.5944. Cellino, A., Hestroffer, D., Tanga, P., Mottola, S., Dell'Oro, A., 2009. Genetic inversion of sparse disk-integrated photometric data of asteroids: application to Hipparcos data. Astronomy and Astro- physics 506, 935–954. doi:10.1051/0004-6361/200912134. Davies, M.E., Colvin, T.R., Belton, M.J.S., Veverka, J., Thomas, the North Pole and the Icarus 120, 33–37. P.C., 1996. Control Network of Asteroid 243 Ida. doi:10.1006/icar.1996.0034. The Direction of Durech, J., Vokrouhlick´y, D., Kaasalainen, M., Weissman, P., Lowry, S.C., Beshore, E., Higgins, D., Krugly, Y.N., Shevchenko, V.G., Gaftonyuk, N.M., Choi, Y.J., Kowalski, R.A., Larson, S., Warner, B.D., Marshalkina, A.L., Ibrahimov, M.A., Molotov, I.E., Micha lowski, T., Kitazato, K., 2008. New photometric observa- tions of asteroids (1862) Apollo and (25143) Itokawa - an analy- sis of YORP effect. Astronomy and Astrophysics 488, 345–350. doi:10.1051/0004-6361:200809663 . Durech, J., Kaasalainen, M., Warner, B.D., Fauerbach, M., Marks, S.A., Fauvaud, S., Fauvaud, M., Vugnon, J.M., Pilcher, F., Bernasconi, L., Behrend, R., 2009. Asteroid models from com- bined sparse and dense photometric data. Astronomy and Astro- physics 493, 291–297. doi:10.1051/0004-6361:200810393. Durech, J., Sidorin, V., Kaasalainen, M., 2010. DAMIT: a database of asteroid models. Astronomy and Astrophysics 513, A46+. doi:10.1051/0004-6361/200912693 . 4http://asteroidsathome.net/scientific_results.html light_curve.php 5http://www.minorplanetcenter.net/light_curve2/ 4 Schroder, S.E., Spjuth, S., Vernazza, P., 2010. E-Type asteroid (2867) Steins as imaged by OSIRIS on board Rosetta. Science 327, 190–193. doi:10.1126/science.1179559. Korpela, E.J., 2012. SETI@home, BOINC, and Volunteer Dis- tributed Computing. Annual Review of Earth and Planetary Sci- ences 40, 69–87. doi:10.1146/annurev-earth-040809-152348. Lamberg, L., Kaasalainen, M., 2001. Numerical solution of the J. Comp. Appl. Math. 137, 213–227. Minkowski problem. doi:10.1016/S0377-0427(01)00360-0. Marchis, F., Kaasalainen, M., Hom, E.F.Y., Berthier, J., Enriquez, J., Hestroffer, D., Le Mignant, D., de Pater, I., 2006. Shape, size and multiplicity of main-belt asteroids. Icarus 185, 39–63. doi:10.1016/j.icarus.2006.06.001 . Marciniak, A., Micha lowski, T., Poli´nska, M., Bartczak, P., Hirsch, R., Sobkowiak, K., Kami´nski, K., Fagas, M., Behrend, R., Bernasconi, L., Bosch, J., Brunetto, L., Choisay, F., Coloma, J., Conjat, M., Farroni, G., Manzini, F., Pallares, H., Roy, R., Kwiatkowski, T., Kryszczy´nska, A., Rudawska, R., Star- czewski, S., Micha lowski, J., Ludick, P., 2011. Photome- try and models of selected main belt asteroids. VIII. Low- pole asteroids. Astronomy and Astrophysics 529, A107+. doi:10.1051/0004-6361/201015365 . Pravec, P., Harris, A.W., Micha lowski, T., 2002. Asteroid rotations, in: Bottke, W.F., Cellino, A., Paolicchi, P., Binzel, R.P. (Eds.), Asteroids III. University of Arizona Press, Tucson, pp. 113–122. Pravec, P., Scheirich, P., Durech, J., Pollock, J., Kusnir´ak, P., Hornoch, K., Gal´ad, A., Vokrouhlick´y, D., Harris, A.W., Jehin, E., Manfroid, J., Opitom, C., Gillon, M., Colas, F., Oey, J., Vrastil, J., Reichart, D., Ivarsen, K., Haislip, J., LaCluyze, A., 2014. The tumbling spin state of (99942) Apophis. Icarus 233, 48–60. doi:10.1016/j.icarus.2014.01.026. Thomas, P.C., Belton, M.J.S., Carcich, B., Chapman, C.R., Davies, M.E., Sullivan, R., Veverka, J., 1996. The Shape of Ida. Icarus 120, 20–32. doi:10.1006/icar.1996.0033. Vinsen, K., Thilker, D., 2013. A BOINC based, citizen-science project for pixel spectral energy distribution fitting of resolved galaxies in multi-wavelength surveys. Astronomy and Computing 3, 1–12. doi:10.1016/j.ascom.2013.10.001, arXiv:1306.1618. Warner, B.D., Harris, A.W., Pravec, P., asteroid doi:10.1016/j.icarus.2009.02.003 . lightcurve database. Icarus 2009. 202, The 134–146. Durech, J., Kaasalainen, M., Herald, D., Dunham, D., Timerson, B., Hanus, J., Frappa, E., Talbot, J., Hayamizu, T., Warner, B.D., Pilcher, F., Gal´ad, A., 2011. Combining asteroid models derived by lightcurve inversion with asteroidal occultation silhou- ettes. Icarus 214, 652–670. doi:10.1016/j.icarus.2011.03.016, arXiv:1104.4227. Durech, J., Delbo, M., Carry, B., 2012. Asteroid models derived from thermal infrared data and optical lightcurves. LPI Contributions 1667, 6118. Hanus, J., Marchis, F., Durech, J., 2013a. asteroids belt adaptive doi:10.1016/j.icarus.2013.07.023. combining observations. optics by shape models 226, Icarus Sizes of main- and Keck 1045–1057. Hanus, J., Durech, J., Broz, M., Marciniak, A., Warner, B.D., Pilcher, F., Stephens, R., Behrend, R., Carry, B., Capek, D., Antonini, P., Audejean, M., Augustesen, K., Barbotin, E., Bau- douin, P., Bayol, A., Bernasconi, L., Borczyk, W., Bosch, J.G., Brochard, E., Brunetto, L., Casulli, S., Cazenave, A., Charbon- nel, S., Christophe, B., Colas, F., Coloma, J., Conjat, M., Cooney, W., Correira, H., Cotrez, V., Coupier, A., Crippa, R., Cristo- fanelli, M., Dalmas, C., Danavaro, C., Demeautis, C., Droege, T., Durkee, R., Esseiva, N., Esteban, M., Fagas, M., Farroni, G., Fauvaud, M., Fauvaud, S., Del Freo, F., Garcia, L., Geier, S., Godon, C., Grangeon, K., Hamanowa, H., Hamanowa, H., Heck, N., Hellmich, S., Higgins, D., Hirsch, R., Husarik, M., Itkonen, T., Jade, O., Kami´nski, K., Kankiewicz, P., Klotz, A., Koff, R.A., Kryszczy´nska, A., Kwiatkowski, T., Laffont, A., Leroy, A., Lecacheux, J., Leonie, Y., Leyrat, C., Manzini, F., Martin, A., Masi, G., Matter, D., Micha lowski, J., Micha lowski, M.J., Micha lowski, T., Michelet, J., Michelsen, R., Morelle, E., Mottola, S., Naves, R., Nomen, J., Oey, J., Og loza, W., Oksa- nen, A., Oszkiewicz, D., Paakkonen, P., Paiella, M., Pallares, H., Paulo, J., Pavic, M., Payet, B., Poli´nska, M., Polishook, D., Poncy, R., Revaz, Y., Rinner, C., Rocca, M., Roche, A., Romeuf, D., Roy, R., Saguin, H., Salom, P.A., Sanchez, S., Santa- cana, G., Santana-Ros, T., Sareyan, J.P., Sobkowiak, K., Sposetti, S., Starkey, D., Stoss, R., Strajnic, J., Teng, J.P., Tr´egon, B., Vagnozzi, A., Velichko, F.P., Waelchli, N., Wagrez, K., Wucher, H., 2013b. Asteroids' physical models from combined dense and sparse photometry and scaling of the YORP effect by the ob- served obliquity distribution. Astronomy and Astrophysics 551, A67. doi:10.1051/0004-6361/201220701, arXiv:1301.6943. Hanus, J., Durech, J., Broz, M., Warner, B.D., Pilcher, F., Stephens, R., Oey, J., Bernasconi, L., Casulli, S., Behrend, R., Polishook, D., Henych, T., Lehk´y, M., Yoshida, F., Ito, T., 2011. A study of asteroid pole-latitude distribution based on an extended set of shape models derived by the lightcurve inversion method. Astronomy and Astrophysics 530, A134. doi:10.1051/0004-6361/201116738, arXiv:1104.4114. Kaasalainen, M., 2001. lightcurves of pre- cessing asteroids. Astronomy and Astrophysics 376, 302–309. doi:10.1051/0004-6361:20010935 . Interpretation of Kaasalainen, M., Mottola, S., Fulchignomi, M., 2002. Asteroid models from disk-integrated data, in: Bottke, W.F., Cellino, A., Paolicchi, P., Binzel, R.P. (Eds.), Asteroids III. University of Ari- zona Press, Tucson, pp. 139–150. Kaasalainen, M., Torppa, J., 2001. Optimization Methods for As- teroid Lightcurve Inversion. I. Shape Determination. Icarus 153, 24–36. doi:10.1006/icar.2001.6673. Kaasalainen, M., Torppa, J., Muinonen, K., 2001. Optimization Methods for Asteroid Lightcurve Inversion. II. The Complete In- verse Problem. Icarus 153, 37–51. doi:10.1006/icar.2001.6674. Keller, H.U., Barbieri, C., Koschny, D., Lamy, P., Rickman, H., Rodrigo, R., Sierks, H., A'Hearn, M.F., Angrilli, F., Barucci, M.A., Bertaux, J., Cremonese, G., Da Deppo, V., Davidsson, B., De Cecco, M., Debei, S., Fornasier, S., Fulle, M., Groussin, O., Gutierrez, P.J., Hviid, S.F., Ip, W., Jorda, L., Knollenberg, J., Kramm, J.R., Kuhrt, E., Kuppers, M., Lara, L., Lazzarin, M., Moreno, J.L., Marzari, F., Michalik, H., Naletto, G., Sabau, L., Thomas, N., Wenzel, K., Bertini, I., Besse, S., Ferri, F., Kaasalainen, M., Lowry, S., Marchi, S., Mottola, S., Sabolo, W., 5
1303.3336	1	1303	2013-03-14T03:07:34	Observing Strategies for the Detection of Jupiter Analogs	[ "astro-ph.EP" ]	To understand the frequency, and thus the formation and evolution, of planetary systems like our own solar system, it is critical to detect Jupiter-like planets in Jupiter-like orbits. For long-term radial-velocity monitoring, it is useful to estimate the observational effort required to reliably detect such objects, particularly in light of severe competition for limited telescope time. We perform detailed simulations of observational campaigns, maximizing the realism of the sampling of a set of simulated observations. We then compute the detection limits for each campaign to quantify the effect of increasing the number of observational epochs and varying their time coverage. We show that once there is sufficient time baseline to detect a given orbital period, it becomes less effective to add further time coverage -- rather, the detectability of a planet scales roughly as the square root of the number of observations, independently of the number of orbital cycles included in the data string. We also show that no noise floor is reached, with a continuing improvement in detectability at the maximum number of observations N=500 tested here.	astro-ph.EP	astro-ph	Observing Strategies for the Detection of Jupiter Analogs Robert A. Wittenmyer1, C.G. Tinney1, J. Horner1, R.P. Butler2, H.R.A. Jones3, S.J. O'Toole4, J. Bailey1, B.D. Carter5, G.S. Salter1, D. Wright1 [email protected] ABSTRACT To understand the frequency, and thus the formation and evolution, of plane- tary systems like our own solar system, it is critical to detect Jupiter-like planets in Jupiter-like orbits. For long-term radial-velocity monitoring, it is useful to estimate the observational effort required to reliably detect such objects, par- ticularly in light of severe competition for limited telescope time. We perform detailed simulations of observational campaigns, maximizing the realism of the sampling of a set of simulated observations. We then compute the detection limits for each campaign to quantify the effect of increasing the number of ob- servational epochs and varying their time coverage. We show that once there is sufficient time baseline to detect a given orbital period, it becomes less effective to add further time coverage -- rather, the detectability of a planet scales roughly as the square root of the number of observations, independently of the number of orbital cycles included in the data string. We also show that no noise floor is reached, with a continuing improvement in detectability at the maximum number of observations N = 500 tested here. Subject headings: planetary systems -- techniques: radial velocities 1Department of Astrophysics, School of Physics, Faculty of Science, The University of New South Wales, Sydney, NSW 2052, Australia 2Department of Terrestrial Magnetism, Carnegie Institution of Washington, 5241 Broad Branch Road, NW, Washington, DC 20015-1305, USA 3University of Hertfordshire, Centre for Astrophysics Research, Science and Technology Research Insti- tute, College Lane, AL10 9AB, Hatfield, UK 4Australian Astronomical Observatory, PO Box 915, North Ryde, NSW 1670, Australia 5Faculty of Sciences, University of Southern Queensland, Toowoomba, Queensland 4350, Australia -- 2 -- 1. Introduction As planet search efforts mature, techniques improve, and ever-larger data sets are ob- tained, the detection of planets like our own Jupiter is becoming more tractable. These "Jupiter analogs" are critically important for understanding the frequency of planetary sys- tems with architectures similar to our own solar system (e.g. Cumming et al. 2008; Gould et al. 2010; Wittenmyer et al. 2011). Jupiter analogs are interesting and important for a number of reasons. Firstly, most exoplanetary systems found to date are markedly different from our own Solar system, featuring Jupiter-mass planets on short-period orbits more reminiscent of the terrestrial planets. While such systems are fascinating in their own right, it is clearly important to search for systems that more closely resemble our own. This is particularly true looking forward, when we consider the ongoing efforts of planet-search programs to find Earth-like planets that could potentially host life. Over the years, the presence of Jupiter-analogs has been invoked as a key factor in determining the habitability of Earth-like planets (Ward & Brownlee 2000; Greaves et al. 2004). Although the idea that Jupiter-analogs are required to shield terrestrial planets from impacts has been conclusively dismantled (e.g. Horner & Jones 2008; Horner et al. 2010; Horner & Jones 2012), the presence of such planets could still prove to be an important mechanism that drives the delivery of water to planets that might otherwise have formed as dry, lifeless husks. For an in-depth discussion of this idea, we direct the interested reader to p.285-6 of Horner & Jones (2010), and references therein. Regardless of their role in ensuring (or threatening) the habitability of telluric worlds, the search for Jupiter-analogs will provide a key datum for models of planetary formation and evolution - attempting to answer the question "how common are planetary systems like our own." To date, the most prolific means of detecting Jupiter-analog planets has been the radial-velocity method. While direct imaging and microlensing have contributed to the pool of known Jupiter analogs (e.g. Chauvin et al. 2004; Marois et al. 2010; Gaudi et al. 2008; Dong et al. 2009), those planets detected by radial velocity remain the most amenable for detailed orbital characterisation. The radial-velocity technique is well-established, having been in use by several planet-search teams for nearly 20 years. A key disadvantage of searching for Jupiter analogs in this way is that one must observe for at least a full orbital cycle to properly constrain the period and amplitude of a planet candidate. For this sort of work, there is no substitute for time. The Anglo-Australian Planet Search (AAPS) has been in operation for nearly 14 years, and has discovered some 40 plan- ets (e.g. Tinney et al. 2001; Butler et al. 2001; Jones et al. 2006, 2010; Tinney et al. 2011). A key strength of this program is that the AAPS has used the same instrumental setup throughout its lifetime: the UCLES spectrograph on the 3.9m Anglo-Australian Telescope. -- 3 -- AAT+UCLES has enabled the AAPS to amass a database of velocities with precisions of 2-3 m s−1 for most of its target stars. This is exactly the type of data set required to robustly de- tect the signals of Jupiter-like planets in Jupiter-like orbits. The long-term velocity stability achieved by the AAPS has enabled the discovery of six such planets to date: HD 30177b (Butler et al. 2006: P = 7.6yr, M sin i=9.7MJup) HD 160691c (McCarthy et al. 2004: P = 11.5yr, M sin i=1.9MJup), GJ 832b (Bailey et al. 2009: P = 9.4yr, M sin i=0.6MJup), HD 134987c (Jones et al. 2010: P = 13.7yr, M sin i=0.8MJup), HD 142c (Wittenmyer et al. 2012: P = 16.4yr, M sin i=5.3MJup), and HD 4732c (Sato et al. 2013: P = 7.5yr, M sin i=2.4MJup). Given the significant and ever-tightening constraints on large-telescope time, it is pru- dent to optimise the observing strategy used to detect particular classes of planets (Ford 2008; Bottom et al. 2013). The AAPS, in recognition of its primary strength in detect- ing Jupiter-like planets, has recently adjusted its target list and observing strategy to be able to make meaningful, quantitative statements on the frequency of these types of planets (Wittenmyer et al. 2011). In this paper, we examine the impact of various observing strate- gies on the detectability of Jupiter analogs. Specifically, we ask what number and duration of observations would most efficiently enable the detection of planets with velocity amplitudes similar to that of Jupiter (K = 12 m s−1). We construct simulated data sets which build on the extant AAPS data (Section 2), and we use these results (Section 3) to draw general conclusions (Section 4) on the optimal approach for detecting Jupiter analogs in the face of extremely limited telescope time. 2. Numerical Methods 2.1. Constructing the simulated data sets From the main AAPS sample of ∼250 stars, we wish to define a subsample of stars which have both low intrinsic stellar activity and a sufficient observational baseline to be useful for the detection of Jupiter analogs. Following our previous work on this subject (Wittenmyer et al. 2011), we chose those stars which have, at present, at least 30 epochs over a time baseline of at least 3000 days. We then excluded those stars which did not satisfy the criterion (RMS/√N ) < ∼ 1, an empirical relation derived from the AAPS data examined in Wittenmyer et al. (2011) which can be used as a simple estimate of whether the 12 m s−1 velocity amplitude of Jupiter would be detectable at 99% confidence. After this cut, 103 stars remained, and all further analysis was done using those data. The epochs of real observational data are never purely random: stars can only be -- 4 -- observed at night, and most targets are unobservable for a portion of the year. When combined with the exigencies of telescope scheduling (planet-search programs are usually assigned time during bright lunations) real data can contain significant gaps. To better simulate the sampling characteristics of real data, we made the following assumptions: (1) one observation in a 10-night block every 30 days (bright-time scheduling), (2) the target is unobservable for four consecutive months every year, and (3) poor weather randomly prevents the observation 33% of the time. These conditions were selected for the following reasons: (1) planet-search programs are usually allocated time in bright lunations owing to the brightness of the targets, (2) for a mid-latitude site such as the Anglo-Australian Telescope (AAT), with planet-survey targets distributed randomly in right ascension, the average target is unobservable for four months in a year1, and (3) the long-term weather conditions at the AAT yield a 33% loss rate (this "weather allowance" is accounted for in the proposal process). Using this procedure, we added simulated observations to the current AAPS data (as of 2012 July) until reaching a desired number of added data points (chosen to lie between 9 and 242). In the interest of informing strategic plans for planet-search programs, we investigated the effect of adding observations over two timescales: three and six years. The frequency of observations (e.g. monthly or alternate months) was adjusted so that the resulting total time baseline of new data remained approximately constant at three and six years. Figure 1 shows the time baseline added using this approach, averaged over the 103 stars. In total, 32 artificial data sets were created for each target considered here: N = 9 − 24 points added over 3 and 6 year periods. The radial velocity of the simulated data point is drawn from a Gaussian distribution with a FWHM equal to the RMS of the existing data for that star. The uncertainty on the simulated radial velocity is drawn at random (with replacement) from the uncertainties of the existing real data. The result is a simulated future data set with the same properties as the original data. 2.2. Assessing Detectabilities Throughout this work, we assessed the detectability of radial-velocity signals using the algorithm described extensively in our previous work (Wittenmyer et al. 2006, 2011). Here we only consider circular orbits (after e.g. Cumming et al. 2008), as previous work has shown that circular-orbit detection limits are a good approximation to the result that would 1Here we define "unobservable" to mean that the target spends less than one hour at an airmass less than 2. 2These values were chosen to reflect the minimum (1.5 epochs/yr in 6 years) and maximum (8 epochs/yr for 3 years) practical cadences for a typical shared large telescope -- 5 -- be obtained for planets with e < ∼ 0.5 (Cumming & Dragomir 2010; Wittenmyer et al. 2010, 2011). In brief, for each of 100 trial periods, a Keplerian signal is added to the data, then a Lomb-Scargle periodogram (Lomb 1976; Scargle 1982) is used to search for the injected signal. The radial velocity semi-amplitude K of the artificial planet is increased until 100% of the signals are recovered with at least 99.9% significance at the correct period. That value of K is then the detection limit for that trial period. As we are concerned with long- period planets, we only consider periods from 1000-5000 days, evenly distributed in log space. Throughout this work, the figure of merit used to assess the impact of added data is the mean detection limit K averaged over 100 trial periods in the range 1000-5000 days. 3. Results 3.1. Duration of Observations The most obvious question here is: "How does the number of added epochs affect the detectability of Jupiter analogs?" A related issue is whether there is a point of diminishing return, where adding more data or more frequent observations does not make meaningful improvements in the overall detectability. To determine the improvement to be had by various levels of observational effort, we also computed the detection limits for the 103-star sample exactly as above, but using only the existing AAPS data. This gives a "benchmark" detectability for each star, against which we can then compare the results from the simulated observations on a star-by-star basis. For each star, we divide the mean K limit (averaged over periods from 1000-5000 days) by that obtained for the "benchmark" or present data set. The grand means of these ratios (averaged over the 103 stars) are plotted in Figure 2. From this figure, we can answer two questions: (1) What is the effect of adding epochs (within a fixed time), and (2) What is the impact of the observational baseline for a given number of epochs? Our results show that, for the range of added epochs simulated here, there remains a linear relationship between the number of epochs and the achievable detection limit. That is, the points shown in Figure 2 do not level off, up to 24 added epochs. On the second question, we see that there is no statistically significant difference between N epochs added over 3 years and N epochs added over 6 years, though this applies only for planets with P < 5000 days (14 years) as this is the maximum period considered in the detection-limit calculation. -- 6 -- 3.2. Number of Observations The results of this experiment, in which we added simulated observations to real AAPS data, show that the amount of added time coverage is less important than the number of added observations. It is worth noting here that the selected stars all had at least 3000 days (8 years) of time coverage before adding new data. Adding about 1000 or 2000 days (3 or 6 years) of new data thus has little effect on the detectability of Jupiter-like planets in the 1000-5000 day period range. We also note that the total number of epochs for each star varies greatly, with some stars at present having ∼35 epochs, and others having well over 100. Since this could have biased the results shown in the Figure 2, we ran a second suite of simulations with slightly different initial conditions -- less "realistic" but more controlled. We created wholly artificial data sets with a fixed number of observations N, ranging from 40 to 500 epochs. The time sampling was done exactly as described above, with the natural result that data sets with more observations had a longer timespan. The average time coverage of the simulated data sets ranged from 7.2 years (N = 40) to 92 years (N = 500). We have chosen the minimum N as 40 epochs, which yields a time baseline of 7.2±0.7 years, consistent with the initial selection of AAPS stars used in the first suite of simulations. Also, in recognition of the fact that radial-velocity noise is almost certainly not Gaussian in nature (O'Toole et al. 2009a,b; Tuomi et al. 2013), we chose instead to draw the value of the simulated data point (and its uncertainty) from a "master" set of 531 observations from a selection of known stable stars in the AAPS catalog. Those velocities have an RMS of 3.0 m s−1, which is typical of the long-term accuracy and precision achieved by AAT/UCLES over the lifetime of the AAPS. In this way, we assure that the simulated data are purely noise, with the characteristics of velocity noise from actual observed stars. The result is 100 simulated data sets each with N epochs, for N between 40 and 500. We computed the detection limits as described above, and calculated the mean velocity amplitude K averaged over 100 trial periods for each data set, again averaging over periods from 1000-5000 days. Trial periods for which all simulated planets were not detected were ignored. Table 1 and Figure 3 show the results, where we have averaged the mean K from each data set over all 100 sets at each value of N. We see from Table 1 and Figure 3 that the detection limit K scales approximately as √N , with the greatest marginal improvement occurring when the total duration extends beyond ∼4000 days (i.e. for N = 40 − 60). This is the expected result for data consisting of Gaussian (white) noise (e.g. Cochran & Hatzes 1996); for the purposes of long-period planet detection, the radial-velocity noise distribution may be closer to Gaussian than is commonly expected. Up to the limit of our tests, N = 500, no noise floor is reached - the detectabilities -- 7 -- continue to improve as √N . We emphasize that this result applies for planets in the period range we have tested here: between 1000 and 5000 days. Obviously, continuing to observe a given star indefinitely would permit the detection of arbitrarily long periods shorter than the total duration of observations. The period dependence of our results is shown in Figure 4. Obviously, for the N = 50 case, where the total time coverage is only about 9 years (3300 days), periods substantially longer than this baseline are virtually undetectable. This artifact disappears once the time coverage exceeds the 5000-day maximum trial period, and we see no significant period dependence for the larger-N trials. The mean detection limit drops below the 3.0 m s−1 scatter of the input data after about 80 points (Figure 3, and reaches K/σ =0.5 when 300-350 epochs are obtained. Our result is applicable to other data sets with different noise properties -- the detection limit would then simply scale with the overall noise. This can be seen from recent HARPS results, where the lowest-amplitude planetary signals are indeed at levels consistent with our results in Figure 3. For example, HD 10180b has K/σ =0.62 with N = 190 measurements (Lovis et al. 2011). Similarly, Pepe et al. (2011) report HD 20794c with K/σ =0.68 and N = 187 measurements. Because most radial-velocity planet search programs on shared large telescopes have similar sampling constraints, our results can thus be applied to other data sets by scaling the noise accordingly. 4. Conclusions Taken together, the tests we have performed can inform the strategies used by long- running radial-velocity planet search programmes such as the AAPS, in order to make best use of limited telescope time to efficiently detect (or rule out) Jupiter-like planets. We have shown that, once there is sufficient observational baseline to detect long-period Jupiter analogs, there is not much added benefit to extending that time base (except to detect ever- longer periods, e.g. Saturn analogs with 30-year periods). That is, for a Keplerian velocity signal with a given orbital period and amplitude K, including more orbital cycles in the observed data does not have a significant effect on the detectability of that signal. We have also performed detailed simulations of radial-velocity observations, sampled according to a realistic schedule and with noise characteristics identical to actual data from the Anglo-Australian Planet Search. Those simulations showed that the Keplerian velocity amplitude K detectable from a data set scales approximately as 1/√N , to the maximum N = 500 tested. That is, the radial-velocity noise may be closer to white than one might have expected. -- 8 -- We note, however, that for extremely low-amplitude signals such as those detected for Alpha Centauri B (Dumusque et al. 2012) and Tau Ceti (Tuomi et al. 2013), stellar noise at the < 1 m s−1 level has a critical impact on the detectability of such small signals. The red noise introduced by star spots, differential rotation, and convective blueshift is of an amplitude too small to be especially relevant to this work. In this work, we have concentrated on the detectability of Jupiter-like planets with relatively large K amplitudes of order 10 m s−1. We conclude that it remains worthwhile to continue radial-velocity observations of suit- ably stable stars to robustly detect or exclude long-period giant planets. However, the targets must be chosen carefully to ensure that the K ∼10-15 m s−1 signals of Jupiter-like planets are reliably detectable. For ever-longer orbital periods, approaching Saturn-like or- bits (P ∼30 yr), opportunities are now emerging to combine the complementary strengths of legacy radial-velocity data with rapidly improving direct-imaging technology from new instruments such as the Gemini Planet Imager. The powerful combination of these two ap- proaches will soon yield direct measurements of the occurrence rate of Jupiter analogs, and the first detailed characterisation of such objects orbiting nearby Sun-like stars. We thank the anonymous referee for a timely and thoughtful report, which improved this manuscript. This research has made use of NASA's Astrophysics Data System (ADS), and the SIMBAD database, operated at CDS, Strasbourg, France. This research has also made use of the Exoplanet Orbit Database and the Exoplanet Data Explorer at exoplanets.org (Wright et al. 2011). REFERENCES Bailey, J., Butler, R. P., Tinney, C. G., et al. 2009, ApJ, 690, 743 Bottom, M., Muirhead, P. S., Johnson, J. A., & Blake, C. H. 2013, arXiv:1302.3910 Butler, R. P., Tinney, C. G., Marcy, G. W., et al. 2001, ApJ, 555, 410 Butler, R. P., Wright, J. T., Marcy, G. W., et al. 2006, ApJ, 646, 505 Chauvin, G., Lagrange, A.-M., Dumas, C., et al. 2004, A&A, 425, L29 Cochran, W. D., & Hatzes, A. P. 1996, Ap&SS, 241, 43 Cumming, A., & Dragomir, D. 2010, MNRAS, 401, 1029 -- 9 -- Cumming, A., Butler, R. P., Marcy, G. W., Vogt, S. S., Wright, J. T., & Fischer, D. A. 2008, PASP, 120, 531 Dong, S., Bond, I. A., Gould, A., et al. 2009, ApJ, 698, 1826 Dumusque, X., Pepe, F., Lovis, C., et al. 2012, Nature, 491, 207 Ford, E. B. 2008, AJ, 135, 1008 Gaudi, B. S., Bennett, D. P., Udalski, A., et al. 2008, Science, 319, 927 Gould, A., Dong, S., Gaudi, B. S., et al. 2010, ApJ, 720, 1073 Greaves, J. S., Wyatt, M. C., Holland, W. S., & Dent, W. R. F. 2004, MNRAS, 351, L54 Horner, J., & Jones, B. W. 2008, International Journal of Astrobiology, 7, 251 Horner, J., & Jones, B. W. 2010, International Journal of Astrobiology, 9, 273 Horner, J., Jones, B. W., & Chambers, J. 2010, International Journal of Astrobiology, 9, 1 Horner, J., & Jones, B. W. 2012, International Journal of Astrobiology, 11, 147 Jones, H. R. A., et al. 2010, MNRAS, 403, 1703 Jones, H. R. A., Butler, R. P., Tinney, C. G., et al. 2006, MNRAS, 369, 249 Lomb, N. R. 1976, Ap&SS, 39, 447 Lovis, C., S´egransan, D., Mayor, M., et al. 2011, A&A, 528, A112 Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., & Barman, T. 2010, Nature, 468, 1080 McCarthy, C., Butler, R. P., Tinney, C. G., et al. 2004, ApJ, 617, 575 O'Toole, S. J., Tinney, C. G., Jones, H. R. A., et al. 2009a, MNRAS, 392, 641 O'Toole, S. J., Jones, H. R. A., Tinney, C. G., et al. 2009b, ApJ, 701, 1732 Pepe, F., Lovis, C., S´egransan, D., et al. 2011, A&A, 534, A58 Sato, B., Omiya, M., Wittenmyer, R. A., et al. 2013, ApJ, 762, 9 Scargle, J. D. 1982, ApJ, 263, 835 Tinney, C. G., Butler, R. P., Marcy, G. W., et al. 2001, ApJ, 551, 507 -- 10 -- Tinney, C. G., Butler, R. P., Jones, H. R. A., Wittenmyer, R. A., O'Toole, S., Bailey, J., & Carter, B. D. 2011, ApJ, 727, 103 Tuomi, M., Jones, H. R. A., Jenkins, J. S., et al. 2013, A&A, in press. arXiv:1212.4277 Ward, P., & Brownlee, D. 2000, Rare earth : why complex life is uncommon in the universe / Peter Ward, Donald Brownlee. New York : Copernicus, c2000. Wittenmyer, R. A., Endl, M., Cochran, W. D., Hatzes, A. P., Walker, G. A. H., Yang, S. L. S., & Paulson, D. B. 2006, AJ, 132, 177 Wittenmyer, R. A., O'Toole, S. J., Jones, H. R. A., Tinney, C. G., Butler, R. P., Carter, B. D., & Bailey, J. 2010, ApJ, 722, 1854 Wittenmyer, R. A., Tinney, C. G., O'Toole, S. J., Jones, H. R. A., Butler, R. P., Carter, B. D., & Bailey, J. 2011, ApJ, 727, 102 Wittenmyer, R. A., Horner, J., Tuomi, M., et al. 2012, ApJ, 753, 169 Wright, J. T., Fakhouri, O., Marcy, G. W., et al. 2011, PASP, 123, 412 This preprint was prepared with the AAS LATEX macros v5.2. -- 11 -- Fig. 1. -- Duration of added simulated observations tested here. Each point represents the mean of the added time coverage over the 103 stars considered. Dashed lines are at 3 and 6 years, which are the nominal durations of the added observations. Filled circles -- 3 added years; open circles -- 6 added years. -- 12 -- Fig. 2. -- Ratio of the mean detection limit K achieved by adding simulated data as compared to the present value (averaged over 103 stars). Error bars represent the standard error of the mean ratios. Filled circles -- 3 added years; open circles -- 6 added years. -- 13 -- Fig. 3. -- Radial-velocity signal-to-noise K/σ detectable in simulated data sets with various numbers of observations. For each of 100 data sets with a given N, we average (over 100 trial periods) the detection limit K for which 99% of injected planets were recovered, then divide by the total RMS of the input data. Each point represents the grand mean of these mean K values derived from the 100 data sets. Error bars are the standard deviation about that grand mean, also normalised by the total RMS of the input data. These results show that a noise floor is reached for N > 250, beyond which further observations do not improve the overall detection limit. -- 14 -- Fig. 4. -- Radial-velocity signal-to-noise K/σ detectable in simulated data sets with various numbers of observations. To show the dependence of our results on the orbital period, we have used the same data from Section 3.2 and Figure 3, but averaged over the 100 simulated data sets at each period. Each solid line thus represents the K/σ at each period, averaged over the 100 distinct trials for each number of observations N. The poor detectabilities at long periods seen for the N = 50 case result from an insufficient time baseline. -- 15 -- Table 1. Detectabilities from Artificial Data Sets N Mean K (m s−1) Ki/Ki−1 pNi−1/Ni 40 50 60 70 80 90 100 150 200 250 300 350 400 500 5.11±0.82 4.33±0.63 3.66±0.51 3.31±0.40 3.08±0.31 2.86±0.30 2.67±0.24 2.21±0.16 2.01±0.14 1.69±0.12 1.58±0.09 1.46±0.11 1.40±0.09 1.28±0.10 · · · 0.85 0.84 0.91 0.93 0.93 0.94 0.83 0.86 0.89 0.94 0.92 0.95 0.91 · · · 0.89 0.91 0.93 0.94 0.94 0.95 0.82 0.87 0.89 0.91 0.93 0.94 0.89
1209.5794	1	1209	2012-09-25T23:35:29	Preliminary Analysis of WISE/NEOWISE 3-Band Cryogenic and Post-Cryogenic Observations of Main Belt Asteroids	[ "astro-ph.EP" ]	We present preliminary diameters and albedos for 13511 MBAs that were observed during the 3-Band Cryo phase of the WISE survey (after the outer cryogen tank was exhausted) and as part of the NEOWISE Post-Cryo Survey (after the inner cryogen tank was exhausted). With a reduced or complete loss of sensitivity in the two long wavelength channels of WISE, the uncertainty in our fitted diameters and albedos is increased to ~20% for diameter and ~40% for albedo. Diameter fits using only the 3.4 and 4.6 um channels are shown to be dependent on the literature optical H absolute magnitudes. These data allow us to increase the number of size estimates for large MBAs which have been identified as members of dynamical families. We present thermal fits for 14 asteroids previously identified as the parents of a dynamical family that were not observed during the fully cryogenic mission.	astro-ph.EP	astro-ph	Preliminary Analysis of WISE/NEOWISE 3-Band Cryogenic and Post-Cryogenic Observations of Main Belt Asteroids Joseph R. Masiero1, A. K. Mainzer1, T. Grav2, J. M. Bauer1,3, R. M. Cutri3, C. Nugent4, M. S. Cabrera1,5 ABSTRACT We present preliminary diameters and albedos for 13511 MBAs that were observed during the 3-Band Cryo phase of the WISE survey (after the outer cryogen tank was exhausted) and as part of the NEOWISE Post-Cryo Survey (after the inner cryogen tank was exhausted). With a reduced or complete loss of sensitivity in the two long wavelength channels of WISE, the uncertainty in our fitted diameters and albedos is increased to ∼ 20% for diameter and ∼ 40% for albedo. Diameter fits using only the 3.4 and 4.6 µm channels are shown to be dependent on the literature optical H absolute magnitudes. These data allow us to increase the number of size estimates for large MBAs which have been identified as members of dynamical families. We present thermal fits for 14 asteroids previously identified as the parents of a dynamical family that were not observed during the fully cryogenic mission. 1. Introduction In Masiero et al. (2011, hereafter Mas11) we presented thermal model fits for 129, 750 Main Belt asteroids that were observed during the fully cryogenic portion of the Wide-field Infrared Survey Explorer (WISE, Wright et al. 2010) mission, which ran from 7 January 2010 to 6 August 2010. Sensitivity to Solar system objects was enabled by the NEOWISE augmentation to the WISE mission (Mainzer et al. 2011a) which provided capability for processing and archiving of single-frame exposures and detection of previously known and new asteroids and comets. On 6 August 2010 the hydrogen ice in the outer cryogen tank was exhausted and the telescope began to warm up, resulting in an almost immediate loss of the W4 (22 µm) channel and a decreasing sensitivity in W3 (12 µm) beginning the 3-Band Cryo portion of the mission. On 29 September 1Jet Propulsion Laboratory/California Institute of Technology, 4800 Oak Grove Dr., MS 321-520, Pasadena, CA 91109, USA, [email protected] 2Planetary Science Institute, Tucson, AZ 85719 USA 3Infrared Processing and Analysis Center, California Institute of Technology, Pasadena, CA 91125 USA 4University of California, Los Angeles, CA, 90095 5California State Polytechnic University Pomona, Pomona, CA, 91768 – 2 – 2010 the hydrogen ice in the inner cryogen reservoir, used to cool the detectors, was exhausted and the W3 channel was lost. From 29 September 2010 to 1 February 2011, WISE continued to survey the sky in the NEOWISE Post-Cryo survey phase (Mainzer et al. 2012), searching for new near-Earth objects (NEOs) and completing the survey of the largest Main Belt asteroids (MBAs) using the two shortest bandpasses: W1 (3.4 µm) and W2 (4.6 µm). MBAs have temperatures of ∼ 200 K, depending on their distance from the Sun and surface properties. This places the peak of their blackbody flux near λpeak ∼ 15 µm. During the fully cryogenic portion of the WISE mission the W3 bandpass straddled this peak and was the primary source of data used for identification and analysis of the thermal emission from MBAs. For objects detected during the 3-Band Cryo portion of the mission we used the W3 data to constrain the thermal emission, and thus the diameter, of the objects observed. As the telescope warmed up, the integration times in W3 were shortened to prevent saturation of the detectors from the increasing thermal emission of the telescope (Cutri et al. 2012), resulting in a decrease in sensitivity to sources in the bandpass. During the Post-Cryo Survey only W1 and W2 were operational: for MBAs W1 was sensitive solely to reflected light, while W2 was a blend of reflected and emitted flux dictated by the object’s physical and orbital parameters (e.g. distance to Sun at the time of observation, surface temperature, albedo, etc.). In this work, we present preliminary thermal model fits for 13511 Main Belt asteroids observed during the 3-Band Cryo phase of the WISE survey and the NEOWISE Post-Cryo Survey. During the fully cryogenic portion of the survey, detectability of most minor planets was dominated by their thermal emission and so was essentially independent of their albedo (Mainzer et al. 2011c). However, the Post-Cryo Survey data at 3.4 µm and 4.6 µm are a mix of reflected and emitted light. Thus detectability is strongly coupled to albedo. Additionally, objects with lower temperatures will have a smaller thermal emission component to their flux in the W2 band, resulting in a less accurate estimate of diameter. In general, diameter fits using either the 3-Band Cryo or the Post- Cryo Survey data will typically have larger errors and lower precision than fits from the fully cryogenic survey given in Mas11, though they still provide useful information about the observed population of MBAs. One of the drivers for completing the NEOWISE survey of the inner Main Belt after the cryogen was exhausted was to have a complete census of the largest asteroids, particularly those that may be members of asteroid families. Having this list allows us to constrain the mass of the pre-breakup body and more precisely model the age of the family (Vokrouhlick´y et al. 2006; Masiero et al. 2012). We present in this work preliminary albedos and diameters for objects observed during the 3-Band Cryo and Post-Cryo Survey and discuss the accuracy of these values because these fits use data processed with the preliminary survey calibration values. Future work by the NEOWISE team will include second-pass processing of the raw data with finalized calibration values as well as extraction of sources at lower signal-to-noise that will precede a final release of albedos and diameters. – 3 – 2. Observations In Mas11 we focused our analysis on data taken during the fully cryogenic portion of the WISE mission. For this work, we analyze the 3-Band Cryo and Post-Cryo Survey observations taken by WISE as part of the NEOWISE survey. Observations obtained between Modified Julian Dates (MJDs) of 55414 and 55468 are available in the 3-Band Cryo Single-Exposure database, served by the Infrared Science Archive (IRSA)1. Post-Cryo data, spanning a MJD range of 55468 to 55593, are archived in the NEOWISE Preliminary Post-Cryo database and also served by IRSA. Data from the 3-Band Cryo survey were released to the public on 29 June 20122 and preliminary data from the NEOWISE Post-Cryo Survey were released to the public on 31 July 20123. We note that the Post-Cryo Survey data have only undergone first-pass processing, and users are strongly encouraged to consult the Explanatory Supplement (Cutri et al. 2012) associated with the database. We follow the same method as described in Mas11 to acquire detections of MBAs that have been vetted both by our internal WISE Moving Object Processing System (WMOPS; Mainzer et al. 2011a) and by the Minor Planet Center (MPC). This includes the use of the same quality flag settings from the pipeline extraction for cleaning of detections before thermal fitting as discussed in Mas11. Of the 14638 objects observed by WISE between MJDs 55414 and 55593, 13511 MBAs had data of sufficient quality to perform thermal model fits. Due to the nature of WISE’s orbit and the synodic period of MBAs, approximately half of the objects observed during the 3-Band Cryo and Post-Cryo Survey had also been observed earlier during the fully cryogenic survey. We use these overlap objects as standards to evaluate the accuracy of the thermal model fits using these data (see Section 4.1). While in some cases extremely irregularly shaped slow-rotating objects may show significant changes in projected area between epochs and thus large variations in both emitted and reflected flux, this is expected to be a small fraction of all objects observed and only to add a small component of random error to the comparison (Grav et al. 2011). 3. Thermal Fitting Following the procedure discussed in Mas11, we use a faceted NEATM thermal model to deter- mine the diameter and albedo of the MBAs observed after the outer cryogen tank was exhausted. In most cases we only have thermal emission data in a single band, and so we are forced to assume a beaming parameter for the models. We use a beaming parameter of η = 1.0 ± 0.2, based on the peak of the distribution for MBAs given in Mas11. Our measured flux in W2 is typically dominated 1http://irsa.ipac.caltech.edu 2http://wise2.ipac.caltech.edu/docs/release/3band/ 3http://wise2.ipac.caltech.edu/docs/release/postcryo prelim/ – 4 – by thermal emission, however the reflected component of the W2 flux will influence our models. In order to remove the reflected component from the measured W2 flux, we need to determine the optical geometric albedo (pV ) and assume a ratio between the near-IR (NIR) and optical albedos. In Mas11 we were able to fit this ratio for objects with with observations in W 3 and/or W 4 as well as W 1 and W 2, however we cannot do this for the Post-Cryo Survey data. Following the best-fit value from Mas11, for those objects we assume a NIR/optical reflectance ratio of 1.4 ± 0.5. In all cases, we also assume that the reflectivity in W1 is identical to that in W2 (pW 1 = pW 2 = pN IR). For objects with very red spectral slopes this may not necessarily be a good assumption (cf. Mainzer et al. 2011b; Grav et al. 2012) however without additional data (e.g. spectral taxonomy) it is impossible to disentangle these two values for this dataset. To determine optical albedo we used the H absolute magnitude and G slope parameter given in the Minor Planet Center’s MPCORB file4, and updated using other databases following Mas11. We note that recent work has shown that these H values may be systematically offset in some magnitude ranges by up to 0.4 mags when comparing predicted and observed apparent magnitudes (Pravec et al. 2012). This will affect the albedos that we calculate for the asteroids presented here, which in turn will change the relative contribution of emitted and reflected light in W2. Unlike the results presented in Mas11, where the diameter determination is independent of the optical H measurement, any future revision to the measured H values will require a refitting of the thermal models and will likely result in an change in modeled diameter. 4. Discussion 4.1. Comparison of Overlap Objects Mainzer et al. (2012) showed a comparison between the thermal fits performed with the Post- Cryo Survey data and non-radiometrically determined diameters for a range of NEOs and MBAs to derive a relative accuracy of ∼ 20% on diameter and ∼ 40% on albedo. As a parallel check we have taken objects that were observed both before and after the exhaustion of the outer cryogen reservoir and compared the diameters and albedos found here to those values given in Mas11. Of the fits presented here, 7222 unique objects also appeared in the fully cryogenic observations that were presented in Mas11. Of these, 2844 were observed during the 3-Band Cryo phase of the survey and 4403 were observed during the Post-Cryo Survey (note that 25 objects appeared in all three phases of the survey). For all objects seen in both the Post-Cryo Survey and in the fully cryogenic 4-band survey, we have refit the 4-band Cryo data using only the W1 and W2 measurements as a way of differentiating changes in the quality of fit due to the loss of W3 and W4 sensitivity from changes due to the 4http://www.minorplanetcenter.net/iau/MPCORB.html – 5 – different observing circumstances. The results of this test are shown in Figure 1. We include a running box average of the data in order to assess the population trends, which bins by 100 objects, in steps of 20. In general these tests follow the expected one-to-one relationship, with the exception of the comparison of the 2-band and 4-band fits of the fully cryogenic data (Figure 1b), which deviates at both high and low albedos, and effect that was also observed for the NEOs in the Post-Cryo Survey data by Mainzer et al. (2012). (Mainzer et al. 2011b) have shown that high albedo objects tend to have optical/NIR reflectance ratios of ∼ 1.6, while objects with low albedos tend to have reflectance ratios of ∼ 1.0 (though D-type objects deviate from this trend and have very large reflectance ratios). As we use a fixed reflectance ratio of 1.4, low albedo objects with W1 measurements will have a final fitted pV below the true value, while high albedo objects will have a pV slightly above, which corresponds to the twist observed in Figure 1b. Fig. 1.— Comparison of thermal fits for objects appearing in both the fully cryogenic data set as well as the Post-Cryo Survey data. The left column shows the comparison of the diameters (log D) while the right column shows the comparison of the visual albedos (log pV ). The top row shows the fractional difference between the 4-band fits presented in Mas11 and refits of those data using only the W1 and W2 bandpasses, while the bottom row shows the fractional difference between the 2-band refit of the fully cryogenic data and the 2-band fits of the Post-Cryo Survey data. The dotted line in each case shows a one-to-one relationship, and the solid red line shows a running box average for each comparison. – 6 – Comparison of the 2-band refits to the results from Mas11 show the uncertainty induced by the loss of W3 and W4 information (Figure 1a-b) results in a 1σ scatter of 16% in diameter and 32% in albedo (three points fall outside the plotted range for Figure 1(a); all other panels show all objects considered). Comparison of the 2-band refits of the fully cryogenic data to the Post-Cryo Survey fits (Figure 1c-d) shows the errors induced by both changes in observing aspect as well as calibration differences between the two data sets, which collectively result in a 1σ scatter of 13% in diameter and 31% in albedo as well. Combined, these two errors result in a measured 21% relative error on diameter and 45% in albedo. These total errors are in line with what was found by Mainzer et al. (2012) when comparing the fits from Post-Cryo Survey data to literature diameters. Our errors are also in line with the uncertainties measured for the ExploreNEOs project which uses a similar pair of bandpasses (3.6 µm and 4.5 µm) from the Warm Spitzer mission to model diameters and albedos for previously known NEOs (Trilling et al. 2010; Harris et al. 2011). Our measured level of error indicates that the random error introduced by the combined effect of irregular shape and observing geometry is below this level. We note that the method of source extraction used for all phases of the WISE data processing relies on the position of the object in all detected bands. In general the W1 and W2 measurements from the 4-band data will be at lower signal-to-noise than the data for those objects extracted from the Post-Cryo Data, inflating the errors quoted above. We show in Figure 2 the comparison between the fits for objects appearing in the fully cryogenic data as well as the 3-Band Cryo or Post-Cryo Survey data. As in Figure 1 we include a running box average using the same parameters as above. We see no large-scale systematic shifts between datasets, however we do confirm the increase in scatter in the fits using the latter data sets. We note that in Figure 2d shows a behavior similar to what we observe in Figure 1b, where the fits of albedo deviate to more extreme values for both high and low albedo objects. As discussed above, this is attributed to the use of an assumed optical/NIR reflectance ratio that is between the values measured for high and low albedo objects when they are considered independently. 4.2. Preliminary Diameters and Albedos We present in Table 1 the preliminary fitted diameters and albedos for all MBAs observed during the 3-Band Cryo and Post-Cryo surveys, along with their associated errors (note that errors do not include the systematic ∼ 20% diameter error or ∼ 40% albedo error discussed above). We also include the number of detections used in each band as well as the H and G values used for the fit. Objects without measured visible magnitudes have “nan” entered for their H, G, and albedo values. The recommended method for extracting fluxes for asteroid detections is discussed in Mainzer et al. (2011a) and Cutri et al. (2012). Figure 3 shows the preliminary diameter and albedo distributions for the asteroids observed during 3-Band Cryo and Post-Cryo Surveys compared to the population presented in Mas11. With the loss of the long wavelength channels the sensitivity to small objects was reduced and peak of the diameter distribution moves to larger sizes. In both – 7 – Fig. 2.— Comparison of thermal fits for objects appearing both in the fully cryogenic data set as well as in the 3-Band Cryo or Post-Cryo Survey data. The left column shows the comparison of the diameters (log D) while the right column shows the comparison of the visual albedos (log pV ). The top row shows the fractional difference between the 3-Band Cryo fits and the values from Mas11, while the bottom row shows the fractional difference between the 2-band Post-Cryo Survey fits and the Mas11 values. The dotted line in each case shows a one-to-one relationship, and the solid red line shows a running box average for each comparison. the 3-Band Cryo and Post-Cryo Survey data we see a shift in the high and low branches of the albedo distribution to more extreme values when compared to the population from Mas11. This shift was also observed for the NEOs by Mainzer et al. (2012), and attributed to the forced values for both beaming and NIR/optical reflectance ratio in the fits of the Post-Cryo Survey data. Reflected light is a much more significant component in the W1 and W2 bandpasses for MBAs than in the W3 and W4 bands used in Mas11 to perform thermal fits. As such, results from the model fits presented here are inherently tied to the optical measurements and cannot be considered insensitive to albedo as was assumed in Mas11. This bias will most strongly affect objects that are small and have low albedos. Thus, care must be taken before extrapolating the trends observed in the these fits to the greater MBA population. – 8 – Table 1: Thermal model fits for MBAs in the 3-Band Cryo and NEOWISE Post-Cryo Survey. Table 1 is published in its entirety in the electronic edition of ApJL; a portion is shown here for guidance regarding its form and content. Name 00003 00005 00011 00014 00016 00017 00018 00019 00020 H 5.33 6.85 6.55 6.30 5.90 7.76 6.51 7.13 6.50 G 0.32 0.15 0.15 0.15 0.20 0.15 0.25 0.10 0.25 D (km) 246.60 ± 10.59 106.70 ± 3.14 154.13 ± 3.92 145.68 ± 5.27 288.29 ± 4.63 69.64 ± 2.26 155.84 ± 5.63 209.81 ± 2.20 135.68 ± 3.67 pV 0.214 ± 0.026 0.282 ± 0.050 0.178 ± 0.030 0.251 ± 0.041 0.093 ± 0.024 0.287 ± 0.051 0.181 ± 0.033 0.056 ± 0.012 0.241 ± 0.018 nW 1 11 14 12 9 5 9 12 11 13 nW 2 11 14 12 9 5 10 12 11 13 nW 3 0 0 0 0 0 0 0 0 0 4.3. Asteroid Family Members One of the primary drivers of the Post-Cryo Survey was to complete the census of large MBAs that are related to dynamically associated asteroid families. The largest body in a family anchors both the mass estimate of the pre-breakup body and the starting point for family age simulations. The 3-Band Cryo and Post-Cryo Survey data contain 3319 objects identified by (Nesvorn´y 2010) as members of asteroid families that were able to have thermal models fit to their measurements. Of these, 14 were identified as family parents and were not observed during the WISE fully cryogenic mission, including (3) Juno, (20) Massalia, (44) Nysa, (170) Maria, (298) Baptistina, (363) Padua, (434) Hungaria, (490) Veritas, (569) Misa, (778) Theobalda, (1270) Datura, (1892) Lucienne, (4652) Iannini, and (7353) Kazuya. Of these 14 bodies that are the largest in their family, only 4 had albedos below pV = 0.1, in contrast with the general population presented here where ∼ 60% of the MBAs had low albedos. This is due to a number of overlapping selection biases, including the dominance of high albedo objects in the literature family lists (cf. Mas11), preferential sensitivity to high albedo objects in the 3-Band Cryo and Post-Cryo Survey data compared to Mas11 (meaning this data set is more likely to miss low albedo asteroids), and the longer synodic periods of MBAs with smaller semimajor axes. The differences in synodic period resulted in a larger fraction of objects in the inner Main Belt that were not observed during the fully cryogenic phase of the WISE survey, compared to the outer Main Belt. Future work in family identification will begin to mitigate these biases. – 9 – Fig. 3.— Preliminary diameter (a) and albedo (b) distributions for all MBAs from Mas11 (black dotted), MBAs from the 3-Band Cryo data (blue solid), and MBAs from the Post-Cryo Survey (red dashed). Note the scales are normalized: the total number of objects presented in Mas11 is over an order of magnitude larger than the other two populations. 5. Conclusions We present preliminary thermal model fits for 13511 MBAs using observations acquired by the WISE and NEOWISE surveys following the exhaustion of the outer cryogen tank that marked the end of the fully cryogenic WISE survey. Accuracy of these fits is degraded with respect to the results discussed in Mas11 due to the loss of the W3 and W4 bandpasses, however fits of diameter with relative accuracy of ∼ 20% are still possible. Unlike the fits presented in Mas11, these determinations depend strongly on the measured value of the optical albedo (as calculated from the H absolute magnitude). Thus, any revision to the H values will require a new thermal model to be fit to the data. This dataset includes detection of 3319 members of previously identified asteroid families, one of the main goals of the Post-Cryo Survey. Future work by the NEOWISE team will include second-pass processing of these data sets using updated calibration products, which is expected to improve the accuracy of diameter and albedo determination. – 10 – Acknowledgments J.R.M. was supported by an appointment to the NASA Postdoctoral Program at JPL, ad- ministered by Oak Ridge Associated Universities through a contract with NASA. We thank the anonymous referee for their helpful comments. This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration. This publication also makes use of data prod- ucts from NEOWISE, which is a project of the Jet Propulsion Laboratory/California Institute of Technology, funded by the Planetary Science Division of the National Aeronautics and Space Ad- ministration. This research has made use of the NASA/IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. – 11 – REFERENCES Cutri, R.M., Wright, E.L., Conrow, T., Bauer, J., et al., 2012, http://wise2.ipac.caltech.edu/docs/release/allsky/expsup/index.html Grav, T., Mainzer, A.K., Bauer, J.M., Masiero, J., Spahr, T, et al., 2011, ApJ, 742, 40. Grav, T., Mainzer, A.K., Bauer, J.M., Masiero, J., 2012, ApJ in press. Harris, A.W., Mommert, M., Hora, J.L., Mueller, M., et al., 2011, AJ, 141, 75. Mainzer, A.K., Bauer, J.M., Grav, T., Masiero, J., et al., 2011a, ApJ, 731, 53. Mainzer, A.K., Grav, T., Masiero, J., Hand, E., et al., 2011b, ApJ, 741, 90. Mainzer, A.K., Grav, T., Bauer, J.M., Masiero, J., et al., 2011c, ApJ, 743, 156. Mainzer, A.K., Grav, T., Bauer, J.M., Masiero, J., et al., 2012, ApJ, submitted to ApJL. Masiero, J.R, Mainzer, A.K., Grav, T., Bauer, J.M., et al., 2011, ApJ, 741, 68. Masiero, J.R, Mainzer, A.K., Grav, T., Bauer, J.M. & Jedicke, R., 2012, ApJ in press. Nesvorny, D., 2010, EAR-A-VARGBDET-5-NESVORNYFAM-V1.0, NASA Planetary Data Sys- tem. Pravec, P., Harris, A.W., Kusnir´ak, P., Gal´ad, A. & Hornoch, K., 2012, Icarus, submitted. Trilling, D., Mueller, M., Hora, J.L., Harris, A.W., et al., 2012, AJ, 140, 770. Vokrouhlick´y, D., Broz, M., Bottke, W.F., Nesvorn´y, D. & Morbidelli, A., 2006, Icarus 182, 118. Wright, E.L., Eisenhardt, P., Mainzer, A.K., Ressler, M.E., Cutri, R.M., Jarrett, T., Kirkpatrick, J.D., Padgett, D., et al., 2010, AJ, 140, 1868. This preprint was prepared with the AAS LATEX macros v5.2.
1210.6035	1	1210	2012-10-22T20:00:02	A limit on eccentricity growth from global 3-D simulations of disc-planet interactions	[ "astro-ph.EP" ]	We present high resolution 3-D simulations of the planet-disc interaction using smoothed particle hydrodynamics, to investigate the possibility of driving eccentricity growth by this mechanism. For models with a given disc viscosity (\alpha = 0.01), we find that for small planet masses (a few Jupiter masses) and canonical surface densities, no significant eccentricity growth is seen over the duration of our simulations. This contrasts with the limiting case of large planet mass (over twenty Jupiter masses) and extremely high surface densities, where we find eccentricity growth in agreement with previously published results. We identify the cause of this as being a threshold surface density for a given planet mass below which eccentricity growth cannot be excited by this method. Further, the radial profile of the disc surface density is found to have a stronger effect on eccentricity growth than previously acknowledged, implying that care must be taken when contrasting results from different disc models. We discuss the implication of this result for real planets embedded in gaseous discs, and suggest that the disc-planet interaction does not contribute significantly to observed exoplanet eccentricities.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 12 (2012) Printed 13 June 2021 (MN LATEX style file v2.2) A limit on eccentricity growth from global 3-D simulations of disc-planet interactions A. C. Dunhill1⋆, R. D. Alexander1 and P. J. Armitage2,3 1Department of Physics & Astronomy, University of Leicester, Leicester, LE1 7RH 2JILA, University of Colorado & NIST, 440 UCB, Boulder, CO 80309-0440, USA 3Department of Astrophysical and Planetary Sciences, University of Colorado, Boulder, USA Accepted 2012 October 19. Received 2012 September 17; in original form 2012 June 6 ABSTRACT We present high resolution 3-D simulations of the planet-disc interaction using smoothed par- ticle hydrodynamics, to investigate the possibility of driving eccentricity growth by this mech- anism. For models with a given disc viscosity (α = 0.01), we find that for small planet masses (a few Jupiter masses) and canonical surface densities, no significant eccentricity growth is seen over the duration of our simulations. This contrasts with the limiting case of large planet mass (over twenty Jupiter masses) and extremely high surface densities, where we find eccen- tricity growth in agreement with previously published results. We identify the cause of this as being a threshold surface density for a given planet mass below which eccentricity growth cannot be excited by this method. Further, the radial profile of the disc surface density is found to have a stronger effect on eccentricity growth than previously acknowledged, imply- ing that care must be taken when contrasting results from different disc models. We discuss the implication of this result for real planets embedded in gaseous discs, and suggest that the disc-planet interaction does not contribute significantly to observed exoplanet eccentricities. Key words: planet-disc interaction -- protoplanetary discs -- planets and satellites: formation -- planets and satellites: dynamical evolution and stability -- methods: numerical -- hydrody- namics. 1 INTRODUCTION Modern theories of planet formation all posit that planets are the end result of the evolution of circumstellar discs of gas and dust, termed protoplanetary discs. This basic idea is consistent with both the observation of such discs around young, solar-like stars (e.g., Sargent & Beckwith 1987; Haisch, Lada, & Lada 2001), and the near co-planar, near circular orbits of the planets in the Solar Sys- tem. However, observations of extra-solar planets made over the last 15 years have revealed that many planetary systems do not share the neat structure of our own. In particular, the abundance of planets with eccentric orbits was unexpected given this formation mechanism. The observed distribution of eccentricities is nearly uniform between eccentricity e = 0 and ∼ 0.6, and stretches all the way up to e ≃ 1 (e.g. Wright et al. 2011; Kane et al. 2012). At small semimajor-axis (a . 0.1au) there is a clear preference towards low eccentricity (e < 0.1) orbits, and this is well explained by tidal circularisation (Rasio et al. 1996). Similarly, the dearth of planets in high eccentricity orbits (e & 0.5) with small semi-major axes (. 1au) is readily understood, as these planets have periastron distances ∼ 1 R⊙ and consequently pass too close to their stars to ⋆ E-mail: [email protected] c(cid:13) 2012 RAS be long-lived. At larger semi-major axes, however, the picture is much less clear, and a number of scenarios have been proposed to to explain the apparent discrepancy between theory and observa- tion. These fall into three broad categories: dynamical interactions of planets in multi-planet systems; secular interactions with com- panion stars; and tidal interactions between planets and their parent protoplanetary discs. resonances over Juri´c & Tremaine 2008; Chatterjee et al. longer Large regions of the observed eccentricity distribution can be populated by invoking interactions between multiple planets, be it via direct close encounters (Ford, Havlickova, & Rasio 2001; or 2008) through mean-motion time-scales (Chiang, Fischer, & Thommes 2002). While simulations of such encounters are able to reproduce the observed distribution to a reasonable accuracy, it is unclear if such close interactions are frequent enough in nature to provide a universal source of planetary eccentricity. Another method for growing eccentricity is through secular interactions with inclined companion stars, which lead to a resonant exchange of angular momentum between the planets and the external body (Kozai 1962; Lidov 1962). This results in long-period changes in inclination and eccentricity and although this mechanism seems inviting as an alternative explanation for the observed eccentricity distribution, numerical 2 A. C. Dunhill, R. D. Alexander & P. J. Armitage work has shown that it does not produce the correct eccentricity distribtion (Takeda & Rasio 2005), although recent work has found that it may explain some eccentric misaligned Hot Jupiters (Naoz, Farr, & Rasio 2012). Similar interactions between planets in the same system have also been suggested as a chaotic formation mechanism for highly eccentric planets (Wu & Lithwick 2011). The gravitational interaction between a young embedded planet and its parent gas disc has also been suggested as a mecha- nism for driving eccentricity growth. For companions with masses comparable to the central body it has long been known that tidal interactions with the disc lead to eccentricity excitation. This result has applications to stellar binaries (e.g., Artymowicz et al. 1991) and binary super-massive black holes (e.g, Cuadra et al. 2009), but how it extends to the more extreme mass ratios of star-planet sys- tems is still not clear. Analytical treatment of this problem begins with the consid- eration of resonant torques between a planet and its parent disc. These torques come in two main flavours, Lindblad and corotation. Lindblad torques occur where the epicyclic frequency in the disc is equal to plus or minus the frequency of a component of the planet's potential. Corotation resonances occur at radii where the compo- nent's orbital frequency matches that of the disc. For a planet with a circular orbit, the potential can be described by a series of com- ponents with the same angular velocity as the planet. This results in a coorbital corotation resonance with combs of Lindblad reso- nances to either side. When the planet's orbit is eccentric, this is no longer the case. A more complex picture emerges where not all res- onances have the same angular velocity, resulting in non-coorbital corotation resonances (Goldreich & Tremaine 1980). The interplay between these resonances, the planet that excites them, and how they are affected by various disc parameters have been discussed in detail by Ogilvie & Lubow (2003) and Goldreich & Sari (2003). A great deal of nuance between competing effects is revealed, and in particular the possibility that corotation torques may begin to sat- urate and weaken once an initial eccentricity is attained is useful when considering the results of numerical work (Ogilvie & Lubow 2003; Masset & Ogilvie 2004). However, the complex, non-linear nature of the planet-disc interaction makes the general problem an- alytically intractable, and numerical solution is required to gain a full understanding of eccentricity growth. have been (2008) somewhat analytic works Semi-analytic calculations combining prescriptions from inconclusive. these Moorhead & Adams found eccentric damping rather than growth in most cases, although in the cases where they did find growth it was extremely strong, leading to e ∼ 1 after only a few thousand orbits. However, as such highly eccentric planets would be unable to maintain an equally eccentric gap their orbits would be circularised as they interact with coorbital disc material (e.g. Bitsch & Kley 2010). By contrast, numerical simulations looking specifically at eccentricity growth have so far shown more consistent and positive results. Papaloizou, Nelson, & Masset (2001, hereafter PNM01) found that relatively massive embedded companions (in the brown dwarf regime) undergo eccentricity growth. Lower- mass planets were not found to experience this growth, although it seems likely that this was for numerical rather than physi- cal reasons (Masset & Ogilvie 2004). Later simulations have in- deed found eccentricity growth, albeit at modest levels, down to MJup(D'Angelo, Lubow, & Bate 2006). Extensive analysis of the behaviour and morphology of the disc by PNM01 and Kley & Dirksen (2006) attributes this eccentricity excitation to an instability launched at the 3:1 outer Lindblad resonance, which drives a large eccentricity at the inner edge of the disc. For the large companion masses considered by PNM01 a wide gap is opened in the disc, so coorbital co-rotation resonances are not present, and non-coorbital ones only operate once the planet's orbit is already eccentric. Kley & Dirksen (2006) also extend this analysis down to planets of a few MJup, and find that this mechanism still oper- ates down to planets of mass ∼ 3 MJup, although the magnitude of the eccentricity induced depends strongly on the disc viscosity and temperature. To date the majority of the numerical simulations of this prob- lem have been performed in only two dimensions (2-D), and all have used Eulerian (grid-based) methods. However, it has been sug- gested that a full three-dimensional (3-D) treatment weakens the effect of resonant torques (Tanaka, Takeuchi, & Ward 2002). Each study has also typically only considered a single disc model, with little consistency in the choice of parameters, and it can be eas- ily seen that the properties of the disc itself plays a large role in the evolution of a planet embedded within it. Moreover, Eulerian methods are not always ideal for following the dynamics of gas on non-circular orbits; in general, one expects Lagrangian methods to track eccentric orbits with greater accuracy. In this paper we present results of high-resolution 3-D smoothed particle hydrodynamics (SPH) simulations of eccentric- ity growth due to planet-disc interactions. In section 2 we describe our numerical method and the initial conditions used. In section 3 we present results of our simulations, comparing them to the results of PNM01 especially, and describing the effect of various model parameters. Section 4 contains a discussion of both the numerical limitations of our simulations, and the physical interpretation of the results. Our summary and concluding remarks are in section 5. 2 SIMULATIONS We have performed a suite of three-dimensional simulations of a planet embedded in a protoplanetary accretion disc. We used a modified version of the publicly available Gadget-2 code (Springel 2005), a hybrid SPH/N-body code. SPH methods are highly suited to tracing the evolution of dynamical systems, as the Lagrangian formulation means that important orbital parameters (energy, linear & angular momentum) are naturally conserved (e.g., Price 2012). We treat the star and planet as point masses, and model the gas disc using SPH. The basic parameters of the code are unchanged throughout our simulations, and were set as follows. The number of nearest neighbours for kernel integration is 50, and smoothing lengths are adjusted accordingly when the number of neighbours changes by more than ±2. The Courant parameter used to deter- mine the time-step for SPH particles was set to 0.1. The gravita- tional softening length for the point mass particles was in each case set to be the same as the sink radius for the planet particle (de- scribed below). SPH particles are therefore accreted by the sink particles before they encounter gravitational softening, so this soft- ening has no effect on our simulations. We have altered the public Gadget-2 code in a number of ways to make it more suitable for simulating planets embedded in discs. Firstly, we note that the standard Barnes-Hut tree is insufficiently accurate when computing the gravitational forces on the N-body particles (particularly the 'planet'). Consequently, following the method of Cuadra et al. (2009), we removed the N-body particles from the tree, and instead computed their gravitational forces by di- rect summation. As this is done for only two particles the additional computational cost is not large, and ensures high accuracy for the c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 orbit of the planet. Additionally, following Cuadra et al. (2006), we treat each N-body particle as a 'sink' for SPH particles, accreting the mass and momentum of the swallowed particles on to the target sink, ensuring conservation of linear and angular momentum to a high accuracy. For the planet particle, the radius within which the sink will accrete SPH particles is set to be 0.4 of the Hill radius, defined as RHill = a Mp 3M⋆!1/3 , (1) for a planet and star of mass Mp and M⋆ respectively with semi- major axis a. The sink radius for the star particle was set to be 7/8 of the initial inner disc radius. We have also substituted the energy and entropy equations in Gadget-2 with a locally isothermal equa- tion of state, where the sound speed cs in the disc is a function only of the cylindrical radius R. We choose a power-law profile for the disc temperature T (R) ∝ R−1/2, consistent with both a linear vis- cosity law (e.g., Hartmann et al. 1998) and a standard 'flaring disc' model (e.g., Kenyon & Hartmann 1987). 2.1 Artificial viscosity and angular momentum transport In order to smooth out the discontinuities associated with shocks in SPH, it is necessary to make use of an artificial viscosity. We use the method of Morris & Monaghan (1997) to dissipate energy in shocks and prevent particle penetration. This contributes a further acceleration term to the equations of motion (Springel 2005)1 dui a dt = −Xb mbΠab∇i aWab. (2) Wab is the SPH kernel used by Gadget-2, a cubic spline that goes to zero at one smoothing length h, while W ab is the arithmetic mean of Wab(ha) and Wba(hb) 2. Πab describes the strength of the artificial viscosity and is given by Πab = ( (−αab csab µab + β µ2 0 ab)/ρab for uab · rab < 0, otherwise. (3) with µ given by µab = hab uab · rab 2 rab2 + εh ab . Here csab, ρab and hab represent the arithmetic means of the SPH sound speed, density and smoothing lengths respectively, between particle a and b while uab and rab are the differences between the velocity and position vectors of those particles. ε = 0.0001 is a numerical 'safety factor', which prevents the artificial viscosity di- verging for very small particle separations. In the Morris & Monaghan (1997) scheme, β in equation 3 is given by β = 2αab, and αab is the mean value (αa + αb)/2. We have used the Price (2004) version of this method, where αa is evolved according to αa = αmin − αa τa + (αmax − αa) S a (5) 1 In this system of equations, i, j and k are vectors using summation nota- tion, and a and b are particle indices. 2 Note that this differs from the 'standard' SPH definition of the smoothing length, where the kernel goes to zero at 2h. This difference in notation has no practical effect, but must be considered when comparing these equations to those in the SPH literature. c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 A limit on eccentricity growth 3 and αmin and αmax are imposed minima and maxima, set as a global parameters. The source term S is given by S a = max{−∇ · ua, 0} and the decay time-scale is τa = ha lcsa . (6) (7) where l is a dimensionless scaling parameter. We follow Price (2004) and chose values l = 0.1, αmin = 0.01 and αmax = 2.0. We also make use of the 'Balsara switch' (Balsara 1995) to limit the spurious detection of shear flow as shocks. This artificial viscosity prescription is necessary to handle shocks correctly, but we also wish to include an explicit physical viscosity to drive angular momentum transport in the disc. To this effect we have modified the SPH equation of motion (equation 7 in Springel 2005) to include a Navier-Stokes shear viscosity term (with the bulk coefficient set to zero; Lodato & Price 2010): dui a dt = −Xb mb S i j a Λaρ2 a ∇ j aWab(ha) + S i j b Λbρ2 b ∇ j bWab(hb) Here Λa is dimensionless and accounts for smoothing-length gra- dients: . (8) (9) . (10) ∂ρa ∂ha! ha 3ρa Λa = 1 + and S i j a = "−(Pa) + 2 S i j 3 a is the stress tensor ηa! ∂uk a ∂rk a# δi j + η ∂ui a ∂r j a + ∂u j a ∂ri a P and ρ are the SPH-estimated pressure and density respec- tively. The standard variable-smoothing-length operator is given by (Lodato & Price 2010) ∂ui a ∂r j a = 1 Λaρa Xb mb(cid:16)ui a − ui b(cid:17) ∂Wab(ha) ∂r j a . (11) ηa is the shear viscosity, which is parametrized in terms of kine- matic shear viscosity νa by In practice we set the viscosity such that each particle a has a viscosity νa which depends on only its (cylindrical) radius, which gives a straightforward means of implementing the power-law vis- cosity prescriptions described in section 2.2. 2.1.1 Viscous ring-spreading tests To test this set-up, we conducted test simulations which model the viscous spreading of a gas ring around a point mass. In these tests a thin ring of initially uniform surface density is allowed to evolve under the action of viscous torques. We expect the ring to spread (see, e.g., Pringle 1981), and by comparing our re- sults against those from a one-dimensional explicit scheme (e.g., Pringle, Verbunt, & Wade 1986) we can test the accuracy of our SPH viscosity prescription. We modelled the ring using 105 and 106 SPH particles and four different levels of viscosity (see table 1). The ring was set up orbiting a single point mass, and was al- lowed to evolve for 200 orbits. The viscous time-scale for a spreading ring with constant vis- cosity is is given by (Pringle 1981) (4) νa = ηa/ρa. (12) 4 A. C. Dunhill, R. D. Alexander & P. J. Armitage Figure 1. Radial surface density evolution of a viscously spreading ring. Panel (a) shows a ring with 105 particles in its initial configuration (solid black line), after 4.9 orbits (dashed black line) and after 39.6 orbits (dash-dotted black line). The corresponding red lines show the best-fitting profile from an explicit 1-D ring code, plotted as a fraction of the viscous spreading time. Panel (b) is as (a), but for a ring with 106 particles, in its initial configuration (solid line), after 12 orbits (dashed line) and after 90 orbits (dash-dotted line). Our Navier-Stokes viscosity approximates a uniform viscosity well except at very low resolution, where the 'best fit' to the 1-D ring is rather poor in both cases. The initial configuration in panel (b) is representative of the resolution we obtain in our full disc models described in section 2.2. τν = R2 0 12ν . (13) We measure the effective viscosity in the SPH simulations by fit- ting the resulting surface density profiles Σ(R, t) to those obtained from the 1-D explicit code. We perform a least-squares fit for Σ(R) at several different times in each simulation, and measure the effec- tive viscous time-scale τν. The ring spreading is initially linear with the 1-D model as expected, but later the approximation of constant viscosity becomes invalid as the artificial viscosity becomes strong at the ring edges. We only fit spreading times during this initial phase where the linear relationship exists. We are thus able to com- pare the imposed viscosity νin with the measured rate of viscous angular momentum transport νout, in order to determine the accu- racy of our numerical scheme. We further parametrize the viscosity in terms of an effective Shakura & Sunyaev (1973) α parameter by assuming that ν = αcs0H0 at R0. Our measured values are given in table 1, and typical best fits are shown in figure 1 (for the runs with νin = 10−5 at different fractions of the viscous timescale τν). From these results we can estimate the true level of angular momentum transport present in our disc simulations. We see from Table 1 that for both the 105- and 106-particle runs, there is essen- tially no difference in the measured viscosity between the runs with νin = 0 (i.e., artificial viscosity only) and νin = 10−5, and that the viscosity in the higher-resolution runs is smaller than that in the lower-resolution runs (by a factor of approximately 101/3 ≃ 2.15), suggesting that in this regime artificial transport of angular momen- tum is dominant. For larger values of νin, however, the angular mo- mentum transport increases as expected, showing that our imposed Navier-Stokes viscosity is the dominant source of angular momen- tum transport for νin & 10−5 (or, equivalently, α & 0.008). An input α of 0.01 gives a value of ν at R0 of 1.5 × 10−4. The SPH smoothing lengths throughout our simulated discs are comparable to those in our 106-particle spreading rings at radius R0, being of order 0.01 in code units, and so we use this set of rings for comparison. This suggests that the artificial viscosity sets a floor to the effective vis- cosity in the SPH simulations, approximately at or slightly below our canonical imposed value of α = 0.01. We are therefore satis- fied that artificial transport of angular momentum does not domi- Table 1. Summary of ring spreading tests. νin denotes the magnitude of the imposed Navier-Stokes viscosity, as specified in equations 10 and 12, while νout is the measured viscosity in the SPH runs, calculated by fitting the viscous time-scale as in Equation 13. The effective α values are calculated by assuming that νout = αcs0 H0. For very small values of νin the artificial viscosity is the dominant source of angular momentum transport, but for νin & 10−5 the measured viscosity increases as expected. N 105 105 105 105 106 106 106 106 νin 0 10−5 10−4 10−3 0 10−5 10−4 10−3 νout Corresponding α 3.01 × 10−4 3.06 × 10−4 3.45 × 10−4 6.92 × 10−4 1.29 × 10−4 1.38 × 10−4 2.25 × 10−4 1.05 × 10−3 0.019 0.020 0.022 0.045 0.008 0.009 0.015 0.068 nate the viscosity in our disc models3. Moreover, it is known that in the case of a shearing disc, the SPH artificial viscosity behaves similarly to a Shakura-Sunyaev α viscosity (Murray 1996). Conse- quently, although the SPH artificial viscosity prevents us from run- ning simulations with very low disc viscosities, we are confident that numerical angular momentum transport does not dominate our results. 2.2 Numerical set-up and initial conditions We choose a system of units (mass M0, distance R0 and time P0) such that for a planet mass MP and a stellar mass M⋆, MP + M⋆ = 1M0. The unit of time P0 is the Keplerian orbital period for a semi- major axis R0. This choice of units fixes the gravitational constant 3 The exception is run PNM, which uses a much lower explicit viscosity. In this case we expect the artificial viscosity to dominate the angular mo- mentum transport. c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 Table 2. Summary of parameters used in our disc models. Σ0 [Code Units] a Σ0 [g/cm2] b 1.12 × 10−5 1.10 × 10−5 1.12 × 10−4 1.10 × 10−4 1.10 × 10−4 7.03 × 10−4 7.03 × 10−4 102 102 103 103 103 6.4 × 103 6.4 × 103 γ 1 1 1 1 0 0 1 q Model Name 0.005 0.025 0.005 0.025 0.025 0.025 0.025 Low5 Low25 High5 High25 Flat PNM c PNMslope a As the unit of mass depends on the planet mass MP, different values of q give a different Σ0 for the same physical model. b These values correspond to a 1M⊙ star and an initial semi-major axis of 1 au. c Corresponds to the disc model used in PNM01. For this model the vis- cosity ν in equation 12 was 1.59 × 10−6 in dimensionless units, far less than the artificial viscosity, see sections 2.1.1 & 3.1.2. This was set in error, and should have been 2.68 × 10−5 in our units to match that used in PNM01. to be G = 4π2. This system is particularly convenient as for a mass M0 = 1M⊙ and radius R0 = 1au it gives P0 = 1 year. Our initial conditions consist of a gas disc which is axisym- metric about the centre of mass and extends radially from 0.4 to 6 R0. It has a power-law surface density such that Σ(R) = Σ0 R R0!−γ . (14) where Σ0 = Σ(R0) is a reference surface density used for normali- sation. The viscosity ν in equation 12 for each disc model (barring that used in run PNM, see table 2) was chosen such that νΣ was con- stant (i.e., the disc is a steady-state accretion disc), and normalised such that ν0 = 0.01cs 0H0 (where the subscript 0 indicates values at R0). This treatment reduces to a Shakura & Sunyaev (1973) alpha- prescription, with α = 0.01, in the canonical case of a Σ ∝ R−1 surface density profile. The vertical scale-height H is determined by the T ∝ R−1/2 temperature profile described above, so our disc has H/R ∝ R5/4. We normalise the temperature profile by setting H/R = 0.05 at R = R0. (15) GM⋆ R2 In each case our disc is modelled using N = 107 SPH parti- cles. We build the initial conditions by randomly distributing the particles in the radial and azimuthal directions according to equa- tion 14. Particles were given zero velocity in the radial and vertical (z) directions, and azimuthal velocities vφ such that v2 φ R (i.e., Keplerian orbital velocities, with a small correction to account for radial pressure gradients). The particles were distributed in the vertical direction by randomly sampling a Gaussian density pro- file with scale-height H. This set-up is not strictly in equilibrium, due to the discontinuities at the inner and outer disc edges, and consequently transient density waves pass through the disc for ap- proximately one outer orbital period (∼ 15P0). These transients are short-lived, however, and tests performed with an initially relaxed disc show that these do not affect our results in any significant way. We ran 7 models, using the disc parameters described in table 2 and using planet masses q = MP/M⋆ = 0.005 & 0.025. The star and planet were initially placed on the x-axis and given Keplerian orbits around the centre of mass, and each model was allowed to evolve for 200 planetary orbital periods (unless otherwise stated). = + 1 ρ ∂P ∂R . c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 A limit on eccentricity growth 5 Figure 2. Comparison between initially eccentric models with just artificial viscosity (black line) and with both artificial viscosity and a Navier-Stokes viscosity (red line). Both used disc Low25 (see table 2) and were given an initial eccentricity e0 = 0.05. The model without the Navier-Stokes vis- cosity shows small initial damping of eccentricity which soon flattens off, while the model with the full viscosity scheme implemented sees continued and increased eccentricity damping as it evolves. This shows that the SPH artificial viscosity is not causing spurious eccentricity damping. In the case of the full Navier-Stokes viscosity model, at later times the eccentricity de- cay reversed and began to grow again. This is due to the eccentric planet causing stronger disc eccentricity, and is in agreement with the findings of D'Angelo et al. (2006). The simulations were run on the alice HPC cluster at the University of Leicester4, using 128 parallel cores. 3 RESULTS 3.1 Code tests 3.1.1 Numerical eccentricity damping As an initial test, and to ensure that the effects we see are physical, we first verified that a planet on an eccentric orbit does not undergo spurious eccentricity damping due to the SPH artificial viscosity (or other numerical effects). To this end we ran 2 realisations the disc model Low25, where in each case the planet was given an initial eccentricity e0 = 0.05. In one version of this model the Navier- Stokes prescription described in equations 8 to 10 was switched off, so that in this case the only source of angular momentum transport was from the artificial viscosity. These were allowed to evolve for 125 orbits, and the eccentricity evolution is shown in figure 2. In the model with no physical (Navier-Stokes) viscosity we see very little initial eccentricity decay during the initial period while the disc settles into an equilibrium state and the planet opens its full gap. A loss of some eccentricity during this phase is fully expected and is in agreement with simulations by Bitsch & Kley (2010), who found that for non-gap opening planets eccentricity is damped by the surrounding gas. With the physical viscosity switched on we see additional damping of eccentricity, at a much more pronounced level. In both cases we see exponential damp- ing after the initial phase (beyond ∼ 75 orbits), after the planet has fully cleared its gap. The rate of eccentricity damping during this 4 See http://go.le.ac.uk/alice 6 A. C. Dunhill, R. D. Alexander & P. J. Armitage Figure 3. Surface density evolution of the central region of our disc model PNM (see table 2), roughly equivalent to run N4 from PNM01. The distance unit is equal to the initial separation between the star and planet. Times shown are in units of the initial orbital period of the planet. The eccentricity evolution for this model is shown in figure 4. After the inner disc clears (upper panels), the presence of the planet drives the inner edge of the disc eccentric (bottom panels). As the disc evolves it exerts torques back upon the planet, causing the planet's eccentricity to grow. [All of the SPH maps presented here were rendered using splash (Price 2007).] phase is similar between the two models. This is because once the gap has fully formed, the planet is not directly interacting with the gas to any great extent so the angular momentum exchange here is due to gravitational resonances. At later times, the eccentricity began to rise again in the case of the full viscosity model. This is expected, as D'Angelo et al. (2006) found that even for very low planet masses an initially eccentric planet can undergo far stronger eccentricity growth than one on an initially circular orbit. Conse- quently we conclude that the SPH artificial viscosity is not causing significant spurious eccentricity damping in our disc models. as possible in form to that used in their calculations5. Using the pa- rameters for run PNM given in table 2 this approximates run N4 from that paper, with the obvious caveat that our simulations are in 3-D. Unlike our other models, the normalisation for the viscosity law described in section 2.2 was taken to be 1.59 × 10−6 in our di- mensionless code units6. This is constant across the disc, fulfilling the steady-state accretion requirement that νΣ be constant. Note, however, that with this setup the angular momentum transport due to the explicit viscosity is smaller than that due to the SPH artificial viscosity (see Section 2.1.1), so in practice our test calculation is somewhat more viscous than that of PNM01 (by a factor of a 2 -- 3). This model was allowed to evolve for 340 orbital periods. The evo- 3.1.2 PNM01 Result As a further test, we have also attempted to reproduce the results of PNM01. To this end we have created a disc model that is as near 5 PNM01 used a 2-D fixed-grid code for their simulations, so it is not pos- sible for us to run a completely identical simulation. 6 Equivalent to 2.50 × 10−6 in the units of PNM01. c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 A limit on eccentricity growth 7 Figure 4. Evolution of planet eccentricity for the disc model PNM, which is approximately equivalent to run N4 from PNM01 (see table 2). We find growth of eccentricity in general agreement with that paper. Surface density plots from this run are shown in figure 3. The ∼ 100 orbital period oscilla- tions are due to the relative precession of the planet and the eccentric inner edge of the disc. lution of the planet's eccentricity is shown in figure 4, while a series of surface density maps as the disc evolves are shown in figure 3. The evolution of our disc structure is broadly in line with that found by PNM01, with the planet rapidly opening a wide gap in the disc, and the inner part of the disc quickly accreting onto the central star. As the system evolves the planet begins to drive ec- centricity in the disc at its inner edge, while its own orbit remains essentially circular. At later times this is no longer the case and the planet's orbit becomes significantly eccentric [above the ∼ 0.01- 0.05 level required by Ogilvie & Lubow (2003) for non-coorbital corotation resonances to saturate, at which point further eccentric- ity growth is expected]. The long-period oscillations in eccentricity seen in figure 4 are due to the relative precession of the planet and the eccentric disc inner edge of the disc. The level of eccentricity growth seen in our simulations is somewhat less than that seen by PNM01, but it is broadly com- parable. Moreover, given our larger effective disc viscosity, and the inherent differences between the methods (2-D fixed-grid versus our full 3-D SPH calculation), we do not expect exact agreement. We also note that our simulations are extremely computationally expensive (using up to approximately 150,000 CPU hours per run), so we are limited in how long we can evolve our models for. Conse- quently we are unable to find a level of eccentricity at which growth saturates (the eccentricity was still growing at the end of our sim- ulation), but otherwise we find good agreement between our 3-D results and the 2-D simulations of PNM01. 3.2 The effect of surface density & planet mass 3.2.1 Planet mass To test the effect of different planet masses, we ran models with both Low and High surface densities (see table 2) with planet-star mass ratios of q = 0.005 & 0.025. For a 1M⊙ star this corresponds to planet masses of approximately 5 and 25MJup respectively. Fig- ure 5 shows the time evolution of the orbital eccentricity in each case. For both of the models with q = 0.005 (Low5 and High5), no eccentricity growth was seen at any level. For the models with c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 Figure 5. Eccentricity evolution for planets of different masses in disc mod- els with different surface densities. Low5 and High5 have a planet-star mass ratio of q = 0.005, while Low25 and High25 have q = 0.025. Low and High refer to the choice of disc surface density -- see table 2 for the values -- but all have a power-law index of γ = 1. q = 0.025 (Low25 and High25) some modest growth was seen, but none above e = 0.005. This is consistent with the conclusions of both PNM01 and D'Angelo et al. (2006), who found that eccentricity is first induced in the disc, driven by the outer 3:1 Lindblad resonance. We clearly see this disc eccentricity in both runs with q = 0.025, but not in either of the lower mass planet cases. A comparison of the surface density maps for both planet masses in the Low model is shown in figure 6. The reason for the lack of eccentricity growth in both of the q = 0.005 models is that the 3:1 outer Lindblad resonance induced by the planet is too weak to affect the disc structure sig- nificantly. The difference in eccentricity evolution between models Low25 and High25 arises because the higher surface density disc is simply more massive, and consequently is able to exert stronger torques upon the planet, resulting in more (but still extremely lim- ited) eccentricity growth. We are confident that in the case of both planet masses, eccen- tric co-rotation resonances are resolved where present. Using the width formula provided by Masset & Ogilvie (2004), our smooth- ing lengths are smaller than the resonant widths at the nominal res- onant locations by a factor of at least two, for resonances that are not fully in the open gap. For the case of the 5 MJupplanet, this is assuming an eccentricity of 10−2, as for the eccentricity measured from the simulations the prescribed widths are vanishing. 3.2.2 Radial surface density profile We now consider the effect of the disc surface density profile on the evolution of the embedded planet. To this end we have run two ad- ditional models with the same star-planet mass ratio of q = 0.025, and different radial surface density profiles (γ = 1 & 0). The disc models were Flat and PNMslope (see table 2), and each was evolved for 200 orbits. The evolution of the orbital eccentricity for these runs, along with two previously described (High25 and PNM, for the purposes of comparison), are shown in figure 7. Again, due to computational limitations we have not run these models for long enough to determine at what level of eccentricity growth saturates; we are instead more concerned here with determining the condi- tions under which growth will occur. We see again the eccentric in- 8 A. C. Dunhill, R. D. Alexander & P. J. Armitage Figure 7. Eccentricity evolution for disc models with different surface den- sity profiles. The black line is the same as in figure 4, truncated at t = 200P0 for reference, and the green line is the same as in figure 5. These models show that while higher values of Σ0 give consistently stronger eccentricity growth, shallower radial profiles (lower values of γ, see table 2) also show stronger growth. The oscillations in the curves are again due to the rela- tive precession of the planet and the eccentric disc. Note also that we ran model PNMslope (red line) for a further 100 orbits (not shown here) to ver- ify that the eccentricity does grow indeed continue to grow; the declining eccentricity at t = 200 is simply due to this precession effect. level of eccentricity growth seen. The two models with γ = 1 (High25 and PNMslope) show slower, weaker growth of eccentric- ity than their counterparts with the same normalisation value of Σ0 but a flat γ = 0 radial power-law dependance. There are two pri- mary reasons for this. Firstly, a flatter radial profile puts less mass inside the planet's orbit. The inner disc therefore accretes on to the star more rapidly for flatter surface density profiles (i.e., lower val- ues of γ), and as eccentricity only begins to grow considerably after the inner disc has accreted, this takes place sooner for flatter surface density profiles. A flatter radial profile also puts more mass into the outer Lindblad resonances, which are responsible for the torques that drive eccentricity growth. This changes the torque balance on the planet, and gives rise to more rapid eccentricity growth. In addi- tion, in the special case when Lindblad torques cancel (as proposed by Goldreich & Sari 2003), the resultant net torque is a function of the surface density gradient, with a steeper power-law dependance damping eccentricity. However, we do not expect this effect to be active in our models, as the planets open gaps sufficiently wide that co-rotation resonances are ineffective. To investigate this effect further we have followed the method of Artymowicz et al. (1991) to calculate values of e, time averaged over several orbits at the end of each simulation. The radial con- tributions to this, normalised against the magnitude of the surface density in each model, are shown in figure 9 for all models with q = 0.025. In the low-surface density limit (models Low25, High25 and Flat), we see that the eccentricity is being damped by a peak at R ∼ 1.8, and the magnitude of the surface density acts as a scal- ing factor with only a very weak dependence on the radial slope (γ). In these three disc cases, e is always negative. By contrast, In the models with higher surface densities, where eccentricity is growing (models PNM and PNMslope), the sense of the contribu- tion reverses, and the total e is positive. Here there is also a more pronounced effect of the radial gradient in the surface density. It is unclear exactly what mechanism causes this effect, but it appears that there is a threshold surface density above which the analysis c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 Figure 6. Surface density maps after 200 planetary orbits: panel (a) shows the central region of model Low5 (q = 0.005) and panel (b) shows Low25 (q = 0.025; see table 2). It is clear from panel (a) that the lower mass planet has a far smaller perturbing effect on its host disc, driving spiral waves but not further disrupting the disc shape or structure. This is in stark contrast to panel (b), where the higher mass planet has driven the inner edge of the disc quite eccentric. ner disc driven by the presence of the massive planet, as described above, and again find that its effect on the eccentricity of the planet depends strongly on the precise disc model used. The results from these models show two things. First, the mag- nitude of the disc surface density has a strong effect on the eccen- tricity evolution of an embedded planet. Figure 7 clearly shows that the two models with Σ0 = 7.03 × 10−4 (PNM and PNMs- lope) see significant eccentricity growth, while the models with Σ0 = 1.10 × 10−4 (High25 and Flat) do not. This suggests that eccentricity can only be excited above a threshold surface density. This behaviour was suggested by PNM01 but not investigated in detail. Comparing the ratio between the planet mass and the local disc mass (approximated by Σπa2 in the unperturbed disc) suggests that values between ∼ 1 − 13.5 may result in eccentricity growth (see figure 8). We also note in passing that this is a surprisingly strong effect for a relatively modest (factor of 6.4) change in the disc surface density; the surface densities in real protoplanetary discs are thought to change by factors & 103 over their lifetimes (e.g., Hartmann et al. 1998; Alexander, Clarke, & Pringle 2006). The power-law index γ also has quite a strong effect on the A limit on eccentricity growth 9 been able to run our simulations at very high resolution, and given the results of the various tests presented in sections 2.1.1 and 3.1.1, we are confident that the SPH artificial viscosity is not a dominant influence on our results. We further note that our approximation of a locally isothermal equation of state is somewhat idealised, and in particular may not give an accurate description of the spiral density waves induced in the disc. Bitsch & Kley (2010) looked at the evolution of initially eccentric planets in fully radiative discs, and made comparisons to models which used an isothermal approximation. They only found the results to be inconsistent for relatively low planet masses (. 0.6 MJup). The planet masses considered here are far above this, well into the regime where the isothermal approximation holds, and so we do not expect our use of an isothermal equation of state to introduce significant uncertainties in our results. Perhaps more of a concern is the viscosity in our simulated discs. We are confident that the SPH artificial viscosity in our sim- ulations does not dominate our results, but it does set a floor to the range of viscosities we can explore: with current computa- tional capabilities we cannot model discs with α . 0.017. Pre- vious simulations using grid-based methods have adopted a wide range of viscosities and temperature prescriptions, with effective α values ranging from α ∼ 10−4 -- 10−2 (e.g., Papaloizou et al. 2001; D'Angelo et al. 2006; Bitsch & Kley 2010). It is not clear which value(s) of α is most appropriate, but we note that our simula- tions have a viscosity that is somewhat larger (by a factor ≃ 2 -- 3) than the canonical values adopted in previous (mostly 2-D) studies. We note also that values inferred from observations of protoplan- etary discs are typically of order α ∼ 0.01 (e.g. Hartmann et al. 1998; King, Pringle, & Livio 2007). Presumably the surface den- sity threshold for eccentricity growth must depend on the disc vis- cosity (and indeed temperature), but unfortunately we are not able to explore this issue further. It must also be borne in mind that the Navier-Stokes viscos- ity used in our models is merely a first-order approximation, at- tempting to mimic the effect globally of a process that occurs far below the scales we are able to resolve -- namely the magnetorota- tional instability (MRI) thought to drive angular momentum trans- port in protoplanetary discs (Balbus & Hawley 1991). By adopt- ing a Shakura & Sunyaev (1973) alpha-prescription we implicitly assume that the small-scale effects of turbulence in the disc be- have like a viscosity on large scales, but it is not clear whether this approximation holds for length scales . H. In our simula- tions the planets open gaps on length scales & H, so turbulent fluctuations on smaller scales are unlikely to have a strong effect. However, there is some overlap between the length-scales consid- ered here and the typical scales of MRI turbulence, so we note that our results may not hold if the MRI drives significant power on moderate or large scales (as suggested by recent simulations; Simon, Beckwith, & Armitage 2012). Detailed investigation of this issue is beyond the scope of this paper, but if the viscous approxi- mation does break down at scales ∼ H then it seems likely that ec- centricity growth could be affected, particularly for low-mass plan- ets. 7 Note that angular momentum transport by the SPH artificial viscosity scales with the smoothing length h (Murray 1996), which in 3-D scales as h ∝ N1/3. Reducing the numerical viscosity to α . 0.001 would therefore require us to increase the SPH particle number N by a factor ∼ 103, which is not currently feasible. Figure 8. Comparison between the local unperturbed disc mass (approxi- mated by πΣa2) and planet mass for simulations presented here and in two other papers. Crosses indicate simulations from this paper, squares from PNM01 and triangles from D'Angelo et al. (2006). Green symbols indi- cate that eccentricity growth was seen, while red indicates that it was not. The choice of units on the x-axis gives the reference surface density at the semimajor-axis of the planet for that model (except for the models of D'Angelo et al. (2006), which placed their planets at 5.2au; in this case the x-axis value is 5.22 times their reference surface density). The dashed lines show where the ratio of Mp/πΣa2 cross values of 1.0 and 13.5. We sug- gest that these values represent cases where the planet is massive enough to significantly perturb the inner edge of the disc, but where the disc is also massive enough that it can exert sufficiently strong torques upon the planet, although we note that this ratio is not the lone deciding factor. We also draw attention to the fact that the single point below the threshold Mp/πΣa2 = 1.0 has been questioned by Masset & Ogilvie (2004) as being affected by spu- rious numerical factors, and thus the lower limit may not be real. We do not expect this to extend into the sub-jovian regime, where gap-opening is not necessarily efficient enough to allow similar modes of eccentricity growth (although this also depends on other factors including disc viscosity). performed in the low disc mass limit (e.g. Goldreich & Tremaine 1980) no longer applies. Our results, and those of previous simula- tions (see Figure 8), suggest that this threshold is given by πΣa2 > Mp/C (16) where C is a constant with a value C ∼ 10. Discs with surface den- sities below this threshold are unable to excite significant eccentric- ity in the planet's orbit. This behaviour was suggested by PNM01, and our results support their tentative prediction. The threshold must presumably depend on several other factors (disc viscosity, H/R, etc.), and a complete exploration of this parameter space is beyond the scope of this work. Nevertheless, our results show that eccentricity is only excited in discs with very high surface densities. 4 DISCUSSION 4.1 Numerical limitations The biggest potential numerical problem with this work is that the SPH artificial viscosity may cause spurious eccentricity damping. Artificial viscosity can be especially problematic for shearing-disc type problems such as this, as the differential rotation is often mis- taken by the algorithm for a shock. The effects of artificial viscosity typically scale with the SPH smoothing length h, and consequently are a strong function of numerical resolution. However, we have c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 10 A. C. Dunhill, R. D. Alexander & P. J. Armitage Figure 9. Radial contributions to e, calculated using a Gaussian perturba- tion approach following the method of Artymowicz et al. (1991). These val- ues are time-averaged over 5 orbital periods of the planet at the end of each simulation. Models Low25, High25 and Flat clearly show that in the low- surface density limit, e is approximately linear with the magnitude of the surface density, with a weak dependance on the radial profile γ. For models PNM and PNMslope it can be seen that above the threshold for eccentricity growth, the sign of the curves flip and the net e becomes positive. In this limit the radial profile of the surface density becomes more important, and the effect is no longer linear with its magnitude. 4.2 Applications to real systems Our major result is that resonant torques are not generally an ef- ficient means of exciting eccentricity, and that planet-disc interac- tions are unlikely to be responsible for the eccentricities seen in the majority of exoplanet systems. While we do find eccentricity growth in agreement with PNM01, we note that their calculations considered a very massive planet in a disc with a very high surface density. Their reference surface density of 6.4 × 104g/cm2 at 1 au is a factor of several larger than predicted by the Minimum Mass So- lar Nebula (Weidenschilling 1977) or more realistic accretion disc models (e.g., Hartmann et al. 1998). We find that a modest reduc- tion in the disc surface density results in no significant eccentricity growth for similarly massive planets, in agreement with behaviour suggested by PNM01. Consequently it is unlikely that this mecha- nism will be able to excite eccentricity in real protoplanetary discs. We also fail to find eccentricity growth for lower planet masses, but in this case we treat our results with more caution. Using 2-D simulations D'Angelo et al. (2006) found eccentricity growth up to e ∼ 0.1 for planets between 1-3 MJup, but typically this occurred on time-scales of thousands of orbits. Given the high computational cost of our 3-D simulations we are not able to follow their evolution for such long time-scales, and consequently we can- not rule out this slower mode of growth. However, we note that the disc model used by D'Angelo et al. (2006) also uses a very large surface density: if we re-scale their model to match the units used here, their surface density normalisation (Σ0) becomes 2.0 × 103 g/cm2 at 1 au, larger than in our High models. We also note that their relatively flat choice surface density profile (γ = 1/2) is likely to promote eccentricity growth. Although our simulations do not allow us to rule out growth on very long time-scales, analysis like that shown in figure 9 shows that the disc gives a negative con- tribution to the planet's e which indicates that eccentricity will not grow unless the disc structure changes significantly. Instead it seems likely that differences in the choice of disc model are respon- sible for the apparent discrepancy between our results and those of D'Angelo et al. (2006). As noted in section 3.2.2, there are two reasons why the radial surface density profile plays a role in exciting eccentricity growth. At a basic level, a lower value of γ puts mass mass into the outer Lindblad resonances and conversely less mass into the inner reso- nances. As the mode of growth relies on the 1:3 outer resonance (PNM01), this favours eccentricity growth by strengthening that resonance. In the case of lower mass planets, where a gap is not fully opened, there is another effect which takes place that brings the radial surface density gradient into play. Goldreich & Sari (2003) suggest that in a near-Keplerian disc where an outer and in- ner Lindblad resonance cancel to a reasonable approximation, the resulting net torque scales with dΣ/dr rather than with Σ. This im- plies that not only the Σ0 level but also the radial profile may be- come as important as the planet mass in discerning the physical contribution of resonant torques for low mass planets. We believe that it is the former effect that we are seeing in our simulations, with both the surface density profile and its normalisation level both having a strong effect on if, and how, the orbital eccentricity of an embedded planet will grow (figure 7). This analysis of the effect of both the magnitude and radial profile of the surface density is supported by looking at the ra- dial contributions to the change in eccentricity in different models, shown in figure 9. Below the threshold for eccentricity growth the surface density profile has only a small effect, and its magnitude is an approximately linear scaling factor, which is in agreement with the findings of Artymowicz et al. (1991). In contrast, above the threshold for growth the sense of the contributions reverse and the difference between models with the same Σ0 but different γ be- comes more pronounced. We are confident that this effect is real, and warrants further study, but detailed investigation is beyond the scope of this paper. Other disc parameters including viscosity and temperature structure must also have an effect on this result, but we have not investigated them here. We can extend this analysis by taking equation 16 to be a nec- essary condition for eccentricity growth: CπΣa2 & Mp (17) where C ∼ 10 is a constant. However, giant planets continue to accrete via tidal streams even after opening a gap in the disc, and in discs which are sufficiently massive to meet this condition they are likely to accrete rapidly. This accretion increases Mp, making it less likely that the threshold for eccentricity growth will be met. The critical issue is therefore one of time-scales: does eccentricity grow on a shorter time-scale than the planet's mass? For massive giant planets (& 5MJup) tidal torques strongly suppress accretion from the disc on to the planet (e.g., Lubow, Seibert, & Artymowicz 1999), and we therefore expect eccentricity to grow more rapidly than the planet mass. However, our results suggest that eccentricity growth in this regime requires very high disc surface densities, > 103g cm−2 (see Fig.8), and such massive discs are not commonly observed (see below). By contrast, lower-mass planets (∼ 1MJup) accrete very effi- ciently from their parent discs (e.g., D'Angelo, Henning, & Kley Mp = 2002). We can parametrize the planetary accretion rate as ǫ Md, where Md = 3πνΣ is the disc accretion rate in the absence of the planet. The parameter ǫ represents the efficiency of accretion on to the planet; simulations show that this efficiency has a peak value of ǫ ∼ 1 for planets of approximately Jupiter mass, and declines to higher planet masses (Lubow et al. 1999; D'Angelo et al. 2002, see c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 also Veras & Armitage 2004). If we substitute ν = αΩH2 we can rearrange equation 16 to find . −2 . (18) C 3ǫ α−1Ω−1(cid:18) H a (cid:19) Mp Mp The quantity on the left-hand-side is the time-scale for planet growth through accretion of gas from the disc, τaccrete, so we see that meeting the condition for eccentricity growth (equation 16) also sets an upper limit to the time-scale for planet growth by ac- cretion. Taking standard values of α = 0.01 and H/a = 0.1, and assuming that ǫ ∼ 1, we find τaccrete . 3 × 104 Ω−1 , (19) for planets of approximately Jupiter mass. This is comparable to the time-scale for eccentricity growth seen in previous studies of Jupiter-mass planets (D'Angelo et al. 2006). We therefore suggest that in this case the planet would accrete rapidly, and 'migrate' out of the region of allowed eccentricity growth highlighted in figure 8 before it attains a significant eccentricity. We have shown that in moderately viscous discs (with α ∼ 0.01), eccentricity growth only occurs if the disc surface density is high. Unfortunately, observational constraints on the surface densi- ties of discs are extremely weak, and at au scales such as those con- sidered here, almost non-existent. Andrews & Williams (2007) and Andrews et al. (2009, 2010) made systematic studies of protoplan- etary discs in Taurus and Ophiuchus, fitting surface density profiles to submillimeter observations of thermal dust emission. However, the angular resolution of such observations means that they can only resolve scales a few tens of au (in the very best cases), and are not sensitive to the region of most interest for planet formation. If we extrapolate their results down to the unresolved inner disc, their fits yield values of Σ0 at au radii between ∼ 30 and ∼ 1700 g/cm2, with most values lying at ∼ 700 − 800 g/cm2. The radial power- law indices (γ) they fit to their observations range between 0.4 and 1.1, with a clear preference towards the upper end of this range. These correspond approximately to our High disc model, and have both lower surface densities and steeper power-law indices than the disc models used by PNM01 and D'Angelo et al. (2006). However it must be stressed that extrapolating these sub-mm observations is by no means justified. It has even been suggested that substan- tial mass reservoirs may exist in 'dead zones' close to the star (e.g., Gammie 1996; Hartmann et al. 2006; Zhu et al. 2010), but there are currently no useful constraints on disc surface densities at au radii. ALMA may soon provide real constraints on protoplanetary discs with far higher angular resolution, but it will be some time before it is able to probe radii of a few au (e.g., Cossins, Lodato, & Testi 2010). Despite this, we argue that significant eccentricity growth due to the planet-disc interaction is unlikely given realistic protoplane- tary disc conditions. There is an increasing consensus in the litera- ture that this is the case -- while D'Angelo et al. (2006) did see ec- centricity growth by this method, it did not rise above ∼ 0.15, lower than the observed values of 0.2 -- 0.3 that this effect is invoked to explain. Similarly, the semi-analytic models Moorhead & Adams (2008) were unable to reproduce the observed low-eccentricity dis- tribution. Coupled with our findings, it seems that the planet-disc interaction alone is incapable of reproducing the eccentricities seen in exoplanet observations. This negative result of course begs the question of what the true origin of exoplanet eccentricities is. An emerging consensus seems to be that scattering events are responsible for most if not all of the distribution (e.g. Chatterjee et al. 2008). This is backed up c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12 A limit on eccentricity growth 11 by a vast number of tightly-packed multi-planet systems observed by Kepler (Batalha et al. 2012) and by a plethora of planetary sys- tems that have clearly undergone strong interactions (e.g. highly in- clined and retrograde planets; Wright et al. 2011). However, there is some suggestion that the low-eccentricity end of the distribution perhaps cannot be explained by this mechanism (Goldreich & Sari 2003; Juri´c & Tremaine 2008), and further work in this area is still required. 5 CONCLUSIONS We have performed high-resolution 3D SPH simulations of giant planets embedded in protoplanetary discs. For high disc surface densities and planet masses we find that the planet-disc interac- tion leads to eccentricity growth, in agreement with previous stud- ies. However, we have shown that for realistic planet masses and disc properties, the planet-disc interaction is incapable of exciting significant orbital eccentricity growth. While we do not investigate the effect of different disc viscosities, we identify a threshold sur- face density for eccentricity growth, and note that this threshold is rarely, if ever, met in real systems, except in cases where the timescale for eccentricity growth is comparable to the timescale for mass accretion by the planet. We conclude from our simula- tions that in the case of a real giant planet, the interaction with its parent disc is unlikely to yield growth of its orbital eccentricity at measurable levels. Therefore we suggest that the low but non-zero exoplanet eccentricities observed, not accounted for by simulations of planet-planet scattering events, must have some other origin. ACKNOWLEDGMENTS We thank the anonymous referee for useful comments that helped clarify the text. ACD is supported by an Science & Tech- nology Facilities Council (STFC) PhD studentship. RDA ac- knowledges support from STFC through an Advanced Fellow- ship (ST/G00711X/1). PJA was supported in part by NASA (NNX09AB90G, NNX11AE12G) and by the NSF (0807471). Theoretical astrophysics research in Leicester is supported by an STFC Rolling Grant. This research used the ALICE High Per- formance Computing Facility at the University of Leicester. Some resources on ALICE form part of the DiRAC Facility jointly funded by STFC and the Large Facilities Capital Fund of BIS. REFERENCES Alexander R. D., Clarke C. J., Pringle J. E., 2006, MNRAS, 369, 229 Andrews S. M., Williams J. P., 2007, ApJ, 659, 705 Andrews S. M., Wilner D. J., Hughes A. M., Qi C., Dullemond C. P., 2009, ApJ, 700, 1502 Andrews S. M., Wilner D. J., Hughes A. M., Qi C., Dullemond C. P., 2010, ApJ, 723, 1241 Artymowicz P., Clarke C. J., Lubow S. H., Pringle J. E., 1991, ApJ, 370, L35 Balbus S. A., Hawley J. F., 1991, ApJ, 376, 214 Balsara D. S., 1995, J. Comput. Phys., 121, 357 Batalha N. M. et al., 2012, arXiv:1202.5852v1 Bitsch B., Kley W., 2010, A & A, 523, A30 Chatterjee S., Ford E. B., Matsumura S., Rasio F. A., 2008, ApJ, 686, 580 12 A. C. Dunhill, R. D. Alexander & P. J. Armitage Chiang E. I., Fischer D., Thommes E., 2002, ApJ, 564, L105 Cossins P., Lodato G., Testi L., 2010, MNRAS, 407, 181 Cuadra J., Armitage P. J., Alexander R. D., Begelman M. C., 2009, MNRAS, 393, 1423 Cuadra J., Nayakshin S., Springel V., Di Matteo T., 2006, MN- RAS, 366, 358 D'Angelo G., Henning T., Kley W., 2002, A&A, 385, 647 D'Angelo G., Lubow S. H., Bate M. R., 2006, ApJ, 652, 1698 Ford E. B., Havlickova M., Rasio F. A., 2001, Icarus, 150, 303 Gammie C. F., 1996, ApJ, 457, 355 Goldreich P., Sari R., 2003, ApJ, 585, 1024 Goldreich P., Tremaine S., 1980, ApJ, 241, 425 Haisch K. E., Jr., Lada E. A., Lada C. J., 2001, ApJ, 553, L153 Hartmann L., Calvet N., Gullbring E., D'Alessio P., 1998, ApJ, 495, 385 Hartmann L., D'Alessio P., Calvet N., Muzerolle J., 2006, ApJ, 648, 484 Juri´c M., Tremaine S., 2008, ApJ, 686, 603 Kane S. R., Ciardi D. R., Gelino D. M., von Braun K., 2012, MN- RAS, 425, 757 Kenyon S. J., Hartmann L., 1987, ApJ, 323, 714 King A. R., Pringle J. E., Livio M., 2007, MNRAS, 376, 1740 Kley W., Dirksen G., 2006, A&A, 447, 369 Kozai Y., 1962, AJ, 67, 591 Lidov M. L., 1962, Planet. Space Sci., 9, 719 Lodato G., Price D. J., 2010, MNRAS, 405, 1212 Lubow S. H., Seibert M., Artymowicz P., 1999, ApJ, 526, 1001 Masset F. S., Ogilvie G. I., 2004, ApJ, 615, 1000 Moorhead A. V., Adams F. C., 2008, Icarus, 193, 475 Morris J. P., Monaghan J. J., 1997, J. Comput. Phys., 136, 41 Murray J. R., 1996, MNRAS, 279, 402 Naoz S., Farr W. M., Rasio F. A., 2012, ApJ, 754, L36 Ogilvie G. I., Lubow S. H., 2003, ApJ, 587, 398 Papaloizou J. C. B., Nelson R. P., Masset F., 2001, A&A, 366, 263 Price D. J., 2004, Ph.D. thesis, Univ. Cambridge Price D. J., 2007, Proc. Astron. Soc. Aust., 24, 159 Price D. J., 2012, J. Comput. Phys., 231, 759 Pringle J. E., 1981, ARA&A, 19, 137 Pringle J. E., Verbunt F., Wade R. A., 1986, MNRAS, 221, 169 Rasio F. A., Tout C. A., Lubow S. H., Livio M., 1996, ApJ, 470, 1187 Sargent A. I., Beckwith S., 1987, ApJ, 323, 294 Shakura N. I., Sunyaev R. A., 1973, A&A, 24, 337 Simon J. B., Beckwith K., Armitage P. J., 2012, MNRAS, 2808 Springel V., 2005, MNRAS, 364, 1105 Takeda G., Rasio F. A., 2005, ApJ, 627, 1001 Tanaka H., Takeuchi T., Ward W. R., 2002, ApJ, 565, 1257 Veras D., Armitage P. J., 2004, MNRAS, 347, 613 Weidenschilling S. J., 1977, Ap&SS, 51, 153 Wright J. T. et al., 2011, PASP, 123, 412 Wu Y., Lithwick Y., 2011, ApJ, 735, 109 Zhu Z., Hartmann L., Gammie C., 2010, ApJ, 713, 1143 c(cid:13) 2012 RAS, MNRAS 000, 1 -- 12
1701.07881	1	1701	2017-01-26T21:26:29	Prevalance of Chaos in Planetary Systems Formed Through Embryo Accretion	[ "astro-ph.EP" ]	The formation of the solar system's terrestrial planets has been numerically modeled in various works, and many other studies have been devoted to characterizing our modern planets' chaotic dynamical state. However, it is still not known whether our planet's fragile chaotic state is an expected outcome of terrestrial planet accretion. We use a suite of numerical simulations to present a detailed analysis and characterization of the dynamical chaos in 145 different systems produced via terrestrial planet formation in Kaib and Cowan (2015). These systems were created in the presence of a fully formed Jupiter and Saturn, using a variety of different initial conditions. They are not meant to provide a detailed replication of the actual present solar system, but rather serve as a sample of similar systems for comparison and analysis. We find that dynamical chaos is prevalent in roughly half of the systems we form. We show that this chaos disappears in the majority of such systems when Jupiter is removed, implying that the largest source of chaos is perturbations from Jupiter. Chaos is most prevalent in systems that form 4 or 5 terrestrial planets. Additionally, an eccentric Jupiter and Saturn is shown to enhance the prevalence of chaos in systems. Furthermore, systems in our sample with a center of mass highly concentrated between 0.8 -1.2 AU generally prove to be less chaotic than systems with more exotic mass distributions. Through the process of evolving systems to the current epoch, we show that late instabilities are quite common in our systems. Of greatest interest, many of the sources of chaos observed in our own solar system (such as the secularly driven chaos between Mercury and Jupiter) are shown to be common outcomes of terrestrial planetary formation.	astro-ph.EP	astro-ph	Draft version January 30, 2017 Preprint typeset using LATEX style emulateapj v. 5/2/11 7 1 0 2 n a J 6 2 . ] P E h p - o r t s a [ 1 v 1 8 8 7 0 . 1 0 7 1 : v i X r a PREVALENCE OF CHAOS IN PLANETARY SYSTEMS FORMED THROUGH EMBRYO ACCRETION Matthew S. Clement1 & Nathan A. Kaib1 Draft version January 30, 2017 ABSTRACT The formation of the solar system's terrestrial planets has been numerically modeled in various works, and many other studies have been devoted to characterizing our modern planets' chaotic dynamical state. However, it is still not known whether our planets fragile chaotic state is an expected outcome of terrestrial planet accretion. We use a suite of numerical simulations to present a detailed analysis and characterization of the dynamical chaos in 145 different systems produced via terrestrial planet formation in Kaib & Cowan (2015). These systems were created in the presence of a fully formed Jupiter and Saturn, using a variety of different initial conditions. They are not meant to provide a detailed replication of the actual present solar system, but rather serve as a sample of similar systems for comparison and analysis. We find that dynamical chaos is prevalent in roughly half of the systems we form. We show that this chaos disappears in the majority of such systems when Jupiter is removed, implying that the largest source of chaos is perturbations from Jupiter. Chaos is most prevalent in systems that form 4 or 5 terrestrial planets. Additionally, an eccentric Jupiter and Saturn is shown to enhance the prevalence of chaos in systems. Furthermore, systems in our sample with a center of mass highly concentrated between ∼0.8–1.2 AU generally prove to be less chaotic than systems with more exotic mass distributions. Through the process of evolving systems to the current epoch, we show that late instabilities are quite common in our systems. Of greatest interest, many of the sources of chaos observed in our own solar system (such as the secularly driven chaos between Mercury and Jupiter) are shown to be common outcomes of terrestrial planetary formation. Thus, consistent with previous studies such as Laskar (1996), the solar system's marginally stable, chaotic state may naturally arise from the process of terrestrial planet formation. Keywords: Chaos, Planetary Formation, Terrestrial Planets Received; Accepted 1. INTRODUCTION Our four terrestrial planets are in a curious state where they are evolving chaotically, and are only marginally stable over time (Laskar 1996, 2008; Laskar & Gastineau 2009). This chaos is largely driven by interactions with the 4 giant planets. However our understanding of the dynamical evolution of the gas giants, particularly Jupiter and Saturn, has changed drastically since the in- troduction of the Nice Model (Gomes et al. 2005; Mor- bidelli et al. 2005; Tsiganis et al. 2005). The classical model of terrestrial planetary formation, where planets form from a large number of small em- bryos and planetesimals that interact and slowly accrete, is the basis for numerous studies of planetary evolution (e.g. Chambers 2001a; O'Brien et al. 2006; Chambers 2007; Raymond et al. 2009b; Kaib & Cowan 2015). Us- ing direct observations of proto-stellar disks (Currie et al. 2009), it is clear that free gas disappears long before the epoch when Earth's isotope record indicates the conclu- sion of terrestrial planetary formation (Halliday 2008). For these reasons, a common initial condition taken when numerically forming the inner planets is a fully formed system of gas giants at their current orbital locations. Many numerical models have produced planets using this method. However, none to date have analyzed the chaotic nature of fully evolved accreted terrestrial plan- ets up to the solar system's current epoch. It should be 1 HL Dodge Department of Physics Astronomy, University of Oklahoma, Norman, OK 73019, USA & corresponding author email: [email protected] noted that other works have modeled the outcome of ter- restrial planetary formation up to 4.5 Gyr. Laskar (2000) evolved 5000 such systems from 10000 planetesimals and showed correlations between the resulting power-law or- bital spacing and the initial mass distribution. Further- more, many works have performed integrations of the current solar system, finding solutions that showed both chaos and a very real possibility of future instabilities (Laskar 2008; Laskar & Gastineau 2009). Our work is unique in that we take systems formed via direct numer- ical integration of planetary accretion, evolve them to the solar system's age, probe for chaos and its source, and draw parallels to the actual solar system. Although the classical terrestrial planet formation model has succeeded in replicating many of the in- ner solar systems features, the mass of Mars remains largely unexplained (Chambers 2001a; O'Brien et al. 2006; Chambers 2007; Raymond et al. 2009b; Kaib & Cowan 2015). Known as the Mars mass deficit prob- lem, most simulations routinely produce Mars analogues which are too massive by about an order of magnitude. Walsh et al. (2011a) argue for an early inward, and subse- quent outward migration of a fully formed Jupiter, which results in a truncation of the proto-planetary disc at 1 AU prior to terrestrial planetary formation. If correct, this "Grand Tack Model" would explain the peculiar mass distribution observed in our inner solar system. Another interesting solution involves local depletion of the disc in the vicinity of Mars's orbit (Izidoro et al. 2014). A detailed investigation of the Mars mass deficit problem is beyond the scope of this paper. It is important, how- 2 ever, to note that accurately reproducing the mass ratios of the terrestrial planets is a significant constraint for any successful numerical model of planetary formation. Through dynamical modeling, we know chaos is preva- lent in our solar system (Sussman & Wisdom 1988; Laskar 1989; Sussman & Wisdom 1992; Laskar 2008). It is important to note the difference between "stabil- ity" and "chaos." While a system without "chaos" can generally be considered stable, a system with "chaos" is not necessarily unstable (Milani & Nobili 1992; Deck et al. 2013). As is convention in other works, in this paper "chaos" implies both a strong sensitivity of out- comes to specific initial conditions, and a high degree of mixing across all energetically accessible points in phase space (Deck et al. 2013) Conversely "Instability" is used to describe systems which experience specific dynamical effects such as ejections, collisions or excited eccentrici- ties. The chaos in our solar system mostly affects the ter- restrial planets, particularly Mercury, and can cause the system to destabilize over long periods of time. Laskar (2008) even shows a 1–2% probability of Mercury's eccen- tricity being excited to a degree which would risk plane- tary collision in the next 5 Gyr. What we still don't fully understand is whether these chaotic symptoms (highly excited eccentricities, close encounters and ejection) are an expected outcome of the planetary formation process as we presently understand it, or merely a quality of our particular solar system. The work of Laskar (2000) showed us that the outcomes of semi-analytic planetary formation models of our own solar system show symp- toms of chaos, and are connected to the particular initial mass distribution which is chosen. However these sys- tems were formed without the presence of the gas giants, and planetesimal interactions were simplified to minimize computing time. Perhaps our solar system is a rare out- lier in the universe, with it's nearly stable, yet inherently chaotic system of orbits occurring by pure chance. Of even greater interest, if it turns out that systems like our own are unlikely results of planetary formation, we may need to consider other mechanisms that can drive the terrestrial planets into their modern chaotic state. This work takes 145 systems of terrestrial planets formed in Kaib & Cowan (2015) as a starting point. The systems are broken into three ensembles. The first set of 50 simulations, "Circular Jupiter and Saturn" (cjs), are formed with Jupiter and Saturn on nearly circular (e<0.01) orbits, at their current semi-major axes. The simulations use 100 self-interacting embryos on nearly circular and coplanar orbits between 0.5 and 4.0 AU, and 1000 smaller non-self-interacting planetesimals. The smaller planetesimals interact with the larger bodies, but not with each other. Additionally, the initial embryo spacing is uniform and embryo mass decreases with semi- major axis to yield an r-3/2 surface density profile. The second ensemble (containing 46 integrations), "Extra Ec- centric Jupiter and Saturn" (eejs) evolve from the same initial embryo configuration as cjs, with Jupiter and Sat- urn initially on higher (e=0.1) eccentricity orbits. The fi- nal batch of integrations (49 systems), "Annulus" (ann), begin with Jupiter and Saturn in the same configuration as cjs, however no planetesimals are used. 400 Plan- etary embryos for ann are confined to a thin annulus between 0.7–1.0 AU, roughly representative of the con- ditions described following Jupiter's outward migration in the Grand Tack Model (Walsh et al. 2011b). After advancing each system to t=4.5 Gyr, we perform detailed 100 Myr simulations and probe multiple chaos indicators. By careful analysis we aim to show whether chaotic systems naturally emerge from accretion models, and whether the source of the chaos is the same as has been shown for our own solar system. 2. METHODS 2.1. System Formation and Evolution We use the simulations modeling terrestrial planet for- mation in Kaib & Cowan (2015) as a starting point for our current numerical work. In Kaib & Cowan (2015), all simulations are stopped after 200 Myrs of evolution, an integration time similar to previous studies of terrestrial planet formation (e.g. Chambers 2001a; Raymond et al. 2004; O'Brien et al. 2006; Walsh et al. 2011b). Because we ultimately want to compare the dynamical state of our solar system (a 4.5 Gyr old planetary system) with the dynamical states of our simulated systems, we begin by integrating the systems from Kaib & Cowan (2015) from t = 200 Myr to t = 4.5 Gyr. Since bodies can evolve onto crossing orbits and collide before t = 4.5 Gyr, accurately handling close encounters between mas- sive objects is essential. Thus, we use the MERCURY hybrid integrator (Chambers 1999) to integrate our sys- tems up to t = 4.5 Gyr. During these integrations, we use a 6-day timestep and remove bodies if their helio- centric distance exceeds 100 AU. Because we are unable to accurately integrate through very low pericenter pas- sages, objects are also merged with the central star if their heliocentric distance falls below 0.1 AU. Though by no means ideal, the process of removing objects at 0.1 AU is commonplace in direct numerical models of plan- etary formation due to the limitations of the integrators used for such modeling. Chambers (2001a) showed that this does not affect the ability to accurately form planets in the vicinity of the actual inner solar system, since ob- jects crossing 0.1 AU must have very high eccentricities. These excited objects interact weakly when encounter- ing forming embryos due to their high relative velocity, and rarely contribute to embryo accretion. It should be noted that many discovered exoplanetary systems have planets with semi-major axis interior to 0.1 AU. However, we are not interested in studying such systems since we aim to draw parallels to our actual solar system. The WHFAST integrator used in the second phase of this work (Section 2.2), however, can integrate the innermost planet to arbitrarily high eccentricities, so the 0.1 AU filter is no longer used. Finally, to assess the dynamical chaos among planetary-mass bodies, any "planetesimal" particles (low-mass particles that do not gravitationally interact with each other) that still survive after 4.5 Gyrs are manually removed from the final system. 2.2. Numerical Analysis Numerical simulations for detailed analysis of the fully evolved systems are performed using the WHFAST in- tegrator in the Python module Rebound (Rein & Liu 2012). WHFAST (Rein & Tamayo 2015) is a freely available, next generation Wisdom Holman symplectic integrator (Wisdom 1981; Wisdom & Holman 1991; Ki- noshita et al. 1991) ideal for this project due to its re- duction on the CPU hours required to accurately simu- late systems of planets over long timescales. WHFAST's reduction in error arising from Jacobi coordinate trans- formations, incorporation of the MEGNO (Mean Expo- nential Growth factor of Nearby Orbits) parameter, im- proved energy conservation error and tunable symplectic corrector up to order 11 motivate the integrator choice. The accuracy of many mixed variable symplectic integra- tion routines are degraded by integrating orbits through phases of high eccentricity and low pericenter. For this reason, in Figure 1 we plot the variation in energy from our simulation with the lowest pericenter (q = 0.136 AU). In the upper panel, we plot the energy of the innermost planet, since this should be fixed in the secular regime. We see that energy variations stay well below one part in 103. In the lower panel, we plot the fractional change in the total energy of this system. Again, we find that energy variations rarely exceed one part in 104. Finally, in Figure 2 we show a histogram of the fractional en- ergy change between the start and end of the integration for all of our simulations. For the vast majority of sys- tems, the fractional energy change is far less than one part in 104, and no simulations exceed 103. Values in excess of 104 are from simulations where the eccentrici- ties of the giant planets were artificially inflated and the performance of the integrator is degraded by close en- counters. These systems with frequent close encounters are obviously chaotic, so our chaos determination is not affected. MEGNO is the primary tool for identifying chaotic sys- tems. Introduced in Cincotta et al. (2003), MEGNO represents the time averaged ratio of the derivative of the infinitesimal displacement of an arc of orbit in N- dimensional phase space to the infinitesimal displace- ment. For quasi-periodic (stable) motion, MEGNO will converge to a value of 2 in the infinite limit. For chaotic systems, however, MEGNO will diverge (Cincotta et al. 2003). Maffione et al. (2013) showed that MEGNO is an extremely useful and accurate tool for detecting chaos. Systems which are non-chaotic will maintain sta- ble MEGNO values of ∼2 for the duration of the simu- lation, while chaotic systems diverge from 2. For this project, systems which attained a maximum value of MEGNO ≥ 3.0 were classified as chaotic. For use in certain analyses, the Lyapunov Timescale (τL) is also output. WHFAST calculates the inverse of τL by least squares fitting the time evolution of MEGNO (Rein & Tamayo 2015). Systems classified as chaotic tended to have a τL less than ∼10–100 Myr. Some simulations which quickly displayed chaos were terminated early to save computing time. Terminating these simulations early did not affect the chaos determi- nation since MEGNO had already clearly diverged, nor did shorter simulations affect follow on data analysis and reduction (such as the detection of resonances described in section 2.4). Another tool we use to characterize the chaos in our systems is Angular Momentum Deficit (AMD) (Laskar 1997). AMD (equation 1) measures the difference be- tween the z-component of the angular momentum of a given system to that of a zero eccentricity, zero inclina- tion system with the same masses and semi-major axes. 3 Figure 1. Upper: Fractional Energy (δE/E) change for the in- nermost planet in the system ann21, the body with the smallest pericenter (q = 0.136 AU) in all of our systems. Lower: Fractional Energy change (δE/E) for all bodies in ann21. Figure 2. Cumulative Distribution of the absolute value of Frac- tional Energy change (δE/E) over the duration of the simulation for all WHFAST integrations performed in this project. 10-1010-910-810-710-610-510-410-3Fractional Change in Energy0102030405060Number of Systems4 Run Name 1 a b c d e Run Time Jupiter Saturn 100 Myr 100 Myr 100 Myr 100 Myr 100 Myr 100 Myr/1 Gyr removed removed removed Outer Planet eccentricity shift unchanged unchanged unchanged 150% 200% + 0.05 Table 1 Run 1 simulates each fully evolved 4.5 Gyr system discussed in section 2.1. Runs a and b evaluate the effect of removing outer planets. Runs c–e investigate the result of inflating the giant planet's eccentricities. Run e is re-performed for cjs and ann systems using the MERCURY hybrid integrator for 1 Gyr in order to detect close encounters (see Section 3.2) Evaluating the evolution of AMD over the duration of a simulation will probe whether angular momentum is being exchanged between giant planets and terrestrial planets as orbits excite and deexcite due to induced chaos (Chambers 2001b). (cid:80) i mi (cid:113) ai[1 −(cid:112)(1 − e2 (cid:80) √ i mi ai Mz(def ) = i ) cos ii] (1) 2.3. Simulation Parameters Simulations are run for 100 Myr, with an integration timestep of 3.65 days. Orbital data is output every 5000 years. The order of the symplectic corrector is the order to which the symplectic correction term in the interac- tion Hamiltonian (dt in Rein & Tamayo (2015)) is ex- panded. Here, we set this to order 1 (WHFAST allows for corrections up to order 11) (Rein & Tamayo 2015). 10 sample systems were integrated at different corrector values in order to determine the lowest corrector order necessary to accurately detect chaos. Additionally, a to- tal of 16 1 Gyr simulations consisting of both chaotic and non-chaotic systems of 3, 4 and 5 terrestrial planets were performed to evaluate long-term behavior and ver- ify the adequacy of 100 Myr runs. Finally, 6 sets of 145 simulations are performed, results and findings for which are reported in section 3. The 6 runs are summarized in Table 1. 2.4. Detecting Mean Motion Resonances A Mean Motion Resonance (MMR) occurs when the periods of orbital revolution of 2 bodies are in integer ra- tio to one another. For a given MMR, the resonant angle will librate between 2 values ( `Angel Jorba 2012). Many possible resonant angles exist. For this paper, however, we only consider 4 of the more common planar resonant angles (Elliot et al. 2005). Only planar resonances are considered because our systems typically have very low inclinations. To detect MMRs, the average keplerian period is cal- culated for all bodies in the simulation. 4 resonant an- gles are calculated for all sets of bodies with period ratios within 5% of a given integer ratio. 21 different MMRs (all possible permutations of integer ratios between 2:1 and 8:7) are checked for. Using a Komolgorov-Smirnov test, each resulting time-resonant angle distribution (e.g. fig- ure 3) is compared to a uniform distribution (e.g. figure 3d), yielding a p-value. All distributions with p-values less than 0.01 are evaluated by eye for libration. Fig- ure 3 shows 4 different example distributions and their classification. 3. RESULTS 3.1. System Evolution Beyond 200 Myrs In Kaib & Cowan (2015), systems of terrestrial plan- ets were generated via simulations of terrestrial planet accretion. These simulations were terminated after 200 Myrs of evolution, as each simulation had evolved into a system dominated by 1–6 terrestrial planet-mass bodies. Terminating accretion simulations after 200–400 Myrs of system evolution is common practice since the great ma- jority of accretion events occur well before these final times are reached. However, it remains unknown how these newly formed systems evolve over the next several Gyrs. Do planetesimals and embryos naturally accrete into indefinitely stable configurations of terrestrial plan- ets? Or are the systems that arise from terrestrial planet accretion often only marginally stable, with major insta- bilities occurring hundreds of Myrs or Gyrs after forma- tion? To begin answering this question, we take the systems from Kaib & Cowan (2015) and integrate them for an- other 4.3 Gyrs with MERCURY. In Figure 4, we show the cumulative distributions of times at which these systems lose their last terrestrial planet mass body (m > 0.055 M⊕). These planets can be lost via collision with a larger planet, collision with the Sun, or ejection from the sys- tem (r > 100 AU). We find that there are many systems that undergo substantial dynamical evolution after their first 200 Myrs. As Figure 4 shows, between 20 and 50% of systems lose at least 1 planet after t = 200 Myrs. This fraction varies with the simulation batch. Systems in the cjs set are the most likely to lose planets at late times. This is likely due to the fact that these systems often form planets well beyond 2 AU (Raymond et al. 2009c), where dynamical timescales are longer and planets re- quire a longer time period to undergo ejections or final collisions compared to those at ∼1 AU. In eejs simula- tions, a smaller fraction of systems (∼25%) lose a planet after their first 200 Myrs of evolution. In these systems, the effects of an eccentric Jupiter and Saturn greatly deplete the mass orbiting beyond 1.5–2 AU (Raymond et al. 2009c), and this absence of more distant material may explain the decrease in late instabilities. Finally, the conditions are even more extreme in the ann simula- tions, where the initial planetesimal region is truncated at 1 AU. These simulations have the lowest rate of late (t > 200 Myrs) instabilities at 18%. It should also be noted that some systems lose planets at extremely late times. 8 out of 150 systems (∼5%) lose planets after t = 1 Gyr. 5 of these systems are from cjs, while eejs and ann yield 1 and 2 systems, respectively. This small, yet non-negligible fraction of systems under- going late instabilities may help explain the existence of transient hot dust around older main sequence stars (Wyatt et al. 2007). These very late instabilities in our systems occur even though the orbits of Jupiter and Sat- urn are effectively fixed for the entire integration. The rate of instabilities would likely be significantly higher if the orbits of the gas giants evolved substantially over time (Brasser et al. 2009; Agnor & Lin 2012; Brasser et al. 2013; Kaib & Chambers 2016). 5 Figure 4. Cumulative distribution of the last times at which sys- tems lose a planet more massive than 0.055 M⊕ via collision or ejection. The distributions for cjs, eejs, and ann are shown with the blue, red, and green lines, respectively. Figure 5. For systems that lose a planet after t = 200 Myrs (more massive than 0.055 M⊕), the distribution of masses for the last lost planet is shown. The distributions for cjs, eejs, and ann are shown with the blue, red, and green lines, respectively. In Figure 5, we look at the mass distributions for the last planet lost from each system with an instability after t = 200 Myrs. In general, we see that the last planets lost from systems with late instabilities have masses be- low ∼0.5 M⊕. This is not surprising, since during an instability event it is typical for the smallest planets to be driven to the highest eccentricities, resulting in their collision or ejection (Rasio & Ford 1996; Chatterjee et al. 2008; Juri´c & Tremaine 2008; Raymond et al. 2009a). However, not all systems abide by this. In particular, 3 of the 13 eejs systems that undergo late instabilities lose planets with masses well over 1 M⊕. This suggests there may be a different instability mech- anism in eejs systems. Indeed, when we look at how the last planets are lost from eejs systems, we find that 10 of the 13 systems with late instabilities lose their plan- ets via collision with the Sun. This contrasts strongly with the cjs and ann systems, where there is only one Figure 3. A shows clear libration between 2 resonant angles. These 2 terrestrial planets are locked in a 2:1 MMR for the duration of the simulation. B: depicts occasional libration where 2 rocky planets are going in and out of a 5:3 MMR. C shows slow circulation of a resonant angle for 2 inner planets. These objects are close to a 2:1 MMR. Finally, D is an example of a uniform distribution of resonant angles where the objects are not in a MMR. 5001000150020002500300035004000Last Instability Time (Myrs)0.50.60.70.80.91.0Cumulative FractionCJSEEJSANN0.00.51.01.52.0MLost (M⊕)0.00.20.40.60.81.0Cumulative FractionCJSEEJSANN6 instance of a planet-Sun collision among the 34 systems that have late instabilities. Moreover, there are 4 eejs systems with only 1 planet at t = 200 Myrs, which go on to have a planet-Sun collision before t = 4.5 Gyrs. In these cases, the gas giants are clearly driving instabili- ties. This is not surprising, since the heightened eccen- tricities of Jupiter and Saturn will enhance the secular and resonant perturbations they impart on the terrestrial planets. When interactions between a gas giant and the terrestrial planets are the main driver of an instability, the relative masses of the terrestrial planets lose their significance because they are all so small relative to the gas giants. This allows for more massive planets to be lost from these systems. Figure 6A shows an example of an instability within a cjs system. In this case, a system of 5 terrestrial plan- ets are orbiting at virtually fixed semi-major axes for 3.8 Gyrs when an instability develops between the in- ner 3 planets. The second and third planets collide and the resulting 4-planet system finishes the simulation with smaller orbital eccentricities than it began with. On the other hand, the evolution of an eejs system is shown in Figure 6B. Here we see the eccentricities of 3 relatively well separated planets driven up around 400–500 Myrs, leading to a collision between the second and third plan- ets. After the collision, the outermost planet's eccentric- ity is again quickly excited and eventually approaches 0.8. Shortly after this point, the planet collides with the Sun (after a scattering event with the inner planet). While the detailed dynamics of this system are undoubt- edly complex, the behavior is clearly different from that of the cjs system, and is almost certainly a consequence of the enhanced gas giant perturbations produced from their increased eccentricities. We also study how the properties of systems with late instabilities differ from systems that do not lose planet- mass bodies after t = 200 Myrs. In Figure 7 we look at the number of planets that each system has. In panels A–C, we see that after 200 Myrs of evolution, cjs sys- tems typically have 4–6 planet-mass bodies (an average of 4.68 planets per system). This is significantly higher than eejs and ann systems, which have an average of 2.45 and 3.00 planets per system, respectively. The differences can largely be attributed to the lack of distant planets in these systems, owing to their initial conditions. Panels A–C also show which systems go on to lose planets at later times. For cjs and ann simulations, these systems tend to have more planets than the overall distribution. In contrast, no such trend is seen among eejs systems. Regardless of planet number, the eejs systems all seem to have roughly the same probability of losing a planet at late times. This is again a symptom of the gas giants driving instabilities within these systems, unlike the cjs and ann systems, where interactions between terrestrial planets play a larger role in late instabilities. Finally, panels D–F show the distributions of planets per system after 4.5 Gyrs of evolution. At the end of our integra- tions, the cjs, eejs, and ann systems have an average of 3.76, 2.12, and 2.78 planets per system respectively. For all of our simulation batches, we see that systems with late instabilities tend to have lower numbers of planets than the overall distribution of systems. Thus, in the case of cjs and ann systems, late instabilities tend to transform systems with relatively high numbers of plan- Figure 6. A: Time evolution of a 5-planet system from the cjs simulation batch. The pericenter, apocenter, and semi-major axis of each terrestrial planet is shown vs time. B: Time evolution of a 3-planet system from the eejs simulation batch. The pericenter, apocenter, and semi-major axis of each terrestrial planet is shown vs time. ets into systems with relatively few planets. We also note that 4 eejs systems finish with no terrestrial planets whatsoever. Finally, we show the AMD of each of our terrestrial planet systems at t = 200 Myrs and t = 4.5 Gyrs in Figure 8. Panels A–C show our systems' AMD distribu- tion at t = 200 Myrs. For each of our simulation batches, the median AMD is greater than the solar system's value. Our cjs, eejs and ann simulations have median AMD val- ues of 2.9, 4.0, and 1.7 times the value of the modern inner solar system. Again, we also show the AMD distri- butions for systems that go on to have late instabilities. These systems tend to have larger AMD values. For the cjs, eejs and ann systems that undergo late instabilities the median AMD values at t = 200 Myrs are 4.1, 14, and 4.3 times the solar system's AMD, respectively. In- terestingly, though, panels D–F demonstrate that these systems are not always destined to maintain a relatively large AMD. Systems in the cjs and ann batches that un- dergo late instabilities have median AMD values of 2.7 and 3.7 times the value of the solar system after 4.5 Gyrs of evolution, respectively. Thus, a late instability does not necessarily increase the AMD of the system, and in some situations can result in moderate decreases. On the other hand, in eejs systems, the excited orbits of Jupiter and Saturn continue to wreak havoc on the terrestrial planets. Systems that experience late instabilities have a median AMD of 132 times that of the solar system! 7 Figure 7. A–C: The distribution of the number of planets in each system at t = 200 Myrs are shown for our cjs, eejs, and ann simulation batches in panels A, B, and C, respectively. The unfilled histograms show the distribution for all systems, and the filled histograms shown the distribution for systems that lose at least 1 planet more massive than 0.055 M⊕ after t = 200 Myrs. D–F: The distribution of the number of planets in each system at t = 4.5 Gyrs are shown for our cjs, eejs, and ann simulation batches in panels D, E, and F, respectively. The unfilled histograms show the distribution for all systems, and the filled histograms shown the distribution for systems that lose at least 1 planet more massive than 0.055 M⊕ after t = 200 Myrs. 3.2. Prevalence of Chaos A selection of results from our simulations are pro- vided in Appendix A (Table 4). τL and MEGNO are listed for run 1 for all systems. Additionally, we pro- vide our chaos determination (yes or no) for runs 1, a and b, as well as MMRs detected for run 1. Figure 9 compares the fraction of all systems which are chaotic between runs 1, a and b. We find that removing Jupiter and Saturn has the greatest effect on reducing chaos in our systems. In general, ∼ 50% of systems exhibit some form of chaos, when Saturn is removed only ∼ 40% of systems are chaotic and when Jupiter is removed that number is only ∼ 20%. This indicates that the chaos in most of our systems is likely driven by perturbations from Jupiter. In fact, when Jupiter was removed, all sys- tems but 1 had τL's which either increase, or are within Figure 8. A–C: The distribution of terrestrial AMD values (nor- malized to the solar system's modern terrestrial AMD) at t = 200 Myrs are shown for our cjs, eejs, and ann simulation batches in panels A, B, and C, respectively. The unfilled histograms show the AMD values for all systems, and the filled histograms shown the AMD values for systems that lose at least 1 planet more mas- sive than 0.055 M⊕ after t = 200 Myrs. D–F: The distribution of terrestrial AMD values (normalized to the solar system's modern terrestrial AMD) at t = 4.5 Gyrs are shown for our cjs, eejs, and ann simulation batches in panels A, B, and C, respectively. The unfilled histograms show the AMD values for all systems, and the filled histograms shown the AMD values for systems that lose at least 1 planet more massive than 0.055 M⊕ after t = 200 Myrs. 1.5 orders of magnitude of the original value. Figure 9 also clearly shows the disparity of chaos be- tween systems with different numbers of terrestrial plan- ets. This is most pronounced in 5-planet configurations, where only 2 such systems are free of chaos with the outer planets in place, and only 3 when they are re- moved. Having 4 terrestrial planets may not to be a significant source of chaos in our own solar system. This can be seen in figure 10, which provides MEGNO plots for the solar system with and without the 4 outer plan- ets. In fact, the solar system may better be described as a 3-planet configuration when compared to our results. Most 4 and 5-planet systems in our study differ greatly from our own since they typically contain only planets with masses comparable to Earth and Venus (see Table 3 for 2 such 5-planet examples). Mars analogues are rare 012345670510152025Number of SystemsACJS012345670510152025DCJS012345670510152025Number of SystemsBEEJS012345670510152025EEEJS01234567Number of Planets at 200 Myrs051015202530Number of SystemsCANN01234567Number of Planets at 4.5 Gyrs051015202530FANN024681012141602468101214NumberACJS024681012141602468101214DCJS05101520253035400510152025NumberBEEJS05101520253035400510152025EEEJS0246810121416AMD200/AMDSS0510152025NumberCANN0246810121416AMD4500/AMDSS0510152025FANN8 Figure 9. Comparison of the prevalence of chaos for runs 1, a and b. Figure 11. Comparison of the prevalence of chaos for runs 1, c, d and e. in our systems, and Mercury sized planets are almost non-existent. If we consider our solar system a 3 terres- trial planet arrangement, it's inherent chaos fits in well with our results; where about half of systems show chaos with the giant planets in place, and only around 1 in 6 when they are removed. A shortcoming of this compar- ison is that many studies have shown that Mercury is a very important source of the chaos in our own solar sys- tem (Laskar 2008; Laskar & Gastineau 2009). However, when we integrate the solar system without Mercury, the system is still chaotic. Therefore, though the actual so- lar system does match our results, this comparison is limited by the fact that present models of the terrestrial planet formation systematically fail to produce Mercury analogs. Figure 10. This figure compares the behavior of MEGNO in two separate simulations of our own solar system, run using the same settings described in section 2.3. With the giant planets in place (red line), the system is chaotic, and when they are removed (green line) the chaos disappears. Though Jupiter is the source of the chaos in our own solar system, when only Saturn, Uranus and Neptune are removed the system is non-chaotic because the preces- sion of Jupiter's orbit due to the outer planets stops. This result is different than the vast majority of the systems in this paper, which remain chaotic when only Saturn is removed. system ann cjs 1 0 5 71% 12% 2% 12% 0% 4% 48% 18% 18% 14% 2% 0% 2 3 4 The percentage of ann and cjs systems which lost a given number of planets when integrated with inflated giant planet eccentricities in run e for 1 Gyr. Table 2 MMRs between planets are common features in many of our chaotic systems, implying that they are often im- portant sources of the dynamical chaos. We detect 365 MMRs among all simulations in this phase of the project, 82% of which occur in chaotic systems. Further analy- sis shows that the MMRs which do occur in non-chaotic systems tend to be of higher order between smaller ter- restrial planets. It should be noted that the vast majority of these MMRs are intermittent, and last only a fraction of the entire simulation duration. Figure 11 shows the fraction of systems which are chaotic in runs 1, c, d and e. It is clear that an eccen- tric Jupiter and Saturn can quickly introduce chaos to an otherwise non-chaotic system. One interesting result from this batch of simulations is that when the eccen- tricities of the outer planets are inflated, the likelihood of a 5:2 MMR between Jupiter and Saturn developing increases. In almost all systems, this resonant perturba- tion introduces chaos, and can possibly destabilize the system. Since systems labeled cjs were formed with the giant planets on near circular orbits, simply multiplying the already low eccentricity by 1.5 or 2 was not enough to produce a noticeable effect. For this reason, run e was performed using a step increase of 0.05. There is a clear parallel between the results of this scenario and a Nice Model instability (Gomes et al. 2005; Morbidelli et al. 2005; Tsiganis et al. 2005), where the outer plan- ets rapidly transition from nearly circular to relatively eccentric orbits. To further probe this effect, we repeat run e for cjs and ann systems using the MERCURY hybrid integra- tor in order to accurately detect collisions and ejections. Systems are integrated for 1 Gyr using simulation param- eters similar to those discussed in section 2.1. We find instabilities are relatively common in these systems. 29% will shape the ultimate chaotic state of a system. 3.4. Mass Concentration Statistic and Center of Mass Analysis 9 To evaluate the degree to which mass is concentrated at a given distance away from the central star, we utilize a mass concentration statistic (Sc) (Chambers 2001b): Sc = M AX i mi i mi[log10( a ai )]2 (2) (cid:80) (cid:18) (cid:80) (cid:19) Figure 12. Cumulative distribution of the last times at which systems' with inflated eccentricities lose a planet more massive than 0.055 M⊕ via collision or ejection. The distributions for cjs and ann are shown with the blue and green lines, respectively. of ann systems and 52% of cjs systems lose one or more planets over the 1 Gyr integration. Table 2 shows the percentage of systems which lose a given number of ter- restrial planets. In fact, the resulting systems are quite similar to those produced after integrating the eejs batch to the current epoch. A small fraction of systems lose all inner planets, and some can have instabilities occur very late in the simulations (Figure 12). Overall we show that an event similar to the Nice Model scenario, where the Giant planets eccentricities quickly inflate, can result in a non-negligible probability of inner planet loss. If we again classify our solar system as a 3 terrestrial planet system, we can draw further parallels between the results of such cjs run e configurations and the Nice Model instability, since Jupiter and Saturn begin on near circular orbits. In run e, 6 out of 8 such systems were chaotic, as compared to only 2 out of 8 in run 1. Indeed, we see that even a perturbation in the giant planet's eccentricities of 0.05 is successful in rapidly making a system chaotic. In fact, when our own solar system is integrated with the outer planets on circular (e<0.001, i∼0) orbits, the chaos disappears. 3.3. Angular Momentum Deficit and Last Loss Analysis For our systems, we evaluate the difference between the average AMD of the inner planets over the first and last 3 Myr of our 100 Myr simulations. Taking the average removes the contributions from periodic forcing in the AMD from Jupiter and Saturn. We find a weak trend for chaotic 4 and 5 planet systems to have larger changes in AMD over the duration of the simulation than their non- chaotic counterparts. Of the 13 systems which had total changes in AMD greater than the actual solar system's value, 9 were classified as chaotic. The largest outlier, a non-chaotic eejs 2 planet system (eejs25), is discussed further in section 3.5.1. We also search for any correla- tion between the time and mass of the last object (m > 0.055 M⊕) lost, and the chaos of a system. Though we show that some unstable, chaotic systems can stabilize after losing a planetary mass body, we are unable to iden- tify any conclusive trends as to whether a late instability The expression in parenthesis in (2) is essentially the level of mass concentration at any point as a function of semi-major axis. Chambers (2001b) utilizes a loga- rithm in the equation since, in our own solar system, the semi-major axes of the planets lie spaced in rough ge- ometric series (the famous Titus-Bode law). Sc is the maximum value of the mass concentration function. A system where most of the mass is concentrated in a single, massive planet would have a very steep mass concentra- tion curve, and a high value of Sc. A system of multiple planets with the same mass would have a smoother curve and yield a lower Sc. As a point of reference, the Sc value of the solar system's 4 inner planets, where most of the mass is concentrated in Venus and Earth, is 90. Sc values are provided in the same format as AMD values in fig- ure 13. A general, weak correlation can be seen between chaotic systems and slightly higher values of Sc, however this trend is not very conclusive. We also provide the center of mass for each system of terrestrial planets in figure 13. A clear trend is visible where non-chaotic sys- tems tend to have a center of mass between ∼0.8–1.2 AU (the value being slightly greater as number of planets in- creases). In general, the more mass is concentrated closer to the central star, or closer to Jupiter, the greater the likelihood of chaos developing. This is likely related to Jupiter's role in introducing chaos to systems. This trend is true for all three simulation subsets, but particularly strong in the cjs and eejs batches. 3.5. Systems of Particular Interest 3.5.1. eejs25 The system with the largest change in AMD over the duration of the simulation is surprisingly non-chaotic. This outlier (eejs25), is a system of just 2 inner planets. The innermost planet is ∼117% the mass of Earth, re- siding at a semi-major axis of 0.62 AU, and the second planet is ∼96% the mass of Venus at a semi-major axis of 1.35 AU. The innermost planet is locked in a strong, secularly driven resonance with Jupiter (figure 14). This causes the eccentricity of the innermost planet to peri- odically oscillate between ∼0.15 and ∼0.7 over a period of ∼8 Myr. These oscillations are remarkably stable. In fact, due to the fortuitous spacing between of the Sun and inner 2 planets, this oscillation does not lead to in- teractions with other bodies in the system. 3.5.2. cjs10 and cjs13 The 2 most stable 5 planet configurations occurred in cjs10 and cjs13, with both systems classified as non- chaotic through runs 1, a and b. In the runs which vary eccentricity, cjs10 started to develop a weak 5:2 MMR between Jupiter and Saturn, causing mild chaos in run 10 Figure 13. A–C: Sc's affect on chaos for simulation batches cjs, eejs and ann (A, B and C respectively). Sc's for 1-planet systems are infinite, and therefore omitted in these graphs. D–F: Relation- ship between center of mass and chaos in simulation batches cjs, eejs and ann (D, E and F respectively). Chaotic systems are desig- nated by green plus signs and non-chaotic systems are represented by red circles. c (where eccentricities are increased by 150%). How- ever, the chaos in this run was mild (MEGNO only rose to 3.703 and τL for this run was 1.57 × 109). A step increase of 0.05 to Jupiter and Saturn's eccentricity in run e was required to fully introduce chaos (maximum MEGNO values 198.9 and 186.4 for cjs10 and cjs13 re- spectively) to both of these systems. This excitation of the eccentricities of the giant planets drove an occasional 3:1 MMR between the second and fourth inner planets in cjs13, possibly contributing to this chaos. The most re- markable similarity between these systems is their mass spacing and distribution (summarized in Table 3). Both have similarly low values of Sc (20.1 and 15.6). In fact, the planet spacing of both systems is somewhat reminis- cent of a Titus-Bode Law series. For example, all orbital locations of cjs13 are within 6% of a Titus-Bode series beginning at the inner planet's semi-major axis. 4. DISCUSSION AND CONCLUSIONS We have presented an analysis of systems of terrestrial planets formed through direct numerical integration of terrestrial accretion, fully evolved to the present epoch. Our work aims to assess whether our solar system, and it's inherently chaotic dynamics, is a likely result of plan- etary formation as we currently understand it. We report that roughly half of our systems display some form of Figure 14. Comparison of eejs25's innermost planet's (E0) incli- nation and eccentricity with Jupiter's eccentricity. The periods of oscillation match, indicating a strong secularly driven resonance. Interestingly, this interaction does not drive any chaos in the sys- tem. 020406080ACJS0.40.81.21.6DCJS04080Mass ConcentrationBEEJS0.40.81.21.6Center of MassEEEJS12345Inner Planets04080120160CANN12345Inner Planets0.40.81.21.6FANNbody 1 2 3 4 5 cjs10 a (AU) 0.51 0.76 1.1 1.9 2.5 cjs13 a (AU) 0.54 0.78 1.1 1.6 2.7 sol a (AU) 0.39 0.72 1.0 1.5 2.8 cjs10 m cjs13 m sol m (M⊕) (M⊕) 0.61 0.055 0.82 0.63 1.0 1.0 0.46 0.11 0.32 (M⊕) 0.68 0.18 0.77 0.59 0.80 0.00015 Comparison of non-chaotic 5 terrestrial planet systems (cjs10 and cjs13) mass and semi-major axis distributions. The fifth body for the solar system is taken to be Ceres. Table 3 chaos. By far, the most common source of this dynami- cal chaos is perturbations from Jupiter. Additionally, we find that systems in our sample with greater numbers of terrestrial planets are far more prone to chaos than those with fewer inner planets. Unfortunately, systems formed through numerical integrations (including those of Kaib & Cowan (2015) which are used for this work) still rou- tinely produce Mercury and Mars analogues which are far too massive (Chambers 2001a; O'Brien et al. 2006; Chambers 2007; Raymond et al. 2009b). Consequently, we find it best to consider our solar system a 3 terres- trial planet system for the purposes of comparison in this work. This classification of the solar system works well with our results that 3-planet systems have an ∼50% chance of being chaotic, and a much lower probability when the giant planets are removed. By varying the eccentricities of Jupiter and Saturn in 3 separate batches of simulations, we show that an ec- centric system of outer planets can quickly introduce dy- namical chaos and trigger instabilities in otherwise stable systems. This result confirms the findings in numerous previous works (e.g. Gomes et al. 2005; Morbidelli et al. 2005; Tsiganis et al. 2005). The inflation in eccentricity required to create such a chaotic system is surprisingly small. By varying the eccentricity of a batch of systems with Jupiter and Saturn on nearly circular orbits, we show that dynamical chaos quickly ensues. This sort of event is akin to a Nice Model-like instability (Gomes et al. 2005; Morbidelli et al. 2005; Tsiganis et al. 2005). We go on to show that in such an instability, the possibil- ity of destabilizing the inner planets to the point where a terrestrial planet is lost by either collision or ejection is fairly high. Additionally, we find that systems most immune to de- veloping dynamical chaos tend to have centers of mass between ∼0.8–1.2 AU, though that range is by no means absolute. This is an interesting result, and likely related to Jupiter's role in driving chaos in many of these sys- tems. We consistently identified systems throughout our suite of simulations which displayed many of the same chaotic dynamics as our own solar system. It is clear that chaotic systems such as our own are common results of planetary formation. The largest source of chaos in our own system, perturbations from Jupiter, is the most common source of chaos observed in our work. The solar system, however, is akin to only a small fraction (∼10%) of our simulations since removing just the planets be- yond Jupiter turns our system non-chaotic. Additionally we show that late instabilities are common among these systems, and it is not far-fetched to imagine a late insta- bility shaping dynamics within our own system. Finally, we find many systems with similar numbers of terrestrial 11 planets, semi-major axis configurations, mass concentra- tions and chaos indicators (τL and MEGNO) as our own. ACKNOWLEDGEMENTS This work was supported by NSF award AST- 1615975. Simulations in this paper made use of the REBOUND code which can be downloaded freely at http://github.com/hannorein/rebound. The bulk of our simulations were performed over a network managed with the HTCondor software package (https://research. cs.wisc.edu/htcondor/). REFERENCES Agnor, C. B., & Lin, D. N. C. 2012, ApJ, 745, 143 `Angel Jorba. 2012, UNESCO Encyclopedia of Life Support Systems, Vol. 6.119.55 Celestial Mechanics Brasser, R., Morbidelli, A., Gomes, R., Tsiganis, K., & Levison, H. F. 2009, A&A, 507, 1053 Brasser, R., Walsh, K. J., & Nesvorn´y, D. 2013, MNRAS, 433, 3417 Chambers, J. E. 1999, MNRAS, 304, 793 -. 2001a, Icarus, 152, 205 -. 2001b, Icarus, 152, 205 -. 2007, Icarus, 189, 386 Chatterjee, S., Ford, E. B., Matsumura, S., & Rasio, F. A. 2008, ApJ, 686, 580 Cincotta, P. M., Giordano, C. M., & Sim´o, C. 2003, Physica D Nonlinear Phenomena, 182, 151 Currie, T., Lada, C. J., Plavchan, P., et al. 2009, ApJ, 698, 1 Deck, K. M., Payne, M., & Holman, M. J. 2013, ApJ, 774, 129 Elliot, J. L., Kern, S. D., Clancy, K. B., et al. 2005, AJ, 129, 1117 Gomes, R., Levison, H. F., Tsiganis, K., & Morbidelli, A. 2005, Nature, 435, 466 Halliday, A. N. 2008, Philosophical Transactions of the Royal Society of London Series A, 366, 4163 Izidoro, A., Haghighipour, N., Winter, O. C., & Tsuchida, M. 2014, ApJ, 782, 31 Juri´c, M., & Tremaine, S. 2008, ApJ, 686, 603 Kaib, N. A., & Chambers, J. E. 2016, MNRAS, 455, 3561 Kaib, N. A., & Cowan, N. B. 2015, Icarus, 252, 161 Kinoshita, H., Yoshida, H., & Nakai, H. 1991, Celestial Mechanics and Dynamical Astronomy, 50, 59 Laskar, J. 1989, Nature, 338, 237 -. 1996, Celestial Mechanics and Dynamical Astronomy, 64, 115 -. 1997, A&A, 317, L75 -. 2000, Physical Review Letters, 84, 3240 -. 2008, Icarus, 196, 1 Laskar, J., & Gastineau, M. 2009, Nature, 459, 817 Maffione, N. P., Darriba, L. A., Cincotta, P. M., & Giordano, C. M. 2013, MNRAS, 429, 2700 Milani, A., & Nobili, A. M. 1992, Nature, 357, 569 Morbidelli, A., Levison, H. F., Tsiganis, K., & Gomes, R. 2005, Nature, 435, 462 O'Brien, D. P., Morbidelli, A., & Levison, H. F. 2006, Icarus, 184, 39 Rasio, F. A., & Ford, E. B. 1996, Science, 274, 954 Raymond, S. N., Armitage, P. J., & Gorelick, N. 2009a, ApJ, 699, L88 Raymond, S. N., O'Brien, D. P., Morbidelli, A., & Kaib, N. A. 2009b, Icarus, 203, 644 -. 2009c, Icarus, 203, 644 Raymond, S. N., Quinn, T., & Lunine, J. I. 2004, Icarus, 168, 1 Rein, H., & Liu, S.-F. 2012, A&A, 537, A128 Rein, H., & Tamayo, D. 2015, MNRAS, 452, 376 Sussman, G. J., & Wisdom, J. 1988, Science, 241, 433 -. 1992, Science, 257, 56 Tsiganis, K., Gomes, R., Morbidelli, A., & Levison, H. F. 2005, Nature, 435, 459 Walsh, K. J., Morbidelli, A., Raymond, S. N., O'Brien, D. P., & Mandell, A. M. 2011a, Nature, 475, 206 -. 2011b, Nature, 475, 206 Wisdom, J., & Holman, M. 1991, AJ, 102, 1528 12 Wisdom, J. L. 1981, PhD thesis, California Institute of Wyatt, M. C., Smith, R., Greaves, J. S., et al. 2007, ApJ, 658, 569 Technology, Pasadena. 13 APPENDIX SUPPLEMENTAL DATA Table 4 Simulation Results MEGNO Inner Planets Run 11 Run a1 Run b1 100.7 45.35 187.2 160.9 183.5 160.0 126.0 2.004 1.997 1.896 1.999 90.97 2.003 1.988 8.794 212.8 4.467 21.04 53.36 2.000 1.996 2.000 169.0 20.29 2.040 183.2 1.999 1.999 12.82 4.816 2.001 1.966 1.999 117.0 2.000 186.5 193.5 5.732 1.999 1.999 113.5 1.998 49.51 189.4 194.1 2.011 129.4 8.483 20.44 6.271 178.4 87.08 202.9 174.4 2.000 29.38 34.65 1.988 5 4 5 5 5 5 5 4 3 5 2 5 5 3 5 4 4 3 4 2 2 3 4 4 4 5 3 2 4 3 2 4 2 4 3 4 5 3 4 4 4 4 4 4 4 4 4 4 4 4 3 4 4 1 2 3 4 3 Continued on next page Y Y Y Y Y Y Y N N N N Y N N Y Y Y Y Y N N N Y Y N Y N N Y Y N N N Y N Y Y Y N N Y N Y Y Y N Y Y Y Y Y Y Y Y N Y Y N Y N Y Y Y Y N N N N N Y N N Y Y N N Y N N N Y N N Y N N Y N N N N Y N N Y N N N Y N N Y N N N Y N N N Y N N N Y N N Y N Y Y Y Y Y N N N N Y N N Y Y N N Y N N N Y Y N Y N N Y N N N N Y N Y Y Y N N Y N Y Y Y N Y Y Y Y Y Y Y N N Y Y N System cjs1 cjs2 cjs3 cjs4 cjs5 cjs6 cjs7 cjs8 cjs9 cjs10 cjs11 cjs12 cjs13 cjs14 cjs15 cjs16 cjs17 cjs18 cjs19 cjs20 cjs21 cjs22 cjs23 cjs24 cjs25 cjs26 cjs27 cjs28 cjs29 cjs30 cjs31 cjs32 cjs33 cjs34 cjs35 cjs36 cjs37 cjs38 cjs39 cjs40 cjs41 cjs42 cjs43 cjs44 cjs45 cjs46 cjs47 cjs48 cjs49 cjs50 eejs1 eejs2 eejs3 eejs4 eejs5 eejs6 eejs7 eejs8 τL (years) 7.09 × 105 1.69 × 107 2.93 × 105 3.69 × 105 4.63 × 105 3.61 × 105 5.04 × 106 1.71 × 1010 5.93 × 1010 8.28 × 108 7.80 × 1010 7.14 × 106 6.10 × 109 1.78 × 1010 7.61 × 107 4.89 × 104 5.97 × 108 5.16 × 107 1.50 × 107 7.23 × 1010 9.03 × 1010 3.86 × 1010 8.83 × 103 1.66 × 106 2.16 × 1010 1.39 × 105 8.50 × 1010 7.22 × 1010 9.22 × 107 7.50 × 108 1.58 × 1010 2.81 × 109 1.31 × 1010 7.89 × 104 2.26 × 1010 3.33 × 106 2.14 × 105 3.50 × 108 6.30 × 1010 6.05 × 1010 6.46 × 106 9.03 × 1010 1.47 × 107 1.93 × 105 1.38 × 105 5.10 × 1010 2.11 × 106 1.68 × 108 1.49 × 107 5.08 × 108 4.36 × 105 1.75 × 105 4.55 × 105 9.62 × 105 1.05 × 1010 4.11 × 107 2.41 × 107 1.33 × 1010 Run 1 MMRs2 7:4 2:1 7:4,5:1, 7:1 (S), 5:2 (J,S) 7:3 (J) 3:1 (J) 3:1 (J) 5:2 (J,S) 2:1 6:1 5:3:1 5:3 5:1 4:1 5:3 8:5,7:1 3:1,7:1 (S) 14 TABLE 4 – Continued from previous page System τL MEGNO eejs9 eejs10 eejs11 eejs13 eejs15 eejs16 eejs18 eejs19 eejs20 eejs21 eejs22 eejs23 eejs24 eejs25 eejs26 eejs27 eejs28 eejs29 eejs30 eejs31 eejs32 eejs33 eejs34 eejs35 eejs36 eejs37 eejs38 eejs39 eejs40 eejs41 eejs42 eejs43 eejs44 eejs45 eejs46 eejs47 eejs49 eejs50 ann1 ann2 ann3 ann4 ann5 ann6 ann7 ann8 ann9 ann10 ann11 ann12 ann13 ann14 ann15 ann16 ann17 ann18 ann19 ann20 ann21 ann22 ann23 ann24 ann25 ann26 ann27 ann28 ann29 ann30 ann31 (years) 1.01 × 108 6.92 × 105 1.95 × 1010 9.19 × 108 2.57 × 109 3.39 × 1010 1.87 × 1010 2.93 × 1010 1.94 × 109 4.17 × 109 1.05 × 1010 3.11 × 106 1.21 × 107 6.18 × 109 6.31 × 109 8.95 × 108 2.48 × 109 3.43 × 105 6.32 × 105 2.45 × 105 1.96 × 1010 3.96 × 1010 2.67 × 105 4.23 × 106 3.39 × 106 1.06 × 106 2.29 × 105 1.84 × 109 1.08 × 105 5.27 × 106 1.31 × 1010 4.94 × 109 7.84 × 104 2.09 × 106 5.52 × 1010 5.37 × 108 3.06 × 1010 2.10 × 1010 3.64 × 109 3.02 × 1010 3.06 × 1010 4.94 × 1010 4.97 × 107 1.66 × 108 3.91 × 109 2.87 × 105 4.67 × 1010 2.06 × 1010 3.17 × 109 6.72 × 103 2.89 × 106 1.68 × 107 9.01 × 109 1.85 × 1010 1.05 × 109 8.23 × 108 3.61 × 105 2.07 × 106 4.55 × 106 2.00 × 107 2.00 × 107 2.53 × 1010 3.59 × 107 1.08 × 108 1.15 × 108 1.90 × 108 5.97 × 106 5.10 × 103 1.32 × 108 9.676 151.8 2.000 7.045 1.999 2.002 2.017 2.000 1.992 1.919 1.998 183.4 57.05 2.041 1.912 1.813 4.370 173.0 158.7 184.6 1.969 2.004 196.5 154.0 180.2 183.2 171.8 2.005 189.1 33.10 2.511 1.998 193.8 178.1 2.009 8.082 1.996 1.929 2.014 1.998 1.988 2.000 29.32 7.441 3.132 185.6 2.010 1.998 2.294 195.1 125.7 60.09 2.103 1.99 4.372 2.159 183.8 175.4 21.76 44.64 53.55 1.999 20.22 9.212 13.65 15.54 115.7 193.1 8.47 Inner Planets Run 11 Run a1 Run b1 5 1 2 4 1 3 4 1 3 3 2 4 5 2 3 3 3 1 2 4 3 3 1 2 1 4 2 2 4 4 4 1 3 4 4 3 3 2 3 3 4 3 3 4 3 3 3 2 3 4 2 4 3 2 4 3 3 4 3 3 3 3 4 4 4 5 4 3 3 Y Y N Y N N N N N N N Y Y N N N Y Y Y Y N N Y Y Y Y Y N Y Y N N Y Y N Y N N N N N N Y Y Y Y N N N Y Y Y N N Y N Y Y Y Y Y N Y Y Y Y Y Y Y Y N N Y N N N N N N N Y Y N N N N N N Y N N Y N N Y N N Y Y N N N N N N N N N N N N N N N N N N N Y Y N N N N N Y Y N N N N Y Y Y Y N Y Y Y N N N N N N N N N N Y Y N N N N N Y Y N N N Y N Y N N Y Y N N Y Y N N N N Y N N N Y N N Y N N N Y Y N N N N N Y Y Y Y Y N Y Y N Y Y Y Y Continued on next page Run 1 MMRs2 8:3,8:5 5:2 (J,S) 8:1 (S) 5:2 (J,S) 7:3 7:2,8:5 7:3 5:3 8:5 5:2 7:3 8:5,7:1 (J) 7:4 2:1 4:1 5:2 (J,S) 3:1 7:4 8:3 5:3 5:2,8:5 15 TABLE 4 – Continued from previous page System τL MEGNO ann32 ann33 ann35 ann36 ann37 ann38 ann39 ann40 ann41 ann42 ann43 ann44 ann45 ann46 ann47 ann48 ann49 ann50 (years) 1.85 × 106 2.36 × 106 3.93 × 107 3.50 × 1010 2.75 × 105 4.23 × 1010 2.50 × 105 6.91 × 105 4.22 × 1010 1.01 × 1010 1.49 × 1011 2.75 × 108 9.52 × 109 4.81 × 106 3.71 × 1010 8.07 × 106 7.25 × 105 5.53 × 1010 182.3 191.6 23.00 1.998 192.5 2.001 189.6 192.9 1.998 2.000 2.000 11.41 2.000 152.4 2.004 54.18 185.7 2.008 Inner Planets Run 11 Run a1 Run b1 4 4 3 3 4 3 3 4 3 2 3 3 2 3 3 4 3 3 Y Y Y N Y N Y Y N N N Y N Y N Y Y N Y Y N Y Y N Y Y N N N N N Y N Y N N Y Y N N Y N Y Y N N N N N Y N Y Y N Run 1 MMRs2 2:1 7:3 5:2 (J,S) 7:4 1 "Y" indicates chaos was detected, "N" indicates it was not. 2 (J) and (S) indicate the resonance was with Jupiter or Saturn respectively.
1812.10436	1	1812	2018-12-26T18:10:14	Stability of a rotating asteroid housing a space station	[ "astro-ph.EP" ]	Today there are numerous studies on asteroid mining. They elaborate on selecting the right objects, prospecting missions, potential asteroid redirection, and the mining process itself. For economic reasons, most studies focus on mining candidates in the 100-500m size-range. Also, suggestions regarding the design and implementation of space stations or even colonies inside the caverns of mined asteroids exist. Caverns provide the advantages of confined material in near-zero gravity during mining and later the hull will shield the inside from radiation. Existing studies focus on creating the necessary artificial gravity by rotating structures that are built inside the asteroid. Here, we assume the entire mined asteroid to rotate at a sufficient rate for artificial gravity and investigate its use for housing a habitat inside. In this study we present how to estimate the necessary spin rate assuming a cylindrical space station inside a mined asteroid and discuss the implications arising from substantial material stress given the required rotation rate. We estimate the required material strength using two relatively simple analytical models and apply them to fictitious, yet realistic rocky near-Earth asteroids.	astro-ph.EP	astro-ph	Stability of a rotating asteroid housing a space station T. I. Maindl ∗, R. Miksch and B. Loibnegger Department of Astrophysics, University of Vienna, Vienna, Austria Correspondence*: Department of Astrophysics, University of Vienna, Turkenschanzstr. 17, 1180 Vienna, Austria [email protected] 8 1 0 2 c e D 6 2 . ] P E h p - o r t s a [ 1 v 6 3 4 0 1 . 2 1 8 1 : v i X r a ABSTRACT Today there are numerous studies on asteroid mining. They elaborate on selecting the right objects, prospecting missions, potential asteroid redirection, and the mining process itself. For economic reasons, most studies focus on mining candidates in the 100 -- 500 m size-range. Also, suggestions regarding the design and implementation of space stations or even colonies inside the caverns of mined asteroids exist. Caverns provide the advantages of confined material in near-zero gravity during mining and later the hull will shield the inside from radiation. Existing studies focus on creating the necessary artificial gravity by rotating structures that are built inside the asteroid. Here, we assume the entire mined asteroid to rotate at a sufficient rate for artificial gravity and investigate its use for housing a habitat inside. In this study we present how to estimate the necessary spin rate assuming a cylindrical space station inside a mined asteroid and discuss the implications arising from substantial material stress given the required rotation rate. We estimate the required material strength using two relatively simple analytical models and apply them to fictitious, yet realistic rocky near-Earth asteroids. Keywords: asteroids, asteroid mining, material stress, arificial gravity, space stations 1 INTRODUCTION Sustaining human life on a station built inside a mined asteroid is a task which will require expertise in many fields. There needs to be air to breathe, water to drink, and the appropriate recycling systems, as well as food and light. Nevertheless, one of the most important prerequisites for a human body to stay healthy is gravity. Taking plants to space and zero gravity is not as big a problem as taking a human body into this hostile environment. Plants do adapt to zero gravity relatively easily which is shown by many experiments performed both on the ISS (International Space Station) as well as on the ground with artificial microgravity (for example, see Kitaya et al., 2000, 2001; Kiss et al., 2009). The human body however, reacts sensitive to zero gravity. The lack of an up and down direction in space affects endothelial cells (Versari et al., 2013), blood distribution to endocrine and reflex mechanisms controlling body water homeostasis and blood pressure (e.g. Taibbi et al., 2013), and it results in muscle wasting (Gopalakrishnan et al., 2010), severe bone loss (Keyak et al., 2009), immune depression (e.g. Cogoli, 1993; Battista et al., 2012; Gasperi et al., 2014), and ophthalmic problems (Nelson et al., 2014). A study on how much gravity is needed to keep the human body upright was performed by Harris et al. (2014). They found that the threshold level of gravity needed to influence a persons orientation judgment 1 Maindl et al. Stability of a rotating asteroid housing a space station is about 15 % of the gravity on Earth's surface, which is approximately the gravity acting on the Lunar surface. Martian gravity, 38 % of Earth's gravity, should be enough for astronauts to orient themselves and maintain balance. As a consequence of a lack of experiments on the influence of reduced gravity on the human body we adopt the value of 38 % of Earth's gravity (gE) as starting point for our theoretical approach. We assume that a rotation of the asteroid has to cause an artificial gravity of minimum 0.38 gE in order to sustain long term healthy conditions for humans on the station. Present suggestions to tackle this challenge of providing sufficient gravity rely on habitats in rotating wheels or tori that create gravity: Grandl and Bazso (2013) suggest self-sustained colonies up to 2000 people. Other studies somewhat vaguely mention augmenting the natural rotation with additional artificial rotation Taylor et al. (2008). We elaborate on the latter and explore the feasibility and viability of creating "artificial" gravity for a habitat by putting the entire asteroid to rotation at a rate sufficient to generate the desired gravity. We start with initial considerations regarding the required spin rate of a space station with sufficient artificial gravity (Sect. 2). Section 3 elaborates on the stress acting on the asteroidal hull and formulates two analytical models for the tensile and shear stresses as a function of the asteroid size, the dimensions of the space station, the required artificial gravity level, and the bulk density of the asteroid material. Section 4 applies our formulation to a near-Earth asteroid maximizing the usable area of the space station while observing maximum stress constraints. Finally, conclusions and an outlook on further related research is given in Sect. 5. 2 INITIAL CONSIDERATIONS Let's assume a cylindrical space station with height hc and radius rc as depicted in Fig. 1. Letting the cylinder rotate about its symmetry axis y with angular velocity ω will create an acceleration of gc = ω2rc (1) acting on objects on the lateral surface. For a certain artificial gravity level the required rotation rate ω and rotation period T are then given by (cid:114) gc rc (cid:114) rc gc ω = , T = 2π ω = 2π . (2) Considering a size range of rc = 50 . . . 250 m for example, the rotation rates would need to be between 1.17 and 2.6 rpm (rotations per minute) to create artificial gravity necessary for sustaining extended stays on the station (assuming 0.38 gE as discussed above). Figure 2 gives an overview of rotation rates for a range of radii and gravity levels. The area usable for a space station subject to artificial gravity is the lateral surface of the cylinder S, S = 2π rc hc. (3) 3 ESTIMATING TENSILE AND SHEAR STRESS The station is to be built inside a mined asteroid. Imposing a spin rate sufficient for providing artificial gravity on the lateral surface of the cylinder will create a substantial load on the asteroid material due 2 Maindl et al. Stability of a rotating asteroid housing a space station to centrifugal forces. While little is known on material properties of small asteroids subject to our study, we rely on assumed material strength. Here, we assume that the asteroid is made of homogeneous, solid material such as basaltic silicate rock, for instance. We estimate the load on the asteroid material in simplified models: the tensile stress acting on the asteroid cross section is related to assumed tensile strength of solid silicate rock. Figure 3 shows our geometrical model of a spheroidal asteroid with semi-axes a and b, respectively. The cavern is centered and cylindrical with respective radius rc and height hc. Due to the elliptical cross-section, the lateral distance d between the cavern and the asteroid's surface is given by d = b a 2 a2 − hc 4 − rc. (4) For a feasible solution, the cavern needs to be inside the asteroid in its entirety, which translates to the condition d > 0 or rc < b a a2 − hc 4 2 . (5) The whole body will rotate about its symmetry axis y at a rate ω providing sufficient artificial gravity on the lateral surface of the space station with radius rc. We assume rigid body rotation. 3.1 Model 1 (cid:115) (cid:115) (cid:90) (cid:90) 0 (cid:90)  π(cid:90) 0 In this model centrifugal forces that are acting on the asteroid material exert a load on an arbitrary symmetry plane. We determine the tensile stress that results from this load. The total centrifugal force F1 pulling two halves of the asteroid apart is given by (cid:90) F1 = 2 dF = 2 d(mω2r) = 2ω2ρ r dV. (6) asteroid half asteroid half asteroid half Here, we use the mass m of a volume element, the uniform asteroid density ρ, and the distance r of a volume element from the rotation axis. Transforming into cylindrical coordinates symmetrical w.r.t. the y-axis (dV = r dr dy dϕ) and considering that there will be no contribution from the void of height hc and radius rc (see Fig. 3) gives √ a2−y2 a(cid:90) b a dy π(cid:90) hc/2(cid:90) rc(cid:90) r2dr − dϕ dy r2dr F1 = 2ω2ρ dϕ −a (cid:112)a2 − y2 holds for the elliptical cross section. Integrating (details given in Appendix A) −hc/2 0 0 (7)  . Note that r(y) = b a yields (cid:18) π (cid:19) F1 = π ω2ρ a b3 − 2 3 rc 3hc 4 . (8) 3 Maindl et al. Stability of a rotating asteroid housing a space station This load acts on the asteroid's cross section A = π a b − 2 rc hc (cf. Fig. 3) resulting in tensile stress σ1 = F1/A of (cid:18)3πa b3 − 8rc 3hc π a b − 2 rc hc (cid:19) σ1 = F1 A = π ω2ρ 12 . (9) As we are interested in the stress resulting from a desired artificial gravity gc we substitute ω2 from (2) and get σ1 = π ρ gc 12rc By introducing the dimensionless quantities (cid:18)3πa b3 − 8rc 3hc π a b − 2 rc hc (cid:19) (cid:48) = hc hc a , (cid:48) = rc rc b . (10) (11) (cid:18)3π − 8 rc (cid:19) (cid:48) (cid:48) 3hc (cid:48) (cid:48) hc we can separate the parameters specific to individual asteroids and show that σ scales linearly with each of (cid:48) into material density ρ, desired artificial gravity gc, and asteroid semi-minor axis b. Inserting hc (10) yields (cid:48) and rc π ρ gc b (cid:48) 12 rc (cid:48), hc σ1 = π − 2 rc (12) (cid:48)) to get estimates for the stresses in asteroids of arbitrary density Hence, it is sufficient to study f (rc and semi-minor axis rotating at a rate providing the desired artificial gravity. Figure 4 shows σ1 contours assuming parameters ρ = 1 g cm−3, gc = 1 gE, and b = 1 m. For different parameter values the numbers scale linearly with ρ, gc, and b. The white area in Fig. 4 corresponds to illegal combinations of rc and hc that violate condition (5) demanding the space station has to be inside the asteroid in its entirety. = ρ gc b · f (rc (cid:48), hc (cid:48)). The required rotation rate as a function of space station radius rc and desired artificial gravity gc is given in (2), scales according to ω = = , (13) and is indicated on the upper x-axis. As we will be interested in the usable surface area of the station S, we indicate its dimensionless variant (14) as red, dotted contour lines in Fig. 4. For any assumed maximum material strength, the solution for maximum S suggests a radius of the cylinder rc that extends all the way to the surface (d = 0, cf. Fig. 3). This is however, the edge case of our model 1 that estimates material load by assuming two halves of the asteroid driven apart by centrifugal forces. As soon as d gets very small the "two halves" assumption fails. Therefore, we estimate material load by another model which will be more accurate at the edge case, i.e. very small values of the distance to the surface d. 3.2 Model 2 Rather than focusing on two entire halves of the hollowed-out asteroid, this model studies the "mantle" outside the space station. This is the solid torus created by sweeping the right part of the hashed surface 4 (cid:114) gc rc (cid:115) (cid:114)gE gc[gE] (cid:48) b rc = 2π rc (cid:48) (cid:48)hc S a b Maindl et al. Stability of a rotating asteroid housing a space station between the red lines at y = ± hc/2 in Fig. 5 around the y-axis. As the radius of the space station approaches the asteroid's hull, the centrifugal forces attempting to shear away this torus may get significant. This shearing load acts on the two annuli resulting from rotating the red lines in Fig. 5 about the y-axis. In addition to overcoming the shear strength of the asteroid material however, the tensile strength of the cross section (the hashed area in Fig. 5) has to be exceeded by the load exerted by the centrifugal force. Similar to calculating F1 in Sect. 3.1, the centrifugal force F2 can be derived by transforming to cylindrical coordinates as follows: (cid:90) F2 = ω2ρ (cid:90) r dV = ω2ρ shaded volume in Fig. 5, i.e. y≤hc/2 y≤ 1 2 hc r2 dr dy dϕ = ω2ρ 2π(cid:90) 0 dϕ 2 hc(cid:90) 1 b a dy √ (cid:90) − 1 2 hc rc a2−y2 dr r2. (15) Integrating (Appendix B gives the detailed steps) yields a rather lengthy expression for the centrifugal force: (cid:19)3(cid:34) (cid:40) (cid:18) b a 1 8 hc (cid:32) 5a2 − hc 2 2 (cid:33)(cid:115) a2 − hc 4 F2 = 2πω2ρ 3 2 + 6a4 arcsin hc 2a (cid:35) (cid:41) . − hc rc 3 (16) This force exerts a tensile load on the asteroid's cross section At between y = −hc/2 and y = hc/2 and a shear load on the two annuli of area As given by rotating the red lines in Fig. 5 about the y-axis. The surface area At is given by (cf. Fig. 5)  = 4 hc/2(cid:90) b a x(y) dy − hcrc (cid:114) At = 2  hc/2(cid:90) (cid:34) (cid:112) −hc/2 a2 − y2 4 dy − 2hcrc 0 y(cid:112)4a2 − y2 (cid:35) h 2 0 − 2hcrc and using the identity arcsin x = arctan(x/(cid:112)1 − x2) we get 4a2 − y2 + a2 arctan = 4b a y 4 (cid:115) 4a2 − hc 4 At = b hc 2a 2 + 4ab arcsin − 2hcrc. hc 4a Each of the annuli has a surface of As, (17) 5 Maindl et al. Stability of a rotating asteroid housing a space station 2(cid:3) = π As = π(cid:2)(rc + d)2 − rc (cid:33) (cid:32) (cid:34) (cid:32) a2 − hc 4 (cid:33) 2 (cid:35) − rc 2 (cid:34) (cid:35) b2 a2 = π b2 2 1 − hc 4a2 − rc 2 . Combining equations (16), (17), and (19), we obtain the average stress σ2 in this model, σ2 = F2 At + 2As , (18) (19) (20) for the complete -- rather unwieldy -- formulation of the stress please refer to Appendix C. Unlike it is the case in model 1, we cannot formulate σ2 in a scaling way using the dimensionless quantities rc and hc. For the asteroids in scope of this study though, σ2 is usually smaller than σ1, but increases if the space station radius gets closer to the asteroid's surface. In the following Sect. 4 we will demonstrate this by comparing the two models for a fictitious, yet realistic asteroid. 4 APPLICATION TO A REALISTIC ASTEROID We will apply the analytic models 1 and 2 to a rocky asteroid with dimensions 500 × 390 m. There is a number of similar-sized rocky near-Earth asteroids, e.g. 3757 Anagolay, 99942 Apophis, 3361 Orpheus, 308635 (2005 YU55), 419624 (SO16), etc. (cf. JPL, 2018). As little is known about the composition and material properties of these objects, we assume they are composed of basaltic rock with a bulk density of ρ = 2.7 g cm−3. Tensile strength values for basalt are in the range of approx. 12. . . 14 MPa Stowe (1969), shear strengths are approx. 8. . . 36 MPa Karaman et al. (2015), which provides an order-of-magnitude framework of the expected material strength data. Finally, we will assume a desired artificial gravity level of gc = 0.38 gE as discussed in Sect. 1. Figure 6 shows the resulting tensile stress along with the usable space station surface and required rotation rates predicted by model 1. The stress levels are mostly of the same order of magnitude as the assumed material strength (∼ 10 MPa) or even smaller. However, the solution resulting in the maximum area S ≈ 0.3 km2 would have the cylindrical station extend to the asteroid's surface, which seems unrealistic and will lead to the asteroid becoming unstable. Also, for realistic scenarios a material stress (≈ 4 MPa in this case) very close to the -- poorly constrained -- material strength will be unacceptable so that a cavern with radius-height data more towards the lower right of the diagram will be desirable. Investigating the combined tensile and shear stresses according to our model 2 results in a different stress-pattern, given in Fig. 7. While the material loads are systematically lower for space stations deeper inside the asteroid in their entirety, stresses for "thinner" tori (i.e., larger rc) are of the same order of magnitude as predicted by model 1. In summary, both models predict stresses that are comparable to anticipated material strength for asteroids made of competent rock. As the assumed material parameters are based on the unknown composition, thorough studies of candidate asteroids will be necessary before considering to set them to rotation to house a space station with artificial gravity inside. 6 Maindl et al. Stability of a rotating asteroid housing a space station 5 CONCLUSIONS AND FURTHER RESEARCH We established two simple analytical models for estimating whether a candidate for asteroid mining may be suitable for hosting a space station with artificial gravity. The novelty in our approach is to investigate whether the asteroidal hull -- once set to rotation as a whole -- can sustain the material loads resulting from a sufficiently high rotation rate. We find that loads resulting from centrifugal forces are in the order of magnitude of material strength of solid rock, which makes a space station in the cavern of a mined asteroid feasible if its dimensions are chosen right and if the material composition and material strength of the asteroid is known to a satisfactory level of accuracy. Practical applications will crucially depend on knowing not only the composition but also the internal structure of candidate bodies. As missions to these asteroids seem inevitable for such studies, decisions on inhabiting such asteroids may only be possible after mining operations have started. Also, the methods of actually initiating the rotation at the required rate is subject to further investigations. Hypothetically, starts and landings of spacecraft during the mining process might contribute to building up angular momentum of the asteroid. Currently, we are working on a more realistic analytic approach for determining the detailed shape of the cavern housing the space station taking into account the internal density profile. In the past, we successfully conducted smooth particle hydrodynamics (SPH) simulations of asteroids (e.g. Maindl et al., 2013; Haghighipour et al., 2018; Maindl et al., 2018). As our analytical study is approximative in nature we plan to conduct a series of SPH simulations with different material models and varying porosity. This will allow to numerically verify the predictions of the simplified analytical models presented here as well as future models and to further investigate the behavior of rotating bodies with substantial internal caverns. CONFLICT OF INTEREST STATEMENT The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. AUTHOR CONTRIBUTIONS TIM developed the analytical models and performed most of the calculations. He wrote about 70 % of the paper and created most of the figures. RM provided various aspects of the analytical approximations, helped in the calculations, and contributed to the representation of the individual equations. He wrote about 15 % of the paper. BL contributed to the quality of the analytical models, researched the details of required artificial gravity levels, and created the pictorials of the asteroid with the cylindrical cavern. She wrote about 15 % of the paper. FUNDING This project received seed funding from the Dubai Future Foundation through Guaana.com open research platform. The authors also acknowledge support by the FWF Austrian Science Fund project S11603-N16. ACKNOWLEDGMENTS The authors are indebted to Dr. C. M. Schafer for carefully checking the formulae and wish to thank Drs. ´A. Bazs´o and C. Lhotka for many fruitful discussions and constructive feedback. 7 Maindl et al. APPENDIX Stability of a rotating asteroid housing a space station A MODEL 1 CENTRIFUGAL FORCE CALCULATION Starting at (7),  π(cid:90) 0 F1 = 2ω2ρ a(cid:90) b a dy dϕ √ (cid:90) −a 0 a2−y2 hc/2(cid:90) dy dϕ −hc/2 rc(cid:90) 0 r2dr  π(cid:90) 0  r2dr − 2 hc(cid:90) 1 3 − 1 2 hc dy (cid:18) b (cid:19)3 a(cid:90)  (cid:19)3 1 (cid:18) b −a a a 8 2 − rc dy(cid:0)a2 − y2(cid:1) 3 (cid:34) y(5a2 − 2y2) (cid:112) a2 − y2 + 3a4 arctan y(cid:112)a2 − y2 (cid:35)a −a  . − rc 3 hc we get F1 = 2π 3 ω2ρ = 2π 3 ω2ρ Considering this simplifies to x→±∞ arctan x = ±π lim 2 (cid:34)(cid:18) b (cid:19)3 3 a 8 (cid:35) = π ω2ρ πa4 − rc 3 hc , (cid:18) π 4 (cid:19) . a b3 − 2 3 rc 3hc F1 = 2π 3 ω2ρ This is the same expression as (8). B MODEL 2 CENTRIFUGAL FORCE CALCULATION Starting at (15), we get (cid:90) 2π 0 dϕ F2 = ω2ρ (cid:90) 1 2 hc − 1 2 hc (cid:90) b a rc dy √ a2−y2 dr r2 (21) (22) (23) (24) (25) (26) 8 Maindl et al. Stability of a rotating asteroid housing a space station (cid:35) (cid:35) (27) (28) F2 = 2πω2ρ = 2πω2ρ a rc dy dr r2 2 hc − 1 √ a2−y2 2 hc 2 hc − 1 (cid:90) b (cid:90) 1 (cid:34)(cid:18) b (cid:19)3(cid:0)a2 − y2(cid:1) 3 (cid:90) 1 (cid:34)(cid:18) b (cid:19)3(cid:90) 1 dy(cid:0)a2 − y2(cid:1) 3 (cid:40)(cid:18) b (cid:34) (cid:19)3 1 y(cid:0)5a2 − 2y2(cid:1)(cid:112) 2 hc − 1 2 hc 2 hc 1 3 dy a 2 − rc 3 2 − hc rc 3 2π 3 ω2ρ = a = (cid:41) With the identity arcsin x = arctan(x/(cid:112)1 − x2) and arcsin(−x) = − arcsin x the integral evaluates to y(cid:112)a2 − y2 a2 − y2 + 3a4 arctan (cid:33)(cid:35) 1 − hc rc (cid:32) 2π 3 ω2ρ − 1 2 hc 2 hc a 8 3 . (30) (29) (cid:33)(cid:115) (cid:19)3(cid:34) (cid:40) (cid:18) b a 1 8 hc 2 (cid:32) F2 = 2π 3 ω2ρ + 3a4 arcsin hc 2 (cid:19)3(cid:34) + (cid:32) hc 2a hc 1 8 (cid:18) b a = 2πω2ρ 3 5a2 − 2 2 hc 4 (cid:32) 2 a2 − hc 4 (cid:33)(cid:115) 5a2 − 2 2 hc 4 (cid:33)(cid:115) 2 a2 − hc 4 − 3a4 arcsin −hc 2a (cid:35) 2 5a2 − hc 2 a2 − hc 4 2 + 6a4 arcsin hc 2a − hc rc − hc rc (cid:35)  . 3 (cid:41) 3 (31) (32) This is the same expression as (16). C MODEL 2 STRESS Combining equations (16), (17), and (19), we get the total stress σ2, σ2 = = At + As 2πω2ρ F2 1  b hc (cid:115) 2a 8 3 × (cid:19)3hc (cid:18) b a (cid:32) 5a2 − hc 2 2 (cid:33)(cid:115) a2 − hc 4 2 + 6a4 arcsin hc 2a (cid:34) (cid:32) 2 4a2 − hc 4 + 4ab arcsin − 2hcrc + 2π hc 4a b2 2 1 − hc 4a2 3   − hc rc (cid:35)−1 (cid:33) − rc 2 (33) . (34) 9 Maindl et al. REFERENCES Stability of a rotating asteroid housing a space station Battista, N., Meloni, M. A., Bari, M., Mastrangelo, N., Galleri, G., Rapino, C., et al. (2012). 5- Lipoxygenase-dependent apoptosis of human lymphocytes in the International Space Station: data from the ROALD experiment. FASEB J. 26, 1791 -- 1798 Cogoli, A. (1993). The effect of space flight on human cellular immunity. Environmental Medicine 37, 107 -- 116 Gasperi, V., Rapino, C., Battista, N., Bari, M., Mastrangelo, N., Angeletti, S., et al. (2014). A functional interplay between 5-lipoxygenase and µ-calpain affects survival and cytokine profile of human Jurkat T lymphocyte exposed to simulated microgravity. Biomedical Results Gopalakrishnan, R., Genc, G. O., Rice, A. J., Lee, S. M., Evans, H. J., Maender, C. C., et al. (2010). Muscle volume, strength, endurance, and exercise loads during 6-month missions in space. Aviation, Space, and Environmental Medicine 81, 91 -- 102 Grandl, W. and Bazso, A. (2013). Near earth asteroids -- prospection, orbit modification, mining and habitation. In Asteroids: Prospective Energy and Material Resources, ed. V. Badescu (Berlin, Heidelberg: Springer Berlin Heidelberg), 415 -- 438. doi:10.1007/978-3-642-39244-3\ 17 Haghighipour, N., Maindl, T. I., Schafer, C. M., and Wandel, O. J. (2018). Triggering the Activation of Main Belt Comets: The Effect of Porosity. ApJ 855, 60. doi:10.3847/1538-4357/aaa7f3 Harris, L. R., Herpers, R., Hofhammer, T., and Jenkin, M. (2014). How much gravity is needed to establish the perceptual upright? PLOS One 9 [Dataset] JPL (2018). JPL Small Body Database (2018-11-10) Karaman, K., Cihangir, F., Ercikdi, B., Kesimal, A., and Demirel, S. (2015). Utilization of the Brazilian test for estimating the uniaxial compressive strength and shear strength parameters. Journal of the Southern African Institute of Mining and Metallurgy 115, 185 -- 192 Keyak, J. H., Koyama, A. K., LeBlanc, A., Lu, Y., and Lang, T. F. (2009). Reduction in proximal femoral strength due to long-duration spaceflight. Bone 44, 449 -- 453 Kiss, J. Z., Kumar, P., Millar, K. D. L., Edelmann, R. E., and Correll, M. J. (2009). Operations of a spaceflight experiment to investigate plant tropisms. Advances in Space Research 44, 879 -- 886. doi:10.1016/j.asr.2009.06.007 Kitaya, Y., Kawai, M., Tsuruyama, J., Takahashi, H., Tani, A., Goto, E., et al. (2001). The effect of gravity on surface temperature and net photosynthetic rate of plant leaves. Advances in Space Research 28, 659 -- 664. doi:10.1016/S0273-1177(01)00375-1 Kitaya, Y., Tani, A., Goto, E., Saito, T., and Takahashi, H. (2000). Development of A Plant Growth Unit for Growing Plants Over A Long-Term Life Cycle under Microgravity Conditions. Advances in Space Research 26, 281 -- 288. doi:10.1016/S0273-1177(99)00572-4 Maindl, T. I., Eggl, S., Schafer, C. M., and Chesley, S. R. (2018). Momentum Transfer Uncertainty in the Double Asteroid Redirection Test and its Consequences for Didymos' Impact Risk Assessment. Icarus subm. Maindl, T. I., Schafer, C., Speith, R., Suli, ´A., Forg´acs-Dajka, E., and Dvorak, R. (2013). SPH- based simulation of multi-material asteroid collisions. Astronomische Nachrichten 334, 996 -- 999. doi:10.1002/asna.201311979 Nelson, E. S., Mulugeta, L., and Myers, J. G. (2014). Microgravity-induced fluid shift and ophthalmic changes. Life (Basel) 4, 621 -- 665 Stowe, R. L. (1969). Strength and Deformation Properties of Granite, Basalt, Limestone, and Tuff at Various Loading Rates. Misc. paper C-69-1, Army Engineer Waterways Experiment Station Vicksburg MS 10 Maindl et al. Stability of a rotating asteroid housing a space station Taibbi, G., Kaplovitz, K., Cromwell, R. L., Godley, B. F., Zanello, S. B., and Vizzeri, G. (2013). Effects of 30-Day Head-Down Bed Rest on Ocular Structures and Visual Function in a Healthy Subject. Aviation, Space, and Environmental Medicine 84, 148 -- 154 Taylor, T. C., Grandl, W., Pinni, M., and Benaroya, H. (2008). Space Colony from a Commercial Asteroid Mining Company Town. In Space Technology and Applications International Forum-STAIF 2008, ed. M. S. El-Genk. vol. 969 of American Institute of Physics Conference Series, 934 -- 941. doi:10.1063/1.2845060 Versari, S., Longinotti, G., Barenghi, L., Maier, J. A., and Bradamante, S. (2013). The challenging environment on board the International Space Station affects endothelial cell function by triggering oxidative stress through thioredoxin interacting protein overexpression: the ESA-SPHINX experiment. Aviation, Space, and Environmental Medicine 99 FIGURE CAPTIONS Figure 1. Sketch of a cylindrical space station of height hc and radius rc. 11 Maindl et al. Stability of a rotating asteroid housing a space station Figure 2. Rotation rates ω in rad s−1 and rpm necessary to achieve an artificial gravity of gc on the lateral surface of a cylinder with radius rc. Figure 3. Sketch of asteroid with cylindrical cavern used to house a space station. Picture: Asteroid Vesta. Credit: NASA/JPL-Caltech/UCAL/MPS/DLR/IDA. 12 dxyF2F2abrchc2Maindl et al. Stability of a rotating asteroid housing a space station Figure 4. Model 1 material stress for different dimensions of the space station (dimensionless radius rc ρ gcb for ρ given in g cm−3, gc measured in units of gE, and height hc). The blue contour lines give the ratio σ and b measured in meters, respectively. The dotted red lines give contours of the ratio S a b obtained via (14), the upper x-axis gives the scaled rotation rate. 13 Maindl et al. Stability of a rotating asteroid housing a space station Figure 5. Simplified sketch of cavity for clarification. Force acting on material. 14 Maindl et al. Stability of a rotating asteroid housing a space station Figure 6. Model 1 results for artificial gravity of 0.38 gE in a space station of radius rc and height hc, the color code and the blue contour lines give the tensile stress σ1 resulting from the required rotation rate obtained via (9). The red dotted lines give the usable surface area S of the space station, the upper x-axis denotes the required rotation rate. 15 Maindl et al. Stability of a rotating asteroid housing a space station Figure 7. Model 2 results for artificial gravity of 0.38 gE in a space station of radius rc and height hc, the color code and the blue contour lines give the combined tensile and shear stress resulting from the required rotation rate obtained via (9). The red dotted lines give the usable surface area S of the space station, the upper x-axis denotes the required rotation rate. 16
1310.2642	1	1310	2013-10-09T21:56:06	KOI-152's Low Density Planets	[ "astro-ph.EP" ]	KOI-152 is among the first known systems of multiple transiting planetary candidates (Steffen et al. 2010) ranging in size from 3.5 to 7 times the size of the Earth, in a compact configuration with orbital periods near a 1:2:4:6 chain of commensurability, from 13.5 days to 81.1 days. All four planets exhibit transit timing variations with periods that are consistent with the distance of each planet to resonance with its neighbors. We perform a dynamical analysis of the system based on transit timing measurements over 1282 days of \textit{Kepler} photometry. Stellar parameters are obtained with a combination of spectral classification and the stellar density constraints provided by light curve analysis and orbital eccentricity solutions from our dynamical study. Our models provide tight constraints on the masses of all four transiting bodies, demonstrating that they are planets and that they orbit the same star. All four of KOI-152's transiting planets have low densities given their sizes, consistent with other studies of compact multiplanet transiting systems. The largest of the four, KOI-152.01, has the lowest bulk density yet determined amongst sub-Saturn mass planets.	astro-ph.EP	astro-ph	KOI-152's Low Density Planets Daniel Jontof-Hutter1, Jack J. Lissauer1, Jason F. Rowe1,2, Daniel C. Fabrycky3 [email protected] Received ; accepted 3 1 0 2 t c O 9 . ] P E h p - o r t s a [ 1 v 2 4 6 2 . 0 1 3 1 : v i X r a 1NASA Ames Research Center, Moffett Field, CA 94035, USA 2SETI Institute/NASA Ames Research Center, Moffett Field, CA 94035, USA 3Department of Astronomy and Astrophysics, University of Chicago, 5640 South Ellis Avenue, Chicago, IL 60637, USA -- 2 -- ABSTRACT KOI-152 is among the first known systems of multiple transiting planetary candi- dates (Steffen et al. 2010) ranging in size from 3.5 to 7 times the size of the Earth, in a compact configuration with orbital periods near a 1:2:4:6 chain of commensurability, from 13.5 days to 81.1 days. All four planets exhibit transit timing variations with peri- ods that are consistent with the distance of each planet to resonance with its neighbors. We perform a dynamical analysis of the system based on transit timing measurements over 1282 days of Kepler photometry. Stellar parameters are obtained with a com- bination of spectral classification and the stellar density constraints provided by light curve analysis and orbital eccentricity solutions from our dynamical study. Our mod- els provide tight constraints on the masses of all four transiting bodies, demonstrating that they are planets and that they orbit the same star. All four of KOI-152's transiting planets have low densities given their sizes, consistent with other studies of compact multiplanet transiting systems. The largest of the four, KOI-152.01, has the lowest bulk density yet determined amongst sub-Saturn mass planets. 1. Introduction Within our Solar System, Earth and smaller bodies are primarily mixtures of refractories, rock and metals. In the outer solar system, bodies that are too small to retain deep atmospheres contain rock and ices. In the larger planets, including Uranus and Neptune, the light elements H and He dominate the volume. There no local examples of bodies intermediate in size or mass between Earth (1R⊕, 1M⊕) and Uranus/Neptune, both of which are larger than 3.8 R⊕ and more massive than 14 M⊕. Mass determinations of transiting exoplanets are beginning to allow the characterization of planets in this size range. As more planetary masses and radii are measured, -- 3 -- their bulk densities will provide more constraints on their compositions. To derive meaningful planetary densities requires both accurate mass and radius determinations. The ratio of the planetary radius to the stellar radius is a direct measurement from transit light curves, where the fraction of starlight blocked during transit is a simple measure of the projected area of the planet. The uncertainty in this measurement typically rests on the accuracy to which the stellar radius can be constrained. The star 55 Cancri is unique as the host of a transiting sub-neptune exoplanet in having a direct measurement of its radius via interferometry (von Braun et al. 2011). High resolution spectral classification of the atmosphere of Kepler host stars gives a model dependent measurement of the stellar radius with an uncertainty of typically 10%. The measurement of stellar radius from spectral classification and modelling can be improved upon with additional information. The gold standard for this purpose is where asteroseismic oscillations are detected in the photometric light curve. Amongst these stars, uncertainties in mass and radius can be reduced to ∼ 1%, although detections are only available for giant stars and the brightest dwarf stars (Huber et al. 2013). Another constraint on the stellar density, and hence its radius, can be gleaned from the orbital constraints of exoplanets. The scaled semi-major axis, a/R⋆, (where R⋆ is the stellar radius) can be estimated roughly from the transit and ingress durations, but accurate measurement of a/R⋆ requires additional information about the orbital eccentricity and alignment (ω), the longitude of pericenter from the passage of the orbiting planet through the sky-plane towards the Earth. For a measured fractional transit depth δ, transit duration T , and ingress or egress duration τ, a R⋆ = δ1/4 π P √T τ √1 - e2 1 + esin ω! , (1) (Winn 2011). For the purposes of measuring the stellar radius, information about the the orbital eccentricity from dynamical fits, used in Eq. 1, can provide an independent constraint on the -- 4 -- stellar bulk density following Kepler's Third Law (Seager & Mallén-Ornelas 2003; Winn 2011): ρ⋆ ≈ 3π R⋆(cid:19)3 GP2(cid:18) a . (2) Here ρ⋆ is the bulk density of the star, G is the gravitational constant, and P is the orbital period of the transiting planet. In their study of Kepler-11, Lissauer et al. (2013) used dynamical solutions for the orbital eccentricities, alongside high resolution spectra to reduce the uncertainty on the stellar radius to 2%. The planetary radii were measured with nearly that precision. Of the transiting exoplanets, the majority of mass determinations to date are the result of radial velocity (RV) spectroscopy, and amongst these, most are the so-called hot jupiters, planets with substantial envelopes that orbit very close to their host star. The transiting neptunes and sub-neptunes with measured RV masses all have short orbital periods, the longest being Kepler-68 b, (Gilliland et al. 2013) at 5.4 days. Note that Kepler-18's planets have had their masses measured by the combined constraints of RV and transit timing variations (TTVs), the farthest one orbiting every 14.9 days (Cochran et al. 2011). TTVs exploit the high degree of accuracy in measuring the transit times of transiting exoplanets, with transit time uncertainties as low as a few minutes in some cases. These probe interplanetary perturbations, and in general are sensitive to the mass ratio of perturbing neighboring planets to the host star. The strongest signals in TTVs occur when planets are near (but not trapped in) mean motion resonances, and the resonant argument cycles with a periodicity that is well sampled over a baseline of transit timing measurements. Near first order resonances, the coherence time is long enough for perturbations to build constructively to an easily detectable amplitude. Too far from resonance, the perturbations lose their coherence rapidly, and the TTV amplitude is reduced. Too near resonance, Kepler's four years of observations do not cover a complete cycle. Nevertheless, since TTVs are sensitive to interplanetary perturbations and not the effect of the planet on the star, planetary masses can be determined to far greater orbital periods than RV, as long as enough transit times have been measured to detect timing variations. The -- 5 -- super-neptune Kepler-30 d, with an orbit period of 143 days (Sanchis-Ojeda et al. 2012) has the longest orbital period for an exoplanet with a mass determination on the mass-radius diagram. The difference in orbital periods probes by these two separate techniques highlight their respective biases. RV planets orbit close to their star, are hotter, and several, particularly the earths; including Kepler-10 b (Batalha et al. 2011), CoRoT-7 b (Ferraz-Mello et al. 2011), KOI-94 b (Weiss et al. 2013), Kepler-20 b (Gautier et al. 2012) and potentially Kepler-18 b (Cochran et al. 2011) seem to lack deep atmospheres. The TTV mass determinations, such as Kepler-11 (Lissauer et al. 2011, 2013), Kepler-18 (Cochran et al. 2011), Kepler-30 (Sanchis-Ojeda et al. 2012) and Kepler-36 (Carter et al. 2012) are all either compact multiplanet systems with a small period ratio between neighboring planets, or near resonance. Although detecting a TTV signal certainly favors larger masses, the orbital stability of compact multiplanet systems like Kepler-11 require smaller masses and/or eccentricies given their neptune-like planetary radii. On a practical note, a compact configuration with high multiplicity reduces the risk of intermediate, non-transiting planets confusing TTV dynamical models. KOI-152 (with a Kepler magnitude 13.9, located at RA = 20h02m04.11s, Dec = 44◦22m53.7s Steffen et al. 2010) is an excellent candidate for TTV modeling. It has four planetary candidates near first order Mean Motion Resonances, with clear evidence of transit timing variations (TTVs). Steffen et al. (2012) first noted TTVs at KOI-152 following 6 quarters of data, though only three candidates were known at the time. These three are near the 1:2:4 resonance suggesting that multi-planet resonances are reasonably common (Wang et al. 2012). Wu & Lithwick (2013) noted the TTVs at KOI-152, and also found evidence that the inner two planetary candidates have significant eccentricities. This suggests that circular fits to the transit times would be hampered by the mass-eccentricity degeneracy (Lithwick et al. 2012; Wu & Lithwick 2013), casting uncertainty on measured masses. In Section 2, we introduce our methodology for measuring transit times, and in Section 3, we -- 6 -- examine the transit timing variations, evaluating the applicability of zero-eccentricity analytical solutions for the planetary masses. In Section 4 we describe our numerical models and fits to the transit timing variations, and present our results for the planetary masses and orbital parameters, and stellar parameters. Our analysis confirms that the candidates of KOI-152 are planets, and we refer to them as such for the remainder of the paper. In Section 5 we consider the potential effects of non-transiting perturbers on our four-planet model. In Section 6 we characterize the planets' masses, radii and by bulk densities, and in Section 7, we compare our results for KOI-152's planets with other known plamets of masses less than 30 M⊕ on the mass-radius diagram. 2. Measurement of Transit Times from Kepler Photometric Time Series Variations in the brightness of KOI-152 were monitored with an effective duty cycle exceeding 90% starting at barycentric Julian date (BJD) 2454964.512, with all data returned to Earth at a cadence of 29.426 minutes (long cadence, LC); data were also returned at a cadence of 58.85 seconds (short cadence, SC) beginning from BJD 2455093.216. Here and throughout we base our timeline for transit data from T = JD-2,454,900. Our analysis uses short cadence data where available, augmented by the long cadence dataset primarily during the epoch prior to T< 193 days, for which no SC data were returned to Earth. We obtained these data from the publicly-accessible MAST archive at http://archive.stsci.edu/kepler/ . To measure the transit times from the light curve, we adopt the procedure explained in detail in Appendix 7.1 of Lissauer et al. (2013). Here and throughout we refer to the planets as 'b' (KOI-152.03), 'c' (KOI-152.02) , 'd' (KOI-152.01), and 'e' (KOI-152.04) in order of their orbital periods. -- 7 -- 3. Analytics We begin with the orbital periods based on a linear fit to the observed transit times, summarized in Table 1. We shall solve for the orbital parameters of these planets at T = 780.0 days, an epoch chosen to be near the middle of our dataset. For each candidate, the first transit time after this chosen epoch, calculated from a linear ephemeris to the set of transit times is at time T0. Planetary candidate Period (days) T0 b KOI-152.03 13.4845 784.3010 c KOI-152.02 27.4023 806.4999 d KOI-152.01 52.0909 821.0171 e KOI-152.04 81.0631 802.1119 Table 1: A linear fit to sixteen quarters of Kepler's observed transit times for the planets at KOI- 152, specified as orbital periods and the first calculated transit time after T = 780 days. This configuration of planetary orbits lies close to a 1:2:4:6 resonance chain of orbital periods, and this system is known to exhibit TTVs (Steffen et al. 2012; Wu & Lithwick 2013; Relles 2013; Mazeh et al. 2013). Following the convention of Lithwick et al. (2012) and Wu & Lithwick (2013), we can measure the proximity, ∆, of each adjacent pair in this chain to the nearest first order ( j : j - 1) resonance as follows: ∆1 = j - 1 j P′ P - 1, (3) where P and P′ are the orbital periods of the inner and outer planets respectively. The expected TTV period in this case is: j P′ PT TV =(cid:12)(cid:12)(cid:12)(cid:12) - 1 j - 1 P (cid:12)(cid:12)(cid:12)(cid:12) . (4) - -- 8 -- We seek a similar measure for proximity to second order resonances (∆2), where the expected TTV period replaces j - 1 with j - 2 in Equations 3 and 4. Table 2 highlights the proximity of each pair to first or second order resonances. Each pair is close to a first order resonance (either 2:1 or 3:2). More distant pairings are close to high order (weaker) resonances, the lowest of which is the near 3:1 resonance between c and e. pair period ratio ∆1 ∆2 Expected TTV period (days) b,c c,d d,e c,e 2.032 1.901 1.557 2.959 0.016 -- -0.050 -- 0.037 -- 852.8 525.9 721.4 -- -0.014 (1942.0) Table 2: Orbital period ratios in the KOI 152 system, and their proximity to first order ( j: j - 1, third column) and second order (fourth column) resonances. The final column denotes expected TTV periodicities for each pair of potential interactions, with the periodicities near resonances that are not first order, and thus likely to produce weak perturbations in parentheses. Note that for comparable TTV amplitudes, and assuming the TTVs are linear, the TTVs on 'c' and 'd' are likely a superposition of the two periodicities caused by their immediate inner and outer neighbors. Assuming there are no unseen perturbers, we seek to assess a model for the observed transit times to each transiting planet as as the sum of perturbations of its nearest neighbors. In Fig. 1, we show the transit timing variations for each transiting planet, and solve for sinusoidal fits to the TTVs. The TTV periods are fixed at their expected values based on the orbital periods, and the best-fit amplitudes and phases are solved by MCMC. The solutions to the amplitudes and phases are in Table 3. Lithwick et al. (2012) define the phase of the TTV for circular orbits φttv = 0, when the TTV crosses from above to below the linear fit to the orbit period. For KOI-152, we calculate time Tφ that TTVs transition from negative to positive for inner planets (φttv = 180◦), and where the transition is from positive to negative (φ′ ttv = 0◦) for outer planets of -- 9 -- the same pair. In this notation, anti-correlated TTVs from pairwise interactions on circular orbits have the same Tφ. The uncertainties were measured by recording all extrema in TTV phase and amplitude for models within one reduced χ2 unit from the best fit solution. The uncertainties were largely symmetric and hence we quote the average of positive and negative uncertainties for simplicity. Note that the amplitude of the TTVs for the inner two candidates are slightly less Candidate TTV period (days) TTV ampl. (mins) Tφ (days) Tφ,circ ∆φttv b c d e 852.8 852.8 525.9 525.9 721.4 721.4 6.24±1.35 14.24 ±1.59 5.67±1.48 6.56±0.96 8.31±0.52 5.49±1.36 231.0±30.4 264.5±14.3 751.9±23.0 775.0±13.1 467.2±14.2 422.6±31.4 329 329 846 846 546 546 41± 13◦ 27± 6◦ 64± 16◦ 49± 9◦ 39± 7◦ 62± 16◦ Table 3: The superposition of TTVs at KOI-152, with their expected TTV periods (second column), and their best fit amplitudes (third column) and times that the TTVs transition across zero (fourth column) as depicted graphically in Fig. 1, here with measured uncertainties. The fifth column lists the times that the TTVs would transition across zero from purely circular orbits, where the longitude of conjunctions crosses the transit line of sight, and the final column measures the phase difference in degrees. All times are measured in days unless otherwise indicated. than than the measured values of Wu & Lithwick (2013), and agree within 1σ uncertainties. For orbits with no free eccentricity, TTV phases are anti-correlated. The phases in the fourth column of Table 3 are anti-correlated and agree at the 1.0σ level, between 'b' and 'c', as well as between 'c' and 'd'. However, Tφ of 'd' and 'e' are separated by 1.3σ, weakly suggestive of a phase shift in their TTVs that is not equal to 180◦. We can perform another test on orbital phases by comparing these times Tφ with the date -- 10 -- that the longitude of conjunctions is closest to the line of sight, for pairs of candidates, assuming circular orbits. These dates are in the fifth column of Table 3. For 'b' and 'c', the nearest conjunction to Tφ occurs at T= 329 days, for 'c' and 'd' at T = 846 days, and for 'd' and 'e' at T = 546 days. In each of these cases, there is a phase shift in the expected phase from circular orbits to the observed TTV phases. The phase shift is evidence of significant free eccentricity in the planetary orbits. Before we relax the assumption of circular orbits, we can calculate a quick estimate of the planetary masses. Nominal estimates of the masses of these planets can be made following the solutions of Lithwick et al. (2012); Wu & Lithwick (2013), assuming the orbits are circular. For an interacting pair of planets of mass m and m′, orbiting a star of mass M⋆, near a ( j: j - 1) resonance at periods P and P′ respectively π j π j2/3( j - 1)1/3, V ′∆ P′g(cid:12)(cid:12)(cid:12)(cid:12) m = M⋆(cid:12)(cid:12)(cid:12)(cid:12) m′ = M⋆(cid:12)(cid:12)(cid:12)(cid:12) P f (cid:12)(cid:12)(cid:12)(cid:12) V ∆ (5) (6) where f (∆) = and g(∆) are numerical coefficients of the disturbing function near resonance, calculated in Lithwick et al. (2012). Here V and V ′ are the TTV amplitudes. For planets 'c' and 'd', we calculate two estimates of the mass. We table these nominal masses in Table 4: Note that for both 'c' and 'd', the nominal masses estimated by the TTVs induced on their inner and outer neighbors are inconsistent. This implies that either the TTVs are caused by unseen perturbers, or that circular orbits are a poor fit to the data. The expectation of significant free eccentricity in the inner two planets of KOI-152 was noted by Wu & Lithwick (2013). Due to the degeneracy between mass and eccentricity in TTVs, the expectation of eccentricity casts doubt upon planetary mass estimates made under an assumption of circular orbits. TTVs with significant non-sinusoidal components contain information that is difficult to probe analytically. Hence we perform numerical fits to the transit times to solve for masses and the osculating orbital parameters of the planets at the epoch T= 780 days. -- 11 -- b c Planetary candidate mp (10- 6M⋆) 98.7±11.0 22.3±4.8 44.5±6.5 27.3±7.1 7.3±1.8 19.1±1.2 d e Table 4: Nominal mass estimates assuming circular orbits for the planets of KOI-152, relative to the host star. For 'c' and 'd', perturbations on two neighboring planets permits two estimates of masses, at the amplitudes calculated in Table 3. Their inconsistency here may be due to significant orbital eccentricities. Our final estimates for the planetary masses are in Section 4. 4. Dynamical models of KOI-152 with four planets Figure 2 shows the transit timing variations for the four known candidates and a dynamical fit to this dataset. Our free parameters were orbital period, the time of the first transit T0 after epoch (T = 780 days), the eccentricity vectors ecos ω and esin ω, (where ω is measured from the sky-plane and reaches 90◦ if the pericenter coincides with the transit line of sight), and planetary mass. The dynamical models measure mass as a fraction of the stellar mass. However, following Lissauer et al. (2013), an accurate constraint on ρ⋆ and hence the stellar mass is one of the benefits of dynamical models. To integrate planetary motions, we adopt the 8th order Runge-Kutta Prince-Dormand method, which has 9th order errors. In all of our models, the orbital period and phase of each of the planets are free parameters. The phase is specified by the midpoint of the first transit subsequent to our chosen epoch. We keep all planetary masses as free parameters. We have assumed co-planarity, i.e., negligible mutual inclinations between planetary orbits, in all of our dynamical models. We make no attempt to model transit durations or impact parameters in our -- 12 -- dynamical simulations. Our integrations produce an ephemeris of simulated transit times, S, and we compare these simulated times to the observed transit times (O). We employ the Levenberg-Marquardt algorithm to search for a local minimum in χ2. The algorithm evaluates the local slope and curvature of the χ2 surface. Once it obtains a minimum, the curvature of the surface is used to populate the covariance matrix and evaluate uncertainties. Other parameters are allowed to float when determining the limits on an individual parameter's error bars. Assuming that the χ2 surface is parabolic in the vicinity of its local minimum, its contours are concentric ellipses centered at the best-fit value. The orientations of these ellipses depend on correlations between parameters. The errors that we adopt account for the increase in uncertainty in some dimensions due to such correlations. Our best fit model for the nominal 190 transit times (shown as TTVs in Fig. 1) of the four known planetary candidates leaves 6 data points that are outliers beyond 3σ (where O - S/σ > 3) (of which there should be ≈ 0.4, i.e., likely zero or one, if the errors were Gaussian), and 20 points that are between 2 and 3σ (of which there should be ≈ 8 if the errors were Gaussian). Clearly the uncertainty in the transit times is either under-estimated or the four-planet model is wrong. The outliers may be due to errors in some of the the measured transit times that are not incorporated into timing uncertainty estimates; sources of such errors could be stellar activity, instrumental effects, etc. To assess our dynamical model with χ2 minimization requires a method for dealing with these outliers. Lissauer et al. (2013), in their TTV analysis of Kepler-11, compared three independent methods of measuring transit times from light curves to filter out outlying or anomalous transit times. They discarded points that did not have overlapping error bars for transit times with at least one of the other two transit time measurements. Here, for KOI-152, we seek a method of self-filtering a single measured set of transit times, noting that the distribution of measured -- 13 -- times has far too many 3σ outliers as well as too many 2σ outliers. Hence, we compare best fit dynamical models against the combined 'raw' set of short and long cadence transit times with best fit models where outliers beyond 3σ or 2σ respectively are removed, and a dataset of short cadence only transit times with outliers beyond 3σ removed. We thus conduct our dynamical fits against four sets of transit times. For each set of transit times, we find a best fit solution and then evaluate which outliers, if any, are beyond our 2σ or 3σ threshold. These points are removed and another best fit solution is found with dynamical models. Iterations continued until there were no more outliers at the best fit solution. To account for multiple local minima in the χ2 surface, we used multiple choices of initial conditions, recording all solutions with χ2 values within one reduced χ2 unit of the best fit model. The outputs of each local minimum were used as initial conditions for all other sets of transit times in the search for alternative local minima in the χ2 surface. The search over different initial conditions continued until at least 25 local minima within one reduced χ2 unit of the best known fit for each of the four sets of transit times were found. Table 5 summarizes the goodness of fit of the best fit model for each set of transit times. Including all measured transit times leads to χ2/(d.o.f.) > 1, whereas, not surprisingly, using the highest threshold of acceptance for the transit times (O - S/σ < 2), causes χ2/(d.o.f.) < 1. We note here that KOI-152 d contributes the least to the χ2 when all measured transit times are included, and appears to be the least sensitive to the removal of outlying transit times. Figure 3 shows the range of masses and eccentricity vectors that are with one reduced χ2 unit of the best known minimum for each dataset of transit times. All model fits within one reduced χ2 unit of the best known fit for each dataset of transit times were included. The nominal uncertainties for model fits σnom, are reduced for all minima apart from the one with the lowest χ2, with uncertainties reduced to σr = σnomq1 - ∆χ2 χ2/(d.o. f .) . This assumes that the local minimum in -- 14 -- Planet χ2 raw χ2 3σ χ2 2σ χ2 3σSC b c d e 173.27 107.51 63.79 102.12 88.95 56.25 32.49 51.96 30.12 30.61 21.73 22.73 43.52 14.74 9.79 13.44 total 335.87 209.11 127.80 190.24 # fitted TTs χ2/(d.o.f.) 190 1.98 181 1.30 167 0.87 163 1.33 Table 5: χ2 contributions from each planet for a suite of best fit models against four sets of transit times: the raw measured transit times of short and long candence data combined (raw: second column), a combined set with 3σ outliers (O - S/σ > 3) removed (3σ: third column), a combined set with 2σ outliers removed (2σ: fourth column), and in the last column, short cadence only data with 3σ outliers removed (3σSC). The degrees of freedom (d.o.f.) in each case are the number of data points in each set of transit times (the penultimate row) subtract the number of free parameters in each fit (twenty free parameters). χ2 is on a parabolic surface, and therefore extends error bars to reach where χ2/(d.o. f .) ≈ 1. The result in Figure 3 shows that a wide range of eccentricities can satisfy the data for KOI-152 d, and that individual best fit solutions are indeed moderately sensitive at the ∼ 1σ level to the manner in which outlying transit time measurements are handled. ∆χ2 Nevertheless, with our exploration of multiple local minima in the χ2 surface, our uncertainties are augmented to ensure that our best fit models over all four sets of transit times. We note that within each dataset, the uncertainties with differing datasets are all consistent at the 1σ level. In Fig. 4, we show the 1σ uncertainties for the masses and the components of the eccentricity vectors for all four planets. For our total uncertainty for each parameter, we adopt the -- 15 -- union of all uncertainties for each parameter as shown in Fig. 4, and the median of the four best fit solutions for each parameter as our nominal solution. The results for all the fitted parameters are in Table 6. Planet b c d e Period (days) 13.4845 +0.0002 - 0.0002 27.4029 +0.0008 - 0.0006 52.0902 +0.0009 - 0.0010 81.0659 +0.0013 - 0.0011 T0 (JD-2,545,900) ecos ω esin ω 784.307 +0.002 - 0.002 806.475 +0.004 - 0.004 821.011 +0.002 - 0.001 802.126 +0.003 - 0.005 -0.015 +0.006 - 0.011 -0.020 +0.013 - 0.016 0.014 +0.039 - 0.013 0.012 +0.032 - 0.013 -0.003 +0.005 - 0.005 -0.022 +0.014 - 0.022 0.020 +0.045 - 0.019 0.002 +0.033 - 0.019 106 Mp M⋆ 28.0 +19.0 - 15.4 15.1 +4.9 - 6.0 15.3+5.5 - 4.2 10.7 +3.0 - 2.9 Table 6: Our solutions for KOI-152 following dynamical fits incorporating uncertainties over four choices of observed transit times with different outliers excluded. The parameters we measure include the orbital periods (second column), time of first transit after epoch (third column), ecos ω (fourth column), esin ω (fifth column), and planetary mass relative to the mass of the star (sixth column). To test these solutions for long-term orbital stability, we simulated the trajectories of the best fit solutions using a symplectic integrator (Rauch & Hamilton 2002) for 400 Myr and found the system to be stable. However, increasing the masses and eccentricities to their 1σ maxima, the system lasted just 70,000 years before planets were expelled, an indicator that, dynamically, KOI-152's planets are packed close to the stability limit. The perturbations of each planet orbiting KOI-152 can be deconvolved as a linear sum of pairwise TTVs to a high degree of accuracy. Figure 5 highlights the contribution of each planetary candidate to the TTVs of the other planets in the system, including the non-sinusoidal components to the TTV signal that are captured by dynamical fits. The remarkable fit of the pairwise perturbations in adding up to match the net TTVs is indicative of the linear nature of KOI-152's TTVs over the Kepler observational baseline. The orbits of KOI-152's planets are -- 16 -- close enough to resonance for the coherence period, the length of time over which perturbations act constructively, to be much longer than synodic period. However, the orbits are not so close to resonance that coherent perturbations reach high amplitude, and the TTVs are always a tiny fraction of an orbital period. Hence the TTVs remain in the linear regime. We also note that the near-second-order resonance TTVs on KOI-152 c induced by 'e' have an amplitude of just one minute and the TTVs of KOI-152 e induced by 'c', have an amplitude of just 2 minutes, even though for these orbital periods near 3:1, ∆2 = 0.014 is lower than each of the first order nearness-to-resonance values (∆1). Note that the expected period of this component of the TTVs is much longer than the baseline of observations and the cycle is incomplete. To characterize the host star, we used the light curve alongside spectral classification. For fitting the light-curve, we adopt the analytic model of Mandel & Agol (2002) for a planet transiting a stellar surface described by a quadratic limb-darkening law, and we adopt the limb-darkening parameters of Claret & Bloemen (2011). We modelled the orbits of each planet as non-interacting Keplerian orbits, and fit the light curve for best fit parameters of the mean stellar density, ρ⋆, the photometric zero point for each planet, the center of transit time for the first observed transit T0, the orbital period P, the impact parameter b, the scaled planetary radius Rp/R⋆, and the components of the eccentricity vector, ecos ω and esin ω. To account for the TTVs, the light curve model cadence was contracted and expanded based on a linear interpolation of measured transit times for each planet to match the observed transit times. To calculate the posterior distributions of model paramters, we used an MCMC routine described in Section 4.1 of Rowe et al. (2013). We used the determination of ecos ω and esin ω from dynamical modeling of the TTVs as constraints characterized by a Gaussian distribution. We generated 4×1,000,000 Markov-Chains and calculated the median value for each model parameter and its 1σ uncertainty interval which we list in Table 7. The light curve model gives a geometrical measurement of ρ⋆, which we combined with the -- 17 -- Planet Rp/R⋆ b c d e 0.02442+0.00018 - 0.00018 0.02618+0.00019 - 0.00016 0.05038+0.00010 - 0.00010 0.02458+0.00079 - 0.00094 depth (ppm) 675.1+8.3 - 9.2 789.9+9.0 - 9.5 2968+11 - 13 453+26 - 28 b 0.410+0.021 - 0.036 0.278+0.047 - 0.037 0.056+0.022 - 0.056 0.963+0.015 - 0.016 i (◦) 88.78+0.07 - 0.09 89.48+0.07 - 0.09 89.93+0.07 - 0.03 89.13+0.02 - 0.02 a/R⋆ 19.26+0.19 - 0.27 30.90+0.30 - 0.43 47.42+0.46 - 0.67 63.68+0.62 - 0.90 Table 7: Transit constraints on the planets of KOI-152, following dynamical models; b signifies impact parameter, i inclination of the orbit to to the plane of the sky and a the orbital semimajor axis. spectroscopic determination of Te f f and [Fe/H]. These we matched to stellar evolution models (Demarque et al. 2004), to estimate the stellar mass and radius. Our uncertainties follow from the posterior distributions of our MCMC analysis. Our results are in Table 8. M⋆(M⊙) R⋆(R⊙) L⋆(L⊙) Teff (K) log g (cm s- 2) Z ρ⋆ (g cm- 3) Age (Gyr) 1.165+0.044 - 0.045 1.302+0.026 - 0.027 2.20+0.18 - 0.22 6174+83 - 117 4.274+0.012 - 0.013 0.0169+0.0026 - 0.0030 0.741+0.026 - 0.034 3.44+0.60 - 0.91 Table 8: The characteristics of the star KOI-152, with 1σ uncertainties. -- 18 -- 5. Non-transiting Perturbers? Near strong mean motion resonances, pairwise interactions cause a TTV frequency at the circulating argument of the nearest resonance. The period of this signal increases closer to the resonance. Thus in Table 2, the highest values in the final column mark the orbit pairs closest to resonance. For the known candidates in this system, all pairwise TTVs due to first order resonance have a complete cycle of observed transit times, hence the well constrained model fit. Figure 6 displays the periodicities detected in the TTVs. For this figure we have constructed periodograms of the TTVs (O-C), as well as the residuals (O-S). For each periodogram the frequency was limited at the Nyquist frequency (half the sample rate as the maximum frequency, corresponding to twice the orbit period). The minimum frequency corresponds to the maximum period that is sampled over the observational baseline. For this limit, we have chosen twice the observational baseline, which would be an lower limit on any incomplete TTV period, or an upper limit on its frequency. In these plots, the absolute power in each peak need not correspond between models, since the combined datasets and the short cadence only dataset have different baselines and therefore different frequency domains. However, the relative height of peaks within each dataset, and their locations in frequency are of interest. In each case the peak in the periodogram corresponds to the expected TTV signal from the last column in Table 2. However, the peaks are broad enough such that for the planets with two neighbors, 'c' and 'd', there are not two clear peaks in the periodogram. For 'c', the single broad peak encompasses the expected TTV signal at 853 days (0.0136 µHz) and 526 days (0.0220 µHz). For 'd' the shape of the broad curve hints at two peaks where we would expect them near 526 days (0.0220 µHz) and 721 days (0.0161 µHz) respectively, although they cannot be resolved. There are no peaks in the residuals that stand out significantly from the background or noise. We expect more residual power at higher frequencies where outliers have been included in the fits, and this is confirmed in Fig. 6. We see no evidence that any unseen planets at KOI-152 contribute -- 19 -- significantly to the TTVs of the four known candidates. Furthermore, it follows from the good fits that we have found to the transit time data at the expected TTV periods that these candidates are very likely to orbit the same star. Of course, we cannot rule out the possibility of other planets perturbing the four known candidates. Since the configuration is so compact, it appears that only an unseen perturber orbiting interior to KOI-152 b or beyond KOI-152 e could have any potential of confusing our four-planet model. We note that KOI-152 e has the highest fraction of its transits that are outliers, and that these are all in the second half of the data. Could this be due to the TTVs of an outer perturber increasing the TTV amplitude over the photometric baseline? Whilst we cannot exclude this possibility, we can consider how this hypothetical planet may affect our solutions. We focus on 'e' since it is more likely to have an effect on our most surprising result, the low mass of KOI-152 d. If a non-transiting planet had an orbital period near 104 days, its orbital period would be near 2:1 with KOI-152 d and 3:2 with KOI-152 e, and it could cause near-first-order resonance TTVs in KOI-152 d and e. Since we see no residual peaks in the periodogram for KOI-152 d, a perturber causing first-order TTVs on 'd' appears unlikely. Hence, only if the TTVs induced on 'd' had a similar period to the TTVs on 'd' caused by 'e' or 'c' would our model for d's TTVs be confused. Any other possible near-first-order resonance with KOI-152 e would leave only high order TTVs on KOI-152 d, which would be an insignificant addition to its TTVs. Our solution to the mass of KOI-152 d is constrained by the well-fitted TTVs it induces on 'c' and 'e'. A false measurement of KOI-152 e's mass would cancel one but not both of these constraints, notwithstanding the freedom for eccentricities to adjust to a different mass for 'e'. It appears unlikely that KOI-152 d, given its position between two transiting planets, would have a substantially different mass if a more distant planet were inducing TTVs in just one of its neighbors, namely, KOI-152 e. The effect of a fifth planet beyond KOI-152 e would have an even smaller effect on our solutions for KOI-152 b and c. -- 20 -- 6. Characterizing the Planets Planet Mass (M⊕) Radius (R⊕) Density (g cm- 3) b c d e 10.9+7.4 - 6.0 5.9+1.9 - 2.3 6.0+2.1 - 1.6 4.1+1.2 - 1.1 3.47+0.07 - 0.07 3.72+0.08 - 0.08 7.16+0.13 - 0.16 3.49+0.14 - 0.14 1.43+0.97 - 0.78 0.62+0.20 - 0.25 0.09+0.03 - 0.02 0.53+0.15 - 0.15 a (AU) 0.117+0.002 - 0.002 0.187+0.002 - 0.003 0.287+0.004 - 0.004 0.386+0.005 - 0.005 e Flux (F⊙,1AU) 0.015+0.012 - 0.006 0.030+0.027 - 0.021 0.025+0.059 - 0.023 0.012+0.044 - 0.005 162 63 27 15 Table 9: The planets of KOI-152. All uncertainties are 1σ confidence intervals, with planetary masses and radii in Earth units, and density in g cm- 3. The flux (final column) is scaled to the flux received from the Sun at 1 AU. Table 6 shows our measured masses, radii, densities and incidence fluxes for each planet based on our best fit solutions, with uncertainties extended to account for all four sets of transit times. The planetary bulk densities follow from the constraint in stellar density: ρp = (ρ⋆)(cid:18)Mp M⋆(cid:19)(cid:18)Rp R⋆(cid:19) - 3 . (7) Each of the terms in parentheses in Equation 7 is an independent source of uncertainty that we add in quadrature. We use this formula because it provides a more appropriate accounting of the uncertainties than the uncertainties in the planetary masses and radii. Due to the tight constraints on stellar and planetary radii, the dominant source of uncertainty here is planet-to-star mass ratios from our dynamical models. Nevertheless, all four candidates can be characterized as having low bulk density, due to the retention of volatiles. KOI-152 b has the highest bulk density (albeit with large uncertainties), despite having a similar radius to KOI-152 c and e. The nominal density of KOI-152 b is slightly less than a pure ice world of the same mass, suggesting either a substantial mass fraction of water and/or a relatively thin H/He envelope, with heavy elements in the interior. KOI-152 c and e have similar characteristics to the planets of Kepler-11 d and e (Lissauer et al. (2013)), and likely have envelopes that are far more significant by volume -- 21 -- than by mass (Lissauer et al. 2013). The bulk density of KOI-152 d is the lowest for a Kepler planet measured to date. Although the temperature of KOI-152 d is unknown, its place on the mass-radius diagram appears to give its envelope a total mass (as a fraction of the planetary mass) of around 50% at T = 500K, roughly 10% at 1000 K, (Rogers et al. 2011). The equilibrium temperature of a planet assumes zero albedo and no internal heat source, and depends solely on the temperature of the host star and the orbital distance. For low eccentricities, Teq ≈ T⋆rR⋆ 2a . (8) Given its distance from the star, the equilibrium temperature of KOI-152 d is 634± 16 K. Hence, it appears reasonable that KOI-152 d has an H/He envelope that contributes significantly more than 10% of the planetary mass, but less than 50% of the mass. 7. KOI-152's Planets on the Mass-Radius Diagram Here we plot neptunes and sub-neptunes on the mass-radius diagram, including all planets with measured radii and masses, with nominal mass determinations up to 30 M⊕. These include mass determinations by radial velocity spectroscopy (RV) as well as TTVs. We adopt the data from published studies that include the most recent stellar spectral classification and mass determinations for: HAT-P-11 b (Kepler-3 b) (Bakos et al. 2010) HAT-P-26 b (Hartman et al. 2011), 55 Cancri e (von Braun et al. 2011; Winn et al. 2011; Gillon et al. 2012), GJ 3470 b (Bonfils et al. 2012), GJ 436 b (Ehrenreich et al. 2011; Ballard et al. 2010), GJ 1214 b (Charbonneau et al. 2009; Valencia et al. 2013), CoRoT-7 b (Ferraz-Mello et al. 2011; Bruntt et al. 2010), Kepler-4 b, (Borucki et al. 2010), Kepler-10 b (Batalha et al. 2011), Kepler-11 b-f (Lissauer et al. 2013), Kepler-18 b-d (Cochran et al. 2011), Kepler-30 b and d (Sanchis-Ojeda et al. 2012), Kepler-36 b and c (Carter et al. 2012), Kepler-68 b (Gilliland et al. 2013), and KOI-94b (Weiss et al. 2013). We exclude KOI-94 c (Mp = 15.6+5.7 (Mp = 35+18 - 15.6M⊕) and KOI-94 e - 28M⊕) from the mass-radius diagram because their masses are poorly constrained, and -- 22 -- in the case of KOI-94 e, its nominal mass is beyond 30 M⊕. For the solar system, we adopt data given in de Pater & Lissauer (2010). The planets in Fig. 7 are all compared to theoretical models of pure water ice, silicate rock, or iron planets. Model curves for planetary radii of planets made from pure ice, rock or iron follow the results of Fortney et al. (2007). Several features stand out from Fig. 7. Firstly, amongst the sub-neptunes, TTV planets are systematically larger and hence less dense than RV planets in the same mass range. This is most likely due to the biases of both techniques. Since RV detectability declines rapidly with orbital distance, few RV masses have been measured beyond a few days orbital period, and several of the planets in the RV sample may well have suffered mass loss from the intense stellar radiation upon their atmospheres. On the other hand, the TTV systems are all dynamically compact multiplanet systems, where for a given planetary size, smaller masses are more likely to permit stable orbits (even though a larger mass makes the TTVs more detectable). Furthermore, for planets in multiplanets systems of a particular mass, the larger (and hence lower density) planets can have their transit times measured more accurately for TTV anlysis. The bottom panel of Fig. 7 highlights the two orders of magnitude in the range in planetary densities that are observed in the mass range up to 30 M⊕. The RV density determinations appear more tightly correlated than the TTV determinations, although the uncertainty in density for the low mass, low density TTV planets are certainly larger. Note that in this plot we show planetary radii from a mass limited sample. For given masses, and thus given detectability considerations, larger planets have more precise transit timing measurements. There is thus a bias in TTV detections to large, low density planets, and not massive, smaller planets. This selection effect is part of the reason that the Rp - ρp curve declines to the right. Nominally, KOI-152 d has the lowest planetary bulk density of exoplanets measured to date, although Kepler-12 b, with a mass of 137 M⊕, and radius of 19 R⊕, has a very similar bulk density within uncertainties. KOI-152 d is twenty times less massive than Kepler-12 b, and also receives roughly 30 times less insolation than Kepler-12 b (Fortney et al. 2011). It is significantly less dense than other characterized sub-Saturn mass -- 23 -- planets. If we parametrize the ratio of the planetary radius to the radius of a planet of the same mass composed purely of ice, or rock, we have a simple test of whether the planet has retained a substantial gaseous envelope. Planets that are less dense than water ice, (above the blue line in the top panel of Fig. 8), are likely to have a volumetrically substantial H/He envelope. Planets that are more dense than pure silicate rock, (below the brown line in the bottom panel of Fig. 8), are less likely to retain a substantial amount of volatiles. Between these two limits, the atmosphere of a planet cannot be determined without more information, since it is unknown in what proportions volatiles are likely to be present in the form of ices and gases. Nevertheless, we surmise from Fig. 8 that planets with Rp < 2R⊕ appear unlikely to retain significant atmosphere, and from the bottom panel that planets with Rp > 4R⊕ are very likely to retain deep atmospheres. All the characterized planets in our chosen mass range show remarkably little scatter in Fig. 8. In Fig. 9, we show the equilibrium temperature of the planet against their radii relative to pure rock or ice. We see an abrupt transition around 1050 K. At higher temperatures, there is no evidence of planets harboring substantial gases, and below this temperature, a wide range of densities appears permissible. One possible exception is Kepler-4b, with a equilibrium temperature of 1614 K, Rp/Rrock = 1.61, and Rp/Rice = 1.12. Borucki et al. (2010) estimated that Kepler-4b retains a deep H/He atmosphere of about 4-6% by mass. Of the "hot" planets below 30 M⊕, Kepler-4b is alone in requiring a deep atmosphere. Nevertheless, we still see a much wider range in bulk densities amongst the cool sub-neptunes. Kepler-36 stands out as having two planets that are close in orbital periods (13.84 and 16.24 days respectively, Carter et al. 2012), with remarkably different densities. These planets appear to straddle the transition in equilibrium temperature at ∼ 1050 K, with the outer planet retaining an atmosphere, and the inner one denser than pure silicate rock (Lopez & Fortney 2013). KOI-152 b lies close to Kepler-36 c on this plot, with a blackbody temperature of 1004 ± 24 K, although it is marginally denser. Amongst the -- 24 -- "cool" sub-neptunes, no exoplanets that are denser than pure rock have yet been discovered. 8. Conclusion Assuming a four-planet model for the TTVs in the four transiting candidates of KOI-152, we have added four sub-Uranus masses to the mass-radius diagram, bringing the total to twenty exoplanets. We confirm the planetary nature of these candidates and that they are planets in the same system. The planets of KOI-152 appear to follow the trend of the planets orbiting Kepler-11, being of low density with significant envelopes. Nevertheless, the planets show remarkable variety in their bulk masses. KOI-152 b, c and e all have radii between 3.5 and 4.0 R⊕, and yet their masses range from ∼3 to ∼13 M⊕, and their bulk densities range from 0.3 to 1.6 g cm- 3. The largest planet, KOI-152 d, (∼ 7R⊕), has a remarkably low mass given its size, and most likely has the lowest nominal bulk density amongst known planets with sub-Saturn masses. D.J. gratefully acknowledges the support of the NASA Postdoctoral Program. -- 25 -- KOI 152 b ) s e t u n i m ( V T T ) s e t u n i m ( V T T ) s e t u n i m ( V T T ) s e t u n i m ( V T T 30 20 10 0 -10 -20 -30 -40 -50 40 30 20 10 0 -10 -20 -30 -40 -50 -60 20 15 10 5 0 -5 -10 -15 50 40 30 20 10 0 -10 -20 -30 -40 0 200 400 600 800 t (days) KOI 152 c 1000 1200 1400 0 200 400 600 800 t (days) KOI 152 d 1000 1200 1400 0 200 400 600 800 t (days) KOI 152 e 1000 1200 1400 0 200 400 600 800 t (days) 1000 1200 1400 Fig. 1. -- Sixteen quarters of transit timing variations for KOI-152 (colored points), using primarily Short Cadence data, supplemented by Long Cadence data where Short Cadence was unavailable. The TTVs are the difference between the observed transit times and a calculated linear ephemeris (O-C). The solid curves are best fit sinusoids ('b' and 'e'), or the sum of two sinusoids ('c' and 'd': shown as dashed curves). The sunusoidal fits solve for amplitude and phase, whilst the periods remain fixed at the expected TTV period, given in Table 2. ) s e t u n i m ( V T T ) s e t u n i m ( V T T ) s e t u n i m ( V T T ) s e t u n i m ( V T T 30 20 10 0 -10 -20 -30 -40 -50 40 30 20 10 0 -10 -20 -30 -40 -50 -60 20 15 10 5 0 -5 -10 -15 50 40 30 20 10 0 -10 -20 -30 -40 b c d e 0 200 400 600 800 1000 1200 1400 -- 26 -- ) s e t u n i m ( s l a u d i s e R ) s e t u n i m ( s l a u d i s e R ) s e t u n i m ( s l a u d i s e R ) s e t u n i m ( s l a u d i s e R 30 20 10 0 -10 -20 -30 -40 30 20 10 0 -10 -20 -30 -40 10 5 0 -5 -10 -15 40 30 20 10 0 -10 -20 -30 -40 -50 b c d e 0 200 400 600 800 1000 1200 1400 t (days) t (days) Fig. 2. -- Observed and simulated transit timing variations for the planetary candidates orbiting KOI-152, using the combined dataset of short and long cadence transit times. The panels on the left side compare O - C (colored data) and model times, S - C (black points). The right hand side plots the residuals with the dynamical model subtracted from the observed transit timing variations. Note the vertical scales for the residuals (right panels) differ from those of the TTVs. -- 27 -- 0.06 0.05 0.04 0.03 0.02 0.01 sw o c e raw 3s 2s SC only 3s 0.07 0.06 0.05 0.04 0.03 0.02 0.01 nw i s e 0 11 12 13 14 15 16 17 18 19 20 21 0 11 12 13 14 15 16 17 18 19 20 21 106Md/M★ 106Md/M★ Fig. 3. -- The results of fitting best fit mass and the components of the eccentricity vector for KOI-152 d using four set of transit times. Each point represents a local minimum in χ2 with uncertainties estimated locally from the covarience matrix. For each set of transit times, local minima were included if, compared to the lowest known χ2 for that dataset, ∆χ2 χ2/(d.o. f .) < 1, and their error bars were reduced such that with uncertainties, ∆χ2 χ2/(d.o. f .) = 1. raw 3s 2s SC only union 3s -- 28 -- 0.08 0.06 0.04 0.02 0 -0.02 -0.04 n i s e 50 45 40 35 30 25 20 15 10 5 M★ / p M 6 0 1 s o c e 10 20 30 40 50 60 70 80 90 -0.06 10 20 30 40 50 60 70 80 90 Period (days) Period (days) 0.06 0.05 0.04 0.03 0.02 0.01 0 -0.01 -0.02 -0.03 -0.04 10 20 30 40 50 60 70 80 90 Period (days) Fig. 4. -- Best fit solutions for planetary masses and the eccentricity vector components esin ω and ecos ω using four different methods of accounting for outliers amongst the measured transit times. The orbital periods are displayed with a stagger to allow comparison between the solutions. Here, the dynamical fit against raw transit times is marked by the grey points, the combined dataset of short and long cadence transit times with 3σ outliers removed is in blue, the same set with 2σ outliers removed is in red, and the short cadence data only with 3σ outliers removed is in green. The black squares mark the median of the four solutions, with error bars showing the union of all uncertainties. w w -- 29 -- ) s e t u n i m ( C - S ) s e t u n i m ( C - S ) s e t u n i m ( C - S ) s e t u n i m ( C - S 10 8 6 4 2 0 -2 -4 -6 -8 20 15 10 5 0 -5 -10 -15 -20 -25 20 15 10 5 0 -5 -10 -15 -20 25 20 15 10 5 0 -5 -10 -15 -20 0 200 400 600 800 t (days) 1000 1200 1400 0 200 400 600 800 t (days) 1000 1200 1400 0 200 400 600 800 t (days) 1000 1200 1400 0 200 400 600 800 t (days) 1000 1200 1400 Fig. 5. -- Pairwise contributions to TTVs from each planet (simulated transit times minus a cal- culated linear ephemeris to the transit times) in the best fit solution to the dataset of transit times with 3σ outliers removed for 'b' (top left), 'c' (top right), 'd' (bottom left), and 'e' (bottom right). In each panel, contributions from 'b' are in blue, 'c' in green, 'd' in orange, and 'e' in red. The contributions are summed and displayed in grey, although these curves are largely obscured by the black curves in most of the panels, which denote the best fit solution, in close agreement with the sum of linear pairwise TTVs. -- 30 -- O-C (TTVs) Residuals (O-S) ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N 0 0 700 600 500 400 300 200 100 0 0 600 500 400 300 200 100 0 bb 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 frequency (m Hz) c 1800 1600 1400 1200 1000 800 600 400 200 0 1400 1200 1000 800 600 400 200 0.05 0.1 0.15 frequency (m Hz) 0.2 0.25 d 0.02 0.04 0.06 0.08 frequency (m Hz) 0.1 0.12 e 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 frequency (m Hz) ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N ) m p p ( r e w o P d e z i l a m r o N b 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 frequency (m Hz) c 450 400 350 300 250 200 150 100 50 0 250 200 150 100 50 0 0 0.05 0.1 0.15 frequency (m Hz) 0.2 0.25 45 40 35 30 25 20 15 10 5 0 d 0 0.02 0.04 0.06 0.08 frequency (m Hz) 0.1 0.12 250 200 150 100 50 0 e 0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 frequency (m Hz) Fig. 6. -- Periodograms of the observed TTVs of the planets and residuals of best fit models for each of the four datasets of transit times. Here, the normalized power measured in parts per million (ppm), and the periodograms have been normalized by summing the power over 10,000 equally-spaced frequency channels between an assumed half-wave over the TTV baseline and the Nyquist frequency. The graphs highlight the dominant frequencies in the bestfit model (left) and the residuals (right). In each graph, the dark grey curves mark the best fit to the raw transit times, the purple curves mark the best fit to the combined SC and LC dataset with 3σ outliers removed, the blue curves mark the best fit to the combined datset with all 2σ outliers removed, and the brown curves mark the best fit to SC only times with 3σ outliers removed. -- 31 -- / R¯ p R 10 8 6 4 2 0 Ice Silicate Rock Iron 20 25 30 0 5 10 15 Mp/M¯ ) 3 - m c g ( 10 1 0.1 1 2 3 4 5 6 Rp/R¯ 7 8 9 10 Fig. 7. -- The measured masses and radii of all known planets with masses < 30 M⊕. Planets of the solar system are depicted by black points, transiting planets with masses measured by radial velocity spectroscopy (RV) are in red, planets with masses measured by TTVs are in blue, and the planets of KOI-152 are in green. The masses of the planets orbiting Kepler-18 are the combined result of RV and TTV observations, and appear as red and blue points above. The curves in the left panel represent solutions for the radii of planets composed of pure water ice, silicate rock, or iron (Fortney et al. 2007). In the bottom panel, we plot bulk densities as a function of planet size. r -- 32 -- e c i R p R / k c o r R p R / 3 2 1 0 5 4 3 2 1 0 0 1 2 3 4 Rp/R¯ 5 6 7 8 0 1 2 3 4 Rp/R¯ 5 6 7 8 Fig. 8. -- Planetary radii compared to solutions of pure ice (top panel) or pure silicate rock (bottom panel). Mass determinations by RV are in red, and TTV in blue, with the planets of KOI-152 in green. Planets above the dotted blue line in the left panel, are less dense than water ice and are likely to have deep H/He envelope. Planets below the dotted brown line in the bottom panel are denser than rock, and are unlikely to have a substantial atmosphere. Planets between these limits must contain some volatiles, but the unknown tradeoff between water or a combination of H/He and rock precludes a definitive answer on whether there is an H/He envelope. -- 33 -- k c o r R p R / 5 4 3 2 1 0 0 500 1000 1500 2000 2500 TBB (K) Fig. 9. -- Planetary radii compared to solutions of pure rock (left panel) as a function of equilibrium temperature. Mass determinations by RV are in red, and TTV in blue, with the planets of KOI-152 in green. A sharp transition to planets with little or no volatiles appears around 1100K. In the cooler regime, no planets that are denser than rock are seen amongst the exoplanets. -- 34 -- REFERENCES Bakos, G. Á., et al. 2010, ApJ, 710, 1724 Ballard, S., et al. 2010, PASP, 122, 1341 Batalha, N. M., et al. 2011, ApJ, 729, 27 Bonfils, X., et al. 2012, A&A, 546, A27 Borucki, W. J., et al. 2010, ApJ, 713, L126 Bruntt, H., et al. 2010, A&A, 519, A51 Carter, J. A., et al. 2012, Science, 337, 556 Charbonneau, D., et al. 2009, Nature, 462, 891 Claret, A., & Bloemen, S. 2011, A&A, 529, A75 Cochran, W. D., et al. 2011, ApJS, 197, 7 Demarque, P., Woo, J.-H., Kim, Y.-C., & Yi, S. K. 2004, ApJS, 155, 667 de Pater, I., & Lissauer, J. 2010, Planetary Sciences (Cambridge University Press) Ehrenreich, D., Lecavelier Des Etangs, A., & Delfosse, X. 2011, A&A, 529, A80 Ferraz-Mello, S., Tadeu Dos Santos, M., Beaugé, C., Michtchenko, T. A., & Rodríguez, A. 2011, A&A, 531, A161 Gautier, III, T. N., et al. 2012, ApJ, 749, 15 Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 659, 1661 Fortney, J. J., et al. 2011, ApJS, 197, 9 -- 35 -- Gilliland, R. L., et al. 2013, ApJ, 766, 40 Gillon, M., et al. 2012, A&A, 539, A28 Hartman, J. D., et al. 2011, ApJ, 728, 138 Huber, D., et al. 2013, ApJ, 767, 127 Lissauer, J. J., et al. 2011, Nature, 470, 53 -- . 2013, ApJ, 770, 131 Lithwick, Y., Xie, J., & Wu, Y. 2012, ApJ, 761, 122 Lopez, E. D., & Fortney, J. J. 2013, ApJ, 776, 2 Mandel, K., & Agol, E. 2002, ApJ, 580, L171 Mazeh, T., et al. 2013, ApJS, 208, 16 Rauch, K. P., & Hamilton, D. P. 2002, DAAS, 33, 938 Relles, http://exoplanet-science/com/KOI-152.html Rogers, L. A., Bodenheimer, P., Lissauer, J. J., & Seager, S. 2011, ApJ, 738, 59 Rowe, J.F., et al. 2013, ApJ, submitted Sanchis-Ojeda, R., et al. 2012, Nature, 487, 449 Seager, S., & Mallén-Ornelas, G. 2003, ApJ, 585, 1038 Steffen, J. H., et al. 2010, ApJ, 725, 1226 Steffen, J. H., et al. 2012, ApJ, 756, 186 Valencia, D., Guillot, T., Parmentier, V., & Freedman, R. S. 2013, ApJ, 775, 10 -- 36 -- von Braun, K., et al. 2011, ApJ, 740, 49 Wang, S., Ji, J., & Zhou, J.-L. 2012, ApJ, 753, 170 Weiss, L. M., et al. 2013, ApJ, 768, 14 Winn, J. N. 2011, in Exoplanets, ed. S. Seager (University of Arizona Press), 55 Winn, J. N., et al. 2011, ApJ, 737, L18 Wu, Y., & Lithwick, Y. 2013, ApJ, 772, 74 This manuscript was prepared with the AAS LATEX macros v5.2.
1102.3872	2	1102	2011-02-22T14:35:41	Physical studies of Centaurs and Trans-Neptunian Objects with the Atacama Large Millimeter Array	[ "astro-ph.EP" ]	Once completed, the Atacama Large Millimeter Array (ALMA) will be the most powerful (sub)millimeter interferometer in terms of sensitivity, spatial resolution and imaging. This paper presents the capabilities of ALMA applied to the observation of Centaurs and Trans-Neptunian Objects, and their possible output in terms of physical properties. Realistic simulations were performed to explore the performances of the different frequency bands and array configurations, and several projects are detailed along with their feasibility, their limitations and their possible targets. Determination of diameters and albedos via the radiometric method appears to be possible on ~500 objects, while sampling of the thermal lightcurve to derive the bodies' ellipticity could be performed at least 30 bodies that display a significant optical lightcurve. On a limited number of objects, the spatial resolution allows for direct measurement of the size or even surface mapping with a resolution down to 13 milliarcseconds. Finally, ALMA could separate members of multiple systems with a separation power comparable to that of the HST. The overall performance of ALMA will make it an invaluable instrument to explore the outer solar system, complementary to space-based telescopes and spacecrafts.	astro-ph.EP	astro-ph	, , , Physical studies of Centaurs and Trans-Neptunian Objects with the Atacama Large Millimeter Array Arielle Moulleta, Emmanuel Lellouchb, Raphael Morenob, Mark Gurwella aHarvard-Smithsonian Center for Astrophysics, Cambridge MA-02138 (U.S.A.) bLESIA-Observatoire de Meudon Abstract Once completed, the Atacama Large Millimeter Array (ALMA) will be the most power- ful (sub)millimeter interferometer in terms of sensitivity, spatial resolution and imaging. This paper presents the capabilities of ALMA applied to the observation of Centaurs and Trans-Neptunian Objects, and their possible output in terms of physical proper- ties. Realistic simulations were performed to explore the performances of the different frequency bands and array configurations, and several projects are detailed along with their feasibility, their limitations and their possible targets. Determination of diameters and albedos via the radiometric method appears to be possible on ∼500 objects, while sampling of the thermal lightcurve to derive the bodies' ellipticity could be performed at least 30 bodies that display a significant optical lightcurve. On a limited number of objects, the spatial resolution allows for direct measurement of the size or even surface mapping with a resolution down to 13 milliarcseconds. Finally, ALMA could separate members of multiple systems with a separation power comparable to that of the HST. The overall performance of ALMA will make it an invaluable instrument to explore the outer solar system, complementary to space-based telescopes and spacecrafts. Keywords: Trans-neptunian objects, Centaurs, Instrumentation 1. Introduction Almost 1400 small bodies orbiting beyond Jupiter have been discovered so far, classi- fied as Centaurs (orbiting within Neptune's orbit) or as Trans-Neptunian Objects (TNOs, orbiting beyond Neptune). Due to their distance to the Sun and hence their low physical and chemical processing (McKinnon et al., 2008), their surfaces are expected to expose some of the most pristine material in the solar system. To relate today's outer solar sys- tem characteristics to those of the primordial disk, we need to better constrain physical properties and composition of these bodies, and understand the physical, chemical and dynamical processes that took place to shape this region. Obtaining thermal emission measurements on these bodies is an essential tool for this purpose, since they can give access to properties such as geometric albedo, size, shape and surface properties (e.g. thermal inertia, emissivity). Building a large database of albedos and sizes is necessary to identify possible correlations with spectral properties Preprint submitted to Icarus August 17, 2018 (infrared and visible colors) or dynamical parameters, that would help to understand the roles of processes such as space weathering and collisions (Doressoundiram et al., 2008). This would also allow refinement of the taxonomy of this population, and to compare it to other populations (comets, asteroids) so as to identify similarities and to trace population histories. Constraints on the size distribution power law can give clues on the planetesimal growth and fragmentation processes (Kenyon and Luu, 1999), and the identification of breaks in the size distribution is a powerful diagnostic of the intrinsic strength of these bodies (Pan and Sari, 2005). Accurate determination of the shape and size of individual bodies, along with mass determination, is important to determine their formation and collisional history, as well as their bulk density and their ability to retain an atmosphere and/or surface ices (Lacerda and Jewitt, 2007; Levi and Podolak, 2009; Schaller and Brown, 2007). Knowledge of surface albedos is also necessary to correctly interpret the ice bands that can be detected in near-infrared and visible spectra, and thus to accurately establish surface composition (Barucci et al., 2008). Measuring the variation of spectral emissivity with wavelength, by combining Herschel and ALMA data, gives access to thermophysical and composition properties of the surface and subsurface. Finally, precise determination of the brightness temperature itself is a key indicator of the physical temperature of the surface and subsurface, establishing the possible presence and abundance of a stable atmosphere sustained by ice sublimation. So far, only about 50 Centaurs and TNOs with diameters generally larger than 100 km have been detected at thermal wavelengths, mostly by space-based infrared telescopes ISO (Thomas et al., 2000) and especially Spitzer (Stansberry et al., 2008; Brucker et al., 2009). With disk-averaged surface temperatures below 130 K for Centaurs and 50 K for TNOs, their thermal emission peaks between 20-100 microns, where the Earth's at- mosphere is opaque. At longer wavelengths, the thermal emission is considerably lower. Only 8 bodies have been detected in the (sub)mm wavelength range (Altenhoff and Stumpff, 1995; Altenhoff et al., 2001; Margot et al., 2002; Lellouch et al., 2002; Altenhoff et al., 2004; Bertoldi et al., 2006; Gurwell et al., 2010), mostly around 250 GHz (1.2 mm) with the MAMBO bolometer on the IRAM-30m antenna. Interferometric facilities, with smaller or comparable total collective area, could only detect the brightest bod- ies (Pluto and Charon) so far (Gurwell et al., 2010), but the recent correlator upgrades on the IRAM-Plateau de Bure array now allows for point-source sensitivities better than MAMBO at 250 GHz. The number of thermal detections is expected to increase significantly in the coming years. At 70-160 microns, Herschel's sensitivity may allow to detect up to 140 Centaurs and TNOs, that are the targets of a large photometric program (Muller et al., 2009); first results have just been presented (Muller et al., 2010; Lim et al., 2010; Lellouch et al., 2010). In the (sub)mm wavelength range, starting in 2012-2013, the Atacama Large Millimeter Array (ALMA) will provide unprecedented sensitivity, allowing in principle the detection of a dramatically larger number of targets. This interferometric facility, under construction in Chile, will offer at completion 50 antennas of 12 m diameter each, along with 12 additional 8-m antennas and four 12-m antennas forming the Atacama Compact Array (ACA). In addition, the array will offer very extended configurations with baselines up to 14 km, that will provide spatial resolution down to 5 milliarcseconds (mas) at 850 GHz (350 microns), corresponding to ∼150 km at 40 AU, which as we will show is sufficient to resolve the largest Centaurs and TNOs. 2 This paper presents an analysis of the capabilities of ALMA for physical studies of Centaurs and TNOs, in terms of sensitivity, resolution and imaging performance. De- tailed simulations were performed, which take into account the array characteristics, atmospheric quality, expected receiver performance and correlator capabilities. Imaging capabilities were simulated using a realistic simulator, developed by the GILDAS team (Pety et al., 2001), to calculate the expected Fourier-plane coverage. Feasibility and ob- serving strategies for a number of detection and imaging projects are detailed, along with their expected products in terms of the bodies' properties and in general outer solar system science. A short study on ALMA capabilities for asteroid science can be found in Busch (2009). 2. ALMA technical characteristics and expected performance 2.1. Point-source sensitivity To characterize the expected noise for a given observation, the point source sensitivity is commonly used in radio astronomy. In the case of continuum emission measurements, this corresponds to the rms noise expected on flux measurements on an unresolved source obtained using the whole bandwidth of the instrument ∆ν in ∆t seconds of time. This quantity depends on the instrumental performances (antennas, receivers, correlator) cou- pled with the atmospheric qualities of the site (sky opacity and phase stability), following the classical formula expressed in flux density units (Thompson et al., 1986): ∆(S)(Jy) = KTsys ηatmηcorp∆t∆νnpN (N − 1) (1) where ηcor is the correlator efficiency, Tsys the system temperature characterizing the receiver and sky noise, np the number of polarizations and N the number of antennas. The K term describes the gain of the antennas in Jy/K, and is defined as 2kFef f with k Acolηa the Boltzmann constant, Fef f the forward efficiency, Acol the collecting area of a single antenna, and ηa the aperture efficiency (e.g. K = 40 Jy/K at 230 GHz at the IRAM- PdBI). Finally, ηatm is the phase decorrelation (equal to e−σ2/2, σ being the phase rms at the observing frequency), that measures the effective signal loss due to the atmospheric phase fluctuation. To calculate the point-source sensitivity expected from the ALMA array for each observ- ing frequency, we will use the following approximate equation from De Breuck (2005) for dual polarization observations (np=2), where the correlator and forward efficiencies are considered independent from frequency : ∆(S)(mJy) = 2.6 × 106Tsys ηaηatmN D2√∆t∆ν (2) We will consider the other parameters as following : - N=50, D=12 m : we thus consider only the main array at completion, excluding the contribution of the adjacent ACA (Atacama Compact Array). - ∆ν=8 GHz per polarization using the full correlator capacity. - ηatm=0.87, corresponding to a phase rms of 30◦ - ηa : this parameter depends mostly on the accuracy σa of the surface of the antennas. 3 λ )2 Following (Ruze, 1966), ηant = η0e−( 4πσa , where λ is the observed wavelength, and η0 is the efficiency of a perfectly smooth antenna. The latest measurements on the already available antennas show that σa is of the order of 20 µm although it could rise to 25 µm in bad conditions (cold weather) (Wooten, private comm). The value of η0 is assumed to be 0.8 following the antenna requirements as defined in Butler and Wootten (1999). - system temperature (Tsys, in K) : this parameter includes the combined effects of the thermal noise from the receivers and the opacity of the atmosphere. The val- ues assumed here are the estimates by Moreno and Guilloteau (2002), calculated for a source at 50◦ elevation, and assuming a water vapor content varying with the observ- ing frequency (0.5 mm for frequencies above 370 GHz and either 2.3 mm or 1.2 mm below), which is realistic since high-frequency observations require better sky condi- tions. The Tsys were lowered for bands 8 and 9 by respectively 33% and 40%, to match the latest sensitivity expectations as specified by the ALMA sensitivity simulator (http://www.eso.org/sci/facilities/alma/observing/tools/etc/). In addition, for band 6, the estimates have been updated using the most recent performance measurements on the receivers (ALMA Newletter, September 2010, www.almaobservatory.org). The values of antenna efficiency, system temperature and point-source sensitivity, de- rived from Equation 2, are gathered in Table 1 for a set of characteristic frequencies. For comparison, the best continuum point-source sensitivities reached by available in- struments are shown. ALMA should provide very significant gains in sensitivity (factors 10-100). However given the minimal characterization of the system so far, the ALMA sensitivity estimates may change as new performance measurements will become avail- able. It is also probable that in average, 10% of the antennas may not be used during standard observations, due to maintainance and/or technical failures. This would de- grade the sensitivity estimates by 10%. Finally we note that the point-source sensitivity should not strongly depend on the array configuration. Indeed, although the atmospheric decorrelation typically increases as the distances between antennas increase (i.e. with baseline length), the ALMA antennas will be equipped with water vapor radiometers at 183 GHz, that will monitor variations of the atmospheric optical depth along the line-of-sight. Application of the derived phase corrections should help to compensate the atmospheric decorrelation. Keeping these lim- itations in mind, we will assume the continuum point-source sensitivities presented in Table 1 in the rest of the paper. 2.2. Array configurations and imaging performance A significant advantage of interferometers when compared to single-dish instruments is their ability to boost spatial resolution. While single-dishes are diffraction-limited by the size of the antenna, interferometers are diffraction-limited by the length of their longest baseline, i.e. the longest distance between two antennas. The resolution unit of an interferometer is the synthesized beam, which is the Fourier transform of the spatial distribution of the antennas, and whose size is characterized by the half power beam width (HPBW). The extent of the baselines is however limited by terrain, atmospheric phase decorrelation and lack of phase correction. Baselines up to ∼1 km only are offered on the present interferometers. The ALMA interferometer has been designed to offer flexible configurations of its 50 antennas, with baselines up to 14 km. This is possible thanks to the good atmospheric 4 Frequency Wavelength ηa (GHz) (mm) 110 (band 3) 230 (band 6) 345 (band 7) 460 (band 8) 675 (band 9) 850 (band 10) 2.7 1.3 0.87 0.65 0.44 0.35 0.79 0.77 0.74 0.69 0.58 0.48 (mJy/hour) Tsys ALMA Sensitivity (K) 81 118 222 460 636 1200 0.008 0.012 0.023 0.051 0.085 0.19 Best present sensitivity (mJy/hour) 0.15 (Plateau de Bure-IRAM) 0.31 (Plateau de Bure-IRAM) 2.8 (LABOCA-APEX) 1.6 (SCUBA2-JCMT, expected) 41 (SABOCA-APEX) Table 1: Estimated aperture efficiencies, system temperature and continuum point source sensitivities for ALMA, for 1 hour observation on-source with the whole 16 GHz equivalent bandwidth, derived from Equation 2. The best sensitivity achieved by present facilities are indicated for comparison. stability on the site and the phase correction allowed by the 183 GHz water radiome- ters. The expected resolution for different bands as a function of observing frequency is shown in Figure 1, along with the resolution available from other facilities. An angular resolution down to 5 mas is expected for the highest frequencies, and down to 20 mas at 230 GHz, which is better by a factor 14 than what the Combined Array for Research in Millimeter Astronomy (CARMA) can reach. However, high spatial resolution is not sufficient to ensure good quality imaging. In- deed, an interferometer measures samples of the Fourier transform of the sky brightness distribution, the visibilities. Each visibility corresponds to the correlation of the signal coming from a pair of antennas at a given time. The coordinates (u, v) of the visibilities in the Fourier-plane (or spatial frequencies plane) depend on the source local coordinates on the sky, the baseline length and its orientation. To retrieve the original brightness distribution, the obtained set of visibilities must undergo an inverse Fourier transform. This process is all the more reliable when the coverage of the Fourier-plane is as complete as possible which implies baselines of different lengths and orientations. Thanks to its high number of baselines, low latitude (-23◦S) and configuration design, the ALMA array will offer, in a very short time, very satisfying Fourier-plane filling, sufficient to ensure a good distribution of visibility data in all directions and spatial frequencies, out to a maximum distance. Figure 2 gives an example of Fourier-plane coverage obtained in just 15 minutes, and the corresponding synthesized beam. The array thus provides almost instantaneous good imaging quality, enabling time-resolved imaging on thermally bright and large enough sources. 3. ALMA performance for Centaurs and TNOs observations 3.1. Adopted thermal emission models To assess the performance of ALMA for the observation of outer solar system small bodies, a model of their thermal emission is required. However to fully describe it, one needs information on many geometric and physical properties (pole orientation, thermal inertia, albedo...) that are still unknown in most cases. The best approximate approach 5 110 GHz 230 GHz 345 GHz 460 GHz 675 GHz 850 GHz 101 101 101 100 100 100 10-1 10-1 10-1 10-2 10-2 10-2 ) ) ) s s s d d d n n n o o o c c c e e e s s s c c c r r r a a a ( ( ( m m m a a a e e e b b b d d d e e e z z z s s s e e e h h h i i i t t t n n n y y y S S S 100 100 100 200 200 200 500 500 500 1000 1000 1000 2000 2000 2000 5000 5000 5000 10000 10000 10000 Longest Baseline (m) Longest Baseline (m) Longest Baseline (m) Figure 1: Spatial resolution offered by ALMA, defined by the synthesized beam half power width, as a function of the longest baseline of the configuration and the operating frequency. Best spatial resolutions offered by other facilities (CARMA at 110 and 230 GHz, SMA at 345, 460 and 675 GHz) are represented with thick black lines. 6 Figure 2: Left : Fourier-plane coverage obtained in 2 minutes on a -10◦ declination transiting source, with a simulated extended configuration of ALMA. Each point corresponds to a visibility, i.e. the cross correlation of the signals from a pair of antennas, whose coordinates in the Fourier plane represent the length and orientation of the corresponding projected baseline, in meters. Right : corresponding synthesized beam, of ∼ 50 mas HPBW. Axes are expressed in arcseconds. is to use empirical thermal models designed to reasonably reproduce the thermal emis- sion from Centaurs and TNOs, and, for some parameters, to assume typical values. The models and assumptions used are described below. In the grey-body emission model, for a single spherical body of apparent radius R (in radians), the thermal flux density emitted at a frequency ν from a body corresponds to the Planck law : 2hν3 F (ν) = ǫ (3) c2 πR2Z sin(φ)dφdθ kT (φ,θ) − 1 e hν where h is Planck's constant, k is Boltzmann's constant, φ and θ are respectively the local latitude and longitude on the body, and T (φ, θ) is the surface temperature distribution. The spectral emissivity ǫ characterizes the departure from black-body emission, which varies with wavelength over the (sub)mm range. In particular, ǫ takes into account that the thermal emission is not entirely reemitted towards space, depending on the dielectric constant of the surface. It also reflects the fact that a surface is not completely opaque at thermal wavelengths (i.e. the absorption coefficient by the surface is finite), which means that the outgoing emission is the sum of contributions from different depths below the surface. To phenomenologically represent these aspects, we use an emissivity factor varying between 0.8 to 1, consistent with the emissivities found by Moreno (2007) on Callisto and Ganymede. The emissivity is assumed to be independent of the position on disk, i.e. we do not consider a Fresnel-like description of emissivity but rather a Lambertian surface model. The temperature distribution on a body depends on the local solar illumination and 7 r2 on the surface response to it. It can be derived from the equations of heat conduction if all geometric (solar distance, latitude of sub-earth point, rotation rate) and physical parameters (albedo, bolometric emissivity, thermal inertia) are known. This is however generally not the case. Instead we have considered three different empirical thermal models, adapted from Spencer et al. (1989): - a "hot" model, the Equilibrium Model (EQM). This describes a body with zero thermal inertia (or a very slow rotator). In this model, the temperature varies only with the insolation angle (solar zenith angle, SZA) as T = Tsscos0.25(SZA), where Tss is the temperature at the sub solar point, defined as Tss = ( (1−Ab)F hǫbσ )1/4, where F is the solar constant at 1 AU (in W.m−2), rh is the heliocentric distance, Ab is the bolometric albedo, and ǫb is the bolometric emissivity, that we will assume equal to 0.9. The bolometric albedo is the product of the phase integral q, assumed equal to the standard value for asteroids (0.4), with the geometric albedo pv, for which, if unknown, we will assume the average value of 0.08 from Stansberry et al. (2008). This assumption is valid for the majority of intermediate and small-sized bodies, while very large bodies tend to present higher albedos, up to 0.9 (e.g. Eris, Brown et al. (2006)). - a "cold" model, the Isothermal Latitude Model (ILM), equivalent to a body with infinite thermal inertia (or a rapid rotator). In that case, temperature varies only with latitude, as T = Tss - an intermediate model, the Standard Thermal Model (STM), whose distribution on the disk follows that of the EQM, but with a corrective factor η−1/4. The η parameter allows in an empirical way to situate the thermal model of a given body in between the hot and cold models. Observationally, η can be constrained from measurements at different wavelengths. In this paper, we will use as reference model, the "TNO-tuned" STM model, i.e. with η equal to 1.25, the average value determined from Spitzer data (Stansberry et al., 2008), and a spectral emissivity ǫ=0.9. The other two models will be used to estimate the range of plausible temperature distributions. π0.25 cos0.25(φ), with Tss as defined above. 3.2. Detection thresholds A number of projects that can be performed with ALMA are based on the measure- ment of the total thermal flux to derive a number of parameters. Detection of the source at 5 σ is the main requirement to perform those projects, and is also a pre-requisite condition to perform all other projects described in the following sections. We want to establish how many Centaurs and TNOs are potentially detectable by ALMA, assuming that their thermal emission follows the adopted reference thermal model. We will con- sider as detectable a body whose thermal emission corresponds to at least 5 times the rms noise (5 σ detection). Figure 3 displays the detection limit at 6 different frequencies, for a 1-hour on-source integration. This integration time is reasonable for large surveys, but longer integration times are possible for specific sources. Figure 3 shows that in the most sensitive band (band 7, around 345 GHz), one could detect bodies with equivalent diameters larger than 75 km at 20 AU, 130 km at 30 AU, 190 km at 40 AU and 250 km at 50 AU. Bands 6 and 9 (respectively 230 GHz and 675 GHz) are slightly less efficient than band 7, and other bands are significantly less sensitive : at 40 AU, only bodies larger than 250 km can be detected with band 10 (850 GHz). To illustrate this, 1360 known objects are plotted on Figure 3. For most of them, 8 ) m k ( n o i t c e t e d r o f r e t e m a d l a m n M i i i 1000 800 600 400 200 0 0 110 GHz 230 GHz 345 GHz 460 GHz 675 GHz 850 GHz 40 mas 5 mas 10 20 30 40 50 60 70 Sun distance (AU) Figure 3: 5 σ detection diameter threshold (in km) for detection with ALMA, as a function of frequency and heliocentric distance. Black points represent the Centaurs and TNOs known as of February 2010 (from Minor Planet Center), along with their measured diameters (when available), or estimated di- ameters assuming a geometrical albedo of 0.08. Light black lines represent the best spatial resolution available at 110 and 850 GHz (respectively 40 and 5 mas). the radius is unknown, and was estimated from their visual magnitudes (taken from http://vo.imcce.fr/webservices/miriade), assuming a geometric albedo of 0.08. Figure 3 shows that more than 500 Centaurs and TNOs could be detected with band 7, 460 with band 6, 430 with band 9 and 330 with band 8 (460 GHz) in 1-hour on-source observation per object. The sensitivity performance of ALMA clearly increases the potential number of detectable targets, considering that only ∼50 bodies were detected with Spitzer, that Herschel is currently targeting 140 detectable objects, and that only 8 bodies were so far detected at (sub)mm wavelengths. As mentioned above, those estimates should not depend on the array configuration, unless the bodies are spatially resolved. In those cases in which the source is spatially resolved, the signal per beam (lower than the total emitted signal) determines the signal- to-noise of a given observation, implying requirements higher than 5 σ for the thermal detection . 9 3.3. Spatially resolved measurements As shown in Figure 3, the high angular resolution offered by ALMA at the highest frequencies and most extended configurations (5 mas) allows for the majority of the de- tectable objects to be spatially resolved, meaning that the synthesized HPBW is smaller or comparable to the object size. When a source is resolved, in addition to total flux measurements, two other methods to analyze the data could be applied: visibility anal- ysis and imaging. The first consists in studying the shape of the visibilities curve in the Fourier plane, where the resolution is apparent when the visibility data vary as a function of projected baseline length (see Figure 4), while unresolved sources result in flat visibility functions. The shape of the visibility curve, and in particular the position of the first null, characterizes the brightness temperature distribution and size, and can be compared to synthetic models (e.g. in Moullet et al. (2008)). This analysis requires that the HPBW is at least similar to the size of the observed structure. Imaging the data requires a HPBW smaller than the source size, and a satisfying Fourier- plane coverage. On the obtained images, the size, structure and position of the source temperature distribution can be directly observed and measured, with a precision limited by both the synthesized beam and the deconvolution errors. For both visibility analysis and imaging, a SNR per beam of 5 is a minimum, and higher requirements may be put depending on the desired measurement. The spatially resolved projects presented in the following sections require for example a SNR per beam from 6 to 80. Since the signal per beam decreases as the synthesized beam gets smaller with respect to the source, for a given target, the requirements on the rms to be reached are higher than for unresolved observations, implying longer on-source integration time. For example, if the synthesized beam is half the size of the source, one needs an integration time at least 16 times longer for an imaging project than for a spatially unresolved ther- mal detection at 5 σ. A compromise between spatial resolution and signal to noise must then be found for each scientific case. 4. Equivalent size and albedo Based on the thermal flux estimates and sensitivities presented above, we now describe two different methods to determine the equivalent size and albedo of an object from ALMA measurements. We explain hereafter the rationale behind these methods, their assumptions and limitations, and perform detailed simulations to assess the optimum strategies. 4.1. Radiometric determination As is well known since the IRAS surveys of asteroids (Lebofsky et al., 1986) and first illustrated by Jewitt et al. (2001) in the case of TNOs, the combination of optical and thermal flux measurements (radiometric method) permits in principle to separately determine the albedo and equivalent diameter of airless bodies , i.e. the diameter of a spherical body with the same projected area as the observed object, along with the average brightness temperature. However, the inferred parameters depend on the details of the temperature distribution models and the uncertainties can be prohibitively large (factors of several or even more) if the wavelength of the thermal measurement is shorter than the peak of the Planck function (Stansberry et al., 2008). These large uncertainties 10 Figure 4: Simulated visibility amplitude curve as a function of projected baseline (dots), for a body of the size (D=1207 km) and distance of Charon, observed during one hour in band 7 (345 GHz) with an extended array configuration. The first null is not reached, but the Bessel curve is apparent, as the object is partially resolved. To illustrate the quality of the observations, the expected thermal noise (from Table 1) was applied to the visibilities. The bold line represents the amplitude curve of the model without added noise. 11 can be drastically decreased if multi-wavelength thermal infrared measurements are used, constraining in effect the applicable thermal model (e.g., in an empirical way, through the beaming factor η as defined in Section 3.1), then providing the best accuracy on the temperature and diameter retrieval. Lying in the (quasi)-linear part of the Planck curve, the interpretation of single-band (sub)mm-wave measurements is not as sensitive to the precise temperature determination as single-band infrared measurements, so that the inferred diameters and albedos are not so much affected by the uncertainty on the thermal model. On the other hand, (sub)mm-wave multi-band measurements will not help to constrain an appropriate thermal model, but indicate the spectral emissivity. Here we investigate the precision that can be expected on diameter/albedo determina- tion based on ALMA single-band radiometric thermal flux measurements. Uncertainties result from measurement error and model uncertainty. Measurement errors are due to the combination of : - the measurement rms noise (±20% for a 5 σ detection, ±10% for a 10 σ detection): this is the uncertainty on the flux emitted by the object, that can be retrieved directly from the visibility data or from the brightness distribution map. Since the choice of the configuration does not have an influence on the sensitivity (when the target is not spatially resolved), the measurement of the total flux can be performed using the most compact configurations, for which phase stability requirements are less demanding. - the absolute flux scale error : this could be as low as ±5% using the calibrations schemes proposed by Butler and Wootten (1999) or Moreno and Guilloteau (2002). The influence of model uncertainty can be assessed by considering the ranges of fluxes predicted for a given diameter and albedo by the three models described in Section 3.1, spanning a wide range of temperature distributions, and corresponding to surface ther- mal inertiae from zero to infinite. The effect of the uncertain spectral emissivity (between 0.8-1) is also included. With this conservative approach, we obtain flux variations from model to model of about ±25%, equivalent to ±12 % change in diameter and to a ±25 % change in the albedo as determined from the visible magnitude. We here neglect any un- certainty on the visible magnitude, as the objects in question (i.e. detectable by ALMA) have visual magnitudes typically brighter than 21.5, permitting absolute photometry at the percent level (Doressoundiram et al., 2005). To reach this goal, supporting optical observations should be performed on several targets. Combining measurement error and model uncertainty, we find that for the bodies that can be detected by ALMA at 5 σ or more in the most sensitive bands (6, 7 and 9, i.e. respectively 230, 345 and 675 GHz), the uncertainties on the retrieved diameter are between 15-25%, and are primarily dominated by the model uncertainty (∼14%). The errors reported by recent Herschel-PACS single-band observations (Muller et al., 2010) are of the same order. The total relative error on the diameter obtained after 1 hour integration for bodies at 40 AU is shown in Figure 5, as a function of diameter. For specific targets, the integration time could be increased to reduce the flux measurements rms, but the result will still be limited by the model uncertainty. The main drawback of the radiometric method then lies in the choice of the thermal model. One way to reduce this uncertainty could be the construction of an ad hoc ther- mal model for each body. The optimal way to do that is to combine ALMA data with Spitzer photometric data at 24 and 70 microns, and possibly Herschel-SPIRE data. If Spitzer data is not available, Herschel-SPIRE and PACS photometry, that will be ob- 12 100 90 80 70 60 50 40 30 20 ) % ( n o i t a m i i t s e r e t e m a d f o r o r r e l a t o T 10 100 110 GHz 230 GHz 345 GHz 460 GHz 675 GHz 850 GHz 200 300 400 500 600 Object diameter (km) Figure 5: Error on the diameter retrieval on targets at 40 AU, as a function of frequency and body diameter, for 1-hour on source. Includes instrumental error and model uncertainty. tained on ∼20 objects, can also be used together to constrain the thermal model. The first object to be detected with both instruments, Makemake (Lim et al., 2010), revealed that a 2-terrain model is required to fit the thermal fluxes at the observed wavelengths. An alternate approach will be to infer from the combined Spitzer and Herschel data, in a statistical sense, the general thermal behavior of the population (e.g. mean value of η and variation trends as a function of orbital parameters) and to use the results for the interpretation of the ALMA data. Finally, for a few objects whose size has been indepen- dently derived from a non-radiometric method (imaging, occultation as in Elliot et al. (2010)), high-quality thermal measurements can in turn allow to constrain their thermal model and spectral emissivity. 4.2. Visibility data analysis As mentioned in Section 3.3, the position of the first null of the visibility data curve depends mostly on the size of the emitting region, thus the visibility curve can be an- alyzed to retrieve this parameter. One way to perform this is to fit the visibility data using a disk model : the fitted variable is the equivalent diameter of the source, and the retrieved error bar corresponds to the fit error. Using this direct size measurement, it is then possible to derive the geometric albedo pv using the visible magnitude. This technique has already been applied on asteroids Gaspra and Barbara with mid-infrared interferometry at the VLTI with a single baseline by Delbo et al. (2009). To determine which frequency/configuration combination is the most efficient, simu- lations of observations were performed with the GILDAS ALMA simulator (Pety et al., 2001). For an input source model, observing frequency and array configuration for 50 13 antennas, this tool calculates the expected visibility data with the appropriate Fourier- plane coverage and noise level. Here all simulations assumed a spherical target at -10◦ declination, with a uniform temperature distribution and no emissivity variations (Lam- bertian surface model). Fitting of the simulated visibility data with a plain disk model was performed through a χ2 fit automated in GILDAS. The error bars were estimated on the one hand from the GILDAS routine and on the other hand from the difference between the known (input) diameters and those retrieved from the simulator. A series of simulations shows that the choice of an appropriate configuration and fre- quency clearly depends on the target size and brightness temperature, as the quality of the fit is driven by both SNR and spatial resolution. The best way to determine the observing strategy for a given project is then to perform such simulations for each target. We could notice that a minimal SNR per beam of ∼20 is necessary to perform the fit. In addition, the best results on diameter determination are constantly obtained when the synthesized beam size resolution is comparable to the size of the target (between 0.6-1.2 times the size of the target), corresponding to an important concentration of visibilities in a region of the Fourier plane that is particularly sensitive to the body's size. For example at 45 AU, for targets larger than 26 mas, it appears that the best results are achieved with band 7 (345 GHz) with very extended configurations (baselines > 6000 m). Good results are also obtained with bands 9 and 10 (675 and 850 GHz), for which baselines of 3000 m can be sufficient. Those however require significantly better weather conditions. For targets with lower sizes, the highest frequencies (bands 9 and 10) provide the best results, combined with array configurations of ∼3000 to 6000 m extent. We note that, although observations at the highest frequencies with the most extended configurations permit the highest spatial resolution, they do not necessarily provide the best results, either because they resolve the source too much or because they do not provide the best SNR per beam. By observing with an appropriate frequency and configuration, this technique allows the retrieval of sizes with an accuracy better than or comparable to that achieved from the total flux radiometric technique for a number of targets, whose apparent size can be as low as 15 mas. The accuracies found on the retrieved diameters are of the order of 1.5% for large (1100-1500 km) and thermally bright bodies such as Quaoar, Makemake, Haumea and 2002TC302, when observed with band 7. Errors lower than 7% should be achieved on 2002AW197, Orcus, Varuna, Chariklo, 2004GV9, Huya and 2003VS2 with observations in bands 7, 9, or 10, and uncertainties lower than 15% are expected for medium-sized (700-200 km class) or very cold bodies, after at least 2 hours of integration in band 10. An improved determination of the equivalent diameters, and hence albedos, is then possible on at least 30 bodies, whose diameters are presently known with a 10-30% uncertainty. Compared to the radiometric technique presented in Section 4.1, results are not as much affected by the uncertainty on the thermal model, so that the diameters retrieved can in turn be used to interpret the simultaneously obtained thermal flux in terms of thermal model and emissivity. However the size of the thermal emission is not necessarily identical to the size of the solid body, since the brightness temperature distri- bution is not uniform over the disk. The difference is in fact again model-dependent, and is estimated to account for an additional 4% model uncertainty that must be added to the fit error. This difference is mostly important at the highest frequencies, since lower frequencies are not as sensitive to the surface temperature distribution. For bodies whose 14 sizes are already known to very good accuracy (e.g. Charon from stellar occultations, see Sicardy et al. (2006)), the measurement of the characteristic thermal emission size can be an excellent diagnostic of brightness temperature distribution, especially diur- nal, latitudinal and with emission angle, that can be in turn tied to properties such as thermal inertia and surface dielectric constant. Finally, this technique cannot be applied directly to multiple systems, for which the separation between members must be known to analyze the visibility curve. This parameter can be estimated by mapping the system, as presented in Section 7. 5. Shapes Both methods presented in Section 4 assume spherical shapes and hence can only determine the equivalent diameter of a body. Although this can be a good approxima- tion for the largest, slowly rotating bodies, many objects are suspected to have elongated or irregular shapes. Since bodies larger than ∼50 km are thought to be controlled by hydrostatic equilibrium, their shape is expected to be close to a Jacobi ellipsoid, and de- termination of their axes ratio can give clues on the material strength and bulk density (Farinella, 1987; Lacerda and Jewitt, 2007). For smaller bodies, irregular shapes can be the result of collisions. Shapes are in general inferred from the observed amplitude and period of optical rota- tional lightcurves (Sheppard et al., 2008). Lightcurves result from the variation of the apparent projected area with rotational phase, giving access to the projected axes ratio, corresponding to a lower limit on the ratio of the two axes perpendicular to the rotation axis of the body. The third (polar) axis is determined from hydrostatic equilibrium. However, albedo markings may affect the shape of lightcurves (see Lacerda et al. (2008) for Haumea) or entirely dominate it (Pluto, Buie et al. (1997)). One way to disentangle between shape and albedo effects in visible lightcurves is to observe thermal lightcurves, which depend on disk-averaged albedo variations in an anti-correlated and, except for very reflective objects, less sensitive manner than optical lightcurves (see e.g. the absence of detected lightcurve amplitude at 1.2 mm on Pluto by (Lellouch et al., 2000b)), since the surface temperature varies as (1 − Ab)0.25, and Ab is often small (≪1). With the sensitivity of ALMA it is possible to sample thermal rotational lightcurves on a large number of Centaurs and TNOs, whose rotation periods are on average ∼8 hours (Sheppard et al., 2008), though not with a time resolution as high as in the optical. As for optical lightcurves, the amplitude of the thermal lightcurve is mostly related to the projected axes ratio. Note however that for a given shape, the amplitude of the thermal lightcurve is usually larger than that of the optical lightcurve. This stems from the fact that the distribution of solar zenith angles and hence temperatures over the visible disk are much "flatter" at lightcurve maximum than at the minimum. A recent example is given by the Herschel-PACS observations of Haumea (Lellouch et al., 2010). The effect is however subdued at the longest wavelengths and we ignore it in what follows. For all targets that can be detected at 5 σ with band 7 (345 GHz), variations of flux larger than 40% can be assessed, corresponding to a projected axes ratio a/b > 1.4. For more than 200 large bodies, detectable at a 10 σ level in less than 1 hour, projected axes ratio larger than 1.2 could be identified. Comparing these results with the list of bodies with a 15 known rotational lightcurve in the optical/near-infrared (Sheppard et al., 2008), we can see that significant variations on thermal lightcurves could be observed with ALMA on at least 30 bodies, then allowing to put tighter constraints on the presence of albedo markings and on their projected axes ratio. The lightcurve method is however limited to bodies presenting a favorable geometric configuration, and in particular cannot detect ellipticity for pole-on orientation or for bodies with long (or very short) rotation periods. The alternative method to determine shapes is to identify ellipticity directly on high spatial resolution interferometric data recorded at one given rotational phase, through visibility data fitting with an elliptic model. The inferred parameters in this case are the semi-major and semi-minor axes in the sky plane and the position angle of the ellipse at the relevant phase, along with their error bars. Typically, fitting errors are approx- imately a factor √2 larger than the fit errors for a simple disk model fit (see Section 4.2). Considering that the elliptical shape can be established if the retrieved axes differ within the error bars, we find that projected axes ratios down to 1.04 for the 6 largest bodies (in two hours from band 7 observations, i.e. 345 GHz), and down to 1.2 for ∼7 additional bodies (from observations in bands 10, 9, or 7, i.e. 850, 675, or 345 GHz), can be detected. The method is therefore applicable to fewer objects than the thermal lightcurve approach. However it appears complementary as the measured axes ratios are projected in a different plane. For example, for an equator-on configuration, the ther- mal lightcurve method determines the two equatorial axes, while the "direct imaging" method determines the polar axis and a combination of the two equatorial axes that de- pends on the rotational phase. In addition, this is the only applicable method to measure non-sphericity for large objects without any measured lightcurve, e.g. Makemake, Orcus and 2002 AW197, whose potential ellipticity may not have been detected because of an unfavorable geometry (pole-on inclination of the rotation axis). 6. Surface mapping From infrared and visible spectroscopy, the surfaces of Centaurs and TNOs have revealed a variety of surface materials. Some bodies display spectral bands characteris- tic of ices (mostly H2O and CH4), while others are featureless, with broadband colors suggesting carbon and organic rich materials (Barucci et al., 2008). A handful of these objects have also revealed rotational variations of their spectra, suggesting an heteroge- neous surface. This is in particular the case of Pluto, for which, based on the variability and monitoring of the visible and near-infrared spectrum, Grundy and Fink (1996) and Grundy and Buie (2001) proposed a possible distribution of tholins, water and methane ices, in general agreement with direct HST imaging (Stern et al., 1995; Buie et al., 2010b) and optical lightcurves (Buie et al., 1997, 2010a). Possible causes behind these surface variations include the resurfacing of fresh and bright ice due to collisions, cryo-volcanism and atmospheric condensation/sublimation cycles, while surfaces exposed for a longer time get redder and darker due to space weathering and irradiation. So far such varia- tions could not be assessed on other bodies, due to the lack of spatial resolution and the paucity of rotationally resolved spectra. Brightness temperature maps of Centaurs or TNOs have to date never been obtained, and could help to distinguish surface features. Although the thermal emission of those 16 bodies is typically only weakly dependent on their albedos, this is not true for the most reflective of them. For those bodies, maps of the thermal emission should permit to map surface reflectivity. In addition, even if small-scale variations of the brightness tem- perature cannot be distinguished (e.g. generally due to too dark a surface), the overall limb-to-limb brightness temperature distribution (e.g. diurnal and equator-to-pole) will be diagnostic of the thermal inertia. In the case of Pluto, thermal models predict, based on ISO observations, very high variations of the surface temperature, up to 50%, pro- duced by the combination of high albedo variations and response to insolation taking into account the thermal inertia (Lellouch et al., 2000a). We will present here the case of maps where the surface is resolved in at least two points in each direction. Thermal flux variations of 10%, corresponding to brightness temperature variations of 5-10%, could be detected with a SNR per beam of 30. With this SNR requirement, mapping is possible on 6 large objects with a 4-hour long observation in bands 7, 9 and 10 (345, 675 and 850 GHz), with a spatial resolution rom 17 to 14 mas. To illustrate the effect of Pluto's temperature variations in ALMA observations, we used the thermal map of Lellouch et al. (2000a), calculated from the surface model of Grundy and Fink (1996) at an orbital longitude of 93◦W. Figure 6 shows synthesized maps obtained using band 7 and band 10 with a very extended configuration. Such observations would provide the first direct thermal mapping of Pluto. Search for spatial albedo variations based on brightness temperature mapping is difficult because they produce relatively small variations of the thermal emission. Indeed, assum- ing the average phase integral value of 0.4, even large albedo variations (from pv=0.7 to pv=1) can only produce 5% variations on the brightness temperature, or, assuming a phase integral of 0.8 (corresponding to the value estimated for Pluto by Lellouch et al. (2000a)), 20% at most. Such brightness temperature variations would result at most in variations of 5-40% of the thermal emission, depending on the observing frequency. We consider here that only variations higher than 3 σ on the thermal emission can be detected, and we assume that the albedo is the only property of the surface that could be varying. If q=0.4, a SNR as a high as 80 is necessary to retrieve albedos variations greater than 15% (band 10) or 20% (band 7). Among the objects that can be mapped, such a SNR per beam can only be obtained, in four hours of observation, on a few very large bodies, i.e. Charon, Quaoar and Makemake. The spatial resolution of 28 mas (band 10), 22 mas (band 9) or 20 mas (band 7), is limited by the SNR per beam, and allows a 2x2 mapping only. Such observations in bands 9 and 10 should be performed with at least 3-4 km wide configurations, while band 7 observations must make use of with very extended (∼8 km) configurations, although with lower sky opacity requirements than higher frequencies. If q=0.8, a SNR of 30 can allow to retrieve albedo variations of 12% (band 10) to 15% (band 7). This SNR can be obtained on 6 large bodies, with a spatial resolution down to ∼ 15 mas. Although no fine albedo mapping with this method is possible, it could reveal a strongly inhomogeneous albedo distribution on Charon and Makemake, as was found on Pluto. We note that in July 2015, during the New Horizons encounter (Young et al., 2008), the LEISA near-infrared spectrometer, part of the Ralph instrument suite (Reuter et al., 2008), will use temperature-sensitive N2, CH4 and H2O bands to map surface tempera- tures, while the radiometric mode of the Radio experiment package (REX) will be used to obtain a direct measurement of Pluto's surface temperature. Although the New Hori- 17 Figure 6: Synthetic maps of a simulated observation of Pluto, assuming the thermal model from Lellouch et al. (2000a) based on the Grundy and Fink (1996) albedo map, for a 4-hour observation with a very extended (10 km) configuration. Left : observations at 345 GHz. Contours are set to 10 σ, the beam size is 24 mas. Right : observations at 850 GHz. Contours are set to 2 σ, the beam size is 10 mas. The synthesized beam is represented in the inset in the bottom-left corner. Axes are expressed in arcseconds. The color scale (Jy/beam) is indicated by the vertical bar on the right. zons observations will obviously afford much higher spatial resolution, the ALMA maps will be complementary, noting in particular that LEISA may not be able to determine temperature of the possible ice-free regions. 7. Multiple system imaging Large imaging surveys, mostly with HST, have revealed that more than 50 TNOs and Centaurs are actually multiple systems, representing a rather high fraction of the whole population of more than 1400 objects (Noll et al., 2008). Most of those consist in nearly equally-sized binaries that suggest formation by dynamic capture, but the measured size ratios can go as low as 1:100. Studying these systems, while leading to insights into the processes of capture and fragmentation, also provides the only access to the system's total mass through mutual orbit determination. If the individual sizes (or at least the total equivalent size) of the system members are known, this leads to determination of the average density, bringing a strong constraint on interior composition and structure. Bulk densities could also be correlated to other properties such as size, albedo and he- liocentric distance. In spite of obvious observational biases, which favor systems with high separations, the frequency of binaries seems to increase as the separation decreases (Noll et al., 2008), warranting the use of the highest possible spatial resolution. At best, the point spread function of HST/STIS allows the detection of equally bright pairs separated by 30 mas, 18 Figure 7: Synthetic maps of a simulated observation of a binary system at 675 GHz with a 10 km wide configuration. Left : two bodies of 8 mas apparent size at 30 AU (174 km diameter), separated by 10 mas. Right : the same bodies with a separation of 12 mas. Contours are set to the rms reached after 2 hours on source (6×10−5 Jy/beam). The synthesized beam (12 mas) is represented in the inset at the bottom-left corner. Axes are expressed in arcseconds. The color scale (Jy/beam) is indicated by the vertical bar on the right. and to determine the separation with an error bar larger than 2 mas. Although one may not immediately think of (sub)mm-wave observations as the most obvious method to enhance spatial resolution, the high angular resolution of ALMA (see Figure 1) makes it competitive with the most performing space telescopes on this partic- ular topic. To estimate the separation power of ALMA for binary Centaurs and TNOs, simulations were performed for two-hour observations of different sets of equally-sized binary systems. We adopted the reference thermal model of Section 3.1 and an albedo of 0.08. We considered that a binary system is resolved when the emission map shows two emission peaks with significant separation (i.e. with a minimum emission level at least 3 σ lower than the peaks between them), as is illustrated in Figure 7. This conservative requirement could be relaxed if the image fidelity, i.e. the imaging quality, is good enough so that the errors due to image deconvolution are low. We observe that, provided that each member of the system is detected with a SNR ratio higher than 6, the separation power of the array is equal to its synthesized beam size. As an example, for a configuration with a maximum 10 km baseline, the separation power reaches 10 mas in band 10 (850 GHz), 12 mas in band 9 (675 GHz), 18 mas in band 8 (460 GHz), and 24 mas in band 7 (345 GHz). However this does not mean that choosing the highest frequency is always the best strategy. Indeed observations in band 7 are more sensitive than observations in band 10, giving access to smaller targets. In band 10 and for a 30 AU distance, in two hours on source, only (individual) targets with apparent sizes larger than 9.5 mas have a sufficient thermal flux, given our assumed thermal model, to be detected at 6 σ. Band 10 could then be possibly used only for a 19 selection of rather large targets whose multiplicity is either highly suspected or already assessed (e.g. 1997 CS29, 2003 QA91, 2001 QY297), so as to get a very good accuracy on their separation. For band 9 and band 7 observations, at 30 AU, targets larger than respectively 6 and 5.5 mas are suitable, and with this limitation, assuming that all known bodies are equally-sized and equally-bright binary systems, as many as 250 among them would be bright enough for each member to be detected and thus binarity to be deter- mined. Once the multiplicity of a system is assessed, it is possible to retrieve its projected sep- aration by measuring directly on the map the relative distance between the two peaks. Following astrometric error budgets from Lestrade (2008), we can estimate that with a minimal SNR of 6 on peak, the astrometric error on each peak position will be lower than 0.7 mas at 345 GHz, and 0.3 mas at 850 GHz, the latter being comparable to the astrometric precision of the FGS on HST. Including deconvolution effects, we estimate conservatively that the error on the separation can be lower than 1.5 mas, corresponding to ∼40 km at 40 AU. Given that the configuration of the system at the moment of detection is unknown, the obtained projected separation is only a lower limit to the actual system separation. Only follow-up observations can allow the retrieval of the separation and the orbiting period (Hestroffer et al., 2005), that are necessary to derive the total system mass. In addition, independent measurements of the flux density of each body can be obtained. If the system is resolved both at visible and thermal wavelengths, it is possible to retrieve each member's diameter using the radiometric method described in Section 4.1. If no visible magnitude for each member is available, an acceptable estimate of the individual sizes can be obtained assuming the albedo value, or using the whole system magnitude to derive an average albedo value. 8. Conclusions The calculations and simulations presented in this paper show that the capabilities of ALMA at completion in terms of sensitivity, spatial resolution and imaging will be well suited to study a large number of Centaurs and TNOs. Determination of equivalent diameters and albedos via the radiometric method may tar- get more than 500 objects using spatially unresolved continuum measurements in bands 6 or 7, allowing improved determination of the size distribution for bodies larger than 100 km, and helping to provide robust conclusions when correlations between physical, spectral and dynamical properties are examined. With the same bands, thermal rotational lightcurves could be sampled on ∼30 targets that present a significant optical lightcurve, allowing one to disentangle the effects of non-spherical shape and surface albedo variations. Using extended configurations, bands 7, 9 and 10 could provide essentially model-independent size and ellipticity measurements, using visibility data fitting, on ∼30 large bodies. The first thermal maps of the 6 largest objects could also be obtained with a spatial resolution down to ∼14 mas. Finally, ALMA will be the only facility that could separate multiple systems as close as 10 mas, and possibly contact binaries. Retrieval of the system orbit and period is key 20 for the determination of the system's mass. Although the first interferometric fringes with a 3-antenna array on site were obtained in November 2009, the completion of ALMA is a few years away. A first call for proposal on early science will be issued in spring 2011 for a partial array (∼16 antennas), that will be sufficient for the detection of the largest sources. Acknowledgments We thank the two anonymous referees for their thorough review of the paper. We also thank S. Guilloteau and A. Dutrey for their useful comments and tips, M. Holman for his proofreading and interesting discussions, J. Berthier from IMCCE for his help on ephemeris computation and the GILDAS team for their support. References Altenhoff, W.J., Bertoldi, F., Menten, K.M., 2004. Size estimates of some optically bright KBOs. A&A 415, 771 -- 775. Altenhoff, W.J., Menten, K.M., Bertoldi, F., 2001. Size determination of the Centaur Chariklo from millimeter-wavelength bolometer observations. A&A 366, L9 -- L12. Altenhoff, W.J., Stumpff, P., 1995. Size estimate of "asteroid" 2060 Chiron from 250 GHz measurements. A&A 293, L41 -- L42. Barucci, M.A., Brown, M.E., Emery, J.P., Merlin, F., 2008. Composition and Surface Properties of Transneptunian Objects and Centaurs, in "The Solar System Beyond Neptune", in: The Solar System Beyond Neptune, pp. 143 -- 160. Bertoldi, F., Altenhoff, W., Weiss, A., Menten, K.M., Thum, C., 2006. The trans-neptunian object UB313 is larger than Pluto. Nature 439, 563 -- 564. Brown, M.E., Schaller, E.L., Roe, H.G., Rabinowitz, D.L., Trujillo, C.A., 2006. Direct Measurement of the Size of 2003 UB313 from the Hubble Space Telescope. ApJ 643, L61 -- L63. Brucker, M.J., Grundy, W.M., Stansberry, J.A., Spencer, J.R., Sheppard, S.S., Chiang, E.I., Buie, M.W., 2009. High albedos of low inclination Classical Kuiper belt objects. Icarus 201, 284 -- 294. Buie, M.W., Grundy, W.M., Young, E.F., Young, L.A., Stern, S.A., 2010a. Pluto and Charon with the Hubble Space Telescope. I. Monitoring Global Change and Improved Surface Properties from Light Curves. AJ 139, 1117 -- 1127. Buie, M.W., Grundy, W.M., Young, E.F., Young, L.A., Stern, S.A., 2010b. Pluto and Charon with the Hubble Space Telescope. II. Resolving Changes on Pluto's Surface and a Map for Charon. AJ 139, 1128 -- 1143. Buie, M.W., Tholen, D.J., Wasserman, L.H., 1997. Separate Lightcurves of Pluto and Charon. Icarus 125, 233 -- 244. Busch, M.W., 2009. ALMA and asteroid science. Icarus 200, 347 -- 349. Butler, B., Wootten, A., 1999. ALMA Sensitivity, Supra-THz Windows and 20 km baselines, in: ALMA Memo 276. De Breuck, C., 2005. Scientific requirements of ALMA, and its capabilities for key-projects: extragalactic, in: A. Wilson (Ed.), ESA Special Publication, pp. 27 -- 32. Delbo, M., Ligori, S., Matter, A., Cellino, A., Berthier, J., 2009. First VLTI-MIDI Direct Determinations of Asteroid Sizes. ApJ 694, 1228 -- 1236. Doressoundiram, A., Boehnhardt, H., Tegler, S.C., Trujillo, C., 2008. Color Properties and Trends of the Transneptunian Objects, in: The Solar System Beyond Neptune, pp. 91 -- 104. Doressoundiram, A., Peixinho, N., Doucet, C., Mousis, O., Barucci, M.A., Petit, J.M., Veillet, C., 2005. The Meudon Multicolor Survey (2MS) of Centaurs and trans-neptunian objects: extended dataset and status on the correlations reported. Icarus 174, 90 -- 104. Elliot, J.L., Person, M.J., Zuluaga, C.A., Bosh, A.S., Adams, E.R., Brothers, T.C., Gulbis, A.A.S., Levine, S.E., Lockhart, M., Zangari, A.M., Babcock, B.A., Dupr´e, K., Pasachoff, J.M., Souza, S.P., Rosing, W., Secrest, N., Bright, L., Dunham, E.W., Sheppard, S.S., Kakkala, M., Tilleman, T., 21 Berger, B., Briggs, J.W., Jacobson, G., Valleli, P., Volz, B., Rapoport, S., Hart, R., Brucker, M., Michel, R., Mattingly, A., Zambrano-Marin, L., Meyer, A.W., Wolf, J., Ryan, E.V., Ryan, W.H., Morzinski, K., Grigsby, B., Brimacombe, J., Ragozzine, D., Montano, H.G., Gilmore, A., 2010. Size and albedo of Kuiper belt object 55636 from a stellar occultation. Nature 465, 897 -- 900. Farinella, P., 1987. Small Satellites, in: M. Fulchignoni & L. Kresak (Ed.), The Evolution of the Small Bodies of the Solar System, p. 276. Grundy, W.M., Buie, M.W., 2001. Distribution and Evolution of CH4, N2, and CO Ices on Pluto's Surface: 1995 to 1998. Icarus 153, 248 -- 263. Grundy, W.M., Fink, U., 1996. Synoptic CCD Spectrophotometry of Pluto Over the Past 15 Years. Icarus 124, 329 -- 343. Gurwell, M.A., Butler, B.J., Moullet, A., 2010. Subarcsecond Scale Imaging of the Pluto-Charon System at 1.1 and 1.4 mm, in: Bulletin of the American Astronomical Society, p. 1014. Hestroffer, D., Vachier, F., Balat, B., 2005. Orbit Determination of Binary Asteroids. Earth Moon and Planets 97, 245 -- 260. Jewitt, D., Aussel, H., Evans, A., 2001. The size and albedo of the Kuiper-belt object (20000) Varuna. Nature 411, 446 -- 447. Kenyon, S.J., Luu, J.X., 1999. Accretion in the Early Outer Solar System. ApJ 526, 465 -- 470. Lacerda, P., Jewitt, D., Peixinho, N., 2008. High-Precision Photometry of Extreme KBO 2003 EL61. AJ 135, 1749 -- 1756. Lacerda, P., Jewitt, D.C., 2007. Densities of Solar System Objects from Their Rotational Light Curves. AJ 133, 1393. Lebofsky, L.A., Sykes, M.V., Tedesco, E.F., Veeder, G.J., Matson, D.L., Brown, R.H., Gradie, J.C., Feierberg, M.A., Rudy, R.J., 1986. A refined 'standard' thermal model for asteroids based on obser- vations of 1 Ceres and 2 Pallas. Icarus 68, 239 -- 251. Lellouch, E., Kiss, C., Santos-Sanz, P., Muller, T.G., Fornasier, S., Groussin, O., Lacerda, P., Ortiz, J.L., Thirouin, A., Delsanti, A., Duffard, R., Harris, A.W., Henry, F., Lim, T., Moreno, R., Mommert, M., Mueller, M., Protopapa, S., Stansberry, J., Trilling, D., Vilenius, E., Barucci, A., Crovisier, J., Doressoundiram, A., Dotto, E., Guti´errez, P.J., Hainaut, O., Hartogh, P., Hestroffer, D., Horner, J., Jorda, L., Kidger, M., Lara, L., Rengel, M., Swinyard, B., Thomas, N., 2010. "TNOs are cool": A survey of the trans-Neptunian region. II. The thermal lightcurve of (136108) Haumea. A&A 518, L147+. Lellouch, E., Laureijs, R., Schmitt, B., Quirico, E., de Bergh, C., Crovisier, J., Coustenis, A., 2000a. Pluto's Non-isothermal Surface. Icarus 147, 220 -- 250. Lellouch, E., Moreno, R., Ortiz, J.L., Paubert, G., Doressoundiram, A., Peixinho, N., 2002. Coordinated thermal and optical observations of Trans-Neptunian object (20000)Varuna from Sierra Nevada. A&A 391, 1133 -- 1139. Lellouch, E., Paubert, G., Moreno, R., Schmitt, B., 2000b. NOTE: Search for Variations in Pluto's Millimeter-Wave Emission. Icarus 147, 580 -- 584. Lestrade, J., 2008. Astrometry with ALMA: a giant step from 0.1 arcsecond to 0.1 milliarcsecond in the sub-millimeter, in: W. J. Jin, I. Platais, & M. A. C. Perryman (Ed.), IAU Symposium, pp. 170 -- 177. Levi, A., Podolak, M., 2009. Corona-like atmospheric escape from KBOs. I. Gas dynamics. Icarus 202, 681 -- 693. Lim, T.L., Stansberry, J., Muller, T.G., Mueller, M., Lellouch, E., Kiss, C., Santos-Sanz, P., Vilenius, E., Protopapa, S., Moreno, R., Delsanti, A., Duffard, R., Fornasier, S., Groussin, O., Harris, A.W., Henry, F., Horner, J., Lacerda, P., Mommert, M., Ortiz, J.L., Rengel, M., Thirouin, A., Trilling, D., Barucci, A., Crovisier, J., Doressoundiram, A., Dotto, E., Guti´errez Buenestado, P.J., Hainaut, O., Hartogh, P., Hestroffer, D., Kidger, M., Lara, L., Swinyard, B.M., Thomas, N., 2010. "TNOs are Cool": A survey of the trans-Neptunian region . III. Thermophysical properties of 90482 Orcus and 136472 Makemake. A&A 518, L148+. Margot, J.L., Trujillo, C., Brown, M.E., Bertoldi, F., 2002. The size and albedo of KBO 2002 AW197, in: Bulletin of the American Astronomical Society, p. 871. McKinnon, W.B., Prialnik, D., Stern, S.A., Coradini, A., 2008. Structure and Evolution of Kuiper Belt Objects and Dwarf Planets, in: The Solar System Beyond Neptune, pp. 213 -- 241. Moreno, R., 2007. Continuum measurements of callisto and Ganymede with IRAM PdB: Application to flux calibration, in: Workshop on flux calibration at IRAM. Moreno, R., Guilloteau, S., 2002. An Amplitude Calibration Strategy for ALMA, in: ALMA Memo 372. Moullet, A., Lellouch, E., Moreno, R., Gurwell, M.A., Moore, C., 2008. First disk-resolved millimeter observations of Io's surface and SO2 atmosphere. A&A 482, 279 -- 292. Muller, T.G., Lellouch, E., Bohnhardt, H., Stansberry, J., Barucci, A., Crovisier, J., Delsanti, A., 22 Doressoundiram, A., Dotto, E., Duffard, R., Fornasier, S., Groussin, O., Guti´errez, P.J., Hainaut, O., Harris, A.W., Hartogh, P., Hestroffer, D., Horner, J., Jewitt, D., Kidger, M., Kiss, C., Lacerda, P., Lara, L., Lim, T., Mueller, M., Moreno, R., Ortiz, J., Rengel, M., Santos-Sanz, P., Swinyard, B., Thomas, N., Thirouin, A., Trilling, D., 2009. TNOs are Cool: A Survey of the Transneptunian Region. Earth Moon and Planets 105, 209 -- 219. Muller, T.G., Lellouch, E., Stansberry, J., Kiss, C., Santos-Sanz, P., Vilenius, E., Protopapa, S., Moreno, R., Mueller, M., Delsanti, A., Duffard, R., Fornasier, S., Groussin, O., Harris, A.W., Henry, F., Horner, J., Lacerda, P., Lim, T., Mommert, M., Ortiz, J.L., Rengel, M., Thirouin, A., Trilling, D., Barucci, A., Crovisier, J., Doressoundiram, A., Dotto, E., Guti´errez, P.J., Hainaut, O.R., Hartogh, P., Hestroffer, D., Kidger, M., Lara, L., Swinyard, B., Thomas, N., 2010. "TNOs are Cool": A survey of the trans-Neptunian region. I. Results from the Herschel science demonstration phase (SDP). A&A 518, L146+. Noll, K.S., Grundy, W.M., Chiang, E.I., Margot, J., Kern, S.D., 2008. Binaries in the Kuiper Belt, in: The Solar System Beyond Neptune, pp. 345 -- 363. Pan, M., Sari, R., 2005. Shaping the Kuiper belt size distribution by shattering large but strengthless bodies. Icarus 173, 342 -- 348. Pety, J., Gueth, F., Guilloteau, S., 2001. Impact of ACA on the Wide-Field Imaging Capabilities of ALMA, in: ALMA Memo 398. Reuter, D.C., Stern, S.A., Scherrer, J., Jennings, D.E., Baer, J.W., Hanley, J., Hardaway, L., Lunsford, A., McMuldroch, S., Moore, J., Olkin, C., Parizek, R., Reitsma, H., Sabatke, D., Spencer, J., Stone, J., Throop, H., van Cleve, J., Weigle, G.E., Young, L.A., 2008. Ralph: A Visible/Infrared Imager for the New Horizons Pluto/Kuiper Belt Mission. Space Science Reviews 140, 129 -- 154. Ruze, J., 1966. Antenna cost, efficiency, and system noise. IEEE Transactions on Antennas and Propa- gation 14, 249 -- 250. Schaller, E.L., Brown, M.E., 2007. Volatile Loss and Retention on Kuiper Belt Objects. ApJ 659, L61 -- L64. Sheppard, S.S., Lacerda, P., Ortiz, J.L., 2008. Photometric Lightcurves of Transneptunian Objects and Centaurs: Rotations, Shapes, and Densities, in: The Solar System Beyond Neptune, pp. 129 -- 142. Sicardy, B., Bellucci, A., Gendron, E., Lacombe, F., Lacour, S., Lecacheux, J., Lellouch, E., Renner, S., Pau, S., Roques, F., Widemann, T., Colas, F., Vachier, F., Martins, R.V., Ageorges, N., Hainaut, O., Marco, O., Beisker, W., Hummel, E., Feinstein, C., Levato, H., Maury, A., Frappa, E., Gaillard, B., Lavayssi`ere, M., di Sora, M., Mallia, F., Masi, G., Behrend, R., Carrier, F., Mousis, O., Rousselot, P., Alvarez-Candal, A., Lazzaro, D., Veiga, C., Andrei, A.H., Assafin, M., da Silva Neto, D.N., Jacques, C., Pimentel, E., Weaver, D., Lecampion, J., Doncel, F., Momiyama, T., Tancredi, G., 2006. Charon's size and an upper limit on its atmosphere from a stellar occultation. Nature 439, 52 -- 54. Spencer, J.R., Lebofsky, L.A., Sykes, M.V., 1989. Systematic biases in radiometric diameter determina- tions. Icarus 78, 337 -- 354. Stansberry, J., Grundy, W., Brown, M., Cruikshank, D., Spencer, J., Trilling, D., Margot, J., 2008. Physical Properties of Kuiper Belt and Centaur Objects: Constraints from the Spitzer Space Tele- scope, in: The Solar System Beyond Neptune, pp. 161 -- 179. Stern, S.A., Buie, M.W., Trafton, L.M., Flynn, B.C., 1995. High-resolution HST Images of the Pluto- Charon System, in: Lunar and Planetary Institute Science Conference Abstracts, p. 1359. Thomas, N., Eggers, S., Ip, W., Lichtenberg, G., Fitzsimmons, A., Jorda, L., Keller, H.U., Williams, I.P., Hahn, G., Rauer, H., 2000. Observations of the Trans-Neptunian Objects 1993 SC and 1996 TL66 with the Infrared Space Observatory. ApJ 534, 446 -- 455. Thompson, A.R., Moran, J.M., Swenson, G.W., 1986. Interferometry and synthesis in radio astronomy. New York, Wiley-Interscience. Young, L.A., Stern, S.A., Weaver, H.A., Bagenal, F., Binzel, R.P., Buratti, B., Cheng, A.F., Cruikshank, D., Gladstone, G.R., Grundy, W.M., Hinson, D.P., Horanyi, M., Jennings, D.E., Linscott, I.R., McComas, D.J., McKinnon, W.B., McNutt, R., Moore, J.M., Murchie, S., Olkin, C.B., Porco, C.C., Reitsema, H., Reuter, D.C., Spencer, J.R., Slater, D.C., Strobel, D., Summers, M.E., Tyler, G.L., 2008. New Horizons: Anticipated Scientific Investigations at the Pluto System. Space Science Reviews 140, 93 -- 127. 23
1511.06556	1	1511	2015-11-20T10:53:49	Fossilized condensation lines in the Solar System protoplanetary disk	[ "astro-ph.EP" ]	The terrestrial planets and the asteroids dominant in the inner asteroid belt are water poor. However, in the protoplanetary disk the temperature should have decreased below water condensation level well before the disk was photoevaporated. Thus, the global water depletion of the inner Solar System is puzling. We show that, even if the inner disk becomes cold, there cannot be direct condensation of water. This is because the snowline moves towards the Sun more slowly than the gas itself. The appearance of ice in a range of heliocentric distances swept by the snowline can only be due to the radial drift of icy particles from the outer disk. However, if a sufficiently massive planet is present, the radial drift of particles is interrupted, because the disk acquires a superKeplerian rotation just outside of the planetary orbit. From this result, we propose that the precursor of Jupiter achieved about 20 Earth masses when the snowline was still around 3 AU. This effectively fossilized the snowline at that location. Although cooling, the disk inside of the Jovian orbit remained ice-depleted because the flow of icy particles from the outer system was intercepted by the planet. This scenario predicts that planetary systems without giant planets should be much more rich in water in their inner regions than our system. We also show that the inner edge of the planetesimal disk at 0.7AU, required in terrestrial planet formation models to explain the small mass of Mercury and the absence of planets inside of its orbit, could be due to the silicate condensation line, fossilized at the end of the phase of streaming instability that generated the planetesimal seeds. Thus, when the disk cooled, silicate particles started to drift inwards of 0.7AU without being sublimated, but they could not be accreted by any pre-existing planetesimals.	astro-ph.EP	astro-ph	Fossilized condensation lines in the Solar System protoplanetary disk A. Morbidelli(1), B. Bitsch(2), A. Crida(1,4), M. Gounelle(3,4), T. Guillot(1), S. Jacobson(1,5), A. Johansen(2), M. Lambrechts(1), E. Lega(1) (1) Laboratoire Lagrange, UMR7293, Universit´e Cote d’Azur, CNRS, Observatoire de la Cote d’Azur. Boulevard de l’Observatoire, 06304 Nice Cedex 4, France. (Email: [email protected] / Fax: +33-4-92003118) (2) Lund Observatory, Department of Astronomy and Theoretical Physics, Lund University, Box 43, 22100, Lund, Sweden (3) IMPMC, Mus´eum national d’histoire naturelle, Sorbonne Universit´es, CNRS, UPMC & IRD, 57 rue Cuvier 75005 Paris, France (4) Institut Universitaire de France 103 bd Saint Michel, 75005 Paris, France (5) Bayerisches Geoinstitut, University of Bayreuth, D-95440 Bayreuth, Germany ABSTRACT The terrestrial planets and the asteroids dominant in the inner asteroid belt are water poor. However, in the protoplanetary disk the temperature should have decreased below water-condensation level well before the disk was photo-evaporated. Thus, the global water depletion of the inner Solar System is puzling. We show that, even if the inner disk becomes cold, there cannot be direct condensation of water. This is because the snowline moves towards the Sun more slowly than the gas itself. Thus the gas in the vicinity of the snowline always comes from farther out, where it should have already condensed, and therefore it should be dry. The appearance of ice in a range of heliocentric distances swept by the snowline can only be due to the radial drift of icy particles from the outer disk. However, if a planet with a mass larger than 20 Earth mass is present, the radial drift of particles is interrupted, because such a planet gives the disk a super-Keplerian rotation just outside of its own orbit. From this result, we propose that the precursor of Jupiter achieved this threshold mass when the snowline was still around 3 AU. This effectively fossilized the snowline at that location. In fact, even if it cooled later, the disk inside of Jupiter’s orbit remained ice-depleted because the flow of icy particles from the outer system was intercepted by the planet. This scenario predicts that planetary systems without giant planets should be much more rich in water in their inner regions than our system. We also show that the inner edge of the planetesimal disk at 0.7 AU, required in terrestrial planet formation models to explain the small mass of Mercury and the absence of planets inside of its orbit, could be due to the silicate condensation line, fossilized at the end of the phase of streaming instability that generated the planetesimal seeds. Thus, when the disk cooled, silicate particles started to drift inwards of 0.7 AU without being sublimated, but they could not be accreted by any pre-existing planetesimals. – 2 – 1. Introduction The chemical structure of a protoplanetary disk is characterized by a condensation front for It marks the boundary beyond which the temperature is low enough to each chemical species. If one allow the condensation of the considered species, given its local partial pressure of gas. assumes that the disk is vertically isothermal and neglects pressure effects, the condensation front is a vertical straight-line in (r, z) space. This is the reason for the wide-spread use of the term “condensation line”. However, the vertical isothermal approximation is in many cases a poor proxy for the thermal structure of the disk (see below), so that in reality the condensation “line” is a curve in (r, z) space, like any other isothermal curve (Isella and Natta, 2005). Probably the most important condensation line is that for water, also called the ice-line or the snowline. In the Solar System water accounts for about 50% of the mass of all condensable species (Lodders, 2003). The fact that the inner Solar System objects (terrestrial planets, asteroids of the inner main belt) are water poor, whereas the outer Solar System objects (the primitive asteroids in the outer belt, most satellites of the giant planets and presumably the giant planets cores, the Kuiper belt objects and the comets) are water rich, argues for the importance of the snowline in dividing the protoplanetary disk in two chemically distinct regions. Thus, modeling the thermal structure of the disk has been the subject of a number of papers. There are two major processes generating heat: viscous friction and stellar irradiation. Chiang and Goldreich (1997), Dullemond et al. (2001, 2002) and Dullemond (2002) neglected viscous heating and considered only stellar irradiation of passive disks. They also assumed a constant opacity (i.e. independent of temperature). Chiang and Goldreich demonstrated the flared structure of a protoplanetary disk while the Dullemond papers stressed the presence of a puffed-up rim due to the face illumination of the disk’s inner edge. This rim casts a shadow onto the disk, until the flared structure brings the outer disk back into illumination. Hueso and Guillot (2005), Davis (2005), Garaud and Lin (2007), Oka et al. (2014a, 2015) and Bailli´e et al. (2015) considered viscous heating also and introduced temperature dependent opacities with increasingly sophisticated prescriptions. They demonstrated that viscous heating dominates in the M > 10−10M(cid:12)/y (Oka et al., 2011), where M is the radial mass-flux of inner part of the disk for gas (also known as the stellar accretion rate) sustained by the viscous transport in the disk. In the most sophisticated models, the aspect ratio of the disk is grossly independent of radius in the region where the viscous heating dominates, although bumps and dips exist (with the associated shadows) due to temperature-dependent transitions in the opacity law (Bitsch et al., 2014a, 2015). The temperature first decreases with increasing distance from the midplane, then increases again due to the stellar irradiation of the surface layer. The outer part of the disk is dominated by stellar irradiation and is flared as predicted earlier; the temperature in that region is basically constant with height near the mid-plane and then increases approaching the disk’s surface. (2011), Bitsch et al. As a consequence of this complex disk structure, the snowline is a curve in the (r, z) plane (see for instance Fig. 4 of Oka et al., 2011). On the midplane, the location of the snowline is at – 3 – M = 3–10 × 10−8M(cid:12)/y. When the accretion about 3 AU when the accretion rate in the disk is rate drops to 5–10 × 10−9M(cid:12)/y the snowline on the midplane has moved to 1 AU (Hueso and Guillot, 2005; Davis, 2005; Garaud and Lin, 2007; Oka et al., 2011; Bailli´e et al., 2015; Bitsch et al., 2015). Please notice that a disk should not disappear before that the accretion rate decreases to M (cid:46) 10−9M(cid:12)/y (Alexander et al., 2014). The exact value of the accretion rate for a given snowline location depends on the disk model (1+1D as in the first four references or 2D as in the last one) and on the assumed dust/gas ratio and viscosity but does not change dramatically from one case to the other for reasonable parameters, as we will see below (eq. 9). The stellar accretion rate as a function of age can be inferred from observations. Hartmann et al. (1998) found that on average M = 10−8M(cid:12)/y at 1 My and M = 1–5×10−9M(cid:12)/y at 3 My. The accretion rate data, however, appear dispersed by more than an order of magnitude for any given age (possibly because of uncertainties in the measurements of the accretion rates and in the estimates of the stellar ages, but nevertheless there should be a real dispersion of accretion rates in nature). In some cases, stars of 3-4 My may still have an accretion rate of 10−8M(cid:12)/y (Hartmann et al., 1998; Manara et al., 2013). The Solar System objects provide important constraints on the evolution of the disk chemistry as a function of time. Chondritic asteroids are made of chondrules. The ages of chondrules span the ∼ 3 My period after the formation of the first solids, namely the calcium-aluminum inclusions (CAIs; Villeneuve et al., 2009; Connelly et al., 2012; Bollard et al., 2014; Luu et al., 2015). The measure of the age of individual chondrules can change depending on which radioactive clock is used, but the result that chondrule formation is protracted for ∼ 3 My seems robust. Obviously, the chondritic parent bodies could not form before the chondrules. Hence, we can conclude that they formed (or continued to accrete until; Johansen et al., 2015) 3-4 My after CAIs. At 3 My (typically M = 1–5×10−9M(cid:12)/y) the snowline should have been much closer to the Sun than the inner edge of the asteroid belt (the main reservoir of chondritic parent bodies). Nevertheless, ordinary and enstatite chondrites contain very little water (Robert, 2003). Some water alteration can be found in ordinary chondrites (Baker et al., 2003) as well as clays produced by the effect of water (Alexander et al. 1989). Despite these observations, it seems very unlikely that the parent bodies of these meteorites ever contained ∼ 50% of water by mass, as expected for a condensed gas of solar composition (Lodders, 2003). One could think that our protoplanetary disk was one of the exceptional cases still showing stellar accretion (cid:38) 10−8M(cid:12)/y at ∼ 3 My. However, this would not solve the problem. In this case the disk would have just lasted longer, while still decaying in mass and cooling. In fact, the photo-evaporation process is efficient in removing the disk only when the accretion rate drops at (cid:46) 10−9M(cid:12)/y (see Fig. 4 of Alexander et al., 2014). Thus, even if the chondritic parent bodies had formed in a warm disk, they should have accreted a significant amount of icy particles when, later on, the temperature decreased below the water condensation threshold, but before the disk disappeared. – 4 – The Earth provides a similar example. Before the disk disappears ( M ∼ 10−9M(cid:12)/y), the snowline is well inside 1 AU (Oka et al., 2011). Thus, one could expect that plenty of ice-rich planetesimals formed in the terrestrial region and our planet accreted a substantial fraction of water by mass. Instead, the Earth contains no more than ∼ 0.1% of water by mass (Marty, 2012). The water budget of the Earth is perfectly consistent with the Earth accreting most of its mass from local, dry planetesimals and just a few percent of an Earth mass from primitive planetesimals coming from the outer asteroid belt, as shown by dynamical models (Morbidelli et al., 2000; Raymond et al., 2004, 2006, 2007; O’Brien et al., 2006, 2014). Why water is not substantially more abundant on Earth is known as the snowline problem, first pointed out clearly by Oka et al. (2011). Water is not an isolated case in this respect. The Earth is depleted in all volatile elements (for lithophile volatile elements the depletion progressively increases with decreasing condensation temperature; McDonough and Sun, 1995). Albarede (2009), using isotopic arguments, demonstrated that this depletion was not caused by the loss of volatiles during the thermal evolution of the planet, but is due to their reduced accretion relative to solar abundances. Furthermore, a significant accretion of oxidized material would have led to an Earth with different chemical properties (Rubie et al., 2015). Mars is also a water-poor planet, with only 70–300ppm of water by mass (McCubbin et al., 2012). Thus, it seems that the water and, more generally, the volatile budget of Solar System bodies reflects the location of the snowline at a time different from that at which the bodies formed. Interestingly and never pointed out before, the situation may be identical for refractory elements. In fact, a growing body of modeling work (Hansen, 2009; Walsh et al., 2011; Jacobson and Morbidelli, 2014) suggests that the disk of planetesimals that formed the terrestrial planets had an inner edge at about 0.7 AU. This edge is required in order to produce a planet of small mass like Mercury (Hansen, 2009). On the midplane, a distance of 0.7 AU corresponds to the condensation line for silicates (condensation temperature ∼ 1300 K) for a disk with accretion rate M ∼ 1.5× 10−7M(cid:12)/y, typical of an early disk. Inside this location, it is therefore unlikely that objects could form near time zero. The inner edge of the planetesimal disk at 0.7 AU then seems to imply that, for some unknown reason, objects could not form there even later on, despite the local disk’s temperature should have dropped well below the value for the condensation of silicates. Clearly, this argument is more speculative than those reported above for the snowline, but it is suggestive that the snowline problem is common to all chemical species. It seems to indicate that the structure of the inner Solar System carries the fossilized imprint of the location that the condensation lines had at an early stage of the disk, rather than at a later time, more characteristic of planetesimal and planet formation; hence the title of this Note. Interestingly, if this analogy between the silicate condensation line and the snowline is correct, the time of fossilization of these two lines would be different (the former corresponding to the time when M ∼ 1.5 × 10−7M(cid:12)/y, the latter when M ∼ 3 × 10−8M(cid:12)/y). The goal of this Note is to discuss how this might be understood. This Note will not present new sophisticated calculations, but simply put together results already published in the literature and connect them to propose some considerations, to our knowledge never presented before, that – 5 – may explain the fossilization of the condensation lines, with focus on the snowline and the silicate line. Below, we start in Sect. 2 with a brief review of scenarios proposed so far to solve the snowline and the 0.7 AU disk edge problems. In Sect. 3 we discuss gas radial motion, the radial displacement of the condensation lines and the radial drift of solid particles. This will allow us to conclude that the direct condensation of gas is not the main process occurring when the temperature decreases, but instead it is the radial drift of particles from the outer disk that can repopulate the inner disk of condensed species. With these premises, in Sect. 4 we focus on the snowline, and discuss mechanisms for preventing or reducing the flow of icy particles, so to keep the Solar System deficient in ice inside ∼ 3 AU even when the temperature in that region dropped below the ice-condensation threshold. In Section 5 we link the inner edge of the planetesimal disk to the original location of the silicate condensation line and we attempt to explain why no planetesimals formed inside this distance when the temperature dropped. A wrap-up will follow in Sect. 6 and an Appendix on planet migration in Sect. 7. 2. Previous models The condensation line problem is a subject only partially explored. For the snowline problem, Martin and Livio (2012, 2013) proposed that the dead zone of the protoplanetary disk piled up enough gas to become gravitationally unstable. The turbulence driven by self-gravity increased the temperature of the outer parts of the dead zone and thus the snowline could not come within 3 AU, i.e. it remained much farther from the star than it would in a normal viscously evolving disk. This model, however, has some drawbacks. First, it predicts an icy region inside of the Earth’s orbit, so that Venus and Mercury should have formed as icy worlds. Second, from the modeling standpoint, the surface density ratio between the deadzone and the active zone of the disk is inversely proportional to the viscosity ratio only in 1D models of the disk. In 2D (r, z) models (Bitsch et al., 2014b) the relationship between density and viscosity is non-trivial because the gas can flow in the surface layer of the disk. Thus, the deadzone may not become gravitationally unstable. Hubbard and Ebel (2014) addressed the deficiency of the Earth in lithophile volatile elements. They proposed that grains in the protoplanetary disk are originally very porous. Thus, they are well coupled with the gas and distributed quite uniformly along the vertical direction. The FU- Orionis events, that our Sun presumably experienced like most young stars, would have heated above sublimation temperature the grains at the surface of the disk. Then, the grains would have recondensed, losing the volatile counterpart and acquiring a much less porous structure and a higher density. These reprocessed grains would have preferentially sedimented onto the disk’s midplane, featuring the major reservoir of solids for the accretion of planetesimals and the planets. Planetesimals and planets would therefore have accreted predominantly from volatile depleted dust, even though the midplane temperature was low. This model is appealing, but has the problem that – 6 – the phase of FU-Orionis activity of a star lasts typically much less than the disk’s lifetime. Thus eventually the devolatilazation of the grains would stop and the planetesimals and planets would keep growing from volatile-rich grains. Also, it neglects the radial drift of icy particles on the mid-plane from the outer disk. Concerning the inner edge of the planetesimal disk at 0.7 AU, an explanation can be found in Ida and Lin (2008). The authors pointed out that the timescale for runaway growth of planetary embryos decreases with heliocentric distance. Because the radial migration speed of embryos is propotional to their mass (Tanaka et al., 2002), the innermost embryos are lost into the star and are not replaced at the same rate by embryos migrating inward from farther out. This produces an effective inner edge in the solid mass of the disk, that recedes from the Sun as time progresses (see Fig. 2 of Ida and Lin, 2008). The major issue here is whether planets and embryos can really be lost into the star. The observation of extrasolar planets has revealed the existence of many “hot” planets, with orbital periods of a few days. Clearly, these planets would be rare if there had existed no stopping mechanism to their inward migration, probably due to the existence of an inner edge of the protoplanetary disk where the Keplerian period is equal to the star’s rotation period (Koenigl, 1991; Lin et al., 1996), acting like a planet-trap (Masset et al., 2006). The presence of planet-trap would change completely the picture presented in Ida and Lin (2008) (see for instance Cossou et al., 2014). More recently, Batygin and Laughlin (2015) and Volk and Gladman (2015) proposed that the the Solar System formed super-Earths inside of 0.7 AU, but these planets were lost, leaving behind only the “edge” inferred by terrestrial planet formation models. In Batygin and Laughlin (2015), the super-Earths are pushed into the Sun by small planetesimals drifting by gas-drag towards the Sun and captured in mean motion resonances. Again, we are faced with the issue of the probable presence of a planet-trap at the inner edge of the protoplanetary disk. With a planet-trap, the super-Earths would probably not have been removed despite of the planetesimals push. In Volk and Gladman (2015), instead, the super-Earths become unstable and start to collide with each other at velocities large enough for these collisions to be erosive. There is no explicit modeling, however, of the evolution of the system under these erosive collisions. We expect that the debris generated in the first erosive collisions would exert dynamical friction on the planets and help them achive a new, stable configuration (see for instance Chambers, 2013). Thus, we think it is unlikely that a system of super-Earths might disappear in this way. From this state-of-art literature analysis it appears that the condensation line problem is still open. Thus, we believe that it is interesting to resume the discussion and approach the problem globally, i.e. addressing the general issue of the “fossilization” of condensation lines at locations corresponding to some “early” times in the disk’s life. – 7 – 3. Relevant radial velocities In this section we review the radial velocities of the gas, of the condensation lines and of solid particles. This will be important to understand how a portion of the disk gets enriched in condensed elements as the disk evolves and cools, and it will give hints on how a region could remain depleted in a chemical species even when the temperature drops beyond its corresponding condensation value. The seminal work for the viscous evolution of a circumstellar disk is Lynden-Bell and Pringle (1974). We consider the disk described in their section 3.2, which can be considered as the archetype of any protoplanetary disk, which accretes onto the star while spreading in the radial direction under the effect of viscous transport. The viscosity ν is assumed to be constant with radius in Lynden-Bell and Pringle’s work, but the results we will obtain below are general for a viscously evolving disk, even with more realistic prescriptions for the viscosity (whenever we need to evaluate the viscosity, we will then adopt the α prescription of Shakura and Sunyaev, 1973). According to eq. (18’) of Lynden-Bell and Pringle (1974), the radial velocity of the gas is (cid:20) ur = − 3 2 ν r 1 − 4a(GM r)2 τ , (1) where G is the gravitational constant, M is the mass of the central star, a is a parameter describing how sharp is initially the outer edge of the disk and τ is a normalized time, defined as τ = βνt + 1 , (2) where t is the natural time and β = 12(GM )2a. Still according to the same paper, the surface density of the gas evolves as: (cid:20) Σ = Cτ−5/4 3πν exp − a(GM r)2 τ , (3) where C is a parameter related to the peak value of Σ at r = 0 and t = 0. The disk described by this equation spreads with time (the term a(GM r)2/τ becoming smaller and smaller with time), while the peak density declines as τ−5/4. The motion of the gas is inwards for r < r0 ≡(cid:112)τ /(4a)/(GM ) and outwards beyond r0, which itself moves outwards as r0 ∝ √ t. We now focus on the inner part of the disk, where r << r0. In this region we can approximate a(GM r)2/τ with 0 and therefore the equations for the radial velocity of the gas and the density become: (cid:21) (cid:21) ur = − 3 2 ν , r Cτ−5/4 3πν . Thus, the stellar accretion rate is Σ = M = −2πrΣur(r) = 3πνΣ = Cτ−5/4 . (4) (5) (6) – 8 – That is, the accretion rate in the inner part of an accretion disk is independent of radius. Eq. (4) gives the radial velocity of the gas, i.e. the first of the expressions we are interested in. Notice that Takeuchi and Lin (2002) found that, in a three dimensional disk, the radial motion of the gas can be outwards in the midplane and inwards at some height in the disk. Nevertheless the global flow of gas is inwards (the inward flow carries more mass than the outwards flow). The velocity ur in (4) can be considered as the radial speed averaged along the vertical direction and ponderated by the mass flow. For our considerations below we can consider this average speed, without worrying about the meridional circulation of the gas. We now compute the speed at which a condensation line moves inwards in this evolving disk. Neglecting stellar irradiation (which is dominant only in the outer part of the disk; Oka et al., 2011, Bitsch et al., 2015), the temperature on the midplane of the disk can be obtained by equating viscous heating (Bitsch et al., 2013): Q+ = 2πrδr ΣνΩ2 , 9 8 (7) with radiative cooling: where Ω = (cid:112)GM/r3 is the orbital frequency, σ is Boltzman constant, κ is the opacity (here Q− = 4πrδrσT 4/(κΣ) , (8) assumed independent of radius and time, for simplicity), T is the temperature and δr is the radial width of the considered annulus. Thus, the expression for the temperature is: T = A[κνΣ2]1/4r−3/4 = A[κΣ M /(3π)]1/4r−3/4 , (9) where A = [9GM/(16σ)]1/4. So, the temperature changes with time (through Σ and M ) and with M , T is weakly dependent (i.e. to the 1/4 radius. Eq. 9 also implies that, for a given value of power) on the product κΣ, namely on the remaining disk parameters. This is why we can link the location of a given condensation line with the disk’s accretion rate with small uncertainty. The derivatives of the temperature with respect to radius and time are: dT dr = ∂T ∂r = − 3 4 A[κνΣ2]1/4r−7/4 , dT dt = ∂T ∂t 1 = 2 = − 5 8 = − 5 8 A[κν]1/4r−3/4Σ−1/2 dΣ dt A[κν]1/4r−3/4Σ−1/2 Cτ−9/4 A[κν]1/4r−3/4Σ−1/2 Cτ−9/4β 3πν dτ dt 3π (10) (11) (in the derivation of the equations above, please remember that we assumed that κ and ν are constant in time and space and we derived in Eq. 5 that, at equilibrium, Σ in the inner part of the disk is independent of radius, so that the total derivatives of T are equal to its partial derivatives). – 9 – Therefore, assuming that the location of a condensation line just depends on temperature (i.e. neglecting the effect of vapor partial pressure), the speed at which a condensation line moves inwards (which is the second expression we are looking for) is: vcond r = − dr dT dT dt = − 5 6 r Σ Cτ−9/4β 3π which, using (5) and approximating τ with βνt, gives: = − 5 6 vcond r r t . By comparing (13) with (4) we find that ur > vcond r for t > 5 9 r2 ν . (12) (13) (14) The inequality (14) implies that, after approximately half a viscous timescale tν ≡ r2/ν, the radial motion of the gas is faster than the displacement of a given condensation line. The lifetime of a disk is typically several viscous timescales for the inner region. In fact, Hartmann et al. (1998) found that the time-decay of the accretion rate on stars implies that, if one adopts an α-prescription for the viscosity (i.e. ν = αH 2Ω; Shakura and Sunyaev, 1973), the value of the coefficient α is 0.001–0.01. At 3 AU, assuming a typical aspect ratio of 5%, the viscous timescale is tν = 3 × 104– 3 × 105 y; at 0.7 AU tν is about 10 times shorter. Both values are considerably shorter than the typical disk’s lifetime of a few My. Thus, for the regions and timescales we are interested in (either the asteroid belt at t ∼ 1–3 My or the region around 0.7 AU at 0.1 My) the condition (14) is fulfilled. This result has an important implication. For t (cid:46) 1/2tν the condensation lines move very quickly. Thus there is the possibility that gas condenses out locally when the temperature drops. But for t (cid:38) 1/2tν this process of direct condensation of gas loses importance. This can be under- stood from the sketch in the left part of Fig. 1. Consider the location r0 of a condensation line at time t0 ∼ 1/2tν, where tν ≡ r2 0/ν. The gas beyond the condensation line (r > r0) is “dry”, in the sense that the considered species is in condensed form; instead the gas at r < r0 is “wet” in the sense that it is rich in the vapor of the considered species. Now, consider first the idealized case where the condensed particles are large enough to avoid radial drift. Because the radial drift of the gas is faster than the radial motion of the condensation line, the outer radial boundary of the wet region moves away from the condensation line, in the direction of the star. In reality the boundary between the two gases in wet and dry form is fuzzy, because it is smeared by turbulent diffusion. But it is clear, from the difference in radial velocities and a simple process of dilution, that the gas just inwards of the condensation lines has to become more and more depleted in the considered species as time progresses. Thus, as the condensation line advances towards the star, the amount of mass that can condense locally is very limited. Thus, the condensed material can be (mostly) found only beyond the original location r0 of the condensation line. – 10 – Fig. 1.— Sketch of the radial motion of a condensation line and of the gas, averaged over the vertical direction in the disk. The left/right parts of the figure differ by the assumption that condensed particles do not drift/drift, respectively. From top to bottom, each panel depicts the situation at different times, labeled on the left of each panel, with t0 < t1 < t2. The time t0 is defined as about half of the viscous timescale at the location of the condensation line r0. The main horizontal arrow indicates the radial direction. The vertical dashed line shows the location of the condensation line as a function of time, approaching the star with a speed vcond . The orange shaded region shows the “wet” gas, rich in the vapor of the considered species. The blue shaded region shows the “dry” gas, depleted in the vapor of considered species because the latter has condensed out. The outer boundary of the wet gas region also moves towards the star, with speed ur. Because ur > vcond this boundary moves away from the condensation line. Thus, if there is no radial drift of particles (left part of the plot) the condensation line moves in a disk of dry gas, and therefore the amount of material that can condense out locally is negligible. Instead, in the case shown in the right part of the plot, the condensed particles repopulate, by radial drift, the disk down to the position of the condensation line. Also, the region (green) in between the “wet” and “dry” domains is resupplied in vapor of the considered species by particles sublimating when they pass through the condensation line (see the symbol for the sublimation front). Thus, the mass-flux of particles governs the abundance of the considered species in gaseous or solid form on both sides of the condensation line. r r Does this mean that a region of the disk originally too hot for a species to condense will remain depleted in that species forever, even if the temperature eventually drops well below the condensation threshold? In principle no, because in a more realistic case (at least some of) the condensed particles are small enough to drift inwards by gas-drag, so that they can populate any region that has become cold enough to host them in solid form (see the right part of Fig. 1). Also, particles drifting through the condensation line can sublimate, thus resupplying the gas of the considered species in vapor form. Thus, particle drift is the key to understanding the condensation line problem. A solid particle can be characterized by a dimensionless parameter called the Stokes number: τf = ρbR ρgcs Ω , (15) where ρb is the bulk density of the particle, ρg is the density of the gas, R is the size of the particle – 11 – and cs is the sound speed. A particle feels a wind from the gas, which has two components. The azimuthal component is due to the fact that the gas rotates around the star at a sub-Keplerian speed due to the pressure gradient in the disk; the radial component is due to the fact that the gas flows towards the Star, due to its own viscous evolution. Both components cause the radial drift of the particles towards the star at the speed (Weidenschilling, 1977; Takeuchi and Lin, 2002): vr = −2 τf τ 2 f + 1 ηvK + ur τ 2 f + 1 , (16) where vK is the Keplerian velocity, ur is the radial velocity of the gas and η is a measure of the gas pressure support: . (17) (cid:18) H (cid:19)2 d log P η = − 1 2 r d log r Eq. (16) is the final radial speed we looked for in this section. For mm-size particles or larger, typically the first term in (16) dominates over the second one. The radial speed of particles is very fast. In fact, the typical value of η is ∼ 3 × 10−3 so that the drift speed at 1 AU is ∼ 4τf × 10−2 AU/y. Even particles as small as a millimeter (τf ∼ 10−3 at ∼ 1 AU) would travel most of the radial extent of the disk within the disk’s lifetime. Solid particles condensed in the outer disk are therefore expected to potentially be delivered in the inner disk. In conclusion, solving the snowline problem, i.e. understanding why Solar System objects remained depleted in species that should have condensed locally before the removal of the gas-disk, requires finding mechanisms that either prevent the radial drift of particles or inhibit the accretion of these particles onto pre-existing objects. Below we investigate some mechanisms, focussing on the cases of the snowline and the silicate line. 4. The snowline In this section we discuss several mechanisms that could have potentially prevented the drift or the accretion of icy particles in the asteroid belt and the terrestrial planet zone even after that the snowline had passed across these regions. 4.1. Fast growth If icy particles had accreted each other quickly after their condensation, forming large objects (km-size or more) that were insensitive to gas drag, the inner Solar System would have received very little flux of icy material from the outer part of the disk (as in the example illustrated in the left part of Fig. 1). We think that this scenario is unlikely. The growth of planetesimals should have been extremely efficient for the fraction of the leftover icy particles to be small enough to have a negligible effect – 12 – on the chemistry of the inner Solar System bodies. Such an efficient accretion has never been demonstrated in any model. Observational constraints suggest the same, by showing that disks are dusty throughout their lifetime (see Williams and Cieza, 2011 or Testi et al., 2014 for reviews), with the exception of the inner part of transitional disks (Espaillat et al., 2014) that we will address in sect. 4.4. The Solar System offers its own constraints against this scenario. In chondrites, the ages of the individual chondrules inside the same meteorite span a few millions of years (Villeneuve et al., 2009; Connelly et al., 2012). Despite this variability, it is reasonable to assume that all particles (chondrules, CAIs,...) in the same rock got accreted at the same time. Thus, the spread in chondrule ages implies that particles were not trapped in planetesimals as soon as they formed; instead they circulated/survived in the disk for a long time before being incorporated into an object. Similarly, CAIs formed earlier than most chondrules (Connelly et al., 2012; Bollard et al., 2014), but they were incorporated in the meteorites with the chondrules; this means that the CAIs also spent significant time in the disk before being incorporated in macroscopic objects. Thus, it seems unlikely that virtually all icy particles had been accreted into planetesimals at early times, given that this did not happen for their refractory counterparts (CAIs and chondrules). 4.2. Inefficient accretion A second possibility could be that the water-rich particles that drifted into the asteroid belt once the latter became cold enough, were very small. Small particles accrete inefficiently on pre- existing planetesimals because they are too coupled with the gas (Lambrechts and Johansen, 2012; Johansen et al., 2015) and they are also very inefficient in triggering the streaming instability (Youdin and Goodman, 2005; Bai and Stone, 2010a,b; Carrera et al., 2015). Also, small particles are not collected in vortices, but rather accumulate in the low-vorticity regions at the dissipation scale of the turbulent cascade (Cuzzi et al., 2001). The levels of concentration that can be reached, however, are unlikely to be large enough to allow the formation of planetesimals (Pan et al., 2011). Thus, if the flux of icy material through the asteroid belt and the terrestrial planet region was mostly carried by very small particles, very little of this material would have been incorporated into asteroids and terrestrial planets precursors. But how small is small? Again, chondrites give us important constraints. Chondrites are made of chondrules, which are 0.1–1 mm particles. Thus, particles this small could accrete into (or onto – Johansen et al., 2015) planetesimals. The ice-rich particles flowing from the outer disk are not expected to have been smaller than chondrules. Lambrechts and Johansen (2014) developed a model of accretion and radial drift of particles in the disk based on earlier work from Birnstiel et al. (2012). They found that the size of particles available in the disk decreased with time (the bigger particles being lost faster by radial drift). They estimated that, at 1 My, the particles at 2 AU were a couple of cm in size, so more than 10 times the chondrule size; the particles would have been chondrule-size at ∼ 10 My. Thus, we don’t see any reason why the asteroids should have accreted chondrules but not – 13 – ice-rich particles, if the latter had drifted through the inner part of the asteroid belt. Consequently, this scenario seems implausible as well. 4.3. Filtering by planetesimals Particles, as they drifted radially, passed through a disk which presumably had already formed planetesimals of various sizes. Each planetesimal accreted a fraction of the drifting particles. If there were many planetesimals and they accreted drifting particles efficiently, the flow of icy material could have been decimated before reaching the inner Solar System region. Guillot et al. (2014) developed a very complete analytic model of the process of filtering of drifting dust and pebbles by planetesimals. Unfortunately, the results are quite disappointing from our perspective. As shown by Figs. 21 and 22 of Guillot et al., in general only large boulders (about 10m in size) drifting from the outer disk (35 AU for the calculations illustrated in those figures, but the result is not very sensitive on this parameter) would have been accreted by planetesimals before coming within a few AU from the Sun. There are a few exceptions to this, however, also illustrated in the Guillot et al. paper. If the disk hosted a population of km-size planetesimals with a total mass corresponding to the solid mass in the Minimum Mass Solar Nebula model (MMSN; Weidenschillig, 1977b; Hayashi, 1981) and the turbulent stirring in the disk was weak (α = 10−4 in the prescription of Shakura and Sunyaev, 1973) particles smaller than a millimeter in size could have been filtered efficiently and failed to reach the inner Solar System (however, see Fig. S2 in Johansen et al., 2015, for a different result). We think that it is unlikely that these parameters are pertinent for the real protoplanetary disk. In fact, we have seen above that icy particles are expected to have had sizes of a few cm at 1 My at 2 AU (Lambrechts and Johansen, 2014). Moreover, we believe it is unlikely that the size of the planetesimals that carried most of the mass of the disk was about 1 km. The formation of km-size planetesimals presents unsolved problems (e.g. the m-size barrier -Weidenschilling, 1977- and the bouncing barrier -Guttler et al., 2009). Instead, modern accretion models (e.g. Johansen et al., 2007, 2015; Cuzzi et al., 2010) and the observed size distributions in the asteroid belt and the Kuiper belt suggest that planetesimals formed from self-gravitating clumps of small particles, with characteristics sizes of 100km or larger (Morbidelli et al., 2009). Therefore, more interesting is the other extreme of the parameter space identified by Guillot et al. (2014). If a MMSN mass was carried by “planetesimals” more massive than Mars and the turbulent stirring was small (again, α = 10−4), particles larger than 1 cm in size would have been filtered efficiently and would have failed to reach the inner Solar System. The lower limit of the mass of the filtering planetesimals decreases to 1/10 of a Lunar mass if the “particles” were meter-size boulders. Clearly, this is an important result. It is unclear, however, whether the protoplanetary disk could host so many planetesimals so big in size. The mass in solids in the MMSN model between 1 and 35 AU is about 50 M⊕. Assuming that this mass was carried by Mars-mass bodies would require the existence of about 500 of these objects. – 14 – 4.4. Filtering by proto-Jupiter At some point in the history of the protoplanetary disk, Jupiter started to form. The formation of the giant planets is not yet very clearly understood, so it is difficult to use models to assert when and where Jupiter had a given mass. However, it has been pointed out in Morbidelli and Nesvorny (2012) that when a planet reaches a mass of the order of 50 Earth masses it starts opening a partial gap in the disk. In an annulus just inside the outer edge of the gap the pressure gradient of the gas is reversed. Therefore, in this annulus the rotation of the gas around the Sun becomes faster than the Keplerian speed. Thus, the drag onto the particles is reversed. Particles do not spiral inwards, but instead spiral outwards. Consequently, particles drifting inwards from the outer disk have to stop near the outer edge of the gap. This process is often considered to be at the origin of the so-called “transitional disks” (Espaillat et al., 2014), which show a strong depletion in mm-sized dust inside of some radius, with no proportional depletion in gas content. This mechanism for stopping the radial drift of solid particles has been revisited in Lambrechts et al. (2014), who used three dimensional hydro-dynamical simulations to improve the estimate of the planet’s mass-threshold for reversing the gas pressure gradient. They found that the mass- threshold scales with the cube of the aspect ratio h of the disk and is: (cid:18) h (cid:19)3 Miso = 20M⊕ 0.05 , (18) quite insensitive to viscosity (within realistic limits). Only particles very small and well-coupled with the gas (about 100µm or less; Paardekooper and Mellema, 2006) would pass through the gap opened by the planet and continue to drift through the inner part of the disk. However, these particles are difficult to accrete by planetesimals, because they are “blown in the wind” (Guillot et al., 2014). Thus, they are not very important for the hydration of inner Solar System bodies. The particles which would be potentially important are those mm-sized or larger, because for these sizes the pebble accretion process is efficient (Ormel and Klahr 2010; Lambrechts and Johansen, 2012); however, for these particles the gaps barrier is effective. According to Bitsch et al. (2015), when the snowline was at 3 AU (disk’s accretion rate M ∼ 3 × 10−8M(cid:12)/y) the aspect ratio of the disk was around 0.05 up to ∼ 10 AU . So, the mass of 20 M⊕ is the minimum value required for the proto-Jupiter in order to stop the drift of icy pebbles and large grains. Basically, the constraint that asteroids and the precursors of the terrestrial planets did not accrete much ice translates into a constraint on the mass and location of the proto-Jupiter. More specifically, the proto-Jupiter needs to have reached 20 M⊕ before the disk dropped below M ∼ 3 × 10−8M(cid:12)/y and it needs to have remained beyond the asteroid belt an accretion rate of – 15 – (i.e. beyond ∼ 3 AU) until all the asteroids formed (about 3 My after CAIs). We think that this scenario is reasonable and realistic given that (i) Jupiter exists and thus it should have exceeded 20 M⊕ well within the lifetime of the disk and (ii) Jupiter is beyond 3 AU today. Nevertheless, there are important migration issues for Jupiter, that we will address in the Appendix. In this scenario, the current chemical structure of the Solar System would reflect the position of the snowline fossilized at the time when Jupiter achieved ∼20 M⊕1. Therefore it makes sense that the fossilized snowline position corresponds to a disk already partially evolved (i.e. with M of a few 10−8M(cid:12)/y, instead of a few 10−7M(cid:12)/y, typical of an early disk), given that it may take considerable time (up to millions of years) to grow a planet of that mass. We also notice that this scenario is consistent, at least at the qualitative level, with the fact that ordinary and enstatite chondrites contain some water (typically less than 1% by mass, well below solar relative abundance of ∼ 50%) and show secondary minerals indicative of water alteration (Baker et al., 2003; Alexander et al., 1989). In fact, it is conceivable that some particles managed to jump across the orbit of the proto-Jupiter (either because they were small enough to be entrained in the gas flow or by mutual scattering, once sufficiently piled up at the edge of Jupiter’s gap). Because the entire asteroid region presumably had a temperature well below ice-sublimation by the time these chondrites formed, the icy particles that managed to pass through the planet’s orbit would have been available in the asteroid belt region to be accreted. Of course, if most of the particles were retained beyond the orbit of Jupiter, the resulting abundance of water in these meteorites would have remained well below 50% by mass, as observed. The barrier to particle radial drift induced by the presence of the proto-Jupiter would not just have cut off the flow of icy material. It would also have cut off the flow of silicates. Thus, once the local particles had drifted away, the accretion of planetesimals should have stopped everywhere inside the orbit of the planet. Thus, it seems natural to expect that the chondrites should have stopped accreting at about the time of fossilization of the snowline position. As we said in the introduction, chondrites accreted until ∼ 3 My after CAI formation. Thus the formation of a 20 M⊕ proto-Jupiter should have occurred at about 3 My. Because the position of the snowline should have been at about 3 AU when this happened, this implies that the solar protoplanetary disk had an accretion rate M ∼ 3 × 10−8M(cid:12)/y (to sustain a snowline at 3 AU) at ∼3 My. Therefore, our disk was not typical (typical disks have a lower accretion rate at that age; Hartmann et al., 1998), although still within the distribution of observed accretion rates at this age (Manara et al., 2013). There is, however, a second possibility. If there had been some mechanism recycling particles (i.e. sending the particles back out once they came close enough to the star), or producing new 1In reality, at this time there may still have been a reservoir of icy particles between the snowline location and the orbit of Jupiter. However, because of the typical drift rate of particles (10cm/s for mm-size dust, 1m/s for cm-size pebbles), if this reservoir was just a few AU wide (see Appendix), it should have emptied in 104–105y, i.e. before that the snowline could move substantially. – 16 – particles in situ, it is possible to envision that chondrules continued to form and chondrites continued to accrete them even after the flow of particles from the outer disk was cut off. In that case, the proto-Jupiter should still have formed when the snowline was at about 3 AU, but we would not have any chronological constraint on when this happened. It could have happened significantly earlier than 3 My. Thus, the solar protoplanetary disk might have had a typical accretion rate as a function of time. A mechanism for producing small particles in situ is obviously collisions between planetesi- mals. Chondrules have been proposed to have formed as debris from collisions of differentiated planetesimals (Libourel and Krot, 2007; Asphaugh et al., 2011), although this is still debated (see e.g. Krot et al., 2009 for a reivew). Several mechanisms leading to a recycling of particles have been proposed, such as x-winds (Shu et al. 1996, 1997, 2001), gas outflow on the midplane (Takeuchi and Lin 2002; Ciesla 2007; Bai and Stone, 2010a) and disk winds (Bai, 2014; Staff et al., 2014). Independently of the correct transport mechanism(s), the very detection of high temperature materials within comets (Brownlee et al. 2006; Nakamura et al. 2008; Bridges et al. 2012) demonstrates strong transport from the inner disk regions to the outer disk region. However, it is not clear at which stage of the disk’s life this outward transport was active and whether it concerned also particles larger than the microscopic ones recovered in the Stardust samples. If a mechanism for recycling/producing particles in the inner disk really existed, another impli- cation would be that planetesimals on either side of Jupiter’s orbit eventually accreted from distinct reservoirs. The planetesimals inside of the orbit of the planet accreted only material recycled from the inner disk; instead the planetesimals outside of the proto-Jupiter’s orbit accreted outer disk material, although possibly contaminated by some inner-disk material transported into the outer disk. In this respect, it may not be a coincidence that ordinary and carbonaceous chondrites appear to represent distinct chemical and isotopic reservoirs (Jacquet et al. 2012) because, in addition to the water content, these meteorites show two very distinct trends in the ∆17O-54Cr∗ isotope space (Warren et al. 2011). Today the parent bodies of both classes of meteorites reside in the asteroid belt, i.e. inside of the orbit of Jupiter. But in the Grand Tack scenario (Walsh et al., 2011) the parent bodies of the carbonaceous chondrites formed beyond Jupiter’s orbit and got implanted into the asteroid belt during the phase of Jupiter’s migration. 5. The silicate line According to terrestrial planet formation models, the small mass of Mercury and the absence of planets inside its orbit can be explained only by postulating that the disk of planetesimals and planetary embryos had an inner edge near 0.7 AU (Hansen, 2009). We argued in the introduction that this inner edge might reflect the location of the silicate condensation line at a very early age of the disk. But what could have prevented particles from drifting inside of 0.7 AU, once the disk – 17 – there had cooled below the silicate condensation temperature? Obviously no giant planets formed near 0.7 AU, so the scenario invoked for the snowline cannot apply to this case. A solution can be found in the results presented in Johansen et al., 2015. The authors pointed out that the very early disk is the most favorable environment for the production of planetesimal seeds via the streaming instability. This is because the streaming instability requires the presence of large particles, with Stokes numbers of the order of 0.1–1 (e.g. significantly larger than chondrule- size particles). These particles drift very quickly in the disk, so they are rapidly lost. As shown in Lambrechts and Johansen (2014) the mass ratio between solid particles and gas, as well as the size of the dominant particles, decrease with time. Thus, the streaming instability becomes more and more unlikely to happen as time progresses. Therefore, Johansen et al. argue that planetesimals formed in two stages. In the first stage planetesimal seeds formed by the streaming instability, triggered by large particles (decimeter across). This stage lasted for a short time only, due to the rapid loss of these large particles by radial drift. In the second stage the planetesimal seeds kept growing by accreting chondrule-sized particles, the only ones surviving and still drifting in the disk after a few My. This model of planetesimal formation would provide a natural explanation for the fossilization of the silicate line. Imagine that, when the phase of streaming instability was over, the accretion rate in the protoplanetary disk was M ∼ 1.5× 10−7M(cid:12)/y. In this case, the silicate sublimation line was at ∼ 0.7 AU. Thus, presumably no planetesimal seeds could have formed within this radius, because until that time only the very refractory material would have been available in solid form there (a small fraction of the total mass, insufficient to trigger the streaming instability). After the streaming instability phase was over, planetesimals continued to grow by the accretion of smaller particles onto the planetesimal seeds (Johansen et al., 2015). Meanwhile the disk cooled so that silicate particles could drift within 0.7 AU and remain solid. However, because of the absence of planetesimal seeds within this radius, they could not be accreted by anything, and thus they simply drifted towards the Sun. Clearly, in this scenario the final planetesimal disk would have an inner edge at 0.7 AU, fossilizing the early location of the silicate line that characterized the formation of the planetesimal seeds. Admittedly, this scenario is still qualitative, but it is seducing in its simplicity. The possibility to explain location of the inner edge of the planetesimal disk by this mechanism is an additional argument in favor of the two-phases model for planetesimal growth, simulated in Johansen et al. (2015). 6. Conclusions The chemical composition of the objects of the Solar System seems to reflect a condensation sequence set by a temperature gradient typical of an “early” disk, still significantly warm in its inner part. Particularly significant is the situation concerning water. The terrestrial planets and – 18 – the asteroids predominant in the inner belt are water-poor. However, the disk’s temperature should have decreased below the water-condensation level even at 1 AU before the disappearance of the gas. So, the question why terrestrial planets and asteroids are not all water rich is a crucial one (Oka et al., 2011). Interestingly, water is not the only chemical species that reveals this conundrum. Inner Solar System objects are in general more depleted in volatile element than they should be, given the temperatures expected in the disk at its late stages. Also, terrestrial planet formation models argue that, in order to explain the small mass of Mercury, the planetesimal disk had to have an inner edge at about 0.7 AU (Hansen, 2009; Walsh et al., 2011). This boundary could correspond to the location of the evaporation front for silicates, but again only for a massive (i.e. “early”) disk. In this Note we have discussed how the composition of Solar System objects could reflect the location of the condensation lines fossilized at some specific stages of the disk’s evolution. Some mechanisms have been quickly dismissed, others look promising and we propose them to the community for further discussion and investigation. First, we have demonstrated that the radial motion of gas towards the central star is faster than the inward motion of the condensation lines. This implies that there cannot be condensation of gas in a region swept by the motion of a condensation line, even if this may sound paradoxical. This is because the gas in the considered region comes from farther out and therefore it should have already condensed there (see Fig. 1, left side). Thus, the enrichment of a disk region in condensed elements when the temperature drops can only be due to the radial drift of solid particles from the more distant disk (see Fig. 1, right side). With this consideration in mind, the scenario that we propose to explain the fossilized con- densation lines is the one sketched in Fig. 2. Planetesimal formation occurred in two stages, as proposed in Johansen et al. (2015). In the very early disk, dust coagulation produced pebbles and boulders of sizes ranging from decimeters to, possibly, a meter. These objects were very effective in triggering the streaming instability (Youdin and Goodman, 2005; Johansen et al., 2007; Bai and Stone, 2010a,b) and areodynamically clumped together forming planetesimal seeds of about 100 km in size. This phase, however, could not last long, because these boulders drifted quickly through the disk and those that were not rapidly incorporated into a planetesimal seed got lost by drifting into the Sun. We propose that, when this stage ended, the silicate condensation line M ∼ 1.5 × 10−7M(cid:12)/y, was at 0.7 AU. This would correspond to a disk with an accretion rate of for a nominal metallicity of 1%. Then, no planetesimal seeds could have formed within this ra- dius, because of the lack of a sufficient amount of solids. In the second stage, the surviving solid particles were too small and their mass ratio with the gas too low to trigger streaming instability (Lambrechts and Johansen, 2014). Thus, these particles could only be accreted onto the already formed planetesimal seeds (Johansen et al., 2015). If no seeds existed inside of 0.7 AU, this radius remained the inner edge of the planetesimal disk, a fossil trace of the silicate line at the end of the first stage of planetesimal growth. Meanwhile the temperature in the disk continued to decrease. The snowline moved towards – 19 – Fig. 2.— Sketch of the solution to the follisized condensation lines problem proposed in this paper. The top and central panels show the situation at the times when the silicate line, first, then the snowline, remain fossilized. The bottom panel sketches the situation after the fossilization of the snowline, under the assumption where accretion of planetesimals in the inner disk continues thanks to the recycling of small particles in outwards flows. If a recycling or particle-generation mechanism did not exist, all planetesimals inside of the orbit of proto-Jupiter should have stopped accreting at the time of fossilization of the snowline. See text for detailed description. – 20 – the Sun. The region swept by the snowline became increasingly enriched in icy material due to the radial drift of particles from the outer disk. When the mass of the proto-Jupiter reached 20 M⊕, however, this flux of icy particles stopped. The opening of a shallow gap by the proto- planet created a barrier to the inward drift of the particles by gas drag. Thus, the flux of icy material across Jupiter’s orbit was interrupted, presumably making our Solar System look like a “transitional disk”. We propose that the proto-Jupiter reached this critical mass when the snowline M ∼ 3× 10−8M(cid:12)/y, was at about 3 AU. This corresponds to a disk with a stellar accretion rate of assuming the canonical metallicity of 1%. Thus, the objects inside 3 AU, which could not accrete ice up to that time because the temperature was too high, could not accrete a significant amount of ice also after that the temperature dropped, because of the interrupted icy-particle flow. Instead, they could have continued to accrete refractory particles if the latter were recycled or continuously reproduced in the disk, thanks to the existence of outwards flows in the midplane (Takeuchi and Lin, 2002; Ciesla, 2007; Bai and Stone, 2010a), x-winds (Shu et al., 1996,1997,2001), disk winds (Bai, 2014; Staff et al., 2014) or collisions (Libourel and Krot, 2007; Asphaugh et al., 2011). Thus, the resulting chemistry of planetesimals reflect the location of the snowline fossilized at the time the proto-Jupiter reached 20 M⊕. Notice that, because it takes more time to form the proto-Jupiter than the planetesimal seeds, it makes sense that the snowline appears fossilized at the location corresponding to a “later” disk than the silicate line ( M ∼ 3 × 10−8M(cid:12)/y instead of M ∼ 1.5 × 10−7M(cid:12)/y). This scenario is much simpler than what has been proposed so far for the snowline problem or the origin of the inner edge of the planetesiamal disk (reviewed in Sect. 2). It is indeed appealing for its simplicity. This scenario leads to a few predictions. For the Solar System it predicts that the condensation lines corresponding to species much more volatile than water (e.g. CO, with a condensation tem- perature of ∼ 25 K) should not have been fossilized because only Jupiter and Saturn are sufficiently massive to stop the flow of drifting particles and these planets should always have been too close to the Sun. Thus, the composition of outer Solar System bodies should reflect the location of these lines at the end of the disk’s lifetime. Perhaps this can explain the compositions of Uranus and Neptune (Ali-Dib et al., 2014). For extrasolar planetary systems, the scenario predicts that systems without giant planets should be much more volatile rich in their inner parts than a system like ours. This seems consistent with the observations, which show a large number of systems of low-density super-Earths in close-in orbits and no giant planets farther out (Fressin et al., 2013). In principle the low bulk-densities could be explained by the presence of extended H and He atmosphere around rocky planets. But Lopez and Fortney (2014) concluded that the observed size distribution of extrasolar planets, with a sharp drop-off above 3 R⊕, is diagnostic that most super-Earths are water-rich. In fact, refractory planets with extended atmospheres would have a more uniform size distribution. Finally, the fossilization of the silicate line should be a generic process, although the location – 21 – at which this condensation line is fossilized may change from disk to disk depending on the duration of the streaming instability stage and the evolution of the temperature in that timeframe. This suggests that “hot” extrasolar planets (with orbital radii significantly smaller than Mercury’s) did not form in situ but migrated to their current orbits from some distance away. Clearly, the scenario proposed in this Note remains speculative. However, with the improved understanding of disk evolution and its chronology, planetesimal accretion and giant planet growth, it will be possible in a hopefully not distant future to test it on more quantitative grounds against the available constraints. 7. Appendix: Planet migration issues Planets are known to migrate in disks (see Baruteau et al., 2014 for a review). Thus, we discuss here possible scenarios that could explain how the proto-Jupiter remained beyond ∼ 3 AU until the chondrite formation time. A first possibility is that Jupiter’s core started to form sufficiently far in the disk, so that it could not reach 3 AU before ∼ 3 My. This is one of the approaches taken in Bitsch et al., 2015b. In their model, Jupiter started growing by pebble accretion at about 20 AU. A second possibility is offered by the subtle action of the entropy-driven corotation torque (Paardekooper and Mellema, 2006b; Paardekooper et al., 2010, 2011; Masset and Casoli, 2009, 2010). This torque can reverse the migration of intermediate-mass planets (several Earth masses) in localized regions of the disk where the temperature gradient is steep. Bitsch et al. (2014a, 2015) showed migration maps as a function of location and planet mass at different evolutionary M ) of the disk. They found that the outward migration region is stages (i.e. different values of adjacent to the snowline. It typically ranges from the snowline location up to a few AUs beyond the snowline. For an early disk ( M = 7 × 10−8M(cid:12)/y) outward migration concerns planets with masses between 5 and 40 Earth masses. A proto-Jupiter in this mass range would therefore be retained at 6–8 AU (depending on its mass), the snowline being at ∼ 4 AU (see Fig. 7 of Bitsch et al., 2015). When the disk loses mass and cools, the outward migration region shifts towards the Sun with the snowline position. But it also shrinks in the planet-mass parameter space. In a late disk with M = 8.75 × 10−9M(cid:12)/y (snowline at ∼ 1AU ) the outward migration region still extends to ∼ 3AU , but only for planet masses smaller than 10 M⊕. Thus, in principle, a pebble-stopping planet of 20 M⊕ should have been released to free Type-I migration and should have penetrated into the inner Solar System, possibly too early relative to the chondrite formation time of ∼ 3 My. We stress, however, that the exact location and shape of the outward migration region depends on the adopted parameters for the disk, as shown in Fig. 3. In particular, the upper limit in planet mass for outward migration is due to torque saturation. It can be increased if the disk is more viscous. The size of the coorbital region of a planet (the one characterized by horseshoe streamlines) has a width xs ∝(cid:112)Mp, where Mp is the planet mass. The timescale for viscous transport across – 22 – Fig. 3.— Contours of the outward migration region in the parameter space heliocentric distance vs. plan- M , α and κ are reported on the etary mass. Each color corresponds to a different disk, whose parameters plot. To help reading this plot, the red arrows show the direction of migration of planets of different masses and locations for the case with M = 10−8M(cid:12)/y, α = 0.01 and κ =1%, corresponding to the red contour. A planet of an appropriate mass (between 2.5 and 20 ME for the case of the red contour) migrating inwards from the outer disk would stop at the right-hand-side boundary of the outward migration region. The black and red contorus are too limited in planet-mass range to be able to trap a 20 M⊕ planet, but the other disks could retain a proto-Jupiter of this mass beyond 2.5 or 3 AU. this region is therefore tν ∝ x2 s/ν ∝ Mp/ν . The libration timescale in the horseshoe region is tlib ∝ 1(cid:112)Mp . (19) (20) Saturation is achieved when tlib << tν. If we want saturation to occur for a planet N times bigger in mass, the scaling of (19) and (20) implies that ν has to be N 3/2 times bigger. However, to have the outward migration region in the same radial range we need that the thermal structure of the disk does not change with the increase in viscosity. Because the viscous transport in the disk is proportional to νΣ, we need that Σ is N 3/2 smaller. And because the temperature in the disk is proportional to (κνΣ2)1/4 (see 9) we need that the opacity κ (basically – 23 – the mass ratio between micron-sized dust and gas) scales as N 3/2. In other words, a planet of 20 M⊕ can be retained at 3 AU in a disk with M = 8.75 × 10−9M(cid:12)/y if the viscosity is ∼ 3 times higher and the opacity ∼ 3 times larger than assumed in Bitsch et al. (2015). Given the uncertainties on disk parameters, we cannot exclude this possibility. Fig. 3 indeed shows “late disks” (i.e. with a M – a few 10−9M(cid:12)/y) with an outward migration region capable of retaining a 20 M⊕ planet small beyond 2.5–3 AU. Nevertheless, however we play with disk parameters, it is clear that a planet is eventually released to inward migration once it becomes massive enough. Thus, Jupiter should have eventually invaded the asteroid belt, (unless it started so far out in the disk that it was not able to reach the asteroid belt within the disk’s lifetime; see some simulations in Bitsch et al., 2015b). The migration of Jupiter through the belt is contemplated in the so-called Grand Tack scenario (Walsh et al., 2011), in which Jupiter reached 1.5 AU before being pulled back to its current distance by the presence of Saturn. This scenario of inward-then-outward migration of Jupiter can explain the excitation and depletion of the asteroid belt and the abortion of the growth of Mars. Again, because chondritic planetesimals accrete until ∼ 3 My, what is important is that Jupiter did not invade the asteroid belt till that time. We acknowledge support by the French ANR, project number ANR-13–13-BS05-0003-01 projet MOJO (Modeling the Origin of JOvian planets). A.J. is grateful for the financial support from the European Research Council (ERC Starting Grant 278675-PEBBLE2PLANET) and the Swedish Research Council (grant 2014-5775). A.J. and B.B. thanks the Knut and Alice Wallenberg Foun- dation for their financial support. S. J. was supported by the European Research Council (ERC) Advanced Grant ACCRETE (contract number 290568). We are grateful to C. Dullemond and an anonymous reviewer for their constructive reviews. 8. References – Albar`ede, F. 2009. Volatile accretion history of the terrestrial planets and dynamic implica- tions. Nature 461, 1227-1233. – Alexander, C. M. O., Barber, D. J., Hutchison, R. 1989. The microstructure of Semarkona and Bishunpur. Geochimica et Cosmochimica Acta 53, 3045-3057. – Alexander, R., Pascucci, I., Andrews, S., Armitage, P., Cieza, L. 2014. The Dispersal of Protoplanetary Disks. Protostars and Planets VI 475-496. – Ali-Dib, M., Mousis, O., Petit, J.-M., Lunine, J. I. 2014. The Measured Compositions of Uranus and Neptune from their Formation on the CO Ice Line. The Astrophysical Journal 793, 9. – 24 – – Asphaug, E., Jutzi, M., Movshovitz, N. 2011. Chondrule formation during planetesimal ac- cretion. Earth and Planetary Science Letters 308, 369-379. – Bai, X.-N., Stone, J. M. 2010a. Dynamics of Solids in the Midplane of Protoplanetary Disks: Implications for Planetesimal Formation. The Astrophysical Journal 722, 1437-1459. – Bai, X.-N., Stone, J. M. 2010b. The Effect of the Radial Pressure Gradient in Protoplanetary Disks on Planetesimal Formation. The Astrophysical Journal 722, L220-L223. – Bai, X.-N. 2014. Hall-effect-Controlled Gas Dynamics in Protoplanetary Disks. I. Wind So- lutions at the Inner Disk. The Astrophysical Journal 791, 137. – Bailli´e, K., Charnoz, S., Pantin, ´E. 2015. Time evolution of snow regions and planet traps in an evolving protoplanetary disk. ArXiv e-prints arXiv:1503.03352. – Baker, L., Franchi, I. A., Wright, I. P., Pillinger, C. T. 2003. Aqueous Alteration on Ordinary Chondrite Parent Bodies- The Oxygen Isotopic Composition of Water.O. EGS - AGU - EUG Joint Assembly 11198. – Baruteau, C., Crida, A., Paardekooper, S.-J., Masset, F., Guilet, J., Bitsch, B., Nelson, R., Kley, W., Papaloizou, J. 2014. Planet-Disk Interactions and Early Evolution of Planetary Systems. Protostars and Planets VI 667-689. – Batygin, K., Laughlin, G. 2015. Jupiter’s Decisive Role in the Inner Solar System’s Early Evolution. ArXiv e-prints arXiv:1503.06945. – Birnstiel, T., Klahr, H., Ercolano, B. 2012. A simple model for the evolution of the dust population in protoplanetary disks. Astronomy and Astrophysics 539, AA148. – Bitsch, B., Crida, A., Morbidelli, A., Kley, W., Dobbs-Dixon, I. 2013. Stellar irradiated discs I. Equilibrium discs. Astronomy and and implications on migration of embedded planets. Astrophysics 549, A124. – Bitsch, B., Morbidelli, A., Lega, E., Crida, A. 2014a. Stellar irradiated discs and implications II. Accreting-discs. Astronomy and Astrophysics 564, on migration of embedded planets. AA135. – Bitsch, B., Morbidelli, A., Lega, E., Kretke, K., Crida, A. 2014b. Stellar irradiated discs and III. Viscosity transitions. Astronomy and implications on migration of embedded planets. Astrophysics 570, AA75. – Bitsch, B., Johansen, A., Lambrechts, M., Morbidelli, A. 2015. The structure of protoplane- tary discs around evolving young stars. Astronomy and Astrophysics 575, AA28. – Bitsch, B., Lambrechts, M., Johansen, A., 2015b. The growth of planets by pebble accretion in evolving protoplanetary discs. Astronomy and Astrophysics, in press. – 25 – – Bollard, J., Connelly, J. N., Bizzarro, M. 2014. The Absolute Chronology of the Early Solar System Revisited. LPI Contributions 1800, 5234. – Bridges, J. C., Changela, H. G., Nayakshin, S., Starkey, N. A., Franchi, I. A. 2012. Chondrule fragments from Comet Wild2: Evidence for high temperature processing in the outer Solar System. Earth and Planetary Science Letters 341, 186-194. – Brownlee, D., and 182 colleagues 2006. Comet 81P/Wild 2 Under a Microscope. Science 314, 1711. – Carrera, D., Johansen, A., Davies, M. B. 2015. How to form planetesimals from mm-sized chondrules and chondrule aggregates. ArXiv e-prints arXiv:1501.05314. – Ciesla, F. J. 2007. Outward Transport of High-Temperature Materials Around the Midplane of the Solar Nebula. Science 318, 613. – Chiang, E. I., Goldreich, P. 1997. Spectral Energy Distributions of T Tauri Stars with Passive Circumstellar Disks. The Astrophysical Journal 490, 368-376. – Connelly, J. N., Bizzarro, M., Krot, A. N., Nordlund, A., Wielandt, D., Ivanova, M. A. 2012. The Absolute Chronology and Thermal Processing of Solids in the Solar Protoplanetary Disk. Science 338, 651. – Cossou, C., Raymond, S. N., Hersant, F., Pierens, A. 2014. Hot super-Earths and giant planet cores from different migration histories. Astronomy and Astrophysics 569, AA56. – Cuzzi, J. N., Hogan, R. C., Paque, J. M., Dobrovolskis, A. R. 2001. Size-selective Concen- tration of Chondrules and Other Small Particles in Protoplanetary Nebula Turbulence. The Astrophysical Journal 546, 496-508. – Cuzzi, J. N., Hogan, R. C., Bottke, W. F. 2010. Towards initial mass functions for asteroids and Kuiper Belt Objects. Icarus 208, 518-538. – Davis, S. S. 2005. Condensation Front Migration in a Protoplanetary Nebula. The Astrophys- ical Journal 620, 994-1001. – Dullemond, C. P., Dominik, C., Natta, A. 2001. Passive Irradiated Circumstellar Disks with an Inner Hole. The Astrophysical Journal 560, 957-969. – Dullemond, C. P., van Zadelhoff, G. J., Natta, A. 2002. Vertical structure models of T Tauri and Herbig Ae/Be disks. Astronomy and Astrophysics 389, 464-474. – Dullemond, C. P. 2002. The 2-D structure of dusty disks around Herbig Ae/Be stars. I. Models with grey opacities. Astronomy and Astrophysics 395, 853-862. – 26 – – Espaillat, C., Muzerolle, J., Najita, J., Andrews, S., Zhu, Z., Calvet, N., Kraus, S., Hashimoto, J., Kraus, A., D’Alessio, P. 2014. An Observational Perspective of Transitional Disks. Proto- stars and Planets VI 497-520. – Fressin, F., Torres, G., Charbonneau, D., Bryson, S. T., Christiansen, J., Dressing, C. D., Jenkins, J. M., Walkowicz, L. M., Batalha, N. M. 2013. The False Positive Rate of Kepler and the Occurrence of Planets. The Astrophysical Journal 766, 81. – Garaud, P., Lin, D. N. C. 2007. The Effect of Internal Dissipation and Surface Irradiation on the Structure of Disks and the Location of the Snow Line around Sun-like Stars. The Astrophysical Journal 654, 606-624. – Guillot, T., Ida, S., Ormel, C. W. 2014. On the filtering and processing of dust by planetes- imals. I. Derivation of collision probabilities for non-drifting planetesimals. Astronomy and Astrophysics 572, AA72. – Guttler, C., Blum, J., Zsom, A., Ormel, C. W., Dullemond, C. P. 2009. The first phase of protoplanetary dust growth: The bouncing barrier. Geochimica et Cosmochimica Acta Supplement 73, 482. – Hansen, B. M. S. 2009. Formation of the Terrestrial Planets from a Narrow Annulus. The Astrophysical Journal 703, 1131-1140. – Hartmann, L., Calvet, N., Gullbring, E., D’Alessio, P. 1998. Accretion and the Evolution of T Tauri Disks. The Astrophysical Journal 495, 385-400. – Hayashi, C. 1981. Structure of the Solar Nebula, Growth and Decay of Magnetic Fields and Effects of Magnetic and Turbulent Viscosities on the Nebula. Progress of Theoretical Physics Supplement 70, 35-53. – Hubbard, A., Ebel, D. S. 2014. Protoplanetary dust porosity and FU Orionis outbursts: Solving the mystery of Earth’s missing volatiles. Icarus 237, 84-96. – Hueso, R., Guillot, T. 2005. Evolution of protoplanetary disks: constraints from DM Tauri and GM Aurigae. Astronomy and Astrophysics 442, 703-725. – Ida, S., Lin, D. N. C. 2008. Toward a Deterministic Model of Planetary Formation. IV. Effects of Type I Migration. The Astrophysical Journal 673, 487-501. – Isella, A., Natta, A. 2005. The shape of the inner rim in proto-planetary disks. Astronomy and Astrophysics 438, 899-907. – Jacobson, S. A., Morbidelli, A. 2014. Lunar and terrestrial planet formation in the Grand Tack scenario. Royal Society of London Philosophical Transactions Series A 372, 0174. – 27 – – Jacquet, E., Gounelle, M., Fromang, S. 2012. On the aerodynamic redistribution of chondrite components in protoplanetary disks. Icarus 220, 162-173. – Johansen, A., Oishi, J. S., Mac Low, M.-M., Klahr, H., Henning, T., Youdin, A. 2007. Rapid planetesimal formation in turbulent circumstellar disks. Nature 448, 1022-1025. – Johansen, A., Mac Low, M.-M., Lacerda, P., Bizzarro, M. 2015. Growth of asteroids, planetary embryos and Kuiper belt objects by chondrule accretion. Science Advances: e1500109 – Koenigl, A. 1991. Disk accretion onto magnetic T Tauri stars. The Astrophysical Journal 370, L39-L43. – Krot, A. N., and 15 colleagues 2009. Origin and chronology of chondritic components: A review. Geochimica et Cosmochimica Acta 73, 4963-4997. – Lambrechts, M., Johansen, A. 2014. Forming the cores of giant planets from the radial pebble flux in protoplanetary discs. Astronomy and Astrophysics 572, AA107. – Lambrechts, M., Johansen, A., Morbidelli, A. 2014. Separating gas-giant and ice-giant planets by halting pebble accretion. Astronomy and Astrophysics 572, AA35. – Libourel, G., Krot, A. N. 2007. Evidence for the presence of planetesimal material among the precursors of magnesian chondrules of nebular origin. Earth and Planetary Science Letters 254, 1-8. – Lin, D. N. C., Bodenheimer, P., Richardson, D. C. 1996. Orbital migration of the planetary companion of 51 Pegasi to its present location. Nature 380, 606-607. – Lodders, K. 2003. Solar System Abundances and Condensation Temperatures of the Elements. The Astrophysical Journal 591, 1220-1247. – Luu, T.H., Young, E.D., Gounelle, M., Chaussidon, M., 2015. Short time interval for con- densation of high-temperature silicates in the solar accretion disk. PNAS 112, 1298-1303. – Lynden-Bell, D., Pringle, J. E. 1974. The evolution of viscous discs and the origin of the nebular variables. Monthly Notices of the Royal Astronomical Society 168, 603-637. – Manara, C. F., and 11 colleagues 2013. X-shooter spectroscopy of young stellar objects. II. Impact of chromospheric emission on accretion rate estimates. Astronomy and Astrophysics 551, AA107. – Martin, R. G., Livio, M. 2012. On the evolution of the snow line in protoplanetary discs. Monthly Notices of the Royal Astronomical Society 425, L6-L9. – Martin, R. G., Livio, M. 2013. On the evolution of the snow line in protoplanetary discs - II. Analytic approximations. Monthly Notices of the Royal Astronomical Society 434, 633-638. – 28 – – Marty, B. 2012. The origins and concentrations of water, carbon, nitrogen and noble gases on Earth. Earth and Planetary Science Letters 313, 56-66. – Masset, F. S., Morbidelli, A., Crida, A., Ferreira, J. 2006. Disk Surface Density Transitions as Protoplanet Traps. The Astrophysical Journal 642, 478-487. – Masset, F. S., Casoli, J. 2009. On the Horseshoe Drag of a Low-Mass Planet. II. Migration in Adiabatic Disks. The Astrophysical Journal 703, 857-876. – Masset, F. S., Casoli, J. 2010. Saturated Torque Formula for Planetary Migration in Viscous Disks with Thermal Diffusion: Recipe for Protoplanet Population Synthesis. The Astrophys- ical Journal 723, 1393-1417. – McCubbin, F. M., Hauri, E. H., Elardo, S. M., Vander Kaaden, K. E., Wang, J., and Shearer, C. K. 2012. Hydrous melting of the martian mantle produced both depleted and enriched shergottites. Geology 40, 683-686. – McDonough, W.F., Sun, S.S, 1995. The composition of the Earth. Chemical Geology 120, 223-253 – Morbidelli, A., Chambers, J., Lunine, J. I., Petit, J. M., Robert, F., Valsecchi, G. B., Cyr, K. E. 2000. Source regions and time scales for the delivery of water to Earth. Meteoritics and Planetary Science 35, 1309-1320. – Morbidelli, A., Bottke, W. F., Nesvorn´y, D., Levison, H. F. 2009. Asteroids were born big. Icarus 204, 558-573. – Morbidelli, A., Nesvorny, D. 2012. Dynamics of pebbles in the vicinity of a growing planetary embryo: hydro-dynamical simulations. Astronomy and Astrophysics 546, AA18. – Nakamura, T., and 11 colleagues 2008. Chondrulelike Objects in Short-Period Comet 81P/Wild 2. Science 321, 1664. – O’Brien, D. P., Morbidelli, A., Levison, H. F. 2006. Terrestrial planet formation with strong dynamical friction. Icarus 184, 39-58. – O’Brien, D. P., Walsh, K. J., Morbidelli, A., Raymond, S. N., Mandell, A. M. 2014. Water delivery and giant impacts in the “Grand Tack” scenario. Icarus 239, 74-84. – Oka, A., Nakamoto, T., Ida, S. 2011. Evolution of Snow Line in Optically Thick Protoplan- etary Disks: Effects of Water Ice Opacity and Dust Grain Size. The Astrophysical Journal 738, 141. – Ormel, C. W., Klahr, H. H. 2010. The effect of gas drag on the growth of protoplanets. Analytical expressions for the accretion of small bodies in laminar disks. Astronomy and Astrophysics 520, A43. – 29 – – Paardekooper, S.-J., Mellema, G. 2006. Dust flow in gas disks in the presence of embedded planets. Astronomy and Astrophysics 453, 1129-1140. – Paardekooper, S.-J., Mellema, G. 2006b. Halting type I planet migration in non-isothermal disks. Astronomy and Astrophysics 459, L17-L20. – Paardekooper, S.-J., Baruteau, C., Crida, A., Kley, W. 2010. A torque formula for non- isothermal type I planetary migration - I. Unsaturated horseshoe drag. Monthly Notices of the Royal Astronomical Society 401, 1950-1964. – Paardekooper, S.-J., Baruteau, C., Kley, W. 2011. A torque formula for non-isothermal Type I planetary migration - II. Effects of diffusion. Monthly Notices of the Royal Astronomical Society 410, 293-303. – Pan, L., Padoan, P., Scalo, J., Kritsuk, A. G., Norman, M. L. 2011. Turbulent Clustering of Protoplanetary Dust and Planetesimal Formation. The Astrophysical Journal 740, 6. – Raymond, S. N., Quinn, T., Lunine, J. I. 2004. Making other earths: dynamical simulations of terrestrial planet formation and water delivery. Icarus 168, 1-17. – Raymond, S. N., Quinn, T., Lunine, J. I. 2006. High-resolution simulations of the final as- sembly of Earth-like planets I. Terrestrial accretion and dynamics. Icarus 183, 265-282. – Raymond, S. N., Quinn, T., Lunine, J. I. 2007. High-Resolution Simulations of The Final Assembly of Earth-Like Planets. 2. Water Delivery And Planetary Habitability. Astrobiology 7, 66-84. – Robert, F. 2003. The D/H Ratio in Chondrites. Space Science Reviews 106, 87-101. – Rubie, D. C., Jacobson, S. A., Morbidelli, A., O’Brien, D. P., Young, E. D., de Vries, J., Nimmo, F., Palme, H., Frost, D. J. 2015. Accretion and differentiation of the terrestrial planets with implications for the compositions of early-formed Solar System bodies and accretion of water. Icarus 248, 89-108. – Shakura, N. I., Sunyaev, R. A. 1973. Black holes in binary systems. Observational appear- ance.. Astronomy and Astrophysics 24, 337-355. – Shu, F. H., Shang, H., Lee, T. 1996. Toward an Astrophysical Theory of Chondrites. Science 271, 1545-1552. – Shu, F. H., Shang, H., Glassgold, A. E., Lee, T. 1997. X-rays and fluctuating X-winds from protostars.. Science 277, 1475-1479. – Shu, F. H., Shang, H., Gounelle, M., Glassgold, A. E., Lee, T. 2001. The Origin of Chondrules and Refractory Inclusions in Chondritic Meteorites. The Astrophysical Journal 548, 1029- 1050. – 30 – – Staff, J., Koning, N., Ouyed, R., Pudritz, R. 2014. Three-dimensional simulations of MHD disk winds to hundred AU scale from the protostar. European Physical Journal Web of Con- ferences 64, 05006. – Takeuchi, T., Lin, D. N. C. 2002. Radial Flow of Dust Particles in Accretion Disks. The Astrophysical Journal 581, 1344-1355. – Tanaka, H., Takeuchi, T., Ward, W. R. 2002. Three-Dimensional Interaction between a Planet and an Isothermal Gaseous Disk. I. Corotation and Lindblad Torques and Planet Migration. The Astrophysical Journal 565, 1257-1274. – Testi, L., and 10 colleagues 2014. Dust Evolution in Protoplanetary Disks. Protostars and Planets VI 339-361. – Villeneuve, J., Chaussidon, M., Libourel, G. 2009. Homogeneous Distribution of 26Al in the Solar System from the Mg Isotopic Composition of Chondrules. Science 325, 985. – Volk, K., Gladman, B. 2015. Consolidating and Crushing Exoplanets: Did it happen here?. ArXiv e-prints arXiv:1502.06558. – Youdin, A. N., Goodman, J. 2005. Streaming Instabilities in Protoplanetary Disks. The As- trophysical Journal 620, 459-469. – Walsh, K. J., Morbidelli, A., Raymond, S. N., O’Brien, D. P., Mandell, A. M. 2011. A low mass for Mars from Jupiter’s early gas-driven migration. Nature 475, 206-209. – Warren, P. H. 2011. Stable-isotopic anomalies and the accretionary assemblage of the Earth and Mars: A subordinate role for carbonaceous chondrites. Earth and Planetary Science Letters 311, 93-100. – Weidenschilling, S. J. 1977. Aerodynamics of solid bodies in the solar nebula. Monthly Notices of the Royal Astronomical Society 180, 57-70. – Weidenschilling, S. J. 1977b. The distribution of mass in the planetary system and solar nebula. Astrophysics and Space Science 51, 153-158. – Williams, J. P., Cieza, L. A. 2011. Protoplanetary Disks and Their Evolution. Annual Review of Astronomy and Astrophysics 49, 67-117.
1312.3959	1	1312	2013-12-13T21:22:39	Terrestrial Planet Formation in a protoplanetary disk with a local mass depletion: A successful scenario for the formation of Mars	[ "astro-ph.EP" ]	Models of terrestrial planet formation for our solar system have been successful in producing planets with masses and orbits similar to those of Venus and Earth. However, these models have generally failed to produce Mars-sized objects around 1.5 AU. The body that is usually formed around Mars' semimajor axis is, in general, much more massive than Mars. Only when Jupiter and Saturn are assumed to have initially very eccentric orbits (e $\sim$ 0.1), which seems fairly unlikely for the solar system, or alternately, if the protoplanetary disk is truncated at 1.0 AU, simulations have been able to produce Mars-like bodies in the correct location. In this paper, we examine an alternative scenario for the formation of Mars in which a local depletion in the density of the protosolar nebula results in a non-uniform formation of planetary embryos and ultimately the formation of Mars-sized planets around 1.5 AU. We have carried out extensive numerical simulations of the formation of terrestrial planets in such a disk for different scales of the local density depletion, and for different orbital configurations of the giant planets. Our simulations point to the possibility of the formation of Mars-sized bodies around 1.5 AU, specifically when the scale of the disk local mass-depletion is moderately high (50-75%) and Jupiter and Saturn are initially in their current orbits. In these systems, Mars-analogs are formed from the protoplanetary materials that originate in the regions of disk interior or exterior to the local mass-depletion. Results also indicate that Earth-sized planets can form around 1 AU with a substantial amount of water accreted via primitive water-rich planetesimals and planetary embryos. We present the results of our study and discuss their implications for the formation of terrestrial planets in our solar system.	astro-ph.EP	astro-ph	Terrestrial Planet Formation in a protoplanetary disk with a local mass depletion: A successful scenario for the formation of Mars A. Izidoro1,2,3 [email protected] N. Haghighipour4,5 [email protected] O. C. Winter1 and M. Tsuchida6 Abstract Models of terrestrial planet formation for our solar system have been suc- cessful in producing planets with masses and orbits similar to those of Venus and Earth. However, these models have generally failed to produce Mars-sized objects around 1.5 AU. The body that is usually formed around Mars' semima- jor axis is, in general, much more massive than Mars. Only when Jupiter and Saturn are assumed to have initially very eccentric orbits (e ∼ 0.1), which seems fairly unlikely for the solar system, or alternately, if the protoplanetary disk is truncated at 1.0 AU, simulations have been able to produce Mars-like bodies in the correct location. In this paper, we examine an alternative scenario for the 1 UNESP, Univ. Estadual Paulista - Grupo de Dinamica Orbital & Planetologia, Guaratinguet´a, CEP 12.516-410, Sao Paulo, Brazil 2Capes Foundation, Ministry of Education of Brazil, Bras´ılia/DF 70040-020, Brazil. 3University of Nice-Sophia Antipolis, CNRS, Observatoire de la Cote d'Azur, Laboratoire Lagrange, BP 4229, 06304 Nice Cedex 4, France. 4Institute for Astronomy and NASA Astrobiology Institute, University of Hawaii-Manoa, Honolulu, HI 96822, USA 5Institute for Astronomy and Astrophysics, University of Tubingen, 72076 Tubingen, Germany 6UNESP, Univ. Estadual Paulista , DCCE-IBILCE, Sao Jos´e do Rio Preto, CEP 15.054-000, Sao Paulo, Brazil – 2 – formation of Mars in which a local depletion in the density of the protosolar nebula results in a non-uniform formation of planetary embryos and ultimately the formation of Mars-sized planets around 1.5 AU. We have carried out exten- sive numerical simulations of the formation of terrestrial planets in such a disk for different scales of the local density depletion, and for different orbital con- figurations of the giant planets. Our simulations point to the possibility of the formation of Mars-sized bodies around 1.5 AU, specifically when the scale of the disk local mass-depletion is moderately high (50-75%) and Jupiter and Saturn are initially in their current orbits. In these systems, Mars-analogs are formed from the protoplanetary materials that originate in the regions of disk interior or exterior to the local mass-depletion. Results also indicate that Earth-sized planets can form around 1 AU with a substantial amount of water accreted via primitive water-rich planetesimals and planetary embryos. We present the re- sults of our study and discuss their implications for the formation of terrestrial planets in our solar system. Subject headings: Planets and satellites: formation; Methods: numerical 1. Introduction A major obstacle in developing a comprehensive model for the formation of the terrestrial planets of our solar system is the planet Mars. Despite more than two decades of efforts in explaining the formation of the inner solar system,(e.g., Chambers & Wetherill 1998; Agnor et al. 1999; Chambers 2001; Chambers & Wetherill 2001), and many sophisticated and high resolution computational simulations of the late stage of terrestrial planet formation (Raymond et al., 2004, 2006; O'Brien et al. 2006; Raymond et al. 2007, 2009), the formation of Mars is still a mystery. While modern simulations have been successful in producing a wide variety of terrestrial planets such as two to four planets in well-separated and stable orbits (Raymond et al., 2004, 2006; O'Brien et al. 2006; Raymond et al. 2007, 2009; Lykawka & Ito, 2013) and have been able to account for the origin of Earth's water and its accretion within the timescale consistent with radioactive chronometers (All´egre et al. 1995; Yin et al. 2002; Touboul et al. 2007; Marty 2012), they have not been able to form a Mars-analog at ∼1.5 AU. The body that is produced around the current Mars' semimajor axis is, in general, too massive when compared to the mass of Mars (See Raymond et. al., 2009 and the review by Morbidelli et al. 2012). An important factor in determining the outcome of the simulations of terrestrial planet formation is the orbital configuration of giant planets. As shown by many authors, the – 3 – mean-motion and secular resonances due to these objects strongly affect the dynamics of planetesimals and protoplanetary bodies which play a significant role in their radial mixing and may also result in their ejection from the system (Chambers 2001; Chambers & Cassen 2002; Levison & Agnor 2003; Raymond et al. 2004, 2006, 2009; Agnor & Lin 2012; Haghigh- ipour et al. 2013). Regarding the formation of Mars, Mars-analogs with appropriate masses have only been produced with Jupiter and Saturn initially in relatively eccentric orbits (e.g., eJup = eSat = 0.1, Thommes et al. 2008, Raymond et. al. 2009). Although the current orbits of Jupiter and Saturn do not carry high eccentricities, early orbital instabilities among these planets (before the beginning of the late stage of terrestrial planet formation) could have increased their eccentricities to high values (Lega et al., 2013). While such an increase in orbital eccentricity is a natural consequence of planets' instability, the damping of these eccentricities to their present values, and ensuring that the outer solar system would in fact develop and maintain its current features, has proven to be a difficult task (Raymond et al. 2009). The high eccentricities of Jupiter and Saturn will also affect the process of the delivery of water-rich planetesimals and planetary embryos to the accretion zone of Earth, causing many of these objects to be scattered out of the system, leaving Earth with less water-carrying material to accrete. Models of the origin of Earth's water suggest that a significant fraction of Earth's water was delivered to its accretion zone by water-rich planetesimals and planetary embryos from past 2 AU (Morbidelli et al., 2000; Raymond et al., 2004; 2006; 2007; 2009; O'Brien et al., 2006; Izidoro et al., 2013). An Eccentric Jupiter and Saturn will very quickly deplete the protoplanetary disk inside Jupiter's orbit from its water-carrying objects and causes planets formed around Earth's semimajor axis to have much less water than expected (Raymond et al., 2009). To overcome the difficulties of forming a Mars-sized planet at 1.5 AU, Hansen (2009) proposed that terrestrial planets might have formed in a narrow annulus of protoplanetary bodies around 1 AU. The results of the simulations by this author showed that they can successfully form Venus, Earth and Mars-analogs in a timescale consistent with isotopic constrains [see Kleine et al. (2009) for the constraints on the time of the formation of Earth, and Dauphas & Pourmand (2011) for the constraints on the time of the formation of Mars]. The small mass of Mars in this model is the natural consequence of its accretion from the material in the edge of the protoplanetary disk, where it is subsequently scattered out of the annulus and isolated from other planetary embryos, keeping a low-mass while other planetary embryos grow. It is important to note that the formation of Mercury-analog planets is still an issue in the models of terrestrial planet formation and was only studied in a very limited sample of the simulations presented in Hansen (2009). The success of the model by Hansen (2009) points to an interesting pathway for the – 4 – formation of terrestrial planets. This model requires as much as 2M⊕ of material to be confined in the region between 0.7 AU and 1.0 AU (equivalent to a disk with 3 times the minimum mass solar nebula) implying that a model for the formation of the planets in the inner solar system has to be able to present a scenario through which such a large amount of mass is confined in a small region. Recently Walsh et al (2011) proposed that it would be possible to create a narrow annulus of mass around 1 AU and also deliver water-carrying objects to the accretion zone of Earth, a topic that is not considered by Hansen (2009), if one allows for Jupiter and Saturn to migrate inward from their regions of formation to ∼ 1.5 AU at which point these planets change direction and migrate outward to a region close to their current orbits. In this model, the delivery of water-rich planetesimals and planetary embryos to the terrestrial zone is a natural consequence of the outward migration of giant planets. Known as the "Grand Tack" model, this scenario has been able to provide the region of terrestrial planets with the necessary amount of mass prior to the onset of their formation. Although the Grand Tack model has been able to produce Mars-analogs and account for the current architecture of the asteroid belt, it may not agree entirely with the models of the formation and migration of giant planets. Hydrodynamical simulations show that the migration in a protoplanetary disk is very sensitive to numerous disk parameters (Morbidelli & Crida, 2007) which are hard to constrain. In order to assess the validity of the Grand- Tack scenario on the formation and migration of Jupiter and Saturn, Pierens & Raymond (2011) simulated the growth of these planets and their dynamical evolution driven by tidal interactions with a gas-dominated disk. The results of simulations by these authors showed that the two-phase migration of giant planets could be a natural outcome of the solar system evolution, probably coinciding with the late phases of the dissipation of the solar nebula. In a recent study of the migration of giant planets in an evolved gaseous disk, however, D'angelo & Marzari (2012) have shown that only under favorable conditions and for a small region of the parameter space, the outward migration of Jupiter and Saturn can reach beyond 5 AU (as proposed by the Grand-Tack Model). It is important to note that if the outward migration of Jupiter and Saturn is not sufficiently fast, the efficiency of the Grand-Tack model decreases. The fast inward-then-outward migration of Jupiter and Saturn is the key to reproducing the architecture of the asteroid belt, with two distinct populations corresponding to the C- and S-type asteroids. Also, if the two-phase migration and its change of direction is not fast enough, the protoplanetary bodies in terrestrial planet region will be subject to the gravitational perturbation of giant planets for a long time which may result in their ejection from that region. The latter may subsequently reduce the mass in the planet-forming annulus to below its critical value disrupting the formation of terrestrial planets in that region. A planet-forming disk is a complex and dynamic environment. During the course of – 5 – planet formation, beginning from the stage when the dust particles coagulate to when the cores of giant planets are formed and terrestrial planets began their accretion, the physical properties of the disk are continuously subject to change. For instance, regions may appear where the disk is turbulent, or at places the density of the gas or solid material may be temporarily enhanced (e.g., Chiang & Goldreich 1997; Papaloizou & Nelson 2003; Laughlin et al. 2004; Garaud & Lin 2007). These processes affect the radial profile and the physical properties of the disk (e.g., temperature and surface density), and may deviate them from simple power-laws. As proposed by Jin et al. (2008), the radial variations in the ionization fraction of the disk, for instance, can separate the disk into an inner high-viscosity part and an outer low-viscosity region. The difference in viscosities in these two regions causes the material in the inner part of the disk to flow faster toward the star compared to the material in the outer part, creating a local minimum in the disk's mass distribution in the boundary between these two regions. As shown by these authors, during the evolution of the nebula around the Sun, two of such regions might have appeared. One region is narrow with a width of approximately 0.1 AU, varying between 1.3 AU and 3.4 AU, with a most probable location in the range of 1.3-2.4 AU. The second region is larger, with a width of ∼ 1 AU centered at about 1.6 AU. Jin et al. (2008) suggested that the appearance of such a local minimum in the disk could result in a non-uniform formation and distribution of planetary embryos around the orbit of Mars, and will ultimately result in the formation of Mars-sized objects. As previously discussed, Hansen (2009) and Walsh et al. (2011) showed that the key to the successful formation of Mars is to accumulate planet-forming material in a narrow region with an outer edge at 1 AU. As a consequence of limiting the protoplanetary disk to such a small annulus, the accretion and growth of an object will stop when it is scattered outside this region. The latter seems to present a pathway to the successful formation of Mars (or, Mars- analogs). In that respect, considering a local depletion in the protoplanetary disk around the Mars' location presents an alternative to the Grand Tack model for forming a Mars-analog using the Hansen-style disk truncation (Hansen, 2009) without the brief migration of Jupiter and Saturn in a gas-rich phase as used by Walsh et al., (2011). Another important aspect of this scenario is that unlike the model by Hansen (2009) which does not track the delivery of water to the terrestrial zone due to its initial adhoc set up of the protoplanetary disk, it presents a water-delivery mechanism that is also different from that of Walsh et al. (2011) who suggested that the water is delivered to around 1 AU very early during the outward migration of Jupiter and Saturn. In this paper, we examine this scenario by simulating the formation of terrestrial planets in a disk with a local minimum in surface density as proposed by Chambers & Cassen (2002) and Jin et al (2008). Our goal is to determine the range of the parameters for which a – 6 – protoplanetary disk with a locally depleted region can form a Mars-sized planet around 1.5 AU, and also produce a planetary system (fairly) consistent with the inner solar system (i.e., with Venus- and Earth-sized planets and with appropriate amount of water on Earth). We consider the local minimum to be around Mars' semimajor axis and use a non-uniform distribution of planetesimals and planetary embryos to simulate the final stage of terrestrial planet formation. Because during the evolution of the disk, the location of the local minimum may vary, we will carry out simulations for different values of the location and depth of this region. It is important to note that according to Jin et al. (2008), the inward flow of the material from outer regions of the nebula to its inner parts due to viscosity differences might have affected the composition of the disk material prior to the accretion of terrestrial planets. For instance, volatile-rich dust grains or planetesimals could have drifted inward much earlier than the onset of terrestrial planet formation, and have played an important role in the final mass and composition of these objects. The study of the changes in the properties of the nebula due to the inward flow of material and appearance of the gap is a complicated task that has not been accounted for in any rigorous ways in current protoplanetary gas-disk models. Such a modeling is also beyond the scope of this paper. In this study, we avoid these complications by focusing solely on the last stage of the accretion of planetesimals and planetary embryos. For simplicity, we also neglect the re-accumulation of the material that was originally removed from the depleted region of the disk during the evolution of the nebula, in particular the material that might have drifted inward into the terrestrial planet region. Due to the lack of a proper model for the evolution of the nebula, an initial condition for planetesimals and planetary embryos that takes into account such a re-accretion of the drifted material would be poorly constrained. We describe our model in section 2 and present the results of our simulations in section 3. Section 4 concludes this study by summarizing the results and discussing their implications for the formation of terrestrial planet in our solar system. 2. The Model As shown by Kokubo & Ida (1998, 2000), during the formation of terrestrial planets, the Runaway and Oligarchic growths of planetesimals and planetary embryos result in a bi-modal distribution of mass in the protoplanetary disk. Since in this study, our focus is on the late stage of terrestrial planet formation, in order to be consistent with the bi-modality of the disk, we consider a protoplanetary disk of planetesimals and planetary embryos with half of its mass from the planetesimals and the other half from planetary embryos. The disk – 7 – extends from 0.5 AU to 4 AU. In such a bimodal disk, the planetesimals provide dynamical friction which is necessary to damp the eccentricities and inclination of the planetary em- bryos embedded in the disk (O'Brien et al. 2006; Morishima et al., 2008). The individual planetesimals are considered to have a mass of 0.0025 Earth-masses and distributed with a surface density profile of r−3/2. These objects only interact with the planetary embryos and the giant planets, and do not see each other. The masses of the planetary embryos scale as M ∼ r3/2(2−α)∆3/2 (Kokubo & Ida 2000; Raymond et al. 2005, 2009) where α is a free pa- rameter and ∆ is the number of mutual Hill radii. We consider the embryo-to-planetesimal mass-ratio to be ∼ 8 around 1.5 AU (Raymond et al. 2009), and the disk surface density to have a radial profile of r−3/2 (i.e., α = 3/2) and be given by Σ(r) =  Σ1(r/1AU)−3/2 ; outside the depleted region, (1 − β)Σ1(r/1AU)−3/2 ; inside the depleted region. (1) In this equation, Σ1 = 8 g/cm2 and the parameter 0 < β ≤ 1 represents the scale of local mass-depletion. Figure 1 shows the distribution of planetary embryos and planetesimals for β = 50%. We chose the location of the local mass-depletion according to the model presented by Jin et al. (2008). However, because as shown by these authors, this location changes in time, in order to better explore the parameter space of the system, we considered two different regions of mass-depletion; one extending from 1.1 AU to 2.1 AU denoted as the disk model A, and one extending from 1.3 AU to 2.0 AU denoted as the disk model B. Table 1 shows these regions and their corresponding parameter β. We assume that in the beginning of our simulations, Jupiter and Saturn are fully formed, and carry out simulations considering different orbital configuration of these planets. We would like to recall that our goal is to present a model capable of forming terrestrial planets, in particular Mars, consistent with the inner planets of our solar system. Our approach is to follow the idea presented by Hansen (2009) (in which terrestrial planets are formed in a truncated disk of protoplanetary bodies), however, instead of appealing to the giant planet migration to create a truncated disk as in the Grand Tack scenario (Walsh et al. 2011), we consider the disk truncation to be a consequence of the disk's natural evolution. In that respect, the protoplanetary disk in the Hansen (2009) model can be considered as an extreme case of our disk model, extending from 0.7 AU to 1 AU with a depletion factor of β = 100% for the region beyond 1 AU. The fact that in our disk model, the region between 2 and 4 AU is initially populated by planetesimals and planetary embryos (Table 1) is a distinct characteristic of our model that makes the results more realistic and, unlike Hansen's simulations, allows for the formation of the asteroid belt and the delivery of water – 8 – to Earth from water-rich planetesimals and protoplanetary embryos (Morbidelli et al., 2000; Raymond et al., 2004). 3. Numerical Simulations We performed a total of 84 simulations considering two different sets of initial orbital configurations for the giant planets. As mentioned before, we assumed that at the beginning of the simulations these planets were fully formed and that the gas disk had been fully dissipated. In the first set of simulations, we considered Jupiter and Saturn to be in their current orbits. In the second set, we assumed their orbital elements to be similar to those proposed in the Nice Model, that is, aJup = 5.45 AU, aSat = 8.18 AU, and both planets to be initially in circular and co-planar orbits (Tsiganis et al., 2005; Gomes et al., 2005; Morbidelli et al., 2005). Each simulation started with 700-950 planetesimals and 110-190 planetary embryos, all initially in circular orbits. The planetary embryos were spaced from one another at distances of 3-6 mutual Hill radii (Kokubo & Ida, 2000). The initial orbital inclinations of all bodies were chosen randomly from the range of 10−4 to 10−3 degrees, and their mean anomalies were taken to be between 0◦ and 360◦. The arguments of periastrons and longitudes of ascending nodes of all objects were initially set to zero. For each value of the depletion parameter β and giant planet configuration, we considered at least three different, randomly generated initial conditions for planetesimals and planetary embryos. In view of the stochastic nature of this kind of simulations, we performed additional simulations using those parameters that produced planetary systems with features close to those of the solar system. We also used the water distribution model by Raymond et al. (2004) and considered planetary embryos and planetesimals inside 2.0 AU to be initially dry, the ones between 2 AU and 2.5 AU to carry 0.1% water, and the water contents of objects between 2.5 AU and 4 AU to be 5%. Using the hybrid integrator of the N-body integration package MERCURY (Chambers 1999), we integrated our systems for 500 Myr. The time-steps of integrations were set to 6 days. At the end of each simulation, we identified those that produced a potential Mars- and/or an asteroid belt analog, and integrated them for another 500 Myr. We would like to emphasize that this time of integration is considerably larger than the time of the simulations of terrestrial planet formation that is usually found in the literature. In general, most of the simulations of the late stage of terrestrial planet formation are carried out for 100 to 200 Myr (e.g., Chambers 2001, Raymond et al. 2004, 2006,2009, O'Brien et al. 2006, Hansen 2009, Walsh et al. 2011). We decided to carry out simulations for longer times because we noticed that after 200 Myr of integration, several of our simulations produced – 9 – small bodies (including two or three potential Mars-analogs) in orbits interior to 2 AU. To examine whether these systems would be stable, we extended the integrations to longer times. Except for planetesimal-planetesimal interactions, we allowed all objects to interact with one another and collide. It has been shown by Kokubo & Genda (2010) that the assumption of inelastic collisions in simulations of terrestrial planet formation has no important effect on the mass and orbital assembly of the final planets (if a collision results in fragmentation, the remaining material will be accreted in subsequent collisions). In a recent study, Chambers (2013) also found that in the simulations of terrestrial planet formation where fragmentation is considered, the final planetary systems are broadly similar to those of the simulations in which fragmentation is neglected. We therefore considered collisions to be perfectly inelas- tic, resulting in the complete merging of the two colliding bodies while conserving linear momentum. 4. Results In analyzing our results, we define an Earth-analog as a planet that is formed in the region of 0.75-1.25 AU and has a mass equal to or near that of Earth. Similarly, a Mars- analog is defined as a planet with a mass equal to or near that of Mars (∼ 0.3M⊕), formed between 1.25 AU and 2.0 AU. We also define a terrestrial planet as an object larger than 0.025-Earth masses (∼0.5 Mercury-mass) with a semimajor axis smaller than 2 AU. As mentioned in section 3, we carried out simulations for two different orbital architec- tures of Jupiter and Saturn (their current orbits as well as when they are initially in circular orbits as in the Nice model), and for different values of the scale and location of the local mass depletion. In the following, we present the results of each of these simulations, and compare the results with the current state of the inner planets of the solar system. 4.1. Jupiter and Saturn in their current orbits The first analysis of our results in this case indicates that in close to 50% of our simula- tions, the final planetary system includes at least one object with a mass smaller than half of the mass of Earth around 1.5 AU. Tables 2 and 3 show the statistics for this body in its final planetary system. Results in Table 2 correspond to the simulations of the disk model A (Table 1) and those in Table 3 correspond to the disk model B. The roman numbers in the first column of these tables refer to the first, second, and third set of initial conditions – 10 – in the simulations for a given mass-depletion coefficient. Figure 2 shows the snapshots of the dynamical evolution of a sample system corre- sponding to a simulation using the disk model A (Table 1) and for a depletion scale of 75% (simulation A-75%-II in Table 2). As shown by the bottom-right panel, integrations resulted in the formation of four planets in the region between 0.5 AU and 1.5 AU with the masses of (from left to right) 0.21, 0.95, 0.55 and 0.08 M⊕, respectively. The color of each body repre- sents its water-mass fraction. The least massive planet at 1.5 AU is an interesting low-mass object that has reached 90% of its mass in less than 2.5 Myr. The relatively fast formation of this body is in agreement with the timescale of the formation of Mars as suggested by Nimmo & Kleine (2007) and Dauphas & Pourmand (2011). As argued in the latter article, isotopic analyses require that Mars to have been accreted rapidly and reached approximately half of its present size in only 1.8+0.9 −1.0 Myr. Figure 3 shows the results of a simulation for similar mass-depletion scale and initial orbital configuration of Jupiter and Saturn as in Figure 2, but using the disk model B (simulation B-75%-III in Table 3). As shown by the bottom-right panel, in this simulation, three terrestrial planets are formed with masses equal to 0.7, 0.98, and 0.06 M⊕. Given the masses and distances of these planets to the central star, this system shows a close similarity to the Venus-Earth-Mars configuration in our solar system. However, unlike the results of the simulation of Figure 2, the planet around 1.5 AU in Figure 3 accreted more slowly and took longer than 80 Myr to reach 90% of its current mass. Figure 3 also demonstrates a case for which the extension of integrations to 1 Gyr was necessary. In this simulation, as shown by the panel corresponding to 200 Myr, a moderate number of small bodies, with two potential Mars-analogs, were formed with orbits between 1.0 AU and 2.0 AU. The extension of the integration to over 500 Myr revealed that only one of these bodies was stable, and the other one was ejected from the system after a close- encounter with the proto-Earth at ∼ 530 Myr. This is a typical outcome that was obtained in several of our simulations. In systems with high scales of depletion, in particular, the depleted region of the disk is initially populated by a large number of small bodies. If these objects are sufficiently small, their mutual gravitational interactions will be weak, and as long as no external perturbation affects their dynamics, their orbits will stay stable for long times. However, if the sizes of these objects allow for their actual interactions to be strong, they may scatter each other, in which case the orbits of some of these bodies may become unstable. One characteristic of the two Mars-analogs shown in Figures 2 and 3 is their moderate orbital inclinations relative to the orbit of the largest body in their systems. We note that we compute orbital inclinations relative to the orbit of the largest planet in the inner part of – 11 – the system in order to be able to compare our results with those in Hansen (2009). Unlike the orbit of Mars that has an inclination of ∼ 2◦ with respect to the orbital plane of Earth, the two Mars-analog planets in Figures 2 and 3 have inclinations in the range of 2◦-13◦ with respect to the plane of the orbit of the largest planet in the region between 0.5 AU and 1.5 AU. Our simulations show that moderate inclinations appear mainly for planets smaller than 0.1 M⊕, probably due to the weak action of the dynamical friction on these bodies. This enhanced inclination has also been reported by Hansen (2009) in his simulations of the formation of terrestrial planets where this author showed that Mars analogs are formed with similar inclinations in almost 50% of the cases. In our simulations, the moderate inclinations of planetesimals and planetary embryos between 1.5 AU and 2.0 AU can be attributed to the effect of the ν16 resonance. This resonance, as well as the secular resonance ν6 occur when the periods of the nodal and apsidal precessions of the orbit of a small body become equal to those of Saturn. As shown in Figure 4, the ν16 resonance significantly increases the inclinations of objects within the first 10 Myr. In addition to the ν16 resonance, Figures 2 and 3 also show the strong effects of the mean- motion resonances with Jupiter as well as the effect of the ν6 secular resonance with Saturn (see the upper right panel in each figure). As shown here, these resonances increase the eccentricities of objects in their vicinities causing many of these bodies to be either scattered out of the system, or collide with the Sun, Jupiter, or Saturn. (Gladman et al., 1997, Levison & Agnor 2003, Raymond et al. 2006, Haghighipour et al. 2013). Although these processes remove a significant amount of water-carrying objects from the protoplanetary disk (see middle-right panels of Figures 2 and 3), simulations indicate that despite a depletion scale of 75%, terrestrial planets can still form around 1 AU and carry a significant amount of water. As a point of comparison, The lower limit of the amount of the Earth's water is 5×10−4 of its water-mass fraction (Raymond et al., 2004). The planets around 1 AU in Figures 2 and 3 carry higher amounts. 4.1.1. The effect of the scale of mass-depletion As expected, the growth of planetary embryos in the mass-depleted region is propor- tional to the amount of mass that is available in that region. In other words, it depends on the depletion factor β. For large values of β (e.g., 75% which corresponds to a total initial mass of ∼ 0.25M⊕ in the region of 1.3-2.0 AU in disk model B), the small amount of mass in the depleted region combined with the effect of ν6 and ν16 resonances has a negative impact on the rapid formation of Mars-analogs during the first few Myr of integration. The increase in the orbital inclinations causes planetesimals and planetary embryos to be in dif- – 12 – ferent orbital planes, decreasing the rate of their collisions and consequently the efficiency of their growth. The orbital excitation of these bodies (as a result of both increasing their eccentricities and inclinations), causes many of these objects to be scattered out of the sys- tem devoiding the region from material necessary to form small planets and Mars-analogs (Haghighipour et al., 2013). For smaller values of β where the mass of the depleted region is larger (e.g., 20% which corresponds to a total initial mass of ∼ 0.82M⊕ in the region of 1.3-2.0 AU in disk model B), the collision and growth of planetary embryos results in the formation of larger objects. For instance, simulations with β = 20% routinely produced planets around 1.5 AU up to 5 times larger than Mars. This result is consistent with those in Raymond et al. (2009). The results of our simulations also show that embryos that originated in the non-depleted region of the disk and were scattered into the depleted region seem to have a better chance of growing to the Mars size and maintaining long-term stability. These objects are considerably larger than the planetary embryos native to the depleted region and less responsive to the perturbing effects of ν6 and ν16 resonances. The latter is primarily due to the effect of the dynamical friction which strongly damps the eccentricities and inclinations of these objects. For instance, both Mars-analogs of Figures 2 and 3 were formed at larger distances (∼ 2.7 AU and 3.1 AU, respectively) and were scattered into the region of ∼1.5 AU as a result of successive close-encounters with other planetary embryos and the effect of resonances (Figure 5). The formation of Mars-analogs outside the region around 1.5 AU and their subsequent scattering into this region has also been observed in the simulations by Hansen (2009) and Walsh et al. (2011). As shown by these authors, in those simulations in which a Mars-analog was formed, this object was accreted in the region around 1 AU and was scattered to the vicinity of the Mars' orbit at 1.5 AU. In that region, the object maintained an isolated orbit, and as a result, keeping a low mass. In our simulations, Mars-analogs are formed both in the region interior to 1.5 AU (e.g., around 1 AU) where they are scattered outward, and exterior to this region at 2.0-2.5 AU, where their interactions with other embryos and resonances causes them to scatter inward. This is a significant result in the sense that it is consistent with the formation of Mars as a stranded planetary embryo, and implies that depending on the original region of the accretion of this object, its internal composition will be different. Those Mars-analogs that are scattered inward from the regions beyond 2.0-2.5 AU appear in 25% of our simulations and in general carry more water than those scattered outward from the inner part of the protoplanetary disk (the remaining 75%). Depending on the regions of their origin, these objects will also have different D/H ratios than that of Mars (Drouart et al. 1999, Morbidelli – 13 – et al. 2000). At the moment, the primordial value of the D/H ratio of Mars' water is unknown (Lunine et al., 2003), however, evidence from Martian meteorites suggests that the initial D/H value for Martian interior water is consistent with a chondritic D/H composition (Usui et al., 2012). The formation of Mars as an inward-scattered planetary embryo can also explain the higher D/H ratio for Mars's water relative to that of Earth (Lunine et al., 2004). The scale of depletion has also an important effect on the radial mixing of the bodies and the material-content of the final planets. Radial mixing is driven by the mutual interactions between protoplanetary bodies as well as the perturbations due to giant planets. Embryo- embryo and embryo-planetesimal interactions increase the eccentricity of these objects and scatter them into other regions of the disk. Mean-motion and secular resonances with Jupiter and Saturn (e.g., in the range of 1.8 AU to 3.7 AU when these planets are in their current orbits) also play important roles, in particular in the delivery of water-carrying asteroid to terrestrial planets (Raymond et al., 2004). Our simulations indicate that when the depletion scale is low, the embryos inside the depleted region are initially larger and as a result, have stronger interactions with their neighboring embryos and planetesimals. In this case, radial mixing is more efficient and the material-contents of the final planets show greater diversity in their origin. Figures 6 and 7 show samples of the results. Figure 6 corresponds to the simulations using the disk model A, and Figure 7 shows the results of the simulations using the disk model B. The size of each body in these figures corresponds to its relative physical size scaled as M 1/3. However, it is not to scale on the x-axis. The color of each object represents the relative contribution of material from different parts of the disk. The eccentricity of each planet is represented by a horizontal line corresponding to its variation in heliocentric distance over the semimajor axis. As shown in these figures, in systems with low mass- depletion scales, the water-mass fraction of the final bodies are larger indicating that in such systems, the final terrestrial planets tend to carry more water (left plots in Figure 8). As shown in Figures 6-8, when the scale of depletion is 100%, terrestrial planets are formed interior to ∼1.3 AU with little or no amount of water. However, a scale of depletion of 75% or smaller produces Earth-analogs with mean water-mass fractions consistent with the value expected for Earth (Table 3). 4.1.2. The effect of the location of the mass-depletion As mentioned earlier, we carried out simulations for different locations of the disk local mass-depletion. Results indicated that the position of the inner edge of this region plays a significant role in the mass and orbital assembly of the final terrestrial planets. A comparison – 14 – between the results shown in Figures 6 and 7 indicates that for instance, in simulations where the depleted region started at 1.1 AU (Figure 6), the planets formed around 1 AU were consistently smaller than Earth. However, in simulations of Figure 7, where the inner edge of the depleted region is at 1.3 AU, these planets are relatively larger. This can be explained noting that for similar mass-depletion factors, in the simulations of Figure 6 (using the disk model A), the amount of the mass available for accretion by embryos around 1.0 AU is smaller and as a result, the final planets have lower masses compared to those in Figure 7 where simulations were run for the disk model B. Figure 9 shows the mean planet mass for all our simulations. The top panel in this figure includes all the planets in the system whereas in the bottom panel only those formed in the region between 0.75 AU and 1.25 AU have been included. As shown by this figure, the disk model B is more efficient in forming Earth-sized planets. It is interesting to note that in contrast to our model, Hansen (2009) and Walsh et al. (2011) used a narrower disk that was truncated at 1 AU and were still able to form Earth-sized planets around that region. However, in their disk models, the initial amount of mass between 0.7 AU and 1 AU is 2 M⊕ which is almost twice higher than the amount of mass between 0.5 AU and 1 AU in our protoplanetary disks. The outer edge of the depleted region (at 2.0 AU or 2.1 AU, see Table 1), did not seem to have a strong influence on the mass and orbital architecture of the final terrestrial planets. However, this boundary may play an important role in the efficiency of the delivery of water-carrying objects to the Earth's accretion zone. A mass-depleted region extending to distances well beyond 2 AU (e.g., along all the asteroid belt) can hinder the delivery of water to Earth. We carried out simulations considering a depleted region extending from 1.5 AU to 2.5 AU and assuming Jupiter and Saturn initially in their current orbits. Results showed that for depletion scales of 50% and 75%, simulations produced Earth-analogs that were mainly dry. 4.2. Jupiter and Saturn in Circular Orbits To examine the role that the current orbital eccentricities of Jupiter and Saturn play in the formation of a Mars-analog and the final assembly of terrestrial planets, we also carried out simulations assuming Jupiter and Saturn to be initially in circular orbits. We considered the initial orbital elements of these planets to be identical to those in the Nice model (Tsiganis et al. 2005) and carried out simulations for 1 Gyr. Figure 10 shows the results of one of such simulations. As shown here, four planets are formed with semimajor axes smaller than 2 AU. Among these planets is an object similar to Mars with a mass – 15 – of ∼ 0.24M⊕ at ∼1.35 AU. Although the formation of this planet can be considered as a success in forming Mars-analogs, there is another terrestrial planet in this system with a mass of 0.7M⊕ in a stable orbit at 1.7 AU which has made its final planetary configuration inconsistent with the current architecture of the solar system. The lack of consistency between the final planetary orbits and the current orbits of terrestrial planets was observed in the results of all simulations in which Jupiter and Saturn were initially in circular orbits. Figure 11 shows the mass and semimajor axes of the final bodies. As shown here, in general, results are very different from those of the previous simulations where Jupiter and Saturn were initially in their current orbits (Figures 2, 3 and 15). Unlike those simulations, when giant planets were assumed to be in circular orbits, the final planetary systems did not contain a Mars-sized planet around 1.5 AU. This was irrespective of the scale of the mass-depletion as well as the disk model. Results show that in all these simulations, only one planet with a mass smaller than 0.3M⊕ was formed around 1.5 AU (Figure 11). The lack of success in forming Mars-sized planets around 1.5 AU in the simulations in which Jupiter and Saturn were initially in circular orbits is in contrasts with the results of the simulations by Walsh et al. (2011) who also considered Jupiter and Saturn to be in circular orbits and were still able to produce Mars-analogs. The reason for this discrepancy can be found in the mass-distribution in the disk models used by these authors. At the end of the inward migration of Jupiter and Saturn, these authors set the total initial mass available for accretion in the terrestrial zone to ∼ 2 M⊕. They also considered that the region of the accretion of terrestrial planets is confined to a narrower region from ∼0.7 AU to 1 AU. After the inward-then-outward migration of Jupiter and Saturn, the total mass of the remaining planetesimals and protoplanetary embryos orbiting from 1.5 AU to 4 AU, in their model becomes very small. As a result, these objects do not contribute significantly to the growth of the planetary bodies in the terrestrial zone. In our simulations, however there always exist a significant amount of mass (∼ 1.5 M⊕) beyond 2.5 AU. When the orbits of Jupiter and Saturn are considered to be circular, their interactions with the protoplanetary disk are minimal and as a result, the disk maintains a large portion of its original mass. Consistent with previous studies in which Jupiter and Saturn were assumed to be in circular orbits (e.g., Wetherill 1996; Raymond et al. 2005, Kokubo et al. 2006), a more massive disk in these simulations produces more massive objects. Figure 12 shows the mean mass of the planets formed in our simulations considering different giant planet configurations. The smaller perturbations of the giant planets on the protoplanetary disk in simulations in which these planets are considered to be in circular orbits also affects the water contents of the terrestrial planets. In these simulations, the radial mixing of the protoplanetary bodies – 16 – is more effective and results in forming planets with higher contents of water (Figure 8, see also Raymond et al. 2009). That is because more water-carrying objects from the outer part of the protoplanetary disk maintain their orbits for longer times. A comparison between the result of the simulations shown in Figures 3 and 10, which correspond to the same disk model, indicates that in the simulations where Jupiter and Saturn are in their current orbits, the protoplanetary disk loses a large portion of its water-carrying objects as a result of the stronger interactions of these giant planets with the planetesimals and planetary embryos. We expect comparable results to be obtained using an updated version of the Nice model (Levison et al., 2011) as well. 4.2.1. The effect of the scale and location of the mass-depletion When Jupiter and Saturn are considered in circular orbits, our simulations do not show a clear correlation between the results and the different values of the scale and location of the mass-depletion (Figures 11-13). As discussed above, when the giant planets are in circular orbits, the protoplanetary disk is perturbed only weakly, and it maintains a higher fraction of its original mass for longer times. The effect of the scale of the mass depletion in this case vanishes probably because a significant part of the material in the neighborhood of the depleted region enters in the depleted area during the evolution of the system. This material comes mainly from the outer part of the protoplanetary disk as the gravitational effects of Jupiter and Saturn are weak, and only remove a small fraction of the disk bodies. 5. Comparison with solar system and other simulations To compare the results of our simulations with the current orbital architecture of ter- restrial planets in our solar system and with the results of simulations by other authors, we calculated the radial mass concentration statistics (RMC) and angular momentum deficit (AMD) of the final planetary systems of our simulations. The value of RMC varies with the semimajor axes of planets and is given by (Chambers 1998, 2001; Raymond et al. 2009) RMC = Max j=1 mj In this equation, mj and aj are the mass and semimajor axis of planet j, and N is the number of final bodies. The AMD of a system represents a measure of the deviation of the actual orbital angular momentum of the planets from the total angular momentum of the PN j=1 mj[ log10 (a/aj) ]2! . PN (2) – 17 – system, had the planets been in circular and co-planar orbits. Following Laskar (1997), we use equation (3) to calculate this value, AMD = PN j=1hmj√aj (cid:16)1 − cos ij p1 − ej 2(cid:17)i . (3) j=1 mj√aj PN In this equation, ej is the orbital eccentricity of planet j, and ij is its inclination with respect to an invariant plane. Figures 6 and 7 show the values of the RMC and AMD of their corresponding planetary systems. As a point of comparison, the inner planets of the solar system are also shown. We recall that in these simulations, Jupiter and Saturn were initially in their current orbits. We would also like to note that, similar to the planet V proposed by Chambers (2007), some of our simulations produced small planets (mainly unaccreted embryos) past 2 AU. However, we did not take these planets into account when calculating RMC and AMD values. Figures 13 and 14 show the mean values of the RMC and AMD of our simulations for different scales of the mass-depletion, disk models, and giant planet configuration. The circular solid points in these figures represent the means of the RMC and AMD calculated from a set of planetary systems produced by at least three different simulations with the same mass-depletion scale. The vertical bars on each point represent the lower and upper values of the RMC and AMD in the sample over which their means were calculated. The values of the RMC and AMD of Mercury, Venus, Earth and Mars (hereafter MVEM) are also shown in these figures. A comparison between these values with the values of the RMC and AMD of the sample results shown in Figures 6 and 7, and the mean values shown in Figures 13 and 14 indicates that the mean RMC of our planetary systems are significantly smaller than that of MVEM. The higher value of the RMC of MVEM is mainly due the orbital proximity of these planets and the comparable masses of Venus and Earth. The lower values of the mean RMC in our results indicate that our simulations form planets in more widely spread orbits than the separation of the orbits of MVEM. Our results also indicate that, except for a few planetary systems, the mean AMD of our simulations is, in general, slightly higher than the AMD of MVEM. This is mainly due to the initial distribution of the total mass of the disk among planetesimals and planetary embryos which can be adjusted to result in systems with lower AMDs. As shown by O'Brien el al. (2006), it is possible to obtain lower AMDs by increasing the initial value of the planetesimal/embryo mass-ratio while keeping similar distribution of disk mass between the embryo and planetesimal populations (e.g., assigning 50% of the disk mass to planetesimals and 50% to the planetary embryos). The higher values of the planetesimal/embryo mass- – 18 – ratio will enhance the dynamical friction which will then be more effective in reducing the value of AMD. The dynamical friction can also be enhanced if a residual population of very small objects still exist in the inner solar system after the terrestrial planet formation is completed (Schlichting et al., 2012). To compare our results with previous studies, we first consider the simulations of the formation of terrestrial planets by Raymond et al. (2009). A comparison between the values of the RMC of our systems and those from these authors indicates that only when in their simulations, Jupiter and Saturn are initially in eccentric orbits (e ∼ 0.1), the values of their RMC are similar to those in our best models in which a Mars-analog is formed. The reason is that in their simulations, the eccentric orbits of Jupiter and Saturn cause these planets to have strong interactions with the objects in the outer part of the disk, removing the majority of them from the system in a short time and creating an edge for the disk at ∼2 AU (few million years, Raymond et al. 2009). As the material is removed from the disk, the ν6 secular resonance shifts interior to 2.0 AU (Gomes, 1997; Haghighipour et al., 2013) and continues to remove material from the inner part of the disk. This causes the disk to develop a local mass-depleted region similar to the one considered in our disk models. We note that although the radial mass concentrations in all our simulations are smaller than that of the terrestrial planets, in general, they are higher than those in most of the simulations by Raymond et. al (2009). A comparison of our results with those in the simulations by Hansen (2009) and Walsh et al. (2011) shows that the values of their RMC are higher than the average value found in our results. The reason can be attributed to the extent of the protoplanetary disks in those studies. Both in simulations by Hansen (2009) and Walsh et al. (2011), the total mass necessary for the accretion of terrestrial planets was distributed over a small region between 0.7 AU to 1 AU. Such a narrow distribution will naturally result in the formation of planetary systems with high values of RMC. Recall that in our simulations, similar to the most of the previous studies of terrestrial planet formation in the solar system (e.g., Raymond et al. 2009), the protoplanetary disk extends from 0.5 AU to 4 AU. As shown by our results, it is possible to form planetary systems with higher values of RMC in such disks when Jupiter and Saturn are considered in their current orbits by using a large mass-depletion scale (Figure 13). The results of our simulations indicated that Mars-sized planets have a better chance of forming around 1.5 AU in systems where Jupiter and Saturn are initially in their current orbits and when the mass-depletion factor has a moderately large value (50% - 75%). For the purpose of extending our analysis, we identified this subset of our simulations, and following Raymond et al. (2009) and recalling our definitions of Mars- and Earth-analogs in Section – 19 – 4, we evaluated their success, quantitatively, in producing • a Mars-analog with a mass smaller than 0.3M⊕ in less than 10 Myr (with half-accretion in less than 2.7 Myr), • an Earth-analog with a mass larger than 0.7M⊕ and a water-mass fraction larger than 5 × 10−4 in 30-150 Myr, • a system of terrestrial planets with AMD < 0.0036 (twice the MVEM value), and • no stranded embryos in the asteroid belt with masses bigger than 0.05M⊕. The results are presented in Tables 4 and 5. As shown in these tables, different combinations of the scale and location of the disk local mass-depletion can result in systems that in many occasions partially, and in some specific cases almost entirely satisfy the above-mentioned requirements. When the scale of the local mass-depletion in the disk is moderately large (e.g., β = 50% − 75%), the final systems are more successful in meeting the above mentioned criteria, especially when the depleted region is considered to be between 1.3 AU and 2 AU (Table 5). 5.1. Formation of asteroid belt analogs One important feature of our solar system that imposes strong constraints on the models of terrestrial planet formation and solar system dynamics is the asteroid belt. Any model for the formation of the inner planets has to also be able to account for the existence of small bodies between 2.1 AU and 3.2 AU, and their orbital architecture. In our simulations, several systems showed signs of asteroid belt analogs. For instance, in the simulations of Figure 3, three planetesimals remained stable in the region of 2 AU to 3.5 AU, for the duration of the integration (1 Gyr). This simulation also produced an Earth-analog (∼ 1M⊕) around 1 AU in less than 150 Myr, which is consistent with the timescale of the formation of the Earth- Moon system (Jacobsen 2005; Touboul et al 2007). The innermost planet in this system is a Venus-analog. The final number of planetesimals in the region between 2 AU and 3.5 AU, i.e. their long-term (> 100 Myr) stability, depends on the interactions between these objects and their neighboring planetary embryos. The latter itself depends on the interaction of giant planets with the protoplanetary disk. Figure 15 shows the final distribution of surviving bodies in those simulations that had full or relative success in producing Mars-analogs around 1.5 AU – 20 – (see also Table 5). As shown here, many embryos with long-term stable orbits are formed in the asteroid belt. Quantitatively, in 14 simulations out of 18 shown in Figure 15, either no embryo was left in the asteroid belt region, or if there was any embryo, its mass was smaller than 0.05M⊕. These results agree with the findings of Raymond et al. (2009) who showed that in systems where giant planets are in slightly excited orbits, some embryos may maintain their orbits in the asteroid belt region for 200 Myr of integration. These authors also showed that it is improbable that during the formation of terrestrial planets, a Mars- sized embryo could have been stranded in the asteroid belt and maintained its orbit for 100 Myr (Raymond et al., 2009; Brasser et al., 2011). Such an embryo would have disturbed the orbits of other bodies in its vicinity, creating a gap in the protoplanetary disk that is not observed in the present day asteroid belt1. Our results also agree with this finding. In regard to the latter, we also analyzed the results of our simulations where Jupiter and Saturn were initially in circular orbits. We found that planetary embryos that survived in the asteroid belt were, in general, much larger than Mars, reaching to a mass equal to almost half of the mass of Earth (see Figure 11). Figure 15 also shows that in the simulations in which the depletion scale is 50%, the orbital elements of the population of asteroids in the region of the asteroid belt are closely similar to those of the real population of the main belt. In the simulations where the depletion scale is 75%, on the other hand, asteroids were produced mainly between 2.8 AU and 3.2 AU, with a few bodies also between 2.2 AU and 2.8 AU. We would like to note that because these simulations are very time-consuming, we adopted a moderate resolution. Higher resolution simulations may produce results that may show better agreement with the structure of the asteroid belt. An interesting result depicted by the top panels of Figure 15 is the appearance of a dual- mass population among the surviving planetesimals in the asteroid belt. One population is at 0.0025 Earth-masses outside the depleted region in both panels. The second population is around 0.00125 Earth-masses in the left panel and 0.000625 Earth-masses in the right panel. The appearance of such a dual-mass population is due to the initial assumption of the existence of a local depletion in the disk. The small objects in the asteroid belt are primarily native to the depleted area whereas the larger ones originated outside of the depletion. The native objects are also small in number (only 4 in each of the 9 simulations shown in Figure 15). The small mass and number of the native planetesimals is due to the effect of resonances which causes many of these bodies to be scattered out of the system. As a result, the asteroid belt analogs that are formed in our simulations do not carry a 1Note that this gap is different from those that are due to mean-motion resonances with giant planets. – 21 – mass-gradient distribution. The small number of the final surviving planetesimals in the region of the asteroid belt in our simulations also points to a strong clearing process during which many objects were scattered out of the system. In several of our simulations, the initial number of planetesimals in the asteroid belt was over 400. However, on average no more than 3 planetesimals survived in the end of the simulations (e.g. Figure 2 and 3). This value corresponds to a very small fraction of the initial number of planetesimals in that region which is in agreement with previous studies indicating that a vast majority of the mass of the asteroid belt was removed during the evolution of the solar system and the formation of terrestrial planets (Petit el al. 2001) 6. Conclusion and Discussion We studied the late stage of the accretion of terrestrial planets in a disk of protoplanetary bodies with a locally depleted region. Following Jin et al (2008), we considered that a depletion in the density of the protosolar nebula will result in a non-uniform formation of planetesimals and planetary embryos, and studied the effects of the scale and location of such a local depletion on the mass, water-content, and final orbital assembly of terrestrial planets. For simplicity, we neglected the reallocation/redistribution of mass from the depleted region into the inner parts of protoplanetary disk due to the early radial in-flow of material (Jin et al., 2008). This enabled us to avoid complications that would rise from considering a more complex compositional and mass-size gradient for the initial distribution of solids in our protoplanetary disk models. As expected, the final mass and orbital assembly of the planets in our simulations were strongly affected by the initial orbital configurations of Jupiter and Saturn. The results of integrations showed a clear distinction between the outcome of simulations in which the orbits of the giant planets were considered to be initially circular and those in which Jupiter and Saturn had slightly eccentric orbits. Similar to previous studies (O'Brien et al., 2006; Raymond et al., 2009), when giant planets were in circular orbits, simulations were system- atically unsuccessful in forming planetary systems with Mars-analogs at 1.5 AU. In this case, even in simulations with high scale of depletion, the effect of the local disk-depletion seemed to vanish and the terrestrial planets that formed around Mars' orbit were more massive than 0.3 M⊕. These results suggest that if Jupiter and Saturn were initially in circular orbits, as in the Nice Model (Tsiganis et al. 2005) or a recent model by Levison et al. (2011), a disk with a local mass-depletion may be able to form a Mars-sized planet around 1.5 AU only if an additional mechanism removes material from the disk in the region of the asteroid – 22 – belt. However, such a mechanism has to also ensure that the delivery of water-carrying planetesimals to the region of Earth accretion will stay efficient. When Jupiter and Saturn were placed in their current orbits, the interactions of these planets with the protoplanetary disk played an important role in forming Mars-analogs. Results of our simulations indicated that in this case, a significant portion of the disk material is removed from its outer regions creating a favorable condition for Mars-analogs to form around 1.5 AU when the depletion factor has a moderately large value of 50% to 75%. In these simulations, when considering the disk model B, in addition to forming a Mars-analog, our models were able to deliver sufficient amount of water to their corresponding Earth- analogs in ∼ 40% of the cases (Table 5). The Mars-analogs in these simulations formed as an embryo that was scattered from the non-depleted region (either the inner or the outer part) into the depleted area. Simulations showed that a depleted region around 1.5 AU with a high scale of depletion increases the stability of this planet, although caution has to be taken since larger values of the mass-depletion will have negative impact on the radial mixing of planetesimals and planetary embryos, and can deprive Earth from having sufficient amount of water. Embryos scattered from the other parts of the disk to the depleted region are substantially more massive than the native planetary embryos, and as a result will have a higher chance of being stable. These embryos also tend to grow slowly since not much material will be available in the depleted area to accommodate their collisional growth. The results of our simulations also showed that as expected, the final semimajor axes and eccentricities of Jupiter and Saturn were slightly different from their initial values. Such changes in the orbital elements of the giant planets have also been reported by Chambers & Wetherill (2001), and Raymond et al. (2004), and are the result of the interaction of these bodies with planetesimals and planetary embryos, and the subsequent decrease in the mass of the protoplanetary disk due to the scattering and ejection of these objects from the system. In this study, at the end of the simulations in which Jupiter and Saturn were initially in their current orbits, the semimajor axis and eccentricity of Jupiter decreased by an average of 0.045 AU and 0.035, respectively. However, no significant changes occurred to the semimajor axis of Saturn. In most case, the final semimajor axis of this planet was within a range of ±0.03 AU from its initial value. The eccentricity of Saturn decreased by 0.035. The above-mentioned damping of the final eccentricities of the giant planets suggests that as the formation of terrestrial planets approaches its final stage (or after these planets are fully formed), a mechanism has to exist to ensure that the orbits of Jupiter and Saturn will have eccentricities similar to their current values. One approach is to consider the initial orbital eccentricities of these planets, at the beginning of each simulation, to be large. – 23 – Raymond et al. (2009) considered, for instance, an initial value of 0.1 for the eccentricities of both Jupiter and Saturn, and showed that the final eccentricities of these planets will be close to their current values (0.05). Whether such large orbital eccentricities are possible is, of course, a matter of debate. As shown by Lega et al. (2013), early dynamical instabilities between Jupiter and Saturn that may occur at the end of the lifetime of the gas disk could increase the orbital eccentricities of these bodies. However, Raymond et al. (2009) have presented a list of arguments as to why Jupiter and Saturn could not have evolved into such a dynamical state. Results of our study suggest that modest values of orbital eccentricities for Jupiter and Saturn are required to ensure the efficient formation of Mars-analogs. Given the subsequent damping of the eccentricities of these planets, our model also requires a late- stage dynamical instability between Jupiter and Saturn, such as that proposed by Tsiganis et al., (2005) to raise the orbital eccentricities of these planets to their current values. At the beginning of each simulation, to create a low-density region, we scaled down the masses of planetesimals and planetary embryos inside the depleted area. However, depending on the scale of the mass-depletion, the efficiency of the growth of planetary embryos from planetesimals (Kokubo & Ida, 2000) inside the depleted area may be so low that during the Runaway and Oligarch phases, either no planetary embryo is formed, or they may not grow to large sizes. This implies that even if we had not scaled down the masses of the objects (e.g. this region was initially populated only by planetesimals), the results would not have changed and Mars-analogs would have formed in the same fashion as in our simulations. As mentioned before, Mars-sized objects originated from the regions outside the mass-depleted area and because they had more mass to accrete, they were larger than the native bodies. Once inside the deplete area, these objects had a better chance of growing to the Mars' size because they dominated this region gravitationally and accreted (or scattered) smaller, native bodies more efficiently. The results of the simulations that were successful in forming Mars-analogs around 1.5 AU also show that unlike simulations with low or no local mass-depletion (e.g., Chambers 2001; Raymond et al. 2009), a scale of depletion of 50% to 75% is crucial for building planetary systems with stronger radial mass concentrations. Although our simulations did not produce RMC values as high as that of the current terrestrial planets, their results point to potential pathways for improving models of terrestrial planet formation using a local mass-depletion. The correlation between the scale of mass-depletion and the RMC value also indicates that the strong radial mass concentration of terrestrial planets and the low mass of Mars are two characteristics of the solar system that are deeply connected. Despite the success of our model in forming planets similar to Mars and ensuring the delivery of sufficient amount of water to Earth, none of our simulations was able to reproduce – 24 – all the features of the inner solar system. The hardest constraints to satisfy was the fast formation of Mars. As suggested by Nimmo & Kleine (2007) and Dauphas & Pourmand (2011), the measurements of the Hf/W ratio in the Martian mantle point to a timescale of 0-10 million years for the formation of the core of Mars, and a time of slightly less than 2 million years during which Mars reaches to 50% of its current mass. Although the formation of Mars-analogs around the current orbit of Mars was successful in many of our simulations (Table 4 and 5), the time during which these objects grew to 50% of their masses was in general (in more than 92% of the simulations of Table 4, and in more than 95% in the simulations of Table 5) much longer than 2 million years. It is important to emphasize that this longer time of Mars accretion should not be considered a weakness for our model. The 0-10 million years timescale for the growth of the core of Mars, as suggested by Nimmo & Kleine (2007), is based on the analysis of a limited number of shergottite–nakhlite–chassignite (SNC) meteorites that are assumed to have formed the bulk of Mars mantle. Also, the time of the growth of Mars to 50% of its current mass as suggested by Dauphas & Pourmand (2011) is based on the assumption that the mass-evolution of planetary embryos during the oligarchic state can be modeled by a simple analytic solution if a uniform size of planetesimals is adopted (Chambers, 2006). However, using N-body simulations, Morishima et al. (2013) have shown that such an analytic approximation for the mass-evolution of planetary embryos does not agree with the final masses of these objects at the late stage of the oligarchic growth. These authors also showed that for a disk with a minimum mass solar nebula, the timescale for the formation of Mars maybe much longer than those derived from the Hf-W chronology – a result that is consistent with our finding as well. When comparing the time of the formation of Mars obtained from numerical simula- tions such as those presented here, with those obtained from cosmochemical studies, another important factor that needs to be taken into consideration is that unlike analytical ap- proximations of the mass-evolution of embryos, numerical simulations include the effects of Jupiter and Saturn that are assumed to have been fully formed in the beginning of simu- lations. This is probably 1-3 Myr after the formation of the first solids (Raymond et al. 2009), the calcium-aluminum inclusions [CAIs, dated at 4.568 billion years ago (Bouvier et al. 2007)]. As shown by many authors [see e.g., Haghighipour & Scott (2012) and references therein], these planets have profound effects on the distribution and growth of planetary embryos during their own formation and after they are fully formed. The perturbations of these planets cause many embryos to be scattered out of the planet-forming region – a process that will have important consequences on the final masses of terrestrial planets. The analytical treatment of the growth of embryos, however, does not take the effects of giant planets into account. – 25 – Numerical simulations of the formation of terrestrial planets such as those presented here and by Morishima et al. (2013), consistently form Mars at timescales larger than that suggested by Nimmo & Kleine (2007) and Dauphas & Pourmand (2011). As argued by Morishima et al. (2013) and Kobayashi & Dauphas (2013), this time can be reduced if Mars grew by the accretion of small fragments and pebbles, or the surface density of the protoplanetary disk around the orbit of Mars is locally enhanced. This latter scenario seems to be consistent with the assumption of a local mass-depletion in the proto-solar nebula due to the variation in gas viscosity as suggested by Jin et al (2009)2. Such a depletion may cause the material to accumulate outside the depleted area, creating regions where the surface density of the disk is locally enhanced. These density-enhanced regions may produce large embryos of the size of Mars in short timescales during the Runaway and Oligarchic growth phases (Kokubo & Ida, 2000). In our simulations, Mars-sized bodies do in fact form in the regions interior or exterior to the local mass-depletion where the disk surface density may be locally enhanced (Jin et al., 2008). These objects are then scattered into the region of mass-depletion where they maintain a stable orbit for a long time. This implies that if Mars were a planetary embryo that grew in a density-enhanced region, it could have formed in a short timescale consistent with the finding of Dauphas & Pourmand (2011). However, we recall that as mentioned in Section 1, we do not use such local surface density enhancements when generating the initial structure of our protoplanetary at the beginning of our simulations. One important aspect of our simulations is the treatment of the two-body collisions. In all our simulations, we assumed that collisions were perfectly inelastic and resulted in the complete merging of the two impacting bodies. We also assumed that no water was lost during an impct (Marty & Yokochi 2006) or hydrodynamic escapes (Matsui & Abe 1986), and the total amount of water in the final body would be equal to the sum of the water contents of the colliding objects. In more realistic simulations, the loss of volatile materials during an impact has to be taken into account (Genda & Abe 2005, Canup & Pierazzo 2006). In closing we would like to emphasize that the works of Hansen (2009) and Walsh et al. (2011) provided significant context and motivation for the development of this study. (2011) combined the narrow-disk idea of Hansen (2009) with giant planet Walsh et al. 2We would like to note that the results presented by Jin et al (2009) and their suggested mechanism for generating a local mass depletion in the disk depend highly on the choice of some poorly constrained parameters such as the rotational velocity of the cloud's core and the viscosity of the gaseous disk. In fact, regions of local mass depletion may appear during the evolution of a disk for a verity of reasons, and may have long enough lifetimes to affect the formation and growth of planetesimals, and appear as local mass depletions in the protoplanetary disk as well. – 26 – migration, and developed a model for the formation of Mars and delivery of water to Earth in which most of the water delivered to terrestrial zone is from primitive asteroids that were scattered inward during the outward migration of giant planets. We, however, showed that it would be possible to form a low-mass planet around Mars' semimajor axis (and deliver sufficient amount of water to Earth) without considering drastic inward migration for Jupiter and Saturn, if the protosolar nebula did have a locally depleted region around the Mars' location. The efficiency of the formation of Mars-sized planets around 1.5 AU will be high when the giant planets are considered to be farther out than initially assumed in the Grand Tack model (e.g., close to their current locations) and in slightly eccentric orbits (e.g., close to their current values). In this model, contrary to what Jin et al. (2008) proposed, that a Mars-analog would form from the embryos native to the depleted region, our simulations show that this object is formed outside the depleted area and is scattered into this region as a result of interacting with giant planets and other planetary embryos. Also, although our model uses the same narrow-disk idea as in Hansen (2009) and Walsh et al (2011), our scenario for water-delivery to Earth is completely different. In our model, similar to previous studies such as those by Raymond et al. (2009), water comes from the asteroid belt region due to the perturbing effects of giant planets. This will possibly result in the accretion of a different fraction of primitive asteroids by the forming Earth, as well as a different time for their accretion. More precise measurements are needed to determine the fraction of the material from different parts of the solar nebula that contributed to the formation of terrestrial planets and the origin of the Earth's water. Such measurements will have profound effects on constraining models of terrestrial planet formation in our solar system and can be used to differentiate between existing scenarios. We would like to thank the referee, Kevin Walsh, for his very constructive comments that greatly improved this manuscript. We are also thankful to Alessandro Morbidelli for his carefully reading of our paper and his helpful comments. AI and OCW would like to thank Rafael Sfair for his assistance with the computing cluster that was used to run part of these simulations. AI and OCW would also like to acknowledge financial support from the Brazilian National Research Council (CNPq), Coordenacao de Aperfeicoamento de Pessoal de N´ıvel Superior (CAPES), and FAPESP - Sao Paulo State Funding Agency, proc. 2011/08171-3. AI wishes to thank the Institute for Astronomy at the University of Hawaii for their kind hospitality during the course of this project. NH acknowledges support from the NASA Astrobiology Institute under Cooperative Agreement NNA09DA77A at the Institute for Astronomy, University of Hawaii, and the Alexander von Humboldt Foundation. NH is also thankful to the Computational Physics group at the Institute for Astronomy and Astrophysics, University of Tubingen for their kind hospitality during the course of this project. – 27 – REFERENCES Agnor, C. B., Canup, R. M., & Levison, H. F. 1999, Icarus, 142, 219 Agnor, C. B., & Lin, D. N. C. 2012, ApJ, 745, 143 All`egre, C. J., Manh`es, G., Goupel, C. 1995, Geochim. Cosmochim. Acta, 59, 1445 Bouvier, A., Blichert-Toft, J., Moynier, F., Vervoort, J. D., & Albar`ede, F. 2007, Geochim. Cosmochim. Acta, 71, 1583 Brasser, R., & Morbidelli, A. 2011, A&A, 535, A41 Canup, R. M. & Pierazzo, E. 2006, LPI conference series, 37, 2146 Chambers, J. E. 1998, Earth Moon and Planets, 81, 3 Chambers, J. E. 1999, MNRAS, 304, 793 Chambers, J. E. 2001, Icarus, 152, 205 Chambers, J. E. 2007, Icarus, 189, 386 Chambers, J. 2006, Icarus, 180, 496 Chambers, J. E. 2013, Icarus, 224, 43 Chambers, J. E., & Cassen, P. 2002, Meteoritics and Planetary Science, 37, 1523 Chambers, J. E., & Wetherill, G. W. 1998, Icarus, 136, 304 Chambers, J. E., & Wetherill, G. W. 2001, Meteoritics and Planetary Science, 36, 381 Chiang, E. I., & Goldreich, P. 1997, ApJ, 490, 368 Craddock, R. A., & Howard, A. D. 2002, Journal of Geophysical Research (Planets), 107, 5111 D'Angelo, G., & Marzari, F. 2012, ApJ, 757, 50 Dauphas, N., & Pourmand, A. 2011, Nature, 473, 489 Drouart, A., Dubrulle, B., Gautier, D., & Robert, F. 1999, Icarus, 140, 129 – 28 – Eisner, J. A., Hillenbrand, L. A., White, R. J., Akeson, R. L., & Sargent, A. I. 2005, ApJ, 623, 952 Garaud, P., & Lin, D. N. C. 2007, ApJ, 654, 606 Genda, H. & Abe, Y. 2005, Nature, 433, 842 Gladman, B. J., Migliorini, F., Morbidelli, A., et al. 1997, Science, 277, 197 Goldreich, P., Lithwick, Y., & Sari, R. 2004, ApJ, 614, 497 Gomes, R. S. 1997, AJ, 114, 396 Gomes, R., Levison, H. F., Tsiganis, K., & Morbidelli, A. 2005, Nature, 435, 466 Haghighipour N., Izidoro A., Winter O. C. 2013, submitted Hansen, B. M. S. 2009, ApJ, 703, 1131 Horner, J., Mousis, O., & Hersant, F. 2007, Earth Moon and Planets, 100, 43 Izidoro, A., de Souza Torres, K., Winter, O. C., & Haghighipour, N. 2013, ApJ, 767, 54 Jacobsen, S. B. 2005, Annual Review of Earth and Planetary Sciences, 33, 531 Jin, L., Arnett, W. D., Sui, N., & Wang, X. 2008, ApJ, 674, L105 Kobayashi, H., & Dauphas, N. 2013, Icarus, 225, 122 Kokubo, E., & Genda, H. 2010, ApJ, 714, L21 Kokubo, E., & Ida, S. 1998, Icarus, 131, 171 Kokubo, E., & Ida, S. 2000, Icarus, 143, 15 Kokubo, E., Kominami, J., & Ida, S. 2006, ApJ, 642, 1131 Laughlin, G., Steinacker, A., & Adams, F. C. 2004, ApJ, 608, 489 Laskar, J. 1997, A&A, 317, L75 Lega, E., Morbidelli, A., & Nesvorn´y, D. 2013, MNRAS, 431, 3494 Levison, H. F., & Agnor, C. 2003, AJ, 125, 2692 Levison, H. F., Morbidelli, A., Tsiganis, K., Nesvorn´y, D., & Gomes, R. 2011, AJ, 142, 152 – 29 – Sofia Lykawka, P., & Ito, T. 2013, arXiv:1306.3287 Lin, D. N. C., & Papaloizou, J. 1986, ApJ, 309, 846 Lunine, J. I., Chambers, J., Morbidelli, A., & Leshin, L. A. 2003, Icarus, 165, 1 Marty, B. & Yokochi, R. 2006, Rev. mineral. Geochem., 62, 421 Marty, B. 2012, Earth and Planetary Science Letters, 313, 56 Morbidelli, A., Chambers, J., Lunine, J. I., et al. 2000, Meteoritics and Planetary Science, 35, 1309 Morbidelli, A., & Crida, A. 2007, Icarus, 191, 158 Morbidelli, A., Levison, H. F., Tsiganis, K., & Gomes, R. 2005, Nature, 435, 462 Morbidelli, A., Lunine, J.I., O'Brien, D.P., Raymond, S.N., Walsh, K.J. 2012, Annual Re- views of Earth and Planetary Science, in press. Morishima, R., Golabek, G. J., & Samuel, H. 2013, Earth and Planetary Science Letters, 366, 6 Morishima, R., Schmidt, M. W., Stadel, J., & Moore, B. 2008, ApJ, 685, 1247 Nimmo, F., & Kleine, T. 2007, Icarus, 191, 497 O'Brien, D. P., Morbidelli, A., & Levison, H. F. 2006, Icarus, 184, 39 Papaloizou, J. C. B., & Nelson, R. P. 2003, MNRAS, 339, 983 Petit, J.-M., Morbidelli, A., & Chambers, J. 2001, Icarus, 153, 338 Pierens, A., & Raymond, S. N. 2011, A&A, 533, A131 Raymond, S. N., O'Brien, D. P., Morbidelli, A., & Kaib, N. A. 2009, Icarus, 203, 644 Raymond, S. N., Quinn, T., & Lunine, J. I. 2004, Icarus, 168, 1 Raymond, S. N., Quinn, T., & Lunine, J. I. 2005, ApJ, 632, 670 Raymond, S. N., Quinn, T., & Lunine, J. I. 2006, Icarus, 183, 265 Raymond, S. N., Quinn, T., & Lunine, J. I. 2007, Astrobiology, 7, 66 Schlichting, H. E., Warren, P. H., & Yin, Q.-Z. 2012, ApJ, 752, 8 – 30 – Thommes, E., Nagasawa, M., & Lin, D. N. C. 2008, ApJ, 676, 728 Touboul, M., Kleine, T., Bourdon, B., Palme, H., & Wieler, R. 2007, Nature, 450, 1206 Tsiganis, K., Gomes, R., Morbidelli, A., & Levison, H. F. 2005, Nature, 435, 459 Usui, T., Alexander, C. M. O., Wang, J., Simon, J. I., & Jones, J. H. 2012, Earth and Planetary Science Letters, 357, 119 Walsh, K. J., Morbidelli, A., Raymond, S. N., O'Brien, D. P., & Mandell, A. M. 2011, Nature, 475, 206 Wetherill, G. W. 1996, Icarus, 119, 219 Yin, Q., Jacobsen, S. B., Yamashita, K., et al. 2002, Nature, 418, 949 This preprint was prepared with the AAS LATEX macros v5.2. – 31 – 1.000 0.100 0.010 s s a M h t r a E 0.001 0 0.5 1 2 1.5 Semi-major Axis (AU) 2.5 3 3.5 4 4.5 Fig. 1.- Initial distribution of 154 embryos (black) and 973 planetesimals (red) considering a mass-depletion of 50% extending from 1.3 AU to 2.0 AU. The masses of planetesimals are smaller than 0.003 Earth masses. – 32 – 0.0 Myr 1 Myr 50 Myr 200 Myr 500 Myr 1 Gyr y t i c i r t n e c c E y t i c i r t n e c c E y t i c i r t n e c c E 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Semimajor Axis (AU) Semimajor Axis (AU) 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.001 0.001 0.001 0.001 0.001 0.001 0.01 0.01 0.01 0.01 0.01 0.01 0.1 0.1 0.1 0.1 0.1 0.1 Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Fig. 2.- Snapshots of the formation and dynamical evolution of planets in a disk with a depletion of 75% extending from 1.1 AU to 2.1 AU. Jupiter and Saturn are in their current orbits. The size of each body corresponds to its relative physical size and is scaled as M 1/3. However, it is not to scale on the x-axis. The color-coding represents the water-mass fraction of the body. – 33 – 0.0 Myr 1 Myr 50 Myr 200 Myr 500 Myr 1 Gyr y t i c i r t n e c c E y t i c i r t n e c c E y t i c i r t n e c c E 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Semimajor Axis (AU) Semimajor Axis (AU) 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.001 0.001 0.001 0.001 0.001 0.001 0.01 0.01 0.01 0.01 0.01 0.01 0.1 0.1 0.1 0.1 0.1 0.1 Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Fig. 3.- Snapshots of the formation and dynamical evolution of planets in a disk with a depletion of 75% extending from 1.3 AU to 2.0 AU. Jupiter and Saturn are in their current orbits. The size of each body corresponds to its relative physical size and is scaled as M 1/3. However, it is not to scale on the x-axis. The color-coding represents the water-mass fraction of the body. – 34 – 1 : 3 1 : 2 1.00 Myr ν 16 n o i t a n i l c n I 40.00 35.00 30.00 25.00 20.00 15.00 10.00 5.00 0.00 0 1 2 3 4 5 Semi-major Axis (AU) 0.005 0.06 0.3 0.5 0.91.1 1.5 2 Mass (Earth Mass) Fig. 4.- Graph of the orbital inclination versus semimajor axis for the first 1 Myr of the simulation of Figure 3. As shown here, ν16 and mean-motion resonances with Jupiter increase the inclinations of planetesimals and planetary embryos. This orbital excitation is more pronounced in the depleted region (1.3-2.1 AU). – 35 – Mars-analog scattered inward Mars-analog scattered outward 3.00 2.00 1.00 1.00 0.10 0.01 ] U A [ s x a i j r o a m m e S i ) ⊕ M ( s s a M 1000 10000 100000 1e+06 Time (years) 1e+07 1e+08 1e+09 Fig. 5.- Graphs of the semimajor axes and masses of two typical Mars-analogs formed in simulations where Jupiter and Saturn were initially in their current orbits. Similar to the results shown here, Mars-analogs in all our simulations were formed in the inner or outer non-depleted regions and were scattered into the depleted area. – 36 – 20% - III AMD=0.0080 AMD=0.0080 RMC=40.72 RMC=40.72 20% - II 20% - I 50% - III 50% - II 50% - I 75% - III 75% - II 75% - I 100% - III 100% - II 100% - I AMD=0.0035 AMD=0.0035 RMC=40.12 RMC=40.12 AMD=0.0012 AMD=0.0012 RMC=41.55 RMC=41.55 AMD=0.0049 AMD=0.0049 RMC=34.9 RMC=34.9 AMD=0.0008 AMD=0.0008 RMC=51.8 RMC=51.8 AMD=0.0047 AMD=0.0047 RMC=63.7 RMC=63.7 AMD=0.0033 AMD=0.0033 RMC=66.9 RMC=66.9 AMD=0.0020 AMD=0.0020 RMC=50.5 RMC=50.5 AMD=0.0012 AMD=0.0012 RMC=53.3 RMC=53.3 AMD=0.0014 AMD=0.0014 RMC=67.4 RMC=67.4 AMD=0.0098 AMD=0.0098 RMC=47.4 RMC=47.4 AMD=0.0054 AMD=0.0054 RMC=54.7 RMC=54.7 RMC=0.0018 RMC=0.0018 RMC=89.9 RMC=89.9 SS 0 0.5 1 1.5 2 Semimajor Axis (AU) 0.5 AU 1.0 AU 1.5 AU 2.0 AU 2.5 AU 3.0 AU 3.5 AU 4.0 AU Source Region Fig. 6.- Final masses and orbital configurations of planets in simulations of disk model A with different mass-depletion scales. The depleted area is from 1.1 AU to 2.1 AU (shown in gray). As shown by the Roman numbers on the vertical axis, each simulation was run for three different distributions of planetesimals and planetary embryos. The size of each body corresponds to its relative physical size scaled as M 1/3. However, it is not to scale on the x-axis. The color of each object represents the relative contribution of material from different parts of the disk. The eccentricity of each planet is represented by its variation in heliocentric distance over the semimajor axis (horizontal bars). – 37 – 20% - III 20% - II 20% - I 50% - III 50% - II 50% - I 75% - III 75% - II 75% - I 100% - III 100% - II 100% - I AMD=0.0062 AMD=0.0062 RMC=33.3 RMC=33.3 AMD=0.0041 AMD=0.0041 RMC=45.6 RMC=45.6 AMD=0.0007 AMD=0.0007 RMC=69.0 RMC=69.0 AMD=0.0023 AMD=0.0023 RMC=41.8 RMC=41.8 AMD=0.0076 AMD=0.0076 RMC=58.4 RMC=58.4 AMD=0.0051 AMD=0.0051 RMC=95.4 RMC=95.4 AMD=0.0023 AMD=0.0023 RMC=55.7 RMC=55.7 AMD=0.0029 AMD=0.0029 RMC=40.4 RMC=40.4 AMD=0.0040 AMD=0.0040 RMC=51.1 RMC=51.1 AMD=0.0023 AMD=0.0023 RMC=58.0 RMC=58.0 AMD=0.0065 AMD=0.0065 RMC=55.6 RMC=55.6 AMD=0.0179 AMD=0.0179 RMC=48.7 RMC=48.7 RMC=0.0018 RMC=0.0018 RMC=89.9 RMC=89.9 SS 0 0.5 1 1.5 2 Semimajor Axis (AU) 0.5 AU 1.0 AU 1.5 AU 2.0 AU 2.5 AU 3.0 AU 3.5 AU 4.0 AU Source Region Fig. 7.- Final masses and orbital configurations of planets in simulations of disk model B with different mass-depletion scales. The depleted area is from 1.3 AU to 2.0 AU (shown in gray). As shown by the Roman numbers on the vertical axis, each simulation was run for three different distributions of planetesimals and planetary embryos. The size of each body corresponds to its relative physical size scaled as M 1/3. However, it is not to scale on the x-axis. The color of each object represents the relative contribution of material from different parts of the disk. The eccentricity of each planet is represented by its variation in heliocentric distance over the semimajor axis (horizontal bars). – 38 – n o i t c a r f s s a m r e t a W n o i t c a r f s s a m r e t a W 1e-01 1e-02 1e-03 1e-04 1e-05 1e-06 1e-01 1e-02 1e-03 1e-04 1e-05 1e-06 Nice Model - Disk A Earth 20 50 75 100 Scale of depletion (%) Nice Model - Disk B Earth 20 50 75 100 Scale of depletion (%) n o i t c a r f s s a m r e t a W n o i t c a r f s s a m r e t a W 1e-01 1e-02 1e-03 1e-04 1e-05 1e-06 1e-01 1e-02 1e-03 1e-04 1e-05 1e-06 Current Orbits - Disk A Earth 20 50 75 100 Scale of depletion (%) Current Orbits - Disk B Earth 20 50 75 100 Scale of depletion (%) Fig. 8.- Graphs of the mean value of the water/mass fraction of the planets formed inside 1.3 AU as a function of the scale of the disk mass-depletion for different disk models and giant planets configuration. The panels on the left show the results of the simulations considering Jupiter and Saturn to be initially in their current orbits. The ones on the right show the results for which Jupiter and Saturn were initially as in the Nice Model (2005). – 39 – Current Orbits Disk A Disk B 20 50 75 100 Scale of depletion (%) 0.75<ap<1.25 AU Current Orbits Disk A Disk B 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 ) ⊕ M ( s s a M t l e n a P n a e M ) ⊕ M ( s s a M t l e n a P n a e M 20 50 75 100 Scale of depletion (%) Fig. 9.- Graphs of the mean value of planet mass as a function of the scale of mass-depletion in different disk models. Jupiter and Saturn were initially in their current orbits. The top panel shows the mean planet mass for planets formed interior to 2 AU and the bottom panel shows similar quantity for planets formed around 1 AU (0.75< ap < 1.25 AU). – 40 – 0.0 Myr 1 Myr 50 Myr 200 Myr 500 Myr 1 Gyr y t i c i r t n e c c E y t i c i r t n e c c E y t i c i r t n e c c E 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 Semimajor Axis (AU) Semimajor Axis (AU) 0.00001 0.00001 0.00001 0.00001 0.00001 0.00001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.001 0.001 0.001 0.001 0.001 0.001 0.01 0.01 0.01 0.01 0.01 0.01 0.1 0.1 0.1 0.1 0.1 0.1 Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Water Mass Fraction Fig. 10.- Snapshots of the formation and dynamical evolution of planets in a disk with a depletion of 75% extending from 1.3 AU to 2.0 AU. Jupiter and Saturn were initially in circular orbits corresponding to the Nice Model. The size of each body corresponds to its relative physical size and is scaled as M 1/3. However, it is not to scale on the x-axis. The color-coding represents the water-mass fraction of the body. – 41 – β=100% 75% 50% 35% 20% 2 1.5 3 Semi-major Axis (AU) 2.5 3.5 4 β=100% 75% 50% 20% 2 1.5 3 Semi-major Axis (AU) 2.5 3.5 4 1.000 s s a M h t r a E 0.100 0.010 0 0.5 1 1.000 s s a M h t r a E 0.100 0.010 0 0.5 1 Fig. 11.- Graphs of the mass-semimajor axis distribution for the surviving bodies (open circles) in the simulations considering the disk model A (top) and disk model B (bottom) for all scales of depletion. Jupiter and Saturn were initially in circular orbits. The solid triangles represent the masses of Mars, Earth, Venus, and Mercury. The Surviving planetesimals are not shown. – 42 – Disk A Current Orbits Nice Model 20 50 75 100 Scale of depletion (%) Disk B Current Orbits Nice Model 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 2 1.8 1.6 1.4 1.2 1 0.8 0.6 0.4 0.2 ) ⊕ M ( s s a M t l e n a P n a e M ) ⊕ M ( s s a M t l e n a P n a e M 20 50 75 100 Scale of depletion (%) Fig. 12.- Graphs of the mean values of the masses of planets in terms of the scale of mass-depletion for different disk models and giant planets configurations. The top panel corresponds to the disk model A and the bottom panel is for the disk model B. – 43 – C M R C M R 100 90 80 70 60 50 40 30 20 100 90 80 70 60 50 40 30 20 C M R C M R 100 90 80 70 60 50 40 30 20 100 90 80 70 60 50 40 30 20 Current Orbits - Disk A MVEM 20 50 75 100 Scale of depletion (%) Current Orbits - Disk B MVEM 20 50 75 100 Scale of depletion (%) Nice Model - Disk A MVEM 20 50 75 100 Scale of depletion (%) Nice Model - Disk B MVEM 20 50 75 100 Scale of depletion (%) Fig. 13.- Graphs of the mean values of RMC as a function of the mass-depletion scale for different disk models and giant planets configuration. The left column shows the results for simulations in which Jupiter and Saturn were initially in their current orbits, and the right column shows the results when these planets were in circular orbits similar to those in the Nice model (2005). – 44 – 2e-02 2e-02 1e-02 1e-02 1e-02 9e-03 7e-03 5e-03 4e-03 2e-03 2e-02 2e-02 1e-02 1e-02 1e-02 9e-03 7e-03 5e-03 4e-03 2e-03 D M A D M A 2x MVEM MVEM 2x MVEM MVEM Current Orbits - Disk A 20 50 75 100 Scale of depletion (%) Current Orbits - Disk B 2e-02 2e-02 1e-02 1e-02 1e-02 9e-03 7e-03 5e-03 4e-03 2e-03 2e-02 2e-02 1e-02 1e-02 1e-02 9e-03 7e-03 5e-03 4e-03 2e-03 D M A D M A Nice Model - Disk A 20 50 75 100 Scale of depletion (%) Nice Model - Disk B 2x MVEM MVEM 2x MVEM MVEM 20 50 75 100 20 50 75 100 Scale of depletion (%) Scale of depletion (%) Fig. 14.- Graphs of the mean values of AMD as a function of the mass-depletion scale for different disk models and giant planets configuration. The left column shows the results for simulations in which Jupiter and Saturn were initially in their current orbits, and the right column shows the results when these planets were in circular orbits similar to those in the Nice model (2005). 1.000 0.100 0.010 s s a M h t r a E 0.001 0.700 0.700 0.600 0.600 0.500 0.500 0.400 0.400 0.300 0.300 0.200 0.200 0.100 0.100 0.000 0.000 y y t t i i c c i i r r t t n n e e c c c c E E 25.000 25.000 20.000 20.000 15.000 15.000 10.000 10.000 ) ) . . g g e e D D ( ( . . l l c c n n I I 5.000 5.000 0.000 0.000 0 0.5 1 0 0 0.5 0.5 1 1 0 0 0.5 0.5 1 1 2 1.5 3 Semi-major Axis (AU) 2.5 2 2 1.5 1.5 3 3 Semi-major Axis (AU) Semi-major Axis (AU) 2.5 2.5 2 2 1.5 1.5 3 3 Semi-major Axis (AU) Semi-major Axis (AU) 2.5 2.5 – 45 – 1.000 0.100 0.010 s s a M h t r a E 0.001 3.5 4 0 0.5 1 y y t t i i c c i i r r t t n n e e c c c c E E 0.700 0.700 0.600 0.600 0.500 0.500 0.400 0.400 0.300 0.300 0.200 0.200 0.100 0.100 0.000 0.000 3.5 3.5 4 4 0 0 0.5 0.5 1 1 25.000 25.000 20.000 20.000 15.000 15.000 10.000 10.000 ) ) . . g g e e D D ( ( . . l l c c n n I I 5.000 5.000 0.000 0.000 3.5 3.5 4 4 0 0 0.5 0.5 1 1 2 1.5 3 Semi-major Axis (AU) 2.5 2 2 1.5 1.5 3 3 Semi-major Axis (AU) Semi-major Axis (AU) 2.5 2.5 2 2 1.5 1.5 3 3 Semi-major Axis (AU) Semi-major Axis (AU) 2.5 2.5 3.5 4 3.5 3.5 4 4 3.5 3.5 4 4 Fig. 15.- Orbital distributions of the surviving bodies in the simulations considering the disk model B with Jupiter and Saturn initially in their current orbits. Open circles correspond to bodies with masses larger than 0.3 MMars ≈ 0.033M⊕. Smaller bodies are labeled with crosses. The left column shows the results of 9 simulations with a depletion scale of 50% and the right column corresponds to those with a depletion scale of 75%. The solid triangles represent the inner planets of the solar system. The gray area shows the asteroid belt. – 46 – Table 1: Regions and scales of mass-depletion Disk Region (AU) Scale (%) A B 1.1 to 2.1 1.3 to 2.0 20, 50, 75, 100 20, 35, 50, 75, 100 – 47 – Table 2: Representative results of the simulations for the disk model A in which at least one body with a mass M < 0.5M⊕ was formed within 1.25 AU and 2.0 AUa . Sim Nc Ngc WMF Body e I(◦) t50% (Myr) t90% (Myr) ainit (AU) afin (AU) Mass (M⊕) A-100%- II A-100%- II A-100%- II A-75%- I A-75%- I A-75%- I A-75%- I A-75%- I A-75%- II A-75%- II A-75%- II A-75%- II A-75%- II A-75%- III A-75%- III A-75%- III A-75%- III A-50%- I A-50%- I A-50%- I A-50%- I A-50%- II A-50%- II A-50%- II A-50%- II A-50%- II A-50%- IV A-50%- IV A-50%- IV EM17 EM36 EM93 EM5 EM52 EM64 EM164 EM173 EM34 EM16 EM53 EM167 EM172 EM12 EM34 EM45 EM161 EM19 EM42 EM113 EM169 EM4 EM45 EM47 EM128 EM148 EM18 EM16 EM133 0.5988 0.7690 3.5587 0.5207 0.8923 1.0636 2.3522 2.7681 0.7457 0.5941 0.9574 2.6842 2.9346 0.5698 0.7334 0.8446 2.2421 0.6052 0.7977 1.7161 3.9663 0.5167 0.8718 0.8949 2.1000 3.0285 0.6036 0.5941 2.3435 0.5772 1.2567 2.6544 0.5516 0.8495 1.2543 2.2403 3.0647 0.4576 0.6979 1.1308 1.5480 2.8572 0.5661 0.8792 1.3370 2.2601 0.5708 1.0492 1.5994 3.9013 0.5274 0.7484 1.0026 1.3759 3.0188 0.4939 0.9080 1.4509 0.1626 0.0647 0.2764 0.0802 0.0288 0.0318 0.1213 0.0698 0.0528 0.0212 0.0423 0.1236 0.1416 0.0197 0.0439 0.1392 0.0261 0.0638 0.0803 0.1008 0.1441 0.0329 0.0320 0.0275 0.0332 0.1541 0.0481 0.0303 0.1171 11.9910 9.9490 4.0460 3.6940 2.9710 2.7140 3.2200 2.9860 4.2490 3.5730 3.5950 13.4810 6.4470 1.1750 6.1250 8.7110 0.5690 2.8780 3.7560 29.1090 1.9040 0.9110 3.6580 3.9390 5.4550 6.0760 8.6880 1.1200 4.8840 0.9479 0.3066 0.0667 0.6412 0.7274 0.4355 0.0256 0.0660 0.2195 0.9520 0.5470 0.0864 0.0645 0.7136 0.7475 0.1563 0.0578 1.0949 0.4318 0.0157 0.0379 0.4821 0.5544 0.5127 0.2402 0.0359 0.5856 0.9066 0.3826 2.581546e-06 7.981451e-06 5.000000e-02 2.013308e-04 0.00000e+00 3.537550e-04 1.000000e-03 4.907275e-02 5.723449e-04 1.572659e-03 4.593763e-04 3.373369e-02 4.952557e-02 1.903639e-03 1.227195e-04 8.532762e-04 9.788227e-04 1.310486e-03 6.430292e-04 7.859689e-03 5.000000e-02 1.522684e-05 2.354406e-04 2.640177e-04 1.363240e-03 5.000000e-02 2.151852e-04 4.376598e-05 4.400611e-03 43 25 1 87 91 33 0 6 40 99 63 7 4 54 77 23 6 82 41 6 0 63 80 57 19 2 78 70 15 10 8 1 10 12 6 0 1 6 15 13 2 1 11 14 4 1 20 11 0 0 9 13 13 7 0 14 17 5 114.90 3.61 3.40 32.83 39.44 62.42 0.00 0.00 4.05 12.34 12.92 2.19 0.60 86.33 12.94 56.91 0.00 52.35 10.94 0.00 0.00 20.48 3.17 5.52 28.94 0.00 4.18 58.97 53.69 562.30 116.70 3.40 138.90 53.38 147.70 0.00 0.00 40.76 64.75 28.36 2.48 0.60 537.20 21.26 74.10 0.00 163.60 71.66 40.56 0.00 53.26 54.09 43.28 216.20 0.00 44.41 86.67 126.40 aFrom left to right, the columns show the simulation, the body, the semimajor axis of body in the beginning of the simulation, its final semimajor axis, eccentricity, inclination (deg), water-mass fraction, total number of collisions, number of collisions with large objects (impactor mass > 0.01M⊕), time of reaching 50% of the mass, and time of reaching 90% of the mass, respectively. – 48 – Table 3: Representative results of the simulations for the disk model B in which at least one body with a mass M < 0.5M⊕ was formed within 1.25 AU and 2.0 AUa . Sim Nc Ngc WMF Body e I(◦) t50% (Myr) t90% (Myr) ainit (AU) af in (AU) Mass (M⊕) B-75%- II B-75%- II B-75%- II B-75%- III B-75%- III B-75%- III B-75%- III B-50%- II B-50%- II B-50%- II B-50%- II B-50%- III B-50%- III B-50%- III B-50%- III B-50%- III B-35%- I B-35%- I B-35%- I EM26 EM60 EM70 EM16 EM5 EM159 EM137 EM12 EM19 EM64 EM136 EM22 EM25 EM34 EM10 EM127 EM38 EM11 EM53 0.6646 1.0287 1.1774 0.5973 0.5233 3.1186 2.1413 0.5742 0.6214 1.0799 2.6272 0.6366 0.6628 0.7396 0.5609 2.4728 0.7734 0.5580 0.9462 0.5338 0.8565 1.6135 0.5641 1.0044 1.5477 2.3173 0.4269 0.7462 1.1638 2.6524 0.5194 0.6688 1.0009 1.5132 2.8621 0.5684 0.9886 1.6636 0.0169 0.0229 0.1501 0.0457 0.0135 0.0987 0.0251 0.1639 0.0894 0.1237 0.0389 0.0577 0.0933 0.0174 0.0319 0.0521 0.0218 0.0331 0.1205 2.6090 5.2200 4.1080 5.3410 3.3400 16.9160 10.0770 10.8200 2.1440 4.6200 10.0330 6.0580 3.5970 1.8090 6.9570 1.4900 1.2620 6.3010 5.6370 0.5749 0.9882 0.2976 0.7419 0.9884 0.0658 0.0345 0.2736 1.1127 0.3797 0.0279 0.5161 0.2841 0.9474 0.2653 0.1222 0.8535 0.9829 0.3405 1.808761e-05 6.242398e-05 2.068806e-04 1.715158e-04 1.789666e-03 4.674640e-02 4.380566e-03 4.650020e-04 2.315770e-04 1.764175e-04 5.000000e-02 7.913091e-05 1.291794e-05 1.656217e-03 0.000000e+00 2.729102e-02 3.620376e-04 2.852670e-04 5.966567e-04 88 61 27 58 81 3 5 37 93 22 0 69 46 85 27 5 88 66 24 14 14 9 8 18 1 0 6 17 7 0 10 7 15 7 3 18 18 6 2.60 72.66 21.85 54.60 6.23 18.16 0.00 24.02 19.00 52.41 0.00 5.34 3.90 17.57 28.67 1.58 11.66 21.94 25.49 41.26 149.60 103.30 93.92 31.69 85.59 7.03 73.82 94.39 70.54 0.00 61.23 45.56 116.50 229.90 25.55 50.99 92.21 52.22 aFrom left to right, the columns show the simulation, the body, the semimajor axis of body in the beginning of the simulation, its final semimajor axis, eccentricity, inclination (deg), water-mass fraction, total number of collisions, number of collisions with large objects (impactor mass > 0.01M⊕), time of reaching 50% of the mass, and time of reaching 90% of the mass, respectively. – 49 – MMars Table 4: Summary of the results of the simulations in disk model A. The heading of each column represents the criterion that was used for assessing the success of a simulation. A success is indicated by (X), a failure is shown by (×), and a (∼) indicates a near successful casea Sim. A-75%- I A-75% - II A-75% - III A-50%- I A-50% - II A-50%- III tform,Earth W MFEarth AMD Mast Nast tform,Mars MEarth ∼ × × × 0 1 6 1 4 9 × × × X × × X X X X X × X X X ∼ X × X X X × X X × × × × × ∼ × × × × X × X × aFrom left to right the columns represent the simulation, mass of the Mars-analog candidate, timescale of the formation of the Mars-analog candidate, mass of the Earth candidate, timescale of the formation of the Earth candidate, water-mass fraction of the Earth candidate, angular momentum deficit of the system (AMD), mass in embryos stranded in the asteroid belt, and the number of asteroids left. – 50 – Table 5: Summary of the results of the simulations in disk model B. The heading of each column represents the criterion that was used for assessing the success of a simulation. A success is indicated by (X), a failure is shown by (×), and a (∼) indicates a near successful casea Sim. tform,Earth W MFEarth AMD Mast Nast MMars tform,Mars MEarth B-75% - I B-75% - II B-75% - III B-75% - IV B-75% - V B-75% - VI B-75% - VII B-75% - VIII B-75% - IX B-50%- I B-50%- II B-50%- III B-50% - IV B-50% - V B-50% - VI B-50% - VII B-50% - VIII B-50% - IX × X X X X X × X × × × X X × X × X × × × × × × × × × × × × × × × X × × × X X X X X X X X × X × X X × X X X X × X X X X × X × × × × X X × ∼ × X X × × X X X × X × × × × X X × X × × X × X X X × X × X × × × X X × × X X X X X X X × X X × X X X × X X X × X X 2 0 3 1 1 2 10 0 0 4 1 2 3 2 2 8 0 2 bFrom left to right the columns represent the simulation, mass of the Mars-analog candidate, timescale of the formation of the Mars-analog candidate, mass of the Earth candidate, timescale of the formation of the Earth candidate, water-mass fraction of the Earth candidate, angular momentum deficit of the system (AMD), mass in embryos stranded in the asteroid belt, and the number of asteroids left.
1803.10787	1	1803	2018-03-28T18:01:01	Improving the Accuracy of Planet Occurrence Rates from Kepler using Approximate Bayesian Computation	[ "astro-ph.EP" ]	We present a new framework to characterize the occurrence rates of planet candidates identified by Kepler based on hierarchical Bayesian modeling, Approximate Bayesian Computing (ABC), and sequential importance sampling. For this study we adopt a simple 2-D grid in planet radius and orbital period as our model and apply our algorithm to estimate occurrence rates for Q1-Q16 planet candidates orbiting around solar-type stars. We arrive at significantly increased planet occurrence rates for small planet candidates ($R_p<1.25 R_{\oplus}$) at larger orbital periods ($P>80$d) compared to the rates estimated by the more common inverse detection efficiency method. Our improved methodology estimates that the occurrence rate density of small planet candidates in the habitable zone of solar-type stars is $1.6^{+1.2}_{-0.5}$ per factor of 2 in planet radius and orbital period. Additionally, we observe a local minimum in the occurrence rate for strong planet candidates marginalized over orbital period between 1.5 and 2$R_{\oplus}$ that is consistent with previous studies. For future improvements, the forward modeling approach of ABC is ideally suited to incorporating multiple populations, such as planets, astrophysical false positives and pipeline false alarms, to provide accurate planet occurrence rates and uncertainties. Furthermore, ABC provides a practical statistical framework for answering complex questions (e.g., frequency of different planetary architectures) and providing sound uncertainties, even in the face of complex selection effects, observational biases, and follow-up strategies. In summary, ABC offers a powerful tool for accurately characterizing a wide variety of astrophysical populations.	astro-ph.EP	astro-ph	DRAFT VERSION MARCH 30, 2018 Typeset using LATEX twocolumn style in AASTeX61 IMPROVING THE ACCURACY OF PLANET OCCURRENCE RATES FROM KEPLER USING APPROXIMATE BAYESIAN COMPUTATION DANLEY C. HSU,1, 2, 3, 4 ERIC B. FORD,1, 2, 3, 4 DARIN RAGOZZINE,5, 6 AND ROBERT C. MOREHEAD7 1Department of Astronomy & Astrophysics, 525 Davey Laboratory, The Pennsylvania State University, University Park, PA, 16802, USA 2Center for Exoplanets and Habitable Worlds, 525 Davey Laboratory, The Pennsylvania State University, University Park, PA, 16802, USA 3Center for Astrostatistics, 525 Davey Laboratory, The Pennsylvania State University, University Park, PA, 16802, USA 4Institute for CyberScience, The Pennsylvania State University 5Department of Physics & Astronomy, N283 ESC, Brigham Young University, Provo, UT 84602, USA 6Department of Physics & Space Science, 150 West University, Florida Institute of Technology, Melbourne, FL 32901, USA 7Texas Tech University, Physics & Astronomy Department, Box 41051, Lubbock, TX 79409, USA ABSTRACT We present a new framework to characterize the occurrence rates of planet candidates identified by Kepler based on hierarchical Bayesian modeling, Approximate Bayesian Computing (ABC), and sequential importance sampling. For this study we adopt a simple 2-D grid in planet radius and orbital period as our model and apply our algorithm to estimate occurrence rates for Q1- Q16 planet candidates orbiting around solar-type stars. We arrive at significantly increased planet occurrence rates for small planet candidates (Rp < 1.25R⊕) at larger orbital periods (P > 80d) compared to the rates estimated by the more common inverse detection efficiency method. Our improved methodology estimates that the occurrence rate density of small planet candidates in the habitable zone of solar-type stars is 1.6+1.2 −0.5 per factor of 2 in planet radius and orbital period. Additionally, we observe a local minimum in the occurrence rate for strong planet candidates marginalized over orbital period between 1.5 and 2R⊕ that is consistent with previous studies. For future improvements, the forward modeling approach of ABC is ideally suited to incorporating multiple populations, such as planets, astrophysical false positives and pipeline false alarms, to provide accurate planet occurrence rates and uncertainties. Furthermore, ABC provides a practical statistical framework for answering complex questions (e.g., frequency of different planetary architectures) and providing sound uncertainties, even in the face of complex selection effects, observational biases, and follow-up strategies. In summary, ABC offers a powerful tool for accurately characterizing a wide variety of astrophysical populations. Keywords: methods: data analysis - methods: statistical - catalogs - planetary systems - stars: statistics 8 1 0 2 r a M 8 2 . ] P E h p - o r t s a [ 1 v 7 8 7 0 1 . 3 0 8 1 : v i X r a 2 HSU ET AL. 1. INTRODUCTION Since the first exoplanets were discovered around the pul- sar PSR B1257+12 (Wolszczan & Frail 1992), the number of exoplanets has increased to a few thousand. The exoplanet population includes planet sizes, masses, and orbital proper- ties that are not found among the Solar System planets. For example, there are classes of exoplanets for which there are no Solar System counterparts, including hot Jupiters, warm Neptunes, and super-Earth-size planets. Further, many exo- planets have highly eccentric orbits, in contrast to our Solar System in which most planets follow nearly circular orbits. The discovery and confirmation of such planets and plane- tary systems has inspired theoretical research in planet for- mation, planetary migration and planet-planet interactions. An essential constraint for such studies is the true underlying occurrence rate (or frequency) of exoplanets as a function of their physical and orbital properties. 1.1. Kepler Results NASA's Kepler mission was launched in 2009 with the primary goal of characterizing the occurrence rate of Earth- size planets around Sun-like stars (Borucki 2016; Borucki et al. 2010). Over the course of four years Kepler observed ∼ 192,000 stars and identified several thousand exoplanet candidates (Batalha 2014), including a significant majority of high-quality exoplanet candidates. These discoveries indi- cate that sub-Neptune size planets represent the majority of the exoplanet population among stars surveyed and likely the Milky Way galaxy (Batalha et al. 2013; Burke et al. 2014; Coughlin et al. 2016). While Kepler has identified a large number of exoplanet candidates, translating the Kepler planet candidate catalog into a true underlying population is a challenging. The ob- served catalog differs from the true population of exoplan- ets due to a variety of factors, including: 1) the geometric transit probability which depends primarily on the orbital pe- riod and star size, as well as the orbit shape, 2) the detection probability which depends primarily on the integrated tran- sit signal-to-noise, and thus indirectly on several factors such as the transit depth, transit duration, number of transits, Ke- pler's photometric measurement precision stellar, and stellar photometric variability, and 3) the choice of stars targeted by Kepler. Transit surveys are inherently more sensitive for detecting planets with short orbital periods due to the increased ge- ometric transit probability and increased number of transits within a given timespan of observations. Transit surveys are also more sensitive for detecting larger planets for a given star, since the transit signal-to-noise is proportional to the square of the planet radius. Similarly, transit surveys are more sensitive to the detection of planets of a given planet- star radius ratio for stars with higher quality photometry. While it is not always easy to separate measurement uncer- tainty from astrophysical photometric variability, the com- bined differential photometric precision (CDPP) reported by the Kepler pipeline provides a useful summary of the effec- tive "noise" due to the combination of astrophysical and in- strumental noise sources. Finally, the target selection pro- cess, including both the field of view of the Kepler space- craft and which target stars were selected for data downlink, affects the planets detected and the interpretation of occur- rence rates. Like previous studies, this study will defer on this last issue and focus on measuring the planet occurrence rate among a subset of Kepler target stars, while accounting for geometric transit probability and detection probability. In this study, we introduce Approximate Bayesian Com- puting (ABC) as a tool for overcoming the above challenges to characterize the exoplanet population in general and planet occurrence rates in particular. Our framework is original in that it can properly account for the interaction between mea- surement uncertainties and the planet detection probability which has been well characterized as a function of transit signal-to-noise in previous studies (e.g., Christiansen et al. 2015). We also develop a practical algorithm for applying ABC to infer planet occurrence rates as a function of planet size and orbital period, allowing for direct comparisons to results using previous methods. We verify and validate the ABC algorithm using simulated data sets and report results of applying our algorithm to a recent catalog of Kepler planet candidates. In order to understand the context for this re- search, we first review several influential planet occurrence rate studies. 1.2. Kepler Occurrence Rate Studies Several groups have estimated planet occurrence rates based on various Kepler planet candidate catalogs (e.g., Batalha et al. 2013; Burke et al. 2014; Silburt et al. 2015; Coughlin et al. 2016). At the time of Youdin (2011) only Kepler planet candidates identified based on the first two months of Kepler data were available. These planet can- didates spanned a relatively small range of orbital periods and sizes, so the population could be reasonably modeled by a joint power law parameterization in planet radius and orbital period or a broken, joint power law. While Youdin (2011) used a simplistic planet detection efficiency model, they demonstrated an efficient methodology for computing maximum likelihood estimates of occurrence rates that could generalize once the planet detection efficiency was better characterized. Another influential early study (Howard et al. 2012) esti- mated the planet occurrence rate for solar type stars at each of several bins of planets, defined in terms of a 2-D grid over planet radius and orbital period. This study assumed planet candidates identified from the list of Kepler objects of inter- PLANET OCCURRENCE RATES USING ABC 3 est (KOIs) from Borucki et al. (2011) were true planets and attempted to "correct" for non-detections due to either non- transiting orbital inclinations or insufficient photometric pre- cision via the "inverse detection efficiency method" (IDEM), so named by Foreman-Mackey et al. (2014). The IDEM as- signs each detected planet candidate with a "weight" that at- tempts to estimate the number of exoplanets with the mea- sured size and period that would need to be distributed among the target stars in the catalog, so as to yield one detection. Several subsequent occurrence rates studies have built on the IDEM. Mann et al. (2012) focused on improving the stel- lar parameters of Kepler target stars, noting that occurrence rates estimated in previous studies were biased due to the use of the Kepler Input Catalog (KIC). Fressin et al. (2013) im- proved the detection efficiency model by incorporating a lin- ear ramp model for the transit detection probability as a func- tion of transit signal-to-noise. Petigura et al. (2013) devel- oped a custom transit search pipeline focusing on a subset of favorable target stars and characterize the detection efficiency of their pipeline by injecting simulated transits into the light curves. Dressing & Charbonneau (2013, 2015); Kopparapu (2013) performed occurrence rate studies similar to Howard et al. (2012), but focused on M dwarf stars. In another particularly influential study, Christiansen et al. (2015) estimated the planet occurrence rate in the Kepler sample for several ranges of planet radii and orbital peri- ods using the IDEM. The main advance of Christiansen et al. (2015) was quantifying the planet detection efficiency of the Kepler pipeline by injecting simulated transit signals into the raw pixel data and reprocessing the simulated data with the Kepler pipeline. They performed Monte Carlo simulations to characterize the planet detection efficiency. They fit the empirical planet detection efficiency curve with the CDF of a Γ distribution. They applied their improved planet detection efficiency model by applying it to Kepler planet candidate catalog. They focused on FGK stars observed by Kepler over Q1-Q12 and estimated planet occurrence rates for bins span- ning planet sizes of 1 − 2R⊕ and orbital periods of 0.5 to 320 days. Christiansen et al. (2015) also computed planet occur- rence rates using two alternative planet detection efficiency models: a "perfect detector" model (an error function with a transition at 7.1σ, assuming that each measurement has inde- pendent white noise) and the linear ramp detection efficiency model of Fressin et al. (2013). While the simplicity and speed of the IDEM is appealing, the reliance on estimated planet properties can lead to a sig- nificant bias. To illustrate this problem, we created two sets of 10 planet catalogs defined by a 0.05 per star occurrence rate in the P = 40 − 80 day and Rp = 1.25 − 1.5R⊕ bin (see §4). Each catalog is then treated as an observed planet can- didate catalog to be used to estimate occurrence rates by the IDEM. In Fig. 1 (top panel) we plot the occurrence rate es- timated by IDEM as a Gaussian PDF with a mean and width calculated according to Appendix A. This demonstrates the IDEM estimated rates are biased and systematically underes- timate the occurrence rate. This can be easily understood as a result of using the estimated planet size to compute the com- pleteness correction, rather than the true planet size. While the resulting bias is small for most bins, it becomes substan- tial for planets near the threshold of detection as is the case for this bin. Since small planets at orbital periods near one year are both near the threshold of detection and of particu- lar interest for both science and mission planning, it is im- portant to develop methods that more accurately characterize planet occurrence rates of such planets. The bottom panel of Fig. 1 shows results from an analogous computation using a more rigorous hierarchical Bayesian model and Approxi- mate Bayesian Computing, as described in §2. We will dis- cuss these results in §4 once we have described the method- ology in detail. For now, we merely note that the ABC pos- teriors are correctly centered on the true occurrence rate, in stark contrast to the estimates based on the IDEM. The bias in the IDEM becomes even more pronounced for larger or- bital periods and/or smaller planets, as expected due to the large measurement uncertainties. Foreman-Mackey et al. (2014) described a more rigor- ous methodology for characterizing the exoplanet occurrence rates based on a hierarchical Bayesian model (HBM). Similar to Howard et al. (2012) and Christiansen et al. (2015), they estimated planet occurrence rates over a 2-D grid in terms of planet radius and orbital period. However, unlike previous studies they assumed that the occurrence rates for nearby bins were correlated, making use of a Gaussian Process (GP) prior for the occurrence rates within each bin. Recognizing the im- portance of an accurate planet detection efficiency method, they applied their methodology to a planet catalog from Pe- tigura et al. (2013), as Christiansen et al. (2015) was not yet available. Foreman-Mackey et al. (2014) also found a signif- icantly increased rate of small, long-period planets compared to the inverse detection efficiency method used in the Chris- tiansen et al. (2015) study, due to a combination of using a HBM and their GP prior. However, their inferred results were less than those of Petigura et al. (2013) who focused their at- tention on a subset of target stars and used planet candidates from a custom detection pipeline. Recently, Burke et al. (2015) characterized the popula- tion of small exoplanets by combining a more recent Ke- pler planet candidate catalog including data from Q1-16 (Tenenbaum et al. 2014), an improved planet detection ef- ficiency model, based on planet injection studies which in- jected transits at the pixel level (Christiansen et al. 2015), and a Bayesian generalization of Youdin (2011). Rather than using a grid in planet size and orbital period, Burke et al. (2015) assume a broken, joint power law planet distribution 4 HSU ET AL. studies based on a forward model, while also providing a solid statistical foundation perform statistical inference. 1.3. Future of Kepler Occurrence Rate Studies Previous studies have focused on the occurrence rate of planets on an individual basis, and side-stepped complica- tions due to multiple planet systems. Previous planet oc- currence rates studies that included all Kepler planet candi- dates around a star can be better interpreted as estimating the average number of planets (within a range of sizes and or- bital periods) per star, which may be quite difference from the fraction of stars with such planets (Youdin 2011; Brak- ensiek & Ragozzine 2016). Given the limitations of early Kepler planet candidate catalogs, some previous studies in- cluded only the "first" or "strongest" detected planet candi- date. These are even more difficult to interpret, since whether any given planet is counted depends on the existence of other planets in the same system that differ in size and/or orbital period. Thus, such studies do not estimate the fraction of stars having at least one planet within that bin (Brakensiek & Ragozzine 2016). In principle, a hierarchical Bayesian framework is well-suited to characterizing the occurrence rate of planetary systems. However, a direct generalization of the Foreman-Mackey et al. (2014) model to multiple planet systems would be computationally prohibitive. The primary purpose of this study is to validate a new sta- tistical framework for inferring planet occurrence rates that can later be generalized to address the occurrence rate of planetary systems. In particular, we evaluate a HBM for the exoplanet occurrence rate, using ABC and sequential impor- tance sampling. This allows us to avoid the bias in planet occurrence rates introduced by the IDEM. We also apply our methods to a recent Kepler planet candidate catalog to pro- vide updated planet occurrence rate estimates as a function of planet size and orbital period. Of course, these estimates could be improved with further, as we make several assump- tions common among published planet occurrence rate stud- ies (e.g., host star properties, catalog reliability). To ease comparisons, our study will compare occurrence rates estimated using ABC to those of the inverse detection efficiency method using a star and planet catalog similar to that of Christiansen et al. (2015) and the Γ cumulative distri- bution function (CDF) detection efficiency model of Chris- tiansen et al. (2015). We describe our statistical methodology in §2. In §3, we provide the details of the forward model for the planet population and detection model used by ABC. We validate our methodology on simulated data in §4 and apply it to actual Kepler data in §5. Finally, we discuss the impli- cations of our results and future prospects for characterizing exoplanet populations in §6. 1.4. Role of ABC Figure 1. Ten simulations where the occurrence rate is estimated for the 40-80d period, 1.25-1.5 R⊕ planet radius bin with a fixed "true" occurrence rate of 0.05 per star (indicated by the red vertical line). (Top) Gaussian PDF of the estimated inverse detection effi- ciency method occurrence rate where the width of the Gaussian σ f is defined in Appendix A. The inverse detection efficiency method systematically underestimates the true occurrence rate. (Bottom) Histogram of 1000 samples drawn from the final ABC posterior. The ABC simulations accurately bracket the true occurrence rate. function over planet radii and orbital period parameter space. Given the limitations of their parameterization, they focus on small planets (0.75− 2.5R⊕) with orbital periods in the range 50-300 days. Burke et al. (2015) neglect uncertainty in planet radius, and do not account for how uncertainty in planet ra- dius affects the transit detection efficiency. Their inferred rate of small planets was also higher than Christiansen et al. (2015). In another recent paper, Silburt et al. (2015) applied meth- ods somewhat similar to those presented in this study to the planet catalog generated by the Kepler pipeline. Here, we describe an approach for performing planet occurrence rate 0246810Occurrence Rate (%)0.00.10.20.30.40.50.60.7PDF0246810Occurrence Rate (%)01020304050607080FrequencyPLANET OCCURRENCE RATES USING ABC 5 Originally, Bayesian inference via forward modeling and rejection was intended as a pedagogical tool for understand- ing Bayesian inference (Rubin 1984). The combination of in- creasing computational power and the complexity of genetic evolution models, led to ABC being considered for practical calculations (Tavare et al. 1997). Several authors have ex- plored the choice of summary statistics (e.g., Fearnhead & Prangle 2012; Marin et al. 2012) as well as sampling algo- rithms to improve the sampling efficiency of ABC. Our se- quential importance sampling algorithm (sometimes known as particle or sequential Monte Carlo in the statistics com- munity) closely follows that of Beaumont et al. (2008) and Ratmann et al. (2013). ABC has had some limited applications in the field of as- tronomy over the past few years. Cameron & Pettitt (2012) gave one of the earliest demonstrations of ABC in astron- omy by making use of a Sequential Monte-Carlo implemen- tation of ABC to determine posteriors for a stochastic model of high-redshift massive galaxy morphological transforma- tion. In a later study, an Markov chain Monte Carlo (ABC- MCMC) variant of the algorithm was used to study the shape of the thick disc of the Milky Way galaxy in order to better constrain formation models (Robin et al. 2014). Following these applications in galaxy evolution, several cosmology papers used applications of ABC. Akeret et al. (2015) detailed a Population Monte-Carlo implementation of ABC (ABC-PMC) and applied it to calibrate image sim- ulations for wide-field cosmological surveys. Ishida et al. (2015) released a public version of ABC-PMC written in Python called COSMOABC and demonstrated it by estimat- ing posteriors for cosmological parameters using galaxy clus- ter number counts. A new implementation of ABC called su- perABC is discussed by Jennings et al. (2016) in their study wherein cosmological parameters are constrained using Type Ia supernovae light curves with the use of two different dis- tance functions. This study applies similar algorithms to characterize the occurrence rate of exoplanets. 2. STATISTICAL FRAMEWORK Bayesian statistical analysis is becoming common in as- tronomy thanks to the increasing availability of practical al- gorithms and computing power to perform the required cal- culations. Unlike traditional means of fitting a best-fit model to data (e.g. using χ2), a Bayesian analysis can estimate the posterior probability distribution (π(θY )) for a given set of model parameters (θ) given the data Y , prior information about the model parameters π(θ), and a likelihood function f (Yθ) which specifies the probability of the data given the model and model parameters: π(θY ) = π(θ) f (Yθ) (cid:82) θ dθ(cid:48)π(θ(cid:48)) f (Yθ(cid:48)) . (1) The primary challenge to widespread application of Bayesian methods is computing the potentially high-dimensional in- tegrals involved. Markov chain Monte Carlo (MCMC) is a powerful tool for Bayesian parameter estimation in exo- planet research. Even simple implementations are often ad- equate for low-dimensional problems, such as characterizing the orbit of a single planet (Ford 2005). In higher dimen- sions (e.g., characterizing multiple planet systems), efficient sampling often demands more sophisticated samplers (e.g., Nelson et al. 2014). 2.1. Hierarchical Bayesian Models HBM provides a rigorous statistical framework for char- acterizing a population. In HBM, we separate model pa- rameters into parameters that describe the population (φ) and parameters that describe the properties of individual objects from that population (θ j), for j = 1...Nt, where Nt is the num- ber of members of the population. For our study, φ will refer to population-level model parameters, such as the occurrence rate of planets (for a given range of planet size, orbital period and host star properties), while θ j would refer to the proper- ties of the jth target (e.g., planet size, orbital period, epoch of transit, host star properties). Similarly, we can divide the observed data (Y ) into subsets (Yj) that depend only on the data for the jth target. If we are interested in characterizing the population, then we marginalize over the properties of each target (i.e., integrate over these parameters, weighting by their relative probability), so Bayes' theorem can then be written as (cid:82) π(φ)(cid:81)Nt φ(cid:48) dφ(cid:48)π(φ(cid:48))(cid:81)Nt i=1 θ j (cid:82) (cid:82) i=1 θ j π(φY ) = dθ jπ(θ jφ) f (Yjθ j) dθ jπ(θ jφ(cid:48)) f (Yjθ j) . (2) Note that Yj depends only on θ j and is conditionally indepen- dent from the population parameters, φ, or the properties of other planets. This conditionally independence means that if we knew θ j, then we would not gain any additional informa- tion about the probability distribution for Yj by making use of knowledge about the distribution or values of φ or θk (cid:54)= j. The conditional independence significantly simplifies the calcula- tions, transforming high dimensional integrals into the prod- uct of many lower dimensional integrals. Another common simplification is to replace the observed data Y with sum- mary statistics that describe the key properties of the data (s = S(Y )). As a simplistic example, one might approximate an entire light curve due to a single planet transiting a photo- metrically quiet star as a constant out-of-transit flux, a transit depth, transit duration, orbital period, epoch of one transit and magnitude of Gaussian measurement noise. If one uses an accurate model for the data, then the true model parame- ters form sufficient summary statistics. However, the model parameters are often not directly observable. In practice, the 6 HSU ET AL. true model parameters must be estimated from the data (Y ) or a set of summary statistics. Despite these simplifications, evaluating Eqn. 2 for either the set of all Kepler light curves or a Kepler planet can- didate catalog would be extremely expensive. First, note that Nt refers to the number of targets and not the num- ber of planets. Even if we reduce the Kepler observations to merely four properties per planet (orbital period, orbital phase, transit depth, transit duration) and two properties per star (star radius, magnitude of measurement noise), evaluat- ing the likelihood in Eqn. 2 would require integrating over (cid:39) Nt(2 + 4(cid:104)Npps(cid:105)) dimensions, where (cid:104)Npps(cid:105) is the average number of planets per star. Since Nt > 100,000 and (cid:104)Npps(cid:105) is likely greater than 2, even a simplistic model would require integrating over ∼ 106 dimensions. For targets where all the star and planet properties are well measured, the integral over θ j could likely be performed ef- ficiently using importance sampling and the measured planet properties. However, even with an ideal observatory, most planets will not be detected in a transit search due to transit geometry. Since many (if not most) targets harbor planets that are not detected, the integrals over θ j need to be evalu- ated over all possible planet sizes and orbital periods for each of the planets. If each target star could harbor only a single planet, then one could treat each planet as independent to approximate this integral once for each target and reuse the result each time the likelihood needed to be evaluated (e.g., Youdin 2011; Foreman-Mackey et al. 2014). This approach would no longer be possible once we allow each host star to have multiple planets which may not be distributed indepen- dent of each other (e.g., avoid each other due to stability or correlations in orbital period). The sizes and orbital periods of planets around a given star are not independent of each other, so the properties of any planets that are detected af- fect the conditional distributions for the undetected planets. The abundance of stars with multiple planets necessitates ac- counting for multiple planets around each star. Thus, evaluat- ing the likelihood in Eqn. 2 in the context of the Kepler planet search and multiple planet systems is intractable, regardless of whether it is to be used for an MCMC simulation to per- form parameter estimation for the population parameters, φ, or for a computing maximum likelihood estimator. These limitations motivate us to consider an alternative approach to characterizing the exoplanet population that we will explore in the next section. 2.2. Theory of Approximate Bayesian Computing Approximate Bayesian Computation (ABC) provides a means to perform Bayesian inference for problems where ei- ther it is not practical to write out the likelihood or evaluating the likelihood is computationally prohibitive. ABC is particularly well-suited for applying HBM to model populations. One specifies a forward model to de- scribe both the population to be characterized and the process of collecting data (see §3) and a process for choosing which simulated datasets (and their associated populations param- eters) are to be accepted. One computes many realizations of the forward model via some variant of Monte Carlo sim- ulation and accumulates an ensemble of model parameters that yielded simulated data sets that are sufficiently similar to the observed data set. The result is an "ABC posterior" (πABC(φY )) that takes the place of a traditional Bayesian posterior. If we retain only the population level parameters, then we effectively marginalize over the unknown physical properties of individual members of the population, includ- ing both those detected and those that escaped detection: πABC(φY ) = π(φρ(S(Y ),S(Yobs)) < ) (3) where ρ is the distance function and is the distance thresh- old. Based on the challenges described in §2.1, we anticipate that this approach will be particularly valuable for character- izing the population of planetary systems. While ABC is motivated by complex problems for which likelihood-based methods are impractical, one can also ap- ply ABC to problems for which one can write down the pri- ors and likelihood. In such cases, one can show that the ABC posterior approaches the true posterior for small if one uses sufficient summary statistics and a reasonable distance function (Marin et al. 2012; Beaumont et al. 2008). There- fore, ABC provides a solid statistical foundation for perform- ing likelihood-based statistical inference for complex models where inference using the full likelihood is impractical. Here we will provide an outline of the general process for choosing which simulated data sets are to be accepted, be- fore addressing the specific implementation for this study in subsequent sections. First, we choose a set of summary statistics (S(Y )) that characterize the key properties of the observed or simulated data (see §2.4). Second, we choose a distance function, ρ(S(Yi),S(Yobs)), that specifies the dis- tance between the summary statistics computed for a sim- ulated dataset (si = S(Yi)) and the summary statistics for the observed data, (sobs = S(Yobs); see 2.5). Third, we specify , the maximum acceptable distance for a simulated data set to be accepted. In this case, the ABC posterior is given by dθ j π(φ)π(θ jφ) f (Yjθ j) i=1 θ j The outer integral samples from the prior for φ, the inner integrals correspond to sampling the possible values for the πABC(φY ) ∝(cid:90) Nt(cid:89) φ (cid:90) dφ δ(ρ(S(Y ),S(Yobs)) < )× (4) PLANET OCCURRENCE RATES USING ABC true parameters of each population member (θ j), and the δ- function selects those draws that result in the distance func- tion meeting the acceptance criterion. As a particularly simple example, one could use a naive Monte Carlo algorithm, repeatedly drawing sets of model pa- rameters (φi,θi j) from the prior distributions π(φi)∼ π(φ) and θi j ∼ π(θ jφi), where "∼" can be read as "is distributed as." For each realization, one would generate a simulated data set (Yi) and compute the associated summary statistics, si = S(Yi). If the distance between the simulated data and the observed data is sufficiently small, i.e., ρ(si,sobs) ≤ , then the model parameters φi and θi j's are added to the ensemble that forms the ABC posterior. In other words, only parameter values that create simulated data that are very close to the observed data are kept in the ABC posterior. In practice, the naive ABC algorithm described above is extremely computationally inefficient for most problems of interest and small . For ABC to be useful on real world problems, one must apply a more efficient sampling strat- egy (e.g., Beaumont et al. 2008; Blum 2009; Ratmann et al. 2013). For this study, we apply a sequential importance sam- pling strategy described in §2.3. 2.3. Approximate Bayesian Computing in Practice This study makes use of an ABC - Population Monte Carlo (ABC-PMC) algorithm which relies on sequential impor- tance sampling to evolve an ABC posterior with the goal of achieving a large number of draws with sufficiently small . Our algorithm closely follows that described in Beaumont et al. (2008) and Ratmann et al. (2013). Each generation in the computed sequence contains an ensemble of model pa- rameters. In statistical parlance, each generation's ensem- ble (or population) consists of multiple "particles". Each en- semble of particles can be used to approximate a probabil- ity distribution over the model parameter space. To initialize the ABC-PMC algorithm, we first apply simple Monte Carlo with a fixed number of draws from the prior distribution, and we select the Npart particles which resulted in the smallest dis- tances to the observed dataset. The selected particles make up the 0th generation in which all particles have a distance less than 0 = maxi(ρ0,i). For this and each subsequent gener- ation, we construct a Gaussian mixture model with the mix- ture components centered on the model parameter values of each particle. For the 0th generation, each particle is assigned an equal mixture weight which will be updated on subsequent generations. Each mixture component shares a common co- variance matrix which is based on the sample covariance of the current particle ensemble. Mathematically, the ABC pos- terior at generation g (pABC,g(φYobs)) is estimated by pABC,g(φYobs) = wg,iNφ(φg,i, Σg) (5) Npart(cid:88) i=1 7 weights at each generation form a simplex (i.e.,(cid:80) where wg,i is the weight of the ith particle in generation g, the i wg,i = 1), Nφ(µ, Σ) is a normal probability distribution for φ with mean µ and variance Σ, and Σg is the sample covariance of the population parameters φg,i at generation g. For each subsequent generation, the ABC-PMC particle population is evolved using importance sampling (IS) to draw trial sets of model parameters (φ∗) for the next generation, φ∗ ∼ pIS,g(φ). When evaluating trial sets of model parame- ters for the gth generation, only trials that result in a distance between the simulated data and observed data that is less than g are accepted. The distance threshold g is gradually de- creased, so that each generation more closely approximates the posterior distribution (which requires → 0). Once the algorithm draws Npart trial particles with ρ∗ < g, the weights for each particle are updated to be proportional to the ra- tio of the prior and the importance sampling density. First, we compute unnormalized weights, w∗ g,i = p(φg,i)/pIS,g(φg,i), before computing the properly normalized weights, wg,i = w∗ g,i. This results in the ABC posterior in sub- sequent generations respecting both the prior for model pa- rameters and the distance constraint (Beaumont et al. 2008) . g,i/(cid:80)Npart i=1 w∗ Beaumont et al. (2008) recommend using a Gaussian mix- ture model for the importance sampling density, φ∗ ∼ pIS,g+1(φ) = wg,iNφ(φg,i, τ Σg), (6) i=1 where the covariance is scaled by τ (cid:39) 2 relative to the ABC posterior, so as to reduce the risk of some particles being assigned large weights. By performing this scaling, we ef- fectively sample generation g+ 1 from an expanded posterior of generation g. We found that sometimes this can result in the particle population being dominated by a small number of particles that are assigned large weights. To reduce this risk and make the ABC-PMC algorithm more efficient, we suggest truncating each component of the mixture model to exclude values of φ∗ that result in (φ∗ − φg,i)(cid:48)(τ Σg)−1(φ∗ − φg,i) ≥ Xcrit, where Xcrit = 2√nparam and nparam is the number of model parameters being characterized by ABC. Formally, this can result in some valid regions of parameter space for φ being assigned zero probability by the importance sampler. In practice, this is not a problem for our application, since any such regions are extremely unlikely to result in simu- lated data set with ρ(S(φ∗),sobs) ≤ g+1. Note that truncation is not used when approximating the final ABC posterior, so the ABC posterior is non-zero for all φ (with prior support). During the course of this study, we found that the dis- tance threshold i can be decreased relatively rapidly in early generations. Eventually, i becomes small enough that sam- pling noise is likely to cause a simulated data set to have a distance larger than i. At this point, further reductions in Npart(cid:88) 8 HSU ET AL. i rapidly become computationally prohibitive. Therefore, we developed a set of heuristics to automatically terminate the sequential importance sampler once the desired preci- sion had been reached or once the algorithm efficiency has dropped to the point where it is no longer making signifi- cant improvements. We halt once of the following events oc- curs: (1) The mean distance between the generated catalogs and the true observed catalog drops below a target distance threshold , (2) the number of generations reaches a maxi- mum limit, Nmax,gen = 200, (3) the number of consecutively repeated states exceeds a threshold which we set equal to the the total number of particles. A particle's state is repeated when the algorithm fails to generate a set of model parame- ters resulting in ρ(s∗,sobs) < i after Nmax trial sets of model parameters. The algorithm tracks how many times each parti- cle has repeated its current set of parameter values, and resets the counter each time the particle is successfully updated. If the number of consecutively repeated states summed over all particles in a generation exceeds the number of particles per generation, then the algorithm halts. (4) during the most re- cent generation, the median number of trial sets of model pa- rameters required to generate a successful set of Npart model parameters (i.e., ρ(s∗,sobs) < i) exceeds 0.2× Nmax, or (5) the algorithm fails to improve the distance threshold for three consecutive generations due to at least one particle in each generation requiring a number of attempted draws greater than 0.75 Nmax. Ideally, one would use only Criteria 1, so as to achieve the desired level of convergence. Often, one does not know a priori what level of convergence is practical. In such cases, one can set to be so small that criterion 1 is un- likely to be met. In this case, the algorithm will proceed until one of criteria 2, 3, 4 or 5 are met, at which point further im- provements in become extremely computationally expen- sive. In our calculations, criteria 2 and 3 were never invoked. Instead, criteria 4 and 5 was found to be effective stopping criteria, recognizing when further calculations would yield negligible improvements in . At this point, the ABC pos- terior width had converged to very near the the Monte-Carlo limit for the vast majority of our calculations (see Fig. 2). 2.4. Summary Statistics The ability of the ABC posterior to approximate the true posterior depends on using a distance function that can iden- tify simulated data sets that are similar to the observed data set. In practice, the distance function (ρ) is usually chosen to be a function of a set of summary statistics (S), rather than a function of all the data (Y ). For example, consider the problem of characterizing the distribution of a population of values (xi), each drawn from a normal distribution with unknown mean (µ) and variance of unity. One can compute the posterior distribution for µ using only (cid:104)x(cid:105), the mean value of the xi's. If one knows (cid:104)x(cid:105), then also knowing the specific value of each of the xi's would not affect one's estimate of µ or the population from which xi's are drawn. For this reason, (cid:104)x(cid:105) is referred to as a "sufficient statistic" to characterize a normal distribution with known variance. When using sufficient statistics and a reasonable distance function, one can prove that the ABC posterior will approach the true Bayesian posterior for sufficiently small . Similarly, if the true distribution were normal with unknown mean and unknown variance, the sample mean and sample covariance would form a set of sufficient statistics. Formally, the full data set is also a sufficient statistic, but this is rarely useful, since it is unlikely that one would draw values close to each of the xi's, even if one were using the true population mean and variance. In this paper, we will characterize planet occurrence rates over a 2-d grid in planet size and orbital period, focusing on a specific set of Ntarg host stars. We define ns,r,p to be the num- ber of planets with size Rp,r < Rp ≤ Rp,r+1 and orbital period Pp < P ≤ Pp+1 orbiting star s. We assume that each ns,r,p is drawn independently from a Poisson distribution with oc- currence rate fr,p that is constant over the relevant range of radii & periods. If we assume a known number of target stars, that this model is the true distribution of planets, and that all planets are detected, then the number of planets in each bin would form a sufficient set of summary statistics. Even if not all planets are detected, the detected number of s ns,r,p are sufficient summary statistics provided that the probability of detection for each planet is known. Therefore, we adopt the ratio of the number of planets in each bin to the overall number of target stars (S(Y ) = {nr,p/Ntarg}) as our summary statistics. Of course, it is usually necessary to perform statistical inference with a model whose assumptions are not exactly true. We discuss some of the more obvious issues related to the choice of sum- mary statistics below and return to discuss the issue of model misspecification further in §2.6. planets in each bin: nr,p =(cid:80) We chose bin boundaries of {0.5, 1.25, 2.5, 5, 10, 20, 40, 80, 160, 320} days & {0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 2.5, 3, 4, 6, 8, 12, 16} R⊕, following Christiansen et al. (2015) and Howard et al. (2012). Of course, future applications could make alternative choices for the boundaries of planet size and orbital period bins. If one were to adopt bins that were very large, then the assumption of a constant occurrence rate within the bin could become inadequate. If one were to adopt bins that were very small, then most bins would have zero to a few planets, causing the variance of the posterior occurrence rate to be large. In principle, one could adopt a prior that assumes occurrence rates in nearby bins are cor- relate (Foreman-Mackey et al. 2014) to reduce the variance while retaining significant model flexibility. In this study, we assume rates are uncorrelated between bins, so as to simplify the interpretation and comparisons to previous studies. PLANET OCCURRENCE RATES USING ABC 9 Since orbital periods are measured precisely, there is es- sentially no error in assigning planets to period bins for any reasonable bin width. Uncertainty in transit depths and stellar radii propagate to cause a modest difference between the true and estimated planet radii, resulting in some planets near a boundary in planet size being assigned to the "wrong" radius bin. ABC can accounts for this effect, by using a forward model that generates both true and observed stellar and plan- etary radii. The main consequence of uncertainty in planet size is that any sharp changes in the true occurrence rate as a function of planet radius will appear to be less sharp in the observed (and simulated) catalogs. A more subtle ef- fect of the uncertainty in stellar radii is that the posterior for the planet occurrence rate in neighboring bins in planet size could be anti-correlated. If a planet were near the bound- ary of two bins in planet size, then it could be assigned to one bin or the other, but not both bins or neither bin. If one asked what is the occurrence rate for a range of planet sizes spanned by both bins, then the sum of the occurrence rates would be accurate, but the uncertainty in the sum of the oc- currence rates for the two bins would not be as large is if the two bins were uncorrelated. We investigate the strength of these effects in §5.2. For more complex problems (e.g., characterizing planetary architectures), it is likely not possible to identify a small set of sufficient summary statistics. Yet, the power of ABC can be traced to the dimensional reduction that comes from a rel- atively small set of summary statistics encapsulating the key features of the data. Just as a standard Bayesian analysis must choose an appropriate prior and likelihood, in an ABC anal- ysis, it is important to choose an appropriate set of summary statistics, so as to encode the key physical properties of the population for the scientific questions at hand. 2.5. Distance Functions The ABC algorithm requires specifying a function to com- pute the distance (ρ) between the summary statistics for ob- served catalog sobs and the summary statistics for each sim- ulated planet population (s∗). For the simulations shown in this paper, we consider the planet occurrence rate for each bin of planet size and orbital period separately. When char- acterizing the occurrence rate of a single bin, the summary statistic is a scalar, so the choice of the distance function is obvious: ρ((sobs,s∗) = (sobs − s∗)2. In other words, the trial parameters (φ∗) will be accepted in the ABC posterior for generation g if the result of the forward model produces a frequency of detected planets (nr,p/Ntarg) that is within √ g of the Kepler observed frequency sobs. In the primary applications for this paper, each bin is inde- pendent of the other bins. However, to illustrate how ABC could be applied to more complex models with multiple in- terdependent summary statistics, we also explore distance functions that simultaneously fit occurrence rates in multi- ple different bins. In simulations that consider the occur- rence rate for multiple bins simultaneously, the summary In preliminary simulations, statistics are no longer scalar. we explored various choices for the distance function, in- cluding: 1) ρmax(sobs,s∗) = maxksobs,k − s∗ k, 2) ρL1(sobs,s∗) = k2, where the maximum or sum over k is over each of the summary statis- tics (i.e., over all the bins). k, and 3) ρL2(sobs,s∗) =(cid:80) (cid:80) ksobs − s∗ ksobs − s∗ When analyzing models that include multiple bins with widely varying values of fr,p, we found that ρL1 or ρL2 could result in less precise estimates of some fr,p's. However, when considering occurrence rates for a large number of bins si- multaneously (i.e. > 5 bins) ρmax struggles significantly in the efficiency of convergence. Due to this poor efficiency for the ρmax distance function, we choose to report results based on the ρL2 distance function in this study. In this study, each of the summary statistics is a sample occurrence rate. In fu- ture studies where summary statistics are not naturally com- parable, it will likely be advantageous to normalize or scale the different summary statistics. Our implementation of the ABC-PMC algorithm described above is provided by the ABC.jl package (Ford et al. 2018a) for the Julia programming language (Bezanson et al. 2014) and is available at https://github.com/eford/ ABC.jl/tree/hsu_etal_2018-v1.0 under the MIT Expat License. ABC.jl was designed to be generally ap- plicable and is not specific to our current applications of characterizing exoplanet populations. 2.6. Model Misspecification One should be concerned about the potential effects of model misspecification when applying any statistical model to a real physical situation. In the case of a traditional Bayesian analysis, this concern is manifest in the choice of the likelihood function. In the case of ABC, concerns about model misspecification can appear in either the for- ward model or the choice of summary statistics. While the potential for model misspecification likely renders the choice of summary statistics imperfect, the underlying concerns of model misspecification are not unique to ABC, but apply to any statistical analysis, whether frequentist or Bayesian, with or without a likelihood. Dealing with model misspecification is a major problem well beyond the scope of this paper. We will discuss two examples of how our model is likely imperfect below. First, we note that either ABC or a hierarchical Bayesian model are less sensitive to deviations of the true and assumed planet de- tection efficiency model than the IDEM method, since IDEM uses a point estimate for the planet detection efficiency. Fur- ther, the bias in IDEM persists, even when planet detection efficiency is known exactly. Using ABC offers the benefit 10 HSU ET AL. of a rigorous statistical foundation for performing inference even in presence of the inevitable uncertainty in the planet size. The distribution of planets likely differs from the assumed piecewise constant model (or any other possible assumed dis- tribution). Simple parametric planet occurrence rate models (e.g., power-law) are prone to bias due to model misspeci- fication, particularly if the empirical constraints are signifi- cantly stronger in one region of parameter space (e.g., where planet detections are more abundant) than another region where one would like to characterize the predictive distri- bution (e.g., Earth-size planets in habitable zone). A non- parametric model for the planet occurrence rate provides in- creased flexibility to reduce such risks of introducing a bias in the occurrence rate of Earth-analogs due to model mis- specification. Of course, the specific choice of number and location of bin boundaries is subjective. We discuss the asso- ciated tradeoffs in §2.4 and adopt bin boundaries following Christiansen et al. (2015) and Howard et al. (2012). Similarly, the planet detection efficiency model is not known perfectly. Just as a standard frequentist or Bayesian analysis using a likelihood based on an imperfect detection model would be biased, the results of ABC would be biased by an inaccurate detection probability model. Fortunately, the magnitude of this bias goes to zero as the assumed planet detection efficiency model approaches the true planet detec- tion efficiency. The Kepler team has performed numerous simulations to characterize the detection probability as a function of the injected planet parameters. We apply a de- tection probability for each of the detected planets based on Monte Carlo simulations from transit injection studies (e.g., Christiansen et al. 2015). This dramatically mitigates the concern about the accuracy of the planet detection efficiency model, particularly relative to analyses which assume a de- tection efficiency model without the benefit of calibration to Monte Carlo simulations. Future studies can explore the sen- sitivity of results to the choice of planet detection efficiency model. 3. PHYSICAL MODEL: SYSSIM Our forward model has two key components: a process for generating a set of target stars and associated planets, and a model for "observing" the targets to obtain a cata- log of detected planets and their estimated properties. We describe the model for generating target stars and planets in this section and the model for generating simulated ob- served catalogs in §3.3. We refer to both parts of the forward model as SysSim, short for the Planetary Systems Simula- tor. SysSim is an empirical model which generates planetary systems using flexible parameters (here fr,p) that can be later compared to planet formation hypotheses. Its primary goal is to aid in interpreting Kepler data to characterize the true, underlying exoplanetary population. SysSim is implemented in Julia (Bezanson et al. 2014) by the ExoplanetsSysSim.jl package (Ford et al. 2018b) and is available at https: //github.com/dch216/ExoplanetsSysSim.jl/ tree/hsu_etal_2018-v1.0 under the MIT Expat Li- cense. SysSim was designed to be highly extensible and also contains many additional features, some tested and some still under development, which we intend to present in future publications. 3.1. Stellar Properties For each simulated planet catalog, we draw a sample of Ntarg = 150,518 target stars from the Q1-Q16 catalog of Ke- pler target stars (Huber et al. 2014). For each target star, we store the estimated star mass, estimated star radius, the robust root mean square combined differential photometric precision on 4.5 hour timescale (CDPP), the timespan from the first to the last Kepler observation of the target, and the duty cycle (i.e., the fraction of above time span during which usable Kepler data is available), as well as upper and lower uncertainty estimates for the star mass and radius. For this study, we populate the catalog of potential tar- get stars with a subset of the Kepler stellar properties cata- log, as obtained from the Exoplanet Archive at NExScI in June 2015, which derive from Huber et al. (2014). Fol- lowing Christiansen et al. (2015), we aim to include main sequence FGK stars by including only target stars that: 1) were observed for at least one quarter by Kepler in Q1-Q12, 2) have estimated properties of 4000K < Teff < 7000K and logg > 4.0, and 3) have valid values and uncertainties for star mass, radius, density, and CDPP. Our resulting target star list is very similar to that of Christiansen et al. (2015), but differs slightly, since some of the data in the star properties catalog at NExScI has been updated since the study. For example, we find 150,518 stars, which is slightly less than the 152,066 found by Christiansen et al. (2015). Our approach for assigning stellar properties differs from that of assigning planet properties (see §3.2), since we are not attempting to infer properties of the Kepler target star popu- lation. In principle, one could try to simultaneously infer the properties of the Kepler host star population and the planet population. However, this would require substantially in- creasing the number of model parameters and observational constraints, as well as modeling the complex target selection process for choosing which targets would be downloaded. Future studies may find it advantageous to model both the host star and planet populations. This could be particularly useful when applied to results from the upcoming TESS mis- sion. Since the TESS mission plans to download full frame images, the target selection process is significantly simplified relative to that of the Kepler planet search. 3.2. Planet Model PLANET OCCURRENCE RATES USING ABC 11 planets from a Poisson distribution with rate, λ =(cid:80) Each planet is assigned physical properties (i.e., radius, period, orbit) by drawing from proposed distributions for a given model. The sampling distribution for each planet is set by the model parameters and is not related to measurements of any specific planet candidate in the Kepler catalog. Planet properties are drawn independently, even for planets orbiting the same star. For each target star, we draw the number of r,p fr,p, where r and p range over all planet radius and orbital period bins being included in the simulation at hand. For this paper, the summary statistics depend only on the total number of planets in each bin. Since planets are drawn independently of each other, we could have used a simpler model that did not allow for multiple planets around the same star. Never- theless, the code implements this feature in preparation for future studies with SysSim that will study the architectures of planetary systems. Note that λ is the average number of planets per star (in the entire radius/period range). Experi- mentation showed that large λ would slow the analysis and was inconsistent with the observations, so the Poisson dis- tribution is truncated at Nmax,pl = 10. For each planet, we first assign a radius-period bin by drawing from a categorical distribution with the probability for each bin proportional to the rate fr,p and normalized so the probabilities sum to unity. Next, we draw the precise orbital period and planet size, uni- formly in log period and log radius constrained to be within the respective ranges of planet's assigned bin. Each planet is assigned a Keplerian orbit, where the eccentricity is drawn from a Rayleigh distribution with scale parameter σhk = 0.03 (Fabrycky et al. 2014). Such small eccentricities have only a modest effect on the planet occurrence rate (Burke et al. 2015), but we include full Keplerian orbital parameters for the sake of completeness and future applications. Next, incli- nations are drawn so that orbital planes are oriented isotrop- ically over the sky (i.e. cosi ∼ U(−1,1)). Since we are con- sidering only the rate of planets per star (per bin in period and radius) and not the properties of multiple planet systems, the distribution of mutual inclinations of planets does not affect our calculations. The angles (argument of periastron, longitude of ascending node, and mean anomaly at the reference epoch) are drawn from uniform distributions. Collectively, each simulated cat- alog of target stars and their planets is known as a "physical catalog", since this contains all the stars and planets (includ- ing those that will not be detectable), along with their true properties (as opposed to their estimated properties). 3.3. Observed Catalog For each simulated physical catalog, we construct a simu- lated "observed catalog" that will be compared to the actual catalog of Kepler planet candidates. Rather than generating full light curves for each target, we compute the primary ob- servables that characterize each transiting planet: the orbital period (P), the time of the 0th transit (t0), the fractional transit depth (d = Rp/R(cid:63)), and the transit duration (D). Next, we compute the probability that each planet would be detected. In practice, the detection probability depends on two parts: the geometric probability that the planet transits the star for a given observer and the probability that the planet would be detected if it were known to transit the star. For the sake of computational efficiency, we first compute the proba- bility that the planet would be detected if the orbital plane re- sulted in the planet transiting across the diameter of the star. For this study, we we adopt the Γ distribution CDF model for the planet detection efficiency (pdet) which accounts for the transit depth, duration, CDPP, orbital period, timespan of observations and duty cycle (Christiansen et al. 2015). We multiply the above detection probability by a binomial ana- lytic window function to ensure that only planets transiting at least three times are considered detectable (pwin,≥3), fol- lowing (Burke et al. 2015). At this point, planets with neg- ligible detection probability (even if observed edge-on) are discarded from the observed catalog. The remaining planets are considered "potentially detectable", meaning there is a non-negligible probability that the planet would be detected if the viewing geometry were favorable. Next, we account for the fact that not all planets transit their star for a given viewing geometry. At this point, there are two possible approaches: adopting a single viewing ge- ometry or averaging over all possible viewing geometries. The simplest approach is to compute which of the potentially detectable planets actually transit the star given their orbits and star properties for a single observer (i.e., using the spe- cific i and ω of each planet). In this case, we simply test whether acos(i)(1 − e2) ≤ R(cid:63)(1 + esin ω), which requires that the center of the planet pass inside the disk of the star, to determine whether the geometric transit probability pgeo for this specific planet and viewing geometry is unity or zero. In our ABC simulations, the stellar radius R(cid:63) used for cal- culating the transit probability is the physical stellar radius which is assigned based on the observed radius and its un- certainty reported in the input stellar catalog. (We ignore the small numbers of planets with impact parameters greater than unity, since such transits are rare and are so hard to dis- tinguish from grazing eclipsing binaries that such a transit would often be dismissed as a likely astrophysical false pos- itive.) For each simulated potentially detectable planet that transits the disk of its host star, we compute an updated tran- sit detection probability that now accounts for its actual im- pact parameter which typically results in reducing the tran- sit duration and the integrated transit signal-to-noise. Each potentially detectable planet that transits its host star is la- beled as either detected or not detected by drawing from a Bernoulli distribution with the updated transit detection prob- 12 HSU ET AL. ability. This results in a combined probability of detecting a planet, pcomb = pgeo pdet pwin,≥3). Each detected planet is added to an observed catalog, along with of its measured transit parameters, referring to the mea- sured orbital period ( P), measured time of the 0th transit (t0), measured fractional transit depth ( d), and measured transit duration ( D). Each measurement is assumed to be normally distributed about the true value with dispersion based on the transit signal-to-noise and uncertainty in stellar parameters. For simplicity, we use only the diagonal terms of the covari- ance matrix for the transit parameters, as given by Eqn. A8 and Table 1 of Price & Rogers (2014). Then, planets are assigned to their radius-period bin based on their observed radius and period, which are calculated using the measured transit parameters in the observed planet and stellar catalogs. If a planet's observed radius falls outside the range of radii under consideration, then it is excluded when counting the number of planets detected in the relevant bin(s). The sum- mary statistics are simply the number of detected planets in each planet radius-period bin divided by the number of target stars within the catalog, {nr,p/Ntarg}. An alternative approach is to generate a probabilistic cat- alog, where each potentially detectable planet is included along with a weight proportional to its combined probabil- ity. Future studies characterizing the distribution of plane- tary architectures could benefit from CORBITS (Brakensiek & Ragozzine 2016) which can efficiently compute the sky- averaged geometric probability of any combination of plan- ets transiting their host star using nearly algebraic formu- lae. While SysSim includes support for probabilistic cata- logs, this feature was not used for this study, since we focus on properties of individual planets rather than planetary sys- tems. We provide a brief overview of this approach in Ap- pendix C. 4. VERIFICATION AND VALIDATION Having described our statistical framework in §2 and the particulars of our physically-motivated forward model in §3, we proceed to the task of verification and validation of our code and algorithms before applying them to actual science data. In this section, we verify that our implementations of ABC-PMC and SysSim, as well as our choices for the sum- mary statistics and distance function, accurately characterize planet occurrence rates for simulated data sets. We also val- idate that our refinements to ABC described in §2.3 actually improve the algorithms' efficiency, at least for our specific problem. 4.1. Methods For our initial verification tests, we generated synthetic "observed" catalogs from our forward model that will be used in place of the actual catalog of Kepler planet candidates for the purpose of validating our algorithm & code through- out this section. Both the synthetic observed catalog and cat- alogs simulated during the ABC process are parameterized by the planet occurrence rates ( fr,p) for population level pa- rameters, φ. We apply ABC-PMC to generate pABC,( fr,pYobs), a Gaus- sian mixture model approximation for the ABC posterior dis- tribution with tolerance for each bin in planet radius and orbital period, indexed by r and p for our synthetic data Yobs. Since we marginalize over the physical properties of each planet, the ABC posterior is a function of only the population-level parameters, in this case fr,p. During the se- quential importance sampling phase of ABC, we use 40 "par- ticles" for our mixture model. We run ABC for several gen- erations, decreasing the distance threshold until we reach the stopping criterion. For these validation runs, we set the tar- get distance threshold, , to zero. In practice, this resulted in ABC stopping once the median number of attempted draws per particle exceeds 20% the maximum number of attempts per particle, Nmax = 50. We compute the mean ( fr,p) and a 68.3% credible interval of the ABC posterior for each fr,p using the weights from importance sampling and compare this credible interval to the true simulated planet occurrence rate. 4.2. Results of V&V 4.2.1. ABC Posteriors are Accurate We confirm that the mean of the ABC posterior provides an accurate estimate of the true occurrence rate. We repeat this experiment ten times for three different bins using dif- ferent simulated catalogs to confirm that that 68.3% credible interval for the occurrence rate accurately approximates the difference between the posterior mean and true occurrence rate and that there is no detectable bias (e.g., Fig. 1, bottom panel). Note that our ABC posteriors behave much better than those of the inverse detection efficiency method which can give significantly biased estimates (see Fig. 1, top panel). We provide a more detailed description of the inverse detec- tion efficiency method in Appendix A. 4.2.2. Width of ABC Posterior & Choice of Distance Tolerance In principle, the width of the ABC posterior should be larger than width of the true posterior when the final distance tolerance is greater than zero. If the distance function is cho- sen well and the ABC simulation is executed effectively, then will be sufficiently small that the width of the ABC poste- rior will be close to that of the true posterior. In this sec- tion, we demonstrate that the width of ABC posteriors very closely approximates the correct posterior width by compar- ing results for different values of and by comparing to a simplified Bayesian model in the limit of low fixed transit PLANET OCCURRENCE RATES USING ABC 13 noise included in the forward model for ABC (see Appendix B). First, we characterize how σ, the half-width of the 68.3% credible interval for the ABC posterior for f , depends on the . For this purpose, we perform tests using a single bin in planet size (Rp = 1 − 1.25R⊕) and period (P = 10 − 20 days). For each simulation we plot the half-width of the ABC pos- terior, σg, as a function of ρg, the median distance of the par- ticles in the gth generation of the sequential importance sam- pler (Fig. 2). During early generations, ρg steadily decreases and σ improves, moving towards the bottom left of the fig- ure. Eventually, randomness inherent in the forward model and the finite number of targets results in the ABC posterior width fluctuating, even as g continues to decrease. As long as one uses an value no greater than the point where σg plateaus, the ABC posterior width is insensitive to the choice of . However the computational cost can increase signifi- cantly, if one insists on a significantly smaller value of . Next, we compare the width of the ABC posterior to es- timates for the standard deviation of the posterior from a Bayesian analysis for a simplified model which accounts for the finite sample size, but does not require ABC (solid lines). We see in Figure 2 that the widths of the ABC posterior for simulations with f = 10−1, 10−2, and 10−3 are very near the theoretical limits for the posterior width of the simplified Bayesian model that we describe in Appendix B. In the case of the f = 10−4 simulation, the ABC stopping criterion was met before σg had clearly plateaued. If we were to continue to run the ABC simulations with even smaller , then the ABC posterior width might decrease further, but this would be computationally expensive. Based on comparing to the posterior width for the simplified Bayesian model, we expect that the posterior width could not decrease by more than a factor of ∼ 2. It is also worth exploring why the ABC poste- rior width is greater than the width of the simplified Bayesian model. For the f = 10−4 simulation, there were 0 planets de- tected, so the data impose a strong limit on the planet occur- rence rate, but the shape of the posterior at smaller rates will be sensitive to the choice of priors. Therefore, if zero planet candidates are detected in any bin, then we report only only upper limits on the planet occurrence rate. For example, in the case of the f = 10−4 simulation, we would report a limit of < 6× 10−4. In contrast, for bins with no detected planets, the IDEM method as commonly used in the literature provides neither an estimate of the rate nor an upper limit. Of course, it is possible to compute upper limits using either a maxi- mum likelihood or a Bayesian analysis without using ABC (e.g., see Appendix A). 4.2.3. Width of ABC Posterior as a Function of Orbital Period Since the transit probability decreases with increasing or- bital period, there are more planets detected at small orbital Figure 2. Standard deviation, σ, of the ABC posterior for the oc- currence rates as a function of ρ, the median distance within each generation for P = 10−20 days and Rp = 1−1.25R⊕ & N(cid:63) = 150,518. The symbol style/color indicates simulations with different under- lying occurrence rates. Each simulation starts from the top right and decreases in σ towards the bottom left over the course of the run. The colored lines correspond to the inevitable Monte Carlo un- certainty, estimated by the posterior width for simulations using the Simplified Bayesian model described in Appendix B. The plateau of the different simulations at their respective Monte Carlo uncertain- ties show that our simulations reach the Monte Carlo noise limit. periods. Therefore, we expect the precision of planet occur- rence rates will increase with orbital period for a fixed oc- currence rate and the width of the ABC posteriors will in- crease for bins with increasing orbital period. To explore this effect, we perform simulations on bins over the full pe- riod range, keeping the radius range of each bin the same (Rp = 1 − 1.25R⊕). For one set of simulations, the true oc- currence rate used to generate the observed catalog is fixed at f = 0.015 per star (Fig. 3). First, we note that the ABC posterior accurately estimates the true occurrence rates (not shown) for all of these cases. As expected, we find more pre- cise estimates of the estimated occurrence rate (i.e., smaller values of σ) at shorter orbital periods. For periods less than 20 days, the width of the ABC posterior has clearly plateaued indicating that further reductions in will not affect the width of the ABC posterior. For orbital periods greater than 80 days, ABC has not reached a plateau, so smaller values of might result in a smaller uncertainty on planet occurrence rates. 4.3. Number of Model Parameters For most of this study, we perform inference on a single population-level model parameter at a time. However, in §5.2 we report results based on simulations simultaneously estimating occurrence rates for multiple bins include differ- ing planet size ranges, but a common orbital period. We in- terpret the results to understand the effect of uncertainty in −6.0−5.5−5.0−4.5−4.0−3.5−3.0−2.5−2.0log10ρ−4.0−3.5−3.0−2.5−2.0−1.5−1.0−0.50.0log10σf=10−1f=10−2f=10−3f=10−414 HSU ET AL. Figure 3. Standard deviation, σ, of the ABC posterior for the oc- currence rates as a function of ρ, the median distance within each generation for N(cid:63) = 150,000. The true occurrence rate used to gen- erate the observed catalog is fixed at f = 0.015. The symbol style indicates simulations with bins of different orbital period ranges. Each simulation starts from the top right and decreases in σ towards the bottom left over the course of the run. As expected, more pre- cise estimate of the occurrence rate (i.e., smaller σ) are achieved at shorter orbital periods, primarily due to the increased transit proba- bility. planet radii on the joint ABC posterior for planet occurrence rates in bins of neighboring planet sizes. Before performing such tests, we need to verify that the ABC-PMC algorithm also performs acceptably when esti- mating multiple fr,p's simultaneously. Therefore, we charac- terize the computational efficiency of our algorithm as a func- tion of the number of planet radius-period bins being char- acterized simultaneously. These simulations also serve as a stepping stone to future studies that include more population- level model parameters. For these experiments, we choose the true values of the planet occurrence rates ( fr,p) to be similar to those in the DE2 detection table in Figure 7 of Christiansen et al. (2015). For the distance function, we use ρmax from §2.5 (i.e. compute the absolute difference between the ratio of the number of detected planets to Ntarg in the "observed" planet catalog and the ratio of the expected number of detected planets to Ntarg in the simulated catalog, and take the maximum over all the bins included in the current simulation). We perform tests inferring values of fr,p for 1, 2, 4, 8, 16 and 32 size-period bins simultaneously. We find that the ABC algorithm successfully converged around the true oc- currence rates for the 1, 2, and 4 bin tests. However, it be- came very inefficient for the 8, 16, and 32 bin tests even be- fore achieving an that is acceptably small (see Fig. 4), i.e., slow was significantly larger than the needed to realize the most precise possible ABC estimates for the fr,p's. 5. APPLICATION TO KEPLER DATA Figure 4. Standard deviation (σ) of the inferred planet occurrence rate as a function of the final distance tolerance () for different choices of the number of occurrence rate parameters, Nbin, with N(cid:63) = 40,000 and a "true" occurrence rate f = 0.015. Each simula- tion starts from the top right and decreases in σ towards the bottom left over the course of the run. Our implementation of ABC works well for inferring up to 4 parameters simultaneously, but modifica- tions would be required to efficiently perform inference with ≥ 8 parameters simultaneously. Having validated that our algorithm and implementations of ABC-PMC and SysSim can accurately characterize planet occurrence rates, we proceed to apply this framework to a re- cent Kepler planet catalog. The catalog of target star proper- ties is described in §3.1. For observed planet candidate cata- log properties, we adopt the Q1-Q16 planet candidate catalog obtained from the NASA Exoplanet Archive in March 2017 (Mullally et al. 2015; Rowe et al. 2014). We find Np = 3380 planet candidates associated with our target stars and having estimated planet radii (Rp) in the range 0.5−16R⊕ and orbital periods (P) in the range 0.5 − 320 days. For this catalog, the detection efficiency curve we apply is a Γ distribution CDF with α = 4.65 and β = 0.98, which is the appropriate detection efficiency given the completeness properties of the Q1-Q16 data release. This input catalog has since been updated to reflect new information since Christiansen et al. (2015). It is worth noting, however, that the Q1-Q16 data release has not had a systematic search across the full parameter space con- sidered in this study because of issues which are discussed in detail in Burke et al. (2015). 5.1. Comparison of ABC and Inverse Detection Efficiency Methods First, we perform separate simulations to characterize the planet occurrence rate for each planet size-period bin individ- ually. For each test, the target was set at an unrealistically small 10−10 and the maximum number of generations was set to 200, so that ABC would continue to run until trigger- −6−5−4−3−2−10log10ρ−4.5−4.0−3.5−3.0−2.5−2.0−1.5−1.0−0.50.0log10σP = 0.5-1.25 daysP = 1.25-2.5 daysP = 2.5-5 daysP = 5-10 daysP = 10-20 daysP = 20-40 daysP = 40-80 daysP = 80-160 daysP = 160-320 days−6−5−4−3−2−1log10ǫ−4.0−3.5−3.0−2.5−2.0−1.5−1.0−0.5log10σN_bin = 1N_bin = 2N_bin = 4N_bin = 8N_bin = 16PLANET OCCURRENCE RATES USING ABC 15 ing one of the termination heuristics that recognize when fur- ther computation will not result in significantly reducing g of the final generation (see §2.3). We find that the ABC pos- terior distributions for all occurrence rates are unimodal. For the vast majority of bins, the ABC posterior appears roughly Gaussian, with the exception of bins with zero to a few plan- ets detected. We summarize our results, by reporting the pos- terior mean of each planet occurrence rate and the 68.27% credible interval based on the 15.87% and 84.13% percentiles in Table 1. In Figure 5 we report the maximum of the differ- ence between the 84.13% percentile and the mean and the difference between the mean and the 15.87% percentile. Period (days) 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 0.50 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 1.25 − 2.50 − 2.50 − 2.50 − 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 1.25 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 2.50 5.00 5.00 5.00 Radius (R⊕) 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 Table 1. Occurrence Rates ABC Bayesian IDEM (Full Model) −2.9 × 10−4 7.3+4.7 −0.27 × 10−3 1.18+0.22 −1.4 × 10−4 8.9+2.2 −1.30 × 10−4 5.19+0.96 −1.1 × 10−4 5.3+1.3 −0.78 × 10−4 3.89+1.24 −0.45 × 10−4 1.05+0.52 −2.7 × 10−5 5.8+4.8 −2.2 × 10−5 3.6+3.5 −1.20 × 10−4 2.06+0.55 < 6.8× 10−5 −2.3 × 10−5 3.9+3.2 −2.7 × 10−5 4.1+3.2 −1.8 × 10−3 6.2+2.7 −0.74 × 10−3 3.76+0.63 −0.38 × 10−3 2.36+0.48 −0.26 × 10−3 1.23+0.15 −0.19 × 10−3 1.29+0.20 −1.2 × 10−4 5.3+1.9 −0.96 × 10−4 4.68+2.30 −1.2 × 10−4 4.8+1.3 −0.82 × 10−4 1.83+0.82 −0.80 × 10−4 1.67+0.76 −3.6 × 10−5 5.6+5.5 −0.46 × 10−4 1.15+1.12 −1.2 × 10−4 3.7+1.5 −0.95 × 10−2 4.94+1.07 −0.26 × 10−2 1.42+0.25 −0.74 × 10−3 7.60+1.01 (Simplified Model) −3.2 × 10−4 7.4+3.2 −0.24 × 10−3 1.20+0.24 −1.7 × 10−4 9.3+1.7 −1.2 × 10−4 5.1+1.2 −1.1 × 10−4 5.1+1.1 −0.96 × 10−4 3.75+0.96 −4.7 × 10−5 9.8+4.7 −4.0 × 10−5 7.3+4.0 −3.1 × 10−5 4.8+3.1 −4.6 × 10−5 9.6+4.6 < 2.4× 10−5 −3.1 × 10−5 4.8+3.1 −3.1 × 10−5 4.8+3.1 −1.7 × 10−3 5.9+1.7 −0.67 × 10−3 3.75+0.67 −0.37 × 10−3 2.19+0.37 −0.24 × 10−3 1.16+0.24 −0.24 × 10−3 1.24+0.24 −1.6 × 10−4 5.7+1.6 −1.4 × 10−4 4.7+1.4 −1.4 × 10−4 4.6+1.4 −0.90 × 10−4 2.08+0.90 −0.79 × 10−4 1.66+0.79 −5.3 × 10−5 8.3+5.4 −0.68 × 10−4 1.24+0.68 −1.1 × 10−4 3.3+1.1 −0.82 × 10−2 4.60+0.82 −0.19 × 10−2 1.30+0.19 −0.96 × 10−3 7.17+0.96 Table 1 continued (6.6± 3.3)× 10−4 (1.44± 0.29)× 10−3 (1.12± 0.21)× 10−3 (6.0± 1.4)× 10−4 (5.4± 1.2)× 10−4 (4.5± 1.2)× 10−4 (8.2± 4.7)× 10−5 (3.7± 2.6)× 10−5 (2.7± 2.7)× 10−5 (7.8± 4.5)× 10−5 N/A (2.8± 2.8)× 10−5 (3.0± 3.0)× 10−5 (1.25± 0.38)× 10−2 (4.90± 0.89)× 10−3 (2.47± 0.43)× 10−3 (1.30± 0.28)× 10−3 (1.39± 0.27)× 10−3 (6.1± 1.8)× 10−4 (4.2± 1.3)× 10−4 (5.0± 1.6)× 10−4 (2.0± 1.0)× 10−4 (1.27± 0.74)× 10−4 (3.4± 3.4)× 10−5 (8.0± 5.6)× 10−5 (3.0± 1.1)× 10−4 (5.7± 1.0)× 10−2 (1.48± 0.22)× 10−2 (7.7± 1.0)× 10−3 16 Period (days) 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 2.50 − 5.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 5.00 − 10.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 10.00 − 20.00 Radius (R⊕) 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 HSU ET AL. Table 1 (continued) ABC Bayesian IDEM (Full Model) −0.95 × 10−3 7.72+0.93 −0.82 × 10−3 6.12+1.09 −0.45 × 10−3 4.02+0.54 −0.45 × 10−3 3.36+0.56 −0.37 × 10−3 2.10+0.45 −0.23 × 10−3 1.33+0.26 −0.33 × 10−3 1.36+0.43 −1.2 × 10−4 5.6+2.2 −2.0 × 10−4 8.6+2.2 −0.19 × 10−3 1.19+0.41 −1.9 × 10−2 9.2+1.9 −0.60 × 10−2 5.35+0.76 −0.28 × 10−2 2.41+0.22 −0.18 × 10−2 1.84+0.21 −0.15 × 10−2 1.24+0.13 −0.69 × 10−3 8.70+0.82 −0.15 × 10−2 1.16+0.14 −1.1 × 10−3 9.6+1.1 −0.73 × 10−3 5.84+0.64 −0.52 × 10−3 2.67+0.83 −0.32 × 10−3 1.01+0.39 −0.29 × 10−3 1.52+0.31 −2.8 × 10−4 7.4+2.2 −0.45 × 10−1 1.19+0.58 −1.3 × 10−2 6.3+1.9 −0.95 × 10−2 5.04+0.64 −0.31 × 10−2 3.63+0.76 −0.24 × 10−2 2.16+0.31 −0.17 × 10−2 1.93+0.15 −0.16 × 10−2 2.59+0.24 −0.18 × 10−2 2.31+0.22 −0.13 × 10−2 1.42+0.11 −0.61 × 10−3 4.78+1.19 −0.52 × 10−3 1.81+0.60 −0.58 × 10−3 2.25+0.49 −0.50 × 10−3 1.56+0.51 (Simplified Model) −0.79 × 10−3 7.00+0.79 −0.67 × 10−3 5.85+0.67 −0.53 × 10−3 3.97+0.53 −0.46 × 10−3 3.11+0.46 −0.38 × 10−3 2.15+0.38 −0.29 × 10−3 1.32+0.29 −0.30 × 10−3 1.38+0.30 −1.9 × 10−4 5.9+1.9 −2.3 × 10−4 8.5+2.3 −0.28 × 10−3 1.25+0.28 −2.0 × 10−2 8.6+2.0 −0.61 × 10−2 4.91+0.61 −0.25 × 10−2 2.17+0.25 −0.17 × 10−2 1.68+0.17 −0.12 × 10−2 1.14+0.12 −1.00 × 10−3 8.29+1.00 −0.11 × 10−2 1.04+0.11 −1.00 × 10−3 9.20+1.00 −0.76 × 10−3 5.50+0.76 −0.53 × 10−3 2.72+0.53 −0.32 × 10−3 1.04+0.32 −0.40 × 10−3 1.56+0.40 −2.9 × 10−4 8.3+2.9 −0.42 × 10−1 1.14+0.42 −1.1 × 10−2 5.8+1.1 −0.53 × 10−2 4.12+0.53 −0.33 × 10−2 3.15+0.33 −0.22 × 10−2 1.93+0.22 −0.18 × 10−2 1.66+0.18 −0.21 × 10−2 2.40+0.21 −0.19 × 10−2 2.11+0.19 −0.15 × 10−2 1.38+0.15 −0.89 × 10−3 4.84+0.89 −0.52 × 10−3 1.66+0.52 −0.61 × 10−3 2.33+0.61 −0.52 × 10−3 1.66+0.52 Table 1 continued (7.59± 0.86)× 10−3 (6.33± 0.73)× 10−3 (4.44± 0.60)× 10−3 (3.31± 0.50)× 10−3 (2.38± 0.43)× 10−3 (1.25± 0.29)× 10−3 (1.36± 0.30)× 10−3 (5.5± 1.9)× 10−4 (8.9± 2.6)× 10−4 (1.11± 0.26)× 10−3 (1.15± 0.28)× 10−1 (4.87± 0.61)× 10−2 (2.09± 0.24)× 10−2 (1.74± 0.18)× 10−2 (1.17± 0.13)× 10−2 (8.7± 1.1)× 10−3 (1.16± 0.12)× 10−2 (9.7± 1.1)× 10−3 (5.70± 0.80)× 10−3 (2.65± 0.53)× 10−3 (9.6± 3.2)× 10−4 (1.38± 0.37)× 10−3 (6.7± 2.5)× 10−4 (8.3± 3.4)× 10−2 (6.8± 1.3)× 10−2 (3.77± 0.49)× 10−2 (3.22± 0.34)× 10−2 (1.87± 0.21)× 10−2 (1.68± 0.19)× 10−2 (2.64± 0.23)× 10−2 (2.21± 0.20)× 10−2 (1.38± 0.15)× 10−2 (4.55± 0.86)× 10−3 (1.46± 0.49)× 10−3 (2.51± 0.70)× 10−3 (1.47± 0.49)× 10−3 PLANET OCCURRENCE RATES USING ABC 17 Period (days) 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 20.00 − 40.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 40.00 − 80.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 Radius (R⊕) 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 Table 1 (continued) ABC Bayesian IDEM (Full Model) < 2.5× 10−1 −0.29 × 10−1 1.60+0.39 −0.87 × 10−2 5.30+1.05 −0.45 × 10−2 3.93+0.50 −0.41 × 10−2 2.99+0.37 −0.26 × 10−2 1.62+0.21 −0.40 × 10−2 4.34+0.34 −0.25 × 10−2 2.94+0.26 −0.17 × 10−2 2.07+0.27 −1.2 × 10−3 9.1+1.7 −0.68 × 10−3 2.20+0.90 −1.0 × 10−3 3.6+1.1 −0.63 × 10−3 1.88+0.68 < 4.8× 10−1 < 1.0× 10−1 −1.9 × 10−2 7.5+2.1 −0.98 × 10−2 5.34+0.77 −0.81 × 10−2 3.80+1.06 −0.53 × 10−2 3.06+1.00 −0.66 × 10−2 4.80+0.60 −0.49 × 10−2 4.05+0.41 −0.31 × 10−2 2.57+0.25 −0.25 × 10−2 1.02+0.21 −1.6 × 10−3 7.4+2.4 −1.2 × 10−3 4.2+1.9 −0.76 × 10−3 1.99+1.20 < 1.4× 100 < 1.8× 10−1 −4.6 × 10−2 9.4+4.1 −2.1 × 10−2 8.0+3.2 −1.2 × 10−2 6.0+1.2 −0.62 × 10−2 3.58+1.05 −0.96 × 10−2 6.15+1.03 −0.63 × 10−2 4.24+0.68 −0.75 × 10−2 4.26+0.60 −0.36 × 10−2 1.47+0.49 (Simplified Model) < 1.1× 10−1 −0.30 × 10−1 1.38+0.30 −0.88 × 10−2 4.53+0.88 −0.51 × 10−2 3.33+0.51 −0.35 × 10−2 2.47+0.35 −0.23 × 10−2 1.36+0.23 −0.34 × 10−2 3.62+0.34 −0.28 × 10−2 2.68+0.28 −0.23 × 10−2 1.90+0.23 −1.5 × 10−3 8.4+1.5 −0.74 × 10−3 2.15+0.74 −0.99 × 10−3 3.75+0.99 −0.69 × 10−3 1.88+0.69 < 1.9× 10−1 < 4.3× 10−2 −1.7 × 10−2 5.9+1.7 −0.85 × 10−2 3.84+0.85 −0.56 × 10−2 3.00+0.56 −0.41 × 10−2 2.29+0.41 −0.47 × 10−2 4.00+0.47 −0.39 × 10−2 3.25+0.39 −0.32 × 10−2 2.26+0.32 −2.1 × 10−3 9.7+2.1 −1.7 × 10−3 7.0+1.7 −1.4 × 10−3 4.4+1.4 −0.95 × 10−3 2.20+0.95 < 4.8× 10−1 < 7.3× 10−2 −3.2 × 10−2 7.5+3.2 −1.7 × 10−2 5.6+1.7 −1.0 × 10−2 4.3+1.0 −0.65 × 10−2 2.69+0.65 −0.68 × 10−2 4.42+0.68 −0.52 × 10−2 3.28+0.52 −0.50 × 10−2 3.35+0.50 −0.30 × 10−2 1.30+0.30 Table 1 continued N/A (1.43± 0.32)× 10−1 (3.88± 0.78)× 10−2 (3.56± 0.56)× 10−2 (2.57± 0.37)× 10−2 (1.49± 0.25)× 10−2 (3.80± 0.36)× 10−2 (2.68± 0.28)× 10−2 (1.90± 0.23)× 10−2 (7.8± 1.4)× 10−3 (2.11± 0.80)× 10−3 (3.49± 0.97)× 10−3 (1.80± 0.73)× 10−3 N/A N/A (6.4± 1.9)× 10−2 (3.73± 0.86)× 10−2 (3.09± 0.59)× 10−2 (2.47± 0.45)× 10−2 (3.92± 0.47)× 10−2 (3.09± 0.38)× 10−2 (2.09± 0.30)× 10−2 (9.1± 2.0)× 10−3 (6.3± 1.6)× 10−3 (4.0± 1.3)× 10−3 (1.84± 0.92)× 10−3 N/A N/A (4.4± 2.2)× 10−2 (5.4± 1.7)× 10−2 (3.76± 0.94)× 10−2 (2.49± 0.62)× 10−2 (4.48± 0.70)× 10−2 (3.51± 0.56)× 10−2 (3.22± 0.49)× 10−2 (1.30± 0.32)× 10−2 18 HSU ET AL. Table 1 (continued) Period (days) 80.00 − 160.00 80.00 − 160.00 80.00 − 160.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 160.00 − 320.00 Radius (R⊕) 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 0.50 − 0.75 0.75 − 1.00 1.00 − 1.25 1.25 − 1.50 1.50 − 1.75 1.75 − 2.00 2.00 − 2.50 2.50 − 3.00 3.00 − 4.00 4.00 − 6.00 6.00 − 8.00 8.00 − 12.00 12.00 − 16.00 ABC Bayesian IDEM (Full Model) −2.0 × 10−3 8.5+2.4 −0.22 × 10−2 1.51+0.36 −0.97 × 10−3 1.79+1.55 < 1.4× 100 < 6.9× 10−1 −1.3 × 10−1 2.5+1.9 −0.44 × 10−1 1.81+0.76 −1.7 × 10−2 6.7+1.9 −1.7 × 10−2 6.0+3.3 −0.18 × 10−1 1.29+0.20 −1.27 × 10−2 5.08+0.87 −1.0 × 10−2 7.4+1.0 −0.58 × 10−2 1.85+0.46 −0.83 × 10−2 2.76+0.80 −0.47 × 10−2 1.68+0.52 −2.7 × 10−3 6.9+3.4 (Simplified Model) −2.4 × 10−3 7.9+2.4 −0.32 × 10−2 1.44+0.32 −1.2 × 10−3 2.2+1.2 < 8.0× 10−1 < 2.4× 10−1 −0.86 × 10−1 1.59+0.87 −0.45 × 10−1 1.30+0.45 −1.7 × 10−2 4.3+1.7 −1.3 × 10−2 4.1+1.3 −1.4 × 10−2 8.2+1.4 −0.72 × 10−2 3.19+0.72 −0.86 × 10−2 5.26+0.86 −0.46 × 10−2 1.62+0.46 −0.56 × 10−2 2.42+0.56 −0.46 × 10−2 1.61+0.46 −2.9 × 10−3 6.7+2.9 (7.7± 2.4)× 10−3 (1.44± 0.33)× 10−2 (1.6± 1.1)× 10−3 N/A N/A (6.2± 4.4)× 10−2 (1.74± 0.66)× 10−1 (4.7± 2.1)× 10−2 (3.4± 1.1)× 10−2 (8.7± 1.5)× 10−2 (3.14± 0.74)× 10−2 (5.59± 0.93)× 10−2 (1.62± 0.49)× 10−2 (2.41± 0.59)× 10−2 (1.57± 0.47)× 10−2 (5.8± 2.9)× 10−3 NOTE-ABC, Poisson, and inverse detection efficiency method (IDEM) estimated occurrence rates for Q1-Q16 KOI catalog planet candidates associated with FGK stars. Due to updates to the Kepler planet candidate catalog and stellar catalogs at the Exoplanet Archive, a direct comparison of occurrence rates to those of Christiansen et al. (2015) was not possible. Therefore, we computed occurrence rates using the IDEM (Christiansen et al. 2015) and our exact same cat- alog of target stars and planet candidates. We present the corresponding planet occurrence rates computed using the inverse detection efficiency method in Table 1. Details are provided in Appendix A. For the sake of comparison, Table 1 also presents rates and uncertainties calculated using the simplified Bayesian model described in Appendix B. We find that the three methods yield similar planet occur- rence rates and uncertainties for most of the planet radius- period bins considered. We show the ratio of the planet oc- currence rates estimated by the IDEM to those estimated by our more rigorous ABC method in Fig. 6. The values agree very well for large planet candidates over all period ranges and planet candidates over most of the ra- dius range with P < 5 days. For these bins, there is a suf- ficiently large sample of planet candidates detected at high transit signal-to-noise, that the two methods give similar re- sults, despite the noise induced by low transit signal-to-noise planet candidates. While the ratio of posterior means appears to differ for some bins corresponding to large planets at short orbital periods, these estimates are not not meaningfully in- consistent with each other, since these bins have few or no detections and both methods have large uncertainties. In con- trast, small planets at short periods have occurrence rates that are inconsistent between ABC and IDEM due to IDEM not accounting for measurement uncertainties. In summary, we find substantial differences in the esti- mated planet occurrence rates for planet sizes and orbital periods near the threshold of detectability (i.e. small plan- ets at long periods), which varies slightly depending on the KOI catalog and choice of detection efficiency model. For our baseline case, the differences are particularly notable for planets smaller than 1.5 R⊕ at orbital periods greater than 80 days. Of particular interest, we show that the IDEM can underestimate the occurrence rate of planets with sizes and orbital periods similar to that of the Earth by up to a fac- tor of ∼ 4. An example of how the IDEM underestimates the planet occurrence rate for long orbital period and small planet radius bins is shown in Fig. 1. Note that these esti- mates are based on the same data, so one would expect that the mean rates should agree to better than the width of the posterior, if both methods were providing consistent and un- biased estimates of the same quantity. The bias of the inverse detection efficiency method is to be expected, as described in PLANET OCCURRENCE RATES USING ABC 19 Figure 5. The ABC estimated occurrence rate for the Q1-Q16 planet candidates orbiting FGK stars using the Christiansen et al. (2015) Gamma CDF curve for the detection efficiency. The numerical values of the occurrence rates are stated as percentage (i.e. 10−2). The color coding of each cell is based on (d2 f )/[d(log2 Rp) d(log2 P)], which provides an occurrence rate normalized to the width of the bin and therefore is not dependent on choice of grid density. Cells colored gray have estimated upper limits for the occurrence rate. Note that the bin sizes are not constant. Appendix A. We discuss the significant implications of this result in §6. In the case of the simplified Bayesian model, while the oc- currence rates agree well with the occurrence rates estimated using the full Bayesian model and ABC throughout much of the chosen parameter space, the simplified Bayesian model underestimates the occurrence rate for planet candidates near the edge of detectability (i.e. large period, small radius). In contrast, when estimating occurrence rates in the limit of minimal transit depth uncertainty (as done in §4.2.2 to show ABC posteriors closely approximate the correct poste- rior width) there is close agreement over all parameter space between the two models. As a result, the difference between the occurrence rates can be explained by the inclusion of re- alistic transit depth uncertainty in the ABC forward model. 5.2. Correlations in Occurrence Rates due to Uncertainty in Planet Radii For the planet size-orbital period distribution model con- sidered in this paper, the planet occurrence rates in each bin are very nearly decoupled. The only way that a planet with size and period in one bin can affect the occurrence rate in an- other bin is if the measurement error results in assigning the planet to a different size-period bin. Since orbital periods are measured extremely precisely, planets will almost certainly be assigned to the correct period range. The uncertainties in planet radius are significantly larger, both due to uncertainty in the estimated transit depth and due to the uncertainty in the host star radius. Our ABC approach accounts for this effect, even in simulations that characterize the occurrence rate for a single bin in planet size and orbital period. By perform- ing inference on the occurrence rate for each bin separately, we do lose information about the joint ABC posterior, i.e., correlations between occurrence rates in different bins. To test the significance of this effect, we performed simula- tions that simultaneously estimate occurrence rates for mul- tiple bins include differing planet size ranges, but a common orbital period. To quantify how much these uncertainties have on our occurrence rate estimates, we performed ABC simulations to estimate the joint ABC posterior for the planet occurrence rates for neighboring planet size bins (up to two on each side), each time using a single common orbital pe- riod range. As expected based on results from §4.3, the final is not as small as when characterizing each occurrence rate separately. 0.51.252.5510204080160320Orbital Period (da )0.50.7511.251.51.7522.534681216Planet Radius (R⊕)0.073 ± 0.047 %0.118 ± 0.027 %0.089 ± 0.022 %0.052 ± 0.013 %0.053 ± 0.013 %0.039 ± 0.012 %0.0105 ± 0.0052 %0.0058 ± 0.0048 %0.0036 ± 0.0035 %0.021 ± 0.012 %<0.0068%0.0039 ± 0.0032 %0.0041 ± 0.0032 %0.62 ± 0.27 %0.376 ± 0.074 %0.236 ± 0.048 %0.123 ± 0.026 %0.129 ± 0.02 %0.053 ± 0.019 %0.047 ± 0.023 %0.048 ± 0.013 %0.0183 ± 0.0082 %0.0167 ± 0.008 %0.0056 ± 0.0055 %0.012 ± 0.011 %0.037 ± 0.015 %4.9 ± 1.1 %1.42 ± 0.26 %0.76 ± 0.1 %0.772 ± 0.095 %0.61 ± 0.11 %0.402 ± 0.054 %0.336 ± 0.056 %0.21 ± 0.045 %0.133 ± 0.026 %0.136 ± 0.043 %0.056 ± 0.022 %0.086 ± 0.022 %0.119 ± 0.041 %9.2 ± 1.9 %5.35 ± 0.76 %2.41 ± 0.28 %1.84 ± 0.21 %1.24 ± 0.15 %0.87 ± 0.082 %1.16 ± 0.15 %0.96 ± 0.11 %0.584 ± 0.073 %0.267 ± 0.083 %0.101 ± 0.039 %0.152 ± 0.031 %0.074 ± 0.028 %11.9 ± 5.8 %6.3 ± 1.9 %5.04 ± 0.95 %3.63 ± 0.76 %2.16 ± 0.31 %1.93 ± 0.17 %2.59 ± 0.24 %2.31 ± 0.22 %1.42 ± 0.13 %0.48 ± 0.12 %0.181 ± 0.06 %0.225 ± 0.058 %0.156 ± 0.051 %<25%16 ± 3.9 %5.3 ± 1.1 %3.93 ± 0.5 %2.99 ± 0.41 %1.62 ± 0.26 %4.34 ± 0.4 %2.94 ± 0.26 %2.07 ± 0.27 %0.91 ± 0.17 %0.22 ± 0.09 %0.36 ± 0.11 %0.188 ± 0.068 %<48%<10%7.5 ± 2.1 %5.34 ± 0.98 %3.8 ± 1.1 %3.06 ± 1 %4.8 ± 0.66 %4.05 ± 0.49 %2.57 ± 0.31 %1.02 ± 0.25 %0.74 ± 0.24 %0.42 ± 0.19 %0.2 ± 0.12 %<141%<18%9.4 ± 4.6 %8 ± 3.2 %6 ± 1.2 %3.6 ± 1.1 %6.2 ± 1 %4.24 ± 0.68 %4.26 ± 0.75 %1.47 ± 0.49 %0.85 ± 0.24 %1.51 ± 0.36 %0.18 ± 0.16 %<138%<69%25 ± 19 %18.1 ± 7.6 %6.7 ± 1.9 %6 ± 3.3 %12.9 ± 2 %5.1 ± 1.3 %7.4 ± 1 %1.85 ± 0.58 %2.76 ± 0.83 %1.68 ± 0.52 %0.69 ± 0.34 %10−310−210−1100d2fd(log2Rp)d(log2P)20 HSU ET AL. Figure 6. The ratio of the planet occurrence rate inferred by the inverse detection efficiency method to the planet occurrence rate inferred the ABC method based on the Q1-Q16 candidates orbiting FGK stars. The color indicates the relative difference between the occurrence rates scaled by the uncertainty given by the standard deviation from the ABC method. The two methods give similar results for easily detected planets (most bins with high signal-to-noise are within ∼1 standard deviations), but the inverse detection efficiency method substantially underestimates the frequency of near-Earth-size planets for orbital periods beyond ∼ 80 days. This results in a modest increase in the width of the marginal ABC posterior for each occurrence rate. Next, we investi- gated the correlation between occurrence rate in neighboring planet size bins. We did not fine large correlations for any pair of neighboring bins, with a maximum absolute correla- tion coefficient of 0.4. 6. DISCUSSION NASA's Kepler mission was designed to measure the oc- currence rate of exoplanets, particularly small planets in the habitable zone of solar-like host stars. Kepler was enor- mously successful, finding thousands of strong planet can- didates with sizes less than Neptune. However, since the probability of a planet both transiting and being detectable by Kepler falls off rapidly as one approaches Earth-size plan- ets in the habitable zone of sun-like stars, Kepler identified precious few potential Earth-analogues. The small number of detections and rapidly changing detection efficiency for small planets necessitates careful attention be paid to sta- tistical methodology when interpreting the Kepler results in terms of intrinsic planet populations. Many early investiga- tions of the planet occurrence rates based on Kepler data re- lied on the simplistic IDEM which is known to be signifi- cantly biased for some of the most interesting regions of pa- rameter space. Some more recent studies have attempted to improve on this by using a Bayesian framework for estimat- ing planet occurrence rates (e.g., Burke et al. (2015)) and in- corporating hierarchical models (e.g., Foreman-Mackey et al. (2014)), providing significant advances in statistical rigor and power. We build on these works by introducing Approxi- mate Bayesian Computing to perform inference with hierar- chical Bayesian models for the planet occurrence rates. We emphasize that these calculations are done with several sim- plifying assumptions, particularly in terms of the input cat- alog used (see §6.2). Therefore, we regard these as prelimi- nary estimates which improve on many previous studies, but still leave significant room for further improvement. Sub- stantial further research is needed to improve the accuracy and precision of occurrence rates before making important decisions based on planet occurrence rates (e.g., design of a large space mission). 6.1. Summary of Key Results We have demonstrated that HBM and the ABC-PMC al- gorithm can accurately characterize planet occurrence rates. We confirm that the inverse detection efficiency method can provide accurate estimates of the planet occurrence rate for planets with sufficiently high transit signal-to-noise and short 0.51.252.5510204080160320Orbital Period (day)0.50.7511.251.51.7522.534681216Pla et Radius (R()0.911.221.261.161.021.150.770.640.750.38N/A0.730.722.011.301.051.061.081.150.901.041.110.770.610.690.811.151.051.010.981.031.100.991.130.941.010.971.040.941.250.910.870.950.941.001.001.010.980.990.950.910.910.691.090.750.890.870.871.020.960.980.950.811.120.94N/A0.890.730.910.860.920.880.910.920.860.960.970.96N/AN/A0.840.700.810.810.820.760.810.890.850.950.93N/AN/A0.470.680.630.700.730.830.760.890.910.960.90N/AN/A0.250.960.690.580.670.620.760.870.870.930.84−1.5−1.0−0.50.00.51.01.5(fIDEM−fABC)/σABCPLANET OCCURRENCE RATES USING ABC 21 orbital periods. We also show that the inverse detection effi- ciency method is significantly biased when applied to planets near the threshold of detection, leading to it underestimating the planet occurrence rate by a factor of up to 4 for nearly Earth-size planets with orbital periods greater than 80 days when applied to the Kepler Q1-Q16 catalog. We applied our HBM and ABC methodology to the Kepler Q1-16 planet candidate catalog. We characterized the rate of planet candidates over a larger range of planet sizes and with finer detail than most previous studies. We find a relatively smooth planet occurrence rate in planet size and orbital pe- riod over most of the range explored, even without assuming a smooth functional form. However, we find a "radius valley" (i.e., a local minimum in the marginalized planet occurrence rate per logarithmic interval in planet size), particularly for orbital periods between 20 and 80 days. This finding is con- sistent with other studies which found a radius valley based on smaller samples for which more precisely measured stel- lar properties were available (e.g., Fulton et al. 2017; Van Eylen et al. 2017). Interestingly, we find that this feature can be identified even without the improved spectroscopic or as- teroseismic constraints, albeit not as strongly as if constraints were incorporated. The fact that three studies each find a sig- nificant radius valley, despite substantial differences in the target sample used, planet candidate catalog, and the origin of the stellar properties assumed, points to the robustness of this feature in the Kepler data. We find that a joint power law in planet size and orbital pe- riod is not an adequate model when considering planets rang- ing from Earth-size to ∼ 2R⊕ or larger and periods 20-320 days. Given the observed structure in the planet occurrence rate, even a broken power-law is inadequate when consider- ing planets ranging from Earth-size to ∼ 3R⊕ or larger. We also provide a simplified Bayesian model for planet oc- currence rates that avoids the bias inherent in the IDEM. We recommend that future studies that want a fast method of es- timating occurrence rates in the high signal-to-noise regime adopt this simplified Bayesian model, rather than continu- ing to use the IDEM. Of course, a full hierarchical Bayesian model would be even better, since the simplified Bayesian model neglects measurement uncertainties. 6.2. Limitations of Our Occurrence Rates While these results represent a significant methodological advance, several astrophysical assumptions affect this and other published occurrence rate studies. For example, un- certainties in the radius of host stars directly affects the un- certainty in planet radii. While the ABC method accounts for uncertainty in the stellar properties, we have assumed errors follow a Gaussian distribution with the reported uncertain- ties. In reality there are likely non-Gaussian measurement uncertainties and potentially systematic biases in the mea- Figure 7. ABC estimated occurrence rates from Fig. 5 integrated over three different ranges of period bins to emphasize the existence of a local minimum near 1.75 − 2R⊕. sured stellar properties. While considerable effort has gone into creating and updating the stellar properties catalog, we expect that the stellar properties could be significantly im- proved once parallaxes are available from GAIA. As another example, we have assumed that all KOIs not identified as a likely false positive are strong planet candi- dates. In practice, some are likely astrophysical false posi- tives, such as a brown dwarf or M star that transits one star in a hierarchical triple system. Other planet candidates may be false alarms that survived the vetting process. Similarly, we have neglected the effects of the transit depths being diluted due to more than one star falling into the aperture used for constructing Kepler light curves. We adopted a single fixed distribution for the orbital eccentricity, rather than simulta- neously inferring the eccentricity distribution or allowing for a dependence on planet properties. Based on previous sensi- tivity studies, we expect that this has a modest effect on our results (Burke et al. 2015), smaller than the uncertainties due to the choice of Kepler planet candidate catalog, host stars properties and potential inaccuracy of the model for transit detection and vetting efficiency and reliability. In principle, our ABC framework makes it practical to incorporate these and additional effects into a HBM analysis of exoplanet pop- ulations. However, incorporating such effects is beyond the scope of this study. 6.3. Implications for η⊕ Our results have significant implications for estimates of η⊕, the rate of nearly Earth-size planets in the habitable zone of solar-type stars. Many definitions of η⊕ can be found in the literature. Following the convention of the NASA Exoplanet Program Analysis Group, Study Analysis Group 13's report on Exoplanet Occurrence Rates and Distribu- 124816Planet Radius (R⊕)0.000.050.100.150.200.250.300.350.40Occurrence Rate (f)P=0.5−80dP=20−80dP=20−160d22 HSU ET AL. tions, we compute the occurrence rate for planets with radii 1 − 1.5R⊕ and orbital periods 237-320 days (Belikov et al. 2017). Applying our HBM and ABC method to the Kepler Q1-Q16 catalog, we find a rate of 0.41+0.29 −0.12 per star, larger than the 0.11± 0.08 that would be estimated using the same data set and assumptions with the inverse detection efficiency method. This underscores the importance of quantifying the accuracy and biases of various statistical methodologies for computing planet occurrence rates. As mentioned in §6.2, there are multiple complications that likely affect the inferred rate of small planets and are not re- flected in the uncertainty estimates above. In particular, the above estimates are for the rate of planet candidates, as re- ported in the Q1-Q16 catalog. Any astrophysical false pos- itives or statistical false alarms that managed to survive the vetting process would contribute to increasing our inferred rate of planet candidates. For most of the planet sizes and orbital periods considered, the Kepler pipeline has a low rate of false alarms and good sensitivity to reject astrophysical false positives. However, based on the DR25 completeness products the rate of contamination from false positives and false alarms is greater in the range of periods and sizes af- fecting estimates of η⊕, since the vetting process is more sensitive for planet candidates with larger transit signal-to- noise and shorter orbital periods. Subsequent to the calcu- lations reported in this paper, the Kepler team released ad- ditional data products that could be used to characterize the performance of the Kepler pipeline (Thompson et al. 2017). We recommend that future research to consider how to in- corporate these data products into occurrence rate studies in a statistically valid way. Of course, the largest uncertainty in the rate of small planets in the habitable zone is the size of the habitable zone. Therefore, we also report a differential occurrence rate per star, (d2 f )/[d(log2 P) d(log2 Rp)] = 1.6+1.2 −0.5 for Rp = 1 − 1.5R⊕,P = 237 − 320d, so other researchers can easily scale our result to their preferred ranges of planet sizes and orbital periods. If one were to extrapolate to a wider range of orbital periods and assume that the rate density remains con- stant, then one would arrive at an occurrence rate of 1.0+0.7 −0.3 per star for Rp = 1 − 1.5R⊕,P = 237 − 500d which covers an optimistic range for the habitable zone. Of course, one should recognize the limitations of such an extrapolation, as physical processes could cause the planet occurrence rate to vary significantly over this range. 6.4. Future Prospects for Occurrence Rate Studies Future studies will further improve on planet occurrence rates based on the Kepler data set. While there is clearly room for improvements in the population modeling (e.g., §6.2), we anticipate that these will have relatively minor ef- fects compared to the anticipated improvements in the cat- alogs of planet candidates and star properties. In particu- lar, we are currently incorporating results from the improved pipeline detection efficiency released for the DR25 KOI cat- alog as provided by the Kepler Science Team. The DR25 KOI catalog is based on the complete data set, a final Ke- pler pipeline and an automated vetting process (Thompson et al. 2017). We have shown that the inferred planet occur- rence rates are sensitive to the assumed detection efficiency. Therefore, we refrain from presenting results based on the DR25 catalog until the pipeline detection efficiency is well- characterized and implemented in our model. We also antic- ipate significant improvements due to improved characteri- zation of host stars once GAIA parallaxes become available. We hope that this study will make it practical for future planet occurrence rate studies that incorporating these improved data sets to apply a rigorous hierarchical Bayesian model. In particular, we provide open source codes (at GitHub, see §2.5 to facilitate the use of ABC in future studies, including estimating planet occurrence rates, as well as any number of problems that involve population inference. 6.5. Future Prospects for Characterizing Population of Exoplanet Architectures While numerous studies have attempted to characterize the rate of individual planets, the frequency of various planetary system architectures is an even more powerful constraint on planet formation theories. Perhaps the most important con- tribution of this study is to provide a practical framework for performing statistical inference on the population of plane- tary systems. While the geometric transit probability for a single planet is relatively simple (P ∼ sin−1(R(cid:63)/a)) for circular orbits with a > R(cid:63), the transit probability for a system with multiple planets is much more complicated. Additional selection ef- fects (e.g., the detection efficiency and window function) further complicate comparing the predictions of theoretical models to the Kepler results. Since these considerations are extremely difficult to incorporate in occurrence rate analy- ses of exoplanet catalogs, previous studies have neglected the effects of multiple planet systems. While previous stud- ies could only focus on overall occurrence rates based on planetary property distributions, SysSim and ABC provide a means for rigorously addressing questions like: "How com- mon are systems with at least one pair of planets with orbital periods within 2% of a 2:1 orbital resonance?" This ability to characterize the intrinsic rate of various planetary architec- tures will enable a wide range of theoretical studies to better understanding of planet formation. To address such questions, one must model the joint dis- tribution for planetary system properties, e.g., allowing for potential correlations in the orbital period and sizes of plan- ets in the same planetary system. With SysSim and the ABC PLANET OCCURRENCE RATES USING ABC 23 approach, one has access to the true physical parameters for each planetary system in the simulated catalog. Thus, one can compute the transit and detection probabilities for any combination of the planets in a system. In the most straightforward approach, one can construct simulated cat- alogs based on which planets would transit for a single spec- ified viewing geometry and their detection efficiencies. Sys- Sim can also includes advanced functionality to increase the computational efficiency of ABC when characterizing plan- etary architectures. For many summary statistics related to multiple planet systems (e.g., number of detected planets, ra- tios of their orbital periods), SysSim can include weights, analogous to how this study weighted each potentially de- tectable planet by a detection probability that was averaged over all viewing geometries. The required sky-averaged probabilities can be calculated efficiently using nearly alge- braic formulae that integrate over all viewing directions for any subset of the planets in each system using CORBITS (Brakensiek & Ragozzine 2016). Realizing this potential will require further methodologi- cal research. For example, when using multiple qualitatively different summary statistics, it will be important to explore the impact of various choices for distance functions. An ade- quate model for the population of exoplanetary systems may require several parameters, motivating further research in im- proving the efficiency of the sequential importance sampler used by ABC-PMC. 6.6. Future Prospects for Characterizing Exoplanet Populations Including Follow-Up Observations In §2.1, we highlighted the difficulties of characterizing a population of planetary systems when most of the plan- ets are undetected. Another application of ABC is in cases where practical considerations make it unrealistic to even write down a likelihood function. For example, in princi- ple, a study characterizing the planet occurrence rate around a given type of star should account for the target selection criteria. That would require a likelihood that integrates over all possible target stars, dramatically increasing the compu- tational complexity. The choice of target stars for the Kepler planet search depended on a complex merit function that was based on the estimates of stellar properties prior to launch which are now out-of-date. Further complicating matters, whether a star was targeted depended not only on its own properties, but also the properties of other stars that were be- ing considered for targeting. Without conditional indepen- dence, the integrals over properties of each potential target can not be separated. As another example, if one wanted to include constraints from ground-based follow-up observations, one should ac- count for the process of choosing targets worthy of follow-up observations, the whims of time allocation committees se- lecting successful proposals, the limitations of weather, how observers change their plans based on preliminary data anal- ysis, etc. While writing down a likelihood that accounts for such effects is not realistic, it can be feasible to implement a forward model that reasonably simulates such effects. Our ABC approach has the potential to enable future studies that account for target selection and even follow-up observations. We anticipate that this could be particularly important for future analyses of planet populations identified by NASA's Transiting Exoplanet Survey Satellite (TESS) before being targeted by ground-based Doppler follow-up. We thank the entire Kepler team for the many years of work that has proven so successful and was critical to this study. D.C.H, E.B.F., R.C.M and D.R. acknowledge support from NASA Origins of Solar Systems grant # NNX14AI76G and helpful discussions with Keir Ashby. E.B.F and R.C.M. acknowledge support from NASA Kepler Participating Sci- entist Program Cycle II grant # NNX14AN76G, D.C.H, E.B.F. acknowledge support from the Penn State Eberly Col- lege of Science and Department of Astronomy & Astro- physics, the Center for Exoplanets and Habitable Worlds and the Center for Astrostatistics. The citations in this paper have made use of NASA's Astrophysics Data System Bibli- ographic Services. This research has made use of the NASA Exoplanet Archive, which is operated by the California Insti- tute of Technology, under contract with the National Aero- nautics and Space Administration under the Exoplanet Ex- ploration Program. We acknowledge the Institute for Cy- berScience (http://ics.psu.edu/) at The Pennsylva- nia State University, including the CyberLAMP cluster sup- ported by NSF grant MRI-1626251, for providing advanced computing resources and services that have contributed to the research results reported in this paper. This material was based upon work partially supported by the National Sci- ence Foundation under Grant DMS-1127914 to the Statistical and Applied Mathematical Sciences Institute (SAMSI). Any opinions, findings, and conclusions or recommendations ex- pressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Founda- tion. This study benefited from the 2013 SAMSI workshop on Modern Statistical and Computational Methods for Anal- ysis of Kepler Data, the 2016/2017 Program on Statistical, Mathematical and Computational Methods for Astronomy, and their associated working groups. 24 HSU ET AL. APPENDIX A. INVERSE DETECTION EFFICIENCY METHOD The inverse detection efficiency method (IDEM) attempts to calculate weights from observed planet candidates in order to "correct" for the planets which weren't observed by Kepler. Following closely to the methodology of Christiansen et al. (2015), the IDEM used in this study is as follows: 1. Calculate the geometric transit probability of each planet candidate i about its host star: pg,i = R(cid:63)/a 2. Calculate the probability each planet candidate i has of being detected if it transits around each target star j in the stellar catalog (given the planet has the same orbital period but a uniformly chosen random impact parameter between 0 and 1) using the chosen detection efficiency curve: pd,i, j. 3. Sum pd,i, j over all target stars and divide by the total number of target stars to get the fraction of stars around which a given (cid:16)(cid:80)Ntarg (cid:17) planet candidate i would have been detected: pd,i/Ntarg = j=1 pd,i, j /Ntarg. 4. Calculate a weight Ci = 1/(pg,i pd,i) that is often interpreted as an estimate of the number of planets of the same size and period as planet i in the Kepler survey, attempting to account for incompleteness due to geometric transit probability and detection efficiency. 5. Summing the weights Ci for planets with periods and radii within the limits of a bin and dividing by the total number of target stars gives an estimate of the occurrence rate within that bin: fr,p = i in bin r,pCi /Ntarg (cid:16)(cid:80) (cid:17) For the associated uncertainty, we report the inverse of the Fischer information (curvature of the log likelihood) for the rate parameter under a Poisson distribution, where the rate parameter is the estimated planet occurrence rate and the number of planet detections within a bin Nr,p from the Kepler data. (A1) fr,p(cid:112)Nr,p σ fr,p = A significant shortcoming of this technique for estimating occurrence rates lies with the fact that it does not properly marginalize over the uncertainty in the transit parameters. Inevitably, measurement errors cause the estimated transit properties (e.g., transit depth, duration) to deviate from the true transit parameters. If the estimated transit SNR is smaller (larger) than the true transit SNR, then the planet is less (more) likely to be detected, i.e., measurement noise reduces (increases) the measured signal. For a planet with transit SNR near the threshold of detection, the detection probability changes rapidly as a function of transit SNR, so there can be a large difference between the detection probability for the true transit SNR and the estimated transit SNR. This difference impacts the occurrence rates estimated by IDEM in two ways. First, it affects which planets are included in the catalog of planet detections. The smallest detected planets are likely to have their sizes over-estimated than under-estimated, since this dramatically increases their detection probability. Second, it affects the values of pd,i and hence Ci which are assigned to the planets. For the smallest detected planets, the estimates for the planet size, transit SNR and detection probability are all biased to be larger than the true values, causing IDEM to systematically overestimate pd,i and thus systematically underestimate Ci, the occurrence rate, fr,p, and the uncertainty in the occurrence rate, σ fr,p. Because the detection probability changes strongly and non-linearly as a function of planet size, this decrease in the estimated occurrence rate from smaller planets is not fully compensated for by an increase in the estimated occurrence rate larger from larger planets (whose true SNR may be overestimated). For larger planets, the transit SNR is larger than the threshold of detection, and the the detection probability is changing much more slowly, so the magnitude of these effects are dramatically reduced. We demonstrate the resulting bias in planet occurrence rates by analyzing simulated data sets with the IDEM in Figure 1. This effect is strongest for occurrence rate estimates that include planets near the threshold of detection, such as nearly Earth-size planets near the habitable zone (see Fig. 6). It is important to note that this bias can often be partially corrected with a few modifications as we have done with a simplified Bayesian model in Appendix B. See also the appendix of Foreman-Mackey et al. (2014). B. SIMPLIFIED BAYESIAN MODEL FOR PLANET OCCURRENCE RATES Here we present a simplified Bayesian model for estimating planet occurrence rates. We assume that the log period and log radius of planets are drawn from a Poisson point process with constant rate ((d2 f )/[d(log2 P) d(log2 Rp)]) over each bin defined by a range of orbital periods and planet sizes. Then the number of planets around each star within the bin is distributed as a Poisson random variable with a rate parameter ( fr,p = (d2 f )/[d(log2 P) d(log2 Rp)](cid:12)(cid:12)r,p ∆r∆p) proportional to the area of the bin PLANET OCCURRENCE RATES USING ABC 25 (∆r∆p). If we could observe Ntarg targets and detect all planets, then the sum of the number of planets in the bin (Nall,r,p) over all target stars would also a Poisson random variable with rate parameter fr,pNtarg. If we let the prior for the rate parameter ( fr,p) be a Gamma distribution with shape parameter α0 and rate parameter (also known as the inverse scale parameter) β0, then the posterior distribution for fr,p will be a Gamma distribution with shape parameter α0 + Nall,r,p and rate parameter β0 + Ntarg. In practice, we do not detect all planets due to a combination of geometric transit probability and detection efficiency. Instead, we are only sensitive to planets around a subset of the the target stars, primarily dictated by the orientation of the orbital plane and the photometric precision for each star. If we assume that, for each target star (indexed by j) and bin in planet size and orbital period, the combined probability of detecting its planets (pr,p, j) is nearly constant, then the effective number of stars searched (NESS,r,p, where r and p indicate the range of planet sizes and orbital periods) is the sum of Ntarg Bernoulli random variables with success probabilities pr,p, j. In this model, the posterior distribution for fr,p conditioned on detecting Nr,p planets will again be a Gamma distribution, p( fr,pNr,p,NESS,r,p) ∼ Gamma(α0 + Nr,p, β0 + NESS,r,p). (B2) Since NESS,r,p is itself a random variable, one would like to marginalize over NESS,r,p, but the resulting posterior for fr,p is not analytic and is computationally expensive. We can approximate the posterior for fr,p by substituting the expected value, E(cid:2)NESS,r,p the average value of pr,p, j (cid:39) (1/Nsamp)(cid:80)Nsamp stars searched is simply the sum of these probabilities over all target stars, E(cid:2)NESS,r,p Eqn. B2. We estimate the expected value of the effective number of stars searched via Monte Carlo. For each star and range of planet sizes and orbital periods, we draw Nsamp = 100 planets (uniformly in logP and logRp within the bin limits) and estimate i=1 pgeo,i, j pdet,i, j via Monte Carlo, where pgeo,i, j and pdet,i, j are the geometric transit probability and detection probability for the ith draw of planet parameters. The expected value for the effective number of j=1 pr,p, j. Then we approximate the (cid:3) for the random variable NESS,r,p in posterior for the intrinsic occurrence rate of planets in each range of planet sizes and orbital periods by p( fr,pNr,p,E(cid:2)NESS,r,p To help build intuition for this result is useful to consider the mean and standard deviation of the gamma distribution posterior, (cid:3) =(cid:80)Ntarg (cid:3)) ∼ Gamma(α0 + Nr,p, β0 + E(cid:2)NESS,r,p (cid:3)). β0 + E(cid:2)NESS,r,p (cid:3) (cid:113) (cid:3) . β0 + E(cid:2)NESS,r,p α0 + Nr,p µ fr,p µ fr,p = σ fr,p = (B3) (B4) (B5) If α0 and β0 are much less than Nr,p and E(cid:2)NESS,r,p (cid:3), then the posterior mean for the occurrence rate is essentially the ratio of the number of planets detected to the effective sample size. Similarly, the width of the posterior behaves as expected for a Monte Carlo estimate of a mean, in the limit of a large effective sample size. The primary difference from IDEM is that our simplified model provides a principled definition of the effective sample size that differs from that assumed arbitrarily by IDEM. In Table 1 we report occurrence rates estimated from this simplified Bayesian model, using the gamma distribution of Eqn. B3 in App. B and α0 = β0 = 1, which corresponds to a prior p( fr,p) = exp(− fr,p). For radius-period bins where either of these quantities is zero or of order unity, then the observations provide limited information, so the resulting posterior for fr,p will be sensitive to the choice of prior. Fortunately, for most bins, the results are insensitive to the choice of prior, since both Nr,p and E(cid:2)NESS,r,p (cid:3) are much larger than unity. The occurrence rates computed using the simplified Bayesian model agree well with the occurrence rates estimated using the full Bayesian model and ABC throughout most of the parameter space we consider. This represents a significant improvement upon the IDEM for which we find significant differences occur across the full range of periods for planets smaller than Neptune. Therefore, we recommend that future studies that want a fast method of estimating occurrence rates in the high signal-to-noise regime of parameter space adopt this simplified Bayesian model, rather than continuing to use the IDEM. Upon more detailed inspection, one can observe that at long orbital periods, the simplified Bayesian model appears to slightly underestimate the planet occurrence rate of small planets relative to ABC. The source of the difference can be traced to how the two models decides which occurrence rate bin a planet contributes to. The simplified Bayesian model always assigns the planet to a period-radius bin based on the planet candidate's observed parameters as reported in the planet candidate catalog. In contrast, the ABC method allows for a planet to be assigned to a given been to be due to a planet whose true parameters would place it in 26 HSU ET AL. a different bin. This strict placement of planets by the simplified Bayesian model results in an overestimated E(cid:2)NESS,r,p overestimate of E(cid:2)NESS,r,p (cid:3) when (cid:3) leads to an underestimate of the occurrence rate for those bins. This demonstrates the importance the transit measurement noise has a significant chance of causing a planet to be assigned to a different period-radius bin. The of applying a hierarchical Bayesian model, regardless of whether computed via ABC or more traditional methods, to accurately infer the occurrence rate of small, long-period planets. C. PROBABILISTIC CATALOGS In the present study, we generate a single observed catalog, where each planet is either detected or not detected. SysSim also has the capability to generate a probabilistic catalog, where each potentially detectable planet is included along with a weight proportional to its detection probability. This approach could be useful for accelerating calculations or for comparing the frequency of planetary systems that are intrinsically rare or rarely detected. In the probabilistic catalog approach, we average over all possible observer orientations by using the sky-averaged geometric transit probability for each planet, R(cid:63)/[a(1 − e2)]. As in the previous case, the geometric transit probability is multiplied by an updated transit detection probability. In this approach, we draw a single value of the impact parameter from a uniform distribution between zero and unity for the purposes of computing the transit detection probability. Each potentially detectable planet (indexed by l) is added to the observed planet catalog, along with its measured transit parameters and a weight (wl = pcomb,l) corresponding to the product of its sky-averaged geometric transit probability, the window function for it transiting at least three transits, and the detection efficiency conditioned on the planet transiting at least three times. The expected number of detected planets within each planet size-period divided by the number of target stars is given by E[nr,p]/Ntarg (cid:39) (cid:88) wl/Ntarg (C6) l in bin r,p Comparing this approach to the single viewing geometry approach, using a single viewing geometry has the advantage of simplic- ity and being readily extensible to incorporating more complex summary statistics (e.g., abundance of multiple detected transiting planets). Averaging over viewing geometries has the advantage of enabling a much more accurate estimate of the expected num- ber of detected planets for a given Ntarg which can then be chosen to be significantly less than the number of stars targeted by Kepler. However, one should not simply replace the number of planets in a period-radius bin with the expected value for the number of planets within that bin, as the process of averaging over the sky results in the expected value having less variance than the actual number of planets in the same bin for any given observer. Further, the sky-averaging feature could complicate the interpretation of correlations of planets within one planetary system. Therefore, considerable care is necessary before applying SysSim's sky averaging inside an ABC calculation. A mathematical derivation necessary to motivate the choice of summary statistics and distance function is beyond the scope of this paper. This appendix merely documents this feature of SysSim. REFERENCES Akeret, J., Refregier, A., Amara, A., Seehars, S., & Hasner, C. 2015, JCAP, 8, 043 Burke, C. J., Bryson, S. T., Mullally, F., et al. 2014, ApJS, 210, 19 Burke, C. J., Christiansen, J. L., Mullally, F., et al. 2015, ApJ, 809, Batalha, N. M. 2014, Proceedings of the National Academy of 8 Science, 111, 12647 Batalha, N. M., Rowe, J. F., Bryson, S. T., et al. 2013, ApJS, 204, Cameron, E., & Pettitt, A. N. 2012, MNRAS, 425, 44 Christiansen, J. L., Clarke, B. D., Burke, C. J., et al. 2015, ApJ, 24 810, 95 Beaumont, M. A., Cornuet, J.-M., Marin, J.-M., & Robert, C. P. Coughlin, J. L., Mullally, F., Thompson, S. E., et al. 2016, ApJS, 2008, ArXiv e-prints, arXiv:0805.2256 Belikov, R., Stark, C., Batalha, N., & Burke, C. 2017 Bezanson, J., Edelman, A., Karpinski, S., & Shah, V. B. 2014, ArXiv e-prints, arXiv:1411.1607 Blum, M. 2009, ArXiv e-prints, arXiv:0904.0635 Borucki, W. J. 2016, Reports on Progress in Physics, 79, 036901 Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977 Borucki, W. J., Koch, D. G., Basri, G., et al. 2011, ApJ, 736, 19 Brakensiek, J., & Ragozzine, D. 2016, ApJ, 821, 47 224, 12 Dressing, C. D., & Charbonneau, D. 2013, ApJ, 767, 95 -. 2015, ApJ, 807, 45 Fabrycky, D. C., Lissauer, J. J., Ragozzine, D., et al. 2014, ApJ, 790, 146 Fearnhead, P., & Prangle, D. 2012, Journal of the Royal Statistical Society: Series B (Statistical Methodology), 74, 419. http:// dx.doi.org/10.1111/j.1467-9868.2011.01010.x Ford, E. B. 2005, AJ, 129, 1706 Ford, E. B., He, M., Hsu, D. C., & Ragozzine, D. 2018a, Mullally, F., Coughlin, J. L., Thompson, S. E., et al. 2015, ApJS, PLANET OCCURRENCE RATES USING ABC 27 Approximate Bayesian Computing with Julia, v1.0, Zenodo, doi:10.5281/zenodo.1198716. https://doi.org/10.5281/zenodo.1198716 -. 2018b, Planetary Systems Simulation & Model of Kepler Mission for characterizing the Occurrence Rates of Exoplanets and Planetary Architectures, v1.0, Zenodo, doi:10.5281/zenodo.1205172. https://doi.org/10.5281/zenodo.1205172 Foreman-Mackey, D., Hogg, D. W., & Morton, T. D. 2014, ApJ, 795, 64 Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81 Fulton, B. J., Petigura, E. A., Howard, A. W., et al. 2017, AJ, 154, 109 Howard, A. W., Marcy, G. W., Bryson, S. T., et al. 2012, ApJS, 201, 15 Huber, D., Silva Aguirre, V., Matthews, J. M., et al. 2014, ApJS, 211, 2 Ishida, E. E. O., Vitenti, S. D. P., Penna-Lima, M., et al. 2015, Astronomy and Computing, 13, 1 Jennings, E., Wolf, R., & Sako, M. 2016, ArXiv e-prints, arXiv:1611.03087 Kopparapu, R. K. 2013, ApJL, 767, L8 Mann, A. W., Gaidos, E., Lépine, S., & Hilton, E. J. 2012, ApJ, 753, 90 Marin, J.-M., Pudlo, P., Robert, C., & Ryder, R. 2012, Statistics and Computing, 22, 1167. https://doi.org/10.1007/s11222-011-9288-2 217, 31 Nelson, B., Ford, E. B., & Payne, M. J. 2014, ApJS, 210, 11 Petigura, E. A., Marcy, G. W., & Howard, A. W. 2013, ApJ, 770, 69 Price, E. M., & Rogers, L. A. 2014, ApJ, 794, 92 Ratmann, O., Camacho, A., Meijer, A., & Donker, G. 2013, ArXiv e-prints, arXiv:1305.4283 Robin, A. C., Reylé, C., Fliri, J., et al. 2014, A&A, 569, A13 Rowe, J. F., Bryson, S. T., Marcy, G. W., et al. 2014, ApJ, 784, 45 Rubin, D. B. 1984, Ann. Statist., 12, 1151. https://doi.org/10.1214/aos/1176346785 Silburt, A., Gaidos, E., & Wu, Y. 2015, ApJ, 799, 180 Tavare, S., Balding, D. J., Griffiths, R. C., & Donnelly, P. 1997, Genetics, 145, 505. https://www.ncbi.nlm.nih.gov/ pmc/articles/PMC1207814/ Tenenbaum, P., Jenkins, J. M., Seader, S., et al. 2014, ApJS, 211, 6 Thompson, S. E., Coughlin, J. L., Hoffman, K., et al. 2017, ArXiv e-prints, arXiv:1710.06758 Van Eylen, V., Agentoft, C., Lundkvist, M. S., et al. 2017, ArXiv e-prints, arXiv:1710.05398 Wolszczan, A., & Frail, D. A. 1992, Nature, 355, 145 Youdin, A. N. 2011, ApJ, 742, 38
1604.02310	1	1604	2016-04-08T11:24:37	Detection of H2O and evidence for TiO/VO in an ultra hot exoplanet atmosphere	[ "astro-ph.EP" ]	We present a primary transit observation for the ultra hot (Teq~2400K) gas giant expolanet WASP-121b, made using the Hubble Space Telescope Wide Field Camera 3 in spectroscopic mode across the 1.12-1.64 micron wavelength range. The 1.4 micron water absorption band is detected at high confidence (5.4 sigma) in the planetary atmosphere. We also reanalyze ground-based photometric lightcurves taken in the B, r', and z' filters. Significantly deeper transits are measured in these optical bandpasses relative to the near-infrared wavelengths. We conclude that scattering by high-altitude haze alone is unlikely to account for this difference, and instead interpret it as evidence for titanium oxide and vanadium oxide absorption. Enhanced opacity is also inferred across the 1.12-1.3 micron wavelength range, possibly due to iron hydride absorption. If confirmed, WASP-121b will be the first exoplanet with titanium oxide, vanadium oxide, and iron hydride detected in transmission. The latter are important species in M/L dwarfs, and their presence is likely to have a significant effect on the overall physics and chemistry of the atmosphere, including the production of a strong thermal inversion.	astro-ph.EP	astro-ph	Draft version April 11, 2016 Preprint typeset using LATEX style emulateapj v. 5/2/11 DETECTION OF H2O AND EVIDENCE FOR TIO/VO IN AN ULTRA HOT EXOPLANET ATMOSPHERE Thomas M. Evans1,†, David K. Sing1, H. R. Wakeford2, N. Nikolov1, G. E. Ballester3, B. Drummond1, T. Kataria1, N. P. Gibson4, D. S. Amundsen5,6, J. Spake1 Draft version April 11, 2016 ABSTRACT We present a primary transit observation for the ultra hot (Teq ∼ 2400 K) gas giant expolanet WASP- 121b, made using the Hubble Space Telescope Wide Field Camera 3 in spectroscopic mode across the 1.12–1.64µm wavelength range. The 1.4µm water absorption band is detected at high confidence (5.4σ) in the planetary atmosphere. We also reanalyze ground-based photometric lightcurves taken in the B, r(cid:48), and z(cid:48) filters. Significantly deeper transits are measured in these optical bandpasses relative to the near-infrared wavelengths. We conclude that scattering by high-altitude haze alone is unlikely to account for this difference, and instead interpret it as evidence for titanium oxide and vanadium oxide absorption. Enhanced opacity is also inferred across the 1.12–1.3µm wavelength range, possibly due to iron hydride absorption. If confirmed, WASP-121b will be the first exoplanet with titanium oxide, vanadium oxide, and iron hydride detected in transmission. The latter are important species in M/L dwarfs, and their presence is likely to have a significant effect on the overall physics and chemistry of the atmosphere, including the production of a strong thermal inversion. Subject headings: planets and satellites: atmospheres - stars: individual (WASP-121) - techniques: photometric - techniques: spectroscopic 1. INTRODUCTION Observations have revealed a diversity of atmospheres across the population of transiting gas giant exoplanets (Sing et al. 2016; Stevenson 2016). Transmission spectra at optical wavelengths show evidence for Rayleigh scat- tering by molecular hydrogen and high-altitude aerosols, and absorption by alkali metals (e.g. Charbonneau et al. 2002; Pont et al. 2008; Sing et al. 2013, 2015; Nikolov et al. 2014, 2015). In the near-infrared, water absorp- tion has now been robustly measured for a number of planets, both in transmission and emission (e.g. Deming et al. 2013; Wakeford et al. 2013; Stevenson et al. 2014; McCullough et al. 2014; Kreidberg et al. 2015). At temperatures above 2000 K, models predict that gaseous titanium oxide (TiO) and vanadium oxide (VO) are important absorbers, especially at optical wave- lengths (e.g. Hubeny et al. 2003; Sharp & Burrows 2007; Fortney et al. 2008). Indeed, TiO/VO absorption is prominent in late M and early L dwarf atmospheres at these temperatures (Burrows & Sharp 1999; Kirkpatrick et al. 1999; Burrows et al. 2001). If TiO and VO were present in the upper atmosphere of an irradiated gas giant, it would likely generate a thermal inversion, as incoming stellar radiation is absorbed at low pressures (Hubeny et al. 2003; Fortney et al. 2008). Although 1 School of Physics, University of Exeter, EX4 4QL Exeter, UK 2 NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA 3 Lunar and Planetary Laboratory, University of Arizona, Tucson, Arizona 85721, USA 4 Astrophysics Research Centre, School of Mathematics and Physics, Queens University Belfast, Belfast BT7 1NN, UK 5 Department of Applied Physics and Applied Mathematics, Columbia University, New York, NY 10025, USA 10025, USA 6 NASA Goddard Institute for Space Studies, New York, NY † [email protected] thermal inversions have previously been claimed for a number of exoplanets, most of these results were based on Spitzer Space Telescope secondary eclipse measure- ments since called into question (Diamond-Lowe et al. 2014; Hansen et al. 2014; Evans et al. 2015). Further- more, searches for TiO/VO in the hottest gas giants have failed to detect either species in transmission (Huitson et al. 2013; Sing et al. 2013). One possibility is that TiO/VO is depleted from the upper atmosphere by cold-trapping, either deeper in the dayside atmosphere or on the cooler nightside (Hubeny et al. 2003; Spiegel et al. 2009). The hottest planets, however, might avoid cold-traps. Haynes et al. (2015) have recently published near-infrared secondary eclipse observations for the gas giant WASP-33b, which has an equilibrium temperature of Teq ∼ 2700 K.8 These data reveal excess emission around 1.2µm, which is consistent with TiO and would imply a thermal inversion. To con- firm this hypothesis, further observations are required to spectrally resolve the TiO bandheads. In this Letter, we present near-infrared transmission spectroscopy observations for another ultra hot gas giant, WASP-121b (Delrez et al. 2016). This extreme planet is in a 1.3 day polar orbit around an F6V host star, with a semimajor axis of 0.025 AU. This puts WASP-121b just beyond the Roche limit, where it is subject to intense tidal forces. With an equilibrium temperature of Teq ∼ 2400 K, the atmosphere may be hot enough for gaseous TiO/VO to be abundant. WASP-121b is also one of the most inflated planets known, with mass 1.2MJ and radius 1.8RJ. These properties, combined with the brightness of the host star (J = 9.6 mag), make WASP-121b an 8 We quote equilibrium temperature Teq as the blackbody tem- perature required for planetary thermal emission to balance the stellar irradiation, assuming zero Bond albedo and uniform day- night recirculation. Due to the approximate nature of equilibrium temperature, we round values to the nearest 100 K in this Letter. 2 excellent target for atmospheric characterization. 2. OBSERVATIONS A single transit of WASP-121b was observed on 2016 February 6 UT using the Hubble Space Telescope (HST ) Wide Field Camera 3 (WFC3) for Program 14468 (P.I. Evans). Spectroscopic mode was used with grism G141 and forward scanning to allow longer exposures with- out saturating the detector (McCullough & MacKenty 2012). To reduce overheads, a 256 × 256 subarray con- taining the target spectrum was sampled with 15 non- destructive reads per exposure. Integration times were 103 seconds, using the SPARS10 readout mode.9 A scan rate of 0.12 arcsec/sec was adopted, giving scans along 100 pixel rows of the cross-dispersion axis for each ex- posure. Typical count levels remained below 2.2 × 104 electrons, well within the linear regime of the detector. Exposures were taken contiguously over five HST or- bits, with 17 exposures obtained per orbit. The first orbit allowed the telescope to settle into its new pointing position, and was discarded during the lightcurve analy- sis due to the large amplitude instrumental systematics it exhibited, as is standard practice for exoplanet tran- sit observations (e.g. Deming et al. 2013; Huitson et al. 2013; Stevenson et al. 2014; Haynes et al. 2015). Of the remaining four HST orbits, the first and fourth provided out-of-transit baseline, while the second and third oc- curred during the transit. 3. DATA REDUCTION Our analysis commenced with the IMA data files pro- duced by the CALWF3 pipeline (v3.1.6), which already have basic calibrations such as flat fielding and bias sub- traction applied. We extracted flux for WASP-121 from each exposure by taking the difference between succes- sive non-destructive reads. For each read difference, we removed the background by taking the median flux in a box of pixels well away from the stellar spectra. Typical background levels were 115 electrons. We then deter- mined the flux-weighted center of the WASP-121 scan, and set to zero all pixel values located more than 35 pixels above and below along the cross-dispersion axis. Application of this tophat filter had the effect of masking the flux contributions from nearby contaminants, includ- ing a faint star separated by 7 arcsec on the sky. It had the additional advantage of eliminating many of the pix- els affected by cosmic rays. Final reconstructed images were produced by adding together the read differences for each exposure. We extracted spectra from the recon- structed images by summing the flux within rectangular apertures centered on the scanned spectra and spanning the full dispersion axis. We experimented with aperture widths along the cross-dispersion axis ranging from 100 to 200 pixels in increments of 10 pixels, and found that an aperture of 160 pixels minimized the residuals in the final white lightcurve fit (Section 4). The wavelength solution was determined by cross-correlating the first spectrum, with dispersion 4.64 nm/pixel, against an ATLAS stellar model (Kurucz 1993) with properties similar to WASP- 121 (Teff = 6500 K, log g = 4.0 cgs, vturb = 2 km/s) mod- ulated by the throughput of the G141 grism. Prior to 9 See the WFC3 handbook at http://www.stsci.edu/hst/wfc3 White lightcurve fit results and adopted parameters TABLE 1 Parameter Value Rp/R(cid:63) 0.12109+0.00031 −0.00032 Tmid (HJDUTC) 2457424.88307+0.00010 −0.00011 a/R(cid:63) b i (◦) c1 c2 c3 c4 3.754 0.160 87.557 0.582 0.151 −0.435 0.199 cross-correlation, both the measured spectrum and AT- LAS spectrum were smoothed using a Gaussian filter with full-width half maximum of 20nm. As a result, the cross-correlation was most sensitive to the steep edges of the G141 response curve, rather than individual stellar lines. We repeated this process for all remaining spec- tra to determine shifts along the dispersion axis over the course of the observations, which were found to be within 0.23 pixels. 4. LIGHTCURVE ANALYSIS White lightcurves were generated for each trial aper- ture by summing the flux for each spectrum along the dis- persion axis. We discarded the first exposure from each orbit, as these had significantly lower counts than subse- quent exposures. Two additional frames were flagged as outliers, with closer inspection revealing that one was af- fected by a cosmic ray and the other exhibited an anoma- lous scan. The resulting lightcurve obtained for the 160 pixel aperture is shown in Figure 1. Instrumental sys- tematics that correlate with the HST orbital phase are evident, caused by the varying thermal environment ex- perienced by the spacecraft during its orbit (e.g. Sing et al. 2013; Wakeford et al. 2016). To model the white lightcurves, we adopted a Man- del & Agol (2002) analytic function for the planet signal. For stellar limb darkening, we used the four-parameter nonlinear limb law of Claret (2000) with coefficients (c1, c2, c3, c4) listed in Table 1. The latter were obtained by fitting to the limb-darkened intensities of the ATLAS stellar model described in Section 3 multiplied by the G141 throughput profile. For the white lightcurve fitting, we treated the data as a Gaussian process (GP), using the approach described in Gibson et al. (2012). For the mean function, we adopted a Mandel & Agol (2002) transit model multiplied by a linear time trend, and for the covariance matrix, we used a Mat´ern ν = 3/2 kernel (see Gibson et al. 2013) with HST orbital phase and dispersion shift as the input vari- ables. We allowed the linear trend and covariance pa- rameters to vary in the fitting, along with the planet- to-star radius ratio Rp/R(cid:63) and transit mid-time Tmid. The orbital period P , normalized semimajor axis a/R(cid:63), orbital inclination i, and eccentricity e were fixed to val- ues reported in Delrez et al. (2016) and listed in Table 1. Uniform priors were adopted for Rp/R(cid:63), Tmid, and 3 mented by the emcee software package (Foreman-Mackey et al. 2013). This was done by initializing 100 walkers in close proximity to the maximum likelihood solution, and allowing them to run for 300 steps. Correlation length scales were calculated for each parameter, and a burn-in phase of three times the longest correlation length scale was discarded from all walker chains before combining them into a single chain. The resulting posterior samples displayed good mixing and convergence. Best-fit model residuals for the lightcurve generated with the 160 pixel aperture (Section 3) had the lowest scatter, within 5% of the photon noise floor. We adopt this reduction for all subsequent analysis, and report the results in Table 1 with best-fit model shown in Figure 1. After fitting the white lightcurve, spectroscopic lightcurves were produced using a similar approach to Deming et al. (2013). First, a reference spectrum was produced by taking the average of the out-of-transit spec- tra. Each individual spectrum was then shifted laterally in wavelength and stretched vertically in flux to match the reference spectrum, using linear least squares. The residuals of these fits were binned into 28 spectroscopic channels across the 1.12–1.64 µm wavelength range, each spanning 4 columns of the dispersion axis. The spec- troscopic residuals were then added to a transit signal with Rp/R(cid:63) and Tmid set to the white lightcurve best- fit values (Table 1), and limb darkening appropriate to the wavelength channel, giving the final spectroscopic lightcurves shown in Figure 2. This process reduces sys- tematics that are common-mode in wavelength, as well as those arising due to the spectra drifting along the disper- sion axis. It retains, however, channel-to-channel differ- ences in the flux level, including possible transit depth variations caused by the wavelength-dependent opacity of the planetary atmosphere. We fit the spectroscopic lightcurves using the same GP treatment described above for the white lightcurve. For these fits, Rp/R(cid:63) was allowed to vary separately for each channel, while Tmid was held fixed to the best-fit white lightcurve value. Again, a nonlinear limb darkening law was used with coefficients fixed to the values reported in Table 2, which were determined using the same method described in Section 3 for the white lightcurve. We report the results for Rp/R(cid:63) in Table 2 and show best-fit models in Figure 2. Following Wakeford et al. (2016), we also performed an independent analysis of the white and spectroscopic lightcurves using the method of Gibson (2014) in which marginalization was performed over a grid of parametric systematics models. This gave results in good agreement with the GP analysis. In addition to the spectroscopic WFC3 data, we fit the ground-based photometric transit data presented in Delrez et al. (2016), comprised of three B band lightcurves, two r(cid:48) lightcurves, and four z(cid:48) lightcurves. For all ground-based lightcurves we used a GP system- atics model, with a transit signal multiplied by a linear time trend as the mean function. For the covariance matrix, we adopted a Mat´ern ν = 3/2 kernel with mul- tiple input variables such as the xy coordinates of the star on the detector, seeing, and airmass. Lightcurves in the same bandpass were fit simultaneously, with Rp/R(cid:63) shared but Tmid, linear trends, covariance parameters, and white noise levels allowed to vary separately. Non- Fig. 1.- HST /WFC3 G141 white transit lightcurve for WASP- 121b. (a) Raw flux time series, with solid lines indicating the pre- dictive mean of the maximum likelihood GP model. (b) Residuals after removing the transit and linear time trend, which correlate with HST phase and dispersion drift. (c) Relative flux as a func- tion of time after removing the linear time trend and systematics component of the GP model, with solid line indicating the inferred transit signal. (d) Residuals as a function of time with photon noise errorbars after removing the combined transit, linear time trend, and GP systematics model. the linear trend parameters. Gamma priors of the form Gam(α = 1, β = 100) and Gam(α = 1, β = 1) were adopted for the covariance amplitude and inverse cor- relation length scales, respectively, giving preference to simpler systematics models, i.e. smaller covariance am- plitudes and longer correlation length scales. Maximum likelihood solutions were located using nonlinear opti- mization, as implemented by the fmin routine of the scipy.optimize Python software package.10 The like- lihood distribution was then marginalized using affine- invariant Markov chain Monte Carlo (MCMC), as imple- 10 http://scipy.org −150−100−500501001500.9850.9900.9951.000RelativefluxTimefrommid-transit(min)(a)0.30.40.50.60.7−1000−50005001000Residuals(ppm)HSTorbitalphase(b)orbit1orbit2orbit3orbit40.9850.9900.9951.000Relativeflux(c)−150−100−50050100150−300−200−1000100200300Residuals(ppm)Timefrommid-transit(min)(d)4 Fig. 2.- (Top) Raw spectroscopic lightcurves with solid lines showing best-fit transit signals multiplied by linear time trends. (Bottom) Model residuals with photon noise errorbars in gray and solid lines showing GP model fits. Note that the transit signals, linear time trends, and GP systematics models were fit simultaneously in practice, but have been separated here for illustrative purposes. 0.950.960.970.980.991.001.011.021.031.13µm1.15µm1.17µm1.19µm1.20µm1.22µm1.24µm1.26µm1.28µm1.30µm1.32µm1.33µm1.35µm1.37µm1.39µm1.41µm1.43µm1.45µm1.46µm1.48µm1.50µm1.52µm1.54µm1.56µm1.58µm1.59µm1.61µm1.63µm−140−70070140−0.8−0.40.00.40.8−140−70070140−140−70070140−140−70070140RelativefluxResiduals(%)Timefrommid-transit(min)Inferred transmission spectrum and adopted nonlinear limb darkening coefficients TABLE 2 λ (µm) Rp/R(cid:63) c1 c2 c3 c4 0.392–0.481 0.555–0.670 0.836–0.943 1.121–1.139 1.139–1.158 1.158–1.177 1.177–1.195 1.195–1.214 1.214–1.232 1.232–1.251 1.251–1.269 1.269–1.288 1.288–1.307 1.307–1.325 1.325–1.344 1.344–1.362 1.362–1.381 1.381–1.399 1.399–1.418 1.418–1.437 1.437–1.455 1.455–1.474 1.474–1.492 1.492–1.511 1.511–1.529 1.529–1.548 1.548–1.567 1.567–1.585 1.585–1.604 1.604–1.622 1.622–1.641 0.12375+0.00061 −0.00069 0.12521+0.00065 −0.00069 0.12298+0.00114 −0.00117 0.12033+0.00038 −0.00038 0.12107+0.00035 −0.00036 0.12122+0.00035 −0.00033 0.12149+0.00035 −0.00034 0.12116+0.00034 −0.00034 0.12113+0.00033 −0.00036 0.12143+0.00034 −0.00035 0.12070+0.00037 −0.00036 0.12036+0.00036 −0.00035 0.12033+0.00036 −0.00037 0.12035+0.00033 −0.00033 0.12118+0.00034 −0.00034 0.12201+0.00035 −0.00034 0.12193+0.00034 −0.00035 0.12138+0.00035 −0.00036 0.12118+0.00037 −0.00035 0.12156+0.00035 −0.00036 0.12164+0.00035 −0.00036 0.12167+0.00035 −0.00036 0.12088+0.00039 −0.00040 0.12160+0.00037 −0.00037 0.12101+0.00039 −0.00039 0.12053+0.00042 −0.00039 0.12123+0.00044 −0.00042 0.12057+0.00041 −0.00041 0.12030+0.00042 −0.00041 0.12071+0.00042 −0.00041 0.11954+0.00042 −0.00043 0.227 0.205 0.194 0.259 0.266 0.186 0.205 0.139 0.151 0.109 0.124 0.083 0.086 0.465 0.492 0.507 0.356 0.499 0.364 0.500 0.343 0.543 0.236 0.496 0.325 0.487 0.338 0.491 0.319 0.497 0.270 0.500 0.255 0.502 0.239 0.501 0.233 0.507 0.201 0.337 0.653 0.309 0.553 −0.098 −0.041 −0.408 0.104 −0.311 −0.268 −0.318 −0.346 −0.377 −0.404 −0.482 −0.519 −0.518 −0.486 −0.565 −0.620 −0.628 −0.630 −0.609 −0.659 −0.634 −0.601 −0.534 −0.521 −0.474 −0.459 0.634 0.667 −0.002 −0.388 0.690 −0.058 −0.352 0.715 −0.098 −0.334 0.723 −0.147 −0.276 0.723 −0.170 −0.241 0.751 −0.227 −0.201 0.754 −0.214 −0.214 0.516 0.342 0.528 0.314 0.534 0.327 0.549 0.293 0.567 0.248 0.590 0.180 0.608 0.154 0.622 0.107 0.150 0.154 0.293 0.286 0.232 0.229 0.199 0.176 0.271 0.264 0.278 0.252 0.082 0.249 0.209 0.202 0.161 linear limb darkening laws were adopted, with coefficients fixed to the values reported in Table 2, which were taken from the catalogues of Claret (2000, 2003). Remaining transit parameters were fixed to the values adopted for the WFC3 analysis, namely, those reported by Delrez et al. (2016) and listed in Table 1. Fitting was performed by locating the maximum likelihood and marginalizing over the parameter space with affine-invariant MCMC, as described above. Results are reported in Table 2. 5. TRANSMISSION SPECTRUM The measured transmission spectrum for WASP-121b is plotted in Figure 3. The WFC3 data show struc- ture, with two broad features centered around 1.2 µm and 1.4 µm. The scale of these variations is on the order of a few atmospheric scale heights, consistent with expec- 5 tations for molecular absorption. Additionally, the effec- tive planetary radius is found to be significantly larger at optical wavelengths relative to near-infrared wave- lengths. Before exploring implications for the planetary atmo- sphere, we considered the possibility of stellar activity causing the different radii measured at optical and near- infrared wavelengths. Making the conservative assump- tion that star spots are non-luminous, we calculate that the unocculted spot coverage would need to decrease by an amount equivalent to 7% of the visible stellar disc be- tween the optical and near-infrared observation epochs. This seems highly unlikely, especially given that WASP- 121 appears photometrically stable at the millimag level (Delrez et al. 2016). Also, analyses of the individual lightcurves for each bandpass produced consistent results at the 1σ level (top panel of Figure 3), suggesting that stellar activity does not affect the inferred radius signifi- cantly from epoch to epoch. We therefore conclude that variations in the inferred radius shown in Figure 3 are due to the planetary atmosphere. We used the 1D radiative transfer code ATMO to gen- erate model spectra and investigate atmospheric opac- ity sources (Amundsen et al. 2014; Tremblin et al. 2015, 2016). Isothermal temperature-pressure profiles were as- sumed, with temperature allowed to vary in the fitting. We produced models with and without TiO, VO, and FeH. The latter species are observed in M/L dwarfs (Bur- rows et al. 2001) and models predict they may also be present in hot gas giant planets (e.g. Sharp & Burrows 2007). For models including TiO/VO/FeH, we varied the relative abundances of TiO/H2O, VO/H2O, and Fe/H2O during fitting. Solar abundances under thermochemical equilibrium were adopted for other gas species, including H2O. We note that for transmission spectra, the absolute abundances of individual species are often unconstrained owing to a well-known degeneracy with the absolute pres- sure level, but the relative abundances between species can be well measured (Lecavelier Des Etangs et al. 2008). In our fitting, TiO/VO/FeH abundances were adjusted using the analytical relation of Lecavelier Des Etangs et al. (2008) for the wavelength-dependent transit ra- dius of an isothermal atmosphere. We also considered a simple haze scattering model in the form of a Rayleigh profile with collisional cross-section area allowed to vary as a free parameter, as described in Sing et al. (2016). Best fits for a number of illustrative models are shown in Figure 3, obtained using the nonlinear least squares IDL routine mpfit (Markwardt 2009). We find that a model excluding TiO, VO, FeH, and haze is unable to account for the data (purple line). In particular, it fails to reproduce the optical photometry and short wave- length WFC3 data. A good fit is achieved, however, when TiO, VO, and FeH opacity is included (red line). For this model, we find a temperature of 1500 ± 230 K, and abundances relative to thermochemical equilibrium of ∼ 7× solar for TiO/H2O, ∼ 5× solar for VO/H2O, and ∼ 0.2× solar for FeH/H2O. The inferred tempera- ture is lower than the dayside equilibrium temperature of Teq ∼ 2400 K, which may indicate a significantly cooler nightside hemisphere. The strong absorption of TiO and VO in the optical accounts for the larger radii measured at these wavelengths relative to the near-infrared. To illustrate the effect of FeH, we also generated a model 6 Fig. 3.- Measured transmission spectrum for WASP-121b, with models. In the top panel, unfilled gray data points show results for fits to individual photometric lightcurves, with horizontal offsets applied for clarity, and filled black data points show results for joint fits to all lightcurves in the corresponding bandpass. The lower panel shows a zoomed-in view of the WFC3 transmission measurements. Models assume solar abundances and chemical equilibrium, with the exception of TiO, VO, and FeH, the abundances of which were adjusted to fit the data. (Red line) Clear atmosphere including TiO, VO, and FeH opacity. (Orange line) Same as previous, but excluding FeH opacity to illustrate its effect. (Green line) Hazy atmosphere with enhanced Rayleigh scattering, excluding TiO, VO, and FeH opacity. (Purple line) Clear atmosphere excluding TiO, VO, and FeH opacity. 0.40.60.81.01.21.41.60.1180.1200.1220.1240.126clear,withTiO/VOandFeH,BIC=42.6haze,noTiO/VO,noFeH,BIC=50.4clear,withTiO/VO,noFeHclear,noTiO/VO,noFeH−2024681.11.21.31.41.51.60.1190.1200.1210.122−10123Rp/R?PressurescaleheightsRp/R?PressurescaleheightsWavelength(µm)Wavelength(µm)with it removed, but TiO and VO retained (orange line). Such a model is unable to reproduce the WFC3 transmis- sion spectrum at wavelengths near 1.3µm, where FeH has prominent absorption. Lastly, a model excluding TiO, VO, and FeH, but including haze, gives a relatively poor fit to the data (green line). Most significantly, haze scat- tering underpredicts the r(cid:48) opacity by 2.6σ. To assess the relative quality of fit between the haze and TiO/VO/FeH models, we computed the Bayesian information criterion (BIC) for each, obtaining BIC = 50.4 for the haze model and BIC = 42.6 for the TiO/VO/FeH model. Following Kass & Raftery (1995), we approximate the Bayes factor as exp(−∆BIC/2) = 0.02 in favor of the TiO/VO/FeH model over the haze model. This strongly supports the hypothesis that TiO and VO, rather than haze, are re- sponsible for the enhanced opacity measured at optical wavelengths. All models favor H2O absorption, which matches the measured spectrum well between 1.3–1.64µm. To quan- tify the significance of this detection, we compared the quality of fit obtained for the best-fit TiO/VO/FeH model described above with that obtained for the same model but H2O removed. We determine that H2O is detected at a confidence level of 5.4σ. 6. DISCUSSION Our observations reveal H2O absorption in the atmo- sphere of WASP-121b with an amplitude of > 2 gas pres- sure scale heights. This result is consistent with a clear atmosphere, assuming TiO/VO and FeH absorption is re- sponsible for the additional opacity measured across the 1.12–1.3µm wavelength range. In contrast, WASP-12b is observed to have a hazy atmosphere without TiO/VO (Sing et al. 2013), despite having similar properties to WASP-121b, reinforcing the emerging picture of hot gas giant diversity (Sing et al. 2016). We note, however, that the optical data cannot currently exclude a model including both TiO/VO absorption and haze scattering for WASP-121b. Such a model would be very similar to our best-fit model (red line in Figure 3), except that the effective radius would continue increasing for wave- lengths shortward of the B bandpass. To confidently discount such a scenario, it will be necessary to spec- trally resolve the characteristic steep rise in opacity at ∼ 0.35µm due to TiO/VO with further observations. Until then, the evidence for TiO/VO absorption in the atmosphere of WASP-121b remains tentative, as it pri- marily hinges upon the relative radius measured from the ground in the r(cid:48) bandpass. We emphasize, however, that this measurement is based on two lightcurves that produce consistent results when analyzed separately and simultaneously (top panel of Figure 3). In addition, Del- rez et al. (2016) report ground-based secondary eclipse measurements in the z(cid:48) bandpass that imply a bright- ness temperature higher than the equilibrium temper- ature predicted for zero Bond albedo and instantaneous heat re-radiation. This could potentially be explained by a thermal inversion, with TiO being observed in emission in the z(cid:48) bandpass. Again, follow-up observations that spectrally resolve the TiO/VO features will be necessary for confirmation. To date, HST spectroscopy data have been published for three exoplanets with Teq > 2000 K. Of these, 7 TiO/VO absorption has been ruled out for WASP-12b (Teq ∼ 2600 K; Sing et al. 2013) and WASP-19b (Teq ∼ 2100 K; Huitson et al. 2013), while evidence for TiO emis- sion has been reported for WASP-33b (Teq ∼ 2700 K; Haynes et al. 2015). It is possible that WASP-33b is hot enough to maintain TiO/VO in the gas phase throughout its dayside atmosphere, whereas the day-night termina- tors of WASP-12b and WASP-19b are too cool, or cold- trapping occurs elsewhere in their atmospheres. How- ever, the fact that WASP-121b has a lower equilibrium temperature (Teq ∼ 2400 K) than WASP-12b, and yet shows evidence for TiO/VO absorption at low pressures, indicates that additional factors are at play. For instance, models predict larger temperature contrasts between the dayside and nightside hemispheres with increasing stellar irradiation and planet metallicity (Fortney et al. 2008; Dobbs-Dixon & Lin 2008; Kataria et al. 2015). It is thus plausible that WASP-12b, with its higher irradi- ation and host star metallicity, has a cooler nightside than WASP-121b, which could result in a nightside cold- trap for WASP-12b but not WASP-121b. Compositional differences could also have a significant influence on the temperature profiles for each planet (Parmentier & Guil- lot 2014; Tremblin et al. 2016), in turn determining the susceptibility to cold-trapping. 7. CONCLUSION We have presented a transmission spectrum for WASP- 121b spanning the 1.12–1.64µm wavelength range. Ab- sorption by H2O is detected at high confidence (5.4σ), consistent with predictions of clear atmosphere models. Deeper transits measured at optical wavelengths rela- tive to the near-infrared strongly favor models including TiO/VO absorption, but scattering by a high-altitude haze cannot yet be definitively excluded. We also find evidence for FeH absorption in the WFC3 bandpass. WASP-121b is one of the most favorable targets avail- able for both transmission and emission spectroscopy, and offers particular promise for exploring the link be- tween strong optical absorbers, such as TiO/VO, and thermal inversions in hot gas giant atmospheres. Based on observations made with the NASA/ESA Hubble Space Telescope, obtained at the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., under NASA contract NAS 5-26555. The authors are grateful to the WASP-121 discovery team for generously provid- ing the ground-based photometric lightcurves. Support for this work was provided by NASA through grants un- der the HST-GO-14468 program from the STSci. The research leading to these results has received funding from the European Research Council under the Euro- pean Union Seventh Framework Program (FP7/2007- 2013) ERC grant agreement no. 336792. HRW acknowl- edges support by an appointment to the NASA Postdoc- toral Program at Goddard Space Flight Center, admin- istered by ORAU and USRA through a contract with NASA. NPG gratefully acknowledges support from the Royal Society in the form of a University Research Fel- lowship. 8 REFERENCES Amundsen, D. S., Baraffe, I., Tremblin, P., Manners, J., Hayek, W., Mayne, N. J., & Acreman, D. M. 2014, A&A, 564, A59 Burrows, A., Hubbard, W. B., Lunine, J. I., & Liebert, J. 2001, Reviews of Modern Physics, 73, 719 Burrows, A. & Sharp, C. M. 1999, ApJ, 512, 843 Charbonneau, D., Brown, T. M., Noyes, R. W., & Gilliland, R. L. 2002, ApJ, 568, 377 Claret, A. 2000, A&A, 363, 1081 -. 2003, A&A, 401, 657 Delrez, L., Santerne, A., Almenara, J.-M., Anderson, D. R., Collier-Cameron, A., D´ıaz, R. F., Gillon, M., Hellier, C., Jehin, E., Lendl, M., Maxted, P. F. L., Neveu-VanMalle, M., Pepe, F., Pollacco, D., Queloz, D., S´egransan, D., Smalley, B., Smith, A. M. S., Triaud, A. H. M. J., Udry, S., Van Grootel, V., & West, R. G. 2016, MNRAS Deming, D., Wilkins, A., McCullough, P., Burrows, A., Fortney, J. J., Agol, E., Dobbs-Dixon, I., Madhusudhan, N., Crouzet, N., Desert, J.-M., Gilliland, R. L., Haynes, K., Knutson, H. A., Line, M., Magic, Z., Mandell, A. M., Ranjan, S., Charbonneau, D., Clampin, M., Seager, S., & Showman, A. P. 2013, ApJ, 774, 95 Diamond-Lowe, H., Stevenson, K. B., Bean, J. L., Line, M. R., & Fortney, J. J. 2014, ApJ, 796, 66 Dobbs-Dixon, I. & Lin, D. N. C. 2008, ApJ, 673, 513 Evans, T. M., Aigrain, S., Gibson, N., Barstow, J. K., Amundsen, D. S., Tremblin, P., & Mourier, P. 2015, MNRAS, 451, 680 Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306 Fortney, J. J., Lodders, K., Marley, M. S., & Freedman, R. S. 2008, ApJ, 678, 1419 Gibson, N. P. 2014, MNRAS, 445, 3401 Gibson, N. P., Aigrain, S., Barstow, J. K., Evans, T. M., Fletcher, L. N., & Irwin, P. G. J. 2013, MNRAS, 428, 3680 Gibson, N. P., Aigrain, S., Roberts, S., Evans, T. M., Osborne, M., & Pont, F. 2012, MNRAS, 419, 2683 Hansen, C. J., Schwartz, J. C., & Cowan, N. B. 2014, MNRAS, 444, 3632 Haynes, K., Mandell, A. M., Madhusudhan, N., Deming, D., & Knutson, H. 2015, ApJ, 806, 146 Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011 Huitson, C. M., Sing, D. K., Pont, F., Fortney, J. J., Burrows, A. S., Wilson, P. A., Ballester, G. E., Nikolov, N., Gibson, N. P., Deming, D., Aigrain, S., Evans, T. M., Henry, G. W., Lecavelier des Etangs, A., Showman, A. P., Vidal-Madjar, A., & Zahnle, K. 2013, MNRAS, 434, 3252 Kass, R. E. & Raftery, A. E. 1995, Journal of the American Statistical Association, 90, 773 Kataria, T., Showman, A. P., Fortney, J. J., Stevenson, K. B., Line, M. R., Kreidberg, L., Bean, J. L., & D´esert, J.-M. 2015, ApJ, 801, 86 Kirkpatrick, J. D., Reid, I. N., Liebert, J., Cutri, R. M., Nelson, B., Beichman, C. A., Dahn, C. C., Monet, D. G., Gizis, J. E., & Skrutskie, M. F. 1999, ApJ, 519, 802 Kreidberg, L., Line, M. R., Bean, J. L., Stevenson, K. B., D´esert, J.-M., Madhusudhan, N., Fortney, J. J., Barstow, J. K., Henry, G. W., Williamson, M. H., & Showman, A. P. 2015, ApJ, 814, 66 Kurucz, R. 1993, ATLAS9 Stellar Atmosphere Programs and 2 km/s grid. Kurucz CD-ROM No. 13. Cambridge, Mass.: Smithsonian Astrophysical Observatory, 1993., 13 Lecavelier Des Etangs, A., Pont, F., Vidal-Madjar, A., & Sing, D. 2008, A&A, 481, L83 Mandel, K. & Agol, E. 2002, ApJ, 580, L171 Markwardt, C. B. 2009, in Astronomical Society of the Pacific Conference Series, Vol. 411, Astronomical Data Analysis Software and Systems XVIII, ed. D. A. Bohlender, D. Durand, & P. Dowler, 251 McCullough, P. & MacKenty, J. 2012, Considerations for using Spatial Scans with WFC3, Tech. rep. McCullough, P. R., Crouzet, N., Deming, D., & Madhusudhan, N. 2014, ApJ, 791, 55 Nikolov, N., Sing, D. K., Burrows, A. S., Fortney, J. J., Henry, G. W., Pont, F., Ballester, G. E., Aigrain, S., Wilson, P. A., Huitson, C. M., Gibson, N. P., D´esert, J.-M., Etangs, A. L. d., Showman, A. P., Vidal-Madjar, A., Wakeford, H. R., & Zahnle, K. 2015, MNRAS, 447, 463 Nikolov, N., Sing, D. K., Pont, F., Burrows, A. S., Fortney, J. J., Ballester, G. E., Evans, T. M., Huitson, C. M., Wakeford, H. R., Wilson, P. A., Aigrain, S., Deming, D., Gibson, N. P., Henry, G. W., Knutson, H., Lecavelier des Etangs, A., Showman, A. P., Vidal-Madjar, A., & Zahnle, K. 2014, MNRAS, 437, 46 Parmentier, V. & Guillot, T. 2014, A&A, 562, A133 Pont, F., Knutson, H., Gilliland, R. L., Moutou, C., & Charbonneau, D. 2008, MNRAS, 385, 109 Sharp, C. M. & Burrows, A. 2007, ApJS, 168, 140 Sing, D. K., Fortney, J. J., Nikolov, N., Wakeford, H. R., Kataria, T., Evans, T. M., Aigrain, S., Ballester, G. E., Burrows, A. S., Deming, D., D´esert, J.-M., Gibson, N. P., Henry, G. W., Huitson, C. M., Knutson, H. A., Etangs, A. L. D., Pont, F., Showman, A. P., Vidal-Madjar, A., Williamson, M. H., & Wilson, P. A. 2016, Nature, 529, 59 Sing, D. K., Lecavelier des Etangs, A., Fortney, J. J., Burrows, A. S., Pont, F., Wakeford, H. R., Ballester, G. E., Nikolov, N., Henry, G. W., Aigrain, S., Deming, D., Evans, T. M., Gibson, N. P., Huitson, C. M., Knutson, H., Showman, A. P., Vidal-Madjar, A., Wilson, P. A., Williamson, M. H., & Zahnle, K. 2013, MNRAS, 436, 2956 Sing, D. K., Wakeford, H. R., Showman, A. P., Nikolov, N., Fortney, J. J., Burrows, A. S., Ballester, G. E., Deming, D., Aigrain, S., D´esert, J.-M., Gibson, N. P., Henry, G. W., Knutson, H., Lecavelier des Etangs, A., Pont, F., Vidal-Madjar, A., Williamson, M. W., & Wilson, P. A. 2015, MNRAS, 446, 2428 Spiegel, D. S., Silverio, K., & Burrows, A. 2009, ApJ, 699, 1487 Stevenson, K. B. 2016, ApJ, 817, L16 Stevenson, K. B., D´esert, J.-M., Line, M. R., Bean, J. L., Fortney, J. J., Showman, A. P., Kataria, T., Kreidberg, L., McCullough, P. R., Henry, G. W., Charbonneau, D., Burrows, A., Seager, S., Madhusudhan, N., Williamson, M. H., & Homeier, D. 2014, Science, 346, 838 Tremblin, P., Amundsen, D. S., Chabrier, G., Baraffe, I., Drummond, B., Hinkley, S., Mourier, P., & Venot, O. 2016, ApJ, 817, L19 Tremblin, P., Amundsen, D. S., Mourier, P., Baraffe, I., Chabrier, G., Drummond, B., Homeier, D., & Venot, O. 2015, ApJ, 804, L17 Wakeford, H. R., Sing, D. K., Deming, D., Gibson, N. P., Fortney, J. J., Burrows, A. S., Ballester, G., Nikolov, N., Aigrain, S., Henry, G., Knutson, H., Lecavelier des Etangs, A., Pont, F., Showman, A. P., Vidal-Madjar, A., & Zahnle, K. 2013, MNRAS, 435, 3481 Wakeford, H. R., Sing, D. K., Evans, T., Deming, D., & Mandell, A. 2016, ApJ, 819, 10
1211.6033	1	1211	2012-11-26T17:38:57	WASP-77 Ab: A transiting hot Jupiter planet in a wide binary system	[ "astro-ph.EP" ]	We report the discovery of a transiting planet with an orbital period of 1.36d orbiting the brighter component of the visual binary star BD -07 436. The host star, WASP-77A, is a moderately bright G8V star (V=10.3) with a metallicity close to solar ([Fe/H]= 0.0 +- 0.1). The companion star, WASP-77B, is a K-dwarf approximately 2 magnitudes fainter at a separation of approximately 3arcsec. The spectrum of WASP-77A shows emission in the cores of the Ca II H and K lines indicative of moderate chromospheric activity. The WASP lightcurves show photometric variability with a period of 15.3 days and an amplitude of about 0.3% that is probably due to the magnetic activity of the host star. We use an analysis of the combined photometric and spectroscopic data to derive the mass and radius of the planet (1.76+-0.06MJup, 1.21+-0.02RJup). The age of WASP-77A estimated from its rotation rate (~1 Gyr) agrees with the age estimated in a similar way for WASP-77B (~0.6 Gyr) but is in poor agreement with the age inferred by comparing its effective temperature and density to stellar models (~8 Gyr). Follow-up observations of WASP-77 Ab will make a useful contribution to our understanding of the influence of binarity and host star activity on the properties of hot Jupiters.	astro-ph.EP	astro-ph	WASP-77 Ab: A transiting hot Jupiter planet in a wide binary system.⋆ P. F. L. Maxted1 , D. R. Anderson1 , A. Collier Cameron2 , A. P. Doyle1 , A. Fumel3 , M. Gillon3 , C. Hellier1 , E. Jehin3 , M. Lendl4 , F. Pepe4 , D. L. Pollacco5,6 , D. Queloz4 , D. S´egransan4 , B. Smalley1 , J. K. Southworth 1 , A. M. S. Smith1 , A. H. M. J. Triaud4 , S. Udry4 , R. G. West7 , ABSTRACT We report the discovery of a transiting planet with an orbital period of 1.36 d or- biting the brighter component of the visual binary star BD −07◦ 436. The host star, WASP-77 A, is a moderately bright G8 V star (V=10.3) with a metallicity close to solar ([Fe/H]= 0.0 ± 0.1). The companion star, WASP-77 B, is a K-dwarf approximately 2 magnitudes fainter at a separation of approximately 3′′ . The spectrum of WASP-77 A shows emission in the cores of the Ca II H and K lines indicative of moderate chromo- spheric activity. The WASP lightcurves show photometric variability with a period of 15.3 days and an amplitude of about 0.3% that is probably due to the magnetic activity of the host star. We use an analysis of the combined photometric and spectroscopic data to derive the mass and radius of the planet (1.76 ± 0.06MJup , 1.21 ± 0.02RJup ). The age of WASP-77 A estimated from its rotation rate (∼ 1 Gyr) agrees with the age estimated in a similar way for WASP-77 B (∼ 0.6 Gyr) but is in poor agreement with the age inferred by comparing its effective temperature and density to stellar models (∼ 8 Gyr). Follow-up observations of WASP-77 Ab will make a useful contribution to our understanding of the influence of binarity and host star activity on the properties of hot Jupiters. Subject headings: Extrasolar planets 1Astrophysics Group, Keele University, Staffordshire, ST5 5BG, UK 2SUPA, School of Physics and Astronomy, University of St. Andrews, North Haugh, Fife, KY16 9SS, UK 3 Institut d’Astrophysique et de G´eophysique, Universit´e de Li`ege, All´ee du 6 Aout, 17, Bat. B5C, Li`ege 1, Belgium 4Observatoire astronomique de l’Universit´e de Gen`eve 51 ch. des Maillettes, 1290 Sauverny, Switzerland 5Department of Physics, University of Warwick, Coventry, CV4 7AL 6Astrophysics Research Centre, School of Mathematics & Physics, Queen’s University, University Road, Belfast, BT7 1NN, UK 7Department of Physics and Astronomy, University of Leicester, Leicester, LE1 7RH, UK ⋆Based on observations made with ESO Telescopes at the La Silla Paranal Observatory under programme ID 088.C-0011. – 2 – 1. Introduction Ground-based wide-angle surveys such as WASP (Pollacco et al. 2006) and HATnet (Bakos et al. 2004) have now discovered more than 100 transiting hot Jupiter exoplanets around moderately bright stars (8.5 . V . 12.5). The occurrence rate for hot Jupiters around solar-type stars is known to be about 1% from radial velocity surveys of bright stars (Wright et al. 2012). The transit probability for a typical hot Jupiter with an orbital period ≈ 3 days orbiting a solar-type star is ≈ 10%. There are at least 340 000 single FGK-type dwarf stars bright enough to be accessible to survey such as WASP and HATnet (Ammons et al. 2006), so many more transiting hot Jupiters re- main to be discovered. Increasing the sample of well-characterised transiting hot Jupiters will clarify relations that may exist between parameters such as period, mass, radius, etc., and so enable us to better understand the formation, evolution and destruction of hot Jupiters (e.g., Matsumura et al. 2010; Knutson et al. 2010; Davis and Wheatley 2009; Batygin et al. 2011). Finding new transiting hot Jupiters will also reveal systems that have extreme properties or unusual configurations that en- able them to be characterised in ways not possible for typical hot Jupiters, e.g., low density, bright planets such as WASP-17 that can be characterised by transmission spectroscopy (Wood et al. 2011). Detailed characterisation of a hot Jupiter using ground based observations is always chal- lenging given the small size of the signal due to the secondary eclipse (. 0.1%) or the variation of transit depth with wavelength (. 0.01%). Such observations are made easier by the availability of a nearby comparison star that can be used as a comparison source, particularly if the companion is close enough to be included in the same entrance slit for spectroscopic observations (e.g., Sing et al. 2012). Here we report the discovery by the WASP survey of a companion to the brighter component of the visual binary star BD −07◦ 436 with a mass ≈ 1.7MJup and a radius ≈ 1.2RJup . We find that this star (WASP-77 A) is a G8 V star showing moderate chromospheric activity. The planet, WASP-77 Ab, is a typical hot Jupiter planet with an orbital period of 1.36 days. The companion star, WASP-77 B, is a K-dwarf approximately 2 magnitudes fainter then WASP-77 A at a separation of approximately 3′′ . We show that this star is physically associated with the star-planet system WASP-77A + WASP-77Ab. 2. Observations The WASP survey is described in Pollacco et al. (2006) and Wilson et al. (2008) while a discus- sion of our candidate selection methods can be found in Collier Cameron et al. (2007), Pollacco et al. (2008), and references therein. The star BD −07◦ 436 (WASP-77, 1SWASP J022837.22−070338.4) was observed 5594 times by one camera on the WASP-South instrument from 2008 July 30 to 2008 December 12. The synthetic aperture radius used to measure the flux of BD −07◦ 436 (48′′ ) includes the flux from both components of this visual binary star. We selected this star for follow-up observations based – 3 – on the characteristics of the periodic transit-like features detected in these data using the de- trending and transit detection methods described in Collier Cameron et al. (2007). The transits are also detected with the same period from 3316 observations obtained with the same camera from 2009 August 2 to 2009 December 12. We have also analysed 898 observations obtained with a different camera obtained from 2008 August 18 to 2008 December 25. The WASP photometry is shown as a function of phase for a period of 1.36003 days in Fig. 1. We obtained 11 radial velocity measurements for WASP-77 A using the fibre-fed CORALIE spectrograph on the Euler 1.2-m telescope located at La Silla, Chile. Details of the instrument and data reduction can be found in Queloz et al. (2000) and Wilson et al. (2008). The RV measurements were performed using cross-correlation against a numerical mask generated from a G2-type star and are given in Table 1, where we also provide the bisector span, BS, which measures the asymmetry of the cross-correlation function. The standard error of the the bisector span measurements is estimated to be 2σRV . The amplitude and phase of the radial velocity variations and the lack of any significant variation in the bisector span from these data are consistent with the hypothesis that the transit signal in the WASP photometry is due to a planetary mass ob ject. However, the diameter of the entrance aperture to the CORALIE spectrograph (2′′ ) is not quite sufficient to completely exclude light from the fainter component contaminating the spectrum of the brighter component and vice versa, so the CORALIE data by themselves are not sufficient to exclude the possibility that the transit signal originates from WASP-77 B. We also obtained 4 radial velocity measurements of WASP-77 B with CORALIE, but these are not reported here because the spectra are clearly affected by contamination from the brighter component. Confirmation that the transit signal originates from the brighter component of BD −07◦ 436 was obtained using the 60 cm TRAPPIST telescope (Jehin et al. 2011; Gillon et al. 2012) located at ESO La Silla Observatory (Chile). We obtained a sequence of 671 images of BD −07◦ 436 covering the egress of a transit in good seeing conditions using a z′ filter on the night 2011 November 02. From a selection of these images obtained in the best seeing we estimate that the fainter component is 3.22 ± 0.05 times fainter than the brighter component. If the fainter star were responsible for the transit signal in the WASP photometry then the eclipse in the lightcurve of this star would have a depth of about 7%. The stars are not completely resolved in these images, but lightcurves of the two stars obtained using a synthetic aperture with a radius of 3 pixels show a clear transit signal on the brighter component while the fainter component is constant to within 2%. This excludes the possibility that the transit seen in the WASP photometry is due to a deep eclipse in the lightcurve of the fainter component of the visual binary. We observed 3 further transits of BD −07◦ 436 using TRAPPIST on the nights 2011 November 01 (544 images), 05 November 2011 (1079 images) and 2011 December 01 (925 images). We also obtain a V-band lightcurve on the night 01 November 2011 (327 images) using the EulerCAM instrument on the Euler 1.2-m telescope (Lendl et al. 2012). The flux ratio of the binary in the V-band measured from these images is 5.00 ± 0.13. We used 8 spectra of WASP-77 A obtained with the HARPS spectrograph on the ESO 3.6m telescope (ESO programme ID 088.C-0011) to confirm that the radial velocity signal seen in our – 4 – CORALIE spectra originates from this star and not the fainter component. The entrance aperture to HARPS has a diameter of 1′′ and the spectra were all obtained in good seeing so there is negligible contamination of these spectra by light from the fainter component. Radial velocities measured from these spectra using the same method as for our CORALIE spectra are reported in Table 1. Also reported in this table are 4 radial velocities for WASP-77 B measured in the same way. The radial velocity of WASP-77 B is constant to within 10 m s−1 . The radial velocities of both stars and the bisector span measurements for WASP-77 A are shown as a function of the transit phase in Fig. 2. We obtained a series of 15,000 images of WASP-77 with an exposure time of 40 milliseconds using the 3-channel photometer ULTRACAM (Dhillon et al. 2007) mounted on the 4.2-m William Herschel Telescope on the night 2012 September 10. The pixel scale is approximately 0.30 arc- seconds per pixel and we used the u’, g’ and r’ filters. We then selected 1% of the images in each channel with the best seeing and combined them into three high-resolution images, one for each channel. We used these images to measure the following magnitude difference between the components of the binary: ∆u’ = 2.961 ± 0.015; ∆g’ = 2.156 ± 0.004; ∆r’ = 1.701 ± 0.007. The separation of the components is 3.3 arcseconds. There are no other stars visible in the small images we obtained so we do not have an accurate astrometric solution for the images that we can use to estimate robust errors on this value. 3. Analysis 3.1. Stellar Parameters Eight individual HARPS spectra of WASP-77 A were co-added to produce a single spectrum with an average S/N of around 80:1. Four co-added HARPS spectra of WASP-77 B yielded a spectrum with an average S/N of 30:1. The standard pipeline reduction products were used in the analysis. The analysis was performed using the methods given in Doyle et al. (2012). The Hα and Hβ lines were used to give an initial estimate of the effective temperature (Teff ). The surface gravity (log g ) was determined from the Ca i lines at 6162A and 6439A (Bruntt et al. 2010b), along with the Na i D lines. Additional Teff and log g diagnostics were performed using the Fe lines. An ionisation balance between Fe i and Fe ii was required, along with a null dependence of the abundance on either equivalent width or excitation potential (Bruntt et al. 2008). This null dependence was also required to determine the microturbulence (ξt ). The parameters obtained from the analysis are listed in Table 2. The elemental abundances were determined from equivalent width measurements of several clean and unblended lines, and additional least squares fitting of lines was performed when required. The quoted error estimates include that given by the uncertainties in Teff , log g , and ξt , as well as the scatter due to measurement and atomic data uncertainties. The pro jected stellar rotation velocity (v sin i⋆ , where i⋆ is the inclination of the star’s rotation axis) was determined by fitting the profiles of several unblended Fe i lines. For WASP-77 A, a – 5 – value for macroturbulence (vmac ) of 1.7 ± 0.3 km s−1 was assumed, based on the calibration by Bruntt et al. (2010a). An instrumental FWHM of 0.04 ± 0.01 A was determined from the resolution of the spectrograph. A best fitting value of v sin i⋆ = 4.0 ± 0.2 km s−1 was obtained for WASP-77 A. For WASP-77 B the macroturbulence was assumed to be zero, since for mid-K stars it is expected to be lower than that of thermal broadening (Gray 2008). A best fitting value of v sin i⋆ = 2.8 ± 0.5 km s−1 was obtained for WASP-77 B. There are emission peaks evident in the Ca ii H+K lines of WASP-77 A. The signal-to-noise of the WASP-77 B spectra is too low to discern any emission peaks. 3.2. Rotation period The WASP lightcurves show a weak, periodic modulation with an amplitude of about 0.3 per cent and a period of about 15 days. This is likely to be a signal of magnetic activity in WASP-77 A caused by star-spots modulating the apparent brightness as the star rotates. WASP-77B may contribute to the variability of the lightcurve on timescales of 10 – 20 d, but it is unlikely to be the source of the 15 day modulation (see Section 3.4). We used the sine-wave fitting method described in Maxted et al. (2011) to refine this estimate of the amplitude and period of the modulation. Vari- ability due to star spots is not expected to be coherent on long timescales as a consequence of the finite lifetime of star-spots and differential rotation in the photosphere so we analysed the two seasons of data separately. We removed the transit signal from the data prior to calculating the pe- riodograms by subtracting a simple transit model from the lightcurve. We calculated periodograms over 4096 uniformly spaced frequencies from 0 to 1.5 cycles/day. The results for the two seasons of data are shown in Fig. 3. The false alarm probability (FAP) levels shown in these figures are calculated using a boot-strap Monte Carlo method also described in Maxted et al. (2011). There is a clear detection of a periodic modulation with a period of 15.09 days in the 2008 data set (FAP=0.006). This is confirmed by a detection at a period of 15.78 days, though with lower sig- nificance, in the 2009 data set (FAP=0.052). We adopt a value of Prot = 15.4 ± 0.4 days for the period in the discussion below. 3.3. Mass and radius of WASP-77 Ab To determine the planetary and orbital parameters, the HARPS radial velocity measurements were combined with the photometry from TRAPPIST and EulerCAM in a simultaneous least- squares fit using the Markov Chain Monte Carlo (MCMC) technique. The details of this process are described in Collier Cameron et al. (2007) and Pollacco et al. (2008). Briefly, the radial velocity data are modelled with a Keplerian orbit and the model of Mandel and Agol (2002) is used to fit the transits in the lightcurves. We used the coefficients from Claret (2000) for the four-coefficient limb-darkening model. We did not include the CORALIE radial velocity data in the least-squares – 6 – fit because it is unclear whether they are affected by contamination from the companion star. The TRAPPIST and EulerCAM lightcurves were generated using a synthetic aperture radius large enough to include the light from both stars so we applied a correction for the dilution of the transit depth due to the light from the companion prior to including them in the MCMC analysis. The baseline of the TRAPPIST and EulerCAM observations is rather short, so we used a measurement of the orbital period from an analysis of the WASP photometry as an additional constraint in the MCMC analysis to the TRAPPIST and EulerCAM data. The parameters of the model are given in Table 3 and the model fits to the lightcurves are shown in Fig. 4. We have assumed that the orbit is circular because the Lucy-Sweeney F-test applied to the results of a least- squares fit for an eccentric orbit (Lucy and Sweeney 1971) shows no evidence for a non-circular orbit (p = 0.18, e = 0.008 ± 0.005). The parameters of the transit model and the Keplerian orbit for the host star provide direct estimates for the density of the star and the surface gravity of the planet. To estimate the mass and radius of the planet we require an additional constraint. In this case we derived the following relation specifically for use with WASP-77 A and appropriate for stars with 0.8 < M /M⊙ < 1.2, −0.8 < [Fe/H] < 0.3 and 5000 K < Teff < 6000 K: log(M /M⊙ ) = 0.0213 + 1.570 log (Teff /5781 K) + 0.037 log (ρ/ρ⊙ ) + 0.097[Fe/H] . This equation is the result of a least-squares fit the parameters of 19 stars in eclipsing binary systems with accurately measured masses and radii.1 The standard deviation of the residuals from the fit is 0.051M⊙ . The standard error estimates for the mass and radius of the star and planet given in Table 3 include this contribution to the error budget. We created two sets of EulerCAM and TRAPPIST lightcurves where the correction for the dilution was increased or decreased by its standard error and performed an MCMC analysis of these data in the same way as above. The change in the system parameters between these two sets of data was used to estimate the additional uncertainty on the system parameters due to the uncertainty on the dilution factor. The additional uncertainty is found to be small, e.g., the change in (Rp /R∗ )2 due to the uncertainty in the dilution factor is 0.00007. This small additional uncertainty is included in the standard errors given for all parameters affected in Table 3. 3.4. Discussion The density of WASP-77 A derived from our MCMC analysis is independent of any assumption about the evolutionary state of the star, but our estimates for the mass and radius for the planet do assume that the stellar mass derived from the empirical relation above is accurate. To test this assumption we also compared the effective temperature and density of WASP-77 A to the stellar models of Girardi et al. (2000). The results are shown in Fig. 5, where it can be seen the mass 1http://www.astro.keele.ac.uk/jkt/debcat/ – 7 – estimated from our MCMC analysis (M∗ = 1.00 ± 0.05M⊙ ) is consistent with these stellar models, although the stellar models suggest a slightly lower mass (≈ 0.89M⊙ ) and also suggest that the star is rather old (∼ 8 Gyr), i.e., slightly evolved. We discuss the reliability of these stellar models further below. The mass and radius of WASP-77 A derived in our MCMC analysis (Table 3) agree well with the mass and radius expected for a main sequence star of the same spectral type (Table 2). The period of the modulation in the WASP lightcurve together with our estimate for the radius of WASP-77 A implies a rotation velocity Vrot = 3.1 ± 0.1 km s−1 if we assume that this signal is due to the rotation of this star. This compares with the spectroscopic estimate for the pro jected rotation velocity Vrot sin i⋆ = 4.0 ± 0.2 km s−1 . The difference between these two values cannot be explained by a mis-alignment between the star’s rotation axis and the orbital axis of WASP-77 Ab, but may be explained by an underestimate for the macroturbulence (vmac ) used in our analysis of the spectrum for WASP-77 A. The calibration of Valenti and Fischer (2005) suggests a value of vmac=3.5km s−1 . This additional line broadening would reduce the value of Vrot sin i⋆ estimated from the spectra sufficiently to make it consistent with the hypothesis that the rotational signal in the WASP lightcurves originates from WASP-77 A. The parameters listed in Table 2 are not affected by this uncertainty in the value of vmac . The amplitude of the rotation signal in WASP lightcurves for K-type stars with rotation periods ∼ 15 days can be as much a few percent (Collier Cameron et al. 2009) so it also possible that the fainter component of the visual binary contributes to the variability of the lightcurve, but it is unlikely that the modulation of the lightcurve is due to the fainter component alone. WASP-77 A and WASP-77 B appear to form a genuine physical binary rather than a visual double star. The position angle (PA) and separation of the stars estimated from the images we obtained in good seeing conditions with EulerCAM and TRAPPIST and our high-resolution UL- TRACAM images are consistent with the values reported in The Washington Visual Double Star Catalog (WDS 02286−0704; Mason et al. 2001). That catalog reports that 7 observations were obtained between 1930 and 1933, during which time the recorded separation varied from 2.9′′ to 3.2′′ . For comparison, the proper motion of WASP-77 A implies a change of position of 7.8′′ between 1930 and 2011. The PA of the binary is ≈ 150◦ whereas the proper motion vector is approximately east-west, so the two components of the binary clearly share a common proper motion. To estimate the distance to WASP-77, we used a least-squares fit to the data from Boya jian et al. (2012) to establish the following simple relation between the angular diameter (θLD ), H-band mag- nitude and effective temperature of G-type dwarf stars. θLD = 3.206 − 0.679 log Teff − 0.2H We then used this relation together with the radius of WASP-77 A from Table 3 and the apparent H-band magnitude corrected for contribution from the fainter component to estimate a distance d = 93 ± 5 pc. We used the apparent V-band magnitude of WASP-77 (10.30 ± 0.05, Høg et al. 2000) together with the V-band flux ratio measured from our EulerCAM images to estimate an apparent V-band magnitude for WASP-77 B of 12.05±0.06, corresponding to an absolute magnitude – 8 – MV = 7.2 ± 0.1, which is a typical value for a mid-K-type dwarf star (Gray 2008). There is a small difference between the systemic velocity of WASP-77 A and the radial velocity of WASP-77 B (≈ 1km s−1 ) but this is consistent with the orbital velocity expected for stars of this mass and separation at the estimated distance of the binary. There is no significant detection of lithium in either WASP-77 A or WASP-77 B. The equivalent width upper limits of 9 mA and 13 mA, correspond to an abundance upper limit of log A(Li) < 0.76 ± 0.08 and log A(Li) < 0.10 ± 0.14 respectively. This implies an age at least ∼0.5 Gyr (Sestito and Randich 2005) for both stars. We can also estimate the ages of WASP-77 A and WASP-77 B based on their rotation periods (gyrochronological age). The values of v sin i⋆ and R⋆ for WASP-77A imply a rotation period Prot = 14.2 ± 1.7 days (assuming i⋆ ≈ 90), which corresponds to a gyrochronological age ≈ 1.0+0.5 −0.3 Gyr using the Barnes (2007) relation. Similarly, an age ≈ 0.4+0.3 −0.2 Gyr is obtained for WASP-77 B from the rotation rate of Prot = 12.3 ± 3.0 days. These age estimates are clearly incompatible with the age for WASP-77 A we have derived using the stellar models of Girardi et al. (2000). In this case, the agreement between the gyrochronological ages of WASP-77 A and WASP-77 B suggest that gyrochronological age estimates for hot Jupiter host stars are reliable, whereas ages estimated from the effective temperature and density from isochrones may not be. We checked that this is not a specific problem with this particular set of stellar models by estimating the age of WASP-77 A using 5 additional sets of stellar models. The results are shown in Table 4 with stellar models noted as follows: Claret (Claret 2005), Y 2 (Demarque et al. 2004), Teramo (Pietrinferni et al. 2004), VRSS (VandenBerg et al. 2006) and DSEP (Dotter et al. 2008). The models and method used are described in more detail in Southworth (2012). It can be seen that stellar models consistently over-estimate the age of WASP-77 A compared to the gyrocronological age. This anomaly may well be related to the poor agreement between the observed radii of some low mass stars and the radii predicted by stellar models, which in-turn is thought to be a related to the rotation and/or magnetic activity in these stars (Kraus et al. 2011; Morales et al. 2010). The mass of the star derived from these stellar models is consistently lower than the value derived from our empirical calibration, although not significantly different. The mass and radius of the planet derived using these stellar models are consistent with the values derived using our empirical calibration. 4. Conclusions The brighter component of the visual binary star BD −07◦ 436, WASP-77 A, shows transits every 1.36 days caused by a hot Jupiter companion, WASP-77 Ab, with a mass ≈ 1.8MJup and a radius ≈ 1.2RJup . The radial velocities of the two components of the binary reported here strengthen the conclusion that the two stars are physically associated. The gyrochronological ages for the two stars agree and suggest an age ∼ 0.5 – 1 Gyr for the stars. If the gyrochronological age for WASP-77 A is correct, then this star has a lower density than predicted by standard stellar – 9 – models. This is a phenomenon seen in other stars of similar mass and is likely to be related to the magnetic activity observed in this star. WASP-South is hosted by the South African Astronomical Observatory and we are grateful for their ongoing support and assistance. Funding for WASP comes from consortium universities and from the UK’s Science and Technology Facilities Council. We thank the ULTRACAM team taking the observations of WASP-77 presented here. TRAPPIST is funded by the Belgian Fund for Scientific Research (Fond National de la Recherche Scientifique, FNRS) under the grant FRFC 2.5.594.09.F, with the participation of the Swiss National Science Fundation (SNF). M. Gillon and E. Jehin are FNRS Research Associates. REFERENCES Ammons, S. M., Robinson, S. E., Strader, J., Laughlin, G., Fischer, D., and Wolf, A. 2006, ApJ, 638, 1004 Asplund, M., Grevesse, N., Sauval, A. J., and Scott, P. 2009, ARA&A, 47, 481 Bakos, G., Noyes, R. W., Kov´acs, G., Stanek, K. Z., Sasselov, D. D., and Domsa, I. 2004, PASP, 116, 266 Barnes, S. A. 2007, ApJ, 669, 1167 Batygin, K., Stevenson, D. J., and Bodenheimer, P. H. 2011, ApJ, 738, 1 Boya jian, T. S. et al. 2012, ApJ, 746, 101 Bruntt, H. et al. 2010a, MNRAS, 405, 1907 Bruntt, H., De Cat, P., and Aerts, C. 2008, A&A, 478, 487 Bruntt, H. et al. 2010b, A&A, 519, A51 Claret, A. 2000, A&A, 363, 1081 Claret, A. 2005, A&A, 440, 647 Collier Cameron, A. et al. 2009, MNRAS, 400, 451 Collier Cameron, A. et al. 2007, MNRAS, 380, 1230 Davis, T. A. and Wheatley, P. J. 2009, MNRAS, 396, 1012 Demarque, P., Woo, J.-H., Kim, Y.-C., and Yi, S. K. 2004, ApJS, 155, 667 Dhillon, V. S. et al. 2007, MNRAS, 378, 825 – 10 – Dotter, A., Chaboyer, B., Jevremovi´c, D., Kostov, V., Baron, E., and Ferguson, J. W. 2008, ApJS, 178, 89 Doyle, A. P., Smalley, B., Maxted, P. F. L., Anderson, D. R., Collier-Cameron, A., Gillon, M., and et al. 2012, MNRAS, submitted. Gillon, M. et al. 2012, A&A, 542, A4 Girardi, L., Bressan, A., Bertelli, G., and Chiosi, C. 2000, A&AS, 141, 371 Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres (Cambridge University Press) Høg, E. et al. 2000, A&A, 355, L27 Jehin, E. et al. 2011, The Messenger, 145, 2 Knutson, H. A., Howard, A. W., and Isaacson, H. 2010, ApJ, 720, 1569 Kraus, A. L., Tucker, R. A., Thompson, M. I., Craine, E. R., and Hillenbrand, L. A. 2011, ApJ, 728, 48 Lendl, M. et al. 2012, A&A, 544, A72 Lucy, L. B. and Sweeney, M. A. 1971, AJ, 76, 544 Mandel, K. and Agol, E. 2002, ApJ, 580, L171 Mason, B. D., Wycoff, G. L., Hartkopf, W. I., Douglass, G. G., and Worley, C. E. 2001, AJ, 122, 3466 Matsumura, S., Peale, S. J., and Rasio, F. A. 2010, ApJ, 725, 1995 Maxted, P. F. L. et al. 2011, PASP, 123, 547 Morales, J. C., Gallardo, J., Ribas, I., Jordi, C., Baraffe, I., and Chabrier, G. 2010, ApJ, 718, 502 Pietrinferni, A., Cassisi, S., Salaris, M., and Castelli, F. 2004, ApJ, 612, 168 Pollacco, D. et al. 2008, MNRAS, 385, 1576 Pollacco, D. L. et al. 2006, PASP, 118, 1407 Queloz, D. et al. 2000, A&A, 354, 99 Sestito, P. and Randich, S. 2005, A&A, 442, 615 Sing, D. K. et al. 2012, arxiv:1208.4982 Southworth, J. 2012, MNRAS, 426, 1291 – 11 – Valenti, J. A. and Fischer, D. A. 2005, ApJS, 159, 141 VandenBerg, D. A., Bergbusch, P. A., and Dowler, P. D. 2006, ApJS, 162, 375 Wilson, D. M. et al. 2008, ApJ, 675, L113 Wood, P. L., Maxted, P. F. L., Smalley, B., and Iro, N. 2011, MNRAS, 412, 2376 Wright, J. T., Marcy, G. W., Howard, A. W., Johnson, J. A., Morton, T. D., and Fischer, D. A. 2012, ApJ, 753, 160 This preprint was prepared with the AAS LATEX macros v5.2. – 12 – Table 1: Radial velocity measurements for WASP-77 A and WASP-77 B. BS σRV RV BJD (km s−1 ) (km s−1 ) (km s−1 ) −2 450 000 WASP-77 A, CORALIE 1.3246 5069.7803 1.3362 5163.6091 5169.6920 1.9673 1.5706 5170.6625 1.9696 5188.6258 5856.7095 1.8655 1.7105 5914.6783 1.3328 5915.6775 5916.6681 1.7087 1.9544 5917.6500 5918.6645 1.5542 0.0049 −0.02828 0.0052 −0.04828 0.0049 −0.03717 0.0053 −0.03203 0.0054 −0.04100 0.0053 −0.03387 0.0043 −0.03349 0.0044 −0.02568 0.0048 −0.01432 0.0047 −0.01638 0.0065 −0.01908 WASP-77 A, HARPS 5832.8615 1.4626 1.5111 5832.9040 1.5233 5832.9110 5889.7458 1.4063 2.0220 5890.5370 1.8656 5890.7386 5891.5721 1.7670 1.9885 5891.7468 WASP-77 B, HARPS 5832.8705 2.7508 2.7521 5832.8885 2.7577 5832.8960 5890.5450 2.7586 0.0020 0.0024 0.0025 0.0036 0.0037 0.0041 0.0044 0.0046 0.0047 0.0061 0.0063 0.0057 0.0365 0.0352 0.0410 0.0307 0.0375 0.0180 0.0260 0.0432 0.0168 0.0373 0.0172 0.0203 – 13 – Table 2: Stellar parameters of WASP-77 A and WASP-77 B. Abundances are relative to the solar values obtained by (Asplund et al. 2009). Parameter Teff [K] log g ξt [km s−1 ] v sin i⋆ [km s−1 ] [Fe/H] [Mg/H] [Ca/H] [Sc/H] [Ti/H] [V/H] [Cr/H] [Mn/H] [Co/H] [Ni/H] [Y/H] log A(Li) Mass [M⊙ ] Radius [R⊙ ] Sp. Typea WASP-77 A WASP-77 B 5500 ± 80 4700 ± 100 4.33 ± 0.08 4.6 ± 0.15 0.8 ± 0.1 2.8 ± 0.5 4.0 ± 0.2 0.00 ± 0.11 −0.12 ± 0.19 0.23 ± 0.04 0.09 ± 0.10 −0.02 ± 0.13 −0.01 ± 0.13 −0.03 ± 0.07 0.14 ± 0.18 0.15 ± 0.21 −0.02 ± 0.10 0.39 ± 0.15 −0.03 ± 0.09 0.00 ± 0.06 0.07 ± 0.22 0.06 ± 0.34 0.14 ± 0.14 0.24 ± 0.27 −0.08 ± 0.10 −0.01 ± 0.08 0.00 ± 0.19 0.04 ± 0.08 < 0.76± 0.08 < 0.10± 0.14 0.71 ± 0.06 1.00 ± 0.07 1.12 ± 0.12 0.69 ± 0.12 K5 G8 a Spectral type estimated from Teff using the table in Gray (2008). – 14 – Table 3: System parameters and 1σ error limits derived from the MCMC analysis. Symbol Parameter Value Units 1.3600309 ± 0.0000020 Orbital period P days 2455870.44977 ± 0.00014 BJD T0 Transit epoch (Rp/R∗ )2 0.01693 ± 0.00017 Planet/star area ratio Transit duration tT 0.09000 ± 0.00035 0.06+0.07 −0.05 0.3219 ± 0.0039 1.6845 ± 0.0004 89.4+0.4 −0.7 1.157+0.016 ρ⊙ −0.020 1.002 ± 0.045 M⊙ Stellar reflex velocity Centre-of-mass velocity Stellar density Stellar mass Orbital inclination Impact parameter b K1 γ i ρ∗ M∗ days km s−1 km s−1 ◦ Stellar radius Orbital semi-ma jor axis Planet mass Planet radius Planet surface gravity Planet density R∗ a Mp Rp log gp ρp 0.955 ± 0.015 R⊙ 0.0240 ± 0.00036 AU 1.76 ± 0.06 MJup a 1.21 ± 0.02 RJup [cgs] 3.441 ± 0.008 1.00 ± 0.03 ρJup aRJup = 71492 km Table 4: Mass, radius and age of WASP-77 A and WASP-77 Ab derived using different sets of stellar models. MA (M⊙ ) RA (R⊙ ) Mb (MJup ) Rb (RJup ) Age (Gyr) Claret 0.948 ± 0.055 0.938 ± 0.019 1.695 ± 0.069 1.186 ± 0.025 6.8+4.7 −2.4 Y 2 0.940 ± 0.048 0.935 ± 0.017 1.686 ± 0.061 1.183 ± 0.022 5.8+2.9 −1.8 Teramo 0.889 ± 0.050 0.918 ± 0.018 1.624 ± 0.064 1.161 ± 0.024 10.4+1.6 −4.3 VRSS 0.893 ± 0.048 0.919 ± 0.017 1.629 ± 0.062 1.163 ± 0.023 9.5+1.6 −4.4 DSEP 0.900 ± 0.008 0.922 ± 0.007 1.637 ± 0.024 1.166 ± 0.011 8.2+0.9 −1.6 – 15 – Fig. 1.— WASP photometry of WASP-77 plotted as a function of phase for a period of 1.36003 days. – 16 – Fig. 2.— Upper panel: Radial velocities of WASP-77 A relative to the centre-of-mass velocity measured using HARPS (filled symbols) and CORALIE (open symbols). Also shown are radial velocities of WASP-77 B relative to their weighted mean value (crosses) and the best-fit circular orbit (solid line) for WASP-77 A. Lower panel: Lower panel: bisector span measurements for WASP-77 A (symbols as in upper panel). – 17 – Fig. 3.— Left panels Periodograms for the WASP data from two different observing seasons for WASP-77. Horizontal lines indicate false alarm probability levels FAP=0.1,0.01,0.001. The year of observation is noted in the title to each panel. Right panels Lightcurves folded on the best period as noted in the title. – 18 – Fig. 4.— Photometry of WASP-77 A corrected for the dilution due to WASP-77 B. Data from TRAPPIST are plotted with filled circles in date order from top-to-botton. EulerCAM data are shown with open circles. The solid line shows the lightcurve model for the parameters in Table 3. – 19 – Fig. 5.— Comparison of the effective temperature and stellar density of WASP-77 A to the stellar models of Girardi et al. (2000). Dashed lines show isochrones for age of 10 Myr and 10 Gyr. Solid lines are evolutionary tracks for stellar masses as indicated.
1905.11301	1	1905	2019-05-27T15:38:21	Hazes and clouds in a singular triple vortex in Saturn's atmosphere from HST/WFC3 multispectral imaging	[ "astro-ph.EP", "physics.ao-ph" ]	In this paper we present a study of the vertical haze and cloud structure over a triple vortex in Saturn's atmosphere in the planetographic latitude range 55N-69N (del Rio- Gaztelurrutia et al. , 2018) using HST/WFC3 multispectral imaging. The observations were taken during 29-30 June and 1 July 2015 at ten different filters covering spectral range from the 225 nm to 937 nm, including the deep methane band at 889 nm. Absolute reflectivity measurements of this region at all wavelengths and under a number of illumination and observation geometries are fitted with the values produced by a radiative transfer model. Most of the reflectivity variations in this wavelength range can be attributed to changes in the tropospheric haze. The anticyclones are optically thicker ($\tau \sim$ 25 vs $\sim$ 10), more vertically extended ($\sim$ 3 gas scale heights vs $\sim$ 2) and their bases are located deeper in the atmosphere (550 mbar vs 500 mbar) than the cyclone.	astro-ph.EP	astro-ph	Hazes and Clouds in a Singular Triple Vortex in Saturn's Atmosphere from HST/WFC3 multispectral imaging , c J.F. Sanz-Requenaa,b, S. Pérez-Hoyosc, A. Sánchez-Lavega c, T. del Rio-Gaztelurrutia Patrick G.J. Irwind a) Departamento de Ciencias experimentales. Universidad Europea Miguel de Cervantes, Valladolid, Spain. b) Departamento de Física Teórica, Atómica y Optica. Universidad de Valladolid, Spain c) Departamento de Física Aplicada I, Escuela Técnica Superior de Ingeniería, Universidad del País Vasco, Bilbao, Spain d) Atmospheric, Oceanic and Planetary Physics, University of Oxford, Clarendon Laboratory, Parks Road, Oxford OX1 3PU, UK Abstract In this paper we present a study of the vertical haze and cloud structure over a triple vortex in Saturn's atmosphere in the planetographic latitude range 55ºN-69ºN (del Rio- Gaztelurrutia et al. , 2018) using HST/WFC3 multispectral imaging. The observations were taken during 29-30 June and 1 July 2015 at ten different filters covering spectral range from the 225 nm to 937 nm, including the deep methane band at 889 nm. Absolute reflectivity measurements of this region at all wavelengths and under a number of illumination and observation geometries are fitted with the values produced by a radiative transfer model. Most of the reflectivity variations in this wavelength range can be attributed to changes in the tropospheric haze. The anticyclones are optically thicker (τ ~ 25 vs ~ 10), more vertically extended (~ 3 gas scale heights vs ~ 2) and their bases are located deeper in the atmosphere (550 mbar vs 500 mbar) than the cyclone. Keywords: Saturn, Saturn Atmospheres, Radiative Transfer, Hubble Space Telescope Observations. Highlights  We present the cloud and haze structure in a triple vortex in Saturn´s atmosphere.  Most of the reflectivity changes are related to the properties of the tropospheric haze.  In the anticyclonic region we find a higher particle number density than in the cyclonic region  The aerosol optical thickness and vertical extent agree with upwelling in the anticyclones. 1. Introduction The zonal wind profile of Saturn's upper clouds is approximately symmetrical, with a strong prograde equatorial jet and four other eastward jets in the northern and southern hemispheres (Sánchez-Lavega et al., 2000; García-Melendo et al., 2011). The jet at 65ºN planetographic latitude (PG) (in this paper, all latitudes are given in planetographic units, except stated otherwise) has a singular structure, with a double peak (del Rio-Gaztelurrutia et al., 2018, del Genio et al., 2009) that marks two different dynamical regions that are very close in latitude. Both have a similarly high eastward velocity and the ambient vorticity facilitates the coupling of opposite voriticity ovals located to the north and south of the velocity local minimum, as shown in Figure 1 (del Rio-Gaztelurrutia et al., 2018). This double jet seems to be permanent having been observed since Voyager times (Sánchez-Lavega et al., 2000; García-Melendo et al., 2011). Figure 1: Zonal wind profile for the region of interest (del Río-Gaztelurrutia et al., 2018). The blue points indicate the location of the anticyclones and the green point indicates the location of the cyclone. In fact, at the latitude of this double peak, a system of three vortices, a cyclone and two anticyclones can be tracked in Cassini ISS images since the beginning of 2012 (del Rio- Gaztelurrutia et al., 2018), confirming that vortices in Saturn can be long-lived (Trammel et al., 2016; del Rio-Gaztelurrutia et al., 2010). We shall refer to the triple vortex system as Anticyclone-Cyclone-Anticyclone abbreviated as the ACA system. In Saturn, the detection of vortices using ground-based telescopes used to be complicated (del Rio-Gaztelurrutia et al., 2018), and most of our knowledge of these systems comes from space-based observations. Vortices were first detected during the Voyager flybys in 1980 -- 81 (Smith et al., 1981, 1982; Ingersoll et al., 1984; García- Melendo et al., 2007), and then by the Hubble Space Telescope (HST) (Sánchez-Lavega et al., 2004) and the Cassini spacecraft (Vasavada et al., 2006; Trammel et al, 2016; Ingersoll et al., 2018; Sayanagi et al., 2019; Sánchez-Lavega et al., 2019). In more recent times, the improvement of observation techniques has allowed the observation of vortices from Earth even with small sized telescopes and in May 2015, amateur observers detected a disturbance that started at the location of the triple vortex system, which had been previously observed in their images as a dark spot. The perturbation evolved fast, extending rapidly in longitude. The orbits of the Cassini spacecraft at the time were not favorable for the observation of the region, and so we were granted Director Discretionary Time at the Hubble Space Telescope (HST) to observe the region before the perturbation faded away (del Rio-Gaztelurrutia et al., 2018). More recently, in 2018, a convective outbreak occurred in the cyclonic side of the poleward jet disturbing the latitude band from  65°N to 76°N (Sánchez-Lavega et al., 2019). The study of the existence of long-lived vortices and their evolution is an excellent way to increase our understanding of the atmospheric conditions below the observable upper clouds (García-Melendo et al., 2007; del Río-Gaztelurrutia et al., 2010). An essential factor in understanding the atmospheric dynamics of vortices is the knowledge of the vertical distribution of the haze and clouds used as tracers, and to achieve this knowledge we need to determine also the physical and optical properties of the haze particles and clouds in Saturn's stratosphere and upper troposphere (Sanz-Requena et al., 2018). Our current understanding of Saturn's clouds and hazes is constrained by several decades of remote sensing data (e.g., Pérez-Hoyos et al., 2005; Karkoschka 2005; West et al., 2009) and a usual model is to consider a three-layered aerosol structure formed by a thin stratospheric haze and a denser tropospheric haze, both above a thick cloud layer (Roman et al., 2013). The goal of this paper is to evaluate the structure of the clouds and hazes and the distribution of aerosols and particles and their properties in the upper troposphere and lower stratosphere in the region of the triple vortex and its environs area, using HST/WFC3 multispectral imaging. The paper is organized as follows: Section 2 is devoted to a short description of the observations used in this work. Section 3 covers the radiative transfer model, including a description of the vertical cloud structure model and its a priori assumptions. Results are presented and discussed in Section 4, including an analysis of the sensitivity to the model parameters. Results are discussed in Section 5 in terms of the local dynamics and a summary of the main conclusions of this work is presented in Section 6. 2. Observations 2.1. Description of the observations In this study, we have used 42 images taken with the Wide Field Camera 3 (WFC3) onboard HST. The images were taken with a variety of filters in three different orbits, on June 29 -- 30 and July 1, 2015. We show in Figure 2 a representative set of these images. Figure 2: Images taken on July 1, 2015 for the ten filters used in this work. Note that quad filters (FQ727N, FQ750N, FQ889N and FQ937N) are binned for a better signal-to-noise ratio. Table 1 summarizes the observations used in this work and some of the geometric parameters that characterize them. All images have been photometrically calibrated (Dressel, 2019), and navigated and cylindrically projected with the LAIA software, developed by J.A. Cano (Grup d'Estudis Astronomics, GEA) (Sanz-Requena et al., 2012 and Pérez-Hoyos et al., 2005). We assumed a 10% error in absolute calibration (Dressel, 2019), without taking into account whether these are random or systematic or even known variations from one filter to the other. It must be noted that systematic errors, particularly those regarding absolute calibration, are usually substantially higher than random errors. The former ones can reach up to 10-20%, while the latter ones can be assumed to be always around 1%. This will be of interest later on, when the fitting algorithm is described in section 3.3. The filters used in this work are F225W, F336W, F410M, F502N, F547M, F689M, FQ727N, FQ750N, FQ889N, FQ937N (where the 3-digit numbers in the filter name refer to each filter's effective wavelength) and the pixel scale of the images (300 km/pixel without binning, proportionally increased in quad filters for optimizing signal to noise and exposure time) are described in Dressel (2019). The ultraviolet filters are, generally speaking, sensitive to Rayleigh scattering by the atmospheric gas and to the properties of the sub-micron sized particles at the upper atmosphere. On the other hand, narrow filters covering methane absorption (FQ727N -- intermediate, FQ889N -- deep), when used together with near continua (FQ750N, FQ937N) are able to provide an altimetry of the cloud tops. A first estimation of the relative altitudes of the triple vortex, with the anticyclones bright at methane bands and dark at short wavelengths, and the opposite for the cyclone, provides a crude picture of the two anticyclones located higher than the cyclone. A similar behavior is observed in images taken with the Imaging Science Subsystem (ISS) onboard the Cassini spacecraft (del Rio-Gaztelurrutia et al., 2010; Vasavada et al., 2006) where cyclones appear bright in the BL1 filter and dark in the MT2 and MT3 filters. However, in order to retrieve a more detailed description of the vertical structure, we need to perform a detailed radiative transfer analysis of the data. Date 2015/06/29 B 28.74 B' α 29.62 3.65 Filters F225W, F336W, F410M, F502N, F547M, F689M, FQ727N, FQ750N, FQ889N, FQ937N 2015/06/30 28.63 2015/07/01 28.73 29.62 3.72 FF225W, F336W, F410M, FF502N, F547M, F689M, FFQ727N, FQ750N, FQ889N, FFQ937N F225W, F336W, F410M, F502N, F547M, F689M, FQ727N, FQ750N, FQ889N, FQ937N 29.63 3.79 Table 1: Observations: B (sub-earth planetocentric latitude); B' (sub-solar planetocentric latitude; α (phase angle). 2.2. Data selection. From all the available data, we have selected a range of latitudes, from 55ºN to 69ºN where we can observe the atmospheric feature of interest and the structure of bands and zones in the surrounding background atmosphere. Since we have images taken on different days, there is a longitude drift in the positions of individual features (such as the vortices) following the zonal wind profile (García- Melendo et al., 2011). We have considered such a drift, and studied the longitude box surrounding the triple vortex that is visible at the three visits, at least in one image for each of them. When more than one observation was available, the values of reflectivity and geometry were averaged, since the differences in observing conditions were small for each visit. This allowed us to obtain three spectra for every point of the region of interest, each one at a different viewing and illumination conditions (Figure 3). Figure 3: Maps for the values of the cosine of the emission (µ) and emission (µ0 ) angles for the different dates of the images used in this work. a) June-29, b) June-30 c) July-1. The location of the triple vortex is indicated on each map with red stars. In Figure 4 we show cylindrical projections of the region of interest in every filter. These images have been corrected for limb-darkening only for display purposes, as the limb-darkening information will be used in the following sections to constrain atmospheric properties. These images are averages from one or more original images, depending on the latitude and longitude coverage of the HST observations for each case. Figure 4: Cylindrical maps at the available wavelengths showing the region under study with A- C-A system included corrected for limb-darkening. The position of the triple vortex is centered at approximately 85ºE and is most apparent in filters F336W, FQ889N and FQ937N. 3. Methodology 3.1 Radiative Transfer code Our goal is to reproduce the observed dependence of absolute reflectivity with geometry (three combinations of incidence and emission angles) for all wavelengths at the same time, so we can deduce the values of different parameters that give us information about the atmosphere such as the optical thicknesses of aerosol layers, the mean size of the particles, the height at which they are found the different layers and so on. To do this we used the radiative transfer code and retrieval suite NEMESIS, developed by Irwin et al. (2008). This code uses the optimal estimator scheme to find the most likely model that best accounts for the observations. The version of the code used here is based in a doubling-adding scheme that assumes a plane -- parallel atmosphere to compute the emergent intensity of reflected sunlight due to scattering and absorption from atmospheric aerosols and gases. In our model we also take into account the Rayleigh scattering due to the mixture of H2 and He, as well as the absorption due to CH4. The general assumptions (temperature-pressure profile and gaseous abundances) used in this work are the same as in Sanz-Requena et al. (2018). 3.2 Vertical cloud structure model Previous works (Pérez-Hoyos et el. 2005, Sanz-Requena et al. 2018) have found that a vertical structure consisting of three distinct layers of particles is good enough to reproduce the spectral and geometric variations of the absolute reflectivity at visible wavelengths. The overall vertical distribution of particles assumed in the present work is similar to that of Sanz-Requena et al., 2018, as shown in Figure 5. In Table 2, we summarize the list of free and fixed parameters, which have been chosen according to previous works (Pérez-Hoyos et el. 2016, Sanz-Requena et al. 2018). The same is true for the description of the gaseous scattering (by a mixture of H2 and He, with a volume mixing ratio of 0.124 relative to H2, (de Pater and Lissauer, 2001) and absorption. We only considered absorption by CH4, using pre-computed k-tables based on the absorption coefficients given by Karkoschka and Tomasko (2010). Figure 5: A priori assumed particle density profiles in our three assumed cloud/haze layers (solid lines) and their corresponding uncertainties (dotted lines). Parameters of the model for each layer are also indicated (see the text for a full explanation). Adapted from Sanz-Requena et al., 2018. The uppermost aerosol layer corresponds to the stratospheric haze that is located between P1 = 1 mbar and P2 = 100 mbar (Pérez-Hoyos et al., 2005). In this layer, we assume a constant refractive index with real part and imaginary parts for all wavelengths, which we set to the average of that for ammonia ice (mr = 1.43 and mi = 10-3; Pérez-Hoyos et al., 2016). We also set the effective radius and the effective variance to be 0.1 µm and 0.1, respectively, and assume that the particle size distribution follows a log-normal distribution (Hansen and Travis, 1974). The only free parameter in this layer is the optical thickness, for which we have set a starting point τstr= 0.01±0.01 (Sanz-Requena et al., 2018), at 900 nm, which will be used as the reference wavelength in the following analysis, except where stated otherwise. The second layer, corresponding to the tropospheric haze, is characterized by a variable optical thickness (τtrop = 10 ± 2) (Karkoschka and Tomasko, 2005), as well as a parameterization of its vertical distribution. This is defined by the pressure corresponding to the lower base (600 ± 100 mbar; Fletcher et al., 2007, Roman et al., 2013). The particle-to-gas scale height ratio of this aerosol layer is taken initially as Haerosol/Hg = 0.7 ±0.1 (Pérez-Hoyos et al., 2016) and the value of the initial maximum concentration of particles is N = 20 ± 10 particles/cm3 (Sanz-Requena et al., 2018). Here Hg is the atmospheric (gas) scale height  38 km. We assume that the particles are spherical and we use Mie theory to compute the phase function. Since all the observations are made at a similarly low phase angle value (~ 3.6º), this assumption is not critical. We have taken the initial values of reff = 1.5 ± 0.5 and σeff = 0.1 ± 0.1. (Ortiz et al., 1996). Our model calculates the real refractive index from Kramers-Kronig's relation (Lucarini et al., 2005) from an initial value mr = 1.43. The imaginary refractive index is set as a free parameter taking as initial value mi = 10-3 ± 10-3 for all wavelengths (Roman et al., 2013). The lower layer is fixed between pressures P5 = 1.0 bar and P6 = 1.4 bar and corresponds to the cloud putatively formed by ammonia ice (Roman et al., 2013). It must be noted that such ammonia ice has been very rarely spectroscopically identified, with a few exceptions (Baines et al., 2009; Sromovsky et al., 2015). The refractive indices are fixed (mr = 1.43 and mi = 10-3) as in the stratospheric haze, and the optical thickness is the only free parameter (with a priori values of τcloud = 10 ± 2 from Pérez- Hoyos et al., 2016). As in stratospheric haze, the particle size distribution is log-normal with an effective radius of 10 μm and an effective variance of 0.1 μm (West et al., 2009). Layer StratosphericHaze Tropospheric Haze Bottom Cloud Parameter P1 P2 τstr mr mi reff σeff Pbot N H τtrop reff σeff mr mi(225 µm) mi(336 µm) mi(410 µm) mi(502 µm) mi(547 µm) mi(689 µm) mi(727 µm) mi(750 µm) mi(889 µm) mi(937 µm) P5 P6 τcloud mr mi reff σeff Type Fixed Fixed Free Fixed Fixed Fixed Fixed Free Free Free Computed Free Free Fixed Free Free Free Free Free Free Free Free Free Free Fixed Fixed Free Fixed Fixed Fixed Fixed Value 1mbar 100 mbar 0.01±0.01 Amonnia ice Amonnia ice 0.1 µm 0.1 µm 600±100 mbar 20 ±10 particles/cm3 25±5 km 10±5 1.5±0.5 µm 0.1 ±0.1µm 1.43 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 10-3±10-3 1.0 bar 1.4 bar 10±5 1.43 10-3 10µm 0.1 µm Table 2: Model atmosphere parameters. Please note that τtrop is in fact computed from the other parameters describing the vertical distribution. 3.3. Fitting strategy To estimate the goodness of fit between the observed and modelled reflectivities, we evaluated the error function 𝜒2/𝑛 at every point of the free-parameter space. The error function is defined at each filter observation as: 2 𝜒𝜆 𝑛 = 1 𝑛 𝑛 𝑖=1 1 2 𝜎𝑖 𝐼 𝐹 𝑜𝑏𝑠 − 𝐼 𝐹 𝑚𝑜𝑑 2 where n is the number of points to be fitted by the model (e.g., the number of points scanning over longitude at a given latitude); σi is the error in the ith measurement; and (I/F)obs is the observed and (I/F)mod the modeled reflectivity at a given point. Then we calculate a reduced 𝜒2/𝑛, an average of all the filters and positions over the disk being modeled. When 𝜒2/𝑛 is larger than one, the profiles deviate systematically from the data. When 𝜒2/𝑛 is smaller than 1, we can accept the model as it is, on average, inside the data error bars. Our goal is to find models close enough to the data for all geometries and wavelengths (𝜒2/𝑛 < 1) at every point of the region of interest. This approach does not take into account any differences between systematic and random errors and only minimizes the overall model deviation from the data. As low pixel-to- pixel noise allows determining limb-darkening more precisely than absolute reflectivity values (mostly affected by systematic errors), we will discuss in section 4.1 how well best-fitting models are able to match the observed limb-darkening. We first fitted center-to-limb brightness profiles for the range of selected latitudes 55º to 69ºN with intervals of 0.5º, excluding the regions where the triple vortex is present. In this way, we obtain a reference model that we use later, including the regions where the triple vortex is located. We initially fitted filters FQ727N, FQ750N, FQ889N and FQ937N with a constant value of mi = 10-3. Once the result was acceptable we added the shortest wavelengths (filters F225W, F336W, F410M, F502N, F547M and F689M) leaving also the refractive indices as free parameters. This procedure avoids overfitting with the imaginary refractive index (whose values are forced to be nearly flat for longer wavelengths), provides good results and it is very similar to the strategy in Sanz- Requena et al. (2018). The result of this analysis is presented in Fig. 6, where we show the ten filters covering a wide geometrical range. Figure 6: Best-fitting models for center-to-limb variations of reflectivity. Blue circles represent the modeled values and the red circles and lines correspond to the observed reflectivity and its corresponding error bar. (a) latitude 56ºN , (b) latitude 61ºN , and (c) latitude 69ºN . In this figure, longitudes are measured in degrees from an arbitrary reference longitude. Once we have a good model for the overall limb-darkening behavior of the reference atmosphere, we want to fit individually all the points in the region of interest, including particular features such as the vortices. For doing so we run new retrievals using as input the reference atmosphere model at a given latitude. In figure 7 we show best- fitting model results for each of the three vortices at some locations of interest. Figure 7: Examples of best-fitting results. Blue circles indicate the modeled values and the red circles and error bars correspond to the observed reflectivities and their corresponding error bars. Panel a) show longitude 60º- latitude 68º, panel b) show longitude 66º- latitude 64º (anticyclone), panel c) show longitude 81º - latitude 63º (cyclone), panel d) show longitude 50º - latitude 58º and panel e) show longitude 60º - latitude 64º. 4. Modelling Results 4.1. General results and limb-darkening We have assumed error bars for the data that do not separate random from systematic errors. In order to rule out and quantify a possible systematic deviation of the limb- darkening in the different regions and filters and in order to improve our model we fitted both observations and best- fitting models to a Minnaert law. 𝐼 𝐹 = 𝐼 𝐹 0 𝑘µ𝑘−1 µ0 Where (I/F)0 is the reference reflectivity value at perfect nadir geometry (µ=1 and µ0=1) and k is a wavelength-dependent limb-darkening coefficient (Sanz-Requena et al., 2018). This analysis corresponds to the background models where center-to-limb data is available, and not to the specific cyclonic features. While previous figure 7 explicitly shows the deviation between models and data, we show in Figure 8 the results corresponding to the variation of the values of k for the different latitudes and for all wavelengths. As previously stated, the limb-darkening can be more precisely determined from the original data, as it is mostly affected by pixel-to-pixel and random errors, which are substantially lower than systematic calibration uncertainties. Although it would be desirable to be able to fit independently the limb-darkening with stronger constraints, our current version of the retrieval code does not have this capability. We want to show here that, at least, our best-fitting models are close enough to the measured limb-darkening values. Figure 8: Values of the limb darkening coefficient k for all latitudes (55ºN to 69ºN and filters (F225W, F336W, F410M, F502N, F547M, F689M, FQ727N, FQ750N, FQ889N, FQ937N). Blue circles correspond to the best fits and the red circles and lines correspond to the HST data and their error bar. Error bars for the observed limb-darkening is taken as a 5% of the value of the HST data. This includes not only pixel-to-pixel random noise but also navigation uncertainties. There is no systematic deviation in the limb-darkening of the models from that of the data, as the same filter or region can give differences as low as 0.15 % or as high as 7%, with average differences of 3%. Taking into account not only the errors in relative photometry but also the navigation uncertainties (including the longitudinal drift due to zonal winds) the level of discrepancy between observations and models is acceptable and indicates that our model atmosphere reproduces well the observed limb-darkening. 4.2. Best-fitting results In figure 9 we show the goodness of the fit (𝜒2/𝑛 ) for all the points of the region of study. We find that at all points 𝜒2/𝑛 <1. However, there is a longitude trend in the goodness of fits, which implies that the overall limb-darkening behaviour is not perfectly constrained when we study each point of the grid separately. This implies that the state derived from the limb-darkening analysis is not sufficiently well constrained and the optimal results retrieved from a point-by-point basis do not fully reproduce the observed limb-darkening, as we will discuss later. Figura 9: Reduced value of 𝜒2/𝑛 for all the points of the grid. All the fittings are satisfactory although limb-darkening seems not to be perfect in our models, as there is a longitude trend in the goodness of fit. Figures 10 to 14 show the best-fitting parameters as a function of latitude and longitude. The optical thickness (at 0.9 μm) of the stratospheric haze (Figure 10a) shows no zonal variation and depends mostly on latitude. We find an increase with latitude that varies from τstr = 0.01 ± 0.01 (~55º N ) to τstr = 0.025± 0.01 (~69ºN ). This implies an increase of a factor 3 in the stratospheric particle density from the lower to the upper latitudes, similar to that presented in Sanz-Requena et al. (2018). It is important to notice that the models do not require differences in the thickness of the upper haze between the anticyclone-cyclone system and the surrounding regions (τstr ~ 0.015± 0.01). In Figure 10b we show the optical thickness down to the bottom cloud at the ammonia condensation levels. In this case we find that the values are quite homogeneous, with hardly any spatial variation ( τcloud ~ 9.2 ±2 to ~55ºN and τcloud ~ 8.4 ±2 to ~ 69ºN). The pressure level for the base of the tropospheric haze (Figure 10c) varies from 700 ± 100 mbar for ~ 55ºN to 300 ± 100 mbar for ~ 69ºN. The values are very homogeneous in longitude. However, there are small differences in the region of the triple vortex. The value in the anticyclones is approximately 550 ± 100 mbar while in the cyclone region the average value is 500 ± 100 mbar. This implies that in terms of altitude above the 1 bar level, the tropospheric haze is located at 40 ± 5 km and 50 ± 5km for the anticyclones and the cyclone respectively. On the other hand, as we approach higher latitudes the height of the base of the tropospheric haze increases from 20 ± 5 km at 55ºN to 60 ± 5km at 70ºN . The maximum particle concentration (Figure 10d) follows a latitudinal behavior similar to that of the base pressure of the tropospheric haze. Its value ranges from 100 ± 10 particles /cm3 for 55ºN to 50 ± 10 particles/cm3 for 70ºN. Again, the behavior is quite homogeneous in longitude, with small differences at the anticyclones (80 ± 10 particles/cm3) and the cyclone (55 ± 10 particles/cm3), while in the cyclone we can see a decrease in the maximum average concentration. Figure 10: a) Optical thickness of the stratospheric haze. b) Optical thickness down to the bottom cloud. c) Pressure (mbar) of the base of the tropospheric haze. d) Height (km) of the base of the tropospheric haze. e) Maximum particle concentration (particles/cm3). The location of the triple vortex is indicated on each map. (Anticyclone (A), Cyclone (C), Anticyclone (A)) In Figure 11a we show the variation of particle density with height for 6 different regions. We observe that for low latitudes (58ºN ) maximum concentrations (~ 110 ± 10 particles/cm3) are located at pressures of ~ 900 ± 100 mbar. The maximum concentrations in the two anticyclonic regions (64ºN and 65 ºN) have similar values (~85 ± 10 particles / cm3) and both are at the same pressure level (~550 ± 100 mbar). In the cyclonic region (63ºN) we observe a smaller peak concentration (~ 55 ± 10 particles / cm3) at a lower pressure (~ 500 ± 100 mbar). Outside the triple vortex, we find that at the same latitudes the maximum concentration (~ 70 ± 10 particles /cm3 and ~600 ± 100 mbar is smaller than that of the anticyclones and bigger than that of the cyclone. The value of the maximum concentration at higher latitudes (68ºN) is ~ 60 ± 10 particles /cm3 and is located at a pressure similar to that of the cyclone. Figure 11: a) Vertical distribution of the tropospheric particles for the vortices and a reference region. b) Location of vortices and reference regions Figure 12a shows the fractional scale height of the tropospheric haze, Haerosol/Hgas. We observe a decrease of this parameter with latitude, 0.65 ± 0.1 (Haerosol ~ 23± 2 km) for ~ 55ºN and 0.45 ± 0.1 (Haerosol ~16± 2 km) for ~69ºN . In the latitudes where the triple vortex is found, we do not observe substantial differences between the values corresponding to the anticyclones and the cyclone (0.55 ± 0.1) (Haerosol ~ 18± 2 km). The optical thickness of the tropospheric haze is also quite homogeneous in longitude except at the triple vortex (Figure 12b). In latitude we observe an increase of optical thickness from τtrop ~ 28± 2 at 55ºN to a maximum τtrop ~ 35± 2 at 61º N , and then the magnitude decreases down to τtrop ~ 10± 2 at 69ºN . This is consistent with the belt and zone structure of the region. The values of the optical thickness in the anticyclones is τtrop ~ 25± 2, similar to the average at their latitude, while the optical thickness of the cyclone it is ~ 10± 2. As we will see, this parameter accounts for most of the spectral and geometrical variation in this data set. We have calculated the vertical thickness of the tropospheric haze (Figure 12c) from the height corresponding to the optical thickness equal to 1 down to the base level, as a proxy to the vertical extension of the haze. We observe that the tropospheric haze thickness decreases polewards from 80 ±5 km at ~55º N to 40 ±5 km at ~69ºN. Again, the behavior in longitude is quite homogeneous. However, we do find differences between the anticyclonic region and the cyclonic region with thicknesses of 60±5 km and 50±5 km respectively. Regarding the particles size, we do not find significant differences for the range of latitudes that we are considering, being ~ 0.15 ± 0.1 μm. (Figure 12d), with a subtle belt/zone structure that is inside the parameter error bars and thus not statistically significant. Figure 12: a) Scale height of the tropospheric haze, (Haerosol/Hgas). b) Optical thickness of the tropospheric haze. c) Vertical thickness of the tropospheric haze (km). d) Particle size (μm). The location of the triple vortex is indicated on each map. (Anticyclone (A), Cyclone (C), Anticyclone (A)) Figure 13 shows the imaginary refractive indices of the tropospheric haze for six different wavelengths. We observe that the values are quite homogeneous in longitude. We have omited the values for wavelengths 725 nm, 750 nm, 889 nm and 937 nm since they are practically constant. We do not appreciate significant differences at the locations of the anticyclones and the cyclone at any wavelength. The relative errors of all parameters retrievals are displayed in Figure 14. Figure 13: Imaginary refractive indexes for the tropospheric haze retrieved for the first six filters wavelengths. The location of the triple vortex is indicated on each map. (Anticyclone (A), Cyclone (C), Anticyclone (A)) Figure 14: Relative errors. a) Optical thickness of the stratospheric haze. b) Optical thickness down to the bottom cloud. c) Optical thickness of the tropospheric haze. d) Pressure of the base of the tropospheric haze. e) Height of the base of the tropospheric haze. f) Vertical thickness of the tropospheric haze. g) Maximum particle concentration . h) Particle size, i) Scale height of the tropospheric haze, (Haerosol/Hgas). j) Imaginary refractive index (410 nm) In Figure 15, we show the variation of the imaginary refractive indexes mi with wavelength for six selected regions, including the locations of the cyclone and anticyclones. The behavior at all the six regions is similar. At visible and infrared wavelengths, mi decreases with wavelength, from ~ 28 ± 0.1 10-3 (410 nm) to 5 ± 0.1 10-4 (937 nm). In the shortest wavelengths, it increases from ~7 ± 0.1 10-3 (225 nm) to ~16 ± 0.1 10-3 (336 nm) and ~ 28 ± 0.1 10-3 (410 nm). In this range of wavelegths, mi is slightly higher at higher latitudes. Figure 15: The imaginary refractive indexes respect to the wavelength for six selected regions. Table 3 shows the best-fitting values of the parameters for different regions. anticyclone cyclone Region 1 Region 2 Stratospheric Haze τstr(0.9 µm ) TroposphericHaze z(km) Pbot(mbar) N(part/cm3) H(km) τtrop(0.9 µm) Thicknes tropospheric haze(km) reff(µm) σeff(µm) mi(225 µm) mi(336 µm) mi(410 µm) mi(502 µm) mi(547 µm) mi(689 µm) mi(727 µm) mi(750 µm) mi(889 µm) mi(937 µm) Cloud τcloud(0.9 µm ) 0.015±0.01 0.015±0.01 0.01±0.01 0.025±0.01 40±5 550±50 78±10 18±2 25 ±2 70±2 0.14±0.1 0.05±0.01 7±0.1E-03 15±0.1E-03 26±0.1E-03 7±0.1E-03 1.7±0.1E-03 1.3±0.1E-03 5±0.1E-04 5±0.1E-04 5±0.1E-04 5±0.1E-04 9±2 50±5 500±50 55±10 18±2 10 ±2 50±2 0.15±0.1 0.05±0.01 7±0.1E-03 15±0.1E-03 26±0.1E-03 7±0.1E-03 1.7±0.1E-03 1.3±0.1E-03 5±0.1E-04 5±0.1E-04 5±0.1E-04 5±0.1E-04 9.5±2 20±5 750±50 100±10 23±2 28 ±2 78±2 60±5 300±50 50±10 17±2 10 ±2 35±2 0.15±0.1 0.05±0.01 6±0.1E-03 16±0.1E-03 26±0.1E-03 8±0.1E-03 3±0.1E-03 1.3±0.1E-03 5±0.1E-04 5±0.1E-04 5±0.1E-04 5±0.1E-04 9.5±2 0.15±0.1 0.05±0.01 12±0.1E-03 18±0.1E-03 32±0.1E-03 9±0.1E-03 1.6±0.1E-03 1.7±0.1E-03 5±0.1E-04 5±0.1E-04 5±0.1E-04 5±0.1E-04 8.8±2 Table 3: Region 1 corresponds to longitude 80º - latitude 60ºN and region 2 corresponds to longitude 80º- latitude 68ºN . According to our results, the behavior of the parameters is quite homogeneous in longitude both in the stratospheric haze and at cloud level. This same behavior is found in the tropospheric haze, except in the latitudes where the triple vortex appears. 4.3. Sensitivity to the Model Parameters. It is possible to evaluate the information gain during the retrieval process by comparing the relative errors between the a priori assumption and the a posteriori best-fitting value. For doing so, we evaluate the improvement factor as defined by Irwin et al. (2015). A low improvement factor indicates that the a posteriori result of the free parameter is not giving us substantial information regarding the a priori uncertainty, while a high improvement factor indicates that we have significantly reduced the a priori uncertainty during the retrieval. In table 4 we show the results of the improvement factors for the different free parameters. improvement Stratospheric Haze τstr(0.9 µm ) TroposphericHaze Pbot(mbar) N(part/cm3) Haerosol/Hgas reff(µm) σeff(µm) mi(lambda) Cloud τcloud(0.9 µm ) factor 2% 92% 95% 15% 12% 3% 10% 5% Table 4: Improvement factor of free parameters. According to these results, we can assess that the base altitude and peak concentration, and the scale height to a lesser extent are the most important parameters, and the retrieval more informative about their values. To address the importance of these most informative free parameters in both the nadir- viewing reflectivity and the limb-darkening, we made a new Minnaert fit to models computed at nominal values as well as 1-σ above and below the nominal result. In figure 16 we show how the average value of the values of I/F0 and of k for the different wavelengths varies as a given parameter is changed. From these results we observe that the most affected filters are F336W, FQ750N, FQ889N and FQ937N. Figure 16: Sensitivity analysis based on changing a given parameter of the tropospheric haze parameterization by 1-σ above (red line) and below (blue line) of the nominal value for the following parameters: a) altitude of the haze base; b) particle peak concentration. The black lines are the values of the best fit data. 5. Discussion. We have found that in the region where ACA system is located, the optical thickness of the stratospheric and tropospheric hazes depends to a large extent on the latitude. This variation is not so pronounced in the lower cloud. As a general rule, we observe that the pressure of the base of the tropospheric haze decreases northwards from north 700 ± 100 mbar to 300 ± 100 mbar. According to our modelling there are no significant differences between the anticyclones (A) and cyclones (C) either in the stratospheric haze (at pressures levels above tropopause  60 -- 100 mbar) or in the lower cloud (at 1-1.4 bar). If we compare the reflectivity of the three vortices with the surrounding regions in the range of selected latitudes, we find that in the FQ889N and FQ937N filters the cyclone has low brightness relative to the surroundings, but turns bright in F336W, consistent with low particle density in the tropospheric haze and deeper clouds. A similar behavior has been found in other cyclones (Sánchez-Lavega et al., 2006; Baines et al., 2009). This is in agreement with our result since we found a minimum of optical thickness in the tropospheric haze (~ 10± 2) and a lower concentration of particles (~ 55 ± 10 particles / cm3). On the other hand, anticyclones are brighter than their surroundings in the FQ889N and FQ937N filters and darker in the F336N filter. This situation is associated to two regions with greater optical thickness (~ 25± 2) and a higher concentration of particles (~ 80 ± 10 particles / cm3). Similar observations have been made by Roman et al. (2013) in long-lived cyclones located at 51ºS and by del Río- Gaztelurrutia et al. (2010), who investigated the vertical structure as well as the winds and dynamics. The particle properties in both regions (effective radius and imaginary refractive index) are similar within our model sensitivity, indicating that for the ACA, no particular difference in microphysics processes and chromophore agents exit. The main difference in the cloud structure between the A and C occurs in the tropospheric haze and affects only to the particle number density and vertical thickness, higher in A than in C. No difference is found between A and C in the base location of this haze, in both cases at  500 mbar. The excess of tropospheric haze particles in A compared to C can be due to a higher ammonia ice condensation in the anticyclones, due to differences in temperature or in the vapor abundance at this level. For Jupiter's anticyclones, dynamical modelling proposes the existence of a cold core above the main cloud deck (Marcus et al., 2012) probably favoring haze formation. This could be also the situation for Saturn anticyclones. On the other hand, the fact that particle density is lower in C than in A could also be due to vertical motions, with subsidence in C at the tropospheric haze, also suggested in a previous work (del Rio-Gaztelurrutia et al., 2010). Note that this is at odds with the behavior that dominates Earth's vortices: subsidence and ample cloud free areas in anticyclones, and cloudy extratropical and tropical cyclones where cloud formation by baroclinic frontal systems and massive moist convection occurs. 6. Conclusions. We report a photometric analysis and radiative transfer modelling of a triple vortex (ACA system) in Saturn's atmosphere using images taken with the HST/WFC3 at a spectral range from 225 nm to 937 nm, including an intermediate and a deep methane absorption band. We retrieve the vertical distribution and properties of the upper cloud and hazes at a region which includes this ACA system, covering a range of latitudes from 55ºN to 69ºN. Below we list the most important conclusions.  Most atmospheric parameters seem to be zonally homogenous except in the region of the ACA system, where a few of them have significant variations.  These variations correspond to characteristic parameters of the tropospheric haze, in particular, the particle number density and the base height.  The optical thickness of the stratospheric haze is quite homogeneous in longitude while increasing with latitude. The optical thickness of the cloud is nearly constant  The optical thickness of the tropospheric haze increases from τtrop ~ 28±2 at 55ºN up to τtrop ~35± 2 at 61ºN , and then decreases down to τtrop ~10±2 at 69ºN . The greatest variability is found in the range of latitudes of the ACA system. At the anticyclones τtrop ~20±2, while in the cyclone τtrop ~ 15±2.  Both the anticyclones and the cyclone display a base pressure of the tropospheric haze (~550±50 mbar with a greater thickness ~70±2 km = 3H and ~500 ± 50 mbar with a thickness of 50±2 km = 2H), lower than the base pressures of the regions at lower latitudes (750±50 mbar with a thickness 78±2 km = 3.25H to 55ºN ) and higher than the base pressures at higher latitudes (300±50 mbar with a thickness 35±2 km = 1.5H ). H is the atmospheric scale height.  The maximum particle number density is higher in the anticyclones (~ 78±10 particles/cm3) than in the cyclonic region (~ 50 ± 10 particles/cm3).  The low values of optical thickness, the concentrations of particles found, as well as the base height of the tropospheric haze in the cyclone suggest that it is a subsidence region.  The vortices show no significative variations in the scale height, particle size or refractive indices of the haze.  The properties of the anticyclones and cyclone are compatible with the general picture of upwelling in the former and downwelling in the latter. Acknowledgments This work was (MINECO/FEDER, UE) and Grupos Gobierno Vasco IT-765-13. the Spanish project AYA2015-65041-P supported by References Baines, K. H., Delitsky, M. L., Momary, T. W., Brown, R. H., Buratti, B. J., Clark, R. N., and Nicholson, P. D., 2009. Storm clouds on Saturn: Lightning-induced chemistry and associated materials consistent with Cassini/VIMS spectra. Planetary and Space Sci. 57, 1650-1658. doi:10:1016/j.pss.2009.06.025 Del Genio, A.D. , Achterberg, R.K. , Baines, K.H. , Flasar, F.M. , Read, P.L. , Sánchez- Lavega, A. , Showman, A.P. , 2009. Saturn atmospheric structure and dynamics. In: Dougherty, M., Esposito, L., Krimigis, T. (Eds.), Saturn from Cassini-Huygens. Springer-Verlag, pp. 113 -- 159. Chapter 6 . De Pater, I. and Lissauer, J., 2001. Planetary Science. Cambridge University Press, Cambridge, U.K del Río-Gaztelurrutia, T., Legarreta, J., Hueso, R., Pérez-Hoyos, S., Sánchez-Lavega, A., 2010. A long-lived cyclone in Saturn's atmosphere: observations and models. Icarus 209, 665 -- 681. doi: 10.1016/j.icarus.2010.04.002 . del Río-Gaztelurrutia, T., Sánchez-Lavega, A., Antuñano, A., Legarreta,J., García- Melendo, E., Sayanagi, K.M, Hueso, R., Wong, M., Pérez-Hoyos, S., Rojas, J.F., Simon, A.A., de Pater, I., Blalock, J., Barry, T. 2018. A planetary-scale disturbance in a long living three vortex coupled system in Saturn's atmosphere. Icarus 302, 499-513. Dressel, L., 2019 "Wide Field Camera 3 Instrument Handbook, Version 11.0" (Baltimore: STScI) Fletcher, L.N., Irwin, P.G.J., Teanby, N.A., Orton, G.S., Parrish, P.D., de Kok, R., Howett, C., Calcutt, S.B., Bowles, N., Taylor, F.W., 2007. Characterising Saturn's vertical temperature structure from Cassini/CIRS.Icarus 189,457 -- 478. García-Melendo, E., Sánchez-Lavega, A., Hueso, R., 2007. Numerical models of Sat- urn's long-lived anticyclones. Icarus 191, 665 -- 677. doi: 10.1016/j.icarus.2007.05. 020 . García-Melendo, E., Pérez-Hoyos, S., Sánchez-Lavega, Hueso,R., 2011. Saturn's zonal wind profile in 2004-2009 from Cassini ISS images and its long-term variability. Icarus 215, 62 -- 74. doi: 10.1016/j.icarus.2011.07.005 . Hansen, J.E., Travis, L.D., 1974. Light scattering in planetary atmospheres. SpaceSci. Rev. 16, 527 -- 610 Ingersoll, A. P., R. F. Beebe, B. J. Conrath, and G. E. Hunt, 1984. Structure and dynamics of Saturn's atmosphere. In Saturn (T. Gehrels and M. S. Matthews,Eds.), pp. 195 -- 238. Univ. of Arizona Press, Tucson. Ingersoll, A. P., Ewald, S. P.Sayanagi, K. M., & Blalock, J. J. , 2018. Saturn's atmosphere at 1 -- 10 kilometer resolution. Geophysical Research Letters, 45. https://doi.org/10.1029/2018GL079255 Irwin, P.G.J. ,Teanby, N.A., de Kok, R., et al., 2008. The NEMESIS planetary atmosphere radiative transfer and retrieval tool. J. Quant. Spectrosc. Radiat. Transf. 109, 1136 -- 1150. Irwin, P.G.J. , Fletcher, L.N. , Read, P.L. , et al. , 2015. Spectral analysis of Uranus' 2014 bright storm with VLT/SINFONI. Icarus 264, 72 -- 89 . Karkoschka, E., Tomasko, M.G., 2005. Saturn's vertical and latitudinal cloud structure 1991 -- 2004 from HST imaging in 30 filters. Icarus 179, 195 -- 221 Lucarini, V., Saarinen, J.J., Peiponen, K.E., Vartiainen, E.M., 2005. Kramers -- Kronig Relations in Optical Materials Research.Springer-Verlag Berlin Heidelberg .Germany. Marcus, P.S., Asay-Davis, X., Wong, M.H., de Pater, I., 2012. Jupiter's new red oval: Dynamics, color, and relationship to jovian climate change. J. Heat Transfer, 135(1), 011007 (9 pages). Ortiz, J.L , Moreno, M. , Molina, AndA. , 1996. Saturn 1991 -- 1993: clouds and hazes. Icarus 119, 53 -- 66 . Pérez-Hoyos, S., Sánchez-Lavega, A., French, R.G., Rojas, J.F., 2005. Saturn's cloud structure and temporal evolution from ten years of Hubble space telescope images (1994 -- 2003). Icarus 176, 155 -- 174. Pérez-Hoyos, S. , Sanz-Requena, J.F. , Sánchez-Lavega, A. , Irwin, P.G.J. , Smith, A. , 2016. Saturn's tropospheric particles phase function and spatial distribution from Cassini ISS 2010 -- 11 observations. Icarus 277, 1 -- 18. Roman, M.T. , Banfield, D. , Gierasch, P.J. , 2013. Saturn ´s cloud structure from Cassini ISS. Icarus 225, 93 -- 110. Sánchez-Lavega, A., Rojas, J.F., Sada, P.V., 2000. Saturn's zonal winds at cloud level. Icarus 147, 405 -- 420. doi: 10.10 06/icar.20 0 0.6449 . Sánchez-Lavega, A., Hueso, R., Pérez-Hoyos, S., Rojas, J.F., French, R.G., 2004. Saturn's cloud morphology and zonal winds before the Cassini encounter. Icarus 170, 519 -- 523. doi: 10.1016/j.icarus.20 04.05.0 02 Sánchez-Lavega, A. , Hueso, R. , Pérez-Hoyos, S. , Rojas, J.F. , 2006. A strong vortex in Saturn's south pole. Icarus 184, 524 -- 531. Sánchez-Lavega A., G. Fisher, L. N. Fletcher, E. Garcia-Melendo, B. Hesman, S. Perez- Hoyos, K. Sayanagi, L. Sromovsky, "The Great Storm of 2010-2011", Chapter 13 of the book Saturn in the 21st Century, eds. K. H. Baines, F. M. Flasar, N. Krupp, T. S. Stallard, Cambridge University Press, pp. 377-416 (2019). ISBN 978-1-107-10677-2 Sánchez-Lavega A., E. García-Melendo, J. Legarreta, R. Hueso, T. del Río- Gaztelurrutia, J. F. Sanz-Requena, S. Pérez-Hoyos, A. A. Simon, M. H. Wong, M. Soria, J. M. Gómez-Forrellad, T. Barry, M. Delcroix, K. M. Sayanagi, J. J. Blalock, J. L. Gunnarson, U. Dyudina, S. Ewald, 2019. A complex storm system and a planetary-scale disturbance in Saturn's north polar atmosphere in 2018, Nature Astronomy (submitted). Sanz-Requena, J.F., Pérez-Hoyos, S., Sánchez-Lavega, A. , et al. , 2012. Cloud structure of Saturn's 2010 storm from ground-based visual imaging. Icarus 219, 142 -- 149. Sanz-Requena, J.F, Pérez-Hoyos, S., Sánchez-Lavega, A., Antuñano, A., Iwing, P. 2018. Haze and cloud structure of Saturn's North Pole and Hexagon Wave from Cassini/ISS imaging. Icarus .305, 284-300 Sayanagi K., R. West, K. Baines, L. Fletcher, P. Read and U. Dyudina, "Saturn's Poles: Vortices, Hexagons, and Auroral Aerosols", Chapter 12 of the book Saturn in the 21st Century, eds. K. H. Baines, F. M. Flasar, N. Krupp, T. S. Stallard, Cambridge University Press, pp. 337-376 (2019). ISBN 978-1-107-10677-2 Smith, B.A, 1981. Encounter with Saturn: Voyager 1 imaging results. Science 212, 163 -- 191. doi: 10.1126/science.212.4491.163. Smith, B.A, 1982. A new look at the Saturn system: the Voyager 2 images. Science 215, 505 -- 537. doi: 10.1126/science.215.4532.504. Sromovsky,L.A., Baines, K.H., Fry, P.M. 2013. Saturn's Great Storm of 2010 -- 2011: Evidence for ammonia and water ices from analysis of VIMS spectra. Icarus.226,402- 418 Trammell, H.J., Li, L., Jiang, X., Pan, Y., Smith, M.A., Bering, E.A., Hörst, S.M., Vasavada, A .R., Ingersoll, A .P., Janssen, M.A ., West, R.A ., Porco, C.C., Li, C., Si- mon, A .A ., Baines, K.H., 2016. Vortices in Saturn's Northern Hemisphere (2008 -- 2015) observed by Cassini ISS. J. Geophys. Res. Planets 121, 1814 -- 1826. doi: 10. 10 02/2016JE0 05122. Vasavada, A.R., Hörst, S.M., Kennedy, M.R., Ingersoll, A.P., Porco, C.C., Del Genio, A.D., West, R.A., 2006. Cassinin imaging of Saturn: Southern hemisphere winds and vortices. J. Geophys. Res. 111, E05004. West,R.A., Baines,K.H., Karkoschka,E., Sánchez-Lavega, A., 2009. Clouds and aerosols in Saturn's atmosphere. In: Saturn from Cassini-Huygens, Dougherty, M.K., Esposito, L.W., Krimigis, S.M.(Eds.).. Springer, Netherlands, pp.161 -- 179.
1207.1987	1	1207	2012-07-09T08:40:42	An insight in the surroundings of HR4796	[ "astro-ph.EP", "astro-ph.IM" ]	HR4796 is a young, early A-type star harbouring a well structured debris disk, shaped as a ring with sharp inner edges. It forms with the M-type star HR4796B a binary system, with a proj. sep. ~560 AU. Our aim is to explore the surroundings of HR4796A and B, both in terms of extended or point-like structures. Adaptive optics images at L'-band were obtained with NaCo in Angular Differential Mode and with Sparse Aperture Masking (SAM). We analyse the data as well as the artefacts that can be produced by ADI reduction on an extended structure with a shape similar to that of HR4796A dust ring. We determine constraints on the presence of companions using SAM and ADI on HR4796A, and ADI on HR4796B. We also performed dynamical simulations of a disk of planetesimals and dust produced by collisions, perturbed by a planet located close to the disk outer edge. The disk ring around HR4796A is well resolved. We highlight the potential effects of ADI reduction of the observed disk shape and surface brightness distribution, and side-to-side asymmetries. No planet is detected around the star, with masses as low as 3.5 M_Jup at 0.5" (58 AU) and less than 3 M_Jup in the 0.8-1" range along the semi-major axis. We exclude massive brown dwarfs at separations as close as 60 mas (4.5 AU) from the star thanks to SAM data. The detection limits obtained allow us to exclude a possible close companion to HR4796A as the origin of the offset of the ring center with respect to the star; they also allow to put interesting constraints on the (mass, separation) of any planet possibly responsible for the inner disk steep edge. Using detailed dynamical simulations, we show that a giant planet orbiting outside the ring could sharpen the disk outer edge and reproduce the STIS images published by Schneider et al. (2009).	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. paperrevfinsansbf May 21, 2018 c(cid:13) ESO 2018 2 1 0 2 l u J 9 . ] P E h p - o r t s a [ 1 v 7 8 9 1 . 7 0 2 1 : v i X r a An insight in the surroundings of HR4796⋆ A.-M. Lagrange1, J. Milli1, A. Boccaletti2, S. Lacour2, P. Thebault2, G. Chauvin1,3, D. Mouillet1, J.C. Augereau1, M. Bonnefoy3, D. Ehrenreich1, and Q. Kral2 1 Institut de Plan´etologie et d'Astrophysique de Grenoble, Universit´e Joseph Fourier, CNRS, BP 53, 38041 Grenoble, France e-mail: [email protected] 2 LESIA-Observatoire de Paris, CNRS, UPMC Univ. Paris 06, Univ. Paris-Diderot, 92195, Meudon, France 3 Max Planck Institut fur Astronomie Konigstuhl 17, D-69117 Heidelberg, Germany Received date / Accepted date ABSTRACT Context. HR4796 is a young, early A-type star harbouring a well structured debris disk, shaped as a ring with sharp inner edges. The inner edge might be shaped by a yet unseen planet inside the ring; the outer one is not well understood. The star forms together with the M-type star HR4796B, a binary system, with a projected separation of ≃ 560 AU. Aims.Our aim is to explore the surroundings of HR4796A and B, both in terms of extended or point-like structures. Methods. Adaptive optics images at L'-band were obtained with NaCo in Angular Differential Mode and with Sparse Aperture Masking (SAM). We analyse the data as well as the artefacts that can be produced by ADI reduction on an extended structure with a shape similar to that of HR4796A dust ring. We determine constraints on the presence of companions using SAM and ADI on HR4796A, and ADI on HR4796B. We also performed dynamical simulations of a disk of planetesimals and dust produced by collisions, perturbed by a planet located close to the disk outer edge. Results. The disk ring around HR4796A is well resolved. We highlight the potential effects of ADI reduction of the observed disk shape and surface brightness distribution, and side-to-side asymmetries. We produce 2D maps of planet detection limits. No planet is detected around the star, with masses as low as 3.5 MJup at 0.5" (58 AU) and less than 3 MJup in the 0.8-1" range along the semi-major axis. We exclude massive brown dwarfs at separations as close as 60 mas (4.5 AU) from the star thanks to SAM data. The detection limits obtained allow us to exclude a possible close companion to HR4796A as the origin of the offset of the ring center with respect to the star; they also allow to put interesting constraints on the (mass, separation) of any planet possibly responsible for the inner disk steep edge. Using detailed dynamical simulations, we show that a giant planet orbiting outside the ring could sharpen the disk outer edge and reproduce the STIS images published by Schneider et al. (2009). Finally, no planets are detected around HR4796B with limits well below 1 MJup at 0.5" (35 AU). Key words. stars: early-type – stars: planetary systems – stars: individual (HR4796) 1. Introduction Understanding planetary systems formation and evolution has become one of the biggest challenges of astronomy, since the imaging of a debris disk around β Pictoris in the 80's (Smith and Terrile 1984) and the discovery of the first exo- planet around the solar-like star 51 Pegasi during the 90's (Mayor and Queloz 1995). Today, about 25 debris disks have been imaged at optical, infrared, or submillimetric wave- lengths (http://astro.berkeley.edu/kalas/disksite). Debris disks trace stages of system evolution where solid bodies with sizes significantly larger than the primordial dust size (larger than me- ters or km sized) are present to account for through collisions, the presence of short lived dust. They are thought to be privi- leged places to search for planets. This is particularly true for those showing peculiar structures (e.g. rings with sharp edges) or asymmetries, spirals even though other physical effects not in- volving planets could also lead to the formation of similar struc- tures. Takeuchi and Artymowicz 2001 for instance showed that relatively small amounts (typ. 1-a few Earth masses) of gas can shape the dust disk through gas-dust interactions into rings (see below). It is remarkable however that all the stars around which relatively close (separations less than 120 AU) planets have been imaged are surrounded by debris disks: a ≤ 3-MJup planetary companion was detected in the outskirts of Fomalhaut's debris disk (119 AU from the star; Kalas et al. 2008), four planetary companions of 7-10 MJup were imaged at 15, 24, 38 and 68 AU (projected separations) from HR 8799 (Marois et al. 2008, Marois et al. 2010). Using VLT/NaCo L'-band saturated im- ages, we detected a 9+ −3 MJup planet in the disk of β Pictoris (≃ 12 Myr) with an orbital radius of 8-12 AU from the star (Lagrange et al. 2010; Chauvin et al. 2012). More recent stud- ies at Ks show that β Pic b is located in the inclined part of the disk (Lagrange et al. 2011), conforting the link between disk morphology and the presence of a planet (Mouillet el al. 1997; Augereau et al. 2001). β Pic b also confirms that giant planets form in a timescale of 10 Myr or less. Interestingly, β Pic b and maybe also HR8799e could have formed in situ via core accre- tion, in contrast with the other, more remote, young companions detected with high-contrast imaging. If formed in situ, the latter probably formed through gravitational instabilities within a disk, or through the fragmentation and collapse of a molecular cloud. Send offprint requests to: A.-M. Lagrange ⋆ Based on observations collected at Observatory, Chile, ESO; run 085.C-0277A. the European Southern There are many exciting questions regarding disks and plan- ets: could different planet formation processes be at work within a given disk? Disks and planets are known to exist in binary 1 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Table 1. HR 4796 stellar properties. (1)Van Leeuwen, 2007, (2) Stauffer et al. 1995. Spectral Type A0V 5.78 V 0.009 B − V 3.74 + π 8+ Age −2 Myr (2) −0.33 mas (1) systems; (how) do massive companions impact on the dynam- ical evolution of inner planets and disks? In a recent study, Rodriguez and Zuckerman 2011 showed that for a binary sys- tem to have a disk, it must either be a very wide binary sys- tem with disk particles orbiting a single star or a small separa- tion binary with a circumbinary disk. Such results can help the search for planets if one relates debris disks with planet forma- tion. However, another question, already mentioned, is to which extent and how can debris disks indicate the presence of already formed planets? A particularly interesting system in the present context is HR4796A, consisting of an early-type (A0), young, close- by star (see Table 1) surrounded by dust, identified in the early 90's (Jura 1991) and resolved by Koerner et al. 1998 and Jayawardhana et al. 1998 at mid-IR from the ground, and at near-IR with NICMOS on the HST (Schneider et al. 1999) as well as from the ground, coupling coronagraphy with adap- tive optics (Augereau et al. 1999). The resolved dust shapes as a narrow ring, with steep inner and outer edges. The steep- ness of the inner edge of the dust ring has been tentatively at- tributed to an unseen planet (Wyatt et al. 1999); however, none has been detected so far. The disk images + SED modeling re- quired at least two populations of grains, one, narrow (a few tens AU) cold ring, located at ≃ 70 AU, and a second one, hotter and much closer to the star (Augereau et al. 1999), al- though the existence of an exozodiacal dust component is de- bated (Li and Lunine 2003). Wahhaj et al. 2005, argued that in addition to the dust responsible for the ring-like structure ob- served at optical/near IR wavelengths, a wider, low-density com- ponent should be present at similar separations to account for the thermal IR images. Recently, higher quality (SN/angular resolu- tion) data were obtained with STIS (Schneider et al. 2009) and from the ground (Thalmann et al. 2011), the latter using perfor- mant AO system on a 8 m class telescope, as well as Angular Differential Imaging (ADI, see below). With the revised dis- tance of HR4796 with respect to earlier results, the ring radius is at about 79 AU, and has a width of 13 AU (STIS 0.2-1 µm data). Furthermore, these authors show a 1.2-1.6 AU physical shift of the projected center of the disk wrt the star position along the major axis and Thalmann et al. 2011 moreover mea- sures a 0.5 AU shift along the disk minor axis. Finally, and very interestingly in the context of planetary system formation, HR4796 has also a close-by (7.7 arcsec, i.e. about 560 AU pro- jected separation), M-type companion (Jura et al. 1993), plus a tertiary one, located much further away (13500 AU in projection; Kastner et al. 2008). The closest companion may have played a role in the outer truncature of the disk, even though, according to Thebault et al. 2010, it alone cannot account for the sharp outer edge of HR4796 as observed by STIS. ADI (Marois et al. 2006) is a technics that has proved to be very efficient in reaching very high contrast from the ground on point-like objects. It has also been used to image disks around HD61005 (the Moth disk, Buenzli et al. 2010), HD32297 2 (Boccaletti et al. 2011), and β Pic (Lagrange et al. 2011), but it should be used with care when observing extended structures, as the morphology of these structures may be strongly impacted by this method (Milli et al, 2012, in prep). We present here new high contrast images of HR4796 ob- tained with NaCo on the VLT at L' band, both in ADI and Sparse Aperture Mode (SAM, Tuthill et al. 2010) aiming at exploring the disk around HR4796A, and at searching for possible com- panions around HR4796A as well as around HR4796B. SAM and ADI are complementary as the first give access to regions in the 40-400 mas range from the star, and ADI further than typi- cally 300 mas. Pedagogical examples of SAM performances and results are reported in Lacour et al. 2011. 2. Data log and Reduction 2.1. Logofobservations VLT/NaCo (Lenzen et al. 2003; Rousset et al. 2003) L'-band data were obtained on April, 6 and 7th, 2010, in ADI mode and with SAM. We used the L27 camera which provides a pixel scale of ≃ 27 mas/pixel. The log of observations is given in Table 2. The precise detector plate scale as well as detector orienta- tion were measured on an θ 1 Ori C field observed during the same run, and with HST (McCaughrean and Stauffer 1994 (with the same set of stars TCC058, 057, 054, 034 and 026). We found −0.04 mas per pixel and the North orienta- a platescale of 27.10+ −0.09 ◦if we do not consider tion offset was measured to be -0.26 + −0.3 ◦otherwise (see a systematics in the North position, or -0.26 + Lagrange et al. 2011 for a detailed discussion on the absolute un- certainty on the detector orientation). 2.1.1. ADI data The principle of ADI imaging is given in Marois et al. 2006 (see also Lafreni`ere et al. 2007). Here, a typical ADI sequence con- sisted in getting sets of saturated images (datacubes of NDIT im- ages) at different positions on the detector, followed and prece- dented by a series of un-saturated PSF images recorded with a neutral density filter (ND Long). These unsaturated images are used to get an estimate of the PSF shape for calibration purposes (photometry, shape), and fake planet simulation. On April 5/6, a few tests were made with different offsets patterns (star centered on either 2 or 4 positions on the detector) so as to test the im- pact of HR4796B (which rotates on the detector during the ADI observations) on the final image quality1. This is important as the field of view (FoV) rotation was fast. On April 6/7, the sat- urated images were recorded with two offsets corresponding to two opposite quadrants on the detector. Both nights, the atmo- spheric conditions were good on average, but variable, and the amplitude of field rotation was larger than 80◦(see Table 2). The comparison between the PSFs taken prior and after the saturated images does not reveal strong variations. 2.1.2. SAM data Sparse aperture masking is obtained on NaCo by insertion of a mask in the cold pupil wheel of the camera (Lacour et al. 2011). The mask acts as a Fizeau interferometer. It forms in the focal 1 As a consequence, in some offset configurations, HR4796B was out of the detector FoV, hence the number of usefull data on HR4796B is smaller than that available on HR4796A. A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Table 2. Log of observations. "Par. range" stands for the parallactic angles at the start and end of observations; "EC mean" for the average of the coherent energy and "t0 mean" for the average of the coherence time during the observations. Date Apr. 07, 2010 Apr. 07, 2010 Apr. 07, 2010 Apr. 06, 2010 Apr. 06, 2010 Apr. 06, 2010 Jun. 07/08, 2011 Jun. 07/08, 2011 UT-start/end DIT (s) 03:08/03:13 03:14/04:56 04:57/05:06 03:07/03:12 03:13/05:05 05:06/05:11 00:28/02:00 00:44/02:12 0.2 (ND Long) 0.2 0.2 (ND Long) 0.2 (ND Long) 0.2 0.2 (ND Long) 0.1 0.1 (Calibrator) NDIT N exp. 150 150 150 150 150 150 500 500 8 160 8 8 175 8 32 32 Par. range ◦ -48.3 to-47.2 -45.3 to 31.9 32.6 to 38.3 -50.5 to -48.4 -47.9 to 35.1 35.8 to 38.8 13/64 19/67 Air Mass range 1.07 1.07 to 1.05 1.05 1.07 1.07 to 1.05 1.05 1.03/1.15 1.04/1.17 EC mean % 44.9 41.4 44.0 25.1 37.9 45.9 - - t0 mean (ms) 3.0 2.6 2.8 6.4 4.9 4.6 3.2 3.2 plane of the camera interference fringes which are used to re- cover the complex visibilities of the astronomical object. Of the four available masks (Tuthill et al. 2010), we used the 7 holes mask which gives the highest throughput (16%). It is made of 1.2 meters wide circular apertures (scaled on M1) positioned in a non-redundant fashion over the pupil. Minimum and maximum baseline lengths are respectively 1.8 and 6.5 meters. Each mask offered by SAM can be used in addition to almost all the spectral filters offered by Conica. The principle of SAM is based on its ability to facilitate the deconvolution of phase aberrations. Phase errors are introduced by i) atmospheric residuals and ii) instrumental aberrations (also called non-common path errors). We used integration times of the order of the typical coherence time of the phase errors. It permits a partial deconvolution of the remaining atmospheric perturbation not corrected by the AO. But most importantly, it gives an excellent correction of the slowly changing instrumental aberrations. This later point is the important factor which makes aperture masking competitive with respect to full aperture AO. In practice, L' SAM data on HR4796 were obtained on June 2011. The adopted DIT was 100 ms, equivalent to a few τ0 in the L' band. Each set of observation consists in 8 ditherings of the telescope to produce 8 datacubes of 500 frames on 8 different positions on the detector. Each dither moves the star by 6 arc- sec in X or in Y on the windowed detector (512 by 512 pixels, equivalent to 14 arcsec on sky). After 8 dithers, the telescope is offset to the K giant star HD110036 for calibration, where the very same observation template is repeated. Four star-calibrator pairs were obtained totalizing 64 datacubes, requiring a total ob- servation time of 2 hours (including overheads). Over this time, the object has rotated by 50 degrees (the variation of the paral- lactic angle). 2.2. Datareduction 2.2.1. ADI data Each individual ADI image was bad pixel-corrected and flat- fielded as usual. Background subtraction was made for each cube using the closest data cubes with the star at a different off- set. Data selection was also made, within each data cube and also for each data cube. Recentering of the images was done using the offsets measured by Moffat fitting of the saturated PSF. The data cubes were then stacked (averaged) and then re- duced with different procedures that are described in details in Lagrange et al. 2011 and reference there-in: cADI, sADI, rADI and LOCI. These procedures differ in the way the star halo is estimated and subtracted. We recall here the differences between these various procedures, as well as new ones developed to limit the disk self-subtraction in cADI and/or LOCI: - In cADI, the PSF is taken as the mean or median of all indi- vidual recentered ADI saturated images. - To remove as much as possible the contribution of the disk from the PSF in the cADI images, we tested two slightly modified cADI reductions. In the first one, we start as usual, i.e. build a PSF from the median of all data, subtract this PSF to all data and rotate back the obtained residual images to align their FoV. The data are then combined (median) to get a first image of the disk. Then, to remove the disk contribution to the PSF, we rotate the disk image back to the n different FoV orientations corresponding to those of the initial images and subtract the median of these rotated disk images to the PSF. We obtain thus a PSF corrected (to first order) from the disk contribution. This disk-corrected PSF is then subtracted to the individual initial images; the individual residuals are then rotated back to be aligned and stacked (median) to get a new disk image (corresponding to one iteration). This ADI procedure is referred to as cADI-disk. In the second one, we mask the disk region in each file when used to compute the reference2. We will call this method mcADI. This method will be described in details in a forthcoming paper (Milli et al, 2012, in prep). - The rADI procedure (identical to Marois et al. 2006 ADI) selects for each frame a given number of images that were recorded at parallactic angles separated by a given value in FWHM (the same value in FWHM for each separation), to build a PSF to be subtracted from the image considered. - In the LOCI approach, for each given image and at each given location in the image, we compute "parts" of "PSFs", using linear combinations of all available data, with coeffi- cients that allow to minimize the residuals in a given portion of the image. - To limit the impact of the disk self-subtraction on the LOCI images, we also modified our LOCI approach, masking the disk in each file whenever the disk appears in the optimiza- tion zone (see Milli et al, 2012, in prep.). We will call this method mLOCI. 2 In practice, we use a synthetic image (without noise) of the disk as described below, and scaled to the observed disk flux, and we define the binary mask by ascribing 0 to the pixels in the disk region that are above a given threshold (20% of the maximum flux). Then the mask is applied to each frame in the cube for the PSF estimation, so that the pixels corresponding to the disk regions in each frame are not taken into account. 3 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 The parameters used for the rADI and LOCI procedures are the following : - LOCI/mLOCI ∆r = 1.4 × FWHM below 1.6" and 5.6 × FWHM beyond (radial extent of the subtraction zones); g = 1 or 0.5 (radial to azimuthal width ratio), NA = 300; separation criteria 1. × FWHM. - rADI: separation criteria: 1.5 × FWHM; number of images used to compute each "PSF" : 20, ∆r = 1.4 × FWHM below 1.6 arcsec and 5.6 × FWHM beyond (radial extent of the psf reconstruction zones) For comparison purposes, we also performed a zero-order reduction (hereafter referred to as "nADI") which consists in, for each image, 1) computing an azimuthal average of the image (with the star position as the center of the image); we get then a 1-D profile, 2) circularizing this 1-D profile to get a 2-D image centered on the star position, 3) subtracting the obtained image to the initial image to get corrected image. We then derotate and stack all the "corrected" images. nADI clearly does not benefit from the pupil stabilization and is not to be considered as a real ADI reduction procedure, but can help in some cases disentan- gling artefacts produced by ADI reductions from real features. The data obtained on the 6th and 7th were reduced separately and then averaged. As they happened to have similar S/N ratio, a simple averaging was made. 2.2.2. SAM data The first step to reduce the SAM data is to clean the frames. This can be done in the same way as any classical imaging method in the infrared. In practice, we flatfielded the data and subtracted the background. The background was estimated by taking the median value of the 8 datacubes of a single observation set. As any interferometric facility, the observable parameters of SAM are fringes. The information lays in the contrast (which, once normalized, is called visibility) and the phase. Contrasts and phases are obtained by least square fitting of the diffrac- tion pattern. Since the fitting of sinusoidal curves is a linear least square problem, a downhill algorithm to find the maximum like- lihood was not required. Instead, inversion was done by projec- tion of the datacubes on a parameter space defined by each com- plex visibility fringes. The matrix used for projection is deter- mined by singular value decomposition of a model of the fringes. In the end, we checked that it gives exactly the same result as a least square minimization algorithm of the kind of conjugate gra- dient (but much faster). The fringes are modeled by cosines of given frequency multiplied by the PSF of the Airy pattern of a single hole. Wavelength bandwith is accounted for by smearing the pattern over the filter bandpass. As a result, we get a single complex value for each baseline and each frame. They are used to compute the bispectrum, which is summed over the 8 datacubes which correspond to a single acquisition. Then, the closure phases are obtained by taking the argument of the bispectrum. One set of closure phase is obtained for an observation set which takes around 8 minutes. Over that time, the parrallactic angle changes less than 6 degrees, which effect is neglected (baselines rotation during an observation set is not accounted for). The final step consists in calibrating the closure phase of HR4796 by subtracting the corresponding val- ues obtained for the red giant (HD110036). 4 3. Data simulations Obviously, ADI affects the resulting disk shape because of disk self-subtraction. This effect is expected to be more important as the disk inclination with respect to line-of-sight decreases. Also, the different ADI reduction procedures will impact differently the disk shape. A general study of the impact of ADI on disk reductions will be presented in a forthcoming paper (Milli et al, 2012, in prep.). In this paper, we concentrate on the HR4796 case and we monitor this impact using fake disks, as done in Lagrange et al. 2011. 3.1. Assumptions To simulate the HR4796 disk, we assumed, following Augereau et al. 1999 a radial midplane number density distribu- tion of grains ∝ ((r/r0)−2αin + (r/r0)−2αout)−0.5. We chose r0 = 77.5 AU, αin =35 to ensure a very sharp inner edge, and αout = -10, as assumed by Thalmann et al. 2011. The vertical distribution is given by: ξ (cid:17)γ Z(r, z) = e−(cid:16) z r0 (cid:17)β is 1 AU at 77.5 AU. where the height scale ξ = ξ0 (cid:16) r The disk flaring coefficient is β=1 and the coefficient γ = 2 ensures a gaussian vertical profile. The disk is inclined by 76 degrees (a pole-on disk would have an inclination of 0 degree), and we assumed an isotropic scattering (g=0), as Hinkley et al. 2009 polarimetric measurements indicate a low value for g (0.1-0.27). The disk was simulated using the GRaTer code (Augereau et al. 1999; Lebreton et al. 2012). It will be re- ferred to as HR4796SD. The ring FWHM thus obtained is 0.14" (before reduction) under such hypothesis. We also con- sidered another disk, with all parameters identical to those of HR4796SD, but with αout = -4; this disk (referred to as HR4796blowoutSD) is representative of the outer density distri- bution that would be observed if the outer brightness distribution was dominated by grains expelled by radiation pressure as in the case of β Pic (Augereau et al. 2001). 3.2. Simulateddiskimages The flux of the simulated disk is scaled so as to have the same number of ADU (at the NE ansae) as in the real disk, once both simulated and real data are reduced by cADI. When brighter disk are needed, a simple scaling factor is applied. The simulated pro- jected disks are then injected in a datacube, at each parallactic angle, corresponding to each real data file, and are then con- volved either by a theoretical PSF matching the telescope and instrument response, or the average of the real PSFs taken prior to and after the saturated images. Each image is added to each frame of the original data cube, with a 130◦or 90◦offset in PA with respect to the real disk, so as to minimize the overlap be- tween both disks. The datacubes are then processed by nADI, CADI, mcADI, LOCI and mLOCI. 4. The HR4796A disk 4.1. Diskimages:qualitativeview Figure 1 shows the images obtained when combining the data obtained on the 6th and 7th of April. Images resulting from the ADI reductions described above are showed: cADI, cADI-disk, mCADI, rADI, LOCI, and mLOCI. We also show for compari- son the image resulting from nADI reduction. A.-M. Lagrange et al.: An insight in the surroundings of HR4796 features. To test whether such features could be due to the ADI reduction and/or data characteristics, we built cADI and LOCI images of a bright3 HR4796SD-like disk, inserted into our data cube, convolved by a theoretical PSF and, in the case of LOCI, reduced with Thalmann et al. LOCI parameters, assuming a 23 degrees FoV rotation. The input disk images are shown in Fig.2, as well as the reduced ones. The reduced images clearly show features similar to those observed by Thalmann et al. 2011 in their Figures 1 and 3. We also show in Fig.2 the same simu- lations of a similar disk inserted into a data cube matching the parallactic angle excursions, but assuming no noise. Note that for the LOCI reduction of noiseless data, we took the coeffi- cients derived from the previous LOCI reduction (taking noise into account); this is necessary to avoid major artefacts when using LOCI on noiseless data. The later simulations (no noise) highlight the reduction artefacts. Other simulations are provided in Milli et al. (2012, in prep.). Hence, we conclude that the fea- tures indicated as "streamers" are in fact artefacts due to the data characteristics (FoV rotation amplitude, number of data, SN) and data reduction. The fact that we do not see them in the present data is due to our relatively lower dynamical range, and to the larger FoV rotation amplitude. To check this, we in- jected a fake disk (HR4796SD), with a flux similar to the ob- served one in the actual datacubes (with a 130◦offset in PA to avoid an overlap of the disks). We processed the new data cubes as described above with cADI, mcADI, LOCI and mLOCI. The resulting images are shown in Figure 3; the artefacts are not detectable. This is also true when considering a fake disk with blowout (HR4796blowoutSD). Finally, we note a small distorsion in the SW disk towards the inner region of the cADI and LOCI images, at (r,PA) be- tween (19 pix, 235◦) and (28 pix, 220◦). The feature, indi- cated by a green arrow in Figure 1, is however close to the noise level. If we go back to the individual images taken on April, 6th and April, 7th, we see that this feature is barely de- tectable on the April 6th eventhough in both cases, there is a very faint signal inside the disk ellipse (see Figure 4 for the cADI images). Hence, in the present data, this feature could be due to noise. However, it seems to be at the same posi- tion as that pointed by Thalmann et al. 2011 in their data as well as Schneider et al. 2009 data as well as, in L'-data ob- tained at Keck by C. Marois and B. McIntosh (priv. com.). In Thalmann et al. 2011 data, it appears as a loss of flux in the an- nulus. In the L' data, we see rather a distorsion in the disk and a possible very faint additional signal at the inner edge of the disk. However, the ring does not appear as azimuthally smooth in the Thalmann et al. 2011 or our data, due to ADI reduction and limited SN, so it is not excluded at this stage that this fea- ture might be an artefact. Clearly, new data are needed to confirm this structure, and if confirmed, to study its shape as a function of wavelength. If confirmed, its origin should be addressed. In the context of the HR4796 system, an interesting origin to be con- sidered is the presence of a planet close to the inner edge of the disk. To test the impact of the ADI reduction procedure on a disk + close planet system at L', we inserted a fake point-like source close to our fake disk HR4796SD (rotated by 130◦with respect to the real disk, convolved by the observed PSF, and inserted in the datacube) inner edge, and processed the data as described previously. For these simulations, we assumed a disk about 10 times brighter than the real one. We run several simulations with 3 we assumed a disk 10 times brighter than the actual one so as to take into account the better SN of Thalmann et al. 2011 H data, compared to the present L' ones. 5 Fig. 1. HR4796A disk at L' (linear scale). The pixel scale is 27 mas/pixel. From top to bottom, the same data reduced with nADI, cADI, cADI-disk, mcADI, rADI, LOCI, and mLOCI (see text). Note that the color codes are identical for all cADI reduc- tions reductions on the one hand, and LOCI reductions on the other hand, to enable comparisons within a given method, but different cuts are used for cADI, rADI and LOCI reductions. The disk is clearly detected at "large" separations from the star with the nADI reduction, and is, expectedly, lost in the Airy rings closer to the star. The parts closer to the star in projection are revealed only by the real ADI reductions, and actually, the disk is more completely detected than in previously published images, in particular the west side is almost continuously de- tected. The masking greatly improves the image quality of the LOCI image; the impact of mcADI with respect to cADI is, expectedly, less important, even though the flux restitution is increased. Nevertheless, the dynamical range of our images is lower than that of the recently AO published images. This is be- cause the present data are obtained at L', with a higher Strehl ratio, whereas the previous ones were obtained at shorter wave- lengths, with lower Strehl ratio, but with detectors that have much lower background levels. The ring appears very narrow in our L'-data, barely resolved: the FWHM measured on the PSF is 4.1 pixels (0.11"), while the ring FWHM is ≃ 5.7 pixels (0.15") (NE) and resp. 5.0 pixels (0.13") (SW) on the cADI data. We will however see below that the ADI reduction has an impact on the observed width. Thalmann et al. 2011 report a relative enhancement of the disk brightness in the outer part of the ring, along the semi- major axis, that they describe as streamers emerging from the ansae of the HR 4796 disk. Our images do not show such strong A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Fig. 2. Left) From top to bottom 1) simulated HR4796SD-like disk (projected, no noise); 2) simulated cADI reduction of this disk once inserted into our data cube, and convolved by a theoretical PSF. We assumed a 23 degrees FoV rotation as Thalmann et al (2011). 3) idem with LOCI reduction. Right) Idem without noise. Fig. 3. Top: simulated disk (model HR4796SD) inserted into the data: images (linear scale) reduced with cADI, mcADI, LOCI and mLOCI, for a rotation angle amplitude of 85 deg similar to that of our data. Bottom: idem for a disk model assuming αout =-4 (model HR4796blowoutSD). North is up and East is to the left. The real disk is NE-SW oriented; the simulated one is NW-SE oriented. 6 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Fig. 4. Top: cADI image of HR4796 on April, 6th. Bottom: idem on April, 7th. different values of planet fluxes and distances to the disk. For some values of the planet position and flux, we were able to re- produce a disk appearent distorsion, especially in the LOCI im- ages. A representative example is given in Figure 5; in this case, the planet flux would correspond to a 2 MJup mass for a disk brightness similar to the HR4796 one. 4.2. Diskproperties 4.2.1. Disk geometry We fitted the observed disk by an ellipse, using the maxi- mum regional merit method, as in Buenzli et al. 2010 and in Thalmann et al. 2011. The resulting semi-major axis a, semi- minor axis b, disk center position along the semi-major axis (xc), and the semi-minor axis (yc), and inclination (defined as arccos(b/a)) are provided in Table 3, and the fit is showed in Figure 6. These parameters are derived from the selection of the best fits, defined as those with parameters within 5% of the best fit (best merit coefficient). The uncertainties associated to these measurments take into account the dispersion within this 5% range, and the other sources of uncertainties that are described hereafter. To estimate the impact of the PSF convolution and ADI process on the ellipse parameters, we used our simulated disk HR4796SD (without noise) and fitted the disk with an ellipse before and after the PSF convolution and the ADI reduction. For (a,b), differences of (-0.03;0.12) pixels were found with cADI, and (-0.06;-0.3) pixels with LOCI. For (xc, yc), no significant differences were found with cADI and very small with LOCI. Fig. 5. cADI, mcADI, rADI, LOCI and mLOCI images of simu- lations of fake disk + fake point-like source (indicated by a green arrow). No significant difference was measured on the inclination with cADI while a difference of -0.4◦was found with LOCI. Finally, no significant differences were found on the PA. We corrected the measured values on the disk from these biases. We also in- serted a model disk HR4796SD in the data cube (at 90 degrees), and processed the data. The differences found between the pa- rameters of the injected disk and the ones of the recovered disk are compatible with the ones obtained in the case "witout noise". Besides, the imperfect knowledge of the PSF center may also affect the results. To estimate this impact quantitatively, we first estimated the error associated to the PSF center, as in Lagrange et al. 2011. The error was found to be [0.,0.27] pixel 7 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 The ring shape seems nevertheless to indicate a sharp outer edge, but we need to check the impact of the ADI reduction pro- cedure on the final shape of the disk. Figure 8 illustrates the evo- lution of the SBD along the semi-major axis, starting from the fake disk HR4796SD, then once the disk is inserted in the real data cubes at 130◦(see the corresponding images in Figure 3), convolved by the observed PSF, and finally when the datacube is reduced with cADI and LOCI. The SBD shape is clearly im- pacted. We note that the effect is stronger in the inner region that in the outer one. To test whether we can discriminate be- tween a steep and a less steep outer profile, we consider the fake disks HR4796SD and HR4796blowoutSD inserted in the real data cubes at 130◦(see the corresponding images in Figure 3), and convolved with the real PSF. The SBD profiles after con- volution and reduction along the semi-major axis are given in Figure 9. We note that the slight shift between the observed and simulated disks SBD is due to an unperfect assumption on the ring position, and is not relevent here. The observed SBD pro- file appears to be more similar in shape to the ones correspond- ing to the HR4796SD case rather than the HR4796blowoutSD one. We conclude then that even when taking into account the possible biases, the data indicate a very steep outer edge, com- patible with αout = -10, as found by Thalmann et al. 2011 rather than a less steep one. We cannot provide precise values to the outer NE and SW slopes with the present data, but they are in any case different from the ones measured in the case of the other A-type stars such as β Pictoris (typ. between -4 and -5; Golimowski et al. 2006) and HD32997 (Boccaletti et al. 2011), and that are expected from a disk which outer part is dominated by grains blown out by radiation pressure from an A-type star (see below). 4.2.3. Disk width Our data indicate a ring width of about 0.154" after cADI re- duction for the NE side and 0.136" for the SW side (data binned over 5 pixels). With mcADI, these values are only marginally changed: 0.147" and 0.135" respectively. However, as seen above, the SBD, especially inner to the ring is impacted by the PSF convolution, the amplitude of the FoV rotation, the ADI procedure, the binning used for the extraction of the SBD, the noise level as well as the zero flux level after reduction. We made several test with fake disks to estimate the impact of these steps on the FWHM. Also, we tested the impact of the evaluation of the zero level after reduction. It appears that the disk width is mainly affected in the present case by the PSF convolution and the zero level. Taking all these parameters into account, we can- not conclude that the disk is significantly narrower than the size found by Schneider et al. (2009) 0.197", which once corrected from the broadening by the STIS PSF became 0.18" (13 AU) at shorter wavelengths. 5. Companions around HR4796A 5.1. SAMdetectionlimits The detection limits from the SAM dataset are derived from a 3D χ2 map. This map has on each axis the three parameters used to model a binary system: the separation, the position angle, and the relative flux. This model, fitted on the closure phase, is de- tailed in Lacour et al. 2011. Visibilities are discarded. Fig. 12 is showing the detection limits as a function of right ascension and declination. It is obtained by plotting the 5 σ isocontour of the 3D map (the isocontour level is given by a reduced χ2 of 25). Fig. 6. Best elliptical fit to the L' cADI data. NE is to the left and SW to the right. on the x-axis and [-0.06, +0.04] pixel on the y-axis of the detec- tor. It appears that this imperfect knowledge on the PSF center does not significantly affect the values of (a,b). It impacts the un- certainty of the ellipse center by up to 0.2 pixel along the major and minor axis, and the disk PA (0.24 ◦). The uncertainty on the PA measurement is found to be 0.15◦in cADI. Finally, the PA measurement is also impacted by the uncer- tainty on the true North Position (0.3◦; see a discussion related to this last point in Lagrange et al. 2011). Our data show an offset from the star center of ≃ 22 mas on the cADI images (≃ 20 mas on the LOCI data) of the center of the fitted ellipse along the major axis, and to the South. Given the uncertainty associated to this measurement, 7 mas, we conclude that the observed offset is real. This off- set along the major axis is in agreement with previous results of −6 mas) and Thalmann et al. 2011 on Schneider et al. 2009 (19+ −5.1 mas). The latter detected more- their LOCI images (16.9 + over an offset of 15.8+ −3.6 mas along the minor axis, which was not detected by Schneider et al. 2009; the measured offset on our cADI images is about 8 mas + −6 mas; hence very close to the er- ror bars, so the present data barely confirm the offset found by Thalmann et al. 2011. 4.2.2. Brightness distribution Figure 7 shows the observed radial surface brightness distribu- tions (SBD) for the HR4796A disk at L' along the major axis af- ter cADI, and LOCI reductions. The SBD extraction was made using a 5 pixel vertical binning. The dynamical range of our data is small (factor of 10); it is improved with masking technics, thanks to a lesser disk self-subtraction (see also Figure 7). Yet, the disk being only slightly resolved, we cannot perform mean- ingful slope measurements on our data. Indeed, the slope of the surface brightness distribution depends on several parameters: the PSF, the amplitude of the FoV rotation, the ADI procedure, the binning used for the extraction of the SBD, and the separa- tion range on which the slopes are measured. In the present case, the separation range is too small to allow a proper measurement of the slope: we run simulations of the HR4796SD disk without noise and checked that indeed, measuring the slopes between the maximum flux and the threshold corresponding to the noise on the actual data gave slopes very different (much higher) from the slope measured on a larger separation range. 8 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Table 3. Ellipse parameters of the observed disk: semi-major axis (a, mas), semi-minor axis (b, mas), position of the center of the ellipse with respect to the star ((xc, yc), expressed in mas), disk PA (deg), and inclination (i, ◦). measured cADI LOCI a (mas) 1077.2+ 1070.2+ −4.1 −5.4 b (mas) 270.5+ 266.9+ −2.4 −3.0 xc (mas) −6.5 22.8 + 19.5+ −6.9 yc (mas) −5.7 7.6+ 10.8+ −5.7 PA (deg) 26.85+ 26.60+ −0.15 −0.32 i (deg) 75.28+ 75.58+ −0.15 −0.3 58 66 73 80 87 95 102 109 116 124 131 138 146 Projected distance to the star (UA) 2.0 1.5 1.0 0.5 U D A n i s s e n t h g i r B 0.0 0.8 CADI NE CADI SW mCADI NE mCADI SW 1.0 1.8 Projected distance to the star (arcsec) 1.2 1.4 1.6 58 2.0 66 73 80 87 Projected distance to the star (UA) 109 116 124 131 138 146 95 102 1.5 1.0 0.5 U D A n i s s e n t h g i r B 0.0 0.8 2.0 loci NE loci SW mLOCI NE mLOCI SW 1.0 1.2 1.6 Projected distance to the star (arcsec) 1.4 1.8 2.0 Fig. 7. Observed SBD for the HR4796A disk at L' along the major axis (after a binning of 5 pixels perpendicular to the major axis): cADI and mcADI reductions (left); LOCI and mLOCI reductions (right). We note that the peaks of the SBD on the NE and SW do not coincide which is due to the shift of the ellipse center along the major axis. 65.5 69.2 72.8 76.4 80.1 83.7 87.4 91.0 94.6 98.3 101.9 105.6 109.2 Projected distance to the star (UA) 65.5 68.4 71.3 74.4 77.7 81.1 84.6 88.3 92.1 96.1 100.3 104.6 109.2 Projected distance to the star (UA) 1.000 0.100 U D A n i s s e n 1.000 0.100 U D A n i s s e n t h g i r B 0.010 0.001 cadi NE cadi SW cadi fakedisk SE cadi fakedisk NW cadi fakedisk blowout SE cadi fakedisk blowout NW t h g i r B 0.010 0.001 loci NE loci SW loci fakedisk SE loci fakedisk NW 0.90 0.95 1.00 1.05 1.10 1.15 1.20 1.25 1.30 1.35 1.40 1.45 1.50 Projected distance to the star (arcsec) 0.90 0.94 0.98 1.02 1.07 1.11 1.16 1.21 1.27 1.32 1.38 1.44 1.50 Projected distance to the star (arcsec) Fig. 9. Left: simulated radial brightness distributions for the HR4796ASD disk (green) and HR4796ASBD disk with blowout (red) at L' along the major axis (log scale) once inserted in the data cube and after cADI reduction has been applied. For comparison, observed SBD (black). Right: simulated radial brightness distributions for the HR4796ASD disk (green) at L' along the major axis (log scale) once inserted in the data cube and after LOCI reduction has been applied. For comparison, observed SBD (black). We did not account for the presence of the disk in the model fitted. We considered that it did not affect the visibilities (because very faint), and did not affect the closure phase (because quasi point-symmetric). Nevertheless, it is not impossible that some of the structures present in Fig. 12 may be caused by a second order effect of the disk on the closure phase. Thus, neglecting the influence of the disk means that we are conservative on the detection limit map. Given these values, and assuming V-L' = 0 for this A0-type star, and an age of 8 Myr for the system, we derive the 2D de- tection limits expressed in Jupiter masses, using DUSTY mod- els (Fig. 12; right). At a separation of about 80 mas (6 AU), we exclude the presence of companions with masses larger than 29 MJup . At a separation of 150 mas, the limit becomes M = 40 MJup (DUSTY). In both cases, COND03 models give simi- lar limits within 1 MJup. At 60 mas, the detection limit is M = 50 MJup (DUSTY) and would be 44 MJup with COND03. Such values fall in the mass range of brown dwarfs and represent un- precedented mass limits for this range of separations. 5.2. ADIdetectionlimits We computed the detection limits using the data obtained on April 6th and 7th, with different reduction methods. To estimate them, we took into account the flux losses due to the ADI re- ductions, either injecting fake planets in cubes of empty frames 9 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 0 2 4 6 8 50 100 150 200 250 300 Fig. 10. Left: 2D map of 5-σ detection limit of point-like structures towards HR4796A, using SAM data. Isocontours for ∆mag = 3.5, 4.5, 5.5, and 6.5. Right: detection limits expressed in Jupiter masses using COND03 models; isocontours for masses of 20, 40, 60, 80 and 100 MJup. In both images, North is up and East to the left. ) g a m ( σ 5 − t s a r t n o C 0 4 6 8 10 12 0.0 11 22 33 44 55 66 76 87 98 109 120 131 Projected distance to the star (AU) SAM NE SAM SW cADI NE cADI SW rADI NE rADI SW LOCI NE LOCI SW 0.5 1.0 Distance to the star (arcsec) 1.5 Fig. 11. 5-σ detection limit of point-like structures around HR4796A, along the major axis in the NE and SW directions using SAM and ADI data. wide box along any given PA. We checked the obtained detection limits by inserting fake planets with fluxes corresponding to the 5σ limit at different separations, and processed again the data, and mesured the resulting S/N ratio on the planets. The S/N ra- tio were close (or sometimes slightly larger than 5) which shows that our limits are properly estimated. The 2D-detection limits are shown in Figure 12 for cADI, expressed either in contrasts or in masses, using the COND03 models (Baraffe et al. 2003) or BT-settl models (Allard et al. 2011) and assuming an age of 8 Myr. Similar (not better) limits were obtained with rADI and LOCI. To check the robustness of the detection limits obtained, we injected fake planets with fluxes corresponding to the 5σ level as well as the fake disks and processed the data cubes as before. The resulting images (see Fig. 13) revealed the planets with at least a 5σ level. 65.5 69.2 72.8 76.4 80.1 83.7 87.4 91.0 94.6 98.3 101.9 105.6 109.2 Projected distance to the star (AU) 1.00 0.10 U D A n i s s e n h g i r t B 0.01 0.9 1.0 unconv. disk conv. disk SE conv. disk NW cADI SE cADI NW LOCI NE LOCI SW 1.1 1.2 1.3 1.4 1.5 Projected distance to the star (arcsec) Fig. 8. Top: from left to right: simulated disk, simulated disk af- ter convolution, after cADI, and after LOCI reductions. Bottom: corresponding SBDs, showing the evolution of the profile after different steps of reduction. See text. obtained with similar FoV rotation for the cADI procedure, or injecting fake planets in the real data cubes in the case of the LOCI procedure. The noise was estimated using a sliding 9 pixel 10 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Table 4. A few detection limits along the semi-major axis of HR4796A (NE side). First lines (top) values are derived from SAM data, using DUSTY models (note that COND models agree within 1-2 MJup in most cases). The other values (bot- tom) are derived from ADI data, using COND03 or BT-SETTLE models. Sep (") 0.06" 0.08" 0.15" 0.3" 0.3" 0.5" 0.8" 1" 1.5" Sep (AU) 4.5 6 11 22 22 36.5 58 73 110 ∆mag (5σ) Mass (MJup ) 50 (DUSTY) 6.9 29 (DUSTY/COND) 7.3 40 (DUSTY/COND) 7.0 7.2 32 (DUSTY/COND) 7 (BTSETTLE, COND) 8.2 3.5 (COND) 10.1 3 (COND) 10.34 3 (COND) 10.23 11.22 2 (COND) Fig. 13. Fake planets inserted at 0.3, 0.5, 0.8, and 1" on both sides of the disk, with fluxes corresponding to the 5-σ detection levels, and reduced with cADI and LOCI. The 1-D limits along the major axis, at a PA of 26 degrees (NE side of the disk) are showed in Figure 11. Similar values are obtained in the SW direction. A few values expressed in jovian masses are given in Table 4. The detection limits are better than the SAM ones further than ≃ 0.25-0.3", with a value of about 7.5 MJup at 0.25-0.3"; they are below 3.5 MJup for separations in the range 0.5-1", and well below further than 1.5". Alltogether, these are to our knowledge the best detection limits obtained in the close surroundings of HR4796A. 6. Companions around HR4796B HR4796B has the following magnitudes: V = 13.3, H = 8.5, K = 8.3. Using the observed contrast between HR4796A and HR4796B (2.6 mag) on the present data, we find L'≃ 8.4 for HR4796B, hence L'abs ≃ 4.1. This value is in agreement with the Lyon's group model (Baraffe et al. 2003), which, given the near-IR colors, predicts an absolute L' magnitude of 3.8. We give in Figure14 the 2D map of the detection limits (5σ), both in terms of contrast magnitudes and Jupiter masses. At 0.3", masses as low as 2 MJup could be detected, and at 0.5", the detection limit is below 1MJup . 7. Discussion 7.1. Theinnerdisksharpedge One of the most remarkable features of the HR4796 disk is cer- tainly the offset of the disk center with respect to the star (also observed in the case of HD141569, and Fomalhaut). Two ex- planations are 1) the presence of a close, fainter companion (in such a case, the disk would be a circumbinary disk and orbit around the binary center of mass), and 2) the presence of a com- panion close to the disk inner edge on an eccentric orbit that induces a forced eccentricity to the disk ring by secular grav- itational interaction, an explanation which was proposed to ex- plain the eccentricity of the Fomalhaut disk (Quillen et al. 2006, Kalas et al. 2005, Kalas et al. 2008, Chiang et al. 2009). We first investigate whether this offset could be due to the presence of a close companion. In such a case, the ellipse cen- ter would mark the center of mass of the binary system. Using the center of mass definition, it appears that the mass of a body necessary to shift the center of mass at the observed position of the ellipse center would be much larger than the detection limit obtained with SAM between 40 and 400 mas or between 400 mas and 1 arcsec (ring position) with the ADI data. A compan- ion located between 23 and 40 mas would have a mass larger or comparable to that of HR4696A; such a scenario must be ex- cluded as under such conditions, the photometric center of the system would also be shifted. Then, the most plausible explana- tion to the offset is a light eccentric planet close to the inner edge of the disk. We now try to use the detection limits found in this pa- per to constrain the properties of an inner planet that could be responsible for the steep inner edge observed with HST/STIS data, which, conversely to ADI data, is not impacted by ADI reduction effects. Wisdom 1980 showed that in the case of a planet and particles on circular orbits, we have the relation δa/a = 1.3.(Mp/Ms)2/7 where Mp and Ms are the planet and star masses, a is the orbital radius of the planet and δ a the distance between the planet and the disk inner edge. Hence if a planet sculpts the inner edge, its mass and distance from the inner edge must satisfy this relation. Assuming an inner edge located at 77 AU, we can derive the mass of the planet necessary to produce this sharp edge, as a function of its distance to the edge, and test whether such a planet would have been detected or not. This is done in Figure15 where we show the region, inside the yellow ellipse, that, given the present detection limits, have to be ex- cluded. Hence the only possible location of the planets respon- sible for the inner edge is between the yellow ellipse and the red one (which traces the inner edge of the disk). We see that along the major axis, only the planets closest to the inner edge (less than ≃ 10 AU) remain out of the present detection capabilities. Hence if a planet is responsible for the inner edge sculpting and is located along or close to the major axis, then it should be a low mass planet, and located further than 63 AU, ie within about 15 AU from the edge. Along the minor axis, due to the projec- tion effects, the presence of planets is much less constrained: only planets at more than 26 AU from the edge would have been detected. The previous constrains were obtained assuming the planet and the perturbated bodies are both on circular orbits. The ac- tual disk eccentricity beeing very small, about 0.02, this assump- tion is reasonable. As an exercice, we investigate the impact of a higher eccentricity, using the results of Mustill and Wyatt 2012, who revisited this scenario, assuming the perturbating bodies were on an eccentric orbit; the relation becomes : δ a/a = 1.8.e1/5.(Mp/Ms)1/5. With the same reasonning, and assuming 11 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 6 7 8 9 10 11 12 13 5 10 15 20 Fig. 12. 2D map of 5-sigma detection limit of point-like structures towards HR4796A, using all available data (CADI). Left: ex- pressed in contrasts (isocontours from 7 to 11 mag., steps of 1 unit). Right: translated in masses using COND03 models (isocontours from 2 to 10 MJup , steps of 1 MJup ). 4 6 8 10 12 0 5 10 15 Fig. 14. 2D map of 5-sigma detection limit of point-like structures around HR4796B (April 6th), expressed in contrast (Left), with contour-levels of 8 to 11 mag (step =1 mag); and in Mass (Right), with contour-levels of 1 to 6 MJup (step =1 MJup). an eccentricity of 0.1, we provide in Figure15 the possible lo- cations of a planet responsible for the inner edge, with the same color conventions as in the circular case. Again, comparison with Figure12 shows that if a planet was responsible for the inner edge sculpting, and located along or close to the semi-major axis, then it would have to be located less than ≃ 25 AU from the edge of the disk. The location of planets along or close to the minor axis would not be significantly constrained. This case, even though not adapted to the present case as the disk eccen- tricity is very small, illustrates the impact of this parameter on the planet detection capabilities. 7.2. Theouterdisksharpedge Another striking feature of the system is the very steep disk outer edge. Radiation pressure from A-type stars in- duces surface brightness distributions in the outer part of disks with slopes of -3.5 to -5, depending on the assump- tions related to the production laws of small grains (see instance Lecavelier Des Etangs and Vidal-Madjar, 1996, for Thebault and Wu 2008 and ref. In particular, Thebault and Wu 2008 modeled the outer parts of collision rings with an initially steep outer edge, following the motion of the small grains produced through collisions and submitted there-in). 12 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 SBD profile beyond the main dust ring, so that it remains to be see whether slopes in the -10 range are possible. Finally, it is worth noting that so far no circumstellar gas has been detected either in atomic species through absorption spectroscopy (but the system being inclined, the non detection is not a strong con- straint) or molecular species, either CO (Liseau, 1999) or H2 (N(H2)≤ 1015cm−2 Martin-Zaiedi et al. 2008). In any case, such a scenario can be tested in the forthcoming years with high an- gular resolution observations on a wide range of wavelengths. We finally study the possibility that the outer disk is sculpted by massive bodies. The first candidate we might think of is HR4796B. Thebault et al. 2010 investigated the possibility that the disk could be sculpted by HR4796B, if orbiting on a rather eccentric orbit (e≥ 0.45) but again showed that, even under such conditions the outer profile would not be so steep. This is mainly because the companion star is not able to dynamically remove small grains from the outer regions at a pace that can compen- sate for their steady collisional production in the parent body ring. An alternative explanation could be the presence of a close, unseen outer planet. We investigate this scenario using the new code developed by Thebault (2012) to study perturbed collision- ally active debris discs. The code computes the motion of plan- etesimals submitted to the gravitational perturbation of a planet; and follows the evolution of small dust realeased through colli- sions among the planetesimals and submitted to radiation pres- sure and Poynting-Robertson effect (note that in the present case, radiation pressure largely drives the grains dynamics once pro- duced). The configuration we consider is a narrow ring of large parent bodies, a birth ring of width ∼ 8AU centered on 71 AU. The collisional production and destruction rates of small grains (those which contribute to the luminosity beyond the main ring) in this parent body ring is parameterized by the average verti- cal optical depth within the main ring, taken to be τ = 5 × 10−3 4. The resulting SBD (case face-on) is derived assuming grey scattering. It can be directly compared to the curve presented in Figure 6 of Schneider et al (2009) paper (de-projected curve5). For the perturbing planet's mass, we consider 3 different val- ues: 8MJup, 5MJup and 3MJup, which are consistent with the constraints imposed by our observational non-detection. Note that 8MJup is only marginally possible in a very narrow region along the disc's semi-minor axis, but we have to keep in mind that the detection limits are derived from masses-brightness re- lationships that are debated at young ages. We assume a circu- lar orbit for the planet (the most favourable case for cleaning out the region beyond the ring, see Thebault (2012)) and place it as close as possible to the parent body ring, i.e., so that the outer edge of the observed ring (around 75 AU) corresponds to the outer limit for orbital stability imposed by planetary pertur- bations. This places the planet at a distance to the central star comprised between 92 and ∼ 99 AU depending on its mass. In Figure 16, we show the SBD obtained for such a config- uration. Note that, for each planet mass, we are not showing an azimutal average but the "best" radial cut, i.e. the one that gives the closest match to the deprojected NE side SBD obtained in Figure 6 of Schneider et al. 2009. As can be clearly seen, the 8MJup case provides a good fit to the observed profile: the max- imum of the SBD roughly corresponds to the outer edge of the parent body disk and is followed by a very sharp brightness de- crease, with a slope ≤ -10 between 75 and 95 AU, i.e. between 4 for more detail on the procedure, see Thebault (2012) 5 the FWHM of the STIS PSF being quite narrow compared to the disk width, its impact is very limited. 13 Fig. 15. Possible remaining locations of a planet inner to the disk that could produce an inner sharp edge at 77 AU given the present detections limits (Left: case of a circular orbit; Right: case of an elliptical orbit with e=0.1). The possible locations are restricted to the region between the yellow curve and the red ellipse that mimics the disk itself. North is up, East to the left. to radiation pressure and showed that, as already proposed, the profile of the resulting SBD in the outer part of the disk followed a r−3.5 law. They showed that unless the disks are extremely and unrealistically dense, and prevent the small grains from escaping, the disks have to be extremely "cold" (with an average free eccentricity of ≤ 0.0035) to explain an outer power law of r−6 (which was the value adopted at this time for the HR4796 profile). AO data suggest that the situation could be even more radical with an even steeper outer edge than previously thought. Also, if confirmed, the fact that we possibly find at 3.8 µm a disk width different from that found in the optical (0.2-1µm) with STIS would argue against an extremely dense disk, as, in such a case, large bodies and small grains would be in the same regions. The cold disk scenario can nonetheless be also prob- lematic as, in such a case small grains should be underabundant (Thebault and Wu 2008) and the optical/near-IR fluxes would be produced by large particules (typ. sizes 50 µm). We should expect then a color index different from that observed (see Debes et al. 2008). The latter rather predict that the scattered light flux is dominated by 1.4 µm dust, which seems difficult to explain within the dynamically-cold-disk scenario. Could gas be responsible for the observed steep outer edge? Takeuchi and Artymowicz 2001 investigated the impact of the gas in such debris disks, and showed that even small amount of gas, 1- a few Earth masses, could partly balance the effects of gravitation, radiation pressure and Poynting-Robertson drag and alter the grains dynamics differentially, and lead to grains spatial distributions, that, depending on the grain sizes, could be differ- ent from those expected in a disk-free gas. Under such processes, the gas could be responsible for ring-like structures at distances depending on the dust size considered. In their attempt to investi- gate disks roughly similar to HR4796 and HD141569, assuming 1 MEarth gas, they showed that, conversely to large grains which occupy the whole gas disk, grains with sizes (≃ 10-200 µm) tend to concentrate in the outer gas disk, where the gaseous density sharply decreases. Hence these grains would form a narrow ring which position traces the change in the radial distribution the gaseous disk. Under this a priori attractive scenario, grains with sizes 1-10 µm would still be blown away. The main problem with this hypothesis is that it requires a gas disk with a relatively sharp outer edge, and thus an explanation for such an edge. Another is- sue is that Takeuchi and Artymowicz 2001 did not explore the A.-M. Lagrange et al.: An insight in the surroundings of HR4796 brightnesses of 1 and 0.1, a range corresponding approximately to the dynamical range accessible to the available images. This is significantly steeper than the one that would be expected if no planet was present (-3.5 according to Thebault and Wu 2008) and is fully compatible with the observed sharp luminosity de- crease. Longwards 95 AU, the flux level is lower (≤ 0.1 ADU) and the SBD is flatter (slope ≃ -3.8). We also note a plateau inside the parent body ring at a level of ∼ 0.2, due to both the inward drift of small grains because of the Poynting-Robertson effect and to the dynamical injection of particles after close encounters with the planet. Of course, not too much significance should be given to the SBD obtained inwards of the disc since our simulations (focused on the outer regions) do not consider any inner planet shaping the inner edge of the disc. Nevertheless, they show that, should "something" have truncated the disc at around 67 AU in the past, then the effect of one external planet on such a truncated disc could lead to an SBD compatible with observations in the inner regions. For the 5MJup case, the fit of the observed SBD is slightly degraded, but mostly in the region beyond 90 AU where flux levels are close to the 0.1 threshold. For the 3MJup case, however, the fit gets very poor for almost the whole outer region (keeping in mind that we are here showing the best radial cut). We conclude that a 8MJup planet located on a circular orbit at ∼ 25 AU from the main ring provides a satisfying fit (especially considering the non-negligeable uncertainties regarding flux val- ues far from the main ring) to the observed SBD. According to the derived detection limits, such a massive planet would have been detected almost everywhere except in a very narrow region along the disc's semi-minor axis; however, we remind the un- certainties inherent to the models used to link planet masses and luminosities as a function of the system's age. In any case, even a less massive perturber of e.g. 5MJup would still give an accept- able fit of the observed luminosity profile. The external planet scenario thus seems the most likely one for shaping the outer regions of the disc. Of course, these results are still preliminary and should be taken with caution. A more thorough numerical investigation should be carried out, exploring a much wider pa- rameter space for planet masses and orbit, as well as deriving other outputs that can be compared to observations, such as 2-D synthetic images. Such a large scale numerical study exceeds the scope of the present work and will be the purpose of a forthcom- ing paper. Note also that the more general issue of how planets shape collisionally active debris disks will be thoroughly inves- tigated in a forthcoming paper (Thebault, 2012b, in prep.). 8. Summary and future prospects In this paper, we have provided the first high-resolution images of the HR4796A disk at L' band. They allow us to see a narrow disk at almost all PA. As the technics used, Angular Differential Imaging is expected to impact the final disk shape and appeare- ance, we have developped simulations to investigate quantita- tively the impact of the reduction procedures on the disk pa- rameters. We conclude that the information on the inner part of the disk is significantly impacted, and that the procedure may in some cases (depending on the amplitude of rotation of the field of view), produce important artefacts. This is specially true for LOCI reduction, while classical ADI affects the data to a lower extent. We showed in particular that the streamers detected by Thalmann et al. 2011 at the outer edge of the disk are probably due to such artefacts. Using both ADI and SAM data, we have derived unprece- dented lower limits to the presence of planets/companions down 14 to 25 mas from the star. The present data allowed then to put first interesting constrains on the location of the possible planet that could produce the inner edge of the disk. We showed that the planet responsible for the inner edge must be closer than 15 AU from the ring if located along or close to the semi-major axis. The forthcoming high dynamics instruments such as SPHERE on the VLT and GPI on GEMINI will allow to test this hypothe- sis with much more accuracy, and be able to actually detect this planet in most cases. We have discussed several hypotheses to explain the sharp outer edge of the disk: gaseous disk, dynamically cold disk, planet on the outer edge. Using detailed simulations, we showed that a planet located outside the planetesimal ring could nicely reproduce the STIS data. Further simulations will help to better constrain the planet and parent bodies characteristics. In any case, this work shows how disks characteristics can help constraining possible planet properties. A very important information can be brought by the dependance of the disk prop- erties (ring width, SBD) as a function of wavelength. Resolved images in the future will be crucial to further understand this system. Acknowledgements. We acknowledge financial support from the French Programme National de Plan´etologie (PNP, INSU). We also acknowledge sup- port from the French National Research Agency (ANR) through project grant ANR10-BLANC0504-01 and ANR-2010 BLAN-0505-01 (EXOZODI). We also thank C. Marois, B. McIntosch and R. Galicher for discussing their L' Keck data with us. We also thank the anonymous referee for his/her helpful comments. References Allard, F., Homeier, D., Freytag, B2011, Philosophy Transaction, AarXiv1112.3591A Augereau, J. C., Lagrange, A. M., Mouillet, D., Papaloizou, J. C. B., Grorod, P. A., 1999, A&A, 348, 557 Augereau, J.-C., Nelson, R. P., Lagrange, A.M., et al, 2001, A&A, 370, 447 Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, A&A, 402, 701 Boccaletti, A, Augereau, J.-C., Baudoz, P. et al, 2009, A&A, 495, 523 Boccaletti, A, Augereau, J.C., Lagrange, A.M., et al., 2011, A&A, to be submit- ted Buenzli, E., Thalmann, C., Vigan, A., et al, 2010, A&A, 524, L1 Chauvin G., Lagrange, A.M., Beust, H., et al, 2011, A&A, arXiv1202.2655 in press, Chiang, E., Kite, E., Kalas, P. et al, 2009, ApJ, 693, 734 Currie, T., Thalmann, T., Matsumura, S. et al, 2011, ApJ, 736, L33 Debes, J.H., Weinberger, A.J., Schneider, G., 2008, ApJ, 673, L191 Golimovski, D. A., Ardila, D. R., Krist, J. E., et al, 2006, AJ, 131, 3109 Hinkley, S., Oppenheimer, B. R., Soummer, R. et al, 2009, ApJ, 701, 804 Jayawardhana, R., Fisher, R. S., Hartmann, L., et al, 1998, ApJ, 503, 79 Jura, M., 1991, ApJ, 383, L79 Jura M., Zuckerman B., Becklin E.E., Smith R.C., 1993, ApJ, 418, L37 Kalas, P., Graham, J. R., Clampin, M, 2005, Nature, 435, 1067 Kalas, P., Graham, J. R., Chiang, E., et al, 2008, Science, 322, 1345 Kastner, J., Zuckerman, B., Bessell, M., 2008, A&A, 491, 829 Koerner, D. W.; Ressler, M. E.; Werner, M. W.; Backman, D. E., 1998, ApJ, 503, L83 Tuthill, P., Lacour, S., Amico, P., et al, 2010, SPIE, 7735, 56 Lacour, S., Tuthill, P., Amico, P., et al, 2011, A&A, 532, 72 Lafreni`ere, D., Marois, C., Doyon, R. et al, 2007, ApJ, 660, 770 Lafreni`ere, D., Jayawardhana, R., van Kerkwijk, M. H., 2010, ApJ, 719, L497 Lagrange, A.M., Gratadour, D;, Chauvin, G. et al, 2009, A&A, 493, L21 Lagrange, A.M., Bonnefoy, M., Chauvin, G., et al, 2010, Science, 329, L57 Lagrange, A.M, Boccaletti, A., Milli, J., et al, 2011, A&A, in press, arXiv1202.2578 Lecavelier Des Etangs, A., Vidal-Madjar, A., and Ferlet, R., 1996, A&A, 307, 542 Lebreton, J., Augereau, J.-C., Thi, W.-F., et al, 2012, A&A, arXiv:1112.3398 in press; Lenzen, R., Hartung, M., Brandner, W., et al. 2003, SPIE, 4841, 944 Li, A., and Lunine, J.I., 2003, ApJ, 590, L368 Liseau, R., 1999, A&A, 348, 133 Martin-Zaiedi C., Deleuil M., Le Bourlot J. et al, 2008, A&A, 484, 225 A.-M. Lagrange et al.: An insight in the surroundings of HR4796 Fig. 16. Synthetic surface brightness profiles obtained, using Thebault (2012)'s numerical model, for 3 different masses of a putative perturbing outer planet: 8MJup, 5MJup and 3MJup. For each case, the planet has a circular orbit and is placed as close as possible to the main ring of large parent bodies in order to truncate it at about 75AU without destroying it (see text for more details). The observed profile, derived from Schneider et al. 2009, is shown for comparison. For each planet mass, we show the radial cut (i.e., for one position angle along the disc) that provides the best fit to this observed profile. The horizontal line delineates approximately the part (above this line) of the SBD accessible to the observations assuming a dynamical range of 10. McCaughrean, M. J., and Stauffer, J. R., 1994, AJ, 108, 1382 Marois, C., Lafreniere, D., Doyon, R., Macintosh, B. and Nadeau, D., 2006, ApJ, 641, 556 Marois, C., MacIntosh, B., Barman, T., Zuckerman, B., Inseok, S., Patience, J., Lafreniere, D., Doyon, R. 2008, Science, 322, 1348 Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., Barman, T., 2010, Nature, 468, 1080 Mayor, Michel; Queloz, Didier, 1995, Nature, 378, 355 Mouillet D.; Lagrange A.-M.; Beuzit J.-L.; Renaud N. 1997, A&A, 324, 1083 Mustill, A.J., and Wyatt, M.C., 2012, MNRAS, 419, 3074 Quillen, A., 2006, MNRAS, 372, L14 Tuthill, P., Lacour, S., Amico, P., et al, 2010, SPIE, 7735, 1O Rodriguez, David R.; Zuckerman, B., 2011, ApJ, in press, 2011arXiv1111.5618 Rousset, G., Lacombe, F., Puget, P., et al. 2003, SPIE 4839, 140 Schneider, G., Smith, B.A., BBecklin, E.E., 1999, ApJ, 513, L127 Schneider, G., Weinberger, A.J., Becklin, E.E., Debes, J.H. and Smith, B.A. , 2009, AJ, 137, 53 Smith, B. A., Terrile, R. J., 1984, Science, 226, 1421 Stauffer J.R., Hartmann L.W.; Barrado Y Navascues, D., 1995, ApJ, 454, 910 Takeuchi, T., and Artymowicz, P., 2001, ApJ, 557, 990 Thalmann, C., Janson, M., Buenzli, E., et al, 2011, ApJ, 743, 6 Thebault P., Wu, Y., 2008, A&A, 481, 713 Thebault P., Marzari, F.; Augereau, J.-C., 2010, A&A, 524, 13 Thebault, P., 2012, A&A, 537, 65 Van Leeuwen, F., 2007, A&A, 474, 653 Wahhaj, Z., Koerner, D. W., Backman, D. E., et al, 2005, ApJ, 618, 385 Wisdom, J., 1980, AJ, 85, 1122 Wyatt, M.C., Dermott, S.F., Telesco, C.M., 1999, ApJ, 527, 918 15
1103.4255	1	1103	2011-03-22T12:37:57	Dependence of GCRs influx on the Solar North-South Asymmetry	[ "astro-ph.EP" ]	We investigate the dependence of the amount of the observed galactic cosmic ray (GCR) influx on the solar North-South asymmetry using the neutron count rates obtained from four stations and sunspot data in archives spanning six solar cycles from 1953 to 2008. We find that the observed GCR influxes at Moscow, Kiel, Climax and Huancayo stations are more suppressed when the solar activity in the southern hemisphere is dominant compared with when the solar activity in the northern hemisphere is dominant. Its reduction rates at four stations are all larger than those of the suppression due to other factors including the solar polarity effect on the GCR influx. We perform the student's t-test to see how significant these suppressions are. It is found that suppressions due to the solar North-South asymmetry as well as the solar polarity are significant and yet the suppressions associated with the former are larger and more significant.	astro-ph.EP	astro-ph	Dependence of GCRs influx on the Solar North-South Asymmetry Il-Hyun Choa,b, Young-Sil Kwaka, Heon-Young Changc,∗, Kyung-Suk Choa, Young-Deuk Parka, and Ho-Sung Choia,b aKorea Astronomy and Space Science Institute, Daejeon, 305-348, Korea bUniversity of Science and Technology, Daejeon, 305-348, Korea cDepartment of Astronomy and Atmospheric Sciences, Kyungpook National University, Daegu, 702-701, Korea 1 2 3 4 5 6 7 8 9 Abstract We investigate the dependence of the amount of the observed galactic cosmic ray (GCR) influx on the solar North-South asymmetry using the neutron count rates obtained from four stations and sunspot data in archives spanning six solar cycles from 1953 to 2008. We find that the observed GCR influxes at Moscow, Kiel, Climax and Huancayo stations are more suppressed when the solar activity in the southern hemisphere is dominant compared with when the solar activity in the northern hemisphere is dominant. Its reduction rates at four stations are all larger than those of the suppression due to other factors including the solar polarity effect on the GCR influx. We perform the student's t-test to see how significant these suppressions are. It is found that suppressions due to the solar North-South asymmetry as well as the solar polarity are significant and yet the suppressions associated with the former are larger and more significant. Keywords: galactic cosmic rays, solar north-south asymmetry, climate change 10 11 Preprint submitted to Journal of Atmosphere and Solar-Terrestrial Physics June 8, 2018 ∗corresponding author Email address: [email protected] (Heon-Young Chang) 12 1. Introduction 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 It is well known that the level of galactic cosmic ray (GCR) influx to the terrestrial atmosphere is variable in time. The observed GCR influx is anti-correlated with the observed sunspot number[Forbush 1954; Usoskin 2008]. The higher level of the solar activity induces a stronger magnetic field strength in the heliosphere, and consequently reduces the amount of GCRs observed in the vicinity of the Earth. For instance, during the increasing phase of the solar cycle, the observed level of GCRs near the Earth has been gradually reduced. A long-term effect is also studied by Lockwood et al. (1999) using several geo- and solar-magnetic indices from 1964 to 1995 cor- responding to solar cycles from 20 to 22. They showed that measurements of the near-Earth interplanetary magnetic field reveal that the total magnetic flux leaving the Sun has risen by a factor of 1.4 since 1964, which interest- ingly reflects the increase of solar source magnetic flux, which also indirectly implies the decrease of GCR influx. In addition to the dependence of the GCR flux on the strength of the solar activity, the heliospheric current sheet is asymmetric with respect to the heli- ographic equatorial plane [Krymsky et al. 2009; Hiltula and Mursula 2005; Munakata et al. 1983]. For instance, a southward offset of the current sheet has been reported while ULYSSES monitored the cosmic-ray proton inten- sity from September 1994 to May 1995[Smith et al. 2000; Simpson et al. 1996]. We note that the observing epoch corresponds to the period when the solar southern hemisphere is magnetically dominant. As one of the solar northern- and southern-hemispheres is dominant in a given solar activity, the magnetic field of the heliosphere near the Sun is likely to follow a similar 2 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 behavior. Zhang et al. [2002] have also reported that the electron temper- ature and magnetic field in the solar southern coronal hole are higher and stronger than those in the northern coronal hole during the period from 1992 to 1997. This is also the time interval when the solar activity in the southern hemisphere is dominant, as we mentioned. In this paper, we investigate whether the level of the GCR influx shows any dependence associated with the solar North-South asymmetry using the neutron count rates obtained from 4 stations and sunspot data archived span- ning six solar cycles from 1953 to 2008. This is an interesting question not only in itself but also because of the following reason. According to what is mentioned above one may expect that the average location of the cur- rent sheet around the Earth can be quasi-periodically varying with the solar North-South asymmetry. Thus, the observed GCR influx may be changing in magnitude not only with the solar cycle but also with the solar North-South asymmetry. 52 2. Data and Results 53 54 55 56 57 58 The GCR data are recorded at ground-based neutron monitors, which detect variations in the low energy part of the primary cosmic ray spectrum. The lowest energy that can be detected at the top of the atmosphere depends on the geomagnetic latitude, and ranges from 0.01 GeV at stations near the geomagnetic poles to about 15 GeV near the geomagnetic equator. The National Geophysical Data Center(NGDC) tabulates the daily average GCR 3 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 count rate per hour for a number of stations. From the website1, we have taken the yearly mean of the GCR influx observed at 4 stations, where GCR data sets having a time-span longer than 20 years in both prior and post to ∼ 1980 are available: Moscow(55◦N, 37◦E, 2.42GV), Kiel(54◦N, 10◦E, 2.32GV), Climax(39◦N, 106◦W, 2.99GV) and Huancayo(12◦S, 75◦W, 12.92GV). We have adopted the yearly averaged wolf number (SN) from 1953 to 2008 provided by Solar Influences Data Analysis Center (SIDC)2 for the GCR-SN plot. We have also taken sunspot area data for the calculation of the solar North-South asymmetry from the NASA website3, which is regularly updated by Hathaway at the Marshall Space Flight Center(MSFC). Text files are there available containing the yearly averages of the daily sunspot areas (in units of millionths of a hemisphere) in the northern and solar southern hemispheres separately. In Figure 1 we show the running average of the solar North-South asym- metry, which is defined as the difference of the sunspot area between so- lar northern and southern hemispheres normalized by their sum, (AN − AS)/(AN + AS), for the period of 1953 − 2008 as a function of time. In this plot the length of the moving window is 7 years. When values of the solar North-South asymmetry is positive the northern solar hemisphere is more active, and vice versa. Note that the dominance of the active hemi- sphere is reversed at ∼ 1980. That is, in general the solar northern and southern hemispheres are magnetically dominant before and after ∼ 1980, 1http://www.ngdc.noaa.gov/ 2http://sidc.oma.be/index.php3 3http : //solarscience.msfc.nasa.gov/greenwich.shtml 4 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 respectively. In Figure 2, we show yearly GCR fluxes at four neutron monitor stations as a function of the sunspot number. We divide the observed GCR flux and SN data sets into two sub-sets, respectively, separated at ∼ 1980 for the present analysis. Open and filled circles represent the observed GCR count rates in the sub-sets with time intervals before and after ∼ 1980, respectively. Straight lines in each panel result from the least square fits of two sub-sets. It should be noted that the averaged value of the observed GCR count rates represented by filled circles is smaller than that represented by open circles while the slopes of the best fits in each panel are more or less the same. The mean differences of GCR count rates between two sub-sets in Moscow, Kiel, Climax and Huancayo are 2.83±0.33%, 2.50±0.44%, 2.72±0.18% and 0.72±0.04%, respectively. What it means is that while the anticorrelation between the GCR flux and the sunspot number holds good regardless of ob- servation durations, the averaged GCR flux is apparently more reduced when the solar southern hemisphere is more active than the northern hemisphere. The relative difference shown is calculated from the best fit. We compute stu- dent's t and its probability for each station to see how significantly two mean values differ from. Student's t values for Moscow, Kiel, Climax and Huan- cayo are 9.50, 7,38, 6.28 and 5.21. Their false-alarm levels are all less than 0.01. It should be noted that the percentage variations in absolute counting rates during the solar cycle are different for different geomagnetic latitudes. The effect is smaller at lower latitudes in agreement with the shielding effect of the Earth's magnetic field on high-energy charged particles. Note that the vertical rigidity cut-offs for Moscow, Kiel, Climax, and Huancayo are 5 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 2.42GV, 2.32GV, 2.99GV, and 12.92GV, respectively. As a result, the mean value of data points in each sub-set is considered as significantly different. This conclusion is found to be insensitive to the epoch when we cut the data set. We have repeated our analysis with a few more sub-sets separated at different times rather than ∼ 1980, and obtained similar results. To compare our findings with results from other possible factors causing the suppression we have observed in the present work, we have carried on our analysis with the data sets divided by three other criteria. By doing so we may wish to demonstrate that the solar North-South asymmetry is a more critical factor in reducing the observed GCR flux than any other solar variables compared in this paper. In Figure 3, we show the GCR influx observed at the Moscow station as a function of the sunspot number, as an example. In panel (a), the data set is divided by the level of the solar activity. That is, open and filled circles represent the observed GCR count rates in the period of strong and weak solar cycles, respectively. Open circles belong to the cycles 19, 21 and 22. Filled circles belong to the cycles 20 and 23. Note that cycles 19 and 20 correspond to the period prior to ∼ 1980, and cycles 21, 22 and 23 to the period post to ∼ 1980. The solar cycle is defined such that it begins and ends at the solar minimum. A clear difference in the mean value cannot be seen. In panel (b), we show the observed GCR influx as a function of the sunspot number with respect to the individual solar cycle. Different symbols represent different solar cycles as indicated by the number. Note that the solar cycle is defined such that it begins and ends at the solar maximum unlike the case in panel (a). Having picked up the cycles 20 and 23, and compared them with other 6 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 solar cycles by averaging, the results can be contrasted with panel (a). By defining the solar cycle in this way one may also see the polarity effect. In panel (c) the polarity effect on the observed GCR influx is shown. Open and filled circles represent the opposite polarity. Two consecutive solar cycles have the opposite polarity. As reported earlier[Nagashima and Morishita 1980; Jokipii 1989; Sabbah and Duldig 2007], we confirm that the amount of the observed GCR influx varies as the polarity of the solar magnetic field is reversed. For comparison in magnitude, in panel (d) we show the result from the solar North-South asymmetry. It should be noted that the solar North-South asymmetry effect on the GCR influx is larger than any other effects. That is, the values of dI/I due to the solar North-South asymmetry and the polarity are 2.83±0.33 % and 1.36±0.50 %, respectively, as shown in Table 1. Similar conclusions can be drawn from other stations. It is also interesting, for instance, to note that the mean difference in the observed GCR influx is very large in cycles 20 and 23. These are the solar cycles whose signs both in the solar polarity and in the solar asymmetry are simultaneously opposite. These two factors apparently compete with each other so that they can be either amplified or compensated accordingly. Hence we conclude that both effects should be considered with due care, particularly when one wishes to employ one of these effects into a sophisticated algorithm to predict anything accurately. In Table 1, to summarize results, we list the difference in mean values and results of the student's t-test for sub-sets divided according to the so- lar polarity and the solar North-South asymmetry. In the first and second columns, neutron monitor stations and the criteria in dividing the data sub- 7 156 157 158 159 160 161 162 163 164 165 sets are shown. In the third column the relative difference in the mean GCR flux is given. In the fourth, fifth, and sixth columns, results of student's t-test are given. We conclude that the suppression associated with the solar North-South asymmetry is larger and more significant than that with the solar polarity. We only consider in the present study the average long-term effect of the solar North-South asymmetry on the GCR influx. Though the solar North-South asymmetry also has shorter-term periodicities than we re- ported in this paper[e.g., Chang 2007a, b, 2008, 2009], those short-terms may not be notable due to the other effect such as a fluctuating shape of current sheet. 166 3. Summary and Discussion 167 168 169 170 171 172 173 174 175 176 177 178 179 We investigate whether the level of the GCR influx shows any dependence associated with the solar North-South asymmetry using neutron count rate data obtained from 4 stations and sunspot data archived spanning six solar cycles from 1953 to 2008. We have found that the GCR influx observed at 4 neutron monitor stations are all suppressed when the solar southern hemisphere is more active. The amount of suppressions are 2.83±0.33% at Moscow, 2.50±0.44% at Kiel, 2.72±0.18% at Climax and 0.72±0.04% at Huancayo. These values are larger than the solar polarity effect observed at 4 stations (1.36±0.50% at Moscow, 1.67±0.57% at Kiel, 2.22±0.89% at Climax and 0.39±0.04% at Huancayo). We compute student's t values and its probability for each case, and come to conclusions that the solar North- South asymmetry effect on the GCR influx is larger than any other effects including the solar polarity effect. 8 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 Finally, we would like to add some speculation of its possible implica- tion on the terrestrial climate. As Svensmark and Friis-Christensen (1997) pointed out, the decrease in the observed GCR influx may cause the low terrestrial albedo through reducing low level cloud cover, and consequently lead to the higher heat content at the surface of the Earth. In a sense, therefore, the heat content at the surface of the Earth could be different from that deduced by the measured TSI alone[Goode et al. 2003]. We sus- pect that the net-radiative energy at the surface of the Earth should be determined by two parameters even in a simple climate model based on an assumption that greenhouse fraction is constant[e.g., Dutton 1995]. For instance, two parameters can be the total solar irradiance (TSI), which is modulated with the solar cycle[Scafetta 2009 and references therein], and the Earth's albedo, which Svensmark and Friis-Christensen (1997) show vary- ing with the observed GCR influx modulated. Hence one may wish to re- examine the temperature anomaly as a function of solar activity including, for example, the solar North-South asymmetry since the correlation between the GCR influx and the sunspot number is subject to the sign of the solar asymmetry[Cho et al. 2009; Georgieva 2002; Georgieva et al. 2005, 2007]. 198 4. Acknowledgments 199 200 201 202 203 We appreciate Katsuhide Marubashi for constructive comments on the GCRs distribution in the heliosphere and Kate Connors for careful reading the manuscript. We thank the anonymous referees for constructive comments which improve the original version of the manuscript. This work was sup- ported by the 'Development of Korean Space Weather Center' of KASI and 9 204 205 206 KASI basic research funds. HYC was supported by Basic Science Research Program through the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science and Technology (2009-0071263). 207 References 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 Chang, H.-Y., 2007a. Variation in North-South Asymmetry of Sunspot Area. Journal of Astronomy and Space Sciences 24 (2), 91–98 Chang, H.-Y., 2007b. A New Method for North-South Asymmetry of Sunspot Area Analysis 24 (4), 261–268 Chang, H.-Y., 2008. Stochastic properties in North-South asymmetry of sunspot area. New Astronomy 13 (4), 195-201 Chang, H.-Y., 2009. Periodicity of North-South asymmetry of sunspot area revisited: Cepstrum analysis. New Astronomy 14 (2), 133-138 Cho, I. H., Kwak, Y. S., Cho, K. S., Choi, H. S., and Chang, H.-Y., 2009. On the relation between the Sun and climate change with the solar North- South asymmetry. Journal of Astronomy and Space Sciences 26 (1), 25–30 Dutton, J. A., 1995. Scientific priorities in global change research. In: Asrar and D. J. Dokken(Ed), The state of earth science from space. American Institute of Physics. Forbush, S. E., 1954. World-Wide Cosmic-Ray Variations, 1937-1952. Journal of Geophysical Research 59 (4), 525-542 10 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 Georgieva, K., 2002. Long-term changes in atmospheric circulation, earth rotation rate and North-South solar asymmetry. Physic and Chemistry of the Earth 27 (6–8), 433–440 Georgieva, K., Kirov, B. and Bianchi, C., 2005. Long-term variations in the correlation between solar activity and climate. Memorie della Societa Astronomica Italiana 76 (4), 965 Georgieva, K., Kirov, B. Tonev, P., Guineva, V., and Atanasov, D., 2007. Long-term variations in the correlation between NAO and solar activity: The importance of north south solar activity asymmetry for atmospheric circulation. Advances in Space Research 40 (7), 1152–1166 Goode, P. R., Palle, E., Yurchyshyn, V., Qiu, J., Hickey, J., Rodriguez, P. Montanes, Chu, M.-C., Kolbe, E., Brown, C. T., and Koonin, S. E., 2003. Sunshine, earthshine and climate change: II. Solar origins of variations in the earth's albedo. Journal of the Korean Astronomical Sociery 36 (S1), S83-S91 Hiltula, T. and Mursula, K., 2005. Heliospheric current sheet North-South asymmetry since 1926. In: D. Danesy, S. Poedts, A. De Groof and J. Andries(Ed), Proceedings of the 11th European Solar Physics Meeting "The Dynamic Sun: Challenges for Theory and Observations", 600E, 123H Jokipii, J. R., 1989. The physics of cosmic-ray modulation. Advanced Space Research 9 (12),105–119 245 Krymsky, G. F., Krivoshapkin, P. A., Mamrukova, V. P., and Gerasimova, 11 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 S. K., 2009. North-south asymmetry of the heliosphere from cosmic-ray observations. Astronomy Letters 35 (5), 333–337 Lockwood, M., Stamper, R., and Wild, M. N., 1999. A doubleing of the Sun's coronal magnetic field during the past 100 years. Nature 399, 437–439 Munakata, Y., Hakamada, K., Mori, S., Yasue, S., and Ichinose, M., 1983. Cosmic ray North-South asymmetry related with the latitudinal angular distance of the earth from the heliospheric current sheet. Proceedings from the 18th International Cosmic Ray Conference 3, 358M Nagashima, K. and Morishita, I., 1980. Twenty-two year modulation of cos- mic rays associated with polarity reversal of polar magnetic field of the sun. Planetary and Space Science 28 (2), 195-205 Sabbah, I. and Duldig, M. L., 2007. Solar polarity dependence of cosmic ray power spectra observed with Mawson underground muon telescopes. Solar Physics 243 (2), 231-235 Scafetta, N., 2009. Total solar irradiance satellite composites and their phe- nomenological effect on climate. eprint Arxiv 0908.0792v1, submitted Simpson, J. A., Zhang, M., and Bame, S., 1996. A solar polar North-South asymmetry for cosmic-ray propagation on the heliosphere: The ULYSSES pole-to-pole rapid transit. Astrophysical Journal Letters 465, L69–L72 Smith, E. J., Jokippi, J. R., Kota, J., Lepping, R. P., and Szabo, A., 2000. Evidence of a North-South asymmetry in the heliosphere associated with a southward displacement of the heliospheric current sheet. The Astrophys- ical Journal 533 (2), 1084–1089 12 269 270 271 272 273 274 275 276 Svensmark, H., Friis-Christensen, E., 1997. Variation of cosmic ray flux and global cloud coverage - a missing link in solar-climate relationships. Journal of Atmosphere and Solar-Terrestrial Physics 59 (11), 1225 - 1232 Usoskin, I. G., 2008. A History of Solar Activity over Millennia. Living Re- views in Solar Physics 5 (3) Zhang, J., Woch, J., Solanki, S. K., and von Steiger, R., 2002. The sun at solar minimum: North-South asymmetry of the polar coronal holes. Geophysical Research Letters 29 (8), 77-1 13 Figure 1: The running average of the solar North-South asymmetry with a 7-year window. The vertical line is given to separate the data set into two sub-sets. Note that the solar northern- and southern-hemisphere is seen more active prior to ∼ 1980 and post to ∼ 1980, respectively. Error bars are included. 14 Figure 2: Observed GCR count rates measured at Moscow(a), Kiel(b), Climax(c) and Huancayo(d) as a function of sunspot numbers. Open and filled circles represent the obseved GCR influx in the period before and after 1980. Straight lines in each panel result from the least square fits of two sub-sets. 15 Figure 3: GCRs count rates measured at Moscow station as a function of sunspot numbers. Each straight line is the result of the least square fit. Different panels result from sub- sets divided by different criteria, i.e., (a) by the activity level of the solar cycles, (b) by the solar cycles, defined by the period between solar maxima, (c) by the solar magnetic polarity, (d) by the solar North-South asymmetry. 16 Table 1: Relative difference by the solar North-South asymmetry and polarity. Station Factors dI/I Student-t Deg. of freedom Prob.(%) Moscow Polarity 1.36±0.50% Asymmetry 2.83±0.33% Kiel Polarity 1.67±0.57% Asymmetry 2.50±0.44% Climax Polarity 2.22±0.89% Asymmetry 2.72±0.18% Huancayo Polarity 0.39+0.04% Asymmetry 0.72±0.04% 2.83 9.50 3.89 7.38 4.78 6.28 2.21 5.21 48 48 49 49 51 51 51 51 >99.09 >99.99 >99.95 >99.99 >99.99 >99.99 >99.31 >99.99 17
1205.6554	2	1205	2012-05-31T00:29:18	The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b	[ "astro-ph.EP" ]	Possible bulk compositions of the super-Earth exoplanets, CoRoT-7b, Kepler-9d, and Kepler-10b are investigated by applying a commonly used silicate and a non-standard carbon model. Their internal structures are deduced using the suitable equation of state of the materials. The degeneracy problems of their compositions can be partly overcome, based on the fact that all three planets are extremely close to their host stars. By analyzing the numerical results, we conclude: 1) The iron core of CoRoT-7b is not more than 27% of its total mass within 1 $\sigma$ mass-radius error bars, so an Earth-like composition is less likely, but its carbon rich model can be compatible with an Earth-like core/mantle mass fraction; 2) Kepler-10b is more likely with a Mercury-like composition, its old age implies that its high iron content may be a result of strong solar wind or giant impact; 3) the transiting-only super-Earth Kepler-9d is also discussed. Combining its possible composition with the formation theory, we can place some constraints on its mass and bulk composition.	astro-ph.EP	astro-ph	Research in Astronomy and Astrophysics manuscript no. (LATEX: raa896.tex; printed on February 19, 2018; 9:55) 2 1 0 2 y a M 1 3 . ] P E h p - o r t s a [ 2 v 4 5 5 6 . 5 0 2 1 : v i X r a The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b ∗ Yan-Xiang Gong1 , 2 and Ji-Lin Zhou1 1 Department of Astronomy & Key Laboratory of Modern Astronomy and Astrophysics in Ministry of Education, Nanjing University, Nanjing 210093, China; [email protected] 2 College of Physics and Electronic Engineering, Taishan University, Taian 271021, China Abstract Possible bulk compositions of the super-Earth exoplanets, CoRoT-7b, Kepler-9d, and Kepler-10b are investigated by applying a commonly used silicate and a non-standard carbon model. Their internal structures are deduced using the suitable equation of state of the materials. The degeneracy problems of their compositions can be partly overcome, based on the fact that all three planets are extremely close to their host stars. By analyzing the numerical results, we conclude: 1) The iron core of CoRoT-7b is not more than 27% of its total mass within 1 σ mass-radius error bars, so an Earth-like composition is less likely, but its carbon rich model can be compatible with an Earth-like core/mantle mass fraction; 2) Kepler-10b is more likely with a Mercury-like composition, its old age implies that its high iron content may be a result of strong solar wind or giant impact; 3) the transiting-only super- Earth Kepler-9d is also discussed. Combining its possible composition with the formation theory, we can place some constraints on its mass and bulk composition. Key words: planets and satellites: general -- methods: numerical -- planets and satellites: individual: CoRoT-7b, Kepler-9d, Kepler-10b 1 INTRODUCTION Rocky planets (like the Earth in our solar system) located in the habitable zone of their stars are presently the best candidates for harbouring extra-terrestrial life (L´eger et al. 2011). Therefore, the search for rocky exoplanets plays an important role in the present exoplanet detections. Recent observations of exoplanets have revealed 758 exoplanets (http://exoplanet.eu/, updated to 03 November 2011), mainly through stel- lar radial velocity measurements or photometric detection of planets transiting their host stars. Among them, more than 30 exoplanets are with minimum mass < 10 Earth masses (M⊕). They are called 'super- Earths' and a faction of them may possibly be rocky planets.Recently, the Kepler mission revealed 1235 planetary candidates with transit-like signatures detected during its first 4-month operation (Borucki et 2 Y.-X. Gong & J.-L. Zhou al. 2011 ). Among them, 68 candidates of approximately Earth-sizes (Rp < 1.25 R⊕), 288 super-Earth sizes (1.25 R⊕ < Rp < 2 R⊕). Beyond all doubt, we are entering the era of small exoplanets. Some of them are expected to be terrestrial in nature. The understanding of super-Earths is of great interest to us due to their possible rocky nature. What is the bulk composition of these new planets? How were they formed? These are all interesting yet puzzling questions. However, the information presently available based on observations is still limited for under- standing their composition. Through transit and radial velocity detections, planetary radii and masses are the only windows presently to investigate their composition for most exoplanets. Among the detected super- Earths, their masses and the radii show quite different signatures when compared to terrestrial planet of our solar system. Their low average density tells us that they must primarily contain some light element which is lighter than water. They may have a significant gas envelope like Jupiter or Neptune in our solar system. However, super-Earths lacking of gas envelops, or 'rocky super-Earths', may be also present in some cases, as we will discuss below. CoRoT-7b is the first super-Earth with a measured radius, R = 1.68 ± 0.09 R⊕ (L´eger et al. 2009). Its obtained mass combined with radial velocity measurements is M = 4.8 ± 0.8 M⊕ (Queloz et al. 2009). After 2009, the initial results have been revised by other authors. For example, after revision of CoRoT-7 stellar parameter by Bruntt et al. (2010), it yields M = 5.2±0.8 M⊕. Hatzes et al. (2011) performed another analysis of the data and got a larger mass, M = 7.0 ± 0.5 M⊕. The works done by Boisse et al. (2011) and Ferraz-Mello et al. (2011) got different results. In this paper, we do not consider those masses and use the initial data of CoRoT-7b only. Recently, two transiting super-Earths, Kepler-9d (Torres et al. 2011) and Kepler-10b (Batalha et al. 2011) were discovered. Kepler-9d was first discovered in 2010 as a planet candidate named KOI-377.03 (Holman et al. 2010). It was validated as a super-Earth and the reported radius is R = 1.64+0.19 −0.14 R⊕ (Torres et al. 2011). Although current spectroscopic observations are yet insufficient to establish its mass, a possible mass range can be found by Holman et al. (2010). The upper mass limit of M = 7 M⊕ corresponds to the maximum mantle-stripping limit (Marcus et al. 2010) for a maximally iron- rich super-Earth. The lower mass limit is ∼3.5 M⊕ as a volatile-poor rocky planet with a Ganymede-like Fe/Si ratio. So we consider all possible compositions of Kepler-9d using this mass range in our paper. In this paper, we use Kepler-9d as an example to show what can be inferred for 'transiting-only' hot super-Earths. Kepler-10b was discovered in early 2011, the reported mass and radius are Mp = 4.56+1.17 Rp = 1.416+0.033 are summarized in Table 1. −1.29 M⊕ and −0.036 R⊕ (Batalha et al. 2011), respectively. The main parameters of these three exoplanets The bulk composition of an exoplanet can not be uniquely determined by the measured mass and ra- dius. When more than two chemical constituents are taken into consideration, a measured radius and mass correspond to more than one possible bulk composition, in other words, it is a degeneracy problem. The degeneracy problems of bulk compositions of low-mass exoplanets can be partly overcome if they are very close to their host stars because some chemical constituents may be ruled out a priori. Valencia et al. (2010) discussed the evolution of close-in low-mass planets like CoRoT-7b. Due to the intense stellar irradiation and its small size, it is unlikely to possess an envelope of hydrogen and helium of more than 1/10000 of its total mass. A relatively significant mass loss of ∼ 1011 g s−1 is expected and the result should prevail The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 3 Table 1 Planet Parameters for Kepler-9d (Torres et al. 2011), Kepler-10b (Batalha et al. 2011), and CoRoT-7b (L´eger et al. 2009). The Ages of Their Host Stars are Also Listed Here. Planet Name Mass (M⊕) Radius (R⊕) Orbital Period (days) Equilibrium Temperature (K) Orbital Semimajor Axis (AU) Host Star Age (Ga) (a) Holman et al. (2010). (b) Queloz et al. (2009). Kepler-10b 4.56+1.17 −1.29 1.416+0.033 −0.036 Kepler-9d 3.5 − 7.0(a) 1.64+0.19 −0.14 CoRoT-7b 4.8 ± 0.8(b) 1.68 ± 0.09 0.837495+0.000004 −0.000005 1833(c) 1.592851 ± 0.000045 0.853585 ± 0.000024 2026 ± 60 1800-2600 0.01684+0.00032 −0.00034 0.02730+0.00042 −0.00043 0.0172 ± 0.00029 11.9 ± 4.5 3.0 ± 1.0 1.5+0.8 −0.3 (d) (c) Calculated value -- assuming a Bond albedo of 0.1 and a complete redistribution of heat for re-radiation. (d) http://www.exoplanet.eu/. independently of the planet's composition, the hydrogen-helium gas envelope or water vapour atmosphere would escape within ∼1 Ga, which is shorter than the calculated age of 1.2-2.3 Ga for CoRoT-7b. The age of Kepler-9 is 3±1 Ga and Kepler-10 is a relatively old star (11.9±4.5 Ga), so volatile-rich solutions are less likely if the Kepler-9 and Kepler-10 planetary systems are old and evaporation has been substantial. Jackson et al. (2010) studied the coupling of tidal evolution and evaporative mass loss on CoRoT-7b. Their investigation manifests that the orbital decay caused by the star-planet tidal interaction enhanced its mass loss rate. Such a large mass loss also suggests that, even if CoRoT-7b began as a gas giant planet, its original atmosphere has been completely evaporated now. As Kepler-9d and Kepler-10b are all small size close-in planets (with orbital periods 1.59 and 0.84 days, respectively), evaporation on the planets is serious due to high stellar irradiation. Thus, the origin of low-mass exoplanets, like Kepler-9d and CoRoT-7b cannot be inferred from the present observations: they may have always had a rocky composition; they may be remnants of a Uranus-like ice giant, or a gas giant with a small core that has been stripped of its gaseous envelope (Valencia et al. 2010). The latest results from Leitzinger et al. (2011) indicated that hydrogen-rich gas giants within the mass domain of Jupiter and Saturn cannot thermally lose such an amount of mass that CoRoT-7b and Kepler-10b would result in a rocky residue. They also concluded that these planets most likely were always rocky planets. In this paper, we assume that CoRoT-7b, Kepler-9d, and Kepler-10b do not have significant gas en- velopes and consider them as fully differentiated with all possible iron-to-silicate ratios. The differentiation assumption is not restrictive since all the terrestrial planets and some large satellites in the solar system are known to be differentiated. Using a relatively simple but well suited model for terrestrial planets in solar system, we discuss the interior structures of CoRoT-7b, Kepler-9d and Kepler-10b. We compare the silicate model and the carbon model (Seager et al. 2007). Our main aims are: 1) infer their bulk composition and sharpen the observational constraints on their masses and radii, 2) compare their interior structures and infer some clues about their formation. 4 Y.-X. Gong & J.-L. Zhou 2 MODEL AND METHOD 2.1 Model Equations and Numerical Method If the rotation rate is not extreme, self-gravitating bodies are close to spherical. Spherical structures are certainly an adequate starting place for studies of exoplanets. Three equations must be satisfied. They are the equation of hydrostatic equilibrium-Equation (1), the mass conservation equation-Equation (2) and the equation of state (EOS)-Equation (3) for the material: dP (r) dr = −G m (r) ρ (r) r2 , dm (r) dr = 4πr2ρ (r) , P (r) = f [ρ (r) , T (r)] . (1) (2) (3) Where m (r) is the mass contained within radius r, P (r) is the pressure, T (r) is the temperature, and ρ (r) is the density of a spherical planet. The thermal contributions to the conclusion can be ignored which has been testified in Seager et al. (2007) and this approximation has been often used in other works (Fortney et al. 2007; L´eger et al. 2004). As a reasonable approximation, the temperature-independent EOS is also used in this work. Using variable transition, if pressure is a nonlinear function of density, Equation (1) can be rewritten as, dρ dr = −G m (r) ρ (r) r2 · 1 dP (r)/dρ (r) . (4) We numerically integrate Equation (2) and Equation (4) starting at the planet's center r = 0 and using the inner boundary condition m (0) = 0 and ρ (0) = ρc, where ρc is a chosen central density. For the outer boundary condition, we use P (Rp) = 0. The choice of ρc at the inner boundary and the outer boundary condition P (Rp) = 0 define the planetary radius Rp and total mass Mp = m (Rp). Integrating Equation (2) and Equation (4) over and over for a range of ρc provides the mass-radius relationship (see 3.1.1) for a given material. For differentiated model containing more than one kind of material, we specify the desired fractional mass of the core and of each shell. We then integrate Equation (2) and Equation (4) as specified above, given a ρc and outer boundary condition. Using the planetary mass, we switch from one material to the next where the desired fractional mass is reached. Since we do not know the total mass that a given integration will yield ahead of time, we generally need to iterate a few times (change the value of ρc) in order to produce a model with the desired distribution of material. 2.2 The Choice of the EOS The materials used in this paper are ice VII, SiC, MgSiO3 (enstatite, ab. en), (Mg0.88, Fe0.12)SiO3 (per- ovskite, ab. pv), Fe(α), Fe(ε). Ice VII is used to model the ice composition in large moons of giant planets. MgSiO3 and (Mg0.88, Fe0.12)SiO3 are used to model the silicate mantles of terrestrial planets or large moons in our solar system. Enstatite is a low-pressure mineralogical phase of silicate, so it is only used for the validation of the model (see section 3.1.2). Fe(α) and Fe(ε) are different phases of iron under different pressure, Fe(ε) is the phase occurring at higher pressures. We compare their differences in mass-radius rela- tion for homogeneous planets. For CoRoT-7b, Kepler-9d, and Kepler-10b, we use Fe(ε) to model their iron The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 5 Table 2 Parameters for the Birch-Murnaghan EOS (BME) or Vinet EOS (VE) Fits. Materials K0 (GPa) K ′ 0 ρ0 (kg/m3) EOS References ice VII SiC MgSiO3 (en) (Mg0.88, Fe0.12)SiO3 (pv) Fe(α) Fe(ε) 23.7 ± 0.9 4.15 ± 0.07 227 ± 3 4.1 ± 0.1 125 5 266 ± 6 3.9 ± 0.04 162.5 ± 5 5.5 ± 0.8 156.2 ± 1.8 6.08 ± 0.12 1460 3220 3220 4260 7860 8300 BME BME BME BME BME VE 1 2, 3 2, 4 2, 5 2, 6 7 References: (1) Hemley et al. 1987, (2) Ahrens 2000, (3) Alexsandrov et al. 1989, (4) Olinger 1977, (5) Knittle & Jeanloz 1987, (6) Takahashi & Spain 1989, (7) Anderson et al. 2001. cores. Let us consider the pressure scope in our topic. Pc, the central pressure of the planet can be estimated as following Pc ∼ GM 2 p. p(cid:14)4R4 (5) It is about 103 GPa for the three exoplanets. For P ≤ 200 GPa we use fits to experimental data, either the Vinet EOS (VE) Equation (6) or the Birch-Murnaghan EOS (BME) Equation (7). In general, when P ≥ 104 GPa, where electron degeneracy pressure becomes increasingly important, the Thomas-Fermi- Dirac (TFD) theoretical EOS (Salpeter & Zapolsky 1967) must be employed, this is out of the range we need. The pressure range from approximately 200 to 103 GPa is not easily accessible to experiment nor is it well described by the TFD EOS (Seager et al. 2007). For all materials in this pressure regime we simply use the VE or BME. For a derivation of these EOSs, see Poirier (2000). The VE is P = 3K0(cid:18) ρ ρ0(cid:19)2/3"1 −(cid:18) ρ ρ0(cid:19)−1/3# exp( 3 2 (K ′ 0 − 1)"1 −(cid:18) ρ ρ0(cid:19)−1/3#) , and the third-order finite strain BME is P = 3 2 K0"(cid:18) ρ ρ0(cid:19)7/3 −(cid:18) ρ ρ0(cid:19)5/3#(1 + 3 4 (K ′ 0 − 4)"(cid:18) ρ ρ0(cid:19)2/3 − 1#) . (6) (7) Where ρ0 is the reference density, K0 = −V (∂P /∂V ) P =0 is the bulk modulus of the material, K ′ 0 = (∂K/∂P ) P =0 is its pressure derivative. Their values for the materials used in this paper are summarized in Table 2. The VE is used for the ε-phase of iron because it is more suitable than the BME for extrapolation to high pressures (Poirier 2000). The third-order BME is used for other materials. 2.3 Carbon Model There is an assumption used in modeling the interior structure of solid exoplanets: their constituents are similar to the terrestrial planets in our solar system. They have an iron core and a silicate envelope for the differentiated model. But based on current observation, other possibilities can not be ruled out. The facts tell us that there are large differences between extrasolar planets and planets in our solar system. For example, research about hot Jupiter tells us it is danger to assume some planets can only form in particular area. It is useful to take as broad a view as possible (Fortney et al. 2007). To other possible composition, Seager et al. (2007) presented the idea of carbon planets (see also Gaidos 2000). The planets in our solar 6 Y.-X. Gong & J.-L. Zhou system were born in an environment where the carbon-to-oxygen ratio C/O = 0.5. Their bulk compositions were determined by the high-temperature chemical equilibrium in protoplanetary disk. Silicates (i.e., Si-O compounds) are the dominant constituents in terrestrial mantles. But in an environment where C/O & 1, the condensation sequence changes dramatically (Larimer 1975; Lewis 1974; Wood et al. 1993). As a result, the high-temperature condensates will be carbon-rich compounds. Kuchner & Seager (2006) have explored some planet-formation scenarios and they proposed that carbon planets composed largely of SiC or other carbides should form in such environment. Carbon-rich nebula created by the disruptions of carbon-rich stars or white dwarfs is a natural cradle for carbon planets, so planets around pulsars and white dwarfs are candidates of carbon planets. For example, the planets around pulsar PSR1257+12 (Wolszczan et al. 1992) may be carbon-rich exoplanets. In the well-known β Pictoris debris disk (Roberge et al. 2006), an exoplanet named β Pic b was discovered by Lagrange et al. ( 2008), it is most likely a carbon-rich planet also because it is located in a carbon-rich environment. Carbon planets can also form in a local area enriched in C or depleted in H2O in an otherwise solar abundance protoplanetary disk (Seager et al. 2007). Lodders (2004) suggested that the planetary embryo that grew into Jupiter may have formed in such a locally carbon-rich area, and that Jupiter's embryo was a carbon planet. In this paper, we also consider the carbon model for CoRoT-7b, Kepler-9d, and Kepler-10b, in which SiC is the major mantle constituent. The detailed formation scenarios for Earth-mass carbon planets can be learned from Kuchner & Seager (2006), the latest research about the carbon-rich giant planets can be found in Madhusudhan et al. (2011). 3 NUMERICAL RESULTS 3.1 Validation in the Solar System 3.1.1 Planets in the Mass-radius Relationship Diagram A mass-radius diagram is a useful tool when inferring a planet's bulk composition. Building on the work done by Zapolsky & Salpeter (1969), we first consider planets of uniform composition. The positions of CoRoT-7b, Kepler-9d, and Kepler-10b in mass-radius diagram can quickly tell us whether we can model them as solid planets. In Figure 1, the lines are mass-radius relationship curves for homogeneous planets. From top to bottom the homogeneous planets consist of ice VII, SiC, MgSiO3 (enstatite, ab. en), (Mg0.88, Fe0.12)SiO3 (perovskite, ab. pv), Fe(α), Fe(ε), respectively. Here, MgSiO3 and (Mg0.88, Fe0.12)SiO3 are usually looked as representative material in mantle. The phase diagram of iron in Valencia et al. (2010) can help us to understand Fe(α) and Fe(ε) phase of iron (under different pressure scope). CoRoT-7b, Kepler- 9d, Kepler-10b and the Earth are marked on the mass-radius diagram. Two notable exoplanets GJ 1214b (Charbonneau et al. 2009) and Kepler-11b (Lissauer et al. 2011) are introduced here as a comparison. Their measured masses and radii tell us that Kepler-9d, Kepler-10b and CoRoT-7b are candidates of solid super- Earths. GJ 1214b is above the water ice VII line, this indicates it must contain some light elements such as H and He, most likely it has a substantial gas envelope. Kepler-11b is above the solid SiC line, it should have a substantial water vapour or H/He gas envelope because it is so close to host star (10.3 days, Lissauer et al. 2011). As the validation, we also mark some planets and satellites (M ≤ 1 M⊕) in our solar system in Figure 2. The Earth and the Venus fall between silicate and iron line, but close to silicate line, it implies they contain more rock composition than iron. Ganymede is found between silicate and the water ice VII The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 7 Table 3 Relative Error Between Calculated Radii and Actual Radii. Celestial Body Radius (km)(a) Assumed Composition(b) Model Radius (km) Relative Error (%) Mercury 2439.70 Venus 6051.80 Moon 1737.50 Earth 6371.00 Mars 3389.50 Ganymede 2631.20 60%Fe + 40%en 60%Fe + 40%pv 27.9%Fe + 72.1%en 27.9%Fe + 72.1%pv 3%Fe + 97%en 3%Fe + 97%pv 32.5%Fe + 67.5%en 32.5%Fe + 67.5%pv 20%Fe + 80%en 20%Fe + 80%pv 6.5%Fe + 48.5%en + 45%ice VII 6.5%Fe + 48.5%pv + 45%ice VII 2474.43 2344.78 6283.54 5862.68 1745.05 1652.37 6589.91 6269.81 3409.08 3149.51 2525.19 2454.14 1.40 3.80 3.83 3.13 0.43 4.90 3.40 1.59 0.58 7.00 4.02 6.73 (a) http://ssd.jpl.nasa.gov/. (b) References: Mercury, Solomon (2003); Venus, Zharkov (1983); Earth, Ganymede, Seager et al. (2007); Mars, Rivoldini et al. (2011). line, which indicates it must have significant amounts of ice composition. The Moon just lies on the silicate line, it agrees with the fact that the Moon is composed almost entirely of rock with a small iron core (Hu & Xu 2008). Unlike the Moon, Mercury must have a massive iron core. 3.1.2 Code Test for the Differentiated Planets Mass-radius relationship diagram (or one layer model) can only give us a qualitative estimation, but it can not tell us the details of interior structure. To estimate the precision of our code, we calculate the radii of planets in solar system from their total mass and mass fraction published in relevant literatures. On observation, the planet mass and radius uncertainties are typically 5% to 10% (Selsis et al. 2007), so a robust numerical code must guarantee the error is less than 5%. Table 3 are the results we get. We must emphasize that mass fractions of materials differ in literature, which indicates that there are still some uncertainties in the bulk composition of solar system objects. For example, the iron core mass fraction of the Moon is estimated at less than 10% of its total mass - 3% is found by Canup & Asphaug (2001). We use both pv and en phase of silicate to calculate the radii, the aim is to test the model approach only. Beside for Ganymede and Mars (both with the pv mantle), the relative error of all other planets and the Moon is less than 5%. Large relative error to Mars is understandable because Mars is relatively small, so pv is less suitable. Mercury is small too, but it has a large iron core, so pv or en can not cause large discrepancies. The relatively large error (6.73%) of Ganymede (with pv mantle) may be caused by the relatively low pressures in the deep interior of Ganymede (Sohl et al. 2002), and therefore pv is not very suitable, either. The calculated Earth radius are within 101km (only 1.59% error) of the actual Earth, this means that pv is the predominant mantle mineralogical phase in Earth. 8 Y.-X. Gong & J.-L. Zhou 3.2 The Results for Three Exoplanets We examine the interior composition of Kepler-10b, Kepler-9d and CoRoT-7b under the assumption of an iron core covered by a mantle composed of (Mg0.88, Fe0.12)SiO3 (silicate model, similar to Earth's mantle) or SiC (carbon model). When considering two-layer models, the measured mass and radius uniquely determine the planet's composition. The core mass fraction as a function of planet radius for the three planets are displayed in Figure 3-5. The solid red line denotes the fraction of planet's mass in its iron core according to an errorless planetary mass, while the green, white, and red shaded regions in Figure 3-4 delimit the 1σ, 2σ, and 3σ error bars on Mp, respectively. In Figure 5, the solid red line denotes the iron core mass fraction for Mp = 5.25 M⊕ (middle mass between 3.5-7.0 M⊕). The gray regions in Figure 5 delimit the mass range we considered. The measured planet radius and its 1σ error bars are denoted by solid blue and the dashed black vertical lines, respectively. Interior structure models can strengthen the observational constraints on a planet's mass and radius (Rogers et al. 2010). For example, with the assumption that planets do not have a significant water or gas layer, some of the mass-radius pairs within Mp ± 1σM and Rp ± 1σR (including the 0σ mass-radius pair) can be ruled out because they correspond to bulk densities lower than a pure silicate planet. These excluded mass-radius pairs would necessitate water (or some other component lighter than silicate). In Figure 3-5, all mass-radius pairs in the 1σ error bar (both mass and radius) compose an area. Here, we call them 'effective area'. To CoRoT-7b, if it has an Earth-like silicate mantle, its radius must be less than the measured value (1.68 R⊕) and its mass has to be larger than the measured mass (4.8 M⊕) within 1σ error bars, which implies an iron core mass fraction of less than ∼ 27%. Within its observational value, its radius is unlikely less than the reliable value if it formed in the carbon rich environment and has a Moon-like composition (where the iron content is below 10% in carbon model). Its iron core is not heavier than ∼ 70% of its total mass even at 3σ error bar (see carbon model in Figure 3), it is the largest value for both models. The silicate model of CoRoT-7b with an Earth-like iron core mass fraction (32.5%) is not consistent with the measured mass and radius within 1σ. If its mass is ascertained with in 2σ or 3σ error bar (in the range got by Hatzes et al. 2011), CoRoT-7b can be consistent with an Earth-like composition. From Figure 3, we can clearly see that the 'effective area' in the carbon model is much larger than for the silicate model. Kepler-10b tells us other information. For its carbon model, there is a lower iron mass fraction limit (about 41%) within 1σ error bar (see carbon model in Figure 4). Kepler10 (7.4-16.4 Ga) is older than CoRoT-7 (1.2-2.3 Ga), if CoRoT-7b is differentiated as we assumed, Kepler-10b is more likely to be dif- ferentiated. This means that Kepler-10b may have a larger iron core. Our simulation indicates Kepler-10b is denser than CoRoT-7b, but the detailed calculation can tell us more: if CoRoT-7b has a silicate man- tle, whereas Kepler-10b formed under carbon-rich circumstances, the upper core mass fraction limit of CoRoT-7b (∼ 27%) is below the lower core mass fraction limit (41% mentioned above) of Kepler-10b! To Kepler-10b's carbon model, the upper limit of the iron core mass fraction is ∼ 84% (see carbon model in Figure 4) within 1σ error bar, this value is the largest in its two models and is larger as the suggested value of Mercury (∼ 60%) -- the densest planet in solar system. In Figure 6, we give the histograms of iron content for CoRoT-7b and Kepler-10b. The lower value we take comes from the planet formation scenario The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 9 which we will discuss later. To Kepler-10b, silicate model is compatible with an Earth-like composition within the 1σ error bar, the mass is about 3.27 M⊕ (lower limit of the observational value). The carbon model of Kepler-10b is not compatible with an Earth-like core/mantle mass ratio (32.5% in core). Kepler-9d is shown in Figure 5, the gray band denotes the planetary mass range we are considering. If the maximum mantle-stripping theory is true, silicate model in Figure 5 tells us that Kepler-9d's radius can not be larger than ∼1.75R⊕. In detail, if it has an Earth-like core/mantle mass ratio, its radius is no more than ∼1.7 R⊕ with silicate model. Despite the mass of a transiting-only super-Earth is unknown, model simulations give constraints on their possible masses. For example, if Kepler-9d has an Earth-like mantle composition (silicate model in Figure 5), it has a lower mass limit according to its smallest possible radius, which is larger than the lower mass limit (3.5 M⊕) discussed in Holman et al. (2010). An Earth-like core mass fraction and the lower limit of observational radius (∼1.50 R⊕) yield a total mass of ∼4.1 M⊕. In the mass range discussed in this paper, the carbon model of Kepler-9d is compatible with an Earth-like core/mantle mass ratio in a large scope of radius. Other information can be read from the Figure 3-5. For a given mass, the carbon model yields a larger radius than the silicate model. If the core is not pure iron but also contains a light element such as sulfur (in the Earth's core), all of the slantwise lines will sway to the right a little more (having no effect on the radius at a core mass fraction of 0.0). This indicates that the core mass fraction at a specified planetary radius will be larger. Finally, we calculated the possible interior structure of CoRoT-7b and Kepler-10b using their presumable masses (Figure 7-8). To CoRoT-7b, we take an Earth-like core/mantle mass ratio (core mass is 32.5%). From Figure 7, silicate model of CoRoT-7b can be ruled out because the calculated radius is outside the observational 1σ error bar. To Kepler-10b, we take a Mercury-like core/mantle mass ratio (core mass is 60 %). In Figure 8, Kepler-10b is compatible with a Mercury-like core/mantle mass ratio with its reliable mass for silicate model. 4 DISCUSSION AND CONCLUSION Two recently discovered super-Earths, Kepler-9d and Kepler-10b, and previously found CoRoT-7b are dis- cussed in this paper. Because of their intense irradiation and small size, they are possibly absent of gas envelopes. Under the assumption of that they are mainly composed by refractory materials, their internal structures are deduced using suitable equation of state of the materials. The silicate and the carbon model are discussed in details. By analyzing the numerical results, we find some of the mass-radius pairs within Mp ± 1σM and Rp ± 1σR can be ruled out. Therefore, our interior models sharpen the observational con- straints on their masses and radii. For CoRoT-7b, an Earth-like composition with a silicate (pv) mantle is less likely within both the mass and the radius 1σ error bars, but a Mars-like composition may be suitable. To Kepler-10b's two models, a Mercury-like composition having 60% of its mass in iron core is consis- tent with the measured mass and radius within 1σ error bar. There is an upper and lower limit for the iron core mass fraction of every planet. To CoRoT-7b, the upper iron core mass fraction limits are 27% (silicate model) and 56% (carbon model). For Kepler-10b, the silicate model have an upper limit of about 75% and its iron content is approximately 41% ∼ 84% for the carbon model. Transiting-only exoplanet like Kepler- 10 Y.-X. Gong & J.-L. Zhou 9d can also be studied, for example, its radius is no more than 1.7 R⊕ in silicate model. The lower mass limit (3.5 M⊕ we adopted) is also ruled out in its silicate model. Our quantitative calculation tells us Kepler-10b may be a Mercury-like planet. CoRoT-7b and Kepler- 10b share the similar features: their radii, their period, the type of host star and perhaps their mass. Why is Kepler-10b denser than CoRoT-7b? Their different ages may give us some answers. To the anomalously dense Mercury in solar system, there are three competing viewpoints (Solomon 2003): 1) the solar nebula caused aerodynamic drag on the particles from which Mercury was accreting, the iron and silicate have different response to this drag, so the lighter particles were lost from the accreting material at the onset of accretion. 2) it surface rock could have been vaporized by strong radiation from a hot nebula and have been carried away by the solar wind, 3) much of the original crust and mantle have been stripped away by a giant impact caused by a planetesimal. Because of lacking detailed observational data, which hypothesis is true can not be determined. However, each hypothesis predicts a different surface composition. So two promising space missions, underway MESSENGER (Solomon et al. 2001, NASA) and the upcoming BepiColombo (Grard et al. 2001, ESA), could bring us the answer. The second and third explanations invoke processes late in the planetary accretion process, after the protoplanet had differentiated silicate mantle from metal core. If CoRoT-7b and Kepler-10b were born in a similar environment, Kepler-10b's abnormal iron content can be naturally associated with its older age. We think the postnatal alteration is more likely. The lower limit of iron content of a super-Earth is an interesting question because a pure silicate super- Earth (iron mass fraction is zero) is unimaginable under differentiation hypothesis. Under physically plau- sible conditions analogous to solar system, planet formation will lead to differentiated super-Earth with an iron core covered by a silicate mantle, with the proportions of each determined by the local Si/Fe ratio (Grasset et al. 2009). The only way to significantly increase the mean density of a planet requires removal of an extended part of its silicate mantle. Marcus et al. (2010) proposed an efficient method -- giant impacts between super-Earths. On the other hand, they prefer a lower limit of iron content of 33% by considering the standard cosmic abundances (see also Valencia et al. 2007). The initial Fe/Si ratio (average value) is used to give a lower limit for the iron content of super-Earth. But, in practice, formation progress must be influenced by other factors (such as the survival competition between planets) because some planets in solar system have an iron content lower than this 'lower limit'. We think that referring to the iron content in chon- drites is a sound choice. L chondrites have lower total iron content, which is about 10% (Chen et al. 1994). L chondrites are the largest group in the ordinary chondrites (40%, http://www.nhm.ac.uk/). Therefore, we use 10% as the lower limit for silicate model. Regarding Figure 6, if there is a calculated lower limit of the iron content, we will use it, otherwise we use 10% (only for silicate model). Our main conclusions are summarized as follows: 1) The iron core of CoRoT-7b is not more than 27% of its total mass within 1 σ mass-radius error bars, so an Earth-like composition is less likely, but its carbon rich model can be compatible with an Earth-like core/mantle mass fraction; 2) Kepler-10b is more likely with a Mercury-like composition, its old age implies that its high iron content may be a result of strong solar wind or giant impact; 3) the transiting-only super-Earth Kepler-9d is also discussed. Combining its possible composition with the formation theory, we can place some constraints on the mass and composition of The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 11 Kepler-9d. The radius derived from transit method and the mass got by radial velocity detections together with the model simulation can constrain the possible compositions of exoplanets. Credible conclusions desiderate comprehensive observations. Planetary formation theories, thermal evo- lution models, and studies of the cosmic abundance of elements can be used to place additional constraints on a planet's interior composition. Besides the temperature-independent model used in this paper, more complicated models of solid exoplanets can be found in other literatures. For example, Valencia et al. (2006) and Sotin et al. (2007) considered the temperature profile and mixture of materials in the mantle or core. Wagner et al. (2011) used material laws in the infinite pressure limit to improve the model of solid exoplanets. In the near future, high-sensitivity spectroscopic transit observations of these exoplanets should constrain the compositions of the evaporating flow and therefore allow us to distinguish between rocky or gas-rich planets. In particular for exoplanets, without any opportunity to in situ composition measurements and gravitational moment measurements from spacecraft flybys, we will be permanently limited in what we can infer about the interior composition from their observational mass and radius (Rogers & Seager 2010). Acknowledgements Firstly, we want to thank the anonymous referees for constructive comments and suggestions which improved the paper a lot. This work was supported by the National Natural Science Foundation of China (Nos.10833001 and 10925313), Ph.D training grant of China (20090091110002), and Fundamental Research Funds for the Central Universities (1112020102). Gong Yan-Xiang also acknowl- edge support form the Shandong Provincial Natural Science Foundation, China (ZR2010AQ023). We also thank Dr. Zeng Li from Harvard University for his communication with us about the interior structure of close-in exoplanets. 12 References Y.-X. Gong & J.-L. Zhou Ahrens, T. J. 2000, Mineral Physics and Crystallography: A Handbook of Physical Constants (Washington, DC: AGU) Alexsandrov, I. V., Gocharov, A. F., et al. 1989, J. Exp. Theor. Phys. Lett., 50, 127 Anderson, O. L., Dubrovinsky, L., Saxena, S. K., & LeBihan, T. 2001, Geophys. Res. Lett., 28, 399 Batalha, N. M., Borucki, W. J., Bryson, S. T., et al. 2011, ApJ, 729, 27 Boisse, I., Bouchy, F., H´ebrard, G., et al. 2011, A&A, 528, A4 Bord´e, P., Rouan, D., & L´eger, A. 2003, A&A, 405, 1137 Borucki, W. J., Koch, D. J., Basri, G., et al. 2011, ApJ, 736, 19 Bruntt, H., Deleuil, M., Fridlund, M., et al. 2010, A&A, 519, A51 Canup, R. M., & Asphaug, E. 2001, Nature, 412, 708 Charbonneau, D., Berta, Z. K., Irwin, J., et al. 2009, Nature, 462, 08679 Chen, D. G., Zhi, X. C., & Yang, H. T. 1994, Geochemistry (Hefei: China Science and Technology University Press) Ferraz-Mello, S., Tadeu dos Santos, M., Beaug´e, C., et al. 2011, A&A, 531, A161 Fortney, J. J., Marley, M. S., & Barnes, J. W. 2007, ApJ, 659, 1661 Gaidos, E. J. 2000, Icarus, 146, 637 Grard, R., & Balogh, A. 2001, Planet. Space Sci., 49, 1395 Grasset, O., Schneider, J., Sotin, C. 2009, ApJ, 693, 722 Hatzes, A. P., Fridlund, M. 2011, arXiv:1105.3372 Hemley, R. J., Jephcoat, A. P., Mao, H. K., et al. 1987, Nature, 330, 737 Holman, M. J., Fabrycky, D. C., Ragozzine, D., et al. 2010, Science, 330, 51 Hu, Z. W., & Xu, W. B. 2008, Planetary Science (Beijing: Science Press), 341 Jackson, B., Miller, N., Barnes, R., Raymond, S. N., Fortney, J. J., & Greenberg, R. 2010, MNRAS, 407, 910 Knittle, E., & Jeanloz, R. 1987, Science, 235, 668 Kuchner, M., & Seager, S. 2006, arXiv:astro-ph/0504214 Lagrange, A. -M., Gratadour, D., Chauvin, G., et al. 2008, A&A, 493, L21 Larimer, J. W. 1975, Geochim. Cosmochim. Acta, 39, 389 L´eger, A., Grasset, O., Fegley, B., et al. 2011, Icarus, 213, 1 L´eger, A., Rouan, D., Schneider, J., et al. 2009, A&A, 506, 287 L´eger, A., Selsis, F., Sotin, C., et al. 2004, Icarus, 169, 499 Lewis, J. S. 1974, Science, 186, 440 Leitzinger M., Odert, P., Kulikov, Y. N., et al. 2011, Planet. Space Sci., 10, 1016 Lissauer, J. J., Fabrycky, D. C., Ford, E. B., et al. 2011, Nature, 470, 53 Lodders, K. 2004, ApJ, 611, 587 Marcus, R. A., Sasselov, D., Hernquist, L., & Stewart, S. T. 2010, ApJ, 712, L73 Madhusudhan, N., Mousis, O., Johnson, T. V., et al. 2011, ApJ, 743, 191 Olinger, B. 1977, in High Pressure Physics Research: Applications in Geophysics, ed. M. Manghani & S. Akimoto (New York: Academic), 325 Poirier, J. P. 2000, Introduction to the physics of the Earth's interior (Cambridge: Cambridge University Press), 63 Queloz, D., Bouchy, F., Moutou, C., et al. 2009, A&A, 506, 303 Rivoldini, A., Van Hoolst, T., Verhoeven, O., et al. 2011, Icarus, 213, 451 Roberge, A., Feldman, P. D., Weinberger, A. J., Deleuil, M., & Bouret, J. C. 2006, Nature, 441, 724 Rogers, L. A., & Seager, S. 2010, ApJ, 712, 974 Salpeter, E. E., & Zapolsky, H. S. 1967, Phys. Rev., 158, 876 Seager, S., Kuchner, M., Hier-Majumder, C. A., & Militzer, B. 2007, ApJ, 669, 1279 Selsis, F., Chazelas, B., Bord´e, P., et al. 2007, Icarus, 191, 453 Sohl, F., Spohn, T., Breuer, D., & Nagel, K. 2002, Icarus, 157, 104 Solomon, S. C., McNutt, R. L., Gold, R. E., et al. 2001, Planet. Space Sci., 49, 1445 Solomon, S. C. 2003, Earth Planet Sci. Lett., 216, 441 Sotin, C., Grasset, O., & Mocquet, A. 2007, Icarus, 191, 337 Takahashi, Y. X., & Spain, I. L. 1989, Phys. Rev. B, 40, 993 Torres, G., Fressin, F., Batalha, N. M., et al. 2011, ApJ, 727, 24 Valencia, D., Guillot, T., & Nettelmann, N. 2010, A&A, 516, A20 Valencia, D., O'Connell, R. J., & Sasselov, D. D. 2006, Icarus, 181, 545 The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 13 GJ 1214b Kepler-11b H2O(ice VII) SiC MgSiO3(en) (Mg0.88, Fe0.12)SiO3(pv) Kepler-9d Fe( ) Fe( ) Kepler-10b CoRoT-7b Earth 2.8 2.4 2.0 1.6 1.2 0.8 0.4 0.0 h t r a E R 0 2 4 6 MEarth 8 10 12 Fig. 1 Mass-radius relationship diagram for Mp ≤ 10 M⊕. The solid or dashed lines are ho- mogeneous planets. Five exoplanets and the Earth are marked on the diagram. The mass range of Kepler-9d (red error bar) is taken from Holman et al. (2010). The locations of CoRoT-7b, Kepler-9d and Kepler-10b indicate they can be described as solid planets. Valencia, D., Sasselov, D. D., & O'Connell, R. J. 2007, ApJ, 665, 1413 Wagner, F. W., Sohl, F., Hussmann, H., Grott, M., & Rauer, H. 2011, Icarus, 214, 366 Wolszczan, A., & Frail, D. A. 1992, Nature, 355, 145 Wood, J. A., & Hashimoto, A. 1993, Geochim. Cosmochim. Acta, 57, 2377 Zapolsky, H. S., & Salpeter, E. E. 1969, ApJ, 158, 809 Zharkov, V. N. 1983, M&P, 29, 139 14 Y.-X. Gong & J.-L. Zhou h t r a E R 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 H2O(ice VII) SiC MgSiO3(en) (Mg0.88, Fe0.12)SiO3(pv) Earth Venus Fe( ) Fe( ) Mars Ganymede Mercury Moon 0.01 0.1 1 10 MEarth Fig. 2 Mass-radius relationship diagram for Mp ≤ 1 M⊕. The solid or dashed lines are ho- mogeneous planets. Some planets and satellites in solar system are shown in it. Their major constituents can be estimated using this mass-radius diagram. The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 15 1.0 0.8 0.6 0.4 0.2 0.0 0.8 1.0 0.8 0.6 0.4 0.2 Silicate Model CoRoT-7b 1.0 1.2 1.4 REarth 1.6 1.8 Carbon Model CoRoT-7b n o i t c a r F s s a M e r o C e F n o i t c a r F s s a M e r o C e F 0.0 0.8 1.0 1.2 1.6 1.8 2.0 1.4 REarth Fig. 3 Core mass fraction as a function of planetary radius for CoRoT-7b. Silicate model ((Mg0.88, Fe0.12)SiO3) and carbon model (SiC) are discussed. The planetary mass is M = 4.8 ± 0.8 M⊕. The green, white and red regions denote the core mass fractions obtained by varying its mass within 1σ, 2σ and 3σ error bars. The dashed black vertical lines delimit the measured radius R = 1.68 ± 0.09 R⊕. The Silicate model of CoRoT-7b had been discussed by Rogers & Seager (2010) and Valencia et al. (2010). We redo it for the comparison with its carbon model. 16 Y.-X. Gong & J.-L. Zhou n o i t c a r F s s a M e r o C e F n o i t c a r F s s a M e r o C e F 1.0 0.8 0.6 0.4 0.2 Silicate Model Kepler-10b 0.0 0.4 0.6 0.8 1.0 1.2 REarth 1.4 1.6 1.8 1.0 0.8 0.6 0.4 0.2 0.0 0.4 Carbon Model Kepler-10b 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 REarth Fig. 4 Kepler-10b core mass fraction as a function of planetary radius. Silicate model ((Mg0.88, Fe0.12)SiO3) and carbon model (SiC) are discussed. The lines and areas are similar to Fig.3. The planetary mass is M = 4.56+1.17 −0.036 R⊕. −1.29 M⊕ and the measured radius is R = 1.416+0.033 ∆M = ±1.29 M⊕ is used for 2σ and 3σ error bars of its mass. The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 17 n o i t c a r F s s a M e r o C e F n o i t c a r F s s a M e r o C e F 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 Silicate Model Kepler-9d 1.0 1.2 1.4 REarth 1.6 1.8 Carbon Model Kepler-9d 1.0 1.2 1.4 1.6 1.8 2.0 REarth Fig. 5 Kepler-9d core mass fraction as a function of planetary radius. Silicate model ((Mg0.88, Fe0.12)SiO3) and carbon model (SiC) are discussed. The measured radius is R = 1.64+0.19 −0.14 R⊕. The gray region denotes the core mass fractions obtained by varying the mass within the mass range we considered, M = 5.25 ± 1.75 M⊕. 18 Y.-X. Gong & J.-L. Zhou 1.1 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 ) % 3 ( n o o M 1 0 n o i t c a r F s s a M e r o C e F ) % 0 6 ( y r u c r e M ) b 7 - T o R o C % 7 2 ~ 0 1 ( ) . % 5 2 3 ( h t r a E ) . % 9 7 2 ( s u n e V ) % 0 2 ( s r a M ) . % 5 6 ( e d e m y n a G 2 3 4 6 5 8 Celestial Bodies 7 b 0 1 - r e p e K l ) % 4 8 ~ 1 4 ( b 0 1 - r e p e K l ) % 5 7 ~ 0 1 ( b 7 - T o R o C ) % 6 5 ~ 0 ( 9 10 11 12 Fig. 6 The histograms of iron content for CoRoT-7b, Kepler-10b and other celestial bodies men- tioned in our work. The corresponding references are the same in Table 3. The green part denotes the upper and lower limit for silicate model, the blue part denotes similar limit for carbon model (see details in Section 4). To the grid range on the histogram of Kepler-10b (silicate model), the lower limit (33%) adopted by Marcus et al. (2010) is used. Obviously, from a statistical point of view, CoRoT-7b is compatible with a Mars-like core/mantle mass fraction, whereas Kepler-10b is compatible with a Mercury-like composition. The silicate model and carbon rich model of CoRoT-7b, Kepler-9d and Kepler-10b 19 25000 20000 15000 10000 ) 3 / m g k ( y t i s n e D Fe 22100 22000 21900 21800 21700 21600 ) 3 / m g k ( y t i s n e D 21500 0.00 0.05 0.10 0.15 0.20 0.25 0.30 r (REarth) 5000 CoRoT-7b (Mg0.88, Fe0.12)SiO3(pv) SiC 0 0.0 0.2 0.4 0.6 1.2 1.4 1.6 1.8 0.8 1.0 r (REarth) Fig. 7 The interior structure of CoRoT-7b. We use the reliable mass M = 4.8 M⊕ with an Earth- like core/mantle mass ratio (iron core is 32.5% ). Two mantle compositions are considered, the red line is for (Mg0.88, Fe0.12)SiO3 and the blue line is for SiC. Its radius scope R = 1.68 ± 0.09 R⊕ is shown using vertical lines. Silicate model can be ruled out because the corresponding radius is out the 1σ error bar for its reliable mass. 20 Y.-X. Gong & J.-L. Zhou 25000 20000 15000 10000 ) 3 / m g k ( y t i s n e D 24870 24840 24810 24780 24750 24720 ) 3 / m g k ( y t i s n e D Fe Kepler-10b (Mg0.88, Fe0.12)SiO3(pv) SiC 5000 0.00 0.04 0.12 0.16 0.08 r(REarth) 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 r (REarth) Fig. 8 The interior structure of Kepler-10b. We use a Mercury-like core/mantle mass ratio (iron core is 60% ) and its reliable mass M = 4.56 M⊕. Two mantle compositions are consid- ered, the red line is for (Mg0.88, Fe0.12)SiO3 and the blue line is for SiC. Its radius scope R = 1.416+0.033 −0.036 R⊕ is shown using vertical lines. Carbon model can be ruled out because the corresponding radius is out the 1σ error bar for its reliable mass.
1805.01453	3	1805	2019-10-16T15:59:00	The California Kepler Survey VII. Precise Planet Radii Leveraging Gaia DR2 Reveal the Stellar Mass Dependence of the Planet Radius Gap	[ "astro-ph.EP", "astro-ph.SR" ]	The distribution of planet sizes encodes details of planet formation and evolution. We present the most precise planet size distribution to date based on Gaia parallaxes, Kepler photometry, and spectroscopic temperatures from the California-Kepler Survey. Previously, we measured stellar radii to 11% precision using high-resolution spectroscopy; by adding Gaia astrometry, the errors are now 2%. Planet radius measurements are, in turn, improved to 5% precision. With a catalog of ~1000 planets with precise properties, we probed in fine detail the gap in the planet size distribution that separates two classes of small planets, rocky super-Earths and gas-dominated sub-Neptunes. Our previous study and others suggested that the gap may be observationally under-resolved and inherently flat-bottomed, with a band of forbidden planet sizes. Analysis based on our new catalog refutes this; the gap is partially filled in. Two other important factors that sculpt the distribution are a planet's orbital distance and its host star mass, both of which are related to a planet's X-ray/UV irradiation history. For lower mass stars, the bimodal planet distribution shifts to smaller sizes, consistent with smaller stars producing smaller planet cores. Details of the size distribution including the extent of the `sub-Neptune desert' and the width and slope of the gap support the view that photoevaporation of low-density atmospheres is the dominant evolutionary determinant of the planet size distribution.	astro-ph.EP	astro-ph	Draft version October 17, 2019 Preprint typeset using LATEX style AASTeX6 v. 1.0 9 1 0 2 t c O 6 1 . ] P E h p - o r t s a [ 3 v 3 5 4 1 0 . 5 0 8 1 : v i X r a THE CALIFORNIA-KEPLER SURVEY VII. PRECISE PLANET RADII LEVERAGING GAIA DR2 REVEAL THE STELLAR MASS DEPENDENCE OF THE PLANET RADIUS GAP Benjamin J. Fulton1,2 and Erik A. Petigura1,3 1California Institute of Technology, Pasadena, CA 91125, USA 2IPAC-NASA Exoplanet Science Institute Pasadena, CA 91125, USA 3Hubble Fellow ABSTRACT The distribution of planet sizes encodes details of planet formation and evolution. We present the most precise planet size distribution to date based on Gaia parallaxes, Kepler photometry, and spec- troscopic temperatures from the California-Kepler Survey. Previously, we measured stellar radii to 11% precision using high-resolution spectroscopy; by adding Gaia astrometry, the errors are now 3%. Planet radius measurements are, in turn, improved to 5% precision. With a catalog of ∼1000 planets with precise properties, we probed in fine detail the gap in the planet size distribution that separates two classes of small planets, rocky super-Earths and gas-dominated sub-Neptunes. Our pre- vious study and others suggested that the gap may be observationally under-resolved and inherently flat-bottomed, with a band of forbidden planet sizes. Analysis based on our new catalog refutes this; the gap is partially filled in. Two other important factors that sculpt the distribution are a planet's orbital distance and its host star mass, both of which are related to a planet's X-ray/UV irradiation history. For lower mass stars, the bimodal planet distribution shifts to smaller sizes, consistent with smaller stars producing smaller planet cores. Details of the size distribution including the extent of the 'sub-Neptune desert' and the width and slope of the gap support the view that photoevaporation of low-density atmospheres is the dominant evolutionary determinant of the planet size distribution. 1. INTRODUCTION NASA's prime Kepler mission (2009 -- 2013; Borucki et al. 2010; Borucki 2016) is continuing to revolution- ize our understanding of planetary astrophysics. Ke- pler's success flows from its near continuous high pre- cision photometric monitoring of ∼150,000 stars over a four year mission. Among many discoveries, the large and homogeneous Kepler dataset enabled demographic studies of large numbers of exoplanets as small as Earth (see, e.g., Howard et al. 2012; Fressin et al. 2013; Pe- tigura et al. 2013). One startling result from these stud- ies is that nearly every Sun-like star has a planet larger than Earth but smaller than Neptune. Given the lack of such planets orbiting the Sun, Kepler demonstrated that the Solar System is not a typical outcome of planet formation, in at least that one key respect. Initially, the basic structure of these ubiquitous 1 to 4 R⊕ planets was unknown. It was unclear whether these planets were predominately rocky or had substan- tial gaseous envelopes. Early clues came from mass mea- surements of a few tens of Kepler planets based on the radial velocity (RV) and transit-timing variation (TTV) techniques (Marcy et al. 2014; Holman et al. 2010). These measurements revealed a transition in exoplanet bulk composition at ≈1.5 R⊕, with smaller planets hav- ing bulk densities consistent with rock and larger planets having extended low-density envelopes (Weiss & Marcy 2014; Rogers 2015). The distribution of Kepler planets as a function of size, orbital period, and other properties encodes key as- pects of planet formation physics including the growth of solid cores, the accretion/loss of gaseous envelopes, and the extent to which planets migrate. Insight into these processes requires accurate knowledge of host star prop- erties. Until recently, the properties of the vast majority of Kepler planet host stars were based on photometry alone, i.e. from the Kepler Input Catalog (KIC) and its updates (Brown et al. 2011; Huber et al. 2014; Mathur et al. 2017). Importantly, stellar radii R(cid:63) determined from photometry are uncertain at the ≈40% level, which hides important features in the planet population. To address the limitations of photometric stellar prop- erties, our team conducted the California-Kepler Survey (CKS), which obtained high-resolution optical spectra of 1305 planet hosting stars (Petigura et al. 2017, P17 here- after). Among other properties, these spectra enabled more precise stellar radii with ≈11% precision (Johnson et al. 2017, J17 hereafter). In Fulton et al. (2017), F17 2 Fulton & Petigura hereafter, we recomputed planet occurrence given these improved properties and found that the radius distribu- tion of small planets is bimodal with a paucity of planets between 1.5 and 2.0 R⊕. In previous studies, this feature was washed out to large Rp uncertainties. The radius gap occurs at the transition radius separating planets with and without gaseous envelopes. A gap in the radius distribution was predicted by several groups who considered the effect of photo- evaporation on planetary envelopes by X-ray and ex- treme ultraviolet (XUV) radiation (Lopez & Fortney 2013; Owen & Wu 2013; Jin et al. 2014; Chen & Rogers 2016). The observation of the radius gap lends much credibility to photoevaporation as a key process that sculpts the population of sub-Jovian class planets. How- ever, while photoevaporation is a leading theory explain- ing this feature, alternative mechanisms have been pro- posed, such as mass loss powered by luminosity of a planet's cooling core (Ginzburg et al. 2018). The apparent width of the radius gap in F17 was ≈25% in Rp. Because the gap was only marginally wider than the Rp uncertainties (≈13%), its true width and depth was uncertain. Indeed, Van Eylen et al. (2017) studied a smaller sample of ∼100 planets with percent- level Rp precision enabled by asteroseismology, and sug- gested that the gap may be wider an deeper than it appears in F17. Here, we re-examine the planet population at higher resolution by incorporating recently released parallax measurements from ESA's Gaia mission (Gaia Collabo- ration et al. 2016a). Launched in 2013, Gaia is conduct- ing an all-sky astrometric survey of ∼ 109 stars. Gaia's first data release (DR1) included 14 months of Gaia measurements, leveraging the Tycho catalog to con- strain proper motions (Gaia Collaboration et al. 2016b; Lindegren et al. 2016). Gaia DR1 included parallax measurements of only a handful of Kepler planet hosts and were not precise enough to improve radii over those from spectroscopy alone. Gaia DR2 (Gaia Collabora- tion et al. 2018)1 is the first Gaia-only catalog and is based on 22 months of observations. DR2 provides sub- 1% distances to the majority of Kepler planet hosts, enabling more precise stellar and planetary radii. Our paper is organized as follows: Section 2 describes our sample selection. We derive new stellar radii in Sec- tion 3, with 2.7% precision. In Section 4, we derive new planet radii and examine the exoplanet population with our high-resolution sample. We also characterize astrophysical spread in the planet size distribution and note correlations between the exoplanet population and 1 Released on 2018-04-25 stellar mass. We conclude in Section 5, connecting our observations to planet formation theory. 2. INITIAL SAMPLE SELECTION We began with the sample of planet host stars in the CKS sample. The CKS sample selection, spectro- scopic observations, and spectroscopic analysis are de- scribed in detail in P17. In brief, the sample was ini- tially constructed by selecting all Kepler Objects of In- terest (KOIs) brighter than Kp = 14.2 mag. A KOI is a Kepler target star which showed periodic photomet- ric dimmings indicative of planet transits. However, not all KOIs have received the necessary follow-up attention needed to confirm the planets. Over the course of the CKS project, we included additional targets to cover different planet populations, including multi-candidate hosts, ultra-short period candidates, and habitable zone candidates. We cross-matched the CKS sample with the Gaia DR2 catalog by querying all Gaia sources within 1 arcsec of the KIC coordinates. In rare cases, Gaia detected more than one source within 1 arcsec, and we selected the source with the smallest difference between G and Kp magnitudes. We cross-matched 1257 targets in this way. 3. STELLAR RADII 3.1. Introduction Our re-derived stellar radii (R(cid:63)) follow from the Stefan -- Boltzmann law, (cid:18) Lbol (cid:19)1/2 4πσsbT 4 eff R(cid:63) = , (1) where Lbol is the bolometric stellar luminosity, σsb is the Stefan -- Boltzmann constant, and Teff is the effective temperature. Lbol is related to bolometric magnitude Mbol via Lbol = L010−0.4Mbol, (2) where L0 is defined to be L0 = 3.0128 × 1028 W.2 Mbol may be measured from a single broadband photometric apparent magnitude m, if the distance modulus µ, line- of-sight extinction A, and bolometric correction BC are known: Mbol = m − A − µ + BC. (3) Therefore, our derived stellar radii depend on five pa- rameters: m, A, µ, Teff , BC. We discuss the prove- nance of these in parameters in Sections 3.2 -- 3.6 along with their respective contributions to the R(cid:63) error bud- get, which are summarized in Table 1. Section 3.7 ex- plains our detailed modeling of R(cid:63), which closely fol- 2 See IAU 2015 Resolution B2 and Mamajek et al. (2015). CKS-Gaia 3 lows that of Huber et al. (2017). We validate our stel- lar radii through a comparison with asteroseismology in Section 3.8. We also compare our radii to the purely spectroscopic measurements of J17 in Section 3.9 and to those computed by the Gaia project in Section 3.10. 3.2. Photometry We used K-band photometric measurements because dust extinction is less severe in the infrared (see Sec- tion 3.3). The Two Micron All-Sky Survey (2MASS; Skrutskie et al. 2006) measured mK for our target stars with a median precision of 0.02 mag, which corresponds to ≈1% errors in R(cid:63). We elected to use a single photometric band so that our Teff constraints would depend only on spectroscopy. Compared broadband colors, spectroscopy has the ad- vantage that it yields more precise temperatures that are insensitive to reddening. For Kepler field stars, there are significant degeneracies between reddening and pho- tometric Teff that result in uncertainties of ≈200 K (see Pinsonneault et al. 2012 and P17). Given that Teff un- certainties often dominate the final R(cid:63) uncertainty, we restricted our analysis to mK. As an aside, we expect that Gaia DR2 will transform our understanding of the 3D distribution of dust in the Milky Way Galaxy. This will reduce reddening-Teff de- generacies for Kepler field stars, and result in improved measurements of Teff from broadband colors. However, such a dust modeling effort is beyond the scope of this work. 3.3. Extinction We consulted the 3D dimensional dust map of Green et al. (2018) to quantify and correct for K-band extinc- tion. The map tabulates reddening in PS1 and 2MASS passbands as a function of a function of galactic latitude, galactic longitude, and distance. Our median target has a E(B − V ) reddening of 0.048 mag. To convert between between E(B − V ) and Aλ, one must multiply E(B − V ) by an extinction vector Rλ. Green et al. (2018) adopted Rλ from Schlafly et al. (2016) who studied the variation in observed stellar col- ors with reddening. Unfortunately, the Schlafly et al. (2016) methodology is insensitive to the gray compo- nent of the extinction curve, i.e. Rλ → Rλ + b. As a matter of convenience, Green et al. (2018) resolved this ambiguity by setting RW 2 = 0, which implies RK = 0.161. However, if one adopts AH /AK, one may derive b by solving the following system of equations: AH = (RH + b)E(B − V ) AK = (RK + b)E(B − V ). (2008), which yields b = 0.063 and RK = 0.224. The value of AH /AK itself is uncertain. As a point of ref- erence, Indebetouw et al. (2005) found AH /AK = 1.55, which yields b = 0.141 and RK = 0.302. To account for the uncertainty in RK, we add 30% additional fractional uncertainty to AK. The expected K-band extinction ranges from AK = 0.001 -- 0.054 mag, with a median value of AK = 0.011 mag. The low typical extinction highlights the ad- vantage of K-band. Neglecting extinction entirely would result in a ≈0.5% error in R(cid:63) for our median target, which is smaller than other terms in the R(cid:63) error bud- get. For completeness, we incorporated AK derived from the Green et al. (2018) maps into our radius calculations (Section 3.7). 3.4. Parallaxes We used parallaxes from Gaia DR2 and required that the parallax uncertainties be smaller than 10%. The median parallax precision of the remaining 1189 stars is 1.3 % and contributes 1.3 % to our R(cid:63) error budget. The Gaia team recommends adopting systematic error floor of 0.1 mas, which accounts for zero-point and spa- tially correlated systematics. Fortunately, the Kepler field is one of the best characterized regions of the sky and independent methods may be used to measure and correct for these systematics. Zinn et al. (2018) used pre- cise distances to asymptotic giant branch (AGB) stars to measure offsets in Gaia parallaxes for the Kepler field and found that the Gaia parallaxes were too small by 0.053 mas. In this work, we apply a correction of +0.053 mas to the Gaia parallaxes to account for this offset. 3.5. Effective Temperatures Stellar effective temperatures factor into our measure- ment of stellar radii in two ways: through the Stefan -- Boltzmann law (Section 3.1) and through the bolomet- ric corrections (Section 3.6). We used the CKS spec- troscopic Teff which have an internal precision 60 K (P17). We note that offsets of ≈100 K are often ob- served when comparing different spectroscopic catalogs as well as when comparing spectroscopic temperatures to temperatures determined by other techniques, such as the infrared flux method or interferometry (Brewer et al. 2016). Therefore, these temperatures are accurate on an absolute scale to ≈100 K. However, since our radii are all derived using CKS Teff , the Teff precision, rather than its absolute accuracy, factors into the precision of our stellar radii. A precision of 60 K corresponds to ≈ 2% errors on R(cid:63). We adopted AH /AK = 1.74 from Nishiyama et al. 3.6. Bolometric Corrections 4 Fulton & Petigura With mk, AK, and µ we may compute absolute K- band magnitude, MK. Converting MK to Mbol requires a bolometric correction BCK. We computed BCK us- ing the isoclassify package (Huber et al. 2017) which in- terpolates over the MIST grid of bolometric corrections (Choi et al. 2016). For each star, we found the range of BCK consistent with our spectroscopically determined Teff and [Fe/H]. The uncertainties on Teff dominate the uncertainty of BCK because Teff has the largest influ- ence on shape of the stellar SED. For a Sun-like star a 60 K uncertainty translates to a ≈0.03 mag error on BCK or ≈1.5% errors in R(cid:63). Errors on BCK stemming from uncertain log g and [Fe/H] are negligible by com- parison. We note that Teff errors enter into the Stefan- Boltzmann Law and BCK in ways that largely cancel. For our stars, K-band probes the Rayleigh-Jeans tail of the SED, where flux scales like Teff . Therefore, at a fixed mK, Lbol ∝ T 3 eff , which is largely canceled by the T −4 term in Equation 1. The bolometric corrections also include model- dependent errors. (2017) assessed these errors by comparing stellar radii derived from the MIST grids to those derived using the BASTA grids (Casagrande & VandenBerg 2014) and estimated 0.03 mag errors. As with Teff , we expect these model- dependent errors to be largely common-mode and are thus more relevant for the accuracy rather than the pre- cision of our stellar radii. Huber et al. eff 3.7. Detailed Modeling In the previous sections, we enumerated the various measurements that we used to compute R(cid:63) and esti- mated their final contribution to the R(cid:63) error budget, which we summarize in Table 1. To compute the radii, we used the isoclassify package in its "direct" mode (Huber et al. 2017). For each star, we provided iso- classify with Teff , [Fe/H], π, and mK. Then, isoclassify computed the posterior probability on R(cid:63) using Equa- tions 1 -- 3. As an intermediate step, isoclassify must in- fer a distance given the parallax measurement. This is done using Bayesian inference, incorporating an expo- nentially decreasing volume density prior with a length scale of 1.35 kpc, as recommended by Astraatmadja & Bailer-Jones (2016). Figure 1 shows the distribution of the formal R(cid:63) precision, which have a median value of 2.7%. The radii are provided in Table 2. One advantage of deriving radii from the Stefan- Boltzmann law is they are minimally model-dependent; they rely on models only for the bolometric corrections (Section 3.6). A disadvantage is that this analysis does not constrain stellar mass and age. We performed a parallel analysis with isoclassify using its "grid" mode. In this mode, isoclassify computes the range of masses, radii, and ages that are consistent with the observa- tional constraints and the MIST isochrone grids. We include these parameters in Table 2 as a matter of con- venience. We caution that their formal uncertainties do not include systematic uncertainties associated with the MIST models. Such uncertainties are likely largest for the coolest stars in our sample. Finally, we must consider the effects of flux contami- nation from unresolved binaries on our radius measure- ments. If a given target star has a companion within the 2MASS software aperture, typically 4 arcsec (Skrut- skie et al. 2006), the target star appears brighter and we infer a larger radius. In Section 4.2, we screen out con- taminating sources using the Gaia source catalog, which has an effective angular resolution of 0.4 arcsec (Arenou et al. 2018), and existing high resolution follow-up imag- ing. As an additional check, we ran isoclassify in the "grid" mode while providing Teff , log g, [Fe/H], and mK constraints, but no parallax constraints. For each star, isoclassify returned a parallax consistent with the in- put constraints and the MIST models. If the "isochrone parallax" is significantly larger than the Gaia parallax the star is likely an unresolved binary. We include this "isochrone parallax" in Table 2 and recommend using radii where the two parallax measurements are consis- tent to four sigma. 3.8. Validation with Asteroseismic Radii As in Paper-II, we assessed the final precision and ac- curacy of our stellar radii with a comparison to stellar radii derived using asteroseismology. We first compared against radii from Silva Aguirre et al. (2015), S15 here- after, who performed an asteroseismic analysis of 33 Ke- pler planet hosts and achieved a median radius uncer- tainties of ≈1%. Importantly, the S15 analysis mod- eled individual oscillation frequencies, which achieves higher precision than simpler asteroseismic scaling rela- tionships. We compared the stellar radii for the 29 CKS stars in common with the S15 study (Figure 2). Our radii are 0.9% larger on average, with a 1.9% RMS scat- ter in the ratio, which is consistent with the quadrature sum of the formal uncertainties of both sets of radii. Because the S15 radii span a narrow range in R(cid:63) of 0.7 -- 2.0 R(cid:12), we performed a second comparison against radii from Huber et al. (2013), H13 hereafter, which span 0.7 -- 10 R(cid:12). H13 relied on scaling relationships using the small frequency separation δν and peak frequency νmax. These relationships are lower precision than the more detailed analysis of S15 at 3% fractional precision. Our radii are 0.03% smaller on average, with a 3.5% RMS scatter in the ratio, which is consistent with the quadrature sum of the formal uncertainties of both sets of radii. Our comparisons with S15 and H13 show that our stel- CKS-Gaia 5 lar radius precision is comparable to, or smaller than, those from asteroseismology. In principle our method- ology for measure stellar radii can be used to test sys- tematics in the asteroseismic scaling relationships as in (Huber et al. 2017). Such an effort is beyond the scope of this work. 3.9. Comparison with Johnson et al. (2017) Radii Figure 3 compares our radii against those from J17, which relied on spectroscopy alone. The RMS scatter in the ratios is 13.9%, which is consistent with the 11% median uncertainty quoted in J17. We also note that the J17 radii on average fall below the one-to-one line between 1 and 3 R(cid:12). We observed this trend in J17 when comparing the J17 radii to asteroseismic radii. It is not surprising that we observe this same trend in a larger sample given our new radii closely track astero- seismology. This demonstrates the potential for Gaia to serve as a benchmark with which to test synthetic spec- tra and model atmospheres. We also note a handful of outliers in the comparison. These could be due to stars with unresolved companions contributing extra K-band flux and making the CKS+Gaia radii seem larger. They may also be due to rare and unknown failure modes in the spectroscopic analysis of P17. 3.10. Comparison with Gaia DR2 Stellar Radii The Gaia project also provided radii based on SED modeling that fits for effective temperature, extinction, and radius given the known distance. Figure 3 compares our radii with the Gaia project radii for 1077 stars in common. On average Gaia DR2 radii are 3.0% larger than ours with a 6.2% RMS scatter in the ratio, which is consistent with the formal median uncertainty of 6.9% reported in Gaia DR2. Table 1. Star and Planet Properties Parameter Median Value Median Uncertainty Teff mK AK π µ BC R(cid:63) Rp/R(cid:63) Rp 5698 K 12.24 mag 0.011 mag 1.5 mas 9.26 mag −1.46 mag 1.1 R(cid:12) 1.7% 2.1 R⊕ 60 K 0.02 mag 0.004 mag 1.3% 0.01 mag 0.03 mag 2.7% 4.1% 5.2% 4. PLANET POPULATION 4.1. Distribution of Detected Planets Using our updated stellar radii we derived planet radii using the values of Rp/R(cid:63) tabulated in Mullally et al. (2015). We also computed the incident stellar flux Sinc using our updated R(cid:63) and Teff . These Rp and Sinc mea- surements are listed in Table 3. Figure 4 shows the distribution of planets in the P -Rp and Sinc-Rp planes. As in F17, we observe a narrow gap separating two populations of planets at ≈2 R⊕. While the gap is clearly visible in this sample of 1901 planets, we cau- tion that the distribution of detected planets does not convey the underlying distribution of planets, due to se- lection effects that we discuss in Section 4.2. 4.2. Intrinsic Distribution of Planets Here, we measure planet occurrence, the number of planets per star, as a function of P , Rp, and Sinc. In order to measure the intrinsic distribution of planets, we must account for selection effects in the construction of the CKS target list, geometrical transit probability, and pipeline completeness. Our methodology follows that of F17. We first identified a subset of CKS planets drawn from a well-defined population of parent stars by applying the following cuts to our planet sample: 1. Stellar brightness. We restricted our sample to the magnitude-limited CKS subsample, where Kp < 14.2 mag. 2. Stellar radius. We restricted our analysis to dwarf stars where log10 (cid:18) R(cid:63) R(cid:12) (cid:19) < (cid:18) Teff − 5500 K (cid:19) 4000 K + 0.2. (4) 3. Stellar effective temperature. We restricted our planet sample to stars with Teff = 4700 -- 6500 K, where the CKS temperatures are reliable. 4. Isochrone parallax. For each star, we computed an "isochrone parallax" based on Teff , log g, [Fe/H], and mK (see Section 3.7). We removed stars where the Gaia and isochrone parallaxes differed by more than 4σ, due to likely flux contamination by unre- solved binaries. 5. Stellar dilution (Gaia). Dilution from nearby stars can also alter the apparent planetary radii. For each target, we queried all Gaia sources within 8 arcsec (2 Kepler pixels) and computed the sum of their G-band fluxes. The ratio between this cu- mulative flux and the target flux r8 approximates 6 Fulton & Petigura Figure 1. Left: Distribution of fractional stellar radius uncertainties from this work (black) compared to those from Johnson et al. (2017) (grey). Right: Same as left but comparing fractional planet radius uncertainties. Figure 2. Left: Comparison of stellar radii derived from asteroseismology (Silva Aguirre et al. 2015; S15) and spec- troscopy+astrometry (this work) for 29 stars in common. Equality is represented by the dashed green line. Our radii are 0.9% larger on average and there is a 1.9% RMS dispersion in the ratios. Right: same as left but comparing our radii to Huber et al. 2013 (H13). Our radii are 0.03% smaller on average and there is a 3.5% RMS dispersion in the ratios. the Kp-band dilution for each transiting planet. We required that r8 < 1.1. 6. Stellar dilution (imaging). Furlan et al. (2017) compiled high-resolution imaging observations performed by several groups. When a nearby star is detected, Furlan et al. (2017) computed a ra- dius correction factor (RCF), which accounts for dilution assuming the planet transits the bright- est star. We do not apply this correction factor, but conservatively exclude KOIs where the RCF exceeds 5%. 7. Planet false positive designation. We excluded candidates that are identified as false positives ac- cording to P17. 8. Planets with grazing transits. We excluded stars having grazing transits (b > 0.9), which have sus- pect radii due to covariances with the planet size and stellar limb-darkening during the light curve fitting. After applying these cuts, we are left with 907 planets. Where possible, we applied the same filters on stel- lar properties to the Kepler field star population. For 0.51.02.03.05.010.030.0Fractional Stellar Radius Uncertainty [%]0100200300400500600Number of Starsmedian = 11%median = 3%Johnson+17this work1.02.03.05.010.015.030.0Fractional Planet Radius Uncertainty [%]050100150200250300350Number of Planetsmedian = 13%median = 5%Johnson+17this work0.512351020 [CKS+Gaia] (Solarradii)0.512351020 [S15] (Solarradii)0.90.951.01.051.1Ratio (/)0.512351020 [CKS+Gaia] (Solarradii)0.512351020 [H13] (Solarradii)0.90.951.01.051.1Ratio (/)CKS-Gaia 7 Figure 3. Left: Comparison of stellar radii derived from spectroscopy+astrometry (this work) and spectroscopy alone (Johnson et al. 2017, CKS-II). The CKS-II radii are 0.7% larger on average and there is a 13.9% RMS dispersion in the ratios. Right: same as left but comparing our radii to the Gaia DR2 radii. The Gaia DR2 radii are 3.0% larger on average and there is a 6.2% RMS dispersion in the ratios. the stellar radius and temperature filters we used the Gaia DR2 parameters. We could not apply the imag- ing cut to the parent stellar population because it relies on follow-up resources directed specifically at KOIs not at the parent parent population. After filtering, 24981 stars remain. We calculated planet occurrence using the inverse de- tection efficiency methodology IDEM of F17. In brief, we account for the detection sensitivity of the survey using the injection-recovery tests performed by Chris- tiansen et al. (2015). We calculated planet occurrence as the number of planets per star in discrete bins as npl,bin(cid:88) i=1 fbin = 1 N(cid:63) wi. (5) where N(cid:63) = 24981 and wi is the product of the inverse pipeline detection efficiency pdet and the inverse transit probability ptr for each detected planet. Values of wi, pdet, ptr are listed in Table 4. Computing these weights requires knowledge of the distribution of radii and noise properties of stars in the parent stellar sample. As in F17, we used the Combined Differential Photometric Precision computed by the Ke- pler project (Mathur et al. 2017) as our noise metric. Unlike F17, we used the R(cid:63) from Gaia DR2 as opposed to photometric R(cid:63) to characterize the distribution of parent stellar radii. F17 found that plausible statistical and systematic errors of 40% and 25% respectively in the photometric radii of the parent stellar population led to errors in planet occurrence of up to a factor of two at 1.0 R⊕. Our new occurrence measurements have the major advantage that there are negligible differences between the radii of the field stars and planet hosts; thus our occurrence measurements are up to twice as precise. The IDEM has been used in a number of previous works (e.g. Howard et al. 2010; Howard et al. 2012; Mor- ton & Swift 2014; Fulton et al. 2017). While our results depend on the relative occurrence of planets as a func- tion of host star mass, we wish to remind the reader that additional care is required when computing absolute oc- currence in regions of low completeness. Hsu et al. (2018) performed a comparison of various occurrence estimators including the IDEM used in this study. Hsu et al. (2018) found that the IDEM is a bi- ased estimator in the limit of low completeness. In brief, the bias arises because fluctuations due to photometric noise cause transits to appear larger/smaller, resulting in larger/smaller planet radii. However, near the de- tection limit, there is a bias toward detecting the ap- parently larger planets. Hsu et al. (2018) recommends adopting a different estimator based on approximate bayesian computation (ABC). The ABC method has the advantage that it is less biased in regions of low com- pleteness, but requires significantly more computational effort compared to the IDEM. In this work, we restricted our occurrence analysis to 0.512351020 [CKSII] (Solarradii)0.512351020 [CKS+Gaia] (Solarradii)0.80.91.01.11.2Ratio (/)0.512351020 [GaiaDR2] (Solarradii)0.512351020 [CKS+Gaia] (Solarradii)0.80.91.01.11.2Ratio (/)8 Fulton & Petigura domains of P , Rp, and Sinc where pipeline completeness exceeds 25% for our sample of 1189 stars. Regions that do not meet this threshold are shaded in 5 and shown as gray triangles in Figure 6. We placed an upper bound on the bias introduced by the IDEM estimator, by consider- ing several regions abutting our completeness cutoff. We used the SysSim code from Hsu et al. (2018) to estimate the occurrence using ABC. All values were consistent to within 1 σ. Figure 5 shows the radius distribution of close-in plan- ets, i.e. the number of planets per star with orbital periods less than 100 days. Despite the increased pre- cision relative to F17, the one-dimensional distribution of planet sizes is qualitatively the same. We confirm the presence of a gap in the occurrence distribution of planet radii at 1.5 -- 2.0 R⊕, as seen in F17 and several subsequent works (see, e.g., Van Eylen et al. 2017). The relative heights of the two peaks are similar indicating that the frequency of super-Earths and sub-Neptunes are nearly equal over the full period range analyzed in this work (0-100 days). We do not resolve additional small scale structure in the radius distribution, and the depth of the gap relative to the sub-Neptune and super- Earths peaks remains largely unchanged. This suggests that we are resolving the astrophysical scatter in the radii of the super-Earth and sub-Neptune populations, and that the gap is not completely devoid of planets. We quantify this astrophysical scatter in Section 4.3. Figure 6 shows the two-dimensional occurrence distri- bution of planet radii as a function of orbital period and insolation flux. The contours show the relative planet occurrence computed using the weighted kernel density estimator (wKDE) described in F17. We used a Gaus- sian kernel spanning 40% in Sinc and 40% P and 5% in Rp. Smaller kernels offer higher resolution, but nois- ier contours due to few detected planets. Readers may create their own occurrence contours using the weights provided in Table 4. The radius gap is wider and more empty at P (cid:38) 30 days and Sinc (cid:46) 50 S⊕. While there was tentative evidence for this in F17, the smaller Rp uncertainties lend more confidence to this observation. The gap also appears to slope downward with increasing orbital pe- riod, which is consistent with the observations of Van Eylen et al. (2017) and with several theoretical models discussed in Section 5. 4.3. Intrinsic Spread in the Sizes of Super-Earths and Sub-Neptunes Here, we consider the intrinsic astrophysical spread in the population of super-Earths and sub-Neptune plan- ets and whether planets that appear to reside in the gap spanning 1.5 to 2.0 R⊕ could be explained by measure- ment uncertainties alone. Previous studies have strug- gled to measure the occurrence of planets over this nar- row region of planet size. In F17, planet radius uncer- tainties were marginally smaller than the width of the gap, and the Van Eylen et al. (2017) asteroseismic anal- ysis suffered from a small number of detected planets. We show the filtered distribution of planets in Fig- ure 7, and we identify two fiducial planet classes, "super- Earths" and "sub-Neptunes," as well as the radius gap. The large number of planets (129) residing within the gap appear to be inconsistent with scatter from above and below. To test this, we constructed a toy model to assess whether they could be explained by measurement uncertainties alone. In or toy model, we took the observed planet de- tections and assigned them to one of the two planet classes, based on whether they resided in one of the two boxes shown in Figure 7. For each super-Earth and sub- Neptune we assigned a new radius from uniform distri- butions with centers at 1.2 R⊕ or 2.4 R⊕, respectively, which correspond to the locations of the observed peaks in the radius distribution. The orbital periods were re- tained from the actual detections. The fractional width shared by both distributions W is a free parameter in this model. computed a figure of merit FOM =(cid:80) We simulated planet detections over a range of W and i(Nreal,i−Nsim,i)2, where N is the number of detections within box i (plot- ted in Figure 7). The FOM and visual inspection identified an intrinsic spread of W = 60% as a good match to the data (see Figure 7). In our best-fitting toy model, the super-Earth and sub-Neptune populations span 0.85 to 1.55 R⊕ and 1.7 to 3.1 R⊕, respectively; it is inconsistent with a pop- ulation devoid of planets between 1.5 and 2.0 R⊕. As a limiting case, we show a model with W = 40% in Figure 7. Here, the super-Earth and sub-Neptune popu- lation span 0.96 to 1.44 R⊕ and 1.92 to 2.88 R⊕, respec- tively, which approximate the upper and lower bound- aries of the gap. This model produces a much emptier gap and is an obvious mismatch with the observations. We recognize that this toy model does not capture the detailed radius distribution of planets, most notably the tail of planets larger than 3 R⊕. A more detailed study might use a different distribution to model the planet radii, such as a Gaussian, Rayleigh, or a non-parametric distribution. Nonetheless, we have constrained the in- trinsic dispersion in the size of the super-Earth and sub- Neptune populations to be ≈60%. While there is a dip in the occurrence of close-in planets from 1.5 to 2.0 R⊕, occurrence does not fall to zero. These interpretations were not possible in previous studies, due to larger ra- dius uncertainties or limited numbers of detected plan- ets. CKS-Gaia 9 Figure 4. Left: distribution of planet radii and orbital periods. Right: same as left but with insolation flux relative to Earth on the horizontal axis. In both plots, an underdensity of points appears between 1.5 and 2.0 R⊕. 4.4. Trends with Host-Star Mass We plot planet size vs. stellar mass in Figure 8 in order to investigate potential changes in the structure of the planet radius distribution as a function of stellar host mass. This Figure shows that the transition radius be- tween the two populations increases monotonically with stellar mass. The gap occurs near 1.6 R⊕ for planets or- biting host stars with masses near 0.8 M(cid:12) and moves to ≈2.0 R⊕ for planets orbiting stars with masses above 1.2 M(cid:12). We also split the sample into three bins of stellar mass: M(cid:63) < 0.96 M(cid:12), M(cid:63) = 0.96 -- 1.11 M(cid:12), and M(cid:63) > 1.11 M(cid:12). We chose bin boundaries such that the three bins captured equal numbers of planets. Figure 9 shows the planet population in the P -Rp and Sinc-Rp planes for each of the three mass bins. The gap is clearly vis- ible in each of the three stellar mass bins, and appears wider than the gap from the combined sample shown in Figure 6. We observe several trends with stellar mass. First, the typical size of super-Earth and sub-Neptune planets in- creases with increasing stellar mass, an observation that we quantify later in this section. This explains why the planet populations are better separated when one con- siders a narrow range of stellar mass; when all three mass groups are combined the distributions overlap. It also helps to clarify why the planet populations in Van Eylen et al. (2017) seemed to be more separated com- pared to those in F17. The asteroseismic sample was heavily weighted toward stars more massive than the sun, and is more directly comparable to the P -- Rp dis- tribution of our high mass bin. The top right panel of our Figure 9 is a closer match to Figure 2 from Van Eylen et al. (2017) than the upper left panel of our Figure 9. To quantify the change in typical planet size with stel- lar mass, we calculated the mean planet radius for sub- Neptunes (1.7 -- 4.0 R⊕, and P < 100 days) and super- Earths (1.0 -- 1.7 R⊕, and P < 30 days). We weighted each radius by the wi weights used in the occurrence calculations, described in Section 4.2. Figure 10 shows these mean planet parameters as a function of stellar mass. Consistent with visual inspection of Figure 9, we see monotonically increasing planet size with increasing stellar host star mass in both the super-Earth and sub- Neptune planets. Although the trend with stellar mass is strong, we caution that stellar metallicity may be a confounding factor. More massive stars are younger on average and are thus more metal-rich due to galactic chemical en- richment. Indeed, Petigura et al. (2018) observed a cor- relation between planet size and host star metallicity in the CKS sample and this correlation has been ob- served previously in many different samples (e.g. Santos et al. 2004; Fischer & Valenti 2005; Sousa et al. 2008; Ghezzi et al. 2010; Buchhave et al. 2014; Schlaufman 2015; Wang & Fischer 2015). The solid component of the protoplanetary disk likely tracks both stellar metal- licity and stellar mass. Therefore, we expect planet size to be correlated with both stellar mass and metallicity. Future studies spanning a larger range of stellar mass and metallicity are necessary to resolve this ambiguity. Previous studies have noted a desert of highly- irradiated sub-Neptune planets (see, e.g., Lundkvist et al. 2016 and Mazeh et al. 2016). We observe this sub-Neptune desert in our three mass bins (Figure 9), but note that it shifts to higher incident stellar flux around high mass stars. This trend is highly significant. 0.10.31310301003001000Period [days]0.30.50.7123571020Planet Size [Earth radii]100101102103104Stellar light intensity relative to Earth0.30.50.7123571020Planet Size [Earth radii]10 Fulton & Petigura 5. DISCUSSION AND CONCLUSIONS We analyzed the Kepler planet population after im- proving the radius precision of the stellar hosts and their planets. This improvement leveraged both CKS spec- troscopy and Gaia DR2 parallaxes. Our median stellar radius precision is now 2% compared to 11% in J17. The uncertainty in our planet radii are now typically 5% and are limited by uncertainties in the Kepler transit mod- eling, rather than the stellar radius uncertainties as in J17. With these improved planet radii, we examined the population of small planets at higher resolution. We confirmed the existence of the F17 radius gap between 1.5 and 2.0 R⊕, with more precise and independently derived planet radii. The overall radius distribution is similar to that of F17, which demonstrates that we are resolving the intrinsic spread of the super-Earth and sub-Neptune populations, which span ≈60% in radius. We also demonstrated that the gap from 1.5 to 2.0 R⊕ is not devoid of planets, a conclusion previously obscured by measurement uncertainties or small sample sizes. We observed a correlation of the average planet size and av- erage insolation flux with stellar host mass. However, there is no significant correlation in the average orbital period as a function of average stellar host mass. Here, we interpret our findings in the context of two theories that have been proposed to explain the distri- bution planets between the size of Earth and Neptune: 1. Mass loss by photoevaporation. In this mechanism, X-ray and UV radiation heats the outer layers of a planet's envelope and drives mass loss. Several groups considered photoevaporation and predicted the planet radius gap before it was observed in F17, including Owen & Wu (2013), Lopez & Fort- ney (2013), Jin et al. (2014), and Chen & Rogers (2016). Following F17, Owen & Wu (2017) de- veloped additional analytic photoevaporation the- ory and performed a population synthesis analysis comparing their simulated populations to the F17 occurrence measurements. 2. Core-powered mass loss. In this mechanism, lumi- nosity from a cooling rocky core heats a planet's envelope and drives mass loss. Ginzburg et al. (2016) developed the theory of core-powered mass loss and computed mass loss rates. Ginzburg et al. (2018) performed a population synthesis with com- parisons to the F17 radius distribution and demon- strated that core-powered mass loss could explain the bimodal radius distribution. Both theories can explain a bimodal population of planet sizes composed of two subpopulations: a pop- ulation of bare rocky cores and a population with H/He Figure 5. The distribution of close-in planet sizes. The top panel shows the distribution from Fulton et al. (2017) and the bottom panel is the updated distribution from this work. The solid line shows the number of planets per star with or- bital periods less than 100 days as a function of planet size. A deep trough in the radius distribution separates two pop- ulations of planets with Rp > 1.7 R⊕ and Rp < 1.7 R⊕. As a point of reference, the dotted line shows the size distribu- tion of detected planets, before completeness corrections are made arbitrarily scaled for visual comparison. Figure 10 shows the average Sinc as function M(cid:63). The mean Sinc for both the super-Earths and sub-Neptunes increases by 3× over a relatively narrow range of aver- age M(cid:63), 0.85 M(cid:12) to 1.2 M(cid:12). One explanation could be that the orbital periods of small planets decreases with stellar mass. However, the mean orbital periods for the three different mass bins are consistent to ≈30%. Figure 11 shows the cumulative fraction of hot (Sinc = 30 -- 3000 S⊕) sub-Neptune size planets (Rp = 1.7 -- 4.0 R⊕) as a function of insolation flux, highlighting the shift of the Sinc sub-Neptune desert with stellar mass. In the high stellar mass sample, 10% of planets have Sinc > 300 S⊕. For the low stellar mass sample, one must include planets out to 150 Sinc to encompass 10% of the population. 0.71.01.31.82.43.54.56.08.012.020.0Planet Size [Earth radii]0.000.020.040.060.080.100.120.140.16Number of Planets per Star(Orbital period < 100 days)typicaluncert.0.71.01.31.82.43.54.56.08.012.020.0Planet Size [Earth radii]0.000.020.040.060.080.100.12Number of Planets per Star(Orbital period < 100 days)typicaluncert.CKS-Gaia 11 Figure 6. Left: Two-dimensional distribution of planet size and orbital period. The median uncertainty is plotted in the upper left. Right: same as left but with insolation flux on the horizontal axis. In both plots, the two peaks in the population as observed by F17 are clearly visible, but with greater fidelity. Figure 7. Toy model demonstrating that the two populations of planets have intrinsic widths. Left: Real planet detections with boxes demarking the boundaries defined for the population of large planets (Rp = 2.0 -- 4.0 R⊕), small planets (Rp = 0.7 -- 1.5 R⊕), and the gap between them (Rp = 1.5 -- 2.0 R⊕). We find that the data is well-described by two populations with a 60% intrinsic spread in their radii (middle). Decreasing that width to 40% is a clear mismatch to the data (right). Our toy model is described in Section 4.3. envelopes with mass fractions of a few percent. Because both mass loss mechanisms are more efficient at high levels of incident stellar flux, they both predict that the population of sub-Neptunes should be offset to lower in- solation fluxes compared to the super-Earths A key difference between the two mechanisms is the expected dependence on stellar mass. Core-powered mass loss depends only on properties of the planet and bolometric incident stellar flux. All else being equal, this mechanism predicts no dependence of the planet popu- lation as a function of M(cid:63). In contrast, the efficiency of photoevaporation depends on the time-integrated XUV flux, or "fluence." This quantity is a strong function of (Jackson et al. 2012). Therefore, photoevaporation predicts that the population of sub-Neptunes should shift to lower Sinc stellar mass since(cid:82) (LX /Lbol)dt ∝ M−3 (cid:63) with decreasing stellar mass, due to increased activity around lower mass stars. The shifts in the Sinc-Rp dis- tribution of planets with M(cid:63) are consistent with this prediction from photoevaporation. (cid:63) The lack of a strong P -- M(cid:63) dependence is also consis- tent with photoevaporation. Owen & Wu (2017) showed that the mass loss timescale t = M/ M ∝ P 1.4M−0.48 ∝ inc M 2.2 S1.06 (cid:63) . Photoevaporation thus has a steeper depen- dence on M(cid:63) at fixed Sinc than at fixed P . This naturally explains why we see a strong trend in planet Sinc with stellar mass and no significant trend with P in Figure 10. Other super-Earth formation mechanisms have been proposed that could potentially produce a gap in the size distribution including delayed formation in a gas- poor disk (e.g. Lee et al. 2014; Lee & Chiang 2016), and sculpting by giant impacts (e.g. Liu et al. 2015; 103010030010003000Stellar light intensity relative to Earth1.01.52.43.5Planet Size [Earth radii]typicaluncert.0.0000.0050.0100.0150.0200.0250.030Relative Occurrence0.31310301000.711.52346Planet Size [Earth radii]Real detections0.3131030100Orbital Period [days]Simulation (width=60.0%)0.3131030100Simulation (width=40.0%)12 Fulton & Petigura Figure 8. Two-dimensional distribution of stellar mass and planet size. The median uncertainty is plotted in the upper left. As we see in Figure 9, the position of the gap, and the population of planets on either side of the gap increases monotonically with increasing stellar mass. We plot a dashed line at the location of the gap to guide the eye. For stars with masses of ≈0.8 M(cid:12) the gap falls at ≈1.6 R⊕, while for host stars with masses of ≈1.2 M(cid:12) the gap occurs at ≈2.0 R⊕. The peaks of the two populations of planets on either side of the gap also shift in the same way. Planets smaller than 4 R⊕ tend to be larger around more massive stars and the same is true for the gap between the two populations. The relative occurrence rate between the two populations remains constant for all stellar masses analyzed in this work. Schlichting et al. 2015; Inamdar & Schlichting 2016). Formation in a gas-poor disk without any sculpting by photoevaporation would produce a gap radius that does not change with orbital period and is inconsistent with the results of Van Eylen et al. (2017). Sculpting by gi- ant impacts alone predicts that the gap radius would be found at larger radii at longer orbital periods (Lopez & Rice 2016) which is the opposite of the trend found by Van Eylen et al. (2017). We interpret the observed stellar mass depen- dence of the planet population as evidence supporting photoevaporation model. However, these mechanisms are not mutually exclusive and all of them could be op- erating simultaneously or or at different times during the formation of planetary systems. If photoevapora- tion is the dominate mechanism for sculpting planet en- velopes, one may fit the observed distribution of planets to constrain important quantities like the distribution of planet core masses, envelope fractions, and core compo- sitions (Owen & Wu 2017). Due to the magnitude-limited nature of CKS, our analysis was restricted to a fairly narrow range in M(cid:63), spanning 0.85 to 1.2 M(cid:12). Previous studies of the radius distribution of planets orbiting M dwarfs have shown that these planets tend to be smaller on average (Morton & Swift 2014; Dressing & Charbonneau 2015). This may be an extension of the stellar mass dependence of the planet population observed in this work. However, no study of planets orbiting low-mass stars to date has de- tected a gap in the radius distribution. Such a detection (or lack thereof) would reveal insights into the structure and formation of planets around low-mass stars. This motivates future high precision studies of large samples of planets orbiting K and M dwarfs. Such studies would provide additional leverage on M(cid:63) to test the depen- dence of the planet population on stellar mass and to constrain the mechanisms that form and sculpt planets. Facility: Keck:I (HIRES), Kepler , Gaia We thank the Kepler and Gaia teams for years of work making these precious datasets possible. We are grateful for Andrew Howard's guidance and comments on our manuscript. We thank Eddie Schlafly and Gre- gory Green for guidance regarding the treatment of dust extinction. We thank Alberto Krone-Martins for discus- sions regarding the Gaia mission and its data products. Daniel Huber kindly assisted with the isoclassify pack- 0.81.01.21.5Stellar Mass [Solar masses]1.01.52.43.5Planet Size [Earth radii]typicaluncert.0.000.010.020.030.04Relative Density of PlanetsCKS-Gaia 13 Figure 9. Top row: the two-dimensional distribution of planet size and orbital period for three bins of stellar mass. The typical size of super-Earths (Rp = 1.0 -- 1.7 R⊕) and sub-Neptunes (Rp = 1.7 -- 4.0 R⊕) increases with stellar mass while typical orbital periods are roughly constant. Bottom row: same as top row, but with insolation flux on the horizontal axis. The population of super-Earths and sub-Neptunes shifts to higher incident flux for higher mass stars. age. We thank D. Hsu and E. Ford for help using their SysSym code. EAP acknowledges support from Hubble Fellowship grant HST-HF2-51365.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc. for NASA under contract NAS 5-26555. This research has made use of NASA's Astrophysics Data System Space Agency This work has made use of data from the (ESA) mission Gaia European (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/ dpac/consortium). Funding for the DPAC has been provided by national in particular the institutions, institutions participating in the Gaia Multilateral Agreement. Finally, the authors wish to recognize and acknowl- edge the very significant cultural role and reverence that the summit of Maunakea has long had within the in- digenous Hawaiian community. We are most fortunate to have the opportunity to conduct observations from this mountain. Software: All code used in this paper is available at https://github.com/California-Planet-Search/ cksgaia/. We made use of the following publicly available Python modules: astropy (Astropy Collab- oration et al. 2013), isoclassify (Huber et al. 2017), lmfit (Newville et al. 2014), matplotlib (Hunter 2007), numpy/scipy (van der Walt et al. 2011), and pandas (McKinney 2010). REFERENCES Arenou, F., Luri, X., Babusiaux, C., et al. 2018, ArXiv e-prints, arXiv:1804.09375 Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977 Brewer, J. M., Fischer, D. A., Valenti, J. A., & Piskunov, N. Astraatmadja, T. L., & Bailer-Jones, C. A. L. 2016, ApJ, 833, 2016, ApJS, 225, 32 119 Brown, T. M., Latham, D. W., Everett, M. E., & Esquerdo, Astropy Collaboration, Robitaille, T. P., Tollerud, E. J., et al. G. A. 2011, AJ, 142, 112 2013, A&A, 558, A33 Buchhave, L. A., Bizzarro, M., Latham, D. W., et al. 2014, Borucki, W. J. 2016, Reports on Progress in Physics, 79, 036901 Nature, 509, 593 1310301001.01.52.43.5Planet Size [Earth radii]typicaluncert.<.131030100Orbital period [days]typicaluncert...131030100typicaluncert.>.0.0000.0050.0100.0150.020Relative Occurrence1030100300100030001.01.52.43.5Planet Size [Earth radii]typicaluncert.<.103010030010003000Stellar light intensity relative to Earthtypicaluncert...103010030010003000typicaluncert.>.0.0000.0050.0100.0150.020Relative Occurrence14 Fulton & Petigura Figure 10. Mean planet properties as a function of mean stellar mass for super-Earths and sub-Neptunes (left and right columns respectively). The top, middle, and bottom rows show the weighted average of planet radius, insolation flux, and orbital period, respectively. Planets around more massive stars tend to be larger and hotter than those around lower mass stars, but their orbital periods are similar. Casagrande, L., & VandenBerg, D. A. 2014, MNRAS, 444, 392 Chen, H., & Rogers, L. A. 2016, ApJ, 831, 180 Choi, J., Dotter, A., Conroy, C., et al. 2016, ApJ, 823, 102 Christiansen, J. L., Clarke, B. D., Burke, C. J., et al. 2015, ApJ, 810, 95 Dressing, C. D., & Charbonneau, D. 2015, ApJ, 807, 45 Fischer, D. A., & Valenti, J. 2005, ApJ, 622, 1102 Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81 Gaia Collaboration, Prusti, T., de Bruijne, J. H. J., et al. 2016a, A&A, 595, A1 Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2016b, A&A, 595, A2 Ghezzi, L., Cunha, K., Smith, V. V., et al. 2010, ApJ, 720, 1290 Ginzburg, S., Schlichting, H. E., & Sari, R. 2016, ApJ, 825, 29 -- . 2018, MNRAS, 476, 759 Green, G. M., Schlafly, E. F., Finkbeiner, D., et al. 2018, ArXiv e-prints, arXiv:1801.03555 Fulton, B. J., Petigura, E. A., Howard, A. W., et al. 2017, AJ, Holman, M. J., Fabrycky, D. C., Ragozzine, D., et al. 2010, 154, 109 Science, 330, 51 Furlan, E., Ciardi, D. R., Everett, M. E., et al. 2017, The Howard, A. W., Marcy, G. W., Johnson, J. A., et al. 2010, Astronomical Journal, 153, 71 Science, 330, 653 Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, Howard, A. W., Marcy, G. W., Bryson, S. T., et al. 2012, ApJS, ArXiv e-prints, arXiv:1804.09365 201, 15 0.80.91.01.11.21.31.261.281.301.321.341.361.38average planet size [R]0.0 < P 30 days1.0 < R 1.7 R0.80.91.01.11.21.32.252.302.352.402.452.502.552.602.650.0 < P 100 days1.7 < R 4.0 R0.80.91.01.11.21.3102030405060700.0 < P <= 100 days1.7 < R <= 4.0 R0.80.91.01.11.21.350100150200250300average insolation flux [S]0.0 < P <= 30 days1.0 < R <= 1.7 R0.80.91.01.11.21.3average stellar mass [M]2426283032340.0 < P <= 100 days1.7 < R <= 4.0 R0.80.91.01.11.21.3average stellar mass [M]8.08.59.09.510.010.511.011.512.012.5average orbital period [days]0.0 < P <= 30 days1.0 < R <= 1.7 RCKS-Gaia 15 Lopez, E. D., & Rice, K. 2016, ArXiv e-prints, arXiv:1610.09390 Lundkvist, M. S., Kjeldsen, H., Albrecht, S., et al. 2016, Nature Communications, 7, 11201 Mamajek, E. E., Torres, G., Prsa, A., et al. 2015, ArXiv e-prints, arXiv:1510.06262 Marcy, G. W., Isaacson, H., Howard, A. W., et al. 2014, ApJS, 210, 20 Mathur, S., Huber, D., Batalha, N. M., et al. 2017, ApJS, 229, 30 Mazeh, T., Holczer, T., & Faigler, S. 2016, A&A, 589, A75 McKinney, W. 2010, in Proceedings of the 9th Python in Science Conference, ed. S. van der Walt & J. Millman, 51 -- 56 Morton, T. D., & Swift, J. 2014, ApJ, 791, 10 Mullally, F., Coughlin, J. L., Thompson, S. E., et al. 2015, ApJS, 217, 31 Newville, M., Stensitzki, T., Allen, D. B., & Ingargiola, A. 2014, LMFIT: Non-Linear Least-Square Minimization and Curve-Fitting for Python, , , doi:10.5281/zenodo.11813. https://doi.org/10.5281/zenodo.11813 Nishiyama, S., Nagata, T., Tamura, M., et al. 2008, ApJ, 680, 1174 Owen, J. E., & Wu, Y. 2013, ApJ, 775, 105 -- . 2017, ApJ, 847, 29 Petigura, E. A., Howard, A. W., & Marcy, G. W. 2013, Proceedings of the National Academy of Science, 110, 19273 Petigura, E. A., Howard, A. W., Marcy, G. W., et al. 2017, AJ, 154, 107 Petigura, E. A., Marcy, G. W., Winn, J. N., et al. 2018, AJ, 155, 89 Pinsonneault, M. H., An, D., Molenda- Zakowicz, J., et al. 2012, Figure 11. Cumulative distribution of planets as a function of insolation flux. The solid lines are corrected for complete- ness and the dashed lines are not corrected. Planets residing in very high insolation flux environments tend to be more rare around low mass stars. For example, ∼20% of the planet population orbiting stars more massive than 1.11 M(cid:12) have Sinc > 200 S⊕, compared to only ∼10% of planets orbiting stars less massive than 0.96 M(cid:12). Hsu, D. C., Ford, E. B., Ragozzine, D., & Morehead, R. C. 2018, AJ, 155, 205 Huber, D., Chaplin, W. J., Christensen-Dalsgaard, J., et al. ApJS, 199, 30 2013, ApJ, 767, 127 Huber, D., Silva Aguirre, V., Matthews, J. M., et al. 2014, ApJS, 211, 2 Rogers, L. A. 2015, ApJ, 801, 41 Santos, N. C., Israelian, G., & Mayor, M. 2004, A&A, 415, 1153 Schlafly, E. F., Meisner, A. M., Stutz, A. M., et al. 2016, ApJ, Huber, D., Zinn, J., Bojsen-Hansen, M., et al. 2017, ApJ, 844, 821, 78 102 Hunter, J. D. 2007, Computing In Science & Engineering, 9, 90 Inamdar, N. K., & Schlichting, H. E. 2016, ApJL, 817, L13 Indebetouw, R., Mathis, J. S., Babler, B. L., et al. 2005, ApJ, 619, 931 Schlaufman, K. C. 2015, ApJL, 799, L26 Schlichting, H. E., Sari, R., & Yalinewich, A. 2015, Icarus, 247, 81 Silva Aguirre, V., Davies, G. R., Basu, S., et al. 2015, MNRAS, 452, 2127 Jackson, A. P., Davis, T. A., & Wheatley, P. J. 2012, MNRAS, Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 422, 2024 Jin, S., Mordasini, C., Parmentier, V., et al. 2014, ApJ, 795, 65 Johnson, J. A., Petigura, E. A., Fulton, B. J., et al. 2017, ArXiv e-prints, arXiv:1703.10402 Lee, E. J., & Chiang, E. 2016, ApJ, 817, 90 Lee, E. J., Chiang, E., & Ormel, C. W. 2014, ApJ, 797, 95 Lindegren, L., Lammers, U., Bastian, U., et al. 2016, A&A, 595, A4 Liu, S.-F., Hori, Y., Lin, D. N. C., & Asphaug, E. 2015, ApJ, 812, 164 1163 Sousa, S. G., Santos, N. C., Mayor, M., et al. 2008, A&A, 487, 373 van der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Computing in Science Engineering, 13, 22 Van Eylen, V., Agentoft, C., Lundkvist, M. S., et al. 2017, ArXiv e-prints, arXiv:1710.05398 Wang, J., & Fischer, D. A. 2015, AJ, 149, 14 Weiss, L. M., & Marcy, G. W. 2014, ApJL, 783, L6 Zinn, J. C., Pinsonneault, M. H., Huber, D., & Stello, D. 2018, Lopez, E. D., & Fortney, J. J. 2013, The Astrophysical Journal, ArXiv e-prints, arXiv:1805.02650 776, 2 Table 2. Stellar Properties [Fe/H] mK π KOI Teff K K00001 5819 K00002 6449 dex 0.01 0.20 mag mas R(cid:12) M(cid:12) R Miso Riso R(cid:12) ρiso ageiso πspec r8 RCF g/cc dex mas 9.8 9.3 4.67 1.04 2.96 1.97 K00006 6348 0.04 11.0 2.13 1.30 K00007 K00008 5827 0.18 5891 −0.07 10.8 2.07 1.51 11.0 3.01 0.94 0.99 1.51 1.20 1.15 1.00 1.04 1.96 1.28 1.51 0.93 0.87 0.20 0.57 0.34 1.23 9.74 9.25 9.32 9.78 9.18 4.76 1.00 1.010 3.66 1.00 1.003 2.20 1.01 2.11 1.00 2.90 1.00 1.001 · · · · · · Table 2 continued 3010030010003000Stellar light intensity relative to Earth (S)0.00.20.40.60.81.0Fraction of planets with S<.<<<<>...<.16 Fulton & Petigura Table 2 (continued) KOI Teff [Fe/H] mK π mag mas R(cid:12) M(cid:12) R Miso Riso R(cid:12) ρiso ageiso πspec r8 RCF g/cc dex mas K dex 6181 −0.08 0.36 5660 K00010 K00017 12.3 1.00 1.54 11.6 1.73 1.26 K00018 6332 0.02 11.8 1.14 1.74 K00020 5926 0.03 12.1 1.16 1.50 K00022 5891 0.21 12.0 1.41 1.25 K00041 5854 0.10 9.8 3.31 1.53 K00046 5661 0.39 12.0 1.10 1.66 1.15 1.09 1.31 1.09 1.12 1.10 1.24 1.53 1.25 1.69 1.50 1.25 1.52 1.64 0.32 0.56 0.27 0.32 0.58 0.31 0.28 9.67 9.81 9.46 9.83 9.67 9.84 9.72 1.21 1.01 1.54 1.03 1.25 1.02 1.10 1.00 1.31 1.00 3.41 1.00 1.19 1.00 1.001 · · · 1.005 · · · · · · 1.008 · · · Note -- Properties of planet hosting stars. Teff and [Fe/H] are from P17, mK is the K-band apparent magnitude from 2MASS (Skrutskie et al. 2006, see Section 3.2), π is the trigonometric parallax from Gaia DR2 (Gaia Collaboration et al. 2018, see Section 3.4). The following quantities are described in Section 3.7: R is the adopted stellar radius, computed using the Stefan-Boltzmann law; stellar properties with the 'iso' subscript incorporate constraints from the MIST isochrones; and πspec is the "spectroscopic parallax." r8 encodes contaminating flux from neighboring stars within 8 arcsec in G-band (see Section 4.2). RCF is the "radius correction factor" computed by Furlan et al. (2017) (see Section 4.2). Table 2 is published in its entirety in machine-readable format. A portion is shown here for guidance regarding its form and content. Table 3. Planet Properties Planet candidate K00001.01 K00002.01 K00006.01 K00007.01 K00008.01 K00010.01 K00017.01 K00018.01 K00020.01 K00022.01 P d 2.5 2.2 1.3 3.2 1.2 3.5 3.2 3.5 4.4 7.9 Rp/R(cid:63) Rp R⊕ a AU Sinc S⊕ 0.124 14.14 0.036 882 0.075 16.25 0.038 4161 0.294 41.94 0.025 3852 0.025 0.019 4.08 0.045 1180 1.90 0.022 2031 0.094 15.70 0.047 1375 0.095 13.10 0.044 753 0.080 15.23 0.050 1759 0.118 19.38 0.054 0.094 12.85 0.081 K00041.01 12.8 K00041.02 6.9 K00041.03 35.3 K00046.01 K00046.02 3.5 6.0 0.014 0.008 0.009 0.033 0.007 2.34 0.111 1.34 0.073 1.54 0.217 5.97 0.048 1076 1.24 0.070 520 848 260 201 459 52 Note -- Planetary properties. Period P and planet-to-star ra- dius ratio Rp/R(cid:63) are from Mullally et al. (2015). Planet size Rp, semi-major axis a, and incident stellar flux relative to Earth Sinc are derived from the updated stellar properties in Table 2. Table 2 is published in its entirety in machine- readable format with full numerical precision and uncertain- ties. A portion is shown here for guidance regarding its form and content. CKS-Gaia 17 Table 4. Planet Detection Statistics Planet SNR Detection probability Transit probability Weight candidate mi K00958.01 186.24 K04053.01 21.03 K04212.02 K04212.01 K01001.01 K01001.02 K02534.01 K02534.02 K02403.01 K00988.01 8.77 16.53 37.27 15.49 22.64 11.91 17.98 60.03 pdet 0.97 0.77 0.81 0.93 0.99 0.96 0.94 0.84 0.79 0.97 ptr 0.02 0.17 0.05 0.08 0.03 0.01 0.11 0.08 0.04 0.04 1/wi 49.24 7.71 22.85 13.79 32.14 75.81 9.37 15.49 29.89 28.79 Note -- Table 4 is available in its entirety in machine-readable format. A portion is shown here for guidance regarding its form and content. This table contains only the subset of planet detections that passed the filters described in Section 4.
1811.09174	1	1811	2018-11-22T13:56:58	Search for Exoplanetary Transits in the Galactic Bulge	[ "astro-ph.EP" ]	A search for extrasolar planetary transits using the extended Kepler mission (K2) campaigns 9 and 11 revealed five new candidates towards the Galactic bulge. The stars EPIC 224439122, 224560837, 227560005, 230778501 and 231635524 are found to have low amplitude transits consistent with extrasolar planets, with periods P = 35.1695, 3.6390, 12.4224, 17.9856, and 5.8824 days, respectively. The K2 data and existing optical photometry are combined with the multi-band near-IR photometry of the VVV survey and 2MASS in order to measure accurate physical parameters for the host stars. We then measure the radii of the new planet candidates from the K2 transit light curves and also estimate their masses using mass-radius relations, concluding that two of these candidates could be low mass planets, and three could be giant gaseous planets.	astro-ph.EP	astro-ph	MNRAS 000, 1 -- 7 (2018) Preprint 26 November 2018 Compiled using MNRAS LATEX style file v3.0 Search for Exoplanetary Transits in the Galactic Bulge C.C.Cort´es,1(cid:63) D. Minniti2,3,4 , and S. Villanova1 1Departamento de Astronom´ıa, Casilla 160-C, Universidad de Concepci´on, Concepci´on, Chile. 2Departamento de F´ısica, Facultad de Ciencias Exactas, Universidad Andr´es Bello, Av. Fernandez Concha 700, Las Condes, Santiago, Chile. 3Instituto Milenio De Astrof´ısica, Santiago, Chile. 4Vatican Observatory, V00120 Vatican City State, Italy. Accepted 2018 November 21. Received 2018 November 20; in original form 2018 August 29 ABSTRACT A search for extrasolar planetary transits using the extended Kepler mission (K2) campaigns 9 and 11 revealed five new candidates towards the Galactic bulge. The stars EPIC 224439122, 224560837, 227560005, 230778501 and 231635524 are found to have low amplitude transits consistent with extrasolar planets, with periods P = 35.1695, 3.6390, 12.4224, 17.9856, and 5.8824 days, respectively. The K2 data and existing optical photometry are combined with the multi-band near-IR photometry of the VVV survey and 2MASS in order to measure accurate physical parameters for the host stars. We then measure the radii of the new planet candidates from the K2 transit light curves and also estimate their masses using mass-radius relations, con- cluding that two of these candidates could be low mass planets, and three could be giant gaseous planets. Key words: Kepler -- Exoplanets -- Galactic Bulge 1 INTRODUCTION The Kepler mission (Borucki et al. 2010, 2011) was a clear success and a revolution for extrasolar planet studies. The main mission lasted four years and the data collected is still producing extrasolar planets, that are now counted by the thousands. The extended Kepler mission called K2 consisted of several campaigns, with multiple fields observed along the ecliptic plane since 2014. Among a variety of studies, K2 has discovered more than a hundred transiting extrasolar planets up to now (Montet et al. 2015; Schlieder et al. 2016; Van Eylen et al. 2016; Johnson et al. 2016; Adams et al. 2016; Sinukoff et al. 2016; Barros et al. 2016; Pope et al. 2016; Dressing et al. 2017b,a; Petigura et al. 2018; Wittenmyer et al. 2018; Mayo et al. 2018; Yu et al. 2018; Crossfield et al. 2018; Livingston et al. 2018). We are interested here in the data from campaigns 9 and 11 (hereafter K2C9 and K2C11), that observed the Milky Way bulge. This is a crowded and reddened region of our Galaxy, but of great interest, because it overlaps with our ongoing VVV survey, that has been mapping the whole bulge in the near-IR since 2010 (Minniti et al. 2010; Saito et al. 2012). The challenge is the large Kepler 4 arcsec pixel scale, as discussed extensively elsewhere by Henderson et al. (2016) and Zhu et al. (2017). We use our higher resolution VVV images (with 0.3"/pixel scale) in order to weed out bad candidates (usually blended objects). In particular, Hender- (cid:63) E-mail: [email protected] © 2018 The Authors son et al. (2016) describe in detail the goals, difficulties, and procedures of the K2C9. We started a project to detect and study new transit- ing exoplanets in the Galactic Bulge, using K2 mission and VVV survey data. We report here the discovery and charac- terization of five exoplanetary candidates in the bulge. This paper is organized as follows, in Section 2 we present the K2 photometry for the campaign 9 and 11 and the VVV survey. Section 3 discusses our search. In Section 4 and 5 we give the physical parameters for the sample stars and planets, respectively. The discoveries are discussed in turn in Section 6. Finally, Section 7 outlines our main conclusions. 2 DATA 2.1 K2 Photometry For our study, we used the K2 database, which pro- vides high-precision photometry on the 1 and 30 minute timescales. The Kepler magnitude (Kp) refers to an AB mag- nitude, ranging from 425 to 900 nm. The Kepler photometer consists in multi CCD modules, and each module covers 5 square degrees on the sky. The observations of K2 consist of a series of observation field "Campaigns" distributed in the plane of the ecliptic. The campaign 9 consisting of 19 CCD modules (Fig- ure 1, left) covered part of the Galactic Bulge and was ded- icated to a microlensing study. In order to increase the data storage, the campaign 9 was split into two parts (campaign 2 C.C. Cort´es et al. Figure 1. The figure shows the two K2 campaign fields: campaign 9 (left) and campaign 11 (right), in which we perform the search for extrasolar planets. The red stars represent the position of the exoplanet candidates that we found in our study with their respective numbers (see table 3). The green area in campaign 9 represents the microlensing super apertures. The overlapped coloured region corresponds to the region mapped by the VVV survey. This plot is taken from the Kepler & K2 website. 9a and 9b), with a three-day gap from May 19 to May 22, 2016. The campaign 9a is centred at RA=270.3544823 de- grees, DEC=-21.7798098 degrees and was observed between April 22 and May 19, 2016. The campaign 9b that was ob- served between May 22 and July 1, 2016, and is centred at RA=270.3543824 degrees, DEC=-21.7804700 degrees. For our study, we considered only targets from the microlensing super apertures (green area in the Figure 1, left ), in which 3.3 million pixels were dedicated on five CCD channels. The campaign 11 consists of 18 CCD modules, due to the loss of CCD module 4. This campaign is centred at RA=260.3880064 degrees, DEC=-23.9759578 degrees (Fig- ure 1, right) and covers part of the Galactic Bulge. This campaign was split into two parts with a three-day gap from October 18 to October 21, 2016: the campaign 11a that was observed between September and October 2016 during 23 days and the campaign 11b was observed between October and December 2016 during 48 days. 2.2 VVV survey The VISTA Variables in the V´ıa L´actea (VVV) survey (Min- niti et al. 2010; Saito et al. 2012) that covers a bulge area of 300 square degrees between −10◦ < l < +10◦ and −10◦ < b < +5◦ divided in 196 tiles. This survey provides near-infrared photometry in five broad-band filters: Z (0.87 µm), Y (1.02 µm), J (1.25 µm), H (µm) and, Ks (2.14 µm). We used the VVV photometric catalogue that was obtained from the Cambridge Astronom- ical Survey Unit (CASU) 1 in different tiles in the galactic bulge. The K2-VVV areas of overlap are shown in Figure 1. 3 SEARCH FOR EXOPLANETARY TRANSITS For the exoplanet candidates search, we used 875 light curves from the campaign 9 and 13.607 light curves from the cam- paign 11, that were extracted by Vanderburg & Johnson (2014). Both campaigns were split into two parts, therefore we normalized the flux of the light curve for part a and b (see section 2.1) using a cubic spline function. We choose the order of the polynomials according to the light curve. After the fitting, we removed upwards outliers, which are caused by cosmic rays or asteroids and also, we removed the down- ward outliers making sure that the transit was not removed. After flattening the light curves, we calculated a Box Least Squares (BLS)2 periodogram Kov´acs et al. (2002), to detect a periodic signal. We used the definition of Vanderburg et al. (2016) to perform the period search ranging from 2.4 hours to half the length of the campaign and the spacing between periods expressed as: ∆P = P D N × Ttot where ∆P is the spacing between periods, P is the period tested, D is the transit duration at that period, N is an oversampling factor and Ttot is the total duration of the campaign. After this process, we cleaned our catalogue by apply- ing some restrictions. From the analysis described by Van- derburg et al. (2016), we considered targets that in the BLS periodogram have at least one peak with S/N > 9. Also, we eliminate objects whose duration is greater than 20% of the detected period, and we considered only detections that have two or more transit events. Even when an object passes these tests, there is the possibility that it is a false positive. For this reason, we per- formed a visual inspection to discard obvious false positives 1 http://casu.ast.cam.ac.uk/vistasp/ 2 https://github.com/dfm/python-bls MNRAS 000, 1 -- 7 (2018) such as spurious detections, eclipsing binaries and any other astrophysical false positives. Additionally, we take advantage of the near-infrared data from the VVV survey that overlapped with the K2 data (see Figure 1), to discard false positives, specially blended objects, because with this photometry we can constrain the contaminant stars with a different colour than the target (Fressin et al. 2012). Our final catalogue contains five planet candidates (see table 3). The planetary parameters estimation is explained in section 5. 4 STELLAR PROPERTIES The stellar parameters of the host stars of our exoplanet candidates are summarized in Table 1. Based on previous studies, the host star 224439122 was catalogued as a variable Weak T tau (Prisinzano et al. 2012) with a period 5.8775 days (Henderson & Stassun 2012). The stellar parameters for our exoplanet candidates were estimated through Gaia DR2 (Andrae et al. 2018), whose information was extracted from Gaia Archive3 (Gaia Collaboration et al. 2016, 2018). The radius of the host star 231635524 cannot be derived from data in Gaia DR2, because stars with fractional parallax uncertainties greater than 20 percent are not reliably inverted to yield distances. This particularly affects distant and/or faint stars (Andrae et al. 2018). For the host star 231635524, the fractional par- allax uncertainty is 450 percent (Bailer-Jones et al. 2018 infer a distance in excess of 10 kpc with large errors). It is most likely this host star is at least a giant, since this star is too bright to be a dwarf, given the likely very long distance. We classify our host stars by calculating the reduced proper motion, after properly correcting for extinction. For the passbands in the VVV survey, we measured the extinc- tion using the reddening maps of Gonzalez et al. (2012) by the tool BEAM (Bulge Extinction And Metallicity) calcula- tor4 using the Cardelli et al. (1989) extinction law. In the case of filters in the 2MASS catalogue, we used Schlegel et al. (1998) maps through the tool Galactic DUST Reddening & Extinction5. The photometric parameters are summarized in Table 2. With the proper motion of Gaia DR2 catalogue, we cal- culated the reduced proper motion of our host stars, through the equation defined as: HJ = J + 5 log µ (1) where J is the J-band magnitude and µ is the total proper motion. With the criteria defined by Rojas-Ayala et al. (2014): > H∗ J J = 68.5(J − Ks) − 50.7 H dwar f we classify the host star like dwarf or giant, where H∗ dwarf/giant discriminator. (2) J is the Exoplanet Candidates in the Galactic Bulge 3 5 PLANETARY PARAMETERS We have found five planet candidates, with the period and depth calculated from the output of the BLS algorithm. Assuming that the orbit is circular, we modeled the tran- sit time, the period, the planetary to stellar radius ratio (Rp/R(cid:63)), the semi-major axis normalized to the stellar ra- dius (a/R(cid:63)) and the inclination, through a transit model using the BAsic Transit Model cAlculatioN (BATMAN) 6 Python package (Kreidberg 2015). For the development of our model, we used the quadratic limb darkening law (Kopal 1950) with the coefficient estimated by Kreidberg (2015). In order to take into account the smearing effect of the 30 min cadence of the K2 data (Kipping 2010), we used the supersampling provided by BATMAN, that consists in calculating the average value of the light curve from the evenly spaced samples during an exposure. After that, we measured the transit parameters and their uncertainties of this model using emcee7 Python pack- age (Foreman-Mackey et al. 2013), which is an implemen- tation of the affine-invariant ensemble sampler for Markov Chain Monte Carlo (MCMC) (Goodman & Weare 2010). We implement the same uncertainties to each flux, because the flux error was not calculated in the K2 data reduction process. We estimated the mass of the planet candidates through Forecaster8 Python package developed by Chen & Kipping (2017), which uses a probabilistic mass-radius relationship. With this code it is possible to predict the mass of the can- didates based on a given radius measured previously. The estimation of the planetary parameters is summa- rized in Table 3. The exoplanet candidates with the fitting model are shown in Figure 2 to Figure 6. 6 DISCUSSION In this work, we present the discovery of five exoplanet can- didates, which were detected with the transit method using K2 photometry. One of the parameters that we can obtain with this method is the planetary radius. To determine this parameter, we need the stellar radius, which was calculated photometrically (see Section 4). To definitely classify these candidates is it is necessary to estimate their masses, which are not possible to measure using this technique. Clearly, spectroscopic observations are needed for these candidates. Therefore, we predict the mass through the code Forecaster (see Section 5), which uses a mass-radius relationship. Even though these parameters are not highly accurate, they rep- resent a good initial estimate to perform our analysis and give a preliminary idea about the nature of our candidates. We compared our candidates with the mass-density relationship proposed by Hatzes & Rauer (2015). The relationship is based on the inflections in the mass- density diagram, and shows three regions. The regions are low mass planets, giant gaseous planets and stellar objects. Then, we proceed to analyze each of our candidates: 3 https://gea.esac.esa.int/archive/ 4 http://mill.astro.puc.cl/BEAM/calculator.php 5 https://irsa.ipac.caltech.edu/applications/DUST/ 6 http://astro.uchicago.edu/ kreidberg/batman/tutorial.html 7 http://dfm.io/emcee/current/ 8 https://github.com/chenjj2/forecaster MNRAS 000, 1 -- 7 (2018) 4 C.C. Cort´es et al. EPIC 224439122b is a candidate exoplanet orbiting the host star located in NGC 6530 (open cluster), clas- sified as a variable Weak T Tau star with spectral type M0-M1 V (Prisinzano et al. 2012) with a period 5.8775 days (Henderson & Stassun 2012). This extrasolar planet candidate has two transits and has a period 35.1695 days, indicating that it could be a warm Jupiter (the orbital period between 10 and 100 days). Also, this is the largest candidate in our sample with R=48.1R⊕ and an estimated mass of M=438.0MJ , implying that it could be a stellar object (M > 60MJ , Hatzes & Rauer 2015). As the largest planets known have radii of ∼ 20R⊕, this radius seems to be too large for an exoplanet. Although this candidate has a depth of ∼ 5% and a large mass, we consider that this object passed the test described in Section 3 to discard false positive. Also, as we mentioned above, we do not measure the mass and the stellar parameters. Therefore these are not entirely reliable. EPIC 224560837b is a candidate with nineteen transit events with a period of 3.6390 days, which is within the definition of hot Jupiters (the orbital period between 1 and 10 days). This candidate has a radius of 30.6R⊕ and an estimated mass of 260.8MJ indicating that it could be a stellar object (M > 60MJ , Hatzes & Rauer 2015). Despite this candidate is the second largest in our sample and according to the mass classification this could be a stellar object, we take into account that this target passed the test mentioned in Section 3 and the estimation of the stellar parameters and the planetary mass were not measured. For this reason, we do not discard the possibility that this object could be an extrasolar planet. EPIC 227560005b has a period of 12.4224 days. This candidate has four transits and is our smallest exoplanet candidate with a R=2.0R⊕ and an estimated mass of 0.02MJ indicating that it could be a low mass planet (M < 0.3MJ , Hatzes & Rauer 2015). The same definition would apply for EPIC 230778501b that has three transits with a period of 17.9856 days. This candidate is the second smallest in our sample with R=2.2R⊕ and an estimated mass of 0.02MJ . EPIC 231635524b is a candidate that has eleven transits with a period of 5.8824 days. According to the period, this could be a hot Jupiter. As we explained in Section 4, the radius of the host star 231635524 is not available in Gaia DR2, and we infer that this star could be a giant. Therefore if 231635524 could be a giant and the planetary to stellar radius ratio is 0.132 (see Table 3) probably our candidate could be a giant gaseous planet. 7 CONCLUSIONS We reported the discovery of five exoplanet candidates de- tected in the Galactic Bulge with K2 data with orbital pe- riods between 3.6390 and 35.1695 days and planetary radii in the range of 2.0 to 48.1 R⊕. These planet candidates or- bit host stars with a range of magnitudes and temperatures (12.039 < Kp < 16.072, and 4184 K < Te f f < 4647 K). Additionally, two of our candidates were classified as stellar objects (224439122 and 224560837) and two as low mass planets (227560005 and 230778501) according to Hatzes & Rauer (2015), but we considered that 224439122 and 224560837 passed the test to discard false positive men- tioned in Section 3, therefore there are the possibility that these targets could be planets. Due to the radius for the host star of the candidate 231635524 is not available in Gaia DR2, we can not estimate the planetary radius, and consequently, we can not predict the mass of the planet. Therefore, as we explained in Section 6, we infer that the candidate 231635524 could be a giant gaseous planet. In addition, we emphasize that the stellar parameters were determined using photometry, and that the derived masses in particular are very uncertain. Therefore, we would like to encourage follow-up spectroscopic observations in or- der to confirm our exoplanet candidates and to refine their physical parameters. ACKNOWLEDGEMENTS We would like to thank the anonymous referee for careful review of this manuscript and for giving such valuable comments. C.C.C. is supported by CONICYT (Chile) throught Programa Nacional de Becas de Doc- torado 2014 (CONICYT-PCHA/Doctorado Nacional/2014- 21141084). S.V. and C.C.C. gratefully acknowledge the sup- port provided by Fondecyt reg.n. 1170518. D.M. is sup- ported by FONDECYT Regular grant No. 1170121, by the BASAL Center for Astrophysics and Associated Technolo- gies (CATA) through grant PFB-06, and the Ministry for the Economy, Development and Tourism, Programa Iniciativa Cientifica Milenio grant IC120009, awarded to the Millen- nium Institute of Astrophysics (MAS). This paper includes data collected by the K2 mission. Funding for the K2 mission is provided by the NASA Science Mission directorate. We gratefully acknowledge use of data from the ESO Public Sur- vey programme ID 179.B-2002 taken with the VISTA tele- scope, and data products from the Cambridge Astronomical Survey Unit. This publication makes use of data products from the Two Micron All Sky Survey, which is a joint project of the University of Massachusetts and the Infrared Pro- cessing and Analysis Center/California Institute of Technol- ogy, funded by the National Aeronautics and Space Admin- istration and the National Science Foundation. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Analysis Con- sortium (DPAC, https://www.cosmos.esa.int/web/gaia/ dpac/consortium). Funding for the DPAC has been pro- vided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. This re- search has made use of the VizieR catalogue access tool, CDS, Strasbourg, France. The original description of the VizieR service was published in A&AS 143, 23. REFERENCES Adams E. R., Jackson B., Endl M., 2016, AJ, 152, 47 Andrae R., et al., 2018, A&A, 616, A8 Bailer-Jones C. A. L., Rybizki J., Fouesneau M., Mantelet G., Andrae R., 2018, AJ, 156, 58 Barros S. C. C., Demangeon O., Deleuil M., 2016, A&A, 594, A100 Borucki W. J., et al., 2010, Science, 327, 977 MNRAS 000, 1 -- 7 (2018) Table 1. Stellar parameters Exoplanet Candidates in the Galactic Bulge 5 ID EPIC ID RA DEC (J2000, h:m:s) (J2000, d:m:s) Te f f (K) R(cid:63) (R(cid:12)) RPM HJ classification Campaign 9 1 224439122 18:04:44.11 -24:14:39.06 4348+106−190 (1) 1.96+0.18 −0.09 (1) -2.60(1) Dwarf Campaign 11 2 3 4 5 224560837 17:35:13.92 -24:01:42.35 227560005 17:31:02.07 -17:50:35.39 230778501 17:09:39.53 -19:13:01.41 231635524 17:07:33.40 -29:16:21.27 4191+217−199 4184+124−179 4647+141−134 4322+277−75 (1) 2.75+0.28 −0.26 (1) 0.51+0.05 −0.03 (1) 0.73+0.04 −0.04 (1) - (1) (1) (1) -1.09(1) 4.20(1) 3.24(1) 1.80(1) Dwarf Dwarf Dwarf Giant Note. RPM HJ is the reduced proper motion of the star. RPM HJ was calculated using the equation (1), with the proper motion from Gaia DR2 catalogue. In column 10, we classify the host star as giant and dwarf according to the criteria defined by Rojas-Ayala et al. (2014) equation (2). References. (1) Gaia Collaboration et al. (2016, 2018). Table 2. Photometric parameters ID EPIC ID Kp U B V R I z Y J H Ks (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) (mag) Campaign 9 1 224439122 16.072 17.445(1) 17.445(1) 16.145(1) - 14.560(1) 14.132(2) 13.786(2) 10.395 (2) 10.878(2) 11.257 (2) Campaign 11 2 3 4 5 224560837 227560005 230778501 231635524 14.931 12.039 12.388 14.749 - - - - 16.443(3) 15.174(3) 14.723(3) 14.132(4) 12.706(4) 12.088(4) 11.402(4) 13.794(3) 12.685(3) 12.264(3) 12.007(3) 14.940(6) 14.030(6) - - - 13.068(2) 12.828(2) 11.525(2) 11.514(2) 11.481(2) 9.517(5) 9.006(5) 8.883(5) 10.351(5) 9.879(5) 9.760(5) 12.311(5) 11.787(5) 11.709(5) - - - - - - Note. Kp is the Kepler magnitude described in the section 2.1. The filter J, H, and Ks were corrected for extinction, as explained in section 4. References. (1) Sung et al. (2000); (2) VVV survey (Minniti et al. 2017); (3) Zacharias et al. (2012); (4) APASS catalogue (Henden et al. 2016); (5) 2MASS All-Sky Survey (Cutri et al. 2003); (6) Zacharias et al. (2005). Table 3. Planetary parameters ID EPIC ID Period (days) Transit epoch T0 BJD - 2454833 a/R(cid:63) i (deg) δ (%) RP/R(cid:63) RP (R⊕) MP (M⊕) Campaign 9 1 224439122 35.1695 2682.8129 34.09+0.16 −0.07 88.62+0.05 −0.02 5.05 0.225+0.004 −0.009 48.1+5.4−0.5 139220.7+18450.3 −18814.8 Campaign 11 2 3 4 5 224560837 3.6390 227560005 12.4224 230778501 17.9856 231635524 5.8824 2828.2806 2835.5338 2829.4447 2828.8985 4.95+0.19 −0.21 19.09+1.42 −1.23 27.78+1.37 −1.77 10.17+0.50 −0.35 80.48+0.54 −0.62 88.68+0.30 −0.38 88.49+0.95 −0.67 85.71+0.35 −0.26 1.04 0.13 0.07 1.76 0.102+0.001 −0.001 0.036+0.002 −0.003 0.027+0.011 −0.005 0.132+0.001 −0.001 30.6+3.5−3.2 2.0+0.3−0.3 2.2+1.1−0.5 - 82890.3+14961.3 −14115.1 4.9+3.9−2.1 6.1+4.7−3.3 - Note. The Sun and Earth units were obtained from the International Astronomical Union(Prsa et al. 2016). The planetary parameters were estimated in the section 5. T0 is the time of transit, a/R(cid:63) is the semi-major axis normalized to the stellar radius, i is the inclination, δ is the transit depth, RP/R(cid:63) the planetary to stellar radius ratio, RP is the planetary radius, calculated by multiplying the RP/R(cid:63) values with the stellar radius and MP is the planet mass estimated using mass-radius relationship developed by Chen & Kipping (2017) through Forecaster Python package. MNRAS 000, 1 -- 7 (2018) 6 C.C. Cort´es et al. Figure 2. Top: Light curve from the campaign 9 of the exoplanet candidate 224439122 orbiting a variable star. As we mention in the section 2.1 the campaign was split into two parts: the blue points indicate c9a and the gray points indicate c9b. The red lines show two transit times. Middle: Flattened light curve of EPIC ID 224439122 (see section 3). Bottom: Phase folding of the normalized light curve and residuals. The black points mark the K2 data, and the red line marks the best-fitting transit model. This candidate could be a gaseous giant with a period of 35.1695 days. Figure 3. Top: Flattened light curve from the campaign 11 of the exoplanet candidate 224560837. The blue points indicate c11a and the gray points indicate c11b. The red lines show nineteen transit times. Bottom: Phase folding of the normalized light curve and residuals. The black points mark the K2 data, and the red line marks the best-fitting transit model. This is a possible gaseous giant in a 3.6390 days orbit. Figure 4. Light curve from the campaign 11 of the exoplanet candidate 227560005. See Figure 3 caption. This target could be a low mass planet with period P=12.4224 days and four transit events. Borucki W. J., et al., 2011, ApJ, 728, 117 Cardelli J. A., Clayton G. C., Mathis J. S., 1989, ApJ, 345, 245 Chen J., Kipping D., 2017, ApJ, 834, 17 Crossfield I. J. M., et al., 2018, preprint, (arXiv:1806.03127) MNRAS 000, 1 -- 7 (2018) Exoplanet Candidates in the Galactic Bulge 7 Fressin F., et al., 2012, ApJ, 745, 81 Gaia Collaboration et al., 2016, A&A, 595, A1 Gaia Collaboration et al., 2018, A&A, 616, A1 Gonzalez O. A., Rejkuba M., Zoccali M., Valenti E., Minniti D., Schultheis M., Tobar R., Chen B., 2012, A&A, 543, A13 Goodman J., Weare J., 2010, Communications in Applied Math- ematics and Computational Science, Vol.5, No.1, p.65-80, 2010, 5, 65 Hatzes A. P., Rauer H., 2015, ApJ, 810, L25 Henden A. A., Templeton M., Terrell D., Smith T. C., Levine S., Welch D., 2016, VizieR Online Data Catalog, 2336 Henderson C. B., Stassun K. G., 2012, ApJ, 747, 51 Henderson C. B., et al., 2016, PASP, 128, 124401 Johnson M. C., et al., 2016, AJ, 151, 171 Kipping D. M., 2010, MNRAS, 408, 1758 Kopal Z., 1950, Harvard College Observatory Circular, 454, 1 Kov´acs G., Zucker S., Mazeh T., 2002, A&A, 391, 369 Kreidberg L., 2015, PASP, 127, 1161 Livingston J. H., et al., 2018, AJ, 156, 78 Mayo A. W., et al., 2018, AJ, 155, 136 Minniti D., et al., 2010, New Astron., 15, 433 Minniti D., Lucas P., VVV Team 2017, VizieR Online Data Cat- alog, 2348 Montet B. T., et al., 2015, ApJ, 809, 25 Petigura E. A., et al., 2018, AJ, 155, 21 Pope B. J. S., Parviainen H., Aigrain S., 2016, MNRAS, 461, 3399 Prisinzano L., Micela G., Sciortino S., Affer L., Damiani F., 2012, A&A, 546, A9 Prsa A., et al., 2016, AJ, 152, 41 Rojas-Ayala B., Iglesias D., Minniti D., Saito R. K., Surot F., 2014, A&A, 571, A36 Saito R. K., et al., 2012, A&A, 537, A107 Schlegel D. J., Finkbeiner D. P., Davis M., 1998, ApJ, 500, 525 Schlieder J. E., et al., 2016, ApJ, 818, 87 Sinukoff E., et al., 2016, ApJ, 827, 78 Sung H., Chun M.-Y., Bessell M. S., 2000, AJ, 120, 333 Van Eylen V., et al., 2016, ApJ, 820, 56 Vanderburg A., Johnson J. A., 2014, PASP, 126, 948 Vanderburg A., et al., 2016, ApJS, 222, 14 Wittenmyer R. A., et al., 2018, AJ, 155, 84 Yu L., et al., 2018, AJ, 156, 22 Zacharias N., Monet D. G., Levine S. E., Urban S. E., Gaume R., Wycoff G. L., 2005, VizieR Online Data Catalog, 1297 Zacharias N., Finch C. T., Girard T. M., Henden A., Bartlett J. L., Monet D. G., Zacharias M. I., 2012, VizieR Online Data Catalog, 1322 Zhu W., et al., 2017, PASP, 129, 104501 This paper has been typeset from a TEX/LATEX file prepared by the author. Figure 5. Light curve from the campaign 11 of the exoplanet candidate 230778501. See Figure 3 caption. This target could be a low mass planet with a period of 17.9856 days and three transit events. Figure 6. Light curve from the campaign 11 of the exoplanet candidate 231635524. See Figure 3 caption. This target could be a gaseous giant with a period P=5.8824 days and eleven transit events. Cutri R. M., et al., 2003, VizieR Online Data Catalog, 2246 Dressing C. D., et al., 2017a, AJ, 154, 207 Dressing C. D., Newton E. R., Schlieder J. E., Charbonneau D., Knutson H. A., Vanderburg A., Sinukoff E., 2017b, ApJ, 836, 167 Foreman-Mackey D., Hogg D. W., Lang D., Goodman J., 2013, PASP, 125, 306 MNRAS 000, 1 -- 7 (2018)
1605.00616	2	1605	2016-10-06T12:01:16	Water loss from Earth-sized planets in the habitable zones of ultracool dwarfs: Implications for the planets of TRAPPIST-1	[ "astro-ph.EP" ]	Ultracool dwarfs (UCD; $T_{\rm eff}<\sim3000~$K) cool to settle on the main sequence after $\sim$1 Gyr. For brown dwarfs, this cooling never stops. Their habitable zone (HZ) thus sweeps inward at least during the first Gyr of their lives. Assuming they possess water, planets found in the HZ of UCDs have experienced a runaway greenhouse phase too hot for liquid water prior to entering the HZ. It has been proposed that such planets are desiccated by this hot early phase and enter the HZ as dry worlds. Here we model the water loss during this pre-HZ hot phase taking into account recent upper limits on the XUV emission of UCDs and using 1D radiation-hydrodynamic simulations. We address the whole range of UCDs but also focus on the planets recently found around the $0.08~M_\odot$ dwarf TRAPPIST-1. Despite assumptions maximizing the FUV-photolysis of water and the XUV-driven escape of hydrogen, we find that planets can retain significant amounts of water in the HZ of UCDs, with a sweet spot in the $0.04$-$0.06~M_\odot$ range. We also studied the TRAPPIST-1 system using observed constraints on the XUV-flux. We find that TRAPPIST-1b and c may have lost as much as 15 Earth Oceans and planet d -- which might be inside the HZ -- may have lost less than 1 Earth Ocean. Depending on their initial water contents, they could have enough water to remain habitable. TRAPPIST-1 planets are key targets for atmospheric characterization and could provide strong constraints on the water erosion around UCDs.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1–14 (xx) Printed 7 October 2016 (MN LATEX style file v2.2) Water loss from terrestrial planets orbiting ultracool dwarfs: Implications for the planets of TRAPPIST-1 E. Bolmont1(cid:63), F. Selsis2,3, J. E. Owen4†, I. Ribas5, S. N. Raymond2,3, J. Leconte2,3, & M. Gillon6 1 NaXys, Department of Mathematics, University of Namur, 8 Rempart de la Vierge, 5000 Namur, Belgium 2 Univ. Bordeaux, LAB, UMR 5804, F-33615, Pessac, France 3 CNRS, LAB, UMR 5804, F-33615, Pessac, France 4 Institute for Advanced Study, Einstein Drive, Princeton NJ 08540, USA 5 Institut de Ci`encies de l'Espai (CSIC-IEEC), Carrer de Can Magrans s/n, Campus UAB, 08193 Bellaterra, Spain 6 Institut d'Astrophysique et de G´eophysique, Universit´e de Li`ege, All´ee du 6 Aout 19C, 4000 Li`ege, Belgium Accepted xx xx xx. Received xx xx xx; in original form xx xx xx ABSTRACT Ultracool dwarfs (UCD; Teff <∼ 3000 K) cool to settle on the main sequence after ∼1 Gyr. For brown dwarfs, this cooling never stops. Their habitable zone (HZ) thus sweeps inward at least during the first Gyr of their lives. Assuming they possess water, planets found in the HZ of UCDs have experienced a runaway greenhouse phase too hot for liquid water prior to entering the HZ. It has been proposed that such planets are desiccated by this hot early phase and enter the HZ as dry worlds. Here we model the water loss during this pre-HZ hot phase taking into account recent upper limits on the XUV emission of UCDs and using 1D radiation-hydrodynamic simulations. We address the whole range of UCDs but also focus on the planets recently found around the 0.08 M(cid:12) dwarf TRAPPIST-1. Despite assumptions maximizing the FUV-photolysis of water and the XUV-driven es- cape of hydrogen, we find that planets can retain significant amounts of water in the HZ of UCDs, with a sweet spot in the 0.04 – 0.06 M(cid:12) range. We also studied the TRAPPIST-1 system using observed constraints on the XUV-flux. We find that TRAPPIST-1b and c may have lost as much as 15 Earth Oceans and planet d – which might be inside the HZ – may have lost less than 1 Earth Ocean. Depending on their initial water contents, they could have enough water to remain habitable. TRAPPIST-1 planets are key targets for atmospheric characterization and could provide strong constraints on the water erosion around UCDs. Key words: stars: low-mass – (stars:) brown dwarfs – planets and satellites: terrestrial plan- ets – planet star interactions – planets and satellites: atmospheres – planets and satellites: individual: TRAPPIST-1 6 1 0 2 t c O 6 . ] P E h p - o r t s a [ 2 v 6 1 6 0 0 . 5 0 6 1 : v i X r a 1 INTRODUCTION Earth-like planets have been detected in the HZs (defined in Kast- ing, Whitmire, & Reynolds 1993; Selsis et al. 2007a; Kopparapu et al. 2013) of early type M-dwarfs (e.g., Quintana et al. 2014). Here we address the potential habitability of planets orbiting even less massive objects: ultracool dwarfs (UCDs), which encompass brown dwarfs (BDs) and late type M-dwarfs. The first gas giant was detected around a BD by Han et al. (2013). Very recently, Gillon et al. (2016) discovered 3 Earth-sized planets close to an object of mass 0.08 M(cid:12) (just above the theoretical limit of M(cid:63) ∼ 0.075 M(cid:12) between brown dwarfs and M-dwarfs, Chabrier & Baraffe 1997). (cid:63) E-mail: [email protected] † Hubble Fellow c(cid:13) xx RAS The two inner planets in this system have insolations between 4.25 and 2.26 times Earth's. The orbital period of planet d is still unde- termined but between 4.5 and 73 days. The planet receives a stellar flux in the range 0.02-1 times the Earth one, which includes a large fraction of the habitable zone of Trappist-1 (0.023 au – 0.048 au, Kopparapu et al. 2013). The atmospheres of these planets could be probed with facilities such as the HST and JWST, which makes them all the more interesting to study (Barstow & Irwin 2016). BDs (i.e., objects of mass 0.01 (cid:54) M(cid:63)/M(cid:12) (cid:54) 0.07) do not fuse hydrogen in their cores (Spiegel, Burrows, & Milsom 2011). They contract and become fainter in time. Their HZs therefore move in- ward in concert with their decreasing luminosities (Andreeshchev & Scalo 2004; Bolmont, Raymond & Leconte 2011). Nonetheless, for BDs more massive than 0.04 M(cid:12) a planet on a fixed or slowly- evolving orbit can stay in the HZ for Gyr timescales (Bolmont, 2 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon Raymond & Leconte 2011). In this study we consider UCDs of mass up to 0.08 M(cid:12). A UCD of mass 0.08 M(cid:12) is actually a late- type M-dwarf. As for a BD, its luminosity also decreases with time but its mass is high enough so that it starts the fusion of hydrogen in its core. From that moment on, the HZ stops shrinking, allow- ing close-in planets to stay more than 10 Gyr in the HZ. However, any planet that enters the HZ has spent time in a region that is too hot for liquid water. A planet experiencing a runaway greenhouse around a Sun-like star is expected to lose considerable amounts of water (one Earth ocean in less than one Gyr). This is due to H2O photolysis by FUV photons and thermal escape of hydrogen due to XUV heating of the upper atmosphere. This is the current scenario to explain the water depletion in the atmosphere of Venus and its high enrichment in deuterium (Solomon et al. 1991). Could planets in the HZs of UCDs, like Venus, have lost all of their water during this early hot phase? Barnes & Heller (2013) found that planets entering the BD habitable zone are completely dried out by the hot early phase. They concluded that BDs are unlikely candidates for habitable planets. Here we present a study of water loss using more recent estimates for the X-ray luminosity of very low mass stars and con- fronting results obtained with 1D radiation-hydrodynamic mass- loss simulations. We find that – even in a standard case scenario for water retention – a significant fraction of an initial water reser- voir (equivalent to one Earth ocean, MH2O = 1.3 × 1021 kg) could still be present upon reaching the HZ. Once planets enter the HZ, the water can condense onto the surface (as is thought happened on Earth, once the accretion phase was over; e.g. Matsui & Abe 1986; Zahnle, Kasting, & Pollack 1988). Observation of these objects, such as the TRAPPIST-1 planets, would probably lift this uncer- tainty. 2 ORBITAL EVOLUTION OF PLANETS IN THE ULTRACOOL DWARF HABITABLE ZONE The UCD habitable zone is located very close-in, at just a few per- cent of an au (or less; Bolmont, Raymond & Leconte 2011). Tidal evolution is therefore important. Due to UCDs' atmospheres low degree of ionization (Mohanty et al. 2002) and the high densities, magnetic breaking is inefficient and cannot counteract spin-up due to contraction. UCDs' corotation radii – where the mean motion matches the UCD spin rate – move inward. Figure 1 shows the evolution of the HZ boundaries for two UCDs: one of mass 0.01 M(cid:12) and one of mass 0.08 M(cid:12). The time at which a planet reaches the HZ depends on its tidal orbital evolu- tion as well as the cooling rate of the UCD. The tidal evolution of a planet around a UCD is mainly controlled by its initial semi-major axis with respect to the corotation radius (in red and orange full lines in Figure 1). A planet initially interior to the corotation radius migrates inwards due to the tide raised in the UCD and eventu- ally falls on the UCD. A planet initially outside corotation migrates outward (Bolmont, Raymond & Leconte 2011). There does exist a narrow region in which the moving corotation radius catches up to inward-migrating planets and reverses the direction of migration. The planet's probability of survival is smaller the farther inwards it is from the corotation radius, which is illustrated by the colored gradient area in Figure 1. Surviving planets cross the shrinking HZ at different times depending on their orbital distance. Close-in planets tend to stay longer in the HZ because the UCDs' luminosity evolution slows as it cools (Chabrier & Baraffe 1997). Bolmont, Raymond & Leconte Figure 1. Evolution of the HZ limits for a UCD a) of 0.01 M(cid:12) and b) of 0.08 M(cid:12): the inner edge corresponds to a flux of 1.5 that of the Earth (2049 W.m−2) and the outer edge corresponds to a flux of 240 W.m−2 (Bol- mont, Raymond & Leconte 2011). The full lines correspond to the corota- tion radius. The dashed lines correspond to the radius of the dwarf and the thick dashed lines correspond to the Roche limit (assuming an Earth-like planet). The horizontal full black line represents an orbit of 0.013 au. (2011) showed that for UCDs of mass higher than 0.04 M(cid:12), planets can spend up to several Gyr in the HZ. Although their earlier tidal histories vary, planets are on basically fixed orbits as they traverse the UCD habitable zone (Bolmont, Raymond & Leconte 2011). We computed the mass loss taking into account the orbital tidal evolu- tion, but we found that it has no effect on the result1. For the rest of this study we therefore consider the case of planets on fixed orbits. We consider planets from 0.005 au to 0.05 au. Figure 1 shows an or- bit in this range (0.013 au), this orbit allows the planet to pass some time in the HZ of the 0.08 M(cid:12) UCD. For a fixed initial rotation pe- riod of the UCD, planets around low mass UCDs could survive at smaller initial orbital radius, than around high mass UCDs. 3 MODELING WATER LOSS 3.1 Energy-limited escape formalism In order to place the strongest possible constraints, we calculate the mass loss of the atmosphere via energy-limited escape (Wat- son, Donahue, & Walker 1981; Lammer et al. 2003). During a run- away phase, water is able to reach the stratosphere without con- densing. Assuming photolysis is not a limiting process, water is photo-dissociated. In order to escape, water needs to reach the base of the hydrodynamic wind, which is much higher up. We here make 1 In the worst case, i.e. if the UCD is very dissipative, this would mean that we underestimate the mass loss by ∼ 1%. c(cid:13) xx RAS, MNRAS 000, 1–14 1061071081091010Age (yr)10-310-210-110-410-310-210-110-4Semi-major axis (au)Semi-major axis (au)105Roche limitR(cid:803)Corotation radiusHabitable zoneHabitable zonea) M(cid:803)=0.01M⊙b) M(cid:803)=0.08M⊙Water loss from terrestrial planets orbiting ultracool dwarfs 3 the hypothesis that this is the case. In this work, we do not con- sider the diffusion-limited escape, which would be responsible for a lower escape rate than the energy-limited escape rate. The pres- ence of other gases not considered here could bottleneck the diffu- sion of water vapor into the upper atmosphere. The energy-limited escape mechanism requires two types of spectral radiation: FUV (100–200 nm) to photo-dissociate water molecules and XUV (0.1– 100 nm) to heat up the exosphere. This heating causes atmospheric escape when the thermal velocity of atoms exceeds the escape velocity above the exobase. Non-thermal loss induced by stellar winds can in theory contribute significantly to the total atmospheric loss, either by the quiescent wind or by the coronal ejections asso- ciated with flares (Lammer et al. 2007; Khodachenko et al. 2007). For earlier type UCDs, this might be an issue, however we here consider UCDs of spectral type later than M7 type, for which there is no indication of winds that could enhance the atmospheric loss (due to the low degree of ionization, see Mohanty et al. 2002). Energy-limited escape considers that the energy of incident ra- diation with λ < 100 nm is converted into the gravitational energy of the lost atoms. This formalism was used in many studies (e.g., Barnes & Heller 2013; Luger & Barnes 2015; Heller & Barnes 2015) but we use here the prescription of Selsis et al. (2007) linking the XUV flux FXUV(at d = 1 au) to the mass loss rate m: FXUVπR2 p (a/1au)2 = GMp m Rp , (1) where Rp is the planet's radius, Mp its mass and a its semi-major axis. is the fraction of the incoming energy that is transferred into gravitational energy through the mass loss. This efficiency is not to be confused with the heating efficiency, which is the fraction of the incoming energy that is deposited in the form of heat, as only a fraction of the heat drives the hydrodynamic outflow (some being for instance conducted downward). This fraction has been estimated at about 0.1 by Yelle (2004) and more recently at 0.1 or less in the limit of extreme loss by Owen & Wu (2013). The left term of Equation 1, FXUVπR2 p/(a/1au)2, corresponds to the fraction of the XUV flux intercepted by the planet. Thus the mass loss rate is given by: m = FXUVπR3 p GMp(a/1au)2 . (2) Taking into account that the planet might undergo orbital mi- gration or the evolution of the XUV flux the mass lost by the planet at a time t is of: (cid:90) t 0 m = πR3 pGMp FXUV (a/1au)2 dt. (3) One could argue that the XUV cross section radius of the planet is larger than the planet's radius and is probably similar the Hill radius (as in Erkaev et al. 2007, which considered Jupiter-mass planets). This assumption would mean that the water mass loss would be higher than what we propose to calculate here. In a hydro- dynamic outflow the effect of the Roche Lobe is to weaken the ef- fect of gravity compared to the pressure force, its role is felt through the gradient of the potential. For transonic outflow the distance of the sonic point from the surface of the planet depends on the gra- dient of the potential, with shallow gradients (i.e. closer to the Hill radius) pushing the sonic point closer to the planet increases the mass loss. We tested our hypothesis of taking the real radius of the planet with a hydrodynamical code (see Section 3.3 and Owen & c(cid:13) xx RAS, MNRAS 000, 1–14 Alvarez 2016). We find that, for the small rocky planets we con- sider here, this does not underestimate the mass loss as much as the work of Erkaev et al. (2007) would imply (taking the Roche radius), as the effect is of the order of a few percent. Venus probably lost its water reservoir by this mechanism. Radar observations of the Magellan satellite and ground-based ob- servations showed that the last traces of water on Venus date back at least 1 billion years (this corresponds to the age of the surface; Solomon et al. 1991). Venus certainly experienced energy-limited hydrodynamic loss of water during several billion years Chassefi`ere et al. (2012). As UCDs are less bright than the Sun, we expect their XUV flux to be lower than the Sun's. In order to calculate the mass loss, we therefore need the XUV flux of the dwarfs and estimate the efficiency . In the following, we give the observational constraints on the XUV luminosity of UCDs of spectral type later than M7 type. We then use these values to compute the efficiency using 1D radiation-hydrodynamic mass- loss simulations. 3.2 Physical inputs We first assume that protoplanetary disks around UCDs dissipate after 3 Myr (Pascucci et al. 2009; Pecaut & Mamajek 2016) ; this is our "time zero". This assumption is a strong one, it is possible that disks live much longer than that, ∼ 10 Myr according to Pfalzner, Steinhausen, & Menten (2014). We consider that when the planet is embedded in the disk it is protected and does not experience mass loss. Planets are thought to acquire water during and after accre- tion via impacts of volatile-rich objects condensed at larger orbital radii. On the one hand, if planets form in-situ in the HZs of UCDs then it may be difficult to retain water because of the rapid forma- tion timescale and very high impact speeds (Lissauer 2007; Ray- mond, Scalo, & Meadows 2007). On the other hand, because they form fast while embedded in a disk, they can acquire volatiles di- rectly from the disk itself. It has been suggested that accreting H2 this way can lead to the formation of H2O (Ikoma & Genda 2006). In contrast, if these planets or some of their constituent planetary embryos migrated inward from wider orbital distances then they are likely to have significant water contents (Raymond, Barnes, & Mandell 2008; Ogihara & Ida 2009). Although the origin of hot super-Earths is debated (see Raymond, Barnes, & Mandell 2008; Raymond & Cossou 2014), several known close-in planets are con- sistent with having large volatile reservoirs (e.g., GJ 1214 b; Rogers & Seager 2010; Berta et al. 2012). We therefore consider it plausi- ble for Earth-like planets to form in the UCD habitable zone with a water content at least comparable to Earth's surface reservoir and possibly much bigger. For example, Earth contains several addi- tional oceans of water trapped in the mantle (Marty 2012). Water can also be trapped in the mantle of the planet during the forma- tion process and perhaps released in the atmosphere at later times through volcanic activity. A key input into our model is the stellar flux at high ener- gies. There are no observations of the EUV flux of UCDs. We as- sume here that the loss rate of water is not limited by the photo- dissociation of water. This assumes that FUV radiation is suffi- cient to dissociate all water molecules and produce hydrogen at a rate higher than its escape rate into space. X-ray observations with Chandra/ACIS-I2 exist for objects from M6.5 to L5 (e.g., Berger et al. 2010; Williams, Cook, & Berger 2014) for the range 0.1– 10 keV (0.1–12.4 nm). This only represents a small portion of the 4 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon XUV range considered in Equations 2 and 3. For solar-type stars the flux in the whole XUV range is 2 to 5 times higher than in the X-ray range (Ribas et al. 2005). We thus multiply the value corre- sponding to the X-ray range by 5. This constitute an upper limit of what one might expect, indeed for active UCDs the factor can be lower than 2 (e.g., for TRAPPIST-1 this factor is of 1.78, Wheatley et al. 2016). We consider here that it can be used as a proxy for the whole XUV range. On one hand, observations of objects of spectral types M0 to M7 show that the X-ray luminosity scales as 10−3 the bolometric luminosity Lbol (Pizzolato et al. 2003). On the other hand, more re- cent observations from Berger et al. (2010) and Williams, Cook, & Berger (2014) show that the X-ray luminosity of L dwarfs seems to scale as 10−5× Lbol. However, some of these observations are actually non-detections, so the X-ray luminosity of objects like 2M0523-14 (L2–L3), 2M0036+18 (L3–L4) and 2M1507-16 (L5) must be even lower than 10−5× Lbol. Between the populations of dwarfs of spectral type M0 to M7 and those of spectral type L0 to L6, there are dwarfs of spectral type M8 and M9 which do not really follow either trends. TRAPPIST-1 belongs to this transition popula- tion. That is why to study the water loss for the TRAPPIST planets we considered a wide range of XUV-luminosities that are represen- tative of this transition region: from LX/Lbol < 10−5 (as the analog dwarf VB 10 Williams, Cook, & Berger 2014) to LX/Lbol < 10−3.4 (recent observations from Wheatley et al. 2016). There is no indi- cation of whether the X-ray luminosity varies in time. For the UCDs of mass 0.01 < M(cid:63)/M(cid:12) < 0.08, we consider two limiting cases for our mass loss calculation. In the first we adopt a value of 10−5× Lbol. As the bolometric luminosity changes with time, we consider that the X-ray luminosity does as well. In the second case we assume that the X-ray luminosity does not vary with the bolometric luminosity but rather remains constant. We adopt the value of 1025.4 erg.s−1 = 2.5 × 1018 W from Williams, Cook, & Berger (2014). This value corresponds to a X-ray detec- tion (0.1–10 keV) of the object 2MASS13054019-2541059 AB of spectral type L2. To take into account the fact that the dwarfs with the higher masses considered in the range 0.01 < M(cid:63)/M(cid:12) < 0.08 are possibly part of the transition population, we also tested two ad- ditional cases corresponding to a higher XUV emission: LX/Lbol < 10−4.5 and LX = 1026 erg.s−1. For each case we then use Equation 3 to calculate the mass loss of a planet of 0.1 M⊕, 1 M⊕ and 5 M⊕. We compared the XUV flux received by the planets of Figure 1 to the one Earth receives. Before reaching the HZ, they are at least a few times higher than Earth's incoming XUV flux. 3.3 Estimation of with a hydrodynamical code In order to guide our calculations, we perform a set of 1D radiation- hydrodynamic mass-loss simulations based on the calculations of Owen & Alvarez (2016). The simulations are similar in setup to those described by Owen & Alvarez (2016), where we perform 1D spherically symmetric simulations using a modified version of the zeus code (Stone & Norman 1992; Hayes et al. 2006), along a streamline connecting the star and planet. We include tidal gravity, but neglect the effects of the Coriolis force as it is a small while the outflow remains sub-sonic (Murray-Clay, Chiang, & Murray 2009; Owen & Jackson 2012). The radial grid is non-uniform and consists of 192 cells, the flow is evolved for 40 sound crossing times such that a steady state is achieved (which is checked by making sure the pseudo-Bernoulli potential and mass-flux are constant). We ex- plicitly note that these simulations of a pure hydrogen atmosphere do not include line cooling from oxygen that maybe important in Figure 2. Variation of the efficiency with respect to the incoming flux obtained with the model from Owen & Alvarez (2016), we note the stellar mass makes very little difference in the obtained mass-loss rates. The hori- zontal dashed line represents the value of usually used in energy-limited escape calculations. The vertical dotted line represents the XUV flux corre- sponding to LX = 1025.4 erg/s for an orbital distance of 0.01 au. these flows, and as such these calculations should be considered as a maximum rate, as any other elements would lower the hydrogen loss by cooling the flow and colliding with the hydrogen atoms. We can use the simulations to calculate the efficiency () pa- rameter along with the corresponding mass loss. Figure 2 shows the variation of with the incoming XUV flux, we note that the tidal gravity places a negligible role in changing the mass-loss rates with stellar mass. At high fluxes the efficiency drops off due to the increased radiative cooling that can occur as the flows get more vigorous and dense. The drop off at low fluxes is simply caused by the fact that the heating rate is not strong enough to launch a powerful wind. At high fluxes the temperature peaks at a radii ∼ 1.2 − 1.4 R⊕, indicating some of the XUV photons are absorbed far from the planet, increasing its effective absorbing area, but at low fluxes the temperature peaks close to 1 R⊕. Figure 2 shows that for a X-ray luminosity of LX = 1025.4 erg/s (corresponding to a flux in XUV of ∼ 450 erg/s/cm2), assuming an efficiency of 0.1 as is typically the case in energy-limited escape calculations overesti- mate the mass loss by a factor of ∼ 1.17. Using these 1D radiation-hydrodynamic mass-loss simula- tions (see Section 3.3 and Owen & Alvarez 2016), we find that the temperature T of the wind is of the order of 3000 K, which is much lower than what is calculated for hot Jupiters (of the order of 104 K, e.g., Lammer et al. 2003; Erkaev et al. 2007; Murray-Clay, Chiang, & Murray 2009). 3.4 The joint escape of hydrogen and oxygen The computed mass loss m is linked to a mass flux FM given by: FM = m/(4πR2 p). We consider here a mass loss, but do not com- pute the proportion of hydrogen and oxygen atoms lost. Losing just hydrogen atoms and losing hydrogen and oxygen atoms does not have the same consequence for water loss. For example, if only hy- drogen atoms are lost, losing an ocean means that the planet loses the mass of hydrogen contained in one Earth Ocean (9 times lower that the mass of water in one Earth Ocean). This would change the proportion of H/O thus preventing any later recombination of wa- c(cid:13) xx RAS, MNRAS 000, 1–14 10-210-1100101102103104105Flux [erg s-1 cm-2]10-310-210-1100ϵLX ∝ LbolLX=const.Lx = 1025.4 erg/sLx = 1026 erg/sLx (cid:1) LbolWater loss from terrestrial planets orbiting ultracool dwarfs 5 ter (e.g. for Venus, Gillmann, Chassefi`ere, & Lognonn´e 2009). In contrast, if the planet loses hydrogen and oxygen in stoichiometric proportion, losing an ocean means that the planet is losing the mass of water contained in one Earth Ocean. This is the more favorable case for water retention because it requires a higher energy to lose one ocean. In the following, we estimate the proportion of escaping hy- drogen and oxygen atoms. The mass loss flux FM (kg.s−1.m−2) can be expressed in terms of the particle fluxes (atoms.s−1.m−2): FM = mHFH + mOFO. (4) The ratio of the escape fluxes of hydrogen and oxygen in such an hydrodynamic outflow can be calculated following Hunten, Pepin, & Walker (1987): , (5) rF = FO FH = XO XH mc − mO mc − mH where mH the mass of one hydrogen atom, mO the mass of one oxygen atom and mc is called the crossover mass and is defined by: mc = mH + kT FH bgXH , (6) where T is the temperature in the exosphere, g is the gravity and b is a collision parameter between oxygen and hydrogen. In the oxygen and hydrogen mixture, we consider XO = 1/3, XH = 2/3, which corresponds to the proportion of dissociated water. This leads to XO/XH = 1/2. When mc < mO, only hydrogen atoms are escaping and FH = FM/mH. When mc > mO, hydrogen atoms drag along some oxygen atoms and: FM = (mH + mOrF)FH. (7) With Equations 5, 6 and 7, we can compute the mass flux of hydrogen atoms: FH = FM + mOXO (mO − mH) bg kT mH + mO XO XH . (8) In order to calculate the flux of hydrogen atoms, we need an estimation of the XUV luminosity of the star considered, as well as an estimation of the temperature T. We use the estimations of the XUV flux of Section 3.2 and the estimation of and T obtained with the 1D radiation-hydrodynamic mass-loss simulations (Owen & Alvarez 2016). We assume that the XUV luminosity is here L0 = 5× LX = 5× 1025.4 erg/s, that the planet is at a = 0.013 au and following Section 3.3 (Figure 2), we assume here = 0.097. This is our baseline case. Using these values and Equation 2, we calculate the mass loss flux FM = F0 for a 1 M⊕ planet: F0 = m 4πR2 p = FXUVRp 4GMp(a/1au)2 = 1.02 × 10−10 kg.s−1m−2. (9) Using Equation 8 and the estimation of T, we can compute the flux of hydrogen atoms FH as a function of the XUV luminosity LXUV compared to our baseline case luminosity L0. Figure 3 shows the behavior of the mass loss of the atmosphere in units of Earth Ocean equivalent content of hydrogen (EOH) with respect to the ratio LXUV/L0. This behavior is plotted for two cases: a stoichio- metric mixture of hydrogen and oxygen atoms (XO = XH/2 = 1/3) and a mixture slightly depleted in hydrogen (XO = 2XH = 2/3). We can see that the mass loss is a monotonous function of the total XUV luminosity, and for both cases it saturates much be- low our baseline case. For a stoichiometric mixture, at a ratio c(cid:13) xx RAS, MNRAS 000, 1–14 Figure 3. Top panel: Mass of hydrogen lost during the runaway phase as a function of the XUV luminosity for an Earth-like planet at 0.013 au of a 0.08 M(cid:12) dwarf. The mass loss are in unit of Earth Ocean equivalent content of hydrogen (EOH). The blue dashed line is for the limit case where only hydrogen atoms escape, which is valid at low irradiation, when the escape is not strong enough to drag oxygen atoms. The black line accounts for oxygen dragging in the case where XO = XH/2 = 1/3. The gray line is the same for an atmosphere slightly depleted in hydrogen (XO = 2XH = 2/3). Bottom panel: Ratio of the flux of oxygen atoms over the flux of hydrogen atoms. The colors are the same as for the top panel. rF = FO/FH = 0.5 means that there is stoichiometric loss of hydrogen and oxygen. For the baseline case, rF = 0.20. LXUV/L0 ∼ 0.13, the oxygen atoms start to be dragged along there- fore consuming energy. For a mixture depleted in hydrogen, oxygen atoms start to be dragged along for lower incoming XUV luminos- ity LXUV/L0 ∼ 0.075. Here however, for LXUV = L0, we find that one oxygen atom is lost for about 5 hydrogen atoms: rF = FO/FH = 0.20. To con- clude, for a X-ray luminosity of 1025.4 erg/s, there is no stoichio- metric loss of hydrogen and oxygen. However, the situation is not as catastrophic as it would be if only hydrogen escaped. In the fol- lowing, we give the hydrogen lost by the planet. We assume that rF is constant with time for a given value of the XUV flux, which is equivalent to assume an infinite reservoir of water. With a finite water reservoir one has to account for the evolution of the O/H ra- tio. As it makes the result depend on the initial reservoir, we choose to maximize the effective water loss by using the value of rF calcu- lated for a ratio XO = XH/2. The hydrogen mass loss is thus given by: MH = m mH mH + rFmO , (10) where m is obtained with Equation 3 and is given in units of the mass of hydrogen in one Earth Ocean (EOH, which corresponds to ∼ 1.455 × 1020 kg). Hydrogen is the limiting element for the re- 0.010.101101000.11101001000��/��(��)��⊕0.010.101101000.00.51.01.52.0��/��/��⊕6 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon combination of water, so that the remaining content of hydrogen in EOH actually represents the ocean portion available for precipita- tion once in the HZ. According to Equation 6, rF depends on the gravity of the planet considered. In the following, we also calculate the hydrogen loss from planets of 0.1 M⊕ and 5 M⊕. Computing the hydrogen loss for these cases show us that the higher the mass the lower is rF. For LXUV = L0, we find that rF ∼ 0.45 for the 0.1 M⊕ planet (meaning that the loss is quasi-stoichiometric), and rF = 0 for the 5 M⊕ planet (meaning that the only hydrogen atoms escape). 4 WATER LOSS OF PLANETS IN THE UCD HABITABLE ZONE Using the estimation of as a function of XUV flux (Figure 2), the estimations of rF as calculated in the previous section and the equations of Section 3, we can now estimate the loss of hydrogen from planets around UCDs. Figure 4a) shows the evolution of the hydrogen loss from an Earth-mass planet orbiting a 0.04 M(cid:12) BD as a function of both time and orbital radius (assumed to remain constant). We assume that the X-ray luminosity scales as 10−5 Lbol and we calculate and rF accordingly (typically, for a planet at 0.013 au, varies be- tween 0.045 and 0.111 as the XUV flux decreases and rF decreases from 0.43 to 0.10). Let us consider that the planet is located at 0.013 au from a 3 Myr-old BD. Figure 4a) shows that it would lose only 0.48 EOH in the 156 Myr before entering the HZ (ver- tical red dashed line). This planet could therefore be considered a potentially habitable planet, assuming an appropriate atmospheric composition and structure. As the bolometric luminosity decreases with time, the XUV luminosity decreases as well, which has the effect of increasing the efficiency (Figure 2) and decreasing rF. Consequently, with this compensation, the efficiency of the mass loss do not decrease linearly with decreasing XUV-flux. Figure 4b) shows the evolution of the hydrogen loss from an Earth-mass planet orbiting a 0.08 M(cid:12) BD. A planet located at 0.013 au from a 3 Myr- old BD lose ∼ 3.2 EOH in the 1180 Myr before entering the HZ (vertical red dashed line). Figure 5a) shows the hydrogen loss for planets of 0.1 M⊕, 1 M⊕ and 5 M⊕ orbiting UCDs of different masses at 0.013 au. The figure shows how much water was lost by the time each planet reached the HZ. This time is ∼ 8 Myr for a BD of 0.01 M(cid:12) and ∼ 1180 Myr for a dwarf of 0.08 M(cid:12). The calculations were done for both cases: LX/Lbol = 10−5 (red symbols) and LX = 1025.4 erg/s (blue symbols). As before, and rF were calculated consistently for each planet and for the two X-ray luminosity assumptions. We find that planets orbiting more massive UCDs lose more water. This is mainly because of the much longer time spent by those planets interior to the HZ. Planets at 0.013 au around BDs of mass 0.01 M(cid:12) lose less than 0.04 EOH, whatever their mass. Planets at 0.013 au around BDs of mass 0.05 M(cid:12) lose less than 2 EOH, whatever their mass. Planets at 0.013 au around UCDs of mass 0.08 M(cid:12) lose more than 2 EOH, whatever their mass. We find that higher-mass planets lose their water at a higher rate than lower-mass ones. This is due to the effect of gravity on the cross-over mass that makes rF smaller for a higher gravity. In other words too high a gravity prevents the loss of oxygen and thus enhances the loss of hydrogen (which is not directly affected by gravity, e.g, Luger & Barnes 2015). Consequently, throughout the evolution, we expect more massive planets to lose more than low- mass planets. For example, assuming LX/Lbol = 10−5, we find that Figure 4. Map of hydrogen loss (in EOH) as a function of the age of the BD of a) 0.04 M(cid:12) and b) 0.08 M(cid:12) and the orbital distance of the planet. White lines correspond, from left to right, to a loss of 10 EOH, 5 EOH, 1 EOH, 0.5 EOH and 0.1 EOH. The grey line corresponds the inner edge of the HZ: the HZ lies above this line. Here the X-ray luminosity evolves as 10−5 × Lbol. the 5 M⊕ planet orbiting a UCD of 0.08 M(cid:12) loses 6.7 EOH before reaching the HZ, while the 1 M⊕ loses only 3.2 EOH and the 0.1 M⊕ only 2.0 EOH. Assuming LX/Lbol = 10−5, we find that 1 M⊕ planets orbiting at 0.013 au around BDs with masses smaller than 0.06 M(cid:12) lose less than 1 EOH before reaching the HZ and 1 M⊕ planets orbiting BDs of mass (cid:46) 0.07 M(cid:12) lose less than 2 EOH before reaching the HZ. Whatever the mass of the UCD, low mass planets (Mp (cid:54) 1 M⊕) lose less than 3.2 EOH before reaching the HZ. Whatever the mass of the UCD, whatever the mass of the planet and the XUV-luminosity assumption, all planets lose less than 9 EOH before reaching the HZ. Figure 6 shows the hydrogen loss as a function of the planet's orbital distance and mass of host UCD. The black lines represent different hydrogen loss levels (1, 2, 10 EOHH) and the white lines represent levels of time the planet passes in the HZ (as computed as in Figure 1). The closer the planet and the more massive the UCD, the more hydrogen is lost. The closer the planet and the more mas- sive the UCD, the more time the planet spends in the HZ. However for the more massive UCD we consider (as can be seen in Figure 1 for the UCD of 0.08 M(cid:12)), the planets on the closest orbits are always interior to the HZ and thus stay in a runaway phase during all the time of the evolution (10 Gyr). In Figure 6, these planets are separated from the rest by the blue dashed line. They can lose up to c(cid:13) xx RAS, MNRAS 000, 1–14 0.1 EOH0.5 EOH1 EOH5 EOH HZ inner  edge0.1 EOH0.5 EOH1 EOH10 EOH5 EOH0.0050.010.020.030.05Orbital distance (au)BD age (Myr)BD age (Myr)36102040601002005001000200050003610204060100200500100020005000 HZ inner  edgea) M(cid:803)=0.04M⊙b) M(cid:803)=0.08M⊙Water loss from terrestrial planets orbiting ultracool dwarfs 7 [b] Figure 5. Hydrogen loss (in units of Earth's ocean) as a function of the mass of the UCD for a planet of different masses at 0.013 au. Panel a) corresponds to LX = 1025.4 erg/s (in blue) and LX/Lbol = 10−5 (in red). Panel b) corresponds to LX = 1026 erg/s (in blue) and LX/Lbol = 10−4.5 (in red). 160 EOH. There is a compromise to be found between the hydro- gen loss and the time the planet spends in the HZ: planets around low mass BDs lose little hydrogen but they stay a short time in the HZ, planets around the higher mass UCDs considered here spend a longer time in the HZ but they lose more hydrogen prior to enter- ing. The shaded regions in Figure 6 shows an interesting parameter space for each planet and XUV emission hypothesis: the planets in these regions lose less than 1 EOH before entering the HZ and spend in the HZ more than 1 Gyr. Planets with a similar or larger water content as the Earth would thus enter the HZ with enough water to form oceans and as they spend a long time in the HZ, this gives time for life to eventually appear and modify the environment (Bolmont, Raymond & Leconte 2011) For example, for an Earth mass planet and assuming LX/Lbol = 10−5, we find that the plan- ets around UCDs more massive than ∼ 0.035 M(cid:12) and farther away than 0.007 au fulfill these conditions. Of course, when consider- ing higher mass UCDs, the minimum orbital distance increases: a planet around a 0.08 M(cid:12) UCD has to be between farther away than ∼ 0.02 au to fulfill the conditions. If we consider softer constraints, for example cases for which the planets lose less than 2 EOH and spend more than 500 Myr in the HZ, the parameter space gets much bigger: for example, Earth-mass planets as close as 0.005 au around a 0.02 M(cid:12) BD fulfill these conditions. However these very close-in planets could be in danger of falling onto the BD: they could be interior to the corotation radius (see Figure 1 and Bolmont, Ray- mond & Leconte 2011). If we consider 5 M⊕ planets, the param- eter space corresponding to a loss (cid:54) 1 EOH and a time in the HZ (cid:62) 1 Gyr shrinks towards the higher UCD masses and bigger or- bital distances. If we consider 0.1 M⊕ planets, the parameter space corresponding to a loss (cid:54) 1 EOH and a time in the HZ (cid:62) 1 Gyr ex- tends towards the lower UCD masses and smaller orbital distances. We thus can conclude that with favorable atmospheric conditions and reasonable water content (a few oceans), Earth-mass planets between 0.01 and 0.04 au orbiting UCDs of mass 0.04 – 0.08 M(cid:12) would be good targets for the characterization of a potentially hab- itable planet. on the X-ray luminosity of these faint dwarfs: LX/Lbol = 10−5 and LX = 1025.4 erg/s. These values are probably a good approxima- tion of the X-ray flux received by planets around BDs (Berger et al. 2010; Cook, Williams & Berger 2014; Williams, Cook, & Berger 2014). However, one might want to consider, for the higher mass UCDs we consider here, stronger high-energy irradiation levels, such as LX/Lbol = 10−4.5 and LX = 1026 erg/s, which correspond to the quiescent emission of the object 2MASS 14542923+1606039 Bab (see Williams, Cook, & Berger 2014). Figure 5b) shows the hydrogen loss for these two X-ray luminosity cases. The hydro- gen loss is higher than before. Due to the dependance of the effi- ciency and of rF with the XUV luminosity, the mass loss com- puted with LX/Lbol = 10−4.5 is less than ∼ 3.16 higher than the one calculated with LX/Lbol = 10−5. Similarly, the loss computed with LX = 1026 erg/s is less than ∼ 4 times higher than the one computed with LX = 1025.4 erg/s. Using these values, we find that all planets orbiting BDs of mass lower than 0.03 M(cid:12) at 0.013 au lose less than 1 EOH before reaching the HZ. Note that when the XUV flux is high (red symbols), rF for the two lower mass planets tend to 0.5, counterbalancing the effect of the planet's gravity on the crossover mass and leading here to a very similar mass loss. With these higher X-ray luminosity assumptions, we thus can conclude that with favorable atmospheric conditions and reason- able water content (a few oceans), Earth-mass planets between 0.018 and 0.04 au (instead of 0.01 – 0.04 au) orbiting UCDs of mass 0.06 – 0.08 M(cid:12) (instead of 0.04 – 0.08 M(cid:12)) would be good targets for the characterization of a potentially habitable planet. 5 IMPLICATION FOR THE TRAPPIST-1 PLANETS The three planets of the TRAPPIST-1 system (Gillon et al. 2016) are Earth-sized planets, and thus probably rocky (Weiss & Marcy 2014; Rogers 2015). They orbit a M8-type dwarf of 0.080 ± 0.009 M(cid:12)2 TRAPPIST-1b is located at ab = 0.011 au, TRAPPIST- These calculations were made with the following assumptions 2 TRAPPIST-1 is therefore not a BD but a very low mass star. c(cid:13) xx RAS, MNRAS 000, 1–14 t_init = 3 Myreps OwenMp = 5.0 M⊕Mp = 0.1 M⊕Mp = 1.0 M⊕Hydrogen loss (EOH)UCD mass (M⊙)0.000.020.040.060.08061012LX = cstLX/Lbol = cstUCD mass (M⊙)0.000.020.040.060.08a) LX = 1025.4 erg/s and LX/Lbol = 10-58b) LX = 1026 erg/s and LX/Lbol = 10-4.5sma 0.013428 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon Figure 6. Hydrogen loss (black contour lines) as a function of the mass of the UCD and the orbital distance of the planet. Panels a) correspond to LX = 1025.4 erg/s and panels b) correspond to LX/Lbol = 10−5. The blue lines correspond from left to right to a time spent in the HZ of 500 Myr and 1 Gyr. The dashed blue line corresponds to the limit where the planets never reach the HZ because the dwarf initiated the fusions of hydrogen preventing the inner edge of the HZ to sweep in towards very small orbital distances (see Figure 1). The blue shaded areas represent two interesting parameter spaces: the planets in the light blue area lose less than 2 EOH before reaching the HZ and they will spend more than 500 Myr in it, the planets in the dark blue area lose less than 1 EOH before reaching the HZ and they will spend more than 1 Gyr in it. c(cid:13) xx RAS, MNRAS 000, 1–14 a) LX = 1025.4 erg/sb) LX/Lbol = 10-50.1 M⨁1.0 M⨁5.0 M⨁0.1 M⨁1.0 M⨁5.0 M⨁0.040.030.020.010.005distance (au)0.040.030.020.010.005distance (au)0.050.050.040.030.020.010.005distance (au)0.050.020.030.040.050.060.070.080.010.020.030.040.050.060.070.080.01UCD mass (M⊙)UCD mass (M⊙)500 Myr0.1 EOH0.1 EOH10 EOH2 EOH10 EOH0.1 EOH0.5 EOH10 EOH0.1 EOH10 EOH0.1 EOH10 EOH0.1 EOH0.5 EOH10 EOH0.5 EOH1000 Myr1.0 EOH0.5 EOH1000 Myr1.0 EOH500 Myr2 EOH500 Myr1000 Myr2 EOH1.0 EOH500 Myr1000 Myr0.5 EOH1.0 EOH2 EOH500 Myr1000 Myr0.5 EOH1.0 EOH2 EOH500 Myr1000 Myr1.0 EOH2 EOHWater loss from terrestrial planets orbiting ultracool dwarfs 9 1c at ac = 0.015 au. The orbit of TRAPPIST-1d is poorly con- strained, however it is farther away than ad = 0.022 au. The irra- diation of the planets are respectively: 4.25 S ⊕, 2.26 S ⊕ and 0.02– 1 S ⊕, where S ⊕ is the insolation received by the Earth. Therefore TRAPPIST-1d could be in the HZ. The age of the system has been estimated to be more than 500 Myr. The structural evolution grids we use in this article for a dwarf star of 0.08 M(cid:12) (Chabrier & Baraffe 1997) show that the luminosity and radius of TRAPPIST-1 correspond to a body of ∼ 400 Myr, which is lower than the esti- mated age of the system. This is consistent with the fact that evo- lution models seem to under-estimate the luminosity of low-mass objects (Chabrier, Gallardo & Baraffe 2007). However, we explored the mass range allowed by the observations and found that a dwarf star of 0.089 M(cid:12) can reproduce the characteristics of TRAPPIST- 1 at an age of ∼ 850 Myr. Figure 7 shows the evolutionary tracks we used to simulate the luminosity evolution of TRAPPIST-1. We interpolated the values of radius, luminosity and effective temperature between the evolu- tionary tracks of a 0.08 M(cid:12) dwarf and a 0.1 M(cid:12) dwarf (Chabrier & Baraffe 1997). Figure 7 shows that the characteristics of the star – radius, luminosity and effective temperature – can be repro- duced with our interpolated tracks for ages between 800 Myr and 900 Myr, which is compatible with the estimation of the age of the star made by Gillon et al. (2016). We use here two different assumptions to calculate the HZ in- ner edge: S p = 0.9 S ⊕, which corresponds to the inner edge for a non-synchronized planet (Kopparapu et al. 2013) and S p = 1.5 S ⊕ which corresponds to the inner edge for a synchronized planet (Yang, Cowan, & Abbot 2013). Following this model, the two in- ner planets of TRAPPIST-1 always stay interior to the HZ. As the orbit of planet d is poorly constrained, we considered three differ- ent orbits: 0.022 au (the closest one), 0.058 au (the most probable one) and 0.146 au (the farthest one). A planet at 0.022 au enters the HZ corresponding to S p = 1.5 S ⊕ at an age of 393 Myr (later called THZ(1.5 S ⊕)). However, it never enters the HZ corresponding to S p = 0.9 S ⊕. For a planet at 0.058 au, THZ(1.5 S ⊕) = 29 Myr and THZ(0.9 S ⊕) = 58 Myr. For a planet at 0.146 au, THZ(1.5 S ⊕) = 3.2 Myr and THZ(0.9 S ⊕) = 5.4 Myr. If we had considered more massive stars than what is allowed by Gillon et al. (2016), the en- try in the HZ would have been postponed by at least a few tens of million years increasing the period of time the planet spends in the runaway phase. To calculate the mass loss from the planets of the TRAPPIST- 1 system, we assumed different XUV emissions. As explained in Section 3.2, TRAPPIST-1 is part of the transition population be- tween early M type and late M, early L. In order to treat the whole XUV range possible we assumed the two different XUV luminosity measured by Wheatley et al. (2016): - LX/Lbol = 10−3.7 and LX/Lbol = 10−3.4 Observational studies (e.g., Williams, Cook & Berger 2014; Cook, Williams & Berger 2014) indicate a significantly large scatter at spectral type M8, with values ranging between LX/Lbol = 10−5 and 10−3 in quiescence. Cook, Williams & Berger (2014) mention an- other analog of TRAPPIST-1 , LP 412-313, which has a quiescent emission of 1027.2 erg.s−1 or LX/Lbol = 10−3.1. Furthermore, us- ing XMM-Newton observations, Wheatley et al. (2016) measured 3 Its bolometric luminosity Lbol/L(cid:12) ∼ 10−3.29 is close to the luminosity of TRAPPIST-1 (Lbol/L(cid:12) ∼ 10−3.28), its V sin i of 8 km.s−1(Reid et al. 2002; Newton et al. 2016) is also close to TRAPPIST-1's ∼ 6 km.s−1. c(cid:13) xx RAS, MNRAS 000, 1–14 recently for TRAPPIST-1 LX/Lbol = 10−3.7 to 10−3.4, which is sig- nificantly higher than the value we adopted for the UCDs of the previous Sections. - LX/Lbol = 10−5 This is what we used for UCDs in the previous Sections. As the measurements of Wheatley et al. (2016) could be due to a flare, we consider this much lower flux. This also corresponds to an analog of TRAPPIST-1: the M8 dwarf VB 104. Observations of VB 10 by Fleming, Giampapa & Schmitt (2000) showed that the quiescent emission was LX = 10−5.0 Lbol and the flaring emission was LX = 10−2.8 Lbol. Later, Berger et al. (2008) measured LX = 10−5.0 Lbol for the quiescent emission and LX = 10−4.1 Lbol during flaring events. Finally, Williams, Cook & Berger (2014) and Cook, Williams & Berger (2014) found that the quiescent emission of VB 10 was LX = 10−5.1 Lbol and that the flaring emission was LX = 10−4.6 Lbol. We used the method described in Section 3.1, using an effi- ciency based on the hydrodynamical simulations of Section 3.3. We assumed an Earth-like composition to compute the masses of the planets (Fortney, Marley, & Barnes 2007) and we calculated rF following the method given in Section 3.4 for the three differ- ent XUV luminosity assumptions and for the different planets of the system. We assumed that the semi-major axes of the planets remain constant throughout the evolution. Figure 8 show the hydrogen loss for the planets of the system for the three different XUV-luminosity trends as a function of time. Table 1 summarizes the results. For LX/Lbol = 10−5, we considered that LXUV = 5 LX, as in Section 4. However, for LX/Lbol = 10−3.4 and LX/Lbol = 10−3.7, we used the value of Wheatley et al. (2016): LXUV = 1.78 LX. For LX/Lbol = 10−5, we find that planet b lose less than 4 EOH and planet c lose less than 3 EOH at the age of the system. However, considering a higher XUV flux, this limit goes up to 13.5 EOH for planet b and 9.5 EOH for planet c. Unless those planets have a big water content, they are therefore likely to be desiccated. For planet d, due to the high uncertainty on its orbit, we find that at worst it could lose almost 7 EOH for an orbital separation of 0.022 au (assuming LX/Lbol = 10−3.4 and that the planet never reaches the HZ). For LX/Lbol = 10−5.0, a planet d at 0.022 au loses more hydrogen than planet c at late ages. This is due to a combina- tion of the effect of gravity on the cross-over mass (planet d being bigger than planet c, its rF is smaller) and XUV-flux which is lower for planet d which means that the efficiency is bigger. However, considering TRAPPIST-1d is on the more probable orbit, at 0.058 au, we find that it loses between 0.06 and 0.41 EOH. It loses much less than if it was at 0.022 au because it is much farther away and enters the HZ much earlier. If TRAPPIST-1d is at 0.146, it loses less than 0.01 EOH. As Hydrogen escapes faster than Oxygen, mass loss results in an Oxygen build up in the atmosphere (e.g., Luger & Barnes 2015). We estimate the O2 pressure in the atmosphere of the different plan- ets (see Table 1). The pressure of O2 can be as high as 500 bar for a planet d at 0.022 au. It is even higher than the O2 pressure for the two inner planets, because as the XUV flux planet d receives is lower than for planets b and c, its rF is smaller and it therefore loses much less oxygen than the two inner planets. For a planet d at 4 Its bolometric luminosity Lbol/L(cid:12) ∼ 10−3.3 is close to the luminosity of TRAPPIST-1 (Lbol/L(cid:12) ∼ 10−3.28), its V sin i of 6.5 km.s−1 is also close to TRAPPIST-1's ∼ 6 km.s−1. Their radii are also similar within a few percent. 10 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon Figure 7. Characteristics of the star's model we used for the calculation of hydrogen loss. Top left: Evolution of the radius for the 2 models (in red and orange Chabrier & Baraffe 1997) and our interpolation (in blue). The grey area corresponds to R(cid:63) = 0.117 ± 0.004 R(cid:12) (Gillon et al. 2016). Top right: Evolution of the luminosity, the grey area corresponds to L(cid:63) = 0.000525 ± 0.000036 L(cid:12). Bottom left: Evolution of the effective temperature, the grey area corresponds to Teff = 2550 ± 55 K. Bottom right: Evolution of the inner edge of the HZ for two different assumptions: S p = 1.5 S ⊕ and S p = 0.9 S ⊕. The orbital distances of the three planets of TRAPPIST-1 are also represented. Table 1. Lost hydrogen in EOH and corresponding O2 pressure for the TRAPPIST-1 planets at the time they enter the HZ (with the two assumptions about the inner edge). The numbers in bold correspond to cases for which the planet never reaches the HZ, the hydrogen loss is then given for the age of the star (∼ 850 Myr, according to our model). The two values indicated correspond to t0 = 10 Myr and t0 = 3 Myr. SMA (au) 0.01111 0.01522 0.022 0.058 0.146 T1-b T1-c T1-d T1-d T1-d LX/Lbol = 10−3.4 Hydrogen loss (EOH) LX/Lbol = 10−3.7 LX/Lbol = 10−5.0 THZ(0.9 S ⊕) THZ(1.5 S ⊕) THZ(0.9 S ⊕) THZ(1.5 S ⊕) THZ(0.9 S ⊕) THZ(1.5 S ⊕) 12.76–13.18 9.19–9.53 6.56–6.78 0.32–0.41 < 0.01 3.70–3.93 0.15–0.24 < 0.002 8.96–9.28 6.39–6.63 3.61–3.73 2.60–2.69 4.85–5.01 0.24–0.30 < 0.01 2.69–2.85 0.11–0.17 < 0.001 2.67–2.73 0.14–0.17 < 0.0007 1.34–1.40 0.06–0.09 < 0.0007 PO2 (bar) THZ(0.9 S ⊕) THZ(1.5 S ⊕) 418–422 345–348 489–493 28–32 < 1.4 222–227 11–15 < 0.14 0.058 au up to 30 bar of O2 can build up in the atmosphere by the time it reaches the HZ. The orbit of TRAPPIST-1d is not well constrained, but there is a high probability that it is in the HZ. This calculation shows that there is a non-negligible probability that this planet was able to retain a high fraction of an eventual water reservoir of one Earth Ocean, which makes it a very interesting astrobiology target. Additional measurements of TRAPPIST-1's X-ray luminos- ity are needed in order to establish whether the values of Wheat- ley et al. (2016) correspond to a flare or not (a discussion about the quantitative effect of flares can be found in Section 6.3). Be- sides, what might give us a deeper insight of the escape mecha- nisms for TRAPPIST-1's planets will be the characterization of the atmospheres of the three planets. Indeed, Belu et al. (2013) showed that the atmosphere of the planets of TRAPPIST-1 could be charac- terizable with facilities like JWST. The observation of these planets could therefore bring us informations on water delivery during the formation processes and their capacity to retain water. c(cid:13) xx RAS, MNRAS 000, 1–14 T1-bT1-cT1-dAge (yr)Luminosity (L⊙)HZ inner edge (au)0.010.100.00010.0010.010.1107108109106Radius (au)Age (yr)0.0010.012200240026002800300032001071081091010Teff (K)1061010M(cid:803) = 0.100 M⊙M(cid:803) = 0.089 M⊙M(cid:803) = 0.080 M⊙Sp = 1.5 S⨁Sp = 0.9 S⨁Water loss from terrestrial planets orbiting ultracool dwarfs 11 Figure 8. Hydrogen loss as a function of time for the planets of TRAPPIST-1. Panel a) corresponds to LX/Lbol = 10−3.4, panel b) corresponds to LX/Lbol = 10−3.7 and the panel c) to LX/Lbol = 10−5.0. The results were obtained with Equation 3 for the radiuses and semi-major axes of the planets given by Gillon et al. (2016). The dashed vertical lines represents the time the planets reach the HZ for both assumptions. 6 DISCUSSION 6.1 Why this result likely overestimate the loss In this section, we show that the thought process we performed in the previous section, both following the standard way of computing mass loss (as in Barnes & Heller 2013) and using simple radiation- hydrodynamics calculations (as in Owen & Alvarez 2016) may ac- tually be overestimating the mass loss. • The time of the disk dispersal we consider here might be too short for such low mass objects. The evolution of disks around UCDs is not well constrained, however it is reasonable to assume they dissipate between 3 Myr and 10 Myr (disks around low mass stars tend to have longer lifetimes, e.g. Pascucci et al. 2009; Liu et al. 2015; Downes et al. 2015). Disks around UCDs could very well dissipate at an age of 10 Myr. As young UCDs are brighter than old UCDs, a later dissipation of the disk would mean that a planet is exposed less time and to a weaker XUV radiation, meaning that the planet would lose less water than what was calculated in this work. For example, a 1 M⊕ planet orbiting a 0.04 M(cid:12) BD at 0.013 au would only lose 0.41 EOH by the time it reaches the HZ if the disk dissipates at 10 Myr (instead of 0.48 EOH if the disk dissipates at 3 Myr, see Figure 5, for LX/Lbol < 10−5). And a planet orbiting a 0.06 M(cid:12) BD at 0.013 au would lose only 1.06 EOH if the disk dissipates at 10 Myr (instead of 1.15 EOH if the disk dissipates c(cid:13) xx RAS, MNRAS 000, 1–14 at 3 Myr). Under this assumption of a longer lived protoplanetary disk, planets can therefore keep slightly more water. • Water vapor photolysis is required to feed the loss in hydro- gen atoms and is produced by FUV radiation (100–200 nm). UCD are too cool to produce a significant photospheric FUV flux and H2O-photolysing radiation is likely to be restricted to the Lyman- alpha emission. Recent observations of 11 M-dwarfs by France et al. (2016) showed that the estimated energy flux in the Lyman-alpha band is equal to the flux in the XUV range. Using this constraint, we can estimate the quantity of hydrogen produced by photodisso- ciation. Figure 9 shows the quantity of hydrogen lost according to our calculations of Section 4 and the quantity of hydrogen avail- able. If all the incoming FUV photons do photolyse H2O molecules with a 100% efficiency (α = 1) and all the resulting hydrogen atoms remain available for the escape process then photolysis does not appear to be limiting the loss process. The production rate of hydrogen atoms exceeds the computed thermal loss assuming an hydrogen and oxygen mixture. In reality, however, only a frac- tion of the incoming FUV actually results in the loss of an hy- drogen atom. Part of the incoming photons are absorbed by other compounds (in particular hydrogen in the Lyman alpha line) or backscattered to space. Then, products of H2O photolysis (mainly OH and H) recombine through various chemical pathways. If the efficiency α is less than about 23% then the loss rate becomes LX/Lbol = 10-3.7LXUV= 1.78 LX28101214640Age (Myr)Hydrogen loss (EOH)328101214101001000LX/Lbol = 10-5.0LXUV= 5.00 LX640old sma for dLX/Lbol = 10-3.4LXUV= 1.78 LXHydrogen loss (EOH)T1- bT1- cT1- d, sma = 0.022 auT1- d, sma = 0.058 auT1- d, sma = 0.146 au28101214640Age (Myr)3101001000a)b)c)12 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon even have a large portion of their bulk made out of water (Ocean planets, hypothesized by Kuchner 2003; L´eger et al. 2004; Ogihara & Ida 2009). For example, Kaltenegger, Sasselov, & Rugheimer (2013) have identified Kepler-62e and 62f to be possible ocean planets. Furthermore, as discussed in Marty (2012), a large part of water can be trapped in the mantle to be released by geological events during the evolution of the planet, allowing a replenishment of the surface water content. • Johnson, Volkov, & Erwin (2013) showed using molecular- kinetic simulations that the mass loss saturates for high incoming energy, which could mean that in our case the mass loss would be smaller than what we calculated. However, their results should be applied to our specific problem in order to be sure. 6.2 Why this result is different from previous ones Unlike Barnes & Heller (2013), we find that planets in the HZs of UCDs should in most cases be able to retain a non negligible por- tion of their initial reservoir of water. The main differences between Barnes & Heller (2013) and our work are the following. First, they used for the XUV radiation an observed upper limit for early-type M-dwarfs (Pizzolato et al. 2003, which was the only available study at the time) for which the XUV luminosity scales as 10−3 times the bolometric luminosity. We used here more recent estimates of X-ray emission of later-type dwarfs, which show that UCDs emit much less X-rays than earlier-type M-dwarfs (Berger et al. 2010; Williams, Cook, & Berger 2014; Osten et al. 2015). Second, in addition to the standard method used in Barnes & Heller (2013) and Luger & Barnes (2015), we also improved the robustness of our results obtained with an improved energy-limited escape for- malism using a better estimate of the fraction of the incoming en- ergy that is transferred into gravitational energy through the mass loss () obtained with 1D radiation-hydrodynamic mass-loss sim- ulations (Owen & Alvarez 2016). Third, Barnes & Heller (2013) used a larger XUV cross-section for the planets. However, using our hydrodynamical model, we found that for such small planets the XUV cross-section is very similar to the radius of the planet and only causes a difference of a few percents in the quantity of water lost. Fourth, they considered a loss of only hydrogen atoms, while in this work we estimated the ratio of the escape fluxes of both hydrogen and oxygen atoms (rF). We found that in most of the configurations considered here, oxygen atoms are dragged away by the escaping hydrogen atoms, which is more favorable for water retention. 6.3 The effect of flares We only consider here quiescent energetic emissions. However BDs could emit energetic flares for a significant fraction of their lifetime in the Hα emission line and in the U-band (Schmidt et al. 2014; Schmidt 2014). This would also endanger the survival of a water reservoir. Gizis et al. (2013) showed that these flares can be as frequent as 1-2 times a month (e.g., the L1 dwarf W1906+40). W1906+40 experienced a white flare during ∼ 2 hours, which re- leased an energy of 1031 erg (in a band 400–900 nm). Let us con- sider here that the flare released in the XUV range 20% of the en- ergy it released in the 400–900 nm band (this proportion has been measured for Sun flares in Kretzschmar 2011). This flare would then correspond to an energy ∼ 11 times what we considered for the quiescent emission in the case of the constant XUV emission (LX = 1025.4 erg/s). Using the Equation 3, we find that if such a c(cid:13) xx RAS, MNRAS 000, 1–14 Figure 9. Hydrogen loss (as calculated in Section 4) and hydrogen created by photolysis (dashed lines) as a function of time for an Earth-mass planet orbiting a 0.04 M(cid:12) BD at 0.01 au. The top panel corresponds to LX = 1025.4 erg/s and the bottom panel to LX/Lbol = 10−5. The hydrogen quantity created by photolysis was calculated assuming a different efficiency of the photolysis process α. photolysis-limited. Although efficiency calculations would require detailed FUV radiative transfer and photochemical schemes, we can safely argue that efficiencies much lower than 23% can be ex- pected. It is important to stress that the loss is likely to be photolysis- limited, which would allow us to calculate upper limits of the loss without the need of complex thermal and non-thermal escape mod- els. At this point the FUV flux, which is the key input for H2O pho- tolysis, is only estimated based on the XUV/FUV ratio measured on earlier-type stars. Measuring the FUV of UCDs could allow us to put strong constraints on the water erosion on their planets. • The XUV flux considered here might be much higher than what is really emitted by UCDs. Indeed, all Chandra observations of X-ray emissions of low mass objects (e.g. Berger et al. 2010; Williams, Cook, & Berger 2014) are actually non-detection for the UCD range. New estimations from Osten et al. (2015) show that found upper limits for the X-ray luminosity for the object Luh- man 16AB (WISE J104915.57–531906.1, L7.5 and T0.5 spectral types) are lower than what we used in this study: LX < 1023 erg/s or LX/Lbol < 10−5.7. Besides, Mohanty et al. (2002) show that in BDs' cool atmospheres the degree of ionization is very low so it is very possible that the mechanisms needed to emit X-rays are not efficient enough to produce the fluxes considered in this work. In which case, the computed mass loss would be lower than what we calculated here. • We find that the planets might lose several EOH. Therefore, depending on the initial water content, some of them are in danger to be desiccated. The close-in planets we consider may not have formed in situ but migrated from the outer regions of the disk and could have accumulated a large amount of water. Both on the obser- vational and theoretical side, it has been shown that planets could Hydrogen (EOH)Age (Myr)Hydrogen (EOH)LX = 1025.4 erg/sLX/Lbol = 10-5Hydrogen lostHydrogen available from photolysis0.00.20.40.60.80.00.20.40.60.8ε!=1.00ε!=0.10ε!=0.20ε!=0.40ε!=1.000406080100120140t_init = 3 Myr20ε!=0.25ε!=0.30ε!=0.10ε!=0.20ε!=0.25ε!=0.30ε!=0.40Water loss from terrestrial planets orbiting ultracool dwarfs 13 flare happened during 2 hours every months (as could be the case for W1906+40), would reduce slightly the lifetime of the water reservoir. For example, a 1 M⊕ planet at 0.01 au orbiting a UCD of MBD = 0.04 M(cid:12) would lose 0.191 EOH instead of 0.189 EOH be- fore reaching the HZ at ∼ 60 Myr (assuming LX = 1025.4 erg/s, blue squares on Figure 5a). A 1 M⊕ planet at 0.01 au orbiting a UCD of MBD = 0.08 M(cid:12) would lose 0.899 EOH instead of 0.890 EOH be- fore reaching the HZ at ∼ 300 Myr Taking into account the flares therefore does not significantly change the results. 6.4 Water retention does not equal habitability Water retention is not synonymous with habitability. Given that BD's HZs are very close-in, HZ planets feel strong tidal forces. This may affect their ability to host surface liquid water. For ex- ample, a lone planet would likely be in synchronous rotation. One can imagine that all liquid water might condense onto the night side (cold trap). However, this can be avoided if the atmosphere is dense enough to efficient redistribute heat (e.g., Leconte et al. 2013). In multiple-planet systems, a HZ planet's eccentricity can be excited and lead to significant tidal heating (Barnes et al. 2009, 2010; Bol- mont et al. 2013; Bolmont, Raymond & Selsis 2014). In some cases tidal heating could trigger a runaway greenhouse state (Barnes & Heller 2013). In other situations, such as in the outer parts of the HZ or even exterior to the HZ, tidal heating may be beneficial by pro- viding an additional source of heating and perhaps even by helping to drive plate tectonics (Barnes et al. 2009). 7 CONCLUSIONS Considering a very unfavorable scenario for water retention – com- plete dissociation of water molecules – and assuming different val- ues for the X-ray luminosity of the UCDs, we find regions of pa- rameter space (mass of UCD vs orbital distance of the planets) for which planets lose less than a few EOH (here equal to the hydrogen reservoir in one Earth Ocean) before reaching the HZ and can also spend a long time in the HZ. When reaching the HZ, the remaining hydrogen can recombine with the remaining oxygen to form water molecules that can then condense. The longer the planet spends in the HZ, the more time life has to eventually appear, evolve and be observable. Bolmont, Raymond & Leconte (2011) showed that the more massive the BD, the longer a close-in planet spends in the HZ. The low-mass BDs will always suffer from the fact that they cool down very fast and that at best, planets spend a few 100 Myr in the HZ. Even though this could be enough time for life to appear, its potential detectability would be a rare event. This work therefore shows that there is a potential sweet spot for life around UCDs: planets between 0.01 au and 0.04 au orbit- ing BDs of masses between ∼ 0.04 M(cid:12) and 0.08 M(cid:12) (assuming LX/Lbol = 10−5 or LX = 1025.4 erg/s) lose less than 1 EOH while in runaway AND then spend a long time in the HZ ((cid:62) 1 Gyr, ac- cording to Bolmont, Raymond & Leconte 2011). Considering a higher X-ray luminosity (LX/Lbol = 10−4.5 or LX = 1026 erg/s), this sweet spot shifts towards higher orbital distances and higher UCD masses: planets between ∼ 0.02 au and 0.04 au orbiting UCDs of mass between ∼ 0.06 M(cid:12) and 0.08 M(cid:12) lose less than 1 EOH while in runaway AND then spend a long time in the HZ. Of course, if one of the mechanisms considered here does not take place, or if the real XUV flux of BDs is lower than the upper value we consid- ered, as we discussed in the previous Section 6.1, the sweet spot for life could widen towards the smaller orbital distances. c(cid:13) xx RAS, MNRAS 000, 1–14 We also investigated hydrogen losses in the Trappist-1 system (Gillon et al. 2016). Assuming a X-ray quiescent emission compa- rable to a similar star as TRAPPIST-1 (VB 10), we find that the two inner planets of the system TRAPPIST-1 (Gillon et al. 2016) might have lost up to 4 EOH but that the third planet have lost less than 3 EOH. Assuming a X-ray quiescent emission as high as Wheatley et al. (2016), we find that if planet d has an orbital dis- tance of 0.058 au (the most probably one from Gillon et al. 2016) it would have only lost at worst ∼ 0.40 EOH. If the X-ray luminosity of TRAPPIST-1 measured by Wheatley et al. (2016) is confirmed, the observation of the presence of water on the two inner planets would actually bring us informations about the planets initial water content. Despite the lack of knowledge about escape mechanisms, in particular about the way hydrogen and oxygen jointly escape (or not), we find that there are possibilities that planets around UCDs might arrive in the HZ with an important water reservoir even with- out invoking an initial water reservoir larger than Earth's one. As shown in Section 6.1, there might be even more possibilities if the loss of hydrogen is photolysis-limited, which would happen if the efficiency of this process is below 20%. Furthermore, planets in the HZs of BDs may be easy to detect in transit due to their large tran- sit depths and short orbital periods (at least for sufficiently bright sources; Belu et al. 2013; Triaud et al. 2013). Given their large abundance in the Solar neighborhood (∼1300 have been detected to date; see http://DwarfArchives.org), such planets may be among the best nearby targets for atmospheric characterization with the JWST. In particular, the planets of TRAPPIST are an ideal lab- oratory to test the mechanisms of mass loss. ACKNOWLEDGMENTS The authors would like to thank Rory Barnes and Ren´e Heller for bringing this subject to their attention. The authors would also like to thank Rodrigo Luger for useful comments and for helping im- prove our manuscript. The authors also thank the referee for the useful comments on the manuscript. E. B. acknowledges that this work is part of the F.R.S.-FNRS ExtraOrDynHa research project. The work of E. B. was supported by the Hubert Curien Tournesol Program I. R. acknowledges support of Economy and Competitiveness (MINECO) ESP2014-57495-C2-2-R. from the Spanish Ministry through grant F. S. acknowledges support from the Programme National de Plan´etologie (PNP). S. N. R. thanks the Agence Nationale pour la Recherche for support via grant ANR-13-BS05-0003-002 (project MOJO). J. E. O. acknowledges support by NASA through Hubble Fel- lowship grant HST-HF2-51346.001-A awarded by the Space Tele- scope Science Institute, which is operated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. M. G. is Research Associate at the F.R.S.-FNRS. REFERENCES Andreeshchev A., Scalo J., 2004, IauS, 213, 115 Bauer S. J., Lammer H., 2004, Berlin: Springer, 2004 Barnes R., Jackson B., Greenberg R., Raymond S. N., 2009, ApJ, 700, L30 14 E. Bolmont, F. Selsis, J. E. Owen, I. Ribas, S. N. Raymond, J. Leconte and M. Gillon Barnes R., Jackson B., Greenberg R., Raymond S. N., Heller R., 2010, ASPC, 430, 133 Barnes R., Heller R., 2013, AsBio, 13, 279 Barstow J. K., Irwin P. G. J., 2016, arXiv, arXiv:1605.07352 Belu A. R., et al., 2013, ApJ, 768, 125 Berger E., et al., 2008, ApJ, 676, 1307-1318 Berger E., et al., 2010, ApJ, 709, 332 Berta Z. K., et al., 2012, ApJ, 747, 35 Bolmont E., Raymond S. N., Leconte J., 2011, A&A, 535, A94 Bolmont E., Selsis F., Raymond S. N., Leconte J., Hersant F., Maurin A.-S., Pericaud J., 2013, A&A, 556, A17 Bolmont E., Raymond S. N., Selsis F., 2014, sf2a.conf, 63 Chabrier G., Baraffe I., 1997, A&A, 327, 1039 Chabrier G., Gallardo J., Baraffe I., 2007, A&A, 472, L17 Chassefi`ere E., Wieler R., Marty B., Leblanc F., 2012, P&SS, 63, 15 Cook B. A., Williams P. K. G., Berger E., 2014, ApJ, 785, 10 Downes J. J., et al., 2015, MNRAS, 450, 3490 Erkaev N. V., Kulikov Y. N., Lammer H., Selsis F., Langmayr D., Jaritz G. F., Biernat H. K., 2007, A&A, 472, 329 Fleming T. A., Giampapa M. S., Schmitt J. H. M. M., 2000, ApJ, 533, 372 Fortney J. J., Marley M. S., Barnes J. W., 2007, ApJ, 659, 1661 France K., et al., 2016, ApJ, 820, 89 Gillmann C., Chassefi`ere E., Lognonn´e P., 2009, E&PSL, 286, 503 Gillon M., et al., 2016, Natur, 533, 221 Gizis J. E., Burgasser A. J., Berger E., Williams P. K. G., Vrba F. J., Cruz K. L., Metchev S., 2013, ApJ, 779, 172 Gladstone G. R., et al., 2002, Natur, 415, 1000 Hamano K., Abe Y., Genda H., 2013, Natur, 497, 607 Han C., et al., 2013, ApJ, 778, 38 Hayes J. C., Norman M. L., Fiedler R. A., Bordner J. O., Li P. S., Clark S. E., ud-Doula A., Mac Low M.-M., 2006, ApJS, 165, 188 Heller R., Barnes R., 2015, IJAsB, 14, 335 Hunten D. M., Pepin R. O., Walker J. C. G., 1987, Icar, 69, 532 Ikoma M., Genda H., 2006, ApJ, 648, 696 Johnson R. E., Volkov A. N., Erwin J. T., 2013, ApJ, 768, L4 Kaltenegger L., Sasselov D., Rugheimer S., 2013, ApJ, 775, L47 Kasting J. F., 1988, Icar, 74, 472 Kasting J. F., Whitmire D. P., Reynolds R. T., 1993, Icar, 101, 108 Khodachenko M. L., et al., 2007, AsBio, 7, 167 Kopparapu R. K., et al., 2013, ApJ, 765, 131 Kretzschmar M., 2011, A&A, 530, A84 Kuchner M. J., 2003, ApJ, 596, L105 Lammer H., Selsis F., Ribas I., Guinan E. F., Bauer S. J., Weiss W. W., 2003, ApJ, 598, L121 Lammer H., et al., 2007, AsBio, 7, 185 Leconte J., Forget F., Charnay B., Wordsworth R., Selsis F., Mil- lour E., Spiga A., 2013, A&A, 554, A69 L´eger A., et al., 2004, Icar, 169, 499 Liu Y., Joergens V., Bayo A., Nielbock M., Wang H., 2015, A&A, 582, A22 Lissauer J. J., 2007, ApJ, 660, L149 Luger R., Barnes R., 2015, AsBio, 15, 119 Marty B., 2012, E&PSL, 313, 56 Matsui T., Abe Y., 1986, Natur, 322, 526 Mohanty S., Basri G., Shu F., Allard F., Chabrier G., 2002, ApJ, 571, 469 Murray-Clay R. A., Chiang E. I., Murray N., 2009, ApJ, 693, 23 Newton E. R., Irwin J., Charbonneau D., Berta-Thompson Z. K., Dittmann J. A., West A. A., 2016, ApJ, 821, 93 Ogihara M., Ida S., 2009, ApJ, 699, 824 Osten R. A., Melis C., Stelzer B., Bannister K. W., Radigan J., Burgasser A. J., Wolszczan A., Luhman K. L., 2015, ApJ, 805, L3 Owen J. E., Jackson A. P., 2012, MNRAS, 425, 2931 Owen J. E., Wu Y., 2013, ApJ, 775, 105 Owen J. E., Alvarez M. A., 2016, ApJ, 816, 34 Pascucci I., Apai D., Luhman K., Henning T., Bouwman J., Meyer M. R., Lahuis F., Natta A., 2009, ApJ, 696, 143 Payne M. J., Lodato G., 2007, MNRAS, 381, 1597 Pfalzner S., Steinhausen M., Menten K., 2014, ApJ, 793, L34 Pecaut M. J., Mamajek E. E., 2016, arXiv, arXiv:1605.08789 Pizzolato N., Maggio A., Micela G., Sciortino S., Ventura P., 2003, A&A, 397, 147 Quintana E. V., et al., 2014, Sci, 344, 277 Raymond S. N., Cossou C., 2014, MNRAS, 440, L11 Raymond S. N., Barnes R., Mandell A. M., 2008, MNRAS, 384, 663 Raymond S. N., Scalo J., Meadows V. S., 2007, ApJ, 669, 606 Reid I. N., Kirkpatrick J. D., Liebert J., Gizis J. E., Dahn C. C., Monet D. G., 2002, AJ, 124, 519 Ribas I., Guinan E. F., Gudel M., Audard M., 2005, ApJ, 622, 680 Rogers L. A., Seager S., 2010, ApJ, 716, 1208 Rogers L. A., 2015, ApJ, 801, 41 Schmidt S. J., 2014, arXiv, arXiv:1405.6206 Schmidt S. J., et al., 2014, ApJ, 781, L24 Selsis F., et al., 2007, Icar, 191, 453 Selsis F., Kasting J. F., Levrard B., Paillet J., Ribas I., Delfosse X., 2007, A&A, 476, 1373 Solomon S. C., Head J. W., Kaula W. M., McKenzie D., Parsons B., Phillips R. J., Schubert G., Talwani M., 1991, Sci, 252, 297 Spiegel D. S., Burrows A., Milsom J. A., 2011, ApJ, 727, 57 Stone J. M., Norman M. L., 1992, ApJS, 80, 753 Triaud A. H. M. J., et al., 2013, arXiv, arXiv:1304.7248 Udalski A., et al., 2015, ApJ, 812, 47 Weiss L. M., Marcy G. W., 2014, ApJ, 783, L6 Williams P. K. G., Cook B. A., Berger E., 2014, ApJ, 785, 9 Yang J., Cowan N. B., Abbot D. S., 2013, ApJ, 771, L45 Yelle R. V., 2004, Icar, 170, 167 Watson A. J., Donahue T. M., Walker J. C. G., 1981, Icar, 48, 150 Wheatley P. J., Louden T., Bourrier V., Ehrenreich D., Gillon M., 2016, arXiv, arXiv:1605.01564 Williams P. K. G., Cook B. A., Berger E., 2014, ApJ, 785, 9 Zahnle K. J., Kasting J. F., Pollack J. B., 1988, Icar, 74, 62 This paper has been typeset from a TEX/ LATEX file prepared by the author. c(cid:13) xx RAS, MNRAS 000, 1–14
1508.05575	1	1508	2015-08-23T08:11:08	Asteroid thermophysical modeling	[ "astro-ph.EP" ]	The field of asteroid thermophysical modeling has experienced an extraordinary growth in the last ten years, as new thermal infrared data became available for hundreds of thousands of asteroids. The infrared emission of asteroids depends on the body's size, shape, albedo, thermal inertia, roughness and rotational properties. These parameters can therefore be derived by thermophysical modeling of infrared data. Thermophysical modeling led to asteroid size estimates that were confirmed at the few-percent level by later spacecraft visits. We discuss how instrumentation advances now allow mid-infrared interferometric observations as well as high-accuracy spectro-photometry, posing their own set of thermal-modeling challenges.We present major breakthroughs achieved in studies of the thermal inertia, a sensitive indicator for the nature of asteroids soils, allowing us, for instance, to determine the grain size of asteroidal regoliths. Thermal inertia also governs non-gravitational effects on asteroid orbits, requiring thermophysical modeling for precise asteroid dynamical studies. The radiative heating of asteroids, meteoroids, and comets from the Sun also governs the thermal stress in surface material; only recently has it been recognized as a significant weathering process. Asteroid space missions with thermal infrared instruments are currently undergoing study at all major space agencies. This will require a high level of sophistication of thermophysical models in order to analyze high-quality spacecraft data.	astro-ph.EP	astro-ph	Asteroid thermophysical modeling Laboratoire Lagrange, UNS-CNRS, Observatoire de la Cote d’Azur Marco Delbo Michael Mueller SRON Netherlands Institute for Space Research Joshua P. Emery Dept. of Earth and Planetary Sciences - University of Tennessee Dept. of Earth and Planetary Sciences - University of Tennessee Ben Rozitis Maria Teresa Capria Istituto di Astrofisica e Planetologia Spaziali, INAF The field of asteroid thermophysical modeling has experienced an extraordinary growth in the last ten years, as new thermal infrared data became available for hundreds of thousands of asteroids. The infrared emission of asteroids depends on the body’s size, shape, albedo, thermal inertia, roughness and rotational properties. These parameters can therefore be derived by thermophysical modeling of infrared data. Thermophysical modeling led to asteroid size estimates that were confirmed at the few-percent level by later spacecraft visits. We discuss how instrumentation advances now allow mid-infrared interferometric observations as well as high-accuracy spectro-photometry, posing their own set of thermal-modeling challenges. We present major breakthroughs achieved in studies of the thermal inertia, a sensitive indicator for the nature of asteroids soils, allowing us, for instance, to determine the grain size of asteroidal regoliths. Thermal inertia also governs non-gravitational effects on asteroid orbits, requiring thermophysical modeling for precise asteroid dynamical studies. The radiative heating of asteroids, meteoroids, and comets from the Sun also governs the thermal stress in surface material; only recently has it been recognized as a significant weathering process. Asteroid space missions with thermal infrared instruments are currently undergoing study at all major space agencies. This will require a high level of sophistication of thermophysical models in order to analyze high-quality spacecraft data. 1. INTRODUCTION Asteroid thermophysical modeling is about calculating the temperature of asteroids’ surface and immediate sub- surface, which depend on absorption of sunlight, multiple scattering of reflected and thermally emitted photons, and heat conduction. Physical parameters such as albedo (or re- flectivity), thermal conductivity, heat capacity, emissivity, density and roughness, along with the shape (e.g., elevation model) of the body, its orientation in space, and its previous thermal history are taken into account. From the synthetic surface temperatures, thermally emitted fluxes (typically in the infrared) can be calculated. Physical properties are con- strained by fitting model fluxes to observational data. One differentiates between sophisticated thermophysi- cal models (TPMs; Lebofsky and Spencer, 1989; Spencer, 1990; Spencer et al., 1989; Lagerros, 1997, 1996a, 1998; Delbo, 2004; Mueller, 2007; Rozitis and Green, 2011) and simple thermal models, which typically assume spherical shape, neglect heat conduction (or simplify its treatment), and do not treat surface roughness (see Harris and Lagerros, 2002; Delbo and Harris, 2002, for reviews). In the past, us- age of TPMs was reserved to the few exceptional asteroids for which detailed shape models and high quality thermal infrared data existed (Harris and Lagerros, 2002). In the last ten years, however, TPMs became significantly more applicable (see § 6), thanks both to new spaceborn infrared telescopes (Spitzer, WISE and AKARI; see Mainzer et al., 2015) and to the availability of an ever-growing number of asteroid shape models (Durech et al., 2015). After introducing the motivations and the different con- texts for calculating asteroid temperatures (§ 2), we pro- vide an overview of simple thermal models (§ 3) and of TPMs (§ 4). We describe data analysis techniques based on TPMs (§ 5), then we present the latest results and im- plications on the physics of asteroids (§ 6). In § 7, we dis- cuss temperature-induced surface changes on asteroids; see 1 also the chapter ”Asteroid surface geophysics” by Murdoch et al. (2015). All used symbols are summarized in Tab. 1. Note that we do not discuss here the so-called asteroid thermal evolution models that are generally used to compute the temperature throughout the body as a function of time, typically taking into account internal heat sources such as the decay of the radiogenic 26Al. Such models allow one to estimate the degree of metamorphism, aqueous alteration, melting and differentiation that asteroids experienced dur- ing the early phases of the solar system formation (see Mc- Sween et al., 2002, for a review). 2. MOTIVATIONS AND APPLICATIONS OF TPMs Thermophysical modeling of observations of asteroids in the thermal infrared (λ (cid:38) 4 µm) is a powerful tech- nique to determine the values of physical parameters of as- teroids such as their sizes (e.g., Muller et al., 2014a), the thermal inertia and the roughness of their soils (e.g., Muller and Lagerros, 1998; Mueller, 2007; Delbo and Tanga, 2009; Matter et al., 2011; Rozitis and Green, 2014; Capria et al., 2014) and in some particular cases also of their bulk density and their bulk porosity (Rozitis et al., 2013, 2014; Emery et al., 2014; Chesley et al., 2014) Knowledge of physical properties is crucial to under- stand asteroids: for instance, size information is fundamen- tal to constrain the asteroid size frequency distribution that informs us about the collisional evolution of these bodies (Bottke et al., 2005); is paramount for the study of aster- oid families, for the Earth-impact risk assessment of near- Earth asteroids (NEAs; see Harris et al., 2015, for a review), and for the development of asteroid space mission scenarios (§ 5.7). Accurate sizes are also a prerequisite to calculate the volumes of asteroids for which we know the mass, al- lowing us to derive the bulk density, which inform us about the internal structure of these bodies (e.g., Carry, 2012). Thermal inertia, the resistance of a material to temper- ature change (§ 5.2), is a sensitive indicator for the prop- erties of the grainy soil (regolith, Murdoch et al., 2015) on asteroids, e.g., the typical grain size (Gundlach and Blum, 2013) and their degree of cementation (Piqueux and Chris- tensen, 2009a,b) can be inferred from thermal-inertia mea- surements. In general, the regolith is what we study by means of remote-sensing observations. Understanding the regolith is therefore crucial to infer the nature of the under- lying body. Regolith informs us about the geological pro- cesses occurring on asteroids (Murdoch et al., 2015) such as impacts, micrometeoroid bombardment (Horz and Cintala, 1997), and thermal cracking (Delbo et al., 2014). Regolith contains records of elements implanted by the solar wind and cosmic radiation, and therefore informs us about the sources of those materials (Lucey, 2006). Regolith porosity can shed light on the role of electrostatic and van-der-Waals forces acting on the surface of these bodies (e.g., Rozitis et al., 2014; Vernazza et al., 2012). Knowledge of surface temperatures is also essential for the design of the instruments and for the near-surface op- eration of space missions, as in the case of the sample- return missions Hayabusa-II and OSIRIS-REx of JAXA and NASA, respectively, In the future, knowledge of asteroid temperatures will be crucial for planning human interaction with asteroids. Another reason to model asteroid surface temperatures is that they affect its orbital and spin state evolution via the Yarkovsky and YORP effects, respectively (§ 5.8 and Vokrouhlick´y et al., 2015). In particular, thermal inertia dictates the strength of the asteroid Yarkovsky effect. This influences the dispersion of members of asteroid families, the orbital evolution of potentially hazardous asteroids, and the delivery of D (cid:46) 40 km asteroids and meteoroids from the main belt into dynamical resonance zones capable of transporting them to Earth-crossing orbits (see Vokrouh- lick´y et al., 2015, and the references therein). The YORP effect is believed to be shaping the distri- bution of rotation rates (Bottke et al., 2006) and spin vec- tor orientation (Vokrouhlick´y et al., 2003; Hanus et al., 2011, 2013); small gravitationally bound aggregates could be spun up so fast (Vokrouhlick´y et al., 2015; Bottke et al., 2006, and references therein) that they are forced to change shape and/or undergo mass shedding (Holsapple, 2010). Approximately 15% of near-Earth asteroids are observed to be binaries (Pravec et al., 2006), and YORP spin up is pro- posed as a viable formation mechanism (Walsh et al., 2008; Scheeres, 2007; Jacobson and Scheeres, 2011). A further motivation to apply TPM techniques is to con- strain the spin-axis orientation and the sense of rotation of asteroids (examples are 101955 Bennu and 2005 YU55, Muller et al., 2012, 2013). Durech et al. (2015) describe how to use optical and thermal infrared data simultaneously to derive more reliable asteroid shapes and spin properties. The temperature and its evolution through the entire life of an asteroid can alter its surface composition and nature of the regolith (§ 7). For example, when the temperature rises above a certain threshold for a sustained period, certain volatiles can be lost via sublimation (Schorghofer, 2008; Capria et al., 2012), dehydration (Marchi et al., 2009), or desiccation (Delbo and Michel, 2011; Jewitt et al., 2015, and references therein). There can be pronounced and fast temperature variations between day and night. Modeling these temperature varia- tions is fundamental to studying the effect of thermal crack- ing of asteroid surface material (§ 7.1), which was found to be an important source of fresh regolith production (Delbo et al., 2014). 3. SIMPLE THERMAL MODELS We start by introducing the near-Earth asteroid thermal model (NEATM, Harris, 1998) that is typically used where the data quality and/or the available knowledge about as- teroid shape and spin preclude the usage of TPMs. Typ- ically, the NEATM allows a robust estimation of asteroid diameter and albedo, but does not provide any direct infor- mation on thermal inertia or surface roughness (see Harris 2 and Lagerros, 2002, for a review). The recent large-scale thermal-emission surveys of asteroids and trans-Neptunian objects (see Mainzer et al., 2015; Lellouch et al., 2013, and references therein) typically use the NEATM in their data analysis, thereby establishing it as the de-facto default among the simple thermal models. The typical NEATM accuracy is 15% in diameter and roughly 30% in albedo (Harris, 2006). Other simple thermal models are the “Stan- dard Thermal Model” (STM; Lebofsky et al., 1986), the “Isothermal Latitude Model” (ILM, also known as the Fast Rotating Model or FRM, Lebofsky and Spencer, 1989), and the night emission simulated thermal model (NESTM by Wolters and Green, 2009). The STM and the ILM, reviewed by, e.g., Harris and Lagerros (2002), have largely fallen out of use. The NEATM assumes that the asteroid has a spherical shape and does not directly account for thermal inertia nor surface roughness. The surface temperature is given by the instantaneous equilibrium with the insolation, which is pro- portional to the cosine of the angular distance between local zenith and the Sun and zero at the night side. The maximum temperature occurs at the subsolar point and it reads: (1 − A)S(cid:12)r−2 = ησ T 4 SS (1) (nomenclature is provided in Tab. 1). The parameter η was introduced in the STM of Lebofsky et al. (1986) as a means of changing the model temperature distribution to take ac- count of the observed enhancement of thermal emission at small solar phase angles due to surface roughness. This is known as the beaming effect. For this reason η is also called the beaming parameter. The η formalism, in the NEATM, allows a first-order description of the effect of a number of geometrical and physical parameters, in particular the thermal inertia and surface roughness on the spectral en- ergy distribution of an asteroid (Delbo et al., 2007). For a large thermal inertia, one would expect η-values signifi- cantly larger than unity (e.g., 1.5–3; with theoretical max- imum values around 3.5; Delbo et al., 2007), whereas for low thermal inertia η (cid:39) 1 (for Γ = 0 and zero surface rough- ness). Roughness, on the other hand, tends to lower the value of η (for observations at low or moderate phase an- gles). For instance, a value of η ∼ 0.8 for a main belt as- teroid indicates that this body has low thermal inertia and significant roughness (with minimum theoretical values of 0.6 - 0.7; Spencer, 1990; Delbo et al., 2007). We note, how- ever, that η is not a physical property of an asteroid, as it can vary due to changing observing and illumination geometry, aspect angle, heliocentric distance of the body, phase angle and wavelength of observation. 4. THERMOPHYSICAL MODELS 4.1. Overview Different TPMs have been proposed to study the ther- mal emission of asteroids, comets, planets, and satellites. The first models were motivated by thermal observations of the lunar surface, which revealed an almost thermally 3 insulating surface that emitted thermal radiation in a non- Lambertian way (Pettit and Nicholson, 1930; Wesselink, 1948a). Heat conduction and radiation scattering models of various rough surfaces were able to reproduce the lu- nar observations to a good degree (e.g., Smith, 1967; Buhl et al., 1968b,a; Sexl et al., 1971; Winter and Krupp, 1971), and the derived thermal inertia and surface roughness val- ues matched in situ measurements by Apollo astronauts (see Rozitis and Green, 2011, and references therein). These early lunar models were adapted to general planetary bod- ies, albeit with an assumed spherical shape, by Spencer et al. (1989) and Spencer (1990). The most commonly used asteroid TPMs of Lagerros (1996a, 1997, 1998), Delbo (2004), Mueller (2007), and Rozitis and Green (2011) are all based on Spencer et al. (1989) and Spencer (1990). Here we present the basic principles utilized in TPMs; for imple- mentation details, the reader is referred to the quoted works. All TPMs represent the global asteroid shape as a mesh of (triangular) facets (see Fig. 1) that rotates around a given spin vector with rotation period P . In general, utilized shape models are derived from radar observations, inversion of optical light-curves, in-situ spacecraft images, or stellar occultation timing (see Durech et al., 2015, for a review of asteroid shape modeling). If no shape model is available, one typically falls back to a sphere or an ellipsoid (e.g., Muller et al., 2013, 2014a; Emery et al., 2014). The goal is to calculate the thermal emission of each facet of the shape model at a given illumination and ob- servation geometry. To this end, the temperature of the sur- face and, in the presence of thermal inertia, the immediate subsurface need to be calculated. Generally, lateral heat conduction can be neglected as the shape model facets are much larger than the penetration depth of the diurnal heat wave (i.e., the thermal skin depth), and only 1D heat con- duction perpendicular to and into the surface needs to be considered. For temperature T , time t, and depth z, 1D heat conduction is described by: ρC ∂T ∂t = ∂ ∂z κ ∂T ∂z (2) where κ is the thermal conductivity, ρ is the material den- sity, and C is the heat capacity. If κ is independent of depth (and, implicitly, temperature independent, see §5.3), Eq. 2 reduces to the diffusion equation: ∂T ∂t = κ ρC ∂T 2 ∂2z (3) It is useful to define the thermal inertia Γ and the thermal skin depth ls (cid:112)κρC (cid:112)κP/2πρC. Γ = (4) ls = (5) These material properties are generally assumed to be constant with depth and temperature in asteroid TPMs, but varying properties has been considered in some Moon, Mars, planetary satellites, and asteroids models (e.g Giese TABLE 1 NOMENCLATURE Symbol Quantity Unit Symbol Quantity T TSS σ S(cid:12) r (cid:126)r ∆ η κ C Γ Θ ρ H G V D A pV z t rp Temperature Subsolar temperature Stefan-Boltzmann constant Solar constant at r=1 au Distance to the Sun Vector to the Sun Distance to the observer Emissivity Beaming parameter Thermal conductivity Heat capacity Thermal inertia Thermal parameter Material density Absolute magnitude of the H, G system Slope parameter of the H, G system Actual magntude in the V-band Diameter Bolometric Bond albedo Geometric visible albedo Depth in the subsoil Time Pore radius of regolith K K (5.67051 10−8) W m−2 K−4 (1329) W m−1 au m au - - W m−1 K−1 J kg−1 K−1 J m-2s-1/2K-1 - kg m−3 m (or km) - - m s m ¯θ ls P ω λp βp φ0 α a S F(cid:126)ı,(cid:126) JV ((cid:126)) JIR((cid:126)) n (cid:126)ı (cid:126) γC ρC or f φ fλ(τ ) λ Mean surface slope Thermal skin depth Rotation period = 2π/P ecliptic longitude of the pole ecliptic latitude of the pole Initial rotational phase at epoch (Phase) angle between asteroid-sun-observer Area of a facet Shadowing function View factor Visible radiosity Infrared radiosity Local normal Vector to the local facet Vector to the remote facet Crater opening angle Area density of craters Emission angle Infrared flux Wavelength Unit deg m s s−1 deg deg deg deg m2 m m m deg rad W m−2 µm−1 µm and Kuehrt, 1990; Urquhart and Jakosky, 1997; Piqueux and Christensen, 2011; Keihm, 1984; Keihm et al., 2012; Capria et al., 2014, see also section 5.3). TPM implementations typically employ dimensionless time and depth variables: τ = 2πt/P and Z = z/ls. Then, √ the only remaining free parameter is the dimensionless ther- mal parameter (Θ = Γ SS, Spencer, 1990) describ- ing the combined effect of thermal inertia, rotation period, and heat emission into space on the surface temperature dis- tribution (see Fig. 2). ω/σT 3 A numerical finite-difference technique is used to solve the 1D heat conduction equation, and an iterative technique is used to solve the surface boundary condition. This re- quires a suitable number of time and depth steps to fully resolve the temperature variations and to ensure model sta- bility (typically, at least 360 time steps and 40 depth steps over 8 thermal skin depths are required). TPMs are run until specified convergence criteria are met (e.g., until tem- perature variations between successive model iterations are below a specified level) and/or until a specified number of model iterations have been made. For applications such as the study of the sublimation of water ice from the shallow subsurface of asteroids (e.g., the Main Belt Comets or 24 Themis) the heat conduction equation must be coupled with a gas diffusion equation (Schorghofer, 2008; Capria et al., 2012; Prialnik and Rosen- berg, 2009). See also Huebner et al. (2006) for a review. The 1D heat conduction equation is solved with inter- nal and surface boundary conditions to ensure conservation of energy. Since the amplitude of subsurface temperature variations decreases exponentially with depth, an internal boundary condition is required to give zero temperature gra- dient at a specified large depth (cid:18) ∂T (cid:19) ∂z z(cid:29)ls = 0. (6) A typical surface boundary condition for a facet at point (cid:126)ı with respect to the asteroid origin, at point (cid:126)r with respect to the Sun, and with surface normal n is then given by (7) (cid:18) ∂T ((cid:126)ı, t) (cid:19) ∂z = z=0 ((cid:126)r · n)(1 − S((cid:126)r,(cid:126)ı))+ JV ((cid:126))F(cid:126)ı,(cid:126) da(cid:48) + JIR((cid:126))F(cid:126)ı,(cid:126) da(cid:48) (cid:90) σT 4((cid:126)ı, t) − (1 − A)S(cid:12) (cid:90) (cid:126)r 3 (1 − A) σ(1 − ) The left-hand side of Eq. (7) gives the thermal energy ra- diated to space and the heat conducted into the subsurface, and the right-hand side gives the input radiation from three different sources: direct solar radiation, multiply scattered solar radiation (i.e., self-illumination), and reabsorbed ther- mal radiation (i.e., self-irradiation). The two last compo- nents are also known as mutual heating (see Fig. 3). The amount of solar radiation absorbed by a facet de- pends on the Bond albedo A and any shadows projected on it, which is dictated by S((cid:126)r,(cid:126)ı) (i.e., S((cid:126)r,(cid:126)ı) = 1 or 0, de- pending on the presence or absence of a shadow). Projected shadows occur on globally non-convex shapes only, which can be determined by ray-triangle intersection tests of the solar illumination ray (e.g., Rozitis and Green, 2011) or by local horizon mapping (e.g., Statler, 2009). Related to shad- owing are the self-heating effects arising from interfacing 4 Fig. 1.— (a) example of a triangulated 3D shape model as typically used in TPMs (asteroid (2063) Bacchus from http://echo.jpl.nasa.gov/asteroids/shapes/shapes.html). Tempera- tures are color coded: white corresponds to the maximum and dark-gray corresponds to minimum temperature. Three different roughness models are sketched in the bottom of the figure: (b) hemispherical section craters; (c) Gaussian surface; (d) fractal sur- face. Sub-figures b and d are adapted from Davidsson et al. (2015), c is from Rozitis and Green (2011). facets, which tend to reduce the temperature contrast pro- duced inside concavities. The problem here is to determine which facets see other facets, and to calculate the amount of radiation exchanged between them. The former can be determined by ray-triangle intersection tests again, and the latter can be solved using view factors. The view factor F(cid:126)ı,(cid:126) is defined as the fraction of the radiative energy leaving the local facet (cid:126)ı that is received by the remote facet (cid:126) as- suming Lambertian emission (Lagerros, 1998). JV ((cid:126)) and JIR((cid:126)) are then the visible and thermal-infrared radiosities of remote facet (cid:126). Either single or multiple scattering can be taken into account, and the latter can be efficiently solved using Gauss-Seidel iterations (Vasavada et al., 1999). Most TPMs neglect shadowing and self-heating effects resulting from the global shape for simplicity, but they can be sig- nificant on asteroids with large shape concavities (e.g., the south pole of (6489) Golevka: Rozitis and Green, 2013). 4.2. Modeling asteroid thermal emission Once the surface temperature distribution across an as- teroid surface has been computed, the emission spectrum (Fig. 4) at a given observation geometry and a specified time can be calculated. The monochromatic flux density can also be calculated at wavelengths of interest. When these model fluxes are plotted as a function of the asteroid rotational Fig. 2.— Synthetic diurnal temperature curves on the equator of a model asteroid for different values of thermal inertia (in units of J m-2s-1/2K-1). Increasing thermal inertia smooths temperature contrasts and causes the temperature peak to occur after the insola- tion peak at 3 h. The asteroid is situated at a heliocentric distance of r = 1.1 au, has a spin period of 6 h, a Bond albedo of A = 0.1, and its spin axis is perpendicular to the orbital plane. phase, one obtains the so-called thermal lightcurves (e.g. Fig. 5), which can be used to test the fidelity of shape and albedo models typically used as input in the TPM. When the temperature for a facet is known, the inten- sity Iλ(τ ) at which it emits at wavelength λ is given by the Planck function. Assuming Lambertian emission, the spec- tral flux of the facet seen by an observer is then fλ(τ ) = Iλ(τ ) a ∆2 cos φ (8) where a is area of the facet, ∆ is the distance to the ob- server, and φ is the emission angle. The total observed flux is obtained by summing the thermal fluxes of all vis- ible shape model facets including any contributions from surface roughness elements contained within them. For disk-integrated measurements, this summation is performed across the entire visible side of the asteroid, whilst for spa- tially resolved measurements it is summed across facets contained within the detector pixel’s field of view. The assumption of Lambertian emission depends on no directionality induced by surface irregularities at scales below the thermal skin depth. Davidsson and Rickman (2014) show that surface roughness at sub-thermal-skin- depth scales is quasi-isothermal and is therefore not likely to deviate from Lambertian emission overall. However, radiative transfer processes between the regolith grains could contribute up to 20% of the observed beaming effects (Hapke, 1996). Rozitis and Green (2011, 2012) investi- gated combined microscopic (regolith grain induced) and macroscopic (surface roughness induced) beaming effects, and demonstrated that the macroscopic effects dominated overall. This was previously found to be the case in di- rectional thermal emission measurements of lava flows on Earth (Jakosky et al., 1990). 5 00.20.40.60.8100.20.40.60.81−0.3−0.2−0.10xyz00.20.40.60.8100.20.40.60.8100.05xyzFigure3:ISSIThermalEmissionTeam6700.20.40.60.8100.20.40.60.81−0.3−0.2−0.10xyz00.20.40.60.8100.20.40.60.8100.05xyzFigure3:ISSIThermalEmissionTeam6799 Figure 5.3: Wireframe renderings of various rough surfaces. (1st row) 30º and 45º craters. (2nd row) 60º crater and 90º high resolution crater. (3rd row) 90º medium resolution and low resolution craters. (4th row) High resolution and low resolution Gaussian random height surfaces. bcdaZYX 50 100 150 200 250 300 350 400 0 1 2 3 4 5 6Temperature (K)Time (h)110100100010000Fig. 3.— Diagram illustrating the energy balance and radiation transfer between facets (copied from Rozitis and Green, 2011). The terms FSUN, FSCAT, FRAD, k(dT /dx) and σT 4 are the di- rect sunlight, multiply scattered sunlight, reabsorbed thermal radi- ation, conducted heat and thermal radiation lost to space, respec- tively. As wavelengths increase to the submillimeter range and above, asteroid regolith becomes increasingly transparent and the observed flux is integrated over increasing depths (Chamberlain et al., 2009; Keihm et al., 2013). Model- ing such fluxes with typical thermal models (which derive fluxes from surface temperatures, only) requires a signifi- cant reduction in effective spectral emissivity. For example, 3.2 mm flux measurements of (4) Vesta require an emissiv- ity of ∼0.6 to match model predictions (Muller and Barnes, 2007). The reduction in emissivity can be explained by lower subsurface emission temperatures (Lagerros, 1996b) and by different subsurface scattering processes dependent on grain size (Redman et al., 1992; Muller and Lagerros, 1998). Keihm et al. (2012, 2013) attribute the reduced emis- sivity at submm/mm wavelengths to a higher thermal inertia value of the subsurface layers. Reduction in emissivity has also been determined at wavelengths shorter than 4.9 µm for disk-resolved regions of (4) Vesta (Tosi et al., 2014). 4.3. Surface Roughness Roughness causes an asteroid surface to thermally emit in a non-Lambertian way with a tendency to reradiate the absorbed solar radiation back towards the Sun, an effect known as thermal infrared beaming (Lagerros, 1998; Rozi- tis and Green, 2011). It is thought to be the result of two different processes: a rough surface will have elements ori- entated towards the Sun that become significantly hotter than a flat surface, and multiple scattering of radiation be- tween rough surface elements increases the total amount of solar radiation absorbed by the surface. The relevant size scale ranges from the thermal skin depth to the linear size of the facets in the shape model. It is included in thermo- physical models by typically modeling an areal fractional Fig. 4.— (a) Example of SED calculated from a TPM and the NEATM compared to a Spitzer spectrum of (87) Sylvia. (b) Spectral emissivity derived from the above: data divided by the NEATM continuum. Figure from Marchis et al. (2012). coverage (f) of spherical-section craters (of opening an- gle γC) within each shape model facet. Other more com- plex forms have been considered, such as Gaussian random (Lagerros, 1998) or fractal (Groussin et al., 2013) surfaces or parallel sinusoidal trenches (see sketch of Fig. 1) , but the spherical-section crater produces similar results (in terms of the disk-integrated beaming effect Lagerros, 1998) and ac- curately reproduces the directionality of the lunar thermal infrared beaming effect (Rozitis and Green, 2011). How- ever, it has been shown that the thermal emission depends also on roughness type in addition to roughness level, for disk resolved data Davidsson et al. (2015). Spherical-section craters are typically implemented, as the required shadowing and view-factor calculations can be performed analytically (Emery et al., 1998; Lagerros, 1998). Heat conduction can be included by dividing the crater into several tens of surface elements where the same equations listed above can be applied. Alternatively, the temperature distribution within the crater resulting from heat conduction, Tcrater(Γ), can be approximated using Trough(Γ) Trough(0) = Tsmooth(Γ) Tsmooth(0) (9) where Trough(0) can be calculated analytically assuming in- stantaneous equilibrium (Lagerros, 1998). Tsmooth(0) and Tsmooth(Γ) are the corresponding smooth-surface tempera- tures, which can be calculated exactly. This approximation is computationally much cheaper than the full implemen- tation. However, it does not work on the night side of the asteroid and temperature ratios diverge near the terminator (Mueller, 2007). An even simpler alternative is to multi- ply the smooth-surface temperatures by a NEATM-like η 6 66 Figure 4.1: Schematic of the Advanced Thermophysical Model (ATPM) where the terms FSUN, FSCAT, FRAD, k(dT/dx), and εσT4 are the direct sunlight, multiple scattered sunlight, reabsorbed thermal radiation, conducted heat, and thermal radiation lost to space respectively. The 3D shape of a planetary body is broken down into a number of discrete triangular surface elements called facets, and can be obtained from lightcurve and radar inversion, or from direct imaging by a spacecraft. Artificial shapes can be easily generated if necessary (see appendix A.1). Each shape facet has its own unique set of thermal properties including thermal inertia and surface roughness, but generally these are assigned the same values across the planetary body. These shape facets are large enough so that lateral heat conduction can be neglected, and only 1D heat conduction perpendicular and into the surface can be considered. A shape facet's surface temperature is determined by solving the 1D heat conduction equation with a surface boundary condition that includes time-varying direct and multiple scattered sunlight, shape shadowing effects, and reabsorbed thermal radiation from interfacing shape facets. Generally, surface roughness that leads to thermal infrared beaming is not resolved in the planetary body shape model, and therefore it is artificially added to each shape facet. A separate topography model representing the unresolved surface roughness is used, and consists of an additional set of roughness facets. These roughness facets are considered to be smaller than the shape facets, but are larger The Sun Planetary Body FSUN FSCAT FRAD k(dT/dx) FSUN εσT4 εσT4 k(dT/dx) Fig.1.(continued)1136F.Marchisetal./Icarus221(2012)1130–1161Fig.1.(continued)1136F.Marchisetal./Icarus221(2012)1130–11610 10 20 30 401.0 0.9 0.8Fig.1.(continued)1136F.Marchisetal./Icarus221(2012)1130–1161Wavelength (micron)Emissivity50 40 30 20 10Flux density (Jy)Data TPM NEATM(a)(b)and Green, 2011). 5. DATA ANALYSIS USING A TPM 5.1. Thermal infrared spectro-photometry Physical properties than can be derived from TPM fits to disk-integrated thermal observations include the diam- eter, geometric albedo, thermal inertia and roughness. In practically all cases, the absolute visual magnitude H is known, establishing a link between D and pV and reduc- ing the number of TPM fit parameters by one: −H/5, −1/2 D(km) = 1329 p V (11) 10 (Fowler and Chillemi, 1992; Vilenius et al., 2012). Fre- quently, the rotational phase during the thermal observa- tions is not sufficiently well known and has to be fitted to the thermal data (e.g., Harris et al., 2005; Al´ı-Lagoa et al., 2014). In some cases, TPMs can be used to constrain the orientation of the spin vector of an asteroid, with λp and βp treated as free parameters (as demonstrated e.g., by Muller et al., 2013, 2012, note that in the case of 101955 Bennu the radar-constrained pole solution was not yet known). More- over, Muller et al. (2014b) successfully performed a TPM analysis of an asteroid (99942 Apophis) in a non-principal axis rotation state for the first time. The thermal effects of thermal inertia and surface rough- ness are difficult to tell apart. A commonly used approach is to use four different roughness models corresponding to no, low, medium, and high roughness, with each model leading to a different thermal-inertia fit (Mueller, 2007; Delbo and Tanga, 2009); frequently, the scatter between these four so- lutions accounts for the bulk of the uncertainty in thermal inertia. However, in some lucky cases, data do allow the effects of roughness and thermal inertia to be disentangled. This requires good wavelength coverage straddling the ther- mal emission peak and good coverage in solar phase angle, such that both the morning and afternoon sides of the aster- oid are seen. See Fig. 6 for an illustration. The best-fitting model parameters are those that mini- mize χ2. Their uncertainty range is spanned by the values that lead to χ2 within a specified threshold of the best fit, depending on the number of free fit parameters. Ideally, the reduced χ2 of the best fit should be around unity. How- ever, due to systematic uncertainties introduced in thermal infrared observations (e.g., flux offsets between different in- struments), and/or in the thermophysical modeling, it is not uncommon to get large reduced χ2 values. Large χ2 val- ues are also obtained when the assumed shape model dif- fers significantly from the asteroid’s true shape (Rozitis and Green, 2014). In particular, if the spatial extent of the shape model’s z-axis is wrong, this can lead to diameter determi- nations that are inconsistent with radar observations (e.g., for 2002 NY40 and (308635) 2005 YU55 in Muller et al., 2004, 2013, respectively), and/or two different thermal in- ertia determinations (e.g., the two different results produced for (101955) Bennu by Emery et al., 2014; Muller et al., Fig. 5.— Example of a TPM generated thermal lightcurve (dashed line) and real data for (101955) Bennu. From Emery et al. (2014). value (e.g. Groussin et al., 2011). Whilst this alternative might produce the correct disk-integrated color tempera- ture of the asteroid, it does not reproduce the directionality of the beaming effect. Indeed, roughness models predict a limb-brightening effect (Rozitis and Green, 2011), which is seen in spatially-resolved measurements of (21) Lutetia by Rosetta (Keihm et al., 2012). The above implementations neglect lateral heat conduc- tion, although the spatial scales representing surface rough- ness can, in some cases, become comparable to the thermal skin depth. Modeling of 3D heat conduction inside rocks the size of the thermal skin depth has demonstrated that their western sides (for a prograde rotator; eastern sides for a retrograde rotator) are generally warmer than their east- ern sides, which could result in a tangential-YORP effect that predominantly spins up asteroids (Golubov and Krugly, 2012). Other than this, it appears that the 1D heat conduc- tion approximation still produces satisfactory results. In thermophysical models, the degree of surface rough- ness can be quantified in terms of the Hapke mean surface slope (cid:90) π/2 tan ¯θ = 2 π 0 a(θ) tan θ dθ (10) where θ is the angle of a given facet from horizontal, and a(θ) is the distribution of surface slopes (Hapke, 1984). Al- ternatively, it can be measured in terms of the RMS sur- face slope (Spencer, 1990). This then allows comparison of results derived using different surface roughness represen- tations (e.g., craters of different opening angles and frac- tional coverages, or different Gaussian random surfaces: Davidsson et al., 2015), and comparison against rough- ness measured by other means. It has been demonstrated that different roughness representations produce similar de- grees of thermal infrared beaming when they have the same degree of roughness measured in terms of these values (Spencer, 1990; Emery et al., 1998; Lagerros, 1998; Rozitis 7 abovethenoiseintheIRSdata(e.g.,Salisburyetal.,1991),suggest-ingthatthisoptionisalsonotviable.Athirdinterpretation,andtheonethatwefavor,isthatacom-binationofsurfacegrainsizeandpackingstateconspirestokeepthespectralcontrastofthesilicatefeaturesverysmall.Laboratoryexperiments(e.g.,HuntandLogan,1972)andMietheorycalcula-tions(e.g.,Harkeretal.,2002)bothrevealthatsingle-grainemis-sioncrosssectionsforsilicatematerialsintheMIRdecreaseinspectralcontrastasparticlesizeisincreasedfromaboutamicronuptoseveraltensofmicrons.Whenthegrainsarepartofarego-lith,thepolarityoftheemissionfeatureneartheSi–Ofundamen-talsswitchesforlargegrains,suchthatforsampleswithaverageparticlesizegreaterthanafewhundredmicrons,thereisaclearemissivitylow(withsuperposedreststrahlenfeatures)reminiscentofthespectrumofapolishedfaceorbedrocksample.Thicklayersofsmallgrainscanalsosomewhatmimicthiseffect,decreasingthespectralcontrastascomparedtosingle-grainemissivity(HuntandLogan,1972).Infact,theonlywaytoproduceaspectrumlikethatofHektorshowninFig.13isforthesurfacetobecomposedoffine-grained(lm-sized)particlesthatarewell-separatedfromoneanother(e.g.,inanextremelyhighporosityfairy-castlestructure,embeddedinamediumthatistransparentintheMIR,ordispersedinanopticallythincloudorcomaabovethesurface;Emeryetal.,2006).TheabsenceofsuchfeaturesfromthespectraofBennuindi-catesthatthesurfaceisnotcoveredinsuchafine-grained,fairy-castleregolithofsilicateparticles(spectralobservationssensetheupperfewtensofmicronstofewmillimeters,dependingonthemeanfreephotonpathatagivenwavelength).Theabsenceoffeaturesofoppositepolarityindicatesthesurfaceisnotcoveredinbedrockorbouldersofsilicatematerial.Theemissivityspectraarethereforeinconclusiveintermsofsurfacemineralogy,buttheydoplacethesebroadconstraintsonthestructureofthesurfacelayer.3.3.ComasearchWeusetheinfraredimagesofBennuobtainedwithSpitzertoperformasensitivesearchforthepresenceofdustaroundtheasteroid.ThemajorityofNEAsdonotshowanyevidenceofsuchextendedemission,buttherearesomeexceptions.Mostpromi-nentamongtheexceptionsis3200Phaethon,whichisinthesametaxonomicclass(B-type)asBennu(deLeónetal.,2010).PhaethonistheparentoftheGeminidmeteorshower,sohasclearlysheddustatsomepointinitsrecentpast.Furthermore,JewittandLi(2010)detectedabrighteningofPhaethonwhenitwas0.14AUfromtheSun,whichtheyinterpretasanimpulsivereleaseofdust.Jewittetal.(2013)furtherreportadusttailimagednearperihelionpassageofPhaethonin2009and2012.Inaddition,adustycomaaroundtheasteroidwouldrepresentapotentialhazardtosafeoperationofthespacecraft.Therefore,eventhoughnodustisanticipatedaroundBennu,asearchforadustcomaisconducted.Thepeak-upimagesat16and22lmprovidethemostsensitivesearchfordust,becausetheysensethepeakoftheblackbodycurvefordustat1AU.RadialprofilesofallPUIframesaregenerated,using0.2-pixelcubicsplineinterpolation,andco-addedbywave-lengthtogenerateasuperprofile.Thesuperprofileiscomparedtoamodelpointspreadfunction(PSF)generatedwiththeSTiny-Tim7softwarefora290Kblackbodyspectrum.The16lmPSFwassmoothedbytheequivalentofa1.9pixelboxcar(1.5pixelsfor0.00.20.40.60.81.0Rotational Phase0.200.250.300.350.40Flux (mJy)4.5 µm(a)0.00.20.40.60.81.0Rotational Phase4.04.55.05.56.0Flux (mJy)8.0 µm(b)0.00.20.40.60.81.0Rotational Phase1011121314Flux (mJy)22 µm (c)Fig.12.ThermallightcurvescalculatedwiththeTPMusingtheradarshapeoverplottedonIRACdataat(a)4.5,(b)8.0,and(c)22lm.ThelightcurvesareinterpolatedusingthethermalinertiasinTable8forreference.7http://irsa.ipac.caltech.edu/data/SPITZER/docs/dataanalysistools/tools/contrib-uted/general/stinytim/.J.P.Emeryetal./Icarus234(2014)17–3529leads to non-zero night-side temperatures, and causes the surface temperature to peak on the afternoon side, as shown in Fig. 2, thereby causing the Yarkovsky effect (§ 5.8). The mass density ρ, the specific heat capacity c, and the thermal conductivity κ, and correspondingly Γ itself, must be thought of as effective values representative of the depth range sampled by the heat wave, which is typically in the few-cm range. In turn, thermal-inertia values inform us about the physical properties of the top few centimeters of the surface, not about bulk properties of the object. As will be discussed below (§ 6.5), ρ and c of an asteroid surface can plausibly vary within a factor of several, while plausible values of κ span a range of more than 4 orders of magnitude. It is therefore not unjustified to convert from Γ to κ and back using reference values for ρ and c (note that Yarkovsky/YORP models tend to phrase the thermal- conduction problem in units of κ, while TPMs tend to be formulated in units of Γ, which is the observable quantity). Importantly, the κ of finely powdered lunar regolith is 3 orders of magnitude lower than that of compact rock (com- pact metal is even more conductive by another order of magnitude). This is because radiative thermal conduction between regolith grains is significantly less efficient than phononic heat transfer within a grain. A fine regolith, an aggregate of very small grains, is a poor thermal conduc- tor and displays a low Γ. Thermal inertia can therefore be used to infer the presence or absence of thermally in- sulating powdered surface material. In extension, thermal inertia can be used as a proxy of regolith grain size. The required calibration under Mars conditions (where the ten- uous atmosphere enhances thermal conduction within pores compared to a vacuum) was obtained by Presley and Chris- tensen (1997a,b) and used in the analysis of thermal-inertia maps of Mars (see Mellon et al., 2000; Putzig and Mel- lon, 2007). Similar progress in asteroid science was slowed down by the lack of corresponding laboratory measure- ments under vacuum conditions (but see below for recent lab measurements of meteorites). However, Gundlach and Blum (2013) provided a calibration relation based on heat- transfer modeling in a granular medium. 5.3. Temperature dependence of thermal inertia Thermal inertia is a function of temperature (Keihm, 1984), chiefly because the thermal conductivity is. In gen- eral, for a lunar-like regolith the thermal conductivity is given by: κ = κb + 4σrpT 3 (12) where κb is the solid-state thermal conductivity (heat con- duction by phonons) and rp is the radius of the pores of the regolith. The term proportional to T 3 is due to the heat con- duction by photons. Equation 12 is often written in the form κ = κ0(1 + χT 3) (e.g., Vasavada et al., 1999). Note that κb is itself a function of T (Opeil et al., 2010). There is ex- tensive literature on the T -dependence of the conductivity of lunar regolith (e.g., Vasavada et al., 1999, and references therein). A theoretical description of the temperature de- Fig. 6.— (21) Lutetia: a TPM fit that allows surface rough- ness to be constrained. The quantity ρ here is not the bulk density of the body, but it is the r.m.s. of the slopes on the surface. It is related to the ratio between the diameter and the depth of spherical section craters (Lagerros, 1998) in this particular case. f is the areal fraction of each facet covered with craters. From O’Rourke et al. (2012). 2012). In some works, the asteroid shape model has also been optimized during the thermophysical fitting to resolve inconsistencies with radar observations (e.g., (1862) Apollo and (1620) Geographos in Rozitis et al., 2013; Rozitis and Green, 2014, respectively). We remind the reader that the accuracy of the physical properties (in particular the value of Γ) of asteroids derived from TPM depends on the quality of the thermal infrared data, coverage in wavelength, phase, rotational, and aspect angle. The accuracy of the shape model and of the H and G values are also important (see e.g. Rozitis and Green, 2014). The derived thermal inertia value often depends on the assumed degree of roughness and it is usually affected by large errors (e.g. 50 or 100 %, see Tab. 2). Care must be used in accepting TPM solutions purely based on the good- ness of fit (e.g. the value of the χ2), as they can be domi- nated by one or few measurements with unreliable small er- rors or calibration offsets between measurements from dif- ferent sources. 5.2. Thermal Inertia and Thermal Conductivity As asteroids rotate, the day-night cycle causes cyclic temperature variations that are controlled by the thermal in- ertia (defined by Eq. 4) of the soil and the rotation rate of the body. In the limit of vanishing thermal inertia, the surface temperature would be in instantaneous equilibrium with the incoming flux, depending only on the solar incidence an- gle (as long as self heating can be neglected); surface tem- peratures would peak at local noon and would be zero at night. In reality, thermal conduction into and from the sub- soil causes a certain thermal memory, referred to as ther- mal inertia. This smoothens the diurnal temperature profile, 8 Asaresult,usingtheLAMmodel,the‘‘true’’H-magforLutetia,basedupontheshapemodelwithaneffectivediameter(ofanequalvolumesphere)of95.97kmandanalbedoof0.2070.01,iscalculatedtobeH-mag¼7.4870.03.ThesetwovalueswerefedintotheradiometricanalysisdescribedinthenextsubsectionviatheTPMcodetoustoderiveaconclusivevalueforthesurfaceroughnessandthermalinertiaof(21)Lutetia.4.3.Surfaceroughnessandthermalinertiaof(21)LutetiaAsdescribedinSections2and3,92observationsfromdifferentobservatorieswerefedintothemodel.ThethermalmodelwasthereforerunnotonlyagainstHerschelPACSandSPIREobservations,thekeydatasetforthisanalysis,butalsoacomprehensivesetof76observationsmadeofLutetiabyotherobservatoriesasdescribedinTable4.ATPMw2testwasrunusingarangeofthermalinertias(from1to50)withthegoaltofindthebestfitobtainedwithaspecificthermalinertiavaluewhencomparingtheobservation/TPMratiowithphaseangles,wavelength,rotationalphaseandaspectangle.Theinitialresultofthisw2testshowedthatadrivingconstraintonallparameterswasthesurfaceroughness.TheanalysisoftheFig.2.ThermalInertiaw2test—thisfigureshowstheimpactofroughnessontheoverallthermalinertiawherebythebestfitliesclearlywithroughnessofr¼0.6andf¼0.7.Theslopeofthisbottomcurvecanbeseentoclearlyriseafterathermalinertiaof5Jm2s0.5K1whichallowsustoconcludeonthisvalueasbeingtheactualthermalinertiaof(21)Luteta.Fig.3.Thesegraphsshowthewiderangeofobservationsbeingfitagainstthemodelwheretheroughnessisr¼0.6andf¼0.7.andthethermalInertiawas5Jm2s0.5K1.ThegraphsshowObservations/Thermalmodelversus(a)phaseangle(b)wavelength(c)rotationphaseand(d)aspectAngle.L.O’Rourkeetal./PlanetaryandSpaceScience66(2012)192–1991971 5 10 50Thermal Inertia (J m-2 s-0.5 K-1)30 !!!!10 2Reduced Χ2pendence of κ in regoliths is given by Gundlach and Blum (2013). Asteroid TPMs typically neglect the temperature depen- dence of Γ. This is uncritical for typical remote observa- tions, which are dominated by the warm sunlit hemisphere (see Fig. 7 and Capria et al., 2014; Vasavada et al., 2012). In the analysis of highly spatially resolved observations, how- ever, the temperature dependence must be considered, cer- tainly when analyzing night-time observations on low-Γ as- teroids. Note that for temperature-dependent κ, Eq. 2 must be used instead of Eq. 3 (see, e.g., Capria et al., 2014). Fig. 7.— Diurnal temperature curves at the equator of an asteroid with A=0.1, P =6h, =1.0 at a heliocentric distance of 1.0 au. Solid curve: constant thermal conductivity κ=0.02 W K-1 m-1. Dashed- curved: temperature-dependent heat conductivity κ = 10−2(1 + 0.5(T /250)3) W K-1 m-1. Caution must be exercised when comparing thermal- inertia results obtained at different heliocentric distances r, i.e., at different temperatures. All other things being equal, T 4 ∝ r−2. Assuming that the T 3 term dominates in Eq. 12, the thermal inertia of a test object scales with (see also Mueller et al., 2010): Γ ∝ √ κ ∝ T 3/2 ∝ r−3/4. (13) 5.4. Binary Asteroid TPM A rather direct determination of thermal inertia can be obtained by observing the thermal response to eclipses and their aftermath, allowing one to see temperature changes in real time. Such observations have been carried out for plan- etary satellites such as the Galilean satellites (Morrison and Cruikshank, 1973), and our Moon (Pettit, 1940; Shorthill, 1973; Lawson et al., 2003; Lucey, 2000; Fountain et al., 1976). Mueller et al. (2010) report the first thermal obser- vations of eclipse events in a binary Trojan asteroid system, (617) Patroclus, where one component casts shadow on the other while not blocking the line of sight toward the ob- server. The thermophysical modeling of eclipse events is rela- tively straightforward, assuming the system is in a tidally locked rotation state typical of evolved binary systems. In 9 that case, the components’ spin rates match one another, and their spin axes are aligned with that of the mutual or- bit. The system is therefore at rest in a co-rotating frame and can be modeled like a single object with a non-convex (disjoint!) global shape. Eclipse effects are fully captured, provided that shadowing between facets is accounted for. The two hemispheres that face one another can, in princi- ple, exchange heat radiatively. This is negligible for typical binary systems, however. As discussed above, the thermal effects of roughness and thermal inertia can be hard to disentangle. In the case of eclipse measurements, which happen at essentially constant solar phase angle, the effect of surface roughness is much less of a confounding factor. This is because the variation in the thermal signal is dominated by the temperature change induced by the passing shadow, which is a strong function of thermal inertia. It must be kept in mind that the duration of an eclipse event is short compared to the rotation period. The eclipse- induced heat wave therefore probes the subsoil less deeply than the diurnal heat wave does (the typical heat penetration depth is given by Eq. 5 with P equal to the duration of the eclipse event). A depth dependence of thermal inertia (see § 6.6) could manifest itself in different thermal-inertia de- terminations using the two different measurement methods. 5.5. Thermal-infrared interferometry Interferometric observations of asteroids in the ther- mal infrared measure the spatial distribution of the ther- mal emission along different directions on the plane of sky, thereby constraining the distribution in surface temperature and hence thermal inertia and roughness. Provided the as- teroid shape is known, interferometry can be used to break the aforementioned degeneracy between thermal inertia and roughness from a single-epoch observation (Matter et al., 2011, 2013). Interferometry also allows a precise determi- nation of the size of an asteroid (Delbo et al., 2009). Spatial resolutions between 20 and 200 milli-arcseconds can be obtained from the ground (see Durech et al., 2015, for a review and future perspectives of the application of this technique). While for the determination of asteroid sizes and shapes from interferometric observations in the thermal infrared, simple thermal models can be used (Delbo et al., 2009; Carry et al., 2015), a TPM was utilized to calculate interfer- ometric visibilities of asteroids in the thermal infrared for the observations of (41) Daphne (Matter et al., 2011) and (16) Psyche (Matter et al., 2013). Mid-infrared interferometric instruments measure the to- tal flux and the visibility of a source, the latter being re- lated to the intensity of the Fourier Transform (FT) of the spatial flux distribution along the interferometer’s baseline projected on the plane of sky. Thus, the data analysis proce- dure consisted in generating images of the thermal infrared emission of the asteroids at different wavelengths as viewed by the interferometer and then in obtaining the model visi- 180 200 220 240 260 280 300 320 340 360 380 0 1 2 3 4 5 6 7Temperature (K)Time after midnight (h)T-constantT-dependentbility and flux for each image. The former is related to the FT of the image, the latter is simply the sum of the pix- els. The free parameters of the TPM (size, thermal inertia value and roughness) are adjusted in order to minimize the distance between the disk integrated flux and visibility of the model, and the corresponding observed quantities (see Matter et al., 2011, 2013; Durech et al., 2015, for further information). Some results from these observational pro- grams are discussed in § 6 and in the chapter by Durech et al. (2015). 5.6. Disk-resolved data and retrieval of temperatures The availability of disk-resolved thermal-infrared obser- vation has been increasingly steadily over the years: the ESA mission Rosetta performed two successful flybys to the asteroids (2867) Steins (in 2008) and (21) Lutetia (in 2010) (Barucci et al., 2015). In 2011, the NASA mission Dawn began its one-year orbiting of (4) Vesta (Russell et al., 2012); Dawn is reaching (1) Ceres at the time of writing. JAXA’s Hayabusa II sample-return mission will map its tar- get asteroid (162173) 1999 JU3 using the thermal infrared imager (TIR) aboard the spacecraft (Okada et al., 2013); NASA’s OSIRIS-REx mission and its thermal spectrome- ter OTES will do likewise for its target asteroid (101955) Bennu (in 2018-2019). These data can be used to derive surface-temperature maps, from which maps of thermal in- ertia and roughness can be derived. Three different methods are used to measure surface temperatures from orbiting spacecraft: bolometry, mid- infrared spectroscopy, and near-infrared spectroscopy. In the following, we will elaborate on the challenges posed by these different methods, and on their dependence on spec- tral features, surface roughness, illumination geometry, and viewing geometry. Bolometers measure thermal flux within a broad band- pass in the infrared, approximating the integral of the Planck function, U = σT 4 e (e.g., Kieffer et al., 1977; Paige et al., 2010). The temperature derived in this way (effective temperature) is directly relevant to the energy balance on the surface. Since the bolometric flux is spectrally integrated, the resulting temperature is fairly insensitive to spectral emissivity variations, as long as the bolomet- ric emissivity (weighted spectrally averaged emissivity) is known or can be reasonably approximated. Temperatures derived from mid-infrared spectrometry, on the other hand, are typically brightness temperatures, i.e., the temperature of a black body emitting at the wave- length in question. It is generally assumed that at some wavelength, the spectral emissivity is very close to 1.0, and the brightness temperature at this wavelength is taken as the surface temperature. Spacecraft sent to asteroids (and/or comets) have more commonly been instrumented with near-infrared spectrom- eters (e.g., λ < 5 µm) rather than mid-infrared spectrome- ters. The long-wavelength ends of these spectrometers of- ten extend into the range where thermal emission dominates the measured flux (for the daytime surface temperatures of most asteroids). At these wavelengths, one cannot assume that the emissivity is close to 1.0. It is therefore not prac- tical to derive brightness temperatures. Instead, the color temperature is derived, that is the temperature of a black body that emits with the same spectral shape. Such deriva- tions have to separate temperature from spectral emissivity. The problem is under-constrained (N+1 unknowns, but only N data points), so there is no deterministic solution. Spec- tral emissivities for fine-grained silicates trend in the same direction as the blackbody curve, so it would be very easy to mistake spectral emissivity variations for different tem- peratures. The most statistically rigorous approach that has been applied to separating temperature and spectral emis- sivity in the 3 – 5 µm region is that of Keihm et al. (2012) and Tosi et al. (2014) for Rosetta/VIRTIS data of Lutetia and Dawn/VIR data of Vesta. Temperatures thus measured represent an average tem- perature in the field of view of a given pixel: illuminated hot zones and shadowed colder parts will both contribute. They do not directly correspond to a physical temperature of the soil; rather, they depend sensitively on the observation and illumination geometry (see Rozitis and Green, 2011, in par- ticular their Fig. 9), especially in the case of large illumina- tion angles. Microwave spectrometers such as MIRO (Gulkis et al., 2007) can provide both day and nightside thermal flux mea- surements. At sub-mm, mm, and longer wavelengths, aster- oid soils become moderately transparent. Subsurface lay- ers contribute significantly to the observable thermal emis- sion, thus providing information on the subsurface temper- ature. Observable fluxes depend on the subsurface temper- ature profile, weighted by the wavelength-dependent elec- trical skin depths, so both a thermal and an electrical model are required to interpret such data (Keihm et al., 2012). We remind here that thermal infrared fluxes should be used as input data for TPMs and not (effective, color, or brightness) temperatures derived from radiometric meth- ods, because of their dependence on illumination and ob- servation angles ! 5.7. Sample-return missions Space agencies across the planet are developing space missions to asteroids, notably sample-return missions to primitive (C and B type) near-Earth asteroids: Hayabusa- 2, was launched by JAXA towards (162173) 1999 JU3 on December 3, 2014, and OSIRIS-REx is to be launched by NASA in 2016 (Lauretta et al., 2012). A good understand- ing of the expected thermal environment, which is governed by thermal inertia, is a key factor in planning spacecraft op- erations on or near asteroid surfaces. E.g., OSIRIS-REx is constrained to sampling a regolith not hotter than 350 K, severely constraining the choice of the latitude of the sam- ple selection area on the body, the local time, and the arrival date on the asteroid. Both Hayabusa-2 and OSIRIS-REx are required to take 10 regolith samples from the asteroid surface back to Earth. Obviously, this requires that regolith be present in the first place, which needs to be ascertained by means of ground- based thermal-inertia measurements. The sampling mecha- nism of OSIRIS-REx, in particular, requires relatively fine (cm-sized or smaller) regolith. 5.8. Accurate Yarkovsky and YORP modeling from TPMs can be used to place constraints on the internal bulk density distribution of an asteroid (Scheeres and Gaskell, 2008). For instance, Lowry et al. (2014) explain the YORP de- tection on (25143) Itokawa, which was opposite in sign to that predicted, by Itokawa’s two lobes having substantially different bulk densities. However, unaccounted lateral heat conduction in thermal skin depth sized rocks could also ex- plain, at least partially, this opposite sign result (Golubov and Krugly, 2012). Scattered and thermally emitted photons carry momen- tum. Any asymmetry in the distribution of outgoing pho- tons can, after averaging over an orbital period, impart a net recoil force (Yarkovsky effect) and/or a net torque (YORP effect) on the asteroid. Both effects are more noticeable as the object gets smaller. For small enough objects, the orbits can be significantly affected by the Yarkovsky effect, and their rotation state by YORP (Bottke et al., 2006; Vokrouh- lick´y et al., 2015). The strength of the Yarkovsky effect is strongly influ- enced by thermal inertia (Bottke et al., 2006, and references therein) and by the degree of surface roughness (Rozitis and Green, 2012). However, the strength and sign of the YORP rotational acceleration on an asteroid is independent of thermal inertia ( Capek and Vokrouhlick´y, 2004), but it is highly sensitive to the shadowing (Breiter et al., 2009), self-heating (Rozitis and Green, 2013), and surface rough- ness effects (Rozitis and Green, 2013) that are incorporated in thermophysical models. Accurate calculations of the instantaneous recoil forces and torques require an accurate calculation of surface tem- peratures as afforded by TPMs; Rozitis and Green (2012, 2013) report on such models. Other than on thermal in- ertia, the Yarkovsky-induced orbital drift depends on the bulk mass density. Therefore, Yarkovsky measurements combined with thermal-inertia measurements can be used to infer the elusive mass density (Mommert et al., 2014b,a; Rozitis and Green, 2014; Rozitis et al., 2014, 2013). In the case of (101955) Bennu, the uncertainties in published val- ues of thermal inertia (Emery et al., 2014) and measured Yarkovsky drift (Chesley et al., 2014) are so small that the accuracy of the inferred mass density rivals that of the ex- pected in-situ spacecraft result (1260 ± 70 kg m−3, i.e., a nominal uncertainty of only 6%). Rozitis et al. (2014) de- rived the bulk density of (29075) 1950 DA and used it to reveal the presence of cohesive forces stabilizing the object against the centrifugal force. In turn, the measured Yarkovsky drift can be used to infer constraints on thermal inertia. This was first done by Ches- ley et al. (2003) studying the Yarkovsky effect on (6489) Golevka and by Bottke et al. (2001) studying the Koro- nis family in the main asteroid belt; both studies revealed thermal inertias consistent with expectations based on the observed correlation between thermal inertia and diameter (see § 6.2). Whilst the YORP effect is highly sensitive to small-scale uncertainties in an asteroid’s shape model (Statler, 2009), it 6. LATEST RESULTS FROM TPMs In the Asteroid III era, thermal properties were known for only a few asteroids, i.e., (1) Ceres, (2) Pallas, (3) Juno, (4) Vesta, (532) Herculina from Muller and Lagerros (2002), and (433) Eros from Lebofsky and Rieke (1979). Since then, the number of asteroids with known thermal properties has increased steadily. We count 59 minor bodies with known value of Γ (see Tab. 2). Of these, 16 are near- Earth asteroids (NEAs), 27 main-belt asteroids (MBAs), 4 Jupiter Trojans, 5 Centaurs, and 7 trans-Neptunian objects (TNOs). These classes of objects present very different physical properties such as sizes, regolith grain size, average value of the thermal inertia, and composition. Other important dif- ferences are their average surface temperature due to their very different heliocentric distances and orbital elements. The illumination and observation geometry are also diverse for different classes of objects. For instance, for TNOs and MBAs the phase angle of observation from Earth and Earth- like orbits is typically between a few and a few tens of de- grees, respectively. On the other hand, NEAs can be ob- served under a much wider range of phase angles than can approach hundred degrees and more Muller (see also 2002). A special care should be used in these cases to explicitly calculate the heat diffusion in craters instead of using the approximation of Eq. 9. 6.1. Ground truth from space missions to asteroids A number of asteroids have been, or will be, visited by spacecraft, providing ground-truth for the application of TPMs to remote-sensing thermal-infrared data. (21) Lutetia: based on ground-based data and a TPM, Mueller et al. (2006) measured Lutetia’s effective diame- ter and pV to within a few percent of the later Rosetta re- sult (Sierks et al., 2011). Their thermal-inertia constraint (Γ < 50 J m-2s-1/2K-1) was refined by O’Rourke et al. (2012) based on the Rosetta shape model and more than 70 thermal-infrared observations obtained from the ground, Spitzer, Akari, and Herschel: Γ = 5 J m-2s-1/2K-1with a high degree of surface roughness. Keihm et al. (2012) used MIRO aboard Rosetta to obtain a surface thermal inertia (cid:46) 30 J m-2s-1/2K-1. The low thermal inertia can be ex- plained by a surface covered in fine regolith; Gundlach and Blum (2013) infer a regolith grain size of about 200 µm. The study of the morphology of craters by Vincent et al. (2012) indicates abundant, thick (600 m), and very fine re- 11 TABLE 2 PUBLISED THERMAL INERTIA VALUES Number Name 1 2 3 4 16 21 22 32 41 44 45 87 107 110 115 121 130 277 283 306 382 433 532 617 694 720 956 1173 1580 Ceres Pallas Juno Vesta Psyche Lutetia Kalliope Pomona Daphne Nysa Eugenia Sylvia Camilla Lydia Thyra Hermione Elektra Elvira Emma Unitas Dodona Eros Herculina Patroclus Ekard Bohlinia Elisa Anchises Betulia D (km) 923 544 234 525 244 96 167 85 202 81 198 300 245 93.5 92 220 197 38 135 56 75 17.8 203 106 109.5 41 10.4 136 4.57 ∆D (km) 20 43 11 1 25 1 17 1 7 1 20 30 25 3.5 2 22 20 2 14 1 1 1 14 11 1.5 1 0.8 15 0.46 Γ (SI) 10 10 5 20 125 5 125 70 25 120 45 70 25 135 62 30 30 250 105 180 80 150 10 20 120 135 90 50 180 ∆Γ (SI) 10 10 5 15 40 5 125 50 25 40 45 60 10 65 38 25 30 150 100 80 65 50 10 15 20 65 60 20 50 Tax C B S V M M M S Ch E C P P M S Ch Ch S P S M S S P - S - P C r (au) 2.767 2.772 2.671 2.3 2.7 2.8 2.3 2.8 2.1 2.5 2.6 2.7 3.2 2.9 2.5 2.9 2.9 2.6 2.6 2.2 2.6 1.6 2.772 5.9 1.8 2.9 1.8 5.0 1.1 Ref. Number Name 1 1 1 2 3 4 5 6 7 6 5 5 5 6 6 5 5 6 5 6 6 8 1 9 6 6 10 11 8 Geographos 1620 Apollo 1862 Chiron 2060 Chiron 2060 Cebriones 2363 Steins 2867 Steins 2867 8405 Asbolus 3063 Makhaon 10199 Chariklo Chariklo 10199 Itokawa 25143 Itokawa 25143 1950 DA 29075 1998 WT24 33342 50000 Quaoar YORP 54509 2002 AW197 55565 Sedna 90377 Orcus 90482 Apophis 99942 101955 Bennu Bennu 101955 Haumea 136108 1999 JU3 162173 1996 FG3 175706 2003 AZ84 208996 308635 2005 YU55 2008 EV5 341843 D (km) 5.04 1.55 142 218 82 4.92 5.2 66 116 236 248 0.32 0.320 1.30 0.35 1082 0.092 700 995 968 0.375 0.495 0.49 1240 0.87 1.71 480 0.306 0.370 ∆D (km) 0.07 0.07 10 20 5 0.4 1 4 4 12 18 0.03 0.029 0.13 0.04 67 0.010 50 80 63 0.014 0.015 0.02 70 0.03 0.07 20 0.006 0.006 Γ (SI) 340 140 4 5 7 150 210 5 15 1 16 700 700 24 200 6 700 10 0.1 1 600 650 310 0.3 400 120 1.2 575 450 ∆Γ (SI) 120 100 4 5 7 60 30 5 15 1 14 100 200 20 100 4 500 10 0.1 1 300 300 70 0.2 200 50 0.6 225 60 Tax S Q B/Cb B/Cb D E E - D D D S S M E - S - - - Sq B B - C C - C C r (au) 1.1 1.0 8-15 13 5.2 2.1 2.1 7.9 4.7 13 13 1.1 1.1 1.7 1.0 43 1.1 47 87 48 1.05 1.1 1.1 51 1.4 1.4 45 1.0 1.0 Ref. 12 13 14 15 16 17 18 19 16 14 13 8 20 21 8 15 8 22 23 13 24 25 26 27 28 29 27 30 31 References: [1] Muller and Lagerros (1998), [2] Leyrat et al. (2012), [3] Matter et al. (2013), [4] O’Rourke et al. (2012), [5] Marchis et al. (2012), [6] Delbo and Tanga (2009), [7] Matter et al. (2011), [8] Mueller (2007), [9] Mueller et al. (2010), [10] Lim et al. (2011), [11] Horner et al. (2012), [12] Rozitis and Green (2014), [13] Rozitis et al. (2013), [14] Groussin et al. (2004), [15] Fornasier et al. (2013), [16] Fern´andez et al. (2003), [17] Lamy et al. (2008), [18] Leyrat et al. (2011), [19] Fern´andez et al. (2002), [20] Muller et al. (2014a), [21] Rozitis et al. (2014), [22] Cruikshank et al. (2005), [23] P´al et al. (2012), [24] Muller et al. (2014b), [25] Muller et al. (2012), [26] Emery et al. (2014), [27] Lellouch et al. (2013), [28] Muller et al. (2011), [29] Wolters et al. (2011), [30] Muller et al. (2013), [31] Al´ı-Lagoa et al. (2014), Note: for Ceres, Pallas, Juno, and Herculina r is assumed equal to the semimajor asxis of the orbit. 12 golith, confirming the TPM results. (433) Eros was studied by the NASA NEAR-Shoemaker space mission that allowed determination of the shape and size of this asteroid (mean radius of 8.46 km with a mean error of 16 m; Thomas et al., 2002). Mueller (2007) per- formed a TPM analysis of the ground-based thermal in- frared data by Harris and Davies (1999), obtaining a best-fit diameter of 17.8 km that is within 5% of the Thomas et al. (2002) result of 16.9 km, and Γ in the range 100 - 200 J m-2s-1/2K-1. The latter value, in agreement with TPM re- sults of Lebofsky and Rieke (1979), implies coarser surface regolith than that on the Moon and larger asteroids (see, e.g., Mueller, 2007; Delbo et al., 2007). From the value of Γ of Mueller (2007), Gundlach and Blum (2013) calcu- lated a 1-3 mm typical regolith grain size for Eros. Optical images of the NEAR-Schoemaker landing site at a resolu- tion of about 1 cm/pixel (Veverka et al., 2001) show very smooth areas at the scale of the camera spatial resolution (Fig. 8), likely implying mm or sub-mm grain size regolith, consistent with TPM results. (25143) Itokawa physical properties were derived in- situ by the JAXA sample-return mission Hayabusa, allow- ing us to compare the size, albedo and regolith nature de- rived from the TPMs with spacecraft results. Muller et al. (2014a) show an agreement within 2% between the size and the geometric visible albedo inferred from TPM analysis of thermal-infrared data and the value of the corresponding parameters from Hayabusa data. The TPM thermal inertia value for Itokawa is around 750 J m-2s-1/2K-1, significantly higher than the value of our Moon (about 50 J m-2s-1/2K-1) and of other large main belt asteroids including (21) Lutetia, implying a coarser regolith on this small NEA. The corre- sponding average regolith grain size according to Gundlach and Blum (2013) is ∼ 2 cm. Hayabusa observations from the optical navigation camera (ONC-T), obtained during the descent of the spacecraft to the “Muses Sea” region of the asteroid, reveal similar grain sizes, at a spatial resolution of up to 6 mm/pixel. In particular, Yano et al. (2006) describe “Muses Sea” as composed of numerous size-sorted granular materials ranging from several centimeters to subcentime- ter scales. Itokawa’s regolith material can be classified as ”gravel”, larger than submillimeter regolith powders filling in ponds on (433) Eros (Fig. 8). It is worth pointing out, however, that “Muses Sea” is not representative of Itokawa’s surface as a whole. Rather, it was selected as a touchdown site because, in earlier Hayabusa imaging, it appeared as particularly smooth (min- imizing operational danger for the spacecraft upon touch- down) and apparently regolith rich (maximizing the chance of sampling regolith). Grain sizes measured at “Muses Sea” are therefore lower limits on typical grain sizes rather than values typical for the surface as a whole. 6.2. Thermal inertia of large and small asteroids An inverse correlation between Γ and D was noticed by Delbo et al. (2007), then updated Delbo and Tanga (2009) 13 Fig. 8.— Higher Γ-values correspond to coarser regoliths. (A) Close-up image of (433) Eros from the NASA NEAR Shoemaker mission reveals coarse regolith with grain size in the mm-range (adapted from Veverka et al., 2001). The value of Γ is ∼150 J m-2s-1/2K-1 for Eros. (B) Image from the JAXA Hayabusa mis- sion (from Yano et al., 2006) of the surface of (25143) Itokawa dis- playing gravel-like regolith. The value of Γ is ∼750 J m-2s-1/2K-1 for Itokawa. and Capria et al. (2014). This supported the intuitive view that large asteroids have, over many hundreds of millions of years, developed substantial insulating regolith layers, re- sponsible for the low values of their surface thermal inertia. On the other hand, much smaller bodies, with shorter col- lisional lifetimes (Marchi et al., 2006; Bottke et al., 2005, and references therein), have less regolith, and or larger re- golith grains (less mature regolith), and therefore display a larger thermal inertia. In the light of the recently published values of Γ (Tab. 2), said inversion correlation between Γ and D is less clear, in particular, when the values of the thermal inertia are tem- perature corrected (Fig. 9). However, the Γ vs D distribu- tion of D > 100 km (large) asteroids is different than that of D < 100 km (small) asteroids. Small asteroids typically have higher Γ-values than large asteroids, which present a large scatter of Γ-values, ranging from a few to a few hun- dreds J m-2s-1/2K-1. This is a clear indication of a diverse regolith nature amongst these large bodies. A shortage of low Γ values for small asteroids is also clear, with the no- table exception of 1950 DA, which has an anomalously low Γ-value compared to other NEAs of similar size (Rozitis et al., 2014). Fig. 9 also shows previously unnoticed high-thermal- inertia C types, maybe related to CR carbonaceous chon- drites, which contain abundant metal phases. We also note that all E types in our sample appear to have a size- independent thermal inertia. 6.3. Very low Γ-values We also note that the some of the C-complex outer main-belt asteroids and Jupiter Trojans have very low ther- mal inertia in the range between a few and a few tens of J m-2s-1/2K-1. In order to reduce the thermal inertia of a ma- terial by at least one order of magnitude (from the lowest measured thermal inertia of a meteorite, ∼650 J m-2s-1/2K-1 at 200 K (Opeil et al., 2010), to the typical values for these Received18June;accepted6August2001.1.Farquhar,R.W.etal.NEARmissionoverviewandtrajectorydesign.J.Astronaut.Sci.43,353±372(1999).2.Veverka,J.etal.NEARatEros:Imagingandspectralresults.Science289,2088±2097(2000).3.Zuber,M.T.etal.Theshapeof433ErosfromtheNEAR-Shoemakerlaserrange®nder.Science289,2097±2101(2000).4.Thomas,P.C.,Veverka,J.,Robinson,M.S.&Murchie,S.Shoemakercraterasthesourceofmostejectablocksontheasteroid433Eros.Nature413,394±396(2001).5.Robinson,M.S.,Thomas,P.C.,Veverka,J.,Murchie,S.&Carcich,B.ThenatureofpondeddepositsonEros.Nature413,396±400(2001).6.Veverka,J.etal.Imagingofsmall-scalefeatureson433ErosfromNEAR:Evidenceforacomplexregolith.Science292,484±488(2001).7.Wolfe,E.M.&Bailey,N.G.LineamentsoftheApennineFront-Apollo15landingsite.Proc.LunarPlanet.Sci.Conf.3,15±25(1972).8.Sullivan,R.etal.Geologyof243Ida.Icarus120,119±139(1996).9.Lee,P.etal.Ejectablockson243Idaandonotherasteroids.Icarus120,87±105(1996).10.Thomas,P.C.etal.Phobos:RegolithandejectablocksinvestigatedwithMarsorbitercameraimages.J.Geophys.Res.105,15091±15106(2000).11.Thomas,P.C.&Veverka,J.DownslopemovementofmaterialonDeimos.Icarus42,234±250(1980).12.Chapman,C.R.etal.ImpacthistoryofEros:cratersandboulders.Icarus(inthepress).13.Hawkins,S.E.etal.MultispectralimagerontheNearEarthAsteroidRendezvousmission.SpaceSci.Rev.82,30±100(1997).14.Veverka,J.etal.AnoverviewoftheNEARmultispectralimager-near-infraredspectrometerinvestigation.J.Geophys.Res.102,23709±23727(1997).AcknowledgementsWethankthemissiondesign,missionoperations,andspacecraftteamsoftheNEARProjectattheAppliedPhysicsLaboratoryofJohnsHopkinsUniversityandthenavigationteamattheJetPropulsionLaboratoryfortheireffortsthatmadeNEARthe®rstorbitertomakeasuccessfullanding.WearegratefulforthehelpfulreviewsprovidedbyA.RivkinandM.Cintala.CorrespondenceandrequestsformaterialsshouldbeaddressedtoJ.V.(e-mail:[email protected]).letterstonatureNATUREVOL41327SEPTEMBER2001www.nature.com393'Pond''Moat'cdab2 mcdaABbFigure4Geologicalinterpretationofthelandingsite.A,NEAR'slasttwoimages:detailsofpondwithcollapsefeatures.Theseframeshaveanaverageresolutionof1.4cm.Twoenclosedcollapsepitsareindicatedbyarrows(aandb)inthesmoothponddeposit.Arrowcpointstoamoat-likedepressionaroundametre-sizedboulder.Arrowdmarksasubdued,somewhatsinuous,channel-likedepressionwhichtrendsinthegeneraldirectionoftheponddeposit.MosaicofframesMET157417178and-98.Rangeandresolutionsare167m(1.6cmperpixel)and129m(1.2cmperpixel),respectively.Scalebar,2m.B,SketchmapofA.Blocksareoutlinedwithcross®lling.Boundarybetweenpondandroughersurfaceismarkedbycrossedline.Twosinuousdepressionsareshownbyheavy,solidlines(d).Collapsefeaturesareoutlinedwithinthepond(aandb).© 2001 Macmillan Magazines LtdREPORTTouchdownoftheHayabusaSpacecraftattheMusesSeaonItokawaHajimeYano,1T.Kubota,1H.Miyamoto,2T.Okada,1D.Scheeres,3Y.Takagi,4K.Yoshida,5M.Abe,1S.Abe,6O.Barnouin-Jha,7A.Fujiwara,1S.Hasegawa,1T.Hashimoto,1M.Ishiguro,8M.Kato,1J.Kawaguchi,1T.Mukai,6J.Saito,1S.Sasaki,9M.Yoshikawa1Afterglobalobservationsofasteroid25143ItokawabytheHayabusaspacecraft,weselectedthesmoothterrainoftheMusesSeafortwotouchdownscarriedouton19and25November2005UTCforthefirstasteroidsamplecollectionwithanimpactsamplingmechanism.Here,wereportinitialfindingsaboutgeologicalfeatures,surfacecondition,regolithgrainsize,compositionalvariation,andconstraintsonthephysicalpropertiesofthissitebyusingbothscientificandhousekeepingdataduringthedescentsequenceofthefirsttouchdown.Close-upimagesrevealedthefirsttouchdownsiteasaregolithfielddenselyfilledwithsize-sorted,millimeter-tocentimeter-sizedgrains.Themostchallengingengineeringdem-onstration,aswellasthemostimpor-tantscientificgoaloftheHayabusaspacecraft(originallycalledMUSES-C)isthesamplingofsurfacematerialsoftheApollo-type,near-Earthasteroid25143Itokawa(previ-ously1998SF36).Tomaximizescientificpromisesoflaboratoryanalysesofthereturnedsamples,itisnecessarytocharacterizephysicalandgeologicalcontextsofsamplingsitesasmuchaspossiblebyusingbothonboardscienceinstrumentsandhousekeepingdataofthespacecraft.TheHayabusaspacecraftarrivedattheasteroidhoveringata20-kmaltitude(gatepo-sition)on12September2005UTC(1,2).Ataltitudesof7toÈ20kmaboveItokawa_ssurface,Hayabusaspent6weeksperformingglobalremote-sensingmeasurements(3–5)thatrevealedacleardichotomybetweenboulder-richroughterrainsandlow-potentialsmoothterrainsoftheasteroid.ShapemodelsshowthatItokawais550mby298mby224minitscircumscribedboxsize(2).Aftercompletionofthescientificobserva-tionphase,theHayabusateamchosetwosampling-sitecandidatesonthebasisofscien-tificmerits,judgedmainlyfromopticalimagesandlightdetectionandranging(LIDAR)topography,aswellastechnicalconstraintssuchasguidance-navigation-control(GNC)accuracyandoperationalsafetyduringthetouchdownsequence(Fig.1A).Thesesiteswere(i)thelargestsmoothterrainareaintheMusesSea,apartoftheadjacentBneck[regionbetweentheBhead[andBbody[parts,whichisaswideasÈ60mfromtheheadtothebodyandÈ100mfromnorthtosouthand(ii)thelargestfacetoftheroughterrainofthebodycalledLittleWoomera(2,3).Bothareasareinthelocaldaysideoftheequatorialregionduringthereal-timetelemetrycoveragefromgroundstations.Theyalsohaverelativelyflatplainswithfewobstaclesaslargeasthespacecraftitself(6)andshowshallowlocalsurfaceincli-nationssuchthatbothhighsolar-powerpro-ductionandbroadtelecommunicationtoEarthareavailableduringalltouchdownsequencesatasolarangleofÈ10-.Theoperationteamperformedtwotouch-downrehearsalson4and12Novemberandtwoimagingnavigationtestson9November.Highspatialresolutionimagesofbothcan-didatesitesalsowereacquiredfromaltitudes1DepartmentofPlanetaryScience,InstituteofSpaceandAstronauticalScience,JapanAerospaceExplorationAgency,3-1-1Yoshinodai,Sagamihara,Kanagawa229-8510Japan.2DepartmentofGeosystemEngineering,UniversityofTokyo,Hongo,Tokyo,113-8656Japan.3UniversityofMichigan,AnnArbor,MI48109–2140,USA.4TohoGakuenUniversity,3-11Heiwagaoka,Meito,Nagoya,465-8515Japan.5DepartmentofAerospaceEngineering,TohokuUniversity,Sendai,Miyagi,980-8579Japan.6DepartmentofEarthandPlanetarySci-ence,KobeUniversity,Kobe,Hyogo,657-8501Japan.7Ap-pliedPhysicsLaboratory,JohnsHopkinsUniversity,Laurel,MD20723,USA.8SeoulNationalUniversity,Seoul,151-742Korea.9NationalAstronomicalObservatoryofJapan,Mizusawa,Iwate,023-0861Japan.Towhomcorrespondenceshouldbeaddressed:E-mail:[email protected],includingthefirsttouchdownsiteonItokawa.Allimagesweretakeninv-band(3).Thesquarein(A)indicatesthesizeof(B);therectanglein(B)indicatesthesizeof(C);therectanglein(C)indicatesthesizeof(D).Scalebarsin(C)and(D),1m.(A)Itokawais550mby298mby224minitscircumscribedboxsize(2).(B)Takenbythewide-fieldopticalnavigationcamera(ONC-W)fromÈ32-maltitudeat20:33UTC.ThecirclenexttoHayabusa’sshadowshowsthetargetmarkerthatlandedontheMusesSeaatTD1.(C)Acompositeofthreeclose-upimagesofST2563511720,ST2563537820,andST2563607030,whichweretakenfrom80-m,68-m,and63-maltitudes,respectively,accordingtoLIDARmeasurements.Thespatialresolutionsare0.8toÈ0.6cm/pixel.Contrastsin(C)arearbitraryandstretchedtomaketheapparentbrightnessofthethreeimagescontinuous,whereasthegray-scalebrightnessof(D)isstretchedaboutfivetimesthatoftheoriginalimage.TheMusesSeaiscomposedofnumerous,size-sortedgranularmaterialsrangingfromseveralcentimeterstosubcentimeterscales.Rockslargerthantensofcentimetersinsizeoftenexhibitbrighterand/ordarkerspotsontheirsurfacesthandosmallerregolithgrains.HAYABUSAATASTEROIDITOKAWA2JUNE2006VOL312SCIENCEwww.sciencemag.org1350 on April 3, 2009 www.sciencemag.orgDownloaded from Received18June;accepted6August2001.1.Farquhar,R.W.etal.NEARmissionoverviewandtrajectorydesign.J.Astronaut.Sci.43,353±372(1999).2.Veverka,J.etal.NEARatEros:Imagingandspectralresults.Science289,2088±2097(2000).3.Zuber,M.T.etal.Theshapeof433ErosfromtheNEAR-Shoemakerlaserrange®nder.Science289,2097±2101(2000).4.Thomas,P.C.,Veverka,J.,Robinson,M.S.&Murchie,S.Shoemakercraterasthesourceofmostejectablocksontheasteroid433Eros.Nature413,394±396(2001).5.Robinson,M.S.,Thomas,P.C.,Veverka,J.,Murchie,S.&Carcich,B.ThenatureofpondeddepositsonEros.Nature413,396±400(2001).6.Veverka,J.etal.Imagingofsmall-scalefeatureson433ErosfromNEAR:Evidenceforacomplexregolith.Science292,484±488(2001).7.Wolfe,E.M.&Bailey,N.G.LineamentsoftheApennineFront-Apollo15landingsite.Proc.LunarPlanet.Sci.Conf.3,15±25(1972).8.Sullivan,R.etal.Geologyof243Ida.Icarus120,119±139(1996).9.Lee,P.etal.Ejectablockson243Idaandonotherasteroids.Icarus120,87±105(1996).10.Thomas,P.C.etal.Phobos:RegolithandejectablocksinvestigatedwithMarsorbitercameraimages.J.Geophys.Res.105,15091±15106(2000).11.Thomas,P.C.&Veverka,J.DownslopemovementofmaterialonDeimos.Icarus42,234±250(1980).12.Chapman,C.R.etal.ImpacthistoryofEros:cratersandboulders.Icarus(inthepress).13.Hawkins,S.E.etal.MultispectralimagerontheNearEarthAsteroidRendezvousmission.SpaceSci.Rev.82,30±100(1997).14.Veverka,J.etal.AnoverviewoftheNEARmultispectralimager-near-infraredspectrometerinvestigation.J.Geophys.Res.102,23709±23727(1997).AcknowledgementsWethankthemissiondesign,missionoperations,andspacecraftteamsoftheNEARProjectattheAppliedPhysicsLaboratoryofJohnsHopkinsUniversityandthenavigationteamattheJetPropulsionLaboratoryfortheireffortsthatmadeNEARthe®rstorbitertomakeasuccessfullanding.WearegratefulforthehelpfulreviewsprovidedbyA.RivkinandM.Cintala.CorrespondenceandrequestsformaterialsshouldbeaddressedtoJ.V.(e-mail:[email protected]).letterstonatureNATUREVOL41327SEPTEMBER2001www.nature.com393'Pond''Moat'cdab2 mcdaABbFigure4Geologicalinterpretationofthelandingsite.A,NEAR'slasttwoimages:detailsofpondwithcollapsefeatures.Theseframeshaveanaverageresolutionof1.4cm.Twoenclosedcollapsepitsareindicatedbyarrows(aandb)inthesmoothponddeposit.Arrowcpointstoamoat-likedepressionaroundametre-sizedboulder.Arrowdmarksasubdued,somewhatsinuous,channel-likedepressionwhichtrendsinthegeneraldirectionoftheponddeposit.MosaicofframesMET157417178and-98.Rangeandresolutionsare167m(1.6cmperpixel)and129m(1.2cmperpixel),respectively.Scalebar,2m.B,SketchmapofA.Blocksareoutlinedwithcross®lling.Boundarybetweenpondandroughersurfaceismarkedbycrossedline.Twosinuousdepressionsareshownbyheavy,solidlines(d).Collapsefeaturesareoutlinedwithinthepond(aandb).© 2001 Macmillan Magazines Ltd1m1mBA6.4. Average thermal inertia of asteroid populations As described before, the thermal inertia of an aster- oid can be directly derived by comparing measurements of its thermal-infrared emission to model fluxes generated by means of a TPM. Typically, more than one observation epoch is required to derive the thermal inertia, in order to ”see” the thermal emission from different parts of the as- teroid’s diurnal temperature distribution. Unfortunately, the large majority of minor bodies for which we have thermal- infrared observations have been observed at a single epoch and/or information about their gross shape and pole orienta- tion is not available, precluding the use of TPMs. However, if one assumes the thermal inertia to be roughly constant within a population of asteroids (e.g., NEAs) one can use observations of different asteroids under non-identical illu- mination and viewing geometries, as if they were from a unique object. Delbo et al. (2003) noted that qualitative in- formation about the average thermal properties of a sample of NEAs could be obtained from the distribution of the η- values of the sample as a function of the phase angle, α. Delbo et al. (2007) and Lellouch et al. (2013) developed a rigorous statistical inversion method, based on the com- parison of the distributions of published NEATM η-values vs α, or vs. r with that of a synthetic population of aster- oids generated through a TPM, using realistic distributions of the input TPM parameters such as the rotation period, the aspect angle etc. Delbo et al. (2007) found that the av- erage thermal inertia value for km-sized NEAs is around 200 J m-2s-1/2K-1. The average thermal inertia of binary NEAs is higher than that of non-binary NEAs, possibly in- dicating a regolith-depriving mechanism for the formation of these bodies (Delbo et al., 2011). The same authors also found that NEAs with slow rotational periods (P >10 h) have higher-than-average thermal inertia. From a sample of 85 Centaurs and trans-Neptunian objects observed with Spitzer/MIPS and Herschel/PACS, Lellouch et al. (2013) found that surface roughness is significant, a mean ther- mal inertia Γ = 2.5 ± 0.5 J m-2s-1/2K-1, and a trend to- ward decreasing Γ with increasing heliocentric distance. The thermal inertias derived by Lellouch et al. (2013) are 2-3 orders of magnitude lower than expected for compact ices, and generally lower than on Saturn’s satellites or in the Pluto/Charon system. These results are suggestive of highly porous surfaces. 6.5. Relevant astronomical and laboratory data Physical interpretations of thermal-inertia estimates de- pend strongly on laboratory and ground-truth measure- ments of relevant material properties. While in the Asteroid III era, we based interpretation of thermal inertia on Earth analog materials, in the last few years laboratory measure- ments were performed on asteroid analog materials, i.e., meteorites. Meteorite grain densities range from ∼2800 kg m-3 for CM carbonaceous chondrites to ∼3700 kg m-3 for enstatite chondrites (Consolmagno et al., 2006; Macke et al., 2010, 2011a,b). Heat capacities have been measured Fig. 9.— Γ values vs. D from Tab. 2 for different taxonomic types (see key). Top plot: original measurements, bottom plot: Γ corrected to 1 au heliocentric distance for temperature dependent thermal inertia assuming Eq. 13 and the heliocentric distance at the time of thermal infrared observations reported in Tab. 2. Trojans, Centaurs and trans-Neptunian objects are not displayed. large asteroids (Tab. 2 and Fig. 9), a very large porosity (>90%) of the first few mm of the regolith is required (Vernazza et al., 2012). This is consistent with the dis- covery that emission features in the mid-infrared domain (7–25 µm, Fig. 4) are rather universal among large aster- oids and Jupiter Trojans (Vernazza et al., 2012), and that said features can be reproduced in the laboratory by sus- pending meteorite and/or mineral powder (with grain sizes < 30 µm) in IR-transparent KBr (potassium bromide) pow- der (Vernazza et al., 2012). As KBr is not supposed to be present on the surfaces of these minor bodies, regolith grains must be ”suspended” in void space likely due to co- hesive forces and/or dust levitation. On the other hand, radar data indicate a significant porosity (40-50 %) of the first ∼1 m of regolith (Magri et al., 2001; Vernazza et al., 2012), indicating decreasing porosity with increasing depth (see Fig. 5 of Vernazza et al., 2012, for a regolith schemat- ics). 14 10-1100101102103Diameter (km)101102103Thermal Inertia (SI)N/AVC,B,D,PSME10-1100101102103Diameter (km)101102103Thermal Inertia (SI)N/AVC,B,D,PSMEfor a wide sampling of meteorites by Consolmagno et al. (2013), who find that values for stony meteorites are be- tween 450 and 550 J kg-2 K-1, whereas C for irons tends to be smaller (330 – 380 J kg-2 K-1). Opeil et al. (2012, 2010) present thermal conductivity measurements of stony meteorites, finding values of 0.5 W K-1 m-1 for the car- bonaceous chondrite Cold Bokkeveld to 5.5 W K-1 m-1 for the enstatite chondrite Pillistfer. Their one iron meteorite sample has a κ of 22.4 W K-1 m-1. They also find a linear correlation between and the inverse of the porosity, from which Opeil et al. (2012) conclude that the measured κ of the samples is controlled more by micro-fractures than by composition. Grain size and packing, more than compositional het- erogeneity, are responsible for different thermal inertias of different surfaces. This also explains why TPMs are capa- ble of deriving asteroid physical parameters independently of the asteroid mineralogy. Conduction between grains is limited by the area of the grain contact (Piqueux and Chris- tensen, 2009b,a). As grain size decreases to diameters less than about a thermal skin depth (few cm on most aster- oids), conduction is more and more limited (e.g., Presley and Christensen, 1997b). On bodies with atmospheres, con- duction through the air in pores can often efficiently trans- port heat. On airless bodies, however, radiation between grains, which is not very efficient, particularly at low T (e.g., Gundlach and Blum, 2012), is the only alternative to conduction across contacts (Fig. 10). Considering these two modes of energy transport and their dependence on grain size, Gundlach and Blum (2013) developed an analytical approach for determining grain size from thermal inertia measurements. They incorporated the measurements of ma- terial properties of meteorites measured above along with results of their own laboratory of heat transport in dusty layers. Additional laboratory measurements of conductiv- ities of powdered meteorites under high vacuum would be valuable for more precise interpretation of asteroid thermal inertias. The classic opportunity for ground-truth thermal mea- surements came with the Apollo missions. Astronauts on Apollo 15 and 17 carried out bore-hole style temper- ature measurements to depths of 1.4 m below the surface on Apollo 15 and 2.3 m below the surface on Apollo 17 (Langeseth and Keihm, 1977; Vaniman et al., 1991). Ther- mal conductivity of about 0.001 W K-1 m-1 was found in the top 2 to 3 cm of the lunar regolith, increasing to about 0.01 W K-1 m-1 over the next few cm, then to values as high as 2 W K-1 m-1 deeper into the surface where the regolith ap- pears to have been very compacted (Langeseth and Keihm, 1977). Low thermal inertias derived from remote thermal infrared measurements (e.g. Wesselink, 1948b; Vasavada et al., 2012) agree with the very low κ in the topmost few cm of the lunar surface, and the Apollo measurements provide the necessary ground-truth for interpreting such low ther- mal inertias as very fine-grained, ”fluffy” regolith. These measurements fostered, for instance, development of de- tailed models of lunar regolith (Keihm, 1984). Detailed Fig. 10.— Diagram of the modes of heat transport in regoliths. On airless bodies, heat can flow by conduction through grain boundaries (solid line) or by radiation between grains (dashed line). The dotted line showing transport by gas diffusion is not relevant to asteroid surfaces. From Gundlach and Blum (2012). thermal infrared observations and thermal models of the lunar regolith allows today estimating the subsurface rock abundance (e.g., Bandfield et al., 2011), allowing geologi- cal studies of the regolith production rate. 6.6. Dependence of Γ with depth The depth dependence of typical asteroid regolith prop- erties is poorly constrained at this point, which is why physical constants are typically assumed to be constant with depth. MIRO observations of (21) Lutetia, how- ever, showed the existence of a top layer with Γ < 30 J m-2s-1/2K-1, while the thermal inertia of subsurface mate- rial appears to increase with depth much like on the Moon (Keihm et al., 2012). 6.7. Infrared limb brightening Recent modeling and observations show that, contrary to expectation, the flux enhancement measured in disk- integrated observations of the sunlit side of an asteroid (e.g., Lebofsky et al., 1986) is dominated by limb surfaces rather than the subsolar region (Rozitis and Green, 2011; Keihm et al., 2012). This suggests that for the sunlit side of an asteroid, sunlit surfaces directly facing the observer in situ- ations where they would not be if the surface was a smooth flat one are more important than mutual self-heating be- tween interfacing facets raising their temperatures. Figure 9 of Rozitis and Green (2011) pictures this effect for a Gaus- sian random surface during sunrise viewed from different directions. The thermal flux observed is enhanced when viewing hot sunlit surfaces (i.e., Sun behind the observer), and is reduced when viewing cold shadowed surfaces (i.e., Sun in front of the observer). Jakosky et al. (1990) also studied the directional thermal emission of Earth-based lava flows exhibiting macroscopic 15 Figure XX. Diagram of the modes of heat transport in regoliths. On airless bodies, heat can flow by conduction through grain boundaries (solid line) or by radiation between grains (dashed line). The dotted line showing transport by gas diffusion is not relevant to asteroid surfaces. (From Gundlach and Blum 2012). Consolmagno, G. J., Macke, R. J., Rochette, P., Britt, D. T., Gattacceca, J. 2006. Density, magnetic susceptibility, and the characterization of ordinary chondrite falls and showers. Meteor. & Planet. Sci. 41, 331-342. Consolmagno, G. J., Schaefer, M. W., Schaefer, B. E., Britt, D. T., Macke, R. J., Nolan, M. C., Howell, E. S. 2013. The measurement of meteorite heat capacity at low temperatures using liquid nitrogen vaporization. Planet. Space Sci. 87, 146-156. Gundlach, B. and Blum, J. 2012. Outgassing of icy bodies in the Solar System – II: Heat transport in dry, porous surface dust layers. Icarus 219, 618-629. Langseth, M.G. and S.J. Keihm 1977. Lunar Heat-Flow Experiment. NASA Technical Report NASA-CR-151619; CU-4-77. Macke, R. J., Consolmagno, G. J., Britt, D. T., Hutson, M. L. 2010. Enstatite chondrite density, magnetic susceptibility, and porosity. Meteor. & Planet. Sci. 45, 1513-1526. Macke, R. J., Consolmagno, G. J., Britt, D. T. 2011a. Density, porosity, and magnetic susceptibility of achondritic meteorites. Meteor. & Planet. Sci. 46, 311-326. Macke, R. J., Consolmagno, G. J., Britt, D. T. 2011b. Density, porosity, and magnetic susceptibility of carbonaceous chondrites. Meteor. & Planet. Sci. 46, 1842-1862. Piqueux, S. and Christensen, P. R. 2009. A model of thermal conductivity for planetary soils: 1. Theory for unconsolidated soils. J. Geophys. Res. 114, E09005, 20pp. Presley, M. A. and Christensen, P. R. 1997. Thermal conductivity measurements of particulate materials 2. Results. J. Geophys. Res. 102 (E3), 6551-6566. roughness. They found that enhancements in thermal emis- sion were caused by viewing hot sunlit sides of rocks and reductions were caused by viewing cold shadowed sides of rocks. This agrees precisely with the model and adds fur- ther evidence that thermal infrared beaming is caused by macroscopic roughness rather than microscopic roughness. The effect of limb brightening has also been measured from disk-resolved thermal infrared data (<5 µm) acquired during sunrise on the nucleus of the comet 9P/Tempel 1 by the Deep Impact NASA space mission (Davidsson et al., 2013), and from VIRTIS and MIRO measurements of the asteroid (21) Lutetia (Keihm et al., 2012). 6.8. Asteroid thermal inertia maps Disk-resolved thermal infrared observations, in the range between 4.5 – 5.1 µm, were provided by the instrument VIR (De Sanctis et al., 2012) on board of the NASA DAWN (Russell et al., 2012) spacecraft (Capria et al., 2014, and references therein). Form TPM analysis of VIR measure- ments, Capria et al. (2014) obtained a map of the roughness and the thermal inertia of Vesta. The average thermal in- ertia of Vesta is 30 ± 10 J m-2s-1/2K-1, which is in good agreement with the values found by ground-based observa- tions (Muller and Lagerros, 1998; Chamberlain et al., 2007; Leyrat et al., 2012). The best analog is probably the sur- face of the Moon, as depicted by Vasavada et al. (2012) and Bandfield et al. (2011): a surface whose thermal response is determined by a widespread layer of dust and regolith with different grain sizes and density increasing toward the in- terior. Exposed rocks are probably scarce or even absent. Capria et al. (2014) also show that Vesta cannot be consid- ered uniform from the point of view of thermal properties. In particular, they found that the thermal inertia spatial dis- tribution follows the global surface exposure age distribu- tion, as determined by crater counting in Raymond et al. (2011), with higher thermal inertia displayed by younger terrains and lower thermal inertia in older soils. Capria et al. (2014) also found higher-than-average ther- mal inertia terrain units located in low-albedo regions that contain highest abundance of OH, as determined by the 2.8 µm band depth (De Sanctis et al., 2012). These terrains are associated with the dark material, thought to be delivered by carbonaceous chondrite like asteroids that have impacted Vesta at low velocity. Note that in general (carbonaceous chondrites) have lower densities and lower thermal conduc- tivity (Opeil et al., 2010) than basaltic material, which con- stitute the average Vestan terrain. This consideration would point to a lower thermal inertia rather than a higher one, as observed on Vesta. Capria et al. (2014) conclude that the factor controlling the thermal inertia in these areas could be the degree of compaction of the uppermost surface layers, which is higher than in other parts of the surface. 6.9. Thermal inertia of metal-rich regoliths In principle, the composition of the regolith and not only its average grain size and the degree of compaction 16 also affects the thermal inertia of the soil (Gundlach and Blum, 2013). For instance iron meteorites have a higher thermal conductivity than ordinary and carbonaceous chon- drites (Opeil et al., 2010). We thus expect that a metal iron rich regolith displays a higher thermal inertia than a soil poor of this component. Harris and Drube (2014) compared values of the NEATM η-parameter derived from WISE data with asteroid taxonomic classifications and radar data, and showed that the η-value appears to be a useful indicator of asteroids containing metal. Matter et al. (2013) performed interferometric observations with MIDI of the ESO-VLTI in thermal infrared of (16) Psyche and showed that Psy- che has a low surface roughness and a thermal inertia value around 120 ± 40 J m-2s-1/2K-1, which is one of the higher values for an asteroid of the size of Psyche (∼ 200 km). This higher than average thermal inertia supports the evi- dence of a metal-rich surface for this body. 7. EFFECTS OF TEMPERATURES ON THE SUR- FACE OF ASTEROIDS 7.1. Thermal cracking The surface temperature of asteroids follows a diur- nal cycle (see Fig. 2) with typically dramatic tempera- ture changes as the Sun rises or sets. The resulting, re- peated thermal stress can produce cumulative damage on surface material due to opening and extension of micro- scopic cracks. This phenomenon is known as thermal fa- tigue (Delbo et al., 2014). Growing cracks can lead to rock break-up when the num- ber of temperature cycles is large enough. For typical aster- oid properties, this process is a very effective mechanism for comminuting rocks and to form fresh regolith (Delbo et al., 2014). For cm-sized rocks on an asteroid 1 au from the Sun, thermal fragmentation is at least an order of magni- tude faster than comminution by micrometeoroid impacts, the only regolith-production mechanism previously consid- ered relevant (Horz and Cintala, 1997; Hoerz et al., 1975). The efficiency of thermal fragmentation is dominated by the amplitude of the temperature cycles and by the temper- ature change rate (Hall and Andr´e, 2001), which in turn de- pend on heliocentric distance, rotation period, and the sur- face thermal inertia. The rate of thermal fragmentation in- creases with decreasing perihelion distance: at 0.14 au from the Sun, thermal fragmentation may erode asteroids such as (3200) Phaethon and produce the Geminids (Jewitt and Li, 2010), whereas in the outer Main Belt this process might be irrelevant. Thermal fragmentation of surface boulders is claimed by Dombard et al. (2010) to be source of fine re- golith in the so-called ”ponds” on the asteroid (433) Eros. Production of fresh regolith originating in thermal fatigue fragmentation may be an important process for the rejuve- nation of the surfaces of near-Earth asteroids (Delbo et al., 2014). Thermal cracking is reported on other bodies, too: on Earth, particularly in super-arid environments (Hall, 1999; Hall and Andr´e, 2001), on the Moon (Levi, 1973; Duen- nebier and Sutton, 1974), Mercury (Molaro and Byrne, 2012), Mars (Viles et al., 2010), and on meteorites (Levi, 1973). Moreover, Tambovtseva and Shestakova (1999) sug- gest that thermal cracking could be an important process in the fragmentation and splitting of kilometer-sized comets while in the inner solar system. Furthermore, Capek and Vokrouhlick´y (2010) initially proposed that slowly rotating meteoroids or meteoroids that have spin vector pointing to- wards the sun can be broken up by thermal cracking. In a further development of their model, Capek and Vokrouh- lick´y (2012) showed that as the meteoroid approaches the Sun, the stresses first exceed the material strength at the sur- face and create a fractured layer. If inter-molecular forces (e.g., Rozitis et al., 2014) are able to retain the surface layer, despite the competing effects of thermal lifting and centrifu- gal forces, the particulate surface layer is able to thermally shield the core, preventing any further damage by thermal stresses. 7.2. Sun-driven heating of near-Earth asteroids and meteoroids It is known that heating processes can affect the physical properties of asteroids and their fragments, the meteorites (see, e.g., Keil, 2000). Internal heating due to the decay of short-lived radionu- clides was considered early on (Grimm and McSween, 1993). Marchi et al. (2009) discuss close approaches to the Sun as an additional surface-altering heating mechanism. In the present near-Earth asteroid population, the fraction of bodies with relatively small perihelion (q) is very small: about 1/2, 1/10, and 1/100 of the population of currently known near-Earth objects (11,000 as of the time of writ- ing) have a perihelion distance below 1, 0.5, and 0.25 au, where maximum temperature are exceeding 400, 550, and 780 K, respectively (see Fig. 11). However, dynamical sim- ulations show that a much larger fraction of asteroids had small perihelion distances for some time, hence experienc- ing episodes of strong heating in their past (Marchi et al., 2009). For instance, the asteroid 2004 LG was approaching the Sun to within only ∼5.6 solar radii some 3 ky ago, and its surface was baked at temperatures of 2500 K (Vokrouh- lick´y and Nesvorn´y, 2012). Solar heating has a penetration depth of typically a few cm (see Eq. 5 and Spencer et al., 1989). Organic compo- nents found on meteorites break up at temperatures as low as 300–670 K (see Fig. 11 and Kebukawa et al., 2010; Frost et al., 2000; Huang et al., 1994), thus solar heating can re- move these components from asteroid surfaces. 7.3. Thermal metamorphism of meteorites Radiative heating from the Sun has been invoked as a mechanism for the thermal metamorphism of metamorphic CK carbonaceous chondrites (Chaumard et al., 2012). The matrix of these chondrites shows textures consistent with a transient thermal event during which temperatures rose be- tween 550 and 950 K. The inferred duration of these events 17 Fig. 11.— Surface temperature of an asteroid or meteoroid as a function of the distance from the Sun. Vertical arrows indicate the threshold temperature for the thermal alteration/desiccation for a variety of chemical compounds discussed in the text (see Delbo and Michel, 2011, and references therein for further information). The temperature range for thermal metamorphism of the CK chon- drites is from Chaumard et al. (2012). is of the order of days to years, much longer than the time scale of shock events but shorter than the time scale for heating by the decay of radiogenic species such as 26Al (e.g., Kallemeyn et al., 1991). 7.4. Subsurface ice sublimation Observational evidence for the presence of ice on aster- oid surfaces stems from the discovery of main belt comets (MBCs; Hsieh and Jewitt, 2006), the localized release of water vapor from the surface of (1) Ceres (Kuppers et al., 2014), and the detection of spectroscopic signatures inter- preted as water ice frost on the surface of (24) Themis (Rivkin and Emery, 2010; Campins et al., 2010) and of (65) Cybele (Licandro et al., 2011). The lifetime of ices on the surface and in the subsurface depends strongly on temperature. TPMs have been used to estimate these temperatures. This requires a modification of the “classical” TPM as presented in section 4, such that heat conduction is coupled with gas diffusion (Schorghofer, 2008; Capria et al., 2012; Prialnik and Rosenberg, 2009). The referenced models assume a spherical shape. As for the interior structure, Capria et al. (2012); Prialnik and Rosen- berg (2009) assume a comet-like structure, i.e., an inti- mate mixture of ice and dust throughout the entire body, while Schorghofer (2008) consider an ice layer underneath a rocky regolith cover. Sublimation of ice and the trans- port of water molecules through the fine-grained regolith is modeled in all cases. All authors agree that water ice exposed on asteroid sur- faces sublimates completely on timescales much shorter than the age of the Solar System. Therefore, asteroid sur- faces were expected to be devoid of water ice, contrary to the observational evidence quoted above. However, water ice can be stable over 4.5 Gy in the shallow subsurface, at a depth of ∼1–10 m. In particular, Fanale and Salvail (1989) showed that ice could have survived in the subsurface at the ��polar regions of Ceres. Large heliocentric distances, slow rotation, and a fine-grained regolith leading to low thermal conductivity and short molecular free path, all favor the sta- bilization of subsurface water ice (Schorghofer, 2008). The same authors conclude that rocky surfaces, in contrast to dusty surfaces, are rarely able to retain ice in the shallow subsurface. To be observable on the surface, buried ice most be ex- posed. Campins et al. (2010) describe several plausible mechanisms such as impacts, recent change in the obliquity of the spin pole, and daily or orbital thermal pulses reaching a subsurface ice layer. 8. FUTURE CHALLENGES FOR TPMs The Spitzer and WISE telescopes have opened a new era of asteroid thermal-infrared observations and the exploita- tion of their data through TPMs has just begun (e.g., Al´ı- Lagoa et al., 2014; Rozitis et al., 2014; Emery et al., 2014). At the moment, the limiting factor is the availability of ac- curate asteroid shape models. However, optical-wavelength all-sky surveys such as PanSTARRS, LSST, and Gaia are expected to produce enormous photometric data sets lead- ing to thousands of asteroid models. We envision the avail- ability of thousands of thermal-inertia values in some years from now, enabling more statistically robust studies of ther- mal inertia as a function of asteroid size, spectral class, albedo, rotation period, etc. For instance, the distribution of Γ within asteroid fam- ilies will be crucial in the search of evidence of aster- oid differentiation: asteroid formation models and mete- orite studies suggest that hundreds of planetesimals experi- enced complete or partial differentiation. An asteroid fam- ily formed from the catastrophic disruption of such a dif- ferentiated asteroid should contain members corresponding to the crust, the mantel and the iron core. However, the observed spectra and albedos are very homogeneous across asteroid families. Thermal inertia might help in separating iron rich from iron poor family members, supposedly orig- inating respectively from the core and mantle of the differ- entiated parent body (e.g., Matter et al., 2013; Harris and Drube, 2014). At any size range, Fig. 9 shows an almost tenfold vari- ability in thermal inertia, corresponding to difference in av- erage regolith grain size of almost two orders of magnitude Gundlach and Blum (2013). For small near-Earth aster- oids, this could be due to a combination of thermal cracking (Delbo et al., 2014), regolith motion (Murdoch et al., 2015), and cohesive forces (Rozitis et al., 2014). Faster rotation periods allow more thermal cycles, which then enhances thermal fracturing. It also encourages regolith to move to- wards the equator where the gravitational potential is at its lowest (Walsh et al., 2008). And for the extremely fast rota- tors, large boulders/rocks could be selectively lofted away, because they stick less well to the surface than smaller par- ticles. For D > 100 km sized asteroids, Γ-values might be help to distinguish between primordial and more recently re-accumulated asteroids. The former had ∼ 4 Gy of re- golith evolution, the latter have a less developed and there- fore coarser regolith. The high-precision thermal-infrared data of WISE and Spitzer pose new challenges to TPMs, as model uncertain- ties are now comparable to the uncertainty of the measured flux. This will become even more important with the launch of the James Webb Space Telescope (JWST). In particular, the accuracy of the shape models might represent a limiting factor (e.g., Rozitis and Green, 2014). The next challenge will be to allow the TPM to optimize the asteroid shape. This seems to be possible as the infrared photometry is also sensitive to shape, provided good-quality thermal data are available (Durech et al., 2015). New interferometric facilities, such as MATISSE, LBTI, and ALMA, will become available in the next years requir- ing TPMs to calculate precise disk-resolved thermal fluxes (Durech et al., 2015). The wavelengths of ALMA, simi- lar to those of MIRO, will allow to measure the thermal- infrared radiation from the subsoil of asteroids, thus pro- viding further information about how thermal inertia varies with depth. Certainly, constraining roughness is one of the future challenges for TPMs. To do so from disk-integrated data re- quires a range of wavelengths and solar phase angles. Low phase angle measurements are enhanced by beaming whilst high phase angle measurements are reduced by beaming. In particular, shorter wavelengths are affected more than longer wavelengths. Moreover, the future availability of precise sizes and cross sections of asteroids from stellar occultation timing (Tanga and Delbo, 2007), combined with shape informa- tion derived from lightcurve inversion (Durech et al., 2015) will allow to remove the need to constrain the object size from TPM analysis. Iinfrared fluxes will thus be converted into highly reliable thermal inertia and roughness values. Acknowledgments We are grateful to S. Green and to an anonymous referee for their thorough reviews. MD thanks J. Hanus and the support from the French Agence National de la Recherche (ANR) SHOCKS. REFERENCES Al´ı-Lagoa V., Lionni L., Delbo M., et al. (2014). Thermophysi- cal properties of near-Earth asteroid (341843) 2008 EV5 from WISE data. Astronomy & Astrophysics, 561, A45. Bandfield J. L., Ghent R. R., Vasavada A. R., et al. (2011). Lunar surface rock abundance and regolith fines temperatures derived from LRO Diviner Radiometer data. Journal of Geophysical Research, 116. Barucci A. M., Fulchignoni M., Ji J., et al. (2015). The Steins, Lutetia, and Toutatis fly-bys. In Asteroids IV. this book. Bottke W. F., Durda D. D., Nesvorn´y D., et al. (2005). The fos- silized size distribution of the main asteroid belt. Icarus, 175, 111–140. 18 Bottke W. F., Vokrouhlick´y D., Broz M., et al. (2001). Dynam- ical Spreading of Asteroid Families by the Yarkovsky Effect. Science, 294, 1693–1696. Bottke W. F. J., Vokrouhlick´y D., Rubincam D. P., et al. (2006). The Yarkovsky and Yorp Effects: Implications for Asteroid Dynamics. Annual Review of Earth and Planetary Sciences, 34, 157–191. Breiter S., Bartczak P., Czekaj M., et al. (2009). The YORP ef- fect on 25143 Itokawa. Astronomy & Astrophysics, 507, 1073– 1081. Buhl D., Welch W. J., and Rea D. G. (1968a). Anomalous Cooling of a Cratered Lunar Surface. Journal of Geophysical Research, 73, 7593–7608. Buhl D., Welch W. J., and Rea D. G. (1968b). Reradiation and Thermal Emission from Illuminated Craters on the Lunar Sur- face. Journal of Geophysical Research, 73, 5281–5295. Campins H., Hargrove K., Pinilla-Alonso N., et al. (2010). Wa- ter ice and organics on the surface of the asteroid 24 Themis. Nature, 464, 1320–1321. Capek D. and Vokrouhlick´y D. (2004). The YORP effect with finite thermal conductivity. Icarus, 172, 526–536. Capek D. and Vokrouhlick´y D. (2010). Thermal stresses in small meteoroids. Astronomy and Astrophysics, 519, A75. Capek D. and Vokrouhlick´y D. (2012). Thermal stresses in small meteoroids. II. Effects of an insulating surface layer. Astron- omy & Astrophysics, 539, A25. Capria M. T., Marchi S., De Sanctis M. C., et al. (2012). The activity of main belt comets. Astronomy & Astrophysics, 537, A71. Capria M. T., Tosi F., De Sanctis M. C., et al. (2014). Vesta sur- face thermal properties map. Geophysical Research Letters, 41, 1438–1443. Carry B. (2012). Density of asteroids. Planetary and Space Sci- ence, 73, 98–118. Carry B., Matter A., Scheirich P., et al. (2015). The small binary asteroid (939) Isberga. Icarus, 248, 516–525. Chamberlain M. A., Lovell A. J., and Sykes M. V. (2007). Sub- millimeter lightcurves of Vesta. Icarus, 192, 448–459. Chamberlain M. A., Lovell A. J., and Sykes M. V. (2009). Sub- millimeter photometry and lightcurves of Ceres and other large asteroids. Icarus, 202, 487–501. Chaumard N., Devouard B., Delbo M., et al. (2012). Radiative heating of carbonaceous near-Earth objects as a cause of ther- mal metamorphism for CK chondrites. Icarus, 220, 65–73. Chesley S. R., Farnocchia D., Nolan M. C., et al. (2014). Orbit and bulk density of the OSIRIS-REx target Asteroid (101955) Bennu. Icarus, 235, 5–22. Chesley S. R., Ostro S. J., Vokrouhlick´y D., et al. (2003). Direct Detection of the Yarkovsky Effect by Radar Ranging to Aster- oid 6489 Golevka. Science, 302, 1739–1742. Consolmagno G. J., Macke R. J., Rochette P., et al. (2006). Den- sity, magnetic susceptibility, and the characterization of ordi- nary chondrite falls and showers. Meteoritics & Planetary Sci- ence, 41, 331–342. Consolmagno G. J., Schaefer M. W., Schaefer B. E., et al. (2013). The measurement of meteorite heat capacity at low tempera- tures using liquid nitrogen vaporization. Planetary and Space Science, 87, 146–156. Cruikshank D. P., Stansberry J. A., Emery J. P., et al. (2005). The High-Albedo Kuiper Belt Object (55565) 2002 AW197. The Astrophysical Journal Letters, 624, L53–L56. Davidsson B., Rickman H., Bandfield J. L., et al. (2015). Inter- pretation of thermal emission. I. The effect of roughness for spatially resolved atmosphereless bodies. Icarus (in press). Davidsson B. J. R., Guti´errez P. J., Groussin O., et al. (2013). Thermal inertia and surface roughness of Comet 9P/Tempel 1. Icarus, 224, 154–171. Davidsson B. J. R. and Rickman H. (2014). Surface roughness and three-dimensional heat conduction in thermophysical models. Icarus, 243, 58–77. De Sanctis M. C., Ammannito E., Capria M. T., et al. (2012). Spectroscopic Characterization of Mineralogy and Its Diver- sity Across Vesta. Science, 336, 697–700. Delbo M. (2004). The nature of near-Earth asteroids from the study of their thermal infrared emission. PhD thesis, pp. 1– 210. Delbo M., dell’Oro A., Harris A. W., et al. (2007). Thermal inertia of near-Earth asteroids and implications for the magnitude of the Yarkovsky effect. Icarus, 190, 236–249. Delbo M. and Harris A. W. (2002). Physical properties of near- Earth asteroids from thermal infrared observations and thermal modeling. Meteoritics & Planetary Science, 37, 1929–1936. Delbo M., Harris A. W., Binzel R. P., et al. (2003). Keck obser- vations of near-Earth asteroids in the thermal infrared. Icarus, 166, 116–130. Delbo M., Libourel G., Wilkerson J., et al. (2014). Thermal fatigue as the origin of regolith on small asteroids. Nature, 508, 233– 236. Delbo M., Ligori S., Matter A., et al. (2009). First VLTI-MIDI Direct Determinations of Asteroid Sizes. The Astrophysical Journal, 694, 1228–1236. Delbo M. and Michel P. (2011). Temperature History and Dynam- ical Evolution of (101955) 1999 RQ 36: A Potential Target for Sample Return from a Primitive Asteroid. The Astrophysical Journal Letters, 728, L42. Delbo M. and Tanga P. (2009). Thermal inertia of main belt as- teroids smaller than 100 km from IRAS data. Planetary and Space Science, 57, 259–265. Delbo M., Walsh K., Mueller M., et al. (2011). The cool surfaces of binary near-Earth asteroids. Icarus, 212, 138–148. Dombard A. J., Barnouin O. S., Prockter L. M., et al. (2010). Icarus, 210, Boulders and ponds on the Asteroid 433 Eros. 713–721. Duennebier F. and Sutton G. H. (1974). Thermal moonquakes. Journal of Geophysical Research, 79, 4351–4363. Durech J., Carry B., Delbo M., et al. (2015). Asteroid Models from Multiple Data Sources. In Asteroids IV. University of Arizona Press, Tucson, AZ. Emery J. P., Fern´andez Y. R., Kelley M. S. P., et al. (2014). Ther- mal infrared observations and thermophysical characterization of OSIRIS-REx target asteroid (101955) Bennu. Icarus, 234, 17–35. Emery J. P., Sprague A. L., Witteborn F. C., et al. (1998). Mercury: Thermal Modeling and Mid-infrared (5-12 µm) Observations. Icarus, 136, 104–123. Fanale F. P. and Salvail J. R. (1989). The water regime of asteroid (1) Ceres. Icarus, 82, 97–110. Fern´andez Y. R., Jewitt D. C., and Sheppard S. S. (2002). Thermal Properties of Centaurs Asbolus and Chiron. The Astronomical Journal, 123, 1050–1055. Fern´andez Y. R., Sheppard S. S., and Jewitt D. C. (2003). The Albedo Distribution of Jovian Trojan Asteroids. The Astro- nomical Journal, 126, 1563–1574. 19 Fornasier S., Lellouch E., Muller T., et al. (2013). TNOs are Cool: A survey of the trans-Neptunian region. VIII. Combined Her- schel PACS and SPIRE observations of nine bright targets at 70-500 µm. Astronomy and Astrophysics, 555, A15. Fountain W. F., Fountain J. A., Jones B. P., et al. (1976). Obser- vational and theoretical temperatures for a total lunar eclipse. The moon, 15, 421–437. Fowler J. W. and Chillemi J. R. (1992). IRAS asteroid data pro- cessing. The IRAS Minor Planet Survey. Frost R. L., Ruan H., Kloprogge J. T., et al. (2000). Dehydration and dehydroxylation of nontronites and ferruginous smectite. Thermochimica Acta, 346, 63–72. Giese B. and Kuehrt E. (1990). Theoretical interpretation of in- frared measurements at Deimos in the framework of crater ra- diation. Icarus, 88, 372–379. Golubov O. and Krugly Y. N. (2012). Tangential Component of the YORP Effect. The Astrophysical Journal Letters, 752, L11. Grimm R. E. and McSween H. Y. (1993). Heliocentric zoning of the asteroid belt by aluminum-26 heating. Science (ISSN 0036-8075), 259, 653–655. Groussin O., Lamy P., Fornasier S., et al. (2011). The properties of asteroid (2867) Steins from Spitzer Space Telescope obser- vations and OSIRIS shape reconstruction. Astronomy & Astro- physics, 529, A73. Groussin O., Lamy P., and Jorda L. (2004). Properties of the nuclei of Centaurs Chiron and Chariklo. Astronomy & Astrophysics, 413, 1163–1175. Groussin O., Sunshine J. M., Feaga L. M., et al. (2013). The tem- perature, thermal inertia, roughness and color of the nuclei of Comets 103P/Hartley 2 and 9P/Tempel 1. Icarus, 222, 580– 594. Gulkis S., Frerking M., Crovisier J., et al. (2007). MIRO: Mi- crowave Instrument for Rosetta Orbiter. Space Science Re- views, 128, 561–597. Gundlach B. and Blum J. (2012). Outgassing of icy bodies in the Solar System - II: Heat transport in dry, porous surface dust layers. Icarus, 219, 618–629. Gundlach B. and Blum J. (2013). A new method to determine the grain size of planetary regolith. Icarus, 223, 479–492. Hall K. (1999). The role of thermal stress fatigue in the breakdown of rock in cold regions. Geomorphology, 31, 47–63. Hall K. and Andr´e M.-F. (2001). New insights into rock weath- ering from high-frequency rock temperature data: an Antarctic study of weathering by thermal stress. Geomorphology, 41, 23–35. Hanus J., Broz M., Durech J., et al. (2013). An anisotropic distri- bution of spin vectors in asteroid families. Astronomy & Astro- physics, 559, A134. Hanus J., Durech J., Broz M., et al. (2011). A study of aster- oid pole-latitude distribution based on an extended set of shape models derived by the lightcurve inversion method. Astronomy & Astrophysics, 530, 134. Hapke B. (1984). Bidirectional reflectance spectroscopy. III - Cor- rection for macroscopic roughness. Icarus, 59, 41–59. Hapke B. (1996). A model of radiative and conductive energy Journal of Geophysical Re- transfer in planetary regoliths. search, 101, 16817–16832. Harris A., Boslough M., Chapman C. R., et al. (2015). Asteroid Impacts and Modern Civilization: Can we Prevent a Catastro- In Asteroids IV. University of Arizona Press, Tucson, phe? AZ. Harris A. W. (1998). A Thermal Model for Near-Earth Asteroids. Icarus, 131, 291–301. Harris A. W. (2006). The surface properties of small asteroids from thermal-infrared observations. Asteroids, Comets, Mete- ors. Proceedings IAU Symposium No. 229, 2005 D. Lazzaro, S. Ferraz-Mello & J.A. Fernandez, eds, 229, 449–463. Harris A. W. and Davies J. K. (1999). Physical Characteristics of Near-Earth Asteroids from Thermal Infrared Spectrophotome- try. Icarus, 142, 464–475. Harris A. W. and Drube L. (2014). How to Find Metal-rich Aster- oids. The Astrophysical Journal Letters, 785, L4. Harris A. W. and Lagerros J. S. V. (2002). Asteroids in the Ther- mal Infrared. Asteroids III, W. F. Bottke Jr., A. Cellino, P. Paolicchi, and R. P. Binzel (eds), University of Arizona Press, Tucson., pp. 205–218. Harris A. W., Mueller M., Delbo M., et al. (2005). The surface properties of small asteroids: Peculiar Betulia—A case study. Icarus, 179, 95–108. Hoerz F., Schneider E., Gault D. E., et al. (1975). Catastrophic rupture of lunar rocks - A Monte Carlo simulation. Lunar Sci- ence Institute, 13, 235–258. Holsapple K. A. (2010). On YORP-induced spin deformations of asteroids. Icarus, 205, 430–442. Horner J., Muller T. G., and Lykawka P. S. (2012). (1173) An- chises - thermophysical and dynamical studies of a dynami- cally unstable Jovian Trojan. Monthly Notices of the Royal Astronomical Society, 423, 2587–2596. Horz F. and Cintala M. (1997). Impact experiments related to the evolution of planetary regoliths. Meteoritics & Planetary Science, 32, 179–209. Hsieh H. H. and Jewitt D. (2006). A Population of Comets in the Main Asteroid Belt. Science, 312, 561–563. Huang W. L., Bassett W. A., and Wu T. C. (1994). Dehydration and hydration of montmorillonite at elevated temperatures and pressures monitored using synchrotron radiation. American Mineralogist, 79, 683–691. Huebner W. F., Benkhoff J., Capria M. T., et al. (2006). Heat and Gas Diffusion in Comet Nuclei. ISSI Scientific report SR-004, vol. 4. International Space Science Institute. Jacobson S. A. and Scheeres D. J. (2011). Dynamics of rotation- ally fissioned asteroids: Source of observed small asteroid sys- tems. Icarus, 214, 161–178. Jakosky B. M., Finiol G. W., and Henderson B. G. (1990). Direc- tional variations in thermal emission from geologic surfaces. Geophysical Research Letters, 17, 985–988. Jewitt D., Hsieh H., and Agarwal J. (2015). The Active Asteroids. In Asteroids IV. University of Arizona Press, Tucson, AZ. Jewitt D. and Li J. (2010). Activity in Geminid Parent (3200) Phaethon. The Astronomical Journal, 140, 1519–1527. Kallemeyn G. W., Rubin A. E., and Wasson J. T. (1991). The com- positional classification of chondrites. V - The Karoonda (CK) group of carbonaceous chondrites. Geochimica et Cosmochim- ica Acta, 55, 881–892. Kebukawa Y., Nakashima S., and Zolensky M. E. (2010). Kinetics of organic matter degradation in the Murchison meteorite for the evaluation of parent-body temperature history. Meteoritics & Planetary Science, 45, 99–113. Keihm S., Kamp L., Gulkis S., et al. (2013). Reconciling main belt asteroid spectral flux density measurements with a self- consistent thermophysical model. Icarus, 226, 1086–1102. 20 Keihm S., Tosi F., Kamp L., et al. (2012). Interpretation of com- bined infrared, submillimeter, and millimeter thermal flux data obtained during the Rosetta fly-by of Asteroid (21) Lutetia. Icarus, 221, 395–404. Keihm S. J. (1984). Interpretation of the lunar microwave bright- ness temperature spectrum - Feasibility of orbital heat flow mapping. Icarus, 60, 568–589. Keil K. (2000). Thermal alteration of asteroids: evidence from meteorites. Planetary and Space Science, 48, 887–903. Kieffer H. H., Martin T. Z., Peterfreund A. R., et al. (1977). Ther- mal and albedo mapping of Mars during the Viking primary mission. Journal of Geophysical Research, 82, 4249–4291. Kuppers M., O’Rourke L., Bockel´ee-Morvan D., et al. (2014). Lo- calized sources of water vapour on the dwarf planet (1) Ceres. Nature, 505, 525–527. Lagerros J. S. V. (1996a). Thermal physics of asteroids. I. Effects of shape, heat conduction and beaming. Astronomy and Astro- physics, 310, 1011–1020. Lagerros J. S. V. (1996b). Thermal physics of asteroids. II. Po- larization of the thermal microwave emission from asteroids. Astronomy & Astrophysics, 315, 625–632. Lagerros J. S. V. (1997). Thermal physics of asteroids. III. Irregu- lar shapes and albedo variegations. Astronomy & Astrophysics, 325, 1226–1236. Lagerros J. S. V. (1998). Thermal physics of asteroids. IV. Thermal infrared beaming. Astronomy & Astrophysics, 332, 1123–1132. Lamy P. L., Jorda L., Fornasier S., et al. (2008). Asteroid 2867 Steins. III. Spitzer Space Telescope observations, size deter- mination, and thermal properties. Astronomy & Astrophysics, 487, 1187–1193. Langeseth M. G. and Keihm S. J. (1977). Lunar Heat-Flow Exper- iment. NASA Technical Report NASA-CR-151619; CU-4-77, pp. 1–289. Lauretta D. S., Barucci M. A., Bierhaus E. B., et al. (2012). The OSIRIS-REx Mission — Sample Acquisition Strategy and Ev- idence for the Nature of Regolith on Asteroid (101955) 1999 RQ36. Asteroids, Comets, Meteors. Proceedings IAU Sympo- sium No. 229, 2005 D. Lazzaro, S. Ferraz-Mello & J.A. Fer- nandez, eds, 1667, 6291. Lawson S. L. S. L., Rodger A. P. A. P., Bender S. C. S. C., et al. (2003). Multispectral thermal imager observations of the moon during total eclipse. Lunar and Planetary Science XXXIV. Lebofsky L. A. and Rieke G. H. (1979). Thermal properties of 433 Eros. Icarus, 40, 297–308. Lebofsky L. A. and Spencer J. R. (1989). Radiometry and a ther- mal modeling of asteroids. In Asteroids II, pp. 128–147. Uni- versity of Arizona Press, Tucson, AZ. Lebofsky L. A., Sykes M. V., Tedesco E. F., et al. (1986). A refined ’standard’ thermal model for asteroids based on observations of 1 Ceres and 2 Pallas. Icarus, 68, 239–251. Lellouch E., Santos-Sanz P., Lacerda P., et al. (2013). ”TNOs are Cool”: A survey of the trans-Neptunian region. IX. Thermal properties of Kuiper belt objects and Centaurs from combined Herschel and Spitzer observations. Astronomy & Astrophysics, 557, 60. Levi F. A. (1973). Thermal Fatigue: A Possible Source of Struc- tural Modifications in Meteorites. Meteoritics & Planetary Sci- ence, 8, 209–221. Leyrat C., Barucci A., Mueller T., et al. (2012). Thermal properties of (4) Vesta derived from Herschel measurements. Astronomy & Astrophysics, 539, A154. Leyrat C., Coradini A., Erard S., et al. (2011). Thermal properties of the asteroid (2867) Steins as observed by VIRTIS/Rosetta. Astronomy & Astrophysics, 531, A168. Licandro J., Campins H., Kelley M., et al. (2011). (65) Cybele: detection of small silicate grains, water-ice, and organics. As- tronomy & Astrophysics, 525, A34. Lim L. F., Emery J. P., and Moskovitz N. A. (2011). Mineralogy and thermal properties of V-type Asteroid 956 Elisa: Evidence for diogenitic material from the Spitzer IRS (5-35 µm) spec- trum. Icarus, 213, 510–523. Lowry S. C., Weissman P. R., Duddy S. R., et al. (2014). The internal structure of asteroid (25143) Itokawa as revealed by detection of YORP spin-up. Astronomy & Astrophysics, 562, A48. Lucey P. G. (2000). Observations of the moon using the Air Force Maui Space Surveillance Complex. Proc. SPIE Vol. 4091, 4091, 216–224. Lucey P. G. (2006). Radiative transfer modeling of the effect of mineralogy on some empirical methods for estimating iron concentration from multispectral imaging of the Moon. Jour- nal of Geophysical Research, 111, 8003. Macke R. J., BRITT D. T., and Consolmagno G. J. (2011a). Den- sity, porosity, and magnetic susceptibility of achondritic mete- orites. Meteoritics & Planetary Science, 46, 311–326. Macke R. J., Consolmagno G. J., and BRITT D. T. (2011b). Density, porosity, and magnetic susceptibility of carbonaceous chondrites. Meteoritics & Planetary Science, 46, 1842–1862. Macke R. J., Consolmagno G. J., BRITT D. T., et al. (2010). En- statite chondrite density, magnetic susceptibility, and porosity. Meteoritics & Planetary Science, 45, 1513–1526. Magri C., Consolmagno G. J., Ostro S. J., et al. (2001). Radar constraints on asteroid regolith compositions using 433 Eros as ground truth. Meteoritics & Planetary Science, 36, 1697– 1709. Mainzer A., Trilling D., and Usui F. (2015). Space-Based Infrared In Asteroids IV. University of Arizona Studies of Asteroids. Press, Tucson, AZ. Marchi S., Delbo M., Morbidelli A., et al. (2009). Heating of near-Earth objects and meteoroids due to close approaches to the Sun. Monthly Notices of the Royal Astronomical Society, 400, 147–153. Marchi S., Paolicchi P., Lazzarin M., et al. (2006). A General Spectral Slope-Exposure Relation for S-Type Main Belt and Near-Earth Asteroids. The Astronomical Journal, 131, 1138– 1141. Marchis F., Enriquez J. E., Emery J. P., et al. (2012). Mul- tiple asteroid systems: Dimensions and thermal properties from Spitzer Space Telescope and ground-based observations. Icarus, 221, 1130–1161. Matter A., Delbo M., Carry B., et al. (2013). Evidence of a metal- rich surface for the Asteroid (16) Psyche from interferometric observations in the thermal infrared. Icarus, 226, 419–427. Matter A., Delbo M., Ligori S., et al. (2011). Determination of physical properties of the Asteroid (41) Daphne from interfer- ometric observations in the thermal infrared. Icarus, 215, 47– 56. McSween H. Y. J., Ghosh A., Grimm R. E., et al. (2002). Thermal Evolution Models of Asteroids. Asteroids III, W. F. Bottke Jr., A. Cellino, P. Paolicchi, and R. P. Binzel (eds), University of Arizona Press, Tucson., pp. 559–571. 21 Mellon M. T., Jakosky B. M., Kieffer H. H., et al. (2000). High- Resolution Thermal Inertia Mapping from the Mars Global Surveyor Thermal Emission Spectrometer. Icarus, 148, 437– 455. Molaro J. and Byrne S. (2012). Rates of temperature change of air- less landscapes and implications for thermal stress weathering. Journal of Geophysical Research, 117, 10011. Mommert M., Farnocchia D., Hora J. L., et al. (2014a). Physical Properties of Near-Earth Asteroid 2011 MD. The Astrophysical Journal Letters, 789, L22. Mommert M., Hora J. L., Farnocchia D., et al. (2014b). Constrain- ing the Physical Properties of Near-Earth Object 2009 BD. The Astrophysical Journal, 786, 148. Morrison D. and Cruikshank D. P. (1973). Thermal Properties of the Galilean Satellites. Icarus, 18, 224–236. Mueller M. (2007). Surface Properties of Asteroids from Mid- Infrared Observations and Thermophysical Modeling. PhD thesis, PhD thesis on arXiv.org. Mueller M., Harris A. W., Bus S. J., et al. (2006). The size and albedo of Rosetta fly-by target 21 Lutetia from new IRTF mea- surements and thermal modeling. Astronomy & Astrophysics, 447, 1153–1158. Mueller M., Marchis F., Emery J. P., et al. (2010). Eclipsing bi- nary Trojan asteroid Patroclus: Thermal inertia from Spitzer observations. Icarus, 205, 505–515. Muller T. G. (2002). Thermophysical analysis of infrared observa- tions of asteroids. Meteoritics & Planetary Science, 37, 1919– 1928. Muller T. G. and Barnes P. J. (2007). 3.2 mm lightcurve observa- tions of (4) Vesta and (9) Metis with the Australia Telescope Compact Array. Astronomy & Astrophysics, 467, 737–747. Muller T. G., Durech J., Hasegawa S., et al. (2011). Thermo- physical properties of 162173 (1999 JU3), a potential flyby and rendezvous target for interplanetary missions. Astronomy and Astrophysics, 525, 145. Muller T. G., Hasegawa S., and Usui F. (2014a). (25143) Itokawa: The power of radiometric techniques for the interpretation of remote thermal observations in the light of the Hayabusa ren- dezvous results*. Publications of the Astronomical Society of Japan, 66, 52. Muller T. G., Kiss C., Scheirich P., et al. (2014b). Thermal infrared observations of asteroid (99942) Apophis with Herschel. As- tronomy & Astrophysics, 566, A22. Muller T. G. and Lagerros J. S. V. (1998). Asteroids as far- infrared photometric standards for ISOPHOT. Astronomy & Astrophysics, 338, 340–352. Muller T. G. and Lagerros J. S. V. (2002). Asteroids as calibra- tion standards in the thermal infrared for space observatories. Astronomy & Astrophysics, 381, 324–339. Muller T. G., Miyata T., Kiss C., et al. (2013). Physical prop- erties of asteroid 308635 (2005 YU55) derived from multi- instrument infrared observations during a very close Earth ap- proach. Astronomy & Astrophysics, 558, A97. Muller T. G., O’Rourke L., Barucci A. M., et al. (2012). Phys- ical properties of OSIRIS-REx target asteroid (101955) 1999 RQ36. Derived from Herschel, VLT/ VISIR, and Spitzer ob- servations. Astronomy & Astrophysics, 548, A36. Muller T. G., Sterzik M. F., Schutz O., et al. (2004). Thermal infrared observations of near-Earth asteroid 2002 NY40. As- tronomy & Astrophysics, 424, 1075–1080. Murdoch N., sanchez P., Schwartz S. R., et al. (2015). Asteroid In Asteroids IV. University of Arizona Surface Geophysics. Press, Tucson, AZ. Okada T., Fukuhara T., Tanaka S., et al. (2013). Thermal-Infrared Imager TIR on Hayabusa2: Science and Instrumentation. 44th Lunar and Planetary Science Conference, 44, 1954. Opeil C. P., Consolmagno G. J., and Britt D. T. (2010). The ther- mal conductivity of meteorites: New measurements and analy- sis. Icarus, 208, 449–454. Opeil C. P., Consolmagno G. J., Safarik D. J., et al. (2012). Stony meteorite thermal properties and their relationship with mete- orite chemical and physical states. Meteoritics & Planetary Science, 47, 319–329. O’Rourke L., Muller T., Valtchanov I., et al. (2012). Thermal and shape properties of asteroid (21) Lutetia from Herschel obser- vations around the Rosetta flyby. Planetary and Space Science, 66, 192–199. Paige D. A., Foote M. C., Greenhagen B. T., et al. (2010). The Lunar Reconnaissance Orbiter Diviner Lunar Radiometer Ex- periment. Space Science Reviews, 150, 125–160. P´al A., Kiss C., Muller T. G., et al. (2012). ”TNOs are Cool”: A survey of the trans-Neptunian region. VII. Size and surface characteristics of (90377) Sedna and 2010 EK139. Astronomy and Astrophysics, 541, L6. Pettit E. (1940). Radiation Measurements on the Eclipsed Moon. Astrophysical Journal, 91, 408–421. Pettit E. and Nicholson S. B. (1930). Lunar radiation and temper- atures. The Astrophysical Journal, 71, 102–135. Piqueux S. and Christensen P. R. (2009a). A model of thermal conductivity for planetary soils: 1. Theory for unconsolidated soils. Journal of Geophysical Research, 114, 9005. Piqueux S. and Christensen P. R. (2009b). A model of thermal conductivity for planetary soils: 2. Theory for cemented soils. Journal of Geophysical Research, 114, 9006. Piqueux S. and Christensen P. R. (2011). Temperature-dependent thermal inertia of homogeneous Martian regolith. Journal of Geophysical Research, 116, 7004. Pravec P., Scheirich P., KUSNIRAK P., et al. (2006). Photometric survey of binary near-Earth asteroids. Icarus, 181, 63–93. Presley M. A. and Christensen P. R. (1997a). Thermal conductivity measurements of particulate materials 1. A review. Journal of Geophysical Research, 102, 6535–6550. Presley M. A. and Christensen P. R. (1997b). Thermal conductiv- ity measurements of particulate materials 2. results. Journal of Geophysical Research, 102, 6551–6566. Prialnik D. and Rosenberg E. D. (2009). Can ice survive in main- belt comets? Long-term evolution models of comet 133P/Elst- Pizarro. Monthly Notices of the Royal Astronomical Society (ISSN 0035-8711), 399, L79–L83. Putzig N. E. and Mellon M. T. (2007). Apparent thermal inertia and the surface heterogeneity of Mars. Icarus, 191, 68–94. Raymond C. A., Jaumann R., Nathues A., et al. (2011). The Dawn Topography Investigation. Space Science Reviews, 163, 487– 510. Redman R. O., Feldman P. A., Matthews H. E., et al. (1992). Mil- limeter and submillimeter observations of the asteroid 4 Vesta. Astronomical Journal (ISSN 0004-6256), 104, 405–411. Rivkin A. S. and Emery J. P. (2010). Detection of ice and organics on an asteroidal surface. Nature, 464, 1322–1323. Rozitis B., Duddy S. R., Green S. F., et al. (2013). A thermophysi- cal analysis of the (1862) Apollo Yarkovsky and YORP effects. Astronomy and Astrophysics, 555, A20. Rozitis B. and Green S. F. (2011). Directional characteristics of thermal-infrared beaming from atmosphereless planetary sur- faces - a new thermophysical model. Monthly Notices of the Royal Astronomical Society, 415, 2042–2062. 22 Rozitis B. and Green S. F. (2012). The influence of rough surface thermal-infrared beaming on the Yarkovsky and YORP effects. Monthly Notices of the Royal Astronomical Society, 423, 367– 388. Rozitis B. and Green S. F. (2013). The influence of global self- heating on the Yarkovsky and YORP effects. Monthly Notices of the Royal Astronomical Society, 433, 603–621. Rozitis B. and Green S. F. (2014). Physical characterisation of near-Earth asteroid (1620) Geographos. Reconciling radar and thermal-infrared observations. Astronomy & Astrophysics, 568, A43. Rozitis B., MacLennan E., and Emery J. P. (2014). Cohesive forces prevent the rotational breakup of rubble-pile asteroid (29075) 1950 DA. Nature, 512, 174–176. Russell C. T., Raymond C. A., Coradini A., et al. (2012). Dawn at Vesta: Testing the Protoplanetary Paradigm. Science, 336, 684–686. Scheeres D. J. (2007). Rotational fission of contact binary aster- oids. Icarus, 189, 370–385. Scheeres D. J. and Gaskell R. W. (2008). Effect of density inhomo- geneity on YORP: The case of Itokawa. Icarus, 198, 125–129. Schorghofer N. (2008). The Lifetime of Ice on Main Belt Aster- oids. The Astrophysical Journal, 682, 697–705. Sexl R. U., Sexl H., Stremnitzer H., et al. (1971). The directional characteristics of lunar infrared radiation. The moon, 3, 189– 213. Shorthill R. W. (1973). Infrared Atlas Charts of the Eclipsed Moon. The moon, 7, 22–45. Sierks H., Lamy P., Barbieri C., et al. (2011). Images of Aster- oid 21 Lutetia: A Remnant Planetesimal from the Early Solar System. Science, 334, 487–490. Smith B. G. (1967). Lunar surface roughness: Shadowing and thermal emission. Journal of Geophysical Research. Spencer J. R. (1990). A rough-surface thermophysical model for airless planets. Icarus, 83, 27–38. Spencer J. R., Lebofsky L. A., and Sykes M. V. (1989). Systematic biases in radiometric diameter determinations. Icarus, 78, 337– 354. Statler T. S. (2009). Extreme sensitivity of the YORP effect to small-scale topography. Icarus, 202, 502–513. Tambovtseva L. V. and Shestakova L. i. (1999). Cometary splitting due to thermal stresses. Planetary and Space Science, 47, 319– 326. Tanga P. and Delbo M. (2007). Asteroid occultations today and tomorrow: toward the GAIA era. Astronomy and Astrophysics, 474, 1015–1022. Thomas P. C., Joseph J., Carcich B., et al. (2002). Eros: Shape, Topography, and Slope Processes. Icarus, 155, 18–37. Tosi F., Capria M. T., De Sanctis M. C., et al. (2014). Thermal measurements of dark and bright surface features on Vesta as derived from Dawn/VIR. Icarus, 240, 36–57. Urquhart M. L. and Jakosky B. M. (1997). Lunar thermal emis- sion and remote determination of surface properties. Journal of Geophysical Research, 102, 10959–10970. Vaniman D., Reddy R., Heiken G., et al. (1991). The Lunar En- vironment. In Lunat Sourcebook (Heiken G. H., Vaniman D., and French K. L., editors), pp. 27–60. Cambridge University Press., Cambridge, UK. Vasavada A. R., Bandfield J. L., Greenhagen B. T., et al. (2012). Lunar equatorial surface temperatures and regolith properties from the Diviner Lunar Radiometer Experiment. Journal of Geophysical Research, 117. Vasavada A. R., Paige D. A., and Wood S. E. (1999). Near-Surface Temperatures on Mercury and the Moon and the Stability of Polar Ice Deposits. Icarus, 141, 179–193. Vernazza P., Delbo M., King P. L., et al. (2012). High surface porosity as the origin of emissivity features in asteroid spectra. Icarus, 221, 1162–1172. Veverka J., Farquhar B., Robinson M., et al. (2001). The landing of the NEAR-Shoemaker spacecraft on asteroid 433 Eros. Nature, 413, 390–393. Vilenius E., Kiss C., Mommert M., et al. (2012). ”TNOs are Cool”: A survey of the trans-Neptunian region. VI. Her- schel/PACS observations and thermal modeling of 19 classical Kuiper belt objects. Astronomy & Astrophysics, 541, A94. Viles H., Ehlmann B., Wilson C. F., et al. (2010). Simulating weathering of basalt on Mars and Earth by thermal cycling. Geophysical Research Letters, 37, 18201. Vincent J.-B., Besse S., Marchi S., et al. (2012). Physical prop- erties of craters on asteroid (21) Lutetia. Planetary and Space Science, 66, 79–86. Vokrouhlick´y D., Bottke W. F., Chesley S. R., et al. (2015). The Yarkovsky and YORP effects. In Asteroids IV. University of Arizona Press, Tucson, AZ. Vokrouhlick´y D. and Nesvorn´y D. (2012). Sun-grazing orbit of the unusual near-Earth object 2004 LG. Astronomy & Astro- physics, 541, A109. Vokrouhlick´y D., Nesvorn´y D., and Bottke W. F. (2003). The vec- tor alignments of asteroid spins by thermal torques. Nature, 425, 147–151. Walsh K. J., Richardson D. C., and Michel P. (2008). Rotational breakup as the origin of small binary asteroids. Nature, 454, 188–191. Wesselink A. J. (1948a). Heat conductivity and nature of the lunar surface material. Bulletin of the Astronomical Institutes of the Netherlands, 10, 351–363. Wesselink A. J. (1948b). Heat conductivity and nature of the lunar surface material. Bulletin of the Astronomical Institutes of the Netherlands, 10, 351–363. Winter D. F. and Krupp J. A. (1971). Directional characteristics of infrared emission from the moon. The moon, 2, 279–292. Wolters S. D. and Green S. F. (2009). Investigation of systematic bias in radiometric diameter determination of near-Earth aster- oids: the night emission simulated thermal model (NESTM). Monthly Notices of the Royal Astronomical Society, 400, 204– 218. Wolters S. D., Rozitis B., Duddy S. R., et al. (2011). Physical characterization of low delta-V asteroid (175706) 1996 FG3. Monthly Notices of the Royal Astronomical Society, 418, 1246– 1257. Yano H., Kubota T., Miyamoto H., et al. (2006). Touchdown of the Hayabusa Spacecraft at the Muses Sea on Itokawa. Science, 312, 1350–1353. This 2-column preprint was prepared with the AAS LATEX macros v5.2. 23
1711.06567	1	1711	2017-11-17T14:47:38	RV-detected Kepler-multi Analogs Exhibit Intra-system Mass Uniformity	[ "astro-ph.EP" ]	Recent work has shown that the planets in the $Kepler$ Mission's population of multi-transiting systems show surprising uniformity in both mass and radius. In this brief note, I show that this intra-system mass uniformity extends to multiple-planet systems detected with the Doppler velocity technique, thereby avoiding possible biases associated with masses determined by transit timing. I also show that intra-system mass uniformity breaks down when a system contains one or more giant planets.	astro-ph.EP	astro-ph	RV-detected Kepler -multi Analogs Exhibit Intra-system Mass Uniformity Songhu Wang1, 2 1Department of Astronomy, Yale University, New Haven, CT 06511 251 Pegasi b Fellow ABSTRACT Recent work has shown that the planets in the Kepler Mission's population of multi-transiting systems show sur- prising uniformity in both mass and radius. In this brief note, I show that this intra-system mass uniformity extends to multiple-planet systems detected with the Doppler velocity technique, thereby avoiding possible biases associated with masses determined by transit timing. I also show that intra-system mass uniformity breaks down when a system contains one or more giant planets. Keywords: planets and satellites: formation -- planets and satellites: general 7 1 0 2 v o N 7 1 . ] P E h p - o r t s a [ 1 v 7 6 5 6 0 . 1 1 7 1 : v i X r a Corresponding author: Songhu Wang [email protected] 2 The Kepler Mission's most startling discovery was the realization that ∼ 50% of Solar-type stars are accompanied by short-period planets with MP (cid:46) 30 M⊕ (Lissauer et al. 2014). These worlds, which often appear in multi-transiting configurations, are quite ordered, displaying low mutual inclinations, low eccentricities, and low spin-orbital misalign- ments (Winn & Fabrycky 2015). In aggregate, however, they generate a highly scattered mass-radius relation. (Weiss & Marcy 2014). A recent article by Weiss et al. (2017) emphasized that the Kepler multi-transiting systems exhibit a surprising degree of intra-system uniformity in the planetary radii. Millholland et al. (2017) extended this statistically significant "peas-in-a-pod" trend to intra-system masses by drawing on the selection of Kepler planets for which transit timing variations (TTVs) can determine masses. Such "TTV" planets, however, are usually near mean-motion resonance (e.g., Lithwick et al. 2012), leading naturally to concern over whether a finding that holds for a potentially special class of planets can be generalized. Indeed, the discrepancy between RV- and TTV-determined masses suggests there may be potential physical differences that are correlated with proximity to resonance (Steffen 2016; Mills & Mazeh 2017). The prospects for usefully extending Millholland et al. (2017)'s set of mass determinations are rather dim. A significant number of additional masses would require many precise Doppler velocity (RV) measurements, which are impractical for the faint Kepler targets. Several groups, however, have consistently achieved long-term 1 − 3 m s−1 Doppler precision on bright, stable stars (Mayor et al. 2009; Vogt et al. 2010; Howard et al. 2011). The extant catalog of Doppler-discovered multi-planet systems thus provides us with an independent sample of Kepler -multi analogs to assess the degree of intra-system mass uniformity among planets that are far from mean-motion resonance. We draw from https://exoplanetarchive.ipac.caltech.edu/, to establish a sample of the 29 known systems with at least three Doppler-detected planets. Two-planet systems are omitted because they often display large period ratios, and may be not co-planar. The set of systems, ordered by M sin(i)max, is shown in Figure 1, and divides naturally into two groups. In the first, the maximum planetary mass is below the core rapid gas-accretion threshold of ∼ 30 M⊕. These systems resemble the dominant planetary population in the Kepler census, with a prototypical example being HD 40307 (Mayor et al. 2009). The second class contains gas giants with planetary masses greater than ∼ 100 M⊕. A representative member of this group is 55 Cnc (Fischer et al. 2008). dispersion D as To quantify whether planets in a given system are preferentially correlated in mass, we define the intra-system mass where M = (1/N )(cid:80)Npl j=1 Mj. Figure 1 shows that D, as exhibited by the small-planet systems (vertical blue line) is a 3.2σ outlier in comparison to a Monte-Carlo control sample (yellow histogram) in which the number of planets in each system and the total number of systems were conserved, but the planets were randomly shuffled 50,000 times. By contrast, the frequent co-existence of small planets and gas giants in the second class produces significant intra-system mass dispersions. For this group, the real systems show no significant departure from random shuffling. The intra-system mass uniformity that characterizes the Kepler TTV sample thus extends to RV-detected Kepler- multi analogs, which contain sets of small close-in planets, but which avoid the TTV mass determination bias toward near-resonances. This result, together with those of Weiss et al. (2017) and Millholland et al. (2017), implies that planet formation within a given system is a coordinated process. One notes a similarity to the Solar System's Jovian satellites the small scale (∼ 0.01 AU) and to the galactic conformity on the large scale (∼ 1 Mpc, Weinmann et al. 2006). The oft-remarked diversity among exoplanets is perhaps better expressed as a diversity among exoplanetary systems. I thank the Heising-Simons Foundation for their generous support. I would also like to thank Greg Laughlin for inspirational conversations and editing. Additionally, I am grateful to Yutong Wu for improving the quality of the Figure, as well as to Sarah Millholland, Song Huang, Zhongyi Man, and Frank van den Bosch for useful discussion. (cid:115)(cid:80)Npl Nsys(cid:88) i=1 D = j=1(Mj − M )2 Npl − 1 , (1) 3 Figure 1. Left: Architectures of RV-detected planetary systems with at least 3 planets. The aggregate is ordered by the maximum M sin(i) in the system. The systems are divided into two groups. In the first group, Mmax (cid:46) 30 M⊕. These systems resemble the population in the Kepler census, and exhibit significant intra-system mass uniformity. The second group has systems with Mmax (cid:38) 100 M⊕. This group shows considerable intra-system dispersion. Right: Comparisons of the dispersion metric D between the real systems (vertical blue lines) and the control populations of 50,000 realizations of shuffled systems (yellow histograms). 4 REFERENCES Fischer, D. A., Marcy, G. W., Butler, R. P., et al. 2008, ApJ, 675, 790-801 Howard, A. W., Johnson, J. A., Marcy, G. W., et al. 2011, Mills, S. M., & Mazeh, T. 2017, ApJL, 839, L8 Steffen, J. H. 2016, MNRAS, 457, 4384 Vogt, S. S., Butler, R. P., Rivera, E. J., et al. 2010, ApJ, ApJ, 730, 10 723, 954 Lissauer, J. J., Dawson, R. I., & Tremaine, S. 2014, Nature, Weinmann, S. M., van den Bosch, F. C., Yang, X., & Mo, 513, 336 Lithwick, Y., Xie, J., & Wu, Y. 2012, ApJ, 761, 122 Mayor, M., Udry, S., Lovis, C., et al. 2009, A&A, 493, 639 Millholland, S., Wang, S., & Laughlin, G. 2017, ApJL, 849, L33 H. J. 2006, MNRAS, 366, 2 Weiss, L. M., & Marcy, G. W. 2014, ApJL, 783, L6 Weiss, L. M., Marcy, G. W., Petigura, E. A., et al. 2017, arXiv:1706.06204 Winn, J. N., & Fabrycky, D. C. 2015, ARA&A, 53, 409
1510.03549	1	1510	2015-10-13T06:46:43	Thermosphere and geomagnetic response to interplanetary coronal mass ejections observed by ACE and GRACE: Statistical results	[ "astro-ph.EP", "physics.space-ph" ]	For the period July 2003 to August 2010, the interplanetary coronal mass ejection (ICME) catalogue maintained by Richardson and Cane lists 106 Earth-directed events, which have been measured in-situ by plasma and field instruments onboard the ACE satellite. We present a statistical investigation of the Earth's thermospheric neutral density response by means of accelerometer measurements collected by the GRACE satellites, which are available for 104 ICMEs in the data set, and its relation to various geomagnetic indices and characteristic ICME parameters such as the impact speed, southward magnetic field strength (Bz). The majority of ICMEs causes a distinct density enhancement in the thermosphere, with up to a factor of eight compared to the pre-event level. We find high correlations between ICME Bz and thermospheric density enhancements (~0.9), while the correlation with the ICME impact speed is somewhat smaller (~0.7). The geomagnetic indices revealing the highest correlations are Dst and SYM-H (~0.9), the lowest correlations are obtained for kp and AE (~0.7), which show a nonlinear relation with the thermospheric density enhancements. Separating the response for the shock sheath region and the magnetic structure of the ICME, we find that the Dst and SYM-H reveal a tighter relation to the Bz minimum in the magnetic structure of the ICME, whereas the polar cap indices show higher correlations with the Bz minimum in the shock sheath region. Since the strength of the Bz component - either in the sheath or the magnetic structure of the ICME - is highly correlated (~0.9) with the neutral density enhancement, we discuss the possibility of satellite orbital decay estimates based on magnetic field measurements at L1, i.e. before the ICME hits the Earth's magnetosphere. This will further stimulate progress in space weather understanding and applications regarding satellite operations.	astro-ph.EP	astro-ph	JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1002/, Thermosphere and geomagnetic response to interplanetary coronal mass ejections observed by ACE and GRACE: Statistical results. S. Krauss,1 M. Temmer,2 A. Veronig,2 O. Baur1 and H. Lammer1 5 1 0 2 t c O 3 1 . ] P E h p - o r t s a [ 1 v 9 4 5 3 0 . 0 1 5 1 : v i X r a Corresponding author: S. Krauss, Space Research Institute, Austrian Academy of Sciences, Schmiedlstrasse 6, A-8042 Graz, Austria ([email protected]) 1Space Research Institute, Austrian Academy of Sciences, Schmiedlstr. 6, A-8042 Graz, Austria 2Institute of Physics, University of Graz, Universitatsplatz 5/II, A-8010 Graz, Austria. D R A F T October 15, 2018, 3:57am D R A F T X - 2 Abstract. KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES For the period July 2003 to August 2010, the interplanetary coronal mass ejection (ICME) catalogue maintained by Richardson and Cane lists 106 Earth-directed events, which have been measured in-situ by plasma and field instruments on -- board the ACE satellite. We present a statistical investigation of the Earth's thermospheric neutral density response by means of accelerometer measurements collected by the GRACE satellites, which are available for 104 ICMEs in the data set, and its relation to various geomag- netic indices and characteristic ICME parameters such as the impact speed (vmax), southward magnetic field strength (Bz). The majority of ICMEs causes a distinct density enhancement in the thermosphere, with up to a factor of eight compared to the pre -- event level. We find high correlations between ICME Bz and thermospheric density enhancements (≈ 0.9), while the correla- tion with the ICME impact speed is somewhat smaller (≈ 0.7). The geo- magnetic indices revealing the highest correlations are Dst and SYM-H (≈ 0.9), the lowest correlations are obtained for kp and AE (≈ 0.7), which show a nonlinear relation with the thermospheric density enhancements. Separat- ing the response for the shock sheath region and the magnetic structure of the ICME, we find that the Dst and SYM-H reveal a tighter relation to the Bz minimum in the magnetic structure of the ICME, whereas the polar cap indices show higher correlations with the Bz minimum in the shock sheath region. Since the strength of the Bz component -- either in the sheath or the magnetic structure of the ICME -- is highly correlated (≈ 0.9) with the neu- tral density enhancement, we discuss the possibility of satellite orbital de- D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 3 cay estimates based on magnetic field measurements at L1, i.e. before the ICME hits the Earth magnetosphere. These results are expected to further stimulate progress in space weather understanding and applications regard- ing satellite operations. D R A F T October 15, 2018, 3:57am D R A F T X - 4 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES 1. Introduction Since the beginning of the space age in the late 1950's, "space weather" is a hot topic. It deals with the space environment of the Earth's atmosphere from approximately 90 km altitude onwards extending all the way to the Sun. Strong disturbances of this space environment are primarily caused by solar flares, coronal mass ejections (CMEs) and associated solar energetic particles (SEPs). Owing to advances in solar research in the recent decades, today we know that CMEs cause the most comprehensive spectrum of space weather disturbances, including geomagnetic storms, aurorae, geomagnetically in- duced currents affecting electric power lines, heating of the outer atmosphere, and many more. CMEs are huge clouds of magnetized plasma propagating from the solar corona into interplanetary space with speeds ranging from a few hundred up to ≈ 3000 km s−1. CMEs often reveal a three-part structure [Illing and Hundhausen, 1985] (Fig. 1): the front (leading edge) with its shock-sheath region [Vourlidas et al., 2013], the void which comprises the magnetic structure of a CME (often with a magnetic flux rope inherited) and the bright core consisting of filament material. The occurrence rate of CMEs is closely related to the solar cycle, and varies from 0.3 per day at solar minimum to 4 -- 5 per day on average at solar maximum [e.g., St. Cyr et al., 2000; Gopalswamy, 2006]. Close to the Sun, the CME dynamics is governed by the driving Lorentz forces resulting from magnetic instabilities in the fields of active regions [Zhang et al., 2001; Vrsnak et al., 2007; Bein et al., 2012]. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 5 In interplanetary space, on the other hand, the CME dynamics is governed by the interaction of the CME with the surrounding solar wind flow, whereby fast CMEs are decelerated and slow CMEs are accelerated [e.g., Gopalswamy, 2000]. This interaction process with the collisionless plasma in the interplanetary space is described by the mag- netohydrodynamic equivalent of the aerodynamic drag force [Cargill, 2004; Vrsnak et al., 2013]. Thus, the transit time and the arrival speed of an Earth-directed CME -- which are both crucial parameters for space weather description and forecasting -- depend on both the initial speed and the state and interaction of the ambient solar wind plasma with the CME. Typical transit times range between 2 -- 5 days, depending on initial speed, mass and size as well as the speed and density of the surrounding solar wind plasma. The fastest CME events observed so far reached the Earth in less than one day [e.g., Liu et al., 2014; Temmer and Nitta, 2015]. Part of the solar wind energy supplied to the Earth's magnetosphere is continuously deposited into the thermosphere via particle precipitation and Joule heating. This energy supply can be enormously enhanced when a fast CME with a strong southward magnetic field component arrives at the Earth's atmosphere, causing a strong geomagnetic storm [Gosling et al., 1991]. The enhanced energy input from the solar wind to the magneto- sphere and its ultimate dissipation in the thermosphere causes heating and expansion of the Earth's thermosphere [Krauss et al., 2012]. During periods of high CME activity, short timescale (1 -- 2 days) variations in the thermosphere are driven by magnetospheric energy input rather than by changes in the solar extreme-ultraviolet (EUV) flux [e.g., Wilson et al., 2006; Guo et al., 2010; Krauss et al., 2012]. Knipp et al. [2004] have shown D R A F T October 15, 2018, 3:57am D R A F T X - 6 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES that in case of extreme events the geomagnetic power contributes two-thirds of the total solar power at the Earth's thermosphere, with the major contribution from Joule heating. The increase in the neutral density affects Earth-orbiting satellites in such a way that the drag force on the spacecraft is enhanced. The fact that some low-Earth orbiters (LEOs) are equipped with accelerometers allows the derivation of in-situ measurements of the current state of the Earth's thermospheric density. To this class of satellites belong the spacecraft of the projects CHAllening Mini satellite Payload [CHAMP; Reigber et al., 2002] and Gravity Recovery And Climate Experiment [GRACE; Tapley et al., 2004]. In this paper, we present a statistical study of the thermospheric response to a sample of 104 interplanetary CMEs (ICMEs); most of these events caused a significant increase in the neutral density. The period under investigation covers seven years during the max- imum and decaying phase of solar cycle 23 (7/2003 to 8/2010), including the "Halloween storms". Thermospheric density changes are derived from GRACE accelerometer mea- surements and compared with ICME properties measured by the Advanced Composition Explorer [ACE; Stone et al., 1998] and various geomagnetic storm indices. Due to the high correlations obtained, we were able to derive regression curve parameters, which we expect to be highly valuable for the improved estimates of future CME-related thermo- spheric density enhancements. 2. Data and Analysis For the present study we analyzed all ICME events during the time range July 13, 2003 until August 04, 2010 as listed in the catalogue of near-Earth interplanetary ICMEs maintained by Richardson & Cane (R&C) [Richardson and Cane, 2003, 2010]. During this period the R&C list contains 106 ICME events, comprising the arrival times of the D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 7 shock-sheath region, the start and end times of the magnetic structure (based on the plasma and magnetic field observations) and the minimum value of the geomagnetic dis- turbance storm time index (Dst) [Sugiura, 1964; Sugiura and Kamei, 1991] within the total ICME duration. For our statistical study we thoroughly analyzed the magnetic field characteristics of the ICMEs making use of measurements by the ACE satellite located at the Lagrange point L1. The response of the Earth's thermosphere in terms of changes in neutral density are derived from acceleration measurements provided by the GRACE mission, which has been orbiting the Earth at an altitude of ∼450-500 km since 2002. Furthermore, we used various geomagnetic indices derived from terrestrial magnetic field measurements collected at ground-based observatories. 2.1. Magnetic field measurements The ICME kinematics and its associated magnetic field were derived from in-situ mea- surements recorded by instruments aboard the ACE spacecraft -- in particular the Solar Wind Electron, Proton, and Alpha Monitor (SWEPAM) [McComas et al., 1998] and the magnetometer (MAG) [Smith et al., 1998]. We used level-2 data of the proton speed (solar wind bulk speed) and the magnetic field (Bz component in Geocentric Solar Ecliptic, GSE, coordinates) with a time resolution of 4 min (OMNI database) [King and Papitashvili, 2004]. In GSE coordinates the z-component is oriented orthogonal to the ecliptic, from which we derived the southward component of the magnetic field. We identified mini- mum values in the Bz component separately for the shock-sheath region and the magnetic structure of the ICME. Bad or missing data were linearly interpolated. Time information, such as start and end of the disturbance, were extracted from the R&C list. D R A F T October 15, 2018, 3:57am D R A F T X - 8 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Figure 2 shows the temporal evolution of the Bz component, exemplary for an ICME event in May 2010. The shock-sheath region covers a period of strongly varying mag- netic field (Fig. 2, yellow area), with a minimum occurring on May 28 at approximately 11:00 UT. The second minimum occurs about 19 hours later when the magnetic structure of the ICME arrived (Fig. 2, green area). During this period the Bz component shows a smooth behavior and changing orientation, indicative of field rotation. The existence of a shock wave depends on the speed of the ICME relative to the ambient solar wind speed. For 82 of the ICME events analyzed in this study, a shock-sheath region could be identified in the in-situ plasma and field data. 2.2. Thermospheric neutral density In order to gain knowledge about the response of the Earth's thermosphere to ICME events we analyzed accelerometer data collected by the US-German GRACE project. The GRACE satellites were launched in March 2002 and injected into a low-altitude (initially 505 km), near-circular (eccentricity ≈ 0.001), and near-polar (inclination ≈ 89◦) orbit around the Earth. The GRACE constellation consists of two identical spacecraft following each other in the same orbit, separated by about 220 km. GRACE is a dedicated gravity field mission, and therefore both satellites are equipped with accelerometers; the GRACE accelerometers were manufactured by the Office Na- tional d'Etudes et de R´echerches Aerospatials (ONERA) and have an accuracy of ≈ 1 × 10−10m s−2. As far as gravity field recovery is concerned, accelerometer mea- surements are used to get rid of the non-gravitational forces acting on the satellites. Here, we use these observations to compute neutral densities (ρ) along the trajectory of D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 9 the spacecraft. The transition from the along track accelerometer measurements (ax) to atmospheric neutral densities can be established by ρ = − 2axm CDAv2 x , (1) where m indicates the satellite mass, CD the highly complex and variable drag coeffi- cient, A the effective cross sectional area and vx the satellite velocity in the along track axis. For a detailed description of the technique we refer the reader to Sutton [2008]; Doornbos et al. [2009]; Krauss [2013]. Due to ellipticity, the true GRACE orbit implies satellite altitude variations of about 50 km per revolution. For this reason, all densities shown in this study are normalized to an average altitude of 490 km (ρ490) using the empirical thermosphere model by Bowman et al. [2008] (ρJB, ρJB490) according to ρ490 = ρJB490 ρJB ρ. (2) We smoothed the density time series with a 5-min moving average filter to avoid single peaks biasing the results. As an example, Fig. 3 depicts neutral densities in a latitude-time plot covering a period of ten days in July 2004; during this period three ICMEs hit the Earth. Compared to the pre -- event level of ≈ 0.3 × 10−12 kg m−3, each ICME increased the neutral density by a factor of up to five. The perturbations started at high latitudes and thereafter propagated towards the equator within a few hours. These phenomena are known as travelling atmospheric disturbances and occur at high latitudes due to sudden energy and momentum deposition, mostly in the form of Joule heating [e.g., Vasyli¯unas and Song, D R A F T October 15, 2018, 3:57am D R A F T X - 10 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES 2005; Bruinsma and Forbes, 2008]. For most of the analyzed ICME events, we found the maximum geomagnetic disturbance at high latitudes near the auroral region (with a latitude root mean square value of 73.2◦). 2.3. Geomagnetic indices Geomagnetic indices play an essential role in the space weather system, describing the response of the Earth's magnetosphere to effects of the changing solar wind. In the course of the past 80 years various indices were developed, which are monitoring the geomagnetic activity on different latitudes and time scales. Figure 4 gives an overview of the global distribution of geomagnetic observatories, which provide the measurements for the different geomagnetic indices. For this study we analyzed ten different geomagnetic indices. Four of them (aa [Mayaud, 1971], am [Mayaud, 1968] ap [Bartels and Veldkamp, 1954] and kp [Bartels, 1949]) are deduced from the quasi logarithmic K-index [Bartels et al., 1939] -- a measure for the variations in the two horizontal field components. The hourly Dst index (derived from observations made by four near-equatorial observatories) describes the global symmetrical geomagnetic perturbations of the ring current -- a current flowing westward around the Earth. Further we analyzed the auroral electrojet index (AE) [Davis and Sugiura, 1966], which is based on observations in the auroral zone and monitors the horizontal field in the east -- west ionospheric current; it is constructed from the AU and AL indices, which rep- resent the largest and lowest deviation from a base value at each station, respectively (AE = AU − AL). The polar cap index (PC) [Troshichev et al., 1988] represents geomag- netic disturbances in the polar cap regions due to ionospheric and field -- aligned currents D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 11 [Vassiliadis et al., 1996]; measurements are performed by the stations Thule and Vos- tok, therefore the index can be divided into a northern (PCN) and southern (PCS) part, respectively. Finally, we included the SYM -- H index which describes the longitudinal symmetric disturbance field in mid-latitudes for the horizontal component. Basically it is the same as the hourly Dst index and describes the same axial symmetric disturbances in the horizontal component [Iyemori et al., 2010]. The main difference is in the temporal resolution of the indices (1h versus 1min), i.e. short-term changes in solar wind parameters are reflected differently in the two indices. For more information concerning geomagnetic indices the reader is referred to Menvielle et al. [2011]. 2.4. Analysis The time period July 2003 to August 2010 was chosen because of the availability of GRACE calibration data [Bruinsma et al., 2007]. During this time span the R&C list contains 106 ICME events. For two of them, no GRACE acceleration measurements are available. The remaining set of 104 ICMEs is studied in this paper. Starting with the disturbance time of the ICME (taken from the R&C list) we defined a time window of 36 hours over which we extracted the evolution of the various geomagnetic indices, the neutral densities (from GRACE) and the magnetic field measurements (from ACE). For every single ICME event, we derived from each parameter the peak value within the predefined time window. We thoroughly checked the data to ensure that the correct peak value is assigned to each ICME event. This is particularly important for time intervals containing more than one ICME event as well as for complex ICME structures. To account for variations in the background thermospheric density (caused by the non- D R A F T October 15, 2018, 3:57am D R A F T X - 12 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES disturbed solar activity) we calculated the absolute density increase, i.e. the difference ∆ρ490 between the peak density measured during the event minus the pre -- event level, as well as the relative density increase in percent; for the correlation studies the ∆ρ490 values were used. Exemplarily for two ICME events, Fig. 5 shows the evolution of the individual parameters. Additionally, each panel includes the progress of the neutral density in terms of an envelope function (red curves) as well as the detected peak values (blue circles). The envelope function was created such that it comprises the maximum density per satellite revolution (Fig. 5, a1, d1). Figure 5 reveals that the temporal evolution of both the geomagnetic indices and the Bz component of the magnetic field is in good agreement with the neutral density. As far as the Dst index and the SYM -- H index (a3,b3; d3,e3) is concerned, the different phases of the geomagnetic storm become clearly visible: a sudden decrease of the indices during the main phase of the storm, followed by a much slower recovery phase. Regarding the absolute values, over the entire range we derive slightly higher values in the SYM -- H index than in the Dst index and in some cases a small offset between these two indices (d3, e3). This behaviour can be explained by the different methods in the baseline preparations for the two indices [Wanliss and Showalter, 2006]. Notice that the a -- indices (row c,f) exhibit a rather rough structure, caused by their low temporal resolution (3h). The AE index shows disturbances throughout the whole ICME events and even beyond that (b2). Comparing the two polar cap indices (b1,e1) we see larger differences during the most disturbed period than before and after the event, implying different conditions in the polar cap regions [Lukianova et al., 2002]. In general, the majority of the analyzed ICME events shows a significant density increase in D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 13 the atmosphere compared to the density level before the perturbation (Fig. 6). The largest relative increase occurred for two events in 2005 (May 15 and August 24); in both cases the density rose by about 750% compared to the pre -- event level. The highest absolute increases in neutral density occurred during the Halloween storms in 2003 (October 29 -- 30 and November 20) with ∆ρ490 > 4 × 10−12kg m−3. 3. Results In a first step, we compare the in-situ measurements by the ACE and GRACE spacecraft in terms of magnetic field strength (Bz in GSE) and thermospheric neutral density (∆ρ490), respectively. Additionally, we use of the maximum speed (vmax), measured over the entire ICME structure as given in the R&C list. Figure 7 shows scatter plots of the thermospheric density increase ∆ρ490 versus Bz, vmax, as well as two combinations of these quantities. On the one hand a proxy for the convective electric field [Burton et al., 1975] represented by E ≈ vmax × Bz, (3) on the other hand a proxy for the energy input via the Poynting flux into the magneto- sphere by magnetic reconnection [Akasofu, 1981] defined as S = E × B ≈ vmax × B2 z . (4) From the regression analysis we derive the lowest correlation coefficient (cc = 0.69) for the maximum solar wind speed during the disturbance period (vmax), inferring that this is not the primary factor for effective driving of geomagnetic storms. In fact, very high impact velocities (> 1500 km s−1) are only present for two of the 104 events. Irrespective of whether a density response occurred or not, the majority of ICMEs has velocities in the D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 14 range between 400 and 800 km s−1 which is of the statistical average [e.g., Yurchyshyn et al., 2005]. In contrast to vmax, the quantities Bz, E and S show a much higher correlation with the thermospheric density with cc = −0.92, cc = −0.88 and cc = −0.81, respectively. The latter correlation coefficient is particularly influenced by one extreme ICME event; excluding the gigantic Halloween ICME (October 29, 2003) from the calculations leads to a significantly increased correlation coefficient between the ∆ρ490 and S of cc = −0.90. Fig. 7 (top left) also indicates that ICME events with hardly any thermospheric response are associated with a small Bz component (<10 nT). In the next step we analyzed the correlation between the neutral density increase ∆ρ490 and the different geomagnetic indices. Figure 8 provides an overview in terms of scatter plots. It can clearly be seen that most of the geomagnetic indices show an excellent correlation with the thermospheric density response. The highest correlations are derived for the indices aa (cc = 0.92), am (cc = 0.92) and Dst (cc = −0.91) together with the SYM -- H index (cc = −0.91). The reason for the slightly lower correlation of the ap index might be due to the worse station distribution (13 in the northern hemisphere, 2 in the southern hemisphere). Significantly lower correlations are present for the kp (cc = 0.71) and AE index (cc = 0.77). In the first case, this behaviour is not surprising since it is well known that the kp index is not linearly correlated with the geomagnetic activity [Bartels and Veldkamp, 1954]; for this reason the planetary ap index was established in 1954. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 15 Between the density increase and the AE index a non-linear relationship can be ob- served. Analyzing Fig. 8 (top, right) we see different behaviours depending on the geo- magnetic activity level. Only minor density increases are visible for AE values up to 1000 nT. Above this value, we recognize larger scattering and some indication of saturation. The lower correlation between the AE index and the thermospheric densities is in agree- ment with previous findings [Akasofu et al., 1983; Baumjohann, 1986; Chen et al., 2014]. These authors revealed some limitations concerning the measured quantity (H -- field), the existence of longitudinal gaps and the small latitudinal range of the contributing obser- vatories. Compared with the other indices the correlations between the density increases and the polar cap indices (PCN, PCS) are smaller. However, it should be noted that for some ICME events no PCS data was available which excludes some of the strongest events ("Halloween storms" in October/November 2003). Table 1 summarizes all determined correlation coefficients with regard to the atmo- spheric density ∆ρ490 and magnetic field component Bz. In general, the correlations between Bz and the geomagnetic indices are very similar to those of the neutral density. Again the highest correlations are found for the aa, am, Dst and the SYM -- H index. Like- wise, the more exponential relation with the AE and kp indices and the slightly poorer correlations with the polar cap indices. Analyzing the minimum in the Bz separately for the shock-sheath and the magnetic structure of the ICME we may deduce some additional information. For some events, no shocks are reported, and the leading edge of the ICME might be stated as the estimated arrival time of the disturbance. See Richardson and Cane [2003] for additional information D R A F T October 15, 2018, 3:57am D R A F T X - 16 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES on the methods used to identify ICMEs. For 82 out of the 104 analyzed events it was possible to detect and distinguish between the shock-sheath region and the magnetic structure of the specific ICME. For this subset of events we identified the peak values in the neutral density and the various geomagnetic indices separately for the shock sheath region and the magnetic structure of the ICME. As a result, Fig. 9 illustrates the minimum Bz component, either found in the shock-sheath region (red squares) or in the magnetic structure part (blue squares) of the ICME, against the neutral density increase and the different geomagnetic indices, for the particular region. For the a -- indices and the AE index no significant differences in the regression line be- tween the two specific regions are visible. However, large differences are present for the polar cap indices (PCN, PCS), indicating that both correlate better with disturbances triggered at the time of arrival of the shock sheath region (red regression line). In con- trast to that, the hourly Dst index as well as the SYM -- H index show better correlations when the minimum Bz component is detected in the magnetic structure of the ICME. Finally, we examine the time of occurrence of the neutral density and the the peak for the geomagnetic indices. We find the peak times coincide significantly better with the times extracted for the minimum Bz value in the magnetic structure rather than for the Bz value in the shock-sheath region (ratio: 4:1). The complete list of the selected peak values of all quantities discussed in this study is available as supporting information. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 17 4. Discussion Based on the ICME catalogue maintained by R&C we have analyzed 104 ICME events for the period July 2003 to August 2010. We studied the relationship between geomagnetic disturbances and thermospheric densities derived from GRACE accelerometer measure- ments. During periods of geomagnetic disturbances the maximum of the signal was mainly found at high latitudes in the auroral zone (see, e.g., case study by Bruinsma et al. [2006]). By correlating magnetic field characteristics of the ICMEs observed by the ACE satellite we were able to derive linear regression parameters between the different measured quan- tities. We found that the majority of these ICMEs, cause a significant increase in the Earth's thermospheric neutral density. From the regression analysis, it turned out that the ICME parameter vmax has a significantly lower correlation with the thermospheric density than the Bz component or the combined parameters E and S, which are directly related to the energy input into the system. It is worth noting that the highest corre- lation coefficient is obtained for Bz (cc ≈ 0.9), whereas the estimates for E and S do not yield improved correlations. Consequently, it is conceivable to obtain an estimate of the forthcoming maximum neutral density increase using in-situ observations of ICME characteristics by appropriate in-situ measurements from satellites at the Lagrange point L1 before the ICME impacts Earth's magnetosphere. Using real-time in-situ data, this would result in a lead time of some tens of minutes (depending on the actual speed of the ICME event). Furthermore, we can conform previous findings by Liu et al. [2010], that all ICMEs causing a significantly density enhancement have shown a strong negative Bz component. D R A F T October 15, 2018, 3:57am D R A F T X - 18 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Regarding the various geomagnetic indices we obtained high correlations with the ther- mospheric density as well as the Bz component; especially the SYM -- H index, hourly Dst index, aa index and am index showed a good accordance (cc ≈ 0.9). However, it is worth noting that, the a -- indices have a lower temporal resolution of three hours, which might lead to limitations in detailed space weather monitoring [Menvielle et al., 2011]. No linear relationship with the geomagnetic activity could be established for the AE and kp index. Subsequent investigations concentrated on the separation of the magnetic field obser- vations Bz in the shock -- sheath region and magnetic structure of the ICME. We found that the hourly Dst index and the SYM -- H index correlate better with the Bz component when the peak values are extracted from the magnetic structure part of the ICME. Con- trary to that, the polar cap indices (PCN, PCS) showed higher correlations with the Bz minimum located in shock-sheath region. From a temporal view the occurrence times of the peak values in the neutral density and the geomagnetic indices coincide better (75% of the cases) with the time of the minimum Bz component in the magnetic structure of the ICME. However, strong perturbations from the shock-sheath region may cause an additional enhancement in the atmospheric neutral density [Yue and Zong, 2011]. Finally, an important field of application of space weather is the induced satellite orbit decay due to variations in the atmospheric density. By using different parameter types (CD, A ∆ρ) elaborated in the present study we determined the orbit decay rate in terms of changes in the mean semi -- major axis following Chen et al. [2012]. Figure 10 shows exemplarily, for the ICME event on November 20 in 2003, variations of the GRACE semi -- major axis. The middle panel illustrates the total orbit decay over the entire event, whereas the bottom panel shows the orbit decay rate per day. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 19 Comparing the evolution of the neutral density (Fig. 10, top) and the orbit decay rate per day (Fig. 10, bottom) it becomes apparent that these quantities reveal an almost one-to-one relation. As a result, the distinct linear relationships derived between thermal density enhancements and the ICME Bz component can in principle be used to predict the orbital decay rates from the ICME measurements at L1 upstream of Earth. An evaluation of this approach will be the topic of a forthcoming study. Supporting data are included as a table in an SI file; alternatively the data may be obtained from S.K. (email: sandro.krauss[at]oeaw.ac.at). Acknowledgments. M.T. and A.V gratefully acknowledge the support from the Aus- trian Science Fund: project nos. FWF V195-N16 and FWF P27292-N20. H.L. acknowl- edges support from the FWF NFN subproject S11607-N16 'Particle/Radiative Interac- tions with Upper Atmospheres of Planetary Bodies Under Extreme Stellar Conditions'. S.K. thanks Sean Bruinsma for his support with the GRACE accelerometer calibration. References Akasofu, S.I., Energy coupling between the solar wind and the magnetosphere. Space Science Reviews, 28, 121-190, 1981. Akasofu, S.I., Ahn, B.H., Kamide, Y., Allen, J.H., A note on the accuracy of the auroral electrojet indices, J. Geophys. Res., 88, A7, 5769-5772, 1983. Bartels, J., Heck, N.H., Johnston, H.F., The three -hour-range index measuring geomag- netic activity. J. Geophys. Res., 44 (4), 411-454, 1939. Bartels, J., The standardized index, Ks, and the planetary index, Kp. IATME Bulletin, 12B, 97, 1949. D R A F T October 15, 2018, 3:57am D R A F T X - 20 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Bartels, J. and Veldkamp, J. International data on magnetic disturbances, fourth quarter, 1953. J. Geophys Res, 59, 295, 1954. Merits and limitations of the use of geomagnetic indices in solar wind-magnetosphere coupling studies. Solar Wind -- Magnetosphere Coupling, edited by Y. Kamida nad J.A. Slavin, 3 -- 15, Terra Sci., Tokyo, 1986. Bein, B.M., Berkebile-Stoiser, S., Veronig, A.M., Temmer, M., Muhr, N., Kienreich, I., Utz, D., Vrsnak, B. Impulsive Acceleration of Coronal Mass Ejections. I. Statistics and Coronal Mass Ejection Source Region Characteristics. The Astrophysical Journal, 738, 191, doi: 10.1088/0004-637X/738/2/191, 2012. Bowman, B.R., Tobiska, W.K., Marcos, F.A., Huang, C.Y., Lin, C.S., Burke, W.J., A New Empirical Thermospheric Density Model JB2008 Using New Solar and Geomagnetic Indices, AIAA/AAS Astrodynamics Specialist Conference, AIAA 2008-6438, 2008b. Bruinsma, S., Forbes, J.M., Nerem, R.S., Zhang, X. Thermosphere density response to the 20 -- 21 November 2003 solar and geomagnetic storm from CHAMP and GRACE accelerometer data. J. Geophys. Res., 111, 1 -- 14, 2006. Bruinsma, S., Biancale, R., Perosanz, F., Calibration parameters of the CHAMP and GRACE accelerometers. IUGG XXIV General Assembly, Perugia, 2007. Bruinsma, S., Forbes, J.M., Properties of traveling atmospheric disturbances (TADs) inferred from CHAMP accelerometer observations. Advances in Space Research, 3, Issue 3, p. 369-376, 2008. Burton, R.K., McPherron, R.L., Russell, C.T., An Empirical Relationship Between Inter- planetary Conditions and Dst. J. Geophys. Res., 80 (31), 4204-4214, 1975. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 21 Cargill, P. J., On the Aerodynamic Drag Force Acting on Interplanetary Coronal Mass Ejections. Solar Physics, 221, 135-149, doi: 10.1023/B:SOLA.0000033366.10725.a2, 2004. Chen, G., Xu, J., Wang, W., Lei, J., Burns, A.G., A comparison of the effects if CIR- and CME -- induced geomagnetic activity on thermospheric densities and spacecraft orbits: Case studies. J. Geophys. Res. Space Physics, 117, A08315, doi: 10.1029/2012JA017782, 2012. Chen, G., Xu, J., Wang, W., Burns, A.G., A comparison of the effects if CIR- and CME -- induced geomagnetic activity on thermospheric densities and spacecraft orbits: Statistical studies. J. Geophys. Res. Space Physics, 119, 7928 -- 7939, doi: 10.1002/2014JA019831, 2014. Cook, G., Satellite drag coefficients. Planetary and Space Science, 13, 929-946, 1965. Davis, T.N., and Sugiura, M., Auroral electroject activity index AE and its universal time variations. J Geophys Res., 71, 785?801, 1966. Doornbos, E.,Forster, M.,Fritsche, B.,Helleputte, T.V.,Ijssel, J.V.D.,Koppenwallner, W., Luhr, H., Rees, D., Visser, P., Air density models derived from multi-satellite drag observations,Technical Report,DEOS TU Delft,317, 2009. Gopalswamy, N., Lara, A., Lepping, R. P., Kaiser, M. L., Berdichevsky, D., St. Cyr, O. C., Interplanetary acceleration of coronal mass ejections. Geophys. Res. Lett., 27, 145-148, doi: 10.1029/1999GL003639, 2000. Gopalswamy, N., Properties of Interplanetary Coronal Mass Ejections, Space Science Reviews, 124, 145-168, doi: 10.1007/s11214-006-9102-1, 2006 D R A F T October 15, 2018, 3:57am D R A F T X - 22 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Gosling, J.T., McComas, D.J., Phillips, J.L., Bame, S.J., Geomagnetic activity associated with Earth passage of interplanetary shock disturbances and coronal mass ejections. J. Geophys. Res. Space Physics, 96, A5, 7831 -- 7839, doi: 10.1029/91JA00316, 1991. Guo, J., Feng, X., Forbes, J.M., Lei, J., Zhang, J., On the relationship between ther- mosphere density and solar wind parameters during intense geomagnetic storms. J. Geophys. Res., 115, A12335, 2010. Illing, R.M.E., Hundhausen, A.J., Observation of a coronal transient from 1.2 to 6 solar radii. J. Geophys. Res. 90: doi: 10.1029/JA090iA01p00275, 1985. Iyemori, T., Takeda, M., Nose M., Odagi Y., Toh, H., Mid-latitude Geomagnetic Indices "ASY" and "SYM" for 2009 (Provisional), Data Anal. Cent. for Geomagn. and Space Magn., Faculty of Sci., Kyoto Univ., Kyoto, Japan. Kamide, Y., Akasofu, S.I., Notes on the auroral electrojet indices, Reviews of Geophysics and Space Physics, 21, 7, 1647-1656, doi: , 1983. Kilpua, E.K.J., Isavnin, A., Vourlidas, A., Koskinen, H.E.J., Rodriguez, L., On the re- lationship between interplanetary coronal mass ejections and magnetic clouds. Ann. Geophys., 31, 1251-1265, 2013. King, J.H.,Papitashvili, N.E., Solar wind spatial scales in and comparisons of hourly Wind and ACE plasma and magnetic field data, J. Geophys. Res., Vol. 110, No. A2, A02209, 10.1029/2004JA010804, 2004. Knipp, D. J., Tobiska, W. K., Emery, B. A., Direct and Indirect Thermospheric Heating Sources for Solar Cycles 21-23. Solar Physics, 224, 495-505, doi: 10.1007/s11207-005- 6393-4, 2004. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 23 Krauss, S., Fichtinger B., Lammer, H., Hausleitner, W., Kulikov, Y.N., Ribas, I., Shema- tovich, V., Bisikalo, D., Lichtnegger, H.I.M., Zaqarashvili, T.V., Khodachenko, M.L., Hanslmeier, A., Solar flares as proxy for young Sun: satellite observed thermosphere response to an X17.2 flare of Earth's upper atmosphere. Ann. Geophys., 30, 1129-1141, doi: 10.5194 \angeo-30-1129-2012, 2012. Krauss, S., Response of the Earth's thermosphere during extreme solar events. A con- tribution of satellite observations to atmospheric evolution studies. PhD. thesis, Graz University of Technology, 2013. Liu, R., Luhr, H., Ma, S.Y., Storm-time related mass density anomalies in the polar cap as observed by CHAMP. Ann. Geophys., 28, 165-180, 2010. Liu, Y.D., Yang, Z., Wang, R., Luhmann, J.G., Richardson, J.D., Lugaz, N., Sun-to-Earth Characteristics of Two Coronal Mass Ejections Interacting near 1 AU: Formation of a Complex Ejecta and Generation of a Two-Step Geomagnetic Storm. The Astrophysical Journal Letters, 793:L41, doi: 10.1088/2041-8205/793/2/L41, 2014. Lukianova, R., Troshichev O., The polar cap magnetic activity indices in the southern (PCS) and northern (PCN) polar caps: Consitency and discrepancy, Geophys. Res. Lett., 29, 18, 1879, doi: 10.1029/2002GL015179, 2002. Mayaud, P.N., Indices Kn, Ks, Km, 1964?1967, Centre National de la Recherche Scien- tifique, Paris, 156p, 1968. Mayaud, P.N., Une mesure plan´etaire d'activit´e magn´etique bas´ee sur deux observatoires antipodaux. Ann. Geophys., 27, 67, 1971. McComas, D.J., Bame, S.J., Barker, P., Feldman, W.C., Phillips, J.L., Riley, P., Grif- fee, J.W., Solar Wind Electron Proton Alpha Monitor (SWEPAM) for the Advanced D R A F T October 15, 2018, 3:57am D R A F T X - 24 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Composition Explorer. Space Science. Rev., 86, 563?612, 1998. Menvielle, M., Iyemori, T., Marchaudon, A., Nos´e, M. Geomagnetic indices in Geomag- netic Observations and Models. Edited by M. Mandea nad M. Korte, IAGA Special Sopron Book Series, Vol 5., Springer+Business Media 183-228, ISBN: 978-90-481-9857- 3, doi 10.1007/978-90-481-9858-0, 2011. Reigber, C., Luhr, H., Schwintzer P., CHAMP mission status. Advances in Space Research, 30(2), 129-134, 2002. Richardson, I.G., Cane, H.V., Interplanetary coronal mass ejections in the near-Earth so- lar wind during 1996-2002. J. Geophys. Res., 108,A4,1156, doi:10.1029/2002JA009817, 2003. Richardson, I.G., Cane, H.V., Near-Earth Interplanetary Coronal Mass Ejections During Solar Cycle 23 (1996 -- 2009): Catalog and Summary of Properties. Solar Physics,264,1, 189-237,2010. Sentman, L., Free molecule flow theory and its application to the determination of aero- dynamic forces, Technical Report, 1961. Smith, C., L'Heureux, J., Ness, N., Acuna, M., Burlaga, L., Scheifele, J., The ACE Magnetic Fields Experiment, Space Science, Rev., 86, 613?632, doi:10.1023/A:1005092216668, 1998. St. Cyr, O.C., Howard, R.A., Sheeley, N.R., Plunkett, S.P., Michels, D.J., Paswaters, S.E., Koomen, M.J., Simnett, G.M., Thompson, B.J., Gurman, J.B., Schwenn, R.,Webb, D.F., Hildner, E., Lamy, P.L., Properties of coronal mass ejections: SOHO LASCO observations from January 1996 to June 1998, J. Geophys. Res., 105, 18169?18186, doi:10.1029/1999JA000381, 2000. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 25 Stone, E.C., Frandsen, A.M., Mewaldt, R.A., Christian E.R., Margolies, D., Ormes, J.F.,Snow, F., The Advanced Composition Explorer, Space Science Reviews, 86, 1-4, pp. 1 -- 22, doi: 10.1023/A:1005082526237, 1998. Sugiura, M. Hourly values of equatorial Dst for the IGY., Annals of International Geo- physics Year, 35, 9,. Pergamon Press, Oxford, 1964. Sugiura, M. and Kamei, T., Equatorial Dst index 1957 -- 1986 IAGA Bulletin No. 40, 1991. Sutton, E.K., Effects of Solar Disturbances on the Thermosphere Densities and Winds from CHAMP and GRACE Satellite Accelerometer Data. PhD. thesis, Department of Aerospace Engineering Sciences, University of Colorado, 2008. Tapley, B.D., Bettadpur, S., Watkins, M., Reigber, C., The Gravity Recovery and Climate Experiment: Mission Overview and Early Results, Geophysical Research Letters, 31, 9, doi:10.1029/2004GL019920, 2004. Temmer, M., Veronig, A.M., Vrsnak, B., Ryb´ak, J., Gomory, P., Stoiser, S., Marici´c, D., Acceleration in Fast Halo CMEs and Synchronized Flare HXR Bursts. The Astrophys- ical Journal Letters, 673, L95, doi:10.1086/527414, 2008. Temmer, M., Nitta, N. V., Interplanetary Propagation Behavior of the Fast Coronal Mass Ejection on 23 July 2012, Solar Physics, 290, 919 -- 932, doi: 10.1007/s11207-014-0642-3, 2015. Thayer, J.P., Liu, X., Lei, J., Pilinski, M., Burns, A.G., The impact of helium on the thermosphere mass density response to geomagnetic activity during the recent solar minimum. J. Geophys. Res., 117, A07315, doi:10.1029/2012JA017832, 2012. Troshichev, O.A., Andrezen, V.G., Vennerstrøm, S., Friis -- Christensen, E., Magnetic ac- tivity in the polar cap -- A new index. Planet Space Science, 36, 1095, 1988. D R A F T October 15, 2018, 3:57am D R A F T X - 26 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Vassiliadis, D., Angelopoulos, V., Baker, D.N., Klimas, A.J., The relation between the northern polar cap and auroral electrojet geomagnetic indices in the wintertime. Geo- physical Research Letters, 23, 20, 2781 -- 2784, 1996. Vasyli¯unas, V.M., Song, P., Meaning of ionospheric joule heating. J. Geophys. Res., 110, A02301, doi: 10.1029/2004JA010615, 2005. Vourlidas, A., Lynch, B.J., Howard, R.A., Li, Y. How Many CMEs Have Flux Ropes? Deciphering the Signatures of Shocks, Flux Ropes, and Prominences in Coronagraph Observations of CMEs. Solar Physics, 284,1,179-201,doi: 10.1007/s11207-012-0084-8, 2013. Vrsnak, B., Processes and mechanisms governing the initiation and propagation of CMEs. Ann. Geophys., 26, 3089-3101, doi:10.5194/angeo-26-3089-2008, 2008. Vrsnak, B., Marici´c, D., Stanger, A. L., Veronig, A. M., Temmer, M., Rosa, D., Acceler- ation Phase of Coronal Mass Ejections: I. Temporal and Spatial Scales. Solar Physics, 241, 85-98, doi:10.1007/s11207-006-0290-3, 2007. Vrsnak, B., Zic, T., Vrbanec, D., Temmer, M., Rollett, T., Mostl, C., Veronig, A., Calogovi´c, J., Dumbovi´c, M., Luli´c, S., Moon, Y.-J., Shanmugaraju, A., Propagation of Interplanetary Coronal Mass Ejections: The Drag-Based Model. Solar Physics, 285, 295-315, doi: 10.1007/s11207-012-0035-4, 2013. Wanliss, J.A., Showalter, K.M., High-resolution global storm index: Dst versus SYM-H. J. Geophys. Res., 111, A02202, doi:10.1029/2005JA011034, 2006. Wilson G.R., Weimer R., Wise J.O., Marcos F.A., Response of the thermo- sphere to Joule heating and particle precipitation. J. Geophys. Res., 111, A10314, doi:10.1029/2005JA011274, 2006. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 27 Yue, C., Qiugang Zong, Q., Solar wind parameters and geomagnetic indices for four different interplanetary shock/ICME structures. J. Geophys. Res., 116, A12, doi:10.1029/2011JA017013, 2011. Yurchyshyn, V., Yashiro, S., Abramenko, V., Wang, H., Gopalswamy, N., Statistical dis- tributions of speeds of coronal mass ejections. The Astrophysical Journal, 619, 599?603, doi:10.1086/426129, 2005. Zhang, J., Dere, K. P., Howard, R. A., Kundu, M. R., White, S. M., On the Temporal Relationship between Coronal Mass Ejections and Flares. The Astrophysical Journal Letters, 559, 452-462, doi:10.1086/322405, 2001. D R A F T October 15, 2018, 3:57am D R A F T X - 28 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Figure 1. Illustration of the CME three-part structure. Sequence of SOHO/LASCO C2 images of a CME that occurred on May 17, 2012 close to the Western limb. Figure 2. Bz observations from ACE during an ICME event in May 2010. Red circles indicate the minimum values used in the further analysis. Yellow and green areas mark the shock-sheath region and magnetic structure, respectively, according to the time information provided in the R&C list. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 29 Figure 3. Latitude-time plot of thermosphere neutral densities derived from GRACE satellite accelerations. The impact of three ICMEs in July 2004 inducing severe geomag- netic storms can be identified in the neutral density measurements. The perturbations are visible on both the day -- side (top: 08:30 mean local time) and night -- side (bottom: 20:30 mean local time). The time interval between two adjacent vertical grid lines covers one satellite revolution (94 min). Figure 4. Location of observatories that provide measurements to construct geomag- netic indices. The indices are provided with different temporal resolution -- aa, am, ap and kp: 3 h; PC: 1 min; Dst: 1 h, AE: 1 min; SYM -- H: 1 min. D R A F T October 15, 2018, 3:57am D R A F T X - 30 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Figure 5. Overview of analyzed parameters during an ICME event on May 15 in 2005 (rows a -- c) and December 14 in 2006 (rows d -- f). Polar cap index (b1,e1) is split into PCN (black) and PCS (green) part; the red curves shows the evolution of the neutral density; blue circles indicated the peak values of the specific parameters independent whether the peak occurred in the shock-sheath region or the magnetic structure of the ICME. D R A F T October 15, 2018, 3:57am D R A F T pcspcsKRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 31 Figure 6. Absolute (bottom) and relative (top) neutral density increase (w.r.t. to pre -- event level) for each of the analyzed ICME events. Marked in red are the most extreme events in both cases. D R A F T October 15, 2018, 3:57am D R A F T absolute density increase [kg m-3]relative density increase [percentage]X - 32 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Figure 7. Scatter plot of thermospheric densities ∆ρ490 and ICME parameters in terms of Bz component in GSE, the maximum ICME velocity (vmax), the convective electric field estimate E and the Poynting flux estimate S. The correlation coefficients and the linear regression coefficients together with the standard deviation for a 90% confidence interval are given in the inset. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 33 Figure 8. Scatter plot of the increase in the neutral thermospheric density and the various geomagnetic indices. The correlation coefficients and the linear regression coeffi- cients together with the standard deviation for a 90% confidence interval are given in the inset. D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 34 Table 1. Correlation coefficient between the various parameters and the increase in neutral density ∆ρ490 (column 2) as well as the Bz component measured by ACE (column 3). ∆ρ490 Bz 1.00 -0.92 -0.92 1.00 0.69 -0.63 -0.88 0.87 -0.81 0.81 -0.91 0.88 0.78 -0.77 0.92 -0.88 0.92 -0.90 0.87 -0.85 0.71 -0.74 ∆ρ490 Bz vmax E S Dst index AE index aa index am index ap index kp index PCN index 0.84 -0.79 PCS index 0.82 -0.76 SYM -- H index -0.91 0.88 D R A F T October 15, 2018, 3:57am D R A F T KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES X - 35 Figure 9. Scatter plot of the minimum Bz component as measured by ACE with the absolute density increase and the various geomagnetic indices. Marked in red are those events where the minimum in the Bz component is found in the shock-sheath region; blue indicates ICMEs with the Bz minimum found in the magnetic structure. The correlation coefficients and the linear regression coefficients are given in the inset. D R A F T October 15, 2018, 3:57am D R A F T cc = -0.92cc = -0.88cc = -0.91cc = -0.79cc = -0.84cc = -0.87cc = 0.68cc = 0.71cc = -0.88cc = -0.86cc = -0.89cc = -0.72cc = -0.67cc = -0.91cc = 0.93cc = 0.93cc = -0.84cc = -0.65X - 36 KRAUSS ET AL.: THERMOSPHERE AND GEOMAGNETIC RESPONSE TO ICMES Figure 10. From top to bottom: Atmospheric neutral density increase, total orbit decay and daily orbital decay rate during an ICME event in November 2003. D R A F T October 15, 2018, 3:57am D R A F T 20.1121.1122.110246x 10−12neutral density [kg m−3]20.1121.1122.11−80−60−40−200total orbit decay [m]20.1121.1122.11−200−150−100−500decay rate [m/day]2003total orbit decay [m]neutral density [kg/m3]orbit decay rate [m/s]
1103.0225	1	1103	2011-03-01T17:08:27	Water ice at low to midlatitudes on Mars	[ "astro-ph.EP" ]	In this paper, we analyze water ice occurrences at the surface of Mars using near-infrared observations, and we study their distribution with a climate model. Latitudes between 45{\deg}S and 50{\deg}N are considered. Data from the Observatoire pour la Min\'eralogie, l'Eau, les Glaces et l'Actitit\'e and the Compact Reconnaissance Imaging Spectrometer for Mars are used to assess the presence of surface water ice as a function of location and season. A modeling approach combining the 1-D and 3-D versions of the General Circulation Model of the Laboratoire de M\'et\'eorologie Dynamique de Jussieu is developed and successfully compared to observations. Ice deposits 2-200 \mu m thick are observed during the day on pole facing slopes in local fall, winter and early spring. Ice extends down to 13{\deg} latitude in the Southern Hemisphere but is restricted to latitudes higher than 32{\deg} in the north. On a given slope, the pattern of ice observations at the surface is mainly controlled by the global variability of atmospheric water (precipitation and vapor), with local ground properties playing a lower role. Only seasonal surface ice is observed: no exposed patches of perennial ground ice have been detected. Surface seasonal ice is however sensitive to subsurface properties: the results presented in this study are consistent with the recent discovery of low latitude subsurface ice obtained through the analysis of CO2 frost.	astro-ph.EP	astro-ph	Water ice at low to midlatitudes on Mars Mathieu Vincendon1, François Forget2, and John Mustard1 1Department of Geological Sciences, Brown University, Providence, RI, USA. 2Laboratoire de Météorologie Dynamique, Université Paris 6, Paris, France. Citation: Vincendon, M., F. Forget, and J. Mustard (2010), Water ice at low to midlatitudes on Mars, J. Geophys. Res., 115, E10001, doi:10.1029/2010JE003584. Abstract: In this paper, we analyze water ice occurrences at the surface of Mars using near-infrared observations, and we study their distribution with a climate model. Latitudes between 45°S and 50°N are considered. Data from the Observatoire pour la Minéralogie, l’Eau, les Glaces et l’Actitité and the Compact Reconnaissance Imaging Spectrometer for Mars are used to assess the presence of surface water ice as a function of location and season. A modeling approach combining the 1-D and 3-D versions of the General Circulation Model of the Laboratoire de Météorologie Dynamique de Jussieu is developed and successfully compared to observations. Ice deposits 2-200 µm thick are observed during the day on pole facing slopes in local fall, winter and early spring. Ice extends down to 13° latitude in the Southern Hemisphere but is restricted to latitudes higher than 32° in the north. On a given slope, the pattern of ice observations at the surface is mainly controlled by the global variability of atmospheric water (precipitation and vapor), with local ground properties playing a lower role. Only seasonal surface ice is observed: no exposed patches of perennial ground ice have been detected. Surface seasonal ice is however sensitive to subsurface properties: the results presented in this study are consistent with the recent discovery of low latitude subsurface ice obtained through the analysis of CO2 frost. 1. Introduction Water on Mars follows a complex seasonal cycle that had varied over geologic times. While liquid water has been involve in the early processing of the martian surface [Bibring et al., 2006; Carr, 1979], it is not observed today, although it could be locally stable [Haberle et al., 2001; Hecht, 2002; Zorzano et al., 2009]. So far, water on Mars has only been detected in the form of vapor, ice, adsorbed molecules, or as a component of minerals [Bibring et al., 2005; Farrell et al., 2009; Jouglet et al., 2007; Kieffer et al., 1976; Spinrad and Richardson, 1963]. The most prominent reservoirs of martian water are ice, mainly located in the northern and southern polar deposits and in the high-latitude permafrost [Clifford et al., 2000; Feldman et al., 2004]. For comparison the reservoir of water vapor is about 6 orders of magnitude lower [Montmessin et al., 2004]. The low to midlatitude regions of Mars are a key environment to study the distribution of surface water ice and its role in the processing of the Martian surface. The stability limits of both seasonal ice and perennial near-surface ice is observed at those 1 latitudes [Carrozzo et al., 2009; Feldman et al., 2004; Schorghofer and Edgett, 2006; Vincendon et al., 2010]. As attested by numerous morphological features [Head et al., 2003; Mustard et al., 2001; Squyres and Carr, 1986] and climate modeling predictions [Forget et al., 2006; Levrard et al., 2004; Mischna et al., 2003; Schorghofer, 2007], these limits has undergone strong variations over geological epochs. Colder, “ice age” era have seen the formation of several landforms still observed today [Head et al., 2003; Mustard et al., 2001; Neukum et al., 2004; Schon et al., 2009], and mineralogical assemblages found at those latitudes could have been influenced by water ice [Mangold et al., 2010; Niles and Michalski, 2009]. This low to midlatitudes water ice, with its potential for melting when the obliquity varies [Christensen, 2003; Costard et al., 2002; Williams et al., 2009], or at present day [Kreslavsky and Head, 2009; Mohlmann, 2008], has been hypothesized to be a conceivable habitable environment on Mars [Morozova et al., 2007; Pavlov et al., 2009; Price, 2007]. Studying surface or subsurface water ice at low to midlatitudes from remotely sensed data is subject to some limitations and constraints. The simultaneous presence of water ice clouds and surface CO2 frost complicates the detectability of surface water ice. Spatial resolution is also a limitation, as favorable conditions required for surface or subsurface ice to be stable may be of limited spatial extent. The imaging spectrometers OMEGA (Observatoire pour la Minéralogie, l’Eau, les Glaces, et l’Activité) and CRISM (Compact Reconnaissance Imaging Spectrometer for Mars) are well suited for the study of water ice as they offer CO2/H2O and cloud/frost separation capabilities, combined with high spatial resolution (down to 350 meters per pixel for OMEGA and 18 meters per pixel for CRISM). The first Mars year of OMEGA observation have been successfully used to study water frost deposits in the southern hemisphere equatorward of 30° [Carrozzo et al., 2009], and subsurface water ice in the 45°S – 25°S latitude range has been discovered through the analysis of OMEGA and CRISM detections of CO2 frost [Vincendon et al., 2010]. In this paper, we perform the mapping of surface water ice from the edge of the polar caps (about 45°-50° latitude) to the equator, in both hemispheres, using OMEGA and CRISM data covering more than 3 Mars years (section 2). We then develop a new modeling approach based on the LMD (Laboratoire de Météorologie Dynamique) GCM (General Circulation Model) used to predict formation of ice at these latitudes (section 3). Observations and modeling results are then compared and discussed in section 4. 2. Observations The OMEGA and CRISM instruments are imaging spectrometers observing in the visible and near-infrared wavelengths with a spectral sampling between 7 nm and 40 nm. Since early 2004, OMEGA has acquired near-global coverage with an average resolution of 1 km per pixel. Because of the orbit of Mars Express, the actual resolution varies between 5 and 0.35 km per pixel and many regions have been repeatedly observed. Since late 2006, CRISM has acquired observations of Mars with a spatial resolution as high as 18 m per pixel in the targeted mode, covering a few percentages of the planet, and 230 m per pixel in the low resolution mode (which plans to acquire global coverage). 2 2.1. Data reduction These observations are sensitive to the presence of a few µm of water ice within the path of electromagnetic radiation reflected from the surface of Mars [Langevin et al., 2007]. Water ice notably absorbs photons around 1.5 and 2 µm, creating characteristic broad absorption features several tenths of µm wide. In the low latitudes, these spectral features are typically a few percentages deep [Carrozzo et al., 2009]. Algorithms to conduct a global, automated search for the presence of these features can be easily implemented. Such an approach, however, highlights several false detections in addition to real surface deposits we are looking for. Among the caveats are the complex structure of the random spectral noise, data anomalies, wavelength-dependent nonlinearity in detectors response, water ice clouds, and surface deposits of minerals such as sulfate, which absorb radiation at similar wavelengths. As a consequence, automatic processing provides a first analysis of the global data and positive detections must be validated in detail. Spectral analysis is used to confirm the presence of water ice, while analyzing the geologic setting makes it possible to distinguish between surface and atmospheric origins. It has been possible to reach a high level of confidence in the automatic detection algorithm applied to the OMEGA dataset, as detailed in Figure 1. Only a few false positive detections that must be manually removed remain after applying this algorithm. They are due to surface minerals with unusual deep spectral features, such as the olivine spots seen in Nili Fossae, or to small extent topographic clouds such as those seen in Peta crater. Due to the much more complex structure of the CRISM noise [Parente, 2008], a similar approach has not been implemented for this data set: all positive CRISM detections have been checked manually. Figure 1. OMEGA detection algorithm. The 1.5 µm water ice band depth is estimated using ) R 1.39µm × R 1.77µm depth = 1 − R 1.50µm ) 0.3 ) 0.7 ) . The noise in OMEGA ( the criteria, ( ( ( data is dominated by the read noise (1.85 DN). The noise limit is therefore of the form, depth > C × 1.85 ) flux , with a constant. To account for non-linearity issues around 1.5 ( € 3 € depth > 6.5 flux + 0.035 , µm, an offset is added. In panel a, the selected detection limit ( about 4% or higher) has been empirically determined based on observations without ice (dots). Isolated spikes due to cosmic rays or data anomalies are removed by selecting only detections of more than two contiguous pixels. Extreme photometric conditions (i >80°, € e >40°) are not considered due to aerosols scattering. As ice deposits are observed on pole facing slopes, most ice clouds of moderate extent are removed by considering only detections with a spatial extent smaller than 5 km in latitude and 50 km in longitude. Finally, the background atmospheric ice opacity due to large ice clouds is removed by dividing the observed 1.5 µm depth by the latitudinal trends in 1.5 µm depth. An example is shown in Figure 1b: the observed band depth (black) increases with latitude due to increasing cloud opacity. The corrected band depth (lighter tone/blue, centered on 0) is above the noise limit (4%) only at 39°N due to surface ice localized on a pole facing slope. 2.2. Spatial and seasonal distribution Surface water ice deposits are mapped in Figure 2. In the Southern Hemisphere, the distribution of ice strongly varies with longitude. Ice is detected down to 13°S around 10°E and on the northern wall of Valles Marineris near 60°W, while it is not observed equatorward of 40°S at 20°W. The major concentrations of surface ice deposits are south of the great volcanoes and west of the Hellas basin. A particularly dry area is observed between 50°W and 0°. In the Northern Hemisphere, the distribution of ice is more homogeneous, but ice is not observed equatorward of 32°N. Figure 2. Observed surface water ice deposits detected between 45°S and 50°N shown on a MOLA shaded relief map. OMEGA, black stars; CRISM Low-Res (“MRDR”), blue dots; CRISM High-Res (“FRT”, “HRS” and “HRL”), red diamonds. Ice is observed down to 13°S and 32°N. 4 The latitude-season diagrams of ice deposits are shown in Figure 3. Ice is observed only during local fall and winter (and early spring for the Northern Hemisphere). These deposits correspond well to expected seasonal frost. We see no deposits that can be interpreted as perennial ice. This is consistent with the expected rapid sublimation of snowpack at those latitudes [Byrne et al., 2009; Williams et al., 2008]. The distribution of frost derived from observations in the southern hemisphere generally agrees with the results of [Carrozzo et al., 2009] when they overlap in time and location (i.e., for LS in the 100°– 150° range and for latitudes in the 15°S–30°S range), except for their detections at LS 20°S and 30°S which are due to clouds according to our study. These authors have observed water ice detection in the northern hemisphere at latitudes lower than 30°N in summer, which where not considered reliable due to the potential contribution of clouds. These detections are not present in our results, which confirm their atmospheric origin. Figure 3. Observed surface water ice deposits detected between 45°S and 50°N shown in a latitude/season diagram (left, south; right, north). OMEGA, black stars; CRISM L-R, blue dots; CRISM H-R, red diamonds. At those latitudes water ice is observed only during local autumn and winter (as well as during early spring for the Northern Hemisphere). 2.3. Observational bias Since early 2004, almost all the midlatitudes have been observed at appropriate winter LS by either CRISM or OMEGA, which rule out any major observational bias. Spatial resolution is not a major issue: the spatial and temporal stability pattern is similar at first order as seen in the three datasets (OMEGA, with its resolution in the 0.35–5 km range; CRISM LR, at 230 m per pixel resolution and CRISM HR, with a 18 m per pixel resolution; see Figures 2 and 3). Second order, slight differences are observed: at the most equatorward latitudes, a greater number of deposits are observed that result from the higher spatial resolution of CRISM (Figure 2). As a result of the orbital properties of Mars Express, southern midlatitudes have not been observed by OMEGA at LS > 130–150°. A similar bias is observed prior to LS 240° in the Northern Hemisphere. 5 Local time is not a major issue in the distribution of frost presented in this study. All CRISM observations are acquired around 3 pm. OMEGA observation can be obtained anytime in daylight. However, we do not consider early morning and late evening observations as we restrict ourselves to solar zenith angles lower than 80° (above that limit the contribution of light scattered by aerosols dominates, see e.g. [Vincendon et al., 2007]). No specific variability has been found in the OMEGA data set for hours surrounding midday. Figure 4. CRISM high-resolution observations (18-40 m per pixels) of low to midlatitude water ice deposits. (a-e) Ice detected using the 1.5 µm water ice feature is mapped in blue above a background visible composite image at five locations. (f) Observation “e” is also shown without ice to highlight the geologic setting. Ice is observed during local winter in both the (a, d) Northern and (b, c, e) Southern hemispheres. Ice forms on pole facing slopes linked with all kind of geomorphology: (a) escarpments, (b, d) mesas, (c) valleys, and (e) crater rims. 6 2.4. Local setting Deposits are found on steep slopes facing the pole of their local hemisphere (Figure 4). The coverage of crater rims by frost can extend to east and west facing walls when small-scaled topographic features such as gullies creates local pole-facing slopes (Figure 4e; see also [Dickson and Head, 2009]). In general, frost is observed on any kind of pole facing slopes (crater walls, mesas, knobs, etc.), regardless of the underlying material. Slopes on which water ice forms are typically in the 20°–30° range according to MOLA topography measurements [Kreslavsky and Head, 2000; Smith et al., 1999; Vincendon et al., 2010]. At the edge of the stability pattern (lowest latitudes, earliest/latest LS), frost tends to be observed mainly on the steepest slopes (30°). 2.5. Link with gullies Several mechanisms could be responsible or participate to the formation and modification of Martian gullies (see, e.g., a review by [Dickson and Head, 2009]), including melting of surface water ice during period of high obliquity or at present day [Dickson and Head, 2009; Hecht, 2002; Williams et al., 2009]. Variations in shape and color have been observed in gully channels during the last years [Dundas et al., 2010; Malin et al., 2006], which correspond to the years of operation of OMEGA and CRISM. In this study, we found frost to be present at the location of several gully clusters as mapped by [Dickson et al., 2007; Heldmann and Mellon, 2004]. The distribution of gullies does, however, not necessarily correlate with the distribution of present-day surface frost. For example, gullies are frequently observed in the dry region at the east of Argyre (30°W–0° / 45°S–30°), while not found in the ice-rich 20°–30° latitude band. Interestingly, the location of recent gully activity determined by [Dundas et al., 2010] corresponds to areas of water frost stability. CO2 frost is also present at most of these sites of recent gully activity (see [Vincendon et al., 2010]). 3. Modeling 3.1. Modeling approach Our modeling approach is based on the General Circulation Model (GCM) of the Laboratoire de Météorologie Dynamique de Jussieu (LMD) [Forget et al., 1999]. This GCM has been extensively validated through comparisons with observations, notably observations related to the water cycle [Forget et al., 2008; Montmessin et al., 2004]. The main version of this GCM is a 3-D grid point dynamic model, from which a 1-D version has been derived. The same physics is used in this 1-D version. In 1-D, however, it is possible to explore the environment on any slope with any orientation. Local properties, including those varying with season such as the aerosols optical depth, have to be prescribed. This 1-D model is therefore suited for the study of local phenomena that cannot be simulated by the global low-resolution model and for which feedbacks on the large-scale properties of the meteorology can be neglected. An illumination scheme designed for surface slopes has been recently included in this 1-D code [Spiga and Forget, 2008] and successfully used to analyze observations of CO2 ice [Vincendon et al., 2010]. Modeling 7 the condensation of ice on Mars is trickier for H2O than for CO2. CO2 is the main component of Mars atmosphere (95%): the partial pressure of CO2 is roughly equal to the total pressure. CO2 condenses on the ground when the temperature of the surface reaches the frost point, a pressure-dependent parameter. Seasonal and spatial variations of pressure can be simply considered: they are primarily due to global change in atmospheric mass resulting from the seasonal condensation/sublimation of CO2 in the polar regions, modulated by the topography [Hourdin et al., 1995]. The situation is more complex for H2O, a minor component of the Martian atmosphere (<0.1%). At Martian pressure, water can be found in the ice phase when surface temperatures are below 260K – 280K, which is common on Mars. The situation is then similar to liquid water on Earth: two mechanisms can lead to the formation of water ice on the ground. First, direct deposition of water vapor from saturated air (i.e., formation of “frost”) occurs when the partial vapor pressure is higher than the saturation vapor pressure at surface temperature. The second mechanism is precipitation of water ice particles that have condensed higher in the atmosphere. Ice formed this way will be stable if the partial pressure of vapor is higher than the saturation vapor pressure; otherwise it will progressively sublimate with a flux proportional to the wind speed and to the difference between the partial and saturation vapor pressures [Madeleine et al., 2009; Montmessin et al., 2004]. As a consequence, modeling water ice requires a good knowledge of the partial vapor pressure and of the amount of precipitation, parameters which are driven by a complex meteorological dynamic on Mars [Montmessin et al., 2004]. We have constructed global maps of the partial vapor pressure and the amount of precipitation with a time step of half an hour using the 3-D GCM. The spatial sampling of these maps is 300 km. Values from these lookup tables are then used as inputs to the 1-D model, which locally computes the temperature on slopes at a given location as a function of time. The main assumption behind this approach is that the presence of frost patches on slopes does not significantly impact the amount of water vapor or water precipitation in the boundary layer. This hypothesis is relevant considering the small extent of these isolated deposits (deposits a few kilometers wide separated by distances greater by more than one order of magnitude) compared to typical atmospheric movements (winds of several meters per second). 3.2. Selected default parameters In addition to the water vapor pressure and the ice sedimentation flux, two other parameters that vary with latitude, longitude, and season are first computed using the 3-D GCM: the total pressure (so as to account for second order mechanisms affecting total pressure [Hourdin et al., 1993]) and the temperature of flat surfaces surrounding slopes. The ground (below potential ice deposits) is parameterized by its surface albedo and by two layers of different thermal inertia. The depth of the boundary between the two layers can be selected. When not otherwise specified, the two layers have the same thermal inertia, and the albedo and thermal inertia as a function of latitude and longitude correspond to average trends derived from TES retrievals [Putzig et al., 2005] ; a subsurface layer with the thermal inertia of water ice (2120 kg K-1 s-5/2) is used in the Southern Hemisphere at latitudes higher than 25°, using the latitude dependent depth of [Vincendon et al., 2010]. 8 Longitudes 10°E in the Southern Hemisphere and 50°W in the northern hemisphere are considered in more detail in section 4: the default surface inertia and surface albedo are 250 kg K-1 s-5/2 and 0.15, respectively, for these longitudes. An angle of 30° and a pole facing orientation are considered for slopes geometric properties. Two parameters describe ice properties: the solar albedo and the infrared emissivity. The water ice/frost emissivity is between 0.95 and 1.00 across the thermal infrared region [MODIS UCBS Emissivity library]: a value of 1.00 will be used as results are weakly affected over that range. The solar albedo of water ice is known to vary significantly on Mars, from low values of about 0.25 to higher values of 0.6 depending on dust contamination and grain size [Langevin et al., 2005]. Assessing the albedo of ice deposits on slopes is especially difficult due to the complex lighting and viewing geometries. A standard value of 0.4 [Montmessin et al., 2004] is a good proxy for winter water frost observed at higher latitudes on flat surfaces [Vincendon et al., 2007]. The model also accounts for the condensation and sublimation of CO2 ice: an albedo of 0.65 and an emissivity of 1 are selected [Vincendon et al., 2010]. The optical depth of aerosols corresponds to the annual variations seen by Opportunity during MY27, scaled by surface pressure (“well-mixed conditions” hypothesis). These variations were shown to provide a relevant estimate of the optical depth at all longitudes for low to midlatitudes [Vincendon et al., 2009]. As indicated previously, the sublimation flux of water is proportional to the wind speed: a value of 20 m s-1 is selected, as mesoscale modeling indicates winds in the 10 – 30 m s-1 range on slopes [Spiga et al., 2010]. Below we present model results corresponding to the equatorward limit of the presence of water ice at a local time of 3 PM as a function of LS. We used in section 2.1 a detection limit of about 4% for the 1.5 µm band depth. Considering laboratory measurements of near-IR spectra of ices [Schmitt et al., 2004] and the decreasing spectral contrast of surface features when seen through the aerosols layer [Vincendon et al., 2007], we should be sensitive to deposits a few µm thick. The model predicts the mass of ice deposits, from which the thickness can be estimated assuming a water ice density of 920 kg m-3. When not otherwise specified, a 2 µm thickness threshold will be considered in the model. We analyze in the next section the impact of changing assumptions regarding most of these parameters: water ice albedo, aerosols optical depth, surface/subsurface thermal inertia, depth of the subsurface layer, surface albedo, slope angle, wind, and thickness threshold. 4. Results Observations and modeling predictions are mapped at different LS in Figure 5 and 6. Overall, the stability pattern predicted by the model using the default set of parameters is consistent with most observations for both the Southern and the Northern hemisphere. We first analyze the global variability of ice stability in subsections 4.1 and 4.2. Afterwards, we study local slope properties using latitude-season diagrams (subsections 4.3 to 4.5). In these 9 sections longitudes corresponding to the equatorward limit of water ice are selected: 10°E for the Southern Hemisphere and 50°W for the Northern Hemisphere. Figure 5. Predicted area of ice deposits on slopes (dark blue, 5µm thick and more; light blue, 2µm thick) compared to observations (dots). Model predictions are shown for 8 solar longitudes. Corresponding observations at ± 9° of LS are indicated. 10 Figure 6. Same as Figure 5, but for the Northern Hemisphere. 4.1. Large-scale variability: Southern Hemisphere In the southern hemisphere (Figure 5), the major longitudinal patterns are well reproduced by the model. The observed dry area between 50°W and 0° is notably predicted. 11 The origin of this asymmetry is illustrated in Figure 7, which shows the wind and atmospheric water content above the surface in early winter. Between 50°W and 0°, the atmospheric circulation is characterized by strong northward flux in the lower branch of the Hadley cell, concentrated in this longitudinal range by the phenomena of western boundary current on the eastern side of the Tharsis bulge [Joshi et al., 1994; 1995]. At this season the amount of atmospheric water is much lower in the Southern Hemisphere (winter). Consequently, the northward flux is relatively dry and reduces the stability of surface water ice. Similarly, the Hellas topographic depression induces a stationary wave, which generates a northward flux of dry air into the Hellas basin. As a results, a lack of surface frost is predicted in north-west Hellas (50°E – 70°E / 30°S – 45°S) and is consistent with observations. Figure 7. Mean wind and atmospheric H2O content (vapor and ice) at 100 m above the surface, averaged over LS 90°–LS 120° (early southern winter). The model routinely predicts ice at low latitudes (down to 13°S) in the Thaumasia area (110°W–60°W) while only a few spots are observed there. Potential thin (~2 µm) ice deposits are also predicted near Arsia Mons, while not observed. Several factors contribute to this discrepancy. First, the region is covered by Hesperian lava and contains fewer slopes compared to the highly cratered Noachian terrains found at other longitudes, which reduce the probability to observe ice. Second, only the steepest slopes (30°) can hold water ice at most LS in this area (Figure 8). Finally, the model over-predicts cloud opacity and therefore precipitation from the aphelion cloud belt [Montmessin et al., 2004]. This is illustrated in Figure 9 where the predicted thickness of ice deposits is mapped. While the density of positive water ice observations generally correlate with the predicted thickness (up to about 100 µm for most longitudes), very thick deposits (> 150 µm) are predicted in Thaumasia where only a few observations of ice are reported. The local latitude maximum observed at 10°E (13°S, compared to about 20°S for surrounding longitudes) is not reproduced by the model with the default set of parameters described in section 3.2. While specific local conditions on slopes (inertia, albedo) favoring surface ice formation can easily explain ice at these lower latitudes (as discussed below), the “peak” structure in the observations (with an equatorward limit that progressively 12 increases from 5°E to 10°E and then decrease from 10°E to 15°) suggests a possible local meteorological condition not reproduced by the model at its 5° spatial resolution. Figure 8. Stability limit of 5 µm thick water ice deposits predicted by the model as a function of slope angle (blue, 30°, green, 20° and red, 10°) for 3 LS. Figure 9. (top) Thickness of water ice deposits predicted by the model at LS 134° in the Southern Hemisphere. Purple, 2 µm; dark blue, 5 µm; light blue, 10 µm; green, 20 µm; yellow, 50 µm; red, 150 µm. (bottom) Observed deposits (all LS) are superimposed for comparison. 13 4.2. Large-scale geographic variability: Northern Hemisphere The Northern Hemisphere (Figure 6) contrasts with the Southern Hemisphere as it is characterized by smooth longitudinal variations of water ice stability both in the data and in the model. The equatorward limit of ice fluctuates mainly between 30°N and 40°N. The driest place is the smooth plain of Amazonis Planitia, with ice not observed equatorward of 45°N. While the presence of fewer favorable slopes in this area could explain the results, the model indeed predict a local stability minimum there. The effect of a lower density of slopes can be seen in Utopia planitia (80°E–150°E): observed deposits are seen at all latitudes where the model predicts ice, but detections are sparse. The Northern Hemisphere is characterized by a quite uniform thickness distribution: during fall the equatorward limit of “2µm” deposits is only 2° further the “5 µm” limit, and both thickness limits are rapidly mixed up as we enter winter. Deposits reach a maximum thickness of about 20–50 µm at most longitudes and latitudes after the northern winter solstice. 4.3. Ground properties below the ice: thermal inertia, albedo, slope angle In the southern hemisphere, a few ice deposits are observed at latitudes significantly lower than predicted (Figure 5): about 17°S in the 150°W–170°W range, and 13°S at 10°E (while the limit predicted using the default set of parameters is about 19°S). The use of high surface inertia (1200 kg K-1 s-5/2), corresponding e.g. to boulder/rock [Christensen et al., 2003], makes it possible to increase the equatorward limit of the stability of ice by 5° (Figure 10a), providing a possible explanation for those deposits. Increasing the albedo of the ground below the ice from 0.15 to 0.35 shifts the equatorward limit of ice stability by 4° (Figure 10b). A localized increase of surface albedo could thus also contribute to locally shift the equatorward limit of ice stability. In the southern hemisphere, ice is observed up to LS 170°, in agreement with modeling predictions. The model predicts surface ice up to LS 220° if we remove subsurface ice (Figure 10a). Indeed, without subsurface ice the model over-predicts the condensation of CO2 ice [Vincendon et al., 2010], a cold trap for H2O ice. While CO2 observations were only sensitive to the upper limit of the depth of subsurface ice at 25°S (found to be about 90 cm, [Vincendon et al., 2010]), H2O ice provides a constraint on the lower limit of subsurface ice: a layer of ice too close to the surface (< 10 cm) indeed reduces the equatorward limit of surface water ice too far (Figure 10a), which is consistent with the expected deeper permafrost at such low latitudes [Aharonson and Schorghofer, 2006]. Model results are not so sensitive to subsurface ice in the Northern Hemisphere (Figure 11), in particular because CO2 ice does not condense on slopes (the solar flux during the northern winter solstice – almost concomitant with perihelion – is about 50% higher than for the southern winter solstice). Similarly to the Southern Hemisphere, a very shallow (< 10 cm) widespread layer of subsurface ice below steep slopes is not consistent with observations at low latitudes (30°N–35°N). 14 Figure 10. Model compared to observations in latitude/season diagrams at 10°E in the Southern Hemisphere. Observed water ice deposits (green stars) correspond to the 0°– 60°E longitude range. For all panels, black dots correspond to the standard parameters set as described in section 3.2. (a) Impact of surface/subsurface thermal inertia I (standard: I = 250 kg K-1 s-5/2 above I = 2120 kg K-1 s-5/2 – subsurface ice, with a depth varying from 6 cm at 45°S to 90 cm at 25°S). Red squares, no subsurface ice; orange triangles: subsurface ice at a constant depth of 10 cm; blue diamonds, I = 1200 kg K-1 s-5/2 for both surface and subsurface. (b) Impact of albedo (standard: substrate albedo AS of 0.15 and ice albedo AL of 0.4). Orange triangles, AL = 0.25; blue diamonds, AL = 0.60; red squares, AS = 0.35. (c) Impact of water ice thickness thresholds and dust aerosols (standard: threshold T = 2 µm and Opportunity dust scenario). Orange triangles, T = 1 µm; red squares, T = 3 µm; blue diamonds, T = 5 µm; purple crosses, Spirit dust scenario (optical depth about 30% lower in southern winter). (d) Impact of winds (standard wind speed: v = 20 m.s-1). Blue diamonds, v = 10 m.s-1; orange triangles, v = 50 m.s-1; red squares, v = 1 m.s-1. Insights into the dependence of model results to slope angle are provided in Figure 8 and 11. In the Northern Hemisphere (Figure 11), the equatorward limit of ice stability is reduced by about 3°–5° for every 10° decrease of the slope angle. In the Southern 15 Hemisphere, the decrease with slope angle is more pronounced (typically about 5°–10° latitude per 10° of slope angle), with significant longitudinal variations (Figure 8). 4.4. Ice properties Changing the albedo of ice (Figure 10b) does not impact the starting date of the condensation, which depends on the properties of the slope before ice formation. On the contrary, increasing the albedo of ice significantly delays the sublimation (by about 20°– 30° of LS between albedo of 0.25 and 0.60). Low to moderate ice albedos (0.3–0.4) are more consistent with observations (Figure 10b), in agreement with the expected properties of water ice: observed seasonal frost is generally low albedo (see section 3.2 and [Vincendon et al., 2007]), and our model shows that most of the ice on slopes is formed via precipitation of atmospheric water ice, which is known to contain atmospheric dust [Vincendon et al., 2008]. The thickness threshold plays a significant role in modeling predictions in fall in the Southern Hemisphere when condensation starts (Figure 10c). Once 1µm of water ice as accumulated at the surface, a 50° LS period is needed to reach 5µm at certain latitudes and longitudes, whereas the accumulation is significantly faster (10° LS only) at other locations. Comparisons of modeling predictions with observations (Figure 10c) show that the relevant threshold is between 2µm and 5µm, in agreement with expected instrumental capabilities (see section 3.2). The exact value of this threshold is a complex function of observation conditions (photometric angles, atmospheric conditions, spatial sampling, etc.). Two thresholds (2 and 5 µm) are therefore shown in maps of Figures 5 and 6 to assess the influence of this parameter as a function of location. The thickness of deposits is also mapped on Figure 8 for LS 134° (at this time most deposits have reached their maximum thickness). The thickness of most deposits on 30° pole facing slopes is 20–50 µm for both the Southern and the Northern Hemisphere. 4.5. Impact of aerosols and wind A significant part of incoming solar radiations on pole-facing slopes is scattered by aerosols in winter at midlatitudes. This component of the radiative budget must therefore be modeled properly, as shown by modeling results corresponding to CO2 ice [Vincendon et al., 2010]. Uncertainties in the amount of atmospheric dust during winter are the most prominent source of model uncertainties related to aerosols [Vincendon et al., 2010]. The optical depth is low in southern winter while northern winter corresponds to the storm season. Optical depth measurements by Spirit and Opportunity [Lemmon et al., 2004] provide a good proxy for low to midlatitudes [Vincendon et al., 2009]. We found that results are only changed by a few degrees of latitude or LS if we change assumptions regarding the amount of aerosols in the atmosphere (Figure 10c and Figure 11). Wind speed controls the sublimation flux of surface ice when the atmosphere is not saturated in water vapor. A constant wind speed is used in our modeling approach, with a default value set to 20 m.s-1 (see section 3.2). As expected, lower winds increase the stability of ice (Figure 10d). Model results are however weakly modified by a change in the wind speed assumption: results are similar for winds in the 10-50 m.s-1 range expected from 16 mesoscale modeling [Spiga et al., 2010]. Very low local winds (~ 1 m.s-1) increases the stability of ice by about 3° latitude and 10–20° of LS, which could also contribute to explain local mismatch of that order between modeling predictions and observations (see section 4.3). Figure 11. Model compared to observations in latitude/season diagrams at 50°W in the Northern Hemisphere with variations in dust aerosols, slopes angle, and subsurface ice. Observed water ice deposits (green stars) correspond to the 80°W–20°W longitude range. Black dots, standard scenario (30° slope angle, Opportunity MY27 dust scenario, no subsurface ice). Orange triangles, Opportunity MY28 dust scenario, including a global dust storm in northern winter. Purple crosses, subsurface water ice at a constant depth of 10 cm is added. Red squares, 20° slope angle. Blue diamonds, 10° slope angle. Conclusion We used near-IR observations by CRISM and OMEGA to map as a function of time the presence of surface water ice during the day at low and midlatitudes (45°S–50°N latitude range). These instruments have observed Mars for 2 and 3 Mars years, respectively, at all locations and seasons, providing a huge data set exempt of major observational bias. Deposits as thin as 2 µm are detected using the vibrational absorptions of water ice located at a wavelength of 1.5 µm. Surface ice is confidently identified among potentially cloudy observations through the use of spatial extent considerations. At these latitudes, ice is observed on pole-facing slopes. Most deposits are found on crater rims; however ice is observed on any kind of slope such as those created by escarpments, mesas and valleys. In the Northern Hemisphere, the equatorward limit of ice shows smooth variations with longitude, with a local maximum of 32°N observed at 50°W and a local minimum near 45°N in Amazonis. The Southern Hemisphere is characterized by a much more complex geographic pattern: ice extends at significantly equatorward latitudes (13°S at 10°E and in 17 Valles Marineris) but is only observed down to 40°S at 20°W. 13°S is the equatorward limit of surface ice reported so far. We developed a modeling approach based on the LMD GCM to predict the stability of water ice and analyze our observations. Our model combines a 1-D energy balance code that can be used to compute the temperature on slopes with seasonal maps of water vapor and water ice precipitation predicted by the 3-D GCM. Overall, we found a very good agreement between modeling predictions and observations. The model shows that the large- scaled geographic variability in surface water ice is mainly driven by the global meteorology. A particularly dry area is notably found in the Southern Hemisphere between longitudes 50°W and 0° and explained by flux of dry air during winter over this area. While being of second order, local properties also impact ice condensation, which is favored on high inertia and high albedo surfaces. A low albedo for water ice (0.4 or less) is most consistent with observations according to the model. Both observations and modeling predictions shows that local time during daylight does not significantly modify the distribution of detectable ice deposits: the ice thickness required for ice to be detected (2–5 µm) is such that ice needs to accumulate several days to be seen. According to the model, deposits are typically a few tens of µm thick. Surface ice has only been found in fall and winter (as well as in early spring for the Northern Hemisphere) at times and places totally consistent with seasonal ice formation predicted by the model. No exposed perennial ice has hence been detected, which is expected at those latitudes. The presence of perennial ice in the shallow subsurface strongly increases its thermal inertia, which significantly impacts surface temperatures. Surface temperatures then modify the capability of gaseous species (CO2 and H2O) to condense [Kossacki and Markiewicz, 2002]. Observations of surface CO2 ice combined with GCM predictions have been previously used to detect the presence of subsurface water ice down to 25° latitude in the southern hemisphere [Vincendon et al., 2010]. Our present study provided interesting complementary evidence for such subsurface ice. First, the same model and parameters used to match CO2 ice observations are successfully used here to model H2O ice without adjustments. In particular, the study of H2O ice has provided an opportunity to verify the ability of the model to match frost observations where subsurface ice can not be invoked (i.e., at latitudes as low as 13°). Second, observations of H2O ice are also, to a lower extent, sensitive to subsurface ice. They notably provides a lower limit for the ice table depth at 25°S (10 cm), which completes the upper limit (90 cm) estimated using CO2 frost. In the Northern Hemisphere, subsurface ice on steep pole facing slopes, if present, have to be buried deeper than 10 cm at 30°–35° latitude. Acknowledgment: We would like to thank the LMD GCM, the OMEGA and the CRISM engineering and scientific teams, for their help and for making this work possible. We thank in particular (in alphabetical order) Francesca Altieri, Jean-Pierre Bibring, Giacomo Carrozzo, Brigitte Gondet, Misha Kreslavsky, Yves Langevin, Jean-Baptiste Madeleine, Ehouarn Millour, and Aymeric Spiga. 18 References: Aharonson, O., and N. Schorghofer (2006), Subsurface ice on Mars with rough topography, J Geophys Res-Planet, 111(E11), E11007, doi:10.1029/2005JE002636. Bibring, J. P., et al. (2006), Global mineralogical and aqueous mars history derived from OMEGA/Mars express data, Science, 312(5772), 400-404. Bibring, J. P., et al. (2005), Mars surface diversity as revealed by the OMEGA/Mars Express observations, Science, 307(5715), 1576-1581. Byrne, S., et al. (2009), Distribution of Mid-Latitude Ground Ice on Mars from New Impact Craters, Science, 325(5948), 1674-1676. Carr, M. H. (1979), Formation of Martian Flood Features by Release of Water from Confined Aquifers, J Geophys Res, 84(Nb6), 2995-3007. Carrozzo, F. G., G. Bellucci, F. Altieri, E. D'Aversa, and J. P. Bibring (2009), Mapping of water frost and ice at low latitudes on Mars, Icarus, 203(2), 406-420. Christensen, P. R. (2003), Formation of recent martian gullies through melting of extensive water-rich snow deposits, Nature, 422(6927), 45-48. Christensen, P. R., et al. (2003), Morphology and composition of the surface of Mars: Mars Odyssey THEMIS results, Science, 300(5628), 2056-2061. Clifford, S. M., et al. (2000), The state and future of Mars polar science and exploration, Icarus, 144(2), 210-242. Costard, F., F. Forget, N. Mangold, and J. P. Peulvast (2002), Formation of recent Martian debris flows by melting of near-surface ground ice at high obliquity, Science, 295(5552), 110-113. Dickson, J. L., and J. W. Head (2009), The formation and evolution of youthful gullies on Mars: Gullies as the late-stage phase of Mars' most recent ice age, Icarus, 204(1), 63-86. Dickson, J. L., J. W. Head, and M. Kreslavsky (2007), Martian gullies in the southern mid- latitudes of Mars: Evidence for climate-controlled formation of young fluvial features based upon local and global topography, Icarus, 188(2), 315-323. Dundas, C. M., A. McEwen, S. Diniega, and S. Byrne (2010), New and Recent Gully Activity on Mars as Seen by HiRISE, Geophys Res Lett, In press. Farrell, W. M., J. J. Plaut, S. A. Cummer, D. A. Gurnett, G. Picardi, T. R. Watters, and A. Safaeinili (2009), Is the Martian water table hidden from radar view?, Geophys Res Lett, 36, L15206, doi:10.1029/2009GL038945. Feldman, W. C., et al. (2004), Global distribution of near-surface hydrogen on Mars, J Geophys Res-Planet, 109(E9), -. Forget, F., R. M. Haberle, F. Montmessin, B. Levrard, and J. W. Heads (2006), Formation of glaciers on Mars by atmospheric precipitation at high obliquity, Science, 311(5759), 368-371. Forget, F., E. Millour, J. Madelaine, F. Lefevre, and F. Montmessin (2008), Simulation of the Mars water cycle with the LM Mars Global Climate model, The Mars water cycle workshop, Paris April 2008. Forget, F., F. Hourdin, R. Fournier, C. Hourdin, O. Talagrand, M. Collins, S. R. Lewis, P. L. Read, and J. P. Huot (1999), Improved general circulation models of the Martian atmosphere from the surface to above 80 km, J Geophys Res-Planet, 104(E10), 24155- 24175. 19 Haberle, R. M., C. P. McKay, J. Schaeffer, N. A. Cabrol, E. A. Grin, A. P. Zent, and R. Quinn (2001), On the possibility of liquid water on present-day Mars, J Geophys Res- Planet, 106(E10), 23317-23326. Head, J. W., J. F. Mustard, M. A. Kreslavsky, R. E. Milliken, and D. R. Marchant (2003), Recent ice ages on Mars, Nature, 426(6968), 797-802. Hecht, M. H. (2002), Metastability of liquid water on Mars, Icarus, 156(2), 373-386. Heldmann, J. L., and M. T. Mellon (2004), Observations of martian gullies and constraints on potential formation mechanisms, Icarus, 168(2), 285-304. Hourdin, F., F. Forget, and O. Talagrand (1995), The Sensitivity of the Martian Surface Pressure and Atmospheric Mass Budget to Various Parameters - a Comparison between Numerical Simulations and Viking Observations, J Geophys Res-Planet, 100(E3), 5501- 5523. Hourdin, F., P. Levan, F. Forget, and O. Talagrand (1993), Meteorological Variability and the Annual Surface Pressure Cycle on Mars, J Atmos Sci, 50(21), 3625-3640. Joshi, M. M., S. R. Lewis, P. L. Read, and D. C. Catling (1994), Western Boundary Currents in the Atmosphere of Mars, Nature, 367(6463), 548-551. Joshi, M. M., S. R. Lewis, P. L. Read, and D. C. Catling (1995), Western Boundary Currents in the Martian Atmosphere - Numerical Simulations and Observational Evidence, J Geophys Res-Planet, 100(E3), 5485-5500. Jouglet, D., F. Poulet, R. E. Milliken, J. F. Mustard, J. P. Bibring, Y. Langevin, B. Gondet, and C. Gomez (2007), Hydration state of the Martian surface as seen by Mars Express OMEGA: 1. Analysis of the 3 mu m hydration feature, J Geophys Res-Planet, 112(E8), E08S06, doi:10.1029/2006JE002846. Kieffer, H. H., S. C. Chase, T. Z. Martin, E. D. Miner, and F. D. Palluconi (1976), Martian North Pole Summer Temperatures - Dirty Water Ice, Science, 194(4271), 1341-1344. Kossacki, K. J., and W. J. Markiewicz (2002), Martian seasonal CO2 ice in polygonal troughs in southern polar region: Role of the distribution of subsurface H2O ice, Icarus, 160, 73–85, doi:10.1006/icar.2002.6936. Kreslavsky, M. A., and J. W. Head (2000), Kilometer-scale roughness of Mars: Results from MOLA data analysis, J Geophys Res-Planet, 105(E11), 26695-26711. Kreslavsky, M. A., and J. W. Head (2009), Slope streaks on Mars: A new "wet" mechanism, Icarus, 201(2), 517-527. Langevin, Y., F. Poulet, J. P. Bibring, B. Schmitt, S. Doute, and B. Gondet (2005), Summer evolution of the north polar cap of Mars as observed by OMEGA/Mars express, Science, 307(5715), 1581-1584. Langevin, Y., J. P. Bibring, F. Montmessin, F. Forget, M. Vincendon, S. Doute, F. Poulet, and B. Gondet (2007), Observations of the south seasonal cap of Mars during recession in 2004-2006 by the OMEGA visible/near-infrared imaging spectrometer on board Mars Express, J Geophys Res-Planet, 112(E8), E08S12, doi:10.1029/2006JE002841. Lemmon, M. T., et al. (2004), Atmospheric imaging results from the Mars exploration rovers: Spirit and Opportunity, Science, 306(5702), 1753-1756. Levrard, B., F. Forget, F. Montmessin, and J. Laskar (2004), Recent ice-rich deposits formed at high latitudes on Mars by sublimation of unstable equatorial ice during low obliquity, Nature, 431(7012), 1072-1075. 20 Madeleine, J. B., F. Forget, J. W. Head, B. Levrard, F. Montmessin, and E. Millour (2009), Amazonian northern mid-latitude glaciation on Mars: A proposed climate scenario, Icarus, 203(2), 390-405. Malin, M. C., K. S. Edgett, L. V. Posiolova, S. M. McColley, and E. Z. N. Dobrea (2006), Present-day impact cratering rate and contemporary gully activity on Mars, Science, 314(5805), 1573-1577. Mangold, N., L. Roach, R. Milliken, S. Le Mouelic, V. Ansan, J. Bibring, P. Masson, J. Mustard, S. Murchie, and G. Neukum (2010), A Late Amazonian alteration layer related to local volcanism on Mars, Icarus, 207, 265–276, doi:10.1016/j.icarus.2009.10.015. Mischna, M. A., M. I. Richardson, R. J. Wilson, and D. J. McCleese (2003), On the orbital forcing of Martian water and CO2 cycles: A general circulation model study with 108(E6), Res-Planet, J Geophys simplified E65062, schemes, volatile doi:10.1029/2003JE002051. Mohlmann, D. T. F. (2008), The influence of van der Waals forces on the state of water in the shallow subsurface of Mars, Icarus, 195(1), 131-139. Montmessin, F., F. Forget, P. Rannou, M. Cabane, and R. M. Haberle (2004), Origin and role of water ice clouds in the Martian water cycle as inferred from a general circulation model, J Geophys Res-Planet, 109(E10), E10004, doi:10.1029/2004JE002284. Morozova, D., D. Mohlmann, and D. Wagner (2007), Survival of methanogenic archaea from Siberian permafrost under simulated Martian thermal conditions, Origins of Life and Evolution of Biospheres, 37(2), 189-200. Mustard, J. F., C. D. Cooper, and M. K. Rifkin (2001), Evidence for recent climate change on Mars from the identification of youthful near-surface ground ice, Nature, 412(6845), 411-414. Neukum, G., et al. (2004), Recent and episodic volcanic and glacial activity on Mars revealed by the High Resolution Stereo Camera, Nature, 432(7020), 971-979. Niles, P. B., and J. Michalski (2009), Meridiani Planum sediments on Mars formed through weathering in massive ice deposits, Nat Geosci, 2(3), 215-220. Parente, M. (2008), A New Approach to Denoising CRISM Images, 39th Lunar and Planetary Science Conference, (Lunar and Planetary Science XXXIX), held March 10-14, 2008 in League City, Texas. LPI Contribution No. 1391., 2528. Pavlov, A., A. Pavlov, and M. Vdovina (2009), Prospect of life on cold planets with low atmospheric pressures, American Geophysical Union, Fall Meeting 2009, abstract # B11E- 08. Price, P. B. (2007), Microbial life in glacial ice and implications for a cold origin of life, Fems Microbiol Ecol, 59(2), 217-231. Putzig, N. E., M. T. Mellon, K. A. Kretke, and R. E. Arvidson (2005), Global thermal inertia and surface properties of Mars from the MGS mapping mission, Icarus, 173(2), 325- 341. Schmitt, B., S. Douté, F. Altieri, G. Bellucci, and Omega (2004), Physical State And Composition Of Mars Polar Caps And Seasonal Condensations By Radiative Transfer Modelling Of Visible-IR Spectra From OMEGA/MEX Observations, American Astronomical Society, DPS meeting #36, #31.08; Bulletin of the American Astronomical Society, Vol. 36, 1137. 21 Schon, S. C., J. W. Head, and R. E. Milliken (2009), A recent ice age on Mars: Evidence for climate oscillations from regional layering in mid-latitude mantling deposits, Geophys Res Lett, 36, L15202, doi:10.1029/2009GL038554. Schorghofer, N. (2007), Dynamics of ice ages on Mars, Nature, 449(7159), 192-U192. Schorghofer, N., and K. S. Edgett (2006), Seasonal surface frost at low latitudes on Mars, Icarus, 180(2), 321-334. Smith, D. E., et al. (1999), The global topography of Mars and implications for surface evolution, Science, 284(5419), 1495-1503. Spiga, A., and F. Forget (2008), Fast and accurate estimation of solar irradiance on Martian slopes, Geophys Res Lett, 35(15), L15201, doi:10.1029/2008GL034956. Spiga, A., S. R. Lewis, F. Forget, E. Millour, L. Montabone and J.-B. Madeleine (2010), The impact of katabatic winds on Martian thermal inertia retrievals, 41th Lunar and Planetary Science Conference, (Lunar and Planetary Science XXXIX), held March 1-5, 2010 in The Woodlands, Texas. LPI Contribution No. 1533. Spinrad, H., and E. H. Richardson (1963), High dispersion spectra of the outer planets. II. A new upper limit for the water vapor content of the Martian atmosphere, Icarus, 2, 49-53. Squyres, S. W., and M. H. Carr (1986), Geomorphic Evidence for the Distribution of Ground Ice on Mars, Science, 231(4735), 249-252. Vincendon, M., Y. Langevin, F. Poulet, J. P. Bibring, and B. Gondet (2007), Recovery of surface reflectance spectra and evaluation of the optical depth of aerosols in the near-IR using a Monte Carlo approach: Application to the OMEGA observations of high-latitude regions of Mars, J Geophys Res-Planet, 112(E8), E08S13, doi:10.1029/2006JE002845. Vincendon, M., Y. Langevin, F. Poulet, J. P. Bibring, B. Gondet, D. Jouglet, and O. Team (2008), Dust aerosols above the south polar cap of Mars as seen by OMEGA, Icarus, 196(2), 488-505. Vincendon, M., J. Mustard, F. Forget, M. Kreslavsky, A. Spiga, S. Murchie, and J. P. Bibring (2010), Near-tropical subsurface ice on Mars, Geophys Res Lett, 37, L01202, doi:10.1029/2009GL041426. Vincendon, M., Y. Langevin, F. Poulet, A. Pommerol, M. Wolff, J. P. Bibring, B. Gondet, and D. Jouglet (2009), Yearly and seasonal variations of low albedo surfaces on Mars in the OMEGA/MEx dataset: Constraints on aerosols properties and dust deposits, Icarus, 200(2), 395-405. Williams, K. E., O. B. Toon, J. L. Heldmann, and M. T. Mellon (2009), Ancient melting of mid-latitude snowpacks on Mars as a water source for gullies, Icarus, 200(2), 418-425. Williams, K. E., O. B. Toon, J. L. Heldmann, C. Mckay, and M. T. Mellon (2008), Stability of mid-latitude snowpacks on Mars, Icarus, 196(2), 565-577. Zorzano, M. P., E. Mateo-Marti, O. Prieto-Ballesteros, S. Osuna, and N. Renno (2009), Stability of liquid saline water on present day Mars, Geophys Res Lett, 36, L20201, doi:10.1029/2009GL040315. 22
1511.05152	4	1511	2018-02-28T23:15:39	Inferring Planetary Obliquity Using Rotational & Orbital Photometry	[ "astro-ph.EP" ]	The obliquity of a terrestrial planet is an important clue about its formation and critical to its climate. Previous studies using simulated photometry of Earth show that continuous observations over most of a planet's orbit can be inverted to infer obliquity. However, few studies of more general planets with arbitrary albedo markings have been made and, in particular, a simple theoretical understanding of why it is possible to extract obliquity from light curves is missing. Reflected light seen by a distant observer is the product of a planet's albedo map, its host star's illumination, and the visibility of different regions. It is useful to treat the product of illumination and visibility as the kernel of a convolution. Time-resolved photometry constrains both the albedo map and the kernel, the latter of which sweeps over the planet due to rotational and orbital motion. The kernel's movement distinguishes prograde from retrograde rotation for planets with non-zero obliquity on inclined orbits. We demonstrate that the kernel's longitudinal width and mean latitude are distinct functions of obliquity and axial orientation. Notably, we find that a planet's spin axis affects the kernel -- and hence time-resolved photometry -- even if this planet is East-West uniform or spinning rapidly, or if it is North-South uniform. We find that perfect knowledge of the kernel at 2-4 orbital phases is usually sufficient to uniquely determine a planet's spin axis. Surprisingly, we predict that East-West albedo contrast is more useful for constraining obliquity than North-South contrast.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1–13 (2015) Printed 2 March 2018 (MN LATEX style file v2.2) Inferring Planetary Obliquity Using Rotational & Orbital Photometry J. C. Schwartz1,2,3(cid:63)†, C. Sekowski4,5, H. M. Haggard4, E. Pall´e6, and N. B. Cowan2,3† 1Department of Physics & Astronomy, Northwestern University, 2145 Sheridan Road, Evanston, IL, 60208, USA 2Department of Earth & Planetary Sciences, McGill University, 3450 rue University, Montreal, QC, H3A 0E8, CAN 3Department of Physics, McGill University, 3600 rue University, Montreal, QC, H3A 2T8, CAN 4Physics Program, Bard College, PO Box 5000, Annandale, NY, 12504, USA 5Department of Physics, Boston University, 590 Commonwealth Ave, Boston, MA, 02215, USA 6Instituto de Astrof´ıscia de Canarias, Via L´actea s/n, La Laguna, Santa Cruz de Tenerife, 35205, ESP Published in MNRAS ABSTRACT The obliquity of a terrestrial planet is an important clue about its formation and critical to its climate. Previous studies using simulated photometry of Earth show that continuous obser- vations over most of a planet's orbit can be inverted to infer obliquity. However, few studies of more general planets with arbitrary albedo markings have been made and, in particular, a simple theoretical understanding of why it is possible to extract obliquity from light curves is missing. Reflected light seen by a distant observer is the product of a planet's albedo map, its host star's illumination, and the visibility of different regions. It is useful to treat the prod- uct of illumination and visibility as the kernel of a convolution. Time-resolved photometry constrains both the albedo map and the kernel, the latter of which sweeps over the planet due to rotational and orbital motion. The kernel's movement distinguishes prograde from ret- rograde rotation for planets with non-zero obliquity on inclined orbits. We demonstrate that the kernel's longitudinal width and mean latitude are distinct functions of obliquity and axial orientation. Notably, we find that a planet's spin axis affects the kernel-and hence time- resolved photometry-even if this planet is East-West uniform or spinning rapidly, or if it is North-South uniform. We find that perfect knowledge of the kernel at 2–4 orbital phases is usually sufficient to uniquely determine a planet's spin axis. Surprisingly, we predict that East-West albedo contrast is more useful for constraining obliquity than North-South contrast. Key words: methods: analytical – methods: statistical – planets and satellites: fundamental parameters. 1 INTRODUCTION The obliquity of a terrestrial planet encodes information about dif- ferent processes. A planet's axial alignment and spin rate inform its formation scenario. Numerical simulations have shown that the spin rates of Earth and Mars are likely caused by a few planetes- imal impacts (Dones & Tremaine 1993), while perfect accretion produces an obliquity distribution that is isotropic (e.g. Kokubo & Ida 2007; Miguel & Brunini 2010). Conversely, Schlichting & Sari (2007) describe how prograde rotation is preferred to retrograde for a formation model with semi-collisional accretion. Obliquity is also important in controlling planetary climate. This has been studied in-depth for Earth under many conditions (e.g. Laskar et al. 2004; Pierrehumbert 2010), and high axial tilts can make planets at large semi-major axes more habitable (cid:63) E-mail: [email protected] † McGill Space Institute (McGill U.); Institute for Research on Exoplanets (UdeM) c(cid:13) 2015 RAS (Williams & Kasting 1997). Furthermore, while the Earth's spin axis is stabilized by the Moon (Laskar et al. 1993), obliquities of several Solar System bodies evolve chaotically (Laskar 1994). This influences searches for hospitable planets, as Spiegel et al. (2009) note that the habitability of terrestrial worlds may depend sensi- tively on how stable the climate is in the short-term. A planet's average insolation is set by stellar luminosity and semi-major axis; insolation at different latitudes is determined by obliquity and (for eccentric orbits) the axial orientation. Non- oblique planets have a warmer equator and colder poles that do not vary much throughout the year. Modest obliquities produce sea- sons at mid-latitudes because the sub-stellar point moves North and South during the orbit (Pierrehumbert 2010). Planets tilted at angles (cid:62) 54◦ receive more overall radiation near their poles and have large orbital variations in temperature (Williams & Pollard 2003). Thus, even limited knowledge of a planet's obliquity can help constrain the spatial dependence of insolation and tempera- ture. 2 J. C. Schwartz et al. Figure 1. The left panel shows apparent albedo as a function of orbital phase for an arbitrary planet with North-South and East-West albedo markings, seen edge-on with zero obliquity (black) or 45◦ obliquity (green). The average albedo of the green planet increases during the orbit because brighter latitudes become visible and illuminated; this does not happen for the black planet. The vertical bands are each roughly two planetary days, enlarged at right, where lighter shades are the fuller phase. For clarity, the rotational curves are shifted and the zero-obliquity planet is denoted by a dashed line. The apparent albedo of both planets varies more over a day when a narrower range of longitudes and albedo markings are visible and illuminated, and vice versa. Note that both light curves are produced by the same albedo map, and are distinct solely because of differences in the kernel for these two geometries. Orbital and/or rotational changes in apparent albedo can help one infer a planet's obliquity. Numerous methods have been proposed for measuring plan- etary obliquities. Seager & Hui (2002) and Barnes & Fortney (2003) demonstrated constraints on oblateness and obliquity us- ing ingress/egress differences in transit light curves; Carter & Winn (2010) extended and applied these techniques to observa- tions of HD 189733b. Kawahara (2012) derived constraints on obliquity from modulation of a planet's radial velocity during or- bit, while Nikolov & Sainsbury-Martinez (2015) examined the Rossiter-McLauglin effect at secondary eclipse for transiting exo- planets. One could also measure obliquity at infrared wavelengths, using polarized rotational light curves (De Kok et al. 2011) and or- bital variations (e.g. Gaidos & Williams 2004; Cowan et al. 2013). A planet's obliquity can also be constrained by changes in reflected light, which will be studied with forthcoming optical and near-infrared space missions, such as ATLAST (Postman et al. 2010), LUVOIR (Kouveliotou et al. 2014), and HDST (Dalcanton et al. 2015). Time-resolved measurements of a rotating planet in one photometric band can reveal its rotation rate (Ford et al. 2001; Pall´e et al. 2008; Oakley & Cash 2009); this helps determine Cori- olis forces and predict large-scale circulation. Multi-band photom- etry can reveal colors of clouds and surface features (Ford et al. 2001; Fujii et al. 2010, 2011; Cowan & Strait 2013), and enables a longitudinal albedo map to be inferred from disk-integrated light (Cowan et al. 2009). High-cadence, reflected light measurements spanning a full planetary orbit constrain a planet's obliquity and two-dimensional albedo map (Kawahara & Fujii 2010, 2011; Fujii & Kawahara 2012). However, these results have not yet been established for lower cadence measurements of non-terrestrial planets. We also hope to establish a conceptual understanding of how photometric measure- ments constrain obliquity. While this is less immediately practical, a deeper understanding of this inversion has the potential to lead to further advances in inferring planetary geometry from limited data sets. In this paper, we study light curve methods for arbitrary albedo maps and viewing geometries, and demonstrate they are useful even for observations at only one or two orbital phases. Light curves of planets encode the viewing geometry and hence a planet's obliquity because different latitudes are impinged by starlight at different orbital phases (this "kernel" is described in Section 2.2). To see this, consider a planet with no obliquity in an edge-on, circular orbit. The star always illuminates the Northern and Southern hemispheres equally, and we never view some lati- tudes more than others. If instead this planet were tilted, the North- ern hemisphere would be lit first, then the Southern hemisphere half an orbit later. If the planet is not North-South uniform, its apparent albedo (Qui et al. 2003; Cowan et al. 2009) would change during its orbit, shown in the left panel of Figure 1. One may also learn about a planet's obliquity as it rotates. Imagine a zero-obliquity planet in a face-on, circular orbit: the ob- server always sees the Northern pole with half the longitudes illu- minated. For an oblique planet, however, more longitudes would be lit when the visible pole leans towards the star, and vice versa. Zero-obliquity planets in edge-on orbits are similar, since more lon- gitudes are lit near superior conjunction, or fullest phase. If the planet has East-West albedo variations, then this longitudinal width will modulate the apparent albedo of the planet as it spins, shown in the right panel of Figure 1. Our work is organized as follows: in Section 2, we summa- rize the observer viewing geometry and explain the reflective ker- nel, both in two- and one-dimensional forms. Section 3.1 intro- duces a case study planet and describes the kernel at single or- bital phases; we consider time evolution in Section 3.2. We discuss real observations in Section 4.1, then develop our case study in Sections 4.2–4.4, demonstrating that even single- and dual-epoch observations could allow one to constrain obliquity. In Section 4.5, we discuss how to distinguish a planet's rotational direction by monitoring its apparent albedo. Section 5 summarizes our conclu- sions. For interested readers, a full mathematical description of the illumination and viewing geometry is presented in Appendix A. Details about the kernel and its relation to a planet's apparent albedo are described in Appendix B. c(cid:13) 2015 RAS, MNRAS 000, 1–13 30◦60◦90◦120◦150◦Orbital Phase0.10.20.30.40.50.6Apparent Albedo0◦2◦4◦6◦8◦Relative Orbital Phase0.10.20.30.40.50.6Relative Apparent AlbedoInferring Planetary Obliquity 3 2 REFLECTED LIGHT 2.1 Geometry & Flux The locations on a planet that contribute to the disk-integrated re- flected light depend only on the sub-observer and sub-stellar po- sitions, which both vary in time. A complete development of this viewing geometry is provided in Appendix A, which we summarize here. We neglect axial precession and consider planets on circular orbits. Assuming a static albedo map, the reflected light seen by an observer is determined by the colatitude and longitude of the sub-stellar and sub-observer points, explicitly θs, φs, θo, and φo. The intrinsic parameters of the system are the orbital and rotational angular frequencies, ωorb and ωrot (where positive ωrot is prograde), and the planetary obliquity, Θ ∈ [0, π/2]. Extrinsic parameters dif- fer from one observer to the next; we denote the orbital inclination, i (where i = 90◦ is edge-on), and solstice phase, ξs (the orbital phase of Summer solstice for the Northern hemisphere). We also define initial conditions for orbital phase, ξ0, and the sub-observer longitude, φo(0). Reflected light is then completely specified by these seven parameters and the planet's albedo map. We consider only diffuse (Lambertian) reflection in our analy- sis. Specular reflection, or glint, can be useful for detecting oceans (Williams & Gaidos 2008; Robinson et al. 2010, 2014), but is a lo- calized feature and a minor fraction of the reflected light at gibbous phases. The reflected flux measured by a distant observer is there- fore a convolution of the two-dimensional kernel (or weight func- tion; Fujii & Kawahara 2012), K(θ, φ, S), and the planet's albedo map, A(θ, φ): F (t) = K(θ, φ, S)A(θ, φ)dΩ, (1) (cid:73) where F is the observed flux, θ and φ are colatitude and lon- gitude, and S ≡ {θs, φs, θo, φo} implicitly contains the time- dependencies in the sub-stellar and sub-observer locations. Recon- structing a map of an exoplanet based on time-resolved photometry can be thought of as a deconvolution (Cowan et al. 2013), while es- timating a planet's obliquity amounts to backing out the kernel of the convolution. The sub-stellar and sub-observer points are completely determined through a function S = f (G, ωrott), where G ≡ {ξ(t), i, Θ, ξs} is the system geometry and ξ(t) is or- bital phase. This is made explicit in Appendix A. For a planet with albedo markings, one would therefore fit the observed flux to in- fer both the planet's albedo map (Cowan et al. 2009) and spin axis (Kawahara & Fujii 2010, 2011; Fujii & Kawahara 2012). To study how these methods work for arbitrary maps and geometries, we will focus on the kernel from Equation 1, which we can analyze independent of a planet's albedo map. 2.2 Kernel The kernel combines illumination and visibility, defined for diffuse reflection in Cowan et al. (2013) as K(θ, φ, S) = 1 π V (θ, φ, θo, φo)I(θ, φ, θs, φs), (2) where V (θ, φ, θo, φo) is the visibility and I(θ, φ, θs, φs) is the il- lumination. Visibility and illumination are each non-zero over one hemisphere at any time, and are further given by V (θ, φ, θo, φo) = max(cid:2) sin θ sin θo cos(φ − φo) + cos θ cos θo, 0(cid:3), (3) c(cid:13) 2015 RAS, MNRAS 000, 1–13 (4) (5) Figure 2. Upper: A kernel, in gray, with contours showing visibility and illumination, in purple and yellow, as in Cowan et al. (2013). The sub- observer and sub-stellar points are indicated by the purple circle and yel- low star, respectively. The orange diamond marks the peak of the kernel. Lower: The mean of the longitudinal kernel and the width from this mean are shown as solid and dashed red lines; the dominant colatitude is shown as a blue line. I(θ, φ, θs, φs) = max(cid:2) sin θ sin θs cos(φ − φs) + cos θ cos θs, 0(cid:3). As noted above, we can express the kernel equivalently as K(θ, φ, S) = K(θ, φ, G, ωrott), though we will drop the rotational dependence for now because it does not affect our analysis. We return to rotational frequency in Section 4.5. The non-zero portion of the kernel is a lune: the illuminated region of the planet that is visible to a given observer. The size of this lune depends on orbital phase, or the angle between the sub- observer and sub-stellar points. A sample kernel is shown at the top of Figure 2, where the purple and yellow contours are visibility and illumination, respectively. The peak of the kernel is marked with an orange diamond. We begin by calculating time-dependent sines and cosines of the sub-observer and sub-stellar angles for a viewing geometry of interest (Appendix A). These are substituted into Equations 3 and 4 to determine visibility and illumination at any orbital phase. The two-dimensional kernel is then calculated on a 101 × 201 grid in colatitude and longitude. 2.3 Longitudinal Width The two-dimensional kernel, K(θ, φ, G), is a function of lati- tude and longitude that varies with time and viewing geometry. For observations with minimal orbital coverage or planets that are North-South uniform, different latitudes are hard to distin- guish (Cowan et al. 2013) and we use the longitudinal form of the 0◦45◦90◦135◦180◦0◦90◦180◦270◦360◦Longitude0◦45◦90◦135◦180◦Colatitude4 J. C. Schwartz et al. Figure 3. Left panels: Contours of longitudinal kernel width at first quarter phase, ξ(t) = 90◦, as a function of planetary obliquity, from face-on i = 0◦ to edge-on i = 90◦. Obliquity is plotted radially: the center is Θ = 0◦ and the edge is Θ = 90◦. The azimuthal angle represents the orientation (solstice phase) of the planet's obliquity. The contours span 20◦–100◦, dark to light colors, in 5◦ increments. Larger kernel widths-and hence muted rotational variability- occur at gibbous phases, and when the peak of the kernel is near a pole. Right panels: Analogous contours of dominant colatitude that span 35◦–145◦. A lower dominant colatitude-and thus more reflection from the Northern hemisphere-occurs when the kernel samples Northern regions more, and vice versa. kernel, K(φ, G), given by K(φ, G) = (cid:90) π 0 K(θ, φ, G) sin θdθ. (6) We can approximately describe K(φ, G) by a longitudinal mean, ¯φ, and width, σφ. These are defined in Appendix B1; examples are shown as vertical red lines in the bottom panel of Figure 2. For any geometry, we can calculate the two-dimensional ker- nel and the corresponding longitudinal width. The mean lon- gitude is unimportant by itself because, for now, we are only concerned with the size of the kernel. We compute a four- dimensional grid of kernel widths with 5◦ resolution in orbital phase (time), inclination, obliquity, and solstice phase. The re- sult is σφ(ξ(t), i, Θ, ξs) ≡ σφ(G), and our numerical grid has size 73 × 19 × 19 × 73 in the respective parameters. Example con- tours from this array at first quarter phase, or ξ(t) = 90◦, are shown in the left panels of Figure 3. In these plots obliquity is radial: the center is Θ = 0◦ and the edge is Θ = 90◦. The azimuthal angle gives the orientation (solstice phase) of the planet's obliquity. 2.4 Dominant Colatitude For a given planet and observer, the sub-observer colatitude is fixed but the sub-stellar point moves North and South throughout the or- bit if the planet has non-zero obliquity. This means different orbital phases will probe different latitudes, as dictated by the kernel. To analyze these variations we use the latitudinal form of the kernel, K(θ, G), explicitly: K(θ, G) = K(θ, φ, G)dφ. (7) We may describe K(θ, G) by its dominant colatitude (Cowan et al. 2012), ¯θ, also defined in Appendix B1 and shown as a horizontal 0 (cid:90) 2π blue line at the bottom of Figure 2. We produce a four-dimensional dominant colatitude array, ¯θ(ξ(t), i, Θ, ξs) ≡ ¯θ(G), similarly to σφ(G) from Section 2.3. Sample contours from this array at first quarter phase are shown in the right panels of Figure 3. 3 KERNEL BEHAVIOR We now consider how the longitudinal width and dominant colati- tude of the kernel depend on a planet's obliquity. As a case study, we will define the inclination and spin axis of a hypothetical planet, Q:  i = 60 Θ = 55 ξs = 260  . ◦ ◦ ◦ planet Q ≡ (8) 3.1 Phases Considering a single orbital phase defines a three-dimensional slice through σφ(G) and ¯θ(G) that describes the kernel at that specific time. We show the longitudinal form of the kernel for planet Q at different phases in the left panel of Figure 4. Lighter colors are fuller phases, indicating the kernel narrows as this planet orbits towards inferior conjunction, or ξ(t) = 180◦. The kernel width influences the rotational light curve at a given phase: narrower ker- nels can have larger amplitude variability in apparent albedo on a shorter timescale (e.g. right of Figure 1). The latitudinal kernel for planet Q is shown similarly in the right panel of Figure 4. We see that the kernel preferentially probes low and mid-latitudes during the first half-orbit. The dominant co- latitude of planet Q, indicated by circles, also fluctuates during this portion of the orbit-and eventually shifts well into the Northern c(cid:13) 2015 RAS, MNRAS 000, 1–13 0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=0◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=0◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=30◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=30◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=90◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦i=90◦Inferring Planetary Obliquity 5 Figure 4. Left: Longitudinal kernel for planet Q, defined in Equation 8, at orbital phases from 30◦ to 150◦, light to dark shades, in 30◦ increments. Values are scaled to the maximum of the lightest curve. Longitude is measured from each kernel mean. The kernel width decreases as this planet approaches inferior conjunction, or as color darkens. Right: Analogous latitudinal kernel for planet Q, where the dominant colatitude, indicated by a circle, increases then returns towards the equator. hemisphere after inferior conjunction (not shown). Note that the dominant colatitude is not always at the peak of the latitudinal ker- nel (see also Figure A2). Since one needs measurements at mul- tiple phases to be sensitive to latitudinal variations in albedo, we will consider changes in dominant colatitude from one phase to the next, ∆¯θ, for planets with North-South albedo markings. Larger changes in dominant colatitude can make the apparent albedo vary more between orbital phases (e.g. left of Figure 1). 3.2 Time Evolution Kernel width and dominant colatitude both vary throughout a planet's orbit. We investigate this by slicing σφ(G) and ¯θ(G) along obliquity and/or solstice phase. To start, we vary planet Q's obliq- uity and track kernel width as shown in the left panel of Figure 5. The actual planet Q is denoted by a dashed green line: this planet has a narrow kernel width during the first half-orbit that widens sharply after inferior conjunction. The largest variations between the traces occur near ξ(t) ≈ {120◦, 210◦}. We also show tracks of dominant colatitude in the right panel of Figure 5. What matters is the change in this characteristic be- tween two epochs; diverse changes in the traces occur between ξ(t) ≈ {135◦, 240◦}. Planet Q is again the dashed green line, and near the middle of all the tracks more often than for kernel width. If one has some prior knowledge of the viewing geometry, then Figure 5 implies at which phases one could observe to best distin- guish obliquities for planet Q-for example, ξ(t) ≈ {120◦, 240◦}. We can instead vary the solstice phase of planet Q while keep- ing its obliquity fixed (not shown). In most cases, solstice phase impacts the kernel width and dominant colatitude as much as the axial tilt. This is expected, since obliquity is a vector quantity with both magnitude and orientation. 4 DISCUSSION 4.1 Observations By analyzing the kernel, we can learn how observed flux may de- pend on a planet's obliquity independent from its albedo map. For c(cid:13) 2015 RAS, MNRAS 000, 1–13 real observations, one would fit the light curve to directly infer the planet's albedo map (Cowan et al. 2009) and spin axis (Kawahara & Fujii 2010, 2011; Fujii & Kawahara 2012). We will use the ker- nel to predict how single- and dual-epoch observations constrain planetary obliquity. We address our assumptions below. The planetary inclination and orbital phase of observation must be known to model the light curve accurately. Both angles might be obtained with a mixture of astrometry on the host star (e.g. SIM PlanetQuest; Unwin et al. 2008), direct-imaging astrom- etry (Bryden 2015), and/or radial velocity. We will assume that in- clination and orbital phase have each been measured with 10◦ un- certainty. Extracting the albedo map and spin axis from a light curve could also be difficult in practice. Planets with completely uni- form albedo are not amenable to these methods. Moreover, one cannot distinguish latitudes for planets that are North-South uni- form, nor longitudes for those that are East-West uniform. Even if a planet has albedo contrast, photometric uncertainty adds noise to the reflected light measurements. Contrast ratios (cid:54) 10−11 are needed to resolve rotational light curves of an Earth-like exoplanet (Pall´e et al. 2008), which should be achievable by a TPF-type mis- sion with high-contrast coronagraph or starshade (Ford et al. 2001; Trauger & Traub 2007; Turnbull et al. 2012; Cheng-Chao et al. 2015). We will implicitly assume that planet Q has both East-West and North-South albedo markings, and thus that the kernel geom- etry impacts the reflected light. In particular, we will assume two scenarios: perfect knowledge of the kernel, or kernel widths and changes in dominant colatitude that are constrained to ±10◦ and ±20◦, respectively, explained in Appendix B2. Note that these un- certainties will depend on the planet's albedo contrast, and the pho- tometric precision, in a non-linear way. We envision a triage ap- proach for direct-imaging missions: planets that vary in brightness the most, and thus have the easiest albedo maps and spin axes to infer, will be the first for follow-up observations. Of course, planetary radii are necessary to convert fluxes into apparent albedos (Qui et al. 2003; Cowan et al. 2009). Radii will likely be unknown, but could be approximated using mass-radius relations and mass estimates from astrometry or radial velocity, or −180◦−90◦0◦90◦180◦Longitude0.250.500.751.00Scaled Kernel0.250.500.751.00Scaled Kernel0◦45◦90◦135◦180◦Colatitude6 J. C. Schwartz et al. Figure 5. Left: Kernel width for planet Q as a function of orbital phase, with obliquity varied in 5◦ increments. The defined planet Q obliquity, Θ = 55◦, is the dashed green line; darker and lighter shades of red denote obliquities closer to 0◦ and 90◦, respectively. Inferior conjunction occurs at ξ(t) = 180◦. The largest variations are near ξ(t) ≈ {120◦, 210◦}. Right: Analogous dominant colatitude for planet Q. Dual-epoch changes are diverse between ξ(t) ≈ {135◦, 240◦}, for example. inferred from bolometric flux using thermal infrared direct-imaging (e.g. TPF-I; Beichman et al. 1999; Lawson et al. 2008). Real plan- ets may also have variable albedo maps, e.g. short-term variations from changing clouds and smaller variations from long-term sea- sonal changes (Robinson et al. 2010), that could influence the ap- parent albedo on orbital timescales. These are difficulties that will be mitigated with each iteration of photometric detectors and theo- retical models. 4.2 Longitudinal Constraints A fit to the rotational light curve can be used to constrain the spin axis (and longitudinal map) of a planet with East-West albedo con- trast. We can demonstrate these constraints using kernel widths in two ways, described in Section 4.1 and shown in the upper panels of Figure 6. The dark dashed lines and square are idealized con- straints when assuming perfect knowledge of the orbital geome- try and two kernel widths: σφ1 = 25.2◦ at ξ(t1) = 120◦, and σφ2 = 51.7◦ at ξ(t2) = 240◦. Alternatively, the red regions have 10◦ uncertainty on each width (Appendix B2), where we use a nor- malized Gaussian probability density and include Gaussian weights for uncertainties on inclination and orbital phase. The green circles represent the true planet Q spin axis, which always lies on the idealized constraints. Only two spin axes are consistent with the ideal kernel widths from both orbital phases. We run more numerical experiments for a variety of system ge- ometries (not shown) and find that perfect knowledge of the kernel width at three orbital phases uniquely determines the planetary spin axis. However, we find a degeneracy for planets with edge-on or- bits, where two different spin configurations will produce the same kernel widths at all phases. As anticipated, we also find planet Q's spin axis (green cir- cle) consistent with the uncertain kernel widths (dark red regions). Imperfect kernel widths at two phases allow all obliquities above 15◦ at 1σ, but exclude nearly one-fifth of spin axes at 3σ. We find similar predictions for other orbital phases and planet parameters. These examples also suggest that obliquity could be con- strained for planets with variable albedo maps. As long as albedo only changes on timescales longer than the rotational period, light curves will constrain both the instantaneous map and the planet's spin axis. A given spin orientation and orbital inclination dictates a specific kernel width as a function of orbital phase (left panel of Figure 5), so we predict that light curves at three phases will be sufficient to pin down the planetary obliquity, even if the planet's map varies between phases. 4.3 Latitudinal Constraints A fit to light curves from different orbital phases can be used to constrain the spin axis (and latitudinal map) of a planet with North- South albedo contrast. As described in Section 4.1, we can demon- strate this constraint using both perfect and uncertain knowledge of the change in dominant colatitude. Our predictions are shown in the lower left panel of Figure 6. The idealized constraint here is ∆¯θ12 = 76.0◦ between ξ(t) = {120◦, 240◦}. Since the change in dominant colatitude is constrained between pairs of epochs, four orbital phases are needed to produce three independent constraints and uniquely determine planet Q's spin axis. We find the same two-fold degeneracy as before for planets in edge-on orbits, even if one knows the change in dominant colatitude between all pairs of phases. For the blue regions, we reapply our probability density from above and assume 20◦ uncertainty on the change in dominant co- latitude (Appendix B2). The distribution is bimodal because only the magnitude of the change can be constrained, not its direction. This means an observer would not know whether more Northern or Southern latitudes are probed at the later phase, affecting which spin axes are inferred. 4.4 Joint Constraints For a planet with both East-West and North-South albedo contrast, one may combine longitudinal and latitudinal information to bet- ter constrain the planet's spin axis (and two-dimensional map). We show this for planet Q in the lower right panel of Figure 6. The idealized constraint shows that only the true spin configuration is c(cid:13) 2015 RAS, MNRAS 000, 1–13 0◦90◦180◦270◦360◦0◦25◦50◦75◦100◦Kernel Width0◦90◦180◦270◦360◦0◦45◦90◦135◦180◦Dominant ColatitudeOrbital PhaseInferring Planetary Obliquity 7 Figure 6. Predicted confidence regions for planet Q's spin axis, from hypothetical single- and dual-epoch observations. The constraints are predicted using the kernel, as described in Sections 4.1–4.4 and Appendix B2: longitudinal widths in the upper row (red), the change in dominant colatitude at the lower left (blue), and joint constraints at the lower right (purple). Obliquity is plotted radially: the center is Θ = 0◦ and the edge is Θ = 90◦. The azimuthal angle represents the planet's solstice phase. The green circles are the true planet Q spin axis, while the dark dashed lines and square show idealized constraints assuming perfect knowledge of the orbital geometry and kernel (i.e. no uncertainties). The upper left and center panels describe planet Q at ξ(t) = {120◦, 240◦}, respectively, while the lower left panel incorporates both phases. For the colored regions, 10◦ uncertainty is assumed on each kernel width, inclination, and orbital phase, while 20◦ uncertainty is assumed on the change in dominant colatitude. Regions up to 3σ are shown, where darker bands are more likely. Observing a planet at just a few orbital phases can significantly constrain both its obliquity and axial orientation. allowed. We find this result for other system geometries-except using orbital phases 180◦ apart, which creates a two-way degener- acy in the spin axis. The confidence regions in purple assume our notional uncer- tainties on both kernel width and change in dominant colatitude (Appendix B2). This prediction is not unimodal, but the 1σ region excludes obliquities below 30◦. A distant observer would know that this planet's obliquity has probably not been eroded by tides (Heller et al. 2011), and that the planet likely experiences obliquity seasons. 4.5 Pro/Retrograde Rotation The sign of rotational angular frequency (positive = prograde) can affect the mean longitude of the kernel, but not its size and shape. There is a formal degeneracy for edge-on, zero-obliquity cases: prograde planets with East-oriented maps have identical light curves to retrograde planets with West-oriented maps. The motion of the kernel peak is the same over either version of the planet, implying the retrograde rotation in an inertial frame is slower (Appendix B3). We show this scenario in the left panel of Figure 7, where the dashed brown line is the difference in prograde and ret- rograde apparent albedo. The orange and black planets are always equally bright because the same map features, in the upper panels, are seen at the same times. However, the spin direction of oblique planets and/or those on inclined orbits may be deduced. Inclinations that are not edge- on most strongly alter a planet's light curve near inferior conjunc- c(cid:13) 2015 RAS, MNRAS 000, 1–13 tion, seen in the center panel of Figure 7: this planet's properties are intermediate between the edge-on, zero-obliquity planet and planet Q. While a typical observatory's inner working angle would hide some of the signal, differences on the order of 0.1 in the ap- parent albedo would be detectable at extreme crescent phases. Al- ternatively, higher obliquity causes deviations that-depending on solstice phase-can arise around one or both quarter phases. This happens for planet Q in the right panel of Figure 7, where both effects combine to distinguish the spin direction at most phases. Inclination and obliquity influence apparent albedo because the longitudinal motion of the kernel peak is not the same at all latitudes. One can break this spin degeneracy in principle, but we have not fully explored the pro/retrograde parameter space. In gen- eral, the less inclined and/or oblique a planet is, the more favorable crescent phases are for determining its spin direction. 5 CONCLUSIONS We have performed numerical experiments to study the problem of inferring a planet's obliquity from time-resolved photometry, for arbitrary albedo maps and viewing geometries. We have demon- strated that a planet's obliquity will influence its light curve in two distinct ways: one involving East-West albedo markings and an- other involving North-South markings. Provided this planet is not completely uniform, one could constrain both its albedo map and spin axis using reflected light. The kernel-the product of visibility and illumination-has a 0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦0◦45◦90◦135◦180◦225◦270◦315◦0◦30◦60◦×=.→8 J. C. Schwartz et al. Figure 7. Apparent albedo as a function of orbital phase, shown for an edge-on, zero-obliquity planet on the left, planet Q on the right, and an intermediate planet in the center. A low rotational frequency is used for clarity. The black and orange curves correspond to prograde and retrograde rotation, respectively, and the differences in apparent albedo are the dashed brown lines. Inferior conjunction occurs at ξ(t) = 180◦. The albedo maps are color-coded at the top, where arrows indicate spin direction and the prime meridians are centered. Note that these maps are East-West reflections of each other. The edge-on, zero-obliquity curves are identical, while the curves for the intermediate planet and planet Q grow more distinct. Edge-on, zero-obliquity planets are hopeless, but one can distinguish pro/retrograde rotation for inclined, oblique planets by monitoring their brightness, particularly near crescent phases. peak, a longitudinal width, and a mean latitude that vary in time and are functions of viewing geometry. Analyzing the kernel enables us to predict constraints on a planet's spin axis from reflected light, including for maps that are East-West uniform (e.g. Jupiter-like) or North-South uniform (e.g. beach ball-like). Curiously, we find that kernel width offers better constraints on obliquity than dominant colatitude, suggesting that East-West albedo contrast is generally more useful than North-South contrast. This is partly because ker- nel width can be constrained even for variable albedo maps. Furthermore, monitoring a planet at only a few epochs could determine its spin direction and significantly constrain its obliq- uity. In our case study of planet Q, we find crescent phases are favorable for telling prograde from retrograde rotation. Similarly, perfect knowledge of the kernel width at two orbital phases nar- rows the possible spin axes for planet Q to two distinct configu- rations, while kernel width uncertainties of 10◦ still exclude about three-quarters of spin axes at 1σ. Adding the constraint on change in dominant colatitude between the same two phases completely specifies the true spin configuration of planet Q. A change in dom- inant colatitude with 20◦ uncertainty excludes five-sixths of spin orientations at 1σ. Most importantly, we also find that perfect knowledge of the kernel width at just three phases, or its change in dominant co- latitude between four phases, is generally sufficient to uniquely determine a planet's obliquity. This suggests that-in principle- rotational light curves at 2–4 distinct orbital phases uniquely con- strain the spin axis of any planet with non-uniform albedo. This is good news for inferring the obliquity of planets with future direct- imaging missions. ACKNOWLEDGMENTS The authors thank the anonymous referee for important sugges- tions that improved the paper, and the International Space Sci- ence Institute (Bern, CH) for hosting their research workshop, "The Exo-Cartography Inverse Problem." Participants included Ian M. Dobbs-Dixon (NYUAD), Ben Farr (U. Chicago), Will M. Farr (U. Birmingham), Yuka Fujii (TIT), Victoria Meadows (U. Washing- ton), and Tyler D. Robinson (UC Santa Cruz). JCS was funded by an NSF GK-12 fellowship, and as a Graduate Research Trainee at McGill University. c(cid:13) 2015 RAS, MNRAS 000, 1–13 0◦90◦180◦270◦360◦0.250.000.250.500.75Apparent Albedo0◦90◦180◦270◦360◦Orbital Phase0.250.000.250.500.750◦90◦180◦270◦360◦0.250.000.250.500.75REFERENCES Barnes J. W., Fortney J. J., 2003, The Astrophysical Journal, 588, 545 Beichman C. A., Woolf N., Lindensmith C., 1999, The Terrestrial Planet Finder (TPF): a NASA Origins Program to search for habitable plan- ets/the TPF Science Working Group; [Washington, DC]: NASA (JPL publication; 99-3) Bryden G., 2015, IAU General Assembly, 22, 58195 Carter J. A., Winn J. N., 2010, The Astrophysical Journal, 709, 1219 Cheng-Chao L., De-Qing R., Jian-Pei D., Yong-Tian Z., Xi Z., Gang Z., Zhen W., Rui C., 2015, Research in Astronomy and Astrophysics, 15, 453 Cowan N. B., Abbot D. S., Voigt A., 2012, The Astrophysical Journal Letters, 752, L3 Cowan N. B., Agol E., Meadows V. S., Robinson T., Livengood T. A., Deming D., Lisse C. M., A'Hearn M. F., Wellnitz D. D., Seager S., et al., 2009, The Astrophysical Journal, 700, 915 Cowan N. B., Fuentes P. A., Haggard H. M., 2013, Monthly Notices of the Royal Astronomical Society, 434, 2465 Cowan N. B., Strait T. E., 2013, The Astrophysical Journal Letters, 765, L17 Dalcanton J., Seager S., Aigrain S., Battel S., Brandt N., Conroy C., Fein- berg L., Gezari S., Guyon O., Harris W., et al., 2015, arXiv preprint arXiv:1507.04779 De Kok R., Stam D., Karalidi T., 2011, The Astrophysical Journal, 741, 59 Dones L., Tremaine S., 1993, Icarus, 103, 67 Ford E., Seager S., Turner E., 2001, Nature, 412, 885 Fujii Y., Kawahara H., 2012, The Astrophysical Journal, 755, 101 Fujii Y., Kawahara H., Suto Y., Fukuda S., Nakajima T., Livengood T. A., Turner E. L., 2011, The Astrophysical Journal, 738, 184 Fujii Y., Kawahara H., Suto Y., Taruya A., Fukuda S., Nakajima T., Turner E. L., 2010, The Astrophysical Journal, 715, 866 Gaidos E., Williams D., 2004, New Astronomy, 10, 67 Heller R., Leconte J., Barnes R., 2011, Astronomy & Astrophysics, 528, A27 Kawahara H., 2012, The Astrophysical Journal Letters, 760, L13 Kawahara H., Fujii Y., 2010, The Astrophysical Journal, 720, 1333 Kawahara H., Fujii Y., 2011, The Astrophysical Journal Letters, 739, L62 Kokubo E., Ida S., 2007, The Astrophysical Journal, 671, 2082 Kouveliotou C., Agol E., Batalha N., Bean J., Bentz M., Cornish N., Dressler A., Figueroa-Feliciano E., Gaudi S., Guyon O., et al., 2014, arXiv preprint arXiv:1401.3741 Laskar J., 1994, Astronomy and Astrophysics, 287, L9 Laskar J., Joutel F., Robutel P., 1993, Nature, 361, 615 Laskar J., Robutel P., Joutel F., Gastineau M., Correia A., Levrard B., 2004, Astronomy & Astrophysics, 428, 261 Lawson P., Lay O., Martin S., Peters R., Gappinger R., Ksendzov A., Scharf D., Booth A., Beichman C., Serabyn E., et al., 2008, in SPIE Astronomical Telescopes + Instrumentation Terrestrial planet finder in- terferometer: 2007-2008 progress and plans. pp 70132N–70132N Miguel Y., Brunini A., 2010, Monthly Notices of the Royal Astronomical Society, 406, 1935 Nikolov N., Sainsbury-Martinez F., 2015, The Astrophysical Journal, 808, 57 Oakley P., Cash W., 2009, The Astrophysical Journal, 700, 1428 Pall´e E., Ford E. B., Seager S., Montan´es-Rodr´ıguez P., Vazquez M., 2008, The Astrophysical Journal, 676, 1319 Pierrehumbert R. T., 2010, Principles of planetary climate. Cambridge University Press Postman M., Traub W., Krist J., Stapelfeldt K., Brown R., Oegerle W., Lo A., Clampin M., Soummer R., Wiseman J., et al., 2010, in Pathways To- wards Habitable Planets Vol. 430, Advanced technology large-aperture space telescope (atlast): Characterizing habitable worlds. p. 361 Qui J., Goode P., Pall´e E., Yurchyshyn V., Hickey J., Montan´es-Rodrıguez P., Chu M., Kolbe E., Brown C., Koonin S., 2003, Journal of Geophysical Research, pp 12–1 Robinson T. D., Ennico K., Meadows V. S., Sparks W., Bussey D. B. J., c(cid:13) 2015 RAS, MNRAS 000, 1–13 Inferring Planetary Obliquity 9 Schwieterman E. W., Breiner J., 2014, The Astrophysical Journal, 787, 171 Robinson T. D., Meadows V. S., Crisp D., 2010, The Astrophysical Journal Letters, 721, L67 Schlichting H. E., Sari R., 2007, The Astrophysical Journal, 658, 593 Seager S., Hui L., 2002, The Astrophysical Journal, 574, 1004 Spiegel D. S., Menou K., Scharf C. A., 2009, The Astrophysical Journal, 691, 596 Trauger J. T., Traub W. A., 2007, Nature, 446, 771 Turnbull M. C., Glassman T., Roberge A., Cash W., Noecker C., Lo A., Mason B., Oakley P., Bally J., 2012, Publications of the Astronomical Society of the Pacific, 124, 418 Unwin S. C., Shao M., Tanner A. M., Allen R. J., Beichman C. A., Boboltz D., Catanzarite J. H., Chaboyer B. C., Ciardi D. R., Edberg S. J., et al., 2008, Publications of the Astronomical Society of the Pacific, 120, 38 Williams D. M., Gaidos E., 2008, Icarus, 195, 927 Williams D. M., Kasting J. F., 1997, Icarus, 129, 254 Williams D. M., Pollard D., 2003, International Journal of Astrobiology, 2, 1 APPENDIX A: VIEWING GEOMETRY A1 General Observer The time-dependence of the kernel is contained in the sub-observer and sub-stellar angles: θo, φo θs, φs. Since they do not depend on planetary latitude or longitude, these four angles may be factored out of the kernel integrals. However, the light curves are still functions of time, so we derive the relevant dependencies here. In particular, we compute the sub-stellar and sub-observer locations for planets on circular orbits using seven parameters. Three are intrinsic to the system: rotational angular frequency, ωrot ∈ (−∞, ∞), orbital an- gular frequency, ωorb ∈ (0, ∞), and obliquity, Θ ∈ [0, π/2]. Rotational frequency is measured in an inertial frame, where positive values are pro- grade and negative denotes retrograde rotation (for comparison, the rota- tional frequency of Earth is ω⊕ rot ≈ 2π/23.93 h−1). Two more parameters are extrinsic and differ for each observer: orbital inclination, i ∈ [0, π/2] where i = 90◦ is edge-on, and solstice phase, ξs ∈ [0, 2π), which is the orbital angle between superior conjunction and the maximum Northern excursion of the sub-stellar point. The remaining parameters are extrinsic initial conditions: the starting orbital position, ξ0 ∈ [0, 2π), and the initial sub-observer longitude, φo(0) ∈ [0, 2π). These parameters are illustrated in Figure A1; other combinations are possible. We define the orbital phase of the planet as ξ(t) = ωorbt+ξ0. Without loss of generality we may set the first initial condition as ξ0 = 0, which puts the planet at superior conjunction when t = 0. With no precession the sub-observer colatitude is constant, (A1) This angle can be expressed in terms of the inclination, obliquity, and sol- stice phase using the spherical law of cosines (bottom of Figure A1): θo(t) = θo. cos θo = cos i cos Θ + sin i sin Θ cos ξs, (A2) 1 − cos2 θo. sin θo = (A3) The sub-observer longitude decreases linearly with time for prograde rota- tion, as we define longitude increasing to the East: φo(t) = −ωrott + φo(0). (A4) The prime meridian (φp ⇒ φ = 0) is a free parameter, which we define to run from the planet's North pole to the sub-observer point at t = 0. This sets the second initial condition, namely φo(0) = 0, and means (cid:112) cos φo = cos(−ωrott), 1 − cos2 φo. (cid:112) (A5) (A6) Hence, the time evolution of the sub-observer point is specified by its colat- itude and the rotational angular frequency. sin φo = (cid:104) plified by using Equation A2: xp(0) = 1 sin θo (cos ξs sin Θ cos θo − sin i)x + sin ξs sin Θ cos θo y + sin Θ(cos i sin Θ − sin i cos ξs cos Θ)z (cid:105) (A10) . We can now find the planetary axes, in terms of the inertial axes, at any time by rotating Equations A9 and A10 about the zp-axis: xp = cos(ωrott)xp(0) + sin(ωrott)yp(0), yp = − sin(ωrott)xp(0) + cos(ωrott)yp(0). (A12) The sub-stellar angles in the planetary coordinates may then be extracted from the relations (A11) sin θs cos φs = rps · xp, sin θs sin φs = rps · yp, cos θs = rps · zp, (A13) (A14) (A15) (A16) (A17) (A18) (A19) (A20) resulting in cos φs = cos θs = sin Θ cos [ξ − ξs] , (cid:113) sin θs = 1 − sin2 Θ cos2 [ξ − ξs], √ √ , cos(ωrott)a(t) + sin(ωrott)b(t) 1 − cos2 θo − sin(ωrott)a(t) + cos(ωrott)b(t) 1 − cos2 θo (cid:112)1 − sin2 Θ cos2 [ξ − ξs] (cid:112)1 − sin2 Θ cos2 [ξ − ξs] (cid:111) (cid:111) sin i cos ξ − cos θo sin Θ cos [ξ − ξs] , , sin φs = (cid:110) (cid:110) a(t) = where the factors a(t) and b(t) are given by b(t) = sin i sin ξ cos Θ − cos i sin Θ sin [ξ − ξs] (A21) Note that when θs = {0, π}, the sub-stellar longitude can be set arbitrarily to avoid dividing by zero in Equations A18 and A19. . A2 Polar Observer Equations A18 and A19 for the sub-stellar longitude apply to most ob- servers. However, the definition of yp(0) in Equation A8 fails when the sub-observer point coincides with one of the planet's poles. Two alternate definitions can be used in these situations. Case 1: If the sub-stellar point will not pass over the poles during orbit, we may define yp(0) = −y, so that xp(0) = yp(0) × zp = − cos Θx − cos ξs sin Θz. (A22) (A23) This results in cos φs = sin φs = cos(ωrott) cos ξ cos Θ + sin(ωrott) sin ξ , (A24) (cid:112)1 − sin2 Θ cos2 [ξ − ξs] (cid:112)1 − sin2 Θ cos2 [ξ − ξs] − sin(ωrott) cos ξ cos Θ + cos(ωrott) sin ξ . (A25) 10 J. C. Schwartz et al. shows a side view of Figure A1. The upper panel the gen- eral planetary system. The rotational and orbital angular frequencies {ωrot, ωorb}, inclination i, obliquity Θ, sub-observer colatitude θo, and observer viewing direction (cid:96) are indicated. The lower panel is an isomet- ric view, showing the solstice phase ξs. Superior conjunction occurs along the positive x-axis. Note the inertial coordinates, how they relate to the observer's viewpoint and planet's spin axis, and the angles between these vectors. The sub-stellar position is more complex for planets with non- zero obliquity. Consider an inertial Cartesian frame centered on the host star with fixed axes as follows: the z-axis is along the orbital angular frequency, z = ωorb, while the x-axis points towards superior conjunction. The y-axis is then orthogonal to this plane using y = z × x (bottom of Figure A1). In these inertial coordinates, the unit vector from the planet center towards the host star is rps = − cos ξ x− sin ξ y. The corresponding unit vector from the star towards the observer is (cid:96) = − sin ix + cos iz. Our approach is to express everything in the inertial coordinate system, then find the sub-stellar point with appropriate dot products. For the planetary surface, we use a second coordinate system fixed to the planet. We align the zp-axis with the rotational angular frequency, zp = ωrot, while the xp-axis is set by our choice for the prime meridian (and initial sub-observer longitude.) The final axis, yp, is again determined by taking yp = zp × xp. We proceed in two steps, first finding the plan- etary axes when t = 0, then using the planet's rotation to describe these axes at any time. Since we disregard precession, the planet's rotation axis is time- independent: zp = ωrot = − cos ξs sin Θx − sin ξs sin Θy + cos Θz. (A7) The sub-observer point is on the prime meridian when t = 0, so that Computing this we find yp(0) = 1 sin θo yp(0) = zp × (cid:96) . sin θo (cid:104) − cos i sin ξs sin Θx (cid:105) + (cos i cos ξs sin Θ − sin i cos Θ)y − sin i sin ξs sin Θz . The starting xp-axis is then found by taking yp(0) × zp. The result is sim- (A8) Case 2: However, if the sub-stellar point will pass over the poles dur- ing orbit, we define instead (A9) such that This produces xp(0) = z, yp(0) = zp × xp(0) = cos ξs y. cos φs = − sin(ωrott) sin ξ cos ξs (cid:112)1 − sin2 Θ cos2 [ξ − ξs] , (A26) (A27) (A28) c(cid:13) 2015 RAS, MNRAS 000, 1–13 Inferring Planetary Obliquity 11 Figure A2. Left: Longitudinal kernels of a planet with different obliquities at first quarter phase, ξ(t) = 90◦, where ξs = 225◦ and i = 60◦. The curves show 0◦–90◦ obliquity in 15◦ increments, where lighter shades of red indicate higher axial tilts. The kernel width of this planet changes as obliquity increases. Right: Analogous latitudinal kernels that indicate the dominant colatitude, shown as a circle, also changes as obliquity increases. − cos(ωrott) sin ξ cos ξs (cid:112)1 − sin2 Θ cos2 [ξ − ξs] . sin φs = (A29) These special cases only impact the sub-stellar longitude: expressions for the other angles are unchanged. As with a general observer, the Case 2 sub-stellar longitude may be set arbitrarily whenever θs = {0, π}. A3 Zero Obliquity For non-oblique planets, Θ = 0◦, the sub-observer colatitude satisfies cos θo = cos i, sin θo = sin i, while the sub-stellar angles become cos θs = 0, sin θs = 1, cos(ωrott)c(t) + sin(ωrott)d(t) − sin(ωrott)c(t) + cos(ωrott)d(t) cos φs = sin φs = √ √ 1 − cos2 i 1 − cos2 i (A30) (A31) (A32) (A33) (A34) , , (A35) where c(t) and d(t) are given by c(t) = sin i cos ξ, d(t) = sin i sin ξ. The sub-stellar longitude is therefore √ cos φs = cos(ωrott) sin i cos ξ + sin(ωrott) sin i sin ξ 1 − cos2 i sin φs = = cos(ωrott) cos ξ + sin(ωrott) sin ξ = cos(ξ − ωrott), − sin(ωrott) sin i cos ξ + cos(ωrott) sin i sin ξ √ 1 − cos2 i = − sin(ωrott) cos ξ + cos(ωrott) sin ξ = sin(ξ − ωrott). (A36) (A37) (A38) (A39) In other words, θo = i, φo = φo(0) − ωrott, θs = 0, and φs = ξ − ωrott. c(cid:13) 2015 RAS, MNRAS 000, 1–13 APPENDIX B: KERNEL DETAILS B1 Characteristics (cid:90) 2π An important measure of the longitudinal kernel is its width, σφ, as shown in the left panel of Figure A2. We treat this width mathematically as a stan- dard deviation. Since K(φ, G) is on a periodic domain, we minimize the variance for each geometry with respect to the grid location of the prime meridian, φp: (cid:20)(cid:90) 2π (cid:16) (cid:17)2 (cid:48) − ¯φ (cid:21) σ2 φ = min where K(φ) = K(φ)/(cid:82) K(φ)dφ is the spherically normalized longitudi- K(φ)dφ (B1) φp φ 0 , nal kernel, φ (cid:48) ≡ φ + φp, and ¯φ is the mean longitude: ¯φ = (cid:48) φ K(φ)dφ. (B2) (cid:73) 0 All longitude arguments and separations in Equations B1 and B2 wrap around the standard domain [0, 2π). Also note the unprimed arguments in- side the kernel: these make computing the variance simpler. The minimum variance determines the standard deviation of the kernel, and thus width, for a given geometry. The dominant colatitude is similarly important for the latitudinal ker- nel, as shown in the right panel of Figure A2. Cowan et al. (2012) defined the dominant colatitude, ¯θ, for a given geometry: ¯θ = θ K(θ, φ)dΩ, Equation B3 is equivalent to where K(θ, φ) = K(θ, φ)/(cid:72) K(θ, φ)dΩ is the normalized kernel. where K(θ) = K(θ)/(cid:82) K(θ) sin θdθ is the spherically normalized lati- θ K(θ) sin θdθ, (cid:90) π (B4) (B3) ¯θ = 0 tudinal kernel. The dominant colatitude is the North-South region that gets sampled most by the kernel (e.g. the circles in Figure A2.) B2 Albedo Variations Figure 1 demonstrates that obliquity can influence a planet's apparent albedo on both rotational and orbital timescales. Quantifying these rela- tions helps predict the obliquity constraints we may expect from real obser- vations. We use a Monte Carlo approach, simulating planets with different 0◦90◦180◦270◦360◦Longitude0.050.100.150.20Kernel0.30.60.9Kernel0◦45◦90◦135◦180◦Colatitude12 J. C. Schwartz et al. Figure B1. Kernel-albedo distributions for two sets of 10,000 planets generated via Monte Carlo, comparing rotational properties in red and orbital properties in blue. Changes in dominant colatitude are absolute values, and planets are binned in 0.1 × 2◦ regions in both panels. Each color scale is logarithmic: the darkest red and blue bins contain 230 and 279 planets, respectively. The scatter in either kernel characteristic decreases as the corresponding albedo variation rises. We can estimate the uncertainty on a kernel width or change in dominant colatitude by analyzing these distributions. maps and viewing geometries. We generate albedo maps from spherical har- (cid:96) (θ, φ), on the same 101× 201 grid in colatitude and longitude monics, Y m from Section 2.2: A(θ, φ) = Cm (cid:96) Y m (cid:96) (θ, φ), (B5) (cid:96)max(cid:88) (cid:96)(cid:88) (cid:96)=0 m=−(cid:96) where (cid:96)max is chosen to be 3, each coefficient Cm is randomly drawn (cid:96) from the standard normal distribution, and the composite map is scaled to the Earth-like range [0.1, 0.8]. Rotational and orbital changes in brightness are caused by East-West and North-South albedo markings, respectively, so (cid:96) (m (cid:54)= (cid:96)) = 0, we make three types of maps: East-West featured with Cm (cid:96) (m (cid:54)= 0) = 0, or no Cm North-South featured with Cm restrictions. For all maps with East-West features, we randomly offset the prime meridian. We generate 5,000 maps of each type. (cid:96) For each map we randomly select an obliquity, solstice phase, in- clination, and two orbital phases. Since inclination and orbital phase can be measured independent of photometry, we choose inclinations sim- ilar to planet Q, i ∈ [50◦,70◦], and orbital phases {ξ1, ξ2} with ∆ξ ∈ [110◦, 130◦]. Both phases are also at least 30◦ from superior and inferior conjunction, which conservatively mimics an inner working angle at the selected inclinations. We assume the planet's rotational and orbital frequencies are known (Pall´e et al. 2008; Oakley & Cash 2009), and use the Earth-like ratio ωrot/ωorb = 360. We divide roughly one planet rotation centered on each orbital phase into 51 time steps, then define the normal- ized amplitude of rotational and orbital albedo variations, Λrot and Λorb, as Ahigh ξ1 Λrot = , − Alow ξ1 ¯Aξ1 (cid:18) ¯Aξ1 + ¯Aξ2 2 (cid:19)−1 , (B6) (B7) Λorb = ¯Aξ1 − ¯Aξ2 ξ1 and Alow ξ1 where Ahigh are the extreme apparent albedos around the first phase, and ¯A is the mean apparent albedo of all time steps around a given phase. For each computed Λrot and Λorb, we calculate the correspond- ing kernel width and absolute value change in dominant colatitude, from Appendix B1. Figure B1 shows the resulting distributions, where rotational and orbital information is colored red and blue, respectively. We find similar results when relaxing constraints on the inclination and orbital phases. We can estimate uncertainties on values of σφ and ∆¯θ using these distributions. The mean rotational and orbital variations are ¯Λrot ≈ 0.54 and ¯Λorb ≈ 0.21; the average kernel width and change in dominant colat- itude are both roughly 38◦. The full distributions have standard deviations of about 17◦ in σφ and 24◦ in ∆¯θ, but roughly 5◦ and 7◦, respectively, when considering only large variations. To predict constraints on obliquity obtained from real data, we will assume there are single- and dual-epoch observations of planet Q that have our mean variations ¯Λrot and ¯Λorb. By considering samples only around these variations, we find about 10◦ and 20◦ standard deviations apiece in the kernel width and change in dominant colatitude. We use these standard deviations as uncertainties when creating the colored regions in Figure 6. B3 Peak Motion Equations C1 and C2 from Cowan et al. (2009) describe the motion of the kernel peak, where specular reflection occurs, for any planetary system. These equations can be written for edge-on, zero-obliquity planets using Section A3: cos θspec = (cid:112)2(1 + cos i) 1 + cos i = 1√ 2 , (B8) (B9) tan φspec = = sin(ξ − ωrott) + sin(φo(0) − ωrott) cos(ξ − ωrott) + cos(φo(0) − ωrott) sin(ωorbt − ωrott) + sin(φo(0) − ωrott) cos(ωorbt − ωrott) + cos(φo(0) − ωrott) . When finding φspec from Equation B9, the two-argument arctangent must be used to ensure φspec ∈ [−π, π). This also means it is difficult to simplify the equation with trigonometric identities. Instead, we can explicitly write the argument of Equation B9 in terms of the first meridian crossing, ξm, the earliest orbital phase after superior conjunction that the kernel peak recrosses the prime meridian: (cid:20) (cid:18) (cid:19)(cid:21)(cid:19) φspec(ξ; ξm) = ∓ π 2 4 ξ ξm + 1 − sgn cos ξ 2 , (B10) (cid:18) where the leading upper sign applies to prograde rotation and vice versa. The first meridian crossing is related to the planet's frequency (or period) ratio by (cid:12)(cid:12)(cid:12)(cid:12) ωrot ωorb (cid:12)(cid:12)(cid:12)(cid:12) = (cid:12)(cid:12)(cid:12)(cid:12) Porb Prot (cid:12)(cid:12)(cid:12)(cid:12) = (cid:18) 4π (cid:19) , ξm ± 1 1 2 c(cid:13) 2015 RAS, MNRAS 000, 1–13 (B11) 0.00.51.01.52.0Rotational Albedo Variation0◦25◦50◦75◦100◦Kernel Width0.00.20.40.60.81.01.21.4Orbital Albedo Variation0◦25◦50◦75◦100◦Change in Dominant Colatitudewhile the number of solar days per orbit is (cid:12)(cid:12)(cid:12)(cid:12) ωrot ωorb (cid:12)(cid:12)(cid:12)(cid:12) ∓ 1, Nsolar = Inferring Planetary Obliquity 13 (B12) following the same sign convention. Note that the frequency/period ratios and the number of solar days do not have to be integers. We reiterate that Equations B8–B12 apply to edge-on, zero-obliquity planets. Equation B11 gives two frequency ratios for each first meridian cross- ing, one prograde and another retrograde that is smaller in magnitude by unity. Equation B12 then states the corresponding difference in solar days is unity but reversed, making the longitudes of both kernel peaks in Equation B10 analogous at each orbital phase. These two versions of the planet have East-West mirrored albedo maps and identical light curves: they are formally degenerate. An inclined, oblique planet has pro/retrograde ver- sions that could be distinguished, as discussed in Section 4.5. This paper has been typeset from a TEX/ LATEX file prepared by the author. c(cid:13) 2015 RAS, MNRAS 000, 1–13
0902.3579	2	0902	2011-01-10T15:04:07	Probing the history of Solar System through the cratering records on Vesta and Ceres	[ "astro-ph.EP", "physics.space-ph" ]	Through its connection with HED meteorites, Vesta is known as one of the first bodies to have accreted and differentiated in the Solar Nebula, predating the formation of Jupiter and surviving the violent evolution of the early Solar System. The formation time of Ceres instead is unknown, but it should not postdate that of Jupiter by far. In this work we modelled the collisional histories of Vesta and Ceres at the time of the formation of Jupiter, assumed to be the first giant planet to form. In this first investigation of the evolution of the early Solar System, we did not include the presence of planetary embryos in the disk of planetesimals but we concentrated on the role of the forming Jupiter and the effects of its possible inward migration due to disk-planet interactions. Our results clearly indicate that the formation of the giant planet caused an intense early bombardment in the orbital region of the Main Asteroid Belt. According to our results, Vesta and Ceres would not have survived the Jovian early bombardment if the disk was populated mainly by large planetesimals like those predicted to form in turbulent circumstellar disks. Disks dominated by small bodies, like those predicted to form in quiescent circumstellar disks, or with a varying fraction of the mass in the form of larger (D \geq 100 km) planetesimals represent more favourable environments for the survival of the two asteroids. In those scenarios where they survive, both asteroids had their surfaces saturated by craters as big as 150 km and a few as big as 200 - 300 km. In the case of Vesta, the Jovian early bombardment would have significantly eroded (locally or globally) the crust and possibly caused effusive phenomena similar to the lunar maria, whose crystallisation time would then be directly linked to the time of the formation of Jupiter.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 34 (2010) Printed November 21, 2018 (MN LATEX style file v2.2) Probing the history of Solar System through the cratering records on Vesta and Ceres D. Turrini1⋆, G. Magni2, A. Coradini1 1Institute for Interplanetary Space Physics, INAF, Via Fosso del Cavaliere 100, 00133, Rome, Italy 2Institute for Space Astrophysics and Cosmic Physics, INAF, Via Fosso del Cavaliere 100, 00133, Rome, Italy Accepted XXX. Received XXX; in original form XXX ABSTRACT Dawn space mission will provide the first, detailed data of two of the major bodies in the Main Asteroid Belt, Vesta and Ceres. Through its connection with HED mete- orites, Vesta is known as one of the first bodies to have accreted and differentiated in the Solar Nebula, predating the formation of Jupiter and surviving the violent evolu- tion of the early Solar System. The formation time of Ceres instead is unknown, but it should not postdate that of Jupiter by far, since the perturbations of the giant planet stopped planetary accretion in the Main Asteroid Belt. In this work we modelled the collisional histories of Vesta and Ceres at the time of the formation of Jupiter, assumed to be the first giant planet to form. In this first investigation of the evolution of the early Solar System, we did not include the presence of planetary embryos in the disk of planetesimals but we concentrated on the role of the forming Jupiter and the effects of its possible inward migration due to disk-planet interactions. Our results clearly indicate that the formation of the giant planet caused an intense early bombardment in the orbital region of the Main Asteroid Belt. We explored the effects of such bom- bardment on Vesta and Ceres assuming different size distributions of the primordial planetesimals. According to our results, Vesta and Ceres would not have survived the Jovian early bombardment if the disk was populated mainly by large planetesimals like those predicted to form in turbulent circumstellar disks. Disks dominated by small bodies, like those predicted to form in quiescent circumstellar disks, or with a vary- ing fraction of the mass in the form of larger (D > 100 km) planetesimals represent more favourable environments for the survival of the two asteroids. The abundance of planetesimals, especially the larger ones, proved a critical factor to this regards. The extent of Jupiter's radial migration due to disk-planet interactions proved itself another critical factor. In those scenarios where they survive, both asteroids had their surfaces saturated by craters as big as 150 km and a few as big as 200 − 300 km. In the case of Vesta, the Jovian early bombardment would have significantly eroded (locally or globally) the crust and possibly caused effusive phenomena similar to the lunar maria, whose crystallisation time would then be directly linked to the time of the formation of Jupiter. Key words: minor planets, asteroids - Solar System: formation - Planets and satel- lites: individual: Jupiter - methods: N -- Body simulations - methods: numerical. 1 1 0 2 n a J 0 1 . ] P E h p - o r t s a [ 2 v 9 7 5 3 . 2 0 9 0 : v i X r a 1 INTRODUCTION Our knowledge of the chronology and the evolution of the early Solar System is limited, particularly for what it con- cerns the first 10 Ma. This is the timespan generally as- sumed as the upper bound to the lifetime of the gaseous component of the Solar Nebula and therefore to the forma- tion of the giant planets. From the observations of circum- ⋆ E-mail: [email protected] stellar disks we know that their median lifetime is about 3 Ma, with the range of observed values spanning between 1− 10 Ma (Haisch, Lada & Lada 2001; Meyer 2008). During this timespan, solid material should accrete to form the first planetesimals and then the planetary embryos from which the cores of the giant planets would originate. Such plane- tary cores should in fact appear in the forming Solar Sys- tem early enough to allow for the accretion of the gaseous envelopes of the giant planets from the Solar Nebula. The chronology of the early Solar System obtained through 2 Turrini et al. radiometric ages of chondrites, achondrites and differenti- ated meteorites indicates that the first solids to form, about 4567.2± 0.6 Ma ago, were the Ca-Al-rich inclusions (Amelin et al. 2002), CAIs in the following. Chondrules, once thought to represent the oldest material that solidified in the Solar Nebula, formed about 1 − 3 Ma later than CAIs (Amelin et al. (2002), Connelly et al. (2008)) while differentiated bod- ies generally appeared in the next few million years after the formation of chondrules (see Scott (2007) and references therein). However, meteoritic evidences suggest that in some cases differentiation of planetesimals took place extremely early in the history of the Solar System, i.e. about 1 Ma af- ter the formation of CAIs (Baker et al. 2005; Bizzarro et al. 2005). Such primordial differentiation was due to the pres- ence of short-lived radionuclides, mainly 26Al and 60Fe (see e.g. Bizzarro et al. (2005)) in bodies larger than 20−30 km in radius (Scott 2007). In particular, the results by Yang, Gold- stein & Scott (2007) obtained in studying the iron meteorites from the IV A group suggest that these meteorites formed in a parent body that was about 300 km wide and lacked an insulating mantle. The authors explained such anoma- lous composition of the parent body through the removal of the silicate-rich mantle from a differentiated protoplanet whose original size was about 103 km (ibid). All these re- sults collectively imply that planetary accretion started at the very beginning of the history of the Solar System and that a first generation of hundreds-of-km-wide bodies formed and differentiated in about 1− 1.5 Ma. Moreover, the study of differentiated meteorites (eucrites, ureilites and angrites, see Scott (2007) and references therein) indicates that the differentiation of primordial planetesimals driven by short- lived radionuclides took place during a timespan covering the first 10 Ma after the formation of CAIs. The asteroid Vesta is of particular interest to this regard. Vesta has been identified as the possible parent body of the HED meteorites, a family of basaltic achondrites composed by howardite, eucrite and diogenite meteorites, and such connection would imply that the asteroid is differentiated (see e.g. Drake (2001), Keil (2002) and references within). Moreover, the 40Ar-39Ar ages of the oldest HED meteorites (see Keil (2002), Scott (2007) and references within) suggest that this asteroid is primordial, i.e. it formed and differenti- ated in less than 4 Ma since the formation of CAIs. In such scenario, Vesta would be the only known surviving primor- dially differentiated planetesimal and its formation would date back prior to or contemporary to the formation of the giant planets. As we previously mentioned, in fact, the gi- ant planets should appear in the Solar System somewhere during the first 10 Ma. The time needed for their formation is, to most practical purposes, the same over which they ac- crete their solid cores, estimated being of about a few Ma. The results of hydrodynamical studies in fact indicate that the phase of gas accretion took place at an extremely rapid pace. According to the simulations performed by Lissauer et al. (2009), the total time it takes for Jupiter to accrete its gaseous envelope varies between several 104 to a few 105 years. In their simulations, Coradini, Magni & Turrini (2010) instead measured the time-scales of gas accretion for Jupiter and Saturn: their results vary between a few 103 to about 105 years depending on the physical parameters of the Solar Nebula. As a consequence of this match in the timing of the early differentiation of planetesimals and the formation of the giant planets, Vesta could bear the marks of Jupiter's birth. The dwarf planet Ceres is of particular interest for a dif- ferent reason. It has been suggested that, after the forma- tion of the giant planets, the surviving planetary embryos influenced the evolution of the planetesimals in the Solar System and caused a depletion in mass (Wetherill 1992) in the orbital region of the Main Asteroid Belt respect to the mass hypothesised to originally reside in that region (see e.g. Weidenschilling (1977)). The results by Wetherill (1992), Chambers & Wetherill (2001), Petit, Morbidelli & Cham- bers (2001) and O'Brien, Morbidelli & Bottke (2007) indi- cate that the combined gravitational perturbations of the giant planets and the planetary embryos reduce the popula- tion of planetesimals in which they are embedded by about a factor 100 in about 108 years. The planetary embryos them- selves are removed on a ∼ 107 years-long timescale by being ejected from the Solar System or being accreted by plane- tary bodies (Petit, Morbidelli & Chambers 2001; O'Brien, Morbidelli & Bottke 2007). Being the most massive object which survived to present time in the Main Asteroid Belt, Ceres represent an important probe of the efficiencies and timescales of planetary accretion and removal in that orbital region. In this first work we address the topic of the collisional evo- lution of primordial planetesimals during the formation of Jupiter, assumed to be the first giant planet to form in the Solar System. As we will describe, the formation of Jupiter causes a brief yet intense primordial bombardment in the Main Asteroid Belt. The bulk of this Jovian early bombard- ment lasts a few 105 years, so our results describe a scenario where the formation of Saturn is delayed respect to that of Jupiter by about the same amount of time. In this first in- vestigation of the evolution of the primordial Solar System, we did not take in account the possible effects of planetary embryos in the disk of planetesimals and, in particular, in the Main Asteroid Belt due to computational constrains. For the same reason, we did not consider the effects of the presence of the nebular gas on the orbital motion of the plan- etesimals. We will address the issues of the contributions of Saturn, the nebular gas and the planetary embryos in future papers. Previous studies addressing the early evolution of the Solar System differ from our approach in several ways. The studies dealing with the collisional history of the early inner Solar System adopted statistical approaches designed to evaluate the evolution of the size-frequency distribution (Bottke et al. 2005a,b; Morbidelli et al. 2009) and the disruption law (Bot- tke et al. 2005a) of the planetesimals. We concentrated our attention on selected bodies, for which we reproduced pos- sible early collisional histories. We chose Vesta and Ceres as our case studies since they will be visited in the next years by the Dawn space mission. The studies investigating instead the dynamical evolution of early inner Solar System due to Jupiter's influence either used a sharp, step-like transition to describe Jupiter's formation (Petit, Morbidelli & Chambers 2001) or directly assumed a fully formed Jupiter (Chambers & Wetherill 2001; O'Brien, Morbidelli & Bottke 2007). In our study we concentrated on the temporal interval cover- ing the accretion of the core and the gaseous envelope of the giant planet, trying to simulate this phase in a more re- alistic way, and took into account also the contribution of planetesimals from the outer Solar System. The presentation of this work is organised in the following way. We describe in detail our physical and dynamical model in Section 2, while in Section 3 we describe the output of our simulations. Finally, in Section 4 we discuss our results and their implications for the comprehension of the early So- lar System and in Section 5 we draw the conclusions about the interpretation of the collisional features that could be revealed by Dawn mission on Vesta and Ceres. 2 DYNAMICAL AND PHYSICAL MODEL To explore the early collisional history of Vesta and Ceres we simulated the dynamical evolution of a section of the young Solar System at the time of the formation of Jupiter's core and the subsequent accretion of its gaseous envelope. Our template of the forming Solar System was composed of the Sun, the accreting Jupiter, Vesta, Ceres and a swarm of massless particles representing the disk of planetesimals. We followed the evolution of our template of the Solar Sys- tem for a temporal interval equal to 2 × 106 years. Vesta and Ceres were assumed already formed at the beginning of our simulations. The massless particles were initially dis- tributed into a spatial region spanning between 2 − 10 AU from the Sun. Such radial interval has been chosen after a set of numerical experiments to determine the region of the Solar System influenced by the forming Jupiter on the considered timespan, to optimise the usage of the compu- tational resources. Being primarily interested to the effects of Jupiter's mass increase on the dynamical stability and the collisional evolution of the inner Solar System, we mod- eled the formation process of the giant planet through a semi-empirical approach. The timescales and the other pa- rameters, on which our modeling was based, were derived from the results of hydrodynamical simulations described in Magni & Coradini (2004) and Coradini, Magni & Turrini (2010) and consistently with the findings of Lissauer et al. (2009). In our semi-empirical model we considered also the effects of planetary migration due to the disk-planet inter- actions during the formation of Jupiter (see e.g. Papaloizou et al. (2007) and references therein). During the dynamical evolution of our template of the Solar System we evaluated the probabilities of planetesimals impacting Vesta and Ceres through a statistical approach. In the following subsections we will describe in detail both the initial conditions and the physical parameters and constrains of our model. 2.1 Vesta and Ceres Vesta and Ceres were assumed to be on circular orbits that lie on the same plane as that of Jupiter and do not change during the simulations. This approximation obviously ne- glects the orbital inclinations of the two asteroids, yet we adopted it since it significantly simplifies the treatment of their collisional evolution during the simulations, as de- scribed in Sect. 2.4. Moreover, we do not know if the present orbits of Vesta and Ceres are representative of their primor- dial ones. The semimajor axes of Vesta and Ceres were obtained by Collisional histories of early Vesta and Ceres 3 the JPL Small-Body Database Browser1: the values adopted were respectively av = 2.362 AU and ac = 2.765 AU. Their mean radii were assumed rv = 258 km (Thomas et al. 1997) and rc = 476 km (Thomas et al. 2005). Finally, their mass values were derived from Michalak (2000) and were mv = 2.70 × 1023 g and mc = 9.25 × 1023 g. These values implies mean density values of ρv = 3.7 g cm−3 (Michalak 2000) and ρc = 2.0 g cm−3 (Thomas et al. 2005). 2.2 Jupiter's formation and migration In our semi-empirical approach, we considered Jupiter's for- mation as composed by 3 different stages: • a core accretion phase; • a fast, exponential gas accretion phase; • a slow asymptotic gas accretion phase. The second and third stages of the formation process are ruled by the same physical process, i.e. Jupiter accreting gas from the Solar Nebula, and are analytically treated as a single stage in the model, yet they are clearly distinguishable from the point of view of the evolution of the early Solar System, as will be shown in Section 3. At the beginning of the simulations, Jupiter is an embryo with mass M0 = 0.1 M⊕ and it grows to the critical mass Mc = 15 M⊕ in τc = 106 years. Since the total accretion time of Jupiter's core is the sum to our τc to the time needed to form the initial Mars-sized core, our choice of τc is consistent with the lower limits indicated by theoretical works for the formation of Jupiter's core (a few Ma, see e.g. Nagasawa et al. (2007) and references therein). The mass growth is governed by the equation Mp = M0 +(cid:18) e e − 1(cid:19) (Mc − M0) ×(cid:16)1 − e−t/τc(cid:17) (1) During most of the first stage, Jupiter's mass is negligible in terms of its perturbing effects on the planetesimals in the disk and the only planetesimals affected by Jupiter are those which undergo a close encounter with the forming planet. As a consequence, in this phase we expect that the number of impacts against the considered asteroids is governed by stochastic collisions of near-by objects. As soon as Jupiter's core neared the critical mass value of 15 M⊕, the nebular gas surrounding it became gravi- tationally unstable and started to be rapidly accreted by the planet to form its massive envelope. During this phase, Jupiter directly perturbed near-by planetesimals, clearing a gap of increasing width in the disk, and more distant planetesimals through orbital resonances. During this phase, Jupiter's mass growth is governed by the equation Mp = Mc + (MJ − Mc) ×(cid:16)1 − e−(t−τc)/τg(cid:17) (2) where MJ = 1.8986 × 1030 g = 317.83 M⊕ is Jupiter's fi- nal mass. As we previously anticipated, the timescale τg was derived from the hydrodynamical simulations described in Magni & Coradini (2004) and Coradini, Magni & Tur- rini (2010). The value we employed in the simulation is 1 http://ssd.jpl.nasa.gov/sbdb.cgi#top 4 Turrini et al. τg = 5 × 103 years, consistently with the findings of Lis- sauer et al. (2009), and is linked to the turbulence parame- ter α = 0.01 used in modelling the disk. In the third and final stage, Jupiter's accretion slows down while the giant planet reaches its final mass value and its gravitational perturbations secularly affect more and more distant planetesimals. As we discussed at the beginning of this subsection, during this phase Jupiter's mass growth is still governed by Eq. 2 and there is no discontinuity, from a numerical point of view, with the previous phase. We followed the evolution of the template of the Solar Sys- tem across the second and third stages for τa = 106 years, i.e. for 200 × τg. We wish to emphasise again that in our modelling Jupiter begins as a Mars-sized embryo. As a con- sequence, our starting time t0 differs from the t0 of the Solar System, i.e. the condensation of the first solids 4567.2 ± 0.6 Ma ago (Amelin et al. 2002), and is located later, possibly by a few Ma. From a physical point of view, therefore, in our simulations we are looking at 2 Ma-wide temporal win- dows located somewhere in the first 10 Ma in the lifetime of the Solar System and centred on the time Jupiter started to efficiently capture the nebular gas. In all our simulations, Jupiter starts on a circular orbit. How- ever, theoretical models indicate that forming giant planets should undergo Type I and II migrations and drift inward due to their interactions with the protoplanetary disk (see e.g. Papaloizou et al. (2007) and references therein). Type I migration is dominant while the mass of the forming planet is lower than ≈ 100 M⊕ and the effects of the planet on the disk can be treated as linear perturbations (ibid). The timescale τM of Type I migration is a nonlinear function of the mass and the heliocentric distance of the forming planet: for a planet at 5.2 AU, τM varies between ∼ 105 − 107 years (D'Angelo, Kley & Henning 2003). Once the forming planet is massive enough to open a gap in the circumstellar disk, Type II migration takes over: the migration timescale be- comes less sensitive to the planetary mass and for a planet at 5.2 AU τM is of the order of ∼ 105 years (ibid). To eval- uate the effects of the radial displacement of Jupiter on the early collisional histories of Vesta and Ceres, we forced an inward migration in Jupiter's motion. As a first approxima- tion, we ignored the distinction between Type I and II mi- grations and started Jupiter's migration as soon as the mass of the forming planet reaches 15 M⊕, which is equivalent to say that the value of τM becomes of the order of ∼ 105 years (ibid). As a consequence, Jupiter moves on a circular orbit while its planetary core is growing to the critical mass and starts to spiral inward once the phase of gas accretion be- gins. The semi-empirical treatment of Jupiter's migration is analogous to Eq. 2 we used to describe the growth of the gaseous envelope: rp = r0 + (rJ − r0) ×(cid:16)1 − e−(t−τc)/τr(cid:17) (3) where r0 is Jupiter's position at the beginning of the simula- tion, rJ is the final position and τr = τg = 5× 103 years. We assumed Jupiter's final semimajor axis equal to the present one, an assumption consistent with both the standard model of planetary formation and the scenario described by the so called Nice Model (Tsiganis et al. 2005). As a consequence, Jupiter's initial semimajor axis depends on the desired ex- tent of radial displacement. In our simulations we considered four different migration scenarios: 0 AU (no displacement), 0.25 AU, 0.5 AU and 1 AU. The timescale τr we used in Eq. 3 is an e-folding time while the timescale τM previously men- tioned is defined as τM = a/ a, so it represents the timescale for the planet to migrate from its initial position to the in- ner edge of the Solar Nebula. For an exponential decay law, about 95% of the displacement is achieved in 3 e-folding times, which in the case of Eq. 3 is equal to 1.5 × 104 years. Therefore, displacements of 0.25 AU, 0.5 AU and 1 AU with the assumed value of τr are equivalent to assuming values of τM at 5.2 AU respectively of ≈ 3.2 × 105 years, ≈ 1.6 × 105 years and ≈ 8 × 104 years, consistently with the results of theoretical studies (Papaloizou et al. 2007). Before proceeding with the description of the model, we would like to emphasise that the displacement we discuss here is not the one invoked by the Nice Model, which is temporally located several 108 years later. 2.3 Dynamical and physical characterisation of the planetesimals As anticipated, we modelled the dynamical evolution of the disk of planetesimals using a swarm of 8× 104 massless par- ticles. However, we associated to each particle a mass value and other physical features, which we used to model in a re- alistic way the effects of the impacts on Vesta and Ceres. The assumptions under which we derived the physical character- istics of the planetesimals will be detailed in the following. The dynamical characteristics of the planetesimals in the disk at the beginning of our simulations are defined as fol- lows: 2 AU 6 ai 6 10 AU 0 6 ei 6 3 × 10−2 0 rad 6 ii 6 3 × 10−2 rad (4) The values of eccentricity and inclination associated to each massless particle were chosen randomly as ei = e0X, ii = i0(1 − X) (5) where e0 = 3 × 10−2, i0 = 3 × 10−2 rad and X is a number extracted from a uniform distribution in the range [0 − 1]. The planetesimals were first assumed to form by gravita- tional instability of the dust in the midplane of a nontur- bulent protoplanetary nebula (Safronov 1969; Goldreich & Ward 1973) having a mass of Mneb = 0.02 M⊙ distributed 1 AU(cid:1)−1.5 between 1 − 40 AU with a density profile σ = σ0(cid:0) r (σ0 = 2700 g cm−2 being the surface density at 1 AU) and a dust to gas ratio ξ = 0.01. Following Coradini, Federico & Magni (1981), the mass spectrum of the planetesimals formed by the gravitational instability mechanism in this protoplanetary nebula spans the range 2×1017−1020 g. Dif- ferent planetesimal formation mechanisms, however, would produce different size distributions and, as a consequence, different planetesimal abundances than the one we consid- ered. As we will describe in Sect. 3.5, therefore, we investi- gated the implications of planetesimal formation in turbu- lent disks by taking advantage of the results of Morbidelli et al. (2009) and Chambers (2010). While the results of Mor- bidelli et al. (2009) directly supply the size-frequency distri- bution (SFD in the following) of planetesimals that formed in the region of the Main Asteroid Belt, we derived the size distribution and the abundance of the planetesimals in the formation scenario described by Chambers (2010) using the same analytical treatment we will now detail for the case of a quiescent disk. From the results of Coradini, Federico & Magni (1981) we can derived the following semi-empirical relationship: mp = m0(cid:16) r 1 AU(cid:17)β (6) where mp and m0 are expressed in g, r is expressed in AU and β = 1.68. The value m0 is the average mass of a plan- etesimal at 1 AU , i.e. 2 × 1017 g. The number surface density in the disk of planetesimals can then be expressed as a function of mass and radial distance as n(m, r) = Q(r)m2e−(m/mp(r))2 (7) where Q(r) represents the radial dependence of the number surface density n(m, r) and we superimposed a Maxwell- Boltzmann distribution to the semi-empirical relationship of Eq. 6. The functional form of Q(r) can be obtained by coupling Eq. 7 with the relationship governing the surface density profile in the Solar Nebula σp(r) = ξσ(r) = ξσ0(cid:16) r 1 AU(cid:17)−ns (8) where σp is the surface mass density of the planetesimals, σ is the surface mass density of the gas in the protoplanetary nebula, σ0 = 2700 g cm−2 is the surface mass density of the gas at 1 AU and ξ is the dust to gas ratio. As anticipated, we used the standard assumptions for the Solar Nebula, setting ns = 1.5 and ξ = 10−2. In the following, when the distance r is implicitly normalised to 1 AU it will be indicated with the capital letter R, while the symbol 1 AU will indicate the value of the astronomical unit expressed in cm, i.e. 1 AU = 1.49597870691 × 1013 cm. By integrating Eq. 7 over m we obtain n∗(r) = Z ∞ n(m, r)dm = √π 4 Q(r)mp(r) dm = 0 Q(r)m2e−(m/mp(r))2 = Z ∞ 0 (9) while for the surface mass density of planetesimals we have σp(r) = Z ∞ 0 n(m, r)mdm = = Z ∞ 0 Q(r)m3e−(m/mp(r))2 dm = 1 2 Q(r)m4 p(r) (10) By equating Eq. 10 with Eq. 8 we get Q(r) = 2 ξσ0 m4 p R−ns (11) which, substituting Eq. 11 into Eq. 9 and applying Eq. 6, gives n∗(r) = √π 2 ξσ0 m0 R−(ns+β) (12) Collisional histories of early Vesta and Ceres 5 Now we can obtain the cumulative distribution of n∗(r) by integrating 2πrn∗(r)dr = N (x) = Z x (1 AU )2(cid:18) rmin = π3/2 ξσ0 m0 (cid:18)(cid:16) x 1 AU(cid:17)2−ns−β 1 2 − ns − β(cid:19) × 1 AU(cid:17)2−ns−β(cid:19) −(cid:16) rmin (13) where rmin = 2 AU is the inner radius of the disk of plan- etesimals. We can obtain the total number of planetesimals in the disk by integrating Eq. 13 between rmin = 2 AU and rmax = 10 AU: 2πrn∗(r)dr = Ntot = Z rmax (1 AU )2(cid:18) rmin = π3/2 ξσ0 m0 (cid:18)(cid:16) rmax 1 AU(cid:17)2−ns−β 1 2 − ns − β(cid:19) × 1 AU(cid:17)2−ns−β(cid:19) −(cid:16) rmin (14) The function X = N(x) uniformly varies in the range 0 6 Ntot X 6 1. Once we invert the previous equation and express the initial position of the planetesimals in the disk as x = f (X) where µ = 2 − ns − β and x = f (X) = hX (cid:16)(cid:16) rmax 1 AU(cid:17)µ −(cid:16) rmin 1 AU(cid:17)µ(cid:17) +(cid:16) rmax 1 AU(cid:17)µi−µ (15) we can use X as the random variable in a Montecarlo ex- traction to spatially populate the disk. Through Eq. 6 we can link the average mass mp(r) of the planetesimals to their initial position r = x (which we im- plicitly assume coinciding with the formation region). To obtain the mass value of each planetesimal, we apply again a Montecarlo method through a uniform random variable Y . To obtain Y we need to compute and normalise the cu- mulative distribution of Eq. 7 considering m as the variable of interest and r as a constant. The cumulative distribution is Z m∗ 0 n(m, r)dm = 2ξσ0 mp(r) R−ns Z y∗ 0 y2e−y2 dy (16) where y = m/mp(r) and y∗ = m∗/mp(r). The integral on the right side of Eq. 16 cannot be solved analytically: substituting z = y2 with dz = 2ydy we obtain Z y∗ 0 y2e−y2 dy = √z∗ 1 2 Z 0 z1/2e−zdz (17) The integral on the right side of previous equation is a lower incomplete Gamma function with real parameter a = 3/2 or γ(cid:0)3/2, √z∗(cid:1). Normalising Eq. 16 over Eq. 12 we get the uniform random variable Y varying in the range [0, 1] Y = 2γ (3/2, y∗) √π = P (3/2, y∗) (18) where P (3/2, y∗) is the lower incomplete Gamma ratio. The inverse of the lower incomplete Gamma ratio can be computed numerically and, by substituting y∗ back with m∗/mp(r) we obtain m(r) = mpinv (P (3/2, Y ))) (19) 6 Turrini et al. We can therefore assign to each planetesimal its own mass value through a second Montecarlo extraction. The use of massless particles assures the linearity of the pro- cesses investigated over the number of considered bodies: we can therefore extrapolate the number of impacts expected in a disk of planetesimals described by the density profiles previously presented by multiplying the number of impacts recorded in our simulations by a factor γ where γ = Ntot/nmp (20) where Ntot is computed through Eq. 14 and nmp = 8 × 104. Note that the γ factor depends only on the radial extension of the considered region (i.e. rmin and rmax): we took advan- tage of this fact in parallelising the algorithm as described in Sect. 2.5. The dynamical evolution of Jupiter and the swarm of mass- less particles is computed through a fourth order Runge- Kutta integrator with a self adjusting time-step. To time- step is chosen by evaluating at each given time the smallest timescale τmin between: • the orbital periods of the massless particles • the orbital periods of Jupiter and the two asteroids • the free-fall time of Jupiter-particle pairs considered as isolated systems The time-step is then computed as tts = τmin/fts (21) where fts = 100 in our simulations. Finally, to estimate the amount of volatiles delivered to the two asteroids Vesta and Ceres by the planetesimals, we as- sumed that planetesimals formed at their initial positions and considered two compositional classes: following) were considered rocky bodies; • planetesimals formed inside the Snow Line (ISL in the • planetesimals formed beyond the Snow Line (BSL in the following) were considered volatile-rich bodies. ISL and BSL planetesimals were characterised by mean den- sity values respectively of ρISL = 3.0 g/cm3 and ρBSL = 1.0 g/cm3. The location of the Snow Line in our simulations was placed at rSL = 4.0 AU (see Encrenaz (2008) and refer- ences therein). 2.4 Collisional history To reproduce the collisional histories of the two asteroids we opted for a statistical approach based on the probabil- ity density distributions of the asteroids along their orbits. Our method is similar to the analytical method developed by Opik (1976), yet the latter (as well as its variants) com- pute an average impact probability by assuming that the longitude of node, the anomaly and the argument of pericen- ter vary uniformly over all possible orientations. Moreover, Opik's method may fail evaluating the impact probability for near-tangent orbits and for very eccentric orbits (Green- berg, Carusi & Valsecchi 1988). Given the wide range of pos- sible impact geometries due to the extension (2 − 10 AU) of the disk and that Jupiter's perturbations may significantly change the orbits of the planetesimals on timescales analo- gous to their precession timescales, we preferred the use of a numerical algorithm. This way we were able to characterise in a semi-deterministic way the impact probabilities of the real orbital configurations. Each asteroid was spread on a torus representing its spatial probability density and characterised by a mean radius RT and a section σT defined as and RT = aA σT = π 4 × (DAfG)2 (22) (23) where aA and DA were respectively the semimajor axis and the physical diameter of the considered asteroid while fG is the gravitational focusing factor fG = 1 +(cid:18) vesc venc(cid:19)2 (24) with vesc being the escape velocity from the asteroid and venc the relative velocity between the asteroid and the plan- etesimal. The gravitational focussing factor fG was intro- duced to account for the perturbations of the asteroids on the orbits of approaching planetesimals, which were not in- cluded in the explicit dynamical model. When a planetesimal crosses one of the two tori, the impact probability is the probability that both the planetesimal and the asteroid will occupy the same spatial region at the same time. This probability can be evaluated as the ratio between the effective collisional time and the orbital period of the as- teroid. The effective collisional time is the amount of time available for collisions and is evaluated as the minimum be- tween the time spent by the asteroid and the planetesimal into the crossed region of the torus. This is equivalent to writing Pcoll = min(τP , τA) TA (25) where TA is the orbital period of the asteroid, τA and τP are respectively the time spent by the asteroid and the planetes- imal into the crossed region of the torus while min(τP , τA) is the effective collisional time. To estimate τP and τA we need to identify the intersections between the orbit of the planetesimal and the torus of the asteroid. Assuming that the orbital path of the planetesimal during the timestep when it approaches the torus is linear, we just need to solve a ray -- torus intersection problem. A torus centred on the origin of the axes and lying on the xy plane is described by the equation f (x, y, z) = (cid:0)x2 + y2 + z2 − (R2 + s2)(cid:1) + 4R2(cid:0)z2 − s2(cid:1)(26) where R = RT and s = 0.5 (DAfG). The surface of such torus is described by the homogeneous equation f (x, y, z) = 0. By substituting the variables (x, y, z) with the vectorial equation of the ray associated to the linearised orbit ~r = ~mt + ~q (27) where m and q are respectively the angular coefficient and the origin of the ray, we obtain the fourth order equation identifying the intersections between the ray and the torus. By writing a = ~m · ~m b = ~m · ~q (28) (29) c = ~q · ~q we can express the fourth order equation as d4t4 + d3t3 + d2t2 + d1t + d0 = 0 where d4 = a2 d3 = 4ab d2 = 4b2 − 2as2 − 2aR2 + 2ac + 4R2m2 d1 = 4bc − 4bs2 − 4bR2 + 8mzqzR2 d0 = c2 + R4 + s4 − 2R2s2 − 2cs2 − 2cR2 + 4R2q2 z . z (30) (31) (32) This fourth order equation can be solved analytically through Ferrari's method or numerically. The quartic equa- tion will have up to four real solutions: by selecting the ap- propriate pair of real solutions, if existing, we can derive the coordinates of the intersection points ~i1 and ~i2. From the intersection points we can derive the length of the planetesimal's path dP through the torus and the crossing time τP which, in the linear approximation, is τP = dP ~vP (33) where ~vP is the modulus of the velocity of the planetesimal since the velocity and the path are parallel. From the intersection points we can also derive the angular width of the crossed section of the torus ∆θA and the time τA the asteroid spends into the crossed region: Collisional histories of early Vesta and Ceres 7 can apply a data parallel approach and split each simulation into a set of sub-problems which can be treated in parallel. In our implementation, we divided the disk of planetesimals into a number of concentric rings containing a fixed amount of test particles. Each ring has been evolved independently under the influence of the forming Jupiter and the collisions with the two asteroids have been recorded. At the end of the set of simulations, we merged and temporally reordered their outputs to obtain a representation of the evolution of the system as a whole. Obviously, we had to use a different γ factor for each ring of massless particles to normalise it to the real population expected in the orbital region con- sidered. However, as anticipated in Sect. 2.3 we can do this in a straightforward way since γ depends only on the radial extension of the ring. Through this approach, we were able to run the equiva- lent of a 6-month long simulation with 8 × 104 massless particles for each migration scenario by running a set of 8 sub-simulations with 104 massless particles each, every sub- simulation taking about 21 days to conclude. This duration of the sub-simulations is the reason why we concentrated on the perturbing role of Jupiter in this first investigation of the evolution of the early Solar System. The inclusion of Saturn as a second massive perturber, in fact, would have implied the doubling of the time required to complete each simulation and the same hold true for each planetary embryo considered. τA = ∆θA TA 2π = ∆θA ωA (34) 3 RESULTS where ωA = nA = 2π is the angular velocity (coinciding TA with the orbital mean motion for circular orbits) of the as- teroid. To evaluate the effects of the impacts on the two asteroids, we computed the average crater diameter produced by each collision using the empirical scaling law (see p. 165 of De Pater & Lissauer (2001)) D = 1.8ρ0.11 i ρ−1/3 a (2Ri)0.13(cid:18) Ek ga (cid:19)0.22 (sin θ)1/3 (35) where the indexes a and i indicate the asteroid and the im- pactor respectively, ρ is the density, R the physical radius, g the gravitational acceleration, Ek is the impact kinetic en- ergy, θ is the impact angle respect to the local horizontal and all quantities are evaluated in mks units. In our first estimation of the surface cratering of the two asteroids we used a fixed value θ = 45◦ as the average impact angle (see Pierazzo & Melosh (2000) and references therein). 2.5 Parallelising the model Before concluding the section devoted to the dynamical and physical model, we would like to briefly discuss one advan- tage of our modeling, i.e. the possibility to parallelise the treatment of the evolution of this template of the Solar Sys- tem in a straightforward way. Since the disk of planetesimals is reproduced by massless particles, which by definition do not interact between them- selves, and since Jupiter's evolution is governed by the mi- gration and accretion rates described by Eqs. 1, 2 and 3, we In this section we will describe the results of our simulations, highlighting the differences we found between the four dy- namical scenarios for the migration of Jupiter and the influ- ence of the primordial size distribution of the planetesimals. We will first (Sect. 3.1) concentrate on detailing the dynami- cal classes of impactors that dominate the different phases of the evolution of our template of the early Solar System, then (Sect. 3.2) we will characterise the impact fluxes recorded for Vesta and Ceres in the scenario of planetesimal forma- tion in a quiescent disk and discuss their effects on the two asteroids (3.3). Finally, we will discuss the influences of the migration timescale of Jupiter (Sect. 3.4) and of the scenario of planetesimal formation in a turbulent disk (Sect. 3.5) on our results. 3.1 Dynamical features of the impactors In our simulations we identified 2 distinct dynamical classes (i.e. resonant and non-resonant impactores) contributing to the cratering of Vesta and Ceres for each of the populations of planetesimals (ISL and BSL) we considered (see Figs. 1 and 2). The non-resonant class of impactors for the ISL population of planetesimals is composed by bodies spatially near each of the two asteroids: during the first 106 years of our simula- tions, these impactors are characterised by the same dynam- ical features they possessed at the beginning of the simula- tions and can be seen as the tail of the accretion process of the two asteroids. Hereafter we indicate this population as primordial impactors. The collisions due to primordial im- pactors would stop once the region of space near each of 8 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) ) E L A C S . G O L ( I Y T I C R T N E C C E ) E L A C S . G O L ( I Y T I C R T N E C C E ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 1 0.1 0.01 0.001 0 2 4 6 8 10 1 0.1 0.01 0.001 0 SEMIMAJOR AXIS (AU) VESTA (0.50 AU MIGRATION) ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) ) E L A C S . G O L ( I Y T I C R T N E C C E ) E L A C S . G O L ( I Y T I C R T N E C C E ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 1 0.1 0.01 0.001 0 2 4 6 8 10 1 0.1 0.01 0.001 0 SEMIMAJOR AXIS (AU) VESTA (1.00 AU MIGRATION) ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) Figure 1. Semilogarithmic plots of the orbital elements of the impactors on Vesta in the a − e plane. Red and green symbols represent respectively the final (at impact) and initial orbits of ISL impactors, while the blue and magenta symbols represent respectively the final and initial orbits of BSL impactors. As is clearly visible from the plots, both ISL and BSL impactors can be divided in two dynamical families: a first non-resonant one (i.e. ISL primordial impactors and BSL scattered impactors) and a second resonant one (i.e. ISL and BSL resonant impactors). Note that these plots show only the dynamical classes of impactors recorded in the simulations: they are not normalised to the real disk population. the two asteroids is depleted, if not for the increasing grav- itational perturbations of Jupiter that inject new massless particles on Vesta-crossing or Ceres-crossing orbits. The second ISL group is composed of families of resonant impactors. The number of active resonances and their rela- tive abundances of impactors strongly depend on the extent of Jupiter's displacement and the target asteroid considered. We also stress that the active resonances in our simulations reflect the orbital configuration of our template of the early Solar System: the inclusion of Saturn, and to a lesser extent that of planetary embryos, would introduce new resonances in the system (see e.g. O'Brien, Morbidelli & Bottke (2007)). The flux of impactors due to the resonances with Jupiter increases with increasing displacements of the giant planet (see Figs. 1 and 2), as it is expected due to the sweeping of the resonances through wider spatial regions. The reso- nances affecting Vesta and Ceres in the temporal framework of our simulations are the 3 : 1 and the 2 : 1 resonances (see Figs. 1 and 2): we identified them using their locations, i.e. about 2.5 AU and 3.3 AU, in the simulations were Jupiter forms at its present position. Resonant impactors appear in the simulations about 105 years after the beginning of the rapid gas accretion phase, when Jupiter's mass is high enough to strongly excite the resonances (see Figs. 3 and 4). In the case of Vesta, resonant impactors from the 3 : 1 reso- nance are present in all scenarios, while those coming from the 2 : 1 resonance appears only in those simulations where Jupiter migrated by 0.5 AU or more. The case of Ceres is the opposite: the 2 : 1 resonance is always present in the four migration scenarios while the 3 : 1 resonance appears only in the scenarios where Jupiter started farther away from the Sun and migrated inward. Concerning the BSL population of planetesimals, the non- resonant class of impactors is populated by bodies randomly injected by Jupiter on Vesta-crossing or Ceres-crossing or- bits. In the following, we will refer to this class as the scat- tered impactors. Their contribution is more significant on Ceres than on Vesta due to the lower orbital excitation re- quired to reach the outer asteroid and their abundance is inversely proportional to that of the radial displacement of Jupiter. Collisions with scattered impactors are randomly distributed across the whole timespan we considered. The second class of BSL impactors is composed, similarly to the case of ISL planetesimals, by resonant impactors. The Collisional histories of early Vesta and Ceres 9 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ) E L A C S . G O L ( I Y T I C R T N E C C E ) E L A C S . G O L ( I Y T I C R T N E C C E ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 1 0.1 0.01 0.001 0 2 4 6 8 10 1 0.1 0.01 0.001 0 SEMIMAJOR AXIS (AU) CERES (0.50 AU MIGRATION) ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) ) E L A C S . G O L ( I Y T I C R T N E C C E ) E L A C S . G O L ( I Y T I C R T N E C C E ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 1 0.1 0.01 0.001 0 2 4 6 8 10 1 0.1 0.01 0.001 0 SEMIMAJOR AXIS (AU) CERES (1.00 AU MIGRATION) ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) Figure 2. Semilogarithmic plot of the orbital elements of impactors on Ceres in the a − e plane. Red and green symbols represent respectively the final (at impact) and initial orbits of ISL impactors, while the blue and magenta symbols represent respectively the final and initial orbits of BSL impactors. As in Fig. 1, both ISL and BSL impactors can be divided into a non-resonant (i.e. ISL primordial impactors and BSL scattered impactors) and a resonant (i.e. ISL and BSL resonant impactors) population. Note that these plots show only the dynamical classes of impactors recorded in the simulations: they are not normalised to the real disk population. resonances involved for the BSL planetesimals are the 3 : 2 (both for Vesta and Ceres) and the 7 : 6 (mainly for Ceres) resonances, which again we identified in the scenario where Jupiter formed at its present position where they are re- spectively located at ≈ 4 and ≈ 4.75 AU (see Figs. 1 and 2). Similarly to their ISL counterparts, BSL resonant im- pactors from the 3 : 2 resonance appear in the simulations about 105 years after the beginning of the rapid gas accre- tion phase (see Figs. 3 and 4). However, in the case of Ceres BSL resonant impactors from the 7 : 6 resonance appear about 4 × 105 years earlier, i.e. when the core of Jupiter is massive enough to excite the resonance (see Fig. 4). Concerning the dynamical features of the different classes of impactors on Vesta (see Fig. 1), the eccentricity values of primordial impactors are, as the name suggests, in the same range as the initial ones. Vesta's ISL resonant im- pactors (Fig. 1) have final eccentricity values distributed in the range 0.05 − 0.2 for objects coming from the 3 : 1 reso- nance and 0.2−0.5 for those coming from the 2 : 1 resonance. Concerning the BSL impactors, the scattered and resonant ones share about the same range of eccentricity values, i.e. 0.3− 0.7, but while the semimajor axes of the BSL resonant impactors concentrate between 3− 3.5 AU those of the scat- tered impactors are more widespread (3− 6 AU, see Fig. 1). As can be expected, the case of Ceres is the same as that of Vesta concerning ISL primordial impactors, i.e. the eccen- tricity values are the same as the initial ones (see Fig. 2). ISL resonant impactors on Ceres coming from the 3 : 1 reso- nance have eccentricity values distributed between 0.1 − 0.2 while those coming from the 2 : 1 resonance cover the range 0.1−0.4 (see Fig. 2). What we observed for Vesta concerning the BSL impactors remains valid also for Ceres: the scattered and resonant ones share about the same range of eccentric- ity values, i.e. 0.2− 0.7, but while the semimajor axes of the BSL resonant impactors concentrate between 3−4 AU those of the scattered impactors are more widespread (3.5 − 5.5 AU, see Fig. 4). 3.2 Characterisation of the flux of impactors The dynamical classes of impactors we showed in Figs. 1 and 2 and discussed in previous section represent the events recorded during our simulations and, as we anticipated, are not representative of the real flux of impactors. In Figs. 5 and 6 we show respectively the normalised distributions of frequency versus formation region of both ISL and BSL 10 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) ) U A ( S I X A R O J A M M E S L A I T I N I I ) U A ( S I X A R O J A M M E S L A I T I N I 10 9 8 7 6 5 4 3 2 1 10 9 8 7 6 5 4 3 2 1 ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS 10 9 8 7 6 5 4 3 2 1 ) U A ( S I X A R O J A M M E S L A I T I N I 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) I VESTA (0.50 AU MIGRATION) VESTA (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS 10 9 8 7 6 5 4 3 2 1 ) U A ( S I X A R O J A M M E S L A I T I N I I I 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 3. Temporal distribution of the impacts on Vesta due to the planetesimals showed in Fig. 1: the red symbols are related to ISL impactors, the blue ones to BSL impactors. ISL primordial impactors and BSL scattered impactors dominate the first 1.1 × 106 years in all scenarios, while ISL and BSL resonant impactors dominate the last 9 × 105 years. Like in Fig. 1, these plots show the temporal distribution of the events recorded in our simulations and they are not normalised to the real population of the disk. Table 1. Total number of impacts on Vesta, normalised to the real population of the disk and weighted on the impact proba- bility, for both the ISL and BSL dynamical families in the four migration scenarios of Jupiter. Migration scenario ISL impactors BSL impactors No migration 0.25 AU 0.50 AU 1.00 AU 17637 18086 20032 22105 199 9 4 7 Table 2. Total number of impacts on Ceres, normalised to the real population of the disk and weighted on the impact proba- bility, for both the ISL and BSL dynamical families in the four migration scenarios of Jupiter. Migration scenario ISL impactors BSL impactors No migration 0.25 AU 0.50 AU 1.00 AU 33502 37355 40622 42522 537 130 26 579 impactors on Vesta and Ceres in all the migration scenario we considered. The normalisation is obtained multiplying each recorded impact event for its characteristic probability, computed as described in Sect. 2.4, and the γ factor associated to each planetesimal as defined in Sect. 2.3. Once normalised to the real population of planetesimals and weighted on the computed impact probabilities, the contribution on Vesta (Ceres) of the ISL and BSL resonant impactors coming respectively from the 2 : 1 (3 : 1) and 7 : 6 resonances is greatly diminished respect to that of the 3 : 1 (2 : 1) and 3 : 2 resonances (see Figs. 5 and 6). The contribution of the 2 : 1 (3 : 1) resonance, however, is still significant, being of the order of a few thousands impact events (see Figs. 5 and 6). On the contrary, the number of BSL impactors is limited to a few impacts (see Figs. 5 and 6 and Tables 1 and 2) with the only exceptions of resonant impactors in the scenario where Jupiter did not migrate (both for Vesta and Ceres) and in that where Jupiter migrated by 1 AU (Ceres only). As is clearly visible from Figs. 3 and 4 and Tables 1 and 2, ISL impactors are 2− 3 orders of magnitude more abundant than BSL impactors and completely dominated the early cratering histories of the two asteroids. The frequency of BSL impactors is greatly enhanced in those migration Collisional histories of early Vesta and Ceres 11 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS 10 9 8 7 6 5 4 3 2 1 ) U A ( S I X A R O J A M M E S L A I T I N I 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) I CERES (0.50 AU MIGRATION) CERES (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS 10 9 8 7 6 5 4 3 2 1 ) U A ( S I X A R O J A M M E S L A I T I N I 10 9 8 7 6 5 4 3 2 1 10 9 8 7 6 5 4 3 2 1 ) U A ( S I X A R O J A M M E S L A I T I N I I ) U A ( S I X A R O J A M M E S L A I T I N I I I 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 4. Temporal distribution of the impacts on Ceres due to the planetesimals showed in Fig. 2: the red symbols are related to ISL impactors, the blue ones to BSL impactors. Differently from the case of Vesta shown in Fig. 3, the first 1.1 × 106 years are dominated by the fluxes of ISL primordial impactors, BSL scattered impactors and BSL resonant impactors linked to the 7 : 6 resonance. ISL and BSL resonant impactors then dominate again the last 9 × 105 years. Like in Fig. 2, these plots show the temporal distribution of the events recorded in our simulations and they are not normalised to the real population of the disk. scenarios where Jupiter excites resonances in the orbital region outside the Snow Line, as we mentioned in previous section. 3.2.1 ISL impactors The flux of ISL impactors on the two asteroids is charac- terised by two phases (see Figs. 7 and 8) strictly linked to the evolution of the forming Jupiter. As we previously mentioned, the first phase covers the time it takes the planet to form a critical mass core and accrete a significant mass of gas, i.e. 1.1×106 years. During this phase the flux of impactors on the asteroids is dominated by pri- mordial impactors. From the point of view of the number of impacts on Vesta due to ISL planetesimals, this phase accounts for 87% − 67% of the total amount of collisional events (see Fig. 7), with a decreasing trend for increasing values of the Jovian displacement. From the point of view of the mass impacting on Vesta the contribution of this first phase varies between 86% − 57% (see Fig. 9), again with a trend inversely proportional to the Jovian displacement. In the case of Ceres, the same ranges vary respectively be- tween 75% − 61% (see Fig. 8) and 73% − 55% (see Fig. 10) with the same inverse proportionality to the Jovian displace- ment. The temporal distributions of the number and mass fluxes of ISL impactors during this phase are characterised by "knees" due to the Jovian gravitational perturbations in- jecting new impactors on Vesta-crossing and Ceres-crossing orbits. The flux of impactors during the second phase, lasting 9×105 years, is instead dominated by resonant impactors originat- ing from the 3 : 1 and 2 : 1 resonances, as described in Sect. 3.1. As is straightforward to derive from the previous discus- sion, this phase accounts for 13% − 33% of the number flux and 14%− 43% of the mass flux on Vesta and for 25%− 39% of the number flux and 27%−45% of the mass flux on Ceres. The trend across the four migration scenarios is increasing for increasing values of the Jovian displacement. The total mass delivered to Vesta by ISL impactors varies between 8%− 12%, while for Ceres the range varies between 6% − 8%. The total number of impacts on Ceres is about twice as high as that on Vesta (see Tables 1 and 2). 12 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 200 150 100 50 0 0 200 150 100 50 0 0 BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) VESTA (0.50 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 Y C N E U Q E R F Y C N E U Q E R F 200 150 100 50 0 0 200 150 100 50 0 0 BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) VESTA (1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) INITIAL SEMIMAJOR AXIS (AU) Figure 5. Frequency versus formation region histograms of the impactors on Vesta, normalised over the real planetesimal population: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability and facilitate the comparison of the plots. Once normalised, ISL planetesimals completely dominate the flux of impactors on Vesta. The peak in BSL impactors in the top left plot is due to the excitation of the 3 : 2 orbital resonance with Jupiter. 3.2.2 BSL impactors The flux of BSL impactors greatly varies depending on the considered asteroid and migration scenario. For Vesta, the only relevant contribution of BSL impactors to the cratering history of the asteroid appears in the sce- nario where Jupiter did not migrate (see Figs. 7 and 9). In all the other scenarios, the contribution of BSL impactors on Vesta is limited to only few events (see Table 1). This implies that, even in the most favourable scenario, the ex- pected flux of volatile-rich material on Vesta is about 0.23% of the mass of the asteroid and is about 3% of the mass de- livered through rocky bodies in the same scenario. For Ceres the variations between the different scenarios are somewhat less erratic, yet they span over about an order of magnitude both concerning the number and the mass fluxes (see Figs. 8 and 10 and Table 2). The 3 : 2 and 7 : 6 res- onances are the major driver of the delivery of volatile-rich material (see Fig. 6). Even for Ceres, however, the deliv- ery of volatile-rich impactors is limited to about 0.24% of the mass of the asteroid or about 3% of the mass delivered through rocky bodies in the same scenario. It must be taken into account that the amount of volatile- rich material delivered by BSL impactors and retained by each target asteroid is only a small fraction of the previously reported values. Impact velocities of BSL planetesimals are always of the order of a few km s−1, therefore collisions will cause the erosion of the target asteroids and, due to its volatile-rich composition, the vaporisation of part or all the impactors. According to the data supplied by the Deep Im- pact experiment (Weiler et al. 2007), the bulk velocity of the gas emitted by comet 9P/Tempel 1 was ≈ 600 m s−1, while the outer regions of the outburst reached velocities of the order of 1 km s−1. These velocities exceed the escape veloc- ities from Vesta and Ceres, which are respectively ≈ 370 m s−1 and ≈ 510 m s−1. As a consequence, volatiles vaporised by the impact will be lost to the target asteroids. 3.3 Characterisation of the impacts From the point of view of their dynamical features, the im- pacts we recorded in our simulations can be divided into three classes. For what it concerns the impact velocity, in fact, there is little or no difference between BSL scattered impactors and BSL resonant impactors. The first class of impacts is that due to ISL primordial im- Collisional histories of early Vesta and Ceres 13 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 600 500 400 300 200 100 0 0 600 500 400 300 200 100 0 0 BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) CERES (0.50 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 Y C N E U Q E R F Y C N E U Q E R F 600 500 400 300 200 100 0 0 600 500 400 300 200 100 0 0 BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) CERES (1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) 2 4 6 8 10 INITIAL SEMIMAJOR AXIS (AU) INITIAL SEMIMAJOR AXIS (AU) Figure 6. Frequency versus formation region histograms of the impactors on Ceres, normalised over the real planetesimal population: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability and facilitate the comparison of the plots. The flux of BSL impactors, while by far inferior than that of ISL bodies, is significantly higher than in the case of Vesta (see the values reported in Tables 1 and 2). The peaks in BSL impactors are due to the excitation of the 3 : 2 and, to a lesser degree, the 7 : 6 orbital resonances with Jupiter. Table 3. Collisional erosion of Vesta due to planetesimals formed in a quiescent disk in the four migration scenarios of Jupiter. N.B.: the excavated depth is estimated is all cases where Vesta is not collisionally ablated assuming that the final radius of the asteroid is the present one. Migration scenario Excavation depth ISL impactors ISL impactors BSL impactors No migration 0.25 AU 0.50 AU 1.00 AU 20.80 km 23.32 km 49.31 km Ablation (post-core) 5.32 km 8.55 km 36.53 km Ablation 10.66 km 1.19 km 0.49 km 0.85 km pactors: it is characterised by low impact velocities, always inferior to 1 km/s for Vesta and to 600 m/s for Ceres, and a continuous mass spectrum up to about 8 × 1018 g (see Figs. 11 and 12). The second class is composed by impacts due to ISL reso- nant impactors: this class too has a continuous mass spec- trum, which extends up to about 2×1019 g (see Figs. 11 and 12). The characteristic impact velocities are systematically higher than those of the first class: they range between 1−10 km/s depending on the resonance originating the impactors (see Figs. 11 and 12). Finally, the third class of impacts is that due to the BSL planetesimals: as we anticipated, in fact, there is little or no evident distinction between the impacts due to BSL res- onant and scattered impactors (see Figs. 11 and 12). The only exception to this rule is the case of the 1 AU migration scenario for Ceres, were BSL resonant impactors cluster be- tween 2− 3 km/s while BSL scattered impactors are spread 14 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) ) . M U C , . M R O N ( S R O T C A P M I F O . N ) . M U C , . M R O N ( S R O T C A P M I F O . N 1.2 1 0.8 0.6 0.4 0.2 0 0 1.2 1 0.8 0.6 0.4 0.2 0 0 ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) VESTA (0.50 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 ) . M U C , . M R O N ( S R O T C A P M I F O . N ) . M U C , . M R O N ( S R O T C A P M I F O . N 1.2 1 0.8 0.6 0.4 0.2 0 0 1.2 1 0.8 0.6 0.4 0.2 0 0 ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) VESTA (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 7. Normalised cumulative distribution of the flux of impactors over time for Vesta: the red curve is related to ISL impactors, the blue one to BSL impactors. In all plots the fluxes have been normalised to their values at the end of the simulations (see Table 1). The flux of ISL impactors is divided into two main phases: a first one due to the contribution of primordial impactors and a second one, starting after Jupiter accreted a significant mass, mainly due to resonant impactors. In both phases, the cumulative distribution becomes more and more shallow towards the end, meaning that the flux eventually slows down and stops. The flux of BSL impactors begin when Jupiter's gaseous envelope reaches a significant mass and the cumulative distribution does not saturate but grows until the end of the simulations. Table 4. Collisional erosion of Ceres due to planetesimals formed in a quiescent disk in the four migration scenarios of Jupiter. N.B.: the excavated depth is estimated is all cases where Ceres is not collisionally ablated assuming that the final radius of the asteroid is the present one. Migration scenario Excavation depth ISL impactors ISL impactors BSL impactors No migration 0.25 AU 0.50 AU 1.00 AU 25.41 km 42.48 km 72.11 km 118.20 km (post-core) 13.80 km 31.45 km 61.91 km 109.60 km 9.70 km 2.81 km 0.74 km 7.19 km between 3 − 8 km/s (see Fig. 12). In general, impact veloci- ties of BSL impactors are spread over the range 2− 10 km/s (see Figs. 11 and 12). For each of the two asteroids, however, we recorded a isolated, extreme case of about 40 km/s (see Fig. 11 and 12). The mass spectrum of BSL impactors mostly overlap with that of ISL resonant impactors, extending up to about 2 − 3× 1019 g. (see Figs. 11 and 12), yet it favours masses higher than 3 × 1017 g. as a consequence of the mass-heliocentric distance relationship assumed in the model through Eq. 6. In the scenario where Jupiter does not migrate while form- ing, the craters produced by ISL and BSL planetesimals form two different populations. For increasing values of the Jovian displacement, however, the contribution of ISL reso- nant planetesimals tends to obliterate that of BSL planetes- imals for crater diameters up to 200 km. Collisional histories of early Vesta and Ceres 15 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ) . M U C , . M R O N ( S R O T C A P M I F O . N ) . M U C , . M R O N ( S R O T C A P M I F O . N 1.2 1 0.8 0.6 0.4 0.2 0 0 1.2 1 0.8 0.6 0.4 0.2 0 0 ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) CERES (0.50 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 ) . M U C , . M R O N ( S R O T C A P M I F O . N ) . M U C , . M R O N ( S R O T C A P M I F O . N 1.2 1 0.8 0.6 0.4 0.2 0 0 1.2 1 0.8 0.6 0.4 0.2 0 0 ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) CERES (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 8. Normalised cumulative distribution of the flux of impactors over time for Ceres: the red curve is related to ISL impactors, the blue one to BSL impactors. In all plots the fluxes have been normalised to their values at the end of the simulations (see Table 2). The considerations exposed for the case of Vesta (see Fig. 7) are valid also for Ceres. The flux of BSL impactors, however, is less erratic due to the lower orbital excitation of the planetesimals needed to reach Ceres respect to Vesta. For both asteroids, ISL planetesimals produce an asymmet- rical size-frequency distribution of craters, centred at d ≈ 30 km, that dominates the low-end tail, i.e. d 6 50 km, of the size spectrum of the craters in all scenarios (see Figs. 13 and 14). ISL resonant impactors are responsible for the asymme- try in the size-frequency distribution, causing a high-end tail that generally reaches up to d ≈ 100 km. For the highest val- ues of the Jovian displacement we considered, this high-end tail extends up to d ≈ 150− 200 km (see Figs. 13 and 14) as craters produced by ISL resonant impactors grow more and more abundant as discussed in Sect. 3.2. The size-frequency distribution of craters produced by BSL planetesimals is also asymmetrical: it extends mainly be- tween 30 − 200 km and is centred about at d ≈ 80 km (see Figs. 13 and 14). For Vesta, the BSL size-frequency distri- bution of craters becomes almost flat in all scenarios where Jupiter migrated (see Fig. 13). For Ceres, the BSL distri- bution remains bell-shaped in three scenarios over four (see Fig. 14). In our simulations, all craters greater than 150 km produced on Vesta and Ceres are due to BSL planetes- imals (see Figs. 13 and 14): the only exception to this rule is the scenario where Jupiter migrate by 1 AU, where also the high-end tail of the crater size spectrum is dominated by ISL impacts. It is interesting to note that, in the scenario where Jupiter migrates by 1 AU, the bombardment of BSL planetesimals produced one cratering event with diameter d ≈ 400 km on Vesta (see Fig. 13). Such event, however, is characterised by an extremely low probability (p ≈ 9%) even when normalised to the real population of planetesimals in the disk. As we previously mentioned, for increasing values of the ra- dial displacement of Jupiter the contribution of ISL resonant planetesimals grow in importance and the high-end tail of the size-frequency distribution of their craters completely overcomes the contribution of BSL planetesimals (see Figs. 13 and 14). To give an estimate of the effects of the bombardment of ISL and BSL planetesimals on the internal structure of the two asteroids, we computed the specific impact energy Q of the planetesimals expressed in units of the catastrophic dis- ruption energy Q∗D of Vesta and Ceres (see Figs. 15 and 16). The specific impact energy is defined as Q = 0.5mpv2 p/MT , where mp, vp are the mass and velocity of the projectile and MT is the mass of the target asteroid, while the catastrophic disruption energy is defined as the specific impact energy leading to the break-up of the target asteroid with a largest fragment containing 50% of its original mass. We evaluated the catastrophic disruption threshold Q∗D of Vesta and Ceres 16 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) ) . g , E V I T A L U M U C ( X U L F S S A M ) . g , E V I T A L U M U C ( X U L F S S A M 1e+21 1e+20 1e+19 1e+18 1e+17 1e+21 1e+20 1e+19 1e+18 1e+17 ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS ) . g , E V I T A L U M U C ( X U L F S S A M 1e+21 1e+20 1e+19 1e+18 1e+17 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) VESTA (0.50 AU MIGRATION) VESTA (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS ) . g , E V I T A L U M U C ( X U L F S S A M 1e+21 1e+20 1e+19 1e+18 1e+17 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 9. Cumulative distribution of the mass flux over time for Vesta: the red curve is related to ISL impactors, the blue one to BSL impactors. The mass flux of ISL impactors in all plots has been divided by a factor 100 to enhance the readability and facilitate the comparison of the plots. The overall features of the cumulative distribution of the mass flux are similar to those of the cumulative distribution of the flux of impactors (see Fig. 7 for details), yet these plots clearly show that the flux of BSL impactors on Vesta is erratic when Jupiter migrates, and the mass flux can consequently varies by more than one order of magnitude. using Eq. 6 from Benz & Asphaug (1999) and the coeffi- cients for basaltic targets computed by these authors (see Table 3, ibid). We used the coefficients of the vi = 3 km s−1 case (ibid) for all impact events with a velocity lower than 3 km s−1 and those of the vi = 5 km s−1 (ibid) for all the other impact events. In the case of Vesta, the impact energy spans in the range 10−10 − 10−2 (Fig. 15) yet impacts delivering more than 0.1% of Q∗D are extremely limited in number. In the case of Ceres, due to its greater mass, the upper limit of the range is shifted by an order of magnitude, i.e. the energy varies be- tween 10−10 − 10−3 (Fig. 16), with very few impact events delivering more than 0.01% of Q∗D. The single event exca- vating a d ≈ 400 km-wide crater on Vesta would actually deliver about 10% of the Q∗D of the asteroid. According to Benz & Asphaug (1999), such an impact would have caused the disruption of the asteroid leaving a largest fragment con- taining about 80% of its original mass (see Fig. 10 and Eq. 8, ibid). Finally, to characterise the cumulative effects of the recorded impacts on the two asteroids, we computed the total volume excavated by the Jovian bombardment under the simplify- ing assumption that the craters were distributed uniformly on the surfaces of Vesta and Ceres. This approach overes- timates the cratering rates at the polar regions respect to those at the equatorial regions, but it provides an interest- ing insight on the collisional evolution of the two asteroids. Our results for both the ISL and BSL populations of im- pactors are summarised in Tables 3 and 4, where they are expressed in terms of the depth of a spherical shell whose volume is equal to that of the excavated material and whose inner radius coincides with the present radius of the relevant asteroid. Since, due to their lower impact velocities and their proximity to each asteroid, ISL primordial impactors would likely have contributed to the accretion of Vesta and Ceres more than to their cratering histories, we also estimate the contribution of the ISL resonant impactors alone by consid- ering only ISL planetesimals impacting Vesta or Ceres after Jupiter started to accrete its gaseous envelope. In the follow- ing, we will discuss only the effects of ISL resonant impactors and BSL planetesimals assuming a simple cratering regime (i.e. no fracture creation and no weakening of the internal structure of the two asteroids). As is straightforward to see, BSL impactors contribute sig- nificantly to the crustal removal only in the scenario where Jupiter formed at its present position, where their contri- Collisional histories of early Vesta and Ceres 17 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS ) . g , E V I T A L U M U C ( X U L F S S A M 1e+22 1e+21 1e+20 1e+19 1e+18 1e+17 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) CERES (0.50 AU MIGRATION) CERES (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS ) . g , E V I T A L U M U C ( X U L F S S A M 1e+22 1e+21 1e+20 1e+19 1e+18 1e+17 ) . g , E V I T A L U M U C ( X U L F S S A M ) . g , E V I T A L U M U C ( X U L F S S A M 1e+22 1e+21 1e+20 1e+19 1e+18 1e+17 1e+22 1e+21 1e+20 1e+19 1e+18 1e+17 0 500000 1e+06 1.5e+06 2e+06 0 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 10. Cumulative distribution of the mass flux over time for Ceres: the red curve is related to ISL impactors, the blue one to BSL impactors. The number of ISL impactors in all plots has been divided by a factor 100 to enhance the readability and facilitate the comparison of the plots. The overall features of the cumulative distribution of the mass flux are similar to those of the cumulative distribution of the flux of impactors (see Fig. 8 for details). While the mass flux of BSL impactors varies by more than one order of magnitude between the different Jovian migration scenarios, its global features change more smoothly than in the case of Vesta. bution is of the same order of magnitude as that of ISL resonant impactors (see Tables 3 and 4). For Ceres, the to- tal excavated depth in this scenario is ≈ 20 km, while for Vesta the total excavated depth is ≈ 15 km. For increasing values of the Jovian displacement, on Ceres we noted an al- most linear increase in the excavated depth, which reaches a value of ≈ 110 km in the scenario where Jupiter migrated by 1 AU (see Table 4). This depth implies an excavated volume equal to 90% the present volume of Ceres, which is equiva- lent to assuming that the original mass of Ceres was about twice the present one. Interestingly, in the same scenario we noticed that the Jovian bombardment collisionally ablates Vesta, removing an amount of material actually twice as great as the mass of the asteroid (see Table 3). Moreover, in the scenario where Jupiter migrates by 0.5 AU both as- teroids are stripped of an amount of material equivalent to roughly 50% their present volumes. If Jupiter migrated by more than 0.5 AU, therefore, our re- sults indicate that to survive the early bombardment caused by the giant planet the primordial Vesta and Ceres should have been between 150−300% their present sizes, even under the simplifying assumption of a simple excavation regime. 3.4 Effects of a slower migration of Jupiter As we discussed in Sect. 2.2, the radial displacement of Jupiter is governed by Eq. 3 and is characterised by an e-folding time τg = 5 × 103 years, which implies a migra- tion timescale consistent with the indications of theoretical models. However, in order to evaluate the dependence of the results on our assumption on the migration timescale, we considered an additional case respect to our four refer- ence scenarios. We forced Jupiter to migrate inward again by 1 AU but with an e-folding time five times as large as in our original simulations, i.e. τg = 2.5 × 104 years. This is equivalent to assuming a value of τM = a/ a ≈ 4 × 105 years at 5.2 AU, which is consistent with the upper boundary to the migration timescale reported by D'Angelo, Kley & Hen- ning (2003). A summary of the results for both Vesta and Ceres is given by Figs. 17 and 18. As can be easily seen by comparing the left-hand side plots in Figs. 17 and 18 with the bottom-right ones of Figs. 1, 9, 13 and 15, the change in the migration timescale does not af- fect in any significant way the collisional evolution of Vesta respect to our reference case with Jupiter migrating by 1 AU. The same hold true for the ISL impactors on Ceres, 18 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 MASS (G.) VESTA (0.50 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 MASS (G.) VESTA (1.00 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 0.01 1e+17 MASS (G.) 1e+18 1e+19 1e+20 MASS (G.) Figure 11. Dynamical characterisation of the impacts on Vesta: red symbols indicated ISL impactors and blue ones BSL impactors. The data reproduced in the plots are the events recorded in our simulations: they have not been normalised to the real population of the planetesimal disk. As can be seen from the plots, ISL resonant and primordial impactors are characterised by different dynamical features, while no difference is evident between BSL scattered and resonant impactors in those scenario with Jupiter migrating. In the no-migration scenario BSL impactors seem to divide in two clusters, a first between 3 − 5 km/s and a second between 8 − 10 km/s. The relative abundance of bodies in the BSL scattered and resonant groups in Fig. 1, however, argues against attributing this effect to the non-resonant versus resonant subdivision. as can be seen by comparing the right-hand side plots in Figs. 17 and 18 with the bottom-right ones of Figs. 2, 10, 14 and 16. The flux of BSL impactors on Ceres, on the con- trary, is enhanced by a factor 4 respect to the reference case. This enhancement, however, affects the results we previously discussed only from a quantitative point of view, not quali- tatively. The major role in the collisional evolution of Ceres is still played by ISL impactors, whose flux during the bom- bardment is an order of magnitude higher than that of their BSL counterparts. While the link between the Jovian migration rate and ef- ficiency of the giant planet in exciting BSL resonant im- pactors needs further investigation, as a consequence of this additional run we feel confident that the global picture con- cerning the collisional evolution and the survival of both Vesta and Ceres does not depend on our assumptions on the migration rate of Jupiter. 3.5 Collisional evolution in a turbulent disk As we described in Sect. 2.3, the size distribution of the planetesimals in our simulations has been derived under the assumption that these bodies formed by gravitational insta- bility of the dust in a quiescent disk (Safronov 1969; Gol- dreich & Ward 1973). According to Coradini, Federico & Magni (1981) and taking into account the change in density we assumed to take place across the Snow Line, the average diameters of such planetesimals roughly vary between 5− 60 km in the spatial range 1 − 40 AU or, equivalently, between 7 − 25 km in the interval 2 − 10 AU. formation mechanisms, however, Different planetesimal would produce different size distributions and, as a conse- quence, different planetesimal abundances than the one we considered. In particular, theoretical models of planetesimal formation in turbulent disks (Johansen et al. 2007; Cuzzi et al. 2008) predict primordial bodies whose average sizes ex- ceed those contemplated by our model. Therefore, we pro- ceeded to test the effects of different size distributions of the planetesimals populating the disk on our results. To do this, we took advantage of the results of Morbidelli et al. (2009) Collisional histories of early Vesta and Ceres 19 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 MASS (G.) CERES (0.50 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I / ) S M K ( Y T I C O L E V T C A P M I ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 MASS (G.) CERES (1.00 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS 100 10 1 0.1 0.01 1e+17 1e+18 1e+19 1e+20 0.01 1e+17 MASS (G.) 1e+18 1e+19 1e+20 MASS (G.) Figure 12. Dynamical characterisation of the impacts on Ceres: red symbols indicated ISL impactors and blue ones BSL impactors. The data reproduced in the plots are the events recorded in our simulations: they have not been normalised to the real population of the planetesimal disk. As can be seen from the plots, ISL resonant and primordial impactors are characterised by different dynamical features, while no difference is evident between BSL resonant and non-resonant impactors in the first three migration scenarios. In the 1 AU migration scenario, resonant BSL impactors from the 3 : 2 resonance cluster in the range 2 − 3 km/s while non-resonant BSL impactors and the few resonant ones from the 7 : 6 resonance are characterised by higher velocities. and Chambers (2010). Morbidelli et al. (2009) did not explore a specific model of planetesimal formation in quiescent or turbulent disks but instead tried to constrain the initial size-frequency distribu- tion (SFD in the following) of planetesimals in the orbital region of the Main Asteroid Belt. Their results suggest that the best match with the present-day SFD of the Main As- teroid Belt is obtained for planetesimal sizes initially span- ning 100 − 1000 km (see Fig. 8, ibid), a range consistent with their formation in a turbulent nebula. Morbidelli et al. (2009) supplies two SFDs associated to this case: a first one describing the primordial SFD of the planetesimals, which spans 100 − 1000 km (see Fig. 8a, black dots, ibid), and a second, collisionally evolved one where accretion and break- up of the primordial planetesimals extended the size distri- bution between 5 − 5000 km (see Fig. 8a, black solid line, ibid). For each ISL impact event in our simulations, we then extracted the mass and the normalisation factor of the im- pacting planetesimal through simple Montecarlo extractions based on the cumulative probability distributions of the two SFDs supplied by Morbidelli et al. (2009). In using the re- sults of Morbidelli et al. (2009) to evaluate the collisional evolution of Vesta and Ceres we did not considered the con- tribution of BSL impactors, since the authors investigated the initial SFD only of planetesimals in the orbital region of the Main Asteroid Belt. Chambers (2010) instead studied the planetesimal accre- tion efficiency following the model by Cuzzi et al. (2008), where turbulence in the solar nebula act to concentrate dust particles in low vorticity regions. For this model, Cham- bers (2010) derived the accretion timescale and the size- heliocentric distance relationship of the planetesimals as a function of the mass and the gas to dust ratio of the solar nebula. We adopted the size distribution associated to a disk with gas surface density at 1 AU σ′0 = 3500 g cm−2, gas to dust ratio ξ′ = 0.3 beyond the Snow Line and ξ′ = 0.15 inside the Snow Line, and a nebula density profile with ex- ponent n′s = −1 (see Fig. 14, gray dot-dashed line, ibid). Chambers (2010) assumed the Snow Line being placed at 2.7 AU for such nebula: due to our division of the disk into concentric rings, we assumed the Snow Line at 3.0 AU when using the results of this author. As we mentioned, the re- sults of Chambers (2010) supply the average diameter of planetesimals as a function of heliocentric distance: from 20 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 100 80 60 40 20 0 100 80 60 40 20 0 ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS Y C N E U Q E R F 100 80 60 40 20 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 CRATER DIAMETER (KM) CRATER DIAMETER (KM) VESTA (0.50 AU MIGRATION) VESTA (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS Y C N E U Q E R F 100 80 60 40 20 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 13. Normalised frequency versus crater size histograms of the impacts on Vesta: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. As can be easily seen, the high-end tail of the ISL crater size distribution moves towards bigger sizes for increasing values of the Jovian radial displacement. The high-end tail of the BSL distribution shows a somewhat opposite behaviour and its contribution to the surface cratering is eventually completely masked by the one of ISL impactors, with the only exception of stochastic events producing craters bigger than 200 km (see bottom right panel). Fig. 14 (gray dot-dashed line) in Chambers (2010) we can derive the following analytical expression: D = D0(cid:16) r 1 AU(cid:17)β ′ (36) where D0 = 70 km is the average diameter of the plan- etesimals at 1 AU and β′ = 0.4935. We can then derive a semi-empirical relationship, analogous to Eq. 6, linking the average mass of planetesimals to heliocentric distance, i.e. m′p = π 6 ρ′D3 0 (cid:16) r 1 AU(cid:17)3β ′ (37) where ρ′ = 3.0 g cm−2 in the ISL region and ρ′ = 1.0 g cm−2 in the BSL region. By substituting the primed quantities to the original ones in Eqs. 13, 14 and 19, we can thus obtain the mass and the normalisation factor for each planetesimal through the same approach we used for our original disk. A summary of the results we obtained in using these three size distributions for the planetesimals in the disk are re- ported in Tables 5 and 6, where we computed the total ex- cavation depth of the Jupiter-driven bombardment after the giant planet started to accrete the nebular gas, and in Figs. 19-26, where we showed the size distributions of the craters and the energies released by the impact events. As can be seen, the possible collisional histories computed for Vesta and Ceres using the primordial SFD from Morbidelli et al. (2009) and the size distribution from Chambers (2010) de- scribed by Eq. 36 are inconsistent with the survival of the asteroids to the Jovian early bombardment due to the fol- lowing reasons. First, impacts capable to disrupt Vesta (i.e. with Q > Q∗D) are present in all our simulations, even if they are charac- terised by very low probabilities (see Figs. 23 and 25). This holds true also for Ceres in the case of the primordial SFD by Morbidelli et al. (2009) (see Fig. 24) but not in the case of the size distribution by Chambers (2010) (see Fig. 26). In addition, impacts delivering a significant fraction (i.e. Q > 10−2Q∗D) of the dispersal energies Q∗D of Vesta and Ceres are abundant (as shown in Figs. 23-26), especially in those scenarios where Jupiter migrated by 0.5 AU or more. The cumulative effects of these energetic impacts, not ac- counted for in our simplified calculations, would likely result in the weakening of the internal structure of the asteroids and in their eventual destruction. Second, even if Vesta and Ceres could survive the Jovian Collisional histories of early Vesta and Ceres 21 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS Y C N E U Q E R F 200 150 100 50 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 CRATER DIAMETER (KM) CRATER DIAMETER (KM) CERES (0.50 AU MIGRATION) CERES (1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS ISL IMPACTORS (x10-2) BSL IMPACTORS Y C N E U Q E R F 200 150 100 50 0 Y C N E U Q E R F Y C N E U Q E R F 200 150 100 50 0 200 150 100 50 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 14. Normalised frequency versus crater size histograms of the impacts on Ceres: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. Similarly to what we observed for Vesta in Fig. 13, the high-end tail of the ISL crater size distribution moves towards bigger sizes for increasing values of the Jovian radial displacement. The contribution of BSL impactors to the surface cratering is eventually masked by the one of ISL impactors. early bombardment without being disrupted, the cumula- tive volume excavated by the impactors would result in the ablation of the two asteroids even assuming a simple crater- ing regime (see Tables 5 and 6). Moreover, for those impacts producing craters whose size is comparable to the diame- ter of the relevant asteroid (see Figs. 19-22), the cratering regime described by Eq. 35 is not appropriate. According to Eq. 8 by Benz & Asphaug (1999), impacts characterised by Q ∼ 10−2Q∗D would remove from the target asteroid a mass of the order of ∼ 1% for impact velocities of about 3 km s−1 and of 15 − 20% for impact velocities of about 5 km s−1 (not taking into account the cumulative effects of such energetic impacts). As a consequence, the survival of Vesta and Ceres in disks where the size distribution of the plan- etesimals is governed by the primordial SFD from Morbidelli et al. (2009) and Eq. 36 as in Chambers (2010) appears as extremely unlikely. Even in the scenario where the bombard- ment is less efficient (i.e. Jupiter migrating inward by 0.25 AU), Ceres would be collisionally ablated while Vesta would be stripped of a mass equivalent to about 80% its present one. The case of the collisionally evolved SFD from Morbidelli et al. (2009) is different. In such collisionally evolved SFD, about 84% of the ISL population is represented by plan- etesimals whose diameter is ∼ 5 km and about 97% of said population is characterised by diameters in the range 5− 10 km, i.e. our original size range in the ISL orbital region. As is shown in Figs. 19 and 20 for Vesta and Ceres respectively, the crater size distributions computed using this SFD are similar to the ones we obtained for our original cases. The bulk of the craters is characterised by diameters lower than 50 − 75 km depending on the migration scenario consid- ered, while the values of the impact energy Q are generally Q 6 10−4Q∗D for Vesta and Q 6 10−5Q∗D for Ceres. Respect to the case of planetesimals formed in a quiescent disk, the collisional histories of Vesta and Ceres estimated using this second SFD from Morbidelli et al. (2009) are characterised by higher probabilities of highly energetic impacts, with en- ergies Q > 10−2Q∗D (see Figs. 23 and 24) and producing craters with diameters bigger than 150 − 200 km according to Eq. 35 (see Figs. 19 and 20). However, even neglecting the contribution of BSL impactors, the collisional histories computed for Vesta using such SFD suggest that the aster- oid could survive the bombardment only if Jupiter migrated by less than 0.5 AU. Even in this case, the equivalent volume excavated by impacts is about twice than that estimated in 22 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 100 80 60 40 20 0 -10 100 80 60 40 20 0 -10 BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 LOG(Q/Q) VESTA (0.50 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 Y C N E U Q E R F Y C N E U Q E R F 100 80 60 40 20 0 -10 100 80 60 40 20 0 -10 BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 LOG(Q/Q) VESTA (1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 LOG(Q/Q) LOG(Q/Q) Figure 15. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Vesta: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. The most energetic impacts (of the order of 10−2) are due to BSL bodies in those scenarios where the Jovian displacement is limited (no migration or 0.25 AU). ISL impacts in all scenarios and BSL impacts in the 0.5 and 1 AU migration scenarios are limited to impact energies inferior to 10−3, with the only exception of the 400 km wide crater reported in Fig. 13. This event, while characterised by an low probability (p ≈ 9%) even after normalisation, would deliver a significant fraction (about 10−1) of the energy needed to disrupt Vesta and orbitally disperse the resulting fragments. the case of our original computations (see Table 5). Accord- ing to our results, the survival of Ceres is even less plausible: being located between the 3 : 1 and the 2 : 1 resonances with Jupiter, the asteroid is collisionally ablated by the Jovian bombardment in three scenarios over four. Moreover, even if Jupiter did not migrate the volume excavated would be of the order of half the present one of the asteroid (see Table 6), thus implying a more massive primordial Ceres. 4 IMPLICATIONS FOR VESTA, CERES AND THE EARLY SOLAR SYSTEM The picture that our results supply of the time of Jupiter's formation clearly states that the formation of the giant planet has been one of the milestones in the evolution of the early Solar System and in particular of the region of the present Main Asteroid Belt. In the following, we will discuss the consequences of our results for Jupiter, Vesta, Ceres and the early Solar System. 4.1 Planetesimals in the primordial Solar System The results we described in Sect. 3.3 and in Sect. 3.5, in par- ticular the collisional erosion and the likelyhood of survival of Vesta and Ceres to the Jupiter-induced bombardment de- picted by Tables 3-6, give us interesting constrains on the population of planetesimals in the Solar Nebula. As we explained in Sect. 2.3, in our simulations we initially assumed that all the mass of the disk was accounted for by planetesimals formed by gravitational instability of the dust in a quiescent nebula and whose diameters spanned the range 7 − 25 km. According to our results, the Jovian early bombardment would cause an intense cratering and a sig- nificant erosion of the surfaces of Vesta and Ceres but the two asteroids would survive it if the giant planet migrated by less than 0.5 AU (see discussion in Sect. 4.2). We then tested the effects of the Jovian early bombardment in disks characterised by different size distributions of the planetesimals: specifically, we took advantage of the results of Morbidelli et al. (2009) and Chambers (2010). According to our results (see Tables 5 and 6), the survival of Vesta and Collisional histories of early Vesta and Ceres 23 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 LOG(Q/Q) CERES (0.50 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 Y C N E U Q E R F Y C N E U Q E R F 300 250 200 150 100 50 0 -10 300 250 200 150 100 50 0 -10 BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 LOG(Q/Q) CERES (1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) -8 -6 -4 -2 0 Y C N E U Q E R F Y C N E U Q E R F 300 250 200 150 100 50 0 -10 300 250 200 150 100 50 0 -10 LOG(Q/Q) LOG(Q/Q) Figure 16. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Ceres: red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. Due to the bigger size of Ceres, the impact energies delivered by both ISL and BSL planetesimals in all migration scenarios are of the order of 10−3 or lower. Table 5. Collisional erosion of Vesta due to planetesimals formed in a turbulent disk in the four migration scenarios of Jupiter considering only the post-core bombardment. N.B.: the excavated depth is estimated is all cases where Vesta is not collisionally ablated assuming that the final radius of the asteroid is the present one. Migration scenario Excavation depth (turbulent disk) Primordial SFD Evolved SFD ISL impactors BSL impactors (Morbidelli et al. 2009) (Morbidelli et al. 2009) (Chambers 2010) (Chambers 2010) No migration 0.25 AU 0.50 AU 1.00 AU Ablation Ablation Ablation Ablation 9.847 km 14.12 km Ablation Ablation 55.88 km 46.13 km Ablation Ablation Ablation 16.69 km 11.62 km 15.94 km Ceres to the Jovian early bombardment is extremely unlikely in disks populated by massive planetesimals like those ex- pected to form in turbulent disks (e.g. the size distribution by Chambers (2010) and the primordial SFD by Morbidelli et al. (2009)). Disks where the population of planetesimals is dominated by small bodies (i.e. D 6 20 km) but where a significant frac- tion of the mass is in the form of larger objects (i.e. D > 1000 km), as the one suggested by the collisionally evolved SFD by Morbidelli et al. (2009), represent a less hostile environ- ment from the perspective of the survival of Vesta and Ceres to the Jovian early bombardment. However, the abundances of planetesimals predicted by Morbidelli et al. (2009) cause a bombardment on Ceres four times as erosive than the one estimated for our original disk, thus making its survival ex- tremely unlikely. Moreover, the collisional erosion of Vesta would be about twice as large as the one we estimated for our original disk, therefore making the survival of Vesta even less plausible if the giant planet migrated by 0.5 AU or more (see discussion in Sect. 4.2). 24 Turrini et al. VESTA (τ=2x104 years, 1.00 AU MIGRATION) CERES (τ=2x104 years, 1.00 AU MIGRATION) ) E L A C S . G O L ( I Y T I C R T N E C C E ) . g , E V I T A L U M U C ( X U L F S S A M 1 0.1 0.01 0.001 0 1e+21 1e+20 1e+19 1e+18 1e+17 0 ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) VESTA (τ=2x104 years, 1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 ) E L A C S . G O L ( I Y T I C R T N E C C E ) . g , E V I T A L U M U C ( X U L F S S A M 1 0.1 0.01 0.001 0 1e+22 1e+21 1e+20 1e+19 1e+18 1e+17 0 ISL (final orbits) ISL (initial orbits) BSL (final orbits) BSL (initial orbits) 2 4 6 8 10 SEMIMAJOR AXIS (AU) CERES (τ=2x104 years, 1.00 AU MIGRATION) ISL IMPACTORS (x10-2) BSL IMPACTORS 500000 1e+06 1.5e+06 2e+06 TIME (YEARS) TIME (YEARS) Figure 17. Jovian early bombardment for Jupiter migrating with an e-folding time of 2.5 × 104 years. Upper plots: semilogarithmic plots of the orbital elements of the impactors on Vesta (left) and Ceres (right) in the a − e plane. Red and green symbols represent respectively the final (at impact) and initial orbits of ISL impactors, while the blue and magenta symbols represent respectively the final and initial orbits of BSL impactors. As in Figs. 1 and 2, these plots show only the dynamical classes of impactors recorded in the simulations: they are not normalised to the real disk population. Lower plots: cumulative distribution of the mass flux over time for Vesta (left) and Ceres (right). The red curve is related to ISL impactors, the blue one to BSL impactors. The mass flux of ISL impactors in all plots has been divided by a factor 100 to enhance the readability and facilitate the comparison of the plots. Table 6. Collisional erosion of Ceres due to planetesimals formed in a turbulent disk in the four migration scenarios of Jupiter considering only the post-core bombardment. N.B.: the excavated depth is estimated is all cases where Ceres is not collisionally ablated assuming that the final radius of the asteroid is the present one. Migration scenario Excavation depth (turbulent disk) Primordial SFD Evolved SFD ISL impactors BSL impactors (Morbidelli et al. 2009) (Morbidelli et al. 2009) (Chambers 2010) (Chambers 2010) No migration 0.25 AU 0.50 AU 1.00 AU Ablation Ablation Ablation Ablation 69.54 km Ablation Ablation Ablation Ablation Ablation Ablation Ablation Ablation 68.54 km 21.28 km Ablation Before proceeding, we must point out that, strictly speak- ing, our results refer to a disk of planetesimals which did not undergo any process of mass depletion like the one im- plied by the "native embryos" model (Wetherill 1992; Petit, Morbidelli & Chambers 2001; O'Brien, Morbidelli & Bottke 2007). Again, we stress that the mass depletion suggested by the Nice Model (Gomes et al. 2005) is located several 108 years in the future respect to the timeframe we explored and thus has no implications for the present work. Accord- ing to Petit, Morbidelli & Chambers (2001), the presence of planetary embryos in the region of the Main Asteroid Belt does not cause a significant depletion of the planetesimals if Jupiter and Saturn are not present. This implies that the mass depletion process should be active during the last 106 Collisional histories of early Vesta and Ceres 25 VESTA (τ=2x104 years, 1.00 AU MIGRATION) CERES (τ=2x104 years, 1.00 AU MIGRATION) ISL IMPACTORS (x30-1) BSL IMPACTORS 50 100 150 200 250 300 CRATER DIAMETER (KM) CERES (τ=2x104 years, 1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-1) ISL IMPACTORS (x10-2) BSL IMPACTORS 50 100 150 200 250 300 CRATER DIAMETER (KM) VESTA (τ=2x104 years, 1.00 AU MIGRATION) BSL IMPACTORS ISL IMPACTORS (x10-2) 700 600 500 400 300 200 100 0 0 1400 1200 1000 800 600 400 200 Y C N E U Q E R F Y C N E U Q E R F -8 -6 -4 -2 0 0 -10 -8 -6 -4 -2 0 LOG(Q/Q) LOG(Q/Q) Y C N E U Q E R F Y C N E U Q E R F 100 80 60 40 20 0 0 100 80 60 40 20 0 -10 Figure 18. Jovian early bombardment for Jupiter migrating with an e-folding time of 2.5 × 104 years. Upper plots: normalised frequency versus crater size histograms of the impacts on Vesta (left) and Ceres (right). Red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors has been divided by a factor 100 for Vesta and 30 for Ceres to enhance the readability of the plots. Lower plots: normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Vesta (left) and Ceres (right). Red bars are those related to ISL impactors, blue ones to BSL impactors. The frequency of ISL impactors has been divided by a factor 100 for Vesta and 10 for Ceres to enhance the readability of the plots. years of our simulations, likely with a lower efficiency than reported by O'Brien, Morbidelli & Bottke (2007) since we assumed that the formation of Saturn is delayed by several 105 years respect to that of Jupiter. According to the re- sults of O'Brien, Morbidelli & Bottke (2007) for the case where Jupiter and Saturn were initially on circular orbits (CJS, ibid), the combined perturbations of the giant planets and the planetary embryos should account for a depletion of about 10% of the original planetesimals on this timespan. As a consequence of the short timescale the Jovian early bom- bardment acts on (few 105 years), the depletion mechanism of the "native embryos" model should not affect our results significantly. It must be noted, moreover, that a higher ef- ficiency of the depletion mechanism on a timespan of 106 years, like the one estimated by O'Brien, Morbidelli & Bot- tke (2007) for the case where Jupiter and Saturn were on or- bits similar to their present ones (i.e. ∼ 30%), would favour the survival of Vesta and Ceres without qualitatively chang- ing the global picture supplied by our results. In the case of disks with a high abundance of small planetesimals and a significant fraction of the mass in the form of large plan- etesimals, e.g. the one described by the collisionally evolved SFD from Morbidelli et al. (2009), depletion efficiencies as high as the one estimated by Petit, Morbidelli & Chambers (2001), i.e. ∼ 70%, could be required to ensure the survival of Ceres. 4.2 Jupiter According to our simulations, the extent of Jupiter's radial migration due to disk-planet interactions has major implica- tions for the survival of Vesta and Ceres to the Jovian early bombardment. The timescale of the migration also plays a role in influencing the intensity of the bombardment (see Sect. 3.4), but mainly for what concerns the flux of BSL im- pactors. Thus, it influences the production of major craters (i.e. D > 100 km) but not the likelyhood of the survival of Vesta and Ceres to the Jupiter-induced bombardment. In the case of our original disk of planetesimals, our results set a upper limit of 0.5 AU to Jupiter's migration. For ra- dial displacements higher than 0.5 AU during its formation, in fact, the sweeping of the mean motion resonances with Jupiter through the disk of planetesimals causes an enhance- ment in the flux of impactors so high that Vesta is effectively 26 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 140 120 100 80 60 40 20 0 0 140 120 100 80 60 40 20 0 0 PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 200 400 600 800 1000 CRATER DIAMETER (KM) VESTA (0.50 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 200 400 600 800 1000 Y C N E U Q E R F Y C N E U Q E R F 140 120 100 80 60 40 20 0 0 140 120 100 80 60 40 20 0 0 PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 200 400 600 800 1000 CRATER DIAMETER (KM) VESTA (1.00 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 200 400 600 800 1000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 19. Normalised frequency versus crater size histograms of the impacts on Vesta computed using the primordial (red bars) and collisionally evolved (blue bars) size-frequency distributions in the inner Solar System from Fig. 8a in Morbidelli et al. (2009). The frequency of collisionally evolved impactors in all plots has been divided by a factor 1000 to enhance the readability of the plots. Note that the craters whose size is comparable to or greater than the diameter of the asteroid are reported only to show the number of highly energetic impacts, since Eq. 35 is not appropriate for such cases. As can be easily seen from the plots, the collisionally history of Vesta due to the primordial size-frequency distribution would result in the destruction of the asteroid. The collisionally evolved size-frequency distribution would result in a impact history more similar to the one obtained for the quiescent disk (see Fig. 13), but with a higher cumulative probability of impacts capable to destroy to asteroid. The survival and global impact erosion outcomes of the models are summarised in Table 5. collisionally ablated and Ceres is stripped of a volume equiv- alent to 90% of its present one (see Table 3). Moreover, in the case of a radial migration of Jupiter of about 0.5 AU, the volume excavated by the impacts on both asteroids is of the order of half their present volumes, implying the complete removal of their crustal layers. As a consequence, the radial migration of Jupiter due to disk-planet interactions should have been inferior to 0.5 AU. For a disk numerically dominated by small bodies but with a significant fraction of the mass in the form of larger objects like in the case of the collisionally evolved SFD by Mor- bidelli et al. (2009), the constrains on the radial migration of Jupiter implied by the survival of Vesta and Ceres are even stricter, i.e. the displacement should have been lower than 0.25 AU. According to our results, in fact, Vesta would survive the Jovian early bombardment for displacements of the giant planet inferior to 0.5 AU. However, being located between the 3 : 1 and the 2 : 1 resonances, Ceres would be subjected to a bombardment so intense that it would be col- lisionally ablated even for displacements of the giant planet of the order of 0.25 AU. 4.3 Ceres As we discussed previously in Sect. 4.1 and 4.2 and is shown in Tables 4 and 6, the survival of Ceres to the Jovian early bombardment is linked to the extent of Jupiter's migration and to the characteristics of the planetesimals populating the Solar Nebula. According to our results, the survival of Ceres to the flux of impactors coming from the 3 : 1 and the 2 : 1 resonances would require a limited displacement of the forming Jupiter, inferior to 0.5 AU and possibly lower than 0.25 AU. If we consider only those scenarios where Ceres survived the bombardment, the flux of impactors would have removed a significant fraction of the original crust of the asteroid, ejecting the fragments in the Main Asteroid Belt. Even in the less collisionally active scenarios, in fact, the primordial Collisional histories of early Vesta and Ceres 27 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 200 150 100 50 0 0 200 150 100 50 0 0 PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 500 1000 1500 2000 CRATER DIAMETER (KM) CERES (0.50 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 500 1000 1500 2000 Y C N E U Q E R F Y C N E U Q E R F 200 150 100 50 0 0 200 150 100 50 0 0 PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 500 1000 1500 2000 CRATER DIAMETER (KM) CERES (1.00 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-3) 500 1000 1500 2000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 20. Normalised frequency versus crater size histograms of the impacts on Ceres computed using the primordial (red bars) and collisionally evolved (blue bars) size-frequency distributions in the inner Solar System from Fig. 8a in Morbidelli et al. (2009). The frequency of collisionally evolved impactors in all plots has been divided by a factor 1000 to enhance the readability of the plots. Note that the craters whose size is comparable to or greater than the diameter of the asteroid are reported only to show the number of highly energetic impacts, since Eq. 35 is not appropriate for such cases. As in the case of Vesta in Fig. 19, the collisionally history of Ceres due to the primordial size-frequency distribution would result in the destruction of the asteroid. Again, the collisionally evolved size-frequency distribution would result in a impact history similar to the one obtained for the quiescent disk (see Fig. 14), but with a higher cumulative probability of impacts capable to destroy to asteroid. The survival and global impact erosion outcomes of the models are summarised in Table 6. Ceres could have been ≈ 15− 20% bigger than it is now. In- dependently by the migration scenario considered, moreover, the Jovian early bombardment would have left the surface of the asteroid saturated with craters spanning up to 150 km, with a small number of bigger craters in the range 150− 300 km. It is reasonable to assume that bodies as big as Ceres popu- lated the Main Asteroid Belt at the time of Jupiter's forma- tion, yet we do not know if Ceres belonged to such popula- tion. The only indication in this sense is probably its mass: it is not easy, in fact, to form planetesimals as large as Ceres in the Main Asteroid Belt once Jupiter started to perturb it. If the accretion of Ceres ended before the formation of Jupiter, the surface of the asteroid could have kept records of the bombardment induced by the giant planet. If the ac- cretion of Ceres ended at a later time, it is likely that its late phases reshaped the surface and erased every vestigial feature of this ancient bombardment. Similar considerations apply also to the reshaping of the as- teroid due to its thermal evolution. Thomas et al. (2005) suggest that Ceres could be a differentiated body composed of water ice and silicate material yet, if true, we do not know if its differentiation ended before the Jovian early bombard- ment or if it ended later, causing a resurfacing that altered its primordial collisional features. Finally, we must take into account that the surface of Ceres should have been altered by its subsequent collisional evolu- tion through the 4 Ga of the life of the Solar System and by the event known as the Late Heavy Bombardment. Before we can apply our results to probe the early history of Ceres, we therefore must assess to which extent later collisional or thermal events modified or erased the primordial features of its surface. 4.4 Vesta Respect to the case of Ceres, the survival of Vesta puts less stringent constrains to the extent of Jupiter's migration: ac- 28 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 25 20 15 10 5 0 25 20 15 10 5 0 ISL IMPACTORS BSL IMPACTORS ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 25 20 15 10 5 0 0 500 1000 1500 2000 0 500 1000 1500 2000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) VESTA (0.50 AU MIGRATION) VESTA (1.00 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 25 20 15 10 5 0 0 500 1000 1500 2000 0 500 1000 1500 2000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 21. Normalised frequency versus crater size histograms of the impacts on Vesta computed using the diameter-heliocentric distance relationship for planetesimals from Fig. 14 in Chambers (2010): red bars are those related to bodies formed in the region of ISL impactors, blue ones to bodies formed in the region of BSL impactors. Note that the craters whose size is comparable to or greater than the diameter of the asteroid are reported only to show the number of highly energetic impacts, since Eq. 35 is not appropriate for such cases. As can be easily seen from the plots and is summarised in Table 5, the collisional history of Vesta in such a disk of primordial planetesimals would result in the destruction of the asteroid. cording to our simulations, the most favourable scenarios are those where the giant planet migrated by less than 0.5 AU (see Tables 3 and 5). In those scenarios where the asteroid survived to the Jovian early bombardment, our results indicate that Vesta under- went to an intense collisional evolution and that its surface was saturated by craters spanning up to 150 km, with a few craters of ≈ 200 km. The total amount of material excavated from Vesta is 10− 20% of its present volume even in the less collisionally active scenarios we considered. As for Ceres, we cannot rely only on these primordial impact features to date the surface of Vesta, since they were altered or possibly erased by the subsequent 4 Ga of collisional evo- lution of the Main Asteroid Belt and the passage of Vesta through the Late Heavy Bombardment. However, the case of Vesta is different from that of Ceres: thanks to its connec- tion to HED meteorites, we know that Vesta differentiated and formed its basaltic crust in less than 4 Ma since the formation of CAIs (Keil 2002; Scott 2007). Such early epoch and short timescale indicate that short-lived radionuclides like 26Al and 60Fe were the drivers of such differentiation. Preliminary modeling of the thermal evolution of Vesta in- dicates that its mantle would have been in a molten state for several Ma (Federico, Coradini & Pauselli, in preparation, but see also Keil (2002) and references therein), i.e. across the timespan of the Jovian early bombardment. Moreover, thermal models (ibid) indicate that the thickness of the crust of Vesta would have varied between a few km to about 10−20 km in the temporal frame of interest, also depending on the amount of 26Al originally available. Our results (see Tables 3 and 5) indicate that the Jovian early bombardment should have excavated partially or com- pletely the primordial crust of Vesta, thus creating fractures or generating uncompensated negative gravity anomalies. These would have caused effusive phenomena from the un- derlying mantle, in analogy with lunar maria, or the solidifi- cation of the exposed layer of the mantle and the formation of a new basaltic crust. The crystallisation epoch of these regions on the surface of Vesta would then be directly con- nected to the time of formation of Jupiter. Dating the crust of Vesta or the possible Vestian maria once Dawn mission will visit the asteroid would then supply the opportunity to date the Jovian early bombardment and to constrain with unprecedented accuracy the formation of Jupiter. Collisional histories of early Vesta and Ceres 29 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 40 35 30 25 20 15 10 5 0 0 500 1000 1500 2000 0 500 1000 1500 2000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) CERES (0.50 AU MIGRATION) CERES (1.00 AU MIGRATION) ISL IMPACTORS BSL IMPACTORS ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 40 35 30 25 20 15 10 5 0 Y C N E U Q E R F Y C N E U Q E R F 40 35 30 25 20 15 10 5 0 40 35 30 25 20 15 10 5 0 0 500 1000 1500 2000 0 500 1000 1500 2000 CRATER DIAMETER (KM) CRATER DIAMETER (KM) Figure 22. Normalised frequency versus crater size histograms of the impacts on Ceres computed using the diameter-heliocentric distance relationship for planetesimals from Fig. 14 in Chambers (2010): red bars are those related to bodies formed in the region of ISL impactors, blue ones to bodies formed in the region of BSL impactors. Note that the craters whose size is comparable to or greater than the diameter of the asteroid are reported only to show the number of highly energetic impacts, since Eq. 35 is not appropriate for such cases. As we pointed out in Fig. 21 for Vesta and is summarised in Table 6, the collisional history of Ceres in such a disk of primordial planetesimals would result in the destruction of the asteroid. 5 CONCLUSION In this work we explored the collisional evolution of Vesta and Ceres at the time of Jupiter's formation, one of the milestones in the history of the early Solar System. The gravitational perturbations of the giant planet, in fact, excite the orbital resonances both inside and outside the Snow Line and trigger an early, intense bombardment in the orbital region of the Main Asteroid Belt. If Vesta or Ceres formed before Jupiter's core reached its critical mass, they would have been subject to such Jovian early bombardment. In this first investigation we did not include the presence of Saturn, which is equivalent to assuming that its formation was delayed by several 105 years respect to that of Jupiter, and we ignored the perturbing effects of gas drag and of possible planetary embryos embedded in the Solar Nebula on the dynamical evolution of the planetesimals. The survival of Vesta and Ceres to this primordial bombard- ment depends on the characteristics of the planetesimals populating the Solar Nebula and specifically on their size distribution. Our results clearly indicate that the abundance of large planetesimals in the disk (and particularly in the region of the Main Asteroid Belt) is a critical factor for the survival of the two asteroids. If the disk of planetesimals was dominated by large bodies (i.e. D > 100 km), like in the case of planetesimals formed in turbulent circumstellar disks (Morbidelli et al. 2009; Chambers 2010), the two asteroids would not have survived to the Jupiter-induced bombardment. Conversely, disks dominated by small plan- etesimals (i.e. D 6 20 km), like those formed in quiescent circumstellar disks (Coradini, Federico & Magni 1981) or produced by collisional evolution of larger bodies (Mor- bidelli et al. 2009), represent more favourable environments for what it concerns the survival of Vesta and Ceres. The early cratering histories of the two asteroids are extremely sensitive also to the radial migration of Jupiter due to disk-planet interactions during the phase of accretion of its gaseous envelope. According to our results, displace- ments of Jupiter of the order of 0.5 AU or more would cause either an unlikely high collisional erosion of the primordial Vesta and Ceres or their destruction. Depending on the abundance of large planetesimals (D > 100 km) in the disk, moreover, the survival of Ceres could imply that Jupiter migrated inward by less than 0.25 AU. These constrains 30 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 600 500 400 300 200 100 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) VESTA (0.50 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 600 500 400 300 200 100 Y C N E U Q E R F Y C N E U Q E R F PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 600 500 400 300 200 100 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) VESTA (1.00 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 600 500 400 300 200 100 Y C N E U Q E R F Y C N E U Q E R F 0 -10 -8 -6 -4 -2 0 2 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) LOG(Q/Q) Figure 23. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Vesta computed using the primordial (red bars) and collisionally evolved (blue bars) size-frequency distributions in the inner Solar System from Fig. 8a in Morbidelli et al. (2009). The frequency of the collisionally evolved impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. As can be seen in the histograms referring to the primordial SFD by Morbidelli et al. (2009), highly energetic impacts (Q > 10−2Q∗D) are abundant in all scenarios where Jupiter migrates. to the extent of the radial displacement of Jupiter are in agreement with the findings of other authors (see e.g. Scott (2006) and references therein). Finally, in all scenarios where they survived to the Jovian bombardment, Vesta and Ceres underwent to an intense cratering that saturated their surfaces with craters as big as 150 km, with a tail of few bigger craters (200 − 300 km). Craters as big as the south pole impact basin on Vesta (D ≈ 400 km) are characterised in our simulations by extremely low probabilities (of the order of a few per cent). Under the simplifying assumption of a uniform distribution of the craters produced in the simple cratering regime described by Eq. 35, our results indicate that the Jovian early bombardment would have excavated a depth of about 10 − 15 km on Vesta and of 20 − 30 km on Ceres. While it is reasonable to assume that planetesimals as big as Ceres were already present in the Main Asteroid Belt at the time Jupiter formed, we have little information on the real timescale of the accretion of Ceres. Therefore, we do not know if Ceres already completed its formation at the time of the events we simulated. Moreover, we ignore the details of the thermal evolution of the asteroid. As a consequence, we cannot rule out that, if the asteroid differentiated, its thermal evolution reshaped both the internal structure and the surface morphology, obliterating the traces of such ancient times as that covered in our simulations. On the contrary, through its connection with HED me- teorites we know that Vesta differentiated and formed a basaltic crust in the same temporal frame as that of Jupiter's formation. Our results suggest that the Jovian early bombardment was intense enough to excavate the Vestian crust and expose, locally or possibly even globally, the underlying mantle. Effusive phenomena would produce basaltic regions similar to lunar maria, whose epoch of crystallisation would then be directly related to that of Jupiter's formation. In all non-destructive cases resulting from our simulations, the crustal erosion would link large regions on the surface of the asteroid to the time of the Jovian early bombardment. ACKNOWLEDGEMENTS The authors wish to thank the reviewer, John Chambers, for his comments and suggestions that helped to improve Collisional histories of early Vesta and Ceres 31 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 1000 800 600 400 200 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) CERES (0.50 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 1000 800 600 400 200 Y C N E U Q E R F Y C N E U Q E R F PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 1000 800 600 400 200 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) CERES (1.00 AU MIGRATION) PRIMORDIAL IMPACTORS EVOLVED IMPACTORS (x10-2) 1000 800 600 400 200 Y C N E U Q E R F Y C N E U Q E R F 0 -10 -8 -6 -4 -2 0 2 0 -10 -8 -6 -4 -2 0 2 LOG(Q/Q) LOG(Q/Q) Figure 24. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Ceres computed using the primordial (red bars) and collisionally evolved (blue bars) size-frequency distributions in the inner Solar System from Fig. 8a in Morbidelli et al. (2009). The frequency of collisionally evolved impactors in all plots has been divided by a factor 100 to enhance the readability of the plots. When considering the primordial SFD by Morbidelli et al. (2009), highly energetic impacts (Q > 10−2Q∗D) are abundant in all scenarios where Jupiter migrates as in the case of Vesta (see Fig. 23). the manuscript and the quality of the results. D.T. also wish to thank Federico Tosi for his useful comments on the manuscript. This research has been supported by the Italian Space Agency (ASI) through the ASI-INAF contract I/015/07/0. The computational resources used in this re- search have been supplied by IFSI-Rome through the project "HPP - High Performance Planetology". References Amelin Y., Krott A. N., Hutcheon I. D., Ulyanov A. A., "Lead Isotopic Ages of Chondrules and Calcium- Aluminum-Rich Inclusions", 2002, Science, 297, 1678- 1683 Asphaug E., "Growth and Evolution of Asteroids", 2009, Annual Reviews of Earth and Planetary Sciences, 37, 413- 448 Baker J. A., Bizzarro M., Wittig N., Connelly J. N., Haack H., "Early planetesimal melting from an age of 4.5662 Gyr for differentiated meteorites", 2005, Nature, 436, 1127- 1131 Benz W., Asphaug E., "Catastrophic Disruptions Revis- ited", 1999, Icarus, 142, 5-20 Bizzarro M., Baker J. A., Haack H., Luundgard K. L., "Rapid timescales for accretion and melting of differen- tiated planetesimals inferred from 26Al-26Mg chronome- try", 2005, The Astrophysical Journal, 632, L41-L44 Bottke W. F., Durda D. D., Nesvorny D., Jedicke R., Mordibelli A., Vokrouhlicky D., Levison H., "The fos- silized size distribution of the main asteroid belt", 2005, Icarus, 175, 111-140 Bottke W. F., Durda D. D., Nesvorny D., Jedicke R., Mordibelli A., Vokrouhlicky D., Levison H., "Linking the collisional history of the main asteroid belt to its dynam- ical excitation and depletion", 2005, Icarus, 179, 63-94 Chambers J. E., "Planetesimal formation by turbulent con- centration", 2010, Icarus, 208, 505-517 Chambers J. E., Wetherill G. W., "Planets in the asteroid belt", 2001, Meteoritics and Planetary Science, 36, 381- 399 Connelly J. N., Amelin Y., Krot A. N., Bizzarro M., "Chronology of the Solar System's oldest solids", 2008, The Astrophysical Journal, 675, L121-L124 Coradini A., Federico C., Magni G., "Formation of Plan- 32 Turrini et al. VESTA (NO MIGRATION) VESTA (0.25 AU MIGRATION) Y C N E U Q E R F Y C N E U Q E R F 140 120 100 80 60 40 20 0 ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 140 120 100 80 60 40 20 0 ISL IMPACTORS BSL IMPACTORS -10 -8 -6 -4 -2 0 2 4 -10 -8 -6 -4 -2 0 2 4 LOG(Q/Q) LOG(Q/Q) VESTA (0.50 AU MIGRATION) VESTA (1.00 AU MIGRATION) 140 120 100 80 60 40 20 0 ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 140 120 100 80 60 40 20 0 ISL IMPACTORS BSL IMPACTORS -10 -8 -6 -4 -2 0 2 4 -10 -8 -6 -4 -2 0 2 4 LOG(Q/Q) LOG(Q/Q) Figure 25. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Vesta computed using the diameter-heliocentric distance relationship for planetesimals from Fig. 14 in Chambers (2010). Red bars are those related to bodies formed in the region of our original ISL impactors (2 − 4 AU), blue ones to bodies formed in the region of our original BSL impactors (4 − 10 AU). Highly energetic impacts (Q > 10−2Q∗D) are present in all scenarios and are particularly abundant when Jupiter migrates by 0.5 AU or more. etesimals in an evolving Protoplanetary Disk", 1981, As- tronomy and Astrophysics, 98, 173-185 Coradini A., Magni G., Turrini D., "From Gas to Satel- litesimals: Disk Formation and Evolution", 2010, Space Science Reviews, DOI: 10.1007/s11214-009-9611-9 Cuzzi J. N., Hogan R. C., Shariff K., "Toward planetesi- mals: Dense chondrule clumps in the protoplanetary neb- ula.", 2008, The Astophysical Journal, 687, 14321447 D'Angelo G., Kley W., Henning T., "Orbital Migration and Mass Accretion of Protoplanets in Three-dimensional Global Computations with Nested Grids", 2003, The As- trophysical Journal, 586, 540-561 De Pater I., Lissauer J.J., "Planetary Sciences", 2001, Cambridge (UK), Cambridge University Press, ISBN 0521482194 Drake M. J., "The eucrite/Vesta story", 2001, Meteoritics and Planetary Science, 36, 501-513 Encrenaz T., "Water in the Solar System", 2008, Annual Review of Astronomy and Astrophysics, 46, 5787 Goldreich P., Ward W. R., "The formation of planetesi- mals", 1973, The Astrophysical Journal, 183, 10511062 of the terrestrial planets", Nature, 2005, 435,466-469 Greenberg R., Carusi A., Valsecchi G. B., "Outcomes of Planetary Close Encounters: A Systematic Comparisons of Methodologies", 1988, Icarus, 75, 1-29 Haghighipour N., Scott E. R. D., "Meteorite constraints on the accretion of planetesimals and protoplanets in the inner Solar System", 2009, 72nd Annual Meteoritical So- ciety Meeting, abstract n. 5429 Haisch K. E., Lada E. A., Lada C. J., "Disk frequencies and lifetimes in young clusters", 2001, The Astrophysical Journal, 553, L153-L156 Ipatov S. I., A'Hearn M. F., "Velocities and relative amount of material ejected from Comet 9P/Tempel 1 after the Deep Impact collision", 2008, arXiv:0810.1294v3 Johansen A., Oishi J. S., Mac Low M.-M., Klahr H., Hen- ning T., Youdin A., "Rapid planetesimal formation in tur- bulent circumstellar disks.", 2007, Nature, 448, 10221025 Keil K., "Geological History of Asteroid 4 Vesta: The Smallest Terrestrial Planet", 2002, Asteroids III, Eds. W. F. Bottke Jr., A. Cellino, P. Paolicchi, and R. P. Binzel, University of Arizona Press, Tucson, 573-584 Gomes R., Levison H.F., Tsiganis K., Morbidelli A., "Ori- gin of the cataclysmic Late Heavy Bombardment period Lissauer J. J., Hubickyi O., D'Angelo G., Bodenheimer P., "Models of Jupiter's growth incorporating thermal and Collisional histories of early Vesta and Ceres 33 CERES (NO MIGRATION) CERES (0.25 AU MIGRATION) 300 250 200 150 100 50 0 ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 300 250 200 150 100 50 0 ISL IMPACTORS BSL IMPACTORS -10 -8 -6 -4 -2 0 2 4 -10 -8 -6 -4 -2 0 2 4 LOG(Q/Q) LOG(Q/Q) CERES (0.50 AU MIGRATION) CERES (1.00 AU MIGRATION) 300 250 200 150 100 50 0 ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F 300 250 200 150 100 50 0 ISL IMPACTORS BSL IMPACTORS Y C N E U Q E R F Y C N E U Q E R F -10 -8 -6 -4 -2 0 2 4 -10 -8 -6 -4 -2 0 2 4 LOG(Q/Q) LOG(Q/Q) Figure 26. Normalised frequency versus impact energy (in units of the dispersal energy Q∗D) histograms of the impacts on Ceres computed using the diameter-heliocentric distance relationship for planetesimals from Fig. 14 in Chambers (2010). Red bars are those related to bodies formed in the region of our original ISL impactors (2 − 4 AU), blue ones to bodies formed in the region of our original BSL impactors (4 − 10 AU). Similarly to the case of Vesta in Fig. 25, highly energetic impacts (Q > 10−2Q∗D) are present in all scenarios and are particularly abundant when Jupiter migrates by 0.5 AU or more. hydrodynamics constraints", 2009, Icarus, 199, 338-350 Magni G., Coradini A., "Formation of Jupiter by nucleated instability", 2004, Planetary and Space Science, 52, 343- 360 Meech K. J., et al., "Deep Impact: Observations from a Worldwide Earth-Based Campaign", 2005, Science, 310, 265-269 Meyer M. R., "Circumstellar Disk Evolution: Constrain- ing Theories of Planet Formation", 2008, Proceedings of the International Astronomical Union, 4, 111-122, DOI: 10.1017/S1743921309031767 Michalak G., "Determination of asteroid masses. I. (1) Ceres, (2) Pallas and (4) Vesta", 2000, Astronomy & As- trophysics, 360, 363-374 Minton D. A., Malhotra R., "A record of planet migration in the main asteroid belt", 2009, Nature, 457, 1109-1111 Morbidelli A., Levison H.F., Tsiganis K., Gomes R., "Chaotic capture of Jupiter's Trojan asteroids in the early Solar System", Nature, 2005, 435, 462-465 Morbidelli A., Bottke W. F., Nesvorny D., Levison H. F., "Asteroids were born big", 2009, Icarus, 204, 558-573 Nagasawa M., Thommes E. W., Kenyon S. J., Bromley B. C., Lin D. N. C., "The Diverse Origins of Terrestrial- Planet Systems", 2007, Protostars and Planets V, Eds. B. Reipurth, D. Jewitt, and K. Keil, University of Arizona Press, Tucson, 639-654 O'Brien D. P., Morbidelli A., Bottke W. F., "The primor- dial excitation and clearing of the asteroid belt Revis- ited", 2007, Icarus, 191, 434-452 Opik E. J., "Interplanetary encounters", 1976, Eds. Kopal Z. & Cameron A. G. W., Elsevier, ISBN 0-444-41371-5 Papaloizou J. C. B., Nelson R. P., Kley W., Masset F. S., Artymowicz P., "Disk-Planet Interactions During Planet Formation", 2007, Protostars and Planets V, Eds. B. Reipurth, D. Jewitt, and K. Keil, University of Arizona Press, Tucson, 655-668 Petit J., Morbidelli A., Chambers J., "The Primordial Ex- citation and Clearing of the Asteroid Belt", 2001, Icarus, 153, 338-347 Pierazzo E., Melosh H.J., "Understanding Oblique Impacts from Experiments, Observations, and Modeling", 2000, Annual Review of Earth and Planetary Sciences, 28, 141- 167 Safronov V. S., "Evolution of the Protoplanetary Cloud and Formation of the Earth and Planets", 1969, Nauka Press, Moscow, English Translation: NASA TTF-677.7 34 Turrini et al. Scott E. R. D., "Meteorite parent bodies: what can they tell us about the formation and evolution of the asteroid belt?", 2008, 71st Annual Meteoritical Society Meeting, abstract n. 5284 Scott E. R. D., "Chondrites and the Protoplanetary Disk", 2007, Annual Reviews of Earth and Planetary Sciences, 35, 577-620 Scott E. R. D., "Meteoritics and dynamical constrains on the growth mechanisms and formation times of asteroids and Jupiter", 2006, Icarus, 185, 72-82 Thomas P. C., Binzel R. P., Gaffey M. J., Zellner B. H., Storrs A. D., Wells E., "Vesta: Spin Pole, Size, and Shape from HST Images", 1997, Icarus, 128, 88-94 Thomas P. C., Parker J. W., McFadden L. A., Russell C. T., Stern S. A., Sykes M. V., Young E. F., "Differentia- tion of the asteroid Ceres as revealed by its shape", 2005, Nature, 437, 224-226 Tsiganis K., Gomes R., Morbidelli A., Levison H.F., "Ori- gin of the orbital architecture of the giant planets of the Solar System", Nature, 2005, 435,459-461 Yang J., Goldstein J. I., Scott E. D. R., "Iron meteorite evidence for early formation and catastrophic disruption of protoplanets", 2007, Nature, 446, 888-891 Weidenschilling S. J., "Aerodynamics of solid bodies in the Solar Nebula", 1977, Monthly Notices of the Royal Astro- nomical Society, 180, 57-70 Weiler M., Rauer H., Knollenberg J., Sterken C., "The gas production of Comet 9P/Tempel 1 around the Deep Im- pact date", 2007, Icarus, 190, 423-431 Wetherill G. W., "An alternative model for the formation of asteroids", 1992, Icarus, 100, 307-325 This paper has been typeset from a TEX/ LATEX file prepared by the author.
1012.1992	1	1012	2010-12-09T12:55:16	Detection limits for close eclipsing and transiting sub-stellar and planetary companions to white dwarfs in the WASP survey	[ "astro-ph.EP", "astro-ph.SR" ]	We used photometric data from the WASP (Wide-Angle Search for Planets) survey to explore the possibility of detecting eclipses and transit signals of brown dwarfs, gas giants and terrestrial companions in close orbit around white dwarfs. We performed extensive Monte Carlo simulations and we found that for Gaussian random noise WASP is sensitive to companions as small as the Moon orbiting a $V\sim$12 white dwarf. For fainter stars WASP is sensitive to increasingly larger bodies. Our sensitivity drops in the presence of co-variant noise structure in the data, nevertheless Earth-size bodies remain readily detectable in relatively low S/N data. We searched for eclipses and transit signals in a sample of 194 white dwarfs in the WASP archive however, no evidence for companions was found. We used our results to place tentative upper limits to the frequency of such systems. While we can only place weak limits on the likely frequency of Earth-sized or smaller companions; brown dwarfs and gas giants (radius$\simeq$ R$_{jup}$) with periods $\leq$0.2 days must certainly be rare ($<10\%$). More stringent constraints requires significantly larger white dwarf samples, higher observing cadence and continuous coverage. The short duration of eclipses and transits of white dwarfs compared to the cadence of WASP observations appears to be one of the main factors limiting the detection rate in a survey optimised for planetary transits of main sequence stars.	astro-ph.EP	astro-ph	Detection limits for close eclipsing and transiting sub-stellar and planetary companions to white dwarfs in the WASP survey F. Faedi∗,†, R. G. West†, M. R. Burleigh†, M. R. Goad† and L. Hebb∗∗,‡ ∗Astrophysics Research Centre, School of Mathematics and Physics, Queens University, Belfast, BT7 1NN, U.K. †Department of Physics and Astronomy, University of Leicester, Leicester, LE1 7RH, U.K. ∗∗Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, U.S.A. ‡School of Physics and Astronomy, University of St. Andrews, North Haugh, Fife, KY16 9SS, U.K. Abstract. We used photometric data from the WASP (Wide-Angle Search for Planets) survey to explore the possibility of detecting eclipses and transit signals of brown dwarfs, gas giants and terrestrial companions in close orbit around white dwarfs. We performed extensive Monte Carlo simulations and we found that for Gaussian random noise WASP is sensitive to companions as small as the Moon orbiting a V ∼12 white dwarf. For fainter stars WASP is sensitive to increasingly larger bodies. Our sensitivity drops in the presence of co-variant noise structure in the data, nevertheless Earth-size bodies remain readily detectable in relatively low S/N data. We searched for eclipses and transit signals in a sample of 194 white dwarfs in the WASP archive however, no evidence for companions was found. We used our results to place tentative upper limits to the frequency of such systems. While we can only place weak limits on the likely frequency of Earth-sized or smaller companions; brown dwarfs and gas giants (radius≃ R jup) with periods ≤0.2 days must certainly be rare (< 10%). More stringent constraints requires significantly larger white dwarf samples, higher observing cadence and continuous coverage. The short duration of eclipses and transits of white dwarfs compared to the cadence of WASP observations appears to be one of the main factors limiting the detection rate in a survey optimised for planetary transits of main sequence stars. Keywords: methods: data analysis stars: white dwarfs stars: planetary systems occultations PACS: 97.82.Cp, 97.82.Fs INTRODUCTION The transit technique involves searching for periodic dips in stellar light-curves due to the orbital revolution of a transiting body, blocking a fraction of the stellar light. For a given planetary radius, the transit depth (d ) is proportional to (Rpl/R∗)2. Therefore, planets orbiting solartype stars have extremely shallow eclipses, blocking ∼1% of the light for a giant planet and ∼0.01% of the light for an Earth-sized planet. Current ground-based wide-field surveys can achieve the necessary photometric accuracy of better than 1%, only for the brightest stars (V ∼9-12 in the case of WASP), so the bulk of the planets discovered by transit surveys around main-sequence stars have radii in the range Rpl ∼0.8−1.8 R jup. A strong advantage over main sequence star primaries is offered by white dwarf stars. White dwarfs (WD) are compact degenerate objects, with approximately the same radius as the Earth, and represent the final stage of evolution of main-sequence stars with masses < 8M⊙ (i.e. ∼97% of all stars in our Galaxy). Any sub-stellar or gas giant companion in orbit around a white dwarf will completely eclipse it, while bodies as small as the Moon will have relatively large transit depths (∼ 3%), with the only caveat being that it remains unclear as to whether any such systems survive beyond the latter stages of stellar evolution. Sub-stellar companions to white dwarfs are rare (Farihi et al. 2005 using 2MASS estimated that < 0.5% of WDs have L dwarf companions). At the time of writing only three wide white dwarf + brown dwarf (WD+BD) systems have been spectroscopically confirmed, GD 165 (Becklin and Zuckerman 1988), PHL5038 (Steele et al. 2009), and LSPM 1459 + 0857 AB (Day-Jones et al. 2010) and two detached, non-eclipsing, short-period WD+BD systems are currently known, WD0137 −349 (Maxted et al. 2006, Burleigh et al. 2006, P≈116mins), and GD1400 (Farihi and Christopher 2004, Dobbie et al. 2005, Burleigh et al. 2010, P≈9.9h). The latter, is currently the lowest mass (∼50M jup) object known to have survived CE evolution. Although infrared surveys such as UKIDSS, VISTA and WISE, and observatories such as Spitzer hope to reveal many more such binaries, they remain difficult to identify either as infra-red excesses or through radial velocity measurements. In addition the detection of a significant number of eclipsing WD+BD binary systems might help uncover the hypothesised population of 'old' cataclysmic variables (CVs) in which the companion has been reduced to a sub-stellar mass (e.g. Patterson 1998; Patterson et al. 2005; Littlefair et al. 2003). These systems are undetectable as X-ray sources and difficult to identify in optical and infra-red surveys. Littlefair et al. (2006) confirmed the first such system through eclipse measurements, while Littlefair et al. (2007) showed that another eclipsing CV, SDSS J150722.30 + 523039.8, was formed directly from a detached WD+BD binary. Several theoretical studies discuss post-main sequence evolution of planetary systems and show that planetary survival is not beyond possibility (Duncan and Lissauer 1998; Debes and Sigurdsson 2002; Burleigh et al. 2002; and Villaver and Livio 2007). Radial velocity observations of red giants indicate that planets in orbits beyond the red giant's envelope can survive stellar evolution to that stage (see Frink et al. 2002; Hatzes et al. 2005, Sato et al. 2003). Moreover, Silvotti et al. (2007) reported the detection of a ∼3M jup planet orbiting an extreme horizontal branch star, and Mullally et al. (2008) found convincing evidence of a 2M jup planet in a 4.5 year orbit around a pulsating WD. Furthermore, Beuermann et al. (2010) reported the detection of two planetary companions (Mc=6.9MJup and Md=2.2MJup) in the post common envelope binary NN Ser (ab) via measurements of a light-travel-time effect superposed on the linear ephemeris of the binary; showing that planets do survive stellar evolution. Short-period rocky companions to white dwarfs may seem less likely. Villaver and Livio (2007) suggested that planets in orbit within the reach of the AGB envelope will either evaporate or in rare cases, more massive bodies may accrete mass and become close companions to the star. Planets in wide orbits that escape engulfment by the red giant or asymptotic giant will move outwards to conserve angular momentum (as described by Jeans 1924). Duncan and Lissauer (1998) found that for WD progenitors experiencing substantial mass loss during the AGB phase, planetary orbits become unstable on timescales of ≤108 year. Debes and Sigurdsson (2002) found that the mass loss is sufficient to destabilise planetary systems of two or more planets and that the most likely result is that one planet would be scattered into an inner orbit (occupied, before the RGB phase, by a 'now evaporated' inner planet), while the other would either be boosted into a larger orbit, or ejected from the system altogether. The above scenario provides a plausible explanation for the recent detection of silicate- rich dust discs around a growing number of white dwarfs at orbital radii up to ∼1R⊙ (e.g. Reach et al. 2005; Farihi et al. 2007, 2008; Jura 2003). Jura (2003) suggests that the formation of dust discs around white dwarfs is most probably due to the tidal disruption of an asteroid or larger body which has strayed too close to the parent star. (Jura et al. 2009) suggest that the disc around GD362 originated from the tidal destruction of a single massive body such as Callisto or Mars. The detection of short period sub-stellar and planetary mass companions to white dwarfs, will open an exciting chapter in the study of exoplanet evolution, constraining theoretical models of CE evolution and helping us to understand the ultimate fate of hot Jupiter systems as well as the fate of our own solar system in the post main-sequence phase. Here we present some of the results of our study which investigated the detection limits for transiting sub-stellar and terrestrial companions in close orbits around white dwarfs (for more details see Faedi et al. 2010). In §2 we discuss the characteristics of the transit signals, the parameter space investi- gated and our detection method. In §3 we analysed a sample of 194 WDs in the WASP archive. Finally in §4 we discuss our conclusions. CHARACTERISTICS OF THE TRANSIT SIGNAL A transit signal is described by its duration, its depth and its shape. Extra-solar planets transiting main-sequence stars show signals characterised by an ingress, a flat bottom and an egress, with a typical durations of 2-3 hours and depths of about 1% (see for example Collier Cameron et al. 2007; Simpson et al. 2010; Barros et al. 2010; and Faedi et al. 2010). We modelled the synthetic dataset assuming circular orbits and fixed stellar parameters. We considered a typical 1 Gyr old carbon-core white dwarf of mass M∗ = 0.6M⊙ and radius R∗ = 0.013R⊙. We explored the detectability of planetary transits across the two-dimensional parameter space defined by the orbital period and the planet radius. Our simulations cover companions ∼ 0.3R⊕ < Rpl < 12R⊕, and orbital periods in the range P ∼ 2 hours to 15 days (equivalent to orbital distances between a ∼ 0.003 and 0.1 AU). We chose the minimum orbital period to yield an orbital separation close to the Roche radius of the WD, and the maximum period in order to have a reasonable chance of detecting five or more transits in a typical WASP season of 150 day. In Figure 1 we show the probability that a given system will transit, and the duration of such transit across the parameter space defined above. It is evident from these diagrams that the signatures of transits of white dwarfs by typical planet-sized bodies will be rather different than those seen for typical transiting hot Jupiters. In particular the transit FIGURE 1. Contours of constant transit probability (left), and duration (right) in the parameter space defined by orbital period and planetary radius. The transit probability is expressed in percentage values. The transit duration is expressed in minutes. duration is much shorter, from ∼1-30 min for companions with sizes ranging from Moon-size to Jupiter-size, compared to 2-3 hours for a typical hot Jupiter. In addition, Figure 2, left-panel, shows that the transit depths are much larger, from around 3% for a Moon-sized to 100% for any companion larger than the Earth, compared to ∼1% for a hot Jupiter. Synthetic WASP light-curves The synthetic light-curves were generated using the time sampling of a typical WASP survey field, and with statistical S/N representative of magnitude spanning the range of brightness of WDs in the WASP survey. The corresponding photometric accuracy of WASP over this range is ∼1% to 10%. Because WASP data show residual covariant- noise structure we have tested the transit recovery rate in the case of both uncorrelated "white" noise and correlated "red" noise. To cover the orbital period-planet radius parameter space we selected seven trial periods spaced approximately logarithmically (P = 0.08, 0.22, 0.87, 1.56, 3.57, 8.30 and 14.72 days), and five planet radii Rpl = 10.0, 1.0, 0.6, 0.34 and 0.27 R⊕. We modelled the set of synthetic light-curves by injecting fake transit signals into phase-folded light-curves at each trial period with a random transit epoch t0 in the range 0 < t0 < P. Because in the case of a WD host star con- sidered here, the ingress and egress duration is typically short compared to cadence of the WASP survey (8-10 minutes), we ignored the detailed shape of the ingress and egress phases and modelled the transit signatures as simple box-like profiles. Figure 2, right-panel shows two examples of our simulated transit light-curves. The top panel shows the synthetic light-curve of an hypothetical eclipsing WD+BD binary system with an orbital period of P = 116 mins, similar to WD0137 − 349 (a non-eclipsing system, Maxted et al. 2006). The lower panel shows the simulated transit light-curve for a rocky body of radius 1.2R⊕ in a 5 hr orbit. FIGURE 2. Left-panel: contours of constant transit depth; right-panel: examples of synthetic light- curves. Top, an eclipsing BD in orbit with 2 hr period. Bottom, a 1.2R⊕ companion in 5 hr orbit. Detecting transit signals To recover the transit signals from the synthetic light-curves we used an implemen- tation of the box-least-squares (BLS) algorithm (Kovács et al. 2002) commonly used to detect transits of main sequence stars. To ensure that the BLS search was sensitive across the expected range of transit durations, we chose to search a grid of box widths Wb = {1, 2, 4, 8, 16, 32} minutes, covering the range in transit durations over most of our parameter space (Figure 1). In addition, we used an optimised version of the BLS code which best accounted for the shape, duration and depth of the signals investigated in this work (Faedi et al. 2010). We used the Signal Detection Efficiency (SDE) metric defined in Kovács et al. (2002) to assess the likely significance of a peak in a BLS periodogram. We evaluated SDE as follows: where Speak is the height of the peak, and ¯S and s S are measures of the mean level and scatter in the noise continuum of the periodogram. A detection is represented by the highest peak in the BLS power spectrum. We regard as a match any trial in which the most significant detected period is within 1% of being an integer fraction or multiple from 1/5× to 5× the injected transit signal. Details of the algorithm false alarm probability can be found in Faedi et al. (2010). The results of our simulations are illustrated in Figure 2, left-panel in the case of a V ∼12 magnitude WD for light- curves with red noise. It is evident from Figure 2; Figure 6 and Table 1, 2, and 3 from Faedi et al. (2010), that transiting companions are essentially undetectable at our longest trial periods (8.30 and 14.72 days) in a WASP-like survey; the transits are too short in duration and too infrequent to be adequately sampled. In addition, we found that for SDE = Speak − ¯S s S FIGURE 3. Left-panel: recovery rate for simulated transit signals injected into synthetic light-curves of a white dwarf of magnitude V ≃ 12; right-panel: upper-limits on companion frequency (95%) folding- in the detectability of transiting systems in a WASP-like survey. In both panels values are expressed in percent. idealised photon-noise-limited cases, objects as small as Mercury could be detected to periods of around 1.5 d, and the Moon for periods less than 1 d. Once red noise is added, Moon-sized companions become almost undetectable. However, for companions around 1R⊕ and larger there is a good chance of detection out to periods of around 4 days. Our key conclusion from these simulations is that for the case of transits of white dwarfs the degree of photometric precision delivered by a survey is of somewhat secondary importance compared to a high cadence and continuous coverage. For planet-sized bodies individual transits will be quite deep and readily detectable in data of moderate photometric quality, however it is the short duration of the transits that is the main factor limiting the transit detection rate in surveys optimised for main sequence stars. SEARCHING FOR TRANSIT SIGNALS IN WASP SURVEY DATA Encouraged by the results of our simulations we selected a sample of 194 WDs (with V < 15) which have been routinely monitored by WASP through the 2004 to 2008 observing seasons, and performed a systematic search for eclipsing and transiting sub- stellar and planetary companions. We selected the sample by cross-correlating the WASP archive with the McCook & Sion catalogue (McCook and Sion 2003). In addition to our authomated search, we have inspected each of the individual light-curves by eye. In both searches we found no evidence for any transiting and eclipsing companions. We have used this null result together with the results of simulations to estimate an upper- limit to the frequency of such close companions for the sample of WDs considered in this study. In order to estimate an upper limit to the frequency of close sub-stellar and planetary companions to white dwarfs, we used the detection limits derived from our simulations and the results obtained from the analysis of the sample of 194 white dwarfs. Although our complete sample numbers N = 194 stars, only a fraction ptr(Rpl, P) will exhibit a transit, and of those only a fraction pdet (Rpl, P) would be detectable in a WASP- like survey. Both these factors act to reduce the total number of transiting companions detected in the survey, or in the case of a null result will tend to weaken the constraints that can be placed on true companion frequency by such a survey. We incorporate these factors and we modified our effective sample size as N′ = N × ptr(Rpl, P) × pdet (Rpl, P). We combined the magnitude-specific pdet maps obtained from our simulations for WDs of magnitude V ∼ 12, 13, 15 into a single map by interpolating/extrapolating according to the magnitude of each object in our sample and combining these to form an averaged map which can be folded in to our calculation of the upper-limits. The resulting limits corresponding to the 95% of the integrated probability, are shown in the right panel of Figure 3. Our results show that for rocky bodies smaller than the size of Mercury no useful upper limits to the frequency of companions to white dwarfs can be found, and that for Earth-sized companions only weak constraints can be imposed. However, it does suggest that objects the size of BDs or gas giants with orbital periods P< 0.1 − 0.2 days must be relatively rare (upper limit of ∼10%). CONCLUSION We have investigated the detection limits for sub-stellar and planetary companions to white dwarfs using in the WASP survey. We found that Mercury-sized bodies at small or- bital radii can be detected with good photometric data even in the presence of red noise. For smaller bodies red noise in the light-curves becomes increasingly problematic, while for larger orbital periods, the absence of significant numbers of in-transit points, signifi- cantly decreases our detection sensitivity. Application of our modified BLS algorithm to search for companions to WDs in our sample of 194 stars available in the WASP archive, did not reveal any eclipsing or transiting sub-stellar or planetary companions. We have used our results, to place upper limits to the frequency of sub-stellar and planetary com- panions to WDs. While no useful limits can be placed on the frequency of Mercury-sized or smaller companions, slightly stronger constraints can be placed on the frequency of BDs and gas giants with periods < 0.1 − 0.2days, which must certainly be relatively rare (< 10%). More stringent constraints would requires significantly larger WD sam- ples. Our key conclusion from simulations and analysis, using WASP data, suggests that photometric precision is of secondary importance compared to a high cadence and con- tinuous coverage. The short duration of eclipses and transits of WDs compared to the WASP observing cadence, appears to be the main factor limiting the transit detection rate in a survey optimised for planetary transits of main sequence stars. Future surveys such as Pan-STARRS and LSST will be capable of detecting tens of thousands of WDs. However, we emphasise the importance of high cadence and long baseline observation when attempting to detect the signature of close, eclipsing and transiting sub-stellar and planetary companions to WDs. Space missions such as COROT, Kepler (see Di Stefano et al. 2010) and, especially, PLATO may therefore be better suited to a survey of white dwarfs as they deliver uninterrupted coverage at high cadence and exquisite photometric precision (∼ 10−4 − 10−5) and could at least in principle detect the transits of asteroid- sized bodies across a white dwarf. REFERENCES J. Farihi, E. E. Becklin, and B. Zuckerman, ApJ 161, 394 -- 428 (2005), arXiv:astro-ph/0506017. E. E. Becklin, and B. Zuckerman, Nature 336, 656 -- 658 (1988). P. R. Steele, M. R. Burleigh, J. Farihi, B. T. Gänsicke, R. F. Jameson, P. D. Dobbie, and M. A. Barstow, A&A 500, 1207 -- 1210 (2009), 0903.3219. A. Day-Jones, et al., MNRAS in press (2010). P. F. L. Maxted, R. Napiwotzki, P. D. Dobbie, and M. R. Burleigh, Nature 442, 543 -- 545 (2006), arXiv: astro-ph/0608054. M. R. Burleigh, E. Hogan, P. D. Dobbie, R. Napiwotzki, and P. F. L. Maxted, MNRAS 373, L55 (2006). J. Farihi, and M. Christopher, AJ 128, 1868 -- 1871 (2004), arXiv:astro-ph/0407036. P. D. Dobbie, M. R. Burleigh, A. J. Levan, M. A. Barstow, R. Napiwotzki, J. B. Holberg, I. Hubeny, and S. B. Howell, MNRAS 357, 1049 (2005). M. R. Burleigh, J. Farihi, R. Napiwotzki, T. R. Marsh, and P. D. Dobbie, MNRAS in prep. (2010). J. Patterson, PASP 110, 1132 -- 1147 (1998). J. Patterson, J. R. Thorstensen, and J. Kemp, PASP 117, 427 -- 444 (2005), arXiv:astro-ph/ S. P. Littlefair, V. S. Dhillon, and E. L. Martín, MNRAS 340, 264 -- 268 (2003), arXiv:astro-ph/ 0502392. 0211475. 0202194. (2002). 743 -- 751 (2005). S. P. Littlefair, V. S. Dhillon, T. R. Marsh, B. T. Gänsicke, J. Southworth, and C. A. Watson, Science 314, 1578 -- (2006), arXiv:astro-ph/0612220. S. P. Littlefair, V. S. Dhillon, T. R. Marsh, B. T. Gänsicke, I. Baraffe, and C. A. Watson, MNRAS 381, 827 -- 834 (2007), arXiv:0708.0097. M. J. Duncan, and J. J. Lissauer, Icarus 134, 303 -- 310 (1998). J. H. Debes, and S. Sigurdsson, ApJ 572, 556 -- 565 (2002), arXiv:astro-ph/0202273. M. R. Burleigh, F. J. Clarke, and S. T. Hodgkin, MNRAS 331, L41 -- L45 (2002), arXiv:astro-ph/ E. Villaver, and M. Livio, ApJ 661, 1192 -- 1201 (2007), arXiv:astro-ph/0702724. S. Frink, D. S. Mitchell, A. Quirrenbach, D. A. Fischer, G. W. Marcy, and R. P. Butler, ApJ 576, 478 -- 484 A. P. Hatzes, E. W. Guenther, M. Endl, W. D. Cochran, M. P. Döllinger, and A. Bedalov, A&A 437, B. Sato, et al., ApJ 597, L157 -- L160 (2003). R. Silvotti, et al., Nature 449, 189 -- 191 (2007). F. Mullally, D. E. Winget, S. Degennaro, E. Jeffery, S. E. Thompson, D. Chandler, and S. O. Kepler, ApJ 676, 573 -- 583 (2008), 0801.3104. K. Beuermann, F. V. Hessman, S. Dreizler, T. R. Marsh, S. G. Parsons, D. E. Winget, G. F. Miller, M. R. Schreiber, W. Kley, V. S. Dhillon, S. P. Littlefair, C. M. Copperwheat, and J. J. Hermes, A&A 521, L60+ (2010), 1010.3608. W. T. Reach, M. J. Kuchner, T. von Hippel, A. Burrows, F. Mullally, M. Kilic, and D. E. Winget, ApJ 635, L161 -- L164 (2005), arXiv:astro-ph/0511358. J. Farihi, B. Zuckerman, E. E. Becklin, and M. Jura, "Spitzer Observations of GD 362 and Other Metal- Rich White Dwarfs," in Astronomical Society of the Pacific Conference Series, edited by R. Napiwotzki, and M. R. Burleigh, 2007, vol. 372, p. 315. M. Jura, ApJ 584, L91 -- L94 (2003), arXiv:astro-ph/0301411. M. Jura, J. Farihi, and B. Zuckerman, AJ 137, 3191 -- 3197 (2009), 0811.1740. F. Faedi, R. G. West, M. R. Burleigh, M. R. Goad, and L. Hebb, MNRAS pp. 1810 -- + (2010), 1008.1089. A. Collier Cameron, et al., MNRAS 380, 1230 -- 1244 (2007), 0707.0417. E. K. Simpson, F. Faedi, S. C. C. Barros, et al., ArXiv e-prints (2010), 1008.3096. S. C. C. Barros, F. Faedi, et al., ArXiv e-prints (2010), 1010.0849. G. Kovács, S. Zucker, and T. Mazeh, A&A 391, 369 -- 377 (2002), arXiv:astro-ph/0206099. G. P. McCook, and E. M. Sion, VizieR Online Data Catalog 3235, 0 (2003). R. Di Stefano, S. B. Howell, and S. D. Kawaler, ApJ 712, 142 -- 146 (2010), 0912.3253.
1301.4453	3	1301	2013-04-29T08:43:54	Habitable Planets Eclipsing Brown Dwarfs: Strategies for Detection and Characterization	[ "astro-ph.EP", "astro-ph.IM" ]	Given the very close proximity of their habitable zones, brown dwarfs represent high-value targets in the search for nearby transiting habitable planets that may be suitable for follow-up occultation spectroscopy. In this paper we develop search strategies to find habitable planets transiting brown dwarfs depending on their maximum habitable orbital period (PHZ out). Habitable planets with PHZ out shorter than the useful duration of a night (e.g. 8-10 hrs) can be screened with 100 percent completeness from a single location and in a single night (near-IR). More luminous brown dwarfs require continuous monitoring for longer duration, e.g. from space or from a longitude-distributed network (one test scheduling achieved - 3 telescopes, 13.5 contiguous hours). Using a simulated survey of the 21 closest known brown dwarfs (within 7 pc) we find that the probability of detecting at least one transiting habitable planet is between 4.5 +5.6-1.4 and 56 +31-13 percent, depending on our assumptions. We calculate that brown dwarfs within 5-10 pc are characterizable for potential biosignatures with a 6.5 m space telescope using approx. 1 percent of a 5-year mission's life-time spread over a contiguous segment only one fifth to one tenth of this duration.	astro-ph.EP	astro-ph	HABITABLE EXOPLANETS AROUND BROWN DWARFS Accepted to The Astrophysical Journal v3 American Astronomical Society. HABITABLE PLANETS ECLIPSING BROWN DWARFS: STRATEGIES FOR DETECTION AND CHARACTERIZATION 5, 1, 2, ENRIC PALLÉ3,4, RACHEL STREET 1,2, SEAN N. RAYMOND 1,2, FRANCK SELSIS ADRIAN R. BELU 7, EMELINE BOLMONT 6, KASPAR VON BRAUN 1, 2 8, G. C. ANUPAMA 6, D. K. SAHU , PEDRO FIGUEIRA 9 IGNASI RIBAS 1 Univ. Bordeaux, LAB, UMR 5804, F-33270, Floirac, France. ² CNRS, LAB, UMR 5804, F-33270, Floirac, France 3 Instituto de Astrofísica de Canarias, La Laguna, E38205 Spain 4 Departamento de Astrofísica, Universidad de La Laguna, Av., Astrofísico Francisco Sánchez, s/n E38206-La Laguna, Spain 5 Las Cumbres Observatory Global Telescope Network, 6740 Cortona Drive, Suite 102, Goleta, CA 93117, USA 6 Indian Institute of Astrophysics, Koramangala, Bangalore 560034, India 7 NASA Exoplanet Science Institute, California Institute of Technology, MC 100-22, Pasadena, CA 91125, USA 8 Centro de Astrofísica, Universidade do Porto, Rua das Estrelas, 4150 -762 Porto, Portugal 9 Institut de Ciències de l’Espai (CSIC-IEEC), Campus UAB, Facultat de Ciències, Torre C5, parell, 2a pl., 08193 Bellaterra, Spain Received 2012 August 17 ;accepted 2013 January 8. Adrian. Belu * centraliens net ABSTRACT Given the very close proximity of their habitable zones , brown dwarfs represent high-value targets in the search for nearby transiting habitable planets that may be suitable for follow-up occultation spectroscopy. In this paper we develop search strategies to find habitable planets transiting brown dwarfs depending on their maximum habitable orbital period (PHZ out). Habitable planets with PHZ out shorter than the useful du- ration of a night (e.g. 8 -10 hrs) can be screened with 100% completeness from a single location and in a single night (near-IR). More luminous brown dwarfs require continuous monitoring for longer duration, e.g. from space or from a longitude-distributed network (one test scheduling achieved - 3 telescopes, 13.5 contiguous hours). Using a simulated survey of the 21 closest known brown dwarfs (within 7 pc) we find that the probability of detecting at least one transiting habitable planet is between 4.5+5.6/-1.4 and 56+31/-13 %, depending on our assumptions. We calculate that brown dwarfs within 5-10 pc are characterizable for potential biosignatures with a 6.5 m space telescope using ~1% of a 5-year mission’s lifetime spread over a contiguous segment only 1/5th to 1/10th of this duration. Key words: astrobiology — brown dwarfs — eclipses — infrared: planetary systems — instrumentation: spectrographs — solar neighborhood 1. INTRODUCTION Together with in-situ robotic exploration within our solar system, observation of terrestrial extra-solar planets’ spec- tra are the current most robust approaches in the search for non-Earth life. The thermal emission from a habitable planet at 10 pc is ~1 photon sec-1 m-2 µm-1; recording such a spectrum is within the capabilities of upcoming and even some existing space telescopes. Free-floating (rogue) plan- ets, if yielding sufficient internal heat flow, could maintain habitable surface conditions if they also have adequate insulation, but this insulation would then limit the levels of photon emission, enabling characterization only for very nearby objects (1,000 AU for the case of solid insulation, Abbot & Switzer 2011). Therefore the main approach until now has been to search for characterizable habitable planets around beacon-primaries. These beacon-primaries then become the dominant noise source when subsequently undertaking the characterization of the exoplanets. 1 Resolving the planet from the primary is therefore the first challenge. The spatial resolution of planets remains a technological challenge (Traub et al. 2007, Cockell et al. 2009). Fortuitously, transiting planets1 can be time-resolved from the brighter primary. Differential eclipse spectroscopy has enabled the identification of molecules in the atmos- pheres of giant planets close to solar-type stars (Tinetti et al. 2007; Grillmair et al. 2008; Swain et al. 2009; Stevenson et al. 2010, Beaulieu et al. 2010). However, even the upcom- ing James Webb Space Telescope (JWST) will be able to detect biomarkers with this technique only up to ~10 pc and for primary dwarves approximately M5 and later (Beckwith 2008, Kaltenegger & Traub 2009, Deming et al. 2009, Belu et al. 2011, Rauer et al. 2011, Pallé et al. 2011). Projects such as MEARTH (Charbonneau et al. 2008) are currently screening the solar neighborhood for these eclipsing planets. 1 The planet’s orbit is passing in front of the star as seen from the tele- scope (primary eclipse) BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Yet the geometric transit likelihood and the local stellar population density & distribution do not guarantee the presence of even a single nearby transiting habitable planet suitable for characterization with JWST (Belu et al. 2011). To help solve this scarcity problem we propose to ex- tend the search to primaries not yet considered by current surveys: brown dwarfs (BDs). Bright primary objects (primaries hereafter) are indeed required for transit spec- troscopy, but as we quantify in this paper, occultation (sec- ondary eclipse) spectroscopy (in emission) of habitable planets around a nearby BD is a favorable scenario. The intrinsic emission of a body which is at given (habitable) equilibrium temperature is independent from the type of primary. On the contrary: the dimmer the primary, the less photon noise added to the planetary photons. As said primaries are convenient ‘signposts’, ‘lighthous- es’ or ‘beacons’ for planets, but once the planets are found the primaries become a barrier to planetary characterization. So what is the dimmest lighthouse? Jupiter-sized objects colder than room temperature have been detected with the Wide-field Infrared Survey Explorer - WISE (Cushing et al. 2011). Brown dwarf primaries therefore should represent the optimal limit for the ‘lighthouse’ search paradigm for habitable exoplanets that can be time-resolved2. But can planets form and remain habitable around BDs? There are significant differences between the potential habitability of planets around BDs and main sequence stars. For instance, BDs cool in time and their habitable zone (HZ) moves inward such that a planet on a stationary orbit sees the HZ sweep by in a much shorter time interval than for stellar dwarfs (Caballero & Rebolo 2002, Andreeschev & Scalo 2004). However, there is no clear “Achilles’ heel” that would rule out BDs as habitable planet hosts; this issue is discussed at length in § 6. We thus continue with the assumption that BDs can indeed host habitable planets. This paper is structured as follows. We first examine the observational characteristics of habitable planets eclips- ing a BD (§ 2), which leads us to draft specific strategies for different regions of the BD parameter space (§ 3). We also find it useful (while keeping in mind the uncertainties on such estimates) to derive the contribution of the popula- tion of BD primaries to the expected number of transiting habitable planets sufficiently close to be characterized through occultation spectroscopy, regardless of the type of their primary (§ 4). We examine the performance of occul- tation spectroscopic characterization for a habitable planet eclipsing a BD over the whole BD parameter space (§ 5). We then review the literature on formation of terrestrial planets in the HZ of BDs, and discuss BD HZs. Finally in the Conclusion we outline the key points relevant to the fast-track roadmap of time-resolving characterization of habitable exoplanets. 2 So, put in another way, we simply attempt here to push the astrobiological “follow the (liquid) water” philosophy to one of its many limits. 2 Roche & tidal limited HZs Roche-limited HZs Figure 1. Orbital periods (in hours) of the effective habitable zone limits for 1 (thin lines) and 10 (thick lines) Gyr-old BDs. The restricted black zones account for depletion of the inner HZs by tidal migration (see text). The dotted line corresponds to the Roche period limit for the 10 M⊕, 1.8 R⊕ super-Earth considered here: 3.6 h. 2. OBSERVATIONAL CHARACTERISTICS OF HABITABLE PLANETS ECLIPSING A BROWN DWARF The detectability of eclipsing habitable planets around BDs has already been considered & attempted (Caballero & Rebolo 2002, Caballero 2010): because of the small radius of the BD, putative terrestrial planets cause 1-5% transit depths. Blake et al. (2008, hereafter BL08) additionally note that the small orbital radius of a habitable planet around a BD (a few times 10-3 AU) increases its likelihood to transit. They also address issues such as the convenient ruling-out of background blend false positives thanks to the high proper motion of a nearby BDs population. Figure 1 shows the orbital periods (in hours) of the ef- fective habitable zones. We must caution here that the for- mula used for computing the habitable zone (Selsis et al. 2007a) is in principle valid for photospheric temperatures down to 3,700 K (fit to models of the Earth around F-G-K dwarfs). For the lower photospheric temperatures of M dwarfs in Belu et al. (2011) the correction for photospheric temperature was fixed at the 3,700 K value, and we do the same here: the scaling of the HZ is performed only through the luminosity of the primary (discussion of BD HZ in § 6). Ford et al. (2006) show that if the planets undergo circu- larization then the limit of the possible orbital distances is twice the Roche limit, but if the planets undergo migration (such as in a disk) with a circular orbit, they can have o r- bital distances up to the Roche limit. We therefore also overplot in Fig. 1 the orbital period at the Roche limit PRoche using the gravitation-only formula for the Roche aRoche limit used in BL08 (citing Faber et al. 2005 and Paczyński 1971). 1/3 (the BD’s mass), Since aRoche ∝ MBD 11010010000.00040.02040.04040.06040.0804M/M⊙orbital period (hrs)BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Figure 2. Transit depth of a 1.8 R⊕ planet as function of the brown dwarf’s (BD) mass for different ages of the BD. (1) is fairly constant with the BDs mass, because the terrestrial planet’s mass MP ≪ MBD. G is the gravitational constant. For all tidal migration aspects (here and hereafter): see § 4.2 hereafter. While photometrically monitoring SIMP J013656.5+093347 for intrinsic variability, Artigau et al. (2009) detect a 50 mmag deep transit-like event in J band, with a precision of 5 mmag over 5 min bins (1.6 m tele- scope at Observatoire du Mont Mégantique). The simulta- neous monitoring in another band yielded a different depth of the event, which, they conclude, would not be the case if an opaque body such as a planet were masking the BD (the signature of a transit is grey). Therefore transit detection around a BD ideally involves simultaneously monitoring in two different bands. For reference, Fig. 2 shows the depth in mmag of the transit of a 1.8 R⊕ planet as function of the BD mass for different ages of the BD, with BD radii values from Baraffe et al. (2003, COND03 model). Brown dwarfs are also fast rotators. As they contract and cool down on sub-Gyr timescales their rotation speeds increase. . We therefore note here that if the BD is signifi- cantly oblate and the planet’s orbit is aligned with the BD’s spin, this could reduce the transit depth because of equato- rial gravity darkening (mention of this phenomenon in Herbst et al. 2007). For BDs older than 1 Gyr and heavier than 0.04 M⊙ Bolmont et al. (2011 – Fig. 1) predict rotation periods below 1 hour (for reference, the rotation period of Jupiter is ~10 hr and its flattening 0.06). However, the authors recognize that measured BD rotational velocities available to fit their model are scarce, so caution is appro- priate for this matter3. 3 Also see Leconte et al. (2011). Beyond this article’s title, this reference provides separate modeling of rotational deformation free from an exterior 3 Figure 3. Minimal habitable planet eclipse duration (in minutes), as function of the brown dwarf’s (BD) mass, for different ages of the BD, without and with tidal migration (optimistic planet formation at 10 Myr considered). For masses below the plotted low mass bounds the radiative habitable zone is entirely below the Roche limit or below the minimal asymptotic final tidal-migration orbit (no effective planetary habitable zone). In the case of unresolvable binary BDs (such as 2MASS 0939-2448 AB, Leggett et al. 2009), an Earth-like planet around the brightest component (S-type orbit) still produces a 3.5% (40 mmag) deep transit in the combined photometry. For other unresolvable binary BDs, if both feature an effective HZ, one survey actually can monitor two habitable zones simultaneously.4 The required cadence of observation is set by the ex- pected minimal duration of a habitable planet eclipse around a BD (Fig. 3): (2) where PHZ min = max (PHZ in; PRoche; Ptides) is the effective minimal habitable circular orbital period. PHZ in is the or- bital period at the inner limit a HZ in of the radiative habita- ble zone (HZ in), which is established for a planetary body. PRoche, Ptides and aRoche, atides are defined in the same way, at the Roche limit and for tidal migration respectively. RBD is the BDs radius, and b is the median impact factor: (3) for a 10 M⊕, 1.8 R⊕ planet, although this median does not take into account the decrease in eclipse depth for high impact factors.5 If the BD is oblate and the planet’s orbit is gravitational influence for BD mass range objects (J. Leconte 2013, pri- vate communication). 4 Also see Eggl et al. 2012, An Analytic Method to Determine Habitable Zones for S-Type Planetary Orbits in Binary Star Systems 5 Also note that Eq. 1 assumes the transit to start when the center of the planet touches the limb of the primary, whereas Eq. 2 assumes transit start at first contact. Assessing the error induced by all these imprecisions in modeling is beyond the scope of the present work. PBD3RocheRoche4MMGaP);;max(1asin1tidesRochein HZBD2min HZminaaaRbP);max(8.1cosa214cos8.1);max(in RocheBDBDin RocheHZHZaaRRRRaabBELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Photon noise-only signal-to-noise ratio (S/N) of the detection of a single transit event in J band for a non exhaustive list of BDs with PHZ out ≤ 8 h Table 1 PHZ (h) in (Roche) out  min (min) Depth (%) SDSS J1416+13 B 2MASS 0939-2448 A AB CFBDS J005910-011401 3.42 3.42 5.99 6.52 12.2 12.5 3.42 5.45 16.1 3.8 4.0 3.5 3.6 mag (J) 17.35b 16.1c 15.98 18.06d photons s-1 m-2 (a) (× 103) photons, per ½ transit @ HZ in (× 103) S/N per transit 3.8 11 13.4 1.98 600 1800 2200 416 21 34 17 Geom. prob. HZ out (%) 16 14 20 Notes PHZ out is the orbital period at the outer edge of the radiative habitable zone. No tidal migration is considered here. 3 .5 m-class telescope. We want the cadence to be half the minimal possible duration of the transi t  min (at least one complete exposure taken during the transit). a 0.82 transmission of the filter included. b Burningham et al. (2010). c The magnitude of the A component alone was estimated using values from Table 2, and it was checked that when doing the same with the B component the fluxes in the next column add up. d Scholz et al. (2009). For references on the remaining parameters of the targets, see Table 2. aligned with the BD’s spin, the duration of the eclipse will be larger. Finally, the fast rotation of BDs separates the transit signal cadence from rotation-induced variability. BDs are thought to feature evolving inhomogeneities in their cloud deck (weather), likely to generate variability in a manner similar to that of star spots. The corotation distance is the orbit at which a planet’s angular speed matches the BD ’s rotation. So as the rotation speed of the BD increases, the corotation distances moves inward with time. It becomes smaller than the Roche limit after 10-100 Myr, which is also the likely formation time-scale of terrestrial planets around a BD (Bolmont et al. 2011). Thus rotation-induced photometric variability of the BD primary would be in a totally different frequency regime than possible transit the evolution time-scales of cadences. Still, if the inhomogeneities are of the order of the transit cadence (planet orbital periods), rotation induced variability due to these features are very likely to increase the false alert rate (compare Fig. 3 with Roche orbital periods) . 3. DETECTION STRATEGIES 3.1. The Importance of Completeness Assessment for Vol- ume-limited Surveys For any given primary, a photometric monitoring of a given sensitivity can yield a single self-significant transit event candidate6 down to a planet radius Rp (we consider a mean transit duration within our region of interest: the habitable zone). Monitoring this primary for a continuous duration T implies that all potentially transiting planets with periods up to T, whatever their epoch (i.e. orbital phase), have been screened for. This can be dubbed ‘100% completeness up to T (and down to Rp)’. Publishing the completeness of a volume-limited survey is particularly important when the final yield of detections is expected to be low, or even be- low unity. Indeed, the yield is necessary for planning fol- 6 By default and unless mention otherwise no shift-adding of multiple observation nights (to increase the signal-to-noise ratio - S/N) is consid- ered in this paper (more on this in § 3.3). 4 Figure 4. Planet orbital periods for which transit screening completeness is 100%, for two adjacent full monitoring nights each of usable duration n (in the case of just one monitoring night the maximum orbital period for which completeness is 100% is of course only n - diamonds). The step in n between each grey shade is 1 hr, values for darkest and lightest are indicated. low-up eclipse spectroscopic characterization (including building of dedicated facilities). We now examine specific strategies for different regions of the BD parameter space. 3.2. Ground based, Single Night, Single Longitude, 100% Completeness If we ignore tidal migration (§ 4.2), the Roche-limited habitable zone (grey contour in Fig. 1) extends to orbital periods shorter than the typical photometric night durations (e.g. but not limited to, airmass < 2). The Roche limit is at 3.6 h for the 10 M⊕, 1.8 R⊕ super-Earth considered in Fig. 1, at 4.5 h for an analog of the Earth, and at 5 h for a 0.1 M⊕, BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS 0.5 R⊕ planet. This means that some BDs can be screened with 100% completeness from the ground, in just a single observing night, with a single telescope at a single longi- tude – a remarkable efficiency. When such a candidate is detected up to three subse- quent follow-up nights are required: the first for determin- ing a period and confirming the alert, a second for confirm- ing the period and the periodic nature of the signal. An additional shifted third night can help ruling out submulti- ples of the initial period. A tradeoff in the shift has to be determined, since the greater the shift the greater the buildup of ephemeris uncertainties. We explore the increase in orbital period screening by monitoring for one adjacent night (Fig. 4). For instance in the case of a 9 hr useful night the addition of one adjacent observation night yields 100% completeness for planet orbital periods up to 11 hr, and also for orbital periods between ∼15and ∼16.5 hr (∼39% increase). Table 1 gives the expected, photon noise-only signal-to- noise ratio (S/N) of the detection of a single transit event with a 3.5 m-class telescope in the J band for several near- by BDs with PHZ out ≤ 8 h. The integration time is derived from the minimal possible duration for the transit of a hab- itable planet (i.e. a transit at the inner edge of the habitable zone – HZ in). We further halve this integration time to take into account alternating between two bands (see the Artigau et al. observation mentioned above), in case no dichroic is available. We consider the overhead per one forth-back filter switching to be 40 s (case for WIRCAM - Wide-field Infra-Red Camera on the Canada France Hawaii Telescope - CFHT). In the last column, we also give the geometric likelihood of transit at the outer limit of the habitable zone (i.e. the lower limit on the transit likelihood of habitable planets around these BDs; evidently, no primary types have higher habitable planet transit probability as the BD type. To conclude, we note that BDs are currently being photometrically monitored with IR telescopes for increas- ingly extended continuous periods in the frame of atmos- pheric (weather) and evolution tracks research. We there- fore call to this community to integrate the science case presented in this subsection in the evolution and further expansion of their field. 3.3 Ground, Multiple Nights Brown dwarf habitable zones extend up to 10 days of o r- bital period (Fig. 2). BL08 suggested the use of a redun- dant, longitude distributed network of telescopes for con- tinuous photometric monitoring, such as the Las Cumbres Global Telescope (LCOGT) network7. We have executed a test of such longitude distributed observation in early 2011. One z = 17 target was scheduled for 13.5 h of continuous monitoring, involving the two 2-m telescopes of the LCOGT in Hawaii and Australia and the 2-m Himalaya 7 lcogt.net 5 Chandra Telescope (HCT) with the Himalaya Faint Object Spectrograph and Camera (HFOSC). Meteorological conditions enabled only observations from HCT, and the overall environmental conditions for that observation caused a high background level. Therefore the signal-to-noise ratio of the final light-curve (not shown) for this very faint target was too low for exploitation. In conclusion, for the moment such a 2 m far red optical -class network may not be yet sufficiently longitude- redundant for robustness against environmental variability. Also, slightly brighter-on-average targets may relax the constraints on environmental conditions. However, these targets would have longer habitable zone outer limit per i- ods, therefore requiring longer monitoring in order to achieve complete habitable zone screening. For instance already for the present test target the longest habitable period was longer than the 13.5 continuous hours we were able to secure. Longer monitoring means more different observatories are stringed together for such an observation. The red spectral energy distribution of BDs also advocates for extending the equipment of the 2 m class collectors worldwide with J-H-K detectors. Last, taking into account the meteorological forecasts at the different observatories and triggering the observing sequence in a Target-of- Opportunity (ToO) fashion could be investigated. If the continuity is disrupted before the longest habitable period can be covered, a scheduling algorithm can enable to optimize the completeness of the screening of a given tar- get (e.g. Saunders et al. 2009). The completeness may not reach near 100% but for a significant increase in observa- tion time cost. However, such multiple observations enable to search for shallower transits and/or primaries with in- creased variability, by phase-folding search techniques, and accounting for the subsequent introduction of correlated (red) noise (von Braun et al. 2009) . Note that flare- variability can be a real challenge for phase folding in M dwarfs light curves. Last, Blake & Shaw (2011) have shown recently that, following the quality of the site, preciptable water vapor (PWV) variability can induce 5 mmag variations in z band on the hour timescale; however they indicate that PWV can be monitored through Global Positioning System signals. 3.4. Intermediate Cases For the BDs with outer habitable period between ~8 and ~20 h (i.e. the cases intermediate to those addressed in §§ 3.2 and 3.3 above), one would require a network such as the one described above, but operating in the infra-red. Such coordinated observations between telescopes usually operated through time allocation committees may prove difficult to set up (considering the very high pressure on these telescopes and constraints on mutual telescope ob- serving coordination). Therefore, in the frame of a cohe- sive grand strategy for ground detection of habitable plan- ets eclipsing BDs, coordinated observations are likely only as a second step, after 1 telescope-, 1 night observations on cooler targets are first demonstrated (§ 3.2 above). If no BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Table 2 Total expected number of eclipsing habitable planets around nearby L, T and Y dwarfs. T (K) M/M⊙ R/R⊙ L/L⊙ (log) Ref. HZRoche 350 0.011 0.01 -6.88 Cushing et al. 2011 1320 0.065 0.0805 -4.699 King et al. 2010 0.050 0.0825 -5.232 0.039 0.091 -5.1 Kasper et al. 2007 WISE 1541-2250 GJ 845 B a b SCR 1845-6357 B UGPS 0722-05 DEN 0817-6155 DEN 0255-4700 2MASS 0939-2448 A B WISE 1741+2553 2MASS 0415-0935 GJ 229 B GJ 570 D SIMP J013656.5+093347.3 2MASS 0937+2931 WISE 0254+0223 WISE 1738+2732 Dist. (pc) 2.8a 3.6 3.85 4.1 4.9 5.0 5.3 5.5a 5.7 5.8 5.9 6b 6.1 6.1a 7c 910 950 505 950 0.005 0.10 0.04 0.089 1300 0.035 0.09 700 500 947 950 948 0.038 0.085 0.024 0.09 0.01 0.12 0.038 0.094 0.019 0.11 1200 0.044 0.097 0.054 0.01 0.08 0.11 950 660 350 -6.13 -3.53 -4.62 -5.8 -6.3 -5.0 -5.2 -5.0 -5.25 -5.33 Leggett et al. 2012 Artigau et al. 2010 Leggett et al. 2009 d Del Burgo et al. 2009 Geissler et al. 2008 Del Burgo et al. 2009 Artigau et al. 2009 Leggett et al. 2010 -5.7 Kirkpatrick et al. 2011 0.019 0.093 -6.94 Cushing et al. 2011 ×0.41 +0.54/-0.13 = 0.78 +1/-0.25 Stephens et al. 2009 1 0.55 0.74 0.023 0.033 HZasymp. trans. prob. 1 Myr 10 Myr 0 1 0.95 1 0.79 0.98 0.64 0.27 1 Myr 10 Myr 0.030 0.019 0.52 0 0.79 1 1 1 0.24 0.55 0.96 0.87 0.95 0 0.14 0.08 0.14 0.021 0.051 0.011 0.006 0.010 0.09 0.11 Expected # 0.036 +0.048/-0.012 0.078 +0.06/-0.014 % probability of at least 1 occurrence 4.5 56 3.9 +4.9/-1.2 +31/-13 +5.6/-1.4 Notes HZRoche is the fraction of circular orbits that are in the habitable zone but outside a 10 M⊕ Roche limit (uniform distribution in radius). Similarly HZasymp is the fraction, from the remaining habitable zone, where planets can exist at the end of the tidal migration process, for two differen t planet formation ages. The last two columns give the corresponding transit probability of the median orbit in the remaining final effective HZ, weighted by HZasymp. Therefore, the total of these two last columns (0.09 and 0.11), multiplied by ⊕= 0.41 +0.54/-0.13 from Bonfils et al., are estimates of the total expected number of habitable planets transiting BDs in this volume (see text for detailed justification). Distances (pc) are RECONS parallaxes unless mentioned otherwise. When parallaxes were not available, we use photometric distances. Values in italic are from the reference. The remaining non-italic parameters (among photospheric temperature T, mass M, radius R and luminosity L) are interpolated from COND03 grids using the parameters from the reference. Note that the purpose of this table is to compute some ensemble averages; therefore the values of the BD parameters should not be reused for the study of individual objects, since the uncertainties on most parameters are quite large (e.g., spectroscop- ic/photometric distance estimates), and because of ongoing refined observations (e.g. parallaxes). a Kirkpatrick et al. 2011, parallax. b Faherty et al. 2009. c Average of quite dissimilar photometric distance (Kirkpatrick et al. 2011) and spectroscopic distance (Cushing et al. 2011). d For this BD, interpolation of the grids to the 2MASS J and H magnitudes (Kirkpatrick et al. 2011, Scholz et al 2011) did not converge (as it was more the case for WISE J0254+0223). We therefore use here the values of the only other T9 BD in the sample, UGPS 0722-05. coordinated observation can be set up observations have to be spread throughout the observing season of the target, arranging them so that together they satisfactorily cover the time-folded range of orbits that is sought, significantly increasing the total cost in telescope time. See also Berta et al. (2012) for a related study deriving from the MEARTH survey for habitable planets transiting M dwarfs. This study includes analytical tools for integrating “lone transit events” (from different telescopes using different filters at different observatories) into coherent planet candidates. The optimal approach for screening these intermediate cases is a dedicated monitoring program from space, where uninterrupted monitoring can be achieved . Since 2011 August the Spitzer Warm Mission Exploration Science Program 80179 “Weather on Other Worlds: A Survey of Cloud-Induced Variability in Brown Dwarfs” (Metchev: PI) is monitoring one after another, for a minimum of 21 con- tinuous hours each, BDs from a list of 44 targets (873 hr awarded in total)8. Unfortunately none of the 25 targets observed until now are at or beyond than 7 pc. Also ob- serving simultaneously in both channels of Warm Spitzer (3.6 and 4.5 µm) is not possible because the arrays of each channel see different non-overlapping parts of the sky. Therefore a prospective interlaced mode is not up for con- 8 http://sha.ipac.caltech.edu/applications/Spitzer/SHA//#id=SearchByProgr am&DoSearch=true&SearchByProgram.field.program=80179 6 BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS sideration at the cadence required both by intrinsic variabi l- ity studies or exoplanet detection. 4. EXPECTED NUMBER OF NEARBY ECLIPSING HABITABLE PLANETS 4.1. Previous Study Due to S/N constraints, habitable planets can be searched for biosignatures with eclipse spectroscopy only out to a limited distance. Belu et al. (2011, Fig. 8) find that at 6.5 pc emission biosignatures can be detected within JWST’s baseline lifetime only around primaries of spectral types later than ~M5. The S/N scales as the planetary radius squared, and inversely with the distance. Also, considering a specific spectral signature a given planet might exhibit (and which would require to be confirmed or rejected), the signature’s S/N scales linearly with the strength of the spec- tral feature (number of atmospheric scale heights for pr ima- ry eclipse and brightness temperature depth for secondary), and with the inverse of the resolution. Belu et al. further showed (Fig. 19 therein) that in a ~7 pc volume 9 , the total expected number of habitable eclipsing planets is ~0.3 for the M5-M9 dwarf primaries population (considering the recent Bonfils et al. (2011) lower bound for ⊕: 0.41). Note that Selsis et al. (2007a) pointed out that a primary may host in principle several habitable planets (as was perhaps the case for the Sun 4 Gyr ago, when Venus, the Earth and Mars were potential- ly habitable). Therefore the case of a final ⊕ > 1 is not to be discarded. For the compact habitable zones of BDs (10 - 3 AU scale), further studies are needed to address the stabil- ity of multiple planets in this zone, and the ir possibility to form and/or migrate (see next subsection). 4.2. Present Study We now extend the previous study to T, L and Y dwarfs; some young BDs could exhibit M spectral type (and vice- versa): they would have been included in the previous study. Table 2 lists the known BDs likely within 7 pc. This table was compiled from RECONS (recons.org), SIMBAD and other recent discoveries. For each object we have collected from the literature estimates of photospheric temperature, mass, radius and luminosity. The missing values were interpolated from the COND03 evolutionary grids (Baraffe et al. 2003) using the available parameters. Table 2 contains two binary systems. GJ 845 Bab ( In- di Bab) is a binary BD system, with a separation between the components of at least 2.1 AU (Volk et al. 2003). The outer limit of the habitable zone for each of the two com- ponents (S-type orbits) is 0.01 and 0.005 AU respectively, so orbital stability at these distances is not at issue. On the other hand 2MASS 0939-2448 is likely a dissymmetric close binary (Burgasser et al. 2008, Leggett et al. 2009) 9 Volume limit of the 100 nearest objects at that time, RECONS (www.recons.org) 7 with a separation of under 0.03 AU. Holman & Wiegert (1999) have derived semi-major axis upper limits for dy- namically stable orbits around each component of a b inary. The outer habitable orbits around each likely component of 2MASS 0939-2448 (0.0029 and 0.0016 AU respectively - or 7.2 and 3.8 times the radii of their respective primaries) are indeed dynamically stable. We then calculate for each BD the fraction of circular orbits that are in the habitable zone but outside the Roche limit (column HZRoche, 10 M⊕ planet). When computing geometric transit probabilities it is usually assumed that the probability of a planet occupying a given orbit is flat across the available parameter space, e.g., the HZ. However, given the close proximity of BDs’ HZs, tidal interactions between the planet and BD act to modify a planet’s orbit after its formation, with conse- quences for its transit probability. Bolmont et al. (2011) studied the tidal evolution of planets orbiting BDs (previously referred to here as tidal migration). The basic concept is as follows. We already mentioned the gradual reduction of the corotation distance (§ 2). This is important because a planet’s position with respect to corotation determines the direction of tidal mi- gration. Given that BDs’ corotation distances shrink in time, almost all planets that survive around BDs experience outward tidal migration. Thus, there is a parameter- dependent orbital radius inside of which planets should not exist. The asymptotic limit is reached after 10-100 Myr. The ages of observed BDs are known with precisions equal to or larger than 0.1 Gyr, and all of the BDs in Table 3 have age estimates larger or equal to 100 Myr. We therefore proceed and apply the asymptotic limit model to all the BDs in Table 2. The key parameters that determine this limit are the BD and planet masses and their internal dissipation rates (Bolmont et al. 2011). From this reference we use Figs. 6 and 8 which give the asymptotic limit (semi-major axis) for planets around BDs of different masses (for a 1 M⊕ planet forming at 1 and 10 Myr respectively). And we give for the remaining BDs in Table 2 that have HZRoche > 0 (3 BDs do not) the non null fractions of circular orbits that are in the habitable zone but with semi-major axis larger than the asymptote (column HZasymp.). Finally, for the remaining BDs that still feature the above effective final habitable zone, we compute the transit likelihood at the middle of the zone (with the same working assumption ⊕ = 1). And we weight (multiply) this likel i- hood by HZasymp. thus obtaining a final transit probability (column transit. prob.). The justification for this weighting is the following. These individual likelihoods are then mul- tiplied by the recently directly measured ⊕ within the M dwarf primary population, 0.41 +0.54/-0.13 (Bonfils et al. 2011, lower limit, as per sensitivity of the technique to the whole range of habitable planet masses). Weighting our individu- al transit likelihoods by HZasymp. is a valid approach because the habitability of individual planets in the sample of Bonfils et al. is based on radiative (from the primary) co n- BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Figure 5. Signal-to-noise ratio (S/N) on the detection of a spectral feature in emission (secondary eclipse spectroscopy), at 10 µm, as function of the mass of the brown dwarf and the orbital period of the planet. The spectral feature has a brightness temper ature depth of 30 K and is 0.1µm wide (i.e. R = 100). The planet is a 1.8 Earth-radii super-Earth, and the system is situated at 6.7 pc. The observations of 90 eclipses are summed for this result. The age of the brown dwarf is 1 Gyr (left) and 10 Gyr (right, note the different abscissa scale). The grayed area is the habitable zone (§ 2). The S/N scales linearly with the square root of the number of summed eclipses (if no correlated noise), with the square of the planet’s radius, with the brigh tness temperature depth, with the inverse of the distance in pc, and with the inverse of the resolution . siderations alone, and their primary population is not sub- ject to the limitations included in the HZasymp. factor (Roche limit and tidal-migration). Their ⊕ contains only infor- mation on planet orbital density as a function of the prima- ry; it can validate formation and migration models, the latter excluding tidal migration because the habitable dis- tances around M dwarfs are too large for this mechanism. It is this formation- and migration-other-than-tidal- planet density function of Bonfils et al. that we extrapolate here to BDs. We must caution however that the figures in Bolmont at al. (2011) seem to indicate that the tidal migra- tion mechanism tends to redistribute orbits (i.e. change the density of orbits, either shepherding together or dispersing, depending on the initial conditions). The final expected number of eclipsing habitable planets around BDs within 7 pc is given in bold at the bottom of Table 2. A more significant number is the probability for a survey to yield at least one transiting habitable planet in this volume (also in Table 2). Assuming the optimistic scenario of late planet formation around BDs ( i.e. at 10 Myr), the survey of the corresponding 6 BDs in Table 2 has 4.5+5.6/-1.4 % chance of yielding at least one habitable eclipsing planet. Given the recent nature of the work on tidal migra tion, we also include the significantly more optimistic figures when tidal-migration is not considered. The probability for the survey of the above 14 BDs with Roche limited-only HZs to yield at least one transiting habitable planet likely within 7 pc is then 56+31/-13 %. 4.3. Discussion Thus, to include BD primaries in the search for nearby, eclipse-characterizable habitable planets is to increase the expected number of occurrences ~2.5-fold (when compared with the late M dwarf-only search). 8 The results on tidal migration depend on some parame- ters which are unknown/poorly constrained for our nearby BDs, such as the dissipation factor in the BD or the initial rotation rate. There are also various rotation braking mech- anisms that are not considered. This is to be combined with the uncertainties on the parameters of which we give the estimates (mass, age). Will future refinements of tidal migration modeling enable to gain back the order of magni- tude between the Roche-only limitation of the habitable zone and the one by tidal migrations? Or will it completely rule out habitable planets around BDs? What about other mechanisms for late migration, or about the frequency of late planet scattering? The planet’s mass is also a strong factor for tidal migra- tion, with planets of 0.1 M⊕ (0.5 R⊕) hardly experiencing any effect. Unfortunately the performance of the subse- quent search for spectral signatures scales with the square of the planet’s radius. But this should also remind us that the volume limit we used for our list is an average estima- tion of spectroscopic characterization capability with the JWST; particularly favorable cases (strong spectral signa- tures, planets close to the inner limit of the habitable zone, etc.) may be characterizable further away. Planet detection surveys should therefore plan a significant margin on these volume estimates (and the number of targets scales with the cube of the distance). Also note that only lower bounds on the local space density of BDs are presently available (Kirkpatrick et al. 2011). The final BD detection count from the ongoing processing of the Wide-field Infrared Survey Explorer (WISE) data is expected to be ~1,000 BDs, which should double or triple10 the number of know primaries within 25 light years (7.6 pc). Our values should be considered there- fore as lower limits. 10 As of December 2009, NASA WISE Launch Press Kit. BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS 5. FUTURE CHARACTERIZATION 6. DISCUSSION We now consider habitable planet secondary eclipse spec- troscopy performance around a BD. We reprise our previ- ous such study around F-M dwarves with the JWST (Belu et al. 2011 for detailed description of the modeling). Brown dwarfs may not exhibit significant near-UV flux. Therefore a (biotic) O2 atmosphere on a BD exoplanet may not generate O3 (ozone) in its stratosphere, which is a con- venient O2 detection proxy around 10 µm (location of thermal emission from a body at habitable temperatures). The question of BD HZs and biosignatures is discussed at length in § 6.2 & 6.3. We therefore consider a fiducial spectral feature in emission (at 10 µm, brightness tempera- ture depth of 30 K, and 0.1 µm wide - i.e. resolution R = 100). The instrument considered is the Mid Infra-Red Instru- ment (MIRI) in Low Resolution Spectroscopy (LRS) mode. For 1 and 10 Gyr-old BDs, Figure 5 shows the signal-to- nose ratio (S/N) on the detection of our fiducial spectral feature from a 1.8 Earth-radii planet at 6.7 pc, summing the observations of 90 secondary eclipses. Program time cost per eclipse is twice eclipse duration (at least ~30 min, Fig. 2), plus the 65 min generic JWST slew time budget, every 10-70 h (period of the planet). Note that even for the long- est period planets (~10 days), 90 transits are well within the telescope’s lifetime. We reprise here our comment from Belu et. al (2011): such an hypothetical observation, which would happen on only one (see § 4) most interesting trans- iting system, represents a total telescope time only a magni- tude larger than the longest exposures made until now with the Hubble Space Telescope (HST, Beckwith et al. 2006). Also note that the 1.8 Earth radii is an upper limit for habit- ability, but could be extremely optimistic in terms of initial mass available in a BDs protoplanetary disk for planet formation. Despite the lower luminosity of the primary, hence the reduced photon noise, shorter orbital periods also mean shorter occultation durations (10-40 minutes). This curbs the gain one could have expected relative to the case around M dwarfs. The discussion at the end of § 2 on the rotation-induced variability of primary and transit detection also applies this eclipse characterization follow-up. One can see that atmospheric absorption features such as the one presented here can be detected on habitable planets eclipsing BDs after a follow-up of a couple of months, for a cost < 2.5 h of observation every 1 – 2 days, so on average about 1% of the 5-year mission time of the JWST. This cost in mission time is about a factor 2 better on average than for M dwarf habitable planets (Belu et al. 2011), and more importantly, spread over only 1/5 th to 1/10th of the 5-year mission time (whereas in the M dwarf case the required number of observations spreads over the entire mission life-time, exposing to the risk of dedicating time and acquiring data that ends up having insufficient S/N if the JWST were to become inoperable too soon). 9 6.1. Terrestrial Planet Formation and Orbital Evolution around BDs BL08 and Bolmont et al. (2011) reviewed the literature on the likelihood of formation of habitable planets around BDs. There is ample evidence in favor of terrestrial planet for- mation around BDs: the same fraction of young BDs has circumstellar disks as do T Tauri stars (Jayawardhana et al. 2003, Luhman 2005), and there is observed evidence of grain growth in BD disks (Apai et al. 2005). Of course, the exact outcome of the accretion process depends on the disk mass and mass distribution (Raymond et al. 2007, Payne & Lodato 2007), which probably scales roughly linearly with the primary mass (Andrews et al. 2010). Regarding for- mation, see also Charnoz et al. (2010) for late accretion at the Roche edge of a debris disk. If tidal migration influences are confirmed there should be no planets with orbital period under 8 h orbiting them (Fig 3, black contours). Only extreme unlikely scenarios could allow such planets, like unusually low dissipation factors, unusually high initial rotation rate, or very recent capture, migration or formation. For instance Fig. 19 in Bolmont et al. (topmost panel for the lowest BD dissipation factor) shows a tremendous sensitivity to the initial semi- major axis when computing the final asymptotic one. In theory there could be an extremely narrow interval of initial positions for which the final asymptotic orbit is as close as desired to the Roche limit. This interval is likely ≪ 10-3 AU (the span of initial semi-major axes for which planets that migrate in are saved, whatever their final asymptotic semi-major axis). However, we recall that the detection of hot jupiters was proposed as feasible almost half a century before their detection (Struve 1952). To summarize, there are signif i- cant theoretical uncertainties associated with each of these questions and no certain answers at the current time. We are of the opinion that, given their exceptional detection advantages (1 night-, 1 location- 100% completeness) we should invest in the presented strategy. Additionally, whole night monitoring of BDs may enable improved BD atmos- pheric studies. Therefore planets transiting BDs with per i- ods under 8 h have a high payoff / screening cost ratio. 6.2. Habitability and BDs A radiative habitable zone (HZ), within which terrestrial planets can sustain surface liquid water, can be defined around BDs. The inner-edge of the HZ corresponds to a H2O-rich atmosphere and the outer edge to a greenhouse efficient gas-rich atmosphere – most likely CO2. The inner limit is reached when the mean stellar flux absorbed by the planet is 300 W m-2 (runaway greenhouse threshold). De- termining the location of the outer edge, which depends on the efficiency of CO2 as a greenhouse gas, will require specific climate modeling (1D & 3D), due to the strong overlap between the thermal emission of the BD and the molecular lines in the planet’s atmosphere (e.g., Words- BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS worth et al. 2011). The full absorption of the continuum of H2O and CO2, and the absence of Rayleigh scattering will likely lead to a planetary albedo close to null (we take 0.1 in § 5). Strong stratospheric warming and inefficient green- house is expected, possibly leaving the surface at a lower temperature than the globally perceived brightness tem- perature of the stratosphere. We are currently developing 1D and 3D codes suitable for BD planets. The contribution of internal heating due to tidal effects (Jackson et al. 2008 for stellar masses down to 0.1 M⊙) may also be significant. Several threats against surface habitability exist within the HZ of BDs. One is the tidal spin-orbit synchronization. Planets on circular orbits inside the HZ of M stars and BDs are expected to have a permanently dark hemisphere and a zero obliquity. If zonal and meridional heat transport is insufficient the water and the atmosphere can end as con- densed caps at the poles and night side of the planet. Simu- lations done for GJ 581 d (Wordsworth et al. 2011) show however that a dense atmosphere can provide enough heat transport to homogenize the temperature over the whole surface (see also Joshi et al. 2003). Another threat is the rapid cooling of the BD (Fig. 5), which has two implications. The first one is that, coupled with the tidal migration outward drift, a planet remains habitable during only a fraction of the BD’s life (more than 1 Gyr only for BDs > 0.04 M⊙ - Bolmont et al. 2011). Nonetheless, life on Earth is thought to have existed within 1 Gyr of its formation. Thus, although planets have a short habitable window around BDs, it is of great scientific inter- est to search for them (see Lopez et al. 2005 for a similar discussion about the HZ around red giants). The second implication is that habitable planets were initially on the hot side of the HZ. Around a Sun-like star such a hot loca- tion would imply atmospheric losses of water. Venus for instance has kept little of its initial water reservoir, as shown by the high D/H ratio of its remaining water (a few tens of cm precipitable). If planets lose most of their water content during their pre-habitable history, they are unlikely to become habitable worlds when the HZ catches up with them. The case of Venus, however, may not be a relevant analog for BD planets. Indeed, the Sun emits significant UV and XUV(EUV) fluxes, respectively able to photolyse H2O and to drive the atmospheric escape by heating the exosphere. It is unclear whether BDs have enough activity to pro- duce such fluxes that would result in a significant water loss. For G, K and early M stars, magnetic activity and resulting XUV emission is correlated with rotation rate (Ribas et al. 2005; Scalo et al. 2007) and thus the XUV levels in the HZ of early M stars remain very high (higher than in the HZ of the Sun) for 1 to a few Gyr. For late M stars and BDs, and despite their high rotation rate, there is a steep drop-off of activity, which may be explained by their lower atmospheric temperature and ionization fraction (Mohanty et al. 2002). Very young BDs do exhibit observ- able X rays (Preibisch et al. 2005) but likely to come from the accretion of a protoplanetary disk ergo predating the formation of planets. Therefore in the absence of signifi- 10 cant photolysis and exospheric heating, it is possible for planets on the hot side of the HZ of a BD to keep a steam atmosphere long enough to become habitable. Note also that, even in the case of significant atmospheric and water erosion, the amount of water that remains for the habitabil- ity window depends on the initial reservoir. Volatile-rich planets, or so called ocean-planets, that have formed in the cold outer part of the protoplanetary disk and migrated toward inner regions, can keep more than a terrestrial ocean for Gyrs even if located close to a Sun-like star (Selsis et al. 2007b). It is therefore possible to have oceans at the sur- face of planets in the HZ of BDs. Note: Since the submission of this article, Barnes & Heller (2013) have independently addressed & quantitative- ly furthered some of the questions raised above in this Sec- tion; especially they include the potential effect of tidal heating when the planet eccentricity is forced to non-zero values by planet-planet interactions. 6.3. Biosignatures from BD planets The atmospheric biosignatures paradigm rests on the ability to detect an out-of-equilibrium thermodynamical state and that and that all simpler physical processes fail to reproduce this state. For instance the photolysis and escape men- tioned above can lead to abiotic O2 buildup. Unfortunately a back of the envelope calculation shows that a detectable level of UV emission from nearby BDs (with the HST) corresponds to levels in their habitable zone significantly higher that the ones required for such O2 buildup. Supposing that all simpler physical causes for the ob- served out-of-equilibrium state are ruled out it is then likely that a more complex process is responsible. On Earth it is oxygenic photosynthesis (and not chemoautotrophy) that is responsible for the out-of-equilibrium state of our atmos- phere (Rosing 2005). Oxygenic photosynthesis uses pho- tons to break water molecules and the liberated hydrogen to reduce CO2. The water bonding energy corresponds to a wavelength of 240 nm. As no such short wavelength reach- es the surface of the Earth life has adapted to the available, longer wavelengths spectrum and eventually managed to store the energy of the number of photons required for breaking a single H2O molecule. Using the even longer wavelengths of a BD may imply only to store more photons per H2O molecule. Actually, photosynthesis may have evolved from initial infrared sensors used to detect sources of heat (Nisbet et al. 1995), and there are claims that photo- synthetic organisms still use infrared light at great oceanic depth (Beatty et al. 2005). There is, therefore, no reason to rule out the possibility of a photosynthetic activity (oxygen- ic or not) evolving on a BD planet. So going back to BDs possibly lacking photochemical out-of-equilibrium influence on their HZ planets, detection of biosignatures would be a most robust detection of life. Remains the question of how to detect O2? BDs do not provide enough flux at 760 nm for primary eclipse spec- troscopy detection (except perhaps for the brightest BDs). Also no O3 buildup is expected (§ 5). Therefore detecting BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS O2 in the absence of significant photochemistry and visible flux seems to be challenging. One path could be to invest i- gate the observability of the dimers O2-X2 (O2-O2 , O2-N2 , etc.) at 1.26 µm, in the extended column at the limb during primary transits (Pallé et al. 2009 – Fig. S3). 7. CONCLUSION In this paper we have examined three different strategies for detecting eclipsing habitable planets around BDs, de- pending on the maximum habitable orbital period. These group into ground searches and into space based searches (Spitzer Space Telescope). Planets orbiting massive and old BDs can have their maximum habitable orbital period shorter than the duration of the observing night. Though they may be rare they can be screened for with 100% completeness in only one (1) night, from a single location (1 telescope). Conducting the transit search in the near-infrared is mandatory. We have started investigating the monitoring of bright BDs with 2m-class telescope network (multiple institu- tions) in the deep red optical. Coordinated observation with a longitude-distributed network of telescopes can increase the duration of the continuous monitoring sequences there- fore reducing the time-cost for achieving satisfactory screening completeness for a given target. However weath- er statistics at the various sites (and their correlation) should be included in estimating the time-cost of such campaigns, and the need for additional longitude -redundant coordination. It is the third option, of monitoring from space, which appears as the most effective. We also show that the density of habitable eclipsing planets around BDs vary greatly depending on mechanisms that have only recently started to be investigated around BDs. Consider a survey of the habitable orbits for the 21 closest BDs (7 pc). The likelihood to detect at least one (habitable) transiting planet varies between 4.5+5.6/-1.4 and 56+31/-13 %, depending on whether tidal evolution is taken into account or not. Even in the pessimistic case, since these planets are also remarkably easy to find if do exist, a search program is worth the risk since it can val idate these newly investigated mechanisms (tidal-induced migration). Occultation spectroscopic characterization of a habitable planet around a BD within 5 to 10 pc is achievable with the JWST. Such a program would be spread over only 1/5th to 1/10th of the mission’s life-time (instead of the whole mis- sion for planets around M dwarfs). More generally, given uncertainties in the existence of available nearby eclipse-characterizable habitable planets, an effort should be made to upgrade transit surveys around low mass dwarfs projects to increased sensitivity (as planed in recent Berta et al. (2012), but also increased collector diameter, better filters, better site). Setting up new surveys is also to be considered. The completeness of the screening for eclipsing habitable planets around nearby low mass dwarfs should be published at the earliest, to enable if re- 11 quired the adjustment of roadmaps toward their characteri- zation in combined light. In case of absence of eclipsing planets, the effort toward characterization should once again be fully redirected to the spatially resolving track mentioned in the introduction. 11 The interest of a local sample is not only in terms of astronomical photon S/N; but also inspirational, for future generations (e.g., probe send- ing, see also Belu 2011). We are most thankful to Amaury Triaud, Michael Gillon, Roi Alonso & Monika Lendl for helping with this work. AB acknowledges support from CNES. FS acknowledges support from the European Research Council (ERC) star t- ing grant 209622: E3ARTHs. SNR thanks the CNRS's PNP program. PF acknowledges support from the ERC/European Community under the FP7 through Starting Grant agreement number 239953, as well as by Fundação para a Ciência e a Tecnologia (FCT) in the form of grant reference PTDC/CTE-AST/098528/2008. This research made use of www.solstation.com, Aladin and JSkyCalc. AB thanks J.-F. Lecampion for assistance in light curve production. We thank Ludovic Puig for contributing to the inspiration for this work. We thank the anonymous referee for attentive reading that helped improve the manuscript, and the numerous educational and prospective comments. REFERENCES Abbot & Switzer 2011, ApJ, 735L, 27 Andrews, S. M., Wilner, D. J.; Hughes, A. M., Qi, C., Dullemond, C. P. 2010, ApJ, 723, 1241 Apai, D., Pascucci, I., Bouwman, J., et al. 2005, Science, 310, 834 Artigau, E., Bouchard, S., Doyon, R., & Lafrenière, D. 2009, ApJ, 701, 1534 Artigau, E., Radigan, J., Folkes, S., et al. 2010, ApJ, 718L, 38 Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H . 2003, A&A, 402, 701 Beatty, J. T., Overmann, J., Lince, M. T., et al. 2005, PNAS, 102, 9306 Beaulieu, P., Kipping, D. M.; Batista, V. 2010, MNRAS, 409, 963 Beckwith, S. V. W. 2008, ApJ, 684, 1404 Barnes, R. & Heller, R. 2013, Astrobiology, accepted Belu, A. R., Selsis, F., Morales, J.-C. et al. 2011, A&A, 525A, 83 Belu, A. R. 2011, in Springer, Studies in Space Policy, 5, ESPI & ESF – Landfester, U,, Remuss, N.-L., Schrogl, K.-U., & Worms, J.-C. (eds.), 57-64 Berta, Z., Irwin, J., Charbonneau, D., Burke, C., & Falco, E. E. 2012, AJ, 144, 145 Blake, C. H., & Shaw, M. M., 2011, PASP, 123. 1302 Blake, C., Bloom, J. S., Latham, D. W., et al. 2008 PASP, 120 860 Bolmont, E., Raymond, S. N., & Leconte, J. 2011, A&A, 535A, 94 Bonfils, X., Delfosse, X., Udry, S., et al. 2011 arXiv:1111.5019 Burningham, B., Leggett, S. K., Lucas, P. W. et al. 2010, MNRAS, 404, 1952 Burgasser, A. J.; Tinney, C. G., Cushing, M. C., et al. 2008 ApJ 689L 53 Caballero, J. A. & Rebolo, R. 2002, F.Favata, I.W. Roxburgh & D. Galadí Enríquez eds., ESA SP-485 Caballero, J. A. 2010, Highlights of Spanish Astrophysics V, 79 Charbonneau, D., Irwin, J.; Nutzman, P., & Falco, E. E. 2008, AAS, 40, 242 11 Combined light spectroscopic characterization of non-eclipsing habita- ble super-Earths may be achievable with the JWST for extremely favorable (e.g., amongst other, most nearby) cases - see preliminary work by Selsis et al. (2011). Completeness of the screening for this population may be achieved by a mission like the NEAT proposal (Malbet et al. 2011) BELU ET AL. - HABITABLE EXOPLANETS AROUND BROWN DWARFS Charnoz, S., Salmon, J., Crida, A. 2010, Nature, 465, 752 Cockell, C. S., et al., 2009, Astrobiology, 9, 1 Cushing, M. C., Kirkpatrick, J. D., Gelino, C. R., et al. 2011, ApJ, 743, 50 Del Burgo, C., Martín, E. L., Zapatero Osorio, M. R., & Hauschildt, P. H. 2009, A&A, 501, 1059 Deming, D. L., Seager, S., Winn, J., et al. 2009, PASP, 121, 952 Eggl, S., Pilat-Lohinger, E., Georgakarakos, N., Gyergyovits, M., & Funk, B. 2012, ApJ, 752, 74 Faber, J. A., Rasio, F. A., & Willems, B. 2005, Icarus, 175, 248 Faherty, J. K., Burgasser, A. J.. Cruz, K. L. et al. 2009, AJ, 137, 1 Ford, E. B., & Rasio, F. A. 2006, ApJ, 638, L45 Geissler, K., Chauvin, G., & Sterzik, M. F. 2008, A&A, 480, 193 Grillmair, C. J., Burrows, A., Charbonneau, D., et al. 2008, Nature, 456, 767 Herbst, W., Eislöffel, J., Mundt, R., & Scholz, A. 2007, Protostars & Planets V, 297 Holman, M. J. & Wiegert P. A. 1999, AJ, 117, 621 Jackson, B., Barnes, R., & Greenberg, R. 2008, MNRAS, 391, 237 Jayawardhana, R., Mohanty, S., & Basri, G. 2003, ApJ , 592, 282 Joshi, M. 2003, Astrobiology 3-2, 415 Kaltenegger, L., & Traub, W. A. 2009, ApJ, 698, 519 Kasper, M., Biller, B. A.; Burrows, A., et al. 2007, A&A, 471, 655 King, R. R., McCaughrean, M. J., Homeier, D., et al. 2010, A&A, 510A, 99 Kirkpatrick, J., Cushing, M. C., Gelino, C. R., et al. 2011, ApJSS, 197, 19 Knapp, G. R., Leggett, S. K., Fan, X., et al. 2004, AJ, 127, 3553 Lammer, H., Selsis, F., Chassefière, E. et al. 2010, Astrobiology 10-1, 45 Leconte, J., Lai, D., & Chabrier, G. 2011, A&A, 528A, 41 Leggett, S. K., Burningham, B., Saumon, D. et al. 2010, ApJ, 710, 1627 Leggett, S. K., Cushing, M. C.; Saumon, D., et al. 2009, ApJ, 695, 1517 Leggett, S. K., Saumon, D., Marley, M. S., et al. 2012, ApJ, 748, 74 Lopez, B., Schneider, J., & Danchi, W. C. 2005, ApJ, 627, 974 Luhman, K. L., Adame, L., D’Alessio, P., et al. 2005, ApJ, 635, L93 Malbet, F., Léger, A., Shao, M., et al. 2011, ExA Online First, 109 Mohanty, S., Basri, G., Shu, F., Allard, F., & Chabrier, G. 2002, ApJ, 571, 469 Nisbet, E. G., Cann, J. R., & Van Dover, C. L. 1995, Nature, 373, 479 Paczyński, B. 1971, ARA&A, 9, 183 Pallé, E., Zapatero Osorio, M. R., & García Muñoz, A. 2011, ApJ, 728, 19 Pallé, E.; Zapatero Osorio, M. R., Barrena, R. 2009, Nature, 459, 814 Payne, M. J., & Lodato, G. 2007, MNRAS, 381, 1597 Preibisch, T., McCaughrean, M. J.; Grosso, N., et al. 2005, ApJS, 160, 582 Rauer, H., Gebauer, S.. Paris, P. V., et al. 2011, A&A, 529A, 8 Raymond, S. N., Scalo, J., & Meadows, V. S. 2007, ApJ, 669, 606 Ribas, I., Guinan, E. F., Güudel, M., & Audard, M. 2005, ApJ, 622, 680 Rosing, M. T. 2005, International Journal of Astrobiology, 4, 9 Saunders, E. S., Naylor, T., & Allan, A. 2008, AN, 329, 321 Scalo J., Kaltenegger, L., Segura, A., et al. 2007, Astrobiology, 7-1, 85 Selsis, F., Chazelas, B., Bordé, P., et al. 2007b, Icarus, 191, 453 Selsis, F., Kasting, J. F., Levrard, B., et al. 2007a, A&A, 476, 1373 Selsis, F., Wordsworth, R. D., & Forget, F. 2011, A&A, 532A, 1 Scholz R.-D., Bihain, G., Schnurr, O., & Storm, J. 2011, A&A, 532L, 5 Scholz R.-D., Storm, J., Knapp, G. R., & Zinnecker, H. 2009, A&A, 494, 949 Stephens, D. C., Leggett, S. K., Cushing, M. C., et al. 2009, ApJ, 702, 154 Stevenson, K. B., Harrington, J., Nymeyer, S., et al. 2010, Nature, 464, 1161 Struve, Otto, 1952, The Observatory, 72, 199 Swain, M. R., Vasisht, G., Tinetti, G., et al. 2009, ApJ, 690, L114 Tinetti, G., Vidal-Madjar, A., Liang, M.-C., et al. 2007, Nature, 448, 169 Traub, W., Shaklan, S. & Lawson, P. 2007, In the Spirit of Bernard Lyot, ed. P. Kalas Volk, K., Blum, R., Walker, G., & Puxley, P. 2003, IAU Circ., 8188, 2 von Braun, K., Kane, S. R., & Ciardi, D. R. et al. 2009, ApJ, 702, 779 Wordsworth, R., Forget, F., Selsis, F., et al. 2011, ApJ, 733L, 48 12
1105.4616	1	1105	2011-05-23T20:02:01	How common are Earth-Moon planetary systems?	[ "astro-ph.EP" ]	The Earth's comparatively massive moon, formed via a giant impact on the proto-Earth, has played an important role in the development of life on our planet, both in the history and strength of the ocean tides and in stabilizing the chaotic spin of our planet. Here we show that massive moons orbiting terrestrial planets are not rare. A large set of simulations by Morishima et al., 2010, where Earth-like planets in the habitable zone form, provides the raw simulation data for our study. We use limits on the collision parameters that may guarantee the formation of a circumplanetary disk after a protoplanet collision that could form a satellite and study the collision history and the long term evolution of the satellites qualitatively. In addition, we estimate and quantify the uncertainties in each step of our study. We find that giant impacts with the required energy and orbital parameters for producing a binary planetary system do occur with more than 1 in 12 terrestrial planets hosting a massive moon, with a low-end estimate of 1 in 45 and a high-end estimate of 1 in 4.	astro-ph.EP	astro-ph	How common are Earth-Moon planetary systems? S.Elsera,∗, B.Moorea, J.Stadela, R.Morishimab aUniversity of Zurich, Winterthurerstrasse 190, 8057 Zurich, Switzerland bLASP, University of Colorado, Boulder, Colorado 80303-7814, USA 1 1 0 2 y a M 3 2 . ] P E h p - o r t s a [ 1 v 6 1 6 4 . 5 0 1 1 : v i X r a Abstract The Earth's comparatively massive moon, formed via a giant impact on the proto-Earth, has played an important role in the development of life on our planet, both in the history and strength of the ocean tides and in stabilizing the chaotic spin of our planet. Here we show that massive moons orbiting terrestrial planets are not rare. A large set of simulations by Morishima et al. (2010), where Earth-like planets in the habitable zone form, provides the raw simulation data for our study. We use limits on the collision parameters that may guarantee the formation of a circumplanetary disk after a protoplanet collision that could form a satellite and study the collision history and the long term evolution of the satellites qualitatively. In addition, we estimate and quantify the uncertainties in each step of our study. We find that giant impacts with the required energy and orbital parameters for producing a binary planetary system do occur with more than 1 in 12 terrestrial planets hosting a massive moon, with a low-end estimate of 1 in 45 and a high-end estimate of 1 in 4. Keywords: Moon, Terrestrial planets, Planetary formation, Satellites, formation 1. Introduction The evolution and survival of life on a terrestrial planet requires several conditions. A planet orbiting the central star in its habitable zone provides the temperature suit- able for the existence of liquid water on the surface of the planet. In addition, a stable climate on timescales of more than a billion years may be essential to guarantee a suitable environment for life, particularly land-based life. Global climate is mostly influenced by the distribution of solar insolation (Milankovitch, 1941; Berger et al., 1984; Berger, 1989; Atobe and Ida, 2006). The annual-averaged insolation on the surface at a given latitude is, beside the distance to the star, strongly related to the tilt of the ro- tation axis of the planet relative to the normal of its orbit around the star, the obliquity. If the obliquity is close to 0◦, the poles become very cold due to negligible insolation and the direction of the heat flow is poleward. With in- creasing obliquity, the poles get more and more insolation during half of a year while the equatorial region becomes If the obliquity is larger than 57◦, colder twice a year. the poles get more annual insolation than the equator and the heat flow changes. Therefore, the equatorial region can even be covered by seasonal ice (Ward and Brown- lee, 2000). Thus, the obliquity has a strong influence on a planet's climate. The long-term evolution of the Earth's obliquity and the obliquity of the other terrestrial planets ∗Corresponding author Email addresses: [email protected] (S.Elser), [email protected] (B.Moore), [email protected] (J.Stadel), [email protected] (R.Morishima) in the solar system, or planets in general, is controlled by spin-orbit resonances and the tidal dissipation due to the host star and satellites of the planet. Thus, the evolution of the planetary obliquity is unique for each planet. Earth's obliquity fluctuates currently ± 1.3◦ around 23.3◦ with a period of ∼ 41, 000 years (Laskar and Robutel, 1993; Laskar, 1996). The existence of a massive (or close) satel- lite results in a higher precession frequency which avoids a spin-orbit resonance. Without the Moon, the obliquity of the Earth would suffer very large chaotic variations. The other terrestrial planets in the solar system have no massive satellites. Venus has a retrograde spin direction, whereas a possibly initial more prograde spin may have been influenced strongly by spin-orbit resonances and tidal effects (Goldreich and Peale, 1970; Laskar, 1996). Mars' obliquity oscillates ±10◦ degree around 25◦ with a period of several 100, 000 years (Ward, 1974; Ward and Rudy, 1991). Mercury on the other hand is so close to the sun that its rotation period is in an exact 3 : 2 resonance with its orbital period. Mercury's spin axis is aligned with its orbit normal. On larger timescales, the variation of the obliquity can be even more dramatic. It has been shown that the tilt of Mars' rotation axis ranges from 0◦ to 60◦ in less than 50 million years and 0◦ to 85◦ in the case of the obliquity of an Earth without the Moon (Laskar and Robutel, 1993; Laskar, 1996). The main purpose of this report is to explore the giant impact history of the planets in order to calculate the prob- ability of having a giant Moon-like satellite companion, based on simulations done by Morishima et al. (2010). A Preprint submitted to Elsevier August 2, 2018 giant impact between a planetary embryo called Theia, the Greek titan that gave birth to the Moon goddess Selene, first named by Halliday (2000), and the proto-Earth is the accepted model for the origin of our Moon (Hartmann and Davis, 1975; Cameron and Ward, 1976; Cameron and Benz, 1991), an event which took place within about 100 M yr after the formation of calcium aluminum-rich inclusions in chondritic meteoroids, the oldest dated material in the so- lar system (Touboul et al., 2007). After its formation, the Moon was much closer and the Earth was rotating more rapidly. The large initial tidal forces created high tidal waves several times per day, possibly promoting the cyclic replication of early bio-molecules (Lathe, 2004) and pro- foundly affecting the early evolution of life. Tidal energy dissipation has caused the Moon to slowly drift into its cur- rent position, but its exact orbital evolution is still part of an on-going debate (Varga et al., 2006; Lathe, 2006). Cal- culating the probability of life in the Universe (Ward and Brownlee, 2000) as well as the search for life around nearby planets may take into account the likelihood of having a massive companion satellite. This report is structured as follows: In section 2, we give a brief review on the evolution of simulating terres- trial planet formation with N-body codes during the last decades and present the method we used. In section 3, we study the different parameters of a protoplanet collision to identify potential satellite forming events. In section 4, we summarize the different uncertainties from the simulations and our analysis that may affect the final results. Finally, we give a conclusion, we present our results and compare them with previous works in section 5. 2. Simulating terrestrial planet formation There are good observational data on extra-solar gas gi- ant planets, but whilst statistics on extra-solar rocky plan- ets will be gathered in the coming years, for constraints on the formation of the terrestrial planets we rely on our own solar system. The established scenario for the forma- tion of the Earth and other rocky planets is that most of their masses were built up through the gravitational colli- sions and interactions of smaller bodies (Chamberlin, 1905; Safranov, 1969; Lissauer, 1993). Wetherill and Stewart (1989) observed the phase of run-away growth. This phase is characterized by the rapid growth of the largest bodies. While their mass increases, their gravitational cross sec- tion increases due to gravitational focusing. When a body reaches a certain mass, the velocities of close planetes- imals are enhanced, the gravitational focusing decreases and so does the accretion efficiency. This is called the oli- garchic growth phase, first described by (Kokubo and Ida, 1998). During this phase, the smaller embryos will grow faster than the larger ones. At the end, several bodies of comparable size are embedded in a planetesimal disk. These protoplanets merge via giant impacts to form the final planets. Dones and Tremaine (1993) showed that most of a terrestrial planet's prograde spin is imparted by the last major impactor and can not be accumulated via the ordered accretion of small planetesimals. Giant impacts with a certain impact angle and velocity gener- ate a disk of ejected material around the target which is a preliminary step in the formation of a satellite. Usu- ally, the simulations assume perfect accretion in a colli- sion. Tables of the collision outcome can help to improve the simulations or to estimate the errors in the planetary spin, (Kokubo and Genda, 2010). Until recently simula- tions were limited in the number of planetesimal bodies that could be self-consistently followed for time spans of up to billions of years, but recent algorithmic improve- ments by Duncan et al. (1998) in his SyMBA code and by Chambers (1998) in his Mercury code have allowed them to follow over long time spans a relatively large number of bodies (O(1000)) with high precision, particularly during close encounters and mergers between the bodies, where individual orbits must be carefully integrated (Chambers and Wetherill, 1998; Agnor et al., 1999; Raymond et al., 2004; Kokubo et al., 2006). Raymond et al. (2009) have also recently conducted a series of simulations where they varied the initial conditions for the gas giant planets and also track the accretion of volatile-rich bodies from the outer asteroid belt, leading either to "dust bowl" terres- trial planets or "water worlds" and everything between these extremes. All prior simulation methods with full in- teraction among all particles have however been limited in number of particles since their force calculations scale as O(N 2). We have developed a new parallel gravity code that can follow the collisional growth of planetesimals and the subsequent long-term evolution and stability of the result- ing planetary system. The simulation code is based on an O(N ) fast multipole method to calculate the mutual gravitational interactions, while at the same time follow- ing nearby particles with a highly accurate mixed vari- able symplectic integrator, which is similar to the SyMBA (Duncan et al., 1998) algorithm. Since this is completely integrated into the parallel code PKDGRAV2 (Stadel, 2001), a large speed-up from parallel computation can also be achieved. We detect collisions self-consistently and also model all possible effects of gas in a laminar disk: aero- dynamic gas drag, disk-planet interaction including Type- I migration, and the global disk potential which causes inward migration of secular resonances with gas dissipa- tion. In contrast to previously mentioned studies, this code allows us to self-consistently integrate through the last two phases of planet formation with the same numer- ical method while using a large number of particles. Using this new simulation code we have carried out 64 simulations which explore sensitivity to the initial condi- tions, including the timescale for the dissipation of the so- lar nebula, the initial mass and radial distribution of plan- etesimals and the orbits of Jupiter and Saturn (Morishima et al., 2010). All simulations start with 2000 equal-mass particles placed between 0.5 and 4 AU . The initial mass of the planetesimal disk md is 5 or 10 m⊕. The surface 2 density Σ of this disk and of the initial gas disk depends on the radius through Σ ∝ r−p, where p is 1 or 2. The gas disk dissipates exponentially in time and uniformly in space with a gas dissipation time scale τgas = 1, 2, 3 or 5 M yr. After the disappearance of the gas disk (more pre- cisely after time τgas from the beginning of the simulation), Jupiter and Saturn are introduced on their orbits, e.g. cir- cular orbits or the current orbits with higher eccentricities. Figure 1 shows two merger trees: a 1.8 m⊕ planet formed in the simulation with (τgas,p,md)=(1 Myr,1,5 m⊕) and gas giants on the present orbits and a 1.1 m⊕ planet formed in the simulation with (τgas,p,md)=(1 Myr,2,10 m⊕) and gas giants on circular orbits. The red branches are satellite forming impactors both with a mass 0.3 m⊕ and represent two events of the final sample in figure 7. These merger trees with their different morphologies reveal the variety of collision sequences in terrestrial planet forma- tion. They show that a large set of impact histories is generated by these simulations despite the relatively nar- row parameter space for the initial conditions. 3. Satellite formation During the last phase of terrestrial planet formation, the giant impact phase, satellites form. Collisions between planetary embryos deposit a large amount of energy into the colliding bodies and large parts of them heat up to sev- eral 103 K, e.g. Canup (2004). Depending on the impact angle and velocity and the involved masses, hot molten material from the target and impactor can be ejected into an circumplanetary orbit. This forms a disk of ejecta, the disk material is in a partially vapor or partially molten state, around the target planet. The proto-satellite disk cools and solidifies. Solid debris form and subsequently agglomerate into a satellite (Ohtsuki, 1993; Canup and Esposito, 1996; Kokubo et al., 2000). The giant impact which resulted in the Earth-Moon system is a very particular event (Cameron and Benz, 1991; Canup, 2004). The collision parameter space that describes a giant impact can by parametrized by γ = mi/mtot, the ratio of impactor mass mi to total mass in the collision mtot, by v ≡ vimp/vesc, the impact velocity in units of the escape velocity vesc =(cid:112)2Gmtot/(ri + rt), where rt and ri are the radii of target and impactor. Fur- thermore, it is described by the scaled impact parameter b, where b = 0 indicates a head-on collision and b = 1 a graz- ing encounter, and the total angular momentum L. Recent numerical results (Canup, 2008) obtained with smoothed particle hydrodynamic (SPH) simulations for the Moon- forming impact parameters require: γ ∼ 0.11, v ∼ 1.1, b ∼ 0.7 and L ∼ 1.1 LEM, where LEM is the angular mo- mentum of the present Earth-Moon system. These sim- ulations also include the effect of the initial spins of the colliding bodies, but the explored parameter space is re- stricted to being close to the Moon-forming values given above. Figure 1: Two merger trees. They illustrate the accretion from the initial planetesimals to the last major impactors that merge with the planet. Every 'knee' is a collision of two particles and the length between two collisions is given by the logarithm of the time between impacts. The thickness of the lines indicates the mass of the particle (linear scale). The red branch is the identified satellite forming im- pact in the planet's accretion history. Top: a 1.1 m⊕ planet formed in the simulation with (τgas,p,md)=(1 Myr,2,10 m⊕) and gas giants on circular orbits and a 0.3 m⊕ impactor. In this case, the moon forming impact is not the last collision event but it is followed by some major impacts. Right: a 0.7 m⊕ planet formed in the simula- tion with (τgas,p,md)=(3 Myr,1,5 m⊕) and gas giants on the present orbits and a 0.2 m⊕ impactor. It is easy to see that the moon forming impact is the last major impact on the planet. Although this planet is smaller than the upper one, it is composed out of a similar number of particles. In the case of the more massive disk (md = 10), the initial planetesimals are more massive since their number is constant. Therefore, fewer particles are needed to form a planet of comparable mass than in the case of md = 5. 3 If one does not focus on a strongly constrained system like the Earth-Moon system but just on terrestrial planets of arbitrary mass with satellites that tend to stabilize their spin axis, the parameter space is broadened. It becomes difficult to draw strict limits on the parameters because collision simulations for a wider range of impacts were not available for our study. Hence, based on published Moon- forming SPH simulations by Canup (2004, 2008), we use a semi-analytic expression to constrain the mass of a circum- planetary disk that can form a satellite. In addition, we include tidal evolution and study the ability of the satellite to stabilize the spin axis of the planet. 3.1. Satellite mass and collision parameters We do not know the exact outcome of a protoplanet collision, but certainly the satellite mass is related to the collision parameters of the giant impact. Hence, we can draw a connection from these parameters to the mass of the final satellite. Based on the studies of the Earth's Moon formation, we can start with a simple scaling re- lation: a Mars-size impactor gives birth to a Moon-size satellite. Their mass ratio is mMars/mMoon ∼ 10. Thus, to very first approximation, we can assume that an im- pactor mass is usually 10 times larger than the final mass of the satellite. Of course, this shows that only a small amount of material ends up in a satellite, but this state- ment is only valid for a certain combination of mass ratio, impact parameter and impact speed and usually gives an upper limit on the satellite mass. In order to get a better estimation of the satellite mass, we use the method obtained in the appendix of Canup (2008). There, an expression is derived that describes the mass of the material that enters the orbit around the target after a giant impact: (cid:18) mpass (cid:19)2 mtot mdisk mtot ∼ Cγ where the prefactor Cγ ∼ 2.8 (0.1/γ)1.25 has been deter- mined empirically from the SPH data. mpass/mtot is mass of the impactor that avoids direct collision with the tar- get. It depends mainly on the impact parameter b and on the mass ratio γ and can be computed by studying the geometry of the collision. The total impactor volume that collides with the target is: (cid:90) π VT = A(φ) dφ, 0 with A(φ) = r2 t θt(φ) + r2 i θi(φ) − Dri sin φ sin θi(φ), (1) (2) (3) where ri and rt are impactor and target radius, D = b(ri + rt) gives the distance between the centers of the bodies and (cid:20) D2 + r2 i sin2 φ − r2 t 2Dri sin φ (cid:21) θi(φ) = cos−1 , (4) 4 (cid:20) D2 + r2 t − r2 2Drt i sin2 φ (cid:21) . (5) θt(φ) = cos−1 Assuming a differentiated impactor with rcore ∼ 0.5ri and repeating the above integration for this radius, the colliding volume of the impactor mantle is Vmantle = VT − Vcore. If we assume that the core is iron and the mantle dunite, the core density ρcore has roughly twice the density of the mantle ρmantle. The mass of the impactor that hits the target is mhit = ρcoreVcore + ρmantleVmantle. Therefore, the mass that passes the target is mpass = mi − mhit. We use the full expression derived by Canup (2008), equation (1), to estimate the disk mass resulting from the giant impacts in our simulation. Equation (1) is correct to within a factor 2, if v < 1.4 and 0.4 < b < 0.7 or if v < 1.1 and 0.4 < b < 0.8. Figure 2 illustrates the amount of material that is transported into orbit for the parameter range 0.4 < b < 0.8 (see figure B2 in Canup (2008) for more details). It shows that a small impact parameter b reduces the material ejected into orbit significantly. The same holds for a reduction of γ, because those collisions are more grazing. Based of simple arguments, we use the limits above to identify the moon forming collision in the (v, b)-plane. Details on how this assumption affects the result are given in section 4. Ida et al. (1997) and Kokubo et al. (2000) studied the formation of a moon in a circumplanetary disk through N- body simulations. They found that the final satellite mass scales linearly with the specific angular momentum of the disk. The fraction of the disk material that is finally in- corporated into the satellite ranges form 10 to 55%. Thus, we assume that not more than half of the disk material is accumulated into a single satellite. The angular momen- tum of the disk is unknown and we can not use the more exact relationships. 3.2. Spin-orbit resonance Spin-orbit resonance occurs when the spin precession frequency of a planet is close to one of the planet's or- bital precession frequencies. It causes large variation in the obliquity (Laskar, 1996), the angle between this spin axis and the normal of the planet's orbital plane. An obliquity stabilizing satellite increases the spin precession frequency to a non-resonant (spin-orbit) regime. To ensure this, one can set a rough limit on the system parameters (Atobe et al., 2004) through (cid:29) m∗ a3 p ms a3 s (6) , where ms is the mass of the satellite, m∗ the mass of the central star and as the semi-major axis of the satellite's orbit and ap the semi-major axis of the planet. If the left term of the inequality is much larger than the right one, the spin precession frequency of the planet should be high enough to ensure that it is over the up- per limit of the orbital precession frequency so that spin- orbit resonance does not occur. Although this inequality 1 AU , we get a mass ratio of ∼ 10−5, which is smaller than the minimum ratio of the smallest and largest particles in our simulations. This is a lower limit on the stabilizing satellite mass but the tidal evolution of the planet-satellite system can alter this limit dramatically. The orbital precession frequencies of a planet depend on the neighbouring or massive planets in its system. In the case of the Earth, Venus, Jupiter and Saturn cause the most important effects. To keep the spin precession frequency high enough, a spin period below 12 h would have the same effect as the present day Moon (Laskar and Robutel, 1993). Hence, even without a massive satellite, obliquity stabilization is possible as long as the planet is spinning fast enough. However, the moon forming impact provides often a significant amount of angular momentum. A more sophisticated analysis including the precession fre- quencies of the all planets involved or formed in the simu- lations and a better treatment of the collisions to provide better estimates of the planetary spins would clearly be an improvement but this is out of the scope of this work. 3.3. Tidal evolution After its formation, even a small satellite is stabiliz- ing the planet's obliquity. Its fate is mainly controlled by the spin of the planet, the orientation of the spin axis of the planet relative to its orbit around the central star and relative to the orbital plane of the satellite and by possible spin-orbit resonance. Which satellites will con- tinue to stabilize the obliquity as they recede from the planet? Orbital evolution is a complicated issue (Atobe and Ida, 2006) and there is still an ongoing debate even in the case of the Earth-Moon system, as mentioned. To classify the different orbital evolutions, it is helpful to in- troduce the synchronous radius, at which a circular or- bital period equals the rotation period of the planet. In the prograde case, a satellite outwards of the synchronous radius will recede from the planet as angular momentum is tidally transferred from the planet to the satellite and the spin frequency of the planet decreases. In this case the synchronous radius will grow till it equals the satel- lite orbit. Angular momentum is transferred faster if the mass of the satellite is large, since the tidal response in the planet due to the satellite is greater. Large mass satellites will quickly reach this final co-rotation radius, where their recession stops. Even though their orbital radius becomes larger, these moons are massive enough to satisfy inequal- ity (6) and avoid spin-orbit resonances. Small satellites will recede very slowly compared to their heavy broth- ers and eventually fulfill the condition (6) within the host star's main sequence life time. In contrast, low mass satel- lites that form in situ far outside the Roche limit, should this be possible, will probably not stabilize the spin axis. Intermediate mass satellites with ms ∼ mMoon may re- cede fast enough so that they tend to lose their obliquity stabilizing effect during a main sequence life time. The Earth-Moon system shows that even in this intermediate Figure 2: The disk mass resulting after a giant impact in units of the total mass of the colliding system relative to the impact parameter b based on equation (1). The different solid lines belong to different mass ratios γ. From bottom to top: γ = 0.05, 0.1, 0.15, 0.2, 0.3 and 0.5. These mass ratios result in different disk masses. The equation is valid up to a factor 2 in between the two dash lines (0.4 < b < 0.8) for small velocities. The disk mass is an upper bound on the satellite mass. is very simplified, we try to estimate the minimum mass of a satellite such that it is able to stabilize the obliquity of its planet. The exact semi-major axis of a satellite after forma- tion is unknown but the Roche limit is the lower bound of its semi major axis. The Roche limit aR of the planet is (Murray and Dermott, 1999): aR = rs , (7) (cid:18) 3mp (cid:19) 1 3 ms where rs is the radius of the satellite and ms its mass and the mass of the planet is given by mp. Ohtsuki (1993) and Canup and Esposito (1996) provided detailed ana- lytic treatments of the accretion process of satellites in an impact-generated disk. Based on those studies, Kokubo et al. (2000) have shown that the true value of the ra- dius of satellite accretion will not diverge much from the Roche radius in the case of the Earth-Moon system, a typical satellite orbit semi-major axis in their simulations was a (cid:39) 1.3 aR. We used this approximation to estimate a lower bound on the satellite-planet mass ratio. We rewrite the Roche limit as (cid:18) 3 (cid:19) 2 3(cid:18) mp (cid:19) 1 3 2 aR = (8) πρp 3 ρs)−1ms and the fact that ρp = ρs s = ( 4π where we used r3 in our model. We insert this in equation (6) instead of as and get the condition ms mp (cid:29) 9m∗ 4πρpa3 p . (9) Inserting a density of 2 g cm−2, the density of the bodies in the Morishima simulations, and planet semi-major axis of 5 0.10.20.30.40.50.60.70.80.9b0.000.020.040.060.080.10mdisk/mtotmass regime, long term stability can occur, since the Moon has stabilized the Earth's obliquity for billion of years. On the other hand, a satellite inside the synchronous orbit will start to spiral towards the planet, while its an- gular momentum is transferred to the planet. Soon, it is disrupted by tidal forces or will crash on the planet. How do we know if a satellite forms inside or outside the syn- chronous radius? The Roche limit (7) depends on the mass and size of the bodies while the synchronous radius rsync is a function of the planet mass mp and depends inversely proportional on its rotation frequency, which is obtained from equating the gravitational acceleration and the cen- tripetal acceleration: rsync = (Gmp) 1 − 2 3 ω p 3 . (10) Equating this formula with (7) gives: (cid:18) 3 (cid:19) 2 3(cid:18) mp (cid:19) 1 3 = (Gmp) 1 − 2 3 ω p 3 . (11) 2 πρp The planet mass drops out and we get an lower limit on the planet angular velocity to guarantee a satellite outside the Roche radius: ωp,min = 2 3 (πGρp) 1 2 = 0.00043 s−1, (12) which equals a rotation period of 4 h. The final planets of Morishima et al. (2010) have generally a high rotational speed, some of them rotating above break up speed. The rotation period after the moon forming collision is usu- ally around 2-5 hours, see figure 3. A more exact parti- cle growth model without the assumption of perfect stick- ing would lower the rotation frequency by roughly 30% (Kokubo and Genda, 2010), where we can also include the loss of rotational angular momentum due to satellite formation (full circles). On the other hand, the mean dis- tance of satellite formation is around 1.3 aR, and the max- imum rotation period of the planet for a receding satellite changes to ∼ 6 h (dashed line). Hence, applying this to our final sample, 1/4 of all moon forming collisions are excluded. Finally, for the remaining events, we assume that the planet spin is large enough so that the synchronous radius is initially smaller than the Roche radius. Satellites that form behind the Roche limit will start to recede from the planet. We can conclude that almost every satellite-planet system in our simulation will fulfill (6). A special outcome of the tidal evolution of a prograde planet-satellite system is described by Atobe and Ida (2006). If the initial obliquity θ of the planet after the moon form- ing impact is large, meaning that the angle between planet spin axis and planet orbit normal is close to 90◦, results in a very rapid evolution when compared to the previ- ously discussed case of a moon receding to the co-rotating radius and becoming tidally locked. The spin vectors of the protoplanets are isotropically distributed after the gi- ant impact phase (Agnor et al., 1999) and the obliquity Figure 3: The mass of the planet in units of its final mass mtot/mtot,final relative to the rotational period in days of the body after the moon-forming collision. This is the final sample but with- out excluding moon formation inside the synchronous radius. Empty circles: The decrease of the planet spin due to escaping material or the formation of a satellite are not included in the calculation of the rotation period, it is based on the perfect accretion assumption. Full circles: The angular momentum is assumed to be 30% smaller due to a more realistic collision model (Kokubo and Genda, 2010). The dotted line gives the position of the threshold for synchronous rota- tion for the Roche radius aR, the dashed line gives its position for the 1.3 aR. We want to focus on the better estimate of the anuglar momentum (full circles) and the initial moon radius (dashed line). Hence, roughly a quarter of all moon forming collisions are excluded in the final sample, figure 7. π 2 θ < − π 0.2 distribution corresponds to p(θ) = 1 2 sin(θ). In the most extreme scenario, a massive prograde satellite will crash onto the planet in a timescale of order 10,000 years after formation, even though it initially recedes. Hence, without a favorable initial obliquity is an added requirement for the survivability of a close or massive moon. We include this by discarding massive and highly oblique impacts by an approximated inequality, based on area B in figure 13 in Atobe and Ida (2006): ms mp (13) More exactly, this holds best for planets with ap ∼ 1 AU , and it has only to be taken into account if ms/mp < 0.05. The distribution of the ap of the simulated planets ranges for 0.1 to 4 AU. In the case of a smaller distance to the host star, angular momentum is removed faster from the planet- satellite system and the evolution timescales are shorter in general. To use this limit properly, a more general expres- sion for different semi-major axes has to be derived, which is out of the scope of this work. Moreover, it depends linearly on the satellite mass which is overestimated in general. A smaller satellite mass would reduce the num- ber of excluded events in general. However, when applying this constraint on the data, only one out of seven moon forming collisions are affected. Since it is over-simplified, we exclude it from our analysis. For retrograde impacts, where the impactor hits in op- position to the target's spin, two cases result in differing evolution. If the angular momentum of the collision is 6 (cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:231)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)0.20.40.60.81.0051015mtotmtot,finalrotationperioddmuch larger than the initial rotational angular momen- tum, the spin direction of the planet is reversed and any impact-generated disk will rotate in the same direction as the planet's spin. On the other hand, a retrograde collision with small angular momentum will not alter the spin di- rection of the planet significantly and it becomes possible to be left with a retrograde circumplanetary disk(Canup, 2008). After accretion of a satellite, tidal deceleration due to the retrograde protoplanet spin will reduce the orbital radius of the bodies continuously till they merge with the planet. Hence, a long-lived satellite can hardly form in this case. In order to find a reasonable threshold between these two regimes based on the limited information we have, we assume that the sum of the initial angular momenta of the bodies (cid:126)Lt and (cid:126)Li and of the collision (cid:126)Lcol equals the spin angular momentum of the planet and the satellite (cid:126)Lplanet and (cid:126)Lmoon after the collision plus the angular momentum of the orbiting satellite (cid:126)Lorbit at 1.3 aR parallel to the col- lision angular momentum. Comparing initial and final angular momenta gives: (cid:126)Lcol + (cid:126)Lt + (cid:126)Li = (cid:126)Lorbit + (cid:126)Lmoon + (cid:126)Lplanet, (14) where (cid:126)Lorbit is parallel to (cid:126)Lcol: (cid:126)Lorbit = (cid:126)Lorbit (cid:126)Lcol (cid:126)Lcol . (cid:115) Lorbit = msa2 We assume that (cid:126)Lmoon/((cid:126)Lorbit + (cid:126)Lplanet) (cid:28) 1 and (cid:112)Gasmp, (16) mass is much smaller than the planet mass, n ∼(cid:112)Gmp/a3 where n is the orbital mean motion of the satellite if its s, and G is the gravitational constant. With as = aR we get sn = msa2 s = ms Gmp a3 s (cid:19) 1 6 (cid:18) 3 1 ρp 2π mp . (17) Lorbit = ms(Gmp) 1 2 If (cid:112)(L2 (cid:126)Lcol + (cid:126)Lt col + L2 t ) < 1, (15) (18) (19) the collision is retrograde. Hence, (cid:126)Lplanet = (cid:126)Lcol + (cid:126)Lt + Li − (cid:126)Lorbit (cid:126)Lcol (cid:126)Lcol , If (cid:126)Lplanet and (cid:126)Lorbit, which is parallel to (cid:126)Lcol, are retro- grade, < 1, (20) (cid:112)(L2 (cid:126)Lcol + (cid:126)Lorbit orbit) col + L2 the satellite will be tidally decelerated. Those cases are excluded from being moon-forming events. Spin angular momenta are in general overestimated since material is 7 lost during collisions in general, but as mentioned above, the simulations of Morishima et al. (2010) assume perfect accretion. We include this consideration by reducing the involved spins and the orbit angular momentum of the satellite by 30%, (Kokubo and Genda, 2010). Both scenarios described above, a large initial obliquity or a retrograde orbiting planet, might not become impor- tant until subsequent impactors hit the target. Giant im- pacts can change the spin state of the planet in such a way that the satellite's fate is to crash on the planet. This scenario is discussed in the next section. 3.4. Collisional history We exclude all collisions from being satellite-forming impacts whose target is not one of the final planets of a simulation. A satellite orbiting an impact is lost through the collision with the larger target. Multiple giant impacts occur during the formation pro- cess of a planet and it is useful to study the impact his- tory in more detail. Subsequent collisions and accretion events on the planet after the satellite-forming event may have a large effect on the final outcome of the system. We choose a limit of 5mplanetesimal to distinguish between large impacts and impacts of small particles, which are responsible for the ordered accretion. To stay consistent, the same limit is used below to exclude small impactors from our analysis. We divide all identified moon forming events into four groups: a) The moon forming event is the last major impact on the planet. Subsequent mass growth happens basi- cally through planetesimal accretion (see merger tree at the bottom in figure 1). b) There are several moon forming impacts, in which the last impact is the last major impact on the planet. c) The moon forming event is not the last giant im- pact on the planet. The satellite can be lost due to a disruptive near or head-on-collision (Stewart and Leinhardt, 2009) of an impactor and the satellite. A late giant impact on the planet can change the spin axis of the planet and the existing satellite can get lost due to tidal effects (see tree at the top in figure 1). d) There are several moon forming events in the impact history of the planet, followed by additional major impacts. As before, the moon forming collisions can remove previously formed satellites. On the other hand, and existing moon can have an influence on the circumplanetary disk formed by a giant impact and can suppress the formation of multiple satellites orbiting the planet. The final states of the planet-satellite systems in group c are difficult to estimate, since such systems might change significantly by additional giant impacts. To a lesser ex- tent, this holds for d, but those systems are probably more Figure 4: This bar chart shows the distribution of the moon forming collisions in four groups with different impact history with respect to the last major impact. a: the moon forming impact is the last major impact on the planet. b: there are multiple moon forming impacts, but the last one is also the last major impact. c: there is only one moon forming impact and it is followed by subsequent major impacts. d: there are multiple moon forming impacts, but the last of them is followed by subsequent major impacts. resilient to the loss of satellites by direct collisions. The number of events per group is shown in figure 4. Group c and d include more then 2/3 of all collisions. Group c is the most uncertain and we use it to quantify the error on the final sample. Furthermore, we exclude impactors and targets that have masses of the order of an initial planetesimal mass (5 mplanetesimal ∼ 0.0025 m⊕) from producing satellite form- ing events since their masses are discretized and related to the resolution of the simulation. In addition, small im- pactors will probably not have enough energy to eject a significant amount of material into a stable orbit. There- fore, setting a lower limit on the target and impactor mass results in the exclusion of many collisions but not in a sig- nificant underestimation of the true number of satellites, see figure 5. 4. Uncertainties Our final result depends on several assumptions, lim- itations and approximations. In this section we want to quantify them as much as possible and summarize them. The data of Morishima et al. (2010) we are using has two peculiarities worth mentioning: The focus on the Solar System and the small number of simulations per set of initial conditions. The simulations were made in order to reproduce the terrestrial planets of the Solar System. The central star has 1 solar mass and the two gas giants that are intro- duced after the gas dissipation time scale have the mass of Jupiter and Saturn and the same or similar orbital el- ements. Although the initial conditions like initial disk mass or gas dissipation time scale are varied in a certain range, the simulated systems do not represent general sys- tems with terrestrial planets. However, we assume that 8 Figure 5: The target mass mt versus the impactor mass mi in- volved in the satellite-forming collisions. We exclude events that include impactors and targets with masses of the order of the ini- tial planetesimal mass (mi, mt < 5 mplanetesimal) indicated by the dashed line to avoid resolution effects. The line in the case of the target mass is not shown since it is very close the frame. Due to this cut, the total number of satellites decreases significantly while the number of massive satellites in our analysis increases. The shown sample is the final set of events without applying the threshold for small particles. Therefore, the number of accepted events in this plot does not equal the number of satellites in figure 7, since this cut is not applied. If the cut is used, new events are accepted, which where neglected before because they were not the last moon forming impacts on the target. the range of impact histories is representative of the range that would be seen in other systems. Merger trees (figure 1) reveal that those simulations cover a huge diversity of impact histories. But future work will need to investigate the full range of impact histories that could be relevant to the formation of terrestrial planets in other, possibly more exotic, extra-solar systems. However, the set of simulation covers a broad range of initial conditions. But for every set of initial condi- tions, only one simulation exists since they are very time consuming (each simulation requires about 4 months of a quad-core CPU). Therefore, it is difficult to separate ef- fects of the choice of certain initial parameters from effects of stochastic processes. We grouped our moon forming events with respect to the simulation parameter in ques- tion. The only parameter that reveals an effect on the final sample is the gas dissipation time scale τgas. Larger time scales lead to less moon forming collisions. This variation is correlated with mass and number of final planets. If the gas disk stays for several million years, the bodies are affected by the gas drag for a long time, spiral towards the sun and get destroyed. Therefore, there are less giant impacts and smaller planets. One would suppose that the initial mass of the gas disk should be correlated to the mass of the final planets and therefore to the number of giant impacts, but the two initial protoplanetary disk masses of 5 m⊕ and 10 m⊕ show no essential difference. The approach we use to identify events and estimate the mass of the satellite is also based on various approxi- mations and limitations. abcd51015202530(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)0.00.10.20.30.40.50.00.51.01.52.0mim(cid:197)mtm(cid:197)Satellite mass. The method we use to calculate the mass of the circumplanetary disk is valid to better than a factor of 2 within the parameter range we use (Canup, 2008). Only 10-55% of the disk mass are embodied in the satellite (Kokubo et al., 2000). Therefore, the mass we use is just half of the estimated disk mass, and in the worst case, the satellite is ten times less massive than estimated. This uncertainty affects the number of massive satellites but not the number of satellites in general. In the (v,b)-plane, we use the same restrictive limits to constrain the collision events. Figure 6 presents all moon forming events and this parameter range. Hence, this area gives just a lower bound. We see that it covers some of the most populated parts, but a significant amount of the collisions are situated outside this area. Equation (1) can help to constrain the outcome of the collisions close outside of the shaded region. A collision with a small impact pa- rameter (b < 0.4) will bring very little material into orbit, whatever the impactor mass involved. In this regime, we have few events with high impact velocity and even with high velocity it might be very hard to eject a significant amount of material into orbit. Hence, we exclude them from being moon forming events. For intermediate impact parameter (0.4 < b < 0.8), there are high velocity events (v > 1.4). Above a certain velocity threshold, depending on b and γ, most material ejected by the impact will es- cape the system and the disk mass might be too small to form a satellite of interest. A large parameter (b > 0.8) describes a highly grazing collision. It is difficult to extent equation (1) for larger b, since these collisions will proba- bly result in a hit-and-run events for high velocities. SPH simulations (Canup, 2004, 2008) suggest that high rela- tive impact velocities (v > 1.4) will increase rapidly the amount of material that escapes the system. Neverthe- less, these sets of simulations focus not on general impacts and the multi-dimensional collision parameter space is not studied well enough to describe those collisions in more detail. Detailed studies of particle collisions will hopefully be published in the near future (e.g. Kokubo and Genda (2010)). Based on the arguments above, the events inside of the shaded area form our final sample. Most of the colli- sions outside the area will not form a moon. The collisions with 0.4 < b < 0.8 and velocity slightly above v = 1.4 and with b > 0.8 and small velocity (v < 1.4) are events with unknown outcome. Including those events, the final num- ber of possible moon forming events is increased by not more than a factor of 1.5. Tidal evolution - Rotation period. In order to separate re- ceding satellites from satellites which are decelerated after their formation, we check if the initial semi-major axis of the satellite is situated outside or inside of the syn- chronous radius of the planet. In figure 3, the sample is plotted twice, once including a general correction for an- gular momentum loss due to realistic collisions and once without correction. Moreover, two thresholds for the syn- chronous radius are shown. We choose the threshold at 9 1.3aR (dashed line) and the corrected rotation period (full circles) to be the most justified case. There are two most extreme cases: the threshold situated at 1aR (dotted line) and a corrected rotation period (full circles) indicates that only one in five collisions lead to receding moons. On the other hand, the threshold situated at 1.3 aR and a rota- tion period directly obtained form the simulations (empty circles) indicates that only one in eight collisions lead to a non-receding moon. Tidal evolution - Retrograde satellites. To exclude retro- grade orbiting satellites, we use a simple relation between the angular momenta involved in a collision. Since we have no exact data of the angular momentum distribution af- ter the collision, this limit is very approximate. To get an estimate of the quality of this angular momentum ar- gument, we study the effect of this threshold on the final sample. First, roughly half of all moon-forming impacts are retrograde. But in almost every case, Lorbit is much smaller than the initial angular momentum. Only four of the retrograde collisions have not enough angular momen- tum to provide a significant change of the spin axis and those planet-satellite systems remain retrograde. There- fore, our final estimate is not very sensitive to this angular momentum argument. A larger set of particle collision simulations could provide better insight into the angular momentum distribution. Collision history. Two issues affecting the collision history can change the final result. These being the mass cut we choose to avoid resolution effects, where we exclude small impactors and targets, and the uncertainty about the fi- nal state of the planet-satellite system because of multiple and subsequent major impacts. The first limit seems to be well justified. If we change the threshold mass or exclude the cut completely, we change mainly the number of very small, non-satellite forming, impacts since in these cases there is usually insufficient energy to bring a significant amount of material into orbit. The second issue has a sig- nificant effect on the final sample, however. Group c, the single moon forming collisions followed by large impacts, contains the most uncertain set of events: Neither all satel- lites survive the subsequent collisions nor is it likely that all satellites get lost. Hence, in the extreme case, almost half of all initially formed moons, all of Group c, could be lost and our final result reduced significantly. An overview of the above uncertainties in given in ta- ble 1. Additional planet formation simulations are nec- essary to quantify a large part of existing uncertainties. Simulations of protoplanet collisions exploring the multi- dimensional collision parameter space are desirable and will hopefully be published soon and might help to con- strain the parameters for moon formation better. A study of the effect of subsequent accretion of giant bodies after a moon forming impact might give better insight in the evolution of a planet-satellite system. Uncertainty factor Range of initial parameters Number of simulations Collision parameter Tidal evolution - Rotation Tidal evolution - Retrograde orbit Collision history Satellite mass - disk mass Satellite mass - accretion efficiency high high 1-1.5 0.5-2 ∼ 1 0.5-1 0.5-2 0.2-1 Table 1: A list of the different conditions that affect the final number of satellites. The uncertainty factor gives the range in which the final result varies around the most justified. 1 equals the final value. The first two factors can not be estimated, it shows that additional simulations would be helpful. The estimation of the satellite mass is separated from the rest of the list, since the uncertainty on this estimate does not change the number of satellites in the final sample, in contrast to the others. The exclusion of retrograde satellites is very approximate, but it has almost no effect on the final sample and therefore this factor is close to 1. 5. Discussion and results Under these restrictive conditions we identify 88 moon forming events in 64 simulations, the masses of the re- sulting planet-satellite systems are shown in figure 7. On average, every simulation gives three terrestrial planets with different masses and orbital characteristics and we have roughly 180 planets in total. Hence, almost one in two planets has an obliquity stabilizing satellite in its or- bit. If we focus on Earth-Moon like systems, where we have a massive planet with a final mass larger than half of an Earth mass and a satellite larger than half a Lunar mass, we identify 15 moon forming collisions. Therefore, 1 in 12 terrestrial planets is hosting a massive moon. The main source of uncertainties results from the modelling of the collision outcomes and evolution of the planet-satellite system as well as the small number of simulation and the limited range of initial conditions. We do not include the latter in our estimate. Hence, we expect the total number of Earth-Moon like systems in all our simulations to be in a range from 4 to 45. This results in a low-end estimate of 1 in 45 and a high-end estimate of 1 in 4. In addition, taking into account the uncertainties on the estimation of the satellite mass, roughly 60 of those systems are formed in the best case or almost no such massive satellites are formed if the efficiency of the satellite accretion in the cir- cumplanetary disk is very low. There are several papers, where the authors performed N-body simulations and searched for moon-forming colli- sions. Agnor et al. (1999) started with 22-50 planetary embryos in a narrow disk centered at 1 AU. They esti- mated around 2 potentially moon-forming collisions per simulation, where the total angular momentum of the en- counter exceeds the angular momentum of the Earth-Moon system. They pointed out that this number is somewhat sensitive to the number, spacing, and masses of the ini- tial embryos. O'Brien et al. (2006) performed simula- 10 Figure 6: The parameter space in the (b, vimp/vesc)-plane. The darker region gives the region for with equation (1) holds up to a factor of 2. The shown sample is the final sample without the con- straints in the (b, v)-plane. Similar to figure 5, the number of ac- cepted events will increase when applying the cut, since an earlier moon forming impact on the planet can become suitable. As one expects, there are more collisions with larger impact parameter. Ig- noring events with smaller b or higher v (see text), there is still a significant group of events with b > 0.8. Much material escapes in the high velocity cases and no moon will form, but the low velocities are hard to exclude or calculate. (cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)0.00.20.40.60.81.012345bvimpvesctions with 25 roughly Mars-mass embryos embedded in a disk of 1000 non-interacting (with each other) planetes- imals in an annulus from 0.3 to 4.0 AU. They found that giant impact events which could form the Moon occur frequently in the simulations. These collisions include a roughly Earth-size target whose last large impactor has a mass of 0.11 − 0.14ME and a velocity, when taken at in- finity, of 4 km/s, as found by Canup (2004). O'Brien et al. pointed out that their initial embryo mass is close to the impactor mass. Raymond et al. (2009) set up about 90 em- bryos with masses from 0.005 to 0.1ME in a disk of more than 1000 planetesimals, again with non-interaction of the latter. Assuming again Canup's requirements (v/vesc < 1.1, 0.67 < sin θ < 0.76, 0.11 < γ < 0.15), only 4% of their late giant impacts fulfill the angle and velocity cri- teria. They concluded that Earth's Moon must be a cos- mic rarity but a much larger range of late giant collisions would produce satellites with different properties than the Moon. The initial embryo size seems to play a role in those results. In contrast, the simulations of Morishima et al. (2010) start with 2000 fully interacting planetesimals and since embryos form self-consistently out of the plan- etesimals in these simulations, problems with how to seed embryos are completely avoided. To constrain the simula- tions, Morishima et al. (2010) were interested in the tim- ing of the Moon-forming impact. To identify potentially events, they searched for a total mass of the impactor and the target > 0.5m⊕, a impactor mass > 0.05m⊕ and a im- pact angular momentum > LEM. They found almost 100 suitable impacts in the 64 simulations. Since their sample also includes high velocity or grazing impacts, although constrained through the more general angular momentum limit, and does not take into account collision history and tidal evolution, the difference to our result is not surpris- ing. Life on planets without a massive stabilizing moon would face sudden and drastic changes in climate, posing a survival challenge that has not existed for life on Earth. Our simulations show that Earth-like planets are common in the habitable zone, but planets with massive, obliquity stabilizing moons do occur only in 10% of these. Acknowledgments We thank David O'Brien and an anonymous reviewer for many helpful comments. We thank University of Zurich for the financial support. We thank Doug Potter for sup- porting the computations made on zBox at University of Zurich. References Atobe, K., Ida, S., Ito, T., 2004. Obliquity variations of terrestrial planets in habitable zones. Icarus. 168, 223-236 Atobe, K., Ida, S., 2006. Obliquity evolution of extrasolar terrestrial planets. Icarus. 188, 1-17 Figure 7: The masses of the final outcomes of the planets for which we identified satellite forming collisions. msatellite is the mass of the satellite, assuming an accretion efficiency of 50%, and mp is the mass of the planet after the complete accretion. The circle indicates the position of the Earth-Moon system with the assumption mdisk = mMoon. Agnor, C. B., Canup, R. M., Levison, H. F., 1999. On the Character and Consequences of Large Impacts in the Late Stage of Terrestrial Planet Formation. Icarus. 142, 219-237 Berger, A.L., Imbrie, J., Hays, J., Kukla, G., Saltzman B. (Eds.), 1984. Milankovitch and Climate - Understanding the Response to Astronomical Forcing. D. Reidel, Norwell Berger, A. (Ed.), 1989. Climate and Geo-Science - A Challenge for Science and Society in the 21st Century. Kluwer Academic, Dor- drecht Cameron, A. G. W., Ward, W. R., 1976. The origin of the Moon. Proc. Lunar Planet Sci. cd cConf. 7, 120122 Cameron, A.G.W., Benz, W., 1991. The origin of the Moon and the single-impact hypothesis IV. Icarus. 92, 204-216 Canup, M. R., Esposito, L. W., 1996. Accretion of the Moon from an Impact-Generated Disk. Icarus. 119, 427-446 Canup, M. R., 2004. Simulations of a late lunar-forming impact. Icarus. 168, 433-456 Canup, M. R., 2008. Lunar-forming collisions with pre-impact rota- tion. Icarus. 196, 518-538 Chamberlin, T. C., 1905. In Carnegie Institution Year Book 3 for 1904, 195-234, Washington, DC: Carnegie Inst. Chambers, J. E., Wetherill, G. W., 1998. Making the Terrestrial Planets: N-Body Integrations of Planetary Embryos in Three Di- mensions. Icarus 136, 304-327 Chambers, J. E. 1999. A hybrid symplectic integrator that permits close encounters between massive bodies. MNRAS 304, 793-799 Dones, L., Tremaine, S., 1993. On the Origin of Planetary Spin. Icarus. 103, 67-92 Duncan, M., Levison, H.F., Lee, M.H., 1998. A multiple time step symplectic algorithm for integrating close encounters. Astron. J. 116, 2067-2077 Goldreich, P., Peale, S.J., 1970. The obliquity of Venus. Astron. J. 75, 273-284 Halliday, A.N., 2000. Terrestrial accretion rates and the origin of the Moon, Earth and Planetary Science Letters. 176, 17-30 Hartmann, W. K., Davis, D. R., 1975. Satellite-sized planetesimals and lunar origin. Icarus. 24, 504515 Ida, S., Canup, R.M., Stewart, G.R., 1997. Lunar accretion from an impact-generated disk. Nature. 389, 353-357 Kokubo, E., Ida, S., 1998. Oligarchic growth of protoplanets. Icarus. 131, 171-178 Kokubo, E., Ida, S., Makino, J., 2000. Evolution of a Circumterres- trial Disk and Formation of a Single Moon. Icarus. 148, 419-436 Kokubo, E., Kominami, J., Ida, S., 2006. Formation of terrestrial planets from protoplanets. I. Statistics of basic dynamical prop- 11 (cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:230)(cid:231)(cid:231)0.00.51.01.52.00.000.010.020.030.04mpm(cid:197)msatm(cid:197)erties. Astrophys. J. 642, 1131-1139 Kokubo, E., Genda, H., 2010. Formation of Terrestrial Planets from Protoplanets Under a Realistic Accretion Condition. Astrophys. J. 714, 21-25 Laskar, J., Robutel, P., 1993. The chaotic obliquity of the planets. Nature. 361, 608-612 Laskar, J., Joutel, F., Robutel, P., 1993. Stabilization of the Earth's obliquity by the Moon. Nature. 361, 615-617 Laskar, J., 1996. Large scale chaos and marginal stability in the solar system. Celest. Mech. Dynam. Astron. 6, 115-162 Lathe, R., 2004. Fast tidal cycling and the origin of life. Icarus. 168, 18-22 Lathe, R., 2006. Early tides: Response to Varga et al. Icarus 180, 277-280 Lissauer, J. J., 1993 Planet Formation. Ann. Rev. Astron. Astrophys. 31, 129-174 Milankovitch, M., 1941. Kanon der Erdbestrahlung und seine An- wendung auf das Eiszeitproblem. Kanon Koeniglich Serbische Academie Publication. 133 Morishima, R., Stadel, J., Moore, B., 2010. From planetesimals to terrestrial planets: N-body simulations including the effects of nebular gas and giant planets. Icarus. 207, 517-535 Murray, C.D., Dermott, S.F., 1999. Solar System Dynamics. Cam- bridge University Press, Cambridge O'Brien, D. P., Morbidelli, A., Levison, H. F., 2006. Terrestrial planet formation with strong dynamical friction. Icarus. 184, 39-58 Ohtsuki, K., 1993. Capture Probability of Colliding Planetesimals: Dynamical Constraints on Accretion of Planets, Satellites, and Ring Particles. Icarus. 106, 228-246 Raymond, S. N., Quinn, Th., Lunine, J. I., 2004. Making other earths: dynamical simulations of terrestrial planet formation and water delivery. Icarus. 168, 1-17 Raymond, S. N., O'Brien, D.P., Morbidelli, A., Kaib, N.A., 2009. Building the terrestrial planets: constrained accretion in the inner solar system. Icarus. 203, 644-662 Safronov, V. S., 1969. Evolution of the Protoplanetary Cloud and Formation of the Earth and Planets. Moskau, Nauka. Engl. transl. NASA TTF-677, 1972 Stadel, J., 2001. Cosmological N-body simulations and their analysis. University of Washington, Ph.D. Stewart, S.T., Leinhardt, Z.M., 2009. Velocity-dependent catas- trophic disruption criteria for planetesimals. Astrophys. J. 691, 133-137 Touboul, M., Kleine, T., Bourdon, B., Palme, H., Wieler, R., 2007. Late formation and prolonged differentation of the Moon inferred from W isotopes in lunar metals. Nature. 450, 1206-1209 Varga, P., Rybicki, K.R., Denis, C., 2006. Comment on the paper Fast tidal cycling and the origin of life by Richard Lathe. Icarus. 180, 274-276 Ward, W.R., 1974. Climate variations on Mars. 1. Astronomical the- ory of insolation. J. Geophys. Res. 79, 3375-3386 Ward, W.R., Rudy, D.J., 1991. Resonant obliquity of Mars? Icarus. 94, 160-164 Ward, P. D., Brownlee, D., 2000. Rare Earth: Why Complex Live is Uncommon in the Universe. Copernicus, Springer-Verlag, New York Wetherill, G. W., Stewart, G. R., 1989. Accumulation of a swarm of small planetesimals. Icarus. 77, 330-357 12
1711.09397	3	1711	2018-04-09T01:59:44	A general method for assessing the origin of interstellar small bodies: the case of 1I/2017 U1 ('Oumuamua)	[ "astro-ph.EP" ]	With the advent of more and deeper sky surveys, the discovery of interstellar small objects entering into the Solar System has been finally possible. In October 19, 2017, using observations of the Pan-STARRS survey, a fast moving object, now officially named 1I/2017 U1 ('Oumuamua), was discovered in a heliocentric unbound trajectory suggesting an interstellar origin. Assessing the provenance of interstellar small objects is key for understanding their distribution, spatial density and the processes responsible for their ejection from planetary systems. However, their peculiar trajectories place a limit on the number of observations available to determine a precise orbit. As a result, when its position is propagated $\sim 10^5-10^6$ years backward in time, small errors in orbital elements become large uncertainties in position in the interstellar space. In this paper we present a general method for assigning probabilities to nearby stars of being the parent system of an observed interstellar object. We describe the method in detail and apply it for assessing the origin of 'Oumuamua. A preliminary list of potential progenitors and their corresponding probabilities is provided. In the future, when further information about the object and/or the nearby stars be refined, the probabilities computed with our method can be updated. We provide all the data and codes we developed for this purpose in the form of an open source {\tt C/C++/Python package}, {\bf\tt iWander} which is publicly available at http://github.com/seap-udea/iWander.	astro-ph.EP	astro-ph	Draft version September 13, 2018 Preprint typeset using LATEX style emulateapj v. 01/23/15 8 1 0 2 r p A 9 . ] P E h p - o r t s a [ 3 v 7 9 3 9 0 . 1 1 7 1 : v i X r a A GENERAL METHOD FOR ASSESSING THE ORIGIN OF INTERSTELLAR SMALL BODIES: THE CASE OF 1I/2017 U1 ('Oumuamua) Jorge I. Zuluaga, Oscar S´anchez-Hern´andez, Mario Sucerquia & Ignacio Ferr´ın Solar, Earth and Planetary Physics Group & Computational Physics and Astrophysics Group (FACom) Instituto de F´ısica - FCEN, Universidad de Antioquia, Calle 67 No. 53-108, Medell´ın, Colombia Draft version September 13, 2018 ABSTRACT With the advent of more and deeper sky surveys, the discovery of interstellar small objects entering into the Solar System has been finally possible. In October 19, 2017, using observations of the Pan- STARRS survey, a fast moving object, now officially named 1I/2017 U1 ('Oumuamua), was discovered in a heliocentric unbound trajectory suggesting an interstellar origin. Assessing the provenance of interstellar small objects is key for understanding their distribution, spatial density and the processes responsible for their ejection from planetary systems. However, their peculiar trajectories place a limit on the number of observations available to determine a precise orbit. As a result, when its position is propagated ∼ 105 − 106 years backward in time, small errors in orbital elements become large uncertainties in position in the interstellar space. In this paper we present a general method for assigning probabilities to nearby stars of being the parent system of an observed interstellar object. We describe the method in detail and apply it for assessing the origin of 'Oumuamua. A preliminary list of potential progenitors and their corresponding probabilities is provided. In the future, when further information about the object and/or the nearby stars be refined, the probabilities computed with our method can be updated. We provide all the data and codes we developed for this purpose in the form of an open source C/C++/Python package, iWander which is publicly available at http://github.com/seap-udea/iWander. Subject headings: Interstellar small body -- asteroids: individual: 1I/2017 U1 -- Methods: numerical 1. INTRODUCTION The detection of interstellar small objects, wander- ing through the Solar System, has challenged the as- tronomers over the last century (see e.g. Opik 1932; McGlynn & Chapman 1989; Sen & Rama 1993). The de- tection, characterization and investigation of the origin of these objects could shed light on the processes of plane- tary formation, the architecture of their parent planetary systems, as well as on planetary ejection mechanisms (see e.g Raymond et al. 2017). Moreover, interstellar visitors can also be the target for future exploration missions (the first ones traveling to an interstellar object; Mama- jek 2017), become an indirect way to detect planetary systems of nearby stars and to infer some of their prop- erties (Trilling et al. 2017) or help to test bold hypotheses about the origin and distribution of life in the Universe, such as lithopanspermia (Adams & Spergel 2005; Bel- bruno et al. 2012). Recently, Engelhardt et al. 2017 calculated the num- ber density of detectable interstellar objects, obtaining a discouraging value of 1.4 × 10−4 au−3 which seemed to restrict severely the chances of detecting one of these ob- jects in the coming years, at least with the current avail- able instrumentation. Nevertheless, on 19 October 2017 a small object with an uncommon orbit (e = 1.1995 ± 0.0002 and v∞ = 26.33 ± 0.03 km/s) was discovered us- ing the Pan-STARRS Survey (Chambers et al. 2016). The orbital properties determined at that time using a short orbital arc, while the object was still visible after their rapid passage through perihelion, suggested an in- terstellar origin. This result was supported by follow-up spectroscopic observations (Bolin et al. 2018) and other theoretical considerations (Ye et al. 2017; Masiero 2017). The interloper object has been now named "1I/2017 U1 ('Oumuamua)" or simply 'Oumuamua as we will call it consistently through this paper1. The first follow-up observations suggested that the ob- ject was of asteroidal origin, and its composition seemed to be similar to that of transneptunian objects (TNOs) (Merlin et al. 2017). More recently, Jewitt et al. (2017) measured the observed color and suggested a composition overlapping the mean colors of D-type Trojan asteroids and other inner solar system populations, but inconsis- tent with the ultra-red matter found in the Kuiper belt. Ferrin & Zuluaga (2017) compared the color of the object with other 21 active and extinct Solar System comets and suggest that 'Oumuamua could actually be an inactive extrasolar comet. This result could be very important for assessing their dynamical origin. On the theoretical side, Raymond et al. (2017) suggested than from a dy- namical standpoint it is more likely that 'Oumuamua be cometary rather than asteroidal in origin. Only a few days after the announcement of the discov- ery, several works attempting to pinpoint its interstellar origin were published in the form of research notes and short papers (Mamajek 2017; de la Fuente Marcos & de la Fuente Marcos 2017; Gaidos et al. 2017; Portegies Zwart et al. 2017; Dybczy´nski & Kr´olikowska 2017). Most of these initial works were able to restrict the origin of the object to nearby stars and young planetary systems from which it could have escaped via planet-planet scatter- ing. These early attempts, however, either rely on rough comparisons between the heliocentric entrance velocity 1 IAU Minor Planet Center, 2017, U183 2 of the object and that of nearby stars (Mamajek 2017; Gaidos et al. 2017) and/or on the numerical propaga- tion of the orbit of the object among many nearby stars, in the galactic potential (Portegies Zwart et al. 2017; Dybczy´nski & Kr´olikowska 2017; Feng & Jones 2018). These latter efforts, however, originally lack of a rigorous consideration of the orbit uncertainties and the errors in the astrometric and radial velocities of the stars. A more rigorous kinematic treatment was recently presented by Feng & Jones (2018). However, their approach lack of a estimation of the probability that the object were origi- nated in any of their potential progenitors. Several au- thors have even suggested that given the current orbital uncertainties and astrometric information about nearby stars, pinpointing the exact origin of 'Oumuamua (and possibly other future interstellar objects) to a specific stellar system could be impossible (Zhang 2018). The detection of 'Oumuamua however, allowed some authors to update estimations of the number densities of interstellar objects and their detection probability. Thus, for instance, Trilling et al. 2017 recently determined that if during planet formation processes the ejected mass was about 20 Earth's masses, the detection rate of such in- terstellar objects could be at least 0.2 per year with the current instrumentation. This result is consistent with the discovery of 'Oumuamua. Moreover, when the Large Synoptic Survey Telescope (LLST, Ivezic et al. 2008) be- gins its wide, fast and deep all-sky survey, the rate could climb to 1 object per year. As a result, developing a gen- eral and kinematically rigorous strategy for assessing the origin of this and probably other objects detected in the future, is an important goal to pursue. The problem of tracking small Earth's impactors to- wards their progenitor objects in the Solar System has been extensively studied in the literature (see e.g. Strom et al. 2005; Zuluaga et al. 2013; Strom Robert et al. 2015; Zuluaga & Sucerquia 2018). However, the dynamic of the Solar System is well constrained, the timescales are relatively short and the the position of the perturbing objects are known to exquisite levels of precision. This is not the case when reconstructing the orbit of an in- terstellar object. Timescales are huge, amplifying even small uncertainties in initial positions. The perturbing objects (stars) are very distant and small errors in their projected position in the sky correspond to large errors in their position in space. Moreover, the gravitational scat- tering from random encounters with nearby stars, may prevent a successful track-down of their origins. How- ever, with the advent of new precise astrometric data such as that provided by Gaia and the fact that we are now aware that a non-negligible number of detectable interstellar interlopers, could be wandering through the solar system, the situation is improving significantly. Tracking the motion of objects in the interstellar space is not new. Several authors have studied the problem, aiming to determine the past and future close encoun- ters of the Sun with nearby stars (See e.g. Garc´ıa- S´anchez et al. 2001; Bailer-Jones 2015; Berski & Dy- bczy´nski 2016). Of particular interest for this work is the recent paper by Bailer-Jones (2017) which use up- to-date astrometric and radial velocity catalogs to study this problem, while rigorously considering the uncertain- ties involved. The present paper is inspired and mostly based on the models, data and techniques developed in that work. In this paper we present a general methodology and a computer tool designed for assessing the kinematic origin of an interstellar object. Instead of determining which particular stellar system could be the source of a given object, the methodology presented here aims at comput- ing "interstellar origin probabilities" (hereafter IOP) of many nearby stars. For that purpose our work takes into account rigorously the uncertainties in the object and stars kinematic properties. When more and better astrometric information become available, the values of the IOP computed with our methods and software can be improved. This paper is organized as follows: we introduce the basics of our probability model and define the concept of "Interstellar Origin Probability" (IOP) in section 2. In section 3 we outline the method and tools used in this work to estimate probabilities. Sections 4-7 present the details of the implementation, while illustrating each technique in the particular case of 'Oumuamua. Section 9 defines rigorously each of the terms in the calculation of the IOP. The results of applying the methodology to the case of 'Oumuamua are presented in section 10 including a list of the potential progenitors with their respective IOP. Section 11 is devoted at discussing the limitations of the method, but also its future potential as a gen- eral framework to study the orbit of newly discovered interstellar objects. Finally, in section 12 we present a summary of the paper and draw the main conclusions derived from it. 2. PROBABILITY MODEL In this work we aim at assessing the following question: If you observe an interstellar object coming towards the Solar System from a given direc- tion in space, what is the probability (origin probability) that a specific star be the progen- itor of the object? With the exception of trivial cases, defining a mathe- matical model for computing probabilities is tricky. You need to properly define your probability space (Kol- mogorov 1968), including, constructing a probability function that connects "events" with real numbers quan- tifying the "frequency" of those events. What is the probability space in this case? how can we define prob- abilities in the context of our problem?. In Figure 1 we schematically depict the way as the "origin probability" is defined in the context of this work. The uncertainty in the orbit of the interstellar object and the astrometric parameters of the star (position and proper velocity), makes that we cannot define a single trajectory for both. Instead, we have a beam of trajecto- ries that eventually intersect in a relatively small volume in the past (square region in Figure 1). To estimate probabilities, we can imagine that a large number of "parallel" universes exists where the object and the star have one of many particular set of orbital properties inside their respective trajectory beams. In some of those universes, the trajectories may coincide within a given critical radius (eg. the truncation radius of the planetary system, see next section). In others, the properties of the object and the star are such that their 3 Fig. 1. -- Schematic representation of the probability model devised in this model to calculate the Interstellar Origin Probability (IOP). Given the uncertainty in the orbit of the interstellar object (ISO) and a given star, we can imagine a large number of parallel universes where the trajectories of both, the ISO and the star, either coincide in space and time (coincidence) or not (miss). The ratio of the number of those parallel universes where the trajectories coincide give us the position probability. Even in the case of coincidence, the object may come from the background (dashed lines in the rightmost box) instead of being an indigenous object. The IOP is the product of the position probability and a correcting factor (the veloicity probability) taking into account the flux of background objects. trajectories, although reach a minimum distance at some point in the past, have a encounter distance large enough to prevent us thinking that the object could be ejected from the star (we call this condition a "miss"). Under a frequentist approach, the probability of co- incidence in position (hereafter "position probability") will be the ratio of the number of universes in which we have a coincidence, Npos, to the number of all possible trajectory configurations. However, among all those universes where there is a coincidence in position, we cannot ensure that the inter- stellar object was actually ejected from the stellar system (an "indigenous" object). It could happen that the ob- ject be simply the result of the gravitational scattering of a background object by the stellar system (dashed tra- jectories in the leftmost panel of Figure 1). How can we distinguish these two cases?. Since ejec- tion and scattering are different physical phenomena, it is reasonable to expect that the distribution of speeds for indigenous objects be different than that of back- ground objects. Thus for instance, and as we will show in subsection 9.1, while the mechanism ejecting indigenous objects tend to produce Maxwellian-like speed distribu- tions, which are characterized by the fact that low and high speeds have low probabilities, the speeds of scat- tered background objects arises from a combination of ejection and the relative speeds of stars in the Galaxy; for simplicity we can be assumed that background rela- tive speed are nearly uniform, ie. low and large speeds have similar probabilities. If we assume that pind(vrel)dvreldΩ is the number of the total number indigenous objects ejected from the stellar systems with relative speeds between vrel and vrel + dvrel in the direction of the Solar System, and pB(vrel)dvreldΩ is the number of background objects in the same speed interval, the fraction of the number of spatial coinci- dences that actually corresponds to an ejection process, can be estimated as: find(vrel) = pind(vrel) pi(vrel) + pB(vrel) (1) Here, two different extreme situations can be consid- ered: (i) The number of indigenous objects are much larger than that of background objects, ie. pind(vrel) (cid:29) pB(vrel). If this is the case, find(vrel) ≈ 1 This will be the case, for instance, if we have a young stellar system with an ongoing planetary for- mation process. (ii) The number of background objects coming out from the stellar system are much larger than pB(vrel) (cid:29) the number of indigenous ones, pind(vrel). In this case, ie. find(vrel) ≈ pind(vrel) pB(vrel) This will be the case of moderately old stellar sys- tem where scattering still exists but at a more mod- erate rate. Although demonstrating which is the case for a partic- ular star is out of the scope of this work, we will assume that the second situation is more common. Despite this particular assumption, in our detailed model we will still include enough information to estimate origin probabili- ties for both extreme cases. ºSolar SystemStarInterstellarObject TrajectoryCoincidenceMiss0.5 pc0.5 pcPosition ProbabilityVelocity ProbabilityBackgroundX4 In summary, the probability that a given star be the progenitor of an interstellar object, namely its "interstel- lar origin probability" (IOP), is proportional to: IOP ∝ Nposfind (2) In the following paragraphs we first summarize the nu- merical procedure we used to set up our "probability space" and then the detail mathematical models used to estimate the factors Npos and find, required to calculate the IOP of a sample of nearby stars. 3. OUTLINE OF THE METHOD Once a small-body is discovered in the Solar System, the astrometry and a reference orbit (computed from the observation weighting differences) is published by the Mi- nor Planet Center (MPC). An up-to-date best-fit orbital solution of the astrometric data provided by the MPC is then computed and made available publicly in the JPL Small-Body Database Browser2. The JPL orbit solution is provided in the form of a vector of "nominal orbital elements" E0 : (e0, q0, tp,0, Ω0, ω0, i0) (orbital ec- centricity, perihelion distance, time of perihelion passage, longitude of the ascending node, argument of the perihe- lion and inclination, respectively) at a reference epoch t0. Along with this information, the JPL also provides the nominal mean orbital motion n0, and its standard error σn0. The uncertainties in the orbital fit are characterized with an "orbit covariance matrix", Cjk, defined as: (cid:26) σ2 Cjk = j = k ρjkσjσk j (cid:54)= k j (3) where j : e, q, tp, Ω, ω, i and ρjk are the correlations among the orbital elements. Provided the nominal orbit and its associated errors, we need to propagate backward in time the trajectory of the object and the position of a set of nearby stars to set up our probability space and prepare the conditions to compute the IOP. The specific procedure we follow to achieve this goal is summarized as follow: 1. Generate Np clones of the orbital elements vector E of the interstellar object compatible with the latest orbital solution Eo and Cjk (see section 4). These clones describe the orbit of what we will call the "surrogate objects" (following the convention of Bailer-Jones 2017). 2. Integrate backwards the orbits of the surrogate ob- jects, until a time when the object in the nominal orbit reaches a distance from the Sun where the galactic tides become relevant for the dynamic of the object (see section 5). 3. Using the "linear motion approximation" (LMA), identify the stars in an astrometric and radial ve- locity catalog, such that their minimum distance to the nominal object, dLMA,min, be less than or equal to a threshold distance. Stars identified with this procedure are called the "candidates" (see sec- tion 6). 2 https://ssd.jpl.nasa.gov/ 4. For each candidate, compute a more precise en- counter time tmin and minimum distance dmin by integrating the orbit of the star and the nominal object in the galactic potential (see section 7). 5. For each candidate star, generate Ns "surrogate stars" (hypothetical stars having astrometric prop- erties compatible with the observed properties of the star and their errors). The orbital integration of both, the surrogate stars and the surrogate ob- ject, define the intersecting beam of trajectories in Figure 1. 6. Identify those candidates for which the more pre- cisely computed minimum distances is below a tighter distance threshold and the IOP probability is the highest. We call these the "potential progen- itors". All the potential progenitors identified with this proce- dure are tabulated in descending order of IOP probability (Table 4). The idea is not to select an individual star, but to assign a probability to each of them that can be improved with this method as the interstellar object and the stars itself are better known. 4. THE SURROGATE OBJECTS Since the trajectory of the object is uncertain, and small errors inside the Solar System amplify when the orbit is integrated into the interstellar space, assessing its interstellar origin require than more than a single or- bit (the nominal solution) be computed. For this purpose we generate Np vectors of orbital el- ements E compatible with the orbital solution described with E0 and Cjk (see section 3). Random realizations of the orbital elements are computed using a multivari- ate Gaussian random number generator3 having mean values equal to the nominal solution element vector Eo and covariance equal to the orbit covariance published by JPL. In Figure 2 we show the "ingress" position and velocity (see next section) of 1000 surrogate objects whose orbits were generated using this procedure. The strong corre- lation between the orbital parameters places the orbits of the surrogate objects in a very narrow ellipsoid which is projected in the sky at the time of ingress to the Solar System as a narrow line around the nominal radiant of the object (upper panel in Figure 2). The error in the present 'Oumuamua's orbital solution (JPL 15), propa- gates as an error in the position at the time of ingress ( ∼2000 years before present) of ∼7 AU (middle panel in Figure 2), and a dispersion in their velocities of ∼ 0.03 km/s (lower panel in Figure 2). 5. TRAJECTORY IN THE SOLAR SYSTEM Once the orbital elements of the surrogate objects have been generated at the reference epoch, we proceed to cal- culate the trajectory of each object in the gravitational 3 Most numerical libraries are provided with a multivariate ran- dom number generator. In the particular case of this work we use the generator and related routines provided by the GNU Scientific Library GSL (Galassi et al. 2002) field created by the sun and the 8 planets4. For that purpose we use a Gragg-Bulirsch-Stoer algo- rithm (Gragg 1965; Bulirsch & Stoer 1966) adapted from site5. Positions and velocities of the the planets were not computed with the integrator itself, but using NASA NAIF SPICE software (Acton Jr 1996 and Jon D. Giorgini)6. For that purpose we use the latest DE431 planetary ker- nel. We verified that for the case of 'Oumuamua our integrations were close to the precise solutions computed by the JPL Solar System Dynamics group and published in the on-line NASA Horizons system. A maximum error of 0.1% at back to 400 years before the reference epoch were obtained with our integrator. Since the trajectory of the object is hyperbolic, and in the case of 'Oumuamua, highly inclined with respect to the ecliptic (i=122.6◦), the object reached large dis- tances to the Sun and the planets in relatively short times. Given the original uncertainty in the orbit, us- ing the integrator to compute a very precise position of the surrogate objects within the Solar System is point- less. At some time in the past, the accumulated errors in the state vector (x, y, z, vx, vy, vz) due only to the or- bit uncertainty, will be much larger than the errors from assuming that the objects move in an ideal hyperbola. Thus, for instance, in the case of 'Oumuamua, we veri- fied that at t = −100 years, the positions of the surrogate objects were spread inside an ellipsoid with a character- istic size of ∼ 0.5 AU. On the other hand, the osculating elements of the object orbit at t=-6 years, when it was at ∼40 AU from the Sun and above the ecliptic plane, allows us to predict the position at t = −100 years with an uncertainty of ∼0.04 AU. We call the orbital elements at this point, the "asymptotic elements" and use them to compute the position of the object inside the Solar Sys- tem in the far past. The value of the asymptotic elements for 'Oumuamua along with other relevant information about its orbit in the Solar System are presented in Ta- ble 1. Using the asymptotic orbital elements we calculate the position and velocity of the surrogate objects at arbitrary times in the past. However, when the object is at a dis- tance comparable with the truncation tidal radius of the Solar System, dT,(cid:12), the effects of the galactic potential in its motion cannot be neglected. We call this epoch the "time of ingress" ting. The interstellar orbit integra- tion for the surrogate objects and stars, starts precisely at ting. It is important to stress that the particular value as- sumed for dT,(cid:12) does not modify considerably our final results and the conclusions of this work. In a numerical experiment we find that changing the value of dT,(cid:12) from 10,000 AU to 100,000 AU introduces differences below 0.1% in the interstellar positions and velocities of the surrogate objects, an error which is much smaller than their intrinsic orbital uncertainties. In Table 1 we present the time of ingress for 'Oumua- mua and its position and velocity in the ICRS galactic reference frame at that time. Our results are in agree- 4 Although dwarf planets and major asteroids were not included in the integrations presented here, it is easy to add them in the iWander package provided with this work. 5 http://www.mymathlib.com/ 6 http://naif.jpl.nasa.gov/pub/naif Property Value 5 Reference Epoch Nominal elements (JPL 15) Epoch of Asymptotic elements Asymptotic elements Asymptotic covariance ×10−6 Truncation radius Time of ingress Radiant at ingress Velocity at ingress 2458059.5 TDB = 2017-Nov-02.0 q = 0.255343194 AU e = 1.199512420 i = 122.687205 deg Ω = 24.599211 deg ω = 241.702983 deg M = 36.425313 deg µ = 1.327124400 × 1011 km3/s2 2455736.51 TDB = JUN 24.01 2011 q = 0.252440118 AU e = 1.197253807 i = 122.735691 deg Ω = 24.252921 deg ω = 241.680101 deg M = −1544.878492 deg µ = 1.327124400 × 1011 km3/s2 Eccentricity Cee = 0.028 Ceq = 0.010 Cet = 7.558 CeΩ = -0.047 Ceω = 1.905 Cei = 1.002 Perihelion distance Cqq = 0.004 Cet = 2.831 CqΩ = -0.018 Cqω = 0.714 Cqi = 0.375 Periapsis time Ctt = 4847.133 CtΩ = -14.358 Ctω = 514.030 Cti = 271.891 Long. ascending node CΩΩ = 0.080 CΩω = -3.205 CΩi = -1.686 Argument of periapsis Cωω = 130.021 Cωi = 68.363 Inclination Cii = 35.963 50000.0 AU 8994.0 years RA = 279.80 ± 0.03 deg DEC = 33.99 ± 0.01 deg l = 62.90 ± 0.02 deg b = 17.11 ± 0.02 deg U = −11.463 ± 0.011 km/s V = −22.401 ± 0.001 km/s W = −7.748 ± 0.010 km/s Properties of the 'Oumuamua's orbit in the Solar System. TABLE 1 ment to those of Mamajek (2017). In order to illustrate the effect that uncertainties in the orbit solution has in the predicted position of the object in the past, we show in Figure 2 the radiant in the sky of the surrogate objects at ingress time. We also plot the position in space of the objects in the ICRS galactic ref- erence frame, their velocities in the same reference frame and their corresponding errors. The convention stands that (U, V, W ) correspond to (vgal,x, vgal,y, vgal,z). As commented before, even a small error in the orbital elements at the reference epoch are propagated as rela- tively large errors at the ingress time. While at t = −100 yrs the surrogate objects were contained in an ellipsoid with a characteristic size of 0.5 AU, at ting the surrogate 6 Fig. 2. -- Dispersion in the position of the surrogate objects at the time of ingress into the Solar System. Right ascension, RA and declinations, DEC are referred to the Solar System Barycenter and the J2000 equinox. Galactic cartesian coordinates (xgal, ygal, zgal) and velocities (U, V, W ) are referred to the ICRS galactic system of reference. objects have spread as a ∼100 AU cloud (at this time the objects are by definition at a distance equivalent to the truncation radius, namely 50,000 AU ). We have ver- ified that this trend continues into the interstellar space, with a expansion velocity of ∼0.05 km/s (the "size" of the cloud in the velocity U V W space, see bottom panel in Figure 2). At this rate, the cloud characteristic size will grow to ∼2 pc (the typical distance between stars in the solar neighborhood) in ∼40 Myr. Beyond this time, the cloud of surrogate objects will encounter at the same time more than one nearby star, and hence the reliabil- ity of our method will be compromised. We call this the "maximum retrospective time", tret. In the case of 'Oumuamua, tret ∼ 40 Myr. 6. ASTROMETRIC AND RADIAL VELOCITY DATABASES In order to compare the position of our interstellar ob- ject with the position of nearby stars in the far past, we need to know as precisely as possible the position and ve- locities of those stars at present time. For this purpose we have compiled up-to-date astrometric information (posi- tion, parallaxes, proper motion and radial velocities) of 285,114 stars (see Table 2). Compiling precise astrometric and radial velocity mea- surements for a significant number of nearby stars, is tricky. On one hand, precise astrometric databases such as Hipparcos (Perryman et al. 1997) and Tycho-2 (Høg et al. 2000) does not include radial velocities measure- ments. The Gaia mission have the potential to provide this information, but it will only be available since Data Release 2 (DR2) in April 20187. On the other hand, there are several public catalogs that provide precise radial velocities (see Table 2). How- ever, not all the objects in those catalogs are included in the astrometric catalogs, and in some cases they have unique identifications without any reference to the Hipparcos/Tycho-2 ids, which are to the date of writing this paper, the identification for objects having precise astrometric measurements (either from the Hipparcos or the Gaia mission). We compile the information required for this work following the detailed directions recently published by Bailer-Jones 2017. Additionally, and in order to include nearby bright stars (which were not included in the Gaia catalog) we search for other information about the Hip- parcos stars in the Simbad information system8. The importance of having this information in a properly-structured table, led us to create the so-called AstroRV catalog (astrometric and radial velocities cata- log). The catalog is provided with the iWander package developed for this work. For the sake of completeness and reproducibility we outline below the procedure required to compile the AstroRV catalog: 1. Get available astrometric and radial velocities cat- alogs. In Table 2 we summarize the information of the catalogs we use for this work. The most important one is the Gaia DR1 catalog containing over 1 billion objects (Brown et al. 2016). Among those objects, however, only 2,026,210 were pre- viously included in Hipparcos and Tycho-2 cata- logs and a full astrometric solution (with a positive parallax) was obtained (Gaia TGAS catalogue). Among them, 93,398 are in the Hipparcos catalog and 1,932,812 belongs to the Tycho-2 catalog. 2. Obtain the general information available in Simbad for all the stars having an Hipparcos ID. The in- formation includes other designations for the stars (Henry Draper catalog id and proper names for the bright stars) and radial velocities for some of those stars. 3. Find all the objects in the Hipparcos and Tycho Catalogues (ESA 1997) and in the Simbad table compiled before, that are not included in the GAIA TGAS catalog. Append the resulting objects to the latest catalog to create a final table with all the available astrometric information. 4. Create a table including all objects in the ra- dial velocities catalogs that have Hipparcos and/or 7 https://www.cosmos.esa.int/web/gaia/release 8 http://simbad.u-strasbg.fr/simbad/ 279.725279.750279.775279.800279.825279.850279.875279.900RA (deg)33.9533.9633.9733.9833.9934.0034.0134.0234.03DEC (deg)Radiant: RA = 279.80±0.03, DEC = 33.99±0.011I/2017 U12170021720217402176021780218002182021840xgal (AU)146601468014700147201474014760zgal (AU)Position at ingress: (x,y,z) = (21768.2±21.5,42539.4±1.9,14712.9±19.3) AU11.5011.4911.4811.4711.4611.4511.4411.43U (km/s)7.787.777.767.757.747.737.72W (km/s)Velocity at ingress: (U,V,W) = (11.463±0.011,22.401±0.001, 7.747±0.010) pcTycho-2 ids. For those objects not having any of those ids, find the Hipparcos/Tycho-2 objects matching the coordinates within a 50 arcsec ra- dius (we use for this purpose the X-match tool of Simbad). For detailed instructions about how to compile the available radial velocity catalogs please refer to Bailer-Jones (2017). 5. Merge the astrometric and radial velocity tables according to the Hipparcos/Tycho-2 ids. The catalog compiled with this procedure contains 285,114 stars including the nearest and brightest ones. We provide with the iWander package the required scripts for creating the catalog as described before. Those scripts can be modified to include future radial velocity and astrometric catalogs, or to update the infor- mation in older versions. An up-to-date version of the AstroRV catalog will be available at the iWander GitHub repository9. 7. TRAJECTORIES IN INTERSTELLAR SPACE Having the position and velocity of both, the surro- gate objects and the nearby stars, we now may integrate their orbits backward in time to identify any close ap- proach. Since we need to perform a comparison between the position of at least ∼1,000 surrogate objects with O(105) stars, simulation times may become prohibitively large. In a first step we perform a selection of "candidate stars" using the so-called "linear motion approximation" (LMA). 7.1. The LMA approximation Under this approximation, all particles, the surrogate objects and the stars, move in straight lines: (cid:126)ri(t) = (cid:126)ri,0 + (cid:126)vi,0t Here, (cid:126)ri ((cid:126)ri,0) and (cid:126)vi are the position and velocity of the i-th particle. The instantaneous distance between particles j and k (surrogate objects and stars), is given by: ∆(cid:126)rjk(t)2 =∆(cid:126)rjk,0 + ∆(cid:126)vjk,0t2 =∆(cid:126)rjk,02 + 2∆(cid:126)rjk,0 · ∆(cid:126)vjk,0t + ∆(cid:126)vjk,02t2 where ∆(cid:126)rjk,0 = (cid:126)rk,0 − (cid:126)rj,0 and ∆(cid:126)vjk,0 = (cid:126)vk,0 − (cid:126)vj,0. This function has a minimum when d∆(cid:126)rjk(t)2/dt = 0, ie. at the time tjk,min when the distance between the objects is djk,min: tjk,min =− ∆(cid:126)rjk,0 · ∆(cid:126)vjk,0 ∆(cid:126)vjk,02 djk,min =∆(cid:126)rjk(tmin) (4) We compute tjk,min and djk,min for the nominal object and all the stars in our AstroRV catalog in order to se- lect the "candidate stars", namely those stars that could actually have close encounters with the object. After numerically testing several criteria, we found that the most cost-effective (in terms of computational 9 http://github.com/seap-udea/iwander.git (cid:26) (cid:27) 1 5 d0 7 (5) resources in the succeeding steps), albeit simple condi- tion is: djk,min < sup dmax, where dmax is an arbitrary distance threshold, d0 is the present distance of the star and the 1/5 is an adjustable numerical factor. For our present analysis we assume dmax = 10 pc. It should be stressed that with an arbitrary amount of computational resources all the stars in the AstroRV catalog could be integrated in the galactic potential (see next section) and no heuristic selection criterion (such as that in Equation 5) need to be used. This criterion is only intended to save computational resources. In the case of 'Oumuamua we have found that after applying the criterion in Equation 5, only 2560 among the 285114 stars in the input catalog, were selected as candidates (see red circles in Figure 3). We verified with a full simulation (including only the nominal object and stars) that only a handful of suitable candidates (blue triangles in Figure 3) were excluded from the probability calculations. For each "candidate star" selected with the preced- ing criterion, we should convert their observed proper- ties (RA,DEC,,µα,µδ,vr) (right ascension, declination, parallax, projected proper motion in RA, proper mo- tion in declination and radial velocity, respectively) into its spatial cartesian coordinates in the galactic system, (xgal, ygal, zgal, U, V, W ) (and more importantly their cor- responding uncertainties). We use for this purpose the prescripcion Johnson & Soderblom 1987. 7.2. Motion in the galactic potential As mentioned before, the minimum LMA distances are useful at selecting the candidates but are a very crude approximation of the actual encounter conditions. The integrated effect of the galactic potential may modify substantially the position of the objects and the stars, especially after a long time of wandering in the interstel- lar space. In order to obtain a "second order" estimation of the minimum distances, we need to integrate the tra- jectory of the surrogate objects and the stellar candi- dates in the galactic potential. Given the axisymmet- ric nature of the potential, we need to transform the cartesian coordinates of the objects with respect to the Sun (xgal, ygal, zgal, U, V, W ) to cylindrical coordinates re- ferred to the galactocentric reference system (see Fig- ure 4). This coordinate transformation is achieved in three steps: 1. Convert the position and velocities referred to the local standard of rest (LSR) into position and ve- locities relative to the galactic center: (cid:126)vGC = (cid:126)vgal + (cid:126)v(cid:12) + vcirc ygal (cid:126)rGC = (cid:126)rgal − R(cid:12) x where (cid:126)vgal : (U, V, W ) is the velocity of the star with respect to the Sun in galactic coordinates, (cid:126)v(cid:12) : 8 Catalog name Number of objects Hipparcos ID Tycho-2 ID Contribution CDS Code Reference Gaia TGAS Hipparcos Tycho Simbad Totals 2026210 117955 579054 118004 2841223 1167 14139 520701 35493 6028 673 2032 10680 WEB1995 GCS RAVE-DR5 PULKOVO FAMAEY2005 BB2000 MALARODA GALAH MALDONADO 473 APOGEE2 AstroRV 29173 620559 93398 117955 -- 118004 329357 494 12977 121 35493 6028 -- -- -- 473 -- 55586 Astrometric 1932812 -- 579054 -- 2511866 Radial velocities 673 -- 309596 -- -- 673 2032 10680 -- 29173 352827 TABLE 2 2026210 24557 164550 67 2215384 252 7091 217257 23412 5544 503 416 7837 301 22501 285114 I/337/tgas I/239/hip main I/259/tyc2 -- This work III/213 J/A+A/530/A138 III/279/rave dr5 III/252/table8 J/A+A/430/165/tablea1 III/213 III/249/catalog J/MNRAS/465/3203 J/A+A/521/A12/table1 -- This work (1) (2) (2) (3) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) Catalogs used to compile the AstroRV catalog for this work. References: (1) Brown et al. 2016, (2) ESA 1997 (3) Wenger et al. 2000, (5) Barbier-Brossat & Figon 2000a, (6) Casagrande et al. 2011, (7) Kunder et al. 2017, (8) Gontcharov 2006, (9) Famaey et al. 2005, (10) Barbier-Brossat & Figon 2000b, (11) Malaroda et al. 2000, (12) Martell et al. 2016, (13) Maldonado et al. 2010, (14) Bailer-Jones 2017. (U(cid:12), V(cid:12), W(cid:12)) is the velocity of the Sun with respect to the local standard of rest (LSR) and vcirc is the local circular galactic velocity. In Table 3 we show the values adopted in this work for these quantities. 2. (cid:126)vGC and (cid:126)rGC are referred to a system pointing to the galactic center (the unprimed xgal, ygal, zgal in Figure 4). This system is rotated at an angle α = sin−1(z(cid:12)/R(cid:12)) with respect to the plane of the Galaxy. Thus, the actual physical galactocentric coordinates on which we must perform the orbital integration are obtained after the rotation: Property U(cid:12) V(cid:12) W(cid:12) vcirc z(cid:12) R0 Md, ad, bd Mb, ab, bb Mh, ah, bh Value 11.1 km/s 12.24 km/s 7.25 km/s 220.0 km/s 17 pc 8.2 kpc 7.91 × 1010M(cid:12), 3500pc, 250pc 1.40 × 1010M(cid:12), 0, 350pc 6.98 × 1011M(cid:12), 0, 24000pc Reference (1) (1) (1) (2) (3) (3) (4) (4) (4) Properties of the Galaxy. References: (1) Schonrich TABLE 3 et al. 2010, (2) Bovy 2015, (3) Karim & Mamajek 2016, (4) Bailer-Jones 2015. (cid:126)r(cid:48) (cid:126)v(cid:48) GC = Rα(cid:126)rGC GC = Rα(cid:126)vGC  cos α 0 sin α 0 − sin α 0 cos α 0 1  with Rα = 3. Finally, we need to express the resulting galacto- centric position and velocity in cylindrical coordi- nates, ie. (cid:126)r(cid:48) GC:(R,θ,z), (cid:126)v(cid:48) GC:( R,R θ,z). In this coordinate system, the equations of motion of an object moving in the potential of the galaxy Φ(R, θ, z) are given by (see eg. Garc´ıa-S´anchez et al. 2001): + R θ2 R =− ∂Φ ∂R θ =− ∂Φ ∂θ z =− ∂Φ ∂z simplicity an axisymmetric Kuzmin-like potential for the three galactic subsystems Here we assume for R θ R − 2 (6) (Kuzmin 1956; Miyamoto & Nagai 1975), disk (d), bulge (b) and halo (h): Φ(R, θ, z) = − (cid:88) i=d,b,h (cid:113) R2 + (ai +(cid:112)z2 + b2 GMi i )2 (7) The value of the potential parameters Mi, ai, bi as- sumed for each component are summarized in Table 3 Once the trajectory of the objects in the potential of the Galaxy are integrated, we proceed at finding the en- counter distance and time of the nominal object to each stellar candidate. This is a refined estimation of dmin and tmin. 8. THE SURROGATE STARS In the same way as we generate Np surrogate objects to take into account the uncertainties in the orbit so- lution of the interstellar interloper, we need to gener- ate, for each stellar candidate, Ns "surrogate stars" with observed properties compatible with the nominal astro- metric observables, namely (α0,δ0,0,µα,0,µδ,0). For this purpose we build a covariance matrix (see Equation 3) using the errors reported for each astrometric variable and their related correlations10. 10 In the Hipparcos and Gaia databases, correlations are re- ported in the form of fields such as RA DEC CORR, RA PARALLAX CORR, 9 lar system with heliocentric orbital elements (q, e, i, Ω, ω, tp) at a given reference epoch t0. Let's assume first that the orbit of the interstellar ob- ject is known with zero uncertainty. If we propagate all the surrogate stars until the time of minimum distance between the object and the nominal star, tnom min , the object will be surrounded by a cloud of points (see Figure 5). Each of these points represents the position of one sur- rogate star at the time of encounter. In the continuous limit, even if none of the Ns surrogate stars coincide in position with the object, a probability different than zero exists that they were physically connected. Our central assumption here is that the probability that such a re- lationship actually existed will be proportional to the density of surrogate stars at the position of the object. In terms of our probability model in section 2, the num- ber Npos of parallel universes where the trajectory of the star and the object coincide will be proportional to the density of surrogate stars. Computing the number density from a discrete set of positions of the surrogate stars, is challenging. Several numerical techniques have been devised and applied in other areas such as cosmology and hydrodynamics (see eg. Price 2012), to asses similar problems. More recently, Zuluaga & Sucerquia 2018 applied the approach used in Smooth Particle Hydrodynamics in the context of impact probabilities in the the Solar System. According to this approach, the number density of stars around an object with position (cid:126)rp can be computed as: n((cid:126)rp) = W ((cid:126)rp − (cid:126)r∗i, h) (8) Ns(cid:88) i Fig. 3. -- Minimum encounter distances for the nominal orbit of 'Oumuamua and all the stars in the AstroRV catalog (black dots), calculated with the LMA approximation (horizontal axis) and with a precise galactic potential integration (vertical axis of the upper panel). Blue triangles represent the potential progenitors selected from their precise minimum distance, ie. dmin < 10 pc; red circles mark the position of the stars fulfilling the heuristic selection crite- ria in Equation 5. The lower panel is intended to illustrate the way as this criterion was especially devised to minimize the number of actual potential progenitors that are missed when using only the LMA approximation (blue triangles without a corresponding red circle). Summarizing the previous sections, studying the coin- cidence in time and space of an interstellar object with a nearby star, implies dealing properly with the kinemat- ics of Np surrogate objects and Ns surrogate stars. It is quite evident that assessing the probability of a close encounter, using just the nominal solutions for the object and the stars is certainly unrealistic. Moreover, dealing rigorously with the different reference system transfor- mation and the galactic kinematics is neither optional. 9. INTERSTELLAR ORIGIN PROBABILITY We can now reformulate the origin probability question we raise in section 2, in terms of the properties defined in the previous section. The question is now: what is the probability that a star with present astrometric properties (α, δ, , vr) ejected in the past a body that entered into the so- etc. where (cid:126)rp − (cid:126)r∗i is the distance between the object and the ith surrogate star, h is a distance-scale for the dis- tribution, and W (d, h) is called a smoothing kernel. In this work we use for h, the characteristic size of the solar- mass truncation radius, namely h ≈ 0.5 pc (see Figure 1). Other prescriptions can be used, but for the purpose of testing our method we will restrict to this simple ansatz. Different kernel function can be used for calculating n as precisely as possible. Although it is common in SPH to use a B-spline kernel (see eg. Zuluaga & Sucerquia 2018), for the purposes pursued here, the best suited function is one that provide a non-zero, although still very low value of n((cid:126)rp) for large values of (cid:126)rp − (cid:126)r∗i. The kernel function used in this work will be a gaussian one (Price 2012): where σ = ((cid:82) W dV )−1 is a normalization constant. W (d, h) = σ exp(−d2/h2) (9) We assume that the probability that the candidate star min inside a small volume dV around the object be at tnom position will be given by: n((cid:126)rp)dV p((cid:126)rp)dV = tion the property (cid:82) n dV = Ns, the probability density Since the number density in Equation 8 has by defini- (10) Ns function in Equation 10 is properly normalized. In the discrete case, if we take ∆V as the volume of a sphere having the truncation radius, the probability of a 101100101102103dmin (integration,pc)101100101102103dmin (LMA, pc)100101102103d (pc)All starsStars fulfilling criterion in Eq. 3Stars having real dmin<10 pc10 Fig. 4. -- Galactic Reference Systems. i)∆V pos ≈ n( (cid:126)rp P i Ns Finally, the total number of coincidences can be esti- pos, and the mated by summing up the contributions N i position probability can be finally written as: (cid:88) i Ppos ∝ 1 N 2 s n( (cid:126)rp i) (12) Fig. 5. -- Scattering plot of the position of the surrogate stars corresponding to HD 200325 (red star marks its nominal position) at the time of minimum distance with the nominal object (blue dot). Contours show the number density of stars around the object estimated with the methods in this work. coincidence in position between the star and the object at tnom min , P nom pos can be estimated as: pos ≈ n((cid:126)rp)∆V P nom Ns (11) pos ∝ P nom In the language of our probability model, there will be a number N nom pos of parallel universes where the star and the object coincide in position. However, as illustrated in Figure 1, encounter times depend on the region inside the intersection volume where the minimum approach happens. Therefore, the number N nom pos will be just the number of encounters that are produced around that time. We should integrate the cloud of surrogate objects and surrogate stars until a time ti min where the closest approach between the object and the i-th star happens. At that time we estimate the local number i) and the encounter density of stellar trajectories, n( (cid:126)rp probability P i pos: Although the constant factor N 2 s is common to all the stars in our simulations, we preserve it in order to have numbers of a reasonable order of magnitude. 9.1. Relative velocity Only a few processes may lead to the ejection of a a small body from an almost-isolated planetary system (Melosh 2003; Napier 2004). In the low density solar neighborhood the more feasible ejection mechanisms are the particle-particle gravitational scattering, where small bodies receive a gravitational slingshot effect after en- countering a planet or even a star in a multiple stellar system, at the right conditions (Raymond et al. 2017; ´Cuk 2018; Wiegert 2014). We estimate the distribution of excess velocities, v∞, that small bodies (asteroids and comets) receive from their encounters with a giant planet around a solar mass star, using a semi analytical ap- proach inspired in the works by Wiegert (2011, 2014). For this purpose, we first set up a planetary system having a single planet of mass Mp located in a circular orbit with semimajor axis ap. We randomly generate or- bital elements for small bodies such that they intersect the orbit of the planet. For simplicity the small-body semimajor axis, eccentricity and orbital inclinations were uniformly generated in the whole range of possible val- ues, eg. a ∈ (0.5ap − 1.5ap), e ∈ (0, 1), i ∈ (0, 90) deg. Longitude of ascending node and argument of periapsis were calculated imposing the condition that the small body and the planet collide. For each planet-small body orbit configuration we com- pute the relative velocity with which they encounter and the direction with respect to the planet reference frame z(cid:1)z'galz'GCy'GCx'GCx'galxgalygalzgaly'gal432101234xgal (pc)3210123ygal (pc)Nominal objectNominal starHD 200325 and 'Oumuamua at tnommin = -4.2 Myr11 Fig. 7. -- Upper panel: ejection mean velocity for different plane- tary masses and orbital sizes. Lower panel: ratio of the average to the standard deviation of the ejection velocities for different plan- etary masses. Properties are given in canonic unites. If uL =1 AU and uM = 1 M(cid:12), uv = uL/uT = 30 km/s. periments is that the ratio of the standard deviation σv∞ to the mean value of the ejection velocity ¯v∞, is almost independent of planetary orbital distance. For its depen- dency on planetary mass, in the lower panel of Figure 7 we show a plot of the value of σv∞/¯v∞ for different plan- etary masses. It is interesting to notice that in the case of a Maxwell-Boltzmann distribution (MBD) the ratio σ/µ (with µ the mean) is constant and equal to 0.42 which is of the order of σv∞/¯v∞ in our own experiments. This seems to suggest that the ejection velocities can be fit- ted by a MBD with a mean that depends on Mp and ap. This is precisely the fitting functions we have used in Figure 6. Using the average ejection velocities and the dispersion-to-mean ratio in Figure 7, we can estimate the velocity distribution of small-bodies being ejected from a planetary system around a star of a given mass. Since ejection velocities depend on the unknown mass Mp and semimajor axis ap of the largest planet in the system, we estimate the "posterior" ejection velocity probabili- ties, assuming for simplicity uniform "priors" for these quantities. The resulting posterior distribution pv∞(v) for the case of a solar-mass star is shown in Figure 8. Fig. 6. -- Ejection velocity distribution for a planet in a cir- cular orbit at ap = 5 uL and different planetary masses. Continu- ous thick lines are Maxwell-Boltzmann distributions with the same mean as the numerical results. from which the small body approaches. From here we follow the prescription of Wiegert (2011) to compute the outbound velocity of the object after interacting with the planet. A random position (xp, yp) over the tan- gent plane to the Hill sphere is generated (see Figure 1 in Wiegert 2014). From there we compute the impact parameter, scattering angle, planetocentric orbital eccen- tricity and periapsis distance. Finally we rotate the plan- etocentric inbound velocity to compute the outbound ve- locity of the small-body at the Hill radius with respect to the planet and then with respect to the star. Once the synthetic small bodies in our simulation are scattered by the planet, we evaluate if their outbound velocity v with respect to the star is larger than the es- cape velocity vesc at the position of the planet. If this is inf = v2− v2 the case, we compute the excess velocity, v2 esc. In Figure 6 we show the distribution of excess velocities resulting from the interactions of small bodies with plan- ets of different mass Mp located in a circular orbit ap = 5 AU around a solar mass-star. The results are given in canonic units for which we have set G = 1, uL = 1 AU L/(GuM ), and and uM = 1 M(cid:12). In these units, uT =(cid:112)u3 uL/uT = 30 km/s. An interesting advantage of expressing the results in canonic units is that they can be used to predict ejection velocities distribution from planetary systems around stars of arbitrary mass. Thus, for instance, the average ejection velocity for a planet with the mass of Jupiter around a solar mass star, uM = 1 M(cid:12), Mp = 0.001 uM is ¯v∞ = 0.2 uL/uT , or 6 km/s (see the leftmost curve in Figure 6). If we consider now an early M-dwarf with mass M(cid:63) = 0.5 M(cid:12), the same results in Figure 6 will ap- ply, but now for the case of a planet with half the mass of Jupiter. The value of ¯v∞ in km/s, will depend on what value is assumed for uL. If we take uL = 1 AU (as in the case of the solar-mass star), then uL/uT = 20 km/s for the early M-dwarf, and the average ejection velocity will be 4 km/s. In Figure 7 we show contours of ¯v∞ in the ap − Mp plane. We discover that for masses below 10−3 uM ejec- tion velocities are only a function of planetary orbital distance and are very insensitive to planetary mass. Another interesting result from our semi analytical ex- 0.00.20.40.60.8v(uL/uT)ap=5.0uLMp=0.001uMMp=0.005uMMp=0.010uM104103Mp (uM)0.51.02.03.04.05.0a (uL)0.200.300.400.500.600.700.800.900.170.250.330.410.490.570.650.730.810.89v (uL/uT)104103Mp (uM)0.450.500.550.600.65v/v12 Fig. 8. -- Ejection velocity posterior distribution as estimated in this work. Once we have a posterior probability distribution for ejection speeds (see Figure 8), the factor find in the orig- inal IOP expression (Equation 2) can be estimated as: (cid:26) p(vi ind ∝ f i rel) pB (cid:29) pind pind (cid:29) pB 1 (13) "integrating" across the intersection volume and as- suming that the number of background objects com- ing out from the stellar system is much larger than the number of indigenous objects, the joint position-velocity probability Ppos,vel will be proportional to: n( (cid:126)rp i)p(vi rel) (14) Ppos,vel ∝ 1 N 2 s (cid:88) (cid:26) Ppos i Finally the IOP for the star will be: IOP = pind (cid:29) pB Ppos,vel pB (cid:29) pind (15) 9.2. Normalization of the IOP The normalization of the interstellar origin probability will depend on what we define as the "sample universe" Ω of our probability space. Galaxy, the sample universe will be Ω =(cid:83) If we assume for instance, that all interstellar objects coming into the Solar System are ejected from stars in the n sn, where sn is the event "the interstellar object comes from the n-th star". If we assume that all the event sn are independent, then: P (Ω) = IOP(n) = 1 where IOP(n) is the interstellar origin probability of the n-th star as computed with Equation 15. If this case, the normalization constant, following Equation 2, will simply be: (cid:33)−1 Nstellar = IOP(n) (16) (cid:88) n (cid:32)(cid:88) n and the normalized IOP probability of the n-th star can be obtained multiplying the value in the right hand side of Equation 11 or Equation 14 by this constant. If on the other hand, we admit that some interstel- lar objects could come from processes different than the ejection from a stellar system, then the sample space will be larger, and hence Nstellar will be an overestimation of the actual normalization constant. Moreover, since we have in our experiments a limited set of stars (nearby stars with complete astrometric in- formation), the normalization constant estimated with Equation 16 will also represent an overestimation of the true one. Therefore the "normalized" IOP probability will also overestimate the true one. Still, the IOP computed under our assumptions and with a sample-limited normalization, will be good enough to sort-out our potential progenitors and to concentrate our potential follow-up efforts in those having the largest IOP values. 10. RESULTS min < tRet ≈ 40 Myr and dnom As an illustrative example (not necessarily the best one, but the only to date), we apply our methodology to assess the interstellar origin of 'Oumuamua. In Table 4 we present a list of the potential progenitors satisfying the conditions tnom min < 2 pc. For each progenitor we present several statistics of the encounter conditions, namely, encounter time tmin, minimum encounter distance dmin, and relative velocity vrel. Along with the nominal value of these quantities (first row of each entry) we provide the value of the 10%, 50% (median) and 90% percentiles. In the case of min- imum distance, providing the value of the percentiles is uninformative if the cloud of points representing the rel- ative position of surrogate objects and surrogate stars, surrounds the nominal relative position. In this case we have calculated and reported a new statistics, the "cen- tering parameter" f , defined as the fraction of points at a distance less than or equal to dmin,nom from the nom- inal relative position. If the cloud of objects is perfectly centered around the nominal relative position (an ideal configuration for an actual progenitor), f = 0. If on the other hand the cloud is off by several times its own dis- persion, then f ∼ 0.5 (independent of distance). When the cloud is centered, it is better to provide the mini- mum value of dmin which is precisely the second number between brackets in Table 4. When the cloud is decen- tered, it is better to read the 10% and 90% percentiles which are provided as the third and forth figures between brackets. In the last two columns of the table the value of the IOP are reported. We have included both, the value of Ppos and Ppos,vel. Thus, the IOP can be judged accord- ing to the the two extreme cases in Equation 15. In all cases, the IOP values have been normalized following the procedure describe in subsection 9.2. To provide an idea of the astrometric uncertainties in- volved in the calculation of the origin probability, we have tabulated for each potential progenitor, an "astro- metric quality index" q, defined as the minimum ratio between the value of each astrometric key property (par- allax, proper motion and radial velocity) and its standard error. Therefore, a quality factor of 1 implies that one of these quantities has an error of the same order than 0.00.20.40.60.81.0v (uL/uT)0.00.51.01.52.02.53.03.5pv(v)its magnitude (poor astrometry). On the other hand, a large q-value are indicative of the availability of very precise astrometric properties (including radial velocity). Using the available information, we identify only 16 po- tential progenitors for 'Oumuamua fulfilling all the selec- tion criteria. Most of the potential progenitors have mod- erately large q-values and are located at distances beyond 5 pc. As expected the IOP probability has achieved at se- lecting candidates with moderate relative velocities and encounter distances between 0.2 and 5 pc (with a few exceptions, eg. TYC 7582-1449-1 that also has a poor astrometry). The method presented here does not necessarily in- tend to identify a single object as the actual progenitor of 'Oumuamua. Finding the origin would require follow- up observations of the interstellar object (while reach- able), and improving the astrometric properties of the potential progenitors. When more and better informa- tion be available about these and other stars, the list could be extended or reduced, and more importantly the IOP probability could be modified. Still, it is interesting to notice the properties of several of these progenitor candidates. The most interesting case is of course that of HD 200325 (HIP 103749), the first object in the list. The star is probably a double or multiple system (Cvetkovi´c 2011). It is located at the present at a distance of 53.8 pc from the Sun and their physical properties are well constrained (see eg. Cvetkovi´c 2011; Holmberg et al. 2007). Its radial velocity has been measured very pre- cisely (vr = −11.10 ± 0.4 km/s) and their astrometric properties are relatively well known (the object is in the Hipparcos catalog but not in the GAIA TGAS set). Its most uncertain astrometric property is the declination proper motion µdec = 0.90 ± 0.60 mas, which is con- sistent with its low q-value. We expect that better as- trometric parameters be obtained and published in the next Gaia Data Release. The uncertain proper motion is the reason why the minimum encounter distance has a large uncertainty, ie. 0.5-5 pc. The mass of the main component of the HD 200325 system is 1.19 ± 0.1 M(cid:12), and its age is around 3.2 Gyr. The companion seems to be a low mass K-dwarf (Cvetkovi´c 2011) located at ∼ 25 AU from the primary. Although no planet has been discovered yet around the primary star, and its binary nature may prevent the formation of giant planets (The- bault 2011), the stars are far apart and their masses are very dissimilar. Interestingly, there is a known binary system with similar characteristics, HD 41004 (a solar mass evolved primary with a low-mass companion at 20 AU) around which a jupiter-mass planet has been dis- covered at ∼ 1 AU from the primary (Zucker et al. 2004). The existence of this "doppelganger", together with re- cent theoretical evidence that shows that formation of planets around this kind of binaries could not be as im- probable as thought (Higuchi & Ida 2017), lead us to speculate that HD200325 may harbor a planetary system and probably be the source of ejected small bodies. The possibility that 'Oumuamua be an ejecta of a binary sys- tem has been already considered by other authors ( ´Cuk 2018; Raymond et al. 2017) which give some theoretical support to our speculation. Our results match well the works by ´Cuk (2018) and 13 Raymond et al. (2017), that predict a non-catastrophic origin of 'Oumuamua in the neighborhood of a binary stellar system. It is, however, too early to conclude that 'Oumuamua comes actually from our best potential progenitor. It was not either our aim proposing it. However, We expect that improved astrometric information about this and other stellar systems included in our AstroRV catalogue, be available with the Gaia Data Release 2 and help us to improve the IOP for our best candidates or to find even better ones. 11. DISCUSSION When dealing with very uncertain processes such as those involved in this problem, it is important to ask if the identified close encounters could be just the product of chance. Further numerical experiments should be per- formed to test this idea and will be presented in a future work. At least three groups, that of Portegies Zwart et al. (2017), Dybczy´nski & Kr´olikowska (2017) and Feng & Jones (2018) published their own list of candidates. Some of their objects are among the candidates in the list in Table 4, but others are not there. We searched for the "missing" objects among our AstroRV catalog and find that either some of their candidates were not included in our input catalog or have properties (relative velocities, time for minimum approach) too large for our particular selection criteria. This fact put in evidence a limitation of any approach to asses the origin of an interstellar object: the completeness of the database. The approach presented here to estimate ejection ve- locities of small bodies from planetary system, is only our first attempt to model what should be for sure a more complex problem. Although a lot of interest have been paid in to model the flux of planetesimals coming out from young planetary systems, predicting the direction and velocities of these ejected objects has received less attention. The case of interstellar objects and the inves- tigation of their origin could encourage more research in this field. Thus, for instance, improved semi-analytical models and detailed numerical n-body simulations may be required to better constraint the kinematical proper- ties of ejected small bodies from already formed plane- tary systems around single and multiple stellar systems. We have already performed several basic n-body sim- ulations to investigate the problem that confirms some of our semi analytical results but also seems to predict lower ejection velocities in some regions of the parameter space. Although trillions of interstellar small objects are wan- dering around the Solar System and most of them could be there for hundreds of millions if not billions of years, the effort for tracing back the origin of some of those that enter for chance into the inner Solar System, is not irrel- evant. Although many stars may have contributed in the history of Galaxy to populate this graveyard, of course nearby stellar system could be an important source of many of these objects. Assessing the origin of interstellar objects require that the small uncertainties in the initial kinematic parame- ters do not propagate into large errors in the resulting dynamical properties due to factors related with the sim- ulation process. Some sources of errors include but are 14 # Name 1 2 3 4 5 6 7 8 9 HIP 103749 (HD 200325) TYC 3144-2040-1 d∗ (pc) 53.8 4.5 TYC 7069-1289-1 8.3 HIP 3821 (* eta Cas) TYC 3663-2669-1 HIP 91768 (HD 173739) HIP 91772 (HD 173740) TYC 6573-3979-1 HIP 18453 (* 43 Per) 6.0 6.1 3.5 3.5 6.5 37.4 10 TYC 7582-1449-1 192.2 11 TYC 7142-1661-1 21.7 12 HIP 63797 118.1 (HD 113376) 13 HIP 101180 (LHS 3558) 8.1 14 TYC 7093-310-1 6.7 q 1 2 1 85 34 11 10 2 30 1 1 11 52 1 tmin (Myr) −4.22 [−4.41, −4.22, −4.05] −0.12 [−0.12, −0.11, −0.11] −0.39 [−0.41, −0.38, −0.26] −0.17 [−0.17, −0.17, −0.17] −0.17 [−0.17, −0.17, −0.16] −0.03 [−0.03, −0.03, −0.03] −0.03 [−0.03, −0.03, −0.03] −0.18 [−0.18, −0.18, −0.17] −0.86 [−0.87, −0.86, −0.85] −8.96 [−9.20, −8.83, −8.59] −0.62 [−0.62, −0.59, −0.54] −2.90 [−3.33, −2.83, −2.49] −0.17 [−0.17, −0.17, −0.16] −0.19 [−0.21, −0.20, −0.19] −0.03 [−0.03, −0.03, −0.03] −0.24 [−0.25, −0.24, −0.24] dmin (pc) 1.75 vrel (km/s) 12.0 log IOP (cid:104)find(cid:105) Ppos,vel −0.5 Ppos −1.6 −1.9 [f = 0.55, min = 0.31, 0.37, 4.98] [11.4, 12.0, 12.5] 1.00 17.9 [f = 0.60, min = 0.86, 0.88, 1.14] [17.3, 18.0, 18.5] 0.99 24.6 [f = 0.30, min = 0.53, 0.61, 5.25] [23.3, 25.1, 30.6] 1.26 23.5 [f = 0.60, min = 1.23, 1.23, 1.27] [23.3, 23.5, 23.6] 1.34 23.9 [f = 0.65, min = 1.13, 1.20, 1.46] [23.1, 23.8, 24.3] 0.82 36.8 [f = 0.50, min = 0.81, 0.81, 0.82] [36.7, 36.8, 36.9] 0.76 39.3 [f = 0.55, min = 0.75, 0.75, 0.76] [39.1, 39.3, 39.4] 0.95 44.6 [f = 0.55, min = 0.24, 0.27, 1.81] [44.1, 44.7, 45.8] 0.86 41.0 [f = 0.25, min = 0.76, 0.80, 1.42] [40.6, 41.1, 41.5] 1.26 22.1 [f = 0.05, min = 1.26, 2.59, 25.59] [21.8, 22.4, 23.3] 0.75 36.9 [f = 0.06, min = 0.75, 1.43, 14.50] [36.8, 37.9, 47.4] 1.00 40.2 [f = 0.25, min = 0.50, 0.67, 3.28] [35.0, 41.2, 46.8] 1.58 32.6 [f = 0.50, min = 1.57, 1.58, 1.60] [32.4, 32.6, 32.7] 1.96 40.3 [f = 0.50, min = 1.08, 1.56, 2.50] [38.6, 40.2, 41.7] 1.47 38.7 [f = 0.65, min = 1.47, 1.47, 1.47] [38.6, 38.7, 38.8] 1.94 34.8 [f = 0.45, min = 1.90, 1.91, 1.96] [34.6, 34.8, 34.9] −1.8 −2.9 −1.8 −3.7 −2.8 −3.4 −3.0 −3.5 −1.2 −5.5 −1.1 −5.8 −0.8 −6.6 −1.8 −6.1 −4.6 −3.6 −2.9 −5.7 −2.5 −5.7 −4.4 −4.9 −3.7 −5.9 −3.8 −5.8 −6.6 −5.2 −1.6 −2.3 −3.1 −3.4 −3.6 −3.8 −4.3 −4.8 −5.3 −5.4 −5.8 −6.2 −6.5 −6.5 −8.7 Basic properties Encounter conditions 15 HIP 1475 (V* GX And) 16 HIP 21553 (HD 232979) 3.6 9.9 106 171 Interstellar origin probability (IOP) for a selected group of nearby stars. TABLE 4 not restricted to galactic coordinate transformation, un- certainties in the galactic parameters and of course errors in coding and processing the data. In the same way as the trajectory can be propagated backward, it could also be propagated forward in time to predict the fate of these interstellar interlopers. At studying their fate we can also learn interesting things that could shade some light on their own origin. 12. SUMMARY AND CONCLUSIONS In this paper we presented a general method for calcu- lating the probability that nearby stars be the source of an interstellar small object detected inside the Solar Sys- tem. The method relies on the availability of a precisely determined orbit for the object and precise astrometric information about a large enough number of nearby stars. For illustrating the method we applied it for assessing the origin of 'Oumuamua, the first identified interstellar interloper. The application of our method to the case of 'Oumua- mua, allowed us to identify a handful of stars whose kine- matical and physical properties are compatible with the ejection of a small object in the latest couple of Myrs. Of particular interest, at least with the available infor- mation, is the binary(multiple) system HD200325. The system is dominated by a primary star 1.2 M(cid:12) with a K-dwarf companion at ∼ 25 AU. Although no planetary system have been discovered yet around the primary or the secondary star, several similar multiple systems have been discovered in the past with planets; this fact, to- gether with multiple recent theoretical evidences, suggest that the case for HD200325 as 'Oumuamua progenitor is not as unlikely as previously thought. Our method is not intended to identify a single object as the definitive progenitor. Even with small uncertain- ties in the initial orbit and in the astrometric parame- ters of the stars, there will be always large enough un- certainties in the resulting kinematical properties that constrained our capability to pinpoint a single source. Our aim is to identify stars whose properties could be studied in more detail to reduce the uncertainties and increase/decrease the probability that they can be the sources of these objects. One of the most interesting features of our method is the fact that IOP probabilities can be published and updated permanently when new and better information be obtained. A catalog of potential progenitors for this and other future discovered interstellar objects can also be compiled and published with a global ranking of IOP probabilities. The authors believe that it could be a time in the future when this and other efforts could allows us to pinpoint precisely the provenance of an interstellar object. Those will be the times when instead of going to other planetary systems we will be able to study them using natural probes flying through our Solar System. ACKNOWLEDGEMENTS We thank Coryn Bailer-Jones for providing some of the data used in this work to compile the AstroRV cat- alog. We appreciate all the observations and sugges- tions received from our fellow colleagues F. Feng, E. Mamajek, M. ´Cuk and S. Raymond. We express our gratitude to the anonymous referee that carefully re- vised the manuscript and provide insightful comments and corrections that make possible the final version of this work. Some of the computations that made possi- ble this work were performed with Python 3.5 and their related tools and libraries, iPython (P´erez & Granger 2007), Matplotlib (Hunter et al. 2007), scipy and numpy (Van Der Walt et al. 2011). This work has made use of data from the European Space Agency (ESA) mis- sion Gaia (https://www.cosmos.esa.int/gaia), pro- cessed by the Gaia Data Processing and Analysis Con- sortium (DPAC, https://www.cosmos.esa.int/web/ gaia/dpac/consortium). 15 APPENDIX IWANDER PACKAGE We have provided with this work an open source package, iWander, that implements the general methodology described here. Providing a fully fledged computational tool is not only an effort to make these results reproducible, but also to allow the methodology to be applied by any researcher once future interstellar objects be discovered. The package can also serve as a basis or an inspiration to develop better computational tools for this and other related problems. Here we provide some basic information about the package that can be useful for users and developers: • The package is available at GitHub: http://github.com/seap-udea/iWander. • The required NASA NAIF SPICE kernels as well as the libraries required to compile them, are provided with the package while complying the corresponding licenses, in order to ease its compilation and use. • The core modules of the package were written in C/C++ to guarantee computing efficiency. As a result they should be compiled before usage. Other post processing modules were written in Python and are provided as core python scripts as well as iPython notebooks. • One of the key components of the methodology and the package are the astrometric and radial velocities catalogs. Although all of them are publicly available, we also provide them with the package. This is to allow future developers to attempt different merging and filtering strategies when compiling the input AstroRV catalog. • The version of the AstroRV catalog used in this work is also provided with the package. As a result, the full size of the local copy is almost 3 GB in size. A smaller size version of the package with a size of only 450 MB is available at http://github.com/seap-udea/iWander. • Any contribution to the development of the package is welcomed. We can provide full access to the developing branch to any researcher or developer interested in to contribute with this project. • The package, as well as the related databases will be updated as future and best astrometric and radial velocity information be published. REFERENCES Acton Jr, C. H. 1996, Planetary and Space Science, 44, 65 Adams, F. C., & Spergel, D. N. 2005, Astrobiology, 5, 497 Bailer-Jones, C. 2017, Astronomy & Astrophysics Bailer-Jones, C. A. L. 2015, A&A, 575, A35 Barbier-Brossat, M., & Figon, P. 2000a, Astronomy and Astrophysics Supplement Series, 142, 217 -- . 2000b, Astronomy and Astrophysics Supplement Series, 142, 217 Belbruno, E., Moro-Mart´ın, A., Malhotra, R., & Savransky, D. 2012, Astrobiology, 12, 754 Berski, F., & Dybczy´nski, P. A. 2016, A&A, 595, L10 Bolin, B. T., Weaver, H. A., Fernandez, Y. R., et al. 2018, ApJ, 852, L2 de la Fuente Marcos, C., & de la Fuente Marcos, R. 2017, Research Notes of the American Astronomical Society, 1, 5 Dybczy´nski, P. A., & Kr´olikowska, M. 2017, ArXiv e-prints Engelhardt, T., Jedicke, R., Veres, P., et al. 2017, AJ, 153, 133 ESA, ed. 1997, ESA Special Publication, Vol. 1200, The HIPPARCOS and TYCHO catalogues. Astrometric and photometric star catalogues derived from the ESA HIPPARCOS Space Astrometry Mission Famaey, B., Jorissen, A., Luri, X., et al. 2005, A&A, 430, 165 Feng, F., & Jones, H. R. A. 2018, ApJ, 852, L27 Ferrin, I., & Zuluaga, J. 2017, ArXiv e-prints Gaidos, E., Williams, J., & Kraus, A. 2017, Research Notes of the American Astronomical Society, 1, 13 Bovy, J. 2015, The Astrophysical Journal Supplement Series, 216, Galassi, M., Davies, J., Theiler, J., et al. 2002, Network Theory 29 Ltd, 3 Brown, A. G., Vallenari, A., Prusti, T., et al. 2016, Astronomy & Garc´ıa-S´anchez, J., Weissman, P. R., Preston, R. A., et al. 2001, Astrophysics, 595, A2 A&A, 379, 634 Bulirsch, R., & Stoer, J. 1966, Numerische Mathematik, 8, 1 Casagrande, L., Schoenrich, R., Asplund, M., et al. 2011, Astronomy & Astrophysics, 530, A138 Chambers, K. C., Magnier, E. A., Metcalfe, N., et al. 2016, ArXiv e-prints ´Cuk, M. 2018, ApJ, 852, L15 Cvetkovi´c, Z. 2011, The Astronomical Journal, 141, 116 Gontcharov, G. A. 2006, Astronomy Letters, 32, 759 Gragg, W. B. 1965, Journal of the Society for Industrial and Applied Mathematics, Series B: Numerical Analysis, 2, 384 Higuchi, A., & Ida, S. 2017, The Astronomical Journal, 154, 88 Høg, E., Fabricius, C., Makarov, V. V., et al. 2000, A&A, 355, L27 Holmberg, J., Nordstrom, B., & Andersen, J. 2007, Astronomy & Astrophysics, 475, 519 16 Hunter, J. D., et al. 2007, Computing in science and engineering, Portegies Zwart, S., Pelupessy, I., Bedorf, J., Cai, M., & Torres, 9, 90 Ivezic, Z., Tyson, J. A., Abel, B., et al. 2008, ArXiv e-prints Jewitt, D., Luu, J., Rajagopal, J., et al. 2017, ApJ, 850, L36 Johnson, D. R., & Soderblom, D. R. 1987, The Astronomical Journal, 93, 864 S. 2017, ArXiv e-prints Price, D. J. 2012, Journal of Computational Physics, 231, 759 Raymond, S. N., Armitage, P. J., Veras, D., Quintana, E. V., & Barclay, T. 2017, ArXiv e-prints Schonrich, R., Binney, J., & Dehnen, W. 2010, Monthly Notices Karim, M. T., & Mamajek, E. E. 2016, Monthly Notices of the of the Royal Astronomical Society, 403, 1829 Royal Astronomical Society, stw2772 Kolmogorov, A. N. 1968, Foundations of the Theory of Probability (Courier Dover Publications) Sen, A. K., & Rama, N. C. 1993, A&A, 275, 298 Strom, R. G., Malhotra, R., Ito, T., Yoshida, F., & Kring, D. A. 2005, Science, 309, 1847 Kunder, A., Kordopatis, G., Steinmetz, M., et al. 2017, The Strom Robert, G., Renu, M., Zhi-Yong, X., et al. 2015, Research Astronomical Journal, 153, 75 Kuzmin, G. 1956, Astronomicheskii Zhurnal, 33, 27 Malaroda, S., Levato, H., Morrell, N., et al. 2000, A&AS, 144, 1 Maldonado, J., Mart´ınez-Arn´aiz, R. M., Eiroa, C., Montes, D., & Montesinos, B. 2010, A&A, 521, A12 in Astronomy and Astrophysics, 15, 407 Thebault, P. 2011, Celestial Mechanics and Dynamical Astronomy, 111, 29 Trilling, D. E., Robinson, T., Roegge, A., et al. 2017, ApJ, 850, L38 Mamajek, E. 2017, Research Notes of the American Astronomical Van Der Walt, S., Colbert, S. C., & Varoquaux, G. 2011, Society, 1, 21 Martell, S., Sharma, S., Buder, S., et al. 2016, Monthly Notices of the Royal Astronomical Society, stw2835 Masiero, J. 2017, ArXiv e-prints McGlynn, T. A., & Chapman, R. D. 1989, ApJ, 346, L105 Melosh, H. J. 2003, Astrobiology, 3, 207 Merlin, F., Hromakina, T., Perna, D., Hong, M. J., & Alvarez-Candal, A. 2017, A&A, 604, A86 Miyamoto, M., & Nagai, R. 1975, Publications of the Astronomical Society of Japan, 27, 533 Napier, W. M. 2004, MNRAS, 348, 46 Opik, E. 1932, The Scientific Monthly, 35, 109 P´erez, F., & Granger, B. E. 2007, Computing in Science & Engineering, 9, 21 Perryman, M. A. C., Lindegren, L., Kovalevsky, J., et al. 1997, A&A, 323, L49 Computing in Science & Engineering, 13, 22 Wenger, M., Ochsenbein, F., Egret, D., et al. 2000, A&AS, 143, 9 Wiegert, P. A. 2011, in Meteoroids: The Smallest Solar System Bodies, ed. W. J. Cooke, D. E. Moser, B. F. Hardin, & D. Janches, 106 Wiegert, P. A. 2014, Icarus, 242, 112 Ye, Q.-Z., Zhang, Q., Kelley, M. S. P., & Brown, P. G. 2017, ApJ, 851, L5 Zhang, Q. 2018, The Astrophysical Journal Letters, 852, L13 Zucker, S., Mazeh, T., Santos, N., Udry, S., & Mayor, M. 2004, Astronomy & Astrophysics, 426, 695 Zuluaga, J. I., Ferrin, I., & Geens, S. 2013, ArXiv e-prints Zuluaga, J. I., & Sucerquia, M. 2018, Monthly Notices of the Royal Astronomical Society, sty702
1304.2815	1	1304	2013-04-09T23:39:30	Very Low Mass Stellar and Substellar Companions to Solar-Like Stars From MARVELS V: A Low Eccentricity Brown Dwarf from the Driest Part of the Desert, MARVELS-6b	[ "astro-ph.EP" ]	We describe the discovery of a likely brown dwarf (BD) companion with a minimum mass of 31.7 +/- 2.0 M_Jup to GSC 03546-01452 from the MARVELS radial velocity survey, which we designate as MARVELS-6b. For reasonable priors, our analysis gives a probability of 72% that MARVELS-6b has a mass below the hydrogen-burning limit of 0.072 M_Sun, and thus it is a high-confidence BD companion. It has a moderately long orbital period of 47.8929 +0.0063/-0.0062 days with a low eccentricty of 0.1442 +0.0078/-0.0073, and a semi-amplitude of 1644 +12/-13 m/s. Moderate resolution spectroscopy of the host star has determined the following parameters: T_eff = 5598 +/- 63, log g = 4.44 +/- 0.17, and [Fe/H] = +0.40 +/- 0.09. Based upon these measurements, GSC 03546-01452 has a probable mass and radius of M_star = 1.11 +/- 0.11 M_Sun and R_star = 1.06 +/- 0.23 R_Sun with an age consistent with less than ~6 Gyr at a distance of 219 +/- 21 pc from the Sun. Although MARVELS-6b is not observed to transit, we cannot definitively rule out a transiting configuration based on our observations. There is a visual companion detected with Lucky Imaging at 7.7 arcsec from the host star, but our analysis shows that it is not bound to this system. The minimum mass of MARVELS-6b exists at the minimum of the mass functions for both stars and planets, making this a rare object even compared to other BDs.	astro-ph.EP	astro-ph	Accepted by The Astronomical Journal on April 8th, 2013 Preprint typeset using LATEX style emulateapj v. 12/16/11 VERY LOW MASS STELLAR AND SUBSTELLAR COMPANIONS TO SOLAR-LIKE STARS FROM MARVELS V: A LOW ECCENTRICITY BROWN DWARF FROM THE DRIEST PART OF THE DESERT, MARVELS-6b Nathan De Lee1,2,3, Jian Ge3, Justin R. Crepp4, Jason Eastman5,6,7, Massimiliano Esposito8,9, Bruno Femen´ıa8,9, Scott W. Fleming3,10,11,12, B. Scott Gaudi5, Luan Ghezzi13,14, Jonay I. Gonz´alez Hern´andez8,9, Brian L. Lee3,15, Keivan G. Stassun1,2, John P. Wisniewski16, W. Michael Wood-Vasey17, Eric Agol15, Carlos Allende Prieto8,9, Rory Barnes15, Dmitry Bizyaev18, Phillip Cargile1, Liang Chang3, Luiz N. Da Costa13,14, G.F. Porto De Mello19,14, Leticia D. Ferreira19,14, Bruce Gary1, Leslie Hebb1,15, Jon Holtzman20, Jian Liu3, Bo Ma3, Claude E. Mack III1, Suvrath Mahadevan10,11, Marcio A.G. Maia13,14, Duy Cuong Nguyen3,21, Audrey Oravetz18, Daniel J. Oravetz18, Martin Paegert1, Kaike Pan18, Joshua Pepper1, Elena Malanushenko18, Viktor Malanushenko18, Rafael Rebolo 8,9,22, Basilio X. Santiago23,14, Donald P. Schneider10,11, Alaina C. Shelden Bradley18, Xiaoke Wan3, Ji Wang3, Bo Zhao3 Accepted by The Astronomical Journal on April 8th, 2013 ABSTRACT −0.0073, and a semi-amplitude of 1644+12 We describe the discovery of a likely brown dwarf (BD) companion with a minimum mass of 31.7 ± 2.0 MJup to GSC 03546-01452 from the MARVELS radial velocity survey, which we designate as MARVELS-6b. For reasonable priors, our analysis gives a probability of 72% that MARVELS- 6b has a mass below the hydrogen-burning limit of 0.072 M⊙, and thus it is a high-confidence BD companion. It has a moderately long orbital period of 47.8929+0.0063 −0.0062 days with a low eccentricty of 0.1442+0.0078 −13 m s−1. Moderate resolution spectroscopy of the host star has determined the following parameters: Teff = 5598 ± 63, log g = 4.44 ± 0.17, and [Fe/H] = +0.40 ± 0.09. Based upon these measurements, GSC 03546-01452 has a probable mass and radius of M∗ = 1.11 ± 0.11 M⊙ and R∗ = 1.06 ± 0.23 R⊙ with an age consistent with less than ∼6 Gyr at a distance of 219 ± 21 pc from the Sun. Although MARVELS-6b is not observed to transit, we cannot definitively rule out a transiting configuration based on our observations. There is a visual companion detected with Lucky Imaging at 7.7′′ from the host star, but our analysis shows that it is not bound to this system. The minimum mass of MARVELS-6b exists at the minimum of the mass functions for both stars and planets, making this a rare object even compared to other BDs. It also exists in an underdense region in both period/eccentricity and metallicity/eccentricity space. Keywords: stars: individual (GSC 03546-01452) 1 Department of Physics and Astronomy, Vanderbilt Univer- sity, Nashville, TN 37235, USA;[email protected] 2 Department of Physics, Fisk University, Nashville, TN, USA 3 Department of Astronomy, University of Florida, 211 Bryant Space Science Center, Gainesville, FL, 32611-2055, USA 4 Department of Physics, University of Notre Dame, 225 Nieuwland Science Hall, Notre Dame, IN 46556, USA 5 Department of Astronomy, The Ohio State University, 140 West 18th Avenue, Columbus, OH 43210, USA 6 Las Cumbres Observatory Global Telescope Network, 6740 Cortona Drive, Suite 102, Santa Barbara, CA 93117, USA 7 Department of Physics Broida Hall, University of California, Santa Barbara, CA 93106, USA 8 Instituto de Astrof´ısica de Canarias (IAC), E-38205 La La- guna, Tenerife, Spain 9 Departamento de Astrof´ısica, Universidad de La Laguna, 38206 La Laguna, Tenerife, Spain 10 Department of Astronomy and Astrophysics, The Pennsyl- vania State University, 525 Davey Laboratory, University Park, PA 16802, USA 11 Center for Exoplanets and Habitable Worlds, Pennsylvania State University, University Park, PA 16802, USA 12 Space Telescope Science Institute, 3700 San Martin Drive, Baltimore, MD 21218 13 Observat´orio Nacional, Rua Gal. Jos´e Cristino 77, Rio de Janeiro, RJ 20921-400, Brazil 14 Laborat´orio Interinstitucional de e-Astronomia - LIneA, Rua Gal. Jos´e Cristino 77, Rio de Janeiro, RJ 20921- 400, Brazil 15 Astronomy Department, University of Washington, Box 351580, Seattle, WA 98195, USA 16 H L Dodge Department of Physics and Astronomy, Univer- sity of Oklahoma, 440 W Brooks St Norman, OK 73019, USA 17 Pittsburgh Particle physics, Astrophysics, and Cosmology Center (PITT PACC), Department of Physics and Astronomy, University of Pittsburgh, Pittsburgh, PA 15260, USA 18 Apache Point Observatory, P.O. Box 59, Sunspot, NM 88349-0059, USA 19 Universidade Federal do Rio de Janeiro, Observatrio do Va- longo, Ladeira do Pedro Antonio 43, 20080-090 Rio de Janeiro, Brazil 20 Department of Astronomy, New Mexico State University, Box 30001, Las Cruces, NM 880033, USA 21 Department of Physics and Astronomy, University of Rochester, Rochester, NY 14627-0171, USA 22 Consejo Superior de Investigaciones Cient´ıficas, Spain 23 Instituto de Fisica, UFRGS, Porto Alegre, RS 91501-970, Brazil 2 De Lee et al. 1. INTRODUCTION Radial velocity (RV) surveys have provided a wealth of exoplanet discoveries around sun-like stars in recent years (California Planet Survey, Howard et al. (2010), Lick- Carnegie Exoplanet Survey, Haghighipour et al. (2010); CORALIE survey, Udry et al. (2000); and the HARPS survey, Mayor et al. (2003) to name a few), but they have not found a correspondingly large number of brown dwarf (BD) companions (Reid & Metchev 2008). BD companions lie on the mass spectrum between planets and stars and are defined as being between 13MJup and 75.5 MJup (based on the deuterium and hydrogen fusion limits) (Chabrier et al. 2000; Spiegel et al. 2011). The lack of BDs within 3 AU of their host star was first rec- ognized in Marcy & Butler (2000), and is known as the BD "desert". This result is unlikely to be due to obser- vational bias because the RV semiamplitudes of BD are many hundreds to a few thousand meters per second, which is easily detectable by these surveys (Patel et al. 2007). Thus the lack of BD at close to moderate dis- tances from their respective host stars points to an ex- planation based in BD formation mechanisms. Although the formation of low-mass companions (plan- ets through low mass stars) to sun-like stars is still an area of active research, in overview there are two main mechanisms: planets form from a protoplanetary disks and stellar companions [in a similar range of separations] from molecular cloud fragmentation. Given the mass range of BD companions they could form through either mechanism (or both). Understanding the origin of the BD desert can put major constraints on the upper mass limit for companion formation in protoplanetary disks, and a lower mass limit on formation via fragmentation. Recent efforts to quantify the frequency of companions as a function of mass in the BD desert have found that the overall frequency of BD companions at close to mod- erate distances from their host star (≤ 10 AU) is less than < 1% (Grether & Lineweaver 2006), and more recently 0.6% (Sahlmann et al. 2011). This value is low com- pared to ∼ 7% for planetary companions (Udry & Santos 2007) and ∼ 13% for stellar companions in a sim- ilar range of separations (Duquennoy & Mayor 1991; Halbwachs et al. 2003). Grether & Lineweaver (2006) went a step further and defined the driest part of the desert to be where there was a minimum in the number of companions per unit interval in log mass; they found this position to be at a companion mass of 31+25 −18MJup. This is the fifth paper in this series looking at low-mass companions to sun-like stars from the third generation of the Sloan Digital Sky Survey (SDSS-III; Eisenstein et al. (2011)) Multi-Object APO Radial Velocity Exoplanet Large-Area Survey (MARVELS; Ge et al. 2008, 2009; Ge & Eisenstein 2009). The primary goal of this series is to provide a detailed set of well-characterized compan- ions with minimum masses and separations in or near the BD desert, which can be used by future meta-analyzes. Ultimately, this is the sort of groundwork that must be done in order to understand the extent and aridity of the BD desert. The MARVELS survey measured radial velocities of 3,300 unique FGK type stars. MARVELS is a large survey looking for RV companions around bright stars (7.6 ≤ V ≤ 12) with periods below 2 years with well 600 500 400 300 200 100 ) w ( Z 0 1 80 60 40 20 0 1 ) w ( Z e g a r e v A 50% FAP 1% FAP 0.1% FAP Peak Period 10 100 1000 Period (days) Peak Period 10 100 1000 Period (days) Figure 1. Top: The original detection Lomb-Scargle periodogram for GSC 03546-01452 based on the MARVELS data. The two RV points with the lowest spectrum flux were removed for this initial periodogram (as marked in Table 1). The peak in the periodogram (marked by the red dashed line) is quite significant and is at 47.842 days, close to the final adopted value of 47.893 days for the com- bined MARVELS and TNG/SARG data. Bottom: A combined periodogram for all 60 stars on the MARVELS plate. The highest power point in each frequency bin was removed, and the remaining were averaged. The red dashed line shows the peak period from the periodogram on the left.There is no obvious systematic peak on the plate periodogram at the peak period of the GSC 03546- 01452 periodogram. characterized biases; see Lee et al. (2011) for a descrip- tion of the survey design. Other papers in this se- ries (Fleming et al. 2010, 2012; Wisniewski et al. 2012; Ma et al. 2013a, Mack et al. 2013, in press; Jiang et al. 2013, submitted) have helped fill in our understanding of the BD desert, and provided warnings to some of the pitfalls inherent in these analyzes. We will discuss observations of the star GSC 03546- 01452, which has a companion with a period of ∼ 47 days and with a minimum mass of 31.7 ± 2.0MJup placing it near the most "arid" region of the BD desert. In Section 2 we discuss the photometric and spectroscopic observa- tions and basic data processing. In Section 3 we discuss the analysis of this data. Section 4 contains a discussion of the results and places MARVELS-6b within the larger context. Section 5 summarizes our results. 2. OBSERVATIONS AND DATA REDUCTION MARVELS-6b 3 The initial detection and orbitial characterization of MARVELS-6b came from the MARVELS Survey RV data. Once it became a strong candidate further RV, photometric, time-series photometric, and spectroscopic data were taken to confirm this candidate. Each of these data sets will be discussed below. 2.1. Initial Identification of MARVELS Candidates 2.1.1. Survey Summary MARVELS is a multi-epoch radial velocity survey de- signed to detect radial velocity companions around FGK type stars in a magnitude range of (7.6 ≤ V ≤ 12). It uses a dispersed fixed-delay interferometer (DFDI; Ge et al. 2009) on the SDSS 2.5m telescope (Gunn et al. 2006). The DFDI method was introduced for use in a multi-object RV survey by Ge (2002). A single ob- ject version of a DFDI instrument was successfully used to detect a hot Jupiter around HD 102195 (Ge et al. 2006). The DFDI instrument principle was described by (Ge 2002; Ge et al. 2002; Erskine 2003; van Eyken et al. 2010; Wang et al. 2011). The MARVELS interferom- eter delay calibrations were described in Wang et al. (2012a,b). Each MARVELS observation consists of 60 stars spread across 120 spectra (2 spectra for each star). The MARVELS survey started in Oct 2008 and ran through July 2012. It was divided up into two observing sets: year 1-2 fields which were observed from Oct 2008 till Jan 2011 and year 3-4 fields which were observed from Jan 2011 till July 2012. The year 1-2 data set consisted of 43 unique fields (11 of which are in the Kepler [Borucki et al. 2010]) field of view. Year 3-4 field set contained 13 fields (with 1 field overlapping the year 1-2 field set.) This leads to a total of 3,300 unique stars with > 18 epochs. 2.1.2. Radial Velocity Analysis software The large number of RV targets required development a software package to easily display and characterize the radial velocity curves from the MARVELS survey. One of the primary tools used in this process is an IDL-based Keplerian model fitter known as MPRVFIT24. MPRV- FIT starts with the input of RV data in the form of Ju- lian date, RV, and RV error (the Julian date and RVs can be of any type that is appropriate to the analy- sis). In the case of MARVELS, the data from the two beams have been combined as described in Fleming et al. (2010). The software then applies a modified version of the Lomb-Scargle periodogram (Lomb 1976; Scargle 1982) described in Cumming (2004). False alarm proba- bilities were assigned to the highest peaks using the for- mulation of Baluev (2008). Once a number of high prob- ability peaks are identified, the frequency space between the peak and the next nearest frequency point in the periodogram is sub-divided into 10 frequency steps (on both sides of the peak). A Keplerian model is then fit to those frequencies using MPFIT (Markwardt 2009). MP- FIT is a Levenberg-Marquardt non-linear least squares fitter implemented in IDL. The χ2 statistic is determined for each fit, and the best chi-squared fit of the grid is re- tained. Finally, the best fit models are plotted with the 24 http://www.vanderbilt.edu/AnS/physics/vida/mprvfit.htm Table 1 Summary of Radial Velocity Data BJD (days) RV (m s−1) RV err (m s−1) Source · · · 2454956.915966 2454957.946510 2454958.912457 2454959.927110 2454962.924107 2454963.890111 2454964.918085 2454965.910038 2454984.835892 2454986.892198 2454988.905306 2454990.799823 2454993.804470 2454994.793741 2454995.898027 2455014.802962 2455020.845356 2455021.851131 2455023.835931 2455024.788660 2455484.614289 2455436.578251 2455460.467627 2455460.489293 2455460.511619 2455521.551045 2455725.687910 2455760.505697 2455760.694598 2455791.532669 2455791.583907 2455844.434368 2039.20 2150.89 2206.71 2291.25 2568.84 2559.01 2529.69 2579.81 -797.93 -736.07 -444.95 -154.83 255.99 543.09 594.44 2438.46 1639.58 1381.28 792.32 463.02 2165.40 1901.69 -841.33 -833.73 -836.81 722.14 1962.38 66.39 98.71 122.53 82.56 -834.69 72.10 MARVELSa 43.11 MARVELS 39.27 MARVELS 36.06 MARVELS 36.54 MARVELS 48.00 MARVELS 39.97 MARVELS 43.36 MARVELS 41.72 MARVELS 43.95 MARVELS 40.44 MARVELS 64.34 MARVELS 43.11 MARVELS 49.32 MARVELS 43.11 MARVELS 51.41 MARVELS 39.45 MARVELS 50.68 MARVELS 53.52 MARVELS 51.78 MARVELS 59.70 MARVELS 24.22 TNG/SARG 4.49 TNG/SARG 4.44 TNG/SARG 4.76 TNG/SARG 88.25 MARVELSb 6.86 TNG/SARG 9.99 TNG/SARG 5.87 TNG/SARG 5.96 TNG/SARG 5.52 TNG/SARG 5.08 TNG/SARG a Second lowest flux spectrum b The lowest flux spectrum data points for easy reference. For the MARVELS sur- vey, the two best fits for each target were displayed, and the candidates were chosen based on the folded and un- folded radial velocity curves, as well as the significance of the periodogram. An example of this periodogram for GSC 03546-01452 can be found in Figure 1. 2.2. Radial Velocities 2.2.1. SDSS-II MARVELS Radial Velocities Differential RV observations of GSC 03546-01452 were acquired during the first two years of the SDSS-III MAR- VELS survey. A total of twenty-two observations were obtained over the course of 565 days. As discussed in Section 2.1.1, the MARVELS survey uses the DFDI tech- nique which introduces an interferometer into the light path. As a result, each 50 minute observation includes two fringed spectra, or beams, one from each arm of the interferometer. Each spectrum spans a wavelength range roughly 500-570 nm with a R∼12,000 resolution. Each beam is processed individually through the MARVELS pipeline following the methods described in Lee et al. (2011). The formal errors derived from the MARVELS pipeline are known to be underestimates of the true error. This systematic underestimate can be partially corrected for by using the fact that 60 stars (120 beams) are taken in each observation. Following the method outlined in Fleming et al. (2010), it is expected that most stars in 4 De Lee et al. an observation plate are radial velocity stable, so the median RMS is a reasonable estimate of the systematic errors. A quality factor (QF) is derived for each beam, which is the ratio of the radial velocity RMS to the me- dian formal error bar, and the median QF is found for the plate. The formal errors are multiplied by this QF (2.334 for GSC 03546-01452) resulting in the error bars used for the analysis of these observations. During the RV analy- sis, discussed in Section 3.2, it was determined that these scaled error bars were themselves overestimated, leading to a reduction of factor of 0.5794. The final differential RV measurements and final scaled error bars (approxi- mately 40% larger than the formal uncertainties) for GSC 03546-01452 are presented in Table 1. 2.2.2. TNG Differential Radial Velocities Once GSC 03546-01452 was determined to be a can- didate for additional investigation, spectroscopic ob- servations were conducted with the Spettrografo Alta Risoluzione Galileo (SARG) (Gratton et al. 2001) on the 3.6m Telescopio Nazionale Galileo (TNG) located at Roque de Los Muchachos Observatory (ORM). All the spectra were acquired using the same instrumen- tal configuration: a slit with a sky-projected width of 0.8′′ achieving a resolving power of R = 57000; a yellow cross-dispersing grism providing the wavelength range 462 < λ < 792 nm. A total of 15 spectra were taken (11 with and 4 without the iodine cell inserted in the light path). One spectrum with the iodine cell proved to have too low S/N, leaving only 10 spectra suitable for differential velocity measure- ments. The spectra were processed using the standard IRAF25 Echelle reduction packages. The S/N per resolu- tion element at 550 nm ranges from 50 to 130. The spec- tra without I2 lines served as a reference for measuring the relative radial velocities of the 10 spectra with super- imposed I2 absorption lines. The technique adopted to derive RV values is described in Marcy & Butler (1992). For details on how the technique was implemented on SARG spectra, see Fleming et al. (2012). As is the case with the MARVELS RV data, the formal uncertainties for the TNG/SARG data were overestimated, and so the error bars were thus reduced by a factor of 0.2224. The final differential RV and reduced error bars are listed in Table 1. 2.2.3. TNG Absolute Radial Velocities All 15 spectra taken with the TNG/SARG instrument were used to calculate absolute RVs. The stellar spec- tra were cross-correlated with a high resolution solar spectrum26. The cross-correlation was done using the SARG red CCD spectrum with a wavelength coverage of 6200-8000 A. Due to the non-simultaneity of the ThAr wavelength calibration exposures it is possible the slit il- lumination varied between the science spectra and the calibration spectra, which could result in the RV mea- surement being affected by systematic errors. To par- tially mitigate this error source we calculated the cross- correlation function of the telluric lines (Griffin & Griffin 25 IRAF is distributed by the National Optical Astronomy Ob- servatories, which are operated by the Association of Universities for Research in Astronomy, Inc., under cooperative agreement with the National Science Foundation. 26 http://bass2000.obspm.fr/solar spect.php SARG/TNG Absolute Radial Velocities Table 2 BJD days RV (km s−1) RV Err (km s−1) 2455436.550647 2455436.578251 2455460.467627 2455460.489293 2455460.511619 2455698.630462 2455698.654826 2455725.687910 2455760.505697 2455760.694598 2455791.532669 2455791.583907 2455791.610041 2455791.635364 2455844.434368 -12.63 -12.73 -15.43 -15.35 -15.41 -15.28 -15.30 -12.48 -14.39 -14.32 -14.36 -14.22 -14.43 -14.52 -15.48 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 0.13 1973) around 6900 A with a numerical mask. This exer- cise results in a RV correction of a few hundred m s−1. This correction plus the barycentric correction were ap- plied to the radial velocities and the resulting values are shown in Table 2. To validate this method, observations of the RV stan- dard star HD3765 were used. The standard was observed at 8 epochs with a mean result of -63,155 m s−1 ± 128 m s−1. This result is within a 1-σ agreement with the mean 34 observations taken from the ELODIE archive of -63,286 ± 49 m s−1. For these observations, the system- atics caused by the non-simultaneity of the ThAr wave- length calibration is the dominant error source, so we adopt ± 128 m s−1 for the error in these measurements. 2.3. High-Resolution Spectrum for Stellar Classification Two R ∼31,500 optical (∼3600-10,000 A) spectra of GSC 03546-01452 were obtained on UT 2010 June 20 with the Apache Point Observatory 3.5 m telescope and ARC Echelle Spectrograph (ARCES; Wang et al. 2003) to enable accurate characterization of stellar fundamen- tal parameters (Teff , log g, and [Fe/H]). The two spec- tra were obtained using the default 1.6 x 3.2 arcsec slit and an exposure time of 1200 s for each spectra (for a combined exposure time of 2400 s). A ThAr lamp exposure was obtained after each of these integrations to facilitate accurate wavelength calibration. The data were processed using standard IRAF techniques. Fol- lowing heliocentric velocity corrections, each order was continuum-normalized, and the resultant continuum nor- malized data from each observation were averaged. The final spectrum yielded a signal-to-noise ratio (S/N) of approximately 110 per resolution element in the region around 6000 A. 2.4. Photometry 2.4.1. HAO Absolute Photometry We used the Hereford Arizona Observatory (HAO), a private facility in southern Arizona (observatory code G95 in the IAU Minor Planet Center), to measure multi- band, absolute photometry of GSC 03546-01452. HAO employs a 14-inch Meade Schmidt-Cassegrain (model LX200GPS) telescope, fork-mounted on an equatorial MARVELS-6b 5 wedge located in a dome. The telescope's CCD is an SBIG ST-10XME with a KAF-3200ME detector. A 10-position filter wheel accommodates SDSS and John- son/Cousins filter sets. GSC 03546-01452 and standard stars were observed with Johnson B and SDSS u', g', r', i' filters (Fukugita et al. 1996). For the Johnson- Kron-Cousins bands, standard stars are taken from the list published by Landolt & Uomoto (2007) and Landolt (2009). For the SDSS bands, standard stars are taken from the list published by Smith et al. (2002). Observa- tions were conducted on three dates: 24 and 25 April, 2010 and on 7 May, 2010. Between 23 and 75 stan- dard stars (Landolt and SDSS) were used to establish the transformations to the standard photometric systems. 2.4.2. Allegheny Lightcurve We obtained 44 nights of observations of GSC 03546- 01452 with the 16-inch Keeler RCX-400 Meade telescope at the Allegheny Observatory, University of Pittsburgh. The observations span 18 months from May 2010 through November 2011 and were all made through a Cousins R filter onto an SBIG KAF-6303E/LE 2048x3072 pixel CCD with a pixel scale of 0.57′′ pixel−1 for a total FoV of 19.5′ by 29.2′. Exposure times ranged from 30 -- 150 sec- onds depending on conditions, with a median exposure time of 75 seconds. Standard bias and dark subtraction and flatfield cal- ibrations were preformed on the raw data, and the im- ages were astrometrically calibrated to the 2MASS Point- Source Catalog (Skrutskie et al. 2006). Typical seeing conditions were between 3′′ -- 4′′ with a median seeing of 3.5′′. Photometry was accomplished using a 10-pixel (5.7′′) circular aperture and subtractring the estimated sky flux based on a 15 -- 20 pixel (8.6′′ -- 11.4′′) radius sky annulus. Typical 1-σ uncertainties were 5 millimag. Rel- ative photometry was calculated relative to the aver- age counts from two reference stars in the image with J2000 coordinates: (1): 19:11:34.733 +48:34:54.77; and (2) 19:11:56.335 +48:22:55.38. 2.4.3. SuperWASP Lightcurve The SuperWASP survey (Pollacco et al. 2006) is a wide-angle transiting planet survey that monitors the brightness of millions of stars. The survey uses a vi- sual broad band filter that covers the wavelength range from 400-700nm. For details on reduction techniques and survey design please consult Pollacco et al. (2006). The SuperWASP photometry for GSC 03546-01452 consists of 8823 points spanning just over four years from May 2004 to Aug 2008. 2.5. High Spatial Resolution Imaging Two high spatial resolution imaging campaigns of GSC 03546-01452 were undertaken in order to help rule out false positives and to look for hierarchical structure. In particular, these observations were conducted to search for visual companions at large separations that could in- fluence our spectroscopic results, or that might be bound tertiary companions to GSC 03546-01452. The adaptive optics imaging and lucky imaging runs are complimen- tary with the adaptive optics focusing near the star, and lucky imaging covering out to larger distances from the star. GSC 03546 NIRC2 AO, K'−filter August 31, 2011 2 3 4 5 6 7 ) m ∆ ( t s a r t n o C σ 0 1 0 0.5 1.0 1.5 2.0 angular separation (arcseconds) Figure 2. Contrast generated by NIRC2 AO observations of GSC 03546-01452 . Off-axis sources with relative brightness ∆K′ = 7 are ruled out at 10-σ for angular separations beyond ≈ 0.5′′. 2.5.1. Adaptive Optics Imaging We also acquired adaptive optics (AO) observations of GSC 03546-01452 to assess its multiplicity at wide separations. Images were obtained on 31 Aug., 2011 UT using NIRC2 (PI: Keith Matthews) at the 10m Keck II telescope (Wizinowich et al. 2000). GSC 03546- 01452 (V=11.7) is sufficiently bright to serve as its own natural guide star. Our observations consist of 9 dithered images (10 coadds per frame, 0.5s per coadd) taken with the K ′ filter (λc = 2.12 µm). We used NIRC2's narrow camera setting, which has a plate scale of 10 mas pixel−1, to provide fine spatial sampling of the instrument point- spread function. Raw frames were processed by clean- ing hot pixels, flat-fielding, subtracting background noise from the sky and instrument optics, and aligning and coadding the results. No off-axis sources were noticed in individual frames or the final processed image. Figure 2 shows the contrast levels generated by the observations. Our diffraction-limited images rule out the presence of companions down to ∆K ′ = 5.7, 7.1, 7.5 mags at angular separations of 0.25", 0.5", and 1.0" respectively. 2.5.2. Lucky Imaging Lucky imaging (LI) involves acquiring large numbers of observations with a rapid cadence. A subset of these images are then shifted and stacked in order to produce nearly-diffraction-limited images. GSC 03546-01452 was observed over 3 nights separated by 2 years using Fast- Cam (Oscoz et al. 2008) on the 1.5 m TCS telescope at Observatorio del Teide in Spain. The LI frames were ac- quired on 9 Oct., 2010; 25 Aug., 2011 and 11 Sep., 2012 in the I band and spanning ∼ 21 × 21 arcsec2 on sky. The 11 Oct., 2010 observing run had 100,000 frames with 50 msecs per frame, the 25 Aug., 2011 run had 60,000 frames with 60 msecs per frame, and the 11 Sept., 2012 run had 75,000 frames at 60 msecs per frame. The data were processed using a custom IDL software pipeline. After identifying corrupted frames due to cos- mic rays, electronic glitches, etc., the remaining frames are bias corrected and flat fielded. Lucky image selection is applied using a variety of selection thresholds based on the brightest pixel (BP) method. This method involves selecting frames for the final combined LI image based on the BP in that frame. The selected BP must be below a specified brightness 6 De Lee et al. Oct 09th '10 Aug 25th '11 Sep 11th '12 50 msec 60 msec 60 msec 1% LI 1% LI 1% LI 5% LI 5% LI 5% LI 15% LI 15% LI 15% LI 50% LI 50% LI 50% LI 4" N E 70% LI 70% LI 70% LI Figure 3. GSC 03546-01452 was observed on 9 Oct., 2010, 25 Aug., 2011, and 11 Sep., 2012 with LI from the Fastcam imager. The mosaic shows the final LI combined images from each run using different thresholds. LI detected a previously known visual companion 7.7′′ away from GSC 03546-01452 (marked with an ar- row) in the 9 Oct., 2010 and 11 Sep., 2012 runs. The companion was not detected on the 25 Aug., 2011 night because it was not in the field of view of the camera. This visual companion has ∆I ≈ 7.9 magnitudes. Further analysis showed that the companion is in fact a background star. For more details, see Section 3.5. threshold to avoid selecting cosmic rays or other non- speckle features. As a further check, the BP must be consistent with the expected energy distribution from a diffraction speckle under the assumption of a diffraction- limited PSF. The BP's of each frame are then sorted from brightest to faintest, and the best X% are then shifted and added to generate a final combined image. Several values of X, the LI threshold, are tried until the best combined image is generated. The combined images us- ing different LI thresholds are shown for each of the three observing runs in Figure 3. The 11 Sept., 2012 run has a better quality image than the other two runs because of an instrument upgraded to a higher sensitivity CCD. The best LI threshold was determined to be 50%, which is what we used for the rest of the analysis. This results in a total exposure time for each of the final combined images from the three runs in order of date 2500, 1800, and 2250 s respectively. In Figure 4 we show the 3-σ detectability curves com- puted out to 8′′ from GSC 03546-01452. We follow the same procedure as in Femen´ıa et al. (2011) to compute these curves: at a given angular distance ρ from GSC ] n g a m [ d n a b - I . y t i l i b a t c e t e D σ 3 ] O • M [ t i m i l s s a m r e p p u n o n a p m o C i Sept11 '12, Texp=60 msec Aug25 '11, Texp=60 msec Oct09 '10, Texp=50 msec 1 Distance From GSC 03546 [arcsec] 2 3 4 5 6 Sept11 '12, Texp=60 msec Aug25 '11, Texp=60 msec Oct09 '10, Texp=50 msec 3 4 5 6 7 8 9 10 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1 Distance From GSC 03546 [arcsec] 2 3 4 5 6 7 7 Figure 4. Top: 3-σ detectability curves for LI from the FastCam imager. There was significant variation in the curves over the three runs. The best run can detect sources down to ∆I = 8.5 beyond 2′′. No source was detected within 7′′ of GSC 03546-01452. Bot- tom: Conversion of 3-σ detectability curves for the three observing nights into mass sensitivities using empirical Mass-Luminosity re- lationships in the literature. See Section 3.5. 03546-01452 we identify all possible sets of small boxes of a size larger but comparable to the FWHM of the PSF (i.e. 5×5 pixel boxes). Only regions of the image showing structures easily recognizable as spikes due to diffraction of the telescope spider and/or artifacts on the read-out of the detector are ignored. For each of the valid boxes on the arc at angular distance ρ the standard deviation of the image pixels within the 5-pixel boxes is computed. The value assigned to the 3-σ detectability curve at ρ is 3 times the mean value from the standard deviations of all the eligible boxes at ρ. This procedure, using each of the LI % thresholding values provides a detectability curve, while the envelope of all the family of curves for a given night yields the best possible detectability curve to be extracted from the whole data set. Although there are no companions detected within 7′′, we do find a possible companion slightly outside this re- gion at a separation of 7.7′′. This companion is signifi- cantly dimmer than GSC 03546-01452 with a ∆I ≈ 7.9 MARVELS-6b 7 magnitudes. This object was only detected in the 11 Oct., 2010 and 11 Sept., 2012 campaigns. The visual companion was not detected in the 25 Aug., 2012 cam- paign due to the orientation of the camera on the sky. The possible companion can be seen in Figure 3. A dis- cussion of the likelihood that this object is actually a bound companion to GSC 03546-01452 can be found in Section 3.5. 3. RESULTS 3.1. Host Star Characterization 3.1.1. Spectroscopic Analysis The ARCES moderate resolution spectrum (Section 2.3) was analyzed by two independent analysis pipelines. We refer to these pipeline results as the "IAC" (Insti- tuto de Astrof´ısica de Canarias) and "BPG" (Brazilian Participation Group) results. Briefly, both methods are based on the principals of Fe I and Fe II excitation and ionization equilibria. Both techniques employ the 2002 version of the MOOG code (Sneden 1973), but use differ- ent line lists, model atmosphere grids, equivalent width measurements, and convergence criteria. For a com- plete description of the analysis process, please refer to Wisniewski et al. (2012). Our IAC analysis used 204 Fe I lines and 20 Fe II lines, and resulted in the following parameters: Teff = 5502±100, log g = 4.21±0.58, and [Fe/H] = +0.31±0.16. The BPG analysis used 61 Fe I lines and 6 Fe II lines re- sulting in these parameters: Teff = 5652 ± 75, log g = 4.46 ± 0.16, and [Fe/H] = +0.44 ± 0.10. Since the val- ues for both methods are consistent to within 1-σ they were combined using a weighted average as described in Wisniewski et al. (2012). This calculation resulted in the final spectroscopic parameter values of Teff = 5598 ± 63, log g = 4.44 ± 0.17, and [Fe/H] = +0.40 ± 0.09. These values are recorded in Table 3, and are used for the re- maining analysis. From these values of Teff, log g, and [Fe/H] and their er- rors, and using the Torres et al. (2010) relations, includ- ing intrinsic scatter and the uncertainties in the polyno- mial coefficients and covariances, we find that the mean and RMS of the mass and radius of the host are M∗ = 1.11±0.11 M⊙ and R∗ = 1.06±0.23 R⊙ The median and 68 percent confidence intervals are M∗ = 1.11+0.11 −0.10 M⊙ and R∗ = 1.03+0.26 −0.19 R⊙. These spectroscopic values also allows us to transform the detection curves from the left panel of Figure 4 into upper limits on the mass of a possible stellar compan- ion undetected at the 3-σ level. We start by taking the spectroscopic Teff from Table 3, and using the ta- bles from Mamajek (2010), we derive absolute V and I- band magnitudes. We can apply this technique because GSC 03546-01452 has been identified as a main-sequence star by the log g measurement. The MI and the 3-σ de- tectability curves allows the construction of the MI vs. angular distance, ρ from the central star. This curve provides an absolute upper I-band limit to any com- panion, which would also have to be a main-sequence star. From the MV vs. ρ curves we can use empiri- cal Mass Luminosity Relationships (MLR) (Henry et al. 1999; Delfosse et al. 2000; Henry 2004; Xia et al. 2008; Xia & Fu 2010) to derive the upper mass limit. These constraints are plotted for all three observation nights in ) 2 - m c 1 - s g r e ( λ F λ g o l -10 -11 -12 -13 0.1 1.0 λ (µm) 10.0 Figure 5. The spectral energy diagram (SED) for GSC 03546- 01452. The black line shows the best fit NextGen model compared to the observed flux in the photometric passbands from Table 3 (red crosses). The horizontal bars are the approximate passband width for each filter, while the vertical bar is the error in the flux. The blue circles are the expected flux values from the model. The best fit model is fixed to the spectroscopically determined Teff , log g, and [Fe/H] and allowed AV to float. The optical and infrared fit well, but there is a significant excess of flux in the GALEX NUV passband. the right panel of Figure 4. 3.1.2. Photometric Analysis We corroborated the spectroscopic results with a series of photometric tests, including determining the RPM-J statistic from Collier Cameron et al. (2007). For GSC 03546-01452, the RPM-J value was 2.31 and its (J-H) 2MASS color was 0.387, results consistent with GSC 03546-01452 being a dwarf. We also constructed a spectral energy diagram (SED) of GSC 03546-01452 using the GALEX NUV filter (Morrissey et al. 2007); Johnson B and SDSS u′,g′,r′,i′ observations from HAO; 2MASS J, H, and K (Cutri et al. 2003); and WISE1, WISE2, WISE3 (Cutri & et al. 2012a). A summary of the photometric values are pre- sented in Table 3. We used the NextGen model atmo- sphere grid (Hauschildt et al. 1999) to construct theo- retical SEDs. These models were fixed by the spec- troscopic values for TEf f , log g, and [Fe/H] described in Section 3.1.1; the reddening was constrained by the Schlegel et al. (1998) dust maps in the direction of (l,b) = (79.◦339029, 16.◦853379) to be less than AV = 0.213. The resulting fit to the photometry is shown in Figure 5. The best fitting model has a reduced χ2 of 3.91, a negligible reddening of AV = 0.1 ± 0.1, and a distance of 219 ± 21 pc. There appears to be a slight UV excess, which may be the result of modest stellar activity. 3.1.3. Evolutionary Status and and Galactic Population Using the spectroscopic host star parameters and the mass and radius derived from the Torres et al. (2010) relations, GSC 03546-01452 can be placed onto an evolutionary track. We use the Yonsei-Yale ("Y2") model tracks from Demarque et al. (2004) (and refer- ences therein), and select the track corresponding to 1.11 M⊙ with a [Fe/H] = +0.40. Figure 6 shows this track with stellar ages marked as blue dots on the track. The 8 De Lee et al. Parameter GSC1.1 Name GSC2.3 Name KIC Name 2MASS Name α (J2000) [deg] δ (J2000) [deg] µα [mas yr−1] µδ [mas yr−1] Teff [K] log g [cgs] [Fe/H] [dex] vmic [km s−1] vrot sin i [km s−1] vsystemic [km s−1] AV [mag] Distance [pc] M∗ [M⊙] R∗ [R⊙] galNUV B V RC IC SDSS u' SDSS g' SDSS r' SDSS i' 2MASS J 2MASS H 2MASS KS WISE1 WISE2 WISE3 Host Star Properties GSC 03546-01452 Table 3 Value 1-σ Uncertainty Source 3546-01452 N2EA000110 11022130 J19113252+4830436 287.885735 48.512117 -28.6 -8.7 5598 4.44 +0.40 0.38 <9.0 -13.799 0.1 219 1.11 1.06 18.320 12.593 11.799 11.335 10.967 13.483 12.156 11.560 11.400 10.46 10.07 10.010 12.650 13.344 15.181 · · · · · · · · · · · · 0.26′′ 0.25′′ 2.3 1.7 63 0.17 0.09 0.17 · · · 0.034 0.1 21 0.11 0.23 0.100 0.020 0.010 0.012 0.012 0.020 0.012 0.010 0.010 0.030 0.030 0.03 0.023 0.020 0.032 Lasker et al. (2008) Lasker et al. (2008) Brown et al. (2011) Cutri et al. (2003) Lasker et al. (2008) Lasker et al. (2008) Zacharias et al. (2012) Zacharias et al. (2012) This work This work This work This work This work This work This work This work This work This work Morrissey et al. (2007)a HAO (This work)a Derived from HAO Derived from HAO Derived from HAO HAO (This work) a HAO (This work)a HAO (This work)a HAO (This work)a Cutri et al. (2003)a Cutri et al. (2003)a Cutri et al. (2003)a Cutri & et al. (2012a)a Cutri & et al. (2012a)a Cutri & et al. (2012a)a a These passbands were used to generate SED in Figure 5. dashed lines show tracks for the same metallicity, but for ±0.11M⊙, which is the 1-σ uncertainty on our mass es- timate. The gray area shows the region that this family of tracks occupies. The red point shows GSC 03546- 01452 with 1-σ error bars. From the error bars in log g and Teff in Figure 6 we can constrain the age of GSC 03546-01452 to being less than approximately 6 Gyr old. We can determine the Galactic population member- ship of GSC 03546-01452 by using the absolute systemic RV=−13.799 ± 0.128 km s−1 (where the 0.128 km s−1 is a conservative estimate of the systematic error from the comparison with RV standard stars), the proper mo- tions from UCAC4 (Zacharias et al. 2012) (µ = −28.6 ± 2.3, −8.7 ± 1.7 mas yr−1), and the distance from Table 3 (219 ± 21 pc). We find (U, V, W ) = (24.0 ± 2.5, −10.1 ± 1.3, 25.6 ± 3.1) km s−1. This velocity is consistent with membership in the thin disk according to the criteria of Bensby et al. (2003). The space motion velocities were determined using a modification of the IDL rou- tine GAL UVW, which is ultimately based on the method Johnson & Soderblom (1987). We adopt the correction for the Sun's motion with respect to the Local Stan- dard of Rest from Co¸skunoglu et al. (2011), and choose a right-handed coordinate system such that positive U is toward the Galactic Center. 3.0 3.5 4.0 4.5 g g o l M = 1.11 ± 0.11 MO • [Fe/H] = 0.40 9.0 8.0 6.2 4.9 1.0 6000 5500 5000 4500 Teff [K] Figure 6. The black line represents the best fit Yonsei-Yale evolutionary track for a 1.1 M⊙ star with a [Fe/H] of +0.40 (Demarque et al. 2004). The gray area denotes the 1-σ deviation from that track. Ages along the track are denoted by blue dots for 1.0, 4.7, 6.4, 8.0, and 9.0 Gyr. The red cross denotes the loca- tion of GSC 03546-01452 on this diagram with the spectroscopic uncertainties. The orbital parameters of MARVELS-6b were derived using the radial velocity data from both the MARVELS and the SARG spectrographs. The orbital fit was cre- ated using the EXOFAST27 code (Eastman et al. 2013), 3.2. Companion Orbital Analysis 27 http://astroutils.astronomy.ohio-state.edu/exofast/ MARVELS-6b 9 Table 4 Properties for MARVELS-6b Parameter Units TC . . . . . . . . . BJDTDB - 2450000 . . . . . . . . . . . . . P . . . . . . . . . . Period (days) . . . . . . . . . . . . . . . . . . . e . . . . . . . . . . . eccentricity . . . . . . . . . . . . . . . . . . . . . ω . . . . . . . . . . . argument of periastron (radians) K . . . . . . . . . . RV semi-amplitude (m/s). . . . . . . γT N G . . . . . . TNG Systemic velocity (m/s). . . dv/dt . . . . . . . RV slope (m/s/day) . . . . . . . . . . . . γAP O . . . . . . APO Systemic velocity (m/s) . . . Value 5023.377+0.10 −0.099 47.8929+0.0063 −0.0062 0.1442+0.0078 −0.0073 1.998+0.064 −0.061 1644+12 −13 724 ± 15 0.180 ± 0.053 1058+17 −16 e cos(ω) . . . . e sin(ω) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . −0.0597+0.0071 −0.0070 0.1312+0.0093 −0.0092 5025.81+0.46 −0.44 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TP . . . . . . . . . BJDTDB - 2450000 . . . . . . . . . . . . . Mb sin i . . . . . Minimum Mass (MJ up) . . . . . . . . 31.7+2.0 −2.0 which allowed us to determination of the usual Keplerian orbital elements shown in Table 4. The first step in the process was to perform indepen- dent fits to both the MARVELS and SARG data points. Both data sets had error bars that were too large based on the χ2 of their fits, so we re-scaled them to force the probability that the χ2 is greater than or equal to the measured χ2, i.e., P(χ2), = 0.5. The MARVELS error bars were scaled by a factor of 0.579, and the SARG er- ror bars were scaled by a factor of 0.222. The error bars in Table 1 include these scaling factors. Once the error bars were scaled, the two data sets were combined. The final combined radial velocity curve was then run through EXOFAST which is a Markov Chain Monte Carlo (MCMC) based IDL software suite. The basic algorithms of this software follow the work of Ford (2006). The results of the combined fit are listed in Table 4, and the best-fit model is shown in Figures 7 and 8. The radial velocity measurements from both MAR- VELS and SARG are relative radial velocities. As part of the fitting process, EXOFAST fits and subtracts arbi- trary zero points for each data set simultaneously. These zero points are recorded in Table 4. These offsets are based on arbitrary zero-points, and should not be con- fused with the true systemic velocity of the host system −13.799 ± 0.128 km s−1. The systemic velocity of GSC 03546-01452 was derived by taking the orbital solution from EXOFAST for the combined relative radial velocity data set and applying it to the SARG/TNG absolute velocity data set (Table 2.). This calculation was performed by using the MPFIT Levenberg-Marquardt non-linear least squares fitting al- gorithm in IDL (Markwardt 2009). A Keplerian model was fit to the SARG/TNG radial velocity points, holding P, e, ω, K, and TP fixed while allowing γ to float; the results are presented in Table 3. 3.3. Companion Mass 3.3.1. Mass Functions of Secondary 1500 750 0 -750 -1500 300 0 -300 ) s / m ( V R ) s / m ( - C O 1000 1200 1400 1600 1800 BJD TDB - 2454000 Figure 7. All the radial velocity points from both the MARVELS Survey (blue squares) and TNG/SARG follow-up spectroscopy (red circles). The final best fit for a Mb sin i = 31.7 ± 2.0 MJ up com- panion to GSC 03546-01452 (Table 4) is shown with the associ- ated O-C diagram. The observation data points start at BJD = 2454956.916 (5 May, 2009). There is a slight residual linear slope (0.180 m s−1 day−1) after removing the companion orbit. Using the MCMC chain from the joint RV fit, we can derive Mb of the companion, (Mb sin i)3 (M∗ + Mb)2 Mb ≡ = K 3(1 − e2)3/2 P 2πG = (2.136 ± 0.049) × 10−5M⊙, (1) where the uncertainty is essentially dominated by the uncertainty in K, such that σMb /Mb ∼ 3(σK/K) = 3 × 0.8% ∼ 2.4%. 3.3.2. Minimum mass and mass ratio In order to determine the mass or mass ratio of the secondary, we must estimate the mass of the primary, as well as the inclination of the secondary. To estimate the mass and radius of the primary, we proceed as follows. For each link in the MCMC chain from the joint RV fit, we draw a value of Teff , log g, and [Fe/H] for the primary from Gaussian distributions, with means and dispersions given in Table 3. We then use the Torres et al. (2010) relations to estimate the mass M∗ and radius R∗ of the primary, including the intrinsic scatter in these relations. The minimum mass (i.e., Mb if sin i = 1) and minimum mass ratio of the secondary are: Mb,min = 31.7 ± 2.0 MJup = 0.0303 ± 0.0020 M⊙ (2) qmin = 0.02733 ± 0.00092 The uncertainties in these estimates are almost entirely explained by the uncertainties in the mass of the primary: σMb /Mb ∼ (2/3)(σM∗ /M∗) = (2/3) × 9.7% ∼ 6.5%, close to the uncertainty in Mb,min above (6.4%), and σq/q ∼ (1/3)(σM∗ /M∗) ∼ (1/3) × 9.7% ∼ 3.2%, close to the uncertainty in q (3.4%). The mass function (Mb) is the only property of a single-lined radial velocity companion that we can de- rive that is independent of the properties of the primary. 3.3.3. A posteriori distributions of the true mass The a posteriori distribution of the true mass of the companions given our measurements depends on our 10 De Lee et al. TNG/SARG MARVELS 1500 750 0 -750 -1500 300 0 -300 0.0 0.2 0.4 0.6 0.8 1.0 Phase ) s / m ( V R ) s / m ( - C O Figure 8. The phased final RV curve including both the MAR- VELS Survey (blue squares) and the TNG/SARG observations (red circles). The final best fit for a Mb sin i = 31.7 ± 2.0 MJ up companion to GSC 03546-01452 (Table 4), is shown with the asso- ciated O-C diagram. prior distribution for the mass of the companion, which is roughly equivalent our prior on the mass ratio. If we assume a prior that is uniform in the logarithm of the true mass of the companion, then the posterior distribution of cos i will be uniform. More generally, for other priors, cos i is not uniformly distributed. We adopt priors of the form: dN dq ∝ qα (3) where q is the mass ratio between the companion and the primary, and α = −1 for the uniform logarithmic prior discussed above. To include this prior, for each link the MCMC chain we draw a value of cos i from a uniform dis- tribution, but then weight the resulting values of the de- rived parameters for that link (i.e., the companion mass m) by qα+1. For α > 0, the a posteriori distribution does not cov- erage. However, we can rule out equal mass ratio com- panions by the lack of a second set of spectral lines. We therefore assume that the mass ratio cannot be greater than unity; i.e., we give zero weight to inclinations such that q > 1. In doing so, we implicitly assume that the companion is not a stellar remnant. Figures 9 shows the resulting cumulative a posteri- ori distributions of the true mass of the companion, for α = −1 (uniform logarithmic prior on q), α = 0 (uniform linear prior), and α = 1. For α = −1 and α = 0, the inferred median masses are ∼ 37 MJup and ∼ 45 MJup, respectively. For these priors, we conclude that the com- panion has a mass below the hydrogen burning limit and thus is a bona fide brown dwarf at 90% and 72% con- fidence, respectively. For α = 1, the median mass is 172 MJup or ∼ 0.16 M⊙, firmly within the stellar regime. Indeed, for this prior, there is only a ∼ 32% probability that the companion is a brown dwarf. However, this con- clusion is sensitive to the precise form for our constraint that q ≤ 1. With a more careful analysis it may be possible to rule out a wider range of companion masses based on the lack of evidence for a second set of spectral lines. Furthermore, there is little evidence that the mass function of companions to solar type stars is rising as steeply as α = 1 for minimum masses around 30 MJup (Grether & Lineweaver 2006). We conclude that the companion to GSC 03546- 01452 is most likely a brown dwarf. However, we cannot Figure 9. The cumulative a posteriori probability of the true mass of the secondary, for three different priors on the companion mass ratio, dN/dq ∝ qα, with α = −1 (solid), α = 0 (dotted) and α = 1 (dashed). Also indicated are the mass of the primary with uncertainty and the hydrogen burning limit both denoted by shading. For α = −1 and α = 0, the inferred median masses are ∼ 37 MJ up and ∼ 45 MJ up, respectively. For these priors, we conclude that the companion has a mass below the hydrogen burning limit and thus is a bona fide brown dwarf at 90% and 72% confidence, respectively. For α = 1, the median mass is 172 MJ up or ∼ 0.16 M⊙, firmly within the stellar regime. However, it is unlikely that α > 0 in this mass regime. definitively exclude that it is, in fact, a low-mass stellar companion seen at low inclination. 3.4. Time-Series Photometric Analysis 3.4.1. Summary of Datasets The SuperWASP photometric dataset for GSC 03546- 01452 consists of 8823 points spanning just over four years from HJD′ = 3128 to 4690 (where HJD′ ≡ HJD − 2450000). The full, detrended SuperWASP dataset has a relatively high weighted RMS of 1.8% and exhibits evi- dence for systematics. The distribution of residuals from the weighted mean is asymmetric and non-Gaussian, showing long tails containing a much larger number of > 3-σ outliers than would be expected for a normally- distributed population. We therefore cleaned the Super- WASP data as follows. We first add a trial systematic fractional error in quadrature to the photon noise un- certainties σsys. We then compute the error-weighted mean flux, and determine and reject the largest error- normalized outlier from the mean flux. We then recom- pute the mean flux, and scale the uncertainties by a con- stant factor r to force χ2/dof = 1. We iterate this pro- cedure until no more > 4-σ outliers remain. We then repeat the entire procedure to determine the value of σsys that results in a distribution of normalized residu- als that has the smallest RMS from the Gaussian expec- tation. Although 4-σ is a slightly larger deviation than we would expect based on the final number of points, we adopt this conservative threshold in order to avoid removing a potential transit signal. We adopt r = 0.60 and σsys = 0.011. The cleaned light curve has 8635 data points with an RMS of 1.5% and χ2/dof = 1 (by design). The Allengheny photometric dataset consists of 3990 MARVELS-6b 11 spanning roughly a year and a half Figure 10. Cleaned broad band relative photometry of GSC 03546-01452 from SuperWASP (blue) and Allegheny (green) as a function of date. points from HJD′=5341 to 5884. The weighted RMS of the raw light curve is 0.63%. There is mild evidence for systematics errors in this dataset, and thus we repeat the identical procedure as with the SuperWASP data. We choose a more conservative 6-σ cut, given the better precision of the Allengheny observations, to prevent us from remov- ing real transit-like variability. We adopt r = 1.04 and σsys = 0.0028, with a final RMS of 0.4% from 3983 data points. While continuous variability will not be removed by our procedure, variability in the form of a small number of highly-discrepant point will be masked. This possibil- ity is particularly relevant for transit signatures. Given the typical uncertainties of ∼ 0.4% for that data set, we do not expect our cleaning procedure to remove out- liers that differ by less than ∼ 2.4% from the mean, roughly corresponding to the typical depth for a tran- sit of a ∼ 1.5 RJup companion given the primary radius of ∼ 1.03 R⊙. Finally, we combine all the relative photometry af- ter normalizing each individual data set by its mean weighted flux. Figure 10 shows the combined data set, which consists of 12618 data points spanning roughly 7.6 years from HJD′ = 3128 to 5884. The weighted RMS is 0.66%. The resulting light curve is constant to within the uncertainties over the entire time span. Because the individual data sets sample disjoint times and are not relative common set of reference stars, our procedure of normalizing each data set by its mean flux will remove variations on the longest timescales. 3.4.2. Search for Variability We ran a Lomb-Scargle periodogram on the full pho- tometric dataset, testing periods between 1 − 103 days. The resulting periodogram shown in Figure 11 does not exhibit any strong peaks. The maximum fitted ampli- tude over this range is ∼ 0.1%. The inset presents the periodogram of the combined data set for periods within ∼ 10 days of the period of the companion. The individ- ual periodograms for the SuperWASP and Allengheny Figure 11. Lomb-Scargle periodogram of the combined photo- metric data. While a large number of peaks are visible, we do not regard any of them as significant. In particular, while there is a local peak at the period of the secondary (vertical red dotted line), this peak arises from the SuperWASP data alone, and is not con- firmed by the Allegheny data. This result is demonstrated in the inset, which shows the combined periodogram near the period of the companion (grey solid), the periodogram of the SuperWASP data alone (blue dotted), and Allegheny alone (blue long dashed). data are also shown. While the SuperWASP data dis- play a local peak at the period of the companion, this feature is not corroborated by the Allengheny data. Al- though this diffrence could in principle arise from real variability that is present in the SuperWASP data, but not in the Allegheny data (i.e., evolving spots), given the higher precision of the Allengheny data, we believe it is more likely that the peak in the SuperWASP peri- odogram is due to low-level systematics in the form of residual correlations on a range of time scales. Figure 12 shows the combined light curve, folded at the median period and time of conjuction of the companion, (P = 47.8929 ± 0.0063 and TC = 5023.377), and binned 0.05 in phase. The weighted RMS of the binned light curve is ∼ 0.041%. Although the variations are larger than expected from a constant light curve based on the uncertainties (χ2/dof ∼ 2.6), we again suspect that these are due to systematic errors in the relative photometry. In particular, the folded, binned Allegheny light curve shows a somewhat lower RMS of ∼ 0.016%, and is more consistent with a constant flux with a χ2/dof = 0.4. We conclude that there is no strong evidence for vari- ability of GSC 03546-01452 on any time scale we probe. We can robustly constrain the amplitude of any persis- tent periodic variability to be less than . 0.1% over timescales of . 3 years, and we can constrain the ampli- tude of persistent photometric variability at the period of the companion to . 0.02%. 3.4.3. Excluding Transits of the Secondary For a uniform distribution in cos i (corresponding to a prior that is uniform in the logarithm of the mass), the secondary transit probability is ∼ 1.8%. The duration for a central transit is ∼ R∗P/(πa) ∼ 6.6 hours, and the depth for Jupiter-radius body is δ ∼ (Rb/R∗)2 = 12 De Lee et al. determine if the companion transits (b ≤ 1), and, if so, we determine the properties of the light curve using the routines of Mandel & Agol (2002) for a given compan- ion radius Rb. We assume quadratic limb darkening, adopting coefficients appropriate for the R band from Claret & Hauschildt (2003), assuming solar metallicity and the central values of Teff and log g listed in Table 3. Finally, we fit the predicted transit light curve to the combined photometric data, and compute the ∆χ2 be- tween the constant flux fit and the transit model. We perform this calculation for every step in the Markov Chain, and finally repeat this procedure for a range of companion radii. We find a best-fit transit model has ∆χ2 = −9.0 rela- tive to a constant fit. We do not consider this result to be significant, as we find a larger improvement ∆χ2 = −28.1 when we consider signals with the same ephemeris and shape as transits, but corresponding to positive devia- tions (i.e., "anti-transits", see Burke et al. (2006)). Al- though formally statistically significant, both the transit and anti-transit signals are likely caused by residual sys- tematics in the cleaned photometric data set, and thus we conclude there is no evidence for a transit signal in the photometric data. This procedure can be used to determine the confi- dence with which we can rule out transits of a compan- ion with a given radius. Specifically, the confidence with which we can exclude a companion with a given Rb is sim- ply the fraction of the steps in the Markov Chain where the companion transits, which produces a transit with a ∆χ2 relative to the constant fit that is greater than some value of ∆χ2. We consider three different thresholds of ∆χ2 = 25, 49, and 100. The resulting cumulative prob- ability distributions are shown in Figure 13. Given the values of ∆χ2 found for anti-transits, we conservatively consider thresholds of ∆χ2 & 49 to be robust. For this value, we can exclude companions with Rb & 0.7 RJup with 50% confidence (i.e., for 50% of the trials), and com- panions with Rb & 1.2 RJ with 90% confidence. How- ever, there is a long tail toward large companion radii, arising from the conflation of the long period of the com- panion, uncertainties in the emphemeris, and imperfect phase coverage. Therefore, we are unable to exclude transits at > 95% confidence even for Rb & 1.5. For the minimum companion mass of ∼ 32 MJup, Baraffe et al. (2003) predict radii of ∼ 0.9 − 1 RJup for ages between 0.5 and 5 Gyr. Therefore, we cannot definitively exclude the possibility that the companion transits. As discussed above, our 6-σ clipping procedure would in principle remove & 2.4% transit signatures from the Allegheny dataset arising from a large Rb & 1.5 RJup companion. We therefore repeated the transit search on the Allegheny dataset with 10-σ clipping, but again find no significant transits. 3.5. LI Tertiary Companion As discussed in Section 2.5.2 a possible tertiary com- panion was detected at 7.7′′ from GSC 03546-01452 (see Figure 3). The possible tertiary companion has ∆I ≈ 7.9 magnitudes, and at the measured distance to the host star in Table 3 places the companion at a projected dis- tance of 1686 AU from the host star. This possible tertiary companion can be found in both the 2MASS (Cutri et al. 2003) and Kepler Input Catalog Relative photometry folded at the period of sec- Figure 12. ondary, and binned 0.05 in phase. Phase zero corresponds to the expected time of conjunction (and so of transits for the approprate inclinations). Black points are the combined data, blue are Super- WASP, and green are Allegheny. 1.0%(Rb/RJup)2, where Rb is the radius of the compan- ion, and we have adopted the median value for the pri- mary radius of R∗ = 1.03 R⊙. Thus, the expected S/N of a transit in the combined photometric dataset assuming uniform phase coverage is S/N ∼ N 1/2(cid:18) R∗ πa(cid:19)1/2 δ σ ∼ 13(cid:18) Rb RJup(cid:19)2 (4) where N = 12618 is the number of data points, and σ ∼ 0.66% is the RMS of the combined light curve. Thus we would expect to be able to robustly detect or exclude transits from companions with radii as as small as ∼ 0.9 RJup at 10-σ if the transit phase is well covered by the photometric data. However, given the long period of the companion, uniform phase coverage is not necessarily a good approximation. Therefore, in order to account for the actual sam- pling of the photometric data, we perform a quantitative search for transit signals using a modified Monte Carlo method. This method is similar to that described in Fleming et al. (2012) we briefly review the details here. We start with the distributions of the relevant radial ve- locity fit parameters: the period P , semiamplitude K, eccentricity e, and time of conjuction TC. These are ob- tained from the MCMC chain determined by fitting of the joint radial velocity data set as described in 3.2. For each link in this MCMC chain, we also draw values for Teff, log g, and [Fe/H] for the primary from Gaussian distributions, with means and dispersions given in Ta- ble 3. We then use the Torres et al. (2010) relations to estimate the mass M∗ and radius R∗ of the primary, in- cluding the intrinsic scatter in these relations. We then draw a value of cos i from a uniform distribution, assum- ing a prior on the companion mass Mb that is uniform in log Mb. The values of P , K, e, M∗, R∗ and i are then used to determine the secondary mass Mb, semi- major axis a, and impact parameter of the secondary orbit b ≡ a cos i/R∗. From the impact parameter, we MARVELS-6b 13 Difference in Position of GSC 03546-01452 versus Possible Tertiary Table 5 Year Observed ∆α cos(δ) (arcsec) · · · Error ∆δ Error (arcsec) (arcsec) (arcsec) Source · · · 2012 2010 2000 7.232 7.270 6.86 0.049 0.023 0.11 -3.026 -3.163 -3.39 0.043 0.018 0.12 Lucky Lucky 2MASS on the distribution of mass ratios. Working with the minimum mass for MARVELS-6b of 31.7 ± 2.0 MJup, we can compare it to current distribution of masses in the literature. Grether & Lineweaver (2006) investigated the mass function of both stellar and planetary companions and found that they both decreased by a substantial frac- tion as they approached the minimum of the BD desert. In the case of the stellar mass function, it dropped by two orders of magnitude from the 1 M⊙ to the middle of the BD desert. On the planetary mass side, it rises one order of magnitude as one progress away from the BD desert and towards lower mass planets. MARVELS-6b is interesting because its minimum mass, 31.7 ± 2.0 MJup, lies at the very bottom of the two distributions. Grether & Lineweaver (2006) define their mass functions as number of companions per unit interval in log mass, and find the minimum of the distributions to be at 31+25 −18MJup. In order to better understand this minimum,it is necessary to increase the number of ob- jects (or place tighter constraints by non-detections) in this region of the mass function, and MARVELS-6b takes us one step closer to that goal. MARVELS-6b stands out in respect to other brown dwarfs in two significant ways. First, it has a relatively low eccentricity, 0.1442, given its moderately long pe- riod of ≈ 47 days. To place this point in perspective, MARVELS-6b is plotted against a catalog of BDs from Ma & Ge (2013b) in Figure 14. Two aspects of this plot are particularly interesting: First, MARVELS-6b is in an underdense region in period space. In the inset of Figure 14 there is a clear rise in the envelope of eccen- tricity as one moves to larger periods until one reaches roughly 200 days. MARVELS-6b is either on or below the bottom edge of this envelope. There is the possi- bility that this effect is due to observational bias; for low signal to noise ratios (SNR), one often measures a non-zero eccentricity even when the true eccentricity is zero. For a given radial velocity precision and compan- ion mass, longer periods will have lower SNR (since the semi-amplitude is smaller), and so the lack of low eccen- tricity companions at long periods could be due to this effect. In any case, there are so few BD in this plot, that it is not clear whether MARVELS-6b's low eccentricity is significant, so further comment on this will have to wait until more data is available. This BD companion is also an outlier in metallicity space. In Figure 15 MARVELS-6b is again plotted with the BD from Ma & Ge (2013b). GSC 03546-01452's metallicity is significantly super solar, and MARVELS- 6b exists in an empty region of eccentricity-metallicity space. As in the previous figure, the plot is merely sug- gestive of interesting astrophysics, but there are not a sufficient number of BDs to know whether MARVELS- Figure 13. Probability that transits of a companion are excluded at levels of ∆χ2 = 25, 49, 100 based on the analysis of the combined SuperWASP and Allegheny photometric data sets, as a function of the radius of the companion. The gray region represents the 95 percent confidence level. Only the ∆χ2 = 25 model crosses into the gray region, thus we cannot rule out a possible transiting companion even with an unphysically large radii Rb & 1.5 RJ up at the 95% level. (KIC; Brown et al. 2011). The KIC ids of the primary and possible tertiary are 11022130 and 11022139 respec- tively. The first blow to the hypothesis that this is a bound system is found in the KIC itself. The KIC gives the following stellar parameter estimates for the possi- ble tertiary Teff = 5919, log g = 4.361, and AV = 0.515. Comparing these values to the spectroscopic values de- termined for GSC 03546-01452 in Table 3 it is clear that these two objects cannot be bound, and have a ∆I of ≈ 7.9 magnitudes; because two G dwarfs should have ap- proximately the same magnitude. However, the tertiary is faint (J = 16.272), and the KIC spectroscopic determi- nations are photometric in nature, so we considered the possibility that the KIC parameters were incorrect. GSC 03546-01452 has a measured proper motion from the UCAC4 catalog (Zacharias et al. 2012) of µα = -28.6 ± 2.3 mas yr−1 and µδ = -8.7 ± 1.7 mas yr−1 (see Ta- ble 3). This motion is significant given the plate scale of the LI images ≈ 42.56 mas pixel−1, which, given our 2 year baseline for observations, should be sufficient to test whether this is a bound system. Table 5 gives the differ- ence in position between the GSC 03546-01452 and the possible tertiary in RAcos(DEC) and DEC for both LI observations. The two points are distinct to the 2-σ level. We also have observations of both sources in 2MASS, although the astrometric position measurement is sig- nificantly worse, the 12 year baseline between 2MASS and the 2012 observation compensates for the lower ac- curacy. These points are also distinct to the 2-σ level. Given both the astrometric and photometric data, the hypothesis that this is a bound system is very unlikely. 4. DISCUSSION Based on the derived spectroscopic parameters for the host star and the results discussed in Section 3.3.3, it is highly likely that MARVELS-6b is a BD companion to GSC 03546-01452 assuming a logarithmic or linear prior 14 De Lee et al. RV BD Transit BD MARVELS-6b y t i c i r t n e c c E 1 0.8 0.6 0.4 0.2 0 1 0.8 0.6 0.4 0.2 0 1 10 100 1000 10000 0 20 40 60 Orbital Period (days) 80 100 Figure 14. Orbital period and eccentricity of MARVELS-6b com- pared to a subset of literature brown dwarfs found through tran- sit or radial velocity that occupy the period range of 0 to 100 days. The literature brown dwarfs are from the catalog of Ma & Ge (2013b). MARVELS-6b occupies an empty part of orbital period / eccentricty plane. Inset: The full brown dwarf catalog extending out to much larger periods. RV BD Transit BD MARVELS-6b 1 0.8 0.6 0.4 0.2 0 y t i c i r t n e c c E -1 -0.5 0 Host Star [Fe/H] 0.5 Figure 15. The host metallicity and orbital eccentricity of MARVELS-6b compared to the literature brown dwarfs from the catalog of Ma & Ge (2013b) MARVELS-6b has a relatively low eccentricity and its host star has a quite high [Fe/H]. 6b is a significant deviation. MARVELS-6b is helping to fill in the BD parameter space, and with more data its location may become more astrophysically significant. 5. CONCLUSION In this paper, we report the discovery of a BD compan- ion to GSC 03546-01452 with a minimum mass of 31.7 ± 2.0 MJup. This BD, designated MARVELS-6b, has a moderately long period of 47.8929+0.0063 −0.0062 days with a low eccentricty of 0.1442+0.0078 −0.0073. We have analyzed moderate resolution spectroscopy of the host star and have deter- mined the following properties: Teff = 5598 ± 63,log g = 4.44 ± 0.17, and [Fe/H] = +0.40 ± 0.09. From these mea- surements we find that GSC 03546-01452 has a mass and radius of M∗ = 1.11 ± 0.11 M⊙ and R∗ = 1.06 ± 0.23 R⊙. This result combined with photometry, indicates the host star is a G dwarf at 219 ± 21 pc from the Sun with an age less than approximately 6 Gyr based on an evolutionary track analysis. Due to its moderately long period, MARVELS-6b has a transit probability for a uniform distribution of cos i of only 1.8%. Although we have roughly 13,000 pho- tometric data points, we cannot conclusively rule out a transit. In the Keck AO imaging no visual companions were found. However, in the LI a previously known com- panion at 7.7′′ from the host star was detected. This visual companion appears in both the 2MASS and KIC catalogs, and was shown not be a physical companion based upon photometry and astrometry, which is unex- pected given that many of the previous BD found by the MARVELS survey did have tertiary companions. Finally, we found that minimum mass of MARVELS- 6b exists at the minimum of the mass functions of close (orbital period < 5 yr) stellar and planetary compan- ions to stars, making this a rare object even compared to other BDs. It also exists in an underdense region in both period/eccentricity and metallicity/eccentricity space. This ultimately furthers the goal of this series of papers, to help fill out the low-mass companion phase space, which will ultimately help us understand these intriguing objects. The SuperWASP and Allegheny lightcurve data for GSC 03546-01452 , along with the APO spectroscopic data will be made available through the Vizier/CDS28 catalog service. The SuperWASP and Allegheny lightcurve data will also be available with the online edi- tion of this article. Funding for the MARVELS multi-object Doppler in- strument was provided by the W.M. Keck Founda- tion and NSF with grant AST-0705139. The MAR- VELS survey was partially funded by the SDSS-III con- sortium, NSF grant AST-0705139, NASA with grant NNX07AP14G and the University of Florida. Funding for SDSS-III has been provided by the Alfred P. Sloan Foundation, the Participating Institutions, the National Science Foundation, and the U.S. Department of Energy Office of Science. The SDSS-III web site is http://www.sdss3.org/. SDSS-III is managed by the Astrophysical Research Consortium for the Participating Institutions of the SDSS-III Collaboration including the University of Ari- zona, the Brazilian Participation Group, Brookhaven Na- tional Laboratory, University of Cambridge, Carnegie Mellon University, University of Florida, the French Participation Group, the German Participation Group, Harvard University, the Instituto de Astrofisica de Ca- narias, the Michigan State/Notre Dame/JINA Participa- tion Group, Johns Hopkins University, Lawrence Berke- ley National Laboratory, Max Planck Institute for As- trophysics, Max Planck Institute for Extraterrestrial Physics, New Mexico State University, New York Uni- versity, Ohio State University, Pennsylvania State Uni- versity, University of Portsmouth, Princeton University, the Spanish Participation Group, University of Tokyo, University of Utah, Vanderbilt University, University of Virginia, University of Washington, and Yale University. Based on observations made with the Italian Telesco- pio Nazionale Galileo (TNG) operated on the island of La Palma by the Fundacin Galileo Galilei of the INAF (Istituto Nazionale di Astrofisica) at the Spanish Ob- servatoriodel Roque de los Muchachos of the Instituto de Astrofisica de Canarias. The Center for Exoplan- ets and Habitable Worlds is supported by the Penn- sylvania State University, the Eberly College of Sci- ence, and the Pennsylvania Space Grant Consortium. 28 http://vizier.u-strasbg.fr/viz-bin/VizieR MARVELS-6b 15 Keivan Stassun, Leslie Hebb, and Joshua Pepper ac- knowledge funding support from the Vanderbilt Initiative in Data-Intensive Astrophysics (VIDA) from Vanderbilt University, and from NSF Career award AST-0349075. BSG and JDE acknowledge support from NSF CA- REER grant AST-1056524 EA thanks NSF for CAREER grant 0645416. GFPM acknowledges financial support from CNPq grant n◦ 476909/2006-6 and FAPERJ grant n◦ APQ1/26/170.687/2004. LG acknowledges financial support provided by the PAPDRJ CAPES/FAPERJ Fel- lowship Facilities: Sloan () REFERENCES Adelman-McCarthy, J. K., et al. 2008, ApJS, 175, 297 Baluev, R. V. 2008, MNRAS, 385, 1279 Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, A&A, 402, 701 Bensby, T., Feltzing, S., & Lundstrom, I. 2003, A&A, 410, 527 Borucki, W. J., Koch, D., Basri, G., et al. 2010, Science, 327, 977 Brown, T. M., Latham, D. W., Everett, M. E., & Esquerdo, G. A. 2011, AJ, 142, 112 Burke, C. J., Gaudi, B. S., DePoy, D. L., & Pogge, R. W. 2006, AJ, 132, 210 Chabrier, G., Baraffe, I., Allard, F., & Hauschildt, P. 2000, ApJ, 542, 464 Claret, A., & Hauschildt, P. H. 2003, A&A, 412, 241 Co¸skunoglu, B., Ak, S., Bilir, S., et al. 2011, MNRAS, 412, 1237 Collier Cameron, A., Wilson, D. M., West, R. G., et al. 2007, MNRAS, 380, 1230 Cumming, A. 2004, MNRAS, 354, 1165 Cutri, R.M. et al. 2003, 2MASS All-Sky Catalog Cutri, R. M., & et al. 2012a, VizieR Online Data Catalog, 2311, 0 Cutri, R. M., Wright, E. L., Conrow, T., et al. 2012b, Explanatory Supplement to the WISE All-Sky Data Release Products, 1 Delfosse et al, 2000, a, 364, 217 Demarque, P., Woo, J., Kim, Y., & Yi, S.K. 2004, ApJS, 155, 667 Duquennoy, A., & Mayor, M. 1991, A&A, 248, 485 Eastman, J., Gaudi, B. S., & Agol, E. 2013, PASP, 125, 83 Eisenstein, D. J., Weinberg, D. H., Agol, E., et al. 2011, AJ, 142, 72 Erskine, D. J. 2003, PASP, 115, 255 Femen´ıa, B., et al. 2011, MNRAS, 413, 1524 Fleming, S.W. et al. 2010, ApJ, 718, 1186 Fleming, S. W., Ge, J., Barnes, R., et al. 2012, AJ, 144, 72 Ford, E. B. 2006, ApJ, 642, 505 Fukugita, M., Ichikawa, T., Gunn, J. E., et al. 1996, AJ, 111, 1748 Ge, J. 2002, ApJ, 571, L165 Ge, J., Erskine, D. J., & Rushford, M. 2002, PASP, 114, 1016 Ge, J., van Eyken, J., Mahadevan, S., et al. 2006, ApJ, 648, 683 Ge, J., Mahadevan, S., Lee, B., et al. 2008, Extreme Solar Systems, 398, 449 Ge, J. et al. 2009, Proc. SPIE, 7440, 74400L Ge, J., & Eisenstein, D. 2009, astro2010: The Astronomy and Astrophysics Decadal Survey, 2010, 86 Gratton, R.G. et al. 2001, Exp. Astron., 12, 107 Griffin, R., & Griffin, R. 1973, MNRAS, 162, 255 Gunn, J.E. et al. 2006, AJ, 131, 2332 Haghighipour, N., Vogt, S. S., Butler, R. P., et al. 2010, ApJ, 715, 271 Halbwachs, J. L., Mayor, M., Udry, S., & Arenou, F. 2003, A&A, 397, 159 Hauschildt, P. H., Allard, F., & Baron, E. 1999, ApJ, 512, 377 Henry, T. J. 2004, ASP Conf. Series, 318, 159 Henry et al, 1999, ApJ, 512, 864 Howard, A. W., Johnson, J. A., Marcy, G. W., et al. 2010, ApJ, 721, 1467 Johnson, D. R. H., & Soderblom, D. R. 1987, AJ, 93, 864 Landolt, A. U., & Uomoto, A. K. 2007, AJ, 133, 768 Landolt, A. U. 2009, AJ, 137, 4186 Lasker, B. M., Lattanzi, M. G., McLean, B. J., et al. 2008, AJ, 136, 735 Lee, B.L. 2011, ApJ, 728, 32 Lomb, N. R. 1976, Ap&SS, 39, 447 Gratton, R. G., et al. 2001, Grether, D., & Lineweaver, C. H. 2006, ApJ, 640, 1051 Hog, E. et al. 1998, A&A, 335, 65 Ma, B., Ge, J., Barnes, R., et al. 2013, AJ, 145, 20 Ma, B., & Ge, J. 2013, arXiv:1303.6442 Mamajek E., 2010, "A Modern Dwarf Stellar Effective Temperature Scale Based on 17842 Teff and MK Spectral Type Pairs", http://www.pas.rochester.edu/∼emamajek/Teff SpT table.txt Mandel, K., & Agol, E. 2002, ApJ, 580, L171 Marcy, G. W., & Butler, R. P. 2000, PASP, 112, 137 Marcy, G. W. & Butler, R. P. 1992, PASP, 104, 270 Markwardt, C. B. 2009, Astronomical Data Analysis Software and Systems XVIII, 411, 251 Mayor, M., Pepe, F., Queloz, D., et al. 2003, The Messenger, 114, 20 Morrissey, P. et al. 2007, ApJS, 173, 682 Oscoz, A., et al. 2008, Proc. SPIE, 7014,701447-1 -- 701447-12 Patel, S. G., Vogt, S. S., Marcy, G. W., et al. 2007, ApJ, 665, 744 Pollacco, D. L., Skillen, I., Collier Cameron, A., et al. 2006, PASP, 118, 1407 Reid, I. N., & Metchev, S. A. 2008, Exoplanets, 115 Sahlmann, J., S´egransan, D., Queloz, D., et al. 2011, A&A, 525, A95 Scargle, J. D. 1982, ApJ, 263, 835 Schlegel, D. J., Finkbeiner, D. P., & Davis, M. 1998, ApJ, 500, 525 Skrutskie, M. F., Cutri, R. M., Stiening, R., et al. 2006, AJ, 131, 1163 Smith, J. A., Tucker, D. L., Kent, S., et al. 2002, AJ, 123, 2121 Sneden, C. 1973, PhD Thesis, Univ. of Texas-Austin Spagna, A., Lattanzi, M. G., McLean, B., et al. 2006, Mem. Soc. Astron. Italiana, 77, 1166 Spiegel, D. S., Burrows, A., & Milsom, J. A. 2011, ApJ, 727, 57 Torres, G., Andersen, J., & Gim´enez, A. 2010, A&A Rev., 18, 67 Udry, S., Mayor, M., Naef, D., et al. 2000, A&A, 356, 590 Udry, S., & Santos, N. C. 2007, ARA&A, 45, 397 van Eyken, J. C., Ge, J., & Mahadevan, S. 2010, ApJS, 189, 156 Wang, S. et al. 2003, Proc. SPIE, 4841, 1145 Wang, J., Ge, J., Jiang, P., & Zhao, B. 2011, ApJ, 738, 132 Wang, J., Ge, J., Wan, X., Lee, B., & De Lee, N. 2012, PASP, 124, 598 Wang, J., Ge, J., Wan, X., De Lee, N., & Lee, B. 2012, PASP, 124, 1159 Wisniewski, J. P., Ge, J., Crepp, J. R., et al. 2012, AJ, 143, 107 Wizinowich, P., Acton, D. S., Shelton, C., et al. 2000, PASP, 112, 315 Wright, E.L. et al. 2010, AJ, 140, 1868 Xia et al., 2008, Astrophys Space Sci, 314, 51 Xia & Fu, 2010, Chinese Astronomy and Astrophysics, 34, 277 Zacharias, N., Finch, C. T., Girard, T. M., et al. 2012, VizieR Online Data Catalog, 1322, 0
1508.00288	1	1508	2015-08-02T22:20:49	SOFIA Infrared Spectrophotometry of Comet C/2012 K1 (Pan-STARRS)	[ "astro-ph.EP" ]	We present pre-perihelion infrared 8 to 31 micron spectrophotometric and imaging observations of comet C/2012 K1 (Pan-STARRS), a dynamically new Oort Cloud comet, conducted with NASA's Stratospheric Observatory for Infrared Astronomy (SOFIA) facility (+FORCAST) in 2014 June. As a "new" comet (first inner solar system passage), the coma grain population may be extremely pristine, unencumbered by a rime and insufficiently irradiated by the Sun to carbonize its surface organics. The comet exhibited a weak 10 micron silicate feature ~1.18 +/- 0.03 above the underlying best-fit 215.32 +/- 0.95 K continuum blackbody. Thermal modeling of the observed spectral energy distribution indicates that the coma grains are fractally solid with a porosity factor D = 3 and the peak in the grain size distribution, a_peak = 0.6 micron, large. The sub-micron coma grains are dominated by amorphous carbon, with a silicate-to-carbon ratio of 0.80 (+0.25) (- 0.20). The silicate crystalline mass fraction is 0.20 (+0.30) (-0.10), similar to with other dynamically new comets exhibiting weak 10 micron silicate features. The bolometric dust albedo of the coma dust is 0.14 +/- 0.01 at a phase angle of 34.76 degrees, and the average dust production rate, corrected to zero phase, at the epoch of our observations was Afrho ~ 5340~cm.	astro-ph.EP	astro-ph	ApJ Accepted 2015.July.22 SOFIA Infrared Spectrophotometry of Comet C/2012 K1 (Pan-STARRS) Charles E. Woodward1, Michael S. P. Kelley2, David E. Harker3, Erin L. Ryan2, Diane H. Wooden4, Michael L. Sitko5, Ray W. Russell6, William T. Reach7, Imke de Pater8, Ludmilla Kolokolova2, Robert D. Gehrz9 ABSTRACT We present pre-perihelion infrared 8 to 31 µm spectrophotometric and imaging observations of comet C/2012 K1 (Pan-STARRS), a dynamically new Oort Cloud comet, conducted with NASA's Stratospheric Observatory for Infrared Astronomy (SOFIA) facility (+FORCAST) in 2014 June. As a "new" comet (first inner solar system passage), the coma grain population may be extremely pristine, unencumbered by a rime and insufficiently irradiated by the Sun to carbonize its surface organics. The comet exhibited a weak 10 µm silicate feature ≃ 1.18 ± 0.03 above the underlying best-fit 215.32 ± 0.95 K continuum blackbody. Thermal modeling of the observed spectral energy distribution indicates that the coma grains are fractally solid with a porosity factor D = 3 and the peak in the grain size distribution, apeak = 0.6 µm, 1Minnesota Institute for Astrophysics, University of Minnesota, 116 Church St, SE Minneapolis, MN 55455, USA; [email protected] 2University of Maryland, Department of Astronomy, College Park, MD 20742-2421, USA 3University of California, San Diego, Center for Astrophysics & Space Sciences, 9500 Gilman Dr. Dept 0424, La Jolla, CA 92093-0424, USA 4NASA Ames Research Center, MS 245-3, Moffett Field, CA 94035-0001, USA 5Department of Physics, University of Cincinnati, Cincinnati, OH 45221, USA 6The Aerospace Corporation, Los Angeles, CA 90009, USA 7USRA-SOFIA Science Center, NASA Ames Research Center, Moffett Field, CA 94035, USA 8Astronomy Department, 601 Campbell Hall, University of California, Berkeley, CA 94720, USA 9Minnesota Institute for Astrophysics, University of Minnesota, 116 Church St, SE Minneapolis, MN 55455, USA -- 2 -- large. The sub-micron coma grains are dominated by amorphous carbon, with a silicate-to-carbon ratio of 0.80+0.25 −0.20. The silicate crystalline mass fraction is 0.20+0.30 −0.10, similar to with other dynamically new comets exhibiting weak 10 µm silicate features. The bolometric dust albedo of the coma dust is 0.14 ± 0.01 at a phase angle of 34.76◦, and the average dust production rate, corrected to zero phase, at the epoch of our observations was Af ρ ≃ 5340 cm. Subject headings: STARRS) -- ISM: dust comets: general -- comets: individual (C/2012 K1 Pan- 1. INTRODUCTION Solar System formation was an engine that simultaneously preserves and transforms interstellar medium (ISM) ices, organics, and dust grains into cometesimals, planetesimals and, ultimately, planets. Observing and modeling the properties of small, primitive bodies in the solar system whose origins lie beyond the water frost line (> 5 AU) provides critical insight into the formation of Solar System solids and establishes observation constraints for planetary system formation invoking migration -- the 'Grand Tack' epoch (Walsh et al. 2011), followed by the 'Nice Model' events (Levison 2009; Gomes et al. 2005). The characteristics of comet dust can provide evidence to validate the new, emerging picture of small body populations -- including comet families -- resulting from planetary migration in the early Solar System. Inside cometary nuclei, the bulk of the dust likely has been preserved since formation of the nucleus. Comet grains (and ices) also trace the pre-accretion history of comet materials extant in the outer disk. Comet dust composition can be studied via Stardust samples, se- lected collections of Interplanetary Dust Particles (IDPs), and in situ analysis in comet flyby and/or rendezvous missions. Dust species that are best explained as products of aqueous alteration (e.g., magnitite, cubanite, possibly pentlandite) are rare (Stodolna et al. 2012; Berger et al. 2011; Zolensky et al. 2008) and corresponding altered silicates (e.g., phyllosili- cates, smectite) are missing suggesting that aqueous alteration in cometary nuclei is limited, is not well represented in the Stardust samples, or that these minerals have exogenous ori- gins (Brownlee 2014). Thus, the bulk of comet grain properties including dust size, porosity, and composition relate to grain formation, radial mixing, and particle agglomeration in the proto-solar disk (for an extensive review see Brownlee 2014). However, opportunities to study actual samples of cometary dust are rare, motivating the need for telescopic remote sensing observations of dust whenever apparitions are accessible from terrestrial observatories. -- 3 -- In this paper we report our pre-perihelion (TP = 2014 Aug 27.65 UT) infrared 8 to 31 micron spectrophotometric observations of comet C/2012 K1 (Pan-STARRS), a dynamically new (see Oort 1950, for a definition based on orbital elements) Oort Cloud comet -- (1/aorg) = 42.9 × 10−6 AU−1 (Williams 2015) -- conducted with NASA's Stratospheric Observatory for Infrared Astronomy (SOFIA) facility during a series of four flights over the period from 2014 June 04 to 13 UT. Contemporaneous optical imaging observations are also presented. 2. OBSERVATIONS 2.1. Ground-based Optical Imaging Comet C/2012 K1 (Pan-STARRS) was observed on 2012 June 01.22 UT and again on June 04.24 UT with the 2.3-m Bok Telescope at the Kitt Peak National Observatory. The comet was at heliocentric distance (rh) of 1.74 AU and 1.71 AU, a geocentric distance (∆) of 1.66 AU and 1.69 AU, a phase angle of 34.62◦ and 34.76◦, for each date respectively. The images were obtained with the 90Prime camera (Williams et al. 2004), a prime focus imager built for the Bok Telescope. At the time of observation, the 90Prime camera utilized a thinned back-illuminated CCD detector with 4064 × 4064 pixels with a pixel size of 15.0 µm. At prime focus the camera pixel scale is 0.45′′ which yields a field of view of 30.5 × 30.5 square-arcmin. The instrument was equipped with Cousins/Bessel system broadband V and R filters. Multiple exposures (23 images in R band and 9 images in V band of 30 seconds each) were obtained of the nucleus and coma of the comet with the telescope tracking at the non-sidereal rate corresponding to the predicted motion of the comet provided by JPL Horizons1 in an airmass range of 1.40 to 1.74. All images were corrected for overscan, bias and flat-fielding with standard IRAF2 rou- tines. The data was photometrically calibrated using eight field stars of various spectral types with known V and R magnitudes selected from the Naval Observatory Merged As- trometric Dataset (NOMAD) catalog (Zacharias et al. 2004) on the same CCD amplifier as the comet. The standard deviation of the photometric V and R zero points derived from the average of the field stars is of order 1% and no color corrections for spectral type were applied. The average nightly seeing was ∼ 2.2′′ in both bands. A single 30 sec exposure in 1http://ssd.jpl.nasa.gov/horizons.cgi 2IRAF is distributed by the National Optical Astronomy Observatory, which is operated by the Asso- ciation of Universities for Research in Astronomy (AURA) under cooperative agreement with the National Science Foundation. the R band obtained on 2014 June 04.24 UT is shown in Fig. 1. -- 4 -- 2.2. SOFIA Mid-infrared (mid-IR) spectrophotometric observations of comet C/2012 K1 (Pan- STARRS) were obtained using the Faint Object InfraRed CAmera for the SOFIA Telescope (FORCAST; Herter et al. 2012) mounted at the Nasmyth focus of the 2.5-m telescope of the SOFIA Observatory (Young et al. 2012; Gehrz et al. 2009). The data were acquired over a series of four flights, originating from Palmdale, CA at altitudes of ≃ 11.89 km in 2014 June, that were conducted as part of our SOFIA Cycle 2 programs to observe comets (P.I. Woodward, AOR IDs 01 001 and 02 0002). Details of all SOFIA observations and the orbital parameters of comet C/2012 K1 (Pan-STARRS) at those epochs are summarized in Table 1. FORCAST is a dual-channel mid-IR imager and grism spectrometer operating from 5 to 40 µm. Light is fed to two 256 × 256 pixel blocked-impurity-band (BIB) arrays, each with a plate scale of 0.768′′ per pixel and a distortion corrected field of view of 3.2′ × 3.4′. The Short Wavelength Camera (SWC) covers the spectral region from 5 to 25 µm, while the Long Wavelength Camera (LWC) operates at wavelengths from 25 to 40 µm. Imaging data can be acquired in either dual channel mode (with some loss of throughput due to the dichroic) or single channel mode. Imaging observations of C/2012 K1 (Pan-STARRS) in three filters were conducted on the first flight series, prior to three flights dedicated to spectroscopy. Spectroscopic obser- vations of the comet used two grisms, one in the SWC (G111) and one in the LWC (G227), and the instrument was configured using a long-slit (4.7′′ × 191′′) which yields a spectral resolution R = λ/∆λ ∼ 140-300. The comet was imaged in the SWC using the F197 fil- ter to position the target in the slit. Both imaging and spectroscopic data were obtained using a 2-point chop/nod in the Nod-Match-Chop (C2N) mode with 45′′ chop and 90′′ nod amplitudes at angles of 30◦/210◦ in the equatorial reference frame. All FORCAST raw image data products were processed using the FORCAST REDUX Data Pipeline, v1.0.1beta (cf., Clarke, Vacca, & Shuping 2014), which employed the reduc- tion packages FORCAST FSPEXTOOL, v1.1.0, and FORCAST DRIP, v1.1.0. Processing of the raw spectroscopic data was performed using the same packages, with the exception of FORCAST DRIP, which utilized v1.0.4. Details of the FORCAST REDUX Data Pipeline can be found in the Guest Investigator Handbook for FORCAST Data Products, Rev. B3 -- 5 -- 2.3. SOFIA Imagery and Photometry Aperture photometry of the SOFIA image data of comet C/2012 K1 (Pan-STARRS) was performed on the Level 3 pipeline coadded (*.COA) data products using the Aperture Photometry Tool (APT v2.4.7; Laher et al. 2012). At all FORCAST filter wavelengths, the comet exhibited extended emission beyond the PSF of point sources observed with FOR- CAST under optimal telescope jitter performance.4 The photometry was therefore conducted using a circular aperture centroided on the photocenter of the comet nucleus. We used an aperture of radius 13 pixels, corresponding to 9.984′′, with a background aperture annulus of inner radius 30 pixels (23.58′′) and outer radius of 60 pixels (47.16′′). This aperture, which is ≃ 3× the nominal point-source FWHM, encompassed the majority of the emis- sion of the comet and coma. Sky-annulus median subtraction (ATP Model B as described in Laher et al. 2012) was used in the computation of the source intensity. The systematic source intensity uncertainty was computed using a depth of coverage value equivalent to the number of coadded image frames. The dominant source of overall uncertainty in the im- age photometry were image gradients due to imperfect atmospheric background subtraction. The calibration factors (and associated uncertainties) applied to the resultant aperture sums were included in the Level 3 data distribution and were derived from the weighted average of 3 calibrator observations of β And (2 each) and α Boo (1 each). The resultant SOFIA photometry is presented in Table 2. Due to turbulence, telescope jitter, and differing chop-nod patterns, i.e., the chopping difference between beams and the nodding of the entire telescope field-of-view (for a for a discussion and illustration of this standard infrared observing technique with SOFIA -- see Temi et al. 2014; Young et al. 2012) executed in flight, the multi-filter imagery data could not be used to generate color temperature maps due to the unstable PSF. During flights primarily devoted to obtaining grism data (§2.4), images of the comet where obtained through a single filter at 19.7 micron. Figure 2 shows the 19.7 µm surface brightness distribution of comet C/2012 K1 (Pan-STARRS) observed on 2014 June 13.17 UT. The nucleus is unresolved and azimuthally symmetric with a radial profile FWHM of ∼ 1.01′′ 3https://www.sofia.usra.edu/Science/DataProducts/ FORCAST GI Handbook RevA1.pdf 4see http://www.sofia.usra.edu/Science/ ObserversHandbook/FORCAST.html §5.1.2 -- 6 -- and the coma is extended and diffuse. Low surface brightness emission extends in a vector direction commensurate with that expected for a dust tail. 2.4. SOFIA Spectra Three temporally distinct spectra of comet C/2012 K1 (Pan-STARRS) were obtained in both grism over a series of flight sequences spanning 6 days (Table 1). Many comets exhibit temporal variability in the infrared over periods of hours (e.g., Wooden et al. 2004) to days at relatively similar heliocentric distances due to coma jets related to nucleus activity and/or nucleus rotation period (e.g., Keller et al. 2007; Gehrz et al. 1995) that produces observable changes in the observed SEDs. Inter-comparison of each SED over this period showed no substantial changes in overall continuum flux densities nor spectral features to within the uncertainty per spectral resolution element. In addition, the 19.7 µm aperture photometry suggests also that the level of coma emission did not markedly change (cf., Table 2). Apparently, comet C/2012 K1 (Pan-STARRS) was fairly quiescent in its infrared behavior during this epoch given our signal-to-noise ratio and aperture size (12,300 km radius). Thus, the three independent spectra were summed together in pipeline processing to produce an average spectral energy distribution (SED). A 3-point unweighted rectangular smoothing function was applied to this average SED to increase the point-to-point signal-to- noise ratio of the data product used in our thermal model spectral decomposition analysis. The calibrated data products do exhibit a few artifacts near the edges of the 17 -- 27 µm spectral order where a few data points deviate upwards (near 17 µm) or downwards (near 27 µm) from the apparent spectral trend. The spectra of comet C/2012 K1 (Pan-STARRS) are presented in Fig. 3. 3. RESULTS Taxonomically comet C/2012 K1 (Pan-STARRS) is a member of the dynamical comet family denoted as nearly isotropic comets (NICs), also commonly referred to as Oort Cloud comets (cf., Dones et al. 2004). The interior composition of the ecliptic comets (ECs) and the NICs likely are preserved during their residence in the Scattered Disk and the Oort Cloud, but their surfaces are subject to various processing effects. Modeling the coma dust properties provides insight into the origin and evolution of dynamic comet families. -- 7 -- 3.1. Thermal Modeling of the Coma SED Thermal modeling of the observed thermal infrared SED of comets obtained using re- mote sensing techniques enables derivation of coma dust grain properties. In particular, SOFIA (+FORCAST) provides spectroscopic coverage with the G111 grism to the region 9 -- 12 µm which contains features from amorphous and crystalline silicates (e.g., 11.2 µm) and organic species (e.g., PAHs). The G227 grism spans 17.6 -- 27.7 µm, encompassing dis- crete resonances from crystalline silicates as well as spectral signatures from carbonates and phyllosilicates, putatively argued to be extant in comets (Lisse et al. 2006). The SED slope at long thermal ( > ∼ 15 µm) wavelengths provides constraints on the abundance of the larger grain population in the coma. Observations in these spectral regimes are key to ascertaining the origins of silicates within the solar protoplanetary disk, and placing early solar disk evo- lution within the context of other circumstellar disks observed today through comparison to model and laboratory data (cf., Lindsay et al. 2013; Koike et al. 2010). Modeling the mid-IR SED of C/2012 K1 (Pan-STARRS) yields estimates of the coma grain properties. We constrain the grain parameters by chi-squared fitting thermal emis- sion models to the observed spectrum. The grain parameters included in the modeling are size distributions (n(a) da), porosity, the crystalline mass fraction (i.e., the fraction of the coma silicate grains that are crystalline), and relative material abundances. The dust tem- perature is calculated assuming thermal equilibrium of the grains; wherein the composition (mineralogy), size, and heliocentric distance determine the temperature of the grains. Comet grains are dominated in composition by a handful of silicate-type materials (Hanner & Zolensky 2010; Wooden 2008): Mg-Fe olivine- and pyroxene-types in amorphous (glassy) forms and their crystalline Mg-end-members forsterite (Mg2SiO4) and enstatite (MgSiO3). Cometary aggregates also contain organics (Sandford et al. 2006) or amorphous- carbon-like materials (Matrajt et al. 2008; Formenkova 1999) that may be the glue that holds the amorphous and crystalline materials together (Flynn et al. 2013; Ciesla & Sandford 2012). Our model (Harker et al. 2002, and references therein) uses five materials: amorphous olivine and amorphous pyroxene with broad 10, 18, and 20 µm emission features, amorphous carbon with featureless emission, and crystalline olivine (Mg-rich) and orthopryoxene with narrow peaks. Broad and narrow resonances near 10 and 20 µm are modeled by warm chon- dritic (50% Fe; 50% Mg) amorphous silicates (i.e., glasses) and strong 11.25, 19.5, and 24 µm narrow features from cooler Mg-rich crystalline silicate materials. The amorphous carbon component in our dust model is representative of several key dust species -- e.g., elemental carbon dust (Fomenkova et al. 1994), an organic component with C=C bonds, identified by XANES spectra near 285 eV that can include amorphous carbon (Flynn et al. 2013, 2003; Wirick et al. 2009), and possibly other carbonaceous grains; -- 8 -- however, overall model results do not depend on this degeneracy. We do not specifi- cally include Fe-Ni sulfides (such as pyrrhotite or troilite) in our models nor carbonates or phyllosilicate-rich materials. The latter materials have not been detected in track anal- ysis of Stardust samples (Nakamura-Messenger et al. 2011; Wooden 2008; Zolensky et al. 2008) nor are they unequivocally evident in remote sensing data (Bursentova et al. 2012; Woodward et al. 2007). Phyllosilicates, specifically smectites including montmorillonite, chlorite, and serpentine, have 18-23 µm resonances that worsen spectral fitting of comet C/1995 O1 (Hale-Bopp) (Wooden et al. 1999). Hybrid IDPs may contain up to 10% smectite (Nakamura-Messenger et al. 2011). Smectitie is spectrally distinguishable from amorphous anhydrous olivine-type and amorphous pyroxene materials (Nakamura-Messenger et al. 2011; Wooden et al. 1999), yet it is not required for spectral decomposition. While FeS-type grains are present in IDPs (Bradley & Dai 2000), meteoritics samples, and comets grains, such as Wild 2 (Heck et al. 2012; Zolensky et al. 2008; Velbel & Harvey 2007), our SOFIA spectra (Fig. 3) do not exhibit the broad 23 µm spectral features often associated with fine-grained FeS (Bursentova et al. 2012; Min et al. 2005; Hony et al. 2002; Keller et al. 2002). Larger FeS particles would be spectrally indistinguishable from larger amorphous carbon particles at mid- to far-IR wavelengths, yet robust optical constants spanning visible through the far-IR are lacking for FeS due to measurement challenges of an inherently extremely absorbing material (L. Keller, private comm.). Thus, thermal modeling of FeS gains is uncertain. Stardust samples appear to be richer in FeS and poorer in car- bonaceous matter (Joswiak et al. 2012), so there is no basis as yet to make an assumption about the relative abundance of FeS and amorphous carbon-like materials in comet comae. Hence we presume that the majority of absorbing materials in cometary dust re-radiating the observed infrared SED is dominated by olivine, pyroxene, and carbonaceous (amorphous carbon-like) materials (Brownlee 2014; Wooden 2008; Zolensky et al. 2008). This presump- tion provides a foundation for comparing compositional similarities and diversities of comet dust composition derived from thermal models. The best-fit chi-square model results are summarized in Table 3. The model fit to the observed grism spectra with the corresponding spectral decomposition of grain components is presented in Fig. 4. Mineralogically, the grains in the coma of C/2012 K1 (Pan-STARRS) are dominated by amorphous materials, especially carbon. Our models produce a Hanner (modified-power law) differential grain size distribution (HGSD)5 peaking with grains of radii apeak = 0.6 µm, indicating relative moderately larger grains are present, and the grain power-law slope N = 3.4. In a HGSD the small radii grains at the peak of the grain size 5This power law (in grain radii, a) is defined as n(a) da ≡ (1 − ao/a)M (ao/a)N ; where ao = 0.1 µm and apeak = ao(N + M )/N . -- 9 -- distribution dominate the surface area and the flux density. Grains in comets are likely fractal porous aggregates Schulz et al. (2015). The grain porosity (P versus the dust radius a), parameterized by D, is defined as P = (a/0.1 µm)(D−3) with D = 3 for solid and D = 2.5 for highly porous grains (Woodward et al. 2011). Grains in the coma of C/2012 K1 (Pan-STARRS) also are solid (the fractal porosity parameter D = 3.0). Solid grains are not unusual, 65% of the Stardust tracks are carrot-shaped from solid terminal particles (Horz et al. 2006). The sub-micron sized silicate-to-carbon ratio derived from our models is 0.80+0.25 −0.20. The uncertainty in the parameters derived from the thermal models are at the 95% confidence level. 3.2. The 10 µm Silicate Emission Feature The 10 µm silicate feature in comet C/2012 K1 (Pan-STARRS) is quite weak com- pared to comets like C/1995 O1 (Hale-Bopp) or 17P/Holmes (e.g., Wooden et al. 1999; Watanabe et al. 2009). Following (Sitko et al. 2004), at 10.5 µm we find the silicate emission (defined as [F10/F BB continuum]) is 1.18 ± 0.03 above a blackbody curve fit to the observed grism spectra continua longwards of 12.5 µm. The best-fit blackbody is Tbb = 215.32±0.95 K (using Gaussian weighted errors) and color excess, defined as Tbb(fit)/(278K r−0.5 ) is = 0.992±0.004. The normalized (Fλ/Fλ,T ) SED in the region near the silicate feature at 10 µm is presented in Fig. 5. Typically data near 8 µm are used to establish the blue-continua (λλ7.7 − 8.4 µm). Our estimate of the local 10 µm continua may yield slightly lower temperatures than an estimate that included 8 µm photometry. h The large value of apeak inferred from the thermal modeling of the observed SED of comet C/2012 K1 (Pan-STARRS) in 2014 June is commensurate with the weak 10 µm silicate feature. Smaller grains (apeak < ∼ 0.3 µm) produce higher contrast silicate features. Grains of greater porosity also produce higher contrast silicate features in the 10 µm band. Long period NICs have 'typical' HGSD slopes of 3.4 < ∼ 3.7 and silicate-to-amorphous carbon ratios ≫ 1. The grain size distribution slope of C/2012 K1 (Pan-STARRS), N = 3.4, is not atypical. The preponderance of larger sub-micron grains (apeak = 0.6 µm) in the coma of comet C/2012 K1 (Pan-STARRS) results in cooler radiating dust that contributes to the 'continuum' under the 10 µm silicate feature and to the far-infrared flux density (see Fig. 4). The sub-micron mass fraction is dominated by amorphous carbon grains. Amorphous carbon has a featureless emission spectrum that extends through the 10 µm region, so low amorphous silicate-to-carbon ratio also can weaken the silicate feature strength (Wooden 2008; Wooden et al. 2004). ∼ N < -- 10 -- 3.3. The Silicate Crystalline Mass Fraction The mass fraction of silicate sub-micron grains that are crystalline in comet comae is a keystone for models of early planet-forming processes (Bockel´ee-Morvan et al. 2002; Ciesla 2007; Hughes & Armitage 2010). This fraction is defined as f silicates cryst ≡ n X x=1 mcryst,x (mcryst,x + msilicates amorphous,x) (1) where mx is the mass of species x. Crystalline species in comet grains provide a record of the high temperature process that formed dust in the inner disk of the solar system and the large scale mixing that transported these hot nebular products to the cold comet forming zones. Crystals, their composition (e.g., Wooden 2008) and shape (Lindsay et al. 2013) trace inner-solar disk conditions (e.g., Ogliore et al. 2011) and offer a view into the earliest planet-forming processes that occurred in our early Solar System. Crystals from the inner disk were transported out to the comet-forming regime and mixed with "amorphous" silicates (cf. Cielsa 2011). The "amorphous" silicates are thought to be outer disk materials that probably were inherited from the ISM (Brownlee 2014; Watson et al. 2009; Kemper et al. 2004, 2005; Li & Draine 2001) in the infall phase of the disk. They have non-stoichiometric compositions (GEMS-like, Matsuno et al. 2012; Bradley & Dai 2004; Bradley et al. 1999, and references therein) that include the compo- sitional ranges of olivine [(Mgy , Fe(1−y))2 SiO4], with y ≈ 0.5 for amorphous olivine, and [(Mgx , Fe(1−x)) SiO3] with x ≈ 0.5 for amorphous pyroxene-type materials. Crystals are identified by narrow IR emission features (e.g., 11.2, 19, 23.5, 27.5, 33 µm) superposed on an underlying thermal continuum in remote sensing spectra. Crystalline silicates have been detected using remote sensing techniques in the dust comae of all comet classes in- cluding C/1995 O1 (Hale-Bopp) (Wooden et al. 1999; Harker et al. 2002, 2004a) the Deep Impact coma of 9P/Tempel 1 (Harker et al. 2005; Lisse et al. 2006; Harker et al. 2007), the fragmentation outburst of 17P/Holmes (Reach et al. 2010), and several other comets (Kelley & Wooden 2009; Woodward et al. 2011). Amorphous silicates are also detected in these comets as well. Crystalline silicates are found in abundance in the Stardust samples of 81P/Wild 2; however, the amorphous grains are difficult to identify (Ishii et al. 2008) and may be limited to the smallest dust grains (Brownlee 2014). Crystalline silicates are rare in the ISM; however, they account for < ∼ 2.2% of the to- tal silicate component in the direction of the Galactic Center (Kemper et al. 2004, 2005) and < ∼ 5% along other lines-of-sight (Li & Draine 2001). Solar System crystalline silicates detected in comets must be formed in the early stages of our disk's evolution (Brownlee -- 11 -- 2014). Crystalline silicates require T > ∼ 1000K to form through either gas phase condensa- tion or annealing of amorphous (glassy) silicate grains (Wooden 2008; Wooden et al. 2005; Davoisne et al. 2006; Henning 2003; Fabian et al. 2000) implying that the crystalline silicates must have been processed in the disk near the young Sun or in shocks out to a maximum distance of 3 to 5 AU (Harker & Desch 2002; Wehrstedt & Gail 2008). Post-formation, they were transported radially outward into the comet-formation zones (Charnoz & Morbidelli 2007) -- a process that is apparently ubiquitous in observations of external protoplanetary disks Olofssson et al. (2010). Glassy silicate spherules (GEMS) and crystals are seen in ag- gregates in cometary IDPs. Large 'terminal particle' crystals and sub-micron crystals (crys- tallites) are components of aggregate grains captured in Stardust samples (Brownlee et al. 2012, 2006; Zolensky et al. 2008, 2006). Thus to first order, the diversity of comet dust properties reflects the temporal and radial gradients in our Solar System's early history and similarities and differences in dust characteristics, including fcryst, may provide observational tests of of planetary migration models within the early solar system during the epoch of planet formation that resulted in a variety of small body dynamical populations. We find a that the silicate crystalline mass fraction in comet C/2012 K1 (Pan-STARRS) is fcryst = 0.20+0.30 −0.10. This range is similar to that found for comet C/20007 N3 (Lulin), fcryst = 0.48±0.06 (derived from the mass fractions presented in Table 3 of Woodward et al. 2011, and Eqn. 1 of §3.3) which also exhibited a weak 10 µm silicate emission feature. 3.4. EC and NIC Dust Characteristics As a result of giant planet migration, some comet nuclei were dynamically scattered into the Oort cloud to be exposed to the Galactic environment, whereas those bodies comprising the bulk of the ECs population have nuclei exposed and processed (at depths ranging from mm to few cm) by solar insolation, space weathering, and heliocentric activity variations (sublimation of CO, CO2; crystallization of water and other ices) which affects materials lofted into the comae. Although the interior compositions of ECs and NICs likely are pre- served, their surfaces have differing processing histories. Typically, active comets (arising from a population of NICs dynamically derived from the Oort Cloud and moving on long-period orbits) exhibit high contrast 10 µm silicate features. In contrast, short-period ECs (i.e., Jupiter-family comets) have, on average, lower 10 µm silicate features strengths (Sitko et al. 2004), and are thought to have lower activities (cf., the active area and active fraction measurements of A'Hearn et al. 1995). For decades, the low-activity of ECs has been attributed to the accumulation of a rime of insulating -- 12 -- larger grains that were launched on non-escape orbits (Jewitt 2007). Thermal models that fit observed infrared spectra of comets reveal that high contrast silicate features arise from comae having a preponderance of sub-micron grains (Harker et al. 2002; Hanner et al. 1994). Comae without these sub-micron grains have weaker silicate features. In individual comets, variations in the silicate feature strength have been seen on short time scales corresponding to the aperture-crossing times of jets or coma features (Wooden et al. 2004; Harker et al. 2005, 2007; Gicquel et al. 2012). These variations are best explained by changes in the differential grain size (n(a) da) or fluctuations in the silicate-to-carbon grain ratio. Differences between EC and NIC coma grain populations may arise from the surface layers EC nuclei being "processed" or weathered (e.g., Li et al. 2015). Processing of ECs surfaces may result from their frequent perihelion passages that decreases surface volatiles and small grains and leads to the creation of rimes and dust mantles. Evidence suggesting such processing occurs over millennia may be found in the analysis of material excavated from comet 9P/Tempel 1 by Deep Impact: the dust grains in the ejecta were smaller than those in the ambient coma (Harker et al. 2007) and the immediate comet surface contained a layer of carbon rich grains (Sugita et al. 2005) and a dust mantle comprised of compact 20 µm- sized dust aggregates (Kobayashi et al. 2013). However, this conjecture is not definitive as it unknown whether or not the impact location reflects the global surface dust properties of the nucleus. In ECs, the coma 10 µm silicate feature strengths are low (Kelley & Wooden 2009) and the dust production rates are modest. However, when EC nuclei have either fragmented (i.e., 73P/SW3, Harker et al. 2011; Sitko et al. 2011), explosively released materials from subsurface cavities (i.e., 17P/Holmes, Reach et al. 2010), or have had subsurface materials excavated from depth (i.e., the 9P/Tempel 1 Deep Impact encounter, Harker et al. 2005) the IR SEDs exhibit 10 µm relatively strong silicate feature emission ( > ∼ 1.2) arising from a population of sub-micron size silicate grain species. Whether or not the strong 10 µm silicate features arise from the release of sub-micron sized grains or the disruption of loose aggregates of fine particles (e.g., through gas-pressure disruption or impact fragmentation) is not known. Indeed the silicate feature in 9P/Tempel 1 changed from a EC-like spectra to NIC-like spectra immediately after Deep Impact event, returning to an EC-like state several tens of hours later (cf., Harker et al. 2005). NICs are canonically considered to be more pristine with higher surface volatile abun- dance (cf., Wooden 2008) -- the effects of dwell time in the Galactic environment being more benign. Also, NICs are often considered a homologous population lacking significant nucleus evolution. Inner solar system apparitions of these comets frequently result in brilliant comae, with large dust production rates and pronounced silicate feature emission at IR wavelengths. It is not entirely clear whether the highly active nuclear regions of NICs can spawn small sub-micron grains responsible for the silicate feature emission, either by heritage or by frag- -- 13 -- mentation induced within the gas acceleration zone. However, whether comet evolution, such as processing in the Galactic environment, can be ignored when comparing the Oort cloud comet dust composition (including that expressed in fcryst) is an open question. Table 4 present estimates of fcryst and select characteristics of the dust derived from thermal modeling of the mid-IR SEDs for a set of well-studied Oort cloud and "disrupted" Jupiter-family comets. The crystalline silicate fraction ranges appreciably, from ∼ 10% to ∼ 80%. The compositional similarity suggests that Oort cloud and Jupiter-family comets have common origin sites within the early solar system (an argument that parallels that derived from volatile composition studies, A'Hearn et al. 2012), but the range of fcryst values in Oort cloud comets suggest this class may be sampling a particular region that is not represented in the Jupiter-family members. This inference is intriguing; however, limited in robustness as any tentative conclusions are based on a limited sample size. Large sample sizes are required to substantiate or vitiate these trends. Comet C/2012 K1 (Pan-STARRS) and C/2007 N3 (Lulin) have modest mean values for fcryst ( <∼ 48%), low silicate-to-carbon ratios, and grain size distributions that peak at large radii, > ∼ 0.6 µm. The 10 µm silicate feature is weak and/or absent in these NICs. Perhaps these bodies represent a population of more carbon dominated bodies, similar to the dark organic KBOs, whose surfaces are devoid of small grains. Indeed the low albedo (see §3.6) of C/2012 K1 (Pan-STARRS) and the dominance of amorphous carbon grain materials maybe providing clues. 3.5. Dust Production Rates The radial profile of comet C/2012 K1 (Pan-STARRS) was plotted to assess the quality of the data for calculating a dust production rate near the epoch of our SOFIA observations. The radial profile of C/2012 K1 (Pan-STARRS) in the V band shows a deviation from the 1/ρ profile (Gehrz & Ney 1992), suggesting contamination from gas such as C2 (∆ = 0) band(s) near 5141 A. Strong C2 emission is present in spectra (McKay et al. 2014, and also A. McKay, priv. comm.) contemporaneous with our optical imagery. We therefore only calculate the dust production in R band. The R band radial profile of C/2012 K1 (Pan- STARRS) is shown in Fig. 6. To estimate the rate of dust production in comet C/2012 K1 (Pan-STARRS), we utilize the Af ρ quantity introduced by A'Hearn et al. (1984). This quantity serves as a proxy for dust production and when the cometary coma is in steady state, the value for A(Θ)f ρ is an aperture independent parameter, -- 14 -- A(Θ)f ρ = 4 r2 h ∆2 10−0.4(mcomet−m⊙) ρ (cm) (2) where A(Θ) is four times the geometric albedo at a phase angle Θ, f is the filling factor of the coma, mcomet is the measured cometary magnitude, m⊙ is the apparent solar magnitude, ρ is the linear radius of the aperture at the comet's position (cm) and rh and ∆ are the heliocentric and geocentric distances measured in AU and cm, respectively. To correct our comet measurements for phase angle effects we applied the Halley-Marcus (HM) (Marcus 2007a,b; Schleicher et al. 1998) phase angle correction.6 We adopt an interpolated value of 0.3864 (appropriate for the 2014 June 04.24 UT dataset) to normalize A(Θ)f ρ to 0◦ phase angle. Table 5 reports values of Af ρ = [(A(Θ)f ρ/HM] at a selection of distances from the comet photocenter in the R-band. In addition to Af ρ, we also compute the ǫf ρ parameter of comet C/2012 K1 (Pan- STARRS) based on our FORCAST broadband photometry (Table 2). The ǫf ρ parameter (defined by Kelley et al. 2013, Appendix A) can be considered to be the thermal emission corollary to the scattered-light-based Af ρ: ǫf ρ = ∆2 πρ Fν Bν (cm), (3) where ǫ is the effective dust emissivity, Fν is the flux density (Jy) of the comet within the aperture ρ, Bν is the Planck function (Jy/sr) evaluated at the temperature T = Tscale (278 K) r−0.5 , where the scaling factor Tscale = 0.99 based on the 215 K measured continuum temperature discussed in §3.2. Derived values of ǫ f ρ for comet C/2012 K1 (Pan- STARRS) are presented in Table 2. h 3.6. Coma Averaged Dust Albedo Dust albedo is a basic parameter characterizing the size distribution and physical prop- erties of comet dust that is, surprisingly, infrequently measured. Following the convention of Gehrz & Ney (1992) the bolometric albedo, (Abolometric ≡ (Energy scattered/Energy incident) is Abolometric ≃ f (Θ) 1 + f (Θ) , (4) 6see http://asteroid.lowell.edu/comet/dustphase.html -- 15 -- where for comet dust the incident energy is the sum of the energy scattered by the coma plus the total energy of the coma's thermal emission at an observed phase angle Θ (Sun- comet-observer angle). The term f (Θ) can be determined from fitting the observed spectral energy distribution of the comet with appropriate Planck blackbody functions in the infrared (thermal dust emission) and reflected solar spectra at optical (scattering) wavelengths f (Θ) = [λFλ]max,scattering [λFλ]max,IR (5) where the [λFλ]max is the peak of the SEDs in the respective wavelength ranges. Lab ex- periments and theoretical calculations of the scattered light from particles indicate that the total brightness, color, polarization, and polarization color depend on the optical constants, particle size distribution, structure, and porosity of the dust as well as the solar phase angle (Lindsay et al. 2013; Hadamcik et al. 2007; Kolokolova et al. 2004). The spectral shape of the IR thermal emission provides a direct link with the mineralogy and grain size. Both of these processes provide information on the size and composition of the dust. The scattered light and thermal emission are also connected to one another through the grain albedo, the ratio of the scattered light to the total incident radiation. Because light is not isotropically scattered by comet dust the measured albedo will depend not only on the composition and structure of the dust grains, but also on the phase angle (Sun-comet-observer angle) of the observations. The coma SED of comet C/2012 K1 (Pan-STARRS) was measured on 2014 June 04 UT using filter photometry at both mid-IR as well as the optical (scattered sunlight) wavelengths. These data enable computation of the coma averaged bolometric albedo (Gehrz & Ney 1992). Using the integrated flux densities in a circular aperture of radius 9.989′′, [λFλ]max,IR was derived from the SOFIA photometry by χ-square fitting a blackbody to the mid-IR data using Gaussian weighted errors, resulting in Tbb = 214.04 ± 14.94 K with a peak flux of 2.02+0.12 −0.15 × 10−16 W cm−2. The [V] band photometry is contaminated by gas emission (§3.5). However, the C2 bands fall outside the bandpass of the [R] filter and a 5800 K blackbody (the Sun) emission peaks near the [R] filter central wavelength (λc = 0.64 µm) in λFλ (W cm−2) space. Hence, [λFλ]max,scattering = 3.33 ± 0.03 × 10−17 W cm−2 derived the [R] band photometry measured in a circular aperture of radius 9.989′′ (Table 5). The dust bolometric albedo of comet C/2012 K1 (Pan-STARRS) is 0.14 ± 0.01 at phase angle of 34.76◦ (from Eqns. 4 and 5). Kolokolova et al. (2004) reviewed published visual albedos of comets and found only eight 8 comets have measured albedos (excluding comets Kohoutek and Crommelin dis- cussed in Gehrz & Ney 1992), and all were from the NIC dynamical class. Kelley & Wooden -- 16 -- (2009) found only one EC with visual albedo, 21P/Giacobini-Zinner (Pittichov´a et al. 2008). Recently, the albedos of 73P/Schwassmann-Wachmann 3, 103P/Hartley 2, and C/2009 P1 (Garradd) also have been measured (Meech et al. 2011; Sitko et al. 2011, 2013). Fig. 7 is a compilation of the the bolometric albedo data that exists on comets, including our determi- nation for comet C/2012 K1 (Pan-STARRS). There is considerable scatter for multi-epoch observations of individual comets. Such scatter arises from variations in activity of a comet at different epochs of observation. For example, comet C/1995 O1 (Hale-Bopp), whose data are most scattered, had numerous and fast changing morphological structures (jets, shells, envelopes, e.g., Harker et al. 1997; Woodward et al. 1998). All of these features were charac- terized by differing size and particle composition (e.g., Rodriguez et al. 1997; Schleicher et al. 1997). Thus, the difference in the dust albedo for the same comet indicates variations in comet activity, specifically development of jets and other morphological features. The ensemble albedos compiled by Kolokolova et al. (2004) also shows a broad distri- bution of values for each phase angle. The causes for these latter albedo ranges and the scatter in multi-epoch observations of comets are unclear, but must reside in the physical properties of the comet particles, including the grain size distribution, porosity, grain struc- ture (i.e., prolate spheroids, crystalline needles, etc.), and composition (e.g., Lindsay et al. 2013). However, observations have not yet demonstrated to what extent grain structure or grain compositions are important. To assess these latter aspects, thermal emission models and albedo observations of a additional comets are needed. 4. SUMMARY We discuss the pre-perihelion mid-infrared spectrophotometry and narrow band filter imagery obtained in 2014 June with FORCAST on the NASA SOFIA airborne platform of the dynamically new comet C/2012 K1 (Pan-STARRS) at a heliocentric distance of ≃ 1.70 AU. The spectral energy distribution of the comet at this epoch exhibits a 10 µm silicate feature, [F10/F BB continuum] = 1.18 ± 0.03 above a blackbody curve (Tbb = 215.32 ± 0.95 K) fit to the spectra continua longwards of 12.5 µm which is quite weak compared to comets such as C/1999 O1 (Hale-Bopp) or 17P/Holmes. The coma dust bolometric albedo, 0.14 ± 0.01, derived using contemporaneous optical imagery is similar to other comets at the observed phase angle (∼ 35◦), while the dust production rate (Af ρ) from scattered light observations is ≃ 5340 cm. From the observed infrared spectral energy distribution, thermal modeling analysis is used to determine the physical characteristics of the coma dust population and deduce the silicate crystalline mass fraction (0.20+0.30 −0.20). We −0.10) and silicate-to-carbon dust ratio (0.80+0.25 -- 17 -- find that grains in the coma of C/2012 K1 (Pan-STARRS) are dominated by amorphous materials, especially carbon, and the differential grain size distribution peaks at radii of 0.6 µm, the slope of the distribution N = 3.4, and the grains are solid, having a fractal porosity parameter D = 3.0. The bulk grain properties of comet C/2012 K1 (Pan-STARRS) are comparable to other Nearly Isotropic comets (NICs) with weak 10 µm silicate features and similar in respect to coma grains seen in the small-set of Ecliptic family comets (ECs) that have fragmented, explosively released subsurface materials, or have had materials excavated from depth. SOFIA observations of comet C/2012 K1 (Pan-STARRS) and other future comets en- ables characterization grain properties in the NIC and EC dynamical families. These proper- ties, including dust size, porosity, and composition, relate to grain formation, radial mixing, and particle agglomeration in the proto-solar disk and provide insight to the evolution of the early solar system. As the number of well-studied comets increases at infrared wavelengths (from whence dust properties can be characterized), the fundamental differences between comets originating from different regions and times in the solar system may be eventually discerned. 5. Acknowledgments CEW and his team acknowledge support from Universities Space Research Association (USRA)/NASA contract NAS2-97001. CEW, MSK, DEH also acknowledge support from NASA Planetary Astronomy Program grant 12-PAST12-0016, while CEW and ELR also note support from NASA Planetary Astronomy Program grant NNX13AJ11G. The authors would also like to acknowledge the support and insight of Drs. J. DeBuzier and L. A. Helton of the SOFIA Science Ctr. for their assistance with flight planning and data reduction activities. This work is supported at The Aerospace Corporation by the Independent Research and Development program. We also thank the comments and suggestions of an anonymous referee that improved the clarity of our work. Facilities: SOFIA (FORCAST), Bok (90Prime) -- 18 -- REFERENCES A'Hearn, M. F., Feaga, L. M., Keller, H. U., et al. 2012, ApJ, 758, 29 A'Hearn, M. F., Millis, R. L., Schleicher, D. G., Osip, D. J., & Birch, P.V. 1995, Icarus, 118, 223 A'Hearn, M. F., Schleicher, D. G., Feldman, P. D., Millis, R. L., & Thompson, D. T. 1984, AJ, 89, 579 Berger, E. L., Zega, T. J., Keller, L. P., & Lauretta, D. S. 2011, Geochim. Cosmochim. Acta, 75, 3501 Bockel´ee-Morvan, D., Gautier, D., Hersant, F., Hur´e, J.-M., & Robert, F. 2002, A&A, 384, 1107 Bradley, J. P., & Dai, Z. R. 2004, ApJ, 617, 650 Bradley, J. P., & Dai, Z. 2000, Meteoritics and Planetary Science Supplement, 35, 32 Bradley, J. P., Keller, L. P., Gezo, J., et al. 1999, Lunar and Planetary Science Conference, 30, 1835 Brownlee, D. 2014, Ann. Rev. Earth Planet. Sci. 42, 179 Brownlee, D., Joswiak, D., Matrajt, G. 2012, Meteoritics and Planetary Science, 47, 453 Brownlee, D., Tsou, P., Al´eon, et al. 2006, Science, 314, 1711 Brusentsova, T., Peale, R. E., Maukonen, D., et al. 2012, MNRAS, 420, 2569 Ciesla, F. J., Sandford, S. A. 2012 Science, 336, 452 Ciesla, F. J. 2011, ApJ, 740, 9 Ciesla, F. J. 2007, Science, 318, 613 Charnoz, S., & Morbidelli, A. 2007, Icarus, 188, 468 Clarke, M., Vacca, W. D., & Shuping, R. Y. 2014 i nADASS Conf, Ser., ADASS XXIV, eds. A. R. Taylor & J. M. Stil [San Francisco, CA: ASP] Davoisne, G., Djouadi, Z., Leroux, H., et al. 2006, A&A, 448, L1 -- 19 -- Dones, L., Weissman, P. R., Levison, H. F. & Duncan, M. J. 2004, in Comets II, eds. M C . Festou, H. U. Keller, and H. A. Weaver, [University of Arizona press: Tucson AZ], p.153ff Fabian, D., Jager, C., Henning, Th., et al. 2000, A&A,364, 282 Formenkova, M. N. 1999, Space Sci. Rev. 90, 109 Fomenkova, M. N., Chang, S., & Mukhin, L. M. 1994, Geochim. Cosmochim. Acta, 58, 4503 Flynn, G. J., Wirick, S., Keller, L. P., et al. 2013, Earth, Planets, and Space, 65, 1159 Flynn, G. J., Keller, L. P., Feser, M., Wirick, S., & Jacobsen, C. 2003, Geochim. Cos- mochim. Acta, 67, 4791 Gehrz, R. D., Becklin, E. E., de Pater, I., Lester, D. F., Roellig, T. L., & Woodward, C. E. 2009, AdSpR 44, 413 Gehrz, R. D., Johnson, C. H., Magnuson, S. D., & Ney, E. P. 1995, Icarus, 113, 129 Gehrz, R. D., & Ney, E. P. 1992, Icarus, 100, 162 Gicquel, A., Bockel´ee-Morvan, D., Zakharov, V. V., Kelley, M. S., Woodward, C. E., & Wooden, D. H. 2012, A&A, 542, 119 Gomes, R., et al. 2005, Nature 435, 466 Hadamcik, E., et al. 2007, Icarus, 190, 459 Hanner, M. S., Lynch, D. K., & Russell, R. W. 1994, ApJ, 425, 274 Hanner, M. H., & Zolensky, M. E. 2010, Lecture Notes in Physics, Berlin Springer Verlag, 815, 203 Harker, D. E., & Desch, S. J. 2002, ApJ, 565, L109 Harker, D. E., Woodward, C. E., McMurtry, C. W., et al. 1997, Earth Moon and Planets, 78, 259 Harker, D. E., Wooden, D. H., Woodward, C. E., & Lisse, C. M. 2002, ApJ, 580, 579 -- . 2004a, Erratum: ApJ, 615, 1081 Harker, D. E., Woodward, C. E., Wooden, D. H., & Kelley, M. S. 2004b, AAS, 205, 5612H -- 20 -- Harker, D. E., Woodward, C. E., & Wooden, D. H. 2005, Science, 310, 278 Harker, D. E., Woodward, C. E., Wooden, D. H., et al. 2007, Icarus, 190, 432 Harker, D. E., Woodward, C. E., Kelley, M. S., Sitko, M. L., Wooden, D. H., Lynch, D. K., & Russell, R. W. 2011, AJ, 141, 26 Heck, P. R., Hoppe, P., & Huth, J. 2012, Meteoritics and Planetary Science, 47, 649 Henning, T. 2003, in Lecture Notes in Physics, Vol. 609, Astromineralogy, eds. T. K. Henning, (Springer-Verlag: Berlin), pp.266 Henning, Th. 2010, ARA&A, 48, 21 Herter, T. L., Adams, J. D., & de Buizer, J. M. 2012, ApJ, 749, L18 Hill, P. M. et al. 2001, Pub. Nat. Acad. Sci. 91, No.5, 2182 Horz, F., Bastien, R., Borg, J., et al. 2006, Science, 314, 1716 Hony, S., Bouwman, J., Keller, L. P., & Waters, L. B. F. M. 2002, A&A, 393, L103 Hughes, A. L. H., & Armitage, P. J. 2010, ApJ, 719, 1633 Ishii, H. A., et al. 2008, Science, 319, 447 Jewitt, D. 2007, in Trans-Neptunian Objects and Comets, Saas-Fee Advanced Course 35, v35, p1 [Springer-Verlag: Berlin] Joswiak, D. J., Brownlee, D. E., Matrajt, G., et al. 2012, Lunar and Planetary Science Conference, 43, 2395 Laher, R. R., et al. 2012, PASP, 124, 737 Li, J.-Y., Thomas, P. C., Veverka, J., et al. 2015, Highlights of Astronomy, 16, 180 Lindsay, S. S., Wooden, D. H., Harker, D. E., et al. 2013, ApJ, 766, 54 Lisse, C. M., et al. 2006, Science, 313, 635 Kelley, M. S., & Wooden, D. H. 2009, Planet. Space Sci., 57, 1133 Kelley, M. S., Woodward, C. E., Harker, D. E., Wooden, D. H, Sitko, M. L., Russel, R. W., & Kim, D. L. 2015a, AAS, 2254, 305K Kelley, M. S., Fern´andez, Y. R., Licandro, J., et al. 2013 Icarus, 225, 475 -- 21 -- Kelley, M. S., et al. 2006, ApJ, 651, 1256 Keller, H. U., Kuppers, M., Fornasier, S., et al. 2007, Icarus, 191, 241 Keller, L. P., Hony, S., Bradley, J. P., et al. 2002, Nature, 417, 148 Kemper, F., Vriend, W. J., & Tielens, A. G. G. M. 2004, ApJ, 609, 826 -- . 2005, Erratum:, ApJ, 633, 534 Kobayashi, H., Kimura, H., & Yamamoto, S. 2013, aap, 550, 72 Koike, C., et al. 2010, ApJ, 709, 983 Kolokolova, L., et al. 2004, in Comets II, eds. M. C. Festou, H. U. Keller, & H. A. Weaver, [U. of Arizona Press, Tucson], p.577 Levison, H. F. 1996, Comet Taxonomy, in ASPC, Vol. 107, ed. T. Rettig & J. M. Hahn, (ASP: Tucson), pp.173-191 Levison, H. F., et al. 2006, Icarus, 184, 619 Levison, H. F., et al. 2009, Nature, 260, 364 Li, A., & Draine, B. T. 2001, ApJ, 550, L213 Li, A., & Greenberg, J. M. 1998, A&A, 338, 364 Marcus, X. 2007a, International Comet Qrtly April, 39 Marcus, X. 2007b, International Comet Qrtly October, 119 Matrajt, G., Ito, M., Wirick, S., et al. 2008, Meteoritics and Planetary Science, 43, 315 Matsuno, J., Tsuchiyama, A., Koike, C., et al. 2012, ApJ, 753, 141 McKay, A., Kelley, M., Cochran, A., Dello Russo, N., DiSanti, M., Lisee, C., & Chanover, N. 2014, DPS, 46, 11002 Meech, K., et al. 2011, ApJ, 734, L1 Min, M., Hovenier, J. W., de Koter, A., Waters, L. B. F. M., & Dominik, C. 2005, Icarus, 179, 158 Nakamura-Messenger, K., Clemett, S. J., Messenger, S., & Keller, L. P. 2011, Meteoritics and Planetary Science, 46, 843 -- 22 -- Ogliore, R. C., et al. 2011, ApJ, 745, L19 Olofsson, J., et al. 2010, A&A,, 520, A39 Oort, J. H. 1950, Bull. Astron. Inst. Netherlands, 11, 91 Ootsubo, T., Watanabe, J.-I., Kawakita, H., Honda, M., & Furusho, R. 2007, Planet. Space Sci., 55, 1044 Pitticov´a J., Woodward, C. E., Kelley, M. S., & Reach, W. T. 2008, AJ, 136, 112' Reach, W. T., Vaubaillon, J., Lisse, C. M., Holloway, M., & Rho, J. 2010,Icarus, 208, 276 Rodriguez, E., Ortiz, J. L., Lopez-Gonzalez, M. J., et al. 1997, A&A, 324, L61 Sandford, S. A., Al´eon, J., Alexander, C. M. O., et al. 2006, Science, 314, 1720 Schleicher, D. G., Lederer, S. M., Millis, R. L., & Farnham, T. L. 1997, Science, 275, 1913 Schleicher, D. G., et al. 1998, Icarus, 132, 397 Schulz, R., Hilchenbach, M., Langevin, Y., et al. 2015, Nature, 518, 216 Sitko, M. L., Russell, R. W., Woodward, C. E., et al. 2013, Lunar and Planetary Science Conference, 44, 1154 Sitko, M. L., Lisse, C. M., Kelley, M. S., et al. 2011, AJ, 142, 80 Sitko, M. L., Lynch, D. L., Russell, R. W., & Hanner, M. S. 2004, ApJ, 612, 576 Stodolna, J., Jacob, D., & Leroux, H. 2012, Geochim. Cosmochim. Acta, 87, 35 Sugita, S., Ootsubi, T., Kadono, T., et al. 2005, Science, 310, 274 Temi, P., Marcum, P. M., Young, E., et al. 2014, ApJS, 212, 24 Velbel, M. A., & Harvey, R. P. 2007, Lunar and Planetary Science Conference, 38, 1700 Walsh, K. J., et al. 2011, Nature, 475, 206 Watanabe, J.-I., Honda, M., Ishiguro, M., et al. 2009, PASJ, 61, 679 Watson, D. M., Leisenring, J. M., Furlan, E., et al. 2009, ApJS, 180, 84 Wehrstedt, M., & Gail, H.-P. 2008, arXiv:0804.3377 Williams, G. G., Olszewski, E., Lesser, M. P., & Burge, J. H. 2004, Proc. SPIE, 5492, 787 -- 23 -- Williams, G. V. 2015, Observations and Orbits of Comets, Minor Planet Elec. Circ., 2015- A10 Wirick, S., Flynn, G. J., Keller, L. P., et al. 2009, Meteoritics and Planetary Science, 44, 1611 Wooden, D. H. 2008, Space Sci. Rev., 138, 75 Wooden, D. H., Harker, D. E., Woodward, C. E., et al. 1999, ApJ, 517, 1034 Wooden, D. H., Woodward, C. E., & Harker, D. E. 2004, ApJ, 612, L77 Wooden, D. H., Harker, D. E., & Brearley, A. J. 2005, ASP Conf. Series, v341, eds. A. N. Krot, E. R. D. Scott, & B. Reipurth. (San Francisco: ASAP), p.774 Wooden, D. H., Woodward, C. E., Kelley, M. S., Harker, D. E., et al. 2011, EPSC Abs. Vol. 6, EPSC-DPS2011-1557 Woodward, C. E., Gehrz, R. D., Mason, C. G., Jones, T. J., & Williams, D. M. 1998, Earth Moon and Planets, 81, 217 Woodward, C. E., Kelley, M. S., Bockel´ee-Morvan, D., & Gehrz, R. D. 2007, ApJ, 671, 1065 Woodward, C. E., et al. 2011, AJ, 141, 181 Woodward, C. E., Russell, R. W., Harker, D. E., Kim, D. L., Cabreira, B., Sitko, M. L., Wooden, D. H., & Kelley, M. S. 2013, IAUC 9256 Young, E. T., Becklin, E. E., Marcum, P. M., Roellig, T. L., et al. 2012, ApJ, 749, L17 Zacharias, N., Monet, D. G., Levine, S. E., Urban, S. E., Gaume, R., Wycoff, G. L. 2004, Bulletin of the American Astron. Soc. 36, 1418 Zolensky, M. E., Nakamura-Messenger, K., Rietmeijer, F., et al. 2008, Meteoritics and Plan- etary Science, 43, 261 Zolensky, M. E., Zega, T. J., Yano, H., et al. 2006, Science, 314, 1735 This preprint was prepared with the AAS LATEX macros v5.2. -- 24 -- v [R] N E 400 200 0 -200 ) c e s c r a ( t e s f f O C E D 200 0 RA Offset (arcsec) -200 Fig. 1. -- The Bok R-band optical images of comet C/2012 K1 (Pan-STARRS) obtained on 2014 June 04.24 UT, with logarithmic greyscale color map. The vector indicating the direction of the comet's motion and the vector indicating the direction toward the Sun are also provided. -- 25 -- N E v 19.7 µm -100 100 0 RA Offset (arcsec) ) c e s c r a ( t e s f f O C E D 150 100 50 0 -50 -100 -150 200 Fig. 2. -- The SOFIA 19.7 µm image of comet C/2012 K1 (Pan-STARRS) obtained on 2014 June 13.17 UT, with logarithmic color map. The vector indicating the direction of the comet's motion and the vector indicating the direction toward the Sun are also provided. -- 26 -- O3 Fig. 3. -- The SOFIA grism observations of comet C/2012 K1 (Pan-STARRS). Grism ob- servations from three flight series where combined to create a single average SED, and a three-point unweighted rectangular smoothing function was applied to the composite spec- tra to improve the signal-to-noise. The filled circles are the broadband photometric obser- vations obtained on 2015 June 04 UT when each of the filters were observed on a single flight. The photometry is scaled by a factor of 0.75 (which is of the order of absolute photo- metric calibration uncertainty of FORCAST) to match the flux density of the grism spectra and demonstrate that the shape of the photometric SEDs mimics that of the spectroscopic spectral segments. The vertical grey bar indicates the region of terrestrial ozone absorption. -- 27 -- O3 Crys Oliv Model Amor Carbon Amor Pyrox Amor Oliv Fig. 4. -- The spectral decomposition of the SOFIA grism observations (see Fig. 3) of comet C/2012 K1 (Pan-STARRS) derived from thermal modeling. The filled circles are the broad- band photometric observations obtained on 2015 June 04 UT when each of the filters were observed on a single flight. The filter photometry is scaled by a factor of 0.75 (which is of the order of absolute photometric calibration uncertainty of FORCAST) to match the flux density of the grism spectra and demonstrate that the shape of the photometric SEDs mim- ics that of the spectroscopic spectral segments. The vertical grey bar indicates the region of terrestrial ozone absorption. -- 28 -- O3 Fig. 5. -- The observed SOFIA grism flux density of comet C/2012 K1 (Pan-STARRS) near the 10 µm silicate emission feature divided by a ≃ 215 K blackbody continuum (Fλ/Fλ,T ) to highlight the details of the 10 µm silicate feature. The vertical grey bar indicates the region of terrestrial ozone absorption. -- 29 -- Fig. 6. -- Azimuthally averaged radial profile fluxes as a function of linear radius as measured in the R band from the optical photocenter of comet C/2012 K1 (Pan-STARRS) obtained on 2014 June 04.24 UT. The solid blue line denotes a 1/ρ profile. -- 30 -- Fig. 7. -- The bolometric albedo as a function of phase angle for a sample of Nearly Isotropic Comets (NICs; filled orange circles) and Ecliptic Comets (ECs; filled purple squares) derived from the literature following the prescription of (Gehrz & Ney 1992), Our measurement of the NIC C/2012 K1 (Pan-STARRS), 0.14±0.01 at a phase angle of 34.76◦, is indicated by the red star. The phase angles for each comet are obtained from the JPL Horizons ephemerides. -- 31 -- Table 1. SOFIA Observational Summary -- Comet C/2012 K1 (PAN-STARRS)∗ Observation Date 2014 UT (dd-mm hr:min:s) InstCfg Grism or Fltr λc (µm) Total On Src Exp Integ Time Time (sec) (sec) rh ∆ (AU) (AU) Taila Dust Taila Phase Gas Ang PSAng PsAMV (◦) (◦) (◦) FOF176 06-04T03:35:31 06-04T03:35:31 06-04T04:16:38 06-04T04:16:38 FOF177 06-06T04:15:14 06-06T04:29:07 06-06T04:51:42 FOF178 06-11T03:53:40 06-11T03:57:33 06-11T04:32:22 FOF179 06-13T04:04:51 06-13T04:23:04 06-13T04:46:21 Imaging Dual 19.71 Imaging Dual 31.46 Imaging Dual 11.09 Imaging Dual 31.46 29.5 29.5 30.8 30.8 616.0 1.708 1.688 34.76 105.75 616.0 216.0 252.0 78.33 Imaging SWC 19.71 Grism LWC G227 Grism SWC G111 Imaging SWC 19.71 Grism LWC G227 Grism SWC G111 198.0 1.684 1.711 34.76 104.16 45.0 22.5 1800.0 24.0 1920.0 204.0 1.628 1.769 34.46 100.71 42.2 23.0 1748.0 24.0 3336.0 77.47 75.73 Imaging SWC 19.71 Grism LWC G227 Grism SWC G111 330.0 1.605 1.793 34.22 45.1 23.0 1564.0 24.0 2016.0 99.62 75.17 ∗Notes. Orbital elements derived from JPL Horizons, ssd.jpl.nasa.gov/horizons.cgi. aVector direction measured CCW (eastward) from celestial north on the plane of the sky. -- 32 -- Table 2. SOFIA Aperture Photometry and ǫf ρ of Comet C/2012 K1 (PAN-STARRS) Observation Date UT 2014 InstCfg Fltr λc (dd-mm hr:min:s) (Imaging) (µm) Flux Densitya (Jys) λFλ (×10−16 W cm−2) ǫf ρ (cm) FOF176 06-04T03:35:31 06-04T03:35:31 06-04T04:16:38 06-04T04:16:38 FOF177 06-06T04:15:14 FOF178 06-11T03:53:40 FOF179 06-13T04:04:51 DUAL 19.71 15.102 ± 1.117 DUAL 31.46 12.861 ± 2.172 DUAL 11.09 6.119 ± 1.217 DUAL 31.46 11.857 ± 3.149 2.297 ± 0.170 1.226 ± 0.207 1.654 ± 0.329 1.130 ± 0.300 14900 ± 1100 13000 ± 2200 16300 ± 3200 11900 ± 3200 SWC 19.71 18.338 ± 2.343 2.789 ± 0.356 17900 ± 2300 SWC 19.71 16.962 ± 2.246 2.580 ± 0.342 16100 ± 2100 SWC 19.71 16.991 ± 1.777 2.584 ± 0.270 16000 ± 1700 aMeasured in a circular aperture with a radius of 9.984′′ centroided on the photocenter of the comet nucleus. -- 33 -- Table 3. Best-fit Thermal Model Parameters and Derived Grain Mineralogy of Comet C/2012 K1 (Pan-STARRS)∗ Dust component Np a ×1016 Sub-µm mass fraction Amorphous pyroxene 8526.433+1067.813 −1494.938 0.310+0.043 −0.060 Amorphous olivine 1228.326+213.563 −640.688 0.045+0.009 −0.026 Amorphous carbon 15246.560+213.563 −427.125 0.555+0.009 −0.017 Crystalline olivine 2476.880+2135.626 −854.250 0.090+0.078 −0.031 Crystalline pyroxene 0.000+1067.813 −0.000 0.000+0.043 −0.000 Other model parameters χ2 ν Degrees of freedom 0.98 156 Total submicron grain massb (7.663+1.310 −0.952) × 105 kg Silicate/carbon ratio fcryst 0.80+0.25 −0.20 0.202+0.297 −0.099 ∗Notes. Uncertainties represent the 95% confidence level. aNumber of grains at the peak of the grain size distribution. bThe total mass of the sub-µm sized grains contained within the spectral extraction aperture. -- 34 -- Table 4. Thermal Modeling Dust Characteristics of Select Comets Comet Class 1/aorig a SACb fcryst Range Ratio (10−6 AU−1) (%) N apeak (µm) Refs∗ NIC/OCc C/2012 K1 (Pan-STARRS) C/2007 N3 (Lulin) C/2001 Q4 (NEAT) C/2002 V1 (NEAT) C/1995 O1 (Hale-Bopp) EC/JFCc 42.9 32.2 61.2 2279.3 3805.0 9P/Tempel 1 (∼ 1hrs post-impact, ctr) 17P/Holmes 73P/SW3-B (Apert B) 73P/SW3-C (Apert M) . . . . . . . . . . . . 0.6-1.1 10-50 34-51 0.42-0.54 1.7-5.7 0.6 This work 0.9 0.3 0.5 66-69 60-78 8.1-13.3 3.4-3.7 0.2 [1] [2,3] [3] [4,5] 3.4 4.2 3.7 3.7 71 1.38 3.4-4.4 19-25 ∼ 42c 43-69 1.09-1.59 57-69 0.60-0.75 0.2 3.7 . . . 3.4 3.4 0.2 . . . 0.5 0.3 [7] [7] [8] [8] aComputed from thermal model sub-µm silicate mass fractions using Eqn.(1) described in §3.3, where the range includes the model uncertainties when known. bDefined as the silicate-to-amorphous carbon ratio derived from thermal modelling of the SEDs where the range includes the model uncertainties when known. cComet dynamical class divisions are NIC/OC = Nearly Isotropic/Oort Cloud; EC/JFC = Ecliptic/Jupiter-Family dEstimated from Reach et al. (2010) who provide abundances weighted by grain surface area. ∗References. (1) Woodward et al. (2011); (2) Wooden et al. (2004); (3) Ootsubo et al. (2007); (4) Harker et al. (2002); (5) Harker et al. (2004a); (6) Harker et al. (2007); (7) Reach et al. (2010); (8)Harker et al. (2011) -- 35 -- Table 5. Afρ Values for Comet C/2012 K1 (PanSTARRS) on 2014 June 04.24 UT Aperture a (arcsec) ρ (km) R (mag) Afρ b (cm) 11.24 19.97 25.14 33.53 41.91 50.29 6705 11912 14999 19999 24999 29999 12.157±0.008 11.593±0.009 11.344±0.009 11.065±0.009 10.835±0.009 10.674±0.009 5731 ± 42 5424 ± 43 5417 ± 43 5253 ± 41 5194 ± 41 5020 ± 39 aEffective circular aperture diameter. bAfρ values corrected to zero phase (see §3.5).
1607.06697	1	1607	2016-07-22T14:55:28	X-Raying the Dark Side of Venus - Scatter from Venus Magnetotail?	[ "astro-ph.EP" ]	This work analyzes the X-ray, EUV and UV emission apparently coming from the Earth-facing (dark) side of Venus as observed with Hinode/XRT and SDO/AIA during a transit across the solar disk occurred in 2012. We have measured significant X-Ray, EUV and UV flux from Venus dark side. As a check we have also analyzed a Mercury transit across the solar disk, observed with Hinode/XRT in 2006. We have used the latest version of the Hinode/XRT Point Spread Function (PSF) to deconvolve Venus and Mercury X-ray images, in order to remove possible instrumental scattering. Even after deconvolution, the flux from Venus shadow remains significant while in the case of Mercury it becomes negligible. Since stray-light contamination affects the XRT Ti-poly filter data from the Venus transit in 2012, we performed the same analysis with XRT Al-mesh filter data, which is not affected by the light leak. Even the Al-mesh filter data show residual flux. We have also found significant EUV (304 A, 193 A, 335 A) and UV (1700 A) flux in Venus shadow, as measured with SDO/AIA. The EUV emission from Venus dark side is reduced when appropriate deconvolution methods are applied; the emission remains significant, however. The light curves of the average flux of the shadow in the X-ray, EUV, and UV bands appear different as Venus crosses the solar disk, but in any of them the flux is, at any time, approximately proportional to the average flux in a ring surrounding Venus, and therefore proportional to the average flux of the solar regions around Venus obscuring disk line of sight. The proportionality factor depends on the band. This phenomenon has no clear origin; we suggest it may be due to scatter occurring in the very long magnetotail of Venus.	astro-ph.EP	astro-ph	X-Raying the Dark Side of Venus - Scatter from Venus Magnetotail? M. Afshari1,2, G. Peres1,2, P. R. Jibben3, A. Petralia1,2, F. Reale1,2, M. Weber3 1Dipartimento di Fisica e Chimica, Universit`a di Palermo, Piazza del Parlamento 1, 90134, Italy [email protected] 2 INAF- Osservatorio Astronomico di Palermo, Palermo, Piazza del Parlamento 1, 90134, Italy 3Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA Abstract This work analyzes the X-ray, EUV and UV emission apparently coming from the Earth-facing (dark) side of Venus as observed with Hinode/XRT and SDO/AIA during a transit across the solar disk occurred in 2012. We have measured significant X-Ray, EUV and UV flux from Venus' dark side. As a check we have also analyzed a Mercury transit across the solar disk, observed with Hinode/XRT in 2006. We have used the latest version of the Hinode/XRT Point Spread Function (PSF) to deconvolve Venus and Mercury X-ray images, in order to remove possible instrumental scattering. Even after deconvolution, the flux from Venus' shadow remains significant while in the case of Mercury it becomes negligible. Since stray-light contamination affects the XRT Ti-poly filter data from the Venus transit in 2012, we performed the same analysis with XRT Al-mesh filter data, which is not affected by the light leak. Even the Al-mesh filter data show residual flux. We have also found significant EUV (304 A, 193 A, 335 A) and UV (1700 A) flux in Venus' shadow, as measured with SDO/AIA. The EUV emission from Venus' dark side is reduced when appropriate deconvolution methods are applied; the emission remains significant, however. The light curves of the average flux of the shadow in the X-ray, EUV, and UV bands appear different as Venus crosses the solar disk, but in any of them the flux is, at any time, approximately proportional to the average flux in a ring surrounding Venus, and therefore proportional to the average flux of the solar regions around Venus' obscuring disk line of sight. The proportionality factor depends on the band. This phenomenon has no clear origin; we suggest it may be due to scatter occurring in the very long magnetotail of Venus. Subject headings: Venus; Mercury; X-rays; Deconvolution; Hinode/XRT; SDO/AIA; magne- totail -- 2 -- 1. Introduction Transits of Mercury and Venus across the solar disk are well-observed celestial phenomena. Recently, the transit of Mercury observed with Hinode/X-Ray Telescope (XRT; Golub et al. 2007) has been used by Weber et al. (2007) to test the sharpness of the instrument Point Spread Function (PSF). Reale et al. (2015) used Hinode/XRT observations of a Venus transit to measure the size of Venus in the X-ray band thus inferring the extension and optical thickness of Venus' atmosphere. The methods and implications of the latter work reach into planetary physics and hint at similar methods to be potentially used, in the future, for exoplanets. In this work we analyze the same set of observations to explore the residual X-ray emission observed in Venus' shadow and find, with the help of an updated version of the Hinode/XRT PSF, that this emission is not due to instrumental scattering and may have an origin more directly related to Venus. Previous observations with Chandra in 2001 and then in 2006/2007 confirmed the X-ray emission from the sunlit side of the Venus (Dennerl 2002 and Dennerl 2008). In Section 2 we present the observations of Mercury and Venus with a brief summary of the satellites and their instruments; in Section 3 we measure the residual flux in the shadow of Mercury in X-ray and of Venus in X-Ray, EUV and UV bands, and its evolution as Venus crosses the solar disk. In Section 4 we deconvolve X-ray images using the updated PSF and different codes, and again explore similarities and differences among the various observations; in Section 5 we describe the XRT straylight contamination and present our results taken with the Al-mesh filter. In Section 6 we show similar results obtained in EUV and UV bands. Section 7 contains our discussion and the conclusions. 2. Observation: Transit of Mercury and Venus On 2006 Nov 08, Mercury passed across the solar disk. Its transit lasted for almost five hours and was observed with Hinode/XRT in the X-ray band; Fig. 1 shows a selected image of this phe- nomenon. A Venus transit was observed with Hinode/XRT in 2012 while it was crossing the northern hemi- sphere of the Sun; the transit lasted over six hours. On the 5th of June 2012, the Venus transit began at 22:09 UTC and finished on June 6th at 04:49 UTC. The Venus transit was also observed with the Solar Dynamics Observatory/Atmospheric Imaging Assembly (SDO/AIA) (Pesnell et al. 2012) in the Ultraviolet (UV) and Extreme Ultraviolet (EUV) bands. Fig. 1 shows an image taken during this transit. In the following we briefly discuss the satellites and the instruments which took the data used in this work. -- 3 -- Fig. 1. -- Left: Mercury transit across the Sun observed with Hinode/XRT in the X-Ray band. (Time of observation, 2006-11-08 23:51:04.571). Right: Venus (black circle) approaching the Sun, observed with Hinode/XRT in the X-Ray band. (Time of observation 2012-06-05 21:57:39.893). 2.1. Hinode/XRT The Hinode satellite (formerly Solar-B) of the Japan Aerospace Exploration Agency's Institute of Space and Astronautical Science (ISAS/JAXA) was successfully launched in September 2006. There are three instruments onboard: the Solar Optical Telescope (SOT), the EUV Imaging Spec- trometer (EIS), and the X-Ray Telescope (XRT). We used only data from XRT. XRT is a high-resolution grazing-incidence telescope with a modified Wolter-I telescope design that uses grazing incidence optics with an angular resolution consistent with 1.0286 arcsec per pixel at the CCD (Golub et al. 2007). An improved version of the Hinode/XRT PSF has been derived by P. R. Jibben of the XRT in- strument team. The Model PSF has 99% of the encircled energy within a 100 arcsec diameter with the remaining 1% scattered beyond1. The PSF model at 0.56 keV is: P SF =   exp(− 2 r σ2 ) γ 2+r2 a , if r ≤ 3.4176; 0.03 r , 0.15 r2 , if 3.4176 ≤ r ≤ 5; if 5 ≤ r ≤ 11.1; (11.1)2 r4 ×0.15 , if r ≥ 11.1; 1For more information about the derivation of the PSF, the interested reader can refer to Appendix A. -- 4 -- Where r = radial distance in arc seconds, a = 1.31946, σ = 2.19256 and γ = 1.24891. This PSF is planned for distribution in the XRT branch of SolarSoft (Freeland & Bentley 2000) (Bentley & Freeland 1998). 2.2. SDO/AIA The Solar Dynamics Observatory (SDO) was launched on February 11, 2010. The spacecraft includes three instruments: the Extreme Ultraviolet Variability Experiment (EVE), the Helioseis- mic and Magnetic Imager (HMI), and the Atmospheric Imaging Assembly (AIA) (Lemen et al. 2012). We used only data taken with AIA. AIA, with an angular resolution of 0.6 arcsec per pixel, provides narrow-band imaging in seven extreme ultraviolet (EUV) band passes centered on specific lines: (94 A, 131 A, 171 A, 193 A, 211 A, 304 A and 335 A) and in two UV band-passes near 1600 A and 1700 A (Lemen et al. 2012). 2.3. Data sets For Venus' shadow analysis, we used six different data sets in the X-ray band, each with more than 300 images, and four data sets from AIA: at 1700 A, 335 A, 304 A, 193 A, respectively with 114, 169, 118 and 119 images. For the Mercury shadow analysis we used one data set in the X-ray band. A summary of the data sets is presented in Table 1. The filters for all selected images of Venus in the X-ray band are Ti-poly and Al-Mesh, and Al-poly for Mercury images. (Ti-poly and Al-poly are metal foils on a polyimide substrate, and Al-mesh is an Al foil mounted on a fine stainless steel mesh.) The field of view is 384 × 384 pixels for Ti-poly and Al-poly images, and is 192 × 192 pixels for the Al-mesh images (where each CCD pixel has been summed 2 × 2). The AIA and XRT plate scales are 0.6 arcsec per pixel and 1.0286 arcsec per pixel (.0572 arcsec per pixel for Al-mesh), respectively. Hinode/XRT didn't take any full solar disk images of the Venus transit but only partial images of the disk where Venus was. For the data analysis we used the standard instrumental calibration routines provided through SolarSoft. 3. Data Analysis To analyze the features of Venus' and Mercury's shadows in the X-Ray band we have measured, in each image, the flux across the planetary disk and in the nearby solar disk regions. To illustrate the features of such an emission we show the average flux measured along strips 3 pixels wide (in -- 5 -- Table 1: Summary of data sets of Venus and Mercury Planet Filter Instrument Start Time of observation Final Time of observation (UTC Time) (UTC Time) Venus Venus Venus Venus Venus Venus Venus Venus Venus Venus Mercury Ti-poly Ti-poly Ti-poly Ti-poly Ti-poly 193A 304A 335A 1700A Al-mesh Al-poly Hinode/XRT Hinode/XRT Hinode/XRT Hinode/XRT Hinode/XRT SDO/AIA SDO/AIA SDO/AIA SDO/AIA Hinode/XRT Hinode/XRT 2012-06-05T20:03:00.615 2012-06-05T21:58:39.912 2012-06-06T00:23:57.272 2012-06-06T02:06:57.299 2012-06-06T03:51:27.859 2012-06-05T22:23:07.84 2012-06-05T22:23:08.13 2012-06-05T22:25:03.62 2012-06-05T22:32:07.71 2012-06-05T21:06:28.326 2006-11-08T23:50:12.052 2012-06-05T21:58:33.335 2012-06-06T00:23:37.912 2012-06-06T02:06:39.223 2012-06-06T03:51:08.500 2012-06-06T06:47:15.490 2012-06-06T04:17:07.84 2012-06-06T04:17:08.12 2012-06-06T04:01:03.62 2012-06-06T04:11:19.71 2012-06-06T06:44:46.712 2006-11-08T23:59:16.234. order to have a significant S/N ratio). We have considered strips along the planet's diameters, along both the N-S (vertical) and the E-W (horizontal) directions. 3.1. Venus Intensity Profile Analysis In Fig. 2 we plot the Intensity Profile (IP) of Venus' shadow along both the horizontal and vertical directions in the X-Ray band, as collected through the Ti-poly filter of XRT. Venus casts a shadow with an angular diameter of ≈ 60′′. The IP of Venus' shadow consists of three parts: a shadow edge, a region of steep descent on both sides and a residual flux. The regions of steep descent have smooth corners on either side because of the convolution of a step function with the PSF (Reale et al. 2015; Weber et al. 2007). The X-Ray residual flux in Venus' shadow appears too high to be compatible with background signal (Kobelski et al. 2014). We have superimposed in Fig. 3 the IPs taken at different times and positions of Venus on the solar disk. We did not align the borders of Venus, since the purpose here is only to show the level of residual flux (albeit not sampling regularly the whole transit). As we can see the level of residual flux is high at any time; the intensity at the shadow's edge strongly depends on the nearby (along the line of sight) solar emission near Venus at the time the specific frame was taken (Reale et al. 2015). To check the effect of possible instrumental scattering in XRT across Venus' shadow, especially when close to active regions, we took the average flux measured in three regions: in the Venus disk and in two concentric annuli around the Venus disk. Annulus 1 has inner radius Rv, namely the Venusian radius, and outer radius 2Rv, as shown in Fig. 4. Annulus 2 has inner and outer radii Rv and 5Rv, respectively. We plotted the evolution of the mean flux inside each of these annular -- 6 -- Horizontal Cross Section of Venus Shadow 40 30 20 10 ) s / N D ( y t i s n e t n I Vertical Cross Section of Venus Shadow Shadow edge Steep descent Residual Flux 40 30 20 10 ) s / N D ( y t i s n e t n I 0 0 20 40 Pixel 60 80 100 0 0 20 40 Pixel 60 80 100 Fig. 2. -- Top Left: Venus transit above the active region. Top Right: Schematic view of horizontal (E-W) and vertical (N-S) strips, green and blue, respectively. Bottom Left: Vertical IP of Venus' shadow. Bottom Right: Horizontal IP of Venus' shadow. ) s / N D ( x u F l 30 25 20 15 10 5 0 0 Comparison of Venus Shadows in X−Ray Band Horizontal Direction 20 40 Pixel 60 80 ) s / N D ( x u F l 30 25 20 15 10 5 0 0 Comparison of Venus Shadows in X−Ray Band Vertical Direction 20 40 Pixel 60 80 Fig. 3. -- IP of Venus' shadow in both horizontal (Left panel) and vertical (Right panel) directions at different times of observations and positions of Venus on solar disk. -- 7 -- X-Ray Band (Ti-poly) Annulus 2; (Rv-5Rv) Annulus 1; (Rv-2Rv) Venus disk 100 200 Time (min) 300 400 100 ) 1 - s N D ( y t i s n e n t I 10 1 0 (a) (b) Fig. 4. -- Left: Central Black circle: Venus disk; Red annulus: annulus 1 around the Venus disk. Right: the evolution of mean X-ray flux inside Venus disk (Black), annulus 1 (Red) and annulus 2 (Blue) vs. TOBS. Annulus 1 has inner and outer radii Rv and 2Rv. Annulus 2 has inner and outer radii Rv and 5Rv. The vertical bar on the right shows the typical error size. The red vertical line in the lower left marks the first contact. regions versus the time of observation (TOBS) in Fig. 4, along with the flux measured in Venus' shadow. In order to have a comparable time series in all light curves we chose TOBS = 2012-06- 05T21:58:39.912 of one Hinode/XRT image as the reference time. Also, the time of Venus' entrance onto the solar disk is marked with a red vertical line. For each data point the Poisson errors of DN (Digital Numbers) has been used as the error bars in the light curves. This amounts to assume a DN-to-photon conversion factor of 1; according to Narukage et al. (2011) such a factor applies to T ∼ 1.5 MK, typical of the average, or quiet, corona. The conversion factor changes only slightly over the temperature range of interest for the non-flaring corona; since, also, the error depends on the square root of the photon number, the error bar determined is adequate even considering the multi-temperature corona The initial high annulus flux is due to limb brightening, crossed during the initial phase of the Venus transit; then Venus gets close to a big active region, during the central phase of transit, and the mean flux of both the Venus disk and the annuli increases. (The maximum mean flux is measured in this phase.) As Venus moves away from the active region the flux decreases slowly. At the final stage, Venus completes the transit and touches the other limb with a small increase in mean flux at the end of all of the three curves. The blue curve does not cover the full data set: for some images, the annulus with the outer radius 5Rv extends beyond the borders of the X-ray image. -- 8 -- 3.2. Mercury IP Analysis Since the atmosphere of Venus may contribute to -- or be the cause of -- the residual flux in IPs of Venus' shadow, we considered the shadows of other celestial objects occulting the Sun but lacking an atmosphere, in order to remove the possible effects of atmosphere. As a first choice we selected Mercury, already analyzed by Weber et al. (2007). If some effect due to PSF scattering is present in the case of Venus, it should be stronger in the case of the smaller Mercury disk: Mercury casts a shadow with an angular diameter of ≈ 10′′. We have also made some analysis, not reported here, of the Moon's shadow during solar eclipses observed with Hinode/XRT and found almost zero signal coming from the Moon's X-ray shadow. In Fig. 5 we have plotted the IP of Mercury's shadow in the X-Ray band, as taken through the Al-poly filter of XRT, along both the horizontal and the vertical directions. The relevant images were 384 × 384 pixels large. Horizontal Cross Section of Mercury Shadow Vertical Cross Section of Mercury Shadow 60 50 40 30 20 10 ) s / N D ( x u F l 60 50 40 30 20 10 ) s / N D ( x u F l 0 0 5 10 15 Pixel (a) 20 25 30 0 0 5 10 20 25 30 15 Pixel (b) Fig. 5. -- Left: Vertical IP of Mercury's shadow. Right: Horizontal IP of Mercury's shadow. Images taken through the Al-poly filter of XRT. In the case of Mercury we initially find a residual flux, at a level comparable to that in Venus' shadow, as well as a smooth profile. Therefore the effect appears to be, at first sight, the same for Venus and Mercury. As a next step, in order to remove possible instrumental effects due to the PSF, we deconvolved Venus images using the Hinode/XRT PSF and other codes, and compared the relevant results. We also deconvolved Mercury images with the same tools to cross-check the results. -- 9 -- 4. Deconvolution Among different indirect methods of deconvolution such as least-squares fit, Maximum Entropy, Maximum likelihood (Starck et al. 2002), and Richardson-Lucy (Richardson 1972; Lucy 1974), we used the codes based on Maximum Likelihood (M-L) and Richardson-Lucy (AIA Richardson-Lucy; AIA) available in SolarSoft IDL libraries. For a short description of the codes that we used, please refer to Appendix B. With the above codes and the Hinode/XRT PSF we performed deconvolution of the images, and then compared the results to pinpoint similarities and differences; the Venus Ti-poly images were 384 × 384 pixels large. We repeated the cross section analysis, presented before, for the deconvolved images. 4.1. Deconvolution of Mercury shadows Mercury and Venus have been observed at different times, in 2006 and in 2012 respectively, and with different filters. However our aim here is just to check the performance of the updated PSF in removing any emission concerning the instrumental scattering. The cross sections of Mercury shadow, before and after deconvolution, are presented in Fig. 6. After deconvolution, the cross section of Mercury's shadow has practically zero residual flux with edges sharper than those of the original profiles. These results are very important since they not only confirm the accuracy of the updated Hin- ode/XRT PSF but show that, at least in the case of Mercury, the residual flux is due to PSF scattering. An analogous study was done by (Weber et al. 2007), with similar results, using a pre- vious version of the PSF of Hinode/XRT. 4.2. Deconvolution of Venus shadows Cross sections of Venus' shadow after deconvolution are shown in Fig. 7. These cross sections of Venus images deconvolved with the M-L and AIA codes show that: • In some cases the cross sections of images deconvolved with the M-L code have more fluctuations in comparison to those obtained with the AIA code; • For Venus, similarly to the Mercury case, the borders seem to be sharper after deconvolution; -- 10 -- Black= orginal;Red=Max_Likelihood Black=orginal;Red=Max_Likelihood;Blu=aia_deconv 70 ) s / N D ( x u F l 60 50 40 30 20 10 70 60 50 40 30 20 10 ) s / N D ( x u F l 0 0 5 10 15 Pixel (a) 20 25 30 0 0 5 10 20 25 30 15 Pixel (b) Fig. 6. -- Left: Mercury IP before (Black) and after deconvolution with the M-L code (Red). Right: comparison between IP before (Black) and after deconvolution with AIA (Blue) and M-L (Red). 40 30 20 10 ) s / N D ( y t i s n e n I t Black=Original;Red=Max_Likelihood Black=Original;Red=Max_Likelihood;Blue=aia_Deconv 40 Ti − poly Filter Ti − poly Filter ) s / N D ( y t i s n e n I t 30 20 10 0 0 20 40 Pixel (a) 60 80 100 0 0 20 40 60 80 100 Pixel (b) Fig. 7. -- Left: Venus IP before (Black) and after deconvolution made with the M-L code (Red). Right: Comparison between IP before (Black) and after deconvolution made with AIA (Blue) and M-L (Red). . • Residual flux is present in Venus images even after deconvolution; such a flux is significantly higher than the noise. Residual flux present in Venus cross sections after deconvolution does not appear to be due to the Hinode/XRT PSF, since the accuracy of the PSF has been confirmed in the Mercury analysis. -- 11 -- Being that the angular size of Mercury is considerably smaller than that of Venus, any effect of PSF scattering should manifest itself more in the Mercury cross sections. Since both the M-L and AIA codes are iterative we changed the number of iterations during the deconvolution process for some images in each dataset to check the effect of iteration, especially to see whether increasing the number of iterations led to the residual flux being further decreased or removed. The trend is that with increasing the number of iterations the residual flux is still present and its mean value for any reasonable number of iterations is very constant, except that with increasing iterations the fluctuations increase in the IPs. Generally the M-L code is more sensitive to noise and the quality of the images. So the presence of a significant residual flux in Venus' shadow is not due to instrumental scattering but should be related to Venus; for instance, it could originate from some effect occurring in Venus' atmosphere. Comprehensive analysis of deconvolutions show that: • The AIA code does not conserve the total flux, yielding curves with 15% of total flux, so for each image we readjusted the amplitude to conserve the total flux; • Deconvolution causes artifacts and spurious "spikes" at the edge (borders), a common problem in deconvolution which, in the case of Venus, are well identified and do not affect the evaluation of the average flux in the shadow (cf. Fig. 7). We again followed the evolution of the flux in Venus' shadow and in two reference annuli, as done in Section 2, after deconvolution. We plotted the evolution of mean flux inside each of these three regions versus TOBS in Fig. 8. The space-averaged fluxes obtained after deconvolution with the two methods are virtually the same resulting in three light curves, each being two superimposed. The most important points in Fig. 8 are: • The amount of mean flux inside Venus' disk after deconvolution has decreased slightly, especially where close to the active region, therefore deconvolution appears to have removed just a small amount of X-ray flux from Venus' shadow; • The mean fluxes inside the two annuli have not changed after deconvolution (cf. also Fig. 7); • We see that the flux inside Venus' shadow and that inside the two annuli gradually rise as Venus gets more and more inside the solar disk and decreases thereafter; however the flux inside Venus' shadow is not strictly correlated to that inside the two annuli. • There may be some relationship between the observed residual flux and the high surrounding flux; • The mean flux value for both the AIA and M-L deconvolution codes are virtually the same de- spite the fact that the profiles for the deconvolved images are not the same. -- 12 -- X-Ray Band (Ti-poly) Annulus 2; aia-Deconv Annulus 2; Max-Deconv Annulus 1; aia-Deconv Annulus 1; Max-Deconv Venus disk; aia-Deconv Venus disk; Max-Deconv 100 200 Time (min) 300 400 100 ) 1 - s N D ( y t i s n e n t I 10 1 0 Fig. 8. -- Evolution of mean Ti-poly flux after deconvolution: inside Venus disk with AIA (Black) and M-L (Red) codes; inside annulus 1 with AIA (Green) and M-L (Red) codes; inside annulus 2 with AIA (Blue) and M-L (Red) codes. The bar on right shows typical error sizes. The red vertical line in the lower left marks the first contact. As an additional test on the error, we have determined the standard deviation of the resid- ual flux values inside Venus disk and found that it varies between 0.4 -- 1.0 DN s−1, which is negligible in comparison to the observed residual flux (> 5 DN s−1). 5. Light Leak Contamination 5.1. Light Leak Effect on XRT Filters An increase in XRT's straylight was detected on May 9th of 2012, shortly before the Venus transit (5th -- 6th June 2012), which causes significant visible light contributions to the X-Ray images in some filters. In addition, a sudden increase of intensity by a factor of 2 was observed in the visible light measurements (i.e., in the G-band channel). At the same time, the XRT team recognized wood-grain like stripes in daily images taken with the Ti-poly filter (Takeda et al. 2016). The team believes the increase of visible straylight to have been caused by a pinhole puncture in the entrance -- 13 -- aperture filters. The analysis showed that the light leak affects only some of the X-Ray filters: a minor effect was detected for the Al-mesh and Al-poly filters but it was very small (≤ 5 DN s−1), while it strongly affected the Ti-poly and C-poly filters (Takeda et al. 2016). In order to exclude the possibility that Venus residual flux in Ti-poly could be due to the straylight, we used data collected with the Al-mesh filter to repeat the analysis. Importantly, the light leak has a very small effect on the Al-mesh filter, to such a level that it can be neglected (Takeda et al. 2016). 5.2. Al-mesh Filter Analysis In Fig. 9 we present a typical IP of Venus' shadow in both horizontal and vertical directions, taken in an image collected with the Al-mesh filter. 120 100 80 60 40 20 ) s / N D ( y t i s n e t n I Vertical IP Horizontal IP Al − mesh Filter Al − mesh Filter 200 150 100 50 ) s / N D ( y t i s n e t n I 0 0 10 20 Pixel (a) 30 40 0 0 10 30 40 20 Pixel (b) Fig. 9. -- Left: Vertical IP of Venus' shadow. Right: Horizontal IP of Venus' shadow. (XRT Al-mesh filter) . As we can see: • The residual flux is still present in all IP plots. • The intensity profiles of the Al-mesh filter appear approximately 3-5 times higher than Ti-poly ones; the reason is that Al-mesh images are binned 2 × 2 while Ti-poly data are binned 1 × 1 and the filters have different transmissivity; the Al-mesh images are 192 × 192 pixels large. -- 14 -- Also for Al-mesh data we deconvolved images to remove any effect due to the PSF scattering. Sample IP results, after deconvolution, are shown for the vertical direction in Fig. 10. Deconvolution analysis shows that: Black=Original;Red=Max_Likelihood Black=Original;Red=Max_Likelihood;Blue=aia_Deconv Al − mesh Filter 120 100 80 60 40 20 ) s / N D ( y t i s n e t n I Al − mesh Filter 120 100 80 60 40 20 ) s / N D ( y t i s n e t n I 0 0 10 20 Pixel (a) 30 40 0 0 10 30 40 20 Pixel (b) Fig. 10. -- Left: Venus IP before (Black) and after deconvolution with M-L code in the vertical direction (Al-mesh filter). Right: comparison of IP before (Black) and after deconvolution with AIA (Blue) and M-L (Red) in vertical direction. • Artifacts and spurious "spikes" at the edges (borders) of the IPs are much stronger in Al-mesh images in comparison to Ti-poly images. • The AIA code does not conserve the total flux, yielding curves with 60% of total flux (in Ti-poly, 15%), so for each image we rescaled the amplitude to conserve the total flux. • The most important fact is that even after deconvolution residual flux is still present in all of the IPs and is significantly higher than the noise. We have also determined the evolution of the flux in Venus' shadow after deconvolution for Al-mesh images and in two reference annuli, similarly to what we have done for the Ti-poly data. Fig. 11 shows the evolution of the mean flux inside each of these regions versus TOBS. Also for Al-mesh data the DN to photon conversion factor of 1, used to derive the error bars, is appropriate to T ∼ 1 MK and changes slowly over the T range of interest for the non-flaring corona. The ratio of the maximum value of flux inside annulus 1 to the lowest one for Al-mesh data is slightly more than 5, on the average. The ratio is different from that of Ti-poly (≈ 3) probably because the light leak effects (if any) are very small in the case of Al-mesh. We can safely state that the flux detected in Venus' dark side most likely is not due to PSF scat- -- 15 -- 1000 X-Ray Band (Al-mesh); PSF 0.56 keV Annulus 2; aia-Deconv Annulus 1; aia-Deconv Annulus 1; Max-Deconv Annulus 2; Max-Deconv Venus disk; aia-Deconv Venus disk; Max-Deconv ) 1 - s N D ( y t i s n e n t I 100 10 1 -100 0 100 Time (min) 200 300 400 Fig. 11. -- Evolution of mean X-ray flux (as measured through Al-mesh filter) inside Venus disk after deconvolution with AIA (Black) and M-L (Red) codes, inside annulus 1 after deconvolution with AIA (Green) and M-L (Red) codes, and inside annulus 2 after deconvolution with AIA (Blue) and M-L (Red) codes. A typical error bar is shown on the right. The red vertical line in the lower left marks the first contact. tering, noise or light leak, but it may originate from some phenomenon related to Venus. 6. The EUV and UV Flux Analysis We have done a similar analysis of IPs of Venus' shadow in the UV and EUV bands. An image of Venus transit and a sample cross section in the EUV band, taken with SDO/AIA at 193 A, is shown in Fig. 12. Similarly to what was done for the X-Ray band, we deconvolved the EUV images to remove the effects of the PSF. Various works have been dedicated to deriving the PSF of SDO/AIA. Grigis et al. (2012) derived the PSF using pre-flight and post-flight measurements and calibrations. Poduval et al. (2013) derived the in-flight SDO/AIA PSF using several observations, including some of the Moon's limb made during a solar eclipse observed with SDO/AIA. Gonz´alez et al. (2016) -- 16 -- 100 80 60 40 20 ) s / N D ( y t i s n e t n I 193 A 0 0 20 40 60 Pixel 80 100 120 Fig. 12. -- Image in the EUV band (193 A) of Venus transit (Left) and its horizontal IP (Right). used observations of both a solar eclipse and a Venus transit to derive the SDO/AIA PSF. These authors assumed that there is no emission coming from the dark side of Venus during the transit, but then discovered that they needed to include a long range effect, otherwise the parametric PSF they used would not be able to remove "the apparent emission inside the disk of Venus". Interestingly, in a similar but unrelated study, DeForest et al. (2009) used the 2004 Venus transit to determine the TRACE in-flight PSF. They, too, assumed that no EUV radiation comes from Venus' dark side, and then found that "much more scattered light is found than can be accounted for merely by diffraction" and that half of the scattered light was due to some other mechanism. It is quite possible that both Gonz´alez et al. (2016) and DeForest et al. (2009) had discovered, and were trying to account for, some real EUV emission of the kind we find. For the above reasons we decided to adopt the PSF derived in Grigis et al. (2012), which is available in SSW and is a standard in deconvolving AIA EUV images. We have also applied the Poduval et al. (2013) PSF, kindly provided by the author, to test if results are different. This latter PSF is not available for full disk images so we have compared the results obtained with SSW and Poduval PSF only for partial disk images; we found that we get in practice the same average flux with the standard deconvolution in SSW and with the deconvolution which uses Poduval PSF. Being reassured by this result, we resorted to the Grigis et al. (2012) PSF to deconvolve full disk images. We concentrated on full disk images (albeit their number is smaller) because we are thereby certain to remove even possible long range effects. We have, thus, deconvolved each full disk image with the PSF available in SSW and derived the average flux in Venus shadow and in annulus 1. In Fig. 13 we show the evolution of the UV and EUV average flux values as observed with SDO/AIA in the dark side of Venus and in the smallest annulus (annulus 1) taken around Venus, before and after deconvolution performed with the AIA routine and with the Maximum Likelihood (M-L) routine. -- 17 -- ) 1 - s N D ( y t i s n e t n I ) 1 - s N D ( y t i s n e n t I 100.0 10.0 1.0 0.1 0 10.00 1.00 0.10 0.01 0 1700 A 100 200 Time (min) 300 400 193 A 100 200 Time (min) 300 400 1000 100 ) 1 - s N D ( y t i s n e t n I 10 0 1000 100 10 ) 1 - s N D ( y t i s n e n t I 1 0 304 A 100 200 Time (min) 300 400 335 A 100 200 Time (min) 300 400 Fig. 13. -- Mean flux evolution inside Venus disk and annulus 1, before and after deconvolution vs. time in UV and EUV bands. Top Left: 1700 A; Top Right: 304 A; Bottom Left:193 A; Bottom Right: 335 A. Black square: annulus average flux before deconvolution, red X : annulus average flux after ML deconvolution, blue cross: same as red X but after aia deconvolution, black triangle: Venus disk average flux before deconvolution, red star: Venus disk average flux after M-L deconvolution, blue diamond: Venus disk average flux after aia deconvolution. The data in the 335 A band show an unusual behaviour with deconvolution: the average flux inside Venus' disk increases after deconvolution, just the opposite of what one expects and which happens in any other band. The flux in Venus' shadow is rather low before deconvolution, so any signal that is present and not due to noise must be marginal. Indeed, the signal after deconvolution is virtually constant. We are thus forced to consider the results for 335 A images as unreliable; we present all the relevant results just for completeness and as an additional null test, but these results are of little relevance for our problem. No PSF is available for the 1700 A band, so we cannot deconvolve the relevant data; we simply used the non-deconvolved data. -- 18 -- It is immediately apparent that the flux evolution in any EUV band has no evident correlation with that of the X-Ray flux; in the 1700 A band the flux increases as Venus crosses the solar disk, and decreases thereafter. In the 335 A and 193 A bands the opposite occurs; few changes are seen in the 304 A band. However it appears that also in these cases the flux in the shadow clearly follows that in the surrounding ring, almost (but not exactly) by a fixed factor. This approximate proportionality of the flux in Venus' dark side relative to the surrounding regions hints at a strong analogy between the mechanisms generating Venus' dark side emission in X-ray, EUV and UV bands. Fig. 14 shows the evolution of the ratio between flux values inside the annulus and in the Venusian disk, for the original data and for the images deconvolved with both methods. 10 1700 A o i t a R y t i s n e n t I 1 0 100 10 o i t a R y t i s n e t n I 1 0 100 200 Time (min) 300 400 193 A 100 200 Time (min) 300 400 o i t a R y t i s n e n t I o i t a R y t i s n e t n I 100 10 1 0 100 10 1 0 304 A 100 200 Time (min) 300 400 335 A 100 200 Time (min) 300 400 Fig. 14. -- Evolution of (mean annulus 1 flux to mean Venus disk flux) ratio before and after deconvolution in UV and EUV bands. Top Left: 1700 A; Top Right: 304 A; Bottom Left:193 A; Bottom Right: 335 A. Black Triangle: Ratio before deconvolution; Blue cross: Ratio after aia deconvolution; Red X: Ratio after ML deconvolution. Fig. 15 shows the minimum, maximum and mean values of the ratio between flux values inside -- 19 -- the annulus and in the Venus disk versus wavelength and versus temperature (of the plasma which would be observed on the Sun). Since any possible deconvolution would slightly decrease the flux in Venus disk also at 1700 A, we may expect that the data points at 1700 are higher. There appears to be a slight increasing trend with wavelength. 30 20 o i t X-ray(Al-mesh) EUV-193 EUV-304 EUV-335 UV-1700 30 20 o i t X-ray (Al-mesh) EUV-193 EUV-304 EUV-335 UV-1700 l a R x u F n a e M 10 0 1 10 100 Wavelength (A) 1000 10000 l a R x u F n a e M 10 0 103 104 105 Temperature (K) 106 107 (a) (b) Fig. 15. -- Minimum, maximum and mean values of the ratio between the average intensity on solar disk around Venus and average flux inside Venus' shadow vs. wavelength (left panel) and vs. solar plasma temperature (right panel). 7. Conclusion We studied the Venus transit across the solar disk which occurred in 2012 and was observed with Hinode/XRT in the X-Ray band and SDO/AIA in the EUV and UV bands. We have mea- sured a significant X-Ray residual flux from Venus' dark side (i.e., from the Earth-facing side) during the transit that was significantly above the estimated noise level of 2 DN s−1, as reported by Kobelski et al. (2014). Let us discuss the systematic uncertainties of XRT flux. According to Kobelski et al. (2014) there are two kinds of systematic uncertainties for XRT. The first are those which have a reliable quan- titative correction procedure such as: dark current, Fourier, vignetting, and JPEG compression noise sources; their correction procedures have been successfully embedded in xrt prep.pro (the calibration reformatter). Since all of the data we have used for X-ray analysis (both Ti-poly and Al-mesh filters) have been prepared with xrt prep.pro, we expect that this class of uncertainties has been properly corrected and does not explain the observed residual. The second kind of systematic uncertainties are model-dependent and are not included in xrt prep.pro; among them are light scattering by the grazing-incidence mirror of XRT, visible straylight leak and photon counting uncertainty (Kobelski et al. 2014). In this respect the error bars for each flux value have been computed as follows. From the flux values, the exposure times, and the conversion -- 20 -- factor from DN to photons, we have computed the number of photons collected and from them the statistical errors due to Poisson noise. This statistical error has been converted to a flux error bar per data point. The resulting error bar is typically less than the 2 DN s−1 mentioned by (Kobelski et al. 2014). In sections 4 and 5 we comprehensively discuss the PSF scattering and visible light leak effects. To test the performance of the instrument's PSF (i.e., due to instrumental X-ray scattering) and the possible effect of the atmosphere on the residual flux, we studied a Mercury transit across the solar disk, observed with the Hinode/XRT in 2006. We measured an apparent X-Ray residual flux in the case of Mercury before deconvolution. For both Venus and Mercury we used a new version of the Hinode/XRT PSF, selected well illu- minated images in the X-Ray band, and deconvolved them. Even after deconvolution, flux from Venus' shadow has remained significant, while in the Mercury case it has become negligible. So it appears that the observed flux in Venus' shadow is real. As for the Venus case, we have analyzed two X-ray datasets: a set collected with Ti-poly filter and another collected with Al-mesh filter. While the former is potentially strongly affected by a light leak that appeared a short time before that Venus transit, the latter is not. Both datasets, however, clearly show the presence of a significant flux from Venus' dark side, showing the reality of this effect. Although we consider the results from the Al-mesh data set to be a strong confirmation for the observed X-Ray residual flux, the Ti-poly results also provide more confidence about the observed residual flux and prove this effect in more than one filter. The level of the residual flux is not constant: as Venus crosses the solar disk it gradually grows, reaching a maximum roughly halfway through the transit, and then gradually decreases as it approaches the solar limb. The flux changes by an almost fixed factor of the flux of the surrounding solar regions (i.e., along nearby lines of sight) as shown in Figs. 8 and 11. On the other hand the use of the PSF and the test on Mercury convincingly shows the removal of any PSF effect. Furthermore, any light leak effect would instead be expected to be almost uniform or constant in time. The PSF of XRT has also been determined at 1.0 keV. We find that deconvolving the images with this PSF reduces by a factor of about 0.5, on average, the flux inside the Venusian disk. More specifically, the mean flux across the disk before deconvolution is about 24 DN s−1. After decon- volution with the 0.56 keV PSF model it is about 20 DN s−1, but after deconvolution with the 1.0 keV PSF model it is about 10 DN s−1. Therefore a significant flux level still remains, even after deconvolution with the 1.0 keV PSF, showing the reality of the effect nonetheless. In this respect, however, we believe that the PSF at 0.5 keV is more appropriate to our study. In fact, we are detecting photons coming from the corona and re-processed at Venus or in Venus' magnetotail (or something related to Venus), a process which should not raise photon energy. The corona is at a few MK (at most 3 or 5, and only then in some places like active region cores), and no flare appears during the Venus transit. So a relatively smaller fraction of photons are expected even at 0.56 keV. We use the Al-mesh and Ti-poly filters; so considering the coronal spectrum folded with the Al-mesh filter response (Al-mesh data are the most reliable ones for Venus X-ray observations), we may shift the average of the observed plasma emission some 0.1 keVs closer to, -- 21 -- but not at, 0.56 keV. On one hand we are confident that the general result is robust against the choice of PSF model, but the use of the 0.56 keV PSF can be considered to be a conservative evaluation, and so we can use the relevant results quite safely. The analogous kind of analysis made in four EUV bands observed with SDO/AIA has shown that there is also some flux in these bands coming from Venus' night side, and that its evolution clearly follows that of the flux inside an annulus surrounding Venus. The light curves do not show, how- ever, any trend similar to that of the X-Ray flux. Past X-ray observations of Venus were very different, in many respects. In January 2001, Venus was observed for the first time with the Chandra X-ray telescope. Dennerl (2002) proposed that the fluorescent scattering of solar X-rays from Venus' atmosphere was the primary source of the X-ray emission they observed. Not only the morphology, but also the observed X-ray luminosity was consistent with the scattering of solar X-rays (Dennerl 2002). In 2006 and 2007 again with Chandra, besides fluorescent scattering, Solar Wind Charge eXchange (SWCX) emission was clearly detected. Comparison of X-ray images taken in 2006 and 2007 with those obtained in 2001 (taken at a similar phase angle) showed that the limb brightening had increased. This would be the case if the X-ray radiation from Venus was the superposition of scat- tered solar X-rays and SWCX emission. The lack of detection of any SWCX-induced X-ray halo in the first Venus observation was explained by being during a high level of the solar X-ray cycle (Dennerl 2008). Previous X-ray observations, however, have shown X-ray emission from the sunlit side of Venus. The low intensity we detect in X-ray and EUV comes from the dark side of Venus, and appears to have a totally different origin; it appears to evolve during the transit remaining, at any time, approximately proportional to the emission of the solar regions along nearby lines of sight. This intensity cannot be due to scattering in the upper atmosphere of Venus because we should detect a brighter inner rim in Venus' shadow. The effect we are observing could be due to scattering or re-emission occurring in the shadow or wake of Venus. One possibility is due to the very long magnetotail of Venus, ablated by the solar wind and known to reach Earth's orbit (Grunwaldt et al. 1997). This magnetotail could be side- illuminated from the surrounding regions and could scatter, or re-emit, the radiation; the cone of Venus shadow reaches up to 9.6 × 105 km away from Venus, leaving ample space (≈ 4.5 × 107 km) for side-illuminating the magnetotail. The emission we observe would be the reemitted radiation integrated along the magnetotail. One wonders if such an effect is important for exoplanets, in particular for those Jupiter-size planets orbiting very close to their stars; they may have a very large ablated tail, especially if they do not have a magnetic field. To some extent, the study of these tails may help to understand, among other issues, the presence (or lack thereof) of magnetic fields. Future work will study in more detail this phenomenon: we plan to study some faint structures present in the shadow and address possible physical mechanisms involved in generating the residual emission. -- 22 -- Acknowledgments We thank an anonymous referee for suggestions and comments on EUV deconvolution. M.A., G.P., A.P., F.R. acknowledge support from Italian Ministero dell'Universit`a e Ricerca; P.J. and M.W. were supported under contract NNM07AB07C from MSFC/NASA to SAO. Some of the routines for the data analysis and some early evaluations were kindly supplied by A. F. Gambino. SDO data were supplied courtesy of the SDO/AIA consortia. SDO is the first mission to be launched for NASA's Living With a Star Program. Hinode is a Japanese mission developed and launched by ISAS/JAXA, with NAOJ as domestic partner and NASA and STFC (UK) as international partners. It is operated by these agencies in co-operation with ESA and the NSC (Norway). APPENDIX A Metrology data and on-orbit observations are used to model the point spread function of the X-Ray Telescope's (XRT; (Golub et al. 2007)) mirror assuming that XRT is operated at the best on-axis focus. The metrology data estimate encircled energy profiles for two energies, 0.56 keV and 1.0 keV. We develop a PSF for both energies and find the function that returns the encircled energy data as a piecewise continuous function composed of a Lorentzian core and a series of power-law functions as its wings. The PSFs we develop do not consider other sources of scattering such as the effects of changing the focus position, other elements within the optical system, filters, and the CCD camera system, or material contamination on the XRT CCD. We do not incorporate any non-axisymmetric structures although the system PSF is known to vary (see Fig. 4 in Golub et al. (2007)). The XRT mirror is a Wolter Type-I grazing incident optic built by Goodrich. The XRT has 9 broadband filters that sample plasma temperatures from 0.5 -- 10 million Kelvin and is equipped with a 2048x2048 CCD. The XRT has 1.02860 arcsec pixels with a wide field of view of 34x34 arcmin (Kano et al. 2008). The mirror manufacturer provided encircled energy estimates based on the Power Spectral Density (PSD) derived from measurements of the mirror surface roughness. We construct two PSFs for the XRT mirror using a semi-empirical approach. We model the core of the PSF by considering a variety of functions that could reproduce the on-axis encircled energy data. We then estimate the wings of the PSF based on scattering patterns observed in XRT data. The wings of the PSF are modeled assuming a piecewise continuous power law of the form: Pi(r) = r−αi , where αi ≥ 0, and αi+1 ≥ αi, where r is the radial distance from the optical axis. We assume the PSF is spatially invariant and only depends on the radial distance from the scattering source. Scattered light from XRT data are used to determine the breakpoints of the function so that the following criteria are met: 1. The metrology data affirms that 81% of the encircled energy lies within 5 arcsec for a 0.56 keV source and 77% for a 1.00 keV source, meeting design specifications. -- 23 -- 2. The wings of the PSF match and extend the slope of the encircled energy in a smooth and continuous way. 3. In the case when the PSF will be normalized, we assume that 100% of the light will be scattered within the XRT field of view. To gauge how light is scattered far from the source we use full frame images of (a) limb flare data in which a bright flare occurs on the solar limb when the Sun is centered in the field of view, and (b) solar eclipse data when the Moon passes between XRT and the Sun. Because of Hinode's orbit, XRT experiences either a partial or total solar eclipse twice a year and XRT takes full disk data in several filters. We use the Moon as a way to measure scattered light when it partially obscures an active region on the Sun. Scattered light from these bright regions is easily visible across the Moon's shadow. In addition to the scattered light from the mirror, there are at least two other causes of scattered light. First is the scattering due to the entrance apertures that is accentuated during a solar flare, and the second is a pattern of scattered light that is pointing dependent and is always present. The left panel of Fig.16 shows an example of solar eclipse data used in the analysis and demonstrates both patterns of scattered light within the Moon's shadow. The image is scaled on a log scale with Fig. 16. -- Left: Hinode XRT C-Poly (log intensity) eclipse image taken 2008 February 7 at 5:17:53 UT. White arrows point to the shadow of the entrance filters and the gray arrow points to the azimuthal scattered light pattern. Right: Plot of the mean relative normalized intensity as a function of distance from the intersection of the two lines within the region of interest (ROI). Within the ROI, the scattered light follows an inverse square law up through the gray arrow when the intensity increases again. This fit does not distinguish between the mirror scatter and the azimuthal scatter. pixel values above a few DN s−1 saturated to white so that the low-level scatter can easily be seen. Dark bands that appear to emanate from the scattering source are the pattern created from the -- 24 -- entrance filters (white arrows). A gray arrow points to the second scattering pattern. It is a partial dark ring followed by a region of bright light. They appear as partial bands around the scattering source. This pattern is pointing dependent and changes location depending on pointing and the location of the scattering source within the field of view. All XRT data are processed using the standard reduction routines provided by the XRT team in SolarSoft. We use full resolution images. The exposure times of the data vary between 0.5 -- 16 seconds depending on the solar conditions. To estimate the amount of scatter in an XRT image, the average normalized intensity along an arc as a function of radial distance from the scattering source is fit to the general power law function of the form: P (r) = α(r − b)−c + d, where a, b, c, and d are all free parameters. The intensity is normalized to the maximum value set by the data reduction routine, xrt−prep.pro. This fitting method is not able to distinguish between the different sources of scatter. To mitigate the effects of the scattering due to the entrance filters, we select a region between dark bands of scattered light. An example of a region is the one between the two lines on the image on the left of Fig.16. We attempt to deal with the second source of scattered light by considering regions closer to the scattering source rather than farther away. The encircled energy data imply the wings of the PSF do not significantly contribute to the en- circled energy far from the center. At a radial distance of 4 -- 5 arcsec there is little increase in the encircled energies. Therefore, we expect that far from the source, the other scattering elements will dominate the scattering. We use Mathematica to calculate the encircled energy curves for both PSF models, corresponding to the two energy channels, over a spatial grid that is appropriate for the meteorology data and that oversamples the instrument plate scale. Fig.17 shows the encircled energy measurements (squares) for 0.56 keV (a) and 1.0 keV (b). For each of the datasets, we find that a single function could not reproduce the given encircled energies. The two energies have the same functional form but have different parameter values and breakpoints; the relevant values are given in Table 2. We find the inner portion of the PSF is best represented by a Lorentzian function out to an inner radius, r0. From r0 to 5 arcsec, the r−1 function returns the correct encircled energy measurements. We then use the assumption that the PSF will continue to follow the power law trend and fit the following model: P (r) = 2 r σ2 ) exp(− γ 2 + r2 , a br−1, cr−2, dr−4,   r ≤ r0; r0 ≤ r ≤ r1; r1 ≤ r ≤ r2; r2 ≤ r ≤ ∞. -- 25 -- Fig. 17. -- Encircled energy plot of the manufacturer data (squares), the model PSF (dashed line) and the model discretized for XRT pixel size (solid line) for 0.56 keV, (a) and 1.0 keV (b). Table 2: Normalized PSF parameter values. * Denotes exact values and not approximations. Parameter Values 0.56 keV 2.19256 1.24891 1.31946 0.03* 0.15* 1.0 keV 2.36982 0.914686 0.847955 0.038* 0.19* σ γ a b c d r0 r1 r2 18.4815* r0 20.1571* r0 3.4167 5* 11.1* 3.22857 5* 10.3* A plot of the encircled energy (squares) for both channels is given in Figure 2 along with the model PSF (black lines). The models fit the data well. We also discretized the models to the XRT pixel size (solid line). A simple calculation is applied to consider the relative applicability of the two PSF models for a -- 26 -- range of typical plasma temperatures in the corona. We make use of the Astrophysical Plasma Emission Code (APEC, Smith et al. (2001)) to model the plasma emission as a function of wave- length and temperature. We fold this model through the XRT's spectral response and convert the instrument spectral response to a temperature response for each of XRT's filters. We create a spectral response of several plasma temperatures and compare the amount of energy at or below 0.75 keV to the amount of energy above 0.75 keV for a given temperature plasma. Table 3 shows the relative spectral response for each of the XRT's filters. Table 3 shows that, for plasma temperatures above 1MK, a significant portion of the signal will Table 3: Relative spectral response for each of the XRT's filters assuming the specified temperature. 1MK 1MK 3MK 3MK 5MK 5MK 10MK 10MK Filter ≤ 750ev > 750ev ≤ 750ev > 750ev ≤ 750ev > 750ev ≤ 750ev > 750ev Al-mesh Al-poly C-poly Ti-poly Be-thin Be-med Al-med Al-thick Be-thick 0.07 0.05 0.03 0.03 0.00 0.00 0.00 0.00 0.00 0.77 0.83 0.87 0.87 0.98 1.00 1.00 1.00 1.00 0.96 0.97 0.99 0.98 1.00 1.00 1.00 1.00 1.00 0.02 0.05 0.05 0.04 0.38 0.92 0.96 1.00 1.00 0.98 0.95 0.95 0.96 0.62 0.08 0.04 0.00 0.00 0.23 0.17 0.13 0.13 0.02 0.00 0.00 0.00 0.00 0.93 0.95 0.97 0.97 1.00 1.00 1.00 1.00 1.00 0.04 0.03 0.01 0.02 0.00 0.00 0.00 0.00 0.00 come from energies greater than 0.75 keV. The PSFs provided above are designed with normalization in mind but this condition is not neces- sary, and in fact it forces that 100% of the energy is scattered within the XRT field of view. With this condition relaxed, the PSF will scatter light far from the field of view. Table 4 provides the PSF models without normalization. The only difference between these and the normalized models is the value of r2. This causes the slope of the encircled energy to essentially remain flat beyond 5 arcsec. APPENDIX B Among different indirect methods of deconvolution available in SolarSoft IDL libraries, we used codes based on the Maximum Likelihood and Richardson-Lucy methods: • AIA DECONVOLVE RICHARDSONLUCY.pro (AIA) based on Richardson-Lucy al- gorithm. /darts.isas.jaxa.jp/pub/ssw/sdo/aia/idl/psf/PRO/aia deconvolve richardsonlucy.pro The Richardson-Lucy algorithm in this code follows closely the algorithm discussed by Jansson (1997). -- 27 -- Table 4: PSF model without Normalisation. * Denotes exact values and not approximations. Parameter Values σ γ r0 a r1 b c D r2 EE at edge of FOV 0.56 keV 1.0 keV 2.36982 0.914686 3.22857 0.847955 2.19256 1.24891 3.4167 1.31946 5* 0.03* 0.15* 5* 0.038* 0.19* 7.35* r0 10.1251* r0 7 93% 7.3 93% • MAX LIKELIHOOD.pro (M-L) based on Maximum likelihood algorithm. idlastro.gsfc.nasa.gov/ftp/pro/image/max likelihood.pro Based on papers by Richardson (1972) and Lucy (1974). REFERENCES Bentley, R. D. & Freeland, S. L. 1998, in ESA Special Publication, Vol. 417, Crossroads for Euro- pean Solar and Heliospheric Physics. Recent Achievements and Future Mission Possibilities, 225 DeForest, C. E., Martens, P. C. H., & Wills-Davey, M. J. 2009, ApJ, 690, 1264 Dennerl, K. 2002, A&A, 394, 1119 Dennerl, K. 2008, Planet. Space Sci., 56, 1414 Freeland, S. & Bentley, R. 2000, SolarSoft, ed. P. Murdin Grigis, P., Yingna, S., & Weber M. tion and Image Deconvolution Version 2012-Feb-13 - part of the SSW manual, http://hesperia.gsfc.nasa.gov/ssw/sdo/aia/idl/psf/DOC/psfreport.pdf for the AIA team. 2012, AIA PSF Characteriza- in Golub, L., Deluca, E., Austin, G., et al. 2007, Sol. Phys., 243, 63 Gonz´alez, A., Delouille, V., & Jacques, L. 2016, Journal of Space Weather and Space Climate, 6, A1 Grunwaldt, H., Neugebauer, M., Hilchenbach, M., et al. 1997, Geophys. Res. Lett., 24, 1163 -- 28 -- Jansson, P. A. 1997, Deconvolution of images and spectra. Kano, R., Sakao, T., Hara, H., et al. 2008, Sol. Phys., 249, 263 Kobelski, A. R., Saar, S. H., Weber, M. A., McKenzie, D. E., & Reeves, K. K. 2014, Sol. Phys., 289, 2781 Lemen, J. R., Title, A. M., Akin, D. J., et al. 2012, Sol. Phys., 275, 17 Lucy, L. B. 1974, AJ, 79, 745 Narukage, N., Sakao, T., Kano, R., et al. 2011, Sol. Phys., 269, 169 Pesnell, W. D., Thompson, B. J., & Chamberlin, P. C. 2012, Sol. Phys., 275, 3 Poduval, B., DeForest, C. E., Schmelz, J. T., & Pathak, S. 2013, ApJ, 765, 144 Reale, F., Gambino, A. F., Micela, G., et al. 2015, Nature Communications, 6, 7563 Richardson, W. H. 1972, Journal of the Optical Society of America (1917-1983), 62, 55 Smith, R. K., Brickhouse, N. S., Liedahl, D. A., & Raymond, J. C. 2001, ApJ, 556, L91 Starck, J. L., Pantin, E., & Murtagh, F. 2002, PASP, 114, 1051 Takeda, A., Yoshimura, K., & Saar, S. H. 2016, Sol. Phys., 291, 317 Weber, M., Deluca, E. E., Golub, L., et al. 2007, PASJ, 59, S853 This preprint was prepared with the AAS LATEX macros v5.2.
1207.3278	1	1207	2012-07-13T15:20:20	Ohmic Dissipation in the Interiors of Hot Jupiters	[ "astro-ph.EP" ]	We present models of ohmic heating in the interiors of hot jupiters in which we decouple the interior and the wind zone by replacing the wind zone with a boundary temperature Tiso and magnetic field Bphi0. Ohmic heating influences the contraction of gas giants in two ways: by direct heating within the convection zone, and by heating outside the convection zone which increases the effective insulation of the interior. We calculate these effects, and show that internal ohmic heating is only able to slow the contraction rate of a cooling gas giant once the planet reaches a critical value of internal entropy. We determine the age of the gas giant when ohmic heating becomes important as a function of mass, Tiso and induced Bphi0. With this survey of parameter space complete, we then adopt the wind zone scalings of Menou (2012) and calculate the expected evolution of gas giants with different levels of irradiation. We find that,with this prescription of magnetic drag, it is difficult to inflate massive planets or those with strong irradiation using ohmic heating, meaning that we are unable to account for many of the observed hot jupiter radii. This is in contrast to previous evolutionary models that assumed that a constant fraction of the irradiation is transformed into ohmic power.	astro-ph.EP	astro-ph	DRA FT V ER S ION D EC EMB ER 22 , 2013 Preprint typeset using LATEX style emulateapj v. 5/2/11 2 1 0 2 l u J 3 1 ] P E . h p - o r t s a [ 1 v 8 7 2 3 . 7 0 2 1 : v i X r a OHMIC DISSIPATION IN THE INTERIORS OF HOT JUPITERS XU HUANG 1 AND ANDR EW CUMM ING 2 Draft version December 22, 2013 ABSTRACT We present models of ohmic heating in the interiors of hot jupiters in which we decouple the interior and the wind zone by replacing the wind zone with a boundary temperature Tiso and magnetic field Bφ0 . Ohmic heating influences the contraction of gas giants in two ways: by direct heating within the convection zone, and by heating outside the convection zone which increases the effective insulation of the interior. We calculate these effects, and show that internal ohmic heating is only able to slow the contraction rate of a cooling gas giant once the planet reaches a critical value of internal entropy. We determine the age of the gas giant when ohmic heating becomes important as a function of mass, Tiso and induced Bφ0 . With this survey of parameter space complete, we then adopt the wind zone scalings of Menou (2012) and calculate the expected evolution of gas giants with different levels of irradiation. We find that,with this prescription of magnetic drag, it is difficult to inflate massive planets or those with strong irradiation using ohmic heating, meaning that we are unable to account for many of the observed hot jupiter radii. This is in contrast to previous evolutionary models that assumed that a constant fraction of the irradiation is transformed into ohmic power. Subject headings: planets and satellites: magnetic fields — magnetohydrodynamics (MHD) — planets and satellites: atmospheres 1. INTRODUCTION The large radii of many hot jupiters has been a puzzle ever since the discovery of the first transiting planet HD 209458b (Charbonneau et al. (2000); see Baraffe et al. (2010) for a review). Guillot & Showman (2002) pointed out that if a cer- tain amount of the irradiation from the star (∼ 1% of the inci- dent stellar flux) can be deposited deep in the envelope of the planet (pressures of (cid:38) 10 bars), then the inflated radius can be explained. But the physical mechanism by which the required energy is transported into the interior of planet is still an open question. Several explanations have been proposed, such as a downward kinetic flux due to atmosphere circulation, or turbulent transport and shock heating in the flow (Guillot & Showman 2002; Youdin et al. 2010; Perna et al. 2012), or tidal dissipation (Bodenheimer et al. 2001, 2003). But each of these has problems in accounting for all of the observed radii (e.g. Laughlin et al. 2011; Demory et al. 2011). Batygin & Stevenson (2010) suggested that ohmic heating could serve as the heat source in the interior of inflated hot jupiters. The idea is that the shearing of the planetary mag- netic field by the wind driven in the outer layers of the planet by irradiation generates an induced current that flows inwards, dissipating energy by ohmic dissipation in deeper layers (see also Liu et al. 2008). Perna et al. (2010a,b) also pointed out the possible importance of magnetic drag on the dynamics of the flow in the envelope, and found that a significant amount of energy could be dissipated by ohmic heating at depths that could influence the radius evolution of the planet. Batygin et al. (2011) implemented ohmic heating in evo- lutionary models of gas giants and showed that the amount of inflation depends significantly on the amount of irradiation received by the planet (and therefore its equilibrium tempera- ture Teq ). They found that the radii of low mass planets could 1 Department of Astrophysical Sciences, 4 Ivy Lane, Pey- ton Hall, 08544, USA; Princeton, NJ Princeton University, email:[email protected] 2 Department of Physics, McGill University, 3600 rue University, Montreal QC, H3A 2T8, Canada; email:[email protected] run away, increasing dramatically in response to ohmic heat- ing and leading to evaporation of the planet. This same be- havior was not observed in the recent study of Wu & Lithwick (2012), who find that ohmic heating could increase the radius of a planet that had already cooled, but only modestly. On the other hand, if ohmic heating operates early in the lifetime of a hot jupiter, Wu & Lithwick (2012) find that contraction can be halted and large radii obtained, large enough to explain the observed radii of all except a few planets (see their Fig. 4). Both the time evolution models of Batygin et al. (2011) and Wu & Lithwick (2012) calculate the profile of ohmic heating as J 2/σ per unit volume inside the planet (where J is the cur- rent density and σ the electrical conductivity), but adjust the overall level of heating so that the efficiency — the frac- tion of the irradiation that goes into ohmic heating — is fixed at a level of ∼ 1%. In reality, the magnitude of the cur- rent that penetrates into the interior depends on how the flow in the wind zone interacts with the planet’s magnetic field and the feedback from the magnetic field on the flow dynam- ics. Menou (2012) considers scaling arguments for the atmo- spheric flows in a magnetized atmosphere, and argues that the efficiency must decline at large Teq as magnetic drag lim- its the flow velocity in the atmosphere (see also Perna et al. 2010b; Rauscher et al. 2012). In this paper, we take a more general approach to the ques- tion of inflation due to ohmic heating, with the aim of un- derstanding the different evolutions found by Batygin et al. (2011) and Wu & Lithwick (2012), and incorporating the ef- fect of magnetic drag, and therefore variable efficiency, on the evolution. We take a different approach by separately con- sidering the planet interior and the wind zone. We first cal- culate the interior heating by replacing the wind zone with a boundary condition which specifies the toroidal field, or equivalently the radial current, at the base of the wind zone and the temperature there. This allows us to survey the pa- rameters that influence the amount of ohmic heating and its effect on the evolution of the planet. In this way we go beyond the previous assumptions of constant heating efficiency. We 2 then implement the scaling laws derived by Menou (2012), and show that indeed the efficacy of ohmic heating is reduced at high Teq because of increased drag in the wind zone. In this way our time dependent models differ crucially from Batygin et al. (2011) and Wu & Lithwick (2012) in that we find that it is difficult to explain the observed radii of many hot jupiters with ohmic heating under the influence of magnetic drag, par- ticularly those with large masses M (cid:38) MJ . The plan of the paper is as follows. In §2, we review the general mechanism of ohmic heating and how the internal cur- rent is calculated, giving some order of magnitude estimates for the total ohmic power. Next in §3, we present quasi- steady state models of gas giants undergoing ohmic heating as a function of their internal entropy S and the induced mag- netic field Bφ0 and temperature Tiso at the base of the wind zone. We then use these models to follow the time evolu- tion of a given planet under the action of ohmic heating in §4. With this general survey of parameter space in hand, we then use the scalings derived by Menou (2012) for the wind zone to calculate the evolution of observed hot jupiters and com- pare with observed radii (§5). We discuss the limitations of our models and compare to other work in §6. 2. THE GENERAL MECHANISM OF OHMIC HEATING In this section, we review the basic physics of ohmic heat- ing, focussing on the generation of the induced field in the wind zone and radial current that penetrates into the planet interior (§2.1). We then estimate the likely magnetic field strengths that can be generated in the wind zone and the re- sulting ohmic power available for inflation (§2.2). (1) 2.1. Calculation of the induced field and current distribution First we review the general picture put forward by Baty- gin & Stevenson (2010) (See also (Perna et al. 2010a)). Due to strong irradiation from the host star, the hot jupiter has a thermally-ionized atmosphere, coupling the magnetic field and the atmospheric flow. The magnetic field could be either the intrinsic planetary magnetic field generated by a dynamo in the deep interior, or the external field from the host star. In either case, the evolution of the magnetic field in the wind zone is governed by the induction equation = −∇ × η (∇ × (cid:126)B) + ∇ × ((cid:126)v × (cid:126)B) ∂ (cid:126)B ∂ t where (cid:126)v is the wind velocity and η is the magnetic diffusiv- ity of the atmosphere. Assuming a steady state, and that the planet magnetic field in the outer layers is well-represented by a curl-free dipole field (cid:126)Bdip , the induced magnetic field (cid:126)b is given by ∇ × η (∇ × (cid:126)b) = ∇ × ((cid:126)v × (cid:126)Bdip ). φ, the in- If the wind is predominately a zonal flow (cid:126)v = vφ duced field is toroidal, and will penetrate into the interior of the planet, with associated poloidal currents that close in the interior given by (cid:126)J = (c/4π )∇ × (cid:126)b. The internal toroidal field is given by ∇ × (η∇ × (cid:126)b) = 0 (3) below the wind zone where velocities are negligible. For a given poloidal current J , the local ohmic dissipation rate is Pohm = J 2/σ , where σ is the electrical conductivity, related to the magnetic diffusivity by η = c2/4πσ . (2) F IG . 1 .— The general mechanism of ohmic heating illustrated in a plane- parallel model. A vertical field Bz is sheared by the wind vy in the wind zone. An induced field b is produced by the shear that penetrates into the interior. We use the temperature Tiso and induced magnetic field Bφ0 at the base of the wind zone as boundary conditions for our interior solutions. = 0, Some intuition about the solution can be obtained by con- sidering a plane parallel model, which we illustrate in Fig- ure 1. There we divide the planet into three layers, repre- senting the outermost isothermal layer, with pressures lower than 60 mbars, the wind zone, between 60 mbars and 10 bars and the interior of the planet. The vertical field Bz is sheared by the wind with velocity vy (z) exp (ikx). Focusing on the in- nermost layer representing the planet interior, and assuming constant conductivity, equation (3) gives an induced field (cid:126)b ∝ exp(−kz + ikx)y and associated vertical current 4π jz /c = ikb. Both the field and current decrease exponentially in the inte- rior on a length scale 2π/k. When the variation of conductiv- ity with depth is included, we must solve d 2b d b d η d z2 − k2b + d z d z in which case the thickness of the penetration depth depends on the length scale on which the conductivity varies. This simple model illustrates that the interior solution de- pends on three factors: the geometry of the shearing veloc- ity in the wind zone (which is described by k in this simple model), the magnitude of the induced field at the base of the wind zone, and the profile of the electrical conductivity in the interior. The approach that we pursue in this paper is to con- sider the first two factors as boundary conditions on the inte- rior. We choose to parametrize our models by specifying the toroidal magnetic field at a pressure of 10 bars, which we will refer to as Bφ0 . To calculate the distribution of currents inside the planet in detail, we consider a simple geometry with a dipole field and a zonal flow (cid:126)v = v0 sin θ φ. To solve equation (3) in spherical coordinates we write the induced toroidal field: g(r) r in which case the radial dependence part of the field is given by sin θ cos θ φ, (cid:126)Bφ = g(cid:48)(cid:48) (r) − d ln σ g(r) r2 = 0 d r where l = 2 is the index of associated Legendre polynomial P1 l . Having found g(r) and therefore Bφ (r), the currents are g(cid:48) (r) − l (l + 1) (4) (5) (6) Bz b, v Upper Atmosphere Wind Zone Interior z y 60mbar 10bar BΦ0 Tiso determined by Ampère’s law (cid:126)J = (c/4π )∇ × (cid:126)B. The ohmic (cid:90) (cid:104)J (cid:105)2 (cid:90) J 2 power in the interior is dV ∼ 4π r2 d r, P = (7) σ σ where (cid:104)J (cid:105) = (cid:104)J 2 θ (cid:105)1/2 is the effective angle averaged cur- r + J 2 rent at radius r. To get some feeling for the dependence of the field and current on depth, we can consider a power-law (α − 1) + (cid:112)(1 + α)2 + 24 dependence σ ∝ rα . In that case, the solution is Bφ ∝ rβ with β = (8) , 2 and (cid:104)J (cid:105) ∝ rβ−1 . For more complex wind geometries, which involve l > 2, the solution is (cid:126)Bφ ∝ r l P1 l (cosθ) for constant conductivity. For example, this indicates that more zonal jets in the wind zone implies a shallower penetration depth for the induced field. In fact, as we argue in the next section (§2.2), the internal heating is dominated by the lowest densities, since the local heating rate is inversely proportional to the electrical conduc- tivity which increases rapidly with increasing pressure. This means that the current J can be taken as a constant without making a significant error in the heating profile. We have confirmed this by comparing constant current solutions with detailed solutions of equation (6). For the constant current case, we compute the current as c Bφ0 J = (9) , RJ 4π where RJ = 7 × 109 cm is the radius of Jupiter, Bφ0 = Bφ (r, p = 10bars). This means that there is a direct mapping between the chosen value of Bφ0 , the radial current inside the planet Jr , and the local heating rate, taken to be J 2 r /σ (r) per unit volume. Note that we do not take into account the averages over angle in equation (7) nor the true radius of the planet, and so our value of Bφ0 for a given amount of internal heating could be a factor of a few of the toroidal field in a model which self-consistently includes both the wind zone and the interior. Instead, our parameter Bφ0 should be interpreted as a measure of the internal heating (given by eqs. [9] and then J 2 r /σ locally). 2.2. Magnitude of the induced field and ohmic power It is useful to estimate the expected magnitude of ohmic heating and how it scales with parameters such as planet mass M . The first step is to use equation (2) to estimate the expected strength of the induced field by dimensional analysis, 4πσH v b = RM = , (cid:19) (cid:16) (cid:17) (cid:18) H Bdip c2 (cid:16) σ v 1 km s−1 107 s−1 0.01 RJ where v is an average wind speed and σ a typical value of electrical conductivity in the layer3 , and we take the vertical length scale to be the pressure scale height H . b = Bdip (cid:17) (10) or , (11) 3 We give the electrical conductivity in cgs units here. Note that the con- version to SI units is 1 S m−1 = 9 × 109 s−1 . 3 . σ 106 s−1 Pm = 10−1 erg g−1 s−1 In this paper, we take Bdip = 10G as a standard value. Typ- ical dipole field strengths for hot jupiters have been esti- mated from scalings with planet parameters (see Trammell et al. 2011 for a detailed summary and discussion). Sánchez- Lavega (2004) argued that the field is generated by the dy- namo action in the metallic region as in Jupiter (Stevenson 1983), with the field strength closely related to the rotation of the planet, B ∼ (ρΩη ) 1 2 . This predicts that the field on typical hot jupiters should be a factor of few smaller than that on the Jupiter, with a typical value of equatorial mag- netic field Beq ∼ 5G. However, Christensen et al. (2009) ar- gue that the field instead scales with the heat flux escaping from the conductive core at large enough rotation rate, giving B ∼ (ρF 2 core ) 1 3 . This gives a field strength an order of magni- tude larger than estimated with the previous method. Given an induced field b, we can then estimate the expected ohmic power per unit mass, Pm = (cid:104)J (cid:105)2/(ρσ ). Approximating the angle averaged current at the base of the wind zone using (cid:18) Bφ0 (cid:19)2 (cid:16) (cid:17)−1 (cid:18) (cid:19)−1 the constant current case as equation (9) gives ρ 10−4 g cm−3 10G (12) Since σ increases dramatically in the core of the planet, heat- ing at low density dominates. If the conductivity profile scales as exp(−r/H ) for example, where we take the lengthscale as the pressure scale height H , the total power in the interior is (cid:19)2 (cid:16) σt (cid:18) Bφ0 (cid:17)−1 × Pohm,total ≈ 4πPtopR2ρH (cid:19) (cid:18) R (cid:18) H (cid:19)2 = 1023 erg s−1 106 s−1 10 G (cid:19)2 (cid:16) σt (cid:18) Bφ0 (cid:17)−1 × 0.01 RJ RJ = 3 × 1022 erg s−1 (cid:19) (cid:18) R (cid:18) T (cid:19)4 (cid:18) M (cid:19)−1 106 s−1 10 G MJ RJ 1500 K where the subscript t indicates a quantity evaluated in the out- ermost regions of the interior just below the wind zone. Note that H = RT /g, where we adopt R = 3.64 × 107 erg g−1 K−1 for a hydrogen molecule dominated composition with helium fraction Y = 0.25. As we mentioned previously, because the total ohmic power is dominated by the heating at low pres- sure, it is not very sensitive to the radial profile of the current. The scaling in equation (13) implies that Pohm ∝ 1/M when the radius and conductivity of planet is not strongly depend on mass, a scaling that we find in our numerical solutions. More massive planets have less ohmic power for a given Bφ0 and temperature Tiso at the base of the wind zone. (13) (14) 3. OHMIC HEATING AS A FUNCTION OF INTERNAL ENTROPY In this section, we calculate the structure, luminosity, and ohmic heating profile for gas giants as a function of their in- ternal entropy S. We will use these models in §4 to follow the time evolution of the planet by following the decreasing entropy as the planet cools. In this approach, described by Hubbard (1977) (see also Fortney & Hubbard 2003 and Arras & Bildsten 2006), it is assumed that the convective turnover 4 T M ¯T dV, (17) L = = −L(S) (16) (15) = −Lc + dm = M ¯T time is much shorter than the evolution time of the planet, so that the convection zone maintains an adiabatic profile as it cools and lowers its entropy. We also assume that the radia- tive envelope has a thermal timescale much shorter than the evolution time, so that the envelope is in thermal steady-state, carrying the luminosity emerging from the convection zone. The luminosity of the planet is then given by the radiative lu- minosity at the radiative–convective boundary, 16πGcMr aT 4 ∇ad 3κ pc where pc is the pressure at the convective boundary, T is the (cid:90) temperature at that location, and Mr is the enclosed mass. The evolution of the internal entropy is then given by d S d S dt dt (cid:82) T /M dm. where we have assumed d S/dt is constant across the con- vection zone and define the mass-averaged temperature ¯T = The effect of ohmic heating appears in two places in this (cid:90) J 2 approach. The first is that an ohmic heating term must be added to the right hand side of equation (16), d S dt σ where the integral is over the convection zone. For a planet with a given entropy S, the luminosity at the top of the con- vection zone is fixed by the structure and is given by equation (15). However, because some of this luminosity is now pro- vided by ohmic heating, the cooling rate of the convection zone (d S/dt ) is smaller. The second influence of ohmic heat- ing is that it can change the temperature profile in the radiative zone, in particular by pushing the radiative–convective bound- ary to higher pressure and lowering the luminosity (eq. [15]). In this section, we include the first effect by calculating gas giant models without feedback from ohmic heating in the atmosphere (§3.1), and then include the feedback from ohmic heating in the radiative zone to include the second ef- fect (§3.2). In §3.3, we summarize the results. 3.1. Planet models without feedback We now make models of gas giants with given mass M and central entropy S, and use them to calculate the ohmic dissi- pation in the planet, but without including the effect of ohmic heating on the planet structure. This allows us to calculate the ohmic heating within the convection zone and, by including this ohmic power in equation (16), the corresponding slowing of the cooling rate. We present the detail microphysics in our planet model in the Appendix A. Here we only note the differences be- tween our opacity and conductivity profiles and those used in other works. Our opacity profile use a different extrap- olation method in the intermediate pressure range between 103 − 105bars from Paxton et al. (2011), who take the opac- ity for log R > 8 (where R = ρ/T 3 is used in the opacity 6 tables) to be a constant set by the value at log R = 8. As far as we are aware, opacity calculations for this interme- diate pressure range have not been carried out. For plan- ets undergoing a large amount of internal heating, the con- vection zone boundary moves into this region, and so know- ing the opacity there is important for understanding the loca- tion of the convection zone boundary. As we describe later, F IG . 2 .— The temperature profile (top panel) and entropy profile (bot- tom panel) for different treatments of the outer layers, either isothermal (red curve), radiative (green curve), or radiative including ohmic heating (blue curve). For the ohmic heating model, we take Bφ0 = 1000 G. In each case, the radiative–convective boundary is marked with a vertical bar. Note that we do not show the entire structure, but focus on the lower pressures to illustrate the differences in the position of the radiative–convective boundary between models. this controls the contraction rate of the cooling planet. Our potassium conductivity profile (equation [A2]) is the same as used by Perna et al. (2010a), but is different from Batygin & Stevenson (2010), who use a electron-neutron cross-section of π (7.2 × 10−9 cm)2 = 1.6 × 10−16 cm2 rather than 10−15 cm2 (Draine et al. 1983) and a slightly different thermal averag- ing factor for the velocity. The resulting difference is that the conductivity of Batygin & Stevenson (2010) is a factor of 9 times larger than our conductivity. The planet model is calculated by integrating outwards from the center, following the convective adiabat until it in- tersects a radiative zone extending inwards from the surface. When calculating the cooling curve of an irradiated gas gi- ant, a reasonable approximation is to take the outer radiative zone of the planet to be isothermal. However, because ohmic heating is very sensitive to pressure, a small error in the deter- mination of the convective boundary results in a much larger error in the ohmic power. To illustrate this, we have calculated models with an isothermal radiative layer and with a radiative layer that is in thermal equilibrium and carries a constant lu- minosity equal to the luminosity from the convection zone. For the isothermal case, we integrate dm d r = 4πρr2 (18) 103104100101102103104105106T (K)Pressure p (bar) 7.5 8 8.5 9 9.5 10 10.5100101102103104105106Entropy S (kB/mp)Pressure p (bar)5 TABLE 1 MODE L SUMMARY Model Lconv (erg s−1 ) Pohm (erg s−1 )( p > 10 bars) Pohm (erg s−1 )( p > pconv ) pconv (bars) Bφ0 (G) R(RJ ) 1.3 × 1026 1.2 × 1025 8.0 × 1023 62.8 10 1.25 Isothermal 1.3 × 1026 1.2 × 1027 8.0 × 1025 62.8 100 1.25 Isothermal 7.7 × 1025 2.8 × 1024 6.8 × 1022 131.7 10 1.25 Radiative 7.7 × 1025 2.8 × 1026 6.8 × 1024 131.7 100 1.25 Radiative 7.7 × 1025 2.8 × 1024 6.7 × 1022 RadiativeFBb 132.0 10 1.25 5.6 × 1025 1.9 × 1026 3.1 × 1024 RadiativeFBb 176.4 100 1.25 a Model computed with S = 8, Tiso = 1500 K, and M = 0.96 MJ . We refer to this set of input parameters our standard model in the text. b Model including the feedback of ohmic heating in the atmosphere. c Both the cooling luminosity and the ohmic heating in the convective zone reduce while the convective zone boundary move towards deeper pressure due to the feedback in up atmosphere. perature profiles for models with an isothermal and non- isothermal radiative zone. In the isothermal case, we choose Tc = 3 × 104 K, pc = 2 × 107 bars and Tiso = 1500 K, which gives a M = 0.96 MJ , R = 1.25 RJ planet with core entropy S = 7.98 and convective zone boundary pconv = 62.76 bars. The luminosity from the interior is L = 1.28 × 1026 erg s−1 , giving tS = 5.78 Gyr. With a radiative zone, we obtain the same mass, radius and entropy with a convective zone bound- ary pconv = 131.7 bars. The luminosity from the interior is L = 7.7 × 1025 erg s−1 , and tS = 10.5 Gyr to cool. In each case, the magnetic field structure in the planet in- terior is obtained by solving equation (6) using the conduc- tivity profile in the planet (shown in Figure 17), and then the ohmic heating profile is determined. For an induced magnetic field Bφ0 = 10 G at the bottom of the wind zone ( p = 10 bars), we find Pohm ( p (cid:62) pconv ) = 8 × 1023 erg s−1 for the isothermal model and Pohm ( p (cid:62) pconv ) = 6.8 × 1022 erg s−1 in the non- isothermal case. While the cooling time for the planet changes by a factor of two between the two models, the ohmic power changes by more than an order of magnitude. Therefore, it is crucial to locate the convective boundary accurately when calculating the ohmic power inside the convection zone. In Figure 3, we compare the ohmic power calculated in this way, which includes the correct radial distribution of current, with the ohmic power calculated by assuming a constant ra- dial current, independent of depth. The agreement is excel- lent (within a factor of 2) except at the highest pressures in the central regions of the planet. 3.2. Planet models with feedback from ohmic heating The fact that the ohmic heating per unit mass rises rapidly to lower densities (Fig. 3) suggests that the heating in the re- gions lying between the wind zone and the convection zone boundary will be larger than the heating in the convective in- terior. We include the ohmic heating in the radiative layer by allowing L to vary throughout the radiative zone, with J 2 dL d r σ We do not include ohmic heating at pressures less than 10 bars. Instead, we specify the temperature Tiso at p = 10 bars and integrate inwards. Of course, there could be significant ohmic heating within the wind zone at p < 10 bars, but we absorb this into the boundary condition. Note that this means that early in the lifetime of the planet, when the entropy is large enough that pconv < 10 bars, the models here revert back to our previous models with no feedback. However, at those (23) = . F IG . 3 .— The cumulative ohmic power against pressure for the same planet parameters as in Figure 17, with B = 10G (radiative model, no feedback). The solid curve uses a current profile calculated by solving equation (6); the red dashed curve assumes a constant current with depth. (19) Gm d p = −ρ r2 d r T ∇ d p dT (20) = d r , P d r outwards from the center, taking ∇ = ∇ad for T > Tiso (adi- abatic interior) and ∇ = 0 for T < Tiso (isothermal layer). In the non-isothermal case, we take ∇ = min (∇rad , ∇ad ) p 3κL ∇rad = aT 4 . 16πcGM We calculate a model for a given M and S by integrating outwards from the center and inwards from the surface to a matching pressure p = 30 kbars. For the outwards inte- gration, we choose the central pressure pc and cooling rate d S/dt (or equivalently cooling time tS = S/ d S/dt ). For the inwards integration, we start at a pressure of 10 bars and set the temperature there to be Tiso . We then integrate inwards, choosing the luminosity L and radius R. A multi-dimensional Newton-Raphson method is used to find the correct choices of ( pc , tS , R, L) that result in m, r, T and L agreeing to within 1% at the matching pressure. As an example, Figure 2 compares the entropy and tem- (22) (21) 6 F IG . 4 .— The luminosity L (solid curves) and the total ohmic heating in the convection zone Pohm (dashed curves) as a function of central entropy S. The red, yellow, green and blue lines (from top to bottom for the solid lines; inverse for the dashed lines) represent planets with different mass: 3.0 MJ , 1.0 MJ , 0.6 MJ , 0.3 MJ . At larger entropy, where ohmic heating is unimpor- tant, L ∝ M at fixed S, whereas Pohm decreases with increasing M . All the planet models are computed with Tiso = 1750 K and Bφ0 = 100 G. F IG . 5 .— The time history of planet luminosity (solid curves) and ohmic heating in the convection zone Pohm (dashed curves) for M = 0.3 MJ (blue), 0.6 MJ (green), 1 MJ (yellow) and 3 MJ (red curves) (same configuration with Figure 4). The luminosity decreases with time because of cooling, while the ohmic heating either increases or decreases slowly depending on M . At late times, when ohmic heating in the radiative layer becomes important, Pohm decreases because the convective boundary moves inwards. When Pohm be- comes comparable to L, the cooling and contraction of the planet is halted. All the models are calculated with Tiso = 1750 K and Bφ 0 = 100 G. early times, ohmic heating is generally not yet important. Also note that in the models without feedback, the strength of the induced field Bφ does not influence the internal struc- ture of the planet, whereas here a larger Bφ results in more heating in the radiative layer which can push the convective boundary deeper. In Table 1, we compare the models with feedback to our earlier models without feedback. The internal structures are shown in Figure 2. The planet radius does not vary much be- tween different models. The biggest difference is in the posi- tion of convective zone boundaries (marked by black vertical F IG . 6 .— Position of the radiative–convective boundary as a function of internal entropy S for a 1 MJ planet with Tiso = 1750 K. Blue, green, yellow, red, black lines are for Bφ0 =0, 10, 30, 100, 300, 1000 G. At a given entropy, a larger Bφ0 results in more ohmic heating in the radiative zone, moving the convective boundary to a higher pressure. bars in Fig. 2), which results in a difference in the cooling lu- minosity and the ohmic heating both in the convective zone and atmosphere. Note that this means that the cooling history of a planet using these three approaches would be different, especially the time at which ohmic heating begins to become important for evolution. We use this feedback model for all the calculation carried on below. 3.3. Ohmic power as a function of entropy The luminosity and ohmic power is shown as a function of entropy in Figure 4. As noted in particular by Arras & Bildsten (2006), equation (15) shows that L ∝ M at fixed en- tropy, and we see that scaling in Figure 4. On the other hand, the ohmic power decreases with increasing M , as discussed in §2.2. We find that the decrease in Pohm is well described by Pohm ∝ R2.4/M , which has a shallower dependence on R than in equation (13) because of the dependence of the con- ductivity term on mass which compensates the R4 term. In our feedback model, for a lower mass planet the higher atmo- spheric ohmic heating pushes the convective zone boundary slightly deeper, resulting in a higher conductivity at the top of the convection zone. Overall, lower mass planets generally have higher ohmic power deposited in the convection zone. Combined with the mass-luminosity dependence, the de- crease of ohmic power with mass means that ohmic heating becomes important for lower mass planets at a much higher entropy than for more massive planets. The value of entropy at which ohmic heating becomes important depends on the boundary induced field Bφ0 . Turning this around, for an ob- served planet with measured radius and mass, we can infer the entropy and therefore derive a limit on the wind zone Bφ0 required for ohmic heating to be providing a significant part of the luminosity in that object. We carry out this procedure in §5, but first describe our calculations of the time-evolution of planets with ohmic heating. 4. TIME DEPENDENT EVOLUTION OF PLANET STRUCTURE 7 F IG . 7 .— The age of the planet when ohmic heating becomes important (Pohm > 0.1L) versus planet mass, for Bφ0 = 30 G (black), 100 G (red), 300 G (green), 1000 G (blue curves), and for two temperatures Tiso = 1750 K (solid curves) and 2250 K (dashed curves, corresponding to the lower panel of la- bels). F IG . 8 .— The value of Bφ0 required for ohmic heating to become important at an age of 3 Gyr. From bottom to top, M =0.3, 1, 3.0 MJ . Having computed the radiative– the luminosity at convective boundary for a large grid of models with differ- ent M , S, Tiso and Bφ0 , the evolution in time of a planet with fixed mass M , Bφ0 and Tiso then involves stepping in entropy using equation (17). When calculating the ohmic power, we assume constant current J with depth, which as shown earlier is a good approximation. We have checked that for Bφ0 = 0, our cooling models compare well with the earlier results of Burrows et al. (1997) and Baraffe et al. (2003) (see Marleau et al. 2012). We reproduce their cooling curves to within 30% in luminosity, and predict radii that are 0.05-0.08 RJ larger than in those cooling sequences. By integrating in time, we compute the time history of the planet luminosity and ohmic power, shown in Figure 5. As the planet cools, the convection zone ohmic power increases or decreases slightly depending on mass, but always changes more slowly than L, so that ohmic power eventually becomes comparable to the cooling luminosity. At the same time, as F IG . 9 .— Time history of planet radius for (top to bottom) M =0.3, 0.6, 1 and 3 MJ . All the models are calculated with Tiso = 1750 K and Bφ0 = 100 G. the ohmic heating in the upper atmosphere (which is about an order of magnitude larger than convection zone heating) starts to affect the planet structure, the convective zone boundary shrinks inwards. The resulting decrease of both cooling lu- minosity and convection zone ohmic heating result in a rapid increase in cooling time, so that we can view the evolution afterwards as a quasi-steady state. For higher atmospheric ohmic heating, this effect happens at higher entropy, thus the steady radius of planet will be larger. We compare the evo- lutionary tracks of the radiative/convection zone boundary of a 1 MJ planet with different strengths of induced field in Fig- ure 6. As we increase the amount of ohmic heating, the con- vection zone boundary deviates from the no heating path at a higher entropy. In Figure 7, we show the age of a planet with a particular mass when Pohm,c = 0.1Lconv , at which point ohmic heating starts to become significant and the planet contraction slows. In general, the atmospheric heating is an order of magnitude larger, and so comparable to the cooling luminosity at this age. In Figure 7, we report that the result is sensitive to the strength of Bφ0 . With Tiso = 1750 K, for Bφ0 equals 30 G, a 0.3 MJ hot jupiter can reach steady state in 1 Gyr, while a 1 MJ requires 100 G to reach steady state at a similar age. Higher Tiso (dashed line in Figure 7) does not help the planet reach a steady-state radius faster, but on the contrary, it requires a larger induced field to achieve the same result. Figure 8 shows the Bφ0 required to halt contraction within 3 Gyr as a function of Tiso . We see that a hotter planet re- quires a stronger induced field to obtain a significant level of ohmic heating. This is because the interior ohmic heating is closely related to the conductivity at the bottom of the wind zone; the conductivity increases with temperature, reducing the ohmic heating at fixed induced field. But we should also point out that for a hotter planet there is a higher chance to obtain a stronger induced field due to the stronger wind in the atmosphere. So this result does not necessarily imply that it is more difficult to make ohmic heating important in hotter planets. We plot the time evolution of the planet radius for plan- ets with Bφ0 = 100 G and Tiso = 1750 K and different planet masses in Figure 9. In the absence of stellar ages, we shall 8 take a typical age of 3 Gyr, and take the radius at 3 Gyr as the present day radius. For the lowest mass planet, 0.3MJ , the effect on the evolution of the planet radius is significant, and the planet stops cooling around 1 Gyr (with cooling time longer than 10 Gyr) and thereafter maintains a large radius. However, the heating is not as effective at higher masses. For example, HD 209458b has a observed radius of 1.35 RJ with mass 0.7 MJ , while we can only obtain a radius of 1.25 RJ for Bφ0 = 100 G. This is because the power we introduced into the planet interior is far smaller than the received stellar luminosity. In the case of our standard model, the irradiation luminosity from the host star is 1029 erg s−1 , and the heating in the interior is only one 0.01% of it, 1026 erg s−1 . To go further, we must understand what values of Bφ0 might be expected as a function of Teq , and we turn to this in the next section. 5. EVOLUTION INCLUDING WIND ZONE MODEL AND COMPARISON TO OBSERVATIONS In §4, we calculated the time-evolution of cooling gas gi- ants assuming that Tiso and Bφ0 are independent parameters. In reality, they are coupled by the dynamics in the wind zone, since the atmospheric flow, in response to the irradiation, de- termines both the magnetic field in the layer and the tem- perature at depth (the values of Tiso and Bφ0 are specified at p = 10 bars). In this section, we implement the scalings for the wind zone dynamics proposed by Menou (2012) (§5.1) and then compare our results to observed systems (§5.2). − . + 0 = − 5.1. Dynamics of the wind zone and the relation between Tiso and Bφ0 Both Batygin et al. (2011) and Menou (2012) write down simplified models for the wind zone dynamics including the effects of magnetic drag. In both cases, following Perna et al. (2010a) the magnetic drag force is assumed to be (cid:126)J × (cid:126)B/c per unit volume, with the current (cid:126)J set by a balance between the shearing of the magnetic field by the fluid and ohmic diffusion of magnetic field lines against the fluid motion, (cid:126)J = σ(cid:126)v × (cid:126)B/c (§2). However, the dynamical balance in the two models is quite different. Menou (2012) writes the force balance for the equatorial flow as (see also Showman et al. 2010) R∆Thoriz∆ ln p v2 vφB2 r φ RP RP 4πρη The first two terms represent a balance between the advective term and the horizontal driving from the day-night tempera- ture difference ∆Thoriz . This balance is thermal wind-like in that the horizontal pressure gradients require a vertical gradi- ent in the fluid velocity vφ over a vertical pressure scale ∆ ln p. The final term represents the magnetic drag force, again inte- grated over a vertical scale ∆ ln p. Batygin et al. (2011) on the other hand consider the meridional circulation induced by magnetic drag on the azimuthal flow, so that for example the latitudinal force balance is f vy = vφ/τL where f = 2Ω sin θ is the Coriolis parameter and τL the magnetic drag timescale. Their solution represents a thermal wind balance involving the equator-pole temperature gradient, modified by magnetic drag. In both cases, magnetic drag limits the fluid velocity at high temperatures, where the large degree of ionization and therefore large electrical conductivity results in strong cou- pling of the fluid and magnetic field. Balancing the second and third terms in equation (24) gives vφ ∝ η when magnetic (24) drag dominates, and therefore the magnetic Reynolds number RM = vφH /η becomes almost constant, varying only slowly with temperature. Similarly, equation (16) of Batygin et al. (2011) has two possible limits, either vφ ∝ η when the lat- eral temperature gradient is large, in which case RM becomes almost constant at large Tiso , or vφ ∝ η 2 when the drag time scale is comparable to the rotation period, while the lateral gradient of temperature is still small, in which case RM ∝ η declines rapidly at large Tiso . A dynamical model including both day-night driving and meridional circulation with magnetic drag is not yet available. For our purposes, we have implemented the model of Menou (2012) as described by equation (24), with the day-night tem- (cid:40) Tday perature difference given by 2 ( τadv τrad Tday 2 τadv < τrad τadv > τrad ∆Thoriz = (25) ) In equation (25), Tday is the dayside averaged temperature considering a dilution factor of 0.5, T 4 irr = 2T 4 day = 4T 4 eq , and the advective and radiative timescales are RP vφ τadv = (26) τrad = Cp p gσSBT 3 day . (27) Note that these timescales are evaluated at the outermost pres- sure, which following Menou (2012) is taken to be 60 mbars, the estimated location of the thermal photosphere. This is the reason for adopting the thermal timescale appropriate for an optically thin region, so that τrad ∝ p in equation (27); in deeper, optically thick layers, τrad has an extra factor of the optical depth τ , leading to τrad ∝ p2 (e.g. Fig. 3 of Show- man et al. 2008). The magnetic drag term is also evaluated at p = 60 mbars; this term is integrated over height, but since σ decreases with increasing pressure (for an isothermal layer), the dominant contribution to the integral is from the lower limit on pressure, and so η and ρ are evaluated there. This is an important difference from Batygin et al. (2011), who evalu- ated their magnetic drag timescale at p = 10 bars, which gives a drag time an order of magnitude longer than we find here. Based on that estimate, Batygin et al. (2011) concluded that the drag timescale was always much longer than a rotation period. We solve equation (24) for vφ as a function of Teq , and find the corresponding value of the induced field Bφ from equation (10) for different values of the dipole field Bdip . For ∆ ln p = 0.9, we reproduce the results of Menou (2012) (see his Fig. 1), but we also consider larger values of ∆ ln p. Menou (2012) models the weather layer with a modest vertical extension around 1 pressure scale height. We also solve the equation with ∆ ln p = 3 for typical values in hot jupiter atmosphere as reported by Showman et al. (2010), and ∆ ln p = 5 for a wind zone extending to p ∼ 10 bars, for comparison with Batygin & Stevenson (2010) and Batygin et al. (2011). The effect of varying ∆ ln p on Bφ is shown in the left panel of Figure 10. For numerical convenience, we fit the Bφ –Teq relation with the following: 1 Bφ (Teq ) = 1 Badv + 1 Bdrag , (28) 9 and T 4 = f , (31) (29) where F IG . 10 .— The induced field Bφ (left panel) and timescales (right panel) in the wind zone as a function of Teq . In the left panel, the solid, dashed and dotted curves are for wind zone thickness ∆ ln p = 0.9, 3 and 5, and we take Br = 10 G. In the right panel, we show the advection, radiative and drag timescales τadv , τrad and τdrag . The advective timescale is shown for ∆ ln p = 0.9 (solid), 3 (dashed) and 5 (dotted curves) (the radiative and drag timescales are independent of ∆ ln p). (cid:18) (cid:19) − 2.53 × 104 Badv = 2.8 × 106 G Teq exp (cid:19) (cid:19)1/2 (cid:18) Br (cid:18) ∆ ln p Teq 3 10 G (cid:19)−1 (cid:19) (cid:18) Br (cid:19) (cid:18) ∆ ln p (cid:18) Teq Bdrag = 1125 G (30) . 10 G 1000 K 3 This reproduces Bφ to within ≈ 10% for Teq in the range 1100 to 2200 K. The transition to the regime where RM is approximately constant occurs when τdrag exceeds τadv , where the magnetic r = η/v2 drag timescale is τdrag = 4πρη/B2 A , where vA is the Alfven speed, and again the timescale is evaluated at the top of the wind zone ( p = 60 mbars here). These timescales are plotted as a function of Teq in Figure 10. Changing the wind zone thickness from ∆ ln p = 0.9 to ∆ ln p = 5 moves the tran- sition temperature from Teq ≈ 1400 K to 1700 K. To use this value of Bφ as a boundary condition for our evo- lutionary models, we must relate the temperature Teq at low pressure to the temperature Tiso at p = 10 bars. This relation depends on the details of energy transport in the wind zone, including the effects of ohmic heating and needs to be studied further. Here, we adopt the atmospheric temperature profile from Guillot (2010), and keep in mind the uncertainty in the (cid:21) (cid:20) 2 relation between Teq and Tiso when interpreting our results be- low. The relation from Guillot (2010) is (see his eq. [29]) 3T 4 + 1 √ irr 4 3 3 γ where γ is the ratio between visible and infra-red opacities, and f = 1/2 for a dayside average or f = 1/4 for an aver- age over the whole surface. Choosing γ = 0.4, as appropriate for a planet like HD 409658b (e.g. see Fig. 1 of Hubeny et al. 2003) and a dayside average f = 0.5, we obtain Tiso = 0.94Tirr = 1.33Teq . In the following section, we will use this relation to infer the appropriate value of Tiso from the Teq of observed planets. We note here that we don’t have a good knowledge of what the γ parameter would be for most of the observed planets. While γ parameter could vary in a very large numeri- cal range, the ratio between Tiso and Teq only changes within a 5.2. Comparison with Observed Hot Jupiters In Figures 11 to 13, we compare our results with the ob- served properties of transiting planets taken from the TEPcat transiting planet catalog4 , which gives the planet mass, radius, and equilibrium temperature Teq = T(cid:63),eff (R(cid:63)/2a)1/2 where T(cid:63),eff is the stellar effective temperature. As the ages of most stars are unknown or highly uncertain, we assume an age of 3 Gyr when comparing with the observed planets. First, Figure 11 shows the effect of increasing Bφ0 at fixed Tiso on the planet radius. To help compare with the data, we divide the observed sample into two groups with either Teq > 1500 K (green points) or < 1500 K (red points) and show theoretical curves for either Teq = 1316 K or 1692 K (these two temperatures correspond to Tiso = 1750 and 2250 K respectively). We see that for the low Teq group, an induced field of 10–100 G can explain most of the observed radii, 4 http://www.astro.keele.ac.uk/∼jkt/tepcat/ F IG . 11 .— The predicted mass–radius relation at 3 Gyr for Bφ0 =0, 30, 100 and 1000 G and Teq = 1316 K (solid curves) and 1692 K (dashed curves). The data points show observed transiting planets, divided into two temperature groups T > 1500 K (green points) and T < 1500 K (red points). factor of few [0.99(γ → inf) < (Tiso /Teq ) < 3.05(γ = 0.01)]. Since the observed properties of planets gives Teq thus the boundary induced field, changing the value Tiso/Teq is equiva- lent to shift inside the plane Tiso − Bφ given by Figure 8. Gen- erally, a smaller Tiso/Teq is favored to inflate the planet with the same Bφ . 10-210-1100101102103104 1000 1200 1400 1600 1800 2000 2200B(cid:113)(G)Teq (K)103104105106107 1000 1200 1400 1600 1800 2000 2200(cid:111)adv,(cid:111)rad,(cid:111)drag (s)Teq(K)(cid:111)rad(cid:111)drag(cid:111)adv 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.1 1 10R/RJM/MJ10 F IG . 12 .— Comparison with observations using Bφ0 (Teq ) from the wind zone model. Top panel: Radius at 3 Gyr against Teq for M = 0.3 (red), 0.6 (green), 1.0 (blue) and 3.0 MJ (black). The data points are observed planets divided by mass: 0.2 MJ < M < 0.5 MJ (red points), 0.5 MJ < M < 0.9 MJ (green points), 0.9 MJ < M < 1.3 MJ (blue points). Bottom panel: Predicted mass–radius relation at 3 Gyr for Teq = 1316 K and T = 1682 K (bottom to top). In each case, the dashed curve shows the radius without ohmic heating; the solid curve with ohmic heating. The data has been divided by tempera- ture: Teq < 1500 K (red points), T > 1500 K (green points). while the high Teq planets need at least Bφ0 = 1000 G to match the observed radii. It is clear that a higher induced magnetic field is needed to explain a given radius at higher equilibrium temperature. Next, we use the wind zone model described in §5.1 to cal- culate Bφ0 as a function of Teq , assuming canonical values Br = 10 G and ∆ ln p = 3. In the top panel of Figure 12, we show the radius as a function of Teq , with the colors represent- ing three different bins in planet mass. There exists a clear correlation between the radius and Teq , both in the observa- tions and the models. In addition, we see that the amount of inflation is also strongly dependent on the planet mass. Plan- ets within the mass bin 0.3–0.6 MJ agree quite well with our ohmic heating model. However, ohmic heating clearly can- not explain planets with mass ∼ 1 MJ and large inflated radii (cid:38) 1.4 RJ . Ohmic heating can help to increase the radius (for comparison the dashed line shows models with no ohmic heat- ing), but not enough to match the observed value. This is a consequence of the increased power needed to maintain the radius of a massive planet at a particular value, as well as the reduced ohmic heating power at larger masses. In the lower panel of Figure 12, we show the radius as a function of mass. As in Figure 11, we divide the data into two temperature ranges and show the model results for two representative temperatures, now using the wind zone model to specify the value of Bφ0 for each temperature. The param- eter region where ohmic heating has the largest effect is high temperature, low mass planets. The radii of the low tempera- ture group (Teq < 1500 K) can almost all be explained without ohmic heating. For the high temperature group, ohmic heat- ing can explain the observed radii of low mass planets, but most of the radii of the high temperature group lie well above the models, especially at large planet masses (cid:38) 1 MJ . In Figure 13, we show the ratio between observed and pre- dicted radii Robs /Rpred against Teq (upper panel) and against M (lower panel) for each observed planet. In this case, we use the observed values of M and Teq to calculate the evolu- tion of the planet, and, in the absence of stellar ages, we take Rpred to be the radius at 3 Gyr. In the upper panel, we see that for Teq (cid:38) 1600 K, there are many planets whose radii lie above the predicted values. The slow increase of Bφ0 with Tiso at large Tiso due to the magnetic drag term results in a much weaker dependence of Rpred on Teq than observed, and most outliers lie at the highest temperatures. In the lower panel, we see that the majority of the unexplained objects (Robs > Rpred ) are at larger masses M (cid:38) 0.7 MJ . We note that the choice of estimating the planet radius at 3 Gyr is not critical for the above picture. Constraining ourself within the time range of 1 − 5 Gyr, the predicted radius only varies within several per- cent. To look in more detail at the effect of our assumed wind zone model on how successfully we are able to reproduce the observed radii, in Figure 14 we show the results in the Bφ0– Teq parameter space. For each observed planet, we first make a model with no ohmic heating, varying the internal entropy S at the measured M and Teq until we match the measured radius R p . Then we calculate the value of Bφ0 required in that model for the ohmic power in the convection zone Pohm to be 30% of the planet’s luminosity. We colour-code the data points according to whether they are successfully explained by our time evolutions, ie. whether they have Rpred larger or smaller than Robs in Figure 13. These two groups of data points lie on either side of the Bφ0–Teq relation from the wind zone model (solid curve). This shows that the approach of using a struc- tural model with no ohmic heating (we refer to this as a “no feedback” model in §3) to estimate the critical magnetic field is a good approximation of our detailed time-evolution mod- els including feedback. Comparing the red points in Figure 14 with the solid curve gives a sense of how far short the ohmic heating model falls in explaining the most inflated planets. For example, HAT-P-32 is about a factor of 3–4 above the curve, so that the heating rate (∝ B2 ) needs to be increased by about an order of magni- tude to explain the observed radius. It is interesting that most of the unexplained objects lie within a factor of 3 in terms of Bφ0 of the wind zone model. Figure 14 helps to show what changes to the wind zone model would explain more of the observed objects. We have assumed the relation Tiso = 1.33Teq (from eq. [31]); a larger factor between Tiso and Teq would move the solid curve to the left, allowing ohmic heating to explain the radii of low Teq planets such as WASP-06. The dashed and dotted curves show the effect of changing Br . In- creasing Br from 10 to 100 G does increase Bφ0 at low temper- atures, but reduces Bφ0 at high temperatures where magnetic drag is enhanced. A larger depth ∆ ln P would help to reduce 0.6 0.8 1 1.2 1.4 1.6 1.8 2 1200 1400 1600 1800 2000 2200R/RJTeq(K) 0.6 0.8 1 1.2 1.4 1.6 1.8 2 0.1 1 10R/RJM/MJ11 wick (2012)), the feedback of ohmic heating in the region between the wind zone and the convective boundary moves the convective zone boundary deeper (Fig. 6), leading to a re- duced cooling luminosity and reduced internal ohmic heating. Because the electrical conductivity changes dramatically with pressure through the planet, the total ohmic power inside the convection zone is very sensitive to its radial extent. To com- puting the planet age and radius at the late stage when ohmic heating halted the cooling, it is crucial to accurately locate the convective-radiative boundary. 3. A larger Bφ0 is required for ohmic heating to be im- portant in more massive planets or planets with larger Teq . This can be seen in Figures 7 and 8, which show the age of a cooling gas giant when ohmic heating becomes important, and the magnetic field strength required for ohmic heating to be important at different values of Tiso . For example, at a temperature Tiso = 1750 K, Figure 8 shows that Bφ0 ≈ 30 G will halt the contraction of a 0.3 MJ planet in 3 Gyr, whereas Bφ0 ≈ 150 G is required for a 1MJ planet at that temperature. At higher Tiso = 2250 K, the required values are Bφ0 ≈ 100 G for a 0.3MJ planet or Bφ0 ≈ 700 G for a 1MJ planet. 4. With a specific model for the wind zone (§5.1), we can compare to observed systems as a function of their observed equilibrium temperatures Teq . The wind zone model specifies the induced field Bφ0 (or equivalently, the radial current that penetrates into the interior; see eq. [9]) as a function of Teq , and the relation between Teq and the temperature at 10 bars. Using the scaling analysis proposed by Menou (2012) for the dynamics of the wind zone, together with the atmospheric temperature profile from Guillot (2010), we find that it is diffi- cult for ohmic heating to explain the large radii of hot jupiters with large masses and large Teq (see Fig. [13]). 5. A more general approach is to calculate, for each ob- served planet, the Bφ0 that is required if ohmic heating is pro- viding a significant fraction of the luminosity (and therefore able to significantly change the contraction rate of the planet). This is shown in Figure 14 and shows how much the heating rate needs to be increased over the wind zone model in §5.1 to explain particular objects. A modest increase in the wind zone thickness over that assumed here, or larger ratio of the temperature at depth Tiso compared to Teq , would improve the agreement with observed radii (see discussion in §5.2). Even so, several objects require a much more dramatic increase in heating rate (see Fig. 14). The difficulty in explaining many of the observed radii that we have found differs from Batygin et al. (2011) and Wu & Lithwick (2012) who found that they could account for almost all of the observed hot jupiter radii with ohmic heating. The key difference is that we do not assume here that the heat- ing efficiency (the fraction of the irradiation going into ohmic power, typically taken to be ∼ 1%) to be fixed, but instead use the wind zone model to set the induced magnetic field in the wind zone and therefore the magnitude of the heating. It is important to emphasize that our conclusions about the efficacy of ohmic heating depend on the particular prescrip- tion for the magnetic field in the wind zone that we have used. In fact, many complexities underlie the path from the irradi- ation to the properties of the induced magnetic field. More realistic 3D wind zone models may give a different picture than the simple 1D force balance scalings we have used here. For example, in this paper we have assumed the wind zone ex- tends to p = 10 bars. Figure 3 of Wu & Lithwick (2012) nicely illustrates the importance of the depth of the wind zone, show- ing that a shallower wind zone requires a significantly larger F IG . 13 .— The ratio of observed planet radius Robs and predicted radius Rpred (3 Gyr) for observed hot jupiters as a function of Teq (upper panel) or M (lower panel). In the upper panel, the size of the circle scales with planet mass; in the lower panel, the size of the circle scales with Teq . the number of discrepant objects since both Badv and Bdrag in- crease with ∆ ln P (eqs. [29] and [30]). 6. SUMMARY AND DISCUSSION In this paper, we present models of ohmic heating in hot jupiters in which we attempt to decouple the interior and wind zone by replacing the wind zone by a boundary temperature Tiso and magnetic field Bφ0 , both evaluated at a pressure p = 10 bars. This approach allows us to survey the outcomes of ohmic heating, parametrized by Tiso and Bφ0 for planets with different mass M . This is similar in spirit to models of gas giant cooling, which often set an outer boundary pressure of 10 bars and separately integrate a T10 –Teff relation to use as an outer boundary condition. The main conclusions of the paper are: 1. Figure 4 is a key result, showing as a function of en- tropy how the ohmic power compares to the planet luminosity. Only planets with entropy below a critical value have enough ohmic heating to slow their contraction rate. Of particular note are the different mass dependences: at fixed Bφ0 and Tiso , the cooling luminosity L ∝ M whereas the ohmic power de- creases with mass (we find Pohm ∝ R2.4/M ). 2. Ohmic heating has two effects on the thermal state of the planet. As well as providing direct heat input into the adiabatic convective interior (as found by previous works, see Batygin & Stevenson (2010); Perna et al. (2010b); Wu & Lith- 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 800 1000 1200 1400 1600 1800 2000 2200Robs/RpredTeq(K) 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 0.1 1 10Robs/RpredM(MJ)12 F IG . 14 .— For each observed planet, we show the Bφ0 required for ohmic heating in the convection zone to be 30% of the luminosity as estimated from no-feedback planet models, and compare with the results of our time-dependent calculations with feedback included. The curves show the Bφ0 –Teq predicted by the wind zone model for three different values of Br . Black points show planets whose radii can be explained by our model (Rpred > Robs in Fig. 13), red points show planets that cannot be explained (Rpred < Robs in Fig. 13). For clarity, we use the following abbreviations for planet names: W–WASP; H–HAT-P; K–Kepler; OG–OGLE-TR; C–CoRoT-P. overall efficiency to achieve the same interior heating. One situation in which this will break down is for young plan- ets with high entropies when the radiative/convective zone boundary is at lower pressure. More work is needed on what happens when the interior convection zone extends into the wind zone region. Our results emphasize the key inputs that are necessary from atmospheric models: the thermal structure and dynam- ics of the wind zone including a large scale magnetic field, the values of induced magnetic field, or equivalently the mag- netic Reynolds number RM , that can be attained there, and the depth of the wind zone. More studies of the local conductiv- ity profile and magnetic field properties in the high magnetic Reynolds number regime are needed. In particular, it is not clear whether the large values of induced field Bφ0 > 1000 G needed to explain the observed radii (Fig. 17) can be achieved in the wind zone. Furthermore, whether the implied large in- ternal currents affect the planetary dynamo is also an open question. Our results do compare favorably with previous calcula- tions if we use equation (11) to set a value of Bφ0 appropriate for the wind zone conditions assumed in those papers. For ex- ample, we are able to compute the 3% heating profile at pres- sures p > 10 bars in Figure 4 of Batygin & Stevenson (2010) by setting Bφ0 = 300 G; we reproduce the heating profile of Tres-4b from Wu & Lithwick (2012) with Bφ0 ≈ 1500 G. However, a complication in comparing different models is that the heating dissipated in the wind zone is coupled with the heating dissipated in the planet interior. Wu & Lithwick (2012) in particular discuss the expected ratio of heating de- posited in different layers. But this ratio is generally model dependent and varies through the planet lifetime. A direct re- sult of this kind of coupling is that models with same heating efficiency but different wind zone model are not physically comparable. For example, for a given set of planet properties, the radius predicted by Batygin & Stevenson (2010) is larger than in Wu & Lithwick (2012) for the same choice of heat- ing efficiency , because the heating ratio between the wind zone and the interior is much smaller in Batygin & Stevenson (2010), creating a much stronger internal heat source. Simi- larly, although Menou (2012) estimated the total ohmic heat- ing efficiency to be > 1% over a certain range of equilibrium temperatures (with the weather layer between 60 mbars and 150 mbars), the internal heating has a much lower efficiency, consistent with our findings in §5. Another uncertainty is in the microphysics aspects of the electrical conductivity. For example, as we noted in §3.1, Batygin & Stevenson (2010) make a different choice for the electron-neutral cross-section and thermal averaging that re- sults in a factor of 9 difference in electrical conductivity than we adopt here. The estimates in §2.2 show that the amount of ohmic power is sensitive to changes in the electrical con- ductivity (or the ionization fraction) in two ways. At low den- sities in the wind zone, the conductivity determines the size of the current (eq. [10]); in the interior, the ohmic power is ∝ 1/σ (eq. [13]). For a fixed efficiency , a different nor- malization for σ does not change the evolution of the planet, since the normalization of the heating profile is determined 13 F IG . 15 .— The opacity profile in a planet with parameters as in Table 1 tablenote a (no ohmic heating, radiative model), showing the opacity as calculated by combining the Freedman et al. (2008) and Potekhin & Chabrier (2010) tables (solid curve) or from the MESA code (Paxton et al. 2011) (red dashed curve). F IG . 16 .— The contribution to the electron fraction Ye from different alkali metals as a function of temperature and pressure. Solid curves are for potas- sium, dotted for sodium, and dashed for aluminum. In each case, we show (top to bottom) pressures of 1, 100 and 1000 bars. by the choice of , and σ (r) determines only its shape. The normalization of σ is important, however, when going beyond the constant efficiency assumption, making it crucial to under- stand the processes that set the ionization level in hot jupiter atmospheres. This work began as a project at the 2011 International Sum- mer Institute in Modeling in Astrophysics (ISIMA), held at the Kavli Institute of Astronomy and Astrophysics, Beijing, China. We thank ISIMA for support and KIAA for hospital- ity during the program. We would like to thank E. Chiang, D.N.C. Lin, A. P. Showman, Y. Wu and Y. Lithwick for use- ful discussions during 2011 ISIMA. We are also grateful for the helpful suggestions during private communication from T. Guillot and R. Laine, and to G.-D. Marleau for discussions about gas giant models and a thorough reading of the paper. F IG . 17 .— Top panel: the electrical conductivity profile of a planet with parameters as in Table 1 tablenote a (no ohmic heating, radiative model), showing the contributions from alkali metals (dashed green curve) and hy- drogen (dashed blue curve). Bottom panel: the electron fraction Ye as a func- tion of pressure, with the contribution from alkali metal ionization shown as a dashed curve. AC is supported by an NSERC Discovery Grant. APPENDIX M ICROPHYS ICS OF THE PLANET INTER IOR We discuss the microphysics input in our gas giant mod- els here. We adopt the equation of state from Saumon et al. (1995) with helium fraction Y = 0.25. In order to maximize the planet radius, we do not include a solid core or elements heavier than helium. The radiative opacity is taken from Freedman et al. (2008) and in the core we include thermal conduction by electrons from Potekhin & Chabrier (2010). The transition from radia- tive to conducive opacity occurs at a pressure which is greater than the maximum pressure of 300 bars covered by the Freed- man et al. (2008) tables. In the intermediate regime, we as- sume the scaling κ ∝ p0.5 . The opacity profile for our stan- dard model is shown in Figure 15, over-plotted with opacity taken from MESA using the same planet structure (Paxton et al. 2011). The electrical conductivity has contributions from alkali metal ionization in the outer layers, and hydrogen in the inte- rior. In the upper atmosphere of hot jupiters, the conductivity is set by the ionization of alkali metals. For potassium, which has the lowest ionization potential 5 . The potassium only Saha 5 The first ionization potentials of K, Na, Al, Mg and Fe are 4.34, 5.14, 5.99, 7.65 and 7.90 eV respectively (David 2003). 0.001 0.01 0.1 1 10 10010-210-1100101102103104105106107108Opacity (cid:103) (cm2g-1)Pressure p (bar)10-610-410-210010210410610-210-1100101102103104105106107(cid:109) (Sm-1)Pressure p (bar)10-1210-1010-810-610-410-210010-210-1100101102103104105106107YePressure p (bar)14 xk = fk n (A1) e−25.19/T3 = 1.03 × 10−3 T 5/4 3 equation (Balbus & Hawley 2000; Perna et al. 2010a) gives (cid:35) 1 (cid:34) (cid:19)3/2 (cid:18) me kBT the ionization fraction xk = ne/n as 2 e−4.35eV/kB T (cid:19)1/2 (cid:16) p (cid:18) fK (cid:17)−1/2 2π¯h2 10−7 1bar where fK is the number fraction of potassium. Although potassium dominates, we also include the contribution of Na, Mg, and Fe in the ionization balance to sum up the total ion- ization fraction. The ionization fraction of each alkali metal is computed separately by assuming a balance independent on the presents of other elements . We do not include the con- tribution of Al in the calculation, which is likely condensed out (Lodders 1999). But our results are not very sensitive to elements beyond potassium. This is illustrated in Figure 16 which shows the contributions to the ionization level from K, Na and Al at different pressures. Once the ionization fraction is determined, the conductivity is σ = ne e2/me ν where the col- lision frequency of electron-neutral collisions is νen = nn (cid:104)σv(cid:105)e (cid:18) 128kBT (cid:19)1/2 given by Draine et al. (1983) as (cid:104)σv(cid:105)e = 10−15 cm3 s−1 . 9πme (cid:19)−1/2 (cid:17) (cid:18) T σ = 8.8 × 10−2 S m−1 (cid:16) x The conductivity is then 10−7 1500 K . (A3) , (A2) In the deeper part of the planet, the hydrogen is ion- ized by high pressure and the conductivity is dominated by electron-proton collisions. In the fully-degenerate limit, νepd = 4e4meΛ/3π¯h2 = 1.8 × 1016 s−1 . In the non-degenerate limit, νepnd = 6.4 × 1023 s−1ρYeT −3/2 , in which Ye is the electron frac- tion. We interpolate between the two limits to give an esti- mation of the total contribution: ν −2 epd + ν −2 ep = ν −2 epd . We also include the conductivity at intermediate densities as given by Liu et al. (2006). Before the hydrogen molecule is fully ion- ized, the band-gap of hydrogen will diminish with increasing pressure, to a level where there is a significant contribution to the conductivity. Liu et al. (2006) give this as (cid:18) −Eg (ρ) (cid:19) kBT σs = σ0 exp (A4) where between 0.2 Mbars and 1.8 Mbars, Eg = 20.3 − 64.7ρ, where Eg is in eV, and ρ is in mol cm−3 , and σ0 = 3.4 × 1020 s−1 exp(−44ρ). The overall conductivity is constructed as σ = σs + ne e2/me ν = σs + 1.52 × 1032ρYe /ν . The collisional frequency ν is the sum of electron-neutral and electron-proton collisions. A typical conductivity profile and the contribution of different components are shown in Figure 17. REFERENCES Arras, P., & Bildsten, L. 2006, ApJ, 650, 394 Balbus, S. A., & Hawley, J. F. 2000, Space Sci. Rev., 92, 39 Baraffe, I., Chabrier, G., Barman, T. S., Allard, F., & Hauschildt, P. H. 2003, A&A, 402, 701 Baraffe, I., Chabrier, G., & Barman, T. 2010, Reports on Progress in Physics, 73, 016901 Batygin, K., & Stevenson, D. J. 2010, ApJ, 714, L238 Batygin, K., Stevenson, D. J., & Bodenheimer, P. H. 2011, ApJ, 738, 1 Bodenheimer, P., Lin, D. N. C., & Mardling, R. A. 2001, ApJ, 548, 466 Bodenheimer, P., Laughlin, G., & Lin, D. N. C. 2003, ApJ, 592, 555 Burrows, A., Marley, M., Hubbard, W. B., et al. 1997, ApJ, 491, 856 Charbonneau, D., Brown, T. M., Latham, D. W., & Mayor, M. 2000, ApJ, 529, L45 Christensen, U. R., Holzwarth, V., & Reiners, A. 2009, Nature, 457, 167 David R. Lide (ed), CRC Handbook of Chemistry and Physics, 84th Edition. CRC Press. Boca Raton, Florida, 2003 Demory, B.-O., & Seager, S. 2011, ApJ, 197, 12 Draine, B. T., Roberge, W. G., & Dalgarno, A. 1983, ApJ, 264, 485 Fortney, J. J., & Hubbard, W. B. 2003, Icarus, 164, 228 Freedman, R. S., Marley, M. S., & Lodders, K. 2008, ApJS, 174, 504 Guillot, T., & Showman, A. P. 2002, A&A, 385, 156 Guillot, T. 2010, A&A, 520, A27 Hubbard, W. B. 1977, Icarus, 30, 305 Hubeny, I., Burrows, A., & Sudarsky, D. 2003, ApJ, 594, 1011 Laughlin, G., Crismani, M., & Adams, F. C. 2011, ApJ, 729, L7 Liu, J., Goldreich, P. M., & Stevenson, D. J. 2006, Bulletin of the American Astronomical Society, 38, 483 Liu, J., Goldreich, P., & Stevenson, D. J. 2008, Icarus, 196, 653 Lodders, K. 1999, ApJ, 519, 793 Marleau, G.-D., & Cumming, A. 2012, in preparation Menou, K. 2012, ApJ, 745, 138 Paxton, B., Bildsten, L., Dotter, A., Herwig, F., Lesaffre, P., & Timmes, F. 2011, ApJS, 192, 3 Perna, R., Menou, K., & Rauscher, E. 2010, ApJ, 719, 1421 Perna, R., Menou, K., & Rauscher, E. 2010, ApJ, 724, 313 Perna, R., Heng, K., & Pont, F. 2012, arXiv:1201.5391 Potekhin, A. Y. and Chabrier, G. (2010), Thermodynamic Functions of Dense Plasmas: Analytic Approximations for Astrophysical Applications. Contributions to Plasma Physics, 50: 82Ð87. doi: 10.1002/ctpp.201010017 Sánchez-Lavega, A. 2004, ApJ, 609, L87 Rauscher, E., & Menou, K. 2012, ApJ, 745, 78 Saumon, D., Chabrier, G., & van Horn, H. M. 1995, ApJS, 99, 713 Showman, A. P., Cooper, C. S., Fortney, J. J., & Marley, M. S. 2008, ApJ, 682, 559 Showman, A. P., Cho, J. Y.-K., & Menou, K. 2010, Exoplanets, 471 Stevenson, D. J. 1983, Reports on Progress in Physics, 46, 555 Trammell, G. B., Arras, P., & Li, Z.-Y. 2011, ApJ, 728, 152 Youdin, A. N., & Mitchell, J. L. 2010, ApJ, 721, 1113 Wu, Y., & Lithwick, Y. 2012, arXiv:1202.0026
1311.2942	1	1311	2013-11-12T21:00:11	Planet Packing in Circumbinary Systems	[ "astro-ph.EP", "astro-ph.SR" ]	The recent discovery of planets orbiting main sequence binaries will provide crucial constraints for theories of binary and planet formation. The formation pathway for these planets is complicated by uncertainties in the formation mechanism of the host stars. In this paper, we compare the dynamical states of single and binary star planetary systems. Specifically, we pose two questions: (1) What does it mean for a circumbinary system to be dynamically packed? (2) How many systems are required to differentiate between a population of packed or sparse planets? We determine when circumbinary systems become dynamically unstable as a function of the separation between the host-stars and the inner planet, and the first and second planets. We show that these represent unique stability constraints compared to single-star systems. We find that although the existing Kepler data is insufficient to distinguish between a population of packed or sparse circumbinary systems, a more thorough study of circumbinary TTVs combined with an order of magnitude increase in the number of systems may prove conclusive. Future space missions such as TESS provide the best opportunity for increasing the sample size.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, ??–?? (2012) Printed 10 July 2018 (MN LATEX style file v2.2) Planet Packing in Circumbinary Systems Kaitlin M. Kratter1,2, Andrew Shannon3 1 JILA / NIST, CU Boulder, 440 UCB, Boulder, CO, 80309, USA 2 Hubble Fellow 3 Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge, CB3 0HA, UK 10 July 2018 ABSTRACT The recent discovery of planets orbiting main sequence binaries will provide crucial con- straints for theories of binary and planet formation. The formation pathway for these planets is complicated by uncertainties in the formation mechanism of the host stars. In this paper, we compare the dynamical states of single and binary star planetary systems. Specifically, we pose two questions: (1) What does it mean for a circumbinary system to be dynamically packed? (2) How many systems are required to differentiate between a population of packed or sparse planets? We determine when circumbinary systems become dynamically unstable as a function of the separation between the host-stars and the inner planet, and the first and sec- ond planets. We show that these represent unique stability constraints compared to single-star systems. We find that although the existing Kepler data is insufficient to distinguish between a population of packed or sparse circumbinary systems, a more thorough study of circumbinary TTVs combined with an order of magnitude increase in the number of systems may prove conclusive. Future space missions such as TESS provide the best opportunity for increasing the sample size. Key words: planets and satellites: dynamical evolution and stability, planet-star interactions, stars: binaries: general 1 INTRODUCTION The Kepler space satellite has produced a vast new data set of dy- namically interesting exoplanetary systems. In this paper, we are concerned with one of the more complex dynamical configurations observed: multi-planet, circumbinary systems. Currently only one multiplanet circumbinary system has been discovered, Kepler 47 (Orosz et al. 2012). The Kepler 47 system consists of two stars with a 3:1 mass ratio, in a ∼ 7 day orbit, and two planets with pe- riods of ∼ 50 and ∼ 300 days. It seem increasingly likely that the so-called orphan transit identified in Orosz et al. (2012) is indeed a tertiary planet with a period of roughly 186 days (Orosz et al., in prep 2013). There are an additional 5 announced single planet circumbinary systems (Doyle et al. 2011; Welsh et al. 2012; Orosz et al. 2012; Schwamb et al. 2013). Although the high precision era of the Kepler mission has ended, other circumbinary planets may still be lurking in the data (Welsh et al. 2013, W. Welsh, private communication). Of particular interest is the use of circumbinary systems as a testbed for theories of planet formation. Studies of extrasolar planets orbiting single stars show evidence that many of these systems are dynamically packed–that the planets are sufficiently closely spaced that no additional planets would be stable between them (Barnes & Greenberg 2007; Migaszewski et al. 2012; Fang & Margot 2013). Moreover, the eccentricity distribution of plan- ets from radial velocity survey is consistent with expectations for c(cid:13) 2012 RAS systems born overpacked, and reduced to their current configura- tions through scattering events (Juri´c & Tremaine 2008). Studies of the solar system also show that it is dynamically packed (Laskar 1996). Some have even suggested that additional planets were lost due to overpacking (e.g., Chambers & Wetherill (2001); Cham- bers (2007); Ford & Chiang (2007); Yeh & Chang (2009)). These lines of evidence point to a scenario where planet formation occurs around single stars with such efficacy that the systems are saturated with planets. This is the "Packed Planetary Systems" hypothesis. This saturation must be explained by any planet formation theory. Relative to the single star case, forming circumbinary plan- ets presents additional challenges. Processes such as secular per- turbations from the binary, gas drag from an eccentric gas disk, and stochastic gravitational perturbations from the protoplanetary disk can all raise the relative velocities between planetesimals, prevent- ing planetesimal accretion in the inner region of the disk (Moriwaki & Nakagawa 2004; Meschiari 2012; Marzari et al. 2013). Although the exact extent of the region in which planet formation is sup- pressed remains an open question (Rafikov 2013), studies show the region to be a few to several AU in size, covering the locations of the known circumbinary planets. Alternatively, if planet formation occurs predominantly in deadzones, circumbinary disks may have an advantage over circumstellar disks due to the reduction in ac- cretion by the action of the binary torque on the inner disk edge (Martin et al. 2013). Depending on the ease with which planets are formed, and the range of allowed radii, these hurdles may be signif- 2 Kratter & Shannon icant enough that circumbinary planetary systems are dynamically sparse, rather than dynamically packed. Either outcome would rep- resent a significant observational constraint on models of planet formation. 2 DEFINING A PACKED CIRCUMBINARY SYSTEM To compare with single-star systems, we must first determine what constitutes a packed circumbinary system. To understand how planet formation proceeds around binaries, we begin by comparing the population of planets around binaries to that of single stars. Computing the frequency of circumbinary planet occurrence is insufficient as we do not have evidence that circumbinary protoplanetary disks are nearly so universal as cir- cumstellar ones. Recent work shows that the overall disk fraction for binaries is lower (Harris et al. 2012), though this work does not address the properties of circumbinary disks specifically. Ad- ditionally, the mechanism by which the host stellar systems form remains uncertain (Tohline 2002). One of the most commonly in- voked formation mechanisms for close binaries is ruled out in these cases: Kozai circularization via Tidal Friction (KCTF) (Fabrycky & Tremaine 2007). KCTF requires that a distant third body drive the binary to high eccentricity, where the stars interact tidally, lead- ing to orbit circularization on Gyr timescales. Not only does the binary achieve a close orbit on timescales long compared to proto- stellar disk lifetimes, it also would have been on an orbit plunging through the planet forming region on the way to its current loca- tion. Thus the existence of these systems places strong constraints on the processes by which the host stars form themselves. In the absence of information on the natal disks, an alternate approach to understanding the efficacy with which circumbinary disks make planets is to examine their packing properties in com- parison to those of single star systems. By considering only systems with at least one planet, we are assured of only considering systems which did possess protoplanetary disks, and systems which have not lost their planets to binary evolution processes such as KCTF. In this work we entertain the ansatz that the ease with which disk material is converted into planets scales with the packing of the system. The only other reference point for circumbinary multi-planet systems resides in our own solar system. The Pluto-Charon planet- satellite system consists of a binary orbited by 4 low mass satel- lites. The dynamical stability of this system was recently studied by Youdin et al. (2012). They found that circumbinary multi-planet orbital stability cannot be approximated by the constraints on either single-star multi-planet systems, or single-planet circumbinary sys- tems. In this paper we first constrain circumbinary, multi-planet sta- bility by exploring the minimum stable intra-planet separation as a function of star-inner planet separation. These two limits, investi- gated by Holman & Wiegert (1999) and Gladman (1993) respec- tively, are interdependent in the circumbinary, multi-planet case. Close inner planets can only remain stable with more widely spaced secondaries. In Section 2, we conduct a numerical parameter study of two-planet circumbinary systems. In Section 3 we examine the specific case of Kepler 47, and ask whether the system is dynam- ically packed with the three (likely) planets. Finally, in Section 4, we make a simple estimate for the detection frequency of secondary (or tertiary) planets in circumbinary systems, in order to estimate how many detections are required to reliably distinguish between a packed and sparse population. We discuss the implications of our findings, and make suggestions for future observational endeavors in Section 5. 2.1 Previous Constraints We extend the work of three well known dynamical studies: Hol- man & Wiegert (1999), Gladman (1993), and Chambers et al. (1996). Holman & Wiegert (1999) systematically explored the min- imum semi-major axis for stable orbits around either a single star in a binary (s-type orbits) or about both components of a binary (p-type orbits). They found that the critical semi-major axis ratio between the planet and stars varied with stellar mass ratio and ec- centricity. For p-type orbits, with which we are concerned in this work, they found a minimum stable semi-major axis around equal mass, circular binaries of acrit ≡ ap/a∗ = 2.3, where ap is the planetary semi-major axis and a∗ the binary semi-major axis. Gladman (1993) showed analytically that there is minimum separation between two planets orbiting a single star, where Hill stability is guaranteed; Hill Stability guarantees that there will be no close encounters between the two bodies. The analysis assumes a large mass ratio between the star and planets. The critical intra- planet separation can be measured in terms of the planetary Hill radius: (cid:19)1/3 (cid:18) mp 3M∗ RHill = ap (1) Gladman (1993) finds that the minimum stable separation, in units where M∗ = ap = G = 1, is ∆c = 2.4∗(µ1 +µ2)(1/3), where µ1 and µ2 are the planetary masses. This corresponds to a minimum spacing of ∼ 3.5RHill, for equal mass planets. It is convenient to introduce the concept of a mutual Hill ra- dius, as this scale incorporates both planetary masses in estimating the strength of planet-planet perturbations. The mutual Hill radius is defined as1: (cid:18) mp1 + mp2 (cid:19)1/3(cid:16) a1 + a2 (cid:17) (2) RHill,m = 3M∗ 2 The work of Chambers et al. (1996) considers single-star sys- tems with 3 or more planets. While an analytic solution is not pos- sible, numerical work has shown that the average system lifetime increases with increasing spacing. Stability over Gyr timescales re- quires spacings greater than ≈ 8 − 10RHill,m (Chambers et al. 1996). In addition, the timescale on which a system spaced by a fixed number of RHill,m becomes unstable decreases with increas- ing number of planets up to 5, and then plateaus (Chambers et al. 1996; Smith & Lissauer 2009). As shown by Youdin et al. (2012), none of the above limits accurately describe the stability of multi- planet circumbinary systems. 2.2 N-body integrations We conduct a series of N-body integrations to constrain the mini- mum separation allowed between two planets in orbit about a bi- 1 The mutual Hill radius is a somewhat ill-defined metric for arbitrarily large planet separation. Because of the dependence on both the inner and outer planet semi-major axis, for a given system there is a maximum number of mutual Hill radii by which a system can be separated because the outer planet separation increases faster than the number of Hill radii in between the planets. c(cid:13) 2012 RAS, MNRAS 000, ??–?? nary, as a function of the distance of the innermost planet from the binary. We use the publicly available n-body code, Swifter to carry out all of our calculations (Levison & Duncan 2013). We use the Gauss-Radau 15th order integrator as this is relatively efficient, while making no assumptions regarding the system architecture. See the appendix of Youdin et al. (2012) for a discussion of the numerical issues associated with this integrator. We vary two parameters: ain, the distance of the innermost planet from the stellar barycenter in units of the stellar semi-major axis, and β, the spacing between the planets in units of mutual Hill Radii. We explore 2.5 < ain < 3.5 and 3.2 < β < 9. The step size in a is 0.25, and for β as small as 0.052. For each ain and β, we run 100 different realizations of the planetary system with random phases relative to each other and to the binary. For simplicity we use circular, Keplerian, co-planar orbits. Note that such orbits are not precisely circular as the forced eccentricity of the binary can be O(10−2) (see e.g., Mudryk & Wu 2006; Youdin et al. 2012). We consider equal mass stars, and a planet- primary star mass ratio of µ = 10−3, 10−4, 10−5. The minimum ain is just outside of the test particle stability limit of Holman & Wiegert (1999), while the minimum β is just inside the minimum separation for two planets to be Hill stable about a single star. All of the observed systems have inner planets outside of our minimum ap,1/acrit. We focus on the results for the two lower mass ratio cases; as discussed below, the Jupiter mass planets have nearly identical limits to the single-planet / single-star case. Each of the systems is integrated for over 2 × 108 binary or- bits. A system is deemed unstable if either of the planets crosses the stellar orbit, or is ejected from the system. We use the median lifetime of all realizations to compute the instability timescale. 2.3 Numerical Results As seen in Figures 1 and 2, our results are in line with expecta- tions qualitatively: when the inner planet separation is very small, the planets are easily destabilized when β is small. When ain is large, the second planet can remain stable near the critical β = 3.5 of Gladman (1993). The spacing required for stability increases steeply with decreasing ain, although the two planets remain some- what more stable than three planets around a single star at fixed intra-planet separation (for the mass ratios considered here). In some ways this last trend is not surprising: the inner planet is typically 20 − 30Rhill from the barycenter of the stellar orbit. To the extent that the secondary star acts like a massive planet, the system is very well spaced. The most naively surprising trend is that in order to recover the two planet, single-star limit, lower mass planets must be moved further away from the binary than their higher mass counterparts; Jupiter mass planets only 20% outside the single planet critical stability radius are stable when spaced by 3.5RHill,m. When the mass ratio drops to µ = 10−4, the critical in- ner planet separation increases to ∼ 60% of the single planet limit. Going down another order of magnitude, Earth mass planets must be placed roughly 75% further out from the binary to recover the single-star, two planet packing limit. This increase is explained by the larger physical separation of the second planet from the binary: because the Hill radius is larger for more massive planets, the outer planet is 20% further out for Jupiter mass planets than for Earth 2 We have chosen to selectively plot values of β to simplify the figures. For the lowest mass case, we increased the β spacing once monotonic trends became apparent. c(cid:13) 2012 RAS, MNRAS 000, ??–?? Planet Packing in Circumbinary Systems 3 mass ones. Since the quadrupole moment from the binary falls off like r−3, it has decreased by 50% at the location of the second Jupiter mass planet compared to the second Earth mass planet. In order to compare with the previous 3-body results, we fit our data to the following power law formula based on Quillen (2011) for instability generated by three-body resonances: ap,1, AU log(tin) = k1 + k2βµ1/12 + k3log(µ) + k4 (3) where tin is in years, and we have added an additional term to in- clude the inner planet semi-major axis. Our best fit is: log(tin) = −7.75 + 3.05βµ1/12 − 0.28log(µ) + 1.88 , (4) ap,1 AU For comparison, the fit to the Chambers results gives k1 = −9.11, k2 = 4.39, k3 = −1.07 (where there is no k4). In Figures 1 and 2, we show two-dimensional fits for t as a function of β and ap,1 only to demonstrate the applicability of power law scalings. The increasing scatter with mass is consistent with the sin- gle star, 3-planet results (Chambers et al. 1996; Smith & Lissauer 2009). Clearly for the typical masses of circumbinary planets, it is difficult to accurately predict the stability of a given system based on a simple logarithmic scaling. A more in depth exploration of the stability of two-planet sys- tems around a wider variety of binary configurations will be the subject of future work. The precise stability of any given system architecture may not be captured by ain, and β alone. The same is true for single star systems. For example, Kepler 11 (Lissauer et al. 2009) has pairs of planets which would violate the three-planet sta- bility criteria, but they are sufficiently distant from other planets in the system that they remain stable. To understand whether any given system is truly packed requires individual simulations. Nev- ertheless, we can see that for the masses with which we are con- cerned (M⊕ − MJ), the circumbinary systems are typically at least as stable as 3-5 planet systems around single stars, so long as the in- ner planet is more than a factor of ∼ 1.2 outside of the single planet critical radius. One approaches the single star, 2 planet limit out- side of ap,1/acrit = 1.5. All of the observed single planet systems lie in the intermediate regime of 1.1 <∼ ap,1/acrit <∼ 1.5. These as- sumptions allow us to predict the number of expected detections of secondary planets in packed and sparse circumbinary systems in Section 4. An alternative approach is to study individual systems, thus, we now turn to the specific case of Kepler 47, and subsequently the currently known single planet systems. 3 CASE STUDY: KEPLER 47 Kepler 47 is thought to contain three transiting planets, with Kepler 47d being more uncertain (Orosz et al, in prep). Kepler 47d is pre- dicted to orbit between planet b and c, with a period of roughly 186 days. For a mass comparable to c, it is dynamically packed with respect to planet c: it has a separation of roughly β = 10. By con- trast, the separation between planets b and d is rather large, about β = 28. Unlike the single planet systems, the inner planet (b) is rel- atively well separated from the binary with ap,1/a∗ ≈ 3.5, which for a binary mass ratio of M2/(M1 + M2) ≈ 0.25 translates to ap,1 = 1.57 acrit (Holman & Wiegert 1999). Based on our integra- tions in Section 2, there is in principal place for a planet in between b and d, ignoring any contribution from c, especially considering the wide inner spacing. However, comparing the stability of two 4 Kratter & Shannon Figure 1. Instability timescale as a function of β for different inner planet separations. The panels are labeled by the planet-star mass ratio. The power law fits are quite good for the low mass case (top), whereas higher mass planets (bottom) show non-monotonic stability trends. For lower mass plan- ets, the scaling of stability with β is similar for all separations, whereas the more massive planets become more stable at lower β very quickly with in- creasing ap,1. The black dashed line indicates the fit for three planets from Youdin et al (2012) for the data of Chambers (1996). Figure 2. The same data as Figure 1 but now showing instability timescale as a function of inner planet semi-major axis, ap,1 for different values of β. Note that lower mass planets are less stable than higher mass planets for the same parameters. and three planets in the single star case suggests that the required distance will be much greater. Our two planet fit predicts that the lifetime of a planet intermediate between b and d, at β = 14 is 1014 years. In order to gauge the likelihood of such a stable orbit, and quantify whether the system is likely to be packed, we run an inte- gration of the Kepler 47 system to check for stable interior orbits. 3.1 Orbital Parameters We use one of the best fit models from the study of Orosz et al. (2012) to determine the initial orbital configuration for the binary, and planets b and c (J. Carter, private communication). Note that for the existing systems the constraints on eccentricity of the outer planet are limited. We place planet d on a 186 day orbit with zero initial eccentricity. While the fits to the Kepler light curves assumed massless planets, we assign the planets masses based on their in- ferred radii and the planetary radius-mass relationship in (Lissauer et al. 2011). The initial state vectors. planetary masses, and orbital elements are listed in Table 1. To search for stable orbits we populate the region in between Kepler 47b and d with 1000 test particles on circular, coplanar or- bits about the system barycenter. We set the minimum and maxi- mum separations of the test particles to be 5RH outside and inside of planets b and d respectively. The spacing limits are chosen con- servatively based on our two planet integrations above. 3.2 Results After 2 × 106 years (or roughly 108 binary orbits) all of the test particles were lost due to crossing the orbit of either planet b or d. We show both the distribution of particle lifetime with semi-major axis and the distribution of particle ejection in time in Figures 3 and 4. For the masses and orbital parameters chosen, planet c has both a large initial eccentricity, and also a somewhat variable orbit. Based on the variability in planet c, we also predict that either the mass of planet d or the eccentricity of planet c is lower. Although we have clearly not sampled the full phase-space of intermediate orbits, all of those chosen have lifetimes far shorter than the system age. Thus we conclude that with planet d in place, Kepler 47 is most likely dynamically packed. To the extent that β controls stability, changing the planetary mass by a factor of 2 only shifts β by ∼ 1.25, and thus our results are likely to hold for reasonable masses of planet d. 4 INTERPRETING EXISTING SYSTEMS: ARE CIRCUMBINARIES PACKED? Is the packing in Kepler 47 representative of the population as a whole? Clearly our investigation is hindered by poor statistics, but using the results from our n-body integrations we can quantify how many systems are required to distinguish between a packed and sparse population. As demonstrated above, we can roughly consider circumbi- nary two-planet systems packed if they have similar spacings to packed 3-5 planet single-star systems. The nominal minimum sep- aration according to Chambers et al. (1996) or Smith & Lissauer (2009) is of order β = 8 − 10. One should therefore expect subse- quent planets to be located between ∼ 8 < β < 20 away from each other. According to Fang & Margot (2013), the Kepler single star multiple systems are consistent with being packed in ∼ 35 − 45% c(cid:13) 2012 RAS, MNRAS 000, ??–?? 345678456789Log(Binary Orbits)¹=10¡5ap1=a¤3.53.253.02.752.5345678¯456789Log(Binary Orbits)¹=10¡4ap1=a¤3.53.253.02.752.52.62.83.03.23.43456789Log(Binary Orbits)¹=10¡5¯8.07.57.06.56.05.55.04.52.62.83.03.23.4ap;13456789Log(Binary Orbits)¹=10¡4¯4.94.64.34.03.853.73.553.43.25Table 1. State vectors for Kepler 47 system with the primary at the origin with zero velocity. Mass is in units of M(cid:12) = 1, G = 1, and length units are in AU. Planet Packing in Circumbinary Systems 5 M/M(cid:12) 1.043 0.362 2.985e-5 7.295e-5 7.295e-5 x 0.0 -8.638e-2 -0.146 -0.649 -1.575e-2 y 0.0 1.554e-4 -0.271 0.347 -0.985 z 0.0 0.0 1.338e-3 2.843e-8 1.611e-2 vx 0.0 -7.379e-3 1.971 -0.680 1.173 vy 0.0 -3.966 -1.90 -2.248 -0.797 vz 0.0 0.0 2.892e-3 6.115e-3 1.057e-2 a e i 0.083 0.083 0.293 0.712 0.947 0.032 0.032 0.0127 0.0 0.181 0.0 0.0 0.004 0.0 0.008 Figure 3. The distribution of test particles ejected from orbits between Kepler 47b and orphan planet Kepler 47d, as a function of initial semi- major axis. There appears to be a slight increase in stability at early times for orbits roughly equidistant from b and d. Figure 4. Distribution of particle ejections as a function of time. The ma- jority of particles are lost by a few 105 years. The rapid ejection of particles compared to the system lifetime suggests that it is packed with three planets. of the known systems with 3-4 planets. The distribution of separa- tions measured in β is a Rayleigh distribution with σ ≈ 17. Using the Kepler single star sample as an example of a packed population, we now pose the question: how many multi-planet cir- cumbinary systems should Kepler have observed in the current sample if the circumbinary systems are equally packed? If they are sparse? 4.1 Observability of secondary planets via transits Before considering an ensemble population, we need to estimate a rough transit probability Ptran for a second planet in an already detected circumbinary system. We make the simplifying assump- tion that the binary itself is edge-on (consistent with the star eclips- ing), and that any transits will occur over the primary star. From the point of view of the observer, consider a rectangle projected on to the plane of the sky which encompasses all possible locations of the primary star, with a height twice its radius (R1). This rectangle has dimensions Ar = 2a∗ m2 m1 + m2 × 2R1 = 2a1 × 2R1, (5) where a1 now refers to the primary's orbit about the barycenter. If the mutual inclination of the planet and binary is zero, the planet will always transit. As the inclination increases, transits will only occur if the location of the line of ascending nodes guarantees that the projected cord of the planet orbit crosses the projected rectan- gle of the stellar orbit (we shall multiply by 2 later for descending c(cid:13) 2012 RAS, MNRAS 000, ??–?? nodes in this arc as well). Figure 5 illustrates the geometry we con- sider. To determine the fraction of allowed ascending nodes, con- sider the fraction of the planetary orbit, which intersects the stellar rectangle: 2θ = 2 arcsin (6) (cid:18) a1 (cid:19) . ap This angle encompasses most allowed values for the ascending node. However, for small inclinations, there will be an angular off- set on either side of θ where an ascending node location will still send the planet grazing across the top or bottom of the star. These two segments have size δθ = R1 ap tan(i) . (7) Therefore, the total probability (including descending nodes) that such a planet will ever transit the star is: Ptrans = 2 = 2 arcsin + arcsin (8) We use this simplified probability to estimate whether or not a hypothetical secondary planet should be observed to transit. Cir- cumbinary systems with eclipsing binary stars have a much higher transit probability at a given inclination and separation. When con- sidering secondary planet transits only, we have already specified that the stars are eclipsing, and thus even relatively high inclination planets should be expected to cross the area swept out by the stars. It is only in the case where one specifies that the binary eclipses (cid:20) 1 π 2θ + δθ 2π (cid:18) a1 (cid:19) ap (cid:18) R1 (cid:19)(cid:21) ap tan i 0.350.400.450.500.550.600.65Initial Semi-Major Axis (AU)123456Log Ejection Time (yrs)123456Log Ejection Time (yrs)0.000.050.100.150.200.250.30Fraction of Test Particles6 Kratter & Shannon Figure 5. Schematic diagrams. Top panel shows birds eye view of a cir- cumbinary system with the project transit arc, 2θ, while the bottom shows the edge on view, illustrating the extra δθ added to account for some certain combinations of inclinations and lines of ascending node. that the probability greatly exceeds that for the single star case, as noted by Borucki & Summers (1984). Note that we have only defined the probability that a transit is possible in infinite time. De- pending on orbital inclination, the probability of transit per orbit can differ significantly from unity. We do not include this in our calculations at this time, although we do remove planets with pe- riods longer than the Kepler mission lifetime from the detection statistics. In this aspect our calculations are optimistic, although we also pessimistically exclude the possibility of multiple transits per orbit, and transits of the secondary (which have been observed in Kepler 16 by Doyle et al. 2011). 4.2 Observability of secondary planets via TTVs In the absence of (or in addition to) transits, one can also infer the existence of an outer massive body based on its gravitational effect measured through variations in the occurrence time of the transits, so-called TTVs. While there are enormous TTVs due to the stellar motion, there may also be smaller TTVs due to an outer planet. Holman & Murray (2005) estimate the magnitude of the change in transit time occurrence in single star systems to be: √ −2 ∆t ≈ 45π 16 Mp,2 M∗ P1α3 e(1 − e ) 2α3/2 (9) where αe = ap,1/[ap,2(1 − e2)]. We emphasize that this does not include any perturbations due to the binary. However, the TTVs due solely to the motion of the binary as a moving target can in principal be subtracted given a precisely characterized stellar orbit. TTVs due to the binary potential itself should have a significantly higher frequency modulation than that due to an outer planet (D. Nesvorny, private communication). We thus take this as an order Figure 6. Magnitude of predicted TTV signal in log(minutes) for an outer planet perturbing the orbit of Kepler 47b, assuming a zero eccentricity orbit. The location of the orphan transit (47d) is shown for reference. The white region shows orbits prohibited by stability, and the black dashed line shows where β = 20, delineating packed and sparse orbits based on the two planet case. Most packed systems would have TTV signals greater than 1 hour. of magnitude approximation for the amplitude of the variation in- duced by a second planet. A rigorous investigation of the detectabil- ity of TTVs due to an outer perturber in circumbinaries is beyond the scope of this paper, and will be the subject of future work. Because the orbits are long compared to a typical transiting planet, so too are the TTVs. In Figure 6, we show the magnitude of the TTV's induced on Kepler 47b as a function of the mass and semi-major axis of a perturbing outer planet (assuming e = 0). The black dashed lines indicates the boundary in mass and semi-major axis above which the two planets would be separated by more than β = 20. In this case, the detection of a TTV above ∼ 1 hour would indicate a dynamically packed 2 planet system, even without characterizing the mass or orbit of the perturber. Thus even without solving the inverse problem, for a given system it is possible to correlate TTV magnitude with planet packing directly. Based on the stated precision of the single planet transit times of roughly 5 minutes (see, e.g., table S4 of the supplementary infor- mation of Orosz et al. 2012), we adopt 20 minutes as the required size of the TTV to be detectable. For comparison, we also con- sider the detection probabilities for TTV signals that are at least 2.5 hours. Note that based on this formula, Kepler 47b would have TTVs of order 5 minutes due to the posited middle planet d, which have not been reported. Meanwhile, planet c would induce 18 hour TTVs on planet d, which we would expect to be both detectable and frustrating to orbit fitting with only a few transits. For most of the current sample of CBPs, the paucity of transits is likely the limiting factor in detecting TTVs: uncertainties in the orbital pa- rameters are degenerate with the signal. We discuss the importance of the number of transits further below. c(cid:13) 2012 RAS, MNRAS 000, ??–?? 2a1ap2✓ap/(2⇡)i✓=arcsin[R1/(aptani)]2a12R12✓ap/(2⇡)4.3 Monte Carlo generation of multi-planet systems To obtain statistics on the expected detection frequency of sec- ondary planets in existing circumbinary systems we conduct a Monte Carlo simulation of 10,000 possible second planets in each of the 6 detected systems. To increase our statistics we include Ke- pler 47 twice: once using Kepler 47b as the inner planet, and once using the probable orbital location of Kepler 47d, the so-called or- phan transit (Orosz et al. 2012). Thus we have effectively 7 sys- tems. Out of 7 systems, we generously count 2 "outer planet" de- tections (both in Kepler 47)3. We specify a distribution of β's as in Fang & Margot (2013), using a Rayleigh distribution, with a σ = 15 for a packed popu- lations, and σ = 30 for a sparse population. We truncate the dis- tribution at a larger value than these authors of β = 7, because closer planets would be unstable about the binary. This makes our detection predictions somewhat more conservative in that we re- move the most easily detectable companions from the sample. We choose eccentricities and inclinations from Rayleigh distributions with σ = 0.05, 3◦. Note that moderate changes in these values ef- fect the outcome very little. We assume that planetary radii follow a power law distribution in radius (scaling as R−1.97 ), which fits well the planet population from 3R⊕ < Rp < 20R⊕ for periods between 3 − 50 days (Youdin 2011). To convert radii to masses (necessary for TTV calculations), we use the fit to Earth and Saturn from Lissauer et al. (2011) where Mp = (Rp/R⊕)2.06M⊕. For each of the seven systems, we randomly draw planets from these distributions, calculate the transit probability, and based on this probability count the number of expected transiting secondaries. We also calculate the magnitude of the TTV, counting all above 20 minutes as detectable. p 4.4 Results In Figure 7 we show the normalized histogram distribution of the number of expected secondary detections out of 7 total systems via both transits and TTVs in both packed and sparse systems. Based on our sensitivity estimate, TTVs are the preferred detec- tion method for secondary planets, and thus the more efficient way to differentiate between the two dynamical states. As noted above, based on our estimate, Kepler 47d should have a detectable TTV, although given the limited number of transits, characterization may be challenging. We do not claim TTV detections will allow for the determi- nation of a secondary planet's properties. This would require far more transits than feasible for such long period systems. It remains uncertain how many transits will be needed to separate degenera- cies in the orbital properties of the inner planet from true TTVs. For systems like Kepler 47, where the inner planet has a period of 50 days, 2 -3 years of data would provide 10-20 transits. Kepler 47 b was well characterized with 18 transits. Thus characterization is possible in the course of a mission timeline. Ideally one would like to observe the full oscillation period of the TTV (Nesvorn´y & Morbidelli 2008; Meschiari & Laughlin 2010). In practice detect- ing even half of an oscillation might be sufficient to set a lower limit on the magnitude of the TTV. To calculate the sample size needed to distinguish between a population of packed systems and a population of sparse systems, 3 We do not include Kepler 47c as an interior planet as any planet outside of this would be undetectable in the current data c(cid:13) 2012 RAS, MNRAS 000, ??–?? Planet Packing in Circumbinary Systems 7 we use the mean detection probabilities for the packed and sparse cases (Pp and Ps). For transits, these rates are Pp ∼ 0.06 and Ps ∼ 0.03, for TTVs, they are Pp ∼ 0.55 and Ps ∼ 0.18. We use Monte Carlo simulations to determine the chance that a measure- ment of n systems will have k detections. We say that we can dis- tinguish packed and sparse populations at 1 σ if at least 68% of the Monte Carlo realizations have k such that we correctly prefer one model to the other by at least 68% : 32% (and similarly for 2 σ and 3 σ with 96% and 99.7% respectively). Using transit detections re- quires 140, 1027, and 2132 systems for 1σ, 2σ, and 3σ confidence, while using TTVs requires only 5, 34, and 75 systems for 1σ, 2σ, and 3σ confidence. At present, there have been no reports of TTVs in any of the systems. Despite the ∼ 99% probably of detecting at least one > 20 minute duration TTV in the current sample if the systems are packed, we cannot yet draw conclusions about the configura- tion with out a more systematic investigation of the detectability of TTVs. Such an investigation will be the subject of future work. If the true TTV sensitivity is closer to 2.5 hours, the detection prob- abilities shrink by roughly a factor of three: In this case, the likeli- hood of no TTV detections if the population is packed is ∼ 16%, and the number of systems needed to differentiate a population of packed systems from a population of sparse systems is 4, 55, and 120 for 1, 2, and 3 σ confidence respectively. These calculations suggest that continued monitoring of even a few circumbinary sys- tems in order to achieve better orbital parameters might provide substantial insight into the dynamical properties of the systems. 5 DISCUSSION Understanding the stable configurations of circumbinary planets, and comparing with observations, will provide invaluable insight into the formation of these complex systems. The mere existence of planets around binaries places strong constraints on the forma- tion of the binary itself, and on the planet's natal disk. If planet formation in such disks occurs on the same timescale as around a single star (and there is no reason to think the time requirements would be less stringent), the binary must arrive at its current orbital state in a manner that does not disrupt a circumbinary disk, or do so sufficiently early in the formation process as to permit the regrowth of a massive circumbinary disk (Throop & Bally 2008). Observations suggest that massive circumbinary disks are much rarer than those around single stars and wide binaries (Harris et al. 2012). Moreover, the very closest binaries show a bi-modal distribution of disk properties: in a survey of Taurus with the SMA, most tight binaries have, if anything, disks which are too tenuous to detect. The remaining few have disks that are at the high end of the mass function of single stars. Perhaps these are the plane- tary system progenitors? It is interesting to note that the properties of single stars do not converge with very tight binaries, suggesting that the formation of these systems is quite different. We have demonstrated that the stability of two-planet, cir- cumbinary systems is distinct from that of either two or three planet single star systems, where the minimum planet separation β is an increasing function of the inner planet separation. Depending on planet mass, the binary case asymptotes to the single star analytic two-planet results at inner-planet separations of order 1.5-2 acrit. When the inner planet is closer to acrit, the critical intra-planet spacing is of order β = 5 − 7. Kepler 47 is dynamically packed if the third planet exists, but the known population of circumbinary planets is too small to ro- 8 Kratter & Shannon Figure 7. Probability of detecting some number of two planet systems by either transits (top) or through TTVs (bottom) for packed (left) and sparse (right) planetary systems. We compare the transit probability for a single star and the binary to demonstrate the enhanced detection probability for CBPs. bustly distinguish between a generally packed population and a generally sparse population. An increase in the sample of circumbi- nary systems by roughly an order of magnitude may be sufficient to determine the typical dynamical state of such systems, but with the recent demise of the high-precision era of Kepler, the addition to the sample will be slow. Continued monitoring of the existing systems may prove helpful if additional transits are detectable. As noted by Borucki & Summers (1984), eclipsing binaries provide excellent targets for the detection of transiting planets. Moreover, such planets are more likely to be found on wider, Earth-like orbits. We suggest that future missions such as TESS include signifi- cant numbers of eclipsing binary systems among their targets in the areas with continuous coverage. If 20 transits are sufficient to dis- tinguish between degenerate orbital parameters and TTVs, then a 3 year mission may well be long enough to characterize the dynami- cal state of any discovered systems. Moreover, TESS-like missions, which target brighter stars, are particularly valuable because these systems can be followed up from the ground. acknowledegments We thank P. Armitage and W. Welsh for valuable discussions, and the referee for a careful review that sub- stantially improved this paper. Support for this work was pro- vided by NASA through Hubble Fellowship grant #HF-51306.01 awarded by the Space Telescope Science Institute, which is oper- ated by the Association of Universities for Research in Astronomy, Inc., for NASA, under contract NAS 5-26555. AS is supported by the European Union through ERC grant number 279973. REFERENCES Barnes R., Greenberg R., 2007, ApJ, 665, L67 Borucki W. J., Summers A. L., 1984, Icarus, 58, 121 Chambers J. E., 2007, Icarus, 189, 386 Chambers J. E., Wetherill G. W., 2001, Meteoritics and Planetary Science, 36, 381 Chambers J. E., Wetherill G. W., Boss A. P., 1996, Icarus, 119, 261 Doyle L. R., Carter J. A., Fabrycky D. C., Slawson R. W., 2011, Science, 333, 1602 Fabrycky D., Tremaine S., 2007, ApJ, 669, 1298 Fang J., Margot J.-L., 2013, ApJ, 767, 115 Ford E. B., Chiang E. I., 2007, ApJ, 661, 602 Gladman B., 1993, Icarus, 106, 247 Harris R. J., Andrews S. M., Wilner D. J., Kraus A. L., 2012, ApJ, 751, 115 Holman M. J., Murray N. W., 2005, Science, 307, 1288 Holman M. J., Wiegert P. A., 1999, AJ, 117, 621 Juri´c M., Tremaine S., 2008, ApJ, 686, 603 Laskar J., 1996, Celestial Mechanics and Dynamical Astronomy, 64, 115 Levison H. F., Duncan M. J., , 2013, SWIFT: A solar system inte- gration software package Lissauer J. J., Hubickyj O., D'Angelo G., Bodenheimer P., 2009, Icarus, 199, 338 c(cid:13) 2012 RAS, MNRAS 000, ??–?? 01234567Transit Detections0.20.40.60.81.0ProbabilityPackedCBSingle01234567Transit Detections0.20.40.60.81.0ProbabilitySparseCBSingle01234567TTV Detections0.050.100.150.200.250.300.350.40Probability2.5 hours20 min.01234567TTV Detections0.10.20.30.40.50.6Probability2.5 hours20 min.Planet Packing in Circumbinary Systems 9 Lissauer J. J., Ragozzine D., Fabrycky D. C., Steffen J. H., Ford E. B. e. a., 2011, ApJS, 197, 8 Martin R. G., Armitage P. J., Alexander R. D., 2013, ApJ, 773, 74 Marzari F., Thebault P., Scholl H., Picogna G., Baruteau C., 2013, A & A, 553, A71 Meschiari S., 2012, ApJ, 761, L7 Meschiari S., Laughlin G. P., 2010, ApJ, 718, 543 Migaszewski, C., Słonina, M., & Go´zdziewski, K. 2012, MNRAS, 427, 770 Moriwaki K., Nakagawa Y., 2004, ApJ, 609, 1065 Mudryk L. R., Wu Y., 2006, ApJ, 639, 423 Nesvorn´y D., Morbidelli A., 2008, ApJ, 688, 636 Orosz J. A., Welsh W. F., Carter J. A., Brugamyer E., Buchhave L. A., Cochran W. D., Endl M., Ford E. B., MacQueen P., Short D. R., Torres G., Windmiller G., Agol E., Barclay T., Caldwell D. A., Clarke B. D., Doyle L. R., Fabrycky D. C., Geary J. C., Haghighipour N., Holman M. J., Ibrahim K. A., Jenkins J. M., Kinemuchi K., Li J., Lissauer J. J., Prsa A., Ragozzine D., Sh- porer A., Still M., Wade R. A., 2012, ApJ, 758, 87 Orosz J. A., Welsh W. F., Carter J. A., Fabrycky D. C., Cochran W. D., Endl M., 2012, Science, 337, 1511 Orosz J. A. et al. 2013, in prep. Quillen A. C., 2011, MNRAS, 418, 1043 Rafikov R. R., 2013, ApJ, 764, L16 Schwamb M. E., Orosz J. A., Carter J. A., Welsh W. F., Fischer D. A., Torres G., Howard A. W., Crepp J. R., Keel W. C., Lintott C. J., Kaib N. A., Terrell D., Gagliano R., Jek K. J., Parrish M., Smith A. M., Lynn S., Simpson R. J., Giguere M. J., Schawinski K., 2013, ApJ, 768, 127 Smith A. W., Lissauer J. J., 2009, Icarus, 201, 381 Throop H. B., Bally J., 2008, AJ, 135, 2380 Tohline J. E., 2002, ARA&A, 40, 349 Welsh W. F., Orosz J. A., Carter J. A., Fabrycky D. C., 2013, ArXiv e-prints Welsh W. F., Orosz J. A., Carter J. A., Fabrycky D. C., Ford E. B., Lissauer J. J., Prsa A., Quinn S. N., Ragozzine D., Short D. R., Torres G., Winn J. N., Doyle L. R., Barclay T., Batalha N., Bloe- men S., Brugamyer E., Buchhave L. A., Caldwell C., Caldwell D. A., Christiansen J. L., Ciardi D. R., Cochran W. D., Endl M., Fortney J. J., Gautier III T. N., Gilliland R. L., Haas M. R., Hall J. R., Holman M. J., Howard A. W., Howell S. B., Isaacson H., Jenkins J. M., Klaus T. C., Latham D. W., Li J., Marcy G. W., Mazeh T., Quintana E. V., Robertson P., Shporer A., Steffen J. H., Windmiller G., Koch D. G., Borucki W. J., 2012, Nature, 481, 475 Yeh L.-W., Chang H.-K., 2009, Icarus, 204, 330 Youdin A. N., 2011, ApJ, 742, 38 Youdin A. N., Kratter K. M., Kenyon S. J., 2012, ApJ, 755, 17 c(cid:13) 2012 RAS, MNRAS 000, ??–??
0909.2548	1	0909	2009-09-14T13:54:06	Exoplanet Transit Database. Reduction and processing of the photometric data of exoplanet transits	[ "astro-ph.EP" ]	We demonstrate the newly developed resource for exoplanet researchers - The Exoplanet Transit Database. This database is designed to be a web application and it is open for any exoplanet observer. It came on-line in September 2008. The ETD consists of three individual sections. One serves for predictions of the transits, the second one for processing and uploading new data from the observers. We use a simple analytical model of the transit to calculate the central time of transit, its duration and the depth of the transit. These values are then plotted into the observed - computed diagrams (O-C), that represent the last part of the application.	astro-ph.EP	astro-ph	Exoplanet Transit Database. Reduction and processing of the photometric data of exoplanet transits. Stanislav Poddan´y a,b Lubos Br´at c,d Ondrej Pejcha c,e aAstronomical Institute, Faculty of Mathematics and Physics, Charles University Prague, CZ-180 00 Prague 8, V Holesovick´ach 2, Czech Republic b Stef´anik observatory, CZ-118 46 Prague 1, Petr´ın 205, Czech Republic cVariable Star and Exoplanet Section of Czech Astronomical Society dAltan observatory, CZ-542 21 Pec pod Snezkou 193, Czech Republic eDepartment of Astronomy, Ohio State University, Columbus, OH 43210, USA Abstract We demonstrate the newly developed resource for exoplanet researchers - The Ex- oplanet Transit Database. This database is designed to be a web application and it is open for any exoplanet observer. It came on-line in September 2008. The ETD consists of three individual sections. One serves for predictions of the transits, the second one for processing and uploading new data from the observers. We use a simple analytical model of the transit to calculate the central time of transit, its duration and the depth of the transit. These values are then plotted into the ob- served - computed diagrams (O-C), that represent the last part of the application. Key words: exoplanets, planetary systems, techniques: photometric, database PACS: 95.80.+p, 97.82.-j, 97.82.Cp 1 Introduction Research on extrasolar planets is currently one of the most exciting fields in astrophysics. The speculations on the existence of planets orbiting other solar-type stars ended fourteen years ago. In 1995 the discovery of the first extrasolar planet orbiting a solartype star - the well-known 51 Peg b, was Email address: [email protected] (Stanislav Poddan´y). Preprint submitted to Elsevier 3 July 2018 made by Mayor & Queloz (1995). Since then, the number of known planets has been growing quickly. Currently, more than 370 such bodies are known 1 . If a planetary system happens to be oriented in the space so that the or- bital plane is close to the line-of-sight to the observer, a planet periodically passes in front of the stellar disk. Photometric observation of the transit can then be used to derive the orbital and physical parameters of the planet (e.g., Southworth 2008 or Torres et al. 2008). For a review of properties that have been measured, or that might be measured in the future through precise ob- servations of transiting planets see Winn (2008). At the half of the year 2009, more than 60 planets with this special orientation were known. There are many amateur astronomers all over the world that achieved pho- tometric accuracy around the units of percent, which is necessary for quality observing of the exoplanet transit. Unfortunately, up to present day, no exo- planet light curves' database was available that would accept data from both professionals as well as from amateurs. Amateur observers are not constrain by telescope scheduling and often have unlimited access to their instrument which enables them to gather data over a long period. Huge quantity of observations is the key to the search for other planets in already known systems. It is important to monitor possible periodical changes in O-C plots of the planets because as Holman & Murray (2005) demonstrated in their theoretical work, short-term changes of the time of the transit can be caused by the presence of other exoplanets or moons in the system, see also Agol et al. (2005), Kipping (2009). On the other hand, potential long- term changes in the duration of the transit may be the consequence of orbital precession of exoplanets as Miralda-Escud´e (2002) showed in his theoretical work. To perform such effective studies, we need a database which includes all available data divided into groups according their quality. 2 Why ETD The Exoplanet Transit Database 2 (ETD) came on-line in September 2008 as a project of the Variable Star and Exoplanet Section of the Czech Astronomical Society. The ETD includes all known transiting planets that have published ephemerides. 1 see the list http://www.exoplanet.eu 2 http://var2.astro.cz/ETD 2 We have created on-line ETD portal to supply observers with such useful information like transit predictions, transit timing variation (TTV), variation of depth and duration with availability to draw user observation to the plot. Before the ETD, only two other transit databases were available. The Amateur Exoplanet Archive 3 (AXA), lead by Bruce Gary, and the NASA/IPAC/NExScI Star and Exoplanet Database 4 (NStED) (von Braun et al. 2008). AXA strictly accepts data only from amateurs. Unfortunately, the quality of light curves is very diverse. In spite of this fact, all of the available light curves have the same priority grading. On the other hand NStED contains only the light curves that have been already published (in the future some amateur light curves from AXA should be also accepted into the NStED and the AXA should expire). The main goal of the ETD is to gather all available light curves from profes- sional and also amateur astronomers (after one year of the existence of the database, more than one thousand such records are available). We are search- ing for new publications on several open archives to gather all available light curves. We also take over data from the NStED, AXA and from our project TRESCA 5 . It is also possible to upload data into the database directly us- ing a web-form or it can be added to the database by the administrators. All available data are on-line plotted into graphs where we make the provision for the quality of the light curve. All graphs (like TTV) can be easily down- loaded from the database. It is also possible to download light curves from the TRESCA observers and from amateur observers directly from the database. While collecting published data to ETD, we accentuate to have fully referenced its source. Each transit observation we store full reference with URL pointing to the paper or web-page where data were found. When we take over the whole light curve we display it only with reference to source of the data there is no mention of ETD in the picture. 3 Parts of the ETD The ETD is composed from three sections. The first one - Transit prediction - serves for prediction of the transits. The second one - Model fit your data - is a web-form for accepting and processing new data. The last section - O-C plots - contains the observed - computed diagrams of the central times of transits, depths and the transit durations that are generated on-line from the database. 3 http://brucegary.net/AXA/x.htm 4 http://nsted.ipac.caltech.edu 5 http://var2.astro.cz/EN/tresca 3 Fig. 1. A record sample in the ETD after successful processing the light curve. 3.1 Transit prediction This section of the ETD contains one month prediction of observable tran- sits (the starting day is the date two days before current date) and also the prediction for the next 365 days for selected exoplanet. Any observer can find here the time of the transit start/center/end, duration and the depth of each transit for any place in the world. Furthermore the altitude and the cardinal point of the object in the sky are displayed for the first contact, mid-transit time and the last contact. In the one year prediction window you can also see the finding chart 6 (15′ x 15′). 3.2 Model-fit your data This section describes an user-friendly web-form for uploading and process- ing the light curves into the database. To model-fit the transit we assume that the observations consist of N relative magnitudes mi taken at times ti (i = 1, 2..., N) and the photometry software provided measurement errors σi computed most likely from Poisson statistics and read-out noise. We model 6 downloaded from the http://archive.stsci.edu/dss/index.html 4 the dataset by a function m(ti) = A − 2.5 log F (z [ti, t0, D, b] , p, c1) +B(ti − tmean) + C(ti − tmean)2, (1) where F (z, p, c1) is a relative flux from the star due to the transiting planet. We assume that the planet and the star are dark and limb darkened disks, respectively, with radius ratio p = RP /R∗ and that the planet is much smaller than the star, p . 0.2. The projected relative separation of the planet from the star is z. Limb darkening of the star is modeled by the linear law with the coefficient c1. We employ the occultsmall routine of Mandel & Agol (2002) as our F (z, p, c1). We checked that the small planet approximation, p . 0.2, does not produce significant differences from the full model (at least for the typical values of p and having in the mind the typical quality of the photometry) and is much faster to compute, which is the most important factor for on-line processing. We model the planet trajectory as a straight line over the stellar disk with impact parameter b = a cos I/R∗, where a is a semi-major axis and I is the orbit inclination. The mid-transit occurs at t0 and the whole transit lasts D. Based on these assumptions we can compute z[ti, t0, D, b] for every ti. Variable A in the equation (1) descibes the zero-point shift of the magni- tudes, while B and C describe systematic trends in the data. Linear and quadratic terms are computed with respect to the mean time of observations tmean = P ti/N to suppress numeric errors. We do not employ any explicit cor- rection for air-mass curvature as we think a generic second-degree polynomial is sufficient in most cases. We used the Levenberg-Marquardt non-linear least squares fitting algorithm from Press et al. (1992), procedure mrqmin. The algorithm requires initial val- ues of parameters and partial derivatives of the fitted function. We take the initial values from literature (except for c1, see below). We compute all par- tial derivatives of the equation (1) analytically, except for ∂F/∂z, ∂F/∂p and ∂F/∂c1 which were computed numerically using Ridders' method (procedure dfridr of Press et al. 1992). The search for the optimal parameters is done by iterating the fitting proce- dure until the ∆χ2 (between fits) does not change significantly. Usually, with good initial values, about ten iterations are sufficient. Then the error bars σi are re-scaled to make the final χ2 = N − g, where g is a number of free parameters, and we re-run the fitting procedure to obtain final errors of the parameters. Original photometric errors are usually underestimated and this procedure yields more reasonable errors of the output parameters. 5 Table 1 The comparison of our results with Winn et al. (2007). HD189733b ETD Winn et al. (2007) central time [HJD] 2453988.80333 (12) 24543988.80331 (27) duration [min.] 106.01 ± 0.50 109.62 ± 1.74 In the optimal circumstances, one would consider all variables in equation (1), namely A, B, C, t0, D, b, p and c1 free parameters. However, these parameters are correlated to some extent and noisy photometry from a small amateur telescopes does not permit recovery of all of them. We need to fit the zero- point shift A and in most cases also a linear systematic trend B. Our primary goal is to get the central time of the transit t0, duration D and depth of the transit δ. Hence, we set t0 and D as free parameters, by default. However, for a limb-darkened star, the depth of the transit δ is determined by radius ratio p, impact factor b and limb darkening coefficient c1. Primarily, the depth of transit δ is governed by the radius ratio p and we set it as a free parameter. The parameters b and c1 affect the depth and the shape of the transit to a lesser extent and from noisy amateur data we couldn't retrieve meaningful values for the two parameters simultaneously with p. Therefore, we hold b and c1 fixed. We either compute b from orbital parameters of the planet and from the radius of the star or take the value from the literature. The situation is more complicated for limb darkening because c1 should be different for every photometric filter. We decided to keep c1 fixed at a rather arbitrary value c1 = 0.5 in all cases. We experimented with values from 0.2 to 0.9 and found that the effect on other parameters is rather negligible, usually smaller than the error bars. The export value of the depth is then evaluated as δ = −2.5 log(cid:20)min z F (z, p, c1)(cid:21) . (2) At Fig. 1 you can see the example of the record in the ETD after successful processing of the light curve. We tested our algorithm using the data of the exoplanet HD189733b that were published by Winn et al. (2007). When we fitted this precise data using our code we obtained the value of the mid-transit time which was in excellent agreement with the published results. The duration of the transit wasn't inside the error bars. The errors of our fit are lower than those in the paper where the MCMC simulations were used (Tab. 1). We think that this excess is due to red noise which we did not take into account in our calculation and because of impact parameter that is fixed in our case. We also made a test with the data from AXA and we obtained similar results to the results presented in their archive (Tab. 2). 6 Table 2 The comparison of our results with the AXA database. HD189733b ETD AXA central time [HJD] 2454705.40228 (41) 2454705.4023 (5) duration [min.] depth [mag.] 102.09 ± 1.60 0.0287 ± 0.0006 98.4 ± 1.8 0.02895 ± 0.00080 Fig. 2. The central transit time O-C plot for the exoplanet TrES-1 b in the ETD. 3.3 O-C plots The O-C plots section contains the diagrams of the central transit time, the duration of the transit and its depth as a function of time. All available data for the selected planet including the error bars are visible in these graphs (Fig. 2). The quality of individual light curve is indicated by the size of the dot. The records in the ETD are divided into 5 groups according to their data quality index DQ. While computing DQ index of the light curve, the following relation is used α = √ρ δ S (3) where α is a temporary data quality index, S is the mean absolute deviation of the data from our fit and ρ = N/l means the data sampling, where l is the length of observing run in minutes. 7 Table 3 The distribution of the quality of the light curves according to their DQ index (example of the light curves see Figure 3). DQ index 1 2 3 4 5 threshold α ≥ 9.5 9.5 > α ≥ 6.0 6.0 > α ≥ 2.5 2.5 > α ≥ 1.3 1.3 > α Fig. 3. Examples of the light curve with the different DQ index. The number α is further transformed for better lucidity to the scale from 1 to 5 where 1 presents the best quality data and the value 5 the worst data (Tab. 3, Fig. 3). These thresholds are used only if whole transit is observed or when we take over whole light curve (AXA, TRESCA). If some part (egress or ingress) is missing in the dataset, the observation automatically gets the worst DQ index and in the notice column in the summary table the notice "Only partial transit" is generated. When we take over only the results of midtransit time, transit depth and length of the transit (not the whole light curve) we usually give the DQ equal 1. 3.3.1 How to download data In the O-C plots section of the ETD you can also download the data for your further studies. If there is displayed whole light curve you can download it using link in the DQ column. Whole table including the O-C residuals can be downloaded directly using link "Show data as ASCII table separated by semicolon" below the table. If the transit observation that you used in your next study was captured to ETD from literature, you can find reference in column "Author & REFERENCE", so the source paper should be cited in common way. If transit observation was published in some on-line source 8 (AXA, TRESCA), ask observers for a permission and other useful comments about the data. We can supply you with e-mail contact to observer if necessary. 4 Future development and discussion We have created the on-line portal to supply observers with information like transit predictions, TTV, variation of depth and duration plots with avail- ability to draw user observation to the plot. We think that main part of the database is now ready to use and should be useful tool for community. In future we plan some improvements of the database to be more user-friendly. We also plan to implement limb darkening tables into our fits. Further in the future we would like to develop a (semi)automatic procedure for searching of the transit timing variations in the O-C diagrams. 5 Acknowledgments The Exoplanet Transit Database is maintained by the Czech Astronomical Society (CAS), an almost 100 years old astronomical society with hundreds of members - professional and amateur astronomers in the Czech Republic. ETD is stored on the web server of CAS which is periodically backup. The server and CAS itself are supported by grants by Czech national Council of Scientific societies and membership fees. The long history of our organization, large membership platform and financially assured operation of the server are the basic conditions making the ETD permanent source where observers can store their data securely. This investigation was supported by the Grant Agency of the Czech Republic, grant No. 205/08/H005. We also acknowledge the support from the Research Program MSM0021620860 of the Ministry of Education. References Agol, E., Steffen, J., Sari, R., Clarkson, W. 2005, MNRAS, 359, 567 von Braun, K. et al. 2008, The NStED Exoplanet Transit Survey Service, Proceedings of the IAU, Volume 4, IAU Symposium S253, pp 478-481 Henry, G. W., Marcy, G. W., Butler, R. P., Vogt, S. S. 2000, ApJ, 529, L41 Holman, M. J., & Murray, N. W. 2005, Science, 307, 1288 Kipping, D. M. 2009, MNRAS, 392, 181 Mandel, K. & Agol, E. 2002, ApJ, 580, 171 9 Mayor, M. & Queloz, D. 1995, Nature, 378, 355 Miralda-Escud´e, J. 2002, ApJ, 564, 1019 Press, W. H., Teukolsky, S. A., Vetterling, W T., Flannery, B. P. 1992, Cambridge: University Press, Numerical recipes in C. The art of scientific computing Southworth, J. 2008, MNRAS, 386, 1644 Torres, G., Winn, J. N., & Holman, M. J. 2008, ApJ, 677, 1324 Winn, J. N., Holman, M. J., Henry, G. W., Roussanova, A., Enya, K., Yoshii, Y., Shporer, A., Mazeh, T., Johnson, J. A., Narita, N., Suto, Y. 2007, AJ, 133, 1828 Winn, J. N. 2008, Measuring accurate transit parameters. Proceedings of the IAU, Volume 4, IAU Symposium S253, pp 99-109 10
1609.05638	1	1609	2016-09-19T08:48:27	Belt(s) of debris resolved around the Sco-Cen star HIP 67497	[ "astro-ph.EP" ]	In 2015, we initiated a survey of Scorpius-Centaurus A-F stars that are predicted to host warm-inner and cold-outer belts of debris similar to the case of the system HR~8799. The survey aims to resolve the disks and detect planets responsible for the disk morphology. In this paper, we study the F-type star HIP~67497 and present a first-order modelisation of the disk in order to derive its main properties. We used the near-infrared integral field spectrograph (IFS) and dual-band imager IRDIS of VLT/SPHERE to obtain angular-differential imaging observations of the circumstellar environnement of HIP~67497. We removed the stellar halo with PCA and TLOCI algorithms. We modeled the disk emission with the GRaTeR code. We resolve a ring-like structure that extends up to $\sim$450 mas ($\sim$50 au) from the star in the IRDIS and IFS data. It is best reproduced by models of a non-eccentric ring with an inclination of $80\pm1^{\circ}$, a position angle of $-93\pm1^{\circ}$, and a semi-major axis of $59\pm3$ au. We also detect an additional, but fainter, arc-like structure with a larger extension (0.65 arcsec) South of the ring that we model as a second belt of debris at $\sim$130 au. We detect 10 candidate companions at separations $\geq$1''. We estimate the mass of putative perturbers responsible for the disk morphology and compare it to our detection limits. Additional data are needed to find those perturbers, and to relate our images to large-scale structures seen with HST/STIS.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. HIP67497Bonnefoyastroph September 7, 2018 c(cid:13) ESO 2018 Letter to the Editor Belt(s) of debris resolved around the Sco-Cen star HIP 67497(cid:63), (cid:63)(cid:63) M. Bonnefoy1, J. Milli2, F. M´enard1, A. Vigan3, A.-M. Lagrange1, P. Delorme1, A. Boccaletti4, C. Lazzoni5, R. Galicher4, S. Desidera5, G. Chauvin1, J.C. Augereau1, D. Mouillet1, C. Pinte1, G. van der Plas6, R. Gratton5, H. Beust1, and J.L. Beuzit1 (Affiliations can be found after the references) Received May 13, 2016; accepted September 19, 2016 ABSTRACT 6 1 0 2 p e S 9 1 . ] P E h p - o r t s a [ 1 v 8 3 6 5 0 . 9 0 6 1 : v i X r a Aims. In 2015, we initiated a survey of Scorpius-Centaurus A-F stars that are predicted to host warm-inner and cold-outer belts of debris similar to the case of the system HR 8799. The survey aims to resolve the disks and detect planets responsible for the disk morphology. In this paper, we study the F-type star HIP 67497 and present a first-order modelisation of the disk in order to derive its main properties. Methods. We used the near-infrared integral field spectrograph (IFS) and dual-band imager IRDIS of VLT/SPHERE to obtain angular-differential imaging observations of the circumstellar environnement of HIP 67497. We removed the stellar halo with PCA and TLOCI algorithms. We modeled the disk emission with the GRaTeR code. Results. We resolve a ring-like structure that extends up to ∼450 mas (∼50 au) from the star in the IRDIS and IFS data. It is best reproduced by models of a non-eccentric ring with an inclination of 80 ± 1◦, a position angle of −93 ± 1◦, and a semi-major axis of 59 ± 3 au. We also detect an additional, but fainter, arc-like structure with a larger extension (0.65 arcsec) South of the ring that we model as a second belt of debris at ∼130 au. We detect 10 candidate companions at separations ≥1". We estimate the mass of putative perturbers responsible for the disk morphology and compare it to our detection limits. Additional data are needed to find those perturbers, and to relate our images to large-scale structures seen with HST/STIS. Key words. techniques: high contrast imaging- stars: planetary systems - stars: individual:HIP 67497 (HD 120326) 1. Introduction The Scorpius-Centaurus OB association (Sco-Cen) is the near- est site of recent massive star formation (de Zeeuw et al. 1999). The proximity (d=90-200 pc) and young age (∼11-17 Myr) of Sco-Cen all contribute to make it an excellent niche for direct imaging (DI) search of planets and disks. The planet imager in- struments GPI (Macintosh & Graham, J. R., et al. 2008) and SPHERE (Beuzit et al. 2008) have initiated surveys of the cir- cumstellar environment of a few Sco-Cen stars at unprecedented contrasts and angular resolution. They have already resolved several new debris disks around those stars (Currie et al. 2015; Kasper et al. 2015; Kalas et al. 2015; Lagrange et al. 2016; Draper et al. 2016) with wing-tilt and ringed morphologies in- dicative of the presence of planets. We initiated a DI survey with SPHERE to image new giant planets and circumstellar disks around a sample of Sco-Cen F5- A0 stars with infrared excess. The excesses can be modeled by 2 black-body components, each corresponding to a belt of debris (Chen et al. 2014, hereafter C14). This architecture is reminis- cent of the Solar System, and of benchmark systems previously identified by DI like HR 8799 (Marois et al. 2008, 2010), or HD 95086 (Rameau et al. 2013) which is also part of Sco-Cen. In the course of the survey, we resolved a disk around the F0-type star HIP 67497 (HD 120326). This intermediate-mass star (M= 1.6M(cid:12)) is located at a distance of 107.4+10.4−8.7 pc (van Leeuwen 2007) and belongs to the 16 Myr old (Mamajek et al. (cid:63) Based on observations made with ESO Telescopes at the Paranal Observatory under programme ID 097.C-0060(A) (cid:63)(cid:63) This work is based on data products produced at the SPHERE Data Center hosted at OSUG/IPAG, Grenoble. 2002) Upper Centaurus Lupus sub-group (de Zeeuw et al. 1999). C14 modeled the infrared excess of the star with two belts of de- bris: a cold (127 ± 5 K) bright (LIR/L∗ = 1.1 × 10−3) belt at 13.9 au and a second colder (63± 5 K) dimmer (LIR/L∗ = 1.4× 10−4) belt at 116.5 au. The same team produced an alternative model of the excess requiring only one belt located at 8.82±1 au (Jang- Condell et al. 2015). Previous observations with adaptive op- tics systems did not reveal any companion or structure close to HIP 67497 (Janson et al. 2013). Our resolved observations there- fore offer the opportunity to better constrain the radial distribu- tion of the dust around this star and to look for the companions responsible for that distribution. We present the observations and data in § 2, and the models in § 3. We discuss the morphology of the disk and the existence of putative perturbers (planets) in § 4. 2. Observations and data reduction We observed HIP 67497 on April 6, 2016 with the VLT/SPHERE instrument (Beuzit et al. 2008) operated in IRDIFS mode (Tab. 1). The mode enables for pupil-stabilized observations of the source placed behind an apodized Lyot coro- nagraph (92.5 mas radius) with the IRDIS (Dohlen et al. 2008) and IFS (Claudi et al. 2008) sub-instruments operated in paral- lel. IRDIS yielded images of ∼11.1×12.4" centered onto the star in the H2 (λc=1.593µm, width=52 nm) and H3 (λc=1.667µm, width=54 nm) filters (Vigan et al. 2010). The IFS datacubes are made of 39 images covering the 0.96-1.33µm spectral range and a square field-of-view of 1.8 arcseconds. For registration pur- poses -- for both sub-instruments before and after the deep ex- posures -- we obtained frames with satellite spots (Star Center) 1 M. Bonnefoy et al.: Belt(s) of debris resolved around the Sco-Cen star HIP 67497, Table 1. Log of observations, April 6, 2016. UT time HH:MM 04:49 04:50 04:50 06:01 06:02 Neutral density ND 1.0 ND 0.0 ND 0.0 ND 1.0 ND 0.0 IFS IRDIS DIT × NDIT × NEXP (s) (s) 4×2×1 8×1×1 32×1×1 64×1×1 64×4×16 32×8×16 4×2×1 8×1×1 32×1×1 64×1×1 ∆PA <Seeing> Airmass (◦) 0.1 0.1 37.5 0.1 0.1 (") 1.2 1.2 1.1 1.1 1.0 1.1 1.1 1.1 1.1 1.1 τ0 (ms) 3.7 4.6 3.5 2.9 3.0 Remarks Unsat. PSF Star center ADI sequence Unsat. PSF Star center Notes. The seeing is measured at 0.5 µm. DIT (Detector Integration Time) refers to the individual exposure time per frame. NDIT is the number of individual frames per exposure, NEXP is the number of exposures, and ∆PA to the amplitude of the parallactic rotation. Fig. 1. Images of the disk structure around HIP 67497 obtained with the IFS (YJ), and IRDIS (H2H3), and considering PCA and TLOCI reductions. created by a waffle pattern introduced onto the deformable mir- ror of the instrument. Those frames were used to assess the po- sition of the star behind the coronagraphic mask. We recorded additional non-saturated exposures of the star placed outside of the coronagraphic mask with the neutral density filter ND 1.0 to estimate the flux and position of point sources. We reduced the IRDIS data with the Data Reduction Handling software (DRH) of SPHERE (Pavlov et al. 2008). The DRH corrected the raw science frames from the bad pix- els, subtracted the dark, and recentered the images onto a com- mon origin using the Star Center images. The field rotation of 37.5◦ during the sequence of coronagraphic exposures was suf- ficient to apply the angular differential imaging (ADI) technique and reveal point sources down to 0.1". We applied a Principal Component Analysis (PCA) algorithm (Soummer et al. 2012) on the H2 and H3 images separately to attenuate the stellar halo. The small number of modes kept in the PCA (10) enabled to look for extended structures. We used the TLOCI algorithm (Marois et al. 2014) to identify point sources and to confirm the struc- tures. The IFS data were analysed with a custom pipeline (Vigan et al. 2015, hereafter V15). This tool subtracted the background, interpolated the bad pixels, removed the instrument crosstalk, and performed a wavelength calibration of the spectra on the 2D raw frames. The DRH was then used to build the cubes from the 2D calibrated frames. We performed a PCA taking advantage of the wavelength coverage and field rotation to look for point sources (V15). We also performed PCA and TLOCI reductions of the IFS data after collapsing each cube in wavelength in order to reveal the (extended, faint) disk structures. Fig. 2. Radial profile along the disk position angle for the ring (left) and for the arc (right), extracted from the H2H3, and YJ PCA images and normalized to the median emission of the east side from 15 to 40 au. The residuals between the East and West disk profiles are shown at bottom. The normalization is different for the right and left panels. The dot-dashed line correspond to the cut of Fig. 1 We used the True North value of −1.649 ± 0.019◦, and platescales of 12.257±0.004 and 7.46±0.02 mas/pixel for IRDIS and the IFS, respectively. These values are derived from the ob- servations of the NGC3603 astrometric field obtained as part of the SPHERE GTO survey on March 30, 2016 (priv. com.). 3. Disk properties We detect an inclined ring at the same position in our collapsed YJ, H2, and H3 images (Fig. 1). The ring structure is seen both in our PCA and TLOCI reductions. It is also retrieved when ex- ploiting both the ADI technique and the wavelength diversity of the IFS and IRDIS (ASDI). We are confident this is a real struc- ture, not related to the stellar halo. The ring extends up to 450 mas, or 50 au, from the star in the IRDIS images. It has a smaller apparent extension (40 au) in the YJ images. This is likely due to the lower S/N of those data. We derived the flux profile (non cor- rected from artifacts introduced by the ADI; Milli et al. 2012) along the position angle of the ring from the H2H3, and YJ PCA images (Fig. 2). To improve the S/N, we averaged the flux over 2 H2H3H2H3YJYJPCATLOCITLOCIPCAEN300mas32 auRingArc300mas32 auEN102030405001234RING102030405001234Normalized integrated fluxBCK - IFSEast/H2H3West/H2H3East/YJWest/YJBCK - H2H30204060801001200123456ARC0204060801001200123456Normalized integrated fluxEast/H2H3West/H2H3BCKARC0204060801001200123456Normalized integrated flux1020304050-3-2-101231020304050Separation [au]-3-2-10123Difference E-W020406080100120-3-2-10123020406080100120Separation [au]-3-2-10123Difference E-WM. Bonnefoy et al.: Belt(s) of debris resolved around the Sco-Cen star HIP 67497, Fig. 3. Results of the forward modeling for the case of a single belt. Left: PCA image of the disk. Middle: Residuals after subtraction of the best single ring model, with the same linear color scale as the left image. Right: best single ring model (unconvolved with the SPHERE PSF) in a log scale to show the fainter backward side of the disk. circular apertures of one FWHM diameter centered on that axis. The error bars for the East and West sides are computed from the dispersion of the flux within each aperture. They are of com- parable amplitude. We also report the flux profile of the residual background ("BCK") extracted along the same appertures but tilted at a position angle perpandicular to the disk one (green and pink lines). The profiles indicate that the ring is rougthly sym- metric and has a similar flux profile in the YJ and H2/H3 bands from 15 to 50 au. Below 15 au and down to the coronagraph (10 au), the flux profile can be affected by residual speckles. An additional arc-like structure ("Arc" in Fig. 1) is detected at 40 mas South of the main ring. It extends visually up to 650 mas on the west side (70 au). This arc structure shows a strong flux asymmetry compared to the ring (Fig. 2). It is also retrieved at the same position into the IFS and IRDIS image for both TLOCI and PCA reductions. As for the ring, we are confident this feature is real and not a residual from the stellar halo. In the following section, we attempt a first characterization of the observed structures to better understand their nature. A more de- tailed modeling will be done in a forthcoming study. 3.1. Model We injected scattered light disk models into the IRDIS data to interpret the observed features. The disk models were generated with the GRaTeR code (Augereau et al. 1999). GRaTeR consid- ers a power-law for the radial distribution of the dust of index αin > 0 and αout < 0 inward and outward of a reference radius r0(θ) respectively. The disk model is rotated to the angles of the initial frames, convolved with the measured PSF and then sub- tracted from the data. The resulting images are re-reduced using the PCA algorithm used for the data reduction to obtain a disc- subtracted image. This step is repeated iteratively by varying the free parameters of the disk model until a reduced χ2 computed in the part of the model-subtracted image in a zone covering the lo- cation of the disk on the averaged H2+H3 images is minimized. We assumed a grey color for the disk between H2 and H3. To limit the parameter space, we adopted a distribution of optically thin dust with constant effective scattering cross-section. 3.2. Inner ring To model the inner ring, we fixed the inner and outer slope to αin = 10 and αout = −5 to mimic sharp edges. We assumed a circular ring, with anisotropic scattering parametrized by a Henyey-Greenstein parameter g. Therefore, the model has five free parameters (semi-major axis r0, position angle PA, incli- nation itilt, g, and the flux). The χ2 (hereafter χ2 region 1) is com- puted in a region encompassing only the inner belt (see App. A). The parameters of the best-fitting model are reported in Tab. 2 Table 2. Morphology of the inner belt PA (◦) -93±1 r0 (au) 58.6 ± 3 itilt (◦) 80 ± 1 g 0.82±0.02 χ2 region 1 2.6 Table 3. Results of the modeling of the belt and arc Model Two belts Belt+halo αout -8+4−8 -1.3+0.6−0.8 r0 (au) 130±8 n.a. flux ratio 0.08±0.03 0.66±0.40 χ2 region 2 3.5 3.8 and details on the minimization are given in App. A. The initial PCA-reduced image, best model and PCA-reduced image after best model subtraction are shown in Fig. 3. The "arc" structure is still present in the residual image (Fig. 3), thus indicating that it is not a product of self-subtraction ef- fects from the ring. We compare two possible models of that structure in the following section. 3.3. Model of the arc We considered the best model of the inner ring (§ 3.2) and mod- eled the arc with a second, outer belt. We assumed the same inclination, position angle and g parameter as for the inner belt. The outer slope αout, the radius and the flux ratio with respect to the inner belt are left as free parameters. The χ2 (hereafter χ2 region 2) is now computed on a region encompassing the pixels where the ring and the arc are detected (App. A). As an alternative, we investigated whether a narrow belt and a smooth disk halo could reproduce the ring and the arc. We did so by analogy with HR 8799 where such a halo was proposed (Su et al. 2009). We used for the ring the best fit found in § 3.2, and added a smooth halo modeled by an additional disk of outer slope αout,halo < αout and with a flux scaling factor lower than that of the narrow belt. In total, there are 3 free parameters: αout,halo, the narrow belt flux, and the halo flux. The best fit parameters are given in Tab. 3 and the resid- uals after subtraction of the two models are shown in Fig. 4. The residuals are slightly higher for the case of the ring+halo model. The scenario including two belts of debris is favored by the present data. 4. Concluding remarks The disk around HIP 67497 has an infrared luminosity (IL) in the same range as the disks already resolved around the Sco-Cen A-F type stars HD 111520, HIP 79977, HD 106906, and HD 115600 (LIR/L∗ ∼ 10−3−10−4). This could explain why we were 3 !Figure 2. Left: PCA image of the disk. Middle: Residuals after subtraction of the best single ring model, with the same linear color scale as the left image. Right: best single ring model (unconvolved with the SPHERE PSF) in a log scale to show the fainter backward side of the disk. This model only tries to reproduce the inner ring and therefore the arc described in Fig. 1 is still clearly visible in the residuals. !! Figure 3: Left panels: best model of the double ring (top) and residuals after subtraction (bottom) . Right panels: best model of a single ring with a smooth halo (top) and residuals after subtraction. The 2 bottom panels have the same linear colour scale (also identical as that of Fig. 3) while the models have a logarithmic colour scale. !- I strongly recommend adding more discussion and interpretation related to the observed structures. What can be the origin of a single ring vs. ring+arc? Why if there is an external belt, only one part of the belt is observed in form of an arc? I imagine that the authors are concerned about the page limit, but I would suggest moving the section of detected point sources and limits to an appendix and increase the discussion part, if the authors desire to keep the manuscript as a letter. It would be interesting also to compare the results with the structures resolved in other debris disks (which are listed for instance in the conclusion of the abstract). As I mentioned before, a discussion about the possible relation with the observed candidate companions (potential disk truncation and sculpting the observed structures) can be very suitable for a discussion. !!You can discuss the presence of gas that can lead to rings as shown by Lyra and Kuchner. ALMA constraints can constrain such a scenario. Alternatively a planet might have carved a gap, or a resonance might have removed particles from un unstable configuration as is the case in the Saturn rings.!!Minor comments: !!-Conclusion in the abstract: it does not have meaningful information. It reads more like a general remark than a conclusion of the analysis done for the paper. !!M. Bonnefoy et al.: Belt(s) of debris resolved around the Sco-Cen star HIP 67497, Fig. 4. Results of the forward modeling for the case of two belts (left panels, reduced χ2 = 3.5) and a narrow-ring laid over a smooth halo (right, reduced χ2 = 3.8). The models are shown on top and the post-ADI residual images at the bottom. able to resolve it. For HD 106906, and HD 115600, C14 finds outer belts at shorter separations than observed. In the case of HIP 67497, the second belt corresponds roughly to the location of the coldest belt found by C14, but the flux ratio between the two belt models found in § 3.3 is of the same order as the ratio of the IL of the belts found by C14 (7.85). HST/STIS images of HIP 67497 show extended emission at large scales (Padgett & Stapelfeldt 2016), which we compare to the SPHERE images in Fig. 5. The inner structures (≤3") seen with HST have an orientation compatible with the belt and the arc of the SPHERE images. Another asymmetric feature is seen on the east side up to 8 arcsec, but can not be easily re- lated to the rest. Different asymmetries at different scales have already been noticed for HD106906 (Kalas et al. 2015) or HD 32297 (Esposito et al. 2014; Schneider et al. 2014) for instance. One candidate companions (#1, see App. B) could lie within the structures revealed by STIS and the arc seen with SPHERE. The present STIS data are unfortunately affected by blind zones caused by the position of the coronagraphic bars of STIS (only two roll angles). Several options exist to explain the morphology of the disk of HIP 67497. The observed ring-like structures may be caused by dust-gas interactions (Lyra & Kuchner 2013). Unfortunately, the gas content of the disk remains unknown. Alternately, plan- ets with different individual eccentricities and semi-major axes (Lee & Chiang 2016) may also provide an explanation for the double-ring structure. Our observations are sensitive to 1.5 to 15 MJup in-between the ring and the arc when accounting for the disk inclination (Fig. B.1). We explore in App. C the case of one, two, or three perturbers on circular orbit, or one and two planets on eccentric orbits using numerical simulations. The predicted masses can reach ∼21 MJup for the case of a single planet. But the mass estimate of the perturber(s) is sensitive to the eccentric- ity of the orbit(s). New observations with STIS and SPHERE are required to reveal the full morphology of the disk, improve the detection performances, and to clarify the nature of the CCs. Acknowledgements. We thank our referee for his/her constructive review of the manuscript. We also thank the ESO staff for gathering those data. We are grateful to the SPHERE consortium for providing instrument platescale and True North. We ackn. support in France from the ANR (grant ANR-14-CE33-0018), the PNP, and the PNPS. Gvdp ackn. support from the Millennium Science Initiative (grant RC130007), and from FONDECYT (grant 3140393). References Augereau, J. C., Lagrange, A. M., & Mouillet, D., et al. 1999, A&A, 348, 557 Baraffe, I., Chabrier, G., & Barman, T. S., et al. 2003, A&A, 402, 701 4 Fig. 5. Sketch showing the HST/STIS and the SPHERE images of the debris disk around HIP 67497. Beuzit, J.-L., Feldt, M., & Dohlen, K. et al. 2008, in SPIE, Vol. 7014, 18 Chen, C. H., Mittal, T., Kuchner, M., et al. 2014, ApJS, 211, 25 Claudi, R. U., Turatto, M., & Gratton, R. G. et al. 2008, in SPIE, Vol. 7014, 3 Currie, T., Lisse, C. M., Kuchner, M., et al. 2015, ApJ, 807, L7 de Zeeuw, P. T., Hoogerwerf, R., & de Bruijne, J. H. J., et al. 1999, AJ, 117, 354 Dohlen, K., Langlois, M., & Saisse, M., et al. 2008, in SPIE, Vol. 7014, 3 Draper, Z. H., Duchene, G., Millar-Blanchaer, M. A., et al. 2016, ArXiv e-prints Esposito, T. M., Fitzgerald, M. P., & Graham, J. R. et al. 2014, ApJ, 780, 25 Gladman, B. 1993, Icarus, 106, 247 Jang-Condell, H., Chen, C. H., Mittal, T., et al. 2015, ApJ, 808, 167 Janson, M., Lafreni`ere, D., Jayawardhana, R., et al. 2013, ApJ, 773, 170 Kalas, P. G., Rajan, A., Wang, J. J., et al. 2015, ApJ, 814, 32 Kasper, M., Apai, D., Wagner, K., & Robberto, M. 2015, ApJ, 812, L33 Lagrange, A.-M., Langlois, M., Gratton, R., et al. 2016, A&A, 586, L8 Lee, E. J. & Chiang, E. 2016, ApJ, 827, 125 Lyra, W. & Kuchner, M. 2013, Nature, 499, 184 Macintosh, B. A. & Graham, J. R., et al. 2008, in SPIE, Vol. 7015, 701518 Mamajek, E. E., Meyer, M. R., & Liebert, J. 2002, AJ, 124, 1670 Marois, C., Correia, C., & V´eran, J.-P. et al. 2014, in IAU proc., Vol. 299, 48 -- 49 Marois, C., Macintosh, B., Barman, T., et al. 2008, Science, 322, 1348 Marois, C., Zuckerman, B., & Konopacky, Q. M. et al. 2010, Nature, 468, 1080 Marzari, F. 2014, MNRAS, 442, 1110 Mesa, D., Vigan, A., D'Orazi, V., et al. 2016, ArXiv e-prints Milli, J., Mouillet, D., Lagrange, A.-M., et al. 2012, A&A, 545, A111 Morrison, S. & Malhotra, R. 2015, ApJ, 799, 41 Mustill, A. J. & Wyatt, M. C. 2012, MNRAS, 419, 3074 Padgett, D. & Stapelfeldt, K. 2016, in IAU proc., Vol. 314, 175 -- 178 Pavlov, A., Feldt, M., & Henning, T. 2008, in ADASS XVII, Vol. 394, 581 Rameau, J., Chauvin, G., & Lagrange, A.-M., et al. 2013, ApJ, 779, L26 Schneider, G., Grady, C. A., Hines, D. C., et al. 2014, AJ, 148, 59 Soummer, R., Pueyo, L., & Larkin, J. 2012, ApJ, 755, L28 Su, K. Y. L., Rieke, G. H., Stapelfeldt, K. R., et al. 2009, ApJ, 705, 314 van Leeuwen, F. 2007, A&A, 474, 653 Vigan, A., Gry, C., Salter, G., et al. 2015, ArXiv e-prints Vigan, A., Moutou, C., Langlois, M., et al. 2010, MNRAS, 407, 71 Wisdom, J. 1980, AJ, 85, 1122 Appendix A: Details on the model fitting We report in Fig. A.1 the zones used to compute the χ2 when comparing the disk model to the data. The χ2 for the single belt modelling is computed in the region between the plain green el- lipse and the dashed green ellipse. It contains 1041 pixels, or 118 resolution elements. For the case of the model of the outer ring, the χ2 is computed in the region between the plain green ellipse and the black ellipse. It contains 3279 pixels, or 390 resolution elements. The Fig. A.2 show the χ2 minimization for each of the five free parameters of the single ring model (§ 3.2). The Figure A.3 !Figure 2. Left: PCA image of the disk. Middle: Residuals after subtraction of the best single ring model, with the same linear color scale as the left image. Right: best single ring model (unconvolved with the SPHERE PSF) in a log scale to show the fainter backward side of the disk. This model only tries to reproduce the inner ring and therefore the arc described in Fig. 1 is still clearly visible in the residuals. !! Figure 3: Left panels: best model of the double ring (top) and residuals after subtraction (bottom) . Right panels: best model of a single ring with a smooth halo (top) and residuals after subtraction. The 2 bottom panels have the same linear colour scale (also identical as that of Fig. 3) while the models have a logarithmic colour scale. !- I strongly recommend adding more discussion and interpretation related to the observed structures. What can be the origin of a single ring vs. ring+arc? Why if there is an external belt, only one part of the belt is observed in form of an arc? I imagine that the authors are concerned about the page limit, but I would suggest moving the section of detected point sources and limits to an appendix and increase the discussion part, if the authors desire to keep the manuscript as a letter. It would be interesting also to compare the results with the structures resolved in other debris disks (which are listed for instance in the conclusion of the abstract). As I mentioned before, a discussion about the possible relation with the observed candidate companions (potential disk truncation and sculpting the observed structures) can be very suitable for a discussion. !!You can discuss the presence of gas that can lead to rings as shown by Lyra and Kuchner. ALMA constraints can constrain such a scenario. Alternatively a planet might have carved a gap, or a resonance might have removed particles from un unstable configuration as is the case in the Saturn rings.!!Minor comments: !!-Conclusion in the abstract: it does not have meaningful information. It reads more like a general remark than a conclusion of the analysis done for the paper. !!M. Bonnefoy et al.: Belt(s) of debris resolved around the Sco-Cen star HIP 67497, Fig. A.1. Conservative SNR map showing the regions used to compute the chi square . Fig. A.3. Minimization of the free parameters for the model of a single belt and a halo (left column) and of the model with two belts (right column). Fig. A.2. Minimization of the free parameters of the single belt model. show the χ2 minimization for the free parameters of the models with two belts, or one belt and a halo (§ 3.3). The 1σ χ2 levels used to estimate the error bars on each parameters are reported in red. Appendix B: Detected point sources and limits We detect 10 candidate companions (CC) at large separations in the IRDIS H2 and H3 images. Their astrometry and predicted mass -- assuming they are bound -- are reported in Tab. B.1. All those point sources lie outside of the field-of-view of the IFS. None of them is aligned with the disk's position angle. The placement of the candidates in color-magnitude dia- grams can help to determine if they have both the luminosity and colors typical of cold substellar companions, or if they are rather background objects. Such diagrams have been created for the IRDIS filters (see a description in Mesa et al. 2016). We show the placement of the CCs of HIP 67497 in Figure B.2. The CCs #2, 6, and 9 have the luminosities typical of T dwarfs but the colors of MK dwarfs, the most common contaminants. This in- dicates that these CCs are likely to be background objects. All the remaining CCs, but the CC #1 and 2, are retrieved in the Fig. B.1. Detection limits (5σ) for the IFS (spectral PCA) and IRDIS (TLOCI) converted to mass. The candidate companions are reported in red. Top: full separation range. Bottom: zoom at inner working angles. HST/STIS (optical) images of Padgett & Stapelfeldt (2016) and are therefore likely background stars. We estimated the detection limits of the IFS via the injection of fake companions with flat spectra into the datacubes and used the COND evolutionary tracks to convert the derived contrasts to masses (Baraffe et al. 2003). The IRDIS detection limits were 5 ! ! 200400600Projected separation [AU]051015Mass [MJup]IRDIS - H2IFS-YJ12345678910200400600Projected separation [AU]051015Mass [MJup]20406080100Projected separation [AU]051015Mass [MJup]IRDIS - H2IFS-YJ1234567891020406080100Projected separation [AU]051015Mass [MJup]M. Bonnefoy et al.: Belt(s) of debris resolved around the Sco-Cen star HIP 67497, If we make the assumption that one planet on a circular orbit only is responsible for the gap, we find that this planet should have a mass of 20.6 MJup and semi-major axis a=90.5 au fol- lowing Morrison & Malhotra (2015). For the case of an eccentric planet, we used the expressions for the half-width of the chaotic zone ∆a found by Wisdom (1980) (for planet eccentricity ≤0.3) and Mustill & Wyatt (2012) (for eccentricity >0.3) as a basis: ∆a = 1.3 × (Mp/MS tar)2/7 × ap, (C.1) ∆a = 1.8 × (Mp/MS tar)1/5 × e1/5 × ap, (C.2) with Mp the planet mass, Mstar the primary star mass, and e the eccentricity of the disk. In our analysis, we consider the planet to arrive at periastron in the nearest point to the inner belt and at apoastron in the nearest point to the outer one. Then we substitute in the previous expressions for ∆a the value of the semi-major axis with the positions of periastron and apoastron in turns. In that case, we predict that planets with low eccentricities (e = 0.13) can have masses below our detection threshold. We also investigated the case of 2 or 3 equal-mass planets on circular orbits. We considered planets as near as possible to each others (maximum packing, Gladman 1993) in order to have yet a stable system but, at the same time, to obtain completely chaotic regions between the planets such that the disk material could not survive in it. For two planets on circular orbits, we followed the results of Marzari (2014) and found that 2.7 MJup equal-mass perturbers with semi-major axis of 73 and 105 au could clear the gap. For the case of a three planet systems with circular or- bits, the relations of Marzari (2014) give perturber masses of 0.14 MJup and semi-major axis of 64.4, 87.5, and 118.7 au. We considered a last configuration with two equal-mass planets on eccentric orbits, supposing again the maximum packing condi- tion, and following the equations C.1 and C.2. We find that low masses (0.1-0.4 MJup) with quite small eccentricities are needed to dig the entire gap. We recall that multiple assumptions are made here (equal mass pertubers, maximum packing). Therefore, we believe that a more in depth dynamical analysis of this system is requiered to understand the morphology of the disk (N-body simulations; beyond the scope of this letter). 1 Univ. Grenoble Alpes, IPAG, F-38000 Grenoble, France. CNRS, IPAG, F-38000 Grenoble, France e-mail: [email protected] 2 European Southern Observatory (ESO), Alonso de Crdova 3107, Vitacura, Casilla 19001, Santiago, Chile 3 Aix Marseille Universit, CNRS, Laboratoire dAstrophysique de Marseille, UMR 7326, 13388, Marseille, France 4 LESIA, Observatoire de Paris, CNRS, Universit Paris Diderot, Universit Pierre et Marie Curie, 5 place Jules Janssen, 92190 Meudon, France 5 INAF-Osservatorio Astronomico di Padova, Vicolo dellOsservatorio 5, Padova, Italy, 35122-I 6 DAS, Universidad de Chile, camino el observatorio 1515 Santiago, Chile Fig. B.2. Placement of the candidate companions from Tab. B.1 into color-magnitude diagrams. Table B.1. Photometry and astrometry of point sources. The pre- dicted mass if bound to the star is reported into the last column. # Band Contrast 4 2 3 5 1 H2 H3 H2 H3 H2 H3 H2 H3 H2 H3 H2 H3 H2 H3 H2 H3 H2 H3 10 H2 H3 6 7 8 9 13.33±0.09 13.71±0.13 14.62±0.12 14.42±0.13 13.24±0.08 13.10±0.07 10.21±0.07 10.12±0.06 11.20±0.07 11.13±0.07 14.58±0.16 14.12±0.12 12.13±0.07 12.05±0.08 11.58±0.07 11.48±0.07 13.92±0.09 13.87±0.09 13.32±0.09 13.12±0.08 PA (◦) 185.5±0.3 185.3±0.4 71.9±0.1 71.9±0.1 293.3±0.1 293.3±0.1 100.4±0.1 100.4±0.1 155.8±0.1 155.8±0.1 279.9±0.1 280.0±0.1 285.6±0.1 285.7±0.1 283.7±0.1 283.7±0.1 351.8±0.1 351.8±0.1 342.6±0.1 342.6±0.1 Sep (mas) 1092±6 1088±8 4258±7 4264±7 3996±3 3998±3 5626±2 5628±2 6398±2 6400±3 5721±6 5724±7 6462±3 6458±3 6444±3 6439±3 5376±4 5375±5 6539±5 6541±4 MPred (MJup) 2±1 2±1 1±1 1±1 2±1 2±1 7±2 7±2 4±1 4±1 1±1 1±1 3±1 3±1 4±1 4±1 2±1 2±1 2±1 2±1 estimated from the TLOCI coefficients and the local level of the noise. We report those limits in Fig. B.1. The data are sensitive to planetary-mass companions down to 15 au. Appendix C: Putative perturbers We investigate the presence of one, two and three giant planets as responsible for the gap between the two possible belts resolved around HIP 67497 using numerical simulations based on pub- lished analytical formulaes. A detailed description will be made in Lazzoni et al. 2016 in prep. 6 -4-202H2-H320181614121086MH2M0-M5M6-M9L0-L5L6-L9T0-T5T6-T8b/g/VL-G dwarfsHR8799bHR8799cHR8799dHR8799ePZ Tel B1RXS1609b2M1207bCD-352722BHN Peg BUScoCTIO 108BGJ 758bROSS 458C (T8.5p, Fe/H=0.2-0.3)BD+01_2920B (T8, Fe/H=-0.4)HD3651B (T7.5, Fe/H=0.2)G 204-39B (T6.5p, Fe/H=0.0)Gl 570B (T7.5, Fe/Hc=0.0)Wolf 940B (T8.5, Fe/Hc=0.1)CFBDSIR2149-0403 (T7.5p)51 Eri b12345678910>T8-4-202H2-H320181614121086MH2
1910.12889	1	1910	2019-10-28T18:00:36	Effect of wind-driven accretion on planetary migration	[ "astro-ph.EP" ]	Planetary migration is a key link between planet formation models and observed exoplanet statistics. So far the theory of migration has focused on the interaction of planets with an inviscid or viscously evolving disk. Turbulent viscosity is thought to be the main driver of disk evolution and is known to affect the migration process. Recently, the topic of wind-driven accretion is experiencing a renaissance, as evidence is mounting that PPDs may be less turbulent than previously thought, and 3-D non-ideal MHD modeling of the wind-launching process is maturing. Aim: We wish to investigate how wind-driven accretion may affect migration. We aim for a qualitative exploration of the main effects, rather than a quantitative prediction. Methods: We perform 2-D hydrodynamic simulations with the FARGO3D code in the $(r,\phi)$-plane. The vertical coordinate and the launching of the wind are not treated explicitly. Instead, the torque of the wind onto the disk is treated using a simple 2-parameter formula treating the wind mass loss rate and the lever arm. Results: We find that the wind-driven accretion process has a different way of replenishing the co-orbital region than the viscous accretion. The former always injects mass from the outer edge of the co-orbital region and removes mass from the inner edge, while the latter injects or removes mass from the co-orbital region depending on the radial density gradients in the disk. The migration behavior can differ very much and under certain conditions it can drive rapid type-III-like outward migration. We derive an analytic expression for the parameters under which this outward migration occurs. Conclusion: If wind-driven accretion plays a role in the secular evolution of PPDs, migration studies have to include this process as well, because it can strongly affect the resulting migration rate and direction.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. (will be inserted by hand later) 9 1 0 2 t c O 8 2 . ] P E h p - o r t s a [ 1 v 9 8 8 2 1 . 0 1 9 1 : v i X r a Effect of wind-driven accretion on planetary migration C.N. Kimmig1, C.P. Dullemond1, W. Kley2 (1) Zentrum fur Astronomie, Heidelberg University, Albert-Ueberle-Str. 2, 69120 Heidelberg, Germany (2) Institut fur Astronomie und Astrophysik, Universitat Tubingen, Auf der Morgenstelle 10, 72076 Tubingen, Germany October 30, 2019 Abstract. Context: Planetary migration is a key link between planet formation models and observed exoplanet statistics. So far the theory of planetary migration has focused on the interaction of one or more planets with an inviscid or viscously evolving gaseous disk. Turbulent viscosity is thought to be the main driver of secular evolution of the disk, and it is known to affect the migration process for intermediate to high-mass planets. Recently, however, the topic of wind-driven accretion is experiencing a renaissance, now that as evidence is mounting that protoplanetary disks may be less turbulent than previously thought, and 3-D non-ideal magnetohydrodynamic modeling of the wind-launching process is maturing. Aim: We wish to investigate how wind-driven accretion may affect planetary migration. We aim for a qualitative exploration of the main effects, rather than a quantitative prediction. Methods: We perform 2-D hydrodynamic planet-disk interaction simulations with the FARGO3D code in the (r, φ)-plane. The vertical coordinate in the disk, and the launching of the wind, are not treated explicitly. Instead, the torque of the wind onto the disk is treated using a simple 2-parameter formula. The parameters are: the wind mass loss rate and the lever arm. Results: We find that the wind-driven accretion process has a different way of replenishing the co-orbital region than the viscous accretion process. The former always injects mass from the outer edge of the co-orbital region, and always removes mass from the inner edge, while the latter injects or removes mass from the co-orbital region depending on the radial density gradients in the disk. As a consequence the migration behavior can differ very much, and under certain conditions it can drive rapid type-III-like outward migration. We derive an analytic expression for the parameters under which this outward migration occurs. Conclusion: If wind-driven accretion plays a role in the secular evolution of protoplanetary disks, planetary migration studies have to include this process as well, because it can strongly affect the resulting migration rate and migration direction. Key words. protoplanetary disks, planet-disk interactions -- accretion, accretion disks 1. Introduction Planetary migration is an integral part of the theory of planet formation. A planet, once it is formed, can migrate to vastly different radial locations, due to its gravitational interaction with the protoplanetary disk. As a result, the orbital elements of observed exoplanets may not reflect the location where they were born. Any model prediction of exoplanetary orbital statistics must therefore include a treatment of the migration process. Unfortunately, the process of planetary migration is a complex affair (see, e.g., Kley & Nelson 2012, and refer- ences therein). While the effect of the Lindblad torques is fairly well understood (Tanaka et al. 2002), the torques ex- erted by the gas in the co-orbital region has turned out to be hard to predict. They depend on the entropy gradient (Baruteau & Masset 2008), the radiative cooling efficiency (Kley & Crida 2008), the planet mass (Ward 1997), the viscosity of the disk (Masset 2001), and even on the mo- tion of the planet itself (Masset & Papaloizou 2003). Among the above mentioned effects, the role of turbu- lent viscosity needs particular scrutiny, because in recent years doubts have been raised about whether protoplan- etary disks are indeed as turbulent as they were believed to be. Evidence against strong turbulence comes from the velocity dispersion inferred from CO lines (Flaherty et al. 2015, 2018), the observed small scale height of the dust rings in HL Tau (Pinte et al. 2016), as well as from the comparison of planet-disk interaction models with recent ALMA observations (Zhang et al. 2018). Also theoretical considerations about the degree of ionization of the gas in such disks have suggested that the magnetorotational instability, which is a potent driver of turbulence, may not be able to operate in the disk, except in the very up- per layers (Gammie 1996). Instead, a magnetocentrifugal 2 Kimmig, Dullemond & Kley: Winds and migration wind is likely to be launched (Bai & Stone 2013), which in turn exerts a torque back onto the disk, and thus drives inward gas motion in the upper layers of the disk, or in other words: causes accretion within the disk (Simon et al. 2013). If wind-driven accretion becomes the new paradigm and replaces the classic viscous disk theory, then also the theory of planetary migration has to be reconsidered. In the viscous disk picture, the turbulent viscosity has a strong influence on the planetary migration (Masset 2001). When a planet opens a gap, the viscous evolution of the disk acts against this by feeding gas from both the in- ner and the outer disk back into the gap. The depth of the gap, and therefore the amount of material in the co- orbital (corotation) region, thus depends on the equilib- rium between these two counteracting effects. Weak tur- bulence leads to a deep gap and a weak corotation torque. Strong turbulence, on the other hand, keeps feeding gas into the corotation region, leading to a substantial coro- tation torque. In contrast to the viscous disk theory, in the wind- driven accretion picture the radial inward motion of the gas in the disk is much more laminar. Accretion will be only inward, irrespective of the radial gradients of the den- sity. This is because according to the classic Blandford & Payne (1982) model, the acceleration of the wind is driven by the injection of angular momentum into the wind; an- gular momentum that is extracted from the disk. If we now insert a gap-opening planet into such a disk, the wind-driven gas motions in the disk will inject mass into the gap only from the outside. At the inner edge of the gap, in stark contrast to the viscous disk model, the wind- driven accretion will remove gas from the gap. In math- ematical terms the difference is that the viscous model drives accretion as a kind of diffusion process (which can transport gas both inward and outward, dependent on the density gradient), while the wind-driven model drives ac- cretion as an advection process (which transports gas only inward). In this paper we wish to explore how the wind-driven accretion process affects the migration of a planet. The purpose is to gain understanding, not to obtain quanti- tative numbers. We therefore deliberately keep the de- scription of the wind-driven accretion process extremely simple, and focus on the gas motions in the plane of the disk. The structure of this paper is as follows: we first de- scribe the model assumptions, the equations we solve, and the numerical methods we employ (Section 2). Then we describe the setup of the simulations in Section 3 and show the results of a series of model calculations we performed in Section 4. We discuss the meaning of these results, and the caveats of the model in Section 5, and finish with a conclusion (Section 6). Fig. 1. Configuration of the magnetic field in the Blandford & Payne (1982) model (schematic). The blue lines represent the magnetic field lines, the dashed line marks the Alfv´en surface and rA and rF represent the Alfv´en point and the foot point of the (thick) magnetic field line, respectively. Inspired by Spruit (1996). 2. Model For the hydrodynamic simulations we use the code FARGO3D by Ben´ıtez-Llambay & Masset (2016). We in- vestigate the effect of the magneto-centrifugal wind in a two dimensional disk: radial and azimuthal. For simplifi- cation, we implement only the effect of the magnetic wind rather than the magnetic field itself. The wind is mag- netocentrifugally accelerated (Blandford & Payne 1982), meaning that it extracts net angular momentum from the remaining material in the disk, and thus effects a radial in- ward drift of the gas (e.g., Ferreira 2008; Bai 2016; Suzuki et al. 2016). The resulting radial velocity vr can then be written as vr = −2r Σwind Σ (λ − 1), (1) with the convention that a negative velocity implies an inward flow. Here, Σwind is the mass outflow from the disk, Σ the surface density and (cid:18) rA (cid:19)2 rF λ = , (2) the lever arm of a magnetic field with rA and rF being the Alfv´en point and the foot point of the magnetic field line, respectively (see Fig. 1). Two parameters determine the strength of the mag- netic wind: the magnetic lever arm λ and the mass out- flow Σwind. As an assumption for the parameters, we keep the lever arm constant throughout the disk while we scale the mass outflow rate Σwind proportional to the surface density Σwind = b ΩK 2π Σ, (3) xxKimmig, Dullemond & Kley: Winds and migration 3 with ΩK as Keplerian angular velocity. The mass loss parameter b is defined such that it denotes the propor- tional outflow from the disk per orbital period. With this parametrization, Eq. (1) turns into vr = − ΩKr π b (λ − 1). (4) Hence, in our model we have vr ∝ r−0.5. This is steeper than the velocity profiles in the models by Hasegawa et al. (2017) and Suzuki et al. (2016), for which they approxi- mately finds vr ∼ r−0.215 and vr ∼ r−0.25, respectively. We implement this inward drift as an azimuthal torque (5) density on the disk to decelerate the gas: Γ = Σwind ΩKr2 (λ − 1) Kr2 = − Ω2 bΣ (λ − 1). 2π In addition to this torque, we add − Σwind as a sink term in the continuity equation. For a steady radial flow, we need the accretion rate of the gas within the disk to be constant. The accretion rate M = −2πr2vrΣ depends on the radius M ∝ r1/2−p with the assumption of a power law for the surface den- sity Σ(r) = Σ0 · r−p. Therefore, a steady-state solution is possible for p = 1/2. 3. Setup To illustrate the effect of the wind-driven torque on plan- etary migration, we place a Mdisk = 10−2 M(cid:12) disk around a solar mass star. The disk has a surface density profile Σ ∝ r−1/2 and spans the radial range between 0.52 au and 26 au. With this disk mass, powerlaw and inner/outer boundary, we have Σ(5.2au) = 70.4 g/cm2. The radial temperature structure of the disk is taken to be such, that the aspect ratio of the disk is Hp/r = 0.05, where Hp = cs/ΩK, with cs the isothermal sound speed. We set the turbulent viscosity parameter to αturb = 0 (Shakura & Sunyaev 1973). We insert a planet at r = r0 = 5.2 au, and during the first ten orbits we allow the mass of the planet to grow linearly from 0 to the final planet mass Mp. We choose two planetary masses: Mp = 100 M⊕ (about a Saturn mass), and Mp = 1MJup. These amount to q = Mp/M∗ ratios of 3 × 10−4 and 10−3. The smoothing length of the planet potential is s = 0.6 Hp. The sim- ulations with FARGO3D are dimensionless, so that the results can be scaled to other stellar masses, provided q stays the same. For the (r, φ) grid we choose 512×656 grid cells. The radial grid is logarithmically spaced. We run the model for 4000 orbits at r = r0. We employ the GPU acceler- ated mode of FARGO3D. For the boundary conditions we choose the KEPLERIAN2DDENS option of FARGO3D, which is an open boundary condition where the surface density is extrapolated with a powerlaw slope to the ghost cells. As initial condition for the radial velocity vr we use Eq. (4). The initial condition for the azimuthal velocity vφ is set to Kepler rotation corrected by the pressure gradient. as loss parameter We vary the mass b = 10−6, 10−5, 10−4, 10−3, 10−2. The fiducial value of the magnetic lever arm is taken to be λ = 2.25 (i.e., rA/rF = 1.5). To focus on an important parameter regime, we add simulations with b = 2.15 × 10−4, 3.16 × 10−4, 4.64 × 10−4, 6.81 × 10−4, 1.47 × 10−3, 3.16 × 10−3 for the Saturn planet and b = 3.16 × 10−4, 6.81 × 10−4, 1.47 × 10−3, 2.15× 10−3, 3.16× 10−3 for the Jupiter planet. We also experiment with λ = 9, 36, 81 (i.e., rA/rF = 3, 6, 9, respectively). For completeness, we can compute the accretion rate M for our models, given our surface density profile Σ(r) and the equation for the radial velocity (Eq. 4): (cid:112) M ≡ 2πΣr(−vr) = 2Σ0r0.5 0 GM∗ b(λ − 1) = 2.27 × 10−4 b(λ − 1) M(cid:12)/yr (6) M = 2.8 × For λ = 2.25 and b = 10−4 this yields 10−8 M(cid:12)/yr. Comparing this to the viscous disk model with M = 3πνΣ, we would arrive at a α-value of α = 10−2. To test our modifications to FARGO3D, we ran a model without planet, but with a wind. For the simpli- fied case of torque without corresponding mass loss, we can test the numerical result against the analytic solution for vr (Eq. 4). We find a good match. In the other limit- ing case of mass loss without corresponding torque (i.e., λ = 1) an analytic solution of an exponentially dropping surface density can be found. Also here the numerical re- sult matches it well. 4. Results 4.1. Reference case: no wind In order to be able to compare our results to a reference case later on, we first run simulations for the two differ- ent planet masses in a non-viscous disk without magnetic winds. Without viscosity and magnetic winds, the planets should not migrate. Due to the low viscosity, the planets open a gap in the surface density (see Fig. 2). This gap formation takes longer for smaller planet masses. Although the planets slightly change their semi-major axis (see Fig. 3), they do not migrate significantly without viscosity and magnetic winds. The Saturn-like planet (Mp = 100 M⊕) migrates in the first 500 orbits, then stalls. This planet takes longer to clear a gap than the Jupiter-like planet (Fig. 2). The gas exerts torques on the planet and it migrates first in- ward. Once the planet has opened a gap, the corotation and Lindblad torques are reduced. As a result, the planet slows down and comes to a halt. This halting of migration in low-viscosity disks is a known phenomenon: see, e.g., Rafikov (2002), Li et al. (2009) and Yu et al. (2010). For the Jupiter mass planet (Mp = 300 M⊕), the trend looks more complicated. In the first one hundred orbits the planet migrates inward, for the same reason as the Saturn-like planet. It then migrates outward for a period 4 Kimmig, Dullemond & Kley: Winds and migration planet is closer to the inner disk than on the other side. This exerts an asymmetric torque on the planet which ei- ther accelerates or decelerates the planet, causing it to migrate. Thus, the inner disk could also explain the unex- pected behavior of the planet. 4.2. Model with wind-driven accretion Now we switch to the wind-driven accretion. We set λ = 2.25 and vary b = 10−6, 10−5, 10−4, 10−3, 10−2. The resulting disk surface density structures are shown for the Saturn and Jupiter mass planet in Fig. 4 and 5, respectively. For low values of b (the cases b = 10−6, 10−5, 10−4) we see that the planet opens a gap, as expected. In the outer edge of the gap, an elongated vortex forms, which is a known effect for low-viscosity disks (Koller et al. 2003; Li et al. 2000; Ataiee et al. 2013). One can also observe the formation of secondary gaps in the inner disk, consistent with the findings of Bae et al. (2017) and Dong et al. (2017) for low-viscosity disks. These secondary rings take many orbits to form and are most pronounced in the 4000 orbit panels. For the Saturn mass planet models we find clear inward migration for b ≤ 10−4. The planet location (shown in the Figure with a cross) after 4000 orbits is well within the original orbit (shown in the Figure as the dashed circle). For the Jupiter mass planet models the inward migration is weaker, but for the case b = 10−4 it is nevertheless clearly seen (Fig. 5). For large values of b (the cases b = 10−3, 10−2 for the Saturn mass planet, and b = 10−2 for the Jupiter mass planet) we observe a totally different behavior. The gap is opened only partially, and asymmetrically. While the co-orbital region trailing the planet is evacuated, the co- orbital region leading the planet still contains substantial amounts of gas. This leads to a positive corotation torque, which adds angular momentum to the planetary orbit. As a result, the planet rapidly migrates outward. We will discuss this mechanism in more detail in Section 5. In Figs. 6 and 7 we show the migration history of the planets as a function of time. For some wind strengths the Saturn mass planet clearly migrates inward or outward, in comparison to the reference case. It migrates outward for b = 10−2 and b = 10−3, inward for b = 10−4, and slightly inward for b = 10−5. In the case of b = 10−6, the planet initially migrates further than in the reference case, but stalls after 1000 orbits. In both outward migrating cases, the planet stops at 4.5 r0, which is caused by the outer boundary. The Jupiter mass planet migrates outward only in the case of b = 10−2, and inward for b = 10−4. In the case of b = 10−5 and b = 10−6, the planet does not migrate significantly. Particularly, it does not migrate outward as in the reference case without magnetic wind. With a mass loss parameter of b = 10−3, the planet performs an unexpected periodic inward and outward mi- Fig. 2. Results of the inviscid reference models without wind-driven accretion: the surface density of the disk for the two different planet masses, at 100 and 4000 orbits. The white cross denotes the current position of the planet, the dashed white circle marks its initial semi-major axis at r0 = 5.2 au. Fig. 3. Results of the inviscid reference models without wind-driven accretion: shown is the time evolution of the semi-major axes of the embedded planets. of 2500 orbits, migrates back inward for 500 orbits, and stalls there. This migration takes place only in a small scope of 0.05 r0 = 0.26 au. A possible explanation for this unexpected behavior is interaction with vortices that form in the disk, which can be seen in Fig. 2. Such vortices were also found by Koller et al. (2003). The vortices result from perturbations from the planet in the surface density of the disk, potentially caused by the short ramp-up time of the planet mass within ten orbits. These vortices would diffuse in a viscous disk on short time scales. In the non-viscous disk, however, they stay much longer and accelerate or decelerate the planet. They therefore influence its semi- major axis. Another factor influencing its semi-major axis could be the inner disk. In the surface density (Fig. 2), the inner disk becomes eccentric due to the open inner bound- ary condition. The planet's orbit, however, stays nearly circular: it's eccentricity staying below about 0.02. The eccentricity of the inner disk causes that on one side, the 1000200030004000time[orbitsatr0]0.80.91.0semi-maj.axisa[r0]100M⊕1000200030004000time[orbitsatr0]1MjupKimmig, Dullemond & Kley: Winds and migration 5 Fig. 4. Surface density maps for the wind-driven accretion case with a Saturn mass planet. The three columns are for different times. The five rows are for the different values of the mass loss parameter b. The dashed white circle marks the initial orbit of the planet. The cross denotes its current location. gration (see Fig. 7). The surface density map of this model (shown in Fig. 5 in the fourth row), shows a stronger vor- tex in the outer disk than in the other simulations. It does not vanish by time as in the other cases. We have exper- imented with a higher resolution (1024×1312 grid cells) and found that while the vortex remains, the oscillation vanishes, and that the planet migrates somewhat further inward. It is therefore likely that the oscillations we ob- serve are a numerical artifact. We suspect that this numerical artifact occurs when the simulation is set up in the transition regime between inward and outward migration. In this regime the planet is very sensitive to the different torques caused by the gas surface density which could cause the oscillations. To test this, we perform more simulations with the same setup as before, except for the b-parameter. We choose a finer log- arithmic sampling for b to cover the transition regime for both planets, as shown in Fig. 8. We find for the Jupiter 6 Kimmig, Dullemond & Kley: Winds and migration Fig. 5. As Fig. 4, but now for the Jupiter mass planet. mass planet that the setup with b = 10−4 indeed turns out to be the transition between inward and outward mi- gration. For the Saturn planet, the transition takes place at b = 3.16×10−4 in which we observe no migration of the Saturn planet and a light oscillation as in the simulation with the Jupiter planet. We also experiment with different values for the mag- netic lever arm while we keep the mass loss parameter at b = 10−4, as shown in Fig. 9. The torque on the disk is proportional to b(λ − 1), so that the effect of increasing λ is expected to be similar to that of increasing b. However, increasing λ does not increase the wind mass loss rate Σwind, while increasing b does. Hence, some differences in the results are expected. In these simulations, we observe inward migration for low lever arms λ = 2.25, 9 and out- ward migration for high lever arms λ = 36, 81. As in the investigation of the b-parameter, for low wind strength the planet migrates inward while it migrates outward for high wind strengths. Kimmig, Dullemond & Kley: Winds and migration 7 Fig. 6. The planetary migration for the wind-driven accretion case with Saturn mass planet, for the various values of b. Dashed line is the reference model without wind-driven accretion. Fig. 7. As Fig. 6, but now for the Jupiter mass planet. Fig. 8. Direction of migration in simulations with differ- ent mass loss parameters b for both planet masses. The lever arm is λ = 2.25 in all simulations. For the models marked as transition, no clear inward or outward migra- tion was observed. 5. Discussion 5.1. Interpretation of the results Our models show that for low values of b (weak wind loss) the planet migrates inward, while for high values of b (strong wind loss) the planet migrates outward. In the low b case the planet manages to open a gap and clear the co-orbital region of its gas (see, e.g., Kley & Nelson 2012; Kanagawa et al. 2018). Therefore, there are no corotation torques acting on the planet and it experi- ences only the Lindblad torques. The migration is inward and to a certain extent similar to the standard type II viscous migration, in the sense that gap formation plays a key role in the migration process. We therefore call it type IIw migration. The torque by the magnetic wind onto the disk gas causes the disk gas to gradually move in- ward. Consequently, also the inner and outer edges of the gap move inward. Suppose the planet stays on its posi- tion: then the outer edge moves closer to the planet, while the inner edge moves further away from it. The Lindblad torques from the outer gap edge then dominate over the 01000200030004000time[orbitsatr0]012345semi-maj.axisa[r0]b=10−6b=10−5b=10−4b=10−3b=10−201000200030004000time[orbitsatr0]0.20.40.60.81.0referencew/owindb=10−6b=10−5b=10−4b=10−3b=10−201000200030004000time[orbitsatr0]012345semi-maj.axisa[r0]b=10−6b=10−5b=10−4b=10−3b=10−201000200030004000time[orbitsatr0]0.50.60.70.80.91.01.1referencew/owindb=10−6b=10−5b=10−4b=10−3b=10−210−510−410−310−2masslossparameterbJupiterSaturninwardtransitionoutward8 Kimmig, Dullemond & Kley: Winds and migration Fig. 9. As Fig. 7, but now varying the lever arm λ, keeping b = 10−4 in all simulations. This is studied for the Jupiter mass planet. and therefore the gas always moves inward. Thus, type IIw also always has to point inward. Now let us turn to the behavior for large b, where the planet migrates outward at a high speed. In this case, the radial inward motion of the gas is rapid enough that it can enter the horseshoe streamlines, as shown in Fig. 11. This leads to a strong asymmetry in the horseshoe re- gion. When gas enters the corotation region in front of the planet, it performs a horseshoe orbit and moves to- ward the planet. At the U-turn point of the horseshoe orbit an excess of mass occurs. While the gas turns its direction at this point, it moves closer to the star, which means that it loses angular momentum. This angular mo- mentum is transferred to the planet, which in return gains it. Behind the planet, a defect of mass occurs, because the horseshoe orbit transports the gas away from the planet. A key to maintaining this defect is that the gas that per- formed the U-turn in front of the planet now finds itself close to the inner edge of the gap. Due to the rapid inward motion caused by the wind, this gas then quickly leaves the co-orbital region again by entering the inner disk. This means that this gas will not librate all the way to the back side of the planet, leaving this region devoid of gas. In ter- minology of standard migration theory, one can say that this process avoids saturation of the corotation torque. Since only the U-turn in front of the planet is pop- ulated with gas, the planet only gains angular momen- tum, not loses it. The planet thus migrates outward. This outward migration mechanism is very similar to type III "runaway migration" (Masset & Papaloizou 2003). In our case, however, the migration is always outward, while in the case of type III migration, the direction of the initial migration of the planet determines the run-away migra- tion direction. We call this type IIIw migration. We estimate the parameters for the transition from type IIw migration to type IIIw by comparing the libra- tion timescale τlib, which is the time a gas parcel takes to Fig. 10. Migration of the planets compared to the gas motion, for the low b case. Lindblad torques from the inner gap edge, and thus push the planet inward to a new equilibrium position. If the gas drift was the only process determining the migration, the migration rate of the planet could be de- scribed by the radial velocity of the gas due to the wind vmig = vr. The planet would then be coupled to the disk, and essentially follow the disk gas as it accretes inward. A closer inspection of the results, however, shows that some gas leaks through the gap, making the type IIw migration somewhat slower than the gas inward motion. This can be seen in Fig. 10. Similar effects are known to occur in migration of gap-opening planets in viscous disks (Duffell et al. 2014; Durmann & Kley 2017). It is interesting to note that, while type IIw migration seems similar to type II migration, there is a fundamental difference between the two: In a viscous disk, the gas mo- tion can be both inward or outward. In particular the very outer disk regions expand outward (the typical Lynden- Bell & Pringle disk evolution). This would lead to out- ward migration. In the case of wind-driven accretion, the wind always removes angular momentum from the disk, 01000200030004000time[orbitsatr0]012345semi-maj.axisa[r0]λ=2.25λ=9λ=36λ=8101000200030004000time[orbitsatr0]0.20.40.60.81.0referencew/owindλ=2.25λ=9λ=36λ=8101000200030004000time[orbitsatr0]0.00.20.40.60.81.0semi-maj.axisa[r0]100M⊕1Mjupr(t)ofthegasKimmig, Dullemond & Kley: Winds and migration 9 5.2. Caveats One caveat of our flat disk model is that we do not treat the possible vertical stratification of the coupling of the magnetic fields to the gas. It is believed that the well-shielded midplane regions of protoplanetary disks are "dead", in the sense that magnetic fields decouple from the gas. The surface layers are likely sufficiently ionized to be coupled to the field lines. This means that the torque the magnetocentrifugal wind exerts onto the disk is only ap- plied to the very upper layers of the disk, not the full ver- tical extent of the disk. We argue that this does not make a big difference for our model, because by the conservation of angular momentum, the accretion rate driven by this mechanism only depends on the torque. The only effect of the dead midplane zone is that all of this wind-driven accretion will then have to be carried by the surface layer material. For the present model it is irrelevant whether the accretion is carried by the full column of the disk or only by the surface layers. Furthermore, our parametrization of the winds assume a constant lever arm and mass loss parameter throughout the whole disk. A better approximation would be a ra- dial dependency of both parameters. The parameters are not well known so far, therefore the used values are only estimates. The resolution of our models is also an issue. Especially if the mass of the planet is small, the width of the co- orbital region is narrow, requiring high spatial resolution. We have already seen the effect of resolution on the results in Section 4, when we discussed the case of 1 MJup and b = 10−3. The biggest caveat of our simplified approach is that our model does not set the wind rate self-consistently. The wind rate and lever arm, and thereby the torque onto the disk, have to be completely parametrized. Improvement requires a detailed 3-D magnetohydrodynamic simulation of the driving of the wind. Such models exist (see, e.g., B´ethune et al. 2017; Wang & Goodman 2017; Wang et al. 2019), but they are costly. To calculate the effect on plane- tary migration, many thousands of orbits have to be com- puted, which is a challenge for such models. Nevertheless, in the future this is the path that has to be taken. 5.3. Comparison to other work The effect of wind mass loss of the disk on type I migration has been studied before by, e.g., Ogihara et al. (2015). In that study, however, the main effect was the change of the radial profile of the disk surface density Σ(r), which, using the standard type I migration rate formula (e.g., Paardekooper et al. 2011) leads to a modified migration rate. In contrast, in our paper we do not study how the changed Σ(r) profile affects the migration rate (though it is, in a way, obtained for free), but instead focus on the computation of the torque itself. In addition to wind-driven accretion as a replacement for the classic α-viscosity model, there may be other Fig. 11. Cartoon of how the wind-driven inward motion of the gas through the co-orbital region leads to a strongly asymmetric mass distribution, with a mass excess in front of the planet and a mass deficit trailing the planet. complete a horseshoe orbit, to the time it takes to radially cross the horseshoe region driven by the magnetic wind, which we call passing time τpass. The libration timescale is (Kley & Nelson 2012) τlib = 8π 3ΩK,p rp ∆r , (7) where rp is the distance from the planet to the star, ΩK,p is the Kepler frequency at rp and ∆r is the half-width of the horseshoe region. We can estimate the passing time as τpass = 2∆r vr(rp) = 2π ΩK,p b(λ − 1) ∆r rp , using Eq. 4. Comparing the timescales results in (cid:18) ∆r (cid:19)2 rp τpass τlib = 3 4 1 b(λ − 1) . (8) (9) A shorter passing time and a longer libration time means that the gas is faster to cross the corotation region than to complete the horseshoe orbit. This keeps the corota- tion torque unsaturated and we expect outward migra- tion. The half-width of the horseshoe region is found to be ∆ r = C() rp and as smoothing length, q as the planet-to-star mass ratio and h as scale height (see Kley & Nelson 2012). We use this to define a criterion K = τpass/τlib for C() = 1 (cid:112)q/h, with C() as factor of order unity K = 3 4 q h 1 b(λ − 1) . (10) In our simulations, we find that the transition regime be- tween inward and outward migration occurs for K ≈ 10. For K (cid:38) 10 we find inward migration and for K (cid:46) 10 outward migration. Mass Excess10 Kimmig, Dullemond & Kley: Winds and migration drivers of accretion. For instance, McNally et al. (2017) showed that the presence of a vertical magnetic field in the dead zone of a disk could, via the Hall effect, lead to the formation of strong spiral-shaped magnetic fields in the plane of the disk. These fields transport angular mo- mentum within the disk and lead to accretion, even though the disk is laminar. They call this situation a magnetically torqued dead zone. When a planet is inserted in such a disk, this can lead to similar asymmetric filling/depletion of the co-orbital region and the corresponding type-III-like migration effects as we find in our paper (McNally et al. 2017, 2018). Given that, in our model, we need a vertical magnetic field to drive a wind, it is very well possible that both effects act simultaneously. 6. Conclusions Our models show that the effect of magnetocentrifugal wind-driven accretion on planet migration can be pro- found. For very strong winds, it can even lead to rapid outward migration akin to type III migration (which we call type IIIw migration). In this case, however, the direc- tion of migration is deterministic: it does not depend on an initial "seed motion" of the planet. The parameter range for which we find rapid outward migration may occur is, however, coupled to a rapid evolution of the disk. We find that lower mass planets are more prone to the type IIIw ef- fect, and thus more easily migrate outward. We speculate that this type IIIw outward migration mechanism may be a possible origin of the intermediate mass planets at large radii that are thought to be the cause of the multi-ringed ALMA disks (Huang et al. 2018; Zhang et al. 2018). These planets would then have formed at much smaller semi- major axes, and then efficiently transported outward to their final location. Acknowledgements. We thank Leonardo Krapp, Philipp Weber and Thomas Rometsch for assistance with the required mod- ifications to FARGO3D, and Hubert Klahr for useful com- ments. We also thank the referee, John Chambers, for his constructive and helpful comments that substantially im- proved the paper. Part of this work was supported by DFG grant DU 414/18-1 and KL 650/27-1 within the Priority Programme "Exploring the Diversity of Extrasolar Planets" (SPP 1992) and DFG research group FOR 2634 "Planet Formation Witnesses and Probes: Transition Disks" under grant DU 414/23-1 and KL 650/30-1. For the simulations per- formed on the BwForCluster BinAC, we also acknowledge sup- port by the High Performance and Cloud Computing Group at the Zentrum fur Datenverarbeitung of the University of Tubingen, the state of Baden-Wurttemberg through bwHPC and the German Research Foundation (DFG) through grant no INST 37/935-1 FUGG. References Ataiee, S., Pinilla, P., Zsom, A., et al. 2013, Astronomy & Astrophysics, 553, L3 Bae, J., Zhu, Z., & Hartmann, L. 2017, ApJ, 850, 201 Bai, X.-N. 2016, ApJ, 821, 80 Bai, X.-N. & Stone, J. M. 2013, ApJ, 769, 76 Baruteau, C. & Masset, F. 2008, ApJ, 672, 1054 Ben´ıtez-Llambay, P. & Masset, F. S. 2016, ApJS, 223, 11 B´ethune, W., Lesur, G., & Ferreira, J. 2017, A&A, 600, A75 Blandford, R. D. & Payne, D. G. 1982, MNRAS, 199, 883 Dong, R., Li, S., Chiang, E., & Li, H. 2017, ApJ, 843, 127 Duffell, P. C., Haiman, Z., MacFadyen, A. I., D'Orazio, D. J., & Farris, B. D. 2014, The Astrophysical Journal Letters, 792, L10 Durmann, C. & Kley, W. 2017, A&A, 598, A80 Ferreira, J. 2008, New Astronomy Reviews, 52, 42 Flaherty, K. M., Hughes, A. M., Rosenfeld, K. A., et al. 2015, ApJ, 813, 99 Flaherty, K. M., Hughes, A. M., Teague, R., et al. 2018, ApJ, 856, 117 Gammie, C. F. 1996, ApJ, 457, 355 Hasegawa, Y., Okuzumi, S., Flock, M., & Turner, N. J. 2017, ApJ, 845, 31 Huang, J., Andrews, S. M., Dullemond, C. P., et al. 2018, The Astrophysical Journal Letters, 869, L42 Kanagawa, K. D., Tanaka, H., & Szuszkiewicz, E. 2018, ApJ, 861, 140 Kley, W. & Crida, A. 2008, Astronomy & Astrophysics, 487, L9 Kley, W. & Nelson, R. P. 2012, ARAA, 50, 211 Koller, J., Li, H., & Lin, D. N. C. 2003, The Astrophysical Journal, 596, L91 Li, H., Finn, J. M., Lovelace, R. V. E., & Colgate, S. A. 2000, ApJ, 533, 1023 Li, H., Lubow, S. H., Li, S., & Lin, D. N. C. 2009, The Astrophysical Journal Letters, 690, L52 Masset, F. S. 2001, ApJ, 558, 453 Masset, F. S. & Papaloizou, J. C. B. 2003, ApJ, 588, 494 McNally, C. P., Nelson, R. P., & Paardekooper, S.-J. 2018, MNRAS, 477, 4596 McNally, C. P., Nelson, R. P., Paardekooper, S.-J., Gressel, O., & Lyra, W. 2017, MNRAS, 472, 1565 Ogihara, M., Morbidelli, A., & Guillot, T. 2015, A&A, 584, L1 Paardekooper, S.-J., Baruteau, C., & Kley, W. 2011, MNRAS, 410, 293 Pinte, C., Dent, W. R. F., M´enard, F., et al. 2016, ApJ, 816, 25 Rafikov, R. R. 2002, ApJ, 572, 566 Shakura, N. I. & Sunyaev, R. A. 1973, A&A, 24, 337 Simon, J. B., Bai, X.-N., Armitage, P. J., Stone, J. M., & Beckwith, K. 2013, ApJ, 775, 73 Spruit, H. C. 1996, arXiv e-prints, astro Suzuki, T. K., Ogihara, M., Morbidelli, A., Crida, A., & Guillot, T. 2016, A&A, 596, A74 Tanaka, H., Takeuchi, T., & Ward, W. R. 2002, ApJ, 565, 1257 Wang, L., Bai, X.-N., & Goodman, J. 2019, ApJ, 874, 90 Wang, L. & Goodman, J. J. 2017, ApJ, 835, 59 Ward, W. R. 1997, Icarus, 126, 261 Kimmig, Dullemond & Kley: Winds and migration 11 Yu, C., Li, H., Li, S., Lubow, S. H., & Lin, D. N. C. 2010, The Astrophysical Journal, 712, 198 Zhang, S., Zhu, Z., Huang, J., et al. 2018, The Astrophysical Journal Letters, 869, L47
1011.2338	1	1011	2010-11-10T10:52:13	Astrometric search for a planet around VB 10	[ "astro-ph.EP" ]	We observed VB 10 in August and September 2009 using the FORS2 camera of the VLT with the aim of measuring its astrometric motion and of probing the presence of the announced planet VB 10b. We used the published STEPS astrometric positions of VB 10 over a time-span of 9 years, which allowed us to compare the expected motion of VB 10 due to parallax and proper motion with the observed motion and to compute precise deviations. The achieved single-epoch precisions of our observations are about 0.1 mas and the data showed no significant residual trend, while the presence of the planet should have induced an apparent proper motion larger than 10 mas/yr. Subtraction of the predicted orbital motion from the observed data produces a large trend in position residuals of VB 10. We estimated the probability that this trend is caused by random noise. Taking all the uncertainties into account and using Monte-Carlo resampling of the data, we are able to reject the existence of VB 10b with the announced mass of 6.4 M_J with the false alarm probability of only 0.0005. A 3.2 M_J planet is also rejected with a false alarm probability of 0.023.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. vmain October 2, 2018 c(cid:13) ESO 2018 0 1 0 2 v o N 0 1 . ] P E h p - o r t s a [ 1 v 8 3 3 2 . 1 1 0 1 : v i X r a Astrometric search for a planet around VB 10⋆ P. F. Lazorenko1, J. Sahlmann2, D. S´egransan2, P. Figueira2,5, C. Lovis2, E. Martin3, M. Mayor2, F. Pepe2, D. Queloz2, F. Rodler4, N. Santos5,6, and S. Udry2 1 Main Astronomical Observatory, National Academy of Sciences of the Ukraine, Zabolotnogo 27, 03680 Kyiv, Ukraine 2 Observatoire de Gen`eve, Universit´e de Gen`eve, 51 Chemin des Maillettes, 1290 Sauverny, Switzerland 3 INTA-CSIC Centro de Astrobiolog´ıa, 28850 Torrej´on de Ardoz, Madrid, Spain 4 Instituto de Astrof´ısica de Canarias, C/ V´ıa L´actea s/n, E-38200 La Laguna (Tenerife), Spain 5 Centro de Astrof´ısica, Universidade do Porto, Rua das Estrelas, 4150-762 Porto, Portugal 6 Departamento de F´ısica e Astronomia, Faculdade de Ciencias, Universidade do Porto Received ; accepted ABSTRACT We observed VB 10 in August and September 2009 using the FORS2 camera of the VLT with the aim of measuring its astrometric motion and of probing the presence of the announced planet VB 10b. We used the published STEPS astrometric positions of VB 10 over a time-span of 9 years, which allowed us to compare the expected motion of VB 10 due to parallax and proper motion with the observed motion and to compute precise deviations. The achieved single-epoch precisions of our observations are about 0.1 mas and the data showed no significant residual trend, while the presence of the planet should have induced an apparent proper motion larger than 10 mas yr−1. Subtraction of the predicted orbital motion from the observed data produces a large trend in position residuals of VB 10. We estimated the probability that this trend is caused by random noise. Taking all the uncertainties into account and using Monte-Carlo resampling of the data, we are able to reject the existence of VB 10b with the announced mass of 6.4 MJ with the false alarm probability of only 5 · 10−4. A 3.2 MJ planet is also rejected with a false alarm probability of 0.023. Key words. astrometry -- technique: high angular resolution -- planetary systems -- stars: individual: VB 10 1. Introduction The detection of a planetary-mass companion to the nearby M8 ultracool dwarf VB 10 (GJ 752B) was announced by Pravdo & Shaklan (2009). Using ground-based astrometric ob- servations with the STEPS camera of the 5 m Palomar telescope obtained over nine years, the authors derived a full astrometric solution of the system that suggested the presence of a planetary companion with a 271 day orbital period and a 6.4 MJ mass. Even though the VB 10b astrometric signature is large with a peak-to-peak value of 8 mas, the detection is debatable due to imperfect orbital sampling and comparably low astrometric ac- curacy of the individual measurements. Zapatero Osorio et al. (2009) presented high-resolution near-infrared observations of VB 10 obtained between 2001 and 2008 with the NIRSPEC instrument on the Keck II telescope. They emphasized on the necessity to have a better sampling of the orbital phase to precisely constrain the orbital parameters and the individual masses of the system. Bean et al. (2010) used the near-infrared CRIRES spectrograph and did not detect the planet's signature in radial-velocity (RV). Anglada-Escud´e et al. (2010) performed a joint analysis of precise RV observations over 175 days with MIKE/Magellan and ESPaDOnS/CFHT spectrographs, RV measurements by Zapatero Osorio et al. (2009), and astrometric data by Pravdo & Shaklan (2009). They showed that the observed astrometric motion is not due to an unseen companion. We present results of the astrometric search for the planet using the FORS2 camera of the VLT, the only instrument which ⋆ Based on observations made with ESO telescopes at the La Silla Paranal Observatory under programme ID 283.C-5024 can achieve astrometric precision of individual measurements of 0.05 -- 0.1 mas (Lazorenko et al. 2009). Despite the short obser- vation period of two months, we obtained data sufficiently good to verify the hypothesis on the possible existence of VB 10b. 2. Observations Observations were made with the FORS2/VLT camera (Appenzeller et al. 1998) in imaging mode during five nights from 2 August to 25 September 2010. The images cover a 4.2 × 4.2′field of view with a pixel (px) scale of ∼0.1′′/px. For each night, we obtained 21 to 72 frames of 7 s exposures with the Rspecial filter. Seeing varied from 0.55′′to 0.9′′. To reduce dif- ferential chromatic refraction (DCR) of the atmosphere, obser- vations were made near meridian. Because the observed motion of VB 10 in the sky is large (over 2 px), we had to know the precise pixel scale to convert pixel to arcsec units without loss of accuracy. From CMC14 and NOMAD catalogue positions of stars in our reference frame, we derived acceptably precise scales of 0.12538 ± 0.00007 ′′/px and 0.12567 ± 0.00005 ′′/px in RA and Decl, respectively. 3. Data reduction and analysis Our null hypothesis was that the planet VB 10b exists. To test this hypothesis, we computed the residuals of our measurements compared to the position of VB 10 expected from the combina- tion of proper, parallactic, and orbital motions. By determining the probability that the measured residuals are compatible with the expected mean value of zero, we could confirm or reject the 2 P. F. Lazorenko et al.: Astrometric search for a planet around VB 10 null hypothesis. We also performed the corresponding computa- tions assuming that the planet does not exist. In all cases, we considered that the astrometric signal Ψ in VB 10 position induced by the planet is defined by the orbital elements of VB 10b given in the discovery paper by Pravdo & Shaklan (2009). The predicted orbital motion over the measurement timespans of 17 and 54 days is 0.48 and 1.95 mas in RA and 0.92 and 3.28 mas in Decl, respectively. Compared to the astrometric precision of FORS2 of about 0.1 mas per epoch, these displacements are large and should be detected in this study. 3.1. Photocentredetermination Raw images were flat-fielded and bias-subtracted to exclude pixel-to-pixel variations of the CCD sensitivity. To increase the number of reference stars, we measured all star images with R = 14 -- 21. At this faint end, the star field is very crowded. This forced us to improve our procedure of photocentre com- putation, which initially was developed for isolated stellar im- ages (Lazorenko 2006). We scanned images obtained under var- ious seeing conditions and made a detailed census of star po- sitions at subpixel precision, of fluxes, and of the point-spread function variation across the CCD. This information was used to accurately model and subsequently subtract the contamina- tion of background counts caused by distant stars. Wings of star profiles at distances up to 50 px were approximated by expo- nential function with free parameters smoothly varying over the CCD. The photocentres x and y were computed by fitting star profiles in a 11 × 11 px window with a model with 12 free parameters and an auxiliary oscillating function in the central 5 × 5 px window (Lazorenko 2006). Computations based on the Levenberg-Marquardt numerical algorithm of the least squares fit were found to give sufficiently good results for stars of ap- proximately equal brightnesses at separations larger than about 10 px. 3.2. Astrometricmodel Reductions were based on the method previously applied to FORS1/2 observations (Lazorenko et al. 2007, 2009). The method was shown to efficiently mitigate atmospheric image motion, geometric field distortion, DCR, and other effects, thus ensuring precision stable over time scales of a few days to a few years. For every star in the field, i.e. VB 10 and the reference stars, the measured photocenter positions xm and ym in frame m at time t were represented by the model x0 + Φ{x} km(x, y) + µxt + πpx + ρ tan z sin γ + d tan zL sin γ = xm − Ψ(t). (1) The left side contains the free model parameters and the model function Φ{x} km(x, y). The expression for y-data is similar but con- tains cos γ instead of sin γ. Here, x0 is a zero point, k is the mode (arbitrary even integer, usually from 4 to 16), Φ{x} km is a polyno- mial in x and y of order k/2 − 1 that models the sum of atmo- spheric image motion and geometric distortion for each frame m. The parameter µx is the proper motion, π is the parallax, and px is the parallax factor in x. The displacement of the star im- age due to DCR is modelled by a term with leading parameter ρ, which depends on the star's colour, and containing the zenith distance z and the angle γ between a direction to zenith and y- axis. The next term describes an image displacement opposite to that of DCR and introduced by the longitudinal atmospheric a k=6...14 b k=6...14 c k=4...16 60" Fig. 1. Circular reference fields increasing in size with increasing modal number k in the field of view of FORS2 in the case of: a) the full data set when an area unavailable on 2 Aug frames was cut away at all nights; b) the same but the problem area cut only at 2 Aug frames; c) the short data set not using 2 Aug frames. dispersion compensator (LADC) of the VLT (Avila et al. 1997). This displacement also depends on the star colour via the param- eter d ≈ −ρ. Both ρ and d are free model parameters and their computation does not require external colour data. The LADC is automatically adjusted to the average zenith distance zL over a given series of frames by setting a distance b between its two prisms to b ∼ tan zL. Finally, Ψ(t) represents the induced or- bital motion of VB 10 if the planet exists. For reference stars, Ψ(t) = 0. Equation 1 defines a system of equations in the combined {x, y, t}-domain for which the unique solution is derived under the condition that the model parameters of reference stars are orthogonal to each basic function of Φk. It is solved iteratively for all reference stars available in a circular region of radius Rk which increases with k and is centered on the target VB 10. The optimal field size Rk is the size at which the noise from a refer- ence field is equal to the noise from atmospheric image motion. The number of reference stars depending on k varies from a few dozens to 500. Due to the large number of reference stars, the function Φ accurately reproduces the coordinate grid distortion introduced by image motion. Solutions for Φkm at each mode k are then used to form equations for the target in the time domain only. The solution of this new set of equations yields the tar- get's model parameters and position deviations from the model. The final output is the average obtained from the solutions at all modes k. 3.3. ComparisonofSTEPSandFORS2referenceframes Because of different zero-points, STEPS proper motion cannot be directly applied to FORS2 astrometry. The STEPS reference frame is given by 15 bright stars. For the FORS2 reference frame, we use both these (except STEPS star Nr. 10 which is a binary) and a number of much fainter stars. The system of proper motions is defined under the condition P µiwi = 0 where sum- mation is taken over all reference stars i with weights wi approx- imately proportional to their brightness (Lazorenko et al. 2009). In particular, this condition is valid for the proper motions of the STEPS stars which, due to their high brigthness (they contribute to over 40% of the total light flux), are the basis of the FORS2 reference frame. Hence, one may expect that P µiwi/ P wi (the weighted mean proper motion in the system of FORS2 proper motions) taken over these stars, is nearly zero. The computed value of this sum (which is the difference of proper motion zero- points) was found to be small: −0.1 ±1.0 and +0.3 ±1.0 mas yr−1 in RA and Decl. Due to the low precision, we did not apply it but included it in the error budget assuming that the uncertainty in the proper motion zero point is ∼1 mas yr−1. Thus, both proper P. F. Lazorenko et al.: Astrometric search for a planet around VB 10 3 motion systems are well consistent, however at time scales of ∼ 0.1 yr only. 3.4. Shortandfulldataset The data reduction of the first epoch (2 Aug) was problematic because of incorrect telescope pointing, owing to which all stars in a 300 px wide area just below VB 10 (Fig. 1a) were imaged to another CCD chip and could not serve as reference at this particular night. We dealt with this problem in two ways. In the first version, we put aside all stars within this area from frames of all nights, using only stars outside of this area as reference objects. The sizes and shapes of the reference fields in this case are shown in Fig. 1a by five circular segments, each correspond- ing to k increasing from six (small radius) to 14 (large radius). Alternatively, we used circular reference areas, which for the 2 Aug night were of the same Rk size but with the lower half vi- gnetted (Fig. 1b). We also examined a short dataset without the 2 Aug epoch with circular reference area (Fig. 1c). The asym- metry of the first two cases degraded the obtained precision in comparison to the symmetric configurations of the short dataset. We have obtained a solution for all three cases presented in Fig. 1. For the final result, we present two solutions. One is ob- tained from the short dataset (case 'c', MJD between 55082 and 55099 days), and the second one is obtained as the average of the results in case 'a' and 'b'. The latter solution is referred to as obtained from the full dataset (MJD between 55045 and 55099 days). 3.5. RecoveringLADCpositions VB 10 is very red and differs much in colour from the reference stars. Therefore, the DCR displacement of this star is very large and about 30 mas in Decl and 10 mas in RA. This is taken into account by the free model parameters ρ and d, but their adjust- ment requires knowledge of the LADC separation b. Information on b was not part of the obtained fits headers and was not accessible to us. Therefore we had to solve the inverse problem of recovering LADC separations from the observations. Because b is fixed for a series of frames of a single night, we had to determine a small number of values for b (one value per night). Our approach is based on the observation that incorrect values of b bias the average night position residuals hxi and hyi of field stars. According to Eq. 1, this bias linearly depends on the star colour parameter d during one given night. This is illustrated by Fig. 2 which shows the distribution and linear dependencies of hyi on d for field stars in the case of the short dataset (the effect is largest in y). Because all observations were made at small hour angles within ±0.7 h, we initially assumed that LADC was always set to the separation corresponding to the meridian. By applying small corrections to these initial values of b, we iteratively reached a solution without dependence of hyi on colours (lower panel of Fig. 2). For these computations, we used all stars within the en- tire field of view of FORS2, processing them as target objects relative to their own subsets of reference stars. Being based on field stars only, this procedure produced a much smaller disper- sion of hyi values for VB 10 and for reference stars with large colours. Corrections to b were small and within ±4% of the initial values and were found with precision corresponding to ±0.025 mas error in hyi for VB 10. For the full dataset, precision degraded to ±0.068 mas because we could not use stars in the problematic 300 px area below VB 10 (Fig. 1a). ) s a m ( 1 0.5 0 -0.5 1 0.5 0 -0.5 VB 10 VB 10 -1 -100 -50 0 50 100 DCR parameter d (mas / air mass) i l C E D n i s a u d s e r h c o p e Fig. 2. Distribution of epoch position residuals hyi of field stars (dots of different type) as a function of d for four September nights and linear dependences of hyi on d (lines of different type) for each night. Upper panel: with initial LADC separations b. Lower panel: the same with recovered b values. Size of dots is proportional to the star brightness. 3.6. Treatmentofcolour-dependentterms An essential drawback of the above procedure is that the re- covery of b introduces small colour-dependent terms µ′(d) and π′(d), which are similar to proper motion and parallax. This is because the above iterations do not garantee convergence to the actual values of b ∼ tan zL. For example, the restored value of tan zL may differ from its actual value by a term which progres- sively changes in time as µt tan−1 zLhsin γi−1, where µ is an arbi- trary constant and hsin γi is the average value of sin γ during the given night. Hence the term d tan zL sin γ in Eq. 1 generates the term d µt sin γhsin γi−1 ≈ d µt linearly dependent on d and which therefore can be treated as an extra image motion µ′(d) = µd. The term µ′(d) compensates for the linear change of tan zL in time. Similarly, we may assume that the restored values of tan zL contain terms proportional to px and py. In this case, the solution of Eq. 1 for parallax should contain the compensating colour- dependent term π′(d). When recovering b values, we cannot control the ampli- tude of µ and of the equivalent parameter π related to parallax. However, they can be detected as proper motion and parallax de- pendence on colours (i.e. on d) which we model as a linear trend in proper motions of field stars and statistically correct for it. The strong correlation between t, px, and py does not allow us to determine separately µ′(d) and π′(d). However, this is not required because for short times px ∼ t, we can approximate the sum of the proper motion and parallax displacement µ′ xt + π′ px xt, where µ′ is the new effective quantity that substitutes for by µ′ both µ′ and π′. Thus processing the short dataset, we find and use only a single colour term µ′(d). The treatment is similar for the y P. F. Lazorenko et al.: Astrometric search for a planet around VB 10 4 s a m , ) t ( Ψ n o i t o m l a t i b r o 2 1 0 -1 short dataset full dataset 40 50 60 70 80 90 100 MJD-55000 (days) Fig. 3. Orbital motion Ψ(t) of VB 10 in RA (circles) and Decl (triangles) for the full and short FORS2 datasets and the linear approximations (solid and dashed lines) of these motions. components. Subtraction of µ′t from the measured positions thus simultaneously eliminates the colour term π′ (see next section). The value of this correction for VB 10 in Decl is µ′ = 29 ± 4.5 mas yr−1 for the full and µ′ = 27 ± 2.9 mas yr−1 for the short dataset, respectively. In RA, the corrections are an order of magnitude smaller. For the full dataset, the above approximation of µ′ xt + π′ px xt is still valid for reference stars, most of which have mod- by µ′ erate d and therefore very small colour-induced parallaxes π′. But it is not sufficiently precise for VB 10, because of its large colour term d. Consequently, π′ is not accurately eliminated by applying the correction µ′(d) as in the case of the short dataset. Therefore, we had to treat the term π′ of VB 10 as a free model parameter. 3.7. Subtractionofparallaxandpropermotions Since the FORS2 observations cover a small fraction (<20 %) of the orbital period, Ψ(t) can be approximated as a linear function of t: Ψα(t) ∼ Qαt, Ψδ(t) ∼ Qδt, (2) where Qα = ∂Ψα(t)/∂t and Qδ = ∂Ψδ(t)/∂t. With the or- bital elements given by Pravdo & Shaklan (2009), we obtained Qα(PS) = 13.4 and Qδ(PS) = 22.4 mas yr−1 (Fig. 3). Astrometric acceleration terms (deviations from the linearity) are smaller than 0.1 mas for the full dataset, thus are negligible. For the short dataset, they are even smaller than 0.02 mas. In addition, Ψ(t) is approximately a linear function of the parallax factors px(t) and py(t), which themselves have an ap- proximately linear time-dependence. This causes a strong corre- lation between parallax, proper motion, and orbital motion, thus makes Eq. 1 degenerated. Therefore, we subtracted parallax and proper motion from the measured positions of VB 10. Precise values of these parameters were found based on the published STEPS astrometric measurements which cover a 9 years long pe- riod. The best fit of STEPS data yielded µα cos(δ) = −586.8±0.2 mas yr−1, µδ = −1361.0 ± 0.2 mas yr−1, and π = 168.0 ± 1.2 mas. These values are very close to the estimates given by Pravdo & Shaklan (2009) and Anglada-Escud´e et al. (2010). 1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1 1 0 -1 2 Aug 8 Sep 18 Sep 22 Sep 2 Aug 8 Sep 18 Sep 22 Sep 8 Sep 18 Sep 22 Sep precision Ψ=0 a 25 SepΨ=0/ precision Ψ=0 b /Ψ=0 25 Sep precision Ψ=0 c /Ψ=0 25 Sep s a m , l c e D n i s n o i t i a v e d s a m , l c e D n i s n o i t i a v e d s a m , l c e D n i s n o i t i a v e d frame sequence Fig. 4. Frame-to-frame deviations in Decl of VB 10 from its model motion at zero (Ψ = 0, filled circles) and non-zero plan- etary signal (Ψ , 0, open circles), the epoch average deviations (large diamonds), and the single-frame precisions σ1 (asterisks) for the cases 'a', 'b', and 'c' of reference field configurations shown in Fig.1. 3.8. Positionresidualsofmodelswithandwithoutplanet Subtracting the contributions of parallax and proper motion from the measured positions of VB 10 resulted in great simplification of Eq. 1, which for VB 10 took the form x0 + ρ tan z sin γ + d tan zL sin γ + {π′ px} = xm − µxt − πpx − µ′ xt − Φ{x} km(x, y) − Ψx(t) y0 + ρ tan z cos γ + d tan zL cos γ + {π′ py} (3) = ym − µyt − πpy − µ′ yt − Φ{y} km(x, y) − Ψy(t) with only four free parameters x0, y0, ρ, and d in the case of the short dataset. For the full dataset, it incorporates the extra parameter π′ (the term in curly braces, see Sect. 3.6). The right side of Eq. 3 contains the measured coordinates xm and ym, the km(x, y) and Φ{y} in-plate solutions Φ{x} km(x, y) derived from the refer- ence stars, and all other corrections. There is a correlation between the orbital signal and the night average angles hsin γi which are not zero and tend to increase in time. Because of this correlation, the model Eq. 1 filters out any component of the signal (e.g. Ψ), which is linearly dependent on t and enters the right side of Eq. 1. Therefore, the output posi- tion residuals contain only a part of the initial signal amplitude. P. F. Lazorenko et al.: Astrometric search for a planet around VB 10 5 50 25 0 ) s a m ( l c e D 75 50 25 0 RA (mas) Fig. 5. Measured in-frame positions x − Φ{x}, y − Φ{y} (squares) of VB 10, the model motion (solid line) over the CCD between 8 and 25 Sept (short dataset) defined primarily by proper motion and parallax, and the magnified DCR signature for a single night (right upper corner). However, this does not hamper our statistical analysis, which we performed after the subtraction of Ψ(t) from the observed po- sitions, thus assuming zero input signal and consequently zero output signal. In this way, the impact of correlations is mini- mized. If the planet does not exist but the subtraction of Ψ(t) was applied, we should detect the inverse signal −Ψ(t) reduced in amplitude because of the correlation between hsin γi and Ψ(t). This is the case corresponding to the last row of Table 2 (Sect. 4), where the measured signal (expressed by the parameter Q) has about half its expected value −Q(PS). Fig. 4 shows the results in terms of model deviations in Decl, where the expected signal is largest. The single-frame precision σ1 includes errors of photocentre measurements, the reference frame noise, and the atmospheric noise. It varies from 0.4 to 0.7 mas depending on seeing. The effect of the vignetted ref- erence field of the 2 Aug epoch (configurations 'a' and 'b') is seen as a degradation of σ1 to over 1 mas. At other epochs, σ1 is larger compared to configuration 'c' because of a larger Rk. Clear conclusions can be drawn from the short dataset when the model (Eq. 3) is most simple and does not require incorpo- ration of parallax (Sect. 3.6). We considered the cases with the predicted orbital motion subtracted (Ψ , 0) and with Ψ = 0. The epoch average deviations hxi and hyi are very small and ran- domly scattered when assuming Ψ = 0, but display a negative trend in time if Ψ , 0. Note that small position deviations do not correspond to a 'zero' measurement. Instead, they demonstrate very precise position measurements, which track the proper mo- tion and parallax displacement at the daily rate of 2.7 mas and 5.0 mas in RA and Decl, respectively. The motion of VB 10 over the CCD surface (Fig. 5) for the measurement timespans of 17 days is 46 mas in RA and 86 mas in Decl, and is dominated by parallax and proper motion. DCR effects induce a small-scale scatter in the measured positions of one night with an amplitude of about 2 mas. Their structure for a typical night is shown in 5-fold magnification in Fig. 5. Similar computations were performed for the full dataset, where we had to account for the parallax correction π′(d) (see Sect. 3.6), thus used the five parameters x0, y0, ρ, d, and π′ for 0.4 0.2 0 −0.2 0.5 0 −0.5 ) s a m ( l c e D r o A R ) s a m ( l c e D r o A R −1 40 50 60 80 MJD − 55000 (days) 70 90 100 Fig. 6. Astrometric residuals in the scenario with planet (open blue symbols) and without planet (filled black symbols) for the short dataset (top panel) and the full dataset (bottom panel) as a function of time. Circles and triangles mark residuals in RA and Decl, respectively. For clarity, only one errorbar corresponding to σsum is shown at each epoch, but the uncertainties for the four respective measurements are identical. fitting of the model Eq. 3. This introduced a strong correlation between px, py, and Ψ, which reduced the amplitude of the de- tected signal. Table 1. Epoch deviations hxi and hyi in RA and Decl of VB 10 for the models with Ψ = 0 and Ψ , 0 (Eq. 3). Ψ = 0 Ψ , 0 MJD 55000+ 45.163 82.032 92.993 97.014 99.011 82.032 92.993 97.014 99.011 (mas) -0.26 0.20 -0.09 -0.00 -0.01 0.07 -0.02 0.03 -0.06 RA Decl (mas) (mas) -0.79 0.15 -0.06 0.03 0.03 RA Decl σN (mas) (mas) Full dataset, 2 Aug -- 25 Sept 2009 -0.16 0.19 0.10 0.19 0.08 -0.06 0.07 0.18 -0.25 0.07 Short dataset, 8 -- 25 Sept 2009 0.11 -0.09 0.09 -0.04 0.07 0.07 -0.01 0.08 0.41 0.35 0.11 0.08 -0.38 0.43 -0.09 -0.00 -0.15 0.33 -0.12 0.02 -0.09 σsum σfit (mas) (mas) 0.64 0.18 0.11 0.11 0.12 0.19 0.09 0.08 0.09 0.32 0.16 0.10 0.09 0.10 0.16 0.06 0.05 0.07 Table 1 and Fig. 6 summarise the results for the epoch residu- als hxi and hyi. The astrometric precision is described by a nom- inal precision σN based on errors in photocentre determination, the reference frame noise and atmospheric noise. σsum also in- cludes error components which dominate at long time spans and originate from the uncertainties in b, µ′, pixel scale, and proper motion of VB 10. For the full dataset, it also includes the uncer- tainty in the parallax of VB 10. σfit is the mathematical expec- tation of the root-mean-square of hxi and hyi, derived from the least squares fit (Eq. 3). 6 4. Discussion P. F. Lazorenko et al.: Astrometric search for a planet around VB 10 Because the signal Ψ was subtracted in Eq. 3, the epoch de- viations hxi and hyi should have an expectation value of zero and a random scatter irrespective of the model (with or without planet). A small dispersion of hxi and hyi values indicates that the tested hypothesis is correct. The data shown in Fig. 6 and Table 1 is more consistent with Ψ = 0 (the hypothesis of no planetary companion), while incorporating a non-zero planetary signal Ψ , 0 increases the scatter of data points, which does not support the existence of the planet VB 10b. We estimated the probability P that the planet exists in two ways. sum + hyi2/σ2 The first estimate P(χ2) was based on the values of χ2 = P[hxi2/σ2 sum], where the sum is taken over epochs. The number of degrees of freedom (D.o.F.) is equal to the num- ber of epochs minus the number of model parameters (there are two and three free parameters for the short and the full dataset, respectively). The probabilities P(χ2) corresponding to the χ2 values are given in Table 2. For both the short and the full dataset, they are at the equally small at the 2% level if the model assumes Ψ , 0. Because the probability of this event is low, the large scatter of hxi and hyi is unlikely to be caused by random noise in the observations. Table 2. Probabilities P(χ2), P(Q) and linear trends Q (mas yr−1) Dataset full full short short Ψ 0 , 0 0 , 0 χ2 D.o.F. 10.6 18.0 1.9 14.8 7 7 6 6 P(χ2) 0.16 0.02 0.93 0.02 Qα 0.67 4.14 -2.06 -7.82 Qδ -0.35 -5.38 2.47 -11.32 P(Q) 0.36 0.02 0.21 5 · 10−4 Secondly, we noticed that the observed residuals hxi and hyi show a linear trend instead of a random distribution. This can be caused by random errors in the observations, but also may indicate a wrong value of the signal Ψ(t) subtracted in Eq. 3. In the latter case, we would expect that the residuals hxi and hyi show a linear dependence −Qt opposite to Ψ(t) (see Eq. 2). A large negative trend is seen for the short dataset residuals computed with Ψ , 0 (Fig. 6), whereas they should be near zero to support the planetary hypothesis. In contrast, a much smaller trend is observed when assuming Ψ = 0 (no planet). For in- stance, the measured values of Q given in Table 2 for the short dataset and Ψ , 0 are -7.82 and -11.32 mas yr−1 in RA and Decl, respectively. If the null hypothesis Ψ , 0 is correct, their expected values are 0 and 0 mas yr−1, and -13.4 and -22.4 mas yr−1 if the hypothesis is wrong (Ψ = 0). In the last case, the mea- sured Q values are always smaller than the values Q(PS) induced by the orbital motion, because their magnitude is damped by cor- relations between Ψ(t) and model parameters (cf. Sect. 3.8). Because the χ2 criterion is not the most efficient one for the characterisation of linear signals, we developed another, more powerful approach to obtain an alternative estimate of P. We considered Q as quantities describing the dispersion of hxi and hyi better than χ2 and performed Monte Carlo simulations to see if the observed features can be explained by random noise in the observations. We simulated each data frame by a random Gaussian noise with root-mean-square of σ1 and added compo- nents modelling the errors in µ′, b, and in the proper motion and parallax of VB 10. In addition, we included a 1 mas yr−1 error to account for the uncertainties in the zero point of proper motion (cf. Sect. 3.3). After fitting the data with the model Eq. 3, we searched the residuals for linear trends in time and estimated their co- efficients Q∗. The false-alarm probabilities P(Q) that a random noise produces Q∗ values larger than the observed ones are given in Table 2. In the case of the short dataset and Ψ , 0, we find that the observed linear trends can be explained by Gaussian noise with a probability of P(Q) = 5 · 10−4, which does not support the existence of the planet. For the full dataset, we find P(Q) = 0.02, which is not sufficient to draw definite conclusions. In spite of the longer timespan and therefore seemingly better conditions to characterise a planetary signal, the full dataset does not pro- vide a better contraint because of the uncertainty in µ′, whose contribution increases with time. Besides, the planetary signal is substantially damped because of its strong correlation with the colour correction π′ used in Eq. (3) as a free model parameter. 5. Conclusions We conclude that the presence of the announced planet around VB 10 is not supported by astrometry. Even assuming half the planetary mass (i.e. 3.2 MJ), simulations give a low false alarm probability P(Q) = 0.023, which does still not support the ex- istence of VB 10b. Our result obtained from astrometry alone is in agreement with the conclusion of Bean et al. (2010) and Anglada-Escud´e et al. (2010) based on RV data. This study is the first application of the FORS2 camera for the search of exoplanets by means of optical astrometry. Because of the high astrometric precision of FORS2, the availability of external STEPS-based proper motion and parallax of VB 10, and the large expected orbital signal, it was possible to perform the verification of the planetary companion hypothesis within the extremely short observation period of 17 days, which is un- usual for astrometric works of this type. The successful use of the STEPS data for the reduction of FORS2 observations is justi- fied only because we verified that the STEPS and FORS2 proper motion reference frames are consistent within 1 mas yr−1 uncer- tainty. We have demonstrated a mean nominal precision of 0.09 mas per epoch of FORS2/VLT observation for data of reason- able quality, despite problems caused by the uncertainty in the LADC position. This precision is sufficient for astrometric de- tection of planets around ultracool dwarfs. Acknowledgements. We thank Dr. G. Anglada-Escud´e whose comments have helped to improve the paper. PF and NCS would like to thank the support by the European Research Council/European Community under the FP7 through a Starting Grant, as well as the support from Fundac¸ ao para a Ciencia e a Tecnologia (FCT), Portugal, in the form of a grant with reference PTDC/CTE- AST/098528/2008. NCS would further like to thank the support from Fundac¸ ao para a Ciencia e a Tecnologia (FCT), Portugal, through a Ciencia 2007 contract- funded by FCT/MCTES (Portugal) and POPH/FSE (EC). References Anglada-Escud´e, G., Shkolnik, E. L., Weinberger, A. J., et al. 2010, ApJ, 711, L24 Appenzeller, I., Fricke, K., Furtig, W., et al. 1998, The Messenger, 94, 1 Avila, G., Rupprecht, G., & Beckers, J. M. 1997, in SPIE, Vol. 2871 Bean, J. L., Seifahrt, A., Hartman, H., et al. 2010, ApJ, 711, L19 Lazorenko, P. F. 2006, A&A, 449, 1271 Lazorenko, P. F., Mayor, M., Dominik, M., et al. 2007, A&A, 471, 1057 Lazorenko, P. F., Mayor, M., Dominik, M., et al. 2009, A&A, 505, 903 Pravdo, S. H. & Shaklan, S. B. 2009, ApJ, 700, 623 Zapatero Osorio, M. R., Mart´ın, E. L., del Burgo, C., et al. 2009, A&A, 505, L5
1605.02473	1	1605	2016-05-09T08:41:58	Orbital clustering of distant Kuiper Belt Objects by hypothetical Planet 9. Secular or resonant ?	[ "astro-ph.EP" ]	Statistical analysis of the orbits of distant Kuiper Belt Objects (KBOs) have led to suggest that an additional planet should reside in the Solar System. According to recent models, the secular action of this body should cause orbital alignment of the KBOs. It was recently claimed that the KBOs concerned by this dynamics are presumably trapped in mean motion resonances with the suspected planet. I reinvestigate here the secular model underlying this idea. The original analysis was done expanding and truncating the secular Hamiltonian. I show that this is inappropriate here, as the series expansion is not convergent. I present a study based on numerical computation of the Hamiltonian with no expansion. I show in phase-space diagrams the existence of apsidally anti-aligned, high eccentricity libration islands that were not present in the original modelling, but that match numerical simulations. These island were claimed to correspond to bodies trapped in mean-motion resonances with the hypothetical planet, and match the characteristics of the distant KBOs observed. My main result is that regular secular dynamics can account for the anti-aligned particles itself as well as mean-motion resonances. I also perform a semi-analytical study of resonant motion and show that some resonance are actually capable of producing the same libration islands. I discuss then the relative importance of both mechanisms.	astro-ph.EP	astro-ph	Astronomy&Astrophysicsmanuscript no. planet9 October 19, 2018 c(cid:13) ESO 2018 Letter to the Editor Orbital clustering of distant Kuiper Belt Objects by hypothetical Planet 9. Secular or resonant ? H. Beust1,2 1 Universit´e Grenoble Alpes, IPAG, CS 40700, F-38058 Grenoble Cedex 9, France 2 CNRS, IPAG, CS 40700, F-38058 Grenoble Cedex 9, France Received ....; Accepted.... ABSTRACT Context. Statistical analysis of the orbits of distant Kuiper Belt Objects (KBOs) have led to suggest that an additional planet should reside in the Solar System. According to recent models, the secular action of this body should cause orbital alignment of the KBOs. Aims.It was recently claimed that the KBOs concerned by this dynamics are presumably trapped in mean motion resonances with the suspected planet. I reinvestigate here the secular model underlying this idea. Methods. The original analysis was done expanding and truncating the secular Hamiltonian. I show that this is inappropriate here, as the series expansion is not convergent. I present a study based on numerical computation of the Hamiltonian with no expansion. Results. I show in phase-space diagrams the existence of apsidally anti-aligned, high eccentricity libration islands that were not present in the original modelling, but that match numerical simulations. These island were claimed to correspond to bodies trapped in mean-motion resonances with the hypothetical planet, and match the characteristics of the distant KBOs observed . Conclusions. My main result is that regular secular dynamics can account for the anti-aligned particles itself as well as mean-motion resonances. I also perform a semi-analytical study of resonant motion and show that some resonance are actually capable of producing the same libration islands. I discuss then the relative importance of both mechanisms. Key words. Celestial mechanics -- Kuiper belt:general -- Planets and satellites: dynamical evolution and stability 1. Introduction The growing statistics about distant Kuiper Belt Objects (KBOs) recently revived interest into the possible presence of an additional distant planet (hereafter termed Planet 9 or P9) in the Solar System. This came out after the discovery of 2012 VP113, a large KBO with orbital parameters similar to those of Sedna (Trujillo & Sheppard 2014). As noted by de la Fuente Marcos & de la Fuente Marcos (2014), all such ob- jects with large perihelia and eccentricities have arguments of perihelia ω concentrating around 0. Even if the statistics is poor, such a distribution is unlikely, as orbital precession induced by the giant planets is expected to quickly randomize ω values. This led de la Fuente Marcos & de la Fuente Marcos (2014) and Trujillo & Sheppard (2014) to suggest that the perturbing action of a distant Super-Earth sized planet could help maintaining this apsidal clustering. This issue was recently investigated into more details by Batygin & Brown (2016) (hereafter B16). They first note that the distant KBOs not only gather around ω = 318◦ ± 8◦, but that the same applies to their longitudes of ascending nodes Ω which sat- isfy Ω = 113◦ ± 13◦. As a consequence the orbits of all bodies concerned are physically roughly aligned. Based on the idea that this orbital confinement is due to the secular action of the sus- pected P9, they develop a secular dynamical model to constrain its parameters. They come to the conclusions that the planet that best reproduces the observational data has mass m′ = 10 M , ⊕ semi-major axis a′ = 700 au and eccentricity e′ = 0.6. Send offprint requests to: H. Beust Correspondence to: [email protected] Further constraints on this planet are hard to derive. Based on the analysis of residuals in the orbital motion of Saturn recorded by the Cassini spacecraft in the last decade, Fienga et al. (2016) show recently that the residuals are better explained by P9 if we give it a current true anomaly around ∼ 120◦. This can indicate a preferred region to try to detect it. Cowan et al. (2016) claim that if it is large enough, P9 could be detected as a 30 mJy at 1 mm wavelength by existing cosmology experiments. The purpose of this paper is to reinvestigate B16's secular model and conclusions. More specifically, B16 first develop a semi-analytical dynamical model of distant particles as secularly perturbed by the giant planets and the hypothetical P9, and show that if it is eccentric enough, P9 can actually confine the orbits in an apsidally aligned configuration with respect to it. Then they move to numerical simulations. They recover the apsidally reso- nant islands noted in their semi-analytical study, but note that the particles trapped there do not have high enough eccentricities to account for the population of distant KBOs under consideration. Conversely, they notice the presence of new, high eccentricity and anti-aligned libration islands in their numerical work that did not appear in the semi-analytical work. They claim that these anti-aligned particles are presumably trapped in mean-motion resonances (hereafter MMRs) with P9, as resonant dynamics is not taken into account in the secular theory. They show that some particles in their numerical study do exhibit resonant trapping. This issue was further reinvestigated by Malhotra et al. (2016), who found that the orbital periods of the main distant KBOs present commensurabilities that could indicate resonant configurations with P9. According to Malhotra et al. (2016), Sedna is putatively trapped in 3:2 MMR with P9. 1 H. Beust: Orbital clustering of KBO's by Planet 9 I reinvestigate here the latter hypothesis. I show that the semi-analytical analysis of B16 is inappropriate to the present case. The reason is the assumed expansion of the secular Hamiltonian that is not convergent. I thus perform a full numer- ical computation of the secular Hamiltonian, and show that anti- aligned, high eccentricity libration islands that were not present in the simplified analysis appear in the non-resonant phase-space maps. Hence invoking resonant trapping might not be necessary. I subsequently investigate the resonant dynamics in a similar semi-analytical manner, and show that various MMRs can ac- tually also generate high eccentricity, anti-aligned librating par- ticles as well, as suggested by B16. I discuss and compare the relevance of both mechanisms. Table 1. Numerical values of hk's for various values of orbital parameters corresponding to configurations tested by B16. In all cases, I assume ∆ = 0, a′ = 700 au and e′ = 0.6. a e h2 h3 h4 h5 h10 h20 h50 450 au 0.1 −0.046 −0.046 0.189 −0.105 1.15 110.6 7.35 × 108 450 au 0.6 0.311 −0.348 0.690 −1.28 55.0 2.55 × 105 1.33 × 1017 150 au 0.1 2.27 × 10−2 −1.70 × 10−3 2.33 × 10−3 −4.32 × 10−4 1.94 × 10−5 3.17 × 10−8 1.02 × 10−15 150 au 0.6 3.45 × 10−2 −1.29 × 10−2 8.52 × 10−3 −5.26 × 10−3 9.32 × 10−4 7.32 × 10−5 1.86 × 10−7 2. Non-resonant secular dynamics Consider a massless particle test orbiting the Sun, perturbed by Planet 9. In the following, all applications are done with the following parameters for P9 : m′ = 10 M , a′ = 700 au and ⊕ e′ = 0.6, as this appears to be the best model quoted by B16. As long as MMRs are not concerned, the long-term behaviour of the particle's orbit is well described with the secular Hamiltonian, which is obtained taking the time average of the instantaneous interaction Hamiltonian over both orbits independently: Gm′ Hsec = − 1 4 4π2 Z 2π a (cid:16)1 − e2(cid:17)−3/2 0 Z 2π Xi=1 0 4 GM − 1 r − r′ − mia2 i Ma2 r · r′ r′3 ! dl dl′ + ′ √GMa(cid:16)1 − √1 − e2(cid:17) , (1) where M is the mass of the Sun; r and r′ are the instantaneous heliocentric radius vectors of the particle and of P9, respectively; l and l′ are their mean anomalies; a is the particle's semi-major axis and e its eccentricity. See Appendix A for the derivation of this expression. The last two terms account for the perturbing action of the known giant planets on the particle and on P9. Here in the second term, the sum extends over the four giant planets, each of them having mass mi and semi-major axis ai; ′ is the precession rate of P9's perihelion caused by the same planets. This Hamiltonian cannot be expressed in closed form though. However, taking advantage from the fact that the par- ticle's orbit lies inside P9's one, the disturbing function can be expanded in ascending powers of a/a′. The integral appearing in Eq. (1) can be written as: Gm′ 4π2 Z 2π 0 Z 2π 0 − 1 r − r′ − r · r′ r′3 ! dl dl′ = − Gm′ a′ +∞ Xk=2 hk , (2) where the hk's are dimensionless coefficients that can be ex- pressed in closed form (Laskar & Bou´e 2010). Each hk is pro- portional to (a/a′)k. For coplanar orbits, they also are functions of e, e′ and of ∆ = ′ − only, where and ′ are the longitudes of periastron of the particle and of P9 respectively. Whenever a < a′, the hk's are generally assumed to decrease with increasing order k, so that the expansion can be truncated to some finite order n. The Hamiltonian given in Eq. (4) of B16 corresponds to a truncation at n = 3 order (octupole approxima- tion). Two potential problems may arise with this approach, if one wants the truncated Hsec to accurately represent the actual dynamics of the particle. First, to ensure the validity of the secular model, the particle must not be trapped in any MMR, and remain protected from close encounters. Beust et al. (2014) 2 showed however that even in the case of crossing orbits, the sec- ular theory provides a relevant description of the dynamics as long as a close encounter does not occur. Second, one must en- sure that the series expansion of the secular Hamiltonian is con- vergent and that the truncation order is large enough to allow the truncation to accurately approximate to the full sum. Convergence of the series expansion is usually ensured for small enough a/a′ ratio. In the configurations described in B16 with orbits that sometimes cross, this is far from being obvious. Table 1 lists various numerical values of hk's computed for dif- ferent configurations of the particle's orbit. For configurations with a = 150 au, the particle's orbit lies well inside that of the planet, whilst when a = 450 au, both orbits cleary cross. In the former case, the hk's rapidly decrease, ensuring convergence of the expansion, but in the latter case it is obviously divergent. An alternate way to avoid the expansion is to compute nu- merically the double integral appearing in Eq. (1). The integral is computed using Gauss-Legendre numerical quadrature with 70 × 70 points, and with special care of the regions where the orbits cross. This technique was already applied to the semi- analytic study of Fomalhaut b's dynamics in Beust et al. (2014). The result is shown in Fig. 1. This figure was built assuming the same input conditions as Fig. 3 from B16, except that the latter was computed assuming octupolar approximation. On each plot, the red curve marks the separation between regions where both orbits actually cross and regions where they do not. Although the general shape of the various phase portraits agree in both cases, striking differences nevertheless appear. In all situations, libration islands around ∆ = 0◦ (i.e., apsidal alignment) appear, but for large a, they are systematically nar- rower and centered around a secular equilibrium located at lower eccentricity than in the octupolar approximation. Moreover, for a ≥ 250 au, new libration islands appear around ∆ = 180◦ (i.e., antialignement) at high eccentricity. These are not present in Fig. 3 from B16. My phase portraits appear to much more closely match the results of B16's numerical exploration (their Fig. 4) than their original maps. The apsidally aligned libration islands now appear at the right eccentricity with corresponding widths. Moreover, the anti-aligned libration islands reported by B16 in their numerical exploration exactly match those I get in my phase portaits. These anti-aligned librating particles are of prime importance in B16's discussion, as they suggest that the apsi- dally confined distant KBOs could be trapped in this dynamical state. B16 claimed that the particles exhibiting this behaviour in their numerical exploration were actually trapped in a MMR with the perturbing planet. Resonant dynamics in indeed not H. Beust: Orbital clustering of KBO's by Planet 9 Fig. 1. Phase portraits of untruncated, secular averaged Hamiltonian (1) in (∆, e) space, computed in the same conditions as B16, i.e., assuming parameters m′ = 10 M , a′ = 700 au and e′ = 0.6 for P9. Plots are drawn for various semi-major axis values listed ⊕ on top of the panels. On each plot, the red curve separates regions where both orbits cross from regions where they do not. In the a = 50 au case, the orbits never cross; for a = 150 au and a = 250 au, the orbits cross in the regions located above the red curves around ∆ = 180◦; for a = 350 au, a = 450 au and a = 550 au, the orbits cross in most configurations, except above the red curves centered around ∆ = 0◦ where both orbits remain nested. described by the secular theory, irrespective of whether the Hamiltonian is truncated or not. B16 show indeed some parti- cles out of their simulation that appear to be trapped in MMRs, at least temporarily. Whilst there is no doubt about the reality of resonant behaviours, I claim here that secular dynamics is also able to generate the anti-aligned, high eccentricity librating par- ticles they find in their numerical exploration as well. 3. Resonant dynamics In this section, I explore the dynamics of particles trapped in (p+ q : p) MMR with P9 (p and q are integers) and perturbed by the giant planets, in a similar semi-analytic way as detailed above. Appendix B describes the way the secular resonant Hamiltonian is obtained. This goes through the introduction of the criti- cal argument of the resonance σ = ((p + q)/p) λ′ − (p/q) λ − , where λ and λ′ are the mean longitudes of the particle and of P9 respectively. As before, I treat the interaction with the known planet in a secular way. After some algebra (see Appendix B), the instantaneous resonant Hamiltonian can be written as n′ √aGM p + q 1 Hres = − GM 2a − Gm′ r · r′ r′3 ! − p r − r′ − 1 4 − GM a(1 − e2)3/2 4 Xi=1 mia2 i Ma2 + ′ √aGM p + q p − √1 − e2! , (3) where n′ = dλ′/dt is the mean angular velocity of P9. Due to the resonant configuration, the Hamiltonian cannot be averaged over both orbital motions independently. But the critial angle σ is a slow varying variable. Hres can then be averaged for fixed σ over the orbital motion of P9. This leaves an autonomous two degrees of freedom Hamiltonian describing the resonant motion. When e′ = 0, the resulting Hamiltonian does not depend on ∆. In that case, the conjugate action to ∆ N = √aGM √1 − e2 − p + q p ! , (4) is a constant of motion, and the planar problem becomes integrable (Moons & Morbidelli 1995). The resonant motion is characterized by libration of σ around a secular equilib- rium, combined with eccentricity and semi-major axis oscilla- tions. But P9's orbit is eccentric. It can nevertheless be shown (Morbidelli & Moons 1993) that the motion is still character- ized by coupled libations of σ, a, and e. The action N is no longer a constant, and its value is subject to changes on a longer 3 H. Beust: Orbital clustering of KBO's by Planet 9 Fig. 2. Phase portraits of resonant Hamiltonian (3) in (∆, e) space after averaging over the orbital motion of P9, for particle having zero resonant amplitude libration (see text), with the same input parameters in Fig. (1). Each map corresponds to a specific MMR which is specified on top of the plot together with the corresponding semi-major axis value. The red curve has the same meaning as in Fig. 1. timescale. This slower drift of N can drive the particle towards high eccentricity regime. The slower drift of N is characterized by the preservation of a new action J (Morbidelli & Moons 1993; Henrard 1990) that is related to the amplitude of the σ-libration. A good way to investigate the dynamics is then to consider negligible libra- tion amplitude orbits. The preservation of J ensures that an orbit with small initial libration amplitude will keep it during the N- drift process. For such orbits, the semi-major axis remains un- changed. After averaging, the resonant Hamiltonian (3) reduces thus to one degree of freedom so that phase portraits can be drawn. This technique was first used by Yoshikawa (1989) to ex- plore the dynamics of inner resonances with Jupiter, and further- more by Beust & Morbidelli (1996) to study dynamical routes that may generate star-grazing comets in the β Pictoris system. I present in Fig. 2 phase portraits for a selection of MMRs, all computed still assuming the same parameters for P9, except that I have set P9's semi-major axis to 665 au instead of 700 au. This makes the 3:2 MMR to fall at a = 507 au, the semi-major axis value quoted by Malhotra et al. (2016) for Sedna, who sug- gest indeed that Sedna could be trapped in that MMR with P9. I have tested many more MMRs. Those presented here have been selected because they fall in the suitable range of semi-major axis, and because they present anti-aligned libration islands at high eccentricity. Some other resonances, like 9:5, 8:5 and 7:5 also present similar libration islands. Particles trapped in these MMRs evolving in the quoted anti-aligned islands could well match the particles depicted by B16 in their simulations. 4. Discussion The high eccentricity, anti-aligned librating particles quoted by B16 as representative for the distant KBOs in relationship with the hypothetical P9 could well correspond to bodies trapped in some of the MMRs listed above. But they could also be non- resonant particles as well, subject to regular secular dynamics as shown in Fig. 1. It is difficult to state which process is dom- inant. Contrary to resonances, non-resonant configurations have the advantage of not being confined to specific semi-major axis locations with respect to P9. They may therefore concern many more particles. However, the particles in the anti-aligned libra- tion islands move on orbits that cross that of P9. They are thus 4 subject to possible ejection via close encounters, but this may occur after a long time. Numerical simulations by Beust et al. (2014) applied to the case of Fomalhaut showed that particles planet moving on orbits that cross that of an eccentric ∼ 10 M ⊕ may survive hundreds of Myrs before being ejected. The planet considered in the Fomalhaut simulations had an orbital period of 1000 yr. With a′ = 700 au, the hypothetical P9 has a similar mass, but a period of 15,000 yr. The survival time of any par- ticle crossing its orbit should thus scale similarly and could be comparable to the age of the Solar System. Figure 1 also shows aligned libration islands (∆ = 0) next to anti-aligned ones. Bodies trapped in these islands are potentially longer-lived than anti-aligned ones, as they move in non-crossing regions of phase-space (see the red curves in Fig. 1). However, as pointed out by B16, these islands are located too low in eccentricity to match the observed distant KBOs. Nonetheless, is P9 is real, numerous bodies should be present in these islands, but they are beyond observing capabilities. Conversely, as suggested by B16, the survival of resonant particles despite crossing orbits is facilitated by the phase pro- tection mechanism. More specifically, the confinement of the critical angle σ around an equibrium position prevents the par- ticle from encountering the planet at the time it crosses its or- bit (Morbidelli & Moons 1993; Moons & Morbidelli 1995). In some cases however, this situation may not last for ever. Due to perturbations by other planets, particles can have large libration amplitudes and be extracted from resonances. B16 note indeed in their simulations particles exhibiting temporary resonant trap- ping. It turns out that any of the two mechanisms outlined here has its own advantages and disadvantages. Stating which one is dominant is difficult, because it should depend on the specific configuration. The constraints on the hypothetical P9 are still too weak to conclude, but I stress here that the anti-aligned li- brating particles quoted by B16 are not necessarily resonant, but could reside there thanks to regular secular dynamics as well. Hopefully a better statistics on distant KBOs will help refining this analysis in the next future. References Aarseth S.J., 1999, Celest. Mech. 73, 127 H. Beust: Orbital clustering of KBO's by Planet 9 Batygin, K., & Brown, M.E. 2016, ApJ, 151, 22 (B16) Beust, H., & Morbidelli, A. 1996, Icarus, 120, 358 Beust, H., Augereau, J.-C., Bonsor, A., et al. 2014, A&A, 561, A43 Cowan, N.B., Holder, G., & Kaib, N.A., 2016, arXiv:1602.05963v1 Fienga, A., Laskar, J., Manche, H., & Gastineau, M. 2016, A&A, submitted, arXiv:1602.06116v1 de la Fuente Marcos, C., & de la Fuente Marcos, R., 2014, Ap&SS, 352, 409 Henrard, J. 1990, Celest. Mech., 49, 43 Laskar, J., & Bou´e, G. 2010, A&A 522 Malhotra, M., Volk, K., & Wang, X. 2016, submitted, arXiv:1603.02196v2 Moons, M., & Morbidelli, A. 1995, Icarus 114, 33 Morbidelli, A., & Moons, M. 1993, Icarus 102, 316 Trujillo, C.A., & Sheppard, S.S. 2014, Nature, 507, 471 Yoshikawa, M. 1989, A&A, 213, 436 5 Appendix A: The non-resonant secular Hamiltonian where H. Beust: Orbital clustering of KBO's by Planet 9, Online Material p 1 GM I describe here the averaging and expansion process of the non- resonant secular Hamiltonian. In the framework of the restricted three-body problem, the instantaneous interaction Hamiltonian between the particle and P9 reads r · r′ r′3 ! , H = − where M is the mass of the Sun, a is the particle's semi-major axis, and r and r′ are the heliocentric radius vectors of the parti- cle and of P9, respectively. The secular Hamiltonian is obtained taking the time average of H over both orbits independently. This reads 2a − Gm′ r − r′ − (A.1) 1 hk = 1 4π2 Z 2π 0 Z 2π 0 a(cid:19)k r′ (cid:18) r a′!−k−1 Pk(cos β) dl dl′ . (A.8) Each hk is proportional to (a/a′)k and can be expressed in closed form as a function of the orbital elements of both orbits with growing complexity, making use of so-called Hansen coeffi- cients. Details about this process are given in Laskar & Bou´e (2010) with explicit expressions of the hk's up to k = 10. In the case of coplanar orbits, apart from being proportional to (a/a′)k, the hk's are function of e, e′ and of ∆ = ′ − only, where and ′ are the longitudes of periastron of the particle and of P9 respectively. As an example, the first ones read . (A.4) Hres,0 = − 1 4 GM a (cid:16)1 − e2(cid:17)−3/2 2a − Gm′ − GM 4 mia2 i Ma2 Xi=1 r − r′ − 1 r · r′ r′3 ! , (B.1) 1 r · r′ r′3 ! dl dl′ , GM 2a − Gm′ 4π2 Z 2π 0 Z 2π Hsec,0 = − where l and l′ are the mean anomalies of the particle and P9 respectively. Note that a is now a secular invariant. This a con- sequence of the averaging process, as the secular Hamiltonian does not depend of the mean longitudes. r − r′ − (A.2) 0 Then, as explained by B16, terms must be added to Hsec,0 to account for the perturbing action of the giant planets. I first add the direct term, so that we have now GM 1 4 a (cid:16)1 − e2(cid:17)−3/2 Hsec,1 = Hsec,0 − This Hamiltonian is implicitly expressed as a function of the canonically conjugate planar Delaunay orbital elements of the particle, namely Xi=1 (A.3) 4 mia2 i Ma2 . ! Q = λ √aGM P =  √aGM(cid:16)√1 − e2 − 1(cid:17) ≡ Φ  As the semi-major axis is a secular invariant, Hsec,1 reduces to one degree of freedom with (, Φ) as conjugate variable. The constant Keplerian term −GM/2a can also be removed from it. Hsec,1 is nevetheless time-dependent as ′ is not constant. I thus perform a canonical transformation, changing to ∆. To do this, I use the generating function S = −Φ = −Φ(∆ + ′). The new momentum conjugate to ∆ is still Φ, and the new autonomous Hamiltonian reads Hsec = Hsec,1 + ∂S ∂t = Hsec,1 + Φ ′ , (A.5) which is Eq. (1) from the text. Unfortunately, the above expression cannot be computed in closed form, so that to get a closed form expression, expansion of the disturbing function is required before averaging. If the par- ticle's orbit lies inside P9's one (which is the situation supposed here), then the instantaneous Hamiltonian can be expanded using Legendre polynomials Pk (k ≥ 0): r′(cid:19)k Xk=2 (cid:18) r H = − where β is the angle between radius vectors r and r′. Each term of the sum can then be averaged over both orbital motions inde- pendently. The result reads Pk(cos β) GM 2a − Gm′ r′ (A.6)  1 + +∞ , Hsec,2 = − 1 4 GM a (cid:16)1 − e2(cid:17)−3/2 a′  +Φ ′ − Gm′ 4 Xi=1 mia2 i Ma2 1 + . +∞ Xk=2 hk (A.7) 3 1 e e′ h2 = (15/4)e2 + 5 a′(cid:19)2 (3/2)e2 + 1 4 (cid:18) a (1 − e′2)3/2 , a′(cid:19)3 16 (cid:18) a h3 = − a′(cid:19)4 (1 − e′2)7/2 " 15 256 21 h4 = (cid:18) a 64 15 e4 + 5e2 + 1! 1 + 2 1 9 8 + (1 − e′2)5/2 cos (∆) , e2 + 1! e2 e′2 cos (2∆) e′2!# . 3 2 (A.11) (A.9) (A.10) Appendix B: The resonant secular Hamiltonian I consider now a situation where the particle is trapped in (p + q : p) MMR with P9 and perturbed by the giant planets. As above I treat the interaction with the giant planet in a secular way, so that my starting instantaneous Hamiltonian reads This Hamiltonian is still implicitly expressed as a function of the Delaunay elements (A.4), but now it does not immediately reduce to one degree of freedom. Due to the resonance, the semi- major axis can indeed have secular variations. Hres,0 cannot thus be averaged over both orbits independently. I introduce now the critical argument of the resonance σ = p + q p p q λ′ − λ − , (B.2) and the new coordinate vector Q1 = (σ, ∆). I have Q1 = AQ + B(t), where (B.3) A = −p/q −1 1 ! 0 and B(t) = −(p + q)/q λ′ ! the now generating −′ Considering function S (P, Q1) = −tPQ = −tPA−1(Q1 − B(t)), I perform a canon- ical transformation with Q1 as new coordinate vector, and P1 = tA−1P as new momenta vector: −(q/p) √aGM √aGM(cid:16)√1 − e2 − (p + q)/p(cid:17) Q1 = σ = H0 − tPA−1 ∂B √1 − e2! − = Hres,0 + ′ √aGM p + q p − ∆ ! P1 =  = Hres,0 − tP1 p + q The new Hamiltonian then reads n′ √aGM ,(B.5) Hres = Hres,0 + ∂B ∂t ∂S ∂t (B.4) ∂t p . H. Beust: Orbital clustering of KBO's by Planet 9, Online Material p 2 where n′ = dλ′/dt is the mean angular velocity of the perturbing planet. This corresponds to Eq. (3) from the text.
1507.04677	1	1507	2015-07-16T18:03:19	Atmospheric Escape by Magnetically Driven Wind from Gaseous Planets II --Effects of Magnetic Diffusion--	[ "astro-ph.EP" ]	We investigate roles of Alfvenic waves in the weakly-ionized atmosphere of hot Jupiters by carrying out non-ideal magnetohydrodynamic (MHD) simulations with Ohmic diffusion in one-dimensional magnetic flux tubes. Turbulence at the surface excites Alfven waves and they propagate upward to drive hot (~ 10^4 K) outflows. The magnetic diffusion plays an important role in the dissipation of the Alfvenic waves in the weakly ionized atmosphere of hot Jupiters. The mass-loss rate of the spontaneously driven planetary wind is considerably reduced, in comparison with that obtained from ideal MHD simulations because the Alfvenic waves are severely damped at low altitudes in the atmosphere, whereas the wave heating is still important in the heating of the upper atmosphere. Dependence on the surface temperature, planetary radius, and velocity dispersion at the surface is also investigated. We find an inversion phenomenon of the transmitted wave energy flux; the energy flux carried by Alfven waves in the upper atmosphere has a nonmonotonic correlation with the input energy flux from the surface in a certain range of the surface temperature because the resistivity is determined by the global physical properties of the atmosphere in a complicated manner. We also point out that the heating and mass loss are expected only in limited zones if the open magnetic field is confined in the limited regions.	astro-ph.EP	astro-ph	Draft version June 14, 2021 Preprint typeset using LATEX style emulateapj v. 12/14/05 5 1 0 2 l u J 6 1 . ] P E h p - o r t s a [ 1 v 7 7 6 4 0 . 7 0 5 1 : v i X r a ATMOSPHERIC ESCAPE BY MAGNETICALLY DRIVEN WIND FROM GASEOUS PLANETS II -- EFFECTS OF MAGNETIC DIFFUSION -- Yuki A. Tanaka1, Takeru K. Suzuki1 & Shu-ichiro Inutsuka1 Draft version June 14, 2021 ABSTRACT We investigate roles of Alfv´enic waves in the weakly-ionized atmosphere of hot Jupiters by carry- ing out non-ideal magnetohydrodynamic (MHD) simulations with Ohmic diffusion in one-dimensional magnetic flux tubes. Turbulence at the surface excites Alfv´en waves and they propagate upward to drive hot (≈ 104 K) outflows. The magnetic diffusion plays an important role in the dissipation of the Alfv´enic waves in the weakly ionized atmosphere of hot Jupiters. The mass-loss rate of the sponta- neously driven planetary wind is considerably reduced, in comparison with that obtained from ideal MHD simulations because the Alfv´enic waves are severely damped at low altitudes in the atmosphere, whereas the wave heating is still important in the heating of the upper atmosphere. Dependence on the surface temperature, planetary radius, and velocity dispersion at the surface is also investigated. We find an inversion phenomenon of the transmitted wave energy flux; the energy flux carried by Alfv´en waves in the upper atmosphere has a nonmonotonic correlation with the input energy flux from the surface in a certain range of the surface temperature because the resistivity is determined by the global physical properties of the atmosphere in a complicated manner. We also point out that the heating and mass loss are expected only in limited zones if the open magnetic field is confined in the limited regions. Subject headings: planets and satellites: atmospheres -- magnetohydrodynamics (MHD) -- planets and satellites: magnetic fields -- planets and satellites: gaseous planets 1. INTRODUCTION Recently, a large number of exoplanets have been found by various detection methods, and it have been learned that the exoplanets are rich in diversity. Some of them have very small semi-major axes and masses that are sim- ilar to the Jupiter, and they are known as hot Jupiters. Transit observations are a very useful method to know not only the radius of planets and their orbital period, but also information about their atmosphere. For ex- ample, transit observations in the UV band provide in- teresting physical properties of the upper atmosphere. Absorption in the H Lyα have been confirmed in some hot Jupiters, such as HD 209458b (Vidal-Madjar et al. 2003), HD 189733b (Lecavelier des Etangs et al. 2010), and 55 Cnc b (Ehrenreich et al. 2012). The transit depth corresponds to the apparent radius of planets. While the transit depth in optical wavelength at HD 209458b is ∼ 1.5%, that corresponds to ∼ 1.38 RJ , the transit depth in the H Lyα line is ∼ 15%, that corresponds to ∼ 4.3 RJ (Vidal-Madjar et al. 2003). Similar phenomena are reported at HD 189733b (Lecavelier des Etangs et al. 2010), and a hot Neptune, GJ 436b (Kulow et al. 2014). These observations suggest that the existence of the extended exosphere that is filled up with the high- temperature atomic hydrogen gas, and mass loss takes place. Heavier elements such as C, O, Si, and Mg are also detected in the UV band in the escaping atmo- sphere (e.g., Vidal-Madjar et al. 2004; Linsky et al. 2010; Ben-Jaffel & Ballester 2013). Although the detailed mechanism of the strong mass loss from hot Jupiters is still unclear, various theo- Electronic address: [email protected] 1 Department of Physics, Nagoya University, Nagoya, Aichi 464- 8602, Japan retical models have been developed. A leading ex- ample of them is that energy-limited escape by X- ray and extreme ultraviolet (XUV) from a central star, so-called XUV-driven escape (Lammer et al. 2003). The XUV-driven escape have been studied by various ways; for example, including chemical reactions (Yelle 2004, 2006), considering radiation pressure on the es- caping atmosphere and the charge exchange between neutral particles in the escaping atmosphere and pro- tons in the hot stellar wind (Holmstrom et al. 2008; Ekenback et al. 2010; Bourrier & Lecavelier des Etangs 2013; Tremblin & Chiang 2013). Additionally, it is sug- gested that XUV-driven escape can occur not only in hot Jupiters but also in cooler planets. Chadney et al. (2015) demonstrated that hydrodynamic escape driven by XUV can take place at the gaseous planets that orbit beyond 1 AU in some cases if the central stars are active in the X-ray. Furthermore, physical properties and phenomena re- lated to magnetic fields of exoplanets have been inves- tigated in more detail by both theoretical and observa- tional studies. For the theoretical works, for example, Adams (2011); Owen & Adams (2014) suggested out- flow from the upper atmosphere will be magnetically- controlled, thus the outflow becomes asymmetric and a mass-loss rate can be smaller about an order of magni- tude than the value without the magnetic fields, that is based on the assumption of the spherical symmet- ric wind. Also, models of the upper atmosphere with the planetary magnetic fields are proposed and the ef- fects on transit depth and the loss of the angular mo- mentum by the planetary magnetic fields were discussed (Trammell et al. 2011, 2014). Recently, magnetic activi- ties of exoplanets have been observed, and several works suggested that the strength of the magnetic fields of hot 2 Jupiters seems to be smaller than that of Jupiter (e.g., Kislyakova et al. 2014) We have also investigated magnetically-driven mass loss from gaseous planets by extending MHD simulations for winds from stars with surface convection, e.g. solar wind (Tanaka et al. 2014). Alfv´en waves, which are ex- cited by turbulence at the surface, transport the energy upward. They heat up the upper atomosphere and drive outflows. However, in this paper we assumed the ideal MHD approximation, which should be modified to take into account effects of the weak ionization in the planet atmopshere. In this paper, we study non-ideal MHD effects by considering the magnetic diffusion in partially- ionized atmospheres. We explain our concept of the simulations and numer- ical method, and several formulae for analysis in Section 2. Then we show results on the structure of atmopsheres and outflows by inspecting roles of the magnetic resistiv- ity in the propagation and dissipation of Alfv´en waves in Section 3. In Section 4, we discuss effects of Ohmic dissipation in the atmosphere and effects of geometry of the magnetic flux tubes. In Section 5, we summarize the results and refer to our future works. 2. MODEL DESCRIPTION 2.1. Numerical Method In this work we use a numerical simulation code that is originally developed for the calculation of the accel- eration of solar wind (Suzuki & Inutsuka 2005, 2006). At the surface of stars, various types of magnetic waves are excited (Matsumoto & Suzuki 2012, 2014). The ex- cited waves propagate upward from the surface, then heat up the coronal region and drive the stellar wind by various dissipation processes of the waves (Goldstein 1978; Heyvaerts & Priest 1983; Terasawa et al. 1986; Kudoh & Shibata 1999; Matthaeus et al. 1999). It is thought that the Alfv´en wave is especially important for the transfer of the energy. This mechanism of the en- ergy transport from the surface to the upper atmosphere can be applied for objects that have their own magnetic field. Although details are uncertain, hot Jupiters are expected to have magnetic fields, so this mechanism is applicable to hot Jupiters. In our previous research, we have extended the simulation code to calculate the atmo- spheric escape from gaseous planets, especially from hot Jupiters (Tanaka et al. 2014). In addition, this mecha- nism is applied for the calculation of the structure of the atmosphere of brown dwarfs, and it is implied that the heating by the dissipation of MHD waves is also impor- tant for the brown dwarfs' atmosphere (Sorahana et al. 2014). As for our MHD simulations we basically follow (Tanaka et al. 2014), except that we consider magnetic diffusion; we solve time-dependent equations and treat the propagation and dissipation of the MHD waves in a super-radially open magnetic flux tube in the atmo- sphere. We adopt the same functional form for the super- radial expansion factor, f (r), as in (Tanaka et al. 2014), which was originally introduced by (Kopp & Holzer 1976). e f (r) = r−r0−h1 h1 + f0 − (1 − f0) /e r−r0−h1 h1 + 1 e , (1) where h1 is the typical height of closed magnetic loops of, and the subscript 0 means the surface. The effects of resistivity explicitly appear in the energy equation and induction equation: ρ d dt(cid:18)e + v2 2 + B2 8πρ − GM⋆ r (cid:19)+ − Br 4π (B · v) − η 4π ∂ 1 r2f ∂r(cid:26)r2f(cid:20)(cid:18)p + ∂r(cid:16)rpf B⊥(cid:17)#) ∂ B⊥ rpf B2 8π(cid:19) vr (2) ∂ 1 r2f + ∂r(cid:0)r2f Fc(cid:1) + qR = 0, ∂r(cid:20)rpf (v⊥Br − vrB⊥) + η ∂ ∂ ∂r(cid:16)rpf B⊥(cid:17)(cid:21) , ∂B⊥ ∂t = 1 rpf (3) where ρ, v, p, e, and B are the density, velocity, pres- sure, specific energy, and magnetic field strength, re- spectively. The subscripts r and ⊥ denote the ra- dial and perpendicular components; d/dt and ∂/∂t denote the Lagrangian and Eulerian derivatives. G and M⋆ denote the gravitational constant and the mass of a central object, and η is the magnetic re- sistivity. Fc and qR are the thermal conductive flux, and radiative cooling (Landini & Monsignori-Fossi 1990; Sutherland & Dopita 1993; Anderson & Athay 1989, see Tanaka et al. (2014) for detail). In these equa- tions, the effect of the super-radial expansion of the mag- netic flux tubes appears as the factor rpf . Note that the term with η in the energy equation (2) indicates Joule (Ohmic) heating and the term with η in the induction equation (3) denotes diffusion of magnetic field lines. The most dominant origin of the magnetic resistivity in the atmosphere of hot Jupiters is the collision between electrons and neutral particles. In such circumstances the magnetic resistivity η depends on gas temperature and an ionization degree as follows; η ≈ 234pT (K) xe (cid:0)cm2 s−1(cid:1) , (4) where xe = ne/n is the ionization degree (e.g., Blaes & Balbus 1994). ne is number density of elec- trons, and n is that of neutral and ionized hydrogen. If the second term on the right-hand side of Equation (3) is dominant over the first term, that is, the magnetic resis- tivity is large, the magnetic waves will diffuse out. Since the gas temperature and ionization degree will be low in the atmosphere of planets, the magnetic resistivity might not be negligible. Therefore, the resistive MHD calcula- tion is needed to evaluate the non-ideal effects on the structure of the atmosphere and the mass loss. Ionization degrees are determined by a method origi- nally developed for Betelgeuse by Hartmann & Avrett (1984) (see also Harper et al. 2009). We adopt the solar abundance gas (Anders & Grevesse 1989)2, and calcu- late the ionization and recombination of H, C, Na, Mg, Al, Si, S, Ca, Fe, and K in the gas phase. Since the typical surface temperature of hot Jupiters is ∼ 1000 K and it is low to ionize hydrogen, heavy elements such as 2 Note: the "solar composition" is updated recently by Asplund et al. (2009) Na and K are the dominant ionizing sources in the lower atmosphere. The inner boundary of the simulations is set at the photosphere in the solar and stellar wind calcu- lations (Suzuki & Inutsuka 2005, 2006; Suzuki 2007; Suzuki et al. 2013). In this paper, we set the inner boundary at the position where p0 = 105 dyn cm−2(= 0.1bar). We fix the temperature at the inner boundary, and it is treated as a parameter. The density at the inner boundary is determined by the given temperature and p0. The outer boundary radius is set to 360 planetary radii, and the number of the grids is 6000. In the most inner region, dr corresponds to 0.0001 planetary radii, and dr increases gradually as the distance from the surface. We set the planet mass to be the Jupiter mass through- out this paper. We inject velocity perturbations at the inner boundary, which excite upward propagating Alfv´en waves, and their value is treated as a parame- In most cases, amplitude of the injected pertur- ter. bations is 20 % of the sound speed at the surface. In the case of the sun, the velocity dispersion at the sur- face is caused by surface convection. On the other hand, several previous works have suggested that the convective-radiative boundary lies deep region of the at- mosphere(e.g., Burrows et al. 2003). However, three- dimensional calculations of the atmospheric circulation of hot Jupiters suggest the existence of supersonic equa- torial flow(e.g., Showman & Guillot 2002). Therefore, it is expected that turbulence can be created at the sur- face by the strong wind. We assume a broadband spec- trum of the perturbations in proportion to 1/ν, where ν is frequency of perturbations. To solve the MHD equa- tions, we adopt a second-order Godunov scheme with the Method of Characteristics (Sano et al. 1999). 2.2. Useful Formulae We here summarize several useful formula to analyze and evaluate simulation results. 2.2.1. Scale Height We treat the surface temperature T0, the radius of a planet Rp, and the velocity dispersion δv at the surface as parameters. We discuss the dependence of the atmo- spheric structure on these parameters. To evaluate them, we introduce the scale height of the atmosphere. The equation of hydrostatic equilibrium of the atmosphere is 1 ρ dp dr + GMp r2 = 0. (5) Mp is the mass of a planet. If we assume isothermal condition, the density profile can be expressed as follows; ρ = ρ0 exp(cid:20)− = ρ0 exp(cid:20)− r − Rp H0 mg kBT0 Rp r (cid:21) r (r − Rp) Rp (cid:21) , (6) where ρ0 is the density at the surface, r = Rp and H0 is the pressure scale height of the atmosphere that is denoted as H0 = kBT0 mg , (7) where kB is the Boltzmann constant, m is mean molecu- lar weight, and g is the gravitational acceleration at the planetary surface. 3 2.2.2. Poynting Flux Behavior of Alfv´enic wave in the atmosphere, such as propagation, reflection, and dissipation, is a key to un- derstand the atmospheric structure. In the ideal MHD simulations the dissipation of the Alfv´enic waves is done by the nonlinear mode conversion. The fluctuations of the magnetic pressure associated with the Alfv´enic waves excite compressive waves. These waves are eventually steepen to shocklets, which finally dissipate and heat the surrounding gas (Tanaka et al. 2014). This process has been examined in terms of the heating of the solar corona and the acceleration of the solar wind (Hollweg 1982; Kudoh & Shibata 1999; Suzuki & Inutsuka 2005). Re- cently, a radio observation by JAXA's spacecraft Akat- suki actually reveals radial distribution of compressive waves in the solar corona, which supports this scenario (Miyamoto et al. 2014). In the resistive MHD simulations, the magnetic diffu- sion also contributes to the dissipation of MHD waves. When T0 is lower, the dissipation of MHD waves in the low altitude where the magnetic resistivity is high affects the structure of the atmosphere. Therefore, an energet- ics argument of propagating MHD waves in the resistive MHD condition is important to understand the proper- ties of the atmospheric structure and mass loss. To evaluate the propagation, reflection, and dissipa- tion of the Alfv´enic wave energy in the atmosphere, we summarize the Poynting flux carried by Alfv´en waves below. The net Poynting flux arising from magnetic ten- sion, which represents the energy flux of Alfv´en waves measured from the comoving frame, can be written as follows; FP = −Br v⊥B⊥ 4π . (8) The Poynting flux can be divided into an outward (par- allel with Br) part and an inward (antiparallel with Br) part (Jacques 1977; Cranmer et al. 2007; Suzuki et al. 2013). In order to evaluate the inward and outward flux, we introduce Elsasser variables, z± = v⊥ ∓ B⊥ . p4πρ −(cid:1) vA. FP = 1 4 + − z2 ρ(cid:0)z2 (9) (10) By using the Elsasser variables, the net Poynting flux is +vA/4 and ρz2 In this equation, ρz2 −vA/4 correspond to the outward and inward components of the Poynting flux, respectively. Although in our simulations we inject the outward component of Alfv´enic waves from the surface, the waves are eventually reflected inwardly. 3. RESULTS 3.1. Role of Ohmic Dissipation In resistive MHD simulations, the propagation and dis- sipation of MHD waves are affected by magnetic diffu- sivity, and accordingly the atmospheric structure would be modified. We compare the ideal MHD and resistive MHD calculations, and investigate how the magnetic dif- fusion affects the structure of planetary atmospheres and outflows. 4 105 ) K t ( e r u a r e p m e T 104 103 102 T=1000K, resistive T=2000K, resistive T=1000K, ideal T=2000K, ideal (a) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) ) 3 - m c g ( y t i s n e D ) 1 - s m k ( y t i c o e v l a d a R l i 10-4 10-6 10-8 10-10 10-12 10-14 10-16 10-18 10-20 10-22 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 -10.0 (b) T=1000K, resistive T=2000K, resistive T=1000K, ideal T=2000K, ideal 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) T=1000K, resistive T=2000K, resistive T=1000K, ideal T=2000K, ideal (c) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 1. -- Comparison of the atmospheric structure between ideal and resistive MHD calculations. (a) Temperature structure, (b) density profile, and (c) radial velocity profile. The solid and dashed lines correspond to resistive and ideal MHD cases, respec- tively. The red and blue lines correspond to the surface tempera- ture, T0 =1000K and 2000K, respectively. Figure 1 compares the time-averaged atmospheric structure of resistive MHD cases with that of ideal MHD cases. Note that all atmospheric structures in this pa- per are time-averaged, and all properties have large time variability. The gas density in the upper atmosphere de- creases more gradually in the ideal MHD cases because a larger amount of Alfvenic wave energy can reach the upper atmosphere and lifts up the gas by the magnetic pressure associated with the waves. On the other hand, in the resistive MHD cases Alfv´enic waves are damped in the cool atmosphere at low altitudes because of the high resistivity, and then, the gas density sharply drops without the support from the magnetic pressure. The Alfv´enic waves that enter the upper atmosphere con- tribute to the acceleration of planetary winds. The wind speeds are quite slow in the resistive MHD cases, while the wind are accelerated to several tens of kilometer per second in the ideal MHD cases. As a result, the mass-loss rates in the resistive MHD cases are much smaller than those in the ideal MHD cases, because the density in the upper atmosphere is around two orders of magnitude less than that of the ideal MHD cases, and the velocity is also much smaller. If a planet is isolated or have a large dis- tance from the central star, the planetary wind does not stream out in the resistive MHD condition except for in- termittently driven outflows observed in the simulations because the average velocity is smaller than the escape velocity. However, if the planet is a hot Jupiter having small semi-major axis, the gas overflows easily from the small Roche lobe and is blown away by the stellar wind from the central star. In contrast to the large difference in the ρ and vr pro- files, the temperature structures of the resistive and ideal MHD cases are not so different each other. Although the locations of the temperature inversion are slightly differ- ent, the upper atmospheres are heated up to ∼ 104 K in both ideal and resistive MHD cases. Therefore, the heat- ing by the dissipation of MHD waves is still important in the atmospheres of the resistive MHD cases. 3.2. Dependence on Surface Temperature We show the relation between the surface temperature, T0, of gaseous planets and the atmospheric structure, and then we discuss the dependence of the mass-loss rate. In this subsection we fix the radius to the Jupiter radius and the amplitude of the velocity dispersion to 20% of the sound speed at the surface. Therefore, larger T0 cor- responds to larger MHD wave energy that is deposited to the magnetic flux tube. T0 is thought to be mainly determined by the irradiation from the central star, more specificly, determined by the combination of the effective temperature of the central star and the semi-major axis of the planet. This implies that, in the framework of the wave-driven planetary winds, when we take plane- tary systems that have a similar central star, properties of the mass loss from hot Jupiters strongly depend on the distance from a central star. 3.2.1. Atmospheric Structure The dependence of the atmospheric structure on T0 is shown in Figure 2. In these simulations, the radius and mass of the planets are set to the radius and mass of Jupiter. In all cases, the gas temperatures in the lower atmosphere are almost isothermal, and they rise rapidly in the upper atmosphere because of the dissipation of the Poynting flux carried by Alfv´enic waves. The dissi- pation of the energy in the upper atmosphere makes a high temperature (∼10000 K) corona-like region around the hot Jupiters. In all the cases, a low-temperature re- gion forms just below the location of the temperature inversion because of the adiabatic expansion of the gas. As shown in Figure 2 (a), the altitude of the temperature inversion is higher for higher T0. This is mainly caused by the difference of the density in the upper atmosphere. In cases with higher T0, the density in the upper atmo- sphere is higher because of the larger scale height; the density decreases more gradually with altitude (Equa- tion (6)). Since larger heating is required to heat up the denser gas, the temperature inversion is located at a higher altitude in cases with larger T0. 105 ) K t ( e r u a r e p m e T 104 103 102 10-4 10-6 10-8 10-10 10-12 10-14 10-16 10-18 10-20 10-22 12 10 8 6 4 2 0 -2 ) 3 - m c g ( y t i s n e D ) 1 - s m k ( y t i c o e v l a d a R l i 1000K 1600K 2000K 2500K (a) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (b) 1000K 1600K 2000K 2500K 10-3 1000K 1600K 2000K 2500K 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (c) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 2. -- T0 dependence of the atmospheric structure. (a) Temperature structure, (b) density profile, and (c) radial velocity profile. The horizontal axis denotes the distance from the planetary surface that is normalized by Rp in logarithmic scale. The solid, dotted, dashed and dot-dashed lines correspond to T0 = 1000 K, 1600 K, 2000 K, and 2500 K, respectively. While the temperature and density panels in Figure 2 show more or less similar profiles each other, there is a large difference in the radial velocity profiles. The cases with low surface temperature T0 ≤ 2000 K yield quite slow wind velocities < 3 km s−1. In contrast, in the case with high T0 of 2500 K, gas is largely accelerated to at- tain much faster velocity, ≈ 10 km s−1, than the other three cases. These results of the velocity and density profiles suggest that the wind structure can be catego- rized into two regimes; while slow and weak planetary winds stream out from cooler planets, fast and strong planetary winds are accelerated from hotter planets. We discuss this difference in Section 3.2.2. Next, we examine the ionization degree that deter- mines the magnetic resistivity in the atmosphere. To evaluate the non-ideal effects in the weakly-ionized at- 5 (a) (b) 100 10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8 10-9 10-10 10-11 106 105 104 103 102 101 100 10-1 10-2 e e r g e d n o i t i a z n o I l r e b m u n s d o n y e R c i t e n g a M 1000K 1600K 2000K 2500K 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) 1000K 1600K 2000K 2500K 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 3. -- T0 dependence of the (a) ionization degree and (b) magnetic Reynolds number in the atmosphere. The horizontal axis denotes the distance from the planetary surface that is normalized by the Rp in logarithmic scale. The solid, dotted, dashed and dot- dashed lines correspond to T0 = 1000 K, 1600 K, 2000 K, and 2500 K, respectively. mosphere, we introduce a magnetic Reynold number, Rm = vL η (11) where v and L are a typical velocity and length in the sys- tem. We use the scale height of the atmosphere (Equa- tion (7)) and the amplitude of the velocity dispersion at the surface for L and v, respectively. Figure 3 shows the dependence of xe and Rm on T0. Since the typical value of T0 is too low to ionize hydrogen atoms, the ionization degree is very low in the lower atmosphere. xe ∼ 10−10 in the cases with T0 = 1000 K and 1600K, and xe is still ∼ 10−8 even for T0 = 2500 K. In this temperature region, the ionization degree is supported by the ionization of alkali metals, such as sodium and potassium, that have relatively low ionization potential. This low xe leads to high magnetic resistivity and small magnetic Reynolds number. As a result, the magnetic Reynolds number in the lower atmosphere is very small, Rm ∼ 10−2 in the cases with T0 = 1000 K and 1600K, and still Rm ∼ 1 for T0 = 2500 K, although it is ≈100 times larger owing to the higher xe. The resultant low Rm means that the non-ideal effects are important in the lower atmosphere. In the upper atmosphere, however, high gas tempera- ture and low gas density cause a sudden surge of xe. xe increases gradually in the region below the tempera- ture inversion region as ρ decreases with altitude. xe is very high above there because the temperature rises high enough to ionize the hydrogen atoms. 6 3.2.2. Two Regimes of the Poynting Flux Figure 4 shows the net Poynting flux in the atmosphere for cases with different T0. The net outgoing Poynting flux decreases rapidly with an increase of the altitude in the lower atmosphere, and it becomes almost constant in the upper atmosphere. The Poynting flux injected at the surface increases monotonically with T0, but its behavior in the atmosphere is not simple; the net Poynting flux in the upper atmosphere is smaller for higher T0 when T0 ≤ 1600 K (a), but it is larger for higher T0 when T0 > 1600K (b). The Alfv´enic waves are damped in the lower atmosphere where the magnetic resistivity is high. The scale height of the atmosphere plays a key role in the resistive dissipation of the Alfv´enic waves. ) 1 - s 2 - m c g r e ( ) 02 r / f 2 r ( · x u l f g n i t n y o P 108 106 104 102 100 10-2 10-4 ) 1 - s 2 - m c g r e ( ) 02 r / f 2 r ( · x u l f g n i t n y o P 108 106 104 102 100 10-2 10-4 1000 K 1200 K 1400 K 1600 K 10-3 2500 K 2000 K 1800 K 1600 K 10-3 (a) 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (b) 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 4. -- Net outgoing Poynting flux as a function of the distance from the surface for different T0 cases. (a) presents cases with lower T0 ≤ 1600 K and (b) presents cases with higher T0 ≥ 1600 K. The Poynting flux is multiplied by r2f (r) /r2 0 to separate the effect of the adiabatic expansion of the flux tube. The scale height is larger for higher T0, which gives the large density in the upper atmosphere. On the one hand the larger density in the upper atmosphere leads to a larger amount of the mass loss, but on the other hand it gives a negative effect on the wind through the enhance- ment of the resistive dissipation of the Alfv´enic waves. The ionization degree is lower in the higher-density gas because of the efficient recombination, and consequently the magnetic resistivity is relatively higher. In Figure 3 (b), the magnetic Reynolds number of the case with T0 = 1000 K is low in the lower atmosphere, but it in- creases quickly in the upper atmosphere. In the case with T0 = 1600 K, it is as low as the T0 =1000 K case in the lower atmosphere, but it increases more slowly than in the 1000 K case. Therefore, the magnetic Reynolds number remains relatively lower in the almost entire re- gion than in the T0 =1000 K case. Because of the larger resistivity, a smaller fraction of the injected wave energy reaches the upper atmosphere in the T0 = 1600 K case, compared to the T0 = 1000 K case. This is the main reason why the transmitted Poynting flux shows the neg- ative tendency on the surface temperature for T0 < 1600 K. However, for T0 > 1600 K, the Poynting flux exhibits a positive correlation with T0 because the magnetic re- sistivity is lower for higher T0. ) 1 - s 2 - m c g r e ( ) 02 r / f 2 r ( · x u l f g n i t n y o P 1010 108 106 104 102 100 10-2 10-4 10-6 10-3 F+,2000K F-,2000K F+,1600K F-,1600K F+,1000K F-,1000K 10-2 100 Distance from surface ((r-Rp)/Rp) 10-1 101 Fig. 5. -- Comparison of the inward and outward Poynting flux on distance from the surface for the cases with T0 = 1000 K (red), 1600 K (green), and 2000 K (blue). The Poynting flux is multi- plied by r2f (r) /r2 0 in order to remove the effect of the adiabatic expansion of the flux tube. The solid lines are the outward energy flux F+, and the dashed lines are the inward energy flux F−. Figure 5 compares the outward and inward energy flux the Alfv´enic waves (equation 10). Here we show three cases with T0 of 1000 K, 1600 K, and 2000 K. Every case shows that the inward energy flux tracks the outward energy flux with a slightly smaller level in the almost en- tire region of the atmosphere. The net outgoing Poynting flux in Figure 4 is considerably smaller than the outward energy flux in Figure 5. These results indicate that the injected outgoing Alfv´en waves are reflected back down- ward through the propagation in the atmosphere. A tiny fraction of the injected Alfv´en waves can reach the upper atmosphere, which contributes to the heating and the ac- celeration of the gas. In general, a small fraction of the injected Poynting flux by Alfv´en waves is transmitted to the upper atmosphere because they suffer dissipation and reflection, whereas the resistivity is one of the keys that control the dissipation of the waves. The fraction that enters the upper atmosphere determines the properties of the wind. For instance, if a large amount of the MHD wave energy reaches the upper atmosphere, it will rise the density there and drive outflows. The the effect of resistive dissipation of the Alfv´enic waves is also seen in Figure 5. Both dissipation and reflection of the waves decrease the Poynting flux. The Poynting flux in the case T0 =1000 K (red lines) decreases much faster than that of the case T0 =2000 K (blue lines), which is caused by both resistive dissipation and reflection. The higher magnetic resistivity in the 1000 K case makes the Alfvenic waves dissipated faster, and the steeper density gradient results in strong reflec- tion. In the 1600 K case (green lines), the Poynting flux quickly drops, which is much faster than in the other two cases. Since the scale height is larger than that of the T0 =1000 K case, the reflection of the Alfv´enic waves is suppressed. In this case the resistive dissipation of the Alfv´enic waves is the primary reason of the rapid drop of the Poynting flux, and the amount of the energy that can reach the upper atmosphere is quite small. 3.2.3. Time Variability of Mass-loss Rate Our simulations show large time variabilities of the at- mospheric structure. Figure 6 shows the mass-loss rate for different T0 cases with time. One can see that in the cases with lower T0 ≤ 1600 K the outflow is not time- steady, but the atmospheric material infalls during cer- tain periods, which is seen as discontinuous lines. For ex- ample, in the case with T0 = 1000 K, the outflow phases are relatively shorter and it is comparable to the total duration of the infall phases. In the case with T0 = 1600 K, the integrated duration of the outflow phases is longer, and the infall phase disappears for higher T0 ≥ 2000 K. 10-18 1000K 1600K 2000K 2500K ) r a e y / s s a m l r a o S ( e t a r s s o l s s a M 10-19 10-20 108 107 106 ) s / g ( e t a r s s o l s s a M 14 16 18 20 22 24 26 Time (hour) Fig. 6. -- Time evolution of the mass-loss rates for the cases with T0 = 1000 K (red), 1600 K (green), 2000 K (blue), and 2500 K (violet). Thick solid lines show the outflow rates, and dashed lines show the infall rates. On the left the mass-loss rate is shown in units of solar mass per year and on the right in units of gram per second. These results suggest that in the cases with low T0 ≤ 1600 K the atmosphere is almost hydrostatic because the damping of Alfv´enic waves is too severe to drive steady wind, and the outflows occur only intermittently. For higher T0 the intermittency of the planetary winds dis- appears, more steady outflows are obtained. In Figure 7 we compare the Poynting flux by the Alfv´en waves in the active phase during 23.6-31.6 hour and that in the infall phase during 17.8-20.2 hour in Fig- ure 6. In the outflow phase, a larger amount of the Poynt- ing flux is carried by the Alfv´en waves into the upper atmosphere, while in the infall phase, the Poynting flux in the upper atmosphere is strongly reduced around four orders of magnitude. This result implies that the trans- mission of the Poynting flux into the upper layor controls the on-off nature of the atmospheric escape for the low T0 cases. However, we would like to note that the case with T0 = 1000 K (not shown) gives more complicated behavior; Poynting flux during outflow phases does not always exceed Poynting flux during infall phases bucause ) 1 − s 2 − m c g r e ( ) 02 r / f 2 r ( ´ x u l f g n i t n y o P 108 106 104 102 100 10−2 10−4 10−6 10−8 10−10 10−12 7 Active Inactive 10−3 10−2 10−1 100 Distance from surface ((r−Rp)/Rp) Fig. 7. -- Comparison of the Poynting flux in the outflow (solid) and infall (dashed) phases for the case with T0 = 1600K. The Poynting flux is multiplied by r2f (r) /r2 0 in order to remove the effect of the adiabatic expansion of the flux tube. of the rapid temporal variations. 3.3. Dependence on Planet Radius Here we describe the dependence of the atmospheric structure and the mass-loss rate on the radius of gaseous planets Rp. A number of theoretical works on the ther- mal evolution of the gaseous planets suggest that radii of gaseous planets, which tend to be around Jupiter ra- dius, are affected by surrounding circumstances (e.g., Burrows et al. 2007). In addition, observations have shown that some hot Jupiters have enormously larger radii than that are expected (Baraffe et al. 2010). There- fore, we treat Rp as a parameter and investigate the de- pendence on Rp. The mass-loss rate is very small for low T0, therefore we set T0 =2000 K in this subsection. We also fix the amplitude of the velocity dispersion at the surface, δv = 0.2cs. Figure 8 shows the dependence of the atmospheric structures on Rp. At low altitudes the temperature is almost constant (≈ surface temperature) in all the cases. It decreases initially but rises eventually by the heating owing to the wave dissipation. The temperature inver- sion is located at a higher altitude for larger Rp, because the scale height is larger; the slower decrease of the den- sity leads to higher density in the upper region (Figure 8), which is not heated up to high temperatures. Accord- ingly, the bending location of the density gradient, which corresponds to the location of the temperature inversion, is systematically higher for larger Rp by the same reason. In all the cases, the temperature of the upper atmo- sphere is elevated to ∼ 104 K, by the dissipation of MHD waves. However, the wind velocity is quite slow. Partic- ularly in the cases with the large Rp ≥ 1.2RJ , the gas in the atmosphere is almost static or even infalls partly, whereas the cases with the smaller radius show outflows with a few km s−1. Figure 9 shows the Rp dependence of the ionization degree and magnetic Reynolds number. The ionization degree, xe, is quite low at low altitudes because of the low temperature there (≈ 2000 K) in all the cases. The dominant ion sources in the lower atmosphere are the alkali metals. In contrast, xe increases to 10−4 − 10−1 in the upper atmosphere in all cases. In the smaller Rp case, xe starts to increase from a lower altitude and the final xe is also higher in the upper atmosphere, because the recombination is slower there due to the lower density. 8 105 ) K t ( e r u a r e p m e T 104 103 102 10-4 10-6 10-8 10-10 10-12 10-14 10-16 10-18 10-20 3 2.5 2 1.5 1 0.5 0 -0.5 -1 ) 3 - m c g ( y t i s n e D ) 1 - s m k ( y t i c o e v l a d a R l i 0.8RJ 1.0RJ 1.2RJ 1.6RJ 2.0RJ (a) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (b) 0.8RJ 1.0RJ 1.2RJ 1.6RJ 2.0RJ 10-3 0.8RJ 1.0RJ 1.2RJ 1.6RJ 2.0RJ 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (c) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 8. -- Rp dependence of the atmospheric structure. (a) Temperature structure, (b) density profile, and (c) radial velocity profile. The narrow solid, solid, dashed, dotted, and dot-dashed lines correspond to Rp = 0.8RJ , 1.0RJ , 1.2RJ , 1.6RJ , and 2.0RJ , respectively. Distance from the surface (horizontal axis) is normal- ized by Rp each case. The mass-loss rate remains very small in these cases. The case with Rp = RJ gives the mass-loss rate of 2.6 × 10−19 M⊙ yr−1, and the case with 0.8 RJ gives 2.3 × 10−19 M⊙ yr−1, that correspond to 1.7 × 107 g s−1, and 1.4 × 107 g s−1, respectively (here, Mp = MJ and T0 = 2000 K). In the ideal MHD calculations, the mass-loss rate in the case with Rp = RJ and T0 = 2000 K is ∼ 1.9 × 1010 g s−1(Tanaka et al. 2014). Our results show that the large resistivity as a result of the low ionization degree in the low atmosphere significantly suppresses the mass loss from hot Jupiters. However, we would like to note that the wind shows large fluctuations. There is a possibility remained that the planetary wind streams out in an intermittent manner. Also, the Roche lobe of a hot Jupiters is small because of the small separation 100 10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8 10-9 10-10 105 104 103 102 101 100 10-1 10-2 e e r g e d n o i t i a z n o I l r e b m u n s d o n y e R c i t e n g a M (a) (b) 0.8RJ 1.0RJ 1.2RJ 1.6RJ 2.0RJ 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) 0.8RJ 1.0RJ 1.2RJ 1.6RJ 2.0RJ 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 9. -- Rp dependence of the (a) ionization degree and (b) magnetic Reynolds number in the atmosphere. The narrow solid, solid, dashed, dotted, and dot-dashed lines correspond to Rp = 0.8RJ , 1.0RJ , 1.2RJ , 1.6RJ , and 2.0RJ , respectively. Distance from the surface (horizontal axis) is normalized by Rp each case. from the central star. Then, once a slow gas material overflows the Roche lobe, it will be blown away by the stellar wind from the central stars. 3.4. Dependence on Velocity Dispersion The velocity amplitude at the surface, δv, is also an important parameter that controls the structure of the atmosphere and the wind of a planet. In this subsection, we fix Rp = RJ , Mp = MJ , and T0 = 2000 K. In Figure 10, we show the atmospheric structure of the cases with different δv. Cases with different δv give similar density and temperature profiles. On the other hand, larger δv gives faster wind velocity. Because of the similarity of the density and the temperature shown in Figure 10, the ionization degree and the magnetic resistivity are also similar in all the cases. Therefore, it can be said that the difference of δv at the surface has small effect on the structure of the atmosphere except for the radial velocity. Though the difference of the density in the upper at- mosphere is very small, the wind velocity is faster for larger δv, and it causes the increase of the mass-loss rate as shown in Figure 11 for cases with T0 = 1000 K and 2000 K. 4. DISCUSSION 4.1. Ohmic Heating in Planetary Atmosphere We have investigated the effect of the magnetic diffu- sion on the dissipation of the Alfv´enic waves, which sup- presses the planetary winds, so far. On the other hand, the resistive dissipation also heats up the ambient gas 105 ) K t ( e r u a r e p m e T 104 103 102 10-4 10-6 10-8 10-10 10-12 10-14 10-16 10-18 10-20 14 12 10 8 6 4 2 0 -2 ) 3 - m c g ( y t i s n e D ) 1 - s m k ( y t i c o e v l a d a R l i 0.2cs 0.5cs 0.7cs 0.9cs (a) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (b) 0.2cs 0.5cs 0.7cs 0.9cs 10-3 0.2cs 0.5cs 0.7cs 0.9cs 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) (c) 10-3 10-2 10-1 100 Distance from surface ((r-Rp)/Rp) Fig. 10. -- Dependence of the atmospheric structure on the δv at the surface. (a) Temperature structure, (b) density profile, and (c) radial velocity profile. The horizontal axis denotes the distance from the planetary surface that is normalized by the Rp in logarithmic scale. The solid, dashed, dotted, and dot-dashed lines correspond to δv = 0.2cs, 0.5cs, 0.7cs, and 0.9cs, respectively, and cs is the sound speed at the surface. In all cases, Rp = RJ , Mp = MJ , and T0 = 2000 K. via Ohmic (Joule) heating, which plays a positive role in driving outflows. In this subsection, we examine the role of the Ohmic heating in a quantitative manner. Current density is j = c 4π (∇ × B) , (12) and electric conductivity σ is related with the magnetic resistivity η by 1 σ = 4π c2 η. (13) Then, Ohmic heating rate per unit volume, QOhm = j2/σ, is derived as QOhm = η 4π (∇ × B)2 . (14) ) r a e y / s s a M l r a o S t ( e a r s s o l s s a M T=1000K T=2000K 10-17 10-18 10-19 10-20 0.2 0.3 0.5 0.4 0.7 Velocity dispersion (δv/cs) 0.6 9 108 107 ) s / g ( e a r s s o t l s s a M 106 0.8 0.9 Fig. 11. -- Relation between δv/cs and the mass-loss rate for T0 = 1000 K (circle) and 2000 K (filled square). The mass-loss rate is shown in units of solar mass per year (left axis) and gram per second (right axis). ) 1 - s 3 - m c g r e ( e t a r g n i l o o c / g n i t a e H ) 1 - s 3 - m c g r e ( e t a r g n i l o o c / g n i t a e H 105 100 10-5 10-10 10-15 10-20 10-25 10-30 10-35 105 100 10-5 10-10 10-15 10-20 10-25 10-30 10-35 10-3 10-3 (a) T=1000 K 10-2 100 Distance from surface ((r-Rp)/Rp) 10-1 101 (b) T=2000 K 10-2 100 Distance from surface ((r-Rp)/Rp) 10-1 101 Fig. 12. -- Comparison of ideal MHD heating rate (red solid), cooling rate (green dashed), and Ohmic heating rate (blue solid) per unit volume for cases with T0 = 1000 K (a) and 2000 K (b). We set δv = 0.2cs, Mp = MJ , and Rp = Rj . Figure 12 shows the Ohmic heating rate in compari- son with the ideal MHD heating (red solid) and cooling (green dashed) rates of the gas. The ideal MHD heat- ing includes shock heating (see Section 2.2.2) in addi- tion to adiabatic heating. In the lower atmosphere, the Ohmic heating dominates over, or be comparable to the ideal MHD heating because of the large resistivity. How- ever, since the Alfv´enic waves dissipate very rapidly, the Ohmic heating rate drops drastically as the altitude in- creases. 10 ) 1 - s 3 - m c g r e ( e t a r g n i t a e h i c m h O 105 100 10-5 10-10 10-15 10-20 10-25 10-30 10-35 2500 K 2000 K 1600 K 1000 K 10-3 10-2 100 Distance from surface ((r-Rp)/Rp) 10-1 101 Fig. 13. -- Dependence of the Ohmic heating rate per unit volume on T0. The solid, dotted, dashed, and dot-dashed lines correspond to T0 =1000 K, 1600 K, 2000 K, and 2500 K, respec- tively. Resistivity depends sensitively on temperature. Fig- ure 13 compares the Ohmic heating rates for cases with different T0. The higher T0 cases have smaller Ohmic heating rate in the lower atmosphere, but it reverses in the upper atmosphere, because the dissipation of the Alfv´enic waves is more gentle at low altitudes in these cases and a larger amount of the wave energy reaches the upper atmosphere. In addition, as shown in the Figure 3, the magnetic Reynolds number is relatively lower in the upper atmosphere for higher T0, which further increases the resistive heating. However, the Ohmic heating rate is quite small in com- parison to the shock heating except for the region closed to the surface (Fig. 12), so this effect is negligible in the upper atmosphere and gives little contribution to the at- mospheric escape. Even in the low atmosphere where the Ohmic heating is large, it does not play a primary role. The comparison between the ideal and non-ideal calculations in Figure 1 (a) shows that the temperature profiles in the lower atmosphere are almost same. In fact, the temperatures in the lower atmosphere in the resistive cases are barely higher than the ideal cases and the difference is only several tens of kelvin. In summary, the Ohmic heating does not play important roles in the entire structure of the atmosphere and the atmospheric escape. 4.2. A Possibility for Bipolar Outflows In this work and the previous work, we have fixed the shape of the open magnetic flux tube that is described in Section 2.1. We presume that the overall structure of magnetic field lines of gaseous planets is similar to that of the sun. The sun has a multipole magnetic field and the properties of open magnetic flux tubes are observed in detail (e.g., Tsuneta et al. 2008; Ito et al. 2010). How- ever, the actual structures of magnetic field in exoplan- ets are unknown. The overall structure of the planetary magnetic field affects the shape of the open magnetic flux tube. The most critically affected parameter in our mod- eling is the filling factor f (r) (see Equation (1) and (2) of Tanaka et al. (2014)). If the planetary magnetic field has a simple dipole structure, open magnetic flux tubes appear in the magnetic polar regions. In this case, heat- ing in the upper atmosphere and the acceleration of gas flow by MHD wave energy occur only in the polar regions. Therefore, the actual appearance of flow is expected to be bipolar. In the simple dipole case of the magnetic field, the degree of super-radial expansion of the open mag- netic flux tube should be smaller, and thus the mass-loss rate is expected to be smaller because the reflection of the energy flux is expected to be enhanced. To inves- tigate this more quantitatively, a three-dimensional cal- culation is required, which remains difficult at present. On the other hand, the observation of the bipolar outflow from an exoplanet is, in principle, possible in the detailed analysis of the differential spectroscopy for transiting hot Jupiters, unless the magnetic dipole axis is perpendicular to both the orbital plane of the exoplanet and the line- of-sight. This is because a blue-shifted outflow before ingress and a red-shifted outflow after the emersion, or vice versa, can be seen in this case. Once the mass flux of the bipolar outflow from the exoplanet is observed, we will possibly constrain the property of magnetic field of the exoplanet by assuming the driving mechanism as described in this work. 4.3. Limitations in the Present Model As we mentioned in the introduction, high energy ra- diation such as X-ray and extreme ultraviolet from a central star is important for the heating of upper atmo- sphere and atmospheric escape from hot Jupiters. Here we compare the height where the heating by MHD waves is important and the penetration depth of stellar XUV. It is useful to introduce RXUV, the radius at which XUV is absorbed. RXUV means that the optical depth for XUV reaches unity at that radius, and it corresponds to the radius where the column density of hydrogen be- comes ∼ 5 × 1017 cm−2(e.g., Murray-Clay et al. 2009; Rogers et al. 2011). We calculate RXUV in each calcu- lations and compare it with the region where the heating by the dissipation of the MHD waves becomes important. ) p R / ( s u d a R i 1.40 1.35 1.30 1.25 1.20 1.15 1.10 1.05 RXUV Rlow R8000K 1000 1200 1400 1600 2000 Surface temperature (K) 1800 2200 2400 Fig. 14. -- Dependence of RXUV on T0 and comparison with Rlow and R8000K. The black squares with the solid line corre- sponds to RXUV, the circles with the dashed line and the triangles with the dotted lines correspond to Rlow and R8000K , respectively. We set δv = 0.2cs, Rp = RJ , and Mp = MJ for this comparison. Figure 14 shows T0 dependence of RXUV and compar- ison with other considerable radii. As shown in Figure 2(a), the gas temperature slightly drops due to adiabatic expansion and then heated drastically to ∼ 104 K. Rlow is the radius at which the temperature become lowest and heating by MHD waves starts to become important, and R8000K is the radius at which the temperature be- come higher than 8000 K. In this range of T0, RXUV always locates only slightly above Rlow and the value varies between 1.05 ∼ 1.20Rp. This is consistent with previous works on the XUV heating in the atmosphere of hot gaseous planets (e.g., Murray-Clay et al. 2009). In addition, R8000K always lies much higher altitude than RXUV. These result suggest that the region heated to ∼ 104 K by the dissipation of the MHD waves is optically thin to XUV photons, and XUV can reach the altitude that the temperature becomes lowest in our model. Photo- ionization, cooling from metal lines, Lyman α play im- portant roles in the upper atmosphere of hot Jupiters. At this stage, our model does not include the effects of stellar XUV irradiation, the inclusion of which should improve our model. In the present modeling, the results cannot explain the observational results of the mass-loss rates adequately, since the mass-loss rates are reduced largely when we take into account the effects of the magnetic re- sistivity. This seems to indicate that the resulting mass- loss rate strongly depends on the details of the input physics. More realistic calculation is critically needed, and will be out future work. 5. CONCLUSION We have studied the structure of the atmosphere and the wind of hot Jupiters by non-ideal MHD simulations, particularly focusing on the effects of the magnetic diffu- sion. The Alfv´enic waves generated from the surface are strongly damped via resistive dissipation in the weakly- ionized atmosphere at low altitudes and only a tiny frac- tion of the initial energy can reach the upper atmosphere. Consequently, the mass-loss rate is reduced significantly, compared to that obtained from the ideal MHD simula- 11 tions. However, the Alfv´enic waves that survive to trans- mit into the upper atmosphere still support the temper- ature inversion and heat up to ∼ 104 K by the wave dissipation; the Alfv´enic waves are still important in de- termining the temperature profile even though the mag- netic diffusion is taken into account. We would like to note that the wave heating here is not owing to Ohmic heating but the consequence of the shock dissipation of the compressive waves that are nonlinearly excited from the Alfv´enic waves. We also investigated the dependence of atmospheric properties on the surface temperature, the planet radius, and the velocity dispersion at the surface. We found a nonmonotonic dependence of the Poynting flux on the surface temperature (Figure 4), because the resistivity, which controls the dissipation of the Alfv´enic waves, is determined globally the density and temperature of the atmosphere. We point out that the heating and acceler- ation of gas will take place in magnetic polar regions if the planetary magnetic field is simple dipole structure, and non-spherically symmetric bipolar outflow will takes place. In our calculations, the treatment of the radiative cool- ing is simplified especially for low temperature gas (see discussion part of Tanaka et al. (2014)). Therefore, a more accurate treatment is needed for future investiga- tion and now we are going to improve this part. ACKNOWLEDGEMENT We thank G. Harper for providing a useful code to calculate ionization degree in the gas. We also thank the referee for valuable comments for improving the manuscript. This work was supported in part by Grants- in-Aid for Scientific Research from the MEXT of Japan, 22864006 (TKS). REFERENCES Adams, F. C. 2011, ApJ, 730, 27 Anders, E., & Grevesse, N. 1989, Geochim. Cosmochim. Acta, 53, 197 Anderson, L. S., & Athay, R. G. 1989, ApJ, 226, 1089 Asplund, M., Grevesse, N., Sauval, A. J., and Scott, P. 2009, ARA&A, 47, 481 Holmstrom, M., Ekenback, A., Selsis, F., Penz, T., Lammer, H. & Wurz, P. 2008, Nature, 451, 970 Ito, H., Tsuneta, S., Shiota, D., Tokumaru, M., & Fujiki, K. 2010, ApJ, 719, 131 Jacques, S. A. 1977, ApJ, 215, 942 Kislyakova, K. G., Holmstrom, M., Lammer, H., Odert, P., & Baraffe, I., Chabrier, G., and Barman, T. 2010, Reports on Progress Khodachenko, M. L. 2014, Science, 346, 981 in Physics, 73, 016901 Ben-Jaffel, L., & Ballester, G. E. 2013, A&A, 553, A52 Blaes, O. M., & Balbus, S. A. 1994, ApJ, 421, 163 Bourrier, V. & Lecavelier des Etangs, A. 2013, A&A, 557, A124 Burrows, A., Sudarsky, D., and Hubbard, W. B. 2003, ApJ, 594, 545 Burrows, A., Hubeny, I., Budaj, J., and Hubbard, W. B. 2007, ApJ, 661, 502 Chadney, J. M., Galand, M., Unruh, Y. C., Koskinen, T. T., & Sanz-Forcada, J. 2015, Icarus, 250, 357 Cranmer, S. R., van Ballegooijen, A. A., and Edgar, R. J. 2007, ApJS, 171, 520 Ehrenreich, D., Bourrier, V., Bonfils, X., Lecavelier des Etangs, A., H´ebrard, G., Sing, D. K., Wheatley, P. J., Vidal-Madjar, A., Delfosse, X., Udry, S., Forveille, T., & Moutou, C. 2012, A&A, 547, A18 Ekenback, A., Holmstrom, M., Wurz, P., Griessmeier, J. M., Lammer, H., Selsis, F. & Penz, T. 2010, ApJ, 709, 670 Goldstein, M. L. 1978, ApJ, 219, 700 Harper, G. M., Richter, M. J., Ryde, N., Brown, A., Brown, J., Kopp, R. A., & Holzer, T. E. 1976, SoPh, 49, 43 Kudoh, T., Shibata, K. 1999, ApJ, 514, 493 Kulow, J. R., France, K., Linsky, J., & Loyd, R. O. P. 2014, ApJ, 786, 132 Lammer, H., Selsis, F., Ribas, I., Guinan, E. F., Bauer, S. J., & Weiss, W. W. 2003, ApJ, 598, L121 Landini, M., & Monsignori-Fossi, B. C. 1990, A&AS, 82, 229 Lecavelier des Etangs, A., Ehrenreich, D., Vidal-Madjar, A., Ballester, G. E., D´esert, J. M., Ferlet, R., H´ebrard, G., Sing, D. K., Tchakoumegni, K. O., & Udry, S. 2010, A&A, 514, A72 Linsky, J. L., Yang, H., France, K., Froning, C. S., Green, J. C., Stocke, J. T., & Osterman, S. N. 2010, ApJ, 717, 1291 Matsumoto, T., & Suzuki, T. K. 2012, ApJ, 749, 8 Matsumoto, T., & Suzuki, T. K. 2014, MNRAS, 440, 971 Matthaeus, W.H., Zank, G.P., Oughton, S., Mullan, D.J., Dmitruk, P. 1999, ApJ, 523, L93 Miyamoto, M., Imamura, T., Tokumaru, M., Ando, H., Isobe, H., Asai, A., Shiota, D., Toda, T., Hausler, B., Patzold, M., Nabatov, A., & Nakamura, M. 2014, ApJ, 797, 51 Murray-Clay, R. A., Chiang, E. I. & Murray, N. 2009, ApJ, 693, Greathouse, T. K., & Strong, S. 2009, ApJ, 701, 1464 23 Hartmann, L., & Avrett, E. H. 1984, ApJ, 284, 238 Heyvaerts, J. & Priest, E. R. 1983, A&A, 117, 220 Hollweg, J. V. 1982, ApJ, 254, 806 Owen, J. E., & Adams, F. C. 2014, MNRAS, 444, 3761 Rogers, L. A., Bodenheimer, P., Lissauer, J. J., and Seager, S. 2011, ApJ, 738, 59 12 Sano, T., Inutsuka, S., & Miyama, S. 1999, in Astrophysics and Space Science Library, Vol 240, Numerical Astrophysics 1998, ed. S. M. Miyama, K. Tomisaka, & T. Hanawa (Boston, MA: Kluwer), 383 Showman, A. P., and Guillot, T. 2002, A&A, 385, 166 Sorahana, S., Suzuki, T. K., & Yamamura, I. 2014, MNRAS, 440, 3675 Sutherland, R. S., & Dopita, M. A. 1993, ApJS, 88, 253 Suzuki, T. K. 2007, ApJ, 659, 1592 Suzuki, T. K., & Inutsuka, S. 2005, ApJ, 632, L49 Suzuki, T. K., & Inutsuka, S. 2006, JGR, 111, A06101 Suzuki, T. K., Imada, S., Kataoka, R., et al. 2013, PASJ, 65, 98 Tanaka, Y. A., Suzuki, T. K., & Inutsuka, S. 2014, ApJ, 792, 18 Terasawa, T., Hoshino, M., Sakai, J. I., & Hada, T. 1986, J. Geophys. Res., 91, 4171 Trammell, G. B., Arras, P. & Li, Z. Y. 2011, ApJ, 728, 152 Trammell, G. B., Li, Z. Y. & Arras, P. 2014, ApJ, 788, 161 Tremblin, P., & Chiang, E. 2013, MNRAS, 428, 2565 Tsuneta, S., Ichimoto, K., Katsukawa, Y., Lites, B. W., Matsuzaki, K., Nagata, S., Orozco Su´arez, D., Shimizu, T., Shimojo, M., Shine, R. A., Suematsu, Y., Suzuki, T. K., Tarbell, T. D., & Title, A. M. 2008, ApJ, 688, 1374 Vidal-Madjar, A., Lecavelier des Etangs, A., D´esert, J. M., Ballester, G. E., Ferlet, R., H´ebrard, G., & Mayor, M. 2003, Nature, 422, 143 Vidal-Madjar, A., D´esert, J. M., Lecavelier des Etangs, A., H´ebrard, G., Ballester, G. E., Ehrenreich, D., Ferlet, R., McConnell, J. C., Mayor, M., & Parkinson, C. D. 2004, ApJ, 604, L69 Yelle, R. V. 2004, Icarus, 170, 167 Yelle, R. V. 2006, Icarus, 183, 508
1005.1701	1	1005	2010-05-11T01:47:57	A Search for Water Masers in the Saturnian System	[ "astro-ph.EP" ]	We searched for H2O 6(1,6)-5(2,3) maser emission at 22.235 GHz from several Saturnian satellites with the Nobeyama 45m radio telescope in May 2009. Observations were made for Titan, Hyperion, Enceladus and Atlas, for which Pogrebenko et al. (2009) had reported detections of water masers at 22.235 GHz, and in addition for Iapetus and other inner satellites. We detected no emission of the water maser line for all the satellites observed, although sensitivities of our observations were comparable or even better than those of Pogrebenko et al.. We infer that the water maser emission from the Saturnian system is extremely weak, or sporadic in nature. Monitoring over a long period and obtaining statistical results must be made for the further understanding of the water maser emission in the Saturnian system.	astro-ph.EP	astro-ph	A Search for Water Masers in the Saturnian System Shigeru Takahashi [email protected] Shuji Deguchi, Nario Kuno, Tomomi Shimoikura∗ Nobeyama Radio Observatory, National Astronomical Observatory of Japan 462-2 Nobeyama, Minamimaki, Minamisaku, Nagano 384-1305, Japan and Fumi Yoshida National Astronomical Observatory of Japan 2-21-1, Osawa, Mitaka, Tokyo 181-8588, Japan (Received 2010 February 11; accepted ; version 0.0 May. 9, 2010) Abstract We searched for H2O 6(1,6)-5(2,3) maser emission at 22.235 GHz from sev- eral Saturnian satellites with the Nobeyama 45m radio telescope in May 2009. Observations were made for Titan, Hyperion, Enceladus and Atlas, for which Pogrebenko et al. (2009) had reported detections of water masers at 22.235 GHz, and in addition for Iapetus and other inner satellites. We detected no emission of the water maser line for all the satellites observed, although sensitivities of our observa- tions were comparable or even better than those of Pogrebenko et al.. We infer that the water maser emission from the Saturnian system is extremely weak, or sporadic in nature. Monitoring over a long period and obtaining statistical results must be made for the further understanding of the water maser emission in the Saturnian system. Iapetus, Key words: planets and satellites: individual Titan, Hyperion, Enceladus and Atlas 1. Introduction Maser emissions are widely found in celestial objects such as dense cores of molecular clouds and circumstellar envelopes of late-type stars (Reid & Moran 1981). Masers have been used as probes of gas with the H2 number density of typically 104 -- 1010 cm−3. For solar system objects, several maser and laser phenomena have been found; e.g., CO2 (Venus and Mars: ∗ present address: Department of Astronomy and Earth Sciences, Tokyo Gakugei University, 4-1-1 Nukui- Kita-Machi, Koganei, Tokyo 184-8501, Japan 1 Mumma 1992) and OH (for many comets: e.g. Crovisier et al. 2002). Each phenomenon would be induced by different physical processes. While a thermal 22.235 GHz water line was possibly detected for comet Hale-Bopp (Bird et al. 1997), the first detection of H2O maser in the solar system was reported at the catastrophic impact of comet Shoemaker-Levy9 and Jupiter (Cosmovici et al. 1996). This report suggests that such an incident can induce collisional pumping for water masers. Recently, Pogrebenko et al. (2009) (abbreviated as POG hereafter) have reported the detections of H2O masers from the Saturnian satellites (Titan, Hyperion, Enceladus and Atlas) with the Medicina 32m and Metsahovi 14m telescopes. This is interesting because, unlike a temporal phenomenon such as the break-up and disruption of a comet, we can perform long period monitoring of H2O emission using ground-based telescopes, space telescopes (e.g. Herschel Space Telescope) and spacecrafts. So far, we do not have much knowledge about the maser mechanism in the solar system, although a lot of water maser phenomena are observationally studied for extra-solar objects. The combination of ground, space and in-situ observations would contribute to understand the nature of water maser emission if the presence in the Saturnian system is verified. Therefore, we must accumulate more data of the water maser lines in the Saturnian system. In this letter, we report our trial of detecting water maser satellites with the 45m radio telescope at Nobeyama Radio Observatory (NRO). We observed the major Saturnian satellites for which POG reported the detections, and in addition, we observed a few other inner satellites. 2. Observations We observed the Saturnian satellites with the NRO 45m telescope at the water maser frequency 22.23508 GHz on 15-17, 24, 27, 28 May in 2009. We used a cooled HEMT amplifier as the receiver front ends and an acousto-optical spectrometer (AOS-H) as back ends. The total band width of an acousto-optical spectrometer AOS-H was 40 MHz, and frequency resolution was 37 KHz, which corresponds to velocity resolution of ∼ 0.6 km s−1. All the observations were made using the position switching method. We pointed the telescope toward the center of Titan and Saturn. A typical integration time of one scan was 20 sec. As the half-power beam width (HPBW) of the telescope (∼ 72" at 23 GHz) was larger than the angular diameter of the Saturn (41" including the ring), we observed several satellites simultaneously. For example, we could observe Hyperion when we observed Titan on 16, 17, 18 and 28 May, and Enceladus and Atlas when we observed the Saturn throughout the observational days, though a slight offset of pointing resulted in the worse detection upper limit. Furthermore, we select Iapetus as the OFF points of the position switching. We segregated these satellites using the differences of Doppler shift frequency in the data analysis. Telescope pointing was checked with nearby strong H2O maser stars. The antenna temperature, T∗ A, was obtained by the chopper wheel method correction for atmospheric and antenna ohmic losses. A typical range of the system temperature 2 was 120 -- 300K, which depended on weather condition and airmass at the observations. Table 1 gives a summary of the observations. We performed base line fitting with the 3rd order polynomial to eliminate the continuum emission from the sky and Saturn. Then, we co-added and binned these spectra with taking the Doppler shift for each object into consideration. Radial velocities of the Saturnian satellites were calculated using the JPL Horizons On-Line Ephemeris System (Giorgini et al. 1996). The resultant velocity resolution was about ∆v ≃ 1 km s−1. These processes were done on the software package NEWSTAR, and procedures to obtain the final spectra after the Doppler correction were carried out on the software which had been developed for the solar system objects. 3. Results and Discussion Figure 1 shows the obtained spectra (upper two and lower-left panels) and the illustration of the satellite positions (lower-left panel) at the time of observations for Titan, Hyperion and Iapetus. Figure 2 also shows the spectra (upper panels) and illustrations of positions (lower panels) for Enceladus and Atlas. The position of a satellite θ is defined as an angle between the line of sight towards Saturn and the line of a satellite and Saturn (i.e. θ = 0 when the object transits on the Saturn seen from the Earth). Table 2 shows a typical 1σ upper limit of the signal considering the factor of a satellite position inside the telescope beam. The conversion A), 2.8 Jy K−1, was used for the calculations. efficiency of flux density to antenna temperature (T∗ The observational results show the data combined on each day and the entire days during the observation period. POG reported that the water line detections of Titan and Hyperion were ∼30 mK (3.8σ for 8 hour integration) and ∼50 mK (4.0σ for 6 hours integration), respectively, which correspond to 300 mJy and 500 mJy, respectively. They found that these emissions were seen on almost the same satellite positions for both objects, and based on these results, they inferred a common mechanism of the emission such as Saturnian magnetosphere bow shock. On 16-18 and 28 May 2009, Titan and Hyperion were near conjunction so that we could observe them simultaneously in the telescope beam. During 16-18 May, especially, these objects were at almost the same position. Therefore, if the maser emissions had been caused by the common mechanism, we would have detected them simultaneously for these two objects. However, our observations this time could not detect the signals of the emissions, and all the data combined also showed no symptom. We estimated typical values of the 3σ upper limit for each daily data. We also obtained 3σ upper limits during the observations using all the daily data combined. The typical(all) 3σ upper limits were 200(60) mJy for Titan and 240(80) mJy for Hyperion, which means that we could certainly detect the line emissions if the levels of flux densities were as high as those of POG. Although POG did not report the water maser emission on Iapetus, we monitored this 3 satellite during our observation period aiming this object at the OFF-point of the position switching observations. We could not find any signal stronger than ∼210(75) mJy (3σ upper limit) . The H2O ice and vapor plume on Enceladus has been reported by the Cassini Ultraviolet Imaging Spectrometer (Hansen et al. 2006 and Hansen et al. 2008), and the hypothesis that there exists liquid water in the crust has been proposed (Porco et al. 2006). As for the water maser emission, the mass ratio of water vapor to ice in the plume is important because a majority of water molecules must be in the form of vapor in order to have maser emission originated from the Enceladus plume. A recent theoretical study indicates that the plume would be dominated by vapor from the thermodynamics perspective (Kieffer et al. 2009). The maser line intensity reported by POG was 500 mJy (4.2σ), and they estimated the column density of water vapor from the observed maser intensity, which agrees with the column density of the water vapor plume observed in UV (n= 1.5 × 1016 cm−2: Hansen et al. 2006). Nevertheless, we did not find appreciable maser emission in our data. The 3σ upper limits were 780(540) mJy. Our data were worse compared with those of POG because Enceladus was located around the edge of the telescope beam much of the observational time. However, these results might indicate that the flux of the plume is varying, or the plume is sporadic; like a geyser. POG reported the most certain detections for Atlas. The averaged spectrum showed a peak with 32 mK and S/N=7.0. From the satellite positions, they found that the maser emissions occurred on the trailing side which was several thousand km away from Atlas, and suggested that the disturbance of the Atlas's motion had caused the emission in the edge regions of Saturnian rings A and F. We attempted to verify this subject in our data. We divided the positions of Atlas into 3 parts; i.e., position 1 and 3: positions passing before the geometrical edges of the Saturn's rings seen from the Earth, and Position 2: those after the rings (see the positions of Atlas in Figure 2). For each position, we combined the data, and furthermore, we tested the data on 19 May because Atlas passed by the geometrical edges of the rings (θ = 90◦) during the observations. However, we could not have positive results for each case. All the combined data did not show any prominent feature of the emission. The 3σ detection limits for each data set were in the range of 430 -- 520(400) mJy. We also checked several inner satellites which have a diameter larger than ∼10 km; Mimas, Janus, Epimeteus, Prometheus, Pandora and Pan. For all the satellites, the acquired data did not show indicative of the maser emission. The typical value of 3σ upper limits were 500(210) mJy. We had an opportunity to observe the Saturnian water maser line at the time of the ring's disappearance; that is, we see the Saturn ring almost in the edge-on view, and we obtained the data which had comparable or even better sensitivities than those by POG for almost of the satellites, though our observations were made for a limited period in 2009 May. If maser emission is stationary in intensity at the levels as those reported by POG, we could detect them. 4 However the results were negative for all of the observed Saturnian satellites. From these results, we conclude that the water maser in the Saturnian system may be sporadic in nature and it is strongly restricted to the time and position of satellites. We have to monitor the satellites for longer periods and to obtain statistical results. These studies would be useful to figure out the water maser emission reported in the Saturnian system. In addition, we should also perform monitoring observations for other icy bodies; Jovian satellites, comets, outer asteroids and Kuiper belt objects would be inside the scope. As suggested by Cosmovici et al. (1996), the maser emission may be induced by catastrophic events. Such events like a disruption or eruption can occasionally be found among the solar system objects; (e.g. C/1999 S4 Linear: disruption and dissipation, 29P/Schwassmann-Wachmann: outburst, 7968 Elst-Pizarro: impact or cometary activity (Toth 2000 and Hsieh et al. 2004), etc.). We should not miss events which will occasionally happen and carry out the observations to accumulate the data. 4. Summary • A search for water maser emission (22.22351 GHz) from several Saturnian satellites; Titan, Hyperion, Enceladus Atlas, Iapetus, and other inner satellites with a diameter larger than 10 km (Mimas, Janus, Epimeteus, Prometheus, Pandora and Pan) were carried out with the 45m radio telescope at Nobeyama radio observatory. • We could not confirm any emission line for all the satellites. The typical 3σ upper limits of daily(average) data for Titan, Hyperion, Enceladus and Atlas were 200(60), 240(80) 780(540) and 430-520(400) mJy, respectively, and for the other inner satellites, the S/N obtained were almost the same. • The sensitivities were comparable or even better than those of POG for most of the satellites. • From our observations, we infer that water maser emission in the Saturnian system is sporadic in nature. We are grateful to Dr. J. Crovisier, the referee of this paper for suggestions to improve the manuscript, and S. Takano and J. Maekawa for advice and discussions. This research made use of ephemeris given by NASA Jet Propulsion Laboratory, California Institute of Technology. References Bird, M. K., Janardhan, P., Wilson, T. L., Huchtmeier, W. K., Gensheimer, P., & Lemme, C. 1997, Earth Moon and Planets, 78, 2 Cosmovici, C. B., Montebugnoli, S., Orfei, A., Pogrebenko, S., & Colom, P. 1996, Planet. Space Sci., 44, 735 Crovisier, J., Colom, P., G´erard, E., Bockel´ee-Morvan, D., & Bourgois, G. 2002, A&A, 393, 1053 5 Table 1. Observational Summary. Day Time (UT) IT (sec)∗ IS (sec)† May 15 08:49-13:51 05:49-11:12 07:48-13:48 09:28-10:44 5260 3140 2180 2000 08:44-12:37 - 05:19-06:40 06:43-13:29 07:29-11:15 07:16-11:44 4000 2000 2000 2000 16 17 18 19 23 24 27 28 - - - 2000 1560 3000 2020 2020 2320 Total 20580 17100 ∗ ON-source integration time of Titan. † ON-source integration time of Saturn. Table 2. Observational Results. Date (or Data) 1σ U.L. (mJy)∗ Satellite Titan 16-19,24,27,28 All Hyperion 16-18,28 All Iapetus 16-19,24,27,28 All Enceladus 15,19,23,24,27,28 All 65 20 80 25 70 25 260 180 Atlas Pos.1,2,3 170,140,170 19be.,19af. 140,150 Other satellites 16-19,24,27,28 All 130 170 70 ∗ A typical 1σ upper limit of the signal considering the factor of a satellite position inside the telescope beam. The observational results show the data combined on each day and the entire days during the observation period. For Atlas, we show the results at each position. 6 Fig. 1. Observational results and positions of satellite for Titan, Hyperion and Iapetus. Each data set has been shifted and plotted every 50 mK. Horizontal axis is target-centric velocity offset. The observed date in May 2009 (as shown in Table 1) is indicated on each spectrum. The position of a satellite θ is defined as an angle between the line of sight toward Saturn and the line of a satellite and Saturn (i.e. θ = 0 when the object transits on the Saturn seen from the Earth). The radius in the position plot (lower-right panel) does not illustrate a real scale for each satellite. Gerard, E., Bockelee-Morvan, D., Colom, P., & Crovisier, J. 1993, Astrophysical Masers, 412, 468 Giorgini, J. D., et al. 1996, Bulletin of the American Astronomical Society, 28, 1158 Hansen, C. J., Esposito, L., Stewart, A. I. F., Colwell, J., Hendrix, A., Pryor, W., Shemansky, D., & West, R. 2006, Science, 311, 1422 Hansen, C. J., et al. 2008, Nature, 456, 477 Hsieh, H. H., Jewitt, D. C., & Fern´andez, Y. R. 2004, AJ, 127, 2997 Kieffer, S. W., Lu, X., McFarquhar, G., & Wohletz, K. H. 2009, Icarus, 203, 238 Mumma, M. J. 1993, Astrophysical Masers, 412, 455 Pogrebenko, S. V., et al. 2009, A&A, 494, L1 Porco, C. C., et al. 2006, Science, 311, 1393 Toth, I. 2000, A&A, 360, 375 Reid, M. J., & Moran, J. M. 1981, ARA&A, 19, 231 7 Fig. 2. Observational results and positions of satellites for Enceladus and Atlas. The labels 19af. and 19be. mean the combined results of 19 data (θ > 90◦) and (θ < 90◦), respectively. 8
1208.4115	2	1208	2012-09-07T00:46:24	Cyclic Transit Probabilities of Long-Period Eccentric Planets Due to Periastron Precession	[ "astro-ph.EP" ]	The observed properties of transiting exoplanets are an exceptionally rich source of information that allows us to understand and characterize their physical properties. Unfortunately, only a relatively small fraction of the known exoplanets discovered using the radial velocity technique are known to transit their host, due to the stringent orbital geometry requirements. For each target, the transit probability and predicted transit time can be calculated to great accuracy with refinement of the orbital parameters. However, the transit probability of short period and eccentric orbits can have a reasonable time dependence due to the effects of apsidal and nodal precession, thus altering their transit potential and predicted transit time. Here we investigate the magnitude of these precession effects on transit probabilities and apply this to the known radial velocity exoplanets. We assess the refinement of orbital parameters as a path to measuring these precessions and cyclic transit probabilities.	astro-ph.EP	astro-ph	Submitted for publication in the Astrophysical Journal Preprint typeset using LATEX style emulateapj v. 5/2/11 CYCLIC TRANSIT PROBABILITIES OF LONG-PERIOD ECCENTRIC PLANETS DUE TO PERIASTRON PRECESSION Stephen R. Kane1, Jonathan Horner2, Kaspar von Braun1 Submitted for publication in the Astrophysical Journal ABSTRACT The observed properties of transiting exoplanets are an exceptionally rich source of information that allows us to understand and characterize their physical properties. Unfortunately, only a relatively small fraction of the known exoplanets discovered using the radial velocity technique are known to transit their host, due to the stringent orbital geometry requirements. For each target, the transit probability and predicted transit time can be calculated to great accuracy with refinement of the orbital parameters. However, the transit probability of short period and eccentric orbits can have a reasonable time dependence due to the effects of apsidal and nodal precession, thus altering their transit potential and predicted transit time. Here we investigate the magnitude of these precession effects on transit probabilities and apply this to the known radial velocity exoplanets. We assess the refinement of orbital parameters as a path to measuring these precessions and cyclic transit probabilities. Subject headings: planetary systems -- celestial mechanics -- ephemerides -- techniques: photometric 1. INTRODUCTION The realization that we have crossed a technol- ogy threshold that allows transiting planets to be de- tected sparked a flurry of activity in this direction af- ter the historic detection of HD 209458 b's transits (Charbonneau et al. 2000; Henry et al. 2000). This has resulted in an enormous expansion of exoplanetary sci- ence such that we can now explore the mass-radius relationship (Burrows et al. 2007; Fortney et al. 2007; Seager et al. 2007) and atmospheres (Agol et al. 2010; Deming et al. 2007a; Knutson et al. 2009a,b) of planets outside of our Solar System. Most of the known tran- siting planets were discovered using the transit method, but some were later found to transit after first being detected using the radial velocity technique. Two no- table examples are HD 17156 b (Barbieri et al. 2007) and HD 80606 b (Laughlin et al. 2009), both of which are in particularly eccentric orbits. Other radial velocity planets are being followed up at predicted transit times (Kane et al. 2009) by the Transit Ephemeris Refinement and Monitoring Survey (TERMS). Planets in eccentric orbits are particularly inter- esting because of their enhanced transit probabilities (Kane & von Braun 2008, 2009). This orbital eccentric- ity also makes those planets prone to orbital precession. In celestial mechanics, there are several kinds of pre- cession which can affect the orbital properties, spin rotation, and equatorial plane of a planet. These have been studied in detail in reference to known transiting planets, particularly in the context of the precession effects on transit times and duration (Carter & Winn 2010; Heyl & Gladman 2007; Jord´an & Bakos 2008; Miralda-Escud´e 2002; P´al & Kocsis 2008; Ragozzine & Wolf 2009). One Damiani & Lanza 2011; [email protected] 1 NASA Exoplanet Science Institute, Caltech, MS 100-22, 770 South Wilson Avenue, Pasadena, CA 91125 2 Department of Astrophysics & Optics, School of Physics, University of New South Wales, Sydney, 2052, Australia consequence of these precession effects is that a planet that exhibits visible transits now may not do so at a different epoch and vice versa. Here we present a study of some precession effects on known exoplanets. The aspect which sets this apart from previous studies is that we are primarily interested in planets not currently known to transit, particularly long-period eccentric planets which have enhanced tran- sit probabilities and larger precession effects. We inves- tigate the subsequent rate of change of the transit prob- ability to show how they drift in and out of a transit- ing orientation. We calculate the timescales and rates of change for the precession and subsequent transit prob- abilities and discuss implications for the timescales on which radial velocity planets will enter into a transiting configuration, based upon assumptions regarding their orbital inclinations. We finally compare periastron argu- ment uncertainties to the expected precession timescales and suggest orbital refinement as a means to measure this effect. 2. TRANSIT PROBABILITY Here we briefly describe the fundamentals of the geo- metric transit probability for both circular and eccentric orbits. For a detailed description we refer the reader to Kane & von Braun (2008). In the case of a circular orbit, the geometric transit probability is defined as follows Pt = Rp + R⋆ a (1) where a is the semi-major axis and Rp and R⋆ are the radii of the planet and host star respectively. More gen- erally, both the transit and eclipse probabilities are in- versely proportional to the star -- planet separation where the planet passes the star-observer plane that is perpen- dicular to the plane of the planetary orbit. The star -- planet separation as a function of orbital eccentricity e 2 Stephen R. Kane et al. 3. AMPLITUDE OF PERIASTRON (APSIDAL) PRECESSION Periastron (or apsidal) precession is the gradual ro- tation of the major axis which joins the orbital ap- sides within the orbital plane. The result of this pre- cession is that the argument of periastron becomes a time dependent quantity. There are a variety of factors which can lead to periastron precession, such as gen- eral relativity (GR), stellar quadrupole moments, mu- tual star -- planet tidal deformations, and perturbations from other planets (Jord´an & Bakos 2008). For Mercury, the perihelion precession rate due to general relativistic effects is 43′′/century (0.0119◦/century). By compari- son, the precession due to perturbations from the other Solar System planets is 532′′/century (0.148◦/century) while the oblateness of the Sun (quadrupole mo- ment) causes a negligible contribution of 0.025′′/century (0.000007◦/century) (Clemence 1947; Iorio 2005). Here we adopt the formalism of Jord´an & Bakos (2008) in evaluating the amplitude of the periastron precession. We first define the orbital angular frequency as n ≡ r GM⋆ a3 = 2π P (4) where G is the gravitational constant, M⋆ is the mass of the host star, and P is the orbital period of the planet. The total periastron precession is the sum of the individ- ual effects as follows ωtotal = ωGR + ωquad + ωtide + ωpert (5) where the precession components consist of the pre- cession due to GR, stellar quadrupole moment, tidal deformations, and planetary perturbations respectively. Jord´an & Bakos (2008) conveniently express these com- ponents in units of degrees per century. The components of ωquad and ωtide have a−2 and a−5 dependencies respec- tively. Since we are mostly concerned with long-period planets in single-planet systems, we consider here only the precession due to general relativity since this is the dominant component in such cases. This imposes a lower limit on the total precession of the system, particularly for multi-planet systems. This precession is given by the following equation ωGR = 7.78 (1 − e2) (cid:18) M⋆ M⊙(cid:19)(cid:18) a 0.05/AU(cid:19)−1(cid:18) P day(cid:19)−1 (6) with units in degrees per century. To examine this precession effect for the known exo- planets, we use the data extracted from the Exoplanet Data Explorer, described in Section 2. The GR preces- sion rates for these planets are shown in Figure 2 as a function of eccentricity, where the radius of the point for each planet is logarithmically scaled with the or- bital period. As a Solar System example, the precession rate for Mercury is shown using the appropriate symbol. There are two distinct populations apparent in Figure 2 for which the divide occurs at a periastron precession of ∼ 0.1◦/century. It is no coincidence that this divide corresponds to the known relative dearth of planets in the semi-major axis range of 0.1 -- 0.6 AU (Burkert & Ida 2007; Cumming et al. 2008; Currie 2009). As expected from Equation 6, the amplitude of the precession is dominated by the orbital period rather than Fig. 1. -- Transit probability for a sample of the known exoplanets as a function of orbital period. In cases where a change in ω from current to 90◦ results in a transit probability improvement > 1%, a vertical arrow indicates the improvement. is given by r = a(1 − e2) 1 + e cos f . (2) where f is the true anomaly, which describes the location of the planet in its orbit, and so is a time dependent variable as the planet orbits the star. For a transit event to occur the condition of ω + f = π/2 must be fulfilled (Kane 2007), where ω is the argument of periastron, and so we evaluate the above equations with this condition in place. The geometric transit probability may thus be re-expressed as Pt = (Rp + R⋆)(1 + e cos(π/2 − ω)) a(1 − e2) (3) which is valid for any orbital eccentricity. Note that these equations are independent of the true inclination of the planet's orbital plane. Given the sensitivity of transit probability to the argu- ment of periastron, it is useful to assess how the proba- bilities for the known exoplanets would alter if their ori- entation was that most favorable for transit detection: ω = 90◦. We extracted data from the Exoplanet Data Explorer3 (Wright et al. 2011) which include the orbital parameters and host star properties for 592 planets and are current as of 30th June 2012. For each planet, we calculate transit probabilities for two cases: (1) using the current value of ω, and (2) using ω = 90◦. The transit probabilities for case (1) are shown in Figure 1. Those planets whose case (2) probabilities are improved by > 1% are indicated by a vertical arrow to the im- proved probability. There are several features of note in this figure. The relatively high transit probabilities be- tween 100 and 1000 days are due to giant host stars whose large radii dominates the probabilities (see Equation 3)). There are several cases of substantially improved tran- sit probability, most particularly HD 80606 b, which is labelled in the figure. The following sections investigate the periastron precession required to produce such an increase in transit probability. 3 http://exoplanets.org/ Cyclic Transit Probabilities 3 vector, which is usually aligned with the rotation axis of the host star. The precession is caused by the oblate- ness of the star which results in a non-zero gravitational quadropole field. This has the potential to be the dom- inant source of precession when the orbit is polar. For example, the nodal precession for the near-polar retro- grade orbit of WASP-33 b has been calculated by Iorio (2011) to be 9 × 109 times larger than that induced on the orbit of Mercury by the oblateness of the Sun. A description of nodal precession and its effect on transit durations has been provided by Miralda-Escud´e (2002). The frequency of nodal precession can be ex- pressed as Ω = n R2 ⋆ a2 3J2 4 sin 2i (7) where n is the orbital angular frequency described in Equation 4, J2 is the quadrupole moment, and i is the or- bital inclination relative to the stellar equatorial plane. A typical quadrupole moment for the star may be ap- proximated as J2 ∼ 10−6 and one may expect a rela- tively aligned orbit such that sin 2i ∼ 0.1. For a typical hot Jupiter, values for a are 10R⋆, whereas for Mercury a = 83R⋆. Since the nodal precession is in units of the orbital angular frequency, one can see that the result- ing precession rate is typically several orders of magni- tude smaller than that of a hot Jupiter, even at the or- bital distance of Mercury. This effect is generally only considered for circular orbits, most notably for short- period orbits that are the most frequently encountered nature of known transiting planets. Here, we are consid- ering longer period eccentric orbits where this is a much smaller effect on the orbital dynamics of the planet. 4. CYCLIC TRANSIT EFFECTS As discussed in Section 2, the transit probability for a given planet is a function of the periastron argument for orbits with non-zero eccentricity (Kane & von Braun 2008). The precession of the periastron argument thus leads to a cyclic change in the transit probability. Here we quantify this cyclic behaviour and determine rates of change and total timescales. Using the periastron precession rates calculated in Sec- tion 3 and combining these with the transit probability equations of Section 2 allows us to compute the time dependent transit probability for each planet. Recall also that this cyclic behaviour will only occur for planets which have non-zero eccentricities. Shown in Figure 4 are three examples of this time dependence over a pe- riod of 100,000 years. When viewing such a plot one is tempted to interpret the cyclic variability in terms of the orbital period, however this variation is caused by the pe- riastron precession, not the orbital period. There is, of course, some period dependency involved, in that shorter period orbits will tend to have a higher cyclic frequency. The planets shown here (HD 88133 b, HD 108147 b, and HD 190360 c) have orbital periods of 3.4, 10.9, and 17.1 days respectively (Butler et al. 2006; Wright et al. 2009). HD 108147 b, in particular, displays very large amplitude variations due to the relatively high eccentricity of its or- bit (e = 0.53). HD 190360 c has a smaller eccentricity and periastron precession rate, which leads to a cyclic timescale much greater than 100,000 years. We have performed these calculations for a subset of Fig. 2. -- Calculated GR periastron precession rates plotted as a function of eccentricity for the known exoplanets with Keple- rian orbital solutions. The radius of the points is logarithmically proportional to the orbital period of the planet. The symbol for Mercury is used to indicate its position on the plot. Fig. 3. -- Lines of constant GR periastron precession as a function of orbital period and eccentricity, assuming a solar-mass host star. The eccentricity of the orbit only plays a significant role at very large values (e > 0.8). The symbol for Mercury is used to indicate its position on the plot. the orbital eccentricity. Thus, even planets in eccentric orbits do not exhibit significant GR precession at longer periods. This is further demonstrated in Figure 3 where we show lines of constant precession as a function of pe- riod and eccentricity for a solar-mass host star. This shows that the GR periastron precession is almost inde- pendent of orbital eccentricity except at extreme values of e > 0.8. Once again, the location of Mercury on the plot is indicated using the appropriate symbol. As noted the (2002) by Miralda-Escud´e (2008), and Jord´an & Bakos total precession time scales are large. Thus what really matters is the rate of change of the periastron argument and quantifying when it is worth returning to a particular target for re-investigation. This is the context of our analysis in Section 4. 3.1. Nodal (Orbital Plane) Precession For completeness, we briefly consider the effects of nodal precession. Nodal precession occurs when the or- bital plane precesses around the total angular momentum 4 Stephen R. Kane et al. Fig. 4. -- Cyclic transit probabilities resulting from GR periastron precession for three known exoplanets: HD 88133 b, HD 108147 b, and HD 190360 c. This is shown from the present epoch and projected 100,000 years from now. the known exoplanets using the data extracted from the Exoplanet Data Explorer, described in Section 2. We re- strict our sample to those planets which are not known to transit and have non-zero eccentricities. The results of these calculations are shown in Table 1 for 60 of the plan- ets. The calculated values include the periastron pre- cession rate ( ωGR), transit probability (Pt), maximum transit probability at ω = 90◦ (P ′ t ), time from the cur- rent epoch until maximum transit probability (∆t), and the transit probability rate of change (dPt/dt). The table has been sorted according to dPt/dt which is presented in units of %/century. The dPt/dt values have been cal- culated from the current epoch over the coming century and thus represents the present rate of change. The im- portance of this is that dPt/dt is not constant and indeed can have negative values as the periastron argument ro- tates past ω = 90◦. Specifically, dPt/dt will be negative for 90◦ < ω < 270◦ and positive elsewhere. This further restricts the planets considered to those whose current ω falls in this range such that dPt/dt > 0. It can be clearly seen that the time required to reach maximum transit probability is immense, certainly be- yond the lifetime of anyone reading this work. However, the rate of change can yield an improved idea of which planets may have a measurable change in configuration. Consider the case of HD 156846 b, whose orbital param- eters and transit potential have been studied in detail by Kane et al. (2011). This is one of the planets in the table with the longest period and also has one of the highest orbital eccentricities. The transit probability is relatively high for this planet and is close to the max- imum probability since ω only needs to change by 38◦. Even so, observations of the periastron precession are un- likely for the timescales involved. By contrast, the hot Saturn HD 88133 b discovered by Fischer et al. (2005) has the highest transit probability rate of change. 5. CONCLUSIONS Transiting planets have become an essential compo- nent of exoplanetary science due to the exceptional op- portunities they present for characterization of these planets. Many of the known exoplanets discovered through the radial velocity technique are currently not known to transit. However, transit probabilities can be substantially improved if the periastron argument ap- proaches ω = 90◦. Since, for eccentric orbits, the peri- astron argument is time dependent as a result of their precession, planets which do not transit at the present epoch may transit in the future and vice versa. The planet Mercury falls quite central to the current distri- bution of calculated periastron precessions for the known exoplanets. This distribution has an eccentricity depen- dence but is most strongly affected by the orbital period. If a precession rate for a given planet is found to be markedly different from our calculations then this could be indicative of further, as yet undiscovered planets in that system. These additional planets would normally be detected from the radial velocity data unless insuffi- cient observations allow them to remain hidden. The periastron precession leads to a cyclic transit prob- ability variation for all exoplanets with non-zero eccen- tricities. Timescales vary enormously but will likely lead to many of these planets transiting their host stars at some point in the future. A reasonable question to ask at this point is if the periastron arguments of the known planets are known with sufficient precision to detect pre- cession in any acceptable timeframe. Once again, we exploit the data extracted from the Exoplanet Data Ex- plorer, described in Section 2. The uncertainties associ- ated with the values of ω for all these planets have a mean of 28◦ and a median of 15◦. This is much higher than the precession effects shown in Table 1. A program of refining the orbits of the known exoplanets, such as that described by Kane et al. (2009), would result in many of these precession effects to be detectable in reasonable time frames. For example the first planet in the table, HD 88133 b, has a precession rate that will cause a shift of ∼ 0.3◦ per decade. Uncertainties on ω of less than one degree are not unsual and can certainly be achieved for those planets in particularly eccentric orbits. The exo- planet HD 156846 b has a current ω uncertainty of 0.16◦ (Kane et al. 2011) which demonstrates that such refine- ment is possible even for relatively long-period planets. More data and longer time baselines will produce subse- quent improvements for many more planets which can re- sult in the detection of the precession for high-precession cases. The relevance of this work may be extended to the Ke- pler mission which has detected many candidate multi- planet systems (Borucki et al. 2011a,b; Batalha et al. Cyclic Transit Probabilities 5 Exoplanet Periastron Precession, Transit Probabilities, and Timescales TABLE 1 Planet P (days) e ω ωGR (◦/cent) Pt (%) P ′ t (%)1 ∆t (cent)2 dPt/dt (%/cent)3 HD 88133 b HD 76700 b HD 73256 b HD 108147 b HD 102956 b BD -08 2823 b HD 7924 b HD 68988 b HD 1461 b HD 217107 b HD 168746 b HD 149143 b HD 162020 b HD 187123 b HD 47186 b BD -10 3166 b HD 69830 b HD 190360 c upsilon And b HD 179079 b 51 Peg b HD 10180 c HIP 57274 b HD 147018 b HD 16417 b HD 10180 d HD 163607 b HD 224693 b 4 UMa b 61 Vir c HD 102117 b HD 43691 b 70 Vir b HD 156846 b HD 16141 b GJ 785 b HIP 57274 c HD 4113 b rho CrB b HD 45652 b HD 20868 b 61 Vir d 55 Cnc c HD 60532 b HD 145457 b GJ 581 d HD 5891 b HD 1237 b HD 17092 b HD 22781 b HD 107148 b BD +48 738 b HD 180314 b HIP 14810 c HD 8574 b HD 216770 b HD 93083 b HD 11977 b HD 222582 b HD 231701 b 3.42 3.97 2.55 10.90 6.49 5.60 5.40 6.28 5.77 7.13 6.40 4.07 8.43 3.10 4.08 3.49 8.67 17.11 4.62 14.48 4.23 5.76 8.14 44.24 17.24 16.36 75.29 26.73 269.30 38.02 20.81 36.96 116.69 359.51 75.52 74.39 32.03 526.62 39.84 43.60 380.85 123.01 44.38 201.30 176.30 66.64 177.11 133.71 359.90 528.07 48.06 392.60 396.03 147.77 227.00 118.45 143.58 711.00 572.38 141.60 0.13 0.09 0.03 0.53 0.05 0.15 0.17 0.12 0.14 0.13 0.11 0.02 0.28 0.01 0.04 0.02 0.10 0.24 0.01 0.12 0.01 0.08 0.19 0.47 0.20 0.14 0.73 0.05 0.43 0.14 0.12 0.14 0.40 0.85 0.25 0.30 0.05 0.90 0.06 0.38 0.75 0.35 0.05 0.28 0.11 0.25 0.07 0.51 0.17 0.82 0.05 0.20 0.26 0.15 0.30 0.37 0.14 0.40 0.73 0.10 349.0 30.0 337.3 308.0 12.0 30.0 25.0 31.4 58.0 24.4 17.0 0.0 28.4 24.5 59.0 334.0 340.0 5.2 51.0 357.0 58.0 279.0 81.0 336.0 77.0 292.0 78.7 6.0 23.8 341.0 279.0 290.0 358.7 52.2 42.0 15.0 356.2 317.7 303.0 273.0 356.2 314.0 57.4 351.9 300.0 356.0 351.0 290.7 347.4 315.9 75.0 358.9 303.1 327.3 26.6 281.0 333.5 351.5 319.0 46.0 2.9490 2.1838 4.3194 0.5732 1.2451 0.9420 1.0901 1.0214 1.1102 0.8192 0.8587 2.1614 0.5283 3.0953 1.8945 2.3445 0.4923 0.1833 1.8588 0.2481 1.8602 1.1232 0.5075 0.0437 0.1935 0.2001 0.0336 0.1008 0.0025 0.0453 0.1351 0.0612 0.0090 0.0049 0.0163 0.0141 0.0500 0.0031 0.0419 0.0380 0.0019 0.0072 0.0335 0.0040 0.0056 0.0089 0.0049 0.0072 0.0020 0.0014 0.0342 0.0008 0.0019 0.0049 0.0028 0.0075 0.0041 0.0006 0.0010 0.0057 14.6 12.9 16.1 5.1 22.8 11.6 7.2 8.7 9.4 6.9 7.2 15.5 4.7 13.5 11.0 8.7 5.4 4.3 12.0 5.3 10.1 7.6 5.3 2.3 6.3 3.6 8.3 3.2 17.8 2.1 3.3 2.6 1.9 4.4 2.3 1.4 1.9 0.9 2.3 1.7 1.3 0.8 2.1 1.6 5.3 0.7 4.8 0.6 2.8 0.4 2.1 5.5 2.3 1.1 1.1 0.8 1.1 2.2 0.5 1.2 17.0 13.5 16.8 13.4 23.6 12.4 7.9 9.1 9.6 7.4 7.7 15.8 5.3 13.6 11.0 8.9 6.1 5.2 12.0 6.0 10.1 8.8 5.3 4.1 6.4 4.7 8.4 3.4 21.7 2.5 4.2 3.4 2.7 4.8 2.5 1.6 2.0 4.3 2.6 3.8 2.4 1.5 2.1 2.1 6.5 0.9 5.2 1.8 3.4 1.8 2.1 6.7 3.8 1.4 1.3 1.8 1.3 3.3 1.5 1.2 34.2 27.5 26.1 247.7 62.6 63.7 59.6 57.4 28.8 80.1 85.0 41.6 116.6 21.1 16.4 49.5 223.5 462.8 21.0 374.8 17.2 152.2 17.7 2607.5 67.2 789.4 336.7 833.1 26488.3 2405.3 1266.0 2612.5 10097.0 7699.5 2950.5 5332.1 1876.0 42500.7 3510.1 4660.2 48796.3 19010.2 972.3 24667.9 26940.2 10602.7 20191.3 22236.9 52495.0 93315.9 439.0 113252.5 78180.5 25078.6 22763.8 22495.8 28700.1 153321.8 126637.2 7727.6 0.101368 0.038099 0.033421 0.028841 0.022932 0.022925 0.019663 0.015372 0.011871 0.010737 0.010620 0.009380 0.009352 0.006702 0.006690 0.006174 0.004500 0.003171 0.003147 0.002675 0.002169 0.002059 0.001124 0.000922 0.000801 0.000780 0.000406 0.000283 0.000263 0.000227 0.000169 0.000152 0.000124 0.000116 0.000105 0.000089 0.000083 0.000082 0.000055 0.000036 0.000035 0.000034 0.000033 0.000032 0.000032 0.000029 0.000027 0.000027 0.000016 0.000015 0.000015 0.000015 0.000014 0.000013 0.000013 0.000012 0.000010 0.000010 0.000009 0.000008 1 2 3 t refers to the transit probability where ω = 90◦. P ′ ∆t refers to the time until P ′ dPt/dt is calculated over the coming century but is a time dependent quantity. t occurs. 6 Stephen R. Kane et al. 2012), most of which are likely to be real exoplanets (Lissauer et al. 2012). Due to simply geometric transit probabilities, most of these systems will certainly have planets which are not transiting the host star at present. The known transiting multi-planet systems are largely in circular orbits, but may have periastron precession due to perturbations from other planets leading to an eventual transit from currently non-transiting planets in the system. For example, Kepler-19 c is known to exist from Transit Timing Variations of the inner planet, but does not currently have a detectable transit signature. Similarly, some of these planets will cease exhibiting an observable transit signature. Issues such as these are im- portant for considering the completeness of these surveys in determining multi-planetary system architectures. ACKNOWLEDGEMENTS The authors would like to thank David Ciardi and Solange Ramirez for several useful discussions and the numerous people who have requested that we perform this study. We would also like to thank the anonymous referee, whose comments greatly improved the quality of the paper. JH gratefully acknowledges the financial sup- port of the Australian government through ARC Grant DP0774000. This research has made use of the Exo- planet Orbit Database and the Exoplanet Data Explorer at exoplanets.org. This research has also made use of the NASA Exoplanet Archive, which is operated by the Cal- ifornia Institute of Technology, under contract with the National Aeronautics and Space Administration under the Exoplanet Exploration Program. REFERENCES Agol, E., Cowan, N.B., Knutson, H.A., Deming, D., Steffan, J.H., Henry, G.W., Charbonneau, D. 2010, ApJ, 721, 1861 Barbieri, M., et al. 2007, A&A, 476, L13 Batalha, N.M., et al. 2012, ApJS, submitted (arXiv:1202.5852) Borucki, W.J., et al. 2011, ApJ, 728, 117 Borucki, W.J., et al. 2011, ApJ, 736, 19 Burrows, A., Hubeny, I., Budaj, J., Hubbard, W.B. 2007, ApJ, 661, 502 Burkert, A., Ida, S. 2007, ApJ, 660, 845 Butler, R.P. et al. 2006, ApJ, 646, 505 Carter, J.A., Winn, J.N. 2010, ApJ, 716, 850 Charbonneau, D., Brown, T.M., Latham, D.W., Mayor, M., 2000, ApJ, 529, L45 Clemence, G.M. 1947, RvMP, 19, 361 Cumming, A., Butler, R.P., Marcy, G.W., Vogt, S.S., Wright, J.T., Fischer, D.A. 2008, PASP, 120, 531 Currie, T. 2009, ApJ, 694, L171 Damiani, C., Lanza, A.F. 2011, A&A, 535, 116 Deming, D., Richardson, L.J., Harrington, J. 2007, MNRAS, 378, 148 Iorio, L. 2005, A&A, 433, 385 Iorio, L. 2011, Ap&SS, 331, 485 Jord´an, A., Bakos, G.A. 2008, ApJ, 685, 543 Kane, S.R. 2007, MNRAS, 380, 1488 Kane, S.R., von Braun, K. 2008, ApJ, 689, 492 Kane, S.R., von Braun, K., 2009, PASP, 121, 1096 Kane, S.R., Mahadevan, S., von Braun, K., Laughlin, G., Ciardi, D.R. 2009, PASP, 121, 1386 Kane, S.R., et al. 2011, ApJ, 733, 28 Knutson, H.A., et al. 2009a, ApJ, 690, 822 Knutson, H.A., Charbonneau, D., Cowan, N.B., Fortney, J.J., Showman, A.P., Agol, E., Henry, G.W. 2009b, ApJ, 703, 769 Laughlin, G., Deming, D., Langton, J., Kasen, D., Vogt, S., Butler, P., Rivera, E., Meschiari, S. 2009, Nature, 457, 562 Lissauer, J.J., et al. 2012, ApJ, 750, 112 Miralda-Escud´e, J. 2002, ApJ, 564, 1019 P´al, A., Kocsis, B. 2008, MNRAS, 389, 191 Ragozzine, D., Wolf, A.S. 2009, ApJ, 698, 1778 Seager, S., Kuchner, M., Hier-Majumder, C.A., Militzer, B. 2007, ApJ, 669, 1279 Fischer, D.A., et al. 2005, ApJ, 620, 481 Fortney, J.J., Marley, M.S., Barnes, J.W., 2007, ApJ, 659, 1661 Henry, G.W., Marcy, G.W., Butler, R.P., Vogt, S.S., 2000, ApJ, Wright, J.T., Upadhyay, S., Marcy, G.W., Fischer, D.A., Ford, E.B., Johnson, J.A. 2009, ApJ, 693, 1084 Wright, J.T., et al., 2011, PASP, 123, 412 529, L41 Heyl, J.S., Gladman, B.J. 2007, MNRAS, 377, 1511
1202.5314	1	1202	2012-02-23T21:00:09	Searching for Planets During Predicted Mesolensing Events: II. PLAN-IT: An Observing Program and its Application to VB 10	[ "astro-ph.EP", "astro-ph.IM", "astro-ph.SR" ]	The successful prediction of lensing events is a new and exciting enterprise that provides opportunities to discover and study planetary systems. The companion paper investigates the underlying theory. This paper is devoted to outlining the components of observing programs that can discover planets orbiting stars predicted to make a close approach to a background star. If the time and distance of closest approach can be well predicted, then the system can be targeted for individual study. In most cases, however, the predictions will be imprecise, yielding only a set of probable paths of approach and event times. We must monitor an ensemble of such systems to ensure discovery, a strategy possible with observing programs similar to a number of current surveys, including PTF and Pan-STARRS; nova searches, including those conducted by amateurs; ongoing lensing programs such as MOA and OGLE; as well as MEarth, Kepler and other transit studies. If well designed, the monitoring programs will be guaranteed to either discover planets in orbits with semi-major axes smaller than about two Einstein radii, or else to rule out their presence. Planets on wider orbits may not all be discovered, but if they are common, will be found among the events generated by ensembles of potential lenses. We consider the implications for VB 10, the first star to make a predicted approach to a background star that is close enough to allow planets to be discovered. VB 10 is not an ideal case, but it is well worth studying. A more concise summary of this work, and information for observers can be found at https://www.cfa.harvard.edu/~jmatthews/vb10.html.	astro-ph.EP	astro-ph	Submitted to the Astrophysical Journal Searching for Planets During Predicted Mesolensing Events: II. PLAN-IT: An Observing Program and its Application to Rosanne Di Stefano1, James Matthews1,2, and S´ebastien L´epine3,4 VB 10 ABSTRACT The successful prediction of lensing events is a new and exciting enterprise that provides opportunities to discover and study planetary systems. The com- panion paper investigates the underlying theory. This paper is devoted to outlining the components of observing programs that can discover planets or- biting stars predicted to make a close approach to a background star. If the time and distance of closest approach can be well predicted, then the system can be targeted for individual study. In most cases, however, the predictions will be imprecise, yielding only a set of probable paths of approach and event times. We must monitor an ensemble of such systems to ensure discovery, a strategy possible with observing programs similar to a number of current surveys, including PTF and Pan-STARRS; nova searches, including those con- ducted by amateurs; ongoing lensing programs such as MOA and OGLE; as well as MEarth, Kepler and other transit studies. If well designed, the mon- itoring programs will be guaranteed to either discover planets in orbits with semi-major axes smaller than about two Einstein radii, or else to rule out their presence. Planets on wider orbits may not all be discovered, but if they are common, will be found among the events generated by ensembles of potential lenses. We consider the implications for VB 10, the first star to make a pre- dicted approach to a background star that is close enough to allow planets to be discovered. VB 10 is not an ideal case, but it is well worth studying. A more concise summary of this work, and information for observers can be found at https://www.cfa.harvard.edu/∼jmatthews/vb10.html. 1. Introduction 1.1. A Brief History of Prediction When Einstein was first convinced to publish a paper on the generation of gravitational lensing events, he expressed doubt that the effect would be observed (Einstein 1936). The 1Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138 2School of Physics & Astronomy, University of Southampton, Southampton, SO17 1BJ, UK 3Department of Astrophysics, Division of Physical Sciences, American Museum of Natural History, Central Park West 79th Street, New York, NY 10024 4City University of New York, New York, NY -- 2 -- probability of an event occurring is very low. In addition, detection can be made challenging through the "dazzling by the light of the much nearer star", the lens. Einstein's doubts have not deterred astronomers, with their familiarity of the universe of possible lenses and background sources, from predicting specific lensing events. In 1966, Feibelman suggested that the star 40 Eridani-a, a member of a triple which also contains a white dwarf, would pass close enough to a background star in 1988 to produce detectable lensing effects. Given its mass of 0.84 M⊙ and its distance of only 5 pc, 40 Eridani-a has an Einstein angle, θE, of about 37 milliarcseconds (mas) when it lenses a star at much larger distances from us. Unfortunately, though, studies of the motion during the following 20 years found that the distance of closest approach would be approximately 3′′, about 81 θE (Feibelman 1986), too distant for observations at that time to detect the effects of lensing. In addition, the closest approach between 40 Eridani-a and the background star was predicted to occur in June, when Eridani, a winter constellation, would have been unobservable. Any plans to monitor the region were abandoned. In 2001, Paczy´nski suggested that the isolated neutron star RX J185635-3754 would pass close enough to a background star (within about 0.3′′) that its action as a gravitational lens would produce an astrometric shift that could be detected with HST. As reported by Neuhauser et al. 2011, the Ph.D. thesis of Eisenbeiss finds that there is no star close enough to the paths of any of the seven isolated neutron stars known as the Magnificent Seven (Haberl 2007) to produce detectable lensing through the rest of the 21st century. Recently, our group predicted a very close approach between the high-proper-motion dwarf star VB 10, and a dim background star (L´epine & Di Stefano 2011). With two HST images of the region, and the results of previous of astrometric studies of VB 10, we thought at first that we could predict the distance of closest approach to within several mas, and also the approximate date of closest approach. Taking into account the uncertainties, including the uncertainty in the motion of the background star, however, we realized that existing observations are consistent with a bundle of paths, each characterized by a date and distance of closest approach. The most likely date was during December of 2011, when VB 10 is a twilight star, very close to the Sun. Nevertheless, the predicted approach is very close: the probability is ∼ 65% that the distance of closest approach will lie in the range from less than an Einstein angle to 15 θE. The time of closest approach could be during the spring of 2012, even though the most likely dates are earlier. Faced with these uncertainties, and also with the fact that the background star is dim, we set out to assess the value of an observing campaign. To explore the possibilities in detail, we asked whether, if VB 10 has planets, they could enhance the lensing signature or even produce detectable events at different times. The answer to both questions is yes, and the details are reported in the companion paper (Di Stefano et al. 2012a, hereafter Paper I). 1.2. Possible Observing Programs In spite of the difficulty of predicting individual events, Paczy´nski (1986) found a way to transform the study of lensing events into an observational enterprise. He overcame Ein- stein's concern about low probability by pointing out that, fifty years later, computers had made it possible to track the light curves of many millions of stars per night, guaranteeing -- 3 -- the detection of some events if large, dense stellar fields could be monitored. He countered the possibility that dazzling by the lens could prevent the detection of lensing by focusing on the possibility of lensing by compact dark matter objects. Within a few years, several monitoring programs were each monitoring large portions of the Magellanic Clouds and the Galactic bulge to find evidence of microlensing. The challenge that the first generation of microlensing observing programs faced was to prove that they could extract from the haystack of ubiquitous stellar variability, the signal of a relatively small number of lensing events (Udalski et al. 1993, Alcock et al. 1993, Aubourg et al. 1993) They succeeded, and the present generation of lensing monitoring teams are discov- ering more than 1500 events per year1. By demonstrating that long-term, high-cadence monitoring can reveal variability of many types, producing a high science yield, these projects have inspired many other astronomical time domain studies. These include the Palomar Transient Factory (PTF; Rau et al. 2009, Law et al. 2009), the Panoramic Survey Telescope & Rapid Response System (Pan-STARRS; Kaiser et al. 2002), and the Large Synoptic Survey Telescope (LSST; Tyson 2002). One particularly innovative approach has been designed to search for evidence of transits of Earth-like planets around M dwarfs. The MEarth Project is a survey to photo- metrically monitor approximately 2000 nearby M dwarfs, searching for transits by planets that could be habitable Super-Earths (Irwin et al. 2009a). The project has had some success in the detection of Super-Earths (Charbonneau et al. 2009) and also in discovering a number of eclipsing binary systems (Irwin et al. 2009b; Irwin et al. 2010). The unique feature that we will find potentially important in monitoring the sites of predicted lensing events (§5) is that MEarth monitors many directions per night, because their target stars are spread over a large portion of the sky. The same will be true for predicted lensing events. 1.3. Plan of the Paper We develop an observational strategy to systematically study predicted lensing events with the goal of learning about the planetary system of the lens star. The companion paper (Di Stefano et al. 2012) presents the theory. Here we focus on observations. In Sections 2 and 3 we describe what we can learn by monitoring the site of a predicted event during the months and years prior to and after the event, as well as during the time of closest approach. Section 2 is devoted to planning observations for a specific case meant to represent an ideal situation. Section 3 is devoted to planning a real observing campaign for the star VB 10, which is generating the first predicted event about which we can hope to derive observational constraints. In Section 4 we address the issue of how many events are likely to be predicted each year. Since the majority of predictions will not be precise, with the time and distance of closest approach not known in detail, we turn in Section 5 to observational programs designed to monitor the sites of thousands of predicted events. By monitoring a large enough number of events, we can be guaranteed to detect the lensing signatures of some foreground stars and their planetary retinues. The prospects for 1ogle.astrouw.edu.pl/; www.phys.canterbury.ac.nz/moa/ -- 4 -- successful observing programs are discussed in Section 6. In addition to predicted lensing events, there are at least two connections to ongoing observing programs: The first is to current surveys for lensing events; the second connection is to transit studies. 2. Monitoring Individual Predicted Events 2.1. Events to target Multiple precise measurements of the relative positions of source and lens can, in some cases, allow the time and distance of closest approach to be computed with modest uncertainty limits. If the probability is high that the closest approach will lie within θE of the lens, and that the source star will be observable near the time of closest approach, then monitoring the source star during the lensing event can measure the mass of the lens, determine if it has a planetary system, discover planets in a wide variety of orbits and with a broad range of masses, and place limits on the presence of various types of planets (see Paper I for details) . In an ideal case, the source is brighter than the lens in some waveband, bright enough that large telescopes are not needed to detect changes of a few percent in the amount of received light. This circumstance can allow amateurs to play significant roles in the monitoring program, strengthening the results. To make the discussion in this section concrete, we will consider the prediction of an event by a specific star, PMI18537+4929. This is a star with an estimated mass of 0.46 M⊙. PMI18537+4929 is 44.8 pc away and has a proper motion µ = 0.137′′ yr−1. These parameters allow us to compute the value of the Einstein angle, θE = 9.1 mas. We also have the Einstein-diameter crossing time, τE = 49 days. This star falls into the general category of stars expected to produce events at a rate of just under one per century. The Einstein radius, RE = θEDL, where DL is the distance to the lens, is 0.41 AU. If a lensed source is a kpc away, then the value of the Einstein radius projected onto the source plane is 8.9 AU. Thus, even if structures significantly smaller than θE cause short- lived deviations in the lensing light curve, these features are not likely to be washed out by finite-source-size effects. The orbital period of a planet with semi-major axis a = RE is approximately 140 days. If observations are able to detect deviations of 1% in the amount of light received from the source star, then a deviation from baseline is potentially detectable for a maximum of 3.5 τE ∼ 172 days. Thus, rotation could play a role in increasing the detectability of planets in similar orbits. Rotation is less likely to be significant in events caused by wide-orbit planets. On the other hand, deviations caused by close-orbit planets will, in many cases, repeat. (The orbital period for a planet with a = 0.25 RE is about 18 days.) For the purposes of the lensing signatures, the important separation is the projected separation between the lens star and its planet. To simplify this discussion we consider circular face-on orbits. The value of a is constant, and it we define α = a/RE. In Paper I we have shown that it is convenient to consider three classes of orbital separation: wide- orbits (α > 2); orbits in the zone for resonant lensing (0.5 < α < 2), and close-orbit planets (α < 0.5). Events generated by planets in each of these ranges exhibit distinctive characteristics, suggesting that the observing strategy should evolve with time, as orbits -- 5 -- with different values of α become detectable.2 An important parameter in determining what we can learn from a specific predicted event is the angle of closest approach, b, between the foreground and background stars. With β = b/θE, three ranges of values determine which types of planets can be detected. The wide regime covers the entire lens plane; wide-orbit planets can be detected throughout the wide regime, as long as α > β. In addition to wide-orbit planets, close-orbit planets can be detected if β < η, where the value of η depends on the photometric sensitivity and is typically larger than 3 but smaller than 10. In addition to wide-orbit and close-orbit planets, planets in the zone for resonant lensing can be detected in the resonant regime, (β < 2). More details are presented in Paper I. Because the goal of this section is to determine the best day-by-day monitoring strat- egy, we have chosen a star that is more typical than VB 10 of what we can expect in the future. We do, however, assume that the observing conditions will be better than those expected for VB 10. First, we assume that the closest approach occurs when the lens and source are observable: that is, they appear in the night sky, close to neither the Sun nor Moon . Second, we assume that the source star is bright enough that changes in the light we receive from it of about a percent can be detected. Third, we assume a distance of closest approach smaller than θE. We chose a close approach because it allows us to explore the full range of behavior which is potentially detectable. 2.2. Detection as a function of time We considered the possibility that PMI18537+4929 has a Jupiter-mass planet in a face-on, circular orbit. To study detectability, we conducted a simulation in which we generated 20,000 orbits with values of α ranging from 0.1 to 20. For each value of α, we also generated a random starting value for the orbital phase. We then followed the path of both the star and its planet as the center of mass executes a straight-line path making a closest approach of β = 0.5 to a bright background star at time t = t0. We computed the magnification, Apl(t), caused by the planetary system at each time, and compared it with the magnification, Apt(t) expected if PMI18537+4929 alone had been traveling along the line followed by the planetary system's center of mass. At each time, we asked whether the value of Apl(t) − Apt(t) had reached an extremum with a value greater than 0.01. If so, we stored the value of α and (t − t0). In order to display the structures in the densest regions, within about 40 days of t0, we did not plot all of the points. Nevertheless, the relative density of points in a given region roughly corresponds to the probability that a planetary system with a specific value of α will produce a light curve with significant deviations at time t. Points in orange correspond to caustic crossings. The magenta curves that start in the upper right and upper left of the figure were calculated with the analytic formula given in Equation 3 of Paper I, which applies to planets in wide orbits. The cyan curves that start in the lower right and lower left of the figure were calculated with the 2Note that, although this classification is very useful, the values for the boundaries given above are somewhat arbitrary, because event characteristics depend on the mass ratio q = mp/M∗, where mp is the mass of the planet and M∗ is the mass of the star. -- 6 -- analytic formula given in Equation 6 of Paper I, which applies to close-orbit planets. The black points, from our simulation, generally follow these curves, especially for the widest and closest orbits; the curves therefore provide good guides to the positions that would be occupied by low-probability events. To determine what type of event will be detected as time goes on, we consider a ruler, with its straight edge coincident with the vertical axis. As time goes on, the ruler moves to the right. The points and/or curves it intersects at each time show the types of events that can be detected at that time. 2.2.1. Early times and late times Approximately 500 days before the closest approach between the nearby star and the foreground source, a planet with α = 20 could make a very close approach to the background star, producing detectable lensing effects. The radius of the Einstein ring of the planet is θE,pl = r mp M∗ θE = 0.42 mas r mp mJ . (1) The Einstein diameter crossing time, τE,pl is approximately 2.3q mp days. If 1% deviations from baseline are detectable, the planet-induced event could last as long as ∼ 8 days. The peak magnification induced in the source star will be large if the distance of closest approach between the planet and background star is small. mJ One goal of an observing program would be to discover any lensing events generated by wide-orbit planets. In addition, by measuring the magnification as any discovered planet- lens events progress, we can gather enough data to perform a model fit and to extract the gravitational mass of the planet. Furthermore, should any of the planets themselves be binaries, this could affect the lensing signature in a distinctive way. These goals can be achieved in a two-stage process. The first step is monitoring designed to discover evidence of lensing. The second step occurs only if it appears that a lensing event is in progress; in this case, the observing plan can be made more intensive for a limited time interval in order to track the progress of the event. (1) Monitoring to discover wide planet-lens events: It is important to make this stage as easy as possible, because there is a long time (in this case years) during which there is a low but significant probability of detecting wide-orbit planets, should they exist. Fortunately the fact that it takes so long for the lens star to travel ∼ 20 θE means that it can take a few days for a planet-lens event to go from start to finish. Monitoring several times a day is therefore adequate. Ideally, there should not be gaps in coverage; this favors developing a team that can conduct observations across longitudes. If a large enough number of telescopes are involved, each can take observations several times a week. With such infrequent observations by individual teams, it becomes important that the data be quickly communicated to one group for each event, in order to identify trends as they occur for the purposes of calling alerts when needed. An advantage of a long baseline of observations is the opportunity it provides to char- acterize the intrinsic variability of both the lens star and the source to be lensed. We will -- 7 -- discuss this more for the specific case of VB 10 in the next section; VB 10 provides a clear lesson that it can be advantageous to understand the variability in different wavebands. (2) Tracking an event: Once an event is detected, it is important to obtain enough data to provide a model fit. If an event can be identified as such before the magnification has risen by about ∼ 34%, then more intensive monitoring will be able to catch the event during the interval in which it is most variable. During this interval, a total rate of observations on the order of every few hours would be adequate for a Jupiter-mass planet. An alert can encourage more observers to join in, providing even more frequent coverage over an interval of a few days. This could prove helpful if the planet happens to be less massive than Jupiter. The teams observing predicted events have a number of decisions to make as they design a program to monitor a specific predicted event. For example, what is the minimum mass to which they would like to be sensitive? Because the time duration scales as the square root of the mass, the complementary question is: What is an appropriate investment of resources? To decide the answers to such question it can be helpful to consider the evolution of the event probability with time. The probability is ∼ 1% to detect a planet with α = 20 (see Table 1), at t − t0 ∼ 500 days. As time goes on, lensing by planets with smaller values of α becomes detectable, and the probability increases. For example, at t − t0 ∼ 35 days we there is a probability of ∼ 11% of detecting a planet with α = 2. α t − t0 P (A > 1.06) P (A > 1.01) ∼ 35 2 3 ∼ 80 5 ∼ 115 10 ∼ 250 20 ∼ 500 0.18 0.12 0.05 0.02 0.01 0.11 0.07 0.03 0.01 0.01 Table 1: Probability of an event with A > 1.06, 1.01 produced by a Jupiter mass plane- tary companion to PMI18537+492, for a range of orbital separations, α, as predicted by Equation (4) of Paper I. This is for a circular face-on orbit; different eccentricities and orientations will produce different probability distributions. 2.2.2. Intermediate times In the interval within about 80 days of the closest passage, a greater variety of events can be detected3. Detection of planets in certain types of orbits is all but guaranteed, should they exist. A well-designed monitoring program will therefore either make discoveries or place strong limits of the presence of planets with orbits characterized by α < 2. 3We define "intermediate times" to refer to the interval which starts when close-orbit planets can be detected. In principle, this interval can start earlier, if highly sensitive photometric observations are possible. To be conservative, here we take the interval to begin about 80 days prior to closest approach, when close binaries with α = 3 produce a strong signal. -- 8 -- By the time the stellar lens has come within about 3 θE of the background source, four things, summarized below, have changed. (1) The probability of events generated by wide-orbit planets has increased. The increase in probability is shown in Table 1 and can be seen in Figure 1: as time goes on and the ruler discussed in the introduction of §2.2 moves to the right, the density of points lining the magenta curve increases. The increase in probability is partly due to the smaller size of the orbit (see Equation 4 of Paper I) and is partly due to the somewhat larger size of the isomagnification contours associated with detectable events. Furthermore, with a larger event probability, every day a planet is not discovered translates into a significant probability that a planet with a given value of α doesn't exist. (2) Events caused by wide-orbit planets are more likely to exhibit binary effects. This is because the stretching of the isomagnification contours is accompanied by other distortions, making the characteristics of wide-planet-lens events more likely to exhibit deviations from the pure point-lens form. One enhanced binary effect is that the caustic structures are larger. An example of a wide-orbit-planet light curve exhibiting binary effects is shown in the red light curve (α = 2.73) on the lower-left of Figure 2. The deviation of this light curve from the underlying green light curve is caused by a close approach of the Jupiter-mass planet to the background star. Although it is difficult to see on this scale, this event actually exhibits a caustic crossing. (3) The magnification due to the stellar lens has increased. At α = 3, the magnification is about 1.7%, potentially detectable. It is convenient to define δ = (A − 1). From the time the value of δ > 0 becomes measurable, the exact distance between the lens star and the background star is known, in units of θE. Each measurement of the magnification can now be used not only to discover planets, but also to measure the mass of the lens. (4) It has become possible to detect close-orbit planets. In the top panel of Figure 2, the green deviation in the light curve at about 390 days (∼ 25 days after closest approach) is caused by a close-orbit planet. A larger set of such deviations is shown in the bottom panel. More light curves for close-orbit planets are shown in Paper I. The key point to take away from these images is that the phases chosen to generate the light curves were random, yet every light curve exhibits deviations. For the case of Jupiter-mass planets orbiting low- mass dwarfs like PMI18537+492 and VB 10, with α approximately equal to 1/3, these deviations can be easily detected. Furthermore, the values of the orbital separation and mass ratio can be determined. Below we summarize results presented in Di Stefano (2011). For each value of δ, observations are sensitive to planets with α = 0.84 δ0.25. If such a planet exists, it will produce a deviation in the light curve like those that shown in Figure 2, for log10(A − 1) = log10(0.25) = −1.6. The relatively short orbital period of about 26 days, means that the deviations will be repetitive, although their size is maximized in the region just mentioned above. If Porb the planet's orbital period then the time duration of a deviation from the point-lens form is Tdev ≈ 2.5 Porb 10(0.5 Log10(q)−0.2) (2) This predicts a deviation duration of about 2 days for a Jupiter-mass planet in an orbit with α = 3.; this is reflected in the light curves of Figure 2. Thus, by deciding on a sampling frequency and the depth of the observations for each measured value of δ, observers are certain to either detect lensing signatures of close-orbit -- 9 -- planets, or else to be able to determine with certainty that certain orbital separations and mass ratios can be ruled out. (1) Monitoring: Once δ is larger than about 0.017, the monitoring program designed to allow event alerts should be stepped up. This is because the probability of detecting certain close-orbit planets, if they exist, approaches unity. Therefore, at this point in the program it will be productive to increase the numbers of telescopes in use, and to attempt observations two times per night with each, if possible. Fast and efficient calls of alerts are needed, because the deviations from the underlying point-lens light curve associated with lensing by the high-proper-motion star, may exhibit more structure than the planet-lens events expected at early and late times. (2) Tracking an event: The basic principle is the same as for tracking events that occur at early or late times. But, because the light curves will tend to exhibit structure on short time scales, more frequent observations after the alert would be fruitful. This portion of the observing plan would be very similar to plans carried out to discover planets in present-day microlensing programs. 2.2.3. Times near the time of closest approach Once the the stellar lens passes within about 2 θE of the background source (within ∼ 35 days of closest approach) we can start to see features of planets in the zone for resonant lensing, in addition to continuing to find evidence for close-orbit and wide-orbit planets. During the interval from 35 days prior to closest approach until 35 days after closest approach we are almost certain to discover evidence of planets in the zone for resonant lensing, if they exist. Sensitivity to close-orbit planets continues during this time as well. Monitoring and Tracking: The program for monitoring to discover events and tracking the evolution of any events discovered should be the same as described above for close-orbit planets. During the time around closest approach, however, it is especially important to have frequent, deep monitoring: (a) close-orbit planets can cause deviations in the magnification as the lens star wobbles in its orbit on short time scales (Figure 4 of Paper I); (b) wide-orbit planets cause subtle deviations from the point-lens form in the magnification produced by the central star, also associated with stellar wobble (Figure 5 of Paper I). 3. Monitoring VB 10 3.1. VB 10 as an example of a nearby lens VB 10 is a nearby high proper motion M dwarf, whose upcoming predicted close approach with a background star was discussed in detail in L´epine & Di Stefano 2011 and in Paper I. VB 10 has an estimated mass of 0.075M⊙, at a distance from us of 5.82 pc, with a proper motion of about 1.5′′yr−1. These parameters allow us to compute the value of the Einstein angle, θE ≈ 10 mas, and the Einstein diameter crossing time, τE ≈ 5 days. As we did for PMI18537+4929, we conducted simulations considering a hypothetical -- 10 -- Jupiter-mass planet orbiting VB 10 in a face-on, circular orbit. For VB 10 we selected a physical path compatible with astrometric measurements of the system, specifically, the path with β = 0.5 (5 mas), shown in Figure 1 of Paper I. The resulting pattern of points in the plane of α − (t − t0) is shown in Figure 3. The most obvious difference from PMI18537+4929 as shown in Figure 1, is the significantly shorter timescale of the event. The reason the time scale for VB 10's approach is almost ten times shorter than that of PMI18537+4929, is VB 10's relatively high angular speed. To quantify the comparison with other nearby stars, we considered ∼ 7500 stars that happen to lie in the Kepler field and which have measured proper motion. Setting the value of DL/DS to zero, we computed the distributions of the values of both θE and τE. Figure 4 shows that both PMI18537+4929 and VB 10 have large Einstein rings, but that neither is among the largest for nearby stars. Yet, while the value of τE for PMI18537+4929 is fairly typical of values for other nearby stars, the value of τE for VB 10 is smaller than for any star in the sample. This is because it is nearby and also because its transverse speed of about 40 km s−1 is also fairly large. 3.2. Observing the first Predicted Mesolensing Event The relative brightness of VB 10 at long wavelengths and its variability (Cutri et al. 2003; Berger et al. 2008; West et al. 2008; Hilton et al. 2010) pose challenges. VB 10 is an active M dwarf and is known to exhibit flares. It is important to be able to distinguish lensing signatures from possible flare activity. For this we must rely on multi-waveband observations and on the pattern of time variability. Flares have a charateristic fast rise and exponential decay. Although rotation effects can alter the light curve (Berger et al. 2008), it should be rare for a flare to mimic a lensing event. Furthermore, the time scale of most flares is shorter than that of the lensing events we have considered. Thus, should we detect evidence of lensing, we expect to be able to distinguish it from flare activity; this is why we emphasize the need for good time coverage and mutiband observations. In fact, during late November we were able to compare MEarth I-band light curves with a small number of 7-band observations with GROND4. This showed that, even when significant I-band variability of VB 10 was exhibited, the variability was less evident or not evident at shorter wavelengths. The primary challenge we face in designing a program to monitor VB 10 is that the date and distance of closest approach are uncertain (L´epine & Di Stefano 2011; Paper I). Despite the difficulties, we can still learn from the VB 10 lensing event, but what we learn depends on the time of closest approach. It is important to note that, although we do not know at the time this text is being written what the distance and time of closest approach was or will be, future high-resolution images will allow us to determine the relative path of VB 10 and [VB 10]-PMLS-1 with small uncertainties. (1) If the closest approach has already occurred, then we may be able to discover lensing signatures from the time of closest approach in existing data, such as the combination of MEarth and GROND data mentioned above. Since, however, dense monitoring has not yet begun, and VB 10 is still near the Sun, we are not likely to detect 4See Greiner et al. (2008) for details of the GROND program. -- 11 -- lensing signatures associated with the closest approach. This means that we will not measure the mass of VB 10, nor will we be sensitive to signatures from planets in the resonant zone. The signature most likely to be detected, however, is that due to wide-orbit planets. Monitoring that starts now and continues for a few months will either detect or place weak limits on the presence of wide-orbit planets. On every day from now onward, the detection or failure to detect a planet-lens signature either discovers or places a weak limit on the presence of a planet with a well defined value of α. In analogy to the work described in §2.2, we can estimate the probability, P , that an event with A > 1.06 will be produced by a Jupiter-mass planet on a wide orbit. While simulations like those that produced the probability plots of Paper I can provide better estimates, Equation 4 of Paper I provides a reasonable guide to the α dependence of of P , especially for small β. The probability is ∼ 1% to detect a planet with α = 20 at t − t0 ∼ 46 days. As time goes on, lensing by planets with smaller values of α becomes detectable, and the probability increases. For example, at t − t0 ∼ 5 days we have a ∼ 27% chance of detecting a planet with α = 2. The increased probabilities compared to PMI18537+4729 reflect the higher mass ratio for a Jupiter-mass planet orbiting VB 10. α t − t0(days) P (A > 1.06) P (A > 1.01) 2 3 5 10 20 ∼ 5 ∼ 7 ∼ 11 ∼ 23 ∼ 46 0.27 0.16 0.08 0.02 0.01 0.47 0.28 0.14 0.03 0.01 Table 2: Probability of an event produced by a Jupiter mass companion to VB 10. with A > 1.06, 1.01 for a range of orbital separations, α, as predicted by Equation (4) of Paper I. (2) If the approach occurs in the near future, we are unlikely to have ideal monitoring during the event. Nevertheless, if VB 10 is orbited by a planet in the zone for resonant lensing, and if the distance of closest approach lies within about 20 mas, we have a good chance of detecting evidence of the planet. This is because the magnification of the [VB 10]-PMLS-1 will be high enough that we will receive roughly as much light from it in B band as we receive from VB 10 itself. If VB 10 is orbited by a close-orbit planet, we may be able to detect evidence of it, but only when observing conditions become more favorable, because close-orbit planets will produce deviations that are typically small and, for VB 10, short-lived as well. The main channel of detection will be in the wide regime. Fortunately, since we will have continuous monitoring from the time of the event, Table 2 shows that we will be able to place meaningful limits for modest values of α, since the probability that such planets produce events is high. (3) If the event occurs in late winter or early spring, we will be able to learn a good deal. It is therefore important to take high-resolution images as soon as possible, to refine the prediction, as in this case, we should enlist a large number of observers. Close-orbit planets: The orbital period can be very short. For example, with α = 1/3, Porb = 3.6 days, so that orbital motion can increase the detection probability to nearly unity. Equation 2 shows that the deviation duration would be ∼ 0.65 days for a Jupiter- mass planet orbiting VB 10 with α = 1/3. The fractional deviation due to the magnification -- 12 -- of the dim background star is small, so that deep observations are required. Planets in the zone for resonant lensing: Orbital motion is still important enough to significantly increase the probability of detection (see Figure 3 of Paper I). The magnification tends to be large, making it easier to identify the magnification of the dim background star. As is the case for close-orbit planets, frequent observations will either discover planets in the zone for resonant lensing, or else will place strong limits on their existence. Wide-orbit planets: There is a high probability of discovering wide-orbit planets during an interval of weeks from the closest approach. Observing Strategy: While intensive monitoring of a system in which the distance and time of approach is uncertain cannot be justified, a discovery would be important. We therefore propose a compromise between the scientific potential and the uncertainty asso- ciated with the event. First we consider the questions of detectability in more detail. If telescopes at approximately ten locations around the globe can each observe the source star 2 − 5 times per night, a reasonably densely sampled light curve can be developed. Because the background star is so dim, the signal-to-noise ratios required to reliably mea- sure the magnification may require meter-class telescopes. The exception would be for high-magnification. Analysis: For each value of α, a set of orbits are possible. This set includes face-on orbits in either direction, inclined orbits, and eccentric orbits. For each type of orbit, and for a wide range of possible planetary masses, we compute the light curves, specifically computing the value of the magnification at each time an observation occurs. This will identify the cases in which we would have had a high or low probability of planet detection, and may even lead to the discovery of a planet. Due to the proximity in time of the VB 10 event, we have not yet conducted a full analysis of orbits in the general case- i.e. including eccentricity and inclination effects. Throughout this paper and Part 1 we have considered circular face on orbits, and will conduct a full general treatment shortly. 4. Will event prediction be common? Feibelman's identification of the possible 40-Eridani event was almost certainly serendip- itous, as was our discovery that VB 10 would make such a close approach to a background star. The relative ease of these predictions suggests that lensing event prediction should not be too difficult. It is therefore important to make a quantitative estimate. This was first done by Paczy´nski (1995), who calculated the area of the region swept out per year within θE of a typical nearby lens within DL = 10 pc. He found that, if the background source density is 0.3 per sq. arcsecond, then each nearby dwarf should produce, on average, slightly fewer than one event per century. Thus, by monitoring a few hundred of the nearest stars, one can expect that a few of them per year will produce events in which the background star is magnified by ∼ 34% or more. Paczy´nski (1996) later went on to consider events displaying the astrometric effects of lensing, for which the distance of closest approach can be larger, leading to a higher event rate. Here we include for the first time, lensing by possible planetary companions to nearby stars. Every time a nearby star produces an event, there is an opportunity to discover -- 13 -- planets in the zone for resonant lensing as well as close-orbit planets. In addition, the stellar-lens light curve may be distorted by the fact that the lens star is wobbling in response to an orbiting wide-orbit planet. In addition to possibly being discovered through their influence on the motion of the star they orbit, when it serves as a lens, wide orbit planets can be discovered by producing independent events that may or may not be preceded or followed by a detectable stellar-lens event. If a population of N nearby stars is being monitored, and each produces events at a rate Rj, where j labels the star, then the rate of independent wide-planet-lens events is Rind wide = N Xj=1 Rj n(j) Xi=1 r mp,i M∗,j , (3) where i labels the planets associated with an individual star, and n(j) is the number of planets orbiting the j th star. The rate of separate planet-lens events can be a significant fraction, as much as several tenths, of the rate at which nearby stars produce events, depending on the number of planets per star and the mass distribution of the planets. We also extend the reach of these studies by considering lensing by somewhat more distant stars. While the proper motion of stars more than ∼ 10 pc away is generally smaller, producing a smaller event rate per star, there are nevertheless a large number of stars within a hundred parsecs or so that have have proper motions larger than 0.15 arcseconds per year. First we consider the rate at which nearby stars produce events, then turn to the enterprise of prediction. To determine the event rate, we have considered approximately 7500 stars with measured proper motion that happen to lie in the field observed by the Kepler space mission. For each of these stars we have estimates of the mass and distance, in addition to the proper motion. This allows us to compute the size of the Einstein ring of each star and the area covered by the Einstein ring per year. We find that, in total, the Einstein rings of these stars cover approximately 3.6 arcseconds per year. Events can be detectable even when the angle of closest approach is larger than θE. For example, deviations of 6% or more are likely to be detectable, so that the angle of closest approach need only be smaller than 2 θE. Thus, the lensing region of nearby stars in the Kepler field covers about 7.2 sq. arcseconds per year. Over the whole sky (which has about 400 times the area of the Kepler field) the lensing regions of nearby stars cover approximately 3000 sq. arcseconds per year. If the background stellar density is 0.3 per sq. arcsecond, then we expect roughly 1000 events across the sky caused by nearby lenses. These events will be generated by over 2 × 106 stars, most of which have already been identified. The question then becomes, how do we make predictions? There are two types of answers to this. The first is to simply develop an ensemble of possible lenses and to arrange to monitor enough of them that we are guaranteed to detect several events per year. We will know which specific monitored stars produce the events only after the events begin. Wide-field surveys that are more-or-less automated can simply keep track of any changes in the amount of light received from the positions around all high-proper-motion (HPM) stars. Surveys that must select a limited number of directions to monitor must know which of the high-proper-motion stars are most likely to produce events. This can be discovered -- 14 -- by computing, for each HPM star, the area it covers in the sky. Those that cover the largest area are generally the best candidates to produce events. By studying the sample of stars in the Kepler field, we find that a small fraction of the HPM stars have a very high probability of producing lensing events: 0.6% of all of the possible lenses will likely produce 10% of the events. The most likely lenses tend to lie within about 100 pc. Thus, subsets of high-probability lenses can easily be selected; in fact we have already done this for the Kepler field (Di Stefano et al. 2012b). These numbers make it clear that it is possible to identify ensembles of nearby stars, each of which has a high probability of producing a lensing event sometime during the next year. In fact, the total probability is so large that, by monitoring these stars, we are guaranteed to discover dozens of event per year. The reason these events are special is that each can yield the gravitational mass of the lens star and can explore its planetary system in detail, potentially discovering several planets per star and measuring the mass and orbital separation of each. It is clear that, if we focus attention on predicted events, we will have a tool that can be continuously employed to study nearby planetary systems. It is not necessary to rely on only statistical probabilities. We can identify potential lens-source pairs by using catalogs of nearby stars with measured proper motion, and by cross correlating them with catalogs of nearby background stars. In fact catalog matches were suggested by Feibelman5, and implemented by Salim & Gould (2000), who used catalogs to identify pairs of nearby stars that could induce astrometric shifts in more distant stars. The idea was to use the then-proposed Space Interferometry Mission (SIM) to observe these events, due to its excellent astrometric precision of ∼ 4µ arcsec. This discussion leads us to develop a hierarchy of prediction. Level 1: At the lowest level are the stars with measured proper motion, such as those in the LSPM catalog (L´epine 2005; L´epine & Shara 2005; L´epine et al. 2002, 2003). Large-scale semi-automated surveys like PTF and Pan-STARRS can keep track of the light curves of each such star in their fields. Level 2: In order to select the best stars to monitor for smaller or less automated surveys, or to decide on a set of fields that might receive highest priority in large surveys, it is important to know which stars have the highest probability of producing lensing events. To identify them, we must be able to estimate the masses of the stars with measured proper motion, and the distances to them. Using this information, we can identify those whose Einstein rings cut the largest swath across the sky. If, for example, it is feasible to monitor 2000 stars, we may decide to select the top 2000 stars from this list. Level 3: At level 3, we consider the background stellar density. Stars that cut a smaller swath across the sky may nevertheless have a higher probability of producing events if the density of stars behind them is larger. Thus, level 3 is simply a reordering of the systems in level 2, based on a more realistic estimate of the event rate per star. Level 4: We consider all of the stars in level 1 and identify those for which there are high-resolution images of the background extending over time. For this set of systems, we can compute the path of the potential lens and identify stars that lie close to it. In cases in which a full astrometric solution is 5Feibelman wrote (1986) "It is hoped that this 20-year exercise in frustration will encourage others to conduct systematic searches, perhaps by means of computer-based data banks, for stars with large proper motions that eventually may eclipse a background star and give rise to the elusive gravitational lens effect." -- 15 -- possible, we can predict the approximate distance and time of closest approach. If there are too many images to analyze, we can choose to start with the potential lenses at the top of the level 3 list, since they have the highest probabilities of creating lensing events. At the end of this process we should be able to predict a set of specific events. Because the number of such events per year depends not just on the properties of nearby stars and the stellar fields in front of which they travel, but also on the availability of data, it is not possible to predict how many individual events can be predicted each year. It seems likely that at least a handful of individual photometric events will be predicted each year. The ability to detect astrometric events with a mission such as Gaia, should lead to a larger rate of successful event predictions. Since the elements of day-by-day monitoring for individual events were covered in sections 2 and 3, we turn below to the issue of monitoring ensembles of predicted events. 5. Monitoring Ensembles of Sites of Predicted Events Once an ensemble of potential future lensing events is identified, monitoring to detect evidence of lensing can begin. The ideal approach is a partnership among several groups. This minimizes gaps in coverage, while reducing the telescope time needed from any single group. It is important to have a central clearinghouse that receives data soon after each observation, so that photometric changes can be detected quickly enough to call an alert when lensing events start. At present, alerts are called by lensing monitoring teams so that almost continuous coverage of a lensed star can be achieved during a period when the magnification is changing rapidly due to planet-lens effects (see e.g, Janczak et al. 2010; Miyake et al. 2011). The need for continuous observations will be rare for nearby lenses, since effects caused by nearby planets tend to be longer-lived because of the relatively large sizes of the Einstein rings (Figure 4). Instead, an alert will be designed to encourage more observers to join in to achieve monitoring as frequent as a few times per hour (i.e., a few times per day per team) over an interval of several days. With this type of observational plan in mind, the discussion below explores how different observing teams can begin the search for lensing by an ensemble of stars, each having a high probability of causing lensing events in the near future. 5.1. Wide-Field Surveys With 100 − 1000 nearby-lens events per year, and a detection efficiency likely to be smaller than unity, it is important to survey large regions. Fortunately, wide-field surveys are already being conducted. Some of the teams conducting wide-field monitoring are large professional programs. These include PTF and Pan-STARRS. Others are programs of long-standing run by amateurs.6 These surveys differ from each other in the photometric 6For example, an amateur program that surveys 4500 sq. degrees every two nights to find novae was the first to discover a mesolensing event in the field. The Tago event, named for its discoverer, involved a nearby lens that passed in front of an A0 star 1 kpc away (Fukui et al. 2007; Gaudi et al. 2008a). -- 16 -- sensitivities they achieve and in the observing cadences they employ. None of the wide- field surveys currently operating are searching for lensing across the sky7. The quarry considered most exciting by these teams are explosive transients. Lensing is one of many other types of less dramatic variability, and the existing teams have limited resources for research outside their main purview. Event prediction will allow the teams to sidestep the difficult task of characterizing variability and identifying lensing events among the much larger set of all variables. Simply by supplying these teams with the coordinates of the stars likely to cause events within the next few years, we can make it possible for them to begin identifying event candidates immediately. Such a list is now relatively straightforward to develop. Indeed, our group has already identified several hundred high-priority sites to monitor for the Kepler space mission (§5.5; Di Stefano et al. 2012b). . We also plan to publish an all-sky table of the sites of predicted events, prioritized according to levels 2 and 3. The light curves from these locations should be carefully checked with each observation to assess the probability that variability at the site of a predicted event is due to lensing. Because observations by the wide-field monitoring surveys are taken regularly during an interval of years, they have the advantage of developing a long baseline. Information collected over the this timescale can help to determine if variability at the site of a predicted event is unusual in terms of its magnitude, color, or temporal pattern. One caveat is that most fields covered by these surveys are observed at intervals of several days, which is too infrequent. Since, however, several surveys may be observing the same sites of predicted events, a clearinghouse can compile a composite light curve that can be used to identify lensing-like signatures on a shorter time scale. Once a system is determined to be possibly experiencing lensing, additional observations are needed. These can test the lensing hypothesis and measure the mass of the lens star and/or of its planets. We address the issue of follow-up in §5.4. 5.2. Ongoing Lensing Monitoring Programs Ongoing lensing teams are of course the best equipped surveys to discover and identify lensing events. At present, however, it is difficult to predict specific events for two reasons, both related to the fact that these programs monitor dense stellar fields. First, it is difficult to measure the proper motion of stars traveling across regions with a high surface density of stars, and values found in catalogs are often not reliable8. This problem can be overcome by the monitoring teams themselves: with a long baseline, they can measure the proper motion of nearby stars against the background of distant stars. This has so far only been carried out for part of the data (Alcock. et al. 2011; Rattenbury et al. 2008). Second, the background stars are highly blended, making it difficult to know which specific star might serve as a lensed source. 7Pan-STARRS has a sub-program Pandromeda that has identified lensing candidates in M31. 8In a preliminary study that checked images against the published proper motion of stars near lensing events, we found that the images could fail to confirm the result, presumably because variability in a dense stellar field can mimic stellar motion (McCandlish et al. 2010; McCandlish et al. 2012). -- 17 -- Nevertheless, we know that some of the lensing events discovered by the monitoring programs are generated by nearby lenses. At present, lensing teams like OGLE and MOA are monitoring large regions of the sky, on the order of 200 sq. degrees, or ∼ 0.005 of the sky. If, therefore, there are 1000 events per year generated by stars within about 100 pc, roughly 5 per year of them occur in the fields monitored for evidence of lensing. Although these events may not have been predicted in advance, they are ideal for the types of study outlined for individual predicted events in §2 and §3, because we know that β is small and that the background star is bright enough to render the rest of the event detectable. In particular, the proximity of the lens means that RE is likely to be small, and planets in orbits with small values of α, even with α near unity, will have short orbital periods. This means that, if we conduct frequent enough monitoring, and if we can detect small changes in the amount of light received from the lensed source, we will either discover planets or else will know that there are no planets within about 2 θE of the lens star. There is also a significant probability of discovering wide-orbit planets, if they exist, with a day-by-day monitoring program either finding wide-orbit planets, or placing limits on the existence of planets in well-defined orbits. In addition, the value of θE will be large enough to minimize or eliminate finite-source-size effects. If therefore an event posted online in real time by the monitoring teams9 can be identified as being due to a nearby lens, we can immediately start additional observations which can test for close-orbit planets and planets in the zone for resonant lensing. We can also ensure that these observations continue to supplement those of the lensing teams after the event, to search for evidence of wide-orbit planets. (Please see §5.4.) It is important to note that some of the light-curve features associated with close-orbit planets and/or with planets in the zone for resonant lensing, may make the light curve seem different from a typical lensing light curve. This is because the effects of rotation are to distort the features familiar from the more common static case. It is important that the lensing monitoring teams identify these possibly-unusual-looking events as good lensing candidates, so that they will be targeted for further study. In addition to the handful of events per year generated in the lensing fields by stars within about 100 pc, a much larger number of events are generated by stars within about a kpc (Di Stefano 2008a, 2008b). In fact, as many as 10% of the ∼ 1500 events they post per year, may be caused by stars within a kpc. While these lenses will have somewhat larger values of RE, close-orbit planets and wide-orbit planets can be found in the manner outlined in §2 and §3. The search for planets in the resonant zone can be conducted with existing protocols. This has been demonstrated by the discovery, through a lensing event that exhibited evidence of orbital motion, of a system with 2 planets in the zone for resonant lensing, only 1 kpc away (Gaudi et al. 2008b). The key issue is selecting, from among the lens candidates identified by the monitoring teams, those specific events which happen to have been caused by nearby lenses. This can be accomplished using searches through existing catalogs. In fact our group has already studied catalog matches to all of the events discovered that, over a 15 year interval roughly ∼ 8% of the lensing event candidates have matches to catalogued stars that could corre- 9http://ogle.astrouw.edu.pl/; http://www.phys.canterbury.ac.nz/moa/ -- 18 -- spond to nearby lenses (McCandlish et al. 2010; McCandlish et al. 2012). We are refining the assessment procedure. Ideally, data from the monitoring teams can also be used to determine if there is evidence of a high proper motion star along the line of sight to each event. 5.3. MEarth + 5.3.1. The MEarth Concept As described in §1.2, MEarth monitors 2000 stars across the sky (Irwin et al. 2009a). This all-sky coverage is ideal for a survey of predicted lensing events since the sites of predicted events are scattered across the sky. MEarth consists of a bank of small telescopes which switch from field to field during the course of a night. Because the goal is to discover planets transiting M dwarfs, MEarth monitors in I band. If a MEarth target star happens to lens a background star, then MEarth could discover lensing events in the data it collects. This is most likely to happen if the background star also happens to be luminous in I band. As we see with the VB 10 event, however, the lensed star could provide only a tiny fraction of the light received in I band, rendering a lensing event essentially invisible. 5.3.2. Additions to MEarth While the basic design of MEarth is well suited to the study of the sites of thousands of predicted lensing events spread across the sky, some alterations would make a MEarth-like project better suited to monitoring lensing events. If MEarth, or a follow-up project of similar design were to make the discovery of lensing events by M dwarfs one of its scientific goals, it could take several steps to increase the discovery rate. First, the set of stars to monitor could be chosen with the idea of maximizing the area of the sky covered by the Einstein rings of the targets. In fact, the stars already selected are high-proper-motion stars like VB 10, well suited to lensing studies. As additional stars are selected for study, the likelihood that each will produce detectable lensing in the near future could provide an additional selection criterion. Useful information for this purpose would include not only the size and speed of the dwarf star's Einstein ring, but also information about the background stellar field. Higher priority could be assigned to potential targets whose travels are likely to bring them in front of stars that can be detectably lensed. Second, because the background star to be lensed may not be bright in I band, multi- waveband observations are necessary. These could be taken by a new MEarth-like project, if two or more filters can be utilized. They could also be taken by independent telescope participating in the program with the idea of supplementing the basic MEarth coverage to improve the detection efficiency for lensing events. Finally, the analysis of the data should include fits for lensing models. If software suitable to this purpose is applied to existing MEarth data, it may find evidence of past If a typical MEarth target has θE = 10 mas, and µ = 0.15′′ yr−1, then lensing events. over a 5-year period, the total region producing magnifications of 6% covers a significant portion of the sky, about 60′′. Although not every star is observed at all times, the area covered by the targets during MEarth observations is almost certainly large enough that some events occurred. -- 19 -- 5.4. Collaborative Observation: Before, During, and After The idea of "follow-up" is that no single observing program can collect all of the data needed to discover and study planets around nearby stars. It is therefore important to build a team that works together and that can also call occasional alerts as needed to enlist even more observers. Follow-up has been an important part of lensing searches for planets. During certain events deemed as good candidates for planet searches, alerts are called to start intensive world-wide monitoring. Observations that are almost continuous during an interval that is typically longer than several hours and shorter than a few days are taken by different groups, using telescopes across longitudes. The data are combined to provide exquisitely detailed model fits. For predicted lensing events, the concept of follow-up must be generalized. When a specific event, like the VB 10 event, is anticipated, monitoring must begin before lensing is expected, and continue after the time of closest approach. This is needed to discover planets with a range of orbital periods. Monitoring need not be very frequent: a few times per night for each of several teams spread across longitudes will be sufficient. Once evidence of lensing is detected, more frequent coverage may be called for, depending on the situation. In most cases, a total of a few observations per hour, conducted by a combination of teams, will be able to resolve the lensing light curves enough to determine if there are planets and even to discover multiple planets. In some cases, evidence that a short-lived deviation is occurring may trigger intensive monitoring during a limited time interval, similar to what is required in other planet-lens searches. Monitoring in different wavebands is more likely to avoid the "dazzling" effect of the lens. When an ensemble of high-probability events is being monitored, the same considerations apply. Networks of observers have already been set up and have successfully studies a number of different individual systems, including planetary lenses. These projects have engaged a broad range of participants, including the Las Cumbres Observatory Global Telescope Network (LCOGT10), the AAVSO11 and others. A network for predicted events is presently being formed12. 5.5. Kepler The Kepler mission monitors approximately 150, 000 stars, with the goal of identifying transits by Earth-sized planets. It was not designed to identify lensing events, but we have conducted calculations showing that in the Kepler field there are a large number of possible 10http://lcogt.net/ 11http://www.aavso.org/ 12https://www.cfa.harvard.edu/∼jmatthews/vb10.html -- 20 -- lensing candidates (see Figure 4). The excellent photometry from Kepler means that small perturbations can be reliably detected. Such perturbations could be associated with distant approaches by the lens star, possibly including deviations by planets in very close orbits. By conducting a study of the high-proper-motion stars in the Kepler field, we have identified a set of about 700 stars that have the highest probabilities of producing lensing events. We have created a program to make sure that these stars are included among the Kepler targets. It is almost certain that one of these stars, or even some of the other stars already targeted, will produce low-magnification events during the expected 3.5-year duration of the Kepler mission (Di Stefano et al. 2012b) 5.6. Transits and Lensing: A symbiotic connection The observations needed to monitor predicted lensing events are well suited to discov- ering planetary transits. In fact transit-search programs such as MEarth and Kepler are sensitive to lensing events by the target star and/or its planets (Di Stefano et al. 2012b), while the lensing monitoring teams find evidence of transits (Dreizler et al. 2003; Konacki et al. 2005). The observations we propose will test each nearby potential lens for both effects. The probability that the orbital inclination is favorable for transits is R∗/a.. For a 0.25M⊙ star with a planet in an orbit of 0.13 AU, corresponding roughly to the center of the habitable zone (Di Stefano & Night 2008), this probability is roughly equal to 0.009 . The number, NL, of lensing events caused by nearby a high-proper motion star is NL = 0.0018 yr−1(cid:16) θE 10 mas(cid:17)(cid:16) µ 0.15 arcsec yr−1(cid:17)(cid:16) σ 0.3 arcsec−2(cid:17) (4) where σ is the density of background stars on the sky. Depending on the time scale of the observations, the probability of detecting lensing can be comparable to the probability associated with transits described above. Every time we conduct monitoring for a predicted lensing event, there is an opportunity to look for planets through both their possible lensing and transit signatures. 6. PLAN-IT: Prospects The idea of PLAN-IT is to plan observations of nearby stars likely to serve as lenses. This strategy is different from the now common approach of monitoring many stars in a dense field in hope that an unknown mass will happen to lens a more distant star. By focusing on nearby stars that are potential lenses, we are likely to know the proper motion and distance to the lens, so that the light curve fit, which provides an estimate of the Einstein-diameter crossing time, immediately provides an estimate of the gravitational mass of the nearby star. In addition, we have shown that planets orbiting the nearby star can produce lensing signatures, whether or not lensing by the star is detected. When, however, lensing by the star is detected, tests for planets in all orbital ranges can be carried out. It is even possible that the astrometric effects of lensing will be detected in some cases. -- 21 -- The nearby dwarf star VB 10 is the first test case. This text is being completed in mid- February 2012. If the closest approach between VB 10 and the background star [VB 10]- PMLS-1 has not yet occurred, and if the approach between VB 10 and the background star [VB 10]-PMLS-1 is within about two θE, then the upcoming event will measure the gravitational mass of VB 10, determine with certainty whether it has planets orbiting within about two θE, and quantify the probability that VB 10 has planets in wider orbits. Even if the closest approach has already occurred, and even if the closest approach is (or was) more distant, we can still either discover wide-orbit planets or derive constraints on their presence. These constraints will be very weak, but they nevertheless represent the exercise of a new capability. Enhanced monitoring of the region around VB 10 should begin immediately. To bal- ance the excitement of a potential discovery with the uncertainty that detectable lensing will occur, we propose a modest program: observations from 10 or more locations across longitudes, about twice each night from each location. The observations should be com- bined as they are taken to identify trends so that an alert can be called if there is evidence of deviations which may be short-lived. Frequent I-band observations of VB 10 with MEarth will soon resume, as will observations with GROND. GROND, a 2-m telescope, provides 7-band coverage that has a good chance to detect the lensing of the dim blue background star, [VB 10]-PMLS-1. When new telescopes join in, observations in wavebands blueward of I will be useful. The significance of the observing program extends beyond the VB 10 event. Whether or not a discovery is made, the VB 10 event calls our attention to the fact that prediction programs can now be started, and provides a test case about which to organize the first observing campaign. A modest investment of a few observations per telescope per night, spread out across longitudes will allow us to not only check for lensing events but also to test an observing network that can be called into action for future events. We expect future events worthy of monitoring to be identified regularly, by a pro- cess that proceeds systematically by starting with catalogs of high-proper-motion stars. Groups of such stars with the highest probability of producing events can be identified and monitored by programs that observe large regions of the sky. These programs will range from those run by amateur astronomers who hunt for novae, to large semi-automated pro- grams like PTF and Pan-STARRS, to programs like MEarth that systematically target interesting systems spread across the sky. The stars with the highest probability of producing events in the near future can be identified through studies of their proper motion that incorporate cross-correlation with catalogs of possible background stars and analysis of any existing high-resolution images. This process will lead to the identification of individual events that can be monitored to learn about both the high proper motion star and any planetary system it harbors (§2 and §3). It is worth noting that the sets of observations designed to search for lensing by nearby stars are also ideally suited to the search for transits by planets that may orbit these stars in edge-on orientations. Thus, lensing searches and transit searches are automatically carried out for each star studied. In fact, the ongoing MEarth program could find lensing using its current strategy. This is most likely to happen if the lensed background star happens -- 22 -- to have a color similar to that of the M dwarfs that they are targeting for a transit search. Beyond the serendipitous circumstance of being able to search for planets in two ways at once, lensing searches for planets around nearby stars are intimately connected to other planet-search techniques. This is simply because these lensing studies will discover planets orbiting nearby stars or will place quantifiable limits on the masses and orbital separations. These studies can therefore complement and inform radial velocity and direct imaging studies. In summary, we have proposed a monitoring strategy for VB 10. Analogous considera- tions for other individual predicted events will shape future programs to optimize what we learn about the 'planetary systems of nearby stars predicted to produce lensing events. In addition to the monitoring of individual systems, ensembles of future events can be moni- tored by a variety of surveys. This type of program will produce guaranteed results in the form of discoveries of planets of a variety of masses and with a range of orbital separations orbiting specific nearby stars that can be targeted for complementary observations. Acknowledgments We would like to thank Christopher Stubbs; Christopher Crockett; Fred Walters; David Charbonneau, Zachory Berta and the MEarth project; Jochen Greiner at GROND; Matthew Templeton and the AAVSO for their help and advice with observing VB 10. This work was supported in part by NSF under AST-0908878. REFERENCES Alcock, C., Akerlof, C. W., Allsman, R. A., et al. 1993, Nature, 365, 621 Alcock, et al. 2001, ApJ, 562, 337 Aubourg, E., Bareyre, P., Br´ehin, S., et al. 1993, Nature, 365, 623 Berger, E., et al. 2008 ApJ, 676, 1307 Charbonneau, D. et al. 2009, Nature, 462, 7275 Cutri, R. M., et al. 2003, 2MASS Point Source Catalog, University of Massachusetts and Infrared Processing and Analysis Center, (IPAC/California Institute of Technology) Di Stefano, R., 2008a, ApJ, 684, 46 Di Stefano, R., 2008b, ApJ, 684, 59 Di Stefano, R., 2011, ApJ submitted, arXiv:1112.2366v1 Di Stefano, R., 2012a, Paper I, ApJ submitted Di Stefano, R., 2012b, in preparation Di Stefano, R., & Night, C. 2008, arXiv:0801.1510 Di Stefano, R., & Scalzo, R. A. 1999a, ApJ, 512, 564 Di Stefano, R., & Scalzo, R. A. 1999b, ApJ, 512, 579 -- 23 -- Dreizler, S. et al. 2003, A&A, 402, 791 Feibelman, W. A. 1966, Science 151, 73 Feibelman, W. A. 1986, PASP98, 1199 Fukui, A., et al. 2007, ApJ, 670, 423 Gaudi, B. S., et al. 2008a, ApJ, 677, 1268 Gaudi, B. S., et al. 2008b, Science, 319, 927 Gould, A., & Loeb, A. 1992, ApJ, 396, 104 Greiner, J., et al. 2008 PASP, 120, 405 Griest, K., & Safizadeh, N. 1998, ApJ, 500, 37 Haberl, F. 2007, Ap&SS, 308, 181 Hilton, E. J., et al. 2010,AJ, 140, 1402 Irwin, J. et al, 2009a, IAUS, 253, 37 Irwin, J. et al, 2009b, ApJ, 701, 1436 Irwin, J. et al, 2010, ApJ, 718, 1353 Janczak, J., et al. 2010, ApJ, 711, 731 Kaiser, N. et al., 2002, SPIE, 4836, 154 Konacki, M. et al. 2005, ApJ, 624, 372 Law, N. M., Kulkarni, S. R., Dekany, R. G., et al. 2009, PASP, 121, 1395 L´epine, S. 2005, AJ, 130, 1247 L´epine, S. & Di Stefano, R. 2011, ApJ submitted, arXiv:1111.5850 L´epine, S. Shara, M. M., & Rich, R. M. 2002, AJ, 124, 1190 L´epine, S. Shara, M. M., & Rich, R. M. 2003, AJ, 126, 921 L´epine, S., & Shara, M. M. 2005, AJ, 129, 1483 Mao, S., & Paczy´nski, B. 1991, ApJ, 374, 37 McCandlish, S., et al. 2010, AAS Meeting 217, 347.06 -- 24 -- McCandlish, S., et al. 2012, in preparation Miyake, N., et al. 2010, ApJ, 728, 120 Neuhauser, R. et al. 2011, arXiv:1111.0447 Paczy´nski, B. 1996, Acta Astronomica, 46, 291 Paczy´nski, B. 2001, arXiv:0107443v1 Rattenbury, N. J., & Mao S. 2008, MNRAS, 385, 905 Rau, A., et al. 2009 PASP, 121, 1334 Salim, S., & Gould, A. 2000, ApJ, 539, 241 Tyson, A. J. 2002, SPIE, 4836, 10 Udalski, A., Szymanski, M., Kaluzny, J., et al. 1993, Acta Astron., 43, 289 Van Biesbroeck, G. 1944, AJ, 51, 61 West, A. A., et al. 2008, ApJ, 135, 785 This preprint was prepared with the AAS LATEX macros v5.2. -- 25 -- 10 1 0.1 -400 -200 0 200 400 Fig. 1. -- The logarithm of α, the orbital separation in units of RE, versus (t − t0), where t0 is the time of closest approach. For each point shown, t is the time of a peak in the photometric deviation Apl − Apt from the single lens; we consider only those peaks with Apl − Apt > 0.01. The lens is PMI18537+4729 (see §2), orbited by a hypothetical Jupiter-mass planet in a circular face-on orbit. The distance of closest approach is β = 0.5. The smooth magenta curves that start in the upper corners of the plot show the analytic prediction for events produced by wide-orbit planets. Similarly, the smooth cyan curves that start in the lower corners show the analytic prediction for events produced by close- orbit planets. The orange points show the times of events associated with caustic crossings. -- 26 -- 300 350 400 450 t (days) 1 0 -1 -2 0 -0.5 -1 -1.5 -2 250 Fig. 2. -- Light curves for hypothetical planets orbiting PMI18537+4729. Top Panel: Light curves exhibiting caustic crossings for three different values of α; in each case, β = 0.5. Bottom Panel: Light curves for α = 1/3 for a variety of different distances of closest approach between PMI18537+4729 and a background star. In this case the planet's mass is 3 MJ . The orange dashed curve corresponds to β = 0.5, as in Figure 1. The other curves form a sequence of increasing distance of closest approach, with the highest peak having β = 1/3 and β increasing in increments of 1/3 for each subsequent light curve with lower peak magnification. -- 27 -- 10 1 0.1 -40 -20 0 20 40 Fig. 3. -- The logarithm of α, the orbital separation in units of RE, versus (t − t0), where t0 is the time of closest approach. For each point shown, t is the time of a peak in the photometric deviation Apl − Apt from the single lens; we consider only those peaks with Apl − Apt > 0.01. The lens is VB 10 (see §3), orbited by a hypothetical Jupiter-mass planet in a circular face-on orbit. The distance of closest approach is β = 0.5. The smooth magenta curves that start in the upper corners of the plot show the analytic prediction for events produced by wide-orbit planets. Similarly, the smooth cyan curves that start in the lower corners show the analytic prediction for events produced by close-orbit planets. The orange points show the times of events associated with caustic crossings. -- 28 -- Einstein Angle Distribution PM_I18537+4729 VB10 1000 100 10 1 0 1000 100 10 1 0 bin size 0.5 5 10 15 Einstein Diameter Crossing Time Distribution PM_I18537+4729 VB10 bin size 0.5 50 100 150 Fig. 4. -- Top panel: A histogram showing the distribution of Einstein angles across the whole sky of stars with proper motions > 0.04 "yr−1, extrapolated from a sample of 7474 stars across 100 square degrees in the Kepler field of view. The values for VB 10 and PMI18537+4729 are marked with green and red lines respectively; the bin size is 0.5 mas. These nearby stars are mesolenses, having Einstein rings significantly larger than typical microlenses, which have values of θE often less than a mas. They have high probabilities of serving as lenses and may produce detectable astrometric shifts in the positions of the stars they lens. Bottom Panel: A histogram showing the distribution of Einstein diameter crossing times in days, for the same stars. The values for VB 10 and PMI18537+4729 are again marked with green and red lines.
1308.0562	1	1308	2013-08-01T11:24:58	Structure of Surface-H2O Layers of Ice-covered Planets with High-pressure Ice	[ "astro-ph.EP" ]	Many extrasolar (bound) terrestrial planets and free-floating (unbound) planets have been discovered. The existence of bound and unbound terrestrial planets with liquid water is an important question, and of particular importance is the question of their habitability. Even for a globally ice-covered planet, geothermal heat from the planetary interior may melt the interior ice, creating an internal ocean covered by an ice shell. In this paper, we discuss the conditions that terrestrial planets must satisfy for such an internal ocean to exist on the timescale of planetary evolution. The question is addressed in terms of planetary mass, distance from a central star, water abundance, and abundance of radiogenic heat sources. In addition, we investigate the structures of the surface-H2O layers of ice-covered planets by considering the effects of ice under high pressure (high-pressure ice). As a fiducial case, 1M$\oplus$ planet at 1 AU from its central star and with 0.6 to 25 times the H2O mass of Earth could have an internal ocean. We find that high-pressure ice layers may appear between the internal ocean and the rock portion on a planet with an H2O mass over 25 times that of Earth. The planetary mass and abundance of surface water strongly restrict the conditions under which an extrasolar terrestrial planet may have an internal ocean with no high-pressure ice under the ocean. Such high-pressure-ice layers underlying the internal ocean are likely to affect the habitability of the planet.	astro-ph.EP	astro-ph	Structure of surface-H2O layers of ice-covered planets with high-pressure ice S. Ueta and T. Sasaki Earth and Planetary Sciences, Tokyo Institute of Technology, 2-12-1 Ookayama, Meguro-ku, Tokyo 152-8551, Japan [email protected], [email protected] Received ; accepted 3 1 0 2 g u A 1 . ] P E h p - o r t s a [ 1 v 2 6 5 0 . 8 0 3 1 : v i X r a -- 2 -- ABSTRACT Many extrasolar (bound) terrestrial planets and free-floating (unbound) plan- ets have been discovered. The existence of bound and unbound terrestrial planets with liquid water is an important question, and of particular importance is the question of their habitability. Even for a globally ice-covered planet, geothermal heat from the planetary interior may melt the interior ice, creating an internal ocean covered by an ice shell. In this paper, we discuss the conditions that ter- restrial planets must satisfy for such an internal ocean to exist on the timescale of planetary evolution. The question is addressed in terms of planetary mass, distance from a central star, water abundance, and abundance of radiogenic heat sources. In addition, we investigate the structures of the surface-H2O layers of ice-covered planets by considering the effects of ice under high pressure (high- pressure ice). As a fiducial case, 1M⊕ planet at 1 AU from its central star and with 0.6 to 25 times the H2O mass of Earth could have an internal ocean. We find that high-pressure ice layers may appear between the internal ocean and the rock portion on a planet with an H2O mass over 25 times that of Earth. The planetary mass and abundance of surface water strongly restrict the conditions under which an extrasolar terrestrial planet may have an internal ocean with no high-pressure ice under the ocean. Such high-pressure-ice layers underlying the internal ocean are likely to affect the habitability of the planet. Subject headings: Astrobiology -- Planets and satellites: composition -- Planets and satellites: general -- Planets and satellites: interiors -- Planets and satellites: surfaces -- 3 -- 1. Introduction Since the first extrasolar planet was discovered in 1995 (Mayor & Queloz 1995), more than 800 exoplanets have been detected as of March 2013, owing to improvements in both observational instruments and the methods of analysis. Although most known exoplanets are gas giants, estimates based on both theory and observation indicates that terrestrial planets are also common (Howard et al. 2010). Supporting these estimates is the fact that Earth-like planets have indeed been discovered. Moreover, space telescopes (e.g., Kepler) have now released observational data about many terrestrial-planet candidates. Whether terrestrial planets with liquid water exist is an important question to consider because it lays the groundwork for the consideration of habitability. The orbital range around a star for which liquid water can exist on a planetary surface is called the habitable zone (HZ) (Hart 1979; Kasting et al. 1993). The inner edge of the HZ is determined by the runaway greenhouse limit (Kasting 1988; Nakajima et al. 1992), and the outer edge is estimated from the effect of CO2 clouds (Kasting et al. 1993; Mischna et al. 2000). The region between these edges is generally called the HZ for terrestrial planets with plentiful liquid water on the surface (ocean planets). Planets with plentiful water on the surface but outside the outer edge of the HZ would be globally covered with ice, and no liquid water would exist on the surface. These are called "snowball planets" (Tajika 2008). Moreover, an ocean planet could be ice-covered even within the HZ because multiple climate modes are possible, including ice-free, partially ice-covered, and globally ice-covered states (Budyko 1969; Sellers 1969; Tajika 2008). Although such planets would be globally ice-covered, liquid water could exist beneath the surface-ice shell if sufficient geothermal heat flows up from the planetary interior to melt the interior ice. In this scenario, only a few kilometers of ice would form at the surface of the ocean (Hoffman & Schrag 2002), and life could exist in the liquid water under the surface-ice shell (Hoffman et al. 1998; Hoffman & Schrag 2002; Gaidos et al. 1999). -- 4 -- Another possibility is presented by planets that float in space without being gravitationally bound to a star (free-floating planets), as have been found thanks to recent advances in observational techniques (Sumi et al. 2011). Although such planets receive no energy from a central star, even a free-floating Earth-sized planet with considerable geothermal heat could have liquid water under an ice-covered surface. Considering geothermal heat from the planetary interior, Tajika (2008) discusses the theoretical restrictions for ice-covered extrasolar terrestrial planets that, on the timescale of planetary evolution, have an internal ocean. Tajika (2008) shows that an internal ocean can exist if the water abundance and planetary mass are comparable to those of Earth. A planet with a mass less than 0.4M⊕ cannot maintain an internal ocean. For a planet with mass ≥ 4M⊕, liquid water would be stable either on the planetary surface or under the ice, regardless of the luminosity of the central star and of the planetary orbit. These are important conclusions and have important implications for habitable planets. In this paper, we extend the analysis of Tajika (2008) and vary the parameter values such as abundance of radiogenic heat sources and H2O abundance on the surface. Although Tajika (2008) assumed that the mass ratio of H2O on the planetary surface is the same as that on Earth (0.023 wt%), the origin of water on the Earth is not apparent (Genda & Ikoma 2008) so it is possible that extrasolar terrestrial planets have some order of H2O abundance. We investigate this possibility by varying the H2O abundance in our simulation, and also check whether ice appears under H2O layers under high-pressure conditions (see Section 2.2). Therefore, in this work, we consider the effect of high-pressure ice under an internal ocean and discuss its implications for habitability (see Section 4.2). With these considerations, we discuss the conditions required for bound and unbound terrestrial planets to have an internal ocean on the timescale of planetary evolution (owing to geothermal heat -- 5 -- flux from the planetary interior). Our discussion further considers various planetary masses, distances from the central star, water abundances, and the abundances of radiogenic heat sources. Finally, taking into account the effects of high-pressure ice, we investigate the structure of surface-H2O layers of ice-covered planets. 2. Method 2.1. Numerical model To calculate the mass-radius relationships for planets with masses in the range 0.1 M⊕-10 M⊕, we adjust the planetary parameters. We assume R R⊕ = (cid:18) M M⊕(cid:19)0.27 (1) as per Valencia et al. (2006), where R is the planetary radius and M is the planetary mass. The subscript ⊕ denotes values for Earth. The mantle thickness, core size, amount of H2O, average density, and other planetary properties are scaled according to this equation. The planetary surfaces are assumed to consist of frozen H2O and to have no continental crust. We define the planetary radius as R = dw + l, where dw is the H2O thickness and l is the mantle-core radius (see Fig. 1). The mass of H2O on the planetary surface is given by Msw = 4 3 πρw[(dw + l)3 − l3], (2) where ρw is the density of H2O. We vary Msw from 0.1Msw0 to 100Msw0, where Msw0 = 0.00023M with the prefactor being the H2O abundance of Earth (0.023 wt.%). Assuming that the heat flux q is transferred from the planetary interior through the surface ice shell by thermal conduction, the ice thickness dh can be obtained as dh = ki Tib − Ts q , (3) -- 6 -- where ki is the thermal conductivity of ice, Tib is the temperature at the bottom of the ice, and Ts is the temperature at the surface. We assume that the surface ice is hexagonal ice (ice Ih). Between 0.5 K and 273 K, the thermal conductivity of ice Ih is known (Dillard & Timmerhaus 1965; Klinger 1975; Varrot et al. 1978). For temperatures greater than ∼ 25 K, it is given by Klinger (1980) as ki = 567[Wm−1] T . (4) To estimate Tib, we assume that the melting line of H2O is a straight line connecting (0 bar, 273 K) to (2072 bar, 251 K) in the linear pressure-temperature phase diagram. The temperature Tib can be estimated using Tib = 273 − = 273 − 22[K] 2072[bar] 22[K] 2072[bar] pib dhρwg × 10−5, (5) where pib (bar) is the pressure at the bottom of the ice and g is the gravitational acceleration on Earth. Considering energy balance on the planetary surface, the planetary surface temperature Ts is (1 − A) L 4d2 4πD2 0 + q = εσT 4 s , (6) where A is the planetary albedo, d is the distance from the central star in AU, L is the luminosity of the central star, D0 = 1.5 × 1011 m is a distance of 1 AU in meters, ε is the emissivity of the planet, and σ = 5.67 × 10−8 Wm−2K−4 is the Stefan-Boltzmann constant. We assume A = 0.62 and ε = 1.0 (i.e., the planetary atmosphere contains no greenhouse gases, which yields an upper estimate of the ice thickness). The increase in luminosity due to the evolution of the central star as a main sequence star (Gough 1981) is considered using L(t) = (cid:20)1 + 2 5 (cid:18)1 − t t⊙(cid:19)(cid:21)−1 L⊙, (7) where t⊙ = 4.7 × 109 years, and L⊙ = 3.827 × 1026 W. -- 7 -- From these models, we can obtain the H2O thickness dw and the ice thickness dh. The condition for terrestrial planets to have an internal ocean is dw > dh. (8) To estimate the geothermal heat flux q through planetary evolution, we investigate the thermal evolution of terrestrial planets by using a parameterized convection model (Tajika Matsui 1992; McGovern Schubert 1989; Franck Bounama 1995; von Blow et al. 2008; see appendix for details). We assume E, which is the initial heat generation per unit time and volume, is 0.1E0 to 10E0, where the constant E0 is the initial heat generation estimated from the present heat flux of the Earth (see appendix for details). 2.2. High-pressure ice Ice undergoes a phase transition at high pressure (Fig. 2). Unlike ice I, the other phases are more dense than liquid H2O. We call the denser ice "high-pressure ice." Because Tajika (2008) assumes that the amount of H2O on the planetary surface is the same as that on the Earth's surface Msw0 (= 0.00023M), the only possible conditions on the planetary surface are those labeled 1, 2, and 3 in Fig. 3. However, because we consider herein that H2O amounts may range from 0.1Msw0 to 100Msw0, the H2O-rock boundary could move to higher pressure, so we should account for the effect of high-pressure ice (Fig. 3a). Therefore, types 4, 5, and 6 of Fig. 3b are added as possible surface conditions. Types-2 and type-5 planets both have an internal ocean, but high-pressure ice exists in type-5 planets between the internal ocean and the underlying rock. We approximate the melting curve by straight lines connecting the triple points in the linear pressure-temperature phase diagram. We also assume that the amount of heat -- 8 -- flux from the planetary interior that is transferred through the internal ocean by thermal convection is the same as that which is transferred through the surface ice. Here, we presume that temperature gradient in liquid-water part of phase diagram is isothermal (Figs. 3a and 3c), although a gradient for a deeper internal ocean than what is considered in this study should be carefully discussed. The condition for high-pressure ice to exist under the internal ocean is Ph < Pb, (9) where Pb is the pressure at the H2O-rock boundary and Ph is the pressure on the phase diagram where the temperature gradient and the high-pressure melting line cross (Fig. 3c). As a representative value, we assume that high-pressure ice has a density of 1.2 g/cm3. Because the characteristic features of high-pressure ice are poorly understood, we simplified the model; and, in particular, the thermal conductivities of high-pressure ice (see below). When thermal conductivity of high-pressure ice is relatively high and the temperature gradient in the high-pressure-ice part of the phase diagram is less than the gradient of the melting lines, the layer of high-pressure ice continues to the H2O-rock boundary [(i) in Fig. 3c]. However, when the thermal conductivity is comparatively low and the temperature gradient is greater than that of the melting lines, the temperature gradient joins the melting line and goes along with the melting lines to the H2O-rock boundary [dashed arrows (ii) and (iii) in Fig. 3c]. Although little is known about the conditions on the dashed arrows, we assume that layer to be high-pressure ice in this study. Therefore, from the point where the temperature-pressure line crosses into the high-pressure-ice part, the high-pressure layer continues to the H2O-rock boundary. -- 9 -- 3. Results Figures 4a and 4b show the surface conditions for planets with masses from 0.1M⊕ to 10M⊕ at 4.6 billion years after planetary formation, with varying H2O masses on the surface, with initial radiogenic heat sources, and at 1AU from our Sun. We assumed E/E0 = 1 for Fig. 4a and Msw/Msw0 = 1 for Fig. 4b. Because larger planets have larger geothermal heat flux and thicker H2O layers, they could have an internal ocean with less H2O mass on the planetary surface (Fig. 4a) and a weaker initial radiogenic heat source (Fig. 4b). However, larger planets also have larger gravitational acceleration. Thus, on those planets, high-pressure ice tends to appear under the internal ocean with smaller H2O mass on the surface (Fig. 4a). For example, if a planet of mass of 1M⊕ has an H2O mass of 0.6M⊕ to 25Msw0, it could have an internal ocean. However, if a planet has an H2O mass > 25Msw0, high-pressure ice should exist under the ocean (Fig. 4a). Note, however, that an internal ocean can exist on a planet having a mass of 1M⊕ if the initial radiogenic heat source exceeds 0.4E0 (Fig. 4b). Figures 4c and 4d give the temperature profiles of surface-H2O layers. Figure 4c gives the conditions of a planet parameterized by 1M⊕ and 5Msw0, whereas Fig. 4d is for 1M⊕ and 30Msw0. Given the conditions of Fig. 4c, the surface H2O layers consist of a conductive-ice-I layer and a convective-liquid layer (i.e., an internal ocean). When the planet has a greater H2O mass, high-pressure ice could appear under the internal ocean (Fig. 4d). Here, the planetary surface temperature seems very low, so we assume that the planet is covered by ice that has a higher albedo (∼0.6) than ocean/land (∼0.3). Figures 5a and 5b show the surface conditions for free-floating planets (L = 0) with masses from 0.1M⊕ to 10M⊕ at 4.6 billion years after planetary formation. The incident flux from the central star affects the surface temperature, thereby affecting the condition on the surface. Therefore, the conditions, and in particular those shown in Fig. 5a, are -- 10 -- different from those shown in Figs. 4a and 4b. The results of Fig. 5a show that, regardless of the amount of H2O a 1M⊕ planet has, an internal ocean cannot exist under the ice shell. An internal ocean could exist on free-floating planets under certain conditions, but the planetary size and water abundance strongly constrain these conditions (see Fig. 5a). For instance, if a free-floating planet has an initial radiogenic heat source greater than 7E0, it can have an internal ocean (Fig. 5b). Figures 5c and 5d give the temperature gradient of surface-H2O layers for free-floating planets. The parameters are set to the same values as those for Figs. 4c and 4b, except that L = 0. The temperature on the planetary surface, approximately 35 K, is calculated by Eq. (6) on the assumption that L = 0, and the ice-I layer is thicker than that for the conditions of Figs. 4c and 4d. Given the conditions of Fig. 5c, the surface-H2O layers consist only of a conductive-ice-I layer. However, for grater H2O mass, high-pressure ice could appear under the ice-I layer (Fig. 5d). Figure 6 shows the surface conditions for planets with masses from 0.1M⊕ to 10M⊕ at varying distances from a central star, and at 4.6 billion years after planetary formation. The runaway greenhouse limit (Kasting 1988; Nakajima et al. 1992) indicating the inner edge of the HZ is not considered. The effect of the incident flux from the central star on the surface conditions is estimated in each graph of Fig. 6 as a function of distance from the central star. We find that the existence of an internal ocean on planets far from a central star depends on the planetary mass and surface-H2O mass. For the conditions of Fig. 6b on which a planet has two times the H2O mass (i.e., 2Msw/Msw0), for example, a 2M⊕ planet can have an internal ocean under which there is no high-pressure ice only out to approximately 7 AU from the central star. For a planet with five times the H2O mass (i.e., 5Msw/Msw0; Fig. 6c), an internal ocean with no underlying high-pressure ice can exist out to approximately 30 AU. However, if the planet has ten times the H2O mass (i.e., 10Msw/Msw0; Fig. 6d), an internal ocean without underlying high-pressure ice could exist to only approximately 5 AU. The planetary mass and surface-H2O mass strongly constrain -- 11 -- the conditions under which an extrasolar terrestrial planet far from its central star can have an internal ocean with no underlying high-pressure ice. 4. Discussion 4.1. Models of study We expanded on the models of Tajika (2008) by (1) invoking the mass-radius relationship [Eq. (1)] to consider planetary compression by gravity, (2) considering the temperature dependence of the thermal conductivity of ice Ih [Eq. (4)], and varying the (3) abundance of radiogenic heat sources and (4) H2O abundance on the surface, both of which were held constant in Tajika (2008). We used Eq. (1) to determine the mass-radius relationships because we expected more accurate results, although this change makes little quantitative difference in the results. Below, we discuss items (2), (3), and (4) to analyze the models of this study. It is known from experiment that the thermal conductivity of ice Ih depends on temperature [≈ 1/T (Klinger 1980)]. In the present study, the surface-ice shell is thicker than that considered by Tajika (2008). For example, the surface-ice shell considered herein is about 1.1 times thicker at 1 AU and about 3.5 to 4.4 times thicker on the free-floating planets (L = 0). This increased thickness is due to the use of Eq. (4) to describe the thermal conductivity of ice Ih in contrast with Tajika (2008), where a constant thermal conductivity (ki = 2.2) was used. Tajika (2008) treated the abundance of H2O and heat source elements (a ratio of mass against a planetary mass) to be constant, i.e., Msw/Msw0 = 1 (Msw0 = 0.00023M) and E/E0 = 1, although the amounts should vary with a planetary mass, and showed that liquid water would be stable either on the surface or beneath ice for a planet with a mass exceeding 4M⊕, regardless of planetary orbit and luminosity of the central star. -- 12 -- In this paper, however, when we consider a planet with parameters Msw/Msw0 = 1 and E/E0 = 1, the results shown in Fig. 5 are not consistent with those of Tajika (2008). Our results indicate that free-floating planets with masses between 0.1M⊕ and 10M⊕ could not have liquid water on the planetary surface and also under the ice for planetary parameters of Msw/Msw0 = 1 and E/E0 = 1. As just explained, when using Eq. (4) to describe the thermal conductivity of ice Ih, the planetary surface temperature becomes more sensitive to the thickness of the surface-ice layer, which leads to a several-times increase in the thickness of the surface ice. These results thus differ from those of Tajika (2008). In this study, we assume that the high-pressure layer continues to the H2O-rock boundary from the point where the temperature-pressure line crosses into the high-pressure- ice part (see Section 2.2). The results of this paper could depend on this assumption. When the thermal conductivity of high-pressure ice is comparatively low (similar to that of ice Ih), the heat cannot be transported through the high-pressure ice effectively, thus the bottom of the high-pressure ice could be thermally unstable and be melted enough to form internal ocean. The result of the present study indicate that the high-pressure layers under the internal ocean could be from <∼1 km to ∼100 km thick because a super-Earth planet with a few wt.% H2O could have ∼100 km-thick high-pressure ice layers. In high-pressure ice that is ∼100 km thick, it is possible that convective ice layers appear. To estimate whether or not such convection layers arise, we use the Rayleigh number Ra, which is given by Ra = gαρa3∆T κη , (10) where α is the coefficient of thermal expansion, ρ is the density, a is the thickness of the high-pressure-ice layer, ∆T is the temperature difference between top and bottom part of the layer, κ is the thermal diffusivity, and η is the viscosity coefficient. When the Rayleigh number Ra exceeds the critical value for the onset of convection (Racrit ∼ 103), convective -- 13 -- motion spontaneously begins. If we use the parameters from Kubo (2008), i.e., α = 10−4 K−1, ρ = 1000 kgm−3, ∆T = 50 K, κ = 2 × 10−6 m2s−1, and η = 1015 − 1018 Pa s, and use the typical values of this study, g = 10 ms−2 and a = 102 km, the Rayleigh number is ∼ 104 − 107, which indicates that the convective motion is possible. In this study, we consider Msw from 0.1Msw0 (0.0023 wt.%) to 100Msw0 (2.3 wt.%). If the planet has significantly more H2O on the surface, the water layer would be very thick and our model would not apply to that planet. If the planet has significantly less H2O on the surface, the regassing flux of water would change because of insufficient water on the surface, and the planet's thermal evolution would differ from that of Earth. In other words, we consider only those planets that are relatively active geothermally and have an adequate amount of water as Earth-like planets. Improving our model so that it applies to other types of terrestrial planets is an important problem that we leave for future work. 4.2. Habitability of internal ocean For genesis and sustenance of life, we need at least (1) liquid water and (2) nutrient salts because these substances are required to synthesize the body of life (Maruyama et al. 2013). Because nutrient salts are supplied from rocks, it is necessary that liquid water should be in contact with rock to liberate the salts. A type-5 planet (Fig. 3b) is thus not likely to be habitable because the internal ocean does not come in contract with rocks. However, it is possible for a type-2 planet to meet this requirement. Therefore, we presume that only type-2 planets have an internal ocean that is possibly habitable. Therefore, the results of this study indicate that planetary mass and H2O mass constitute two more conditions to add to the previous conditions for an extrasolar planet to have an internal ocean without high-pressure ice. In other words, these considerations -- 14 -- indicate that only a planet with the appropriate planetary mass and H2O mass can have an internal ocean that is possibly habitable. However, it is possible that hydrothermal activities within the rocky crust may transfer nutrient salts to the internal ocean through cracks in the high-pressure ice. Large terrestrial planets such as those considered in this study are likely to have areas of high geothermal activity along mid-oceanic ridges and subduction zones as well as large submarine volcanos whose tops might emerge into the internal ocean from the high-pressure ice. Furthermore, for planets with thick high-pressure-ice layers (Section 4.1), the convective ice layer could transfer nutrient salts from the rock to the internal ocean. In these cases, high-pressure ice might not prevent nutrient flux from the rock floor from reaching the internal ocean. 4.3. Future work As shown in Figs. 4b and 5b, an appropriate initial radiogenic heat source is an important factor in determining whether or not a planet has an internal ocean. Because the variation in the amount of initial heat generation in planets throughout space is not known, we assume in this study that the initial heat generation E per unit time and volume ranges from 0.1E0 to 10E0. Thus, in order to resolve this issue, the general amount of radiogenic heat sources for extrasolar terrestrial planets should be estimated. An Earth-size planet (Msw/Msw0 = 1, E/E0 = 1) orbiting at 1AU around the Sun for 4.6 billion years with no greenhouse gases might be globally covered by ice (Fig. 4). In contrast, Earth currently is partially covered with ice but retains liquid water on its surface. Therefore, it is almost certain that greenhouse gases play a major role in keeping the surface warm. Even free-floating, Earth-sized planets with atmospheres rich in molecular hydrogen could have liquid water on the surface because geothermal heat from the interior would be -- 15 -- retained by the greenhouse-gas effect of H2 (Stevenson 1999). By applying this model here, we can account for the greenhouse-gas effect for various values of emissivity ε in Eq.(6). In future presentations, we will thus discuss how greenhouse gases modify the conditions necessary for the development of a terrestrial ocean planet or ice-covered planet with an internal ocean. In the present study, we assumed pure H2O on the planetary surface. However, even if an extrasolar terrestrial planet has surface H2O, the H2O might not be pure because it might contain dissolved nutrient salts such as those found in Earth's oceans. In this case, the melting point of the solution would differ from that of pure H2O. Moreover, the phase diagram would become more complicated, and the properties of H2O (e.g., thermal conductivity) could be transmuted as a result of the appearance of phases in which H2O ice contains nutrient salts. A very important work would thus be to analyze multicomponent water (e.g., seawater on Earth) to see what qualitative changes such a modified phase diagram would bring to our results. Consider the examples, Europa and Ganymede, which are the two satellites of Jupiter and are thought to have internal oceans. Ganymede is thought to have high-pressure ice under its internal ocean (see, e.g., Lupo 1982). Therefore, determining whether the internal oceans of Europa and Ganymede are suitable for life would be pertinent to the discussion of the habitability of internal oceans with or without high-pressure ice. In addition, a more circumstantial discussion of the habitability of internal oceans would require considering the redox gradient within the internal ocean [as exemplified by the discussions of Europa (Gaidos et al. 1999)], and the effects of the riverine flux of nutrient salts (Maruyama et al. 2013). -- 16 -- 5. Conclusion Herein, we discuss the conditions that must be satisfied for ice-covered bound and unbound terrestrial planets to have an internal ocean on the timescale of planetary evolution. Geothermal heat flow from the planetary interior is considered as the heat source at the origin of the internal ocean. By applying and improving the model of Tajika (2008), we also examine how the amount of radiogenic heat and H2O mass affect these conditions. Moreover, we investigate the structures of surface-H2O layers of snowball planets by considering the effects of high-pressure ice. The results indicate that planetary mass and surface-H2O mass strongly constrain the conditions under which an extrasolar terrestrial planet might have an internal ocean without a high-pressure ice existing under the internal ocean. We thank the reviewer Eiichi Tajika for the constructive comments. We also thank S. Ida and M. Ikoma for valuable discussions. This research was supported by a grant for the Global COE Program "From the Earth to 'Earths'" from the Ministry of Education, Culture, Sports, Science and Technology of Japan. T. S. was supported by a Grant-in-Aid for Young Scientists (B), JSPS KAKENHI Grant Number 24740120. A. Parameterized convection model By applying conservation of energy, we obtain 4πR2 mq + 4 3 πρc(R3 m − R3 c ) dTm dt = 4 3 πE(t)(R3 m − R3 c), (A1) where ρ is the density of the mantle, c is the specific heat at constant pressure, Rm and Rc are the outer and inner radii of the mantle, respectively, Tm is the average mantle -- 17 -- temperature, and E(t) is the rate of energy production by decay of radiogenic heat sources in the mantle per unit volume. The mantle heat flux is parameterized in terms of the Rayleigh number Ra as q = k(Tm − T ′) Rm − Rc (cid:18) Ra Racrit(cid:19)β , (A2) where k is the thermal conductivity of the mantle, T ′ is the temperature at the surface of the mantle, Racrit is the critical value of Ra for convection onset, and β is an empirical constant. The radiogenic heat source is parameterized as E(t) = Ee−λt (A3) where λ is the decay constant of the radiogenic heat source and E is the initial heat generation per unit time and volume. We assume E is 0.1E0 to 10E0, where the constant E0 is the initial heat generation estimated from the present heat flux of Earth. We obtain Ra as Ra = gα(Tm − T ′)(Rm − Rc)3 κν , (A4) where α is the coefficient of thermal expansion, κ is the thermal diffusivity, and ν is the water-dependent kinematic viscosity. The viscosity ν strongly depends on the evolution of the mass Mw of mantle water and the mantle temperature Tm and is parameterized as ν = νm exp(TA/Tm), TA = α1 + α2x, x = Mw Mm , (A5) (A6) (A7) where νm, α1, and α2 are constants, TA is the activation temperature for solid-state creep, and Mm is the mass of the mantle. -- 18 -- The evolution of the mantle water can be described by the regassing flux Freg and outgassing flux Fout as dMw dt = Freg − Fout = fbasρbasdbasSRH2O − Mw m − R3 c ) 4 3π(R3 dmfwS, (A8) where fbas is the water content in the basalt layer, ρbas is the average density, dbas is the average thickness of the basalt layer before subduction, S is the areal spreading rate, RH2O is the regassing ratio of water, dm is the melting generation depth, and fw is the outgassing fraction of water. The regassing ratio of water linearly depends on the mean mantle temperature Tm via RH2O = RT (Tm(0) − Tm) + RH2O,0 (A9) as given by von Bloh et al. (2008), where RT is the temperature dependence of the regassing ratio, and RH2O,0 is the initial regassing ratio. The areal spreading rate S is S = q2πκA0 4k2(Tm − T ′)2 , where A0 is the ocean-basin area. The area A0 can be parameterized as A0 A∗ 0 = (cid:18) R R⊕(cid:19)2 , (A10) (A11) where A∗ 0 is the present ocean-basin area on Earth. By using Eq. (1), we obtain A0 as A0 = A∗ 0(cid:18) M M⊕(cid:19)0.54 . (A12) In Table 1, we summarize selected values for the parameters used in the parameterized convection model. -- 19 -- REFERENCES von Blow, W. et al. 2008, A&A, 476, 1365. Budyko, M. I. 1969, Tellus, 21, 611. Dillard, D. S., & Timmerhaus, K. D. 1965, Bull. Inst. Int. Froid Annexe, part 2, 35. Franck, S., & Bounama, C. 1995, Phys. Earth Planet. Inter., 92, 57. Gaidos, E. J., Nealson, K. H., & Kirschvink, J. L. 1999, Science, 284, 1631. Genda, H., & Ikoma, M. 2008, Icarus, 194, 42. Gough, D. O. 1981, Sol. Phys., 74, 21. Hart, M. H. 1979, Icarus, 37, 351. Hoffman, P. F., & Schrag, D. P. 2002, Terra Nova, 14, 129. Hoffman, P. F. et al. 1998, Science, 281, 1342. Howard, A. W. et al. 2010, Science, 330, 653. Jackson, M. J., & Pollack, H. N. 1984, J. Geophys. Res., 89, 10103. Kamb, B. 1973, Physics and Chemistry of Ice,28. Kasting, J. F. 1988, Icarus, 74, 472. Kasting, J. F. et al. 1993, Icarus, 101, 108. Klinger, J. 1975, J. Glaciol., 14, 517. Klinger, J. 1980, Science, 209, 11. Kubo, T. 2008, Low Temperature Science, 66, 123. -- 20 -- Lupo, M. J. 1982, Icarus, 52, 40. Maruyama, S. et al. 2013, Geoscience Frontiers, 4, 141. Mayor, M., & Queloz, D. 1995, Nature, 378, 355. McGovern, P. J., & Schubert, G. 1989, Earth Planet. Sci. Lett., 96, 27. Mischna, M. A., Kasting, J. F., Pavlov, A., & Freedman, R. 2000, Icarus, 145, 546. Nakajima, S., Hayashi, Y.-Y., & Abe, Y. 1992, J. Atmos. Sci., 49, 2256. Sellers, W. D. 1969, J. Appl. Meteorol., 8, 392. Stevenson, D. J. 1999, Nature, 400, 32. Sumi, T. et al. 2011, Nature, 473, 349. Tajika, E. 2008, ApJ, 680, L53. Tajika, E., & Matsui, T. 1992, Earth Planet. Sci. Lett., 113, 251. Valencia, D., O'Connell, R. J., & Sasselov, D. 2006, Icarus, 181, 545. Varrot, M., Rochas, G., & Klinger, J. 1978, J. Glaciol., 21, 241. This manuscript was prepared with the AAS LATEX macros v5.2. -- 21 -- Fig. 1. -- Schematic of model used in this study -- 22 -- Fig. 2. -- Phase diagram of H2O [after Kamb (1973)] -- 23 -- Fig. 3. -- (a) Schematic phase diagram of H2O (gray lines), temperature gradient (black lines), and H2O-rock boundaries (dashed lines). (b) Types of planets that have H2O on the planetary surface. (c) Schematic of models used to treat high-pressure ice. -- 24 -- Fig. 4. -- Surface conditions for planet at 1AU around a central star (L = L0; the present luminosity of our Sun) and temperature profiles of surface-H2O layers for two cases. (a) Hor- izontal axis is surface-H2O mass, and vertical axis is planetary mass normalized by Earth's mass, assuming E/E0 = 1. The two dots in this panel indicate the conditions corresponding to the temperature profiles shown in panels (c) and (d). (b) Horizontal axis is initial radio- genic heat, and vertical axis is the planetary mass normalized by Earth's mass, assuming Msw/Msw0 = 1. (c) Temperature profile at 1M⊕ and 5Msw0. (d) Temperature profile at 1M⊕ and 30Msw0. -- 25 -- Fig. 5. -- Same as Fig. 4, but for a flee-floating planet (L = 0). -- 26 -- Fig. 6. -- Conditions for planet orbiting central star (E/E0 = 1) for (a) Msw/Msw0 = 1, (b) Msw/Msw0 = 2, (c) Msw/Msw0 = 5, and (d) Msw/Msw0 = 10. Horizontal axes represent the distance from the central star; vertical axes represent the planetary mass normalized by Earth's mass. -- 27 -- Table 1: Values of parameter for parameterized convection model Parameter Value Unit Source Rm(1M⊕) 6, 271 × 103 m Rc(1M⊕) 3, 471 × 103 m von Blow et al. (2008) von Blow et al. (2008) ρc Tm(0) k Racrit β E0 λ g(1M⊕) α κ νm α1 α2 4.2 × 106 J m−3K−1 von Blow et al. (2008) 3, 000 4.2 1, 100 0.3 K von Blow et al. (2008) J s−1m−1K−1 von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) 1.46 × 10−7 J s−1m−3 von Blow et al. (2008) 0.34 9.81 Gyr−1 m s−2 3 × 10−5 K−1 10−6 m2s−1 2.21 × 10−7 m2s−1 6.4 × 104 K von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) McGovern & Schubert (1989) Frank & Bounama (1995) −6.1 × 106 K per weight fraction Frank & Bounama (1995) Mm(1M⊕) 4.06 × 1024 Mw(0)(1M⊕) 4.2 × 1021 kg kg fbas ρbas dbas dm fw RT 0.03 2, 950 5 × 103 40 × 103 0.194 m m 29.8 × 10−5 K−1 RH2O,0 0.001 A∗ 0 3.1 × 1014 m2 McGovern & Schubert (1989) kg m−3 von Blow et al. (2008) Frank & Bounama (1995) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008) von Blow et al. (2008)
1101.4643	2	1101	2011-05-11T21:59:37	WASP-40b: independent discovery of the 0.6-Mjup transiting exoplanet HAT-P-27b	[ "astro-ph.EP" ]	From WASP photometry and SOPHIE radial velocities we report the discovery of WASP-40b (HAT-P-27b), a 0.6 Mjup planet that transits its 12th magnitude host star every 3.04 days. The host star is of late G-type or early K-type and likely has a metallicity greater than solar ([Fe/H] = 0.14 +/- 0.11). The planet's mass and radius are typical of the known hot Jupiters, thus adding another system to the apparent pileup of transiting planets with periods near 3 to 4 days. Our parameters match those of the recent HATnet announcement of the same planet, thus giving confidence in the techniques used. We report a possible indication of stellar activity in the host star.	astro-ph.EP	astro-ph	WASP-40b: Independent Discovery of the 0.6 MJup Transiting Exoplanet HAT-P-27b D.R. Anderson1, S.C.C. Barros2, I. Boisse3,4, F. Bouchy3,5, A. Collier Cameron6, F. Faedi2, G. Hebrard3,5, C. Hellier1, M. Lendl7, C. Moutou8, D. Pollacco2, A. Santerne8, B. Smalley1, A.M.S. Smith1, I. Todd2, A.H.M.J. Triaud7, R.G. West9, P.J. Wheatley10, J. Bento10, B. Enoch6, M. Gillon11, P.F.L. Maxted1, J. McCormac2, D. Queloz7, E.K. Simpson2, I. Skillen12 ABSTRACT From WASP photometry and SOPHIE radial velocities we report the discovery of WASP-40b (HAT-P-27b), a 0.6 MJup planet that transits its 12th magnitude host star every 3.04 days. The host star is of late G-type or early K-type and likely has a metallicity greater than solar ([Fe/H] = 0.14 ± 0.11). The planet's mass and radius are typical of the known hot Jupiters, thus adding another system to the apparent pileup of transiting planets with periods near 3 -- 4 days. Our parameters match those of the recent HATnet announcement of the same planet, thus giving confidence in the techniques used. We report a possible indication of stellar activity in the host star. Subject headings: Extrasolar Planets 1Astrophysics Group, Keele University, Staffordshire, ST5 5BG, UK; [email protected] 2Astrophysics Research Centre, School of Mathematics & Physics, Queen's University, University Road, Belfast, BT7 1NN, UK 3Institut d'Astrophysique de Paris, UMR7095 CNRS, Universite Pierre & Marie Curie, 75014 Paris, France 4Centro de Astrof´ısica, Universidade do Porto, Rua das Estrelas, 4150-762 Porto, Portugal 5Observatoire de Haute-Provence, CNRS/OAMP, 04870 Saint-Michel l'Observatoire, France 6SUPA, School of Physics and Astronomy, University of St. Andrews, North Haugh, Fife, KY16 9SS, UK 7Observatoire Astronomique de l'Universit´e de Gen`eve 51 ch. des Maillettes, 1290 Sauverny, Switzerland 8Laboratoire d'Astrophysique de Marseille, 38 rue Frderic Joliot-Curie, 13388 Marseille Cedex 13, France 9Department of Physics and Astronomy, University of Leicester, Leicester, LE1 7RH, UK 10Department of Physics, University of Warwick, Coven- try CV4 7AL, UK 11Institut d'Astrophysique et de G´eophysique, Universit´e de Li`ege, All´ee du 6 Aout, 17, Bat. B5C, Li`ege 1, Belgium 12Isaac Newton Group of Telescopes, Apartado de Correos 321, E-38700 Santa Cruz de la Palma, Tenerife, Spain 1. Introduction While the Kepler mission is currently produc- ing the most candidates for transiting extraso- lar planets (e.g., Borucki et al. 2010), the ground- based transit-search programs continue to find more planets around stars at brighter magnitudes than those found in the space missions. Of these, Hungarian Automated Telescope Network (HAT- net; Bakos et al. 2004) and Wide Angle Search for Planets (WASP; Pollacco et al. 2006) have been the most successful. Both projects are based on ar- rays of 200 mm f/1.8 lenses backed by CCDs, with the biggest difference being that HATnet operates at several longitudes while WASP consists of one station in each hemisphere. The two projects look at overlapping regions of sky, which has led to some near-simultaneous discoveries, such as the planet WASP-11b (West et al. 2009) also being HAT-P-10b (Bakos et al. 2009). Reporting of such independent discoveries gives important informa- tion on the reliability of the respective techniques and on the completeness of the transit surveys. Recently HATnet announced the planet HAT- 1 P-27b (B´eky et al. 2011), a hot Jupiter in a 3 day orbit around a mV = 12.2 star. This planet had been independently discovered by the WASP project and assigned the name WASP- 40b (Hellier et al. 2011). We report here on the discovery of WASP-40b made using data from SuperWASP-North and WASP-South com- bined, together with radial velocities from the SO- PHIE spectrograph at the Observatoire de Haute- Provence (OHP) observatory. 2. Observations We observed WASP-40, an ∼K0-type star lo- cated in Virgo, with the SuperWASP-North and WASP-South cameras during the three seasons of 2008 -- 2010. A transit search (Collier Cameron et al. 2006) of the resulting 30,260 photometric measure- ments found a strong 3.04 day periodicity. The discovery light curve is displayed in Figure 1a, folded on this period. Using the SOPHIE spectrograph mounted on the 1.93 m OHP telescope (Perruchot et al. 2008; Bouchy et al. 2009), we obtained eight spectra of WASP-40 during 2010 April and May. The high-efficiency mode and slow readout were used, corresponding to a spectral resolving power of 40 000 and lowest readout noise. We acquired a spectrum in both entrance fibers of the spectro- graph to allow monitoring of and correcting for the sky background. However, over the sequence, this background was low enough that we did not have to apply such a correction. Signal-to-noise ratio values range from 22 to 35, for exposure times of 30 to 43 minutes. Radial-velocity (RV) measurements were computed by weighted cross- correlation (Baranne et al. 1996; Pepe et al. 2005) with a numerical G2 spectral template. To ac- count for systematic effects associated with the high-efficiency mode (e.g., Bouchy et al. 2009), we added an uncertainty of 10 m s−1 in quadrature to the formal errors. RV variations were detected with the same period found from the transits and with a semiamplitude of 91 m s−1, consistent with a planetary mass companion. The RV measure- ments are listed in Table 1 and are plotted in Fig- ure 2. To test the hypothesis that the RV varia- tions are due to spectral line distortions caused by a blended eclipsing binary or starspots, we 2 Fig. 1. -- Photometry of WASP-40, with the best- fitting transit model superimposed. (a) WASP discovery light curve, folded on the ephemeris of Table 4. (b) WASP data around the transit, binned in time with a bin width of ∼11 minutes). (c) High-precision transit light curves from Liv- erpool Telescope RISE (top, red) and Euler/C2 (bottom, blue). Table 1: SOPHIE Radial-Velocity Measurements BJD − 2, 400, 000a σRV BS (km s−1) 0.012 0.013 0.012 0.014 0.014 0.014 0.018 0.013 (km s−1) −0.035 −0.043 −0.044 −0.059 −0.025 −0.037 −0.054 −0.014 55,301.4903 55,303.4817 55,305.4918 55,323.4792 55,324.5650 55,334.4374 55,335.5319 55,336.5056 RV (km s−1) −15.796 −15.702 −15.837 −15.858 −15.743 −15.630 −15.809 −15.717 a BJD = Barycentric Julian Date (UTC). Fig. 2. -- (a) SOPHIE RVs with the best-fitting circular Keplerian orbit superimposed. The RV represented by an open, red circle was excluded from the analysis, as it was taken during transit. (b) Residuals of the RVs about the fit. 3 performed a line-bisector analysis (Queloz et al. 2001) of the SOPHIE cross-correlation functions. The lack of correlation between bisector span and RV (Fig. 3) supports our conclusion that the peri- odic dimming of WASP-40 and its RV variations are due to a planet. To refine the system parameters we obtained high-precision transit photometry. On 2010 June 26 we obtained a full transit of WASP-40 with the RISE (rapid imager to search for exoplanets) high- speed CCD camera mounted on the 2.0 m Liv- erpool Telescope (Steele et al. 2008; Gibson et al. 2008). RISE has a wideband filter of ∼ 500 -- 700 nm, which corresponds approximately to V + R. We obtained 285 exposures in the 2 × 2 binning mode with an exposure time of 35 s and effectively no dead time. To minimize the impact of flat- fielding errors and to increase the duty cycle we defocused the telescope by 1 mm. On 2010 June 30 we observed a full transit of WASP-40 through a Gunn r filter with the C2 camera on the 1.2 m Euler Swiss telescope, which was defocused by 0.1 mm. The seeing ranged from 0.5 to 1.1′′, and air mass ranged from 1.2 to 1.5 over the course of the 105 exposures. We performed differential photom- etry relative to several stable reference stars, using the ULTRACAM pipeline (Dhillon et al. 2007) for the RISE data and IRAF for the Euler data. The resulting light curves are displayed in Figure 1c. 3. Stellar parameters from spectra The individual SOPHIE spectra of WASP-40 were co-added to produce a single spectrum with an average S/N of around 60:1. The analysis was performed using the methods given in Gillon et al. (2009). The Hα line was used to determine the ef- fective temperature (Teff ), and the Na i D and Mg i b lines were used as surface gravity (log g) diagnos- tics. The parameters obtained from the analysis are listed in Table 2. The elemental abundances were determined from equivalent-width measure- ments of several clean and unblended lines. A value for microturbulence (ξt) was determined from Fe i using the method of Magain (1984). The quoted error estimates include those given by the uncertainties in Teff, log g and ξt, as well as the scatter due to measurement and atomic data un- certainties. We estimated the sky-projected stellar rotation velocity (v sin I) by fitting the profiles of several unblended Fe i lines. For this, we used an in- strumental FWHM of 0.15 ± 0.01 A, determined from the telluric lines around 6300A and assumed a value for macroturbulence (vmac) of 1.0 ± 0.3 km s−1 (Bruntt et al. 2010). As a check of our analysis we also estimated the metallicity and v sin I directly from the cross- correlation function of the averaged SOPHIE spec- tra, using the methods described in Boisse et al. (2010). These give [Fe/H] = 0.06±0.09 and v sin I = 3.3 ± 1.0 km s−1, which are in agreement with our preceding values. We input our values of Teff, log g and [Fe/H] into the calibrations of Torres et al. (2010) to ob- tain estimates of the stellar mass and radius (Ta- ble 2). 4. System parameters from RV and tran- sit data We determined the system parameters from a simultaneous fit to the data described in § 2. The transit light curve was modeled using the formula- tion of Mandel & Agol (2002) with the assumption that RP≪R∗. Limb-darkening was accounted for using a four-coefficient nonlinear limb-darkening model, using fixed coefficients (Table 3) appropri- ate to the passbands and interpolated in effective temperature, surface gravity and metallicity from the tabulations of Claret (2000). The simultaneous fit was performed using the current version of the Markov-chain Monte Carlo (MCMC) code described by Collier Cameron et al. (2007) and Pollacco et al. (2008). The transit light curve is parameterized by the epoch of mid- transit T0, the orbital period P , the planet-to-star area ratio (RP/R∗)2, the approximate duration of the transit from initial to final contact T14, and the impact parameter b = a cos i/R∗ (the distance, in fractional stellar radii, of the transit chord from the star's center). The radial-velocity orbit is parameterized by the stellar reflex veloc- ity semiamplitude K∗, the systemic velocity γ, Fig. 3. -- Bisector span variation with respect to radial velocity, with approximately the same scale used for each axis. The systemic velocity was subtracted from the radial-velocity values. We adopted uncertainties on the bisector spans that were twice the size of those on the radial velocities. Table 2: Stellar Parameters from Spectra Parameter Teff log g ξt v sin I [Fe/H] log A(Li) M∗ R∗ R.A. (J2000) Decl. (J2000) mV mJ mH mK USNO-B1.0 2MASSa a a a Value 5200 ± 150 K 4.5 ± 0.2 (cgs) 0.9 ± 0.2 km s−1 2.5 ± 0.9 km s−1 0.14 ± 0.11 < 0.5 ′ 0.91 ± 0.08 M⊙ 0.88 ± 0.22 R⊙ 14h51m04.19s +05◦56 50.5′′ 12.2 ± 0.1 10.63 ± 0.03 10.25 ± 0.02 10.11 ± 0.02 0959-0237786 14510418+0556505 a Skrutskie et al. (2006). Table 3: Limb-Darkening Coefficients Light curve (band) WASP/Euler (RC) RISE (V J) a2 a1 0.714 −0.651 0.657 −0.626 a4 a3 1.335 −0.602 1.479 −0.658 4 and √e cos ω and √e sin ω (Anderson et al. 2011), where e is orbital eccentricity and ω is the argu- ment of periastron. The linear scale of the system depends on the orbital separation a which, through Kepler's third law, depends on the stellar mass M∗. At each step in the Markov chain, the latest val- ues of stellar density ρ∗, effective temperature Teff and metallicity [Fe/H] are input to the em- pirical mass calibration of Enoch et al. (2010) to obtain M∗. The shapes of the transit light curve (Seager & Mall´en-Ornelas 2003) and the radial- velocity curve constrain ρ∗, which combines with M∗ to give R∗. Teff and [Fe/H] are proposal parameters constrained by Gaussian priors with mean values and variances derived directly from the stellar spectra (Table 2). As the planet-star area ratio is constrained by the measured transit depth, RP follows from R∗. The planet mass MP is calculated from the mea- sured value of K1 and M∗; the planetary density ρP and surface gravity log gP then follow. We also calculate the blackbody equilibrium temperature TP,A=0, where A is albedo, assuming efficient re- distribution of heat from the planet's presumed permanent day side to its night side. At each step in the MCMC procedure, model transit light curves and radial-velocity curves are computed from the proposal parameter values, which are perturbed from the previous values by a small, random amount. The χ2 statistic is used to judge the goodness of fit of these models to the data and a step is accepted if χ2 is lower than for the previous step. A step with higher χ2 is ac- cepted with a probability exp(−∆χ2/2). To give proper weighting to each transit and RV data set, the uncertainties are scaled at the start of the MCMC so as to obtain a reduced χ2 of unity. From an initial MCMC fit for an eccentric or- bit, we found e = 0.13+0.18 −0.10. The improvement in the fit resulting from the addition of √e cos ω and √e sin ω as fitting parameters is too small to justify adoption of an eccentric orbit. The F -test approach of Lucy & Sweeney (1971) indicates that there is an 84 % probability that the improvement in the fit could have arisen by chance if the under- lying orbit were circular. In the absence of con- clusive evidence to the contrary, we adopted the circular orbit model. One spectrum of the eight SOPHIE spectra was taken during the start of transit. As we did not fit for the Rossiter-McLaughlin effect (e.g., Queloz et al. 2000) we excluded the resulting RV measurement (BJD = 2,455,301.4903) from our analysis. The median values and 1 σ uncertainties of the system parameters derived from the MCMC model fit are presented in Table 4. The corresponding transit and orbit models are superimposed on the transit photometry and radial velocities in Fig- ures 1 and 2. For a transit to be grazing, the grazing criterion (Smalley et al. 2011) must be satisfied: X = b + RP/R⋆ > 1. (1) For WASP-40b X = 0.9895+0.017 −0.014. Figure 4 shows the MCMC posterior distribution of RP/R∗ and b. A total of 26.4% of these data points satisfy the grazing criterion. Using the odds ratio test (e.g., Kipping et al. 2010), we find a 40.5% probability that the system is grazing. 5. System age Assuming aligned stellar-spin and planetary- orbit axes, the measured v sin I of WASP-40 and its derived stellar radius (Table 2) indicate a rota- tional period of Prot = 17.8± 7.8 days. Combining this with the B − V color of a K0 star from Gray (2008), we used the relationship of Barnes (2007) to estimate a gyrochronological age of 1.2+1.3 −0.8 Gyr. Considering that the stellar-spin axis may not be in the sky plane, these are upper limits on the stel- lar rotation period and the gyrochronological age. We found no evidence for rotational modulation in the WASP light curves. We interpolated the stellar evolution tracks of Marigo et al. (2008) and Bertelli et al. (2008) us- ing ρ∗ from the MCMC analysis and using Teff and [Fe/H] from the spectral analysis (Fig. 5). This suggests an age of 6 ± 5 Gyr and a mass of 0.83 ± 0.07 M⊙ for WASP-40. 6. Discussion As we find more exoplanets that transit their host stars we begin to see patterns in their dis- tribution. It is thus important to add new sys- tems to probe such patterns and to understand 5 a Value Table 4: System Parameters from RV and Transit Data Parameter P Tc T14 T12 ≈ T34 ∆F = R2 b i 0.0696 ± 0.0011 days 0.0282+undefined days 0.0152+0.00041 −0.00057 3.0395589 ± 0.0000090 days HJD 2, 455, 362.31489± 0.00023 b P/R2 ∗ K1 γ e M∗ R∗ log g ρ∗ Teff [Fe/H] −0.0037 0.866+0.016 −0.012 85.01+0.20 ◦ −0.26 91 ± 13 m s−1 0 (adopted) −15748.7 ± 1.0 m s−1 0.921 ± 0.034 M⊙ 0.864 ± 0.031 R⊙ 4.529 ± 0.027 (cgs) 1.43 ± 0.13 ρ⊙ 5246 ± 153 K 0.13 ± 0.11 MP 0.617 ± 0.088 MJup 1.038+0.068 RP −0.050 RJup log gP 3.112 ± 0.080 (cgs) ρP 0.54 ± 0.12 ρJup a 0.03995 ± 0.00050 AU TP,A=0 1177 ± 42 K a HJD = Heliocentric Julian Date (UTC). b 1 σ upper limit undefined as system is near- grazing. Fig. 4. -- The MCMC posterior distributions of b and RP/R∗. The solid line indicates the position of the stellar limb, i.e., b + RP/R∗= 1. A total of 26.4% of the points lie above the line and are grazing solutions. 6 the role of selection effects in transit surveys. For example, given the apparent pileup of tran- siting planets with periods near 3 -- 4 days, it is important to investigate the shape of the cut- off toward shorter periods (e.g., Szab´o & Kiss 2011). Further, there are increasing suggestions of correlations between planetary radii, the ir- radiation of the planet and the metallicity of the host star (Enoch et al. 2011; Anderson & Iro 2011, in preparation). For brighter systems, the measurement of the Rossiter -- McLaughlin effect allows us to build up statistics of the alignment of planetary orbits (e.g., Triaud et al. 2010), and to thus test suggestions of a correlation between alignment and the spectral type of the host star (Winn et al. 2010). Since all such comparisons depend on the re- liability of measured parameters of the transit- ing systems, it is worth noting that the param- eters of HAT-P-27b/WASP-40b reported here are in good accord with those from the independent study by B´eky et al. (2011). Most parameters agree to within 1 σ errors, while the impact param- eter, which is strongly dependent on the modeling of the transit light curve and the limb-darkening, agrees to within 2 σ. This adds confidence to the methods used by the two projects. One difference in our analyses is that our radial velocities show more scatter about the model than would be expected from their formal uncertainties (Fig. 6). This could be an indication of stellar ac- tivity. We calculated an activity index suggestive of an active star, log R′ HK = −4.63, directly from the SOPHIE spectra using the methods described in Boisse et al. (2010). The RV dispersion arising from the stellar activity of a K-type star is esti- mated to be ∼10 m s−1 (Santos et al. 2000). (2011), while resulting in a very similar model, do not show a similar scatter, which could indi- cate that the level of stellar activity fluctuates. Alternatively, if the scatter was indicative of an additional planet, the signal may not be present in the radial velocities of B´eky et al. (2011) due to limited sampling. Both the SOPHIE and Keck data sets are sparse, with eight and nine radial velocities, respectively, so more observations are needed to reach a conclusion. We note that the radial-velocity data in B´eky et al. The research leading to these results has re- ceived funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement number RG226604 (OP- TICON). SuperWASP-N is hosted by the Issac Newton Group on La Palma and WASP-South is hosted by the South African Astronomical Obser- vatory. We are grateful for their ongoing support and assistance. Funding for WASP comes from consortium universities and from the UK's Science and Technology Facilities Council. We thank Tom Marsh for the use of the ULTRACAM pipeline. Facilities: SuperWASP, OHP:1.93m, Liver- pool:2m, Euler1.2m REFERENCES Anderson, D. R., & Iro, N. 2011, in preparation Anderson, D. R., et al. 2011, ApJ, 726, L19 Bakos, G., Noyes, R. W., Kov´acs, G., Stanek, K. Z., Sasselov, D. D., & Domsa, I. 2004, PASP, 116, 266 Bakos, G. ´A., et al. 2009, ApJ, 696, 1950 Baranne, A., et al. 1996, A&AS, 119, 373 Barnes, S. A. 2007, ApJ, 669, 1167 B´eky, B., et al. 2011, preprint (arXiv:1101.3511) Bertelli, G., Girardi, L., Marigo, P., & Nasi, E. 2008, A&A, 484, 815 Boisse, I., et al. 2010, A&A, 523, A88 Borucki, W. J., et al. 2010, Science, 327, 977 Bouchy, F., et al. 2009, A&A, 505, 853 Bruntt, H., et al. 2010, MNRAS, 405, 1907 Claret, A. 2000, A&A, 363, 1081 Collier Cameron, A., et al. 2006, MNRAS, 373, 799 -- . 2007, MNRAS, 380, 1230 Dhillon, V. S., et al. 2007, MNRAS, 378, 825 Enoch, B., Collier Cameron, A., Parley, N. R., & Hebb, L. 2010, A&A, 516, A33+ Enoch, B., et al. 2011, MNRAS, 410, 1631 Fig. 5. -- Modified H-R diagram. The evolu- tionary mass tracks (Z = 0.026 ≈ [Fe/H] = 0.14; Y = 0.30) are from Bertelli et al. (2008). The isochrones (Z = 0.026 ≈ [Fe/H] = 0.14) for the ages 1, 4, 7, and 10 Gyr are from Marigo et al. (2008). Fig. 6. -- Residual radial velocities about the best- fitting circular Keplerian orbit. 7 Gibson, N. P., et al. 2008, A&A, 492, 603 Triaud, A. H. M. J., et al. 2010, A&A, 524, A25 Gillon, M., et al. 2009, A&A, 496, 259 West, R. G., et al. 2009, A&A, 502, 395 Winn, J. N., Fabrycky, D., Albrecht, S., & John- son, J. A. 2010, ApJ, 718, L145 Gray, D. F. 2008, The Observation and Analysis of Stellar Photospheres, ed. Gray, D. F. (Cam- bridge: Cambridge Univ. Press) Hellier, C., et al. 2011, EPJ Web Conf. 11, De- tection and Dynamics of Transiting Exoplan- ets, ed. F. Bouchy, R. D´ıaz, & C. Moutou (Provence: St. Michel l'Observatoire), 01004 Kipping, D. M., et al. 2010, ApJ, 725, 2017 Lucy, L. B., & Sweeney, M. A. 1971, AJ, 76, 544 Magain, P. 1984, A&A, 134, 189 Mandel, K., & Agol, E. 2002, ApJ, 580, L171 Marigo, P., Girardi, L., Bressan, A., Groenewegen, M. A. T., Silva, L., & Granato, G. L. 2008, A&A, 482, 883 Pepe, F., et al. 2005, The Messenger, 120, 22 Perruchot, S., et al. 2008, Proc. SPIE, 7014, 17 Pollacco, D., et al. 2008, MNRAS, 385, 1576 Pollacco, D. L., et al. 2006, PASP, 118, 1407 Queloz, D., Eggenberger, A., Mayor, M., Perrier, C., Beuzit, J. L., Naef, D., Sivan, J. P., & Udry, S. 2000, A&A, 359, L13 Queloz, D., et al. 2001, A&A, 379, 279 Santos, N. C., Mayor, M., Naef, D., Pepe, F., Queloz, D., Udry, S., & Blecha, A. 2000, A&A, 361, 265 Seager, S., & Mall´en-Ornelas, G. 2003, ApJ, 585, 1038 Skrutskie, M. F., et al. 2006, AJ, 131, 1163 Smalley, B., et al. 2011, A&A, 526, A130+ Steele, I. A., Bates, S. D., Gibson, N., Keenan, F., Meaburn, J., Mottram, C. J., Pollacco, D., & Todd, I. 2008, Proc. SPIE, 7014, 217 Szab´o, G. M., & Kiss, L. L. 2011, ApJ, 727, L44 Torres, G., Andersen, J., & Gim´enez, A. 2010, A&A Rev., 18, 67 This 2-column preprint was prepared with the AAS LATEX macros v5.2. 8
1001.5022	1	1001	2010-01-27T20:56:44	Dead Zones as Thermal Barriers to Rapid Planetary Migration in Protoplanetary Disks	[ "astro-ph.EP" ]	Planetary migration in standard models of gaseous protoplanetary disks is known to be very rapid ($\sim 10^5$ years) jeopardizing the existence of planetary systems. We present a new mechanism for significantly slowing rapid planetary migration, discovered by means of radiative transfer calculations of the thermal structure of protoplanetary disks irradiated by their central stars. Rapid dust settling in a disk's dead zone - a region with very little turbulence - leaves a dusty wall at its outer edge. We show that the back-heating of the dead zone by this irradiated wall produces a positive gradient of the disk temperature which acts as a thermal barrier to planetary migration which persists for the disk lifetime. Although we analyze in detail the migration of a Super-Earth in a low mass disk around an M star, our findings can apply to wide variety of young planetary systems. We compare our findings with other potentially important stopping mechanisms and show that there are large parameter spaces for which dead zones are likely to play the most important role for reproducing the observed mass-period relation in longer planetary periods.	astro-ph.EP	astro-ph	Dead Zones as Thermal Barriers to Rapid Planetary Migration in Protoplanetary Disks Yasuhiro Hasegawa and Ralph E. Pudritz1 Department of Physics and Astronomy, McMaster University, Hamilton, ON L8S 4M1, Canada YH:[email protected], REP:[email protected] Received ; accepted 0 1 0 2 n a J 7 2 . ] P E h p - o r t s a [ 1 v 2 2 0 5 . 1 0 0 1 : v i X r a 1Origins Institute, McMaster University, Hamilton, ON L8S 4M1, Canada -- 2 -- ABSTRACT Planetary migration in standard models of gaseous protoplanetary disks is known to be very rapid (∼ 105 years) jeopardizing the existence of planetary systems. We present a new mechanism for significantly slowing rapid planetary migration, discovered by means of radiative transfer calculations of the thermal structure of protoplanetary disks irradiated by their central stars. Rapid dust settling in a disk's dead zone - a region with very little turbulence - leaves a dusty wall at its outer edge. We show that the back-heating of the dead zone by this irradiated wall produces a positive gradient of the disk temperature which acts as a thermal barrier to planetary migration which persists for the disk lifetime. Although we analyze in detail the migration of a Super-Earth in a low mass disk around an M star, our findings can apply to wide variety of young planetary systems. We compare our findings with other potentially important stopping mechanisms and show that there are large parameter spaces for which dead zones are likely to play the most important role for reproducing the observed mass- period relation in longer planetary periods. Subject headings: accretion, accretion disks -- radiative transfer -- turbulence -- planets and satellites: formation -- protoplanetary disks -- 3 -- 1. Introduction Extrasolar planets (ESPs) have an unexpected distribution of orbital radii around their host stars (Udry et al. 2009) - ranging from about 0.02 to 70 astronomical units (AU).1 In particular, ESPs are observed to obey a mass-period (M-P) relation wherein lower mass planets end up in short period orbits around their host stars (Udry & Santos 2007). The predominance of very short period planets is generally thought to arise as a consequence of planetary migration. As an example, the tidal interaction of a planet with its surrounding gaseous disk excites density waves in the disk at so-called Lindblad resonances. These waves exert a torque back on the planet which results in a net angular momentum transfer between them (Goldreich & Tremaine 1980). Planets may also exchange angular momentum with the gas inside of their horseshoe region (Ward 1991). For locally isothermal protoplanetary disks with smoothly declining distributions of disk column density and temperature with radius, the net torque generally leaves a planet spiraling inwards through the disk (Tanaka et al. 2002), i.e. the torque exerted by the outer wake is marginally stronger than that of the inner wake (Ward 1997). Many calculations and simulations show that the migration timescale of planets in such "standard" disk models is very short - roughly two orders of magnitude smaller than the disk lifetime (one to ten million years (Myr)) (Ward 1997; Nelson et al. 2000; Tanaka et al. 2002; D'Angelo et al. 2003). Why are there any planetary systems at all? The key to understanding the M-P relation and the survival of planetary systems is in how the dynamics of planetary motion is coupled to the properties and structure of the protoplanetary disks. As an example, Schlaufman et al. (2009) focused on the surface density transition that can be produced at the location of the ice-line, where a local pressure maximum can act as an accumulation point for planetesimals (Kretke & Lin 2007). If it 1See the website: http://exoplanet.eu. -- 4 -- is assumed that type I migration is much slower than predicted in locally isothermal disk models, this feature could account for planets with orbital radii 0.1 - 2 AU. Obviously, a physical explanation for slower migration is needed. In this Letter, we present a new slowing mechanism of rapid type I migration -which may occur for planets with masses that are too low to open up a gap in their disks (massive planets can tidally form a gap and undergo type II migration). We show, by means of Monte Carlo radiative transfer simulations, that dead zones - the dense inner disk region wherein turbulence cannot be readily excited (Gammie 1996) - support a thermal barrier to migration. One of the most important consequences is that the thermal barrier could account for planets at larger orbital radii. In § 2, we outline our disk model and discuss tidal torques. In § 3, we analyze how the presence of a thermal barrier impacts the migration rates of low-mass planets. In § 4, we discuss potential issues for the M-P relation by comparing our stopping mechanism with others. 2. Disk model & tidal torques Protoplanetary disks are known to be heated by radiation from the central star (Chiang & Goldreich 1997; D'Alessio et al. 1998; Hasegawa & Pudritz 2010a, hereafter, HP10). This radiation mainly determines the thermal structure of disks (Kenyon & Hartmann 1987). This is because viscous heating dominates stellar irradiation heating only within about 1 AU for the classical T Tauri star systems (CTTSs; D'Alessio et al. 1998) and only within about 0.1 AU for lower mass stars such as M stars (HP10). Detailed modeling of the spectral energy distributions emitted by disks has shown that s = −1 for disks where the disk surface density Σ ∝ rs (D'Alessio et al. 1998). It is -- 5 -- well established that the sign of a net torque exerted on planets depends on two central properties of disks, their surface density and temperature (e.g., Ward 1997). The thermal structure therefore plays a critical role in controlling the direction of planetary migration. In disks without internal structure, the disk temperature T ∝ rt at the mid-plane steadily decreases (t < 0) and planetary migration is steadily inwards. The point is that disks are not simple power-law structures. The strength of turbulence within them varies considerably, with very low levels occurring in dense regions called dead zones (Gammie 1996) that initially extend over roughly 10 AU in disks (Matsumura & Pudritz 2006) and then gradually shrink in size as disk material is accreted onto the central star (Matsumura et al. 2009, hereafter, MPT09). Turbulence in disks is most likely excited by the magnetorotational instability (MRI) (Balbus & Hawley 1991). The MRI requires good coupling between ions and magnetic fields, which is largely absent in the dead zone - that inner, high density region in which the ionization due to the X-rays from the central star and cosmic rays is suppressed. Dust is the dominant absorber of stellar radiation in disks although its total mass is 100 times smaller than that of gas (Dullemond et al. 2009). Many observations imply that it has a density distribution that is different from the gas distribution, which is derived assuming vertical hydrostatic equilibrium (Kenyon & Hartmann 1987). Dust settling, a consequence of its size distribution (Dullemond & Dominik 2004a), is ubiquitous in protoplanetary disks around any young star (HP10, references herein). The dust scale height depends upon the amplitude of disk turbulence which keeps it suspended in the gravitational field of the disk (which is determined by the central star) (Dubrulle et al. 1995). We demonstrated in HP10 that dust settling in dead zones results in the appearance of a limited region in which the disk temperature can actually increase with radius - a radial temperature inversion. In this Letter, we adopt the disk model developed in HP10 -- 6 -- (see Table. 1) and focus our computations on M dwarf systems, such as the recently discovered Super-Earth (∼ 5M⊕; M⊕ is the Earth's mass) with an orbital radius of 2 AU (Beaulieu et al. 2006). We refer the readers to HP10 for the detail. The low mass of disks around M stars allows much more comprehensive Monte Carlo simulations to be performed, but our analysis in principle applies to any protoplanetary disk. We adopt the torque formula (Ward 1997; Menou & Goodman 2004; Jang-Condell & Sasselov 2005) in which only the Lindblad torque is considered. This is because the corotation torque is readily saturated (i.e. is canceled out) in dead zones in our radiatively heated disk model. We compare the libration timescale (ie the timescale for gas to complete an orbit in the horse- shoe region), τlib ≈ 8πrp/3Ωpxs, with the viscous timescale, τvis ≈ x2 width of the horseshoe region is xs/rp = 1.68(Mprp/M∗hp)1/2 (Paardekooper & Papaloizou s/3ν, where the half- 2009), the kinematic viscosity is ν = αh2ΩKep (Shakura & Sunyaev 1973), and the disk scale height is h. Our numerical results give hp/rp ≃ 0.05, so that the critical value of turbulence, αcrit, below which the corotation torque is saturated (and therefore negligible), is αcrit = 0.01(cid:18) Mp 5M⊕(cid:19)3/2(cid:18) M∗ 0.1M⊙(cid:19)−3/2(cid:18)hp/rp 0.05 (cid:19)−7/2 . (1) Since the dead zone has α = 10−5, we can safely neglect the corotation torque for Mp & 0.5M⊕ in the dead zone of our disk model. In addition, we confirmed that, in the active region where the corotation torque is generally unsaturated, both Lindblad and corotation torques result in inward migration in our disk model (Paardekooper et al. 2009). Thus, exclusion of the corotation torque in the active region does not affect our findings in our disk model. We note that corotation torque may be unsaturated in dead zones for sufficiently small planetary masses, but the exact limit will depends on knowing the disk scale height that is established by the host star. Furthermore, our torque formula takes into account the effects of vertical disk thickness by diluting the gravitational force of a planet by z (Jang-Condell & Sasselov 2005). -- 7 -- 3. Results We performed numerical simulations of the thermal structure of 2D disks by solving the radiative transfer equation with a Monte Carlo method (Dullemond & Dominik 2004b, HP10). We included the effects of vertical hydrostatic balance, dust settling, a dead zone, and the gravitational field of a planet embedded in the disk. The tidal torque is calculated as described in § 2, and incorporates our numerical data, in order to calculate the migration time. Figure 1 presents the thermal and density structure of a disk with a 5M⊕ planet placed at 8 AU. The top and bottom panels show the dust and gas densities by color, respectively. Since we define the disk temperature as the mass-averaged temperature of dust, both panels show the identical temperature structure which is represented by contours. The thick line denotes the Hill radius rH = rp(Mp/3M∗)1/3. In this Letter, we adopt, without loss of generality, a dead zone which is 6 AU in size. Dead zones enhance dust settling because of the low turbulent amplitude there. Since disks have inner dead, and outer active regions for turbulence, the transition of the density distribution of dust occurs at the outer edge of the dead zone. This leaves a marked step in the dust scale height behind - in effect a wall of dust. The additional stellar energy absorbed at the wall is distributed by radiative diffusion (Hasegawa & Pudritz 2010b) since the optical depth at this region is high. The resulting radial "thermal inversion" - ie a region of increasing temperature with increasing disk radius - is shown in Figure 2. An analytic fit to our data shows that this back-heated region in the dead zone has a positive temperature gradient described by a power-law T ∝ rt′ with t′ > 3/2. We show in Figure 3 (top) that this radial thermal inversion causes the migration rate to be positive (the migration time becomes negative - bottom panel), - i.e. planets migrate outward in the region with the positive temperature gradient. The physical explanation of -- 8 -- this behavior is that the increasing function of disk temperature changes the disk's pressure distribution which in turn causes the position of the outer Lindblad resonances to be further from the planet than the inner ones (Artymowicz 1993). This results in outer torques that are much weaker than the inner ones.2 Figure 3 (bottom) also shows that the planets very slowly enter the region of torque reversal - as seen by the strong positive "spike" in the migration time. These regions correspond to radii at which dT /dr ≃ 0 (see Figure 2) at the inner and outer resonances, making the torque difference between them very small. We emphasize that the positive temperature gradient arising from the wall-like dust structure is achieved for the case of a finite transition region, △r ≤ 10h, in the value of turbulent α (HP10). Although, for simplicity, we adopt a sharp spatial transition from the active to the dead zone in this Letter (△r = 0), our above results are valid for the case of △r ≃ h, since the positions of Lindblad resonances are typically offset from the planets by 2h/3 (Artymowicz 1993). In addition, it is interesting that the migration timescale of the M star system is similar to that of CTTS (∼ 104 − 105 years) for the other two cases (well mixed, dust settling). This is because the tidal torque is scaled by Σ(h/r)−2. A more detailed discussion of them is presented in Hasegawa & Pudritz (2010b). 4. Discussion 4.1. A thermal vs a density barrier at outer dead zone radius We find that the dusty wall produces a radial temperature inversion that is a thermal barrier for rapid type I planetary migration. Whereas the torque balance in the well coupled active zone forces planets to migrate inward, once they encounter the radial thermal 2We will show in a forthcoming paper that the Lindblad torque becomes negative when s − t/2 < −7/4 (Hasegawa & Pudritz 2010b). -- 9 -- inversion region, the torque balance reverses, and they move out of the region. Thus, planets are trapped there if they originally migrate from the active region beyond the dead zones, or even if they formed close to the outer edge of the dead zone. The astrophysical implications of this result are very important since we have shown that dusty protoplanetary disks with dead zones possess an innate mechanism for strongly slowing planetary migration within them, provided that the corotation torque is saturated. While such a thermal barrier exists for any size of dead zone (HP10), its effectiveness is probably most important for lower mass disks, as we now show. The density structure of disks evolves with time due to viscous evolution. MPT09 found that the difference of α between the active and dead regions produces a steep density gradient at the boundary and the location of their jump moves inward with time over the long (∼ 10 Myr) viscous timescale of the dead zone. This density gradient at the outer dead zone boundary also plays an important role in slowing down or stopping type I migration, provided that planets migrate from larger disk radii. The inner torques become larger than the outer ones in the density gradient region, resulting in the reflection of migrating planets off the density gradient (MPT09). The relative importance of these two dead zone mechanisms is controlled by the ratio of dust settling τset ≈ Σ/√2πρdaΩKep and the viscous τvis timescales, where ρd is the bulk density of dust and a is the grain size of dust. We find that the critical condition τset/τvis ≥ 1 for dominance of the density vs thermal barriers presented by a dead zone is (2) r(cid:19)2 Σ(cid:18)h ≥ 25(cid:16) α 10−2(cid:17)−1(cid:18) ρd 1g cm−3(cid:19)(cid:16) a 1mm(cid:17) ≡ Ccrit. This implies that a density barrier is dominant for sufficiently large values of Σ and h/r. For disks around CTTSs with a typical dead zone size (≈ 10 AU), Σ(h/r)2 ∼ 300 g cm−2 × (0.4)2 ≈ 2Ccrit (Chiang & Goldreich 1997), which indicates that a density barrier is dominant. For disks around M stars with a typical dead zone size (≈ 5 AU), Σ(h/r)2 ∼ 20 -- 10 -- g cm−2 × (0.05)2 = 2 × 10−3Ccrit (Scholz et al. 2007), which implies that a thermal barrier is dominant. Generally, we find that a thermal barrier becomes weaker in the late stages because of accretion which reduces the density at the outer edge of the dead zones. 4.2. Comparisons with other possible stopping mechanisms and potential effects on the M-P relation It is well known that tidal interaction and angular momentum exchange with the central star (Lin et al. 1996) and the creation of a hole in the inner part of disks by the presence of a stellar magnetosphere (Shu et al. 1994; Lin et al. 1996) cannot reproduce the whole of the observed M-P relation. This is because such barriers become important only for planets approaching very close to the central star. Most ESPs, however, have observed orbital radii from about 0.02 AU to 10 AU (Udry & Santos 2007). Thus, while these barriers may play a role in piling up ESPs in the vicinity of the star, they have difficulty in predicting planets with larger orbital radii. Stochastic migration provides another mechanism for controlling planetary migration. It arises when disks undergo magnetohydrodynamic turbulence (Nelson & Papaloizou 2004; Laughlin et al. 2004), which is the outcome of the MRI (Balbus & Hawley 1991). Stochastic torques tend to reduce the timescale for planetary survival, however (Johnson et al. 2006). In a few exceptional cases, planets can diffuse out to large disk radii where they can survive longer. Planets within the dead zones cannot perform random walks because turbulent torques are reduced by about two orders of magnitude there and consequently would undergo steady inward type I migration (Oishi et al. 2007). Thus, stochastic migration is unlikely to be the main barrier to rapid type I migration. Is our thermal barrier sufficiently wide to stop planets scattered into the dead zone by -- 11 -- stochastic effects? We consider a characteristic length scale for turbulent diffusion defined by △rturb = √ντc = hpατc/Ω−1, where τc is the correlation time of turbulence. We adopt a value τc = 0.5Ω−1 (Nelson & Papaloizou 2004) and find that △rturb ≈ 0.02 AU at r = 6 AU. This is shorter than the width of the thermal barrier (∼ 2 AU). Therefore, stochastic motions due to turbulence are unlikely to affect the migration stopping mechanism at the thermal barrier. Corotation torques are also potentially important for slowing or stopping planets. In both barotropic and adiabatic disks, corotation torque associated with a radial vortensity gradient may work as a barrier around the region of inner edge of the dead zone (∼ 0.01-0.1 AU; Fromang et al. 2002; Masset et al. 2006, MPT09) because it becomes positive due to a positive surface density gradient, resulting in outward planetary migration (Masset et al. 2006). However, the location of the barrier is almost constant with time because stellar irradiation controls it, and consequently this barrier is important only for planets in the vicinity of the star. In adiabatic disks, corotation torque associated with a radial entropy gradient may also act as a barrier because it becomes large and positive around the region with a large (in magnitude), negative entropy gradient (Paardekooper & Mellema 2006; Baruteau & Masset 2008; Paardekooper & Mellema 2008; Paardekooper & Papaloizou 2008). Thus, corotation torque may be important to the M-P relation in certain regimes of planetary mass and disk heating, as noted above. Our mechanism has a movable barrier that shrinks from larger disk radii on the 10 Myr (for CTTSs) viscous timescale of the dead zone. As a concrete example, this shrinkage of the dead zone could explain the recent detected Super-Earth at 2 AU around an M star (Beaulieu et al. 2006). This is because a Super-Earth is likely to be the most massive planet that would surely form in a low mass disk and is therefore most likely to have been formed beyond the outer dead zone radius and left behind as the dead zone shrinks away. -- 12 -- We conclude that the stopping mechanisms arising from dead zones - via thermal and density gradients - are important barriers to rapid planetary migration. We have shown that the thermal barrier that arises from disk irradiation by a heated dust wall is robust and may be most important in the earlier phases of disk evolution and for the evolution of low mass systems as found around M stars as an example. Dead zones may provide the most significant slowing mechanism of type I migration that is needed to explain the longer period of population of the M-P relation, which will be checked in future population synthesis models. The authors thank Kees Dullemond, Thomas Henning, Hubert Klahr, Kristen Menou, Soko Matsumura and Richard Nelson for stimulating discussions. We also thank an anonymous referee for a useful report. Our simulations were carried out on computer clusters of the SHARCNET HPC Consortium at McMaster University. YH is supported by McMaster University, as well as by Graduate Fellowships from SHARCNET and the Canadian Astrobiology Training Program (CATP). REP is supported by Discovery Grant from NSERC. -- 13 -- REFERENCES Artymowicz, P. 1993, ApJ, 419, 155 Balbus, S. A. & Hawley, J. F. 1991, ApJ, 376, 223 Baruteau, C. & Masset, F. S. 2008, ApJ, 672, 1054 Beaulieu, J.-P. et al. 2006, Nature, 439, 437 Chiang, E. I. & Goldreich, P. 1997, ApJ, 490, 368 D'Alessio, P., Cant´o, J., Calvet, N., & Lizano, S. 1998, ApJ, 500, 411 D'Angelo, G., Kley, W., & Henning, T. 2003, ApJ, 586, 540 Dubrulle, B., Morfill, G., & Sterzik, M. 1995, Icarus, 114, 237 Dullemond, C. P. & Dominik, C. 2004a, A&A, 421, 1075 -- . 2004b, A&A, 417, 159 Dullemond, C. P., Hollenbach, D., Kamp, I., & D'Alessio, P. 2009, Protostars and Planets V (Tucson: Univ. Arizona Press) Fromang, S., Terquem, C., & Balbus, S. A. 2002, MNRAS, 329, 18 Gammie, C. F. 1996, ApJ, 457, 355 Goldreich, P. & Tremaine, S. 1980, ApJ, 241, 425 Hasegawa, Y. & Pudritz, R. E. 2010a, MNRAS, 401, 143 -- . 2010b, in preparation Jang-Condell, H. & Sasselov, D. D. 2005, ApJ, 619, 1123 -- 14 -- Johnson, E. T., Goodman, J., & Menou, K. 2006, ApJ, 647, 1413 Kenyon, S. J. & Hartmann, L. 1987, ApJ, 323, 714 Kretke, K. A. & Lin, D. N. C. 2007, ApJ, 664, L55 Laughlin, G., Steinacker, A., & Adams, F. C. 2004, ApJ, 608, 489 Lin, D. N. C., Bodenheimer, P., & Richardson, D. C. 1996, Nature, 380, 606 Masset, F. S., Morbidelli, A., Crida, A., & Ferreira, J. 2006, ApJ, 642, 478 Matsumura, S. & Pudritz, R. E. 2006, MNRAS, 365, 572 Matsumura, S., Pudritz, R. E., & Thommes, E. W. 2009, ApJ, 691, 1764 Menou, K. & Goodman, J. 2004, ApJ, 606, 520 Nelson, R. P. & Papaloizou, J. C. B. 2004, MNRAS, 350, 849 Nelson, R. P., Papaloizou, J. C. B., Masset, F., & Kley, W. 2000, MNRAS, 318, 18 Oishi, J. S., Mac Low, M.-M., & Menou, K. 2007, ApJ, 670, 805 Paardekooper, S.-J., Baruteau, C., Crida, A., & Kley, W. 2009, preprint (astro- ph/arXiv:0909.4552v1) Paardekooper, S.-J. & Mellema, G. 2006, A&A, 459, L17 -- . 2008, A&A, 478, 245 Paardekooper, S.-J. & Papaloizou, J. C. B. 2008, A&A, 485, 877 -- . 2009, MNRAS, 394, 2297 Schlaufman, K., Lin, D. N. C., & Ida, S. 2009, ApJ, 691, 1322 -- 15 -- Scholz, A., Jayawardhana, R., Wood, K., Meeus, G., Stelzer, B., Walker, C., & O'Sullivan, M. 2007, ApJ, 660, 1517 Shakura, N. I. & Sunyaev, R. A. 1973, A&A, 24, 337 Shu, F., Najita, J., Ostriker, E., Wilkin, F., Ruden, S., & Lizano, S. 1994, ApJ, 429, 781 Tanaka, H., Takeuchi, T., & Ward, W. R. 2002, ApJ, 565, 1257 Udry, S., Fischer, D., & Queloz, D. 2009, Protostars and Planets V (Tucson: Univ. Arizona Press) Udry, S. & Santos, N. C. 2007, ARA&A, 45, 397 Ward, W. R. 1991, in Luner and Planetary Institute Conference Abstract, 1463 -- . 1997, Icarus, 126, 261 This manuscript was prepared with the AAS LATEX macros v5.2. -- 16 -- Table 1: Summary of parameters & symbols Symbol M∗ R∗ T∗ Σ Md α Mp Meaning Stellar mass Stellar radius Stellar effective temperature The surface density of (gas + dust) The total disc mass (gas + dust) gas-to-dust ratio Value 0.1 M⊙ 0.4R⊙ 2850 K ∝ r−1 4.5 × 10−3 M⊙ 100 Parameter for turbulence (active/dead) 10−2/10−5 Planetary mass 5 M⊕ M⊙ is one solar mass, and R⊙ is one solar radius. -- 17 -- Fig. 1. -- The density and temperature structures of disks with a 5M⊕ mass planet located at 8 AU. For both panels, the density is denoted by colors [g cm−3] as shown in colorbar, and the disk temperature is denoted by contours [K] (both panels show the identical temperature since the disk temperature is defined by taking the mass-average of the dust temperatures). The size of a dead zone is 6 AU. Top: the dust density. Bottom: the gas density. The thick black circle denotes the Hill sphere rH = rp(Mp/3M∗)1/3, where rp and Mp is the position and mass of a planet, respectively, and M∗ is the stellar mass. A wall-like structure appears at the boundary between the active and dead zones in the dust distribution while it is not in the gas. -- 18 -- Fig. 2. -- The disk temperature in the disk mid-plane. The solid line is for the case of dead zone, the dashed line is the analytical model of disk temperature (Chiang & Goldreich 1997), and the dashed-dot line is for analytical fit to a positive temperature gradient. The size of a dead zone, which is 6 AU, is indicated by the vertical solid line. -- 19 -- Fig. 3. -- Migration rate and time for a 5 mass M⊕ planet at various orbital radii on the top and bottom panels, respectively. For both panels, the solid line denotes the case of a dead zone which is 6 AU in size whose position is indicated by the vertical solid line, the red dashed line is for the well-mixed case, and the blue dashed-dot line is for the dust settling case. The negative migration time region in the bottom panel (outward migration) corresponds to the region with a positive temperature gradient (see Figure 2).
1703.06582	4	1703	2017-05-24T15:54:36	The Apparently Decaying Orbit of WASP-12	[ "astro-ph.EP" ]	We present new transit and occultation times for the hot Jupiter WASP-12b. The data are compatible with a constant period derivative: $\dot{P}=-29 \pm 3$ ms yr$^{-1}$ and $P/\dot{P}= 3.2$ Myr. However, it is difficult to tell whether we have observed orbital decay, or a portion of a 14-year apsidal precession cycle. If interpreted as decay, the star's tidal quality parameter $Q_\star$ is about $2\times 10^5$. If interpreted as precession, the planet's Love number is $0.44\pm 0.10$. Orbital decay appears to be the more parsimonious model: it is favored by $\Delta\chi^2=5.5$ despite having two fewer free parameters than the precession model. The decay model implies that WASP-12 was discovered within the final $\sim$0.2% of its existence, which is an unlikely coincidence but harmonizes with independent evidence that the planet is nearing disruption. Precession does not invoke any temporal coincidence, but does require some mechanism to maintain an eccentricity of $\approx$0.002 in the face of rapid tidal circularization. To distinguish unequivocally between decay and precession will probably require a few more years of monitoring. Particularly helpful will be occultation timing in 2019 and thereafter.	astro-ph.EP	astro-ph	DRAFT VERSION MAY 25, 2017 Preprint typeset using LATEX style AASTeX6 v. 1.0 7 1 0 2 y a M 4 2 . ] P E h p - o r t s a [ 4 v 2 8 5 6 0 . 3 0 7 1 : v i X r a THE APPARENTLY DECAYING ORBIT OF WASP-12b KISHORE C. PATRA1, JOSHUA N. WINN2, MATTHEW J. HOLMAN3, LIANG YU1, DRAKE DEMING4, AND FEI DAI1,2 1Department of Physics, and Kavli Institute for Astrophysics and Space Research, Massachusetts Institute of Technology, Cambridge, MA 02139, USA 2Department of Astrophysical Sciences, Princeton University, 4 Ivy Lane, Princeton, NJ 08540, USA 3Harvard-Smithsonian Center for Astrophysics, 60 Garden Street, Cambridge, MA 02138, USA 4Department of Astronomy, University of Maryland at College Park, College Park, MD 20742, USA ABSTRACT We present new transit and occultation times for the hot Jupiter WASP-12b. The data are compatible with a constant period derivative: P = −29 ± 3 ms yr−1 and P/ P = 3.2 Myr. However, it is difficult to tell whether we have observed orbital decay or a portion of a 14-year apsidal precession cycle. If interpreted as decay, the star's tidal quality parameter Q⋆ is about 2 ×105. If interpreted as precession, the planet's Love number is 0.44 ±0.10. Orbital decay appears to be the more parsimonious model: it is favored by ∆χ2 = 5.5 despite having two fewer free parameters than the precession model. The decay model implies that WASP-12 was discovered within the final ∼0.2% of its existence, which is an unlikely coincidence but harmonizes with independent evidence that the planet is nearing disruption. Precession does not invoke any temporal coincidence, but it does require some mechanism to maintain an eccentricity of ≈0.002 in the face of rapid tidal circularization. To distinguish unequivocally between decay and precession will probably require a few more years of monitoring. Particularly helpful will be occultation timing in 2019 and thereafter. Keywords: planets and satellites: individual (WASP-12 b) - planet-star interactions 1. INTRODUCTION More than 20 years have elapsed since the discovery of hot Jupiters (Mayor & Queloz 1995). The time may be ripe to confirm a long-standing theoretical prediction: the orbits of almost all of these planets should be shrinking due to tidal or- bital decay (Rasio et al. 1996; Sasselov 2003; Levrard et al. 2009). This is because the star's rotational angular momen- tum is typically smaller than one-third of the orbital angular momentum, the critical value beneath which tidal evolution has no stable equilibrium (Hut 1980). Tidal decay of hot Jupiters has been invoked to explain cer- tain properties of the ensemble of star-planet systems. For example, the scarcity of gas giants with periods less than a day is suggestive of orbital decay (see, e.g. Jackson et al. 2008; Hansen 2010; Penev et al. 2012; Ogilvie 2014). The anomalously rapid rotation of some hot-Jupiter host stars has been attributed to transfer of the planet's orbital an- gular momentum (Penev et al. 2016). The absence of hot Jupiters around subgiant stars may be caused by an accel- eration of orbital decay when a star leaves the main sequence (Villaver & Livio 2009; Hansen 2010; Schlaufman & Winn 2013). Tidal decay might also be responsible for the lower occurrence of close-in planets around rapidly rotating stars (Teitler & Königl 2014), or the realignment of stars and their planetary orbits (Matsakos & Königl 2015). However, direct evidence for orbital decay has been lacking: there have been no clear demonstrations of a long-term period decrease due to orbital decay (see, e.g., Hoyer et al. 2016; Wilkins et al. 2017). Another unfulfilled prediction is that the orbits of hot Jupiters should be apsidally precessing on a timescale of decades (Miralda-Escudé 2002; Heyl & Gladman 2007; Pál & Kocsis 2008; Jordán & Bakos 2008), as long as the orbits are at In particular, Ragozzine & Wolf (2009) noted that the theoretical preces- sion rate is dominated by the contribution from the planet's tidally deformed mass distribution. They advocated a search for apsidal precession as a means of probing the interiors of hot Jupiters. least slightly eccentric. With an orbital period of 1.09 days, WASP-12b is one of the shortest-period giant planets known (Hebb et al. 2009), and has been monitored for a decade. It is, therefore, an out- standing target in the search for orbital decay and apsidal pre- cession. Maciejewski et al. (2016) reported a decrease in the apparent period. Despite being the most convincing claim that has yet been presented for orbital decay, those authors could not distinguish between true period shrinkage and a long-term oscillation of the apparent period due to apsidal precession. In this paper, we present new transit and occulta- tion times (§ 2 and 3). We use all of the available data to test which model is favored by the data: a constant period deriva- tive, or sinusoidal variations arising from apsidal precession (§ 4). We also discuss the implications of both models (§ 5) and prospects for future observations (§ 6). 2. NEW TRANSIT TIMES Between 2016 October and 2017 February, we observed seven transits of WASP-12 with the 1.2m telescope at the Fred Lawrence Whipple Observatory on Mt. Hopkins, Ari- zona. Images were obtained with the KeplerCam detector through a Sloan r′-band filter. The typical exposure time was 2 PATRA ET AL. 1.00 0.98 x u l f e v i t l a e R 0.96 0.94 −0.10 −0.05 0.00 0.05 t − t0 [days] 1252 1284 1338 1339 1348 1359 1370 0.10 Figure 1. New transit light curves. Black points are based on ob- servations with the FLWO 1.2m telescope in the Sloan r′ band. Red curves are the best-fit models. Epoch numbers are printed to the right of each curve. Vertical offsets have been applied to separate the light curves. 15 s, chosen to give a signal-to-noise ratio of about 200 for WASP-12. The field of view of this camera is 23.′1 on a side. We used 2x2 binning, giving a pixel scale of 0.′′68. The raw images were processed by performing standard overscan correction, debiasing, and flat-fielding with IRAF1. Aperture photometry was performed for WASP-12 and an en- semble of 7-9 comparison stars of similar brightness. The aperture radius was chosen to give the smallest scatter in the flux outside of the transits, and was generally 7-8 pixels. The reference signal was generated by summing the flux of the comparison stars. The flux of WASP-12 was then divided by this reference signal to produce a time series of relative flux. Each time series was normalized to have unit flux outside of the transit. The time stamps were placed on the BJDTDB sys- tem using the code of Eastman et al. (2010). We fitted a Mandel & Agol (2002) model to the data from each transit. The parameters of the transit model were the midtransit time, the planet-to-star radius ratio (Rp/R⋆), the scaled stellar radius (R⋆/a), and the impact parameter (b = a cos i/R⋆). For given values of R⋆/a and b, the tran- sit timescale is proportional to the orbital period [see, e.g., Eqn. (19) of Winn (2010)]. To set this timescale, we held the period fixed at 1.09142 days, although the individual tran- sits were fitted separately with no requirement for period- icity. To correct for differential extinction, we allowed the apparent magnitude to be a linear function of airmass, giv- ing two additional parameters. The limb darkening law was assumed to be quadratic, with coefficients held fixed at the values (u1 = 0.32, u2 = 0.32) tabulated by Claret & Bloemen (2011) for a star with the spectroscopic parameters given by Hebb et al. (2009).2 To determine the credible intervals for the parameters, we used the emcee Markov Chain Monte Carlo (MCMC) code written by Foreman-Mackey et al. (2013). The transition dis- tribution was proportional to exp(−χ2/2) with χ2 = (cid:18) fobs,i − fcalc,i σi (cid:19)2 , N Xi=1 (1) where fobs,i is the observed flux at time ti and fcalc,i is the cor- responding flux of the model. The uncertainties σi were set equal to the standard deviation of the out-of-transit data. In a few cases, the pre-ingress scatter was noticeably different than the post-egress scatter; for those observations, we as- signed σi by linear interpolation between the pre-ingress and post-egress values. Figure 1 shows the light curves and the best-fit models. Table 1 reports the midtransit times and their uncertainties. For convenience, this table also includes the new occulta- tion times described below, as well as the previously reported times that are analyzed in Section 4. The results for the other transit parameters were consistent with the previous results of Maciejewski et al. (2013), with larger uncertainties. Time-correlated noise is evident in some of the new light curves. Although we made no special allowance for these correlations in our analysis, we have reason to believe that the quoted uncertainties are reliable. When these seven new mid- transit times are fitted with a linear function of epoch, we ob- tain χ2 min = 5.1 with five degrees of freedom. When the period is held fixed at the value derived from all 10 years of timing data, we obtain χ2 min = 7.8 with six degrees of freedom. These tests suggest that the uncertainties are not substantially un- derestimated. Furthermore, spurious timing variations would be random from night to night, whereas our long-term tim- ing analysis (Section 4) reveals that all seven new midtransit times produce residuals of the same sign and amplitude. 1 The Image Reduction and Analysis Facility (IRAF) is distributed by the National Optical Astronomy Observatory, which is operated by the Associa- tion of Universities for Research in Astronomy (AURA) under a cooperative agreement with the National Science Foundation. 2 For this purpose we used the online code of Eastman et al. (2013): http://astroutils.astronomy.ohio-state.edu/exofast/limbdark.shtml APPARENTLY DECAYING ORBIT OF WASP-12B 3 CH2 (4.5 m m) CH1 (3.6 m m) x u l f e v i t l a e R 1.006 1.004 1.002 1.000 0.998 −0.10 −0.05 0.00 0.05 0.10 −0.10 −0.05 0.00 0.05 0.10 t − t0 [days] Figure 2. New occultation light curves. Black points are the binned Spitzer measurements from epochs 305 (left) and 308 (right). Red curves are the best-fit models. 3. NEW OCCULTATION TIMES We measured two new occultation times based on hitherto unpublished Spitzer observations in 2013 December (pro- gram 90186, P.I. Todorov). Two different transits were ob- served, one at 3.6 µm and one at 4.5 µm. The data take the form of a time series of 32x32-pixel subarray images, with an exposure time of 2.0 s per image. The data were acquired over a wide range of orbital phases, but for our purpose, we analyzed only the ≈14,000 images within 4 hours of each occultation. We also reanalyzed the Spitzer occultation pre- sented by Deming et al. (2015) using the technique described below. We determined the background level in each image by fit- ting a Gaussian function to the histogram of pixel values, af- ter excluding the high flux values associated with the star. The centroid of the fitted Gaussian function was taken to be the background value and was subtracted from each image prior to performing aperture photometry. We used two different schemes to choose photometric aperture sizes. In the first scheme, we used 11 apertures ranging in radius from 1.6-3.5 pixels in average increments of 0.2 pixel. In the second scheme, we tried 11 apertures for which the radius was allowed to vary at each time step, based on the procedure described in Appendix A of Lewis et al. (2013). In this procedure, the aperture radius is taken to be the sum of a constant (ranging from 0-2 pixels) and the noise pixel radius, defined as the square root of the ratio of the square of the total flux integrated over all pixels divided by the sum of the squared-fluxes in individual pixels. The noise pixel radius is specific to each image and allows for possible changes in the shape of the pixel response function with position. We also tried two different methods to choose the center of the apertures: fitting a two-dimensional Gaus- sian function to the stellar image, and computing the flux- weighted center-of-light. Hence, there were four versions of the photometry: constant versus variable aperture radii, and Gaussian centroiding versus center-of-light. Each of those four versions contains 11 time series with different aperture sizes. We corrected for the well-known intrapixel sensitivity vari- ations using pixel-level decorrelation [PLD; Deming et al. (2015)]. In PLD, the flux time series is modeled as the sum of the astrophysical variation, a temporal baseline, and a weighted sum of the (normalized) time series of each pixel comprising the point-spread function. Because each pixel value is divided by the total brightness of the star in that im- age, PLD effectively separates astrophysical information and Spitzer detector effects. PLD has also been used to produce high-quality photometry from K2 data (Luger et al. 2016). Our implementation of PLD operates on time-binned data [see Sec. 3.1 of Deming et al. (2015)]. Over a trial range of occultation midpoints and median aperture radii, the code uses linear regression to find the best-fit occultation depth and pixel coefficients. We provisionally adopt the midpoint that produces the best fit (smallest χ2). The code then varies the aperture radius from among the 11 possible values and the duration of the time bins. The optimal values of the radius and bin size are determined by examining the Allan (1966) deviation relation of the residuals and identifying the case 4 PATRA ET AL. that comes closest to the ideal relation.3 Then, an MCMC procedure is used to optimize the light-curve parameters (in- cluding the time of mid-occultation), pixel coefficients, and temporal baseline coefficients. The temporal baseline was taken to be a quadratic function of time, which was sufficient to describe the phase-curve variation in the vicinity of the occultation. After performing these steps for all four different versions of the photometry, we adopted the version that came clos- est to achieving the theoretical photon noise limit. For the 3.6 µm data, the adopted version used 10-frame binning, center-of-light centroiding, and a constant aperture radius of 2.3 pixels. For the 4.5 µm data, the adopted version used 10-frame binning, center-of-light centroiding, and a constant aperture radius of 2.2 pixels. With these choices, we achieved a noise level of 1.29 and 1.24 times the theoretical photon noise limit at 3.6 and 4.5 µm, respectively. The uncertainty in the midpoint of each occultation was determined from the standard deviation of the (very nearly Gaussian) marginal- ized posterior distribution. The new light curves are shown in Figure 2, and the times are given in Table 1. The best-fit cen- tral times are relatively insensitive to the version of the pho- tometry adopted in the final solution. The very worst of the four photometry solutions for the 3.6 and 4.5 µm data gave midpoints differing by 31 and 75 seconds (0.3σ and 0.6σ), respectively. 4. TIMING ANALYSIS Table 1 gathers together all of the times of transits (tt) and occultations (to) used in our analysis. We included all of the data we could find in the literature for which (i) the analysis was based on observations of a single event, (ii) the midpoint was allowed to be a completely free parameter, and (iii) the time system is documented clearly. The tabulated occultation times have not been corrected for the light-travel time across the diameter of the orbit. For the timing analysis described below, the occultation times were corrected by subtracting 2a/c = 22.9 s. We fitted three models to the timing data using the MCMC method. The first model assumes a circular orbit and a con- stant orbital period: ttra(E) = t0 + PE, tocc(E) = t0 + P 2 + PE, (2) (3) where E is the epoch number. Figure 3 displays the residuals with respect to this model. The fit is poor, with χ2 min = 197.6 and 111 degrees of freedom. The transit residuals follow a negative parabolic trend, indicating a negative period deriva- tive. Our new data-the square points at the rightmost ex- treme of the plot-follow the trend that had been established by Maciejewski et al. (2016). Thus, we confirm the finding of Maciejewski et al. (2016) that the transit interval is slowly shrinking. Next we fitted a model that assumes a circular orbit and a constant period derivative: ttra(E) = t0 + PE + 1 2 E2, dP dE 1 2 dP dE E2. tocc(E) = t0 + P 2 + PE + (4) (5) The red curves in Figure 3 shows the best fit, which has χ2 min = 118.5 and 110 degrees of freedom. Both the transit and occultation data are compatible with the model. The im- plied period derivative is dP dP = 1 P = −(9.3 ± 1.1) × 10−10 = −29 ± 3 ms yr−1. (6) dE dt In the third model, the orbit is slightly eccentric and under- going apsidal precession: ttra(E) = t0 + PsE − ePa π cos ω, tocc(E) = t0 + Pa 2 + PsE + ePa π cosω, (7) (8) where e is the eccentricity, ω is the argument of pericenter, Pa is the anomalistic period and Ps is the sidereal period. The argument of pericenter advances uniformly in time, ω(E) = ω0 + dω dE E, and the two periods are related by Ps = Pa(cid:18)1 − dω/dE 2π (cid:19) . (9) (10) are expressions based of These Giménez & Bastero (1995), in the limit of low eccen- tricity and high inclination. This model has 5 parameters: t0, Ps, e, ω0, and dω/dE. on Eqn. (15) The blue curves in Figure 3 show the best-fit precession model. The main difference between the decay and preces- sion models is that apsidal precession produces anticorrelated transit and occultation timing deviations, while the orbital decay model produces deviations of the same sign. The pre- cession fit has χ2 min = 124.0 and 108 degrees of freedom. The model achieves a reasonable fit by adjusting the precession period to be longer than the observing interval. In this way, the parabolic trend can be matched by the downward-curving portion of a sinusoidal function. However, there is tension between the need for enough downward curvature in the tran- sit deviations to fit the earliest data and a small enough up- ward curvature in the occultation deviations to fit the most recent data. The orbital decay model provides the best fit. It is better than the precession model by ∆χ2 = 5.5, despite the handi- cap of having two fewer free parameters. The Akaike Infor- mation Criterion (AIC) and Bayesian Information Criterion (BIC) are widely used statistics to choose the most parsimo- nious model that fits the data: 3 The Allan deviation relation expresses how the standard deviation of the binned residuals varies with bin size. For ideal white noise, it should decrease as the inverse square root of the bin size. α = AIC = χ2 + 2k, β = BIC = χ2 + k log n, (11) (12) APPARENTLY DECAYING ORBIT OF WASP-12B 5 TRANSITS decay OCCULTATIONS precession ] i n m [ d o i r e p t n a t s n o c m o r f n o i t i a v e D 2 0 −2 −4 2 0 −2 −4 −1500 −1000 −500 0 Epoch 500 1000 1500 Figure 3. Timing residuals for WASP-12. Each data point is the difference between an observed eclipse time and the prediction of the best-fit constant-period model. The top panel shows transit data and the bottom panel shows occultation data. Circles are previously reported data, and squares are new data. The blue curves show the best-fit precession model, for which transit and occultation deviations are anticorrelated. The red curves show the best-fit orbital decay model, in which the transit and occultation deviations are the same. where n is the number of data points and k is the num- ber of free parameters. In this case, n = 113, k = 3 for de- cay, and k = 5 for precession. The AIC favors the decay model by ∆α = 9.46, corresponding to a likelihood ratio of exp(∆α/2) = 113. The BIC favors the orbital decay model by ∆β = 14.91, corresponding to an approximate Bayes factor of exp(∆β/2) = 1730. Table 2 gives the best-fit parameters for all three models. In summary, a constant period has been firmly ruled out, and orbital decay is statistically favored over apsidal precession as the best explanation for the timing data. However, the statistical significance of the preference for orbital decay is modest and depends on the reliability of the quoted uncer- tainties for all of the timing data, which come from different investigators using different methods. For example, when the earliest data point is omitted, orbital decay is still preferred but ∆χ2 is reduced to 2.0. For these reasons, and out of gen- eral caution, we do not regard apsidal precession as being definitively ruled out. Further observations are needed. 5. IMPLICATIONS 5.1. Orbital decay To explore the implications of the best-fit models, we as- sume, for the moment, that the orbital decay interpretation is correct. Based on the current decay rate, the period would shrink to zero in P dP/dt = 3.2 Myr. (13) The future lifetime of the planet is likely to be even shorter, because the decay rate is expected to increase rapidly with decreasing period. In the simplified "constant phase lag" model for tidal evo- lution, the period derivative is dP dt = − 27π 2Q⋆ (cid:18) Mp a (cid:19)5 M⋆(cid:19)(cid:18) R⋆ , (14) which we obtained by applying Kepler's third law to Eqn. (20) of Goldreich & Soter (1966). Here, Q⋆ is the "modified quality factor" of the star's tidal oscillations (of- ten designated elsewhere as Q′ ⋆). For the case of WASP- 12, Mp/M⋆ = 9.9 × 10−4 and a/R⋆ = 3.097 (Chan et al. 2011), giving Q⋆ ≈ 2 × 105. (15) This value for Q⋆ is smaller than the typical range of 106−7 that has been inferred through ensemble analyses of binary stars and star-planet systems (see, e.g. Meibom & Mathieu 2005; Hansen 2010; Penev et al. 2012). One exception is Jackson et al. (2008), who found Q⋆ ∼ 105.5 based on the period-eccentricity distribution of hot Jupiters. This is con- sistent with our result. Theoretically, the quality factor should depend on the or- bital period, perturbation strength, and internal structure of the star (Ogilvie 2014). Recently, Essick & Weinberg (2016) calculated Q⋆ for hot Jupiters perturbing solar-type stars, based on the nonlinear interactions and dissipation of tidally driven g-modes. For the mass ratio and period of WASP-12, their Eqn. (26) predicts Q⋆ = 4 × 105, close to the observed 6 PATRA ET AL. value. However, their calculation pertained to stars with a radiative core and a convective envelope, and it is not clear that WASP-12 belongs in this category. With Teff = 6100 K (Torres et al. 2012), WASP-12 is right on the borderline be- tween stars with convective and radiative envelopes. In fact, we wonder if this coincidence-lying right on the Kraft break-could be related to the apparently rapid dissipation rate. The star may have a convective core and a convective envelope, separated by a radiative zone, perhaps leading to novel mechanisms for wave dissipation. 5.2. Apsidal precession Assuming instead that the apsidal precession model is cor- rect, the orbital eccentricity is 0.0021 ± 0.0005. This is com- patible with the upper limit of 0.05 from observations of the spectroscopic orbit (Husnoo et al. 2012). The observed pre- cession rate is ω = 26 ± 3 deg yr−1, corresponding to a pre- cession period of 14 ± 2 years. Ragozzine & Wolf (2009) showed that for systems resem- bling WASP-12, the largest contribution to the theoretical ap- sidal precession rate is from the planet's tidal deformation. The rate is proportional to the planet's Love number kp, a di- mensionless measure of the degree of central concentration of the planet's density distribution. Lower values of kp cor- respond to more centrally concentrated distributions, which are closer to the point-mass approximation and, therefore, produce slower precession. Eqn. (14) of Ragozzine & Wolf (2009) can be rewritten for this case as dω dE = 15πkp(cid:18) M⋆ a (cid:19)5 Mp(cid:19)(cid:18) Rp . (16) Using the measured precession rate and relevant parameters of WASP-12, this equation gives kp = 0.44 ± 0.10. If this in- terpretation is confirmed, it would be a unique constraint on an exoplanet's interior structure, in addition to the usual mea- surements of mass, radius, and mean density. For Jupiter, a value of kp = 0.59 has been inferred from its observed gravity moments (Wahl et al. 2016). Therefore the precession inter- pretation for WASP-12b suggests that its density distribution has a similar degree of central concentration as Jupiter, and perhaps somewhat higher. 5.3. Prior probabilities It is worth contemplating the "prior probability" of each model. By this, we mean the chance that the circumstances required by each model would actually occur, independently of the goodness-of-fit to the data. At face value, both models imply that we are observing WASP-12 at a special time, in vi- olation of the "temporal Copernican principle" articulated by Gott (1993). It is difficult, however, to decide which model requires the greater coincidence. Given the star's main-sequence age of 1700 ± 800 Myr (Chan et al. 2011), the orbital decay model would have us believe we are witnessing the last ∼0.2% of the planet's life. If we observed a single system at a random time, this would require a one-in-500 coincidence. However, WASP-12 is not the only hot Jupiter that we and others have been monitoring. There are about 10 other good candidates with comparably low a/R⋆, increasing the odds of the coincidence by an order of magnitude. It is noteworthy that other investigators have argued on independent grounds that WASP-12b is close to death. Fossati et al. (2010), Haswell et al. (2012), and Nichols et al. (2015) have presented near-ultraviolet transit spectroscopy consistent with an extended and escaping exosphere. The resulting mass loss process has been studied theoretically by Li et al. (2010), Lai et al. (2010), and Bisikalo et al. (2013). Most recently, Jackson et al. (2017) developed a new theory for Roche lobe overflow and identified WASP-12 as a likely case of rapid mass loss. It is also possible that orbital decay occurs in fits and starts, because of strong and erratic variations in the dissipation rate with the forcing period (see, e.g., Ogilvie & Lin 2007; Barker & Ogilvie 2010). Thus, the planet may be experienc- ing a brief interval of rapid decay. This does not eliminate the requirement for a coincidence, because one would expect to discover the system in one of the more prolonged states of slow dissipation. However, it does mean that the planet's future lifetime may be longer than the current value of P/ P. As for apsidal precession, the trouble is the very short ex- pected timescale for tidal orbital circularization. This process is thought to be dominated by dissipation within the planet, rather than the star. Eqn. (25) of Goldreich & Soter (1966), relevant to this case, can be rewritten τe = e de/dt = 2Qp 63π (cid:18) Mp M⋆(cid:19)(cid:18) a Rp(cid:19)5 Porb. (17) For WASP-12, this gives τe ∼ 0.5 Myr, assuming Qp ∼ 106. At this rate, even 4 Myr of tidal evolution would reduce the eccentricity below 10−3. Of course, the planetary quality fac- tor Qp could be larger than the standard value of 106, or the tidal model leading to the preceding equation could be a gross misrepresentation of the actual circularization process. There may also be some process that continually excites the eccentricity. One possibility is gravitational forcing by another planet, although no other nearby planets are known in the WASP-12 system (Knutson et al. 2014). An intriguing possibility is eccentricity excitation by the gravitational per- turbations from the star's convective eddies. In this scenario, proposed by Phinney (1992) to explain the small but nonzero eccentricities of pulsars orbiting white dwarfs, the system reaches a state of equipartition between the energy of eccen- tricity oscillations (epicyclic motion) and the kinetic energy of turbulent convection. To our knowledge, this theory has only been developed for post-main-sequence stars (see, e.g. Verbunt & Phinney 1995; Rafikov 2016). It is not obvious that this theory would apply to WASP-12 and be compatible with e ∼ 10−3. Should further theoretical investigations reveal that this mechanism (or any other) could naturally maintain the orbital eccentricity at the level of 10−3, then the apsidal precession model would require no special coincidence. Neither would it require unique circumstances; it is possible that eccentrici- ties of this order could exist in other hot Jupiter systems and have remained undetected. Thus, the identification of a natu- ral eccentricity-excitation mechanism would swing the prior- APPARENTLY DECAYING ORBIT OF WASP-12B 7 we have not even discussed. The star's equator is likely to be misaligned with the orbital plane (Schlaufman 2010; Albrecht et al. 2012). The star is also part of a hierarchi- cal three-body system, with a tight pair of M dwarfs orbit- ing the planet-hosting star at a distance of about 265 AU (Bechter et al. 2014). Detailed modeling of the star's interior structure and and tidal evolution is warranted, as are contin- ued observations of transits and occultations. We are very grateful to Allyson Bieryla, David Latham, and Emilio Falco for their assistance with the FLWO ob- servations. We thank Nevin Weinberg, Jeremy Goodman, Kaloyen Penev, David Oort Alonso, Heather Knutson, Dong Lai, and the CfA Exoplanet Pizza group for helpful discus- sions. We also appreciate the anonymous reviewer's prompt and careful report. Work by K.C.P. was supported by the MIT Undergraduate Research Opportunities Program and the Paul E. Gray Fund. Note added in proof. D. Lai has reminded us of another possible reason for a cyclic variation in the period: the Applegate (1992) effect, in which a star's quadrupole moment varies over a magnetic activity cycle. For WASP-12, Watson & Marsh (2010) estimated that this effect could produce timing deviations of 4-40 s depending on the cycle duration. The transit and occultation deviations would have the same sign, allowing this effect to be distinguished from apsidal precession. probability balance in the direction of apsidal precession. 5.4. Other possible explanations To this point, we have presented orbital decay and apsi- dal precession as the only possible reasons for an apparent period decrease. Another possibility is that the star is accel- erating toward Earth, due to the force from stellar compan- ions or wide-orbiting planets. This would produce a neg- ative apparent period derivative of vrP/c, where vr is the radial velocity. Based on long-term Doppler monitoring, Knutson et al. (2014) placed an upper limit for WASP-12 of vr < 0.019 m s−1 day−1 (2σ), which, in turn, limits the ap- parent P to be smaller than 7 × 10−11. This is an order of magnitude too small to be responsible for the observed period derivative. Of course, none of these phenomena are mutually exclusive. The system may be experiencing a combination of orbital decay, apsidal precession, and radial acceleration, al- though joint modeling of these effects is not productive with the current data. Rafikov (2009) described two other phenomena that cause changes in the apparent period of a transiting planet. The first is the Shklovskii effect, wherein the star's proper mo- tion leads to a changing radial velocity and a nonzero sec- ond derivative of the light-travel time. This is already ruled out by Doppler observations of the radial acceleration. For completeness, though, we note that the observed distance d and proper motion µ imply a period derivative of Pµ2d/c ∼ 6 × 10−15, too small to explain the data. The second phe- nomenon, also dependent on proper motion, is the appar- ent apsidal precession caused by our changing viewing an- gle. The resulting period derivative is of order ∼(Pµ)2/2π, which in this case is ∼10−21, too small by many orders of magnitude. 6. FUTURE PROSPECTS With WASP-12, we are fortunate that both possibilities- orbital decay and apsidal precession-lead to interesting out- comes. It will soon be possible to measure the tidal dissipa- tion rate of a star, or the tidal deformability of an exoplanet, either of which would be a unique achievement. To help un- derstand the requirements for a definitive verdict, Figure 4 shows the future projections of a sample of 100 models that provide satisfactory fits to the data, drawn randomly from our converged Markov chains. For the transits, the two families of models become sepa- rated by a few minutes by 2021-22. The occultation models diverge earlier, and are separated by a few minutes in 2019- 20. Thus, while continued transit timing is important, the most rapid resolution would probably come from observing occultations a few years from now. In principle, transit dura- tion variations (TDV) would also help to distinguish between the two models, but the expected amplitude is (Pál & Kocsis 2008) TDV ∼ P 2π (cid:18) R⋆ a (cid:19) ecos ω ∼ 10 sec, (18) which will be difficult to detect. In this paper, we have focused on the timing anomalies of WASP-12. This system has other remarkable features 8 PATRA ET AL. ] i n m [ d o i r e p t n a t s n o c m o r f n o i t i a v e D 0 −5 −10 −15 0 −5 −10 −15 TRANSITS precession decay OCCULTATIONS 2010 2015 Year 2020 2025 Figure 4. Possible futures for WASP-12. For each of the two models, we randomly drew 100 parameter sets from our Markov chains. Shown here are the extrapolations of those models to future times. APPARENTLY DECAYING ORBIT OF WASP-12B 9 Table 1. Transit and occultation times. Type of event Midpoint (BJDTDB) tra occ occ tra tra tra tra tra tra occ occ tra tra tra tra tra tra tra tra occ tra tra tra tra occ occ tra tra tra tra tra tra tra tra tra tra tra occ tra tra tra tra occ tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra 2454515.52496 2454769.28131 2454773.64751 2454836.40340 2454840.76893 2455140.90981 2455147.45861 2455163.83061 2455172.56138 2455194.93381 2455202.57566 2455209.66895 2455210.76151 2455230.40669 2455254.41871 2455494.52999 2455498.89590 2455509.80971 2455510.90218 2455517.99455 2455518.54070 2455542.55210 2455542.55273 2455566.56385 2455576.93141 2455587.84671 2455590.57561 2455598.21552 2455600.39800 2455601.49010 2455603.67261 2455623.31829 2455876.52786 2455887.44198 2455888.53340 2455890.71635 2455903.81357 2455910.90841 2455920.18422 2455923.45850 2455924.00411 2455946.37823 2455946.92231 2455947.47015 2455948.56112 2455951.83534 2455952.92720 2455959.47543 2455960.56686 2455970.38941 2455971.48111 2455982.39509 2455983.48695 2455984.57797 2455985.66975 2455996.58378 2456005.31533 2456006.40637 2456245.42729 2456249.79404 2456273.80514 2456282.53584 2456284.71857 2456297.81605 2456302.18179 2456305.45536 2456319.64424 2456328.37556 2456329.46733 2456604.50489 2456605.59624 2456606.68760 2456607.77938 2456629.60726 2456630.69917 Uncertainty (days) 0.00043 0.00080 0.00060 0.00028 0.00062 0.00042 0.00043 0.00032 0.00036 0.00100 0.00220 0.00046 0.00041 0.00019 0.00043 0.00072 0.00079 0.00037 0.00031 0.00118 0.00040 0.00040 0.00028 0.00028 0.00090 0.00170 0.00068 0.00035 0.00029 0.00024 0.00029 0.00039 0.00027 0.00021 0.00027 0.00024 0.00032 0.00130 0.00031 0.00022 0.00210 0.00018 0.00180 0.00017 0.00033 0.00011 0.00010 0.00017 0.00032 0.00039 0.00035 0.00034 0.00035 0.00032 0.00042 0.00037 0.00037 0.00031 0.00033 0.00039 0.00030 0.00030 0.00030 0.00030 0.00046 0.00024 0.00038 0.00027 0.00029 0.00021 0.00030 0.00033 0.00071 0.00019 0.00043 Epoch number -1640 -1408 -1404 -1346 -1342 -1067 -1061 -1046 -1038 -1018 -1011 -1004 -1003 -985 -963 -743 -739 -729 -728 -722 -721 -699 -699 -677 -668 -658 -655 -648 -646 -645 -643 -625 -393 -383 -382 -380 -368 -362 -353 -350 -350 -329 -329 -328 -327 -324 -323 -317 -316 -307 -306 -296 -295 -294 -293 -283 -275 -274 -55 -51 -29 -21 -19 -7 -3 0 13 21 22 274 275 276 277 297 298 Hebb et al. (2009)a Campo et al. (2011) Campo et al. (2011) Copperwheat et al. (2013) Chan et al. (2011) Collins et al. (2017) Maciejewski et al. (2013) Collins et al. (2017) Chan et al. (2011) Croll et al. (2015) Föhring et al. (2013) Collins et al. (2017) Collins et al. (2017) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Sada et al. (2012) Collins et al. (2017) Collins et al. (2017) Deming et al. (2015)b Cowan et al. (2012) Cowan et al. (2012) Maciejewski et al. (2013) Maciejewski et al. (2013) Croll et al. (2015) Croll et al. (2015) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Collins et al. (2017) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Collins et al. (2017) Crossfield et al. (2012) Maciejewski et al. (2013) Maciejewski et al. (2013) Croll et al. (2015) Maciejewski et al. (2013) Croll et al. (2015) Maciejewski et al. (2013) Maciejewski et al. (2013) Stevenson et al. (2014) Stevenson et al. (2014) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2013) Collins et al. (2017) Collins et al. (2017) Collins et al. (2017) Maciejewski et al. (2013) Maciejewski et al. (2013) Maciejewski et al. (2016) Collins et al. (2017) Collins et al. (2017) Maciejewski et al. (2016) Collins et al. (2017) Collins et al. (2017) Maciejewski et al. (2016) Maciejewski et al. (2016) Collins et al. (2017) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Collins et al. (2017) Maciejewski et al. (2016) Maciejewski et al. (2016) Table 1 continued Type of event Midpoint (BJDTDB) Table 1 (continued) Epoch Uncertainty number (days) occ occ tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra tra 2456638.88530 2456642.15848 2456654.71047 2456659.07598 2456662.35014 2456663.44136 2456664.53256 2456674.35560 2456677.63039 2456688.54384 2456694.00161 2456703.82417 2456711.46415 2456719.10428 2456721.28692 2456722.37807 2456986.50195 2457010.51298 2457012.69617 2457045.43831 2457046.53019 2457059.62713 2457060.71839 2457067.26715 2457068.35834 2457103.28423 2457345.57867 2457390.32708 2457391.41818 2457426.34324 2457427.43496 2457671.91324 2457706.83791 2457765.77515 2457766.86633 2457776.68869 2457788.69464 2457800.69978 0.00110 0.00141 0.00034 0.00034 0.00019 0.00019 0.00031 0.00028 0.00032 0.00040 0.00029 0.00029 0.00025 0.00034 0.00034 0.00046 0.00043 0.00039 0.00049 0.00046 0.00049 0.00035 0.00036 0.00022 0.00020 0.00031 0.00042 0.00033 0.00033 0.00055 0.00023 0.00035 0.00037 0.00028 0.00039 0.00029 0.00048 0.00032 305 308 320 324 327 328 329 338 341 351 356 365 372 379 381 382 624 646 648 678 679 691 692 698 699 731 953 994 995 1027 1028 1252 1284 1338 1339 1348 1359 1370 this work this work Collins et al. (2017) Kreidberg et al. (2015) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Kreidberg et al. (2015) Collins et al. (2017) Maciejewski et al. (2016) Kreidberg et al. (2015) Kreidberg et al. (2015) Maciejewski et al. (2016) Kreidberg et al. (2015) Kreidberg et al. (2015) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Collins et al. (2017) Maciejewski et al. (2016) Maciejewski et al. (2016) Collins et al. (2017) Collins et al. (2017) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) Maciejewski et al. (2016) this work this work this work this work this work this work this work a Refers to the light curve obtained by Hebb et al. (2009) with the 2m Liverpool tele- scope, as analyzed by Maciejewski et al. (2013). b Re-analyzed in this work. Table 2. Best-fit model parameters. Parameter Value (Unc.)a Constant period Reference epoch, t0 [BJDTBD] Period, P [days] Orbital decay 2456305.455609(28) 1.091420025(47) Reference epoch, t0 [BJDTBD] Period at reference epoch, P [days] dP/dE [days] 2456305.455790(35) 1.091420078(47) −1.02(11) × 10−9 Apsidal precession Reference epoch, t0 [BJDTBD] Sidereal period, Psid [days] Eccentricity, e A.O.P. at reference epoch, ω0 [rad] Precession rate, dω/dE [rad epoch−1] 2456305.45509(15) 1.09141993(15) 0.00208(47) 2.92(19) 0.00133(18) a The numbers in parenthesis give the 1σ uncertainty in the final two digits. 10 PATRA ET AL. REFERENCES Albrecht, S., Winn, J. N., Johnson, J. A., et al. 2012, ApJ, 757, 18 Allan, D. W. 1966, IEEE Proceedings, 54 Applegate, J. H. 1992, ApJ, 385, 621 Barker, A. J., & Ogilvie, G. I. 2010, MNRAS, 404, 1849 Bechter, E. B., Crepp, J. R., Ngo, H., et al. 2014, ApJ, 788, 2 Bisikalo, D., Kaygorodov, P., Ionov, D., et al. 2013, ApJ, 764, 19 Campo, C. J., Harrington, J., Hardy, R. A., et al. 2011, ApJ, 727, 125 Chan, T., Ingemyr, M., Winn, J. N., et al. 2011, AJ, 141, 179 Claret, A., & Bloemen, S. 2011, A&A, 529, A75 Collins, K. A., Kielkopf, J. F., & Stassun, K. G. 2017, AJ, 153, 78 Copperwheat, C. M., Wheatley, P. J., Southworth, J., et al. 2013, MNRAS, 434, 661 Cowan, N. B., Machalek, P., Croll, B., et al. 2012, ApJ, 747, 82 Croll, B., Albert, L., Jayawardhana, R., et al. 2015, ApJ, 802, 28 Crossfield, I. J. M., Barman, T., Hansen, B. M. S., Tanaka, I., & Kodama, T. 2012, ApJ, 760, 140 Deming, D., Knutson, H., Kammer, J., et al. 2015, ApJ, 805, 132 Eastman, J., Gaudi, B. S., & Agol, E. 2013, PASP, 125, 83 Eastman, J., Siverd, R., & Gaudi, B. S. 2010, PASP, 122, 935 Essick, R., & Weinberg, N. 2016, in APS April Meeting Abstracts Föhring, D., Dhillon, V. S., Madhusudhan, N., et al. 2013, MNRAS, 435, 2268 Foreman-Mackey, D., Hogg, D. W., Lang, D., & Goodman, J. 2013, PASP, 125, 306 Fossati, L., Bagnulo, S., Elmasli, A., et al. 2010, ApJ, 720, 872 Giménez, A., & Bastero, M. 1995, Ap&SS, 226, 99 Goldreich, P., & Soter, S. 1966, Icarus, 5, 375 Gott, III, J. R. 1993, Nature, 363, 315 Hansen, B. M. S. 2010, ApJ, 723, 285 Haswell, C. A., Fossati, L., Ayres, T., et al. 2012, ApJ, 760, 79 Hebb, L., Collier-Cameron, A., Loeillet, B., et al. 2009, ApJ, 693, 1920 Heyl, J. S., & Gladman, B. J. 2007, MNRAS, 377, 1511 Hoyer, S., Pallé, E., Dragomir, D., & Murgas, F. 2016, AJ, 151, 137 Husnoo, N., Pont, F., Mazeh, T., et al. 2012, MNRAS, 422, 3151 Hut, P. 1980, A&A, 92, 167 Jackson, B., Arras, P., Penev, K., Peacock, S., & Marchant, P. 2017, ApJ, 835, 145 Jackson, B., Greenberg, R., & Barnes, R. 2008, ApJ, 678, 1396 Jordán, A., & Bakos, G. Á. 2008, ApJ, 685, 543 Knutson, H. A., Fulton, B. J., Montet, B. T., et al. 2014, ApJ, 785, 126 Kreidberg, L., Line, M. R., Bean, J. L., et al. 2015, ApJ, 814, 66 Lai, D., Helling, C., & van den Heuvel, E. P. J. 2010, ApJ, 721, 923 Levrard, B., Winisdoerffer, C., & Chabrier, G. 2009, ApJL, 692, L9 Lewis, N. K., Knutson, H. A., Showman, A. P., et al. 2013, ApJ, 766, 95 Li, S.-L., Miller, N., Lin, D. N. C., & Fortney, J. J. 2010, Nature, 463, 1054 Luger, R., Agol, E., Kruse, E., et al. 2016, AJ, 152, 100 Maciejewski, G., Dimitrov, D., Seeliger, M., et al. 2013, A&A, 551, A108 Maciejewski, G., Dimitrov, D., Fernández, M., et al. 2016, A&A, 588, L6 Mandel, K., & Agol, E. 2002, ApJL, 580, L171 Matsakos, T., & Königl, A. 2015, ApJL, 809, L20 Mayor, M., & Queloz, D. 1995, Nature, 378, 355 Meibom, S., & Mathieu, R. D. 2005, ApJ, 620, 970 Miralda-Escudé, J. 2002, ApJ, 564, 1019 Nichols, J. D., Wynn, G. A., Goad, M., et al. 2015, ApJ, 803, 9 Ogilvie, G. I. 2014, ARA&A, 52, 171 Ogilvie, G. I., & Lin, D. N. C. 2007, ApJ, 661, 1180 Pál, A., & Kocsis, B. 2008, MNRAS, 389, 191 Penev, K., Jackson, B., Spada, F., & Thom, N. 2012, ApJ, 751, 96 Penev, K., Hartman, J. D., Bakos, G. Á., et al. 2016, AJ, 152, 127 Phinney, E. S. 1992, Philosophical Transactions of the Royal Society of London Series A, 341, 39 Rafikov, R. R. 2009, ApJ, 700, 965 -. 2016, ApJ, 830, 8 Ragozzine, D., & Wolf, A. S. 2009, ApJ, 698, 1778 Rasio, F. A., Tout, C. A., Lubow, S. H., & Livio, M. 1996, ApJ, 470, 1187 Sada, P. V., Deming, D., Jennings, D. E., et al. 2012, PASP, 124, 212 Sasselov, D. D. 2003, ApJ, 596, 1327 Schlaufman, K. C. 2010, ApJ, 719, 602 Schlaufman, K. C., & Winn, J. N. 2013, ApJ, 772, 143 Stevenson, K. B., Bean, J. L., Seifahrt, A., et al. 2014, AJ, 147, 161 Teitler, S., & Königl, A. 2014, ApJ, 786, 139 Torres, G., Fischer, D. A., Sozzetti, A., et al. 2012, ApJ, 757, 161 Verbunt, F., & Phinney, E. S. 1995, A&A, 296, 709 Villaver, E., & Livio, M. 2009, ApJL, 705, L81 Wahl, S. M., Hubbard, W. B., & Militzer, B. 2016, ApJ, 831, 14 Watson, C. A., & Marsh, T. R. 2010, MNRAS, 405, 2037 Wilkins, A. N., Delrez, L., Barker, A. J., et al. 2017, ApJL, 836, L24 Winn, J. N. 2010, Exoplanet Transits and Occultations, ed. S. Seager (University of Arizona Press), 55
1809.07898	1	1809	2018-09-21T00:35:00	MOA-2016-BLG-319Lb: Microlensing Planet Subject to Rare Minor-Image Perturbation Degeneracy in Determining Planet Parameter	[ "astro-ph.EP" ]	We present the analysis of the planetary microlensing event MOA-2016-BLG-319. The event light curve is characterized by a brief ($\sim 3$ days) anomaly near the peak produced by minor-image perturbations. From modeling, we find two distinct solutions that describe the observed light curve almost equally well. From the investigation of the lens-system configurations, we find that the confusion in the lensing solution is caused by the degeneracy between the two solutions resulting from the source passages on different sides of the planetary caustic. These degeneracies can be severe for major-image perturbations but it is known that they are considerably less severe for minor-image perturbations. From the comparison of the lens-system configuration with those of two previously discovered planetary events, for which similar degeneracies were reported, we find that the degeneracies are caused by the special source trajectories that passed the star-planet axes at approximately right angles. By conducting a Bayesian analysis, it is estimated that the lens is a planetary system in which a giant planet with a mass $M_{\rm p}=0.62^{+1.16}_{-0.33}~M_{\rm J}$ ($0.65^{+1.21}_{-0.35}~M_{\rm J}$) is orbiting a low-mass M-dwarf host with a mass $M_{\rm h}=0.15^{+0.28}_{-0.08}~M_\odot$. Here the planet masses in and out of the parentheses represent the masses for the individual degenerate solutions. The projected host-planet separations are $a_\perp\sim 0.95$ au and $\sim 1.05$ au for the two solutions. The identified degeneracy indicates the need to check similar degeneracies in future analyses of planetary lensing events with minor-image perturbations.	astro-ph.EP	astro-ph	DRAFT VERSION SEPTEMBER 24, 2018 Preprint typeset using LATEX style emulateapj v. 12/16/11 MOA-2016-BLG-319Lb: MICROLENSING PLANET SUBJECT TO RARE MINOR-IMAGE PERTURBATION DEGENERACY IN DETERMINING PLANET PARAMETERS CHEONGHO HANA01, IAN A. BONDA02,102 , ANDREW GOULDA03,A04,A05,101 , MICHAEL D. ALBROWA06, SUN-JU CHUNGA03,A07 , YOUN KIL JUNGA03, KYU-HA HWANGA03, CHUNG-UK LEEA03, YOON-HYUN RYUA03, IN-GU SHINA08, YOSSI SHVARTZVALDA09, JENNIFER C. YEEA08, SANG-MOK CHAA03,A10 , DONG-JIN KIMA03, HYOUN-WOO KIMA03, SEUNG-LEE KIMA03,A07 , DONG-JOO LEEA03, YONGSEOK LEEA03,A10, BYEONG-GON PARKA03,A07 , AND RICHARD W. POGGEA04, CHUN-HWEY KIMA11 (THE KMTNET COLLABORATION), FUMIO ABEA12, RICHARD BARRYA13, DAVID P. BENNETTA13,A14, APARNA BHATTACHARYAA13,A14, MARTIN DONACHIEA15, AKIHIKO FUKUIA16, YUKI HIRAOA17, YOSHITAKA ITOWA12, KOHEI KAWASAKIA17, IONA KONDOA17, NAOKI KOSHIMOTOA18,A19, MAN CHEUNG ALEX LIA15, YUTAKA MATSUBARAA12, YASUSHI MURAKIA12, SHOTA MIYAZAKIA17, MASAYUKI NAGAKANEA17, CLÉMENT RANCA13, NICHOLAS J. RATTENBURYA15, HARUNO SUEMATSUA17, DENIS J. SULLIVANA20, TAKAHIRO SUMIA17, DAISUKE SUZUKIA21, PAUL J. TRISTRAMA22, ATSUNORI YONEHARAA23 (THE MOA COLLABORATION), Draft version September 24, 2018 ABSTRACT We present the analysis of the planetary microlensing event MOA-2016-BLG-319. The event light curve is characterized by a brief (∼ 3 days) anomaly near the peak produced by minor-image perturbations. From modeling, we find two distinct solutions that describe the observed light curve almost equally well. From the investigation of the lens-system configurations, we find that the confusion in the lensing solution is caused by the degeneracy between the two solutions resulting from the source passages on different sides of the planetary caustic. These degeneracies can be severe for major-image perturbations but it is known that they are consid- erably less severe for minor-image perturbations. From the comparison of the lens-system configuration with those of two previously discovered planetary events, for which similar degeneracies were reported, we find that the degeneracies are caused by the special source trajectories that passed the star-planet axes at approxi- mately right angles. By conducting a Bayesian analysis, it is estimated that the lens is a planetary system in which a giant planet with a mass Mp = 0.62+1.16 −0.35 MJ) is orbiting a low-mass M-dwarf host with a mass Mh = 0.15+0.28 −0.08 M⊙. Here the planet masses in and out of the parentheses represent the masses for the individual degenerate solutions. The projected host-planet separations are a⊥ ∼ 0.95 au and ∼ 1.05 au for the two solutions. The identified degeneracy indicates the need to check similar degeneracies in future analyses of planetary lensing events with minor-image perturbations. Subject headings: gravitational lensing: micro -- planetary systems −0.33 MJ (0.65+1.21 [email protected] A01 Department of Physics, Chungbuk National University, Cheongju A02 Institute of Natural and Mathematical Sciences, Massey University, A03 Korea Astronomy and Space Science Institute, Daejon 34055, Re- MD 20742, USA A04 Department of Astronomy, Ohio State University, 140 W. 18th Ave., Auckland, New Zealand 8 1 0 2 p e S 1 2 . ] P E h p - o r t s a [ 1 v 8 9 8 7 0 . 9 0 8 1 : v i X r a 28644, Republic of Korea Auckland 0745, New Zealand public of Korea Columbus, OH 43210, USA delberg, Germany A05 Max Planck Institute for Astronomy, Königstuhl 17, D-69117 Hei- A06 University of Canterbury, Department of Physics and Astronomy, Private Bag 4800, Christchurch 8020, New Zealand A07 Korea University of Science and Technology, 217 Gajeong-ro, Yuseong-gu, Daejeon, 34113, Republic of Korea A08 Harvard-Smithsonian Center for Astrophysics, 60 Garden St., Cam- bridge, MA 02138, USA Pasadena, CA 91125, USA Kyeonggi 17104, Korea A10 School of Space Research, Kyung Hee University, Yongin, A11 Department of Astronomy & Space Science, Chungbuk National University, Cheongju 28644, Republic of Korea A12 Institute for Space-Earth Environmental Research, Nagoya Univer- sity, Nagoya 464-8601, Japan A13 Code 667, NASA Goddard Space Flight Center, Greenbelt, MD 20771, USA A14 Department of Astronomy, University of Maryland, College Park, A15 Department of Physics, University of Auckland, Private Bag 92019, A16 Okayama Astrophysical Observatory, National Astronomical Obser- vatory of Japan, 3037-5 Honjo, Kamogata, Asakuchi, Okayama 719-0232, Japan A17 Department of Earth and Space Science, Graduate School of Sci- ence, Osaka University, Toyonaka, Osaka 560-0043, Japan A18 Department of Astronomy, Graduate School of Science, The Uni- versity of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan A19 National Astronomical Observatory of Japan, 2-21-1 Osawa, Mi- taka, Tokyo 181-8588, Japan Wellington, New Zealand A21 Institute of Space and Astronautical Science, Japan Aerospace Ex- ploration Agency, 3-1-1 Yoshinodai, Chuo, Sagamihara, Kanagawa, 252- 5210, Japan A22 University of Canterbury Mt. John Observatory, P.O. Box 56, Lake A23 Department of Physics, Faculty of Science, Kyoto Sangyo Univer- Tekapo 8770, New Zealand sity, 603-8555 Kyoto, Japan 101 KMTNet Collaboration. 102 MOA Collaboration. A09 IPAC, Mail Code 100-22, Caltech, 1200 E. California Blvd., A20 School of Chemical and Physical Sciences, Victoria University, 2 Han et al. 1. INTRODUCTION using the relations Microlensing signals of planets are often described by the phrase "a brief anomaly" to the lensing light curve produced by the host of the planet. However, this phrase is oversimpli- fied because the pattern of planet-induced anomalies greatly varies depending on the configurations of lens systems. Fur- thermore, planetary signals in some cases of lens configura- tions can be confused with anomalies produced by other rea- sons and this induces a degeneracy problem in which multiple interpretations exist for an observed anomaly pattern. There- fore, identifying the origins of degeneracies in various lens configurations is important to find correct interpretations of microlensing planets by enabling one to check similar degen- eracies in future analyses. The degeneracies in the interpretation of planetary signals are broadly classified into two categories: "intrinsic" and "ac- cidental". The intrinsic degeneracies are caused by the sym- metry of the lens equation, which describes the mapping from the source position on the lens plane into the image position on the source plane. The most well known of these is the "close/wide" degeneracy in which a pair of planetary mod- els with projected separations from the host (normalized to the angular Einstein radius θE) s and s−1 result in very sim- ilar anomaly patterns. This degeneracy was first found by Griest & Safizadeh (1998) for a specific case of a planetary lens system and later extended to general binary lenses by Dominik (1999) and further investigated by An (2005). Accidental degeneracies, on the other hand, occur due to the fortuitous alignment of lensing magnification patterns aris- ing from unrelated lens configurations. It was pointed out by Gaudi (1998) that a subset of binary-source events, for which the flux ratio between the binary source stars is small and the lens approaches close to the faint source compan- ion, can produce short-term anomalies, which are similar to those of planet-induced anomalies. These planet/binary- source degeneracies were actually found for MOA-2012- BLG-486 (Hwang et al. 2013) and OGLE-2015-BLG-1459 (Hwang et al. 2018c) for which the degeneracies were diffi- cult to be resolved just based on the lensing light curves and could be resolved with additional data acquired from multi- band observations. Han & Gaudi (2008) pointed out another type of accidental degeneracies in which planetary signals can be imitated by those produced by binaries composed of roughly equal masses. Such degeneracies were demonstrated for OGLE-2011-BLG-0526, OGLE-2011-BLG-0950/MOA- 2011-BLG-336 (Choi et al. 2012) and OGLE-2015-BLG- 1212 (Bozza et al. 2016). In addition, incomplete cover- age of the planet-induced anomalies can cause degenera- cies in interpreting anomalies as demonstrated in the case of OGLE-2012-BLG-0455/MOA-2012-BLG-206 (Park et al. 2014; Hwang et al. 2018a). Gaudi & Gould (1997) (hereafter GG1997) predicted an- other type of accidental degeneracy, in which two planetary lens configurations had similar (s, q, α) but with the source passing on different sides of the caustic. Here q is the planet/host mass ratio and α represents the angle between the trajectory of the source and the axis connecting the planet and its host (source trajectory angle). Gould & Loeb (1992) ar- gued that (under the assumption that the source passed di- rectly over the caustic) one could read off the values of (s, α) from the three Paczy´nski (1986) parameters of the point-lens fit (t0, u0,tE) and the time of the planetary perturbation (tanom) s − 1 s and 0 + τ 2 =(cid:0)u2 1/2 ; anom(cid:1) τanom = tanom − t0 tE , (1) tan α = u0 τanom . (2) If the caustic is relatively small, this approach is approx- imately accurate, even if the source only passes near the caustic. However, GG1997 recognized that this would lead to two slightly different solutions depending on whether the source passes on one side of the caustic or the other. They pointed out that the degeneracy would be severe for perturba- tions produced by planets with projected planet-host separa- tions greater than the angular Einstein radius (s > 1, 'wide' planet): "major image perturbations". For "minor-image per- turbations", which are produced by 'close' planets with s < 1, on the other hand, it was thought that the degeneracy would be considerable less severe. This is mainly because of the qual- itative difference in the caustic structures between the wide and close planetary systems, in which a wide planet induces a single set of planetary caustics and a close planet induces two sets. In Appendix, we review basic facts about the types of planetary anomalies caused by major and minor-image per- turbations for readers who are not familiar with microlensing jargon. Experts will skip this appendix. For the major-image caustic, the magnification pattern on the near and far sides of the caustic are similar, and thus the anomalies produced by the source passing both sides of the caustic are similar to each other. For minor-image perturbations, on the other hand, the source trajectory passing the inner cusp will, in general, approach close to one of the two planetary caustics, while the source trajectory passing the outer cusp will approach the other caustic. As a result, the degeneracy would be generally resolvable from the presence (absence) and/or timing of the anomalies produced by the individual caustics. Degeneracies involved with planetary caustics have been demonstrated for actual lensing events and new types of de- generacies are additionally found with the increasing number of microlensing planets. An example of the GG1997 degener- acy was recently found for the planetary event OGLE-2017- BLG-0173 (Hwang et al. 2018d), for which there existed 3 degenerate solutions and among them 2 solutions were caused by the degeneracy predicted by GG1997. We note that the other solution results from a new discrete degeneracy between the solution in which the caustic is fully enveloped ("Cannae" solution) and the solution in which only one side of the caus- tic is enveloped ("von Schlieffen" solution): "Hollywood" de- generacy. The two solutions resulting from the Hollywood de- generacy have different mass ratios because the source passes through the caustic in different places. Skowron et al. (2018) found a more specific case of GG1997 degeneracies from the analysis of the planetary lensing event OGLE-2017-BLG- 0373. This so-called "caustic-chiral" degeneracy arises when the source passes over the caustic (contrary to the GG1997 degeneracy), but there are gaps in data. In this case, the solu- tions have very similar (s, α) but substantially different q. An- other example of the caustic-chiral degeneracy was found by Hwang et al. (2018a) for KMT-2016-BLG-0212. In addition, there is a case of the source passing through a major-image caustic and being degenerate with passing through a minor- image caustic. This degeneracy was identified for KMT-2016- BLG-1107 by Hwang et al. (2018b). MOA-2016-BLG-319Lb 3 TABLE 1 DATA SETS USED IN THE ANALYSIS Data set Range MOA KMTA (BLG02) KMTA (BLG42) KMTC (BLG02) KMTC (BLG42) KMTS (BLG02) KMTS (BLG42) 7502 . HJD′ . 7569 7440 . HJD′ . 7676 7443 . HJD′ . 7676 7439 . HJD′ . 7675 7439 . HJD′ . 7675 7441 . HJD′ . 7681 7441 . HJD′ . 7681 NOTE. -- HJD′ = HJD − 2450000. Ndata 757 798 701 1143 1006 1394 1472 Network (KMTNet) survey (Kim et al. 2016) conducted using three 1.6 m telescopes. The individual KMTNet telescopes are positioned at the Cerro Tololo Interamerican Observa- tory, Chile (KMTC), the South African Astronomical Obser- vatory, South Africa (KMTS), and the Siding Spring Obser- vatory, Australia (KMTA). The alert system of the KMTNet survey started from the 2018 season (Kim et al. 2018b) and the progress of the event was not known in real time at the time of the event during the 2016 season. From the analysis of lensing events identified by applying the Event Finder al- gorithm (Kim et al. 2018a,b) to the 2016 season data, it was found that KMTNet data densely covered the light curve peak which clearly showed a short-term anomaly. See the zoom of the light curve around the anomaly region presented in the upper panel of Figure 1. The event was identified by KMTNet as KMT-2016-BLG-1816. In Table 1, we present the data sets used in the analy- sis. MOA observations of the event was conducted in a cus- tomized R band with a cadence of 1 hr. KMTNet observa- tions were conducted mostly in I band with occasional ob- servations in V band to measure the source color. The event was in the KMTNet BLG02 and BLG42 fields which were monitored with a cadence of 0.5 hr by the individual tele- scopes. For the period from April 23 (HJD′ ∼ 2457501) to June 16 (HJD′ ∼ 2457555), the cadence of KMTS and KMTA was increased in order to support Kepler K2 C9 campaign (Gould & Horne 2013). While the event does not lie in the K2 field, the anomaly coverage serendipitously benefited from this cadence increase. The columns 'range' and 'Ndata' in- dicate the time range of the data sets used for analysis and the number of data points constituting the individual data sets, re- spectively. We set the range of the MOA data in the region around event because the baseline data exhibit considerable fluctuation. Photometry of data are processed using the codes of the individual groups: Bond et al. (2001) for the MOA survey and Albrow et al. (2009) for the KMTNet survey. Both codes utilize the difference imaging method developed by (Alard & Lupton 1998). Errorbars are normlized using the recipe explained in Yee et al. (2012). 3. LIGHT CURVE ANALYSIS The light curve shows a pronounced dip. Such a dip fea- ture in lensing light curves can only be produced by minor image perturbations. There are two possibilities in the lens- system configuration. First, minor image gives rise to two tri- angular planetary caustics with a magnification dip between them. Second, there is a six-sided resonant caustic whose "back end" consists of two caustic wings separated by a dip. See Figure 4 of Gaudi (2012) for the variation of planetary microlens caustics. FIG. 1. -- Light curve of the lensing event MOA-2016-BLG-319. The zoom of the anomaly near the peak is shown in the upper panel. Superposed on the data points represents the point-source point-lens model. In this work, we analyze the planetary event MOA-2016- BLG-319. The light curve is characterized by a short- term anomaly near the peak produced by the minor-image perturbation. Contrary to the expectation that interpreting minor-image perturbations would not suffer from degenera- cies, we find two discrete solutions that describe the ob- served light curve almost equally well. Similar degeneracies in minor-image perturbations were reported for two planetary events OGLE-2016-BLG-1067 (Calchi Novati et al. 2018b) and OGLE-2012-BLG-0950 (Koshimoto et al. 2017). By an- alyzing the similarity between the anomalies of the events, we investigate the origin of the degeneracy. 2. OBSERVATIONS We identify the case of degeneracies in minor-image per- turbations from the analysis of the lensing event MOA-2016- BLG-319. Figure 1 shows the event light curve. The source star of the event is located at the Galactic bulge field with equatorial coordinates (RA,DEC)J2000 = (17 : 54 : 58.13, −29 : 45 : 01.67). The corresponding galactic coordi- nates are (l, b) = (0.35◦, −2.17◦). The magnification of the source star induced by lensing was first detected and an- nounced to microlensing community on 2016 June 13, HJD′ = HJD − 2450000 ∼ 7552, by the Microlensing Observations in Astrophysics (MOA) survey (Bond et al. 2001; Sumi et al. 2003). The MOA survey used the 1.8 m telescope located at Mt. John Observatory, New Zealand. The day of the event alert approximately corresponds to not only the time of the light curve peak but also the start of a short-term anomaly, which lasted for ∼ 3 days during 7551.5 . HJD′ . 7554.5. However, it was difficult to notice the anomaly because the MOA survey could not cover the event for 4 consecutive nights before the anomaly and the photometry of data during the anomaly was not good enough to delineate the anomaly pattern. As a result, little attention was paid to the event dur- ing the progress of the event. The scientific importance of the event was noticed with the additional data acquired by Korea Microlensing Telescope 4 Han et al. FIG. 2. -- ∆χ2 distributions of points in the MCMC chain on the log s -- log q (left panel) and ∆ξ -- log q (right panel) planes. The parameter q denotes the planet/host mass ratio, s is the projected planet-host separation, and ∆ξ represents the separation between the source and the planetary caustic at the time of the anomaly. Points marked in different colors represents those in the MCMC chain with < 10σ (red), < 20σ (yellow), < 30σ (green), < 40σ (cyan), < 50σ (blue), and < 60σ (purple). Although some lens configurations are excluded in advance based on the previously well-studied origins of degeneracies, interpreting the anomaly may be subject to unknown types of degeneracies. We, therefore, conduct a thorough grid search for the planetary lensing parameters s and q. Besides these planetary parameters, one needs additional lensing pa- rameters to model the observed light curve. These param- eters describe the source star's approach to the lens includ- ing the time of the closest approach, t0, the lens-source sep- aration at that time, u0 (impact parameter), the event time scale, tE (Einstein time scale), and the source trajectory an- gle. The anomaly might be produced by the crossings of the source over caustics. There is no obvious signature of caustic crossings, which usually produce sharp spike fea- ture. However, caustic-crossing features can be smooth if the source is substantially bigger than the caustic. Even if a source is smaller than an overall caustic, it could be big compared to the caustic figure that it is passing over, e.g., OGLE 2016-BLG-1195 (Bond et al. 2017) and OGLE-2016- BLG-1195 (Shvartzvald et al. 2017). These finite-source ef- fects were theoretically predicted by Bennett & Rhie (1996) and observationally confirmed by Beaulieu et al. (2006) for the planetary lensing event OGLE-2005-BLG-390. To ac- count for possible finite-source effects, we include an addi- tional parameter of the normalized source radius ρ, which de- notes the ratio of the angular source radius θ∗ to the angu- lar Einstein radius θE, i.e., ρ = θ∗/θE. For a given set of the planetary parameters s and q, we search for the other parame- ters using the Markow Chain Monte Carlo (MCMC) method. We set the ranges of the grid parameters, i.e., s and q, wide enough to check the possibility that the anomaly is produced by binaries that have similar mass components. In the left panel of Figure 2, we present ∆χ2 distribution of points in the MCMC chain on the log s -- log q plane ac- quired from the preliminary grid search. From the distribu- tion, one first finds that the lens responsible for the anomaly is composed of two masses with a very low mass ratio of q ∼ 4 × 10−3, suggesting that the lower-mass component of the lens is a planet. One also finds that there exist two distinct solutions centered at (log s,log q) ∼ (−0.09, −2.4) (marked by TABLE 2 BEST-FIT LENSING PARAMETERS Parameter Inner solution ("A") Outer solution ("B") χ2 t0 (HJD′) u0 tE (days) s q (10−3) α (rad) Fs,KMTC Fb,KMTC 7308.1 7310.0 7552.737 ± 0.013 7552.742 ± 0.012 0.267 ± 0.012 8.60 ± 0.26 0.817 ± 0.004 3.93 ± 0.11 4.646 ± 0.006 6.25 ± 0.08 -0.10 ± 0.11 0.260 ± 0.012 8.69 ± 0.26 0.945 ± 0.008 4.10 ± 0.12 4.645 ± 0.006 6.23 ± 0.08 -0.10 ± 0.11 NOTE. -- HJD′ = HJD − 2450000. "A") and ∼ (−0.02, −2.4) ("B"). From further refinement of the individual local solutions by letting all parameters vary, it is found that the χ2 difference between the two solutions is merely ∆χ2 ∼ 1.9. This indicates that both solutions describe the observed anomaly almost equally well, although the solu- tion "A" is slightly preferred over the solution "B". To be noted is that the degeneracy between the two solu- tions is different from the previously known 'close/wide' de- generacy. The two solutions resulting from the close/wide de- generacy have planet-host separations s and s−1 and thus one solution has a separation smaller than θE, i.e., s < 1, and the other solution has a separation greater than θE, s > 1. In the case of MOA-2016-BLG-319, both degenerate solutions have separations s < 1 (s ∼ 0.82 for the solution "A" and s ∼ 0.95 for the solution "B") indicating that the origin of the degener- acy does not stem from the symmetry of the lens equation.26 In Table 2, we present the best-fit lensing parameters for the two degenerate solutions. We also present the χ2 values of the fit for the solutions. The uncertainties of the parame- ters correspond to the scatter of points in the MCMC chain. We note that the lensing parameters of the two degenerate so- 26 We note that the red zone in the ∆χ2 distribution (presented in the left panel of Fig. 2) covers s = 1 and even slightly greater. If s ∼ 1, then the lens system forms a resonant caustic, in which the central and planetary caustic merge together. In this case, the back-end of the resonant caustic still induce a dip in the light curve, even if the binary separation s is greater than unity. MOA-2016-BLG-319Lb 5 FIG. 4. -- Lens-system configurations for the two degenerate solutions seen on the source plane. For each panel, the curve with an arrows represents the source trajectory and the red closed curves represent caustics. The two dots marked by M1 and M2 denote the lens components, where M1 > M2. The dashed circle centered at M1 represents the Einstein ring. Contours of lensing magnification are drawn to show the region of anomaly around caustics. FIG. 3. -- Two model light curve resulting from the two degenerate so- lutions: "inner solution" and "outer solution". The lower panels show the residuals from the individual solutions. The lens-system configurations cor- responding to the individual solutions are presented in Fig. 4. lutions are very similar to each other except for the binary separation s. To be also noted is that the time scale of the event, tE ∼ 8.6 days, is short. As a result, higher-order effects induced by the orbital motion of the Earth, microlens-parallax effect (Gould 1992), or that of the lens, lens-orbital effect (Dominik 1998), is not important in describing the observed light curve. Also presented in the table are the flux values of the source, Fs,KMTC, and blend, Fb,KMTC, as measured from the pyDIA photometry of the KMTC data set. We note that the blend flux has a slightly negative value but it is consistent to be zero within the measurement error. These measured values of Fs,KMTC and Fb,KMTC indicate that the flux from the source dominates the blended flux. In Figure 3, we present model light curves for both solu- tions plotted over the observed data points. Except for the very short period around HJD′ ∼ 7553.4, the two model light curves are so similar to each other that it is difficult to distin- guish them within the line width, indicating that the degener- acy between the two solutions is very severe. This can be also seen in the lower two panels in which the residuals from the individual solutions are presented. Figure 4 shows the configurations of the lens system for the individual solutions. In each panel, the line with an ar- row represents the source trajectory, the small dots marked by M1 and M2 indicate the positions of the lens components, and the closed figures composed of concave curves represent the caustic. The upper and lower panels are the configurations for the solutions "A" and "B", respectively. We draw contours of magnification to show the region of anomaly around the caus- tics. For the solution "A", the caustic is composed of 3 sets in which the small caustic located close to the host is the central caustic and the two caustics located away from the host are the planetary caustics. According to this solution, the anomaly was produced by the passage of the source through the region between the central and planetary caustics. The negative devi- ation of the anomaly was produced during the time when the source passed the negative perturbation region between the central and planetary caustics. Because the source trajectory passed the inner region of the planetary caustic with respect to the planet host, we refer to this solution as "inner solution". For solution "B", on the other hand, the lens system produces a single resonant caustic. This results from the merging of the central and planetary caustics because of the proximity of the planet-host separation to unity, s ∼ 0.95. According to this solution, the source trajectory passed the back-end of the caustic and the negative deviation occurred during the time when the source passed the region extending from the caustic end. Because the source trajectory passed the outer region of the caustic (with respect to the planet host), we refer to this solution as "outer solution". For better understanding the origin of the degeneracy be- tween the two solutions, in Figure 5, we present the lens- system configurations seen on the lens plane. In the plot, the planet host is located at the origin and the line with an ar- row represents the path of the source. The two solid curves with arrows represent the paths of the images produced by the host. The red and blue dots represent the planet positions for the inner and outer solutions, respectively. From the con- figurations, it is found that, for both solutions, the planet is located close to the minor image produced by the host at the time of the perturbation. The difference in the configurations between the two solutions is that the planet is located inside the minor image (with respect to the host) for the "inner so- lution", while it is located outside of the image for the "outer solution". This indicates that the similarity between the two solutions is caused by the degeneracy in minor-image pertur- bations. We note that the lens system configuration is similar to that presented in Figure 1 of GG1997 except that the posi- tions of planets are different. The fact that the two solutions are originated from the GG1997 degeneracy can also be seen in the right panel of Figure 2, in which we plot ∆χ2 distribution of MCMC points on the ∆ξ -- log q plane. Here ∆ξ represents the separation be- tween the source and the planetary caustic at the time of the anomaly. The separation ∆ξ is determined from the lensing 6 Han et al. FIG. 5. -- Lensing system configurations for the two degenerate solutions seen on the lens plane. The coordinates are centered at the position of the planet host (black dot) and the dashed circle represents the Einstein ring. The solid line with an arrow represents the path of the source. The two solid curves with arrows represent the paths of the images produced by the planet host. The red and blue dots represent the planet positions for the inner and outer solutions, respectively. parameters by ∆ξ = u0 sin α −(cid:18)s − 1 s(cid:19) , (3) where the former term on the right side, i.e., u0/ sin α, rep- resents the separation between the source and planet host at the time of the anomaly and the latter term, i.e., s − 1/s, denotes the separation between the caustic center and the host. We note that similar plots are presented in Figure 4 of Hwang et al. (2018d) and Figure 5 of Skowron et al. (2018). From the plot, it is found that the degenerate solutions have similar separations ∆ξ ∼ 0.15 but with opposite signs, in- dicating that the source stars of the individual solutions ap- proach the opposite sides of the caustic with similar separa- tions from the caustic. Despite that degeneracies are thought to be considerably less severe for minor-image perturbations compared to major- image perturbations, similar degeneracies in minor-image perturbations were reported by Calchi Novati et al. (2018b) and Koshimoto et al. (2017) for the planetary events OGLE- 2016-BLG-1067 and OGLE-2012-BLG-0950, respectively. We, therefore, compare the lens-system configurations of MOA-2016-BLG-319 with the two other events in order to find the cause of the degeneracy. The lens-system config- urations of OGLE-2016-BLG-1067 are presented in Figures 3 and 4 of Calchi Novati et al. (2018b). We note that they presented 8 degenerate configurations, among which a four- fold degeneracy is caused by the space-based parallax mea- surement (Refsdal 1966; Gould 1994) and the other two- fold degeneracy is relevant to the minor-image perturbation. The configurations of OGLE-2012-BLG-0950 are presented in Figure 2 of Koshimoto et al. (2017). From comparing the lens-system configurations of the events, we find one major difference and one major similar- ity. The difference is that the planetary caustics of the outer FIG. 6. -- Distributions of ∆χ2 of points in the MCMC chain on the s -- ρ parameter plane. Color coding represents points within 1σ (red), 2σ (yel- low), 3σ (green), 4σ (cyan), and 5σ (blue). The left and right panels are the distributions for the inner and outer solutions, respectively. solution is separated from the central caustic for OGLE-2016- BLG-1067, while it is merged with the central caustic for MOA-2016-BLG-319 and OGLE-2012-BLG-0950 (resonant topology). However, in the sense that the source trajectory of MOA-2016-BLG-319 and OGLE-2012-BLG-0950 passed the planetary wing of the resonant caustic, the degeneracies of the events are considered to be of the same type. The similar- ity is that the source stars of the events passed the planet-host axes at about right angles. The source trajectory angles are α ∼ 267◦, ∼ 278◦, ∼ 292◦ for MOA-2016-BLG-319, OGLE- 2016-BLG-1067, and OGLE-2012-BLG-0950, respectively. For general cases in which the entrance angle of the source is substantially different from a right angle, the source tra- jectory passing the inner region of the minor-image caustic produces a light curves that is distinct from a trajectory pass- ing the outer region. On one side of the set of caustics, they will pass closer to one of the two caustics and farther from the other caustic. On the other side, the order will be inverted (for fixed angle). This will result in different anomaly pat- terns, and thus one can easily distinguish the two cases if the anomaly is densely covered. In the case of a right angle source entrance, on the other hand, the source approaches the indi- vidual caustics with approximately same distances. Then, the anomaly patterns resulting from the source trajectories pass- ing the inner and outer regions of the planetary caustic can appear to be similar. We, therefore, conclude that the degen- eracies in the minor-image perturbations of both events occur because the source stars crossed the star-planet axes at ap- proximately right angles. 4. SOURCE STAR We characterize the source star based on the source flux measured in I and V passbands. Besides simply knowing the type of the source star, characterizing the source star is im- portant because it may provide information about the angular Einstein radius in combination with the normalized source ra- dius ρ via the relation θE = θ∗/ρ. For MOA-2016-BLG-319, however, the normalized source radius ρ cannot be measured because deviations in the lensing light curve caused by finite- source effects cannot be firmly detected. See Figure 6, in which we plot the ∆χ2 distributions of points in the MCMC chain on the s -- ρ parameter plane. However, the upper limit on ρ can be measured and this yields a lower limit on θE, which MOA-2016-BLG-319Lb 7 (Gould 2000b). Here κ = 4G/(c2au) ∼ 8.14 mas M−1 ⊙ and πS = au/DS denotes the parallax of the source located at a distance DS. For MOA-2016-BLG-319, however, neither πE nor θE is measured, although the upper limit of θE is set. We, therefore, conduct a Bayesian analysis of Galactic lens- ing events to estimate M and DL based on the measured event time scale. The time scale provides a constraint on the lens parameters because it is related to the parameters by √κMπrel µ ; tE = πrel = au(cid:18) 1 DL − 1 DS(cid:19) . (7) Here µ represents the relative lens-source proper motion. We also use the constraint of θE,min. Implementing a Bayesian analysis requires models describ- ing how lens objects are distributed, i.e., physical distribu- tion, and how they move, i.e., dynamical distribution. One also needs a model mass function of lens objects. We con- struct the lens mass function based on the Chabrier (2003) mass function for stars combined with the Gould (2000a) mass function for stellar remnants including black holes, neu- tron stars, and white dwarfs. Lens and source objects are assumed to be distributed following the physical matter dis- tribution model of Han & Gould (2003), in which the disk has a double-exponential form and the bulge has a triaxial shape. For the dynamical distribution, we adopt Han & Gould (1995) model, in which the motion of disk objects follows a gaussian velocity distribution with a mean corresponding to the rotation speed of the disk, and the motion of bulge ob- jects follows a triaxial gaussian distribution with the velocity components along the individual axes deduced from the bulge shape via the tensor virial theorem. Based on the model dis- tributions, we conduct a Monte Carlo simulation to generate a large number (6× 106) of Galactic lensing events. Then, the the lens mass and distance distributions are constructed based on the events with time scales located within the range of the measured event time scale. With these distributions, we then estimate the representative values of M and DL as the median values. The lower and upper limits of the values are estimated as the 16% and 84% of the distribution. In Figure 8, we present the distributions of the lens mass (upper panel) and distance (lower panel) obtained from the Bayesian analysis. In each panel, the blue curve is the dis- tribution based on only the event time scale tE, while the red curve is the distribution obtained with the additional con- straint of θE,min. It is found that the constraint of θE,min is weak and thus has little effect on the probability distribution. The estimated masses of the lens components are and M1 = 0.15+0.28 −0.08 M⊙ M2 = 0.62+1.16 −0.33 MJ. (8) (9) Therefore, the lens is a planetary system in which a giant planet is orbiting a low-mass M dwarf. Planetary systems be- longing to low-mass hosts are difficult to be detected by other major planet detection methods, e.g., radial velocity or transit methods, due to the faintness of host stars. On the other hand, the microlensing method does not rely on the brightness of the host star, and thus the majority of planetary systems with low- mass stars were found using the microlensing method. See Figure 10 of Han (2018) and Figure 6 of Jung et al. (2018), which show the distribution of planets in the M1 -- M2 plane. The planetary system is estimated to be located at a distance FIG. 7. -- Position of the source star in the instrumental color-magnitude diagram constructed based on the KMTNet data. Also marked is the location of red giant clump (RGC) centroid. may provide a constraint on the physical lens parameters. It is found that the upper limit of the normalized source radius is ρmax ∼ 0.01 as measured at the 3σ level. The de-reddened color (V − I)0 and brightness I0 of the source star are estimated using the known values of the red giant clump (RGC) centroid, (V − I, I)RGC,0 = (1.06,14.41) (Bensby et al. 2011; Nataf et al. 2013), and the offsets in color and magnitude of the source from the RGC centroid. In Fig- ure 7, we mark the positions of the source and RGC cen- troid in the color-magnitude diagram of stars located in the same field of the source. The positions of the source and the RGC centroid are (V − I, I) = (2.00± 0.04,19.01± 0.01) and (V − I, I)RGC = (2.29,15.94), respectively. From the color and brightness offsets, it is found that the de-reddened color and brightness of the source star are (V − I, I)0 = (0.77 ± 0.04,17.49± 0.01), respectively. The estimated de-reddened color and magnitude of the source star indicate that the source is likely to be a turnoff star. We then convert the measured V − I color into V − K color using the (V − I)/(V − K) relation of Bessell & Brett (1988). Finally, we estimate the source an- gular radius using the (V − K)/θ∗ relation of Kervella et al. (2004). The estimated angular source radius is θ∗ = 1.07± 0.09 µas. (4) With the measured angular source radius, the lower limit of the angular Einstein radius is set to be θE,min = θ∗ ρmax ∼ 0.107 mas. (5) 5. BAYESIAN ANALYSIS OF LENS PARAMETERS In order to uniquely determine the mass, M, and distance to the lens, DL, it is required to measure both the microlens parallax πE and the angular Einstein radius θE, from which M and DL are determined by M = θE κπE ; DL = au πEθE + πS (6) 8 Han et al. acy in minor-images perturbations. Because it had been be- lieved that the degeneracy in determining the planet parame- ters would not be severe for minor-image perturbations, such a degeneracy was unexpected. From the comparison of the lens-system configuration with those of OGLE-2016-BLG- 1067 and OGLE-2012-BLG-0950, for which similar degen- eracies were reported, we found that the degeneracies for the events were caused by the special source trajectories that passed the star-planet axis at approximately right angles. By conducting a Bayesian analysis, we estimated that the lens was a planetary system in which a giant planet with a mass Mp = 0.62+1.16 −0.35 MJ) was orbiting a low-mass M dwarf with a mass Mh = 0.15+0.28 −0.08 M⊙, where the planet masses in and out of the parentheses represent the masses for the inner and outer solutions, respectively. The projected host-planet separations were a⊥ ∼ 0.95 au and ∼ 1.05 au for the individual degenerate solutions. The identified degener- acy indicated the need to check similar degeneracies in future analysis of planetary events with minor-image perturbations. −0.33 MJ (0.65+1.21 APPENDIX TYPES OF PLANETARY PERTURBATIONS FIG. 8. -- Probability distributions of the mass (upper panel) and distance (lower panel) of the planet host. The blue curve is the distribution based on only the event time scale tE, while the red curve is the distribution with the additional constraint of the lower limit of the angular Einstein radius, θE,min. TABLE 3 PHYSICAL LENS PARAMETERS Parameter Inner solution Outer solution M1 (M⊙) M2 (MJ) DL (kpc) a⊥ (au) of 0.15+0.28 −0.08 0.62+1.16 −0.33 6.8+1.2 −1.4 0.95+0.17 −0.20 0.15+0.28 −0.08 0.65+1.21 −0.35 6.8+1.2 −1.4 1.09+0.19 −0.22 DL = 6.8+1.2 −1.4 kpc. (10) In Table 3, we list the physical lens parameters. We note that the ranges of the lens mass and distances are consider- able due to the Bayesian nature of determining the lens pa- rameters combined with weak constraint of extra information, e.g., θE or πE. Also presented in the table is the projected sep- aration between the planet and host, which is estimated by a⊥ = sDLθE. The projected separation is a⊥ ∼ 0.95 au and ∼ 1.05 au for the inner and outer solutions, respectively. In both cases, the planet is located away from the snow line at asnow ≃ 2.7 au(M/M⊙) ∼ 0.4 au. 6. SUMMARY We presented the analysis of the planetary lensing event MOA-2016-BLG-319 for which the light curve was charac- terized by a short-term anomaly near the peak produced by the minor-image perturbation. From modeling of the light curve, we found that there existed two distinct solutions that described the observed light curve almost equally well. The planet-host separations of both solutions were smaller than the Einstein radius, indicating that the degeneracy was dif- ferent from the previously known 'close/wide' degeneracy. From the investigation of the lens-system configurations, it was found that the two solutions resulted from the degener- When a source star is microlensed, the image of the source splits into two. One image with a higher magnification ("ma- jor image") appears outside the Einstein ring and the other image with a lower magnification ("minor image") appears in- side the ring. Planetary perturbations occur at the time when the planet is positioned near one of the two microlens images of the primary star and additionally perturbs the nearby image (Mao & Paczy´nski 1991; Gould & Loeb 1992). Depending on which image is perturbed by the planet, planetary perturbations are classified into two types: "major-image perturbation" and "minor-image perturbation" (GG1997). The major-image perturbation indicates the case in which the major image is perturbed by the planet. Because the major image is located outside the Einstein ring, major- image perturbations are caused by planets with separations s > 1. On the other hand, minor-image perturbation indicates the case in which the minor image is perturbed by the planet. The minor image is located inside the Einstein ring and thus minor-image perturbations are caused by planets with sepa- rations s < 1. When a major image is perturbed, the image is further magnified by the planet and thus the lensing light curve always exhibits positive deviations from the light curve produced by the host. In contrast, the minor-image perturba- tion causes the demagnitification of the image, producing, in most cases, a negative deviation in the light curve. In the view point on the source plane, planetary lensing sig- nals are produced when a source approaches the caustic pro- duced by the planet. Planets induce one or two sets of "plane- tary caustics" depending on whether the planetary separation is greater or smaller than the Einstein radius. 27 For the lens system with a wide planet, there exists a single set of planetary caustics with four cusps. For the system with a close planet, on the other hand, there exist two sets of caustics in which one is located above the planet-host axis and the other is located below the axis and each of the caustics is composed of three 27 We note that planets also induce "central caustics" in the region close to the host of the planet. Due to the location of the caustic, planetary signals pro- duced by central caustics always occur near the peak of lensing events with very high magnifications (Griest & Safizadeh 1998). For the dependency of the location, size, and shape of the central caustic on the star-planet separa- tion s and the planet/star mass ratio q, see Chung et al. (2005). MOA-2016-BLG-319Lb 9 cusps. For the detailed properties of planetary caustics, see Han (2006). supported by the Work by CH was grant (2017R1A4A1015178) of National Research Founda- tion of Korea. Work by AG was supported by US NSF grant AST-1516842. Work by IGS and AG were supported by JPL grant 1500811. This research has made use of the KMTNet system operated by the Korea Astronomy and Space Science Institute (KASI) and the data were obtained at three host sites of CTIO in Chile, SAAO in South Africa, and SSO in Australia. The MOA project was supported by JSPS KAKENHI Grant Number JSPS24253004, JSPS26247023, JSPS23340064, JSPS15H00781, and JP16H06287. DPB, AB, and CR were supported by NASA through grant NASA-80NSSC18K0274. The work by CR was supported by an appointment to the NASA Postdoctoral Program at the Goddard Space Flight Center, administered by USRA through a contract with NASA. NJR is a Royal Society of New Zealand Rutherford Discovery Fellow. We acknowledge the high-speed internet service (KREONET) provided by Korea Institute of Science and Technology Information (KISTI). REFERENCES Alard, C., & Lupton, R. H. 1998, ApJ, 503, 325 Albrow, M. D., Horne, K., Bramich, D. M., et al. 2009, MNRAS, 397, 2099 An, J. H. 2005, MNRAS, 356, 1409 Beaulieu, J.-P., Bennett, D. P., Fouqué, P. 2006, Nature, 439, 437 Bennett, D. P., & Rhie, S. H. 1996, ApJ, 472, 660 Bensby, T., Adén, D., Meléndez, J., et al. 2011, PASP, 533, 134 Bessell, M. S., & Brett, J. M. 1988, PASP, 100, 1134 Bond, I. A., Abe, F., Dodd, R. J., et al. 2001, MNRAS, 327, 868 Bond, I. A., Bennett, D. P., Sumi, T., et al. 2017, MNRAS, 469, 2434 Bozza, V., Shvartzvald, Y., Udalski, A., et al. 2016, ApJ, 820, 79 Calchi Novati, S., Suzuki, D., Udalski, A., et al. 2018b, arXiv: 1801.05806 Chabrier, G. 2003, ApJ, 586, L133 Choi, J.-Y., Shin, I.-G., Han, C., et al. 2012, ApJ, 756, 48 Chung, S.-J., Han, C., Park, B.-G., et al. 2005, ApJ, 630, 535 Dominik, M. 1998, A&A, 329, 361 Dominik, M. 1999, A&A, 349, 10 Gaudi, B. S. 1998, ApJ, 506, 533 Gaudi, B. S., 2012, ARA&A, 50, 411 Gaudi, B. S., & Gould, A. 1997, ApJ, 486, 85 Gould, A. 1992, ApJ, 392, 442 Gould, A. 1994, ApJ, 421, L75 Gould, A. 2000a, ApJ, 535, 928 Gould, A. 2000b, ApJ, 535, 928 Gould, A., & Loeb, A. 1992, ApJ, 396, 104 Gould, A., & Horne, K. 2013, ApJ, 779, L28 Griest, K., & Safizadeh, N. 1998, ApJ, 500, 37 Han, C. 2006, ApJ, 638, 1080 Han, C., & Gaudi, B. S. 2008, ApJ, 689, 53 Han, C., & Gould, A. 1995, ApJ, 447, 53 Han, C., & Gould, A. 2003, ApJ, 592, 172 Han, C., Hirao, Y., Udalski, A., et al. 2018, AJ, 155, 211 Hwang, K.-H., Choi, J.-Y., Bond, I. A., et al. 2013, ApJ, 778, 55 Hwang, K.-H., Kim, H.-W., Kim, D.-J., et al. 2018a, arXiv:1802.10246 Hwang, K.-H., Ryu, Y.-H., Kim, H.-W., et al. 2018b, arXiv:1805.08888 Hwang, K.-H., Udalski, A., Bond, I. A. et al. 2018c, AJ, 155, 259 Hwang, K.-H., Udalski, A., Shvartzvald, Y., et al. 2018d, AJ, 155, 20 Jung, Y. K., Udalski, A., Gould, A., et al. 2018, AJ, 155, 219 Kervella, P., Thévenin, F., Di Folco, E., & Ségransan, D. 2004, A&A, 426, 297 Kim, D.-J., Kim, H.-W., Hwang, K.-H., et al. 2018a, AJ, 155, 76 Kim, H.-W., Hwang, K.-H., Kim, D.-J., et al. 2018b, AJ, 155, 186 Kim, S.-L., Lee, C.-U., Park, B.-G., et al. 2016, JKAS, 49, 37 Koshimoto, N., Udalski, A., Beaulieu, J. P., et al. 2017, AJ, 153, 1 Mao, S., & Paczy´nski, B. 1991, ApJ, 374, L7 Nataf, D. M., Gould, A., Fouqué, P., et al. 2013, ApJ, 769, 88 Paczy´nski, B. 1986, ApJ, 304, 1 Park, H., Han, C., Gould, A., et al. 2014, ApJ, 787, 71 Refsdal, S. 1966, MNRAS, 134, 315 Shvartzvald, Y.; Yee, J. C.; Calchi Novati, S., et al. 2017, ApJ, 840, L3 Skowron, J., Ryu, Y.-H., Hwang, K.-H., et al. 2018, AcA, 68, 43 Sumi, T., Abe, F., Bond, I. A., et al. 2003, ApJ, 591, 20 Yee, J. C., Shvartzvald, Y., Gal-Yam, A., et al. 2012, ApJ, 755, 102 Yoo, J., DePoy, D. L., Gal-Yam, A., et al. 2004, ApJ, 603, 139
1310.5048	2	1310	2013-11-30T16:59:42	Extreme orbital evolution from hierarchical secular coupling of two giant planets	[ "astro-ph.EP" ]	Observations of exoplanets over the last two decades have revealed a new class of Jupiter-size planets with orbital periods of a few days, the so-called "hot Jupiters". Recent measurements using the Rossiter-McLaughlin effect have shown that many (~ 50%) of these planets are misaligned; furthermore, some (~ 15%) are even retrograde with respect to the stellar spin axis. Motivated by these observations, we explore the possibility of forming retrograde orbits in hierarchical triple configurations consisting of a star-planet inner pair with another giant planet, or brown dwarf, in a much wider orbit. Recently Naoz et al. (2011) showed that in such a system, the inner planet's orbit can flip back and forth from prograde to retrograde, and can also reach extremely high eccentricities. Here we map a significant part of the parameter space of dynamical outcomes for these systems. We derive strong constraints on the orbital configurations for the outer perturber that could lead to the formation of hot Jupiters with misaligned or retrograde orbits. We focus only on the secular evolution, neglecting other dynamical effects such as mean-motion resonances, as well as all dissipative forces. For example, with an inner Jupiter-like planet initially on a nearly circular orbit at 5 AU, we show that a misaligned hot Jupiter is likely to be formed in the presence of a more massive planetary companion (> 2 MJ) within 140 AU of the inner system, with mutual inclination 50 degrees and eccentricity above 0.25. This is in striking contrast to the test-particle approximation, where an almost perpendicular configuration can still cause large eccentricity excitations, but flips of an inner Jupiter-like planet are much less likely to occur. The constraints we derive can be used to guide future observations, and, in particular, searches for more distant companions in systems containing a hot Jupiter.	astro-ph.EP	astro-ph	Extreme orbital evolution from hierarchical secular coupling of two giant planets Jean Teyssandier1,†, Smadar Naoz2,3, Ian Lizarraga4, Frederic A. Rasio3,5 ABSTRACT Observations of exoplanets over the last two decades have revealed a new class of Jupiter- size planets with orbital periods of a few days, the so-called "hot Jupiters". Recent measure- ments using the Rossiter–McLaughlin effect have shown that many (∼ 50%) of these planets are misaligned; furthermore, some (∼ 15%) are even retrograde with respect to the stellar spin axis. Motivated by these observations, we explore the possibility of forming retrograde orbits in hierarchical triple configurations consisting of a star–planet inner pair with another giant planet, or brown dwarf, in a much wider orbit. Recently Naoz et al. (2011) showed that in such a system, the inner planet's orbit can flip back and forth from prograde to retrograde, and can also reach extremely high eccentricities. Here we map a significant part of the parameter space of dynamical outcomes for these systems. We derive strong constraints on the orbital config- urations for the outer perturber (the tertiary) that could lead to the formation of hot Jupiters with misaligned or retrograde orbits. We focus only on the secular evolution, neglecting other dynamical effects such as mean-motion resonances, as well as all dissipative forces. For ex- ample, with an inner Jupiter-like planet initially on a nearly circular orbit at 5 AU, we show that a misaligned hot Jupiter is likely to be formed in the presence of a more massive planetary companion (> 2MJ) within ∼ 140 AU of the inner system, with mutual inclination > 50◦ and eccentricity above ∼ 0.25. This is in striking contrast to the test-particle approximation, where an almost perpendicular configuration can still cause large eccentricity excitations, but flips of an inner Jupiter-like planet are much less likely to occur. The constraints we derive can be used to guide future observations, and, in particular, searches for more distant companions in systems containing a hot Jupiter. 1Institut d'Astrophysique de Paris, UPMC Paris 06, CNRS, UMR7095, 98 bis bd Arago, F-75014, Paris, France 2Harvard Smithsonian Center for Astrophysics, Institute for Theory and Computation, 60 Garden St., Cambridge, MA 02138 3Center for Interdisciplinary Exploration and Research in Astrophysics (CIERA), Northwestern University, Evanston, IL 60208, USA 4Center for Applied Mathematics, Cornell University, Ithaca, NY 14853-3801 5Department of Physics and Astronomy, Northwestern University †Email: [email protected] – 2 – 1. Introduction To date, about 800 exoplanets have been detected. This number is growing sharply, with more and more planet candidates from the Kepler catalogue being confirmed (there are currently about 3400 unconfirmed candidates, with an overall false-positive rate expected to be 9.4 ± 0.9% according to Fressin et al. 2013). Some of the earliest detections led to the surprising discovery of a new class of Jupiter-like planets in very close proximity to their host star (Mayor & Queloz 1995), the so-called "hot Jupiters" (hereafter HJ). In situ formation at such short distances (just a few stellar radii) from the parent star seems very unlikely. A popular explanation for the presence of a giant gas planet so close to the star is planetary migration, associated with viscous evolution of protoplanetary disk (Lin & Papaloizou 1986; Masset & Papaloizou 2003). This migra- tion should result in orbits with low eccentricities and inclinations (but see Lai et al. 2011; Thies et al. 2011). However, it was shown that other dynamical mechanisms such as planet–planet scattering (Rasio & Ford 1996; Terquem & Papaloizou 2002; Nagasawa et al. 2008; Nagasawa & Ida 2011; Chatterjee et al. 2008; Wu & Lithwick 2011; Beaug´e & Nesvorn´y 2012; Boley et al. 2012) and secular evolution (Holman et al. 1997; Wu & Murray 2003; Takeda & Rasio 2005; Wu et al. 2007; Fabrycky & Tremaine 2007; Takeda et al. 2008; Naoz et al. 2011; Correia et al. 2011; Naoz et al. 2012; Kratter & Perets 2012; Naoz et al. 2013a) also play an important role in the formation of HJs. The Rossiter–McLaughlin effect (Rossiter 1924; McLaughlin 1924; Gaudi & Winn 2007) has enabled measurement of the sky-projected angle between the orbits of several HJs and the spins of their host stars. Surprisingly, about half of these planets are observed to be misaligned and some (about 25%) are even in retrograde orbits with respect to the spin axis of the host star (e.g., Triaud et al. 2010; Albrecht et al. 2012; Brown et al. 2012). These observations suggest that the classical disk migration model is not the only channel to form HJs. The Kepler mission has so far revealed the existence of about 2300 planet candidates, and the number of false positives among this sample is expected to be small (Batalha et al. 2013; Morton 2012). A recent analysis by Steffen et al. (2012) showed that most HJs from the Kepler data appear to have no nearby, coplanar companions (within a period ratio of a few); however, planetary companions at larger separations and large inclinations cannot be excluded (especially since an outer companion with an orbital period ratio of ∼ 10 and a 60◦ mutual inclination would have a detection likelihood of less than 5% by transit methods). Recent developments in direct imaging provide a powerful tool to detect a class of planets that cannot be observed via radial velocity or transit methods: massive planets with large angular separation (i.e., within orbits of tens of astronomical units). For example, Lafreni`ere et al. (2008) (see also Lafreni`ere et al. 2010) 1 and a separation of ∼ 330 AU) around a young found evidence of the first directly imaged planet (of 8 MJ Sun-like star. Shortly after this observation, Marois et al. (2008) announced the discovery of a system of three planets orbiting at several tens of AU of the HR 8799 star, with masses ranging from 5 to 13 MJ (see also Marois et al. 2010; Skemer et al. 2012, for the discovery of a fourth inner planet). More recently, two additional planets (with masses of about 4 MJ) were discovered through direct imaging at projected 1Hereafter we denote by MJ the mass of Jupiter – 3 – distances of few tens of AU from their host stars (Rameau et al. 2013; Kuzuhara et al. 2013). Planets at such distances could have formed in situ through gravitational instabilities in a massive protoplanetary disk (Durisen et al. 2007), or could have been brought there through outward disk-driven migration of planets formed at distances of about 10 AU (Crida et al. 2009). Alternatively it is possible that these planets have migrated there as the result of strong gravitational interactions with (at least) another planet in the system, suggesting a multiple-planet system. Therefore, populations of planets on both close and wide orbits might coexist in planetary systems. These populations in principle could have a large range of eccentricities and inclinations, because of their dynamical history. The direct imaging method is more effective in young systems, for which the planet at large separation still has an important thermal emission, making it easier to observe. Unfortunately, this limitation affects the possibility of detecting any close-in planets (since the star is still very active). Astrometry is another promising method for detecting planets on wide orbits. The efficiency of this method increases with both the orbital separation and the mass of the planet. Therefore a mission such as Gaia could give new insights in the detection of massive planets with orbital periods of several years (see, e.g., Sozzetti et al. 2013, and references therein), with access to a wide range of orbital parameters. In addition, a distant planetary perturber causes a long-term linear trend in the radial velocity curve of its host star, which could appear in long-term radial velocity surveys (see, e.g., Crepp et al. 2012). However this trend does not have a significant effect on the systems we study; we quantify this effect in our results. Different theoretical models have been proposed to explain the presence of HJs and the observed mis- alignments in particular. Some studies proposed that dynamical gravitational scattering in multiplanet sys- tems can lead to large eccentricities and misaligned HJs (Rasio & Ford 1996; Terquem & Papaloizou 2002; Nagasawa et al. 2008; Nagasawa & Ida 2011; Chatterjee et al. 2008; Wu & Lithwick 2011; Beaug´e & Nesvorn´y 2012; Boley et al. 2012). Other studies invoke secular effects (i.e., interactions on timescales that are long compared to the orbital period) by stellar or planet companions in the dynamical evolution of plan- etary systems in the framework of triple systems (Holman et al. 1997; Wu & Murray 2003; Takeda & Rasio 2005; Wu et al. 2007; Fabrycky & Tremaine 2007; Takeda et al. 2008; Naoz et al. 2011; Correia et al. 2011; Naoz et al. 2012; Kratter & Perets 2012; Naoz et al. 2013a). Furthermore, different models suggest that misalignment can be caused by magnetic interactions between the protoplanetary disk and the parent star (Lai et al. 2011) or dynamical interactions with another star that would tilt the disk's axis (Thies et al. 2011). Therefore the planets formed in such disks would be naturally misaligned. Chen et al. (2013) also showed that a combination of disk–planet secular interactions with subsequent Kozai oscillations be- tween the two planets can produce misaligned HJs. In addition, simulations by Teyssandier et al. (2013), Xiang-Gruess & Papaloizou (2013) and Bitsch et al. (2013) also showed that if planets on inclined orbits cohabited with a disk, massive planets were likely to align with the disk, as inclination damping occurs on a timescale shorter than the lifetime of the disk, whereas less massive planets would remain on inclined and eccentric orbits because of Kozai-like excitations emerging from interactions with the disk. Here we study the parameter space of a planetary perturber in the framework of triple-body dynam- ics. For arbitrary inclinations and eccentricities, long-term stability requires the system to be hierarchical. Therefore, the system must consist of an "inner" binary (stellar mass m0 and Jupiter mass m1) in a nearly- – 4 – Keplerian orbit with semi major axis (SMA) a1, and an "outer" binary in which m2 orbits the center of mass of the inner binary, with SMA a2 ≫ a1. Another condition for stability is that the eccentricity of the outer orbit, e2, cannot be too large so that m2 does not make close approaches to the inner binary orbit. In such systems a high mutual inclination between m2 and the (m0, m1) system can produce large-amplitude oscillations of the eccentricity and inclination; this is the so-called Kozai–Lidov mechanism (Kozai 1962; Lidov 1962). Kozai (1962) studied the effects of Jupiter's gravitational perturbation on an inclined asteroid in our own solar system using Hamiltonian perturbation theory. In this influential work, Jupiter was assumed to be on a circular orbit, thus the massless asteroid moved in an axisymmetric gravitational potential. The immediate consequence is that the projection of the inner orbit's angular momentum along the total angular momentum is conserved during the evolution. In fact, at the lowest order of approximation in the ratio of semi-major axes, α = a1/a2, (called the "quadrupole" approximation) in the test particle case (i.e., one of the objects in the inner binary is massless), the component of the inner orbit's angular momentum along the total is conserved even if the outer orbit is not circular (e.g., Lidov & Ziglin 1974). Recently, Naoz et al. (2011, 2013a) showed that these approximations are not appropriate for many systems, particularly in the presence of a (minimally) eccentric outer orbit when the next-order perturbations (octupole) are taken into account, or if the test particle approximation for the inner body is relaxed (at quadrupole or octupole order). As a consequence, the relevant component of the angular momentum is no longer conserved. The lack of conservation of the inner orbit's angular momentum component allows the orbit to reach extremely high eccentricities and can even "flip" the orbit from prograde to retrograde with respect to the total angular momentum. Naoz et al. (2011) considered the secular evolution of a triple system consisting of an inner binary containing a star and a Jupiter-like planet separated by several AU, orbited by a distant Jupiter-like planet or brown dwarf companion. Perturbations from the outer body can drive Kozai-like cycles in the inner binary, which, when planet–star tidal effects are incorporated, can lead to the capture of the inner planet. This leads to a close, highly inclined or even retrograde orbit, similar to the orbits of the observed misaligned HJs. Here we explore the orbital parameter space of a triple-body hierarchical system in the point mass limit (i.e., neglecting tidal dissipation). We focus on planetary systems, where the perturbing object is either a planet or a brown dwarf, but as we will show the system can be scaled to different masses. We show that going beyond the test particle approximation yields qualitatively different results. We map the parameter space of the outer orbit in terms of mass, separation, eccentricity and inclination, and seek the best configurations that would produce retrograde orbits. Thus, we predict the properties of the planet perturber that causes the eccentric Kozai–Lidov evolution. The eccentricity of the inner planet grows large enough to trigger tidal circularization around the host star and could eventually form misaligned HJs. We do not study this process in this paper but give constraints on the perturber that can trigger and cause this behavior. This can help guide future observational programs. This paper is organized as follows: in Section 2 we review the main features of the eccentric Kozai– Lidov mechanism. In Section 3 we present the results of our numerical study: in Section 3.1 we map the – 5 – complete space of parameters, finding the best configurations that could allow the orbit to flip in a retrograde motion and in Section 3.2 we look in closer detail at the inner eccentricity distribution. In Section 4 we run a set of Monte Carlo simulations in order to study precisely the outcome of two representative cases. Finally we discuss these results in section 5. 2. Secular perturbations with an eccentric perturber 2.1. The Eccentric Kozai-Lidov mechanism Throughout the paper we consider the evolution of two planets of mass m1 and m2 orbiting a central star of mass m0. The subscript 1 refers to the inner orbit (consisting of the central mass m0 and the inner planet m1), and the subscript 2 refers to the outer orbit (consisting of the inner orbit's center of mass and the m2 planet). For k = 1, 2, we denote by ak, ek and ik the SMA, eccentricity and inclination of the inner (1) and outer (2) orbits respectively. Throughout the paper we refer to the inclination angle of the inner (outer) orbit with respect to the total angular momentum, i.e., i1 (i2), and to the mutual inclination between the two orbits, which is simply itot = i1 + i2. Relaxing the test particle approximation, Naoz et al. (2011, 2013a) showed that even in the quadrupole level of approximation one finds deviations from the "classical" quadrupole level Kozai evolution. Specif- ically, the inclination can oscillate around 90◦ (e.g., Naoz et al. 2013a), where in the "classical" Kozai mechanism the quantity q1 − e2 1 cos i1 is constant, thus forbidding flips from prograde to retrograde orbits. The "classical" Kozai mechanism is valid for the lowest (quadrupole) order of approximation (if ap- plicable), and if one of the inner orbit members is a test particle. We refer to this limit as the test particle quadrupole (TPQ) approximation. Here we relax the TPQ approximation. In addition we focus on eccentric perturbers, which emphasize the need for the octupole level of approximation (e.g., Naoz et al. 2011, 2013a). In all of our runs we use the Bulirsch–Stoer method in order to numerically solve the octupole-level secular equations following Naoz et al. (2013a), including first-order post-Newtonian relativistic precession of the inner and outer orbits (e.g., Naoz et al. 2013b, note that the interaction term presented there does not affect our results here). We compare our results to the octupole-level test particle approximation (Lithwick & Naoz 2011; Katz et al. 2011). In the octupole level of approximation, the inner orbit's eccentricity can reach very high values, which we map below (see Section 3.2). In addition the inner orbit's inclination can flip its orientation from prograde (itot < 90◦), with respect to the total angular momentum, to retrograde (itot > 90◦). We refer to this process as the eccentric Kozai–Lidov mechanism (hereafter EKL, following the notation of Naoz et al. 2012). In Figure 1 we show an example for the time evolution of a system that is influenced by the EKL mechanism. The system is set initially with the following parameters: m0 = 1 M⊙ , m1 = 1 MJ, m2 = 6 MJ, a1 = 5 AU, a2 = 61 AU, e1 = 0.01, e2 = 0.5, itot = 65◦, with the arguments of pericenters set to be g1 = g2 = 0◦. The longitudes of ascending nodes are set by the relation h2 − h1 = 180◦, (see Naoz et al. 2013a). We choose these parameters as our nominal example and we will often compare our result to it. As can be 160 140 120 100 80 60 40 20 t o t i 1 e - 1 0 2e+09 4e+09 6e+09 8e+09 100 10-1 10-2 10-3 10-4 10-5 10-6 0 2e+09 4e+09 6e+09 8e+09 time (years) – 6 – n o i t u b i r t s d e v i t i l a e r 0.12 0.1 0.08 0.06 0.04 0.02 0 0 20 40 60 80 100 120 140 160 180 inclination Fig. 1.- Time evolution of the nominal example. Left: evolution of the mutual inclination (top) and eccentricity of the inner planet (bottom, as 1 − e1 in log scale) for a two-planet system. The horizontal dashed line shows the separation between prograde and retrograde orbits at 90◦. Right: relative distribution of the mutual inclination during the integration time. The system is the following: a 1 M⊙ star with an inner planet of 1 MJ on an initially circular orbit at 5 AU, and an outer planet of 6 MJ at 61 AU with an eccentricity of 0.5. The two orbits are initially separated by 65◦. The vertical dashed line shows the separation between prograde and retrograde orbits at 90◦. seen in Figure 1 the mutual inclination keeps flipping from a prograde orbit (itot < 90◦) to a retrograde one (itot > 90◦). However those flips are not regularly spaced in time; at every flip the time spent on a prograde or retrograde orbit is not the same. Nevertheless we can see that on average, over the total integration time, the inclination is roughly equally distributed between prograde and retrograde orbits. In addition we note that the eccentricity is mainly distributed between 0 and 0.9 (precisely, 89% of the integration time is spent between these two values in this simulation), but also reaches very high values (up to 0.9999) which, of course, would not make any sense in a system where tidal friction would take place (we quantify the eccentricity distribution in Section 4). In the right panel of Figure 1 we show the distribution of the mutual inclination over the integration time. We also quantify the inclination distribution in Section 4. For this specific example the distribution shows two peaks located at the initial angle and its symmetric with regard to 90◦. Here the inclination tends to be equally distributed between prograde and retrograde orbits. Also, because the system started initially with zero eccentricity the nominal Kozai critical angles (40◦ and 140◦) are limiting the system. – 7 – 2.2. Timescales In Figure 2 we show a close- up of Figure 1, where two distinct timescales appear. Both of them can be associated with a term of the Hamiltonian expansion of the hierarchical three-body problem. The shorter period arises from the quadrupole term, and the longer one arises from the octupole term. These timescales can be estimated by Equations (1) and (2) respectively, where k2 is the gravitational constant (see, e.g., Naoz et al. 2013b, with a modification for the octupole timescale, taking into account the inclination): tquad ∼ 2πa3 2)3/2 √m0 + m1 2(1 − e2 a3/2 1 m2k , 2)5/2 q1 − e2 2(1 − e2 a4 a5/2 1 e2km0 − m1m2 1(m0 + m1)3/2 toct ∼ 2π 4 15 1 G1 G2 + cos itot , where G1 and G2 are the magnitudes of the angular momenta of each orbits, and are given by G1 = G2 = m0m1 m0 + m1 qk2(m0 + m1)a1(1 − e2 1), m2(m0 + m1) m0 + m1 + m2 qk2(m0 + m1 + m2)a2(1 − e2 2). (1) (2) (3) (4) From Equation (2) we see that the octupole timescale increases sharply toward high inclinations. There- fore, systems with very high mutual inclinations (close to polar configurations) are less likely to flip from prograde to retrograde, because the octupole effects take place on a longer timescale than for moderately inclined systems (we refer the reader to ?, for further discussions on the octupole timescale). From Figure 2 we see that in a system that regularly flips from prograde to retrograde configurations, the octupole timescale is of the order of 107 yr. Hence we can expect that for similar initial configurations, this will be the typical timescale for a flip. We also consider the timescale arising from the relativistic precession of the pericenter of the inner orbit (see, e.g, Naoz et al. 2013b), t1PN,1 ∼ 2π a5/2 1 c2(1 − e2 1) 3k3(m0 + m1)3/2 , (5) where c is the speed of light. The same precession timescale can be derived for the outer planet, replacing the subscript 1 by 2, but is negligible because of the large SMA of the outer orbit. If the first-order post- Newtonian (hereafter 1PN) timescale is smaller than the timescale associated with the octupole term, then the latter can be suppressed, leading to no orbital flips of the planet. This will be studied in greater detail in section 3.1. In addition, the EKL mechanism will be completely suppressed if the 1PN timescale is smaller than the timescale associated with the quadrupole term. – 8 – Octupole Period 100 90 80 70 60 50 40 t o t i 30 3.3e+09 3.32e+09 3.34e+09 3.36e+09 3.38e+09 3.4e+09 time (years) Quadrupole Period Fig. 2.- close-up of the time evolution of the inclination of Figure 1. Two periods appear: oscillations due to the quadrupole term, and the oscillation of the octupole envelope. the Kozai 3. Systematic survey of the parameter space In the following section we present numerical results describing the effects of mass, SMA ratio, mutual inclination and the outer orbit's eccentricity. For some specific values of these parameters, the system might be in an unstable configuration. We use the Mardling & Aarseth (2001) stability criterion, which defines a stable three-body system as one that obeys a2 a1 > 2.8(1 + qm)2/5 (1 + e2)2/5 (1 − e2)6/5 (cid:18)1 − 0.3 itot 180(cid:19) , (6) where qm = m2/(m0 + m1) and itot is in degrees. When necessary, we will clearly indicate which region of the parameter space is likely to be unstable. We can already note that the systems are almost always stable for the parameters we have chosen, especially because qm is very small in the case of two planets. The integration time in all our simulations was 8 Gyr. It is important to emphasize that a lower integra- tion time affects the results considerably. In Appendix A.2 we describe our convergence test, which clearly shows that only integration times greater than 5000tquad converge, where tquad is the typical timescale for quadrupole oscillations, and is given by Equation (1). – 9 – 3.1. Likelihood of flipping the orbit In order to estimate the likelihood of this orbital flip, we compute the time spent in a retrograde motion (itot ≥ 90◦) over the total integration time (ttot). We define a new dimensionless parameter f by f = t(itot ≥ 90◦) ttot , (7) and we map this variable over the parameter space. For example a system that never flips from prograde to retrograde has f = 0, and a system that spends exactly half of its time on a retrograde orbit has f = 0.5. We consider our nominal example and systematically vary two parameters in each set of runs (see Table 3.1 for a summary of all the parameters). We plot f as a function of two of these parameters. Results are displayed in Figures 3–10. For each plot, the initial settings are given in the caption of the figure. In Appendix A.1 we map the same numerical experiments as a function of the maximum inclination reached during the integration. ) J M ( 2 m 25 20 15 10 5 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f ) U A ( 2 a 240 220 200 180 160 140 120 100 80 60 40 20 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f 1 2 3 4 5 6 7 8 9 2 4 6 8 m1 (MJ) 10 12 14 16 18 20 a1 (AU) Fig. 3.- Constant parameters are a1 = 5 AU, a2 = 61 AU, e2 = 0.5 and itot = 65◦. The black dashed line represent the m2 = 2m1 function. When q = m1/m2 < 0.5 in- ner orbits start going retrograde, but only for q . 0.3 do they start to converge to f ≃ 0.5. Fig. 4.- Constant parameters are m1 = 1 MJ, m2 = 6 MJ, e2 = 0.5 and itot = 65◦. The lower right black dashed region denotes the region where orbits are likely to be unstable (see Section 2). Large orbital separations (roughly a1/a2 < 1/25 for this set of initial conditions) lead to no formation of retrograde orbits ( f = 0). • Varying m1 and m2. In Figure 3 we find that as long as m2 is at least twice as large as m1, systems always have f > 0. Furthermore, for m2 & 3m1, almost all the systems converge to f ≃ 0.5. A system with m2 ≤ 2m1 will produce retrograde planets for a very limited zone in the phase space. On the contrary, a system with a more massive perturber, even if the latter is on a distant orbit, makes the inner body go into a retrograde motion over the course of 8 billion years. toct=t1PN, e1=0.9 200 180 160 140 120 ) U A ( 2 a 100 80 60 toct=t1PN, e1=0.99 – 10 – 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f 0.7 0.6 0.5 2 e 0.4 0.3 0.2 0.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f 1 2 3 4 5 6 m2 (MJ) 7 8 9 10 1 2 3 4 7 8 9 10 5 6 m2 (MJ) Fig. 5.- Constant parameters are m1 = 1 MJ, a1 = 5 AU, e2 = 0.6 and itot = 65◦. The green dashed lines give the approxi- mate location where the octupole timescale is equal to the 1PN timescale. The top one is for e1 = 0.9 and the bottom one for e1 = 0.99. Systems on the left-hand side of these lines have a 1PN precession time shorter than the octupole time. A close, massive perturber (i.e., a strong perturba- tive potential) induces a longer time in ret- rograde orbits. Fig. 6.- Constant parameters are m1 = 1 MJ, a1 = 5 AU, a2 = 61 AU and itot = 65◦. Here m2 varies from 1 to 10 MJ, and e2 varies from 0.1 to 0.8. The black solid line marks the stability condi- tion according to Equation (6). Above this line the system is unstable according to Mardling & Aarseth (2001). High eccentricities and massive perturbers cause the inner planet to spend more time in retro- grade orbits. • Varying a1 and a2: In Figure 4 we show that for a large range of SMA, the value of f only depends on the ratio between a1 and a2 (rather than the actual value of a1 and a2). This is, of course, not surprising because of the nature of the expansion. As mentioned above we shade in black the possible instability region according to Equation (6). With the parameters used for the runs of Figure 4, we find that there are no more flips when a2 & 25 × a1. Note that this value could be different for other parameters. • Varying a2 and m2. Results in Figure 5 show that the probability of reaching highly inclined orbits strongly depends on these two quantities. Strong outer perturbative potentials produce more flips of the inner orbit. We attribute the sharp transition between flips and no flips to the fact that the post- Newtonian timescale becomes dominant over the octupole timescale. In Figure 5, the green dashed line gives the approximate location for which the octupole timescale is equal to the 1PN timescale. Systems on the left-hand side of this line should not flip, as the post-Newtonian timescale becomes shorter than the octupole timescale. Note that this is just an approximate location, since the octupole timescale gives a rough evaluation for the behavior of the system. Furthermore, the octupole timescale is highly sensitive to the inner orbit eccentricity, which varies during the system evolution. Thus we TP – 11 – 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f ) J M ( 2 m 10 9 8 7 6 5 4 3 2 1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f toct=5.107 years 60 80 100 120 140 160 180 200 40 50 60 70 80 90 a2 (AU) initial mutual inclination 2 e 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 Fig. 8.- Constant parameters are m1 = 1 MJ, a1 = 5 AU, a2 = 61 AU and e2 = 0.5. The green dashed line gives the location where the oc- tupole timescale is equal to 5 × 107 yr, for e1 = 0.9. Mutual inclinations between 55◦ and 85◦ and per- turber's masses between 4 and 10 MJ give the high- est rate of retrograde configurations. Fig. 7.- Constant parameters are m1 = 1 MJ, m2 = 6 MJ and itot = 65◦. Here a2 varies from 51 to 201 AU, and e2 varies from 0.1 to 0.8. The black solid line marks the stability condition according to Equa- tion (6). Above this line the system is unstable according to Mardling & Aarseth (2001). The purple dotted line marks the flip criterion in the test particle limit. Sys- tems above this line are expected to flip in the test particle limit. High outer eccentric- ities and a small SMA ratio cause the inner planet to spend more time in retrograde or- bits. show two possibilities, one with e1 = 0.9 and one for e1 = 0.99. • Varying e2. The eccentricity of the perturber plays a significant part in the evolution of the inner orbit (see Figures 6–9). We find that the eccentricity of the perturber should be higher than 0.2 at least in order to form retrograde inner planets. This is an important constraint on the nature of these systems, and it is interesting to emphasize that a flip can be achieved already for nominally low eccentric perturbers such as e2 = 0.25. We note that for our choice of fiducial parameters, perturbers with eccentricity higher than about 0.68 are unstable according to Equation (6). As shown in Figure 10, relaxing the test particle approximation yields a qualitatively different result. Specifically, in contrast to the test particle case, the flip is, in fact, suppressed at large inclinations, and seems focused (for the nominal example) around initial inclinations itot ∼ 70◦. For comparison the test particle flip criterion is depicted in these figures (see Lithwick & Naoz 2011; Katz et al. 2011). This criterion is symmetric around 90◦, since the outer orbit remains fixed, and is valid only in the regime of very high inclinations (itot > 61.7◦, see Katz et al. 2011). Note that in Figure 10 the corresponding eccentricity TP toct=5.107 years 200 180 160 140 120 ) U A ( 2 a 100 80 60 – 12 – 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f 0.7 0.6 0.5 2 e 0.4 0.3 0.2 0.1 toct=5.107 years TP 40 60 50 initial mutual inclination 70 80 90 40 60 50 initial mutual inclination 70 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 l a t o t / p i l f o i t a r = f 80 90 Fig. 9.- Constant parameters are m1 = 1 MJ and m2 = 6 MJ with e2 = 0.5. The initial mutual inclination varies from 35◦ to 90◦, while a2 varies from 51 to 201 AU. Systems beneath the purple dotted line are expected to flip in the test particle approx- imation. The green dashed line gives the location where the octupole timescale is equal to 5 × 107 yr, for e1 = 0.9. To pro- duce a retrograde orbit, the initial inclina- tion must be in [55◦ : 85◦] for most values of a2. Fig. 10.- Constant parameters are m1 = 1 MJ, m2 = 6 MJ, a1 = 5 AU, a2 = 61 AU. The top solid black line gives the stability limit: systems above this limit are likely to be unstable. Systems above the purple dotted line are expected to flip in the test particle approximation. The green dashed line gives the location where the octupole timescale is equal to 5 × 107 yr, for e1 = 0.9. Highly eccentric and moderately high inclined companions cause the in- ner planet to spend more time in retrograde orbits. for inclinations larger than 80◦ falls below 0.1. If this criterion were valid, all planets above the line labeled "TP" in Figures 7 and 10, and below this line in Figure 9, should flip from prograde to retrograde, and because of the long integration time, converge to f = 0.5. • Varying itot: The ratio of time spent on a retrograde orbit also depends on the initial mutual inclination of the system as seen in Figures 8, 9 and 10. For initial inclinations lower than ∼ 40◦, since we set initially e1 → 0, there are no strong excitations of the inclination and eccentricity, and therefore no possibility of flipping the orbit above 90◦. On the other hand, we find that starting with a very highly inclined orbit (itot > 85◦) does not necessarily imply the formation of retrograde orbits. For most inclinations within the range [55◦, 85◦], the evolution does not show a strong dependence on the initial inclination. This is one of the main differences with the test particle, where both the maximum inclination and eccentricity were well-defined functions of the initial inclination (see Section 2). In the TPQ, f should remain equal to zero, since the initial inclination is lower than 90◦. Thus it appears that an initial inclination between 55◦ and 85◦ is more likely to form retrograde planets. Concerning the issue of why very high initial mutual inclinations (itot > 85◦) do not favor the production of retrograde orbits, we show on Figures 8, 9 and 10 the line at which toct = 5 × 107 yr (taking e1 = 0.9). This line – 13 – 100 10-1 10-2 10-3 , x a m 1 e - 1 10-4 10-5 10-6 2 e 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 60 50 initial mutual inclination 70 80 90 40 60 50 initial mutual inclination 70 2 e 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 40 0.3 0.25 0.2 , 0 2 e , 0 2 e - / ) x a m 2 e ( , 0.15 0.1 0.05 0 80 90 Fig. 11.- Maximum inner eccentricity (given as 1−e1,max in logarithmic scale) for the run in Figure 10. Very high eccentric- ities are associated with flips of the inner orbit. Fig. 12.- Variation of the outer eccentricity e2 for the run in Figure 10. The color scale shows (e2,max− e2,0)/e2,0, where e2,0 is the initial outer eccentricity. This map indicates that the back reaction from the inner planet on the outer planet is more important at high mutual inclination and low eccentricities. strongly suggests that the suppression of the flip at high inclinations is because the octupole timescale becomes too large. Systems set initially with large inclinations typically have toct > 5 × 107 years, which renders the triggering of the EKL mechanism less likely. In such systems, the inner planet does not enter high-eccentricity excitation phases, and is therefore less likely to end up as a HJ. Note that, of course, the inclination also changes as a function of time; however, the system will oscillate between large inclination (minimum eccentricity) and low inclination (large eccentricity). Considering toct for a large eccentricity means that the corresponding inclination should be small, so we approximate it by the initial inclination. We also show in Figures 7, 9 and 10 the analytical prediction for a flip (depicted by purple dashed lines) using the derivation from Katz et al. (2011), which is valid only for inclinations larger than 61.7◦. This presents the qualitatively different results between the test particle approximation and our case. First we find that in our case, unlike the test particle approximation, there is no symmetry of the flip condition around 90◦, and in fact smaller inclinations (around 70◦) are preferable. Furthermore, we can find occasions where a flip can happen in regions unreachable in the test particle approximation, for example the low outer orbit eccentricity case with a2 < 100 AU as seen in Figure 7. It is important to note that the maximum quadrupole inner orbit's eccentricity did not shift from 90◦ to 70◦, however, the contribution of the octupole level of approximation yields a smaller probability for a flip at high inclinations. Furthermore, in the case of small mass of the perturbers the inner orbit torques the outer orbit. This is more apparent in Figure 8 which shows that for larger masses we recover the test particle results. There are three ways to overcome torquing the outer orbit: first, by having a more massive perturber (as in the test particle approximation); second, – 14 – by taking orbits with low initial mutual inclination (since the torque is proportional to sin itot the torque is larger at high inclination), andn finally, by having a larger separation between the inner and outer orbits. The latter not only reduces the torque by reducing the length of the "arm" but also suppresses the octupole contribution. This behavior is apparent, for example, in Figure 9 where an "island" of large probability of flips appears at high inclination and large separations. Of course, large separations also reduce the octupole contribution, resulting in an isolated island. As in the test particle case the system oscillates back and forth from prograde to retrograde. However, unlike the test particle case, the system does not converge to f = 0.5, since the outer orbit reacts to the gravitational perturbations of the inner orbit. In fact one would expect that the system will prefer the retro- grade motion since it is more stable (Innanen et al. 1997), which perhaps can explain the "islands" for which f > 0.5. Note that we have tested in detail the convergence of our systems, and for our integration time (8 Gyr) most of the runs already converged (see Appendix A.2). Another interesting regime that arises from the parameter maps is a "transition zone" where the inner planet spends only about 10% − 20% of its time on a retrograde orbit (colored pale blue in the figures). Also important are the behaviors of the inner and outer orbits' eccentricities. In Figure 11 we show the maximum e1 reached in the corresponding run of Figure 10, and in Figure 12 we show the (relative) maximum e2 for the same run. Not surprisingly, the behavior closely resembles that of the test particle approximation. The probability of flipping the orbits matches the maximum value of e1: flips are associated with excursions to very high eccentricities, which, in fact, happen just before the flip. We find excursions of at least 1 − e1,max . 10−4 when f ≃ 0.5. Furthermore, in our case, the outer orbit's angular momentum is changing too, as can be seen in Figure 12 where we show the maximal relative value reached by the outer eccentricity. This plot shows that the suppression of flips at high initial mutual inclinations is highly related to the outer orbit's evolution. When the outer orbit's eccentricity almost does not change (marked in pale blue) the inner orbit is more likely to flip. These numerical results suggest that HJs that formed through planet–planet secular interactions should have a massive (≥ 3 MJ), eccentric (≥ 0.25), companion with a SMA between 50 and 100 AU, and a mutual inclination between 55◦ and 85◦. A planetary companion like this can drive a Jupiter-like planet in 5 AU to a large eccentricity, which in the presence of dissipation can result in shrinking the orbit to form a HJr (see Naoz et al. 2011). In Appendix A.1, we study the distribution of another variable of interest, the maximum mutual incli- nation reached by the same systems as the ones studied in this section. We show that systems for which f > 0 all reach the same maximum inclination of about 140◦ which is one of the critical Kozai angles. 3.2. Inner orbit eccentricity distribution As noted before, we focus on the dynamical evolution and neglect dissipation throughout the paper. But tidal dissipation will become important when the inner planet reaches very high eccentricities. Therefore in this section we focus specifically on the inner orbit's eccentricity distribution for these systems. In Figure 13 – 15 – we show the cumulative distribution of the inner orbit's eccentricity for different outer orbit configurations. Although a flip (itot > 90◦) happens when the inner orbit's eccentricity reaches a minimum, it also happens right after a large-eccentricity peak (see Lithwick & Naoz 2011; Naoz et al. 2013a, for discussion); thus the large-eccentricity peaks are a good proxy for a flip and vice versa (it is certainly the case for the test particle scenario as shown in Naoz et al. 2012, and we show here that it remains true when this approximation breaks down.) As shown in Figure 13, a systematically low inner orbit eccentricity excitation is achieved for a com- bination of one or more of the following conditions for the outer orbit: low mass, low eccentricity, large orbital separation and low mutual inclinations. However, for high mutual inclinations (& 50◦), outer orbit eccentricities (& 0.25) and massive perturber (& 5 MJ) the cumulative distribution is insensitive to the ini- tial conditions. For these cases, as soon as the octupole effects are triggered, the inner eccentricity reaches extreme values (e1 & 0.99). As a consequence a counterplay may take place between the nearly radial orbit, which drives the planet to the star, and tidal dissipation, which can shrink and circularize the planet's orbit. As shown in Naoz et al. (2011), a fairly high percentage of planets formed by this mechanism end up as hot Jupiters. 4. Statistical estimation through a Monte Carlo experiment We explore the statistical properties of two representative scenarios of systems that are not only signifi- cantly different from the test particle approximation, but also distinct from one another. In the first scenario, we consider a perturber with a mass of 2 MJ (comparable to that of the inner planet, 1 MJ), at a2 = 61 AU. Such a system was shown in the previous section to suppress the EKL behavior. In the second scenario, we consider a system with a perturber with a mass of 6 MJ at a2 = 61 AU. We showed that such a system can undergo large inclination and eccentricity oscillations, but still significantly differs from the test particle approximation since the EKL mechanism is suppressed near initial perpendicular configurations. As shown in Figure 24 in Appendix A.2 most of these systems have converged after 2–6 Gyr. We run our integration up to 8 Gyr and study the distributions of the inclination as well as the inner and outer eccentricities for these systems at this arbitrary time of 8 Gyr, after they have reached a dynamical steady state. For these runs we assume an isotropic distribution of the mutual inclinations (i.e., uniform in cos itot), between 0◦ and 180◦. We make a series of runs with initial conditions m1 = 1 MJ, a1 = 5 AU, a2 = 61 AU, e1 = 0.01 and e2 = 0.3, 0.5, 0.7. We perform 500 runs for each set of parameters (3000 runs total). We show the results of these experiments in Figure 14 for the case where m2 = 2 MJ and Figure 15 for the case where m2 = 6 MJ. Note that the choice of SMA allows the systems to achieve reasonable convergence (see appendix A.2). As long as this convergence condition is fulfilled, the results should not be affected by our choice of SMA. For all the cases with initial mutual inclination lower than 40◦ or above 140◦, no large eccentricity and inclination oscillations occur, and there are no secular changes in the inclination and eccentricity. This is because we set the inner orbit initially with nearly zero eccentricity (we refer the reader to ?, for a discussion on the EKL behavior beyond the Kozai angles). Interesting features appear when the system is initially in the – 16 – m2=2 Mj m2=5 Mj m2=10 Mj m2=20 Mj e2=0.1 e2=0.3 e2=0.5 e2=0.7 100 10-1 10-2 10-3 10-4 10-5 100 10-1 10-2 10-3 10-4 10-5 ) x < 1 e - 1 ( P ) x < 1 e - 1 ( P a2=51 AU a2=81 AU a2=111 AU a2=141 AU i=45 i=55 i=65 i=75 i=85 10-5 10-4 10-3 x 10-2 10-1 100 10-5 10-4 10-2 10-1 100 10-3 x Fig. 13.- Cumulative distribution of the inner eccentricity (represented as 1 − e1 in logarithmic scale) for the following system: a 1 M⊙ star with an inner planet of 1 MJ on an initially circular orbit at 5 AU. When not noted otherwise, the perturber has a mass of 6 MJ at 61 AU with an eccentricity of 0.5, and the two orbits are initially separated by 65◦. For each panel we vary one of these parameters. Top left: we vary the mass of the perturber from 2 to 20 MJ. Top right: we vary the semi-major axis of the perturber from 51 to 141 AU. Bottom left: we vary the eccentricity of the perturber from 0.1 to 0.7. Bottom right: we vary the initial mutual inclination from 45◦ to 85◦. The inner orbit reaches high eccentricities (making orbital flips more likely to happen) for a large set of parameters, almost independently of the exact value of these parameters. Kozai regime. For the m2 = 2 MJ case (Figure 14) with a small outer orbit's eccentricity, the EKL is not very efficient and the final distribution is similar to the initial one. The systems, however, still undergo "classical" quadrupole oscillations, during which they will always reach 40◦ (provided they started on prograde orbits) or 140◦ (provided they started in retrograde orbits) even if they do not flip. This results in the double peak (at 40◦ and 140◦) in the final inclination distribution (as found in Fabrycky & Tremaine 2007), which is also the case for the m2 = 6 MJ case (see Figure 15). As we enter the regime in the parameter space where the EKL mechanism begins to play a significant role (e.g., if the outer orbit's eccentricity is larger, or for a larger perturber's mass, as shown in Figure 15) we deviate from this classical behavior: an additional peak in the final mutual inclination appears around 90◦ because the orbit now flips back and forth. The fraction of time that the orbits spend inclined at 90◦, which is associated with minimum e1, is larger than the fraction of time that the orbits spend at maximum eccentricity (minimum inclination). This, of course, also accounts for the large peak in the inner orbit's eccentricity near zero for the weak EKL case in Figures 14 and 15. As the EKL becomes more significant the minimum eccentricity shifts from zero and becomes wider. As – 17 – expected, the outer orbit's eccentricity is sensitive to the outer planet's mass: for a massive outer planet, e2 almost does not change, but for the m2 = 2 MJ case, e2 oscillates, which results in a suppression of the EKL mechanism. e2=0.5 0 30 60 90 120 150 180 final mutual inclination e2=0.5 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.5 e2=0.7 0 30 60 90 120 150 180 final mutual inclination e2=0.7 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.7 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 e2=0.3 0 30 60 90 120 150 180 final mutual inclination e2=0.3 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.3 n o i t u b i r t s d e v i t i l a e r n o i t u b i r t s d e v i t a e r l i n o i t u b i r t s d e v i t a e r l i 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) Fig. 14.- Final distribution of mutual inclination (top row), inner eccentricity (middle row) and outer eccentricity (bottom row) for a system with two planets. The inner one has m1 = 1 MJ, a1 = 5 AU and e1 = 0.01 and the outer one has m2 = 2 MJ, a1 = 61 AU and three different initial eccentricities: 0.3 (left column), 0.5 (middle column) and 0.7 (right column). The initial mutual inclination is drawn from a distribution uniform in cos itot between 0◦ and 180◦ via a Monte Carlo simulation. When the outer eccentricity is small, the final distribution of the orbital elements remains very close to its initial value. The back reaction from the inner orbit on the outer one is important and only a few systems flip. This is because the masses of the two planets are similar. – 18 – e2=0.5 0 30 60 90 120 150 180 final mutal inclination e2=0.5 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.5 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 e2=0.7 0 30 60 90 120 150 180 final mutal inclination e2=0.7 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.7 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 e2=0.3 0 30 60 90 120 150 180 final mutal inclination e2=0.3 0 0.2 0.4 0.6 0.8 1 final inner eccentricity (e1) e2=0.3 n o i t u b i r t s d e v i t i l a e r n o i t u b i r t s d e v i t a e r i l n o i t u b i r t s d e v i t a e r i l 0.1 0.08 0.06 0.04 0.02 0 0.14 0.12 0.1 0.08 0.06 0.04 0.02 0 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) 0 0.2 0.4 0.6 0.8 1 final outer eccentricity (e2) Fig. 15.- Final distribution of mutual inclination (top panels), inner eccentricity (middle panels) and outer eccentricity (bottom panel) for a system with two planets. The inner one has m1 = 1 MJ, a1 = 5 AU and e1 = 0.01 and the outer one has m2 = 6 MJ, a1 = 61 AU and three different initial eccentricities: 0.3 (left column), 0.5 (middle column) and 0.7 (right column). The initial mutual inclination is drawn from a distribution uniform in cos itot between 0◦ and 180◦ via a Monte Carlo simulation. The distribution of inclination is not uniform anymore and three peaks have appeared around 40◦, 90◦ and 140◦. The inner eccentricity is well distributed between 0 and 1 and the outer one remains close to its original value. This is because of the larger mass of the outer planet, which limits the back reaction and favors orbital flips. – 19 – 5. Summary and discussion We have investigated numerically the dynamics of hierarchical triple systems consisting of a central star, an inner giant planet, and an outer, much more distant perturber that could be another giant planet or a substellar companion. We have varied systematically all important parameters including the separations and masses of the inner and outer planets, the mutual inclination, and the outer orbit's eccentricity (Figures 4 – 10). We showed that relaxing the test particle approximation for this problem results in a much richer variety of dynamical outcomes. In contrast to the test particle case where extreme eccentricity peaks and flips of the inner orbit always happen around a mutual inclination of 90◦, in our systems the behavior is quite different (e.g., Figures 8 and 10; see Section 3.1) and the usual EKL behavior is confined to a smaller region of the parameter space. This study emphasizes the two interesting aspects of the EKL mechanism. One is the importance of the octupole level of approximation, which was explored in detail in the context of HJs and other astrophysical systems in Naoz et al. (2011, 2012, 2013a). The second important aspect is related to relaxing the test particle approximation, which is shown to suppress the EKL mechanism for systems set initially close to perpendicular configurations. In this case, the outer orbit transfers some of its angular momentum to the inner orbit. We showed that the conditions for a flip to occur are sensitive to small changes in the outer orbit's angular momentum. We have also shown that for a large set of parameters (most notably at large orbital separations and large mutual inclinations), the possibility of flipping the orbit from prograde to retrograde is suppressed when the octupole timescale becomes too long (Figures 8–10). Thus, for these systems the inner orbit eccentricity reaches smaller values (see Figure 11). If additional precession effects are present in the system, such as the post-Newtonian precession of pericenters, they can also become more important than the octupole timescale and overcome the octupole variations, suppressing the flips2. The exact location of the limit between flip and no flip is approximated in Figure 5. The Monte Carlo simulations we have conducted (Figures 13–14) suggest that statistically, a hierar- chical triple system that is far from the test particle approximation is most likely to reside in three mutual inclinations regimes, near ∼ 40◦, ∼ 90◦ or ∼ 140◦, almost independently of initial conditions. Furthermore, initial conditions giving low eccentricities of the inner orbit are still more likely configurations to find such a system (although the more interesting behavior will happen in the large-eccentricity peaks). The parameter survey we have conducted gives strong and clear predictions for the orbital parameters of planetary companions that can result in the formation of a HJ. Figures 4–10 suggest that the formation of HJs through planet–planet EKL mechanism predicts a massive (> 2 MJ), eccentric (0.2 − 0.7) planetary companion at separations of ∼ 50−140 AU (for an inner Jupiter-like orbit set initially at 5 AU). Furthermore, as expected for the Kozai mechanism a large (> 50◦) mutual inclination is needed for an inner planet, set initially on a circular orbit. However, orbits close to perpendicular configurations are less likely to flip the 2If the 1PN timescale is shorter the quadrupole timescale further eccentricity excitations are suppressed, when the two timescales are equal a resonant like behavior appears, as shown in Naoz et al. (2013b). – 20 – orbit (unlike the test particle approximation). This means that a planetary companion for misaligned HJs, most likely will not be perpendicular to the HJ orbit, but rather will have a mutual inclination in the range 55◦ − 85◦. A planetary perturber such as described here should, in principle, cause very small variations of the radial velocity curve of the host star, in the form of a small linear trend. A planet of 6 MJ orbiting at 60 star with an eccentricity of 0.5, would cause a semiamplitude variation of 29 m s−1 AU around a 1 M⊙ over its orbital period (465 years), which would appear as a linear acceleration of 0.12 m s−1 year−1. Such perturbations could only appear in long-term, high-precision radial velocity surveys. If we scale down the system slightly, and instead take an 8 MJ planet at 45 AU of a 0.5 M⊙ star with an eccentricity of 0.5, it would cause a semiamplitude variation of 63 m s−1 over its orbital period (427 yr), which would appear as a linear acceleration of 0.3 m s−1 yr−1. Note that for theses calculations, we have assumed the orbital plane of the perturber to be aligned with the line of sight of the observer. If the angle between the two is 45◦, the two trends previously calculated reduce to 0.08 m s−1 yr−1 and 0.21 m s−1 yr−1, respectively. These predictions can also be used as a guide for future direct imaging observations such as the one presented in Macintosh et al. (2006), and can help differentiate between different perturbers (i.e., binary star or faraway planetary or brown dwarf companion). An important caveat is that here we have studied two planet systems. Other routes to the formation of HJs in misaligned orbits exist, including interactions with stellar binary systems (e.g., Naoz et al. 2012), and even primordial misalignment of the disk with respect to the plane of the stellar binary (e.g., Batygin 2012). Achieving high eccentricity peaks (which may result in HJ formation) requires large initial mutual inclination (e.g., Figure 13) which can be a result of planet– planet scattering (e.g., Chatterjee et al. 2008), dynamical relaxation (Papaloizou & Terquem 2001) or early disk accretion from the surrounding gas envelope (Thies et al. 2011). Acknowledgements We thank Guillaume H´ebrard for useful discussions, and the anonymous reviewer for valuable com- ments and suggestions that improved the quality of the paper. Simulations for this project were performed on the HPC cluster fugu funded by an NSF MRI award. This work was supported by NASA Grant NNX12AI86G at Northwestern University. SN was supported by NASA through an Einstein Postdoc- toral Fellowship awarded by the Chandra X-ray Center, which is operated by the Smithsonian Astrophysical Observatory for NASA under contract PF2-130096. REFERENCES Albrecht, S., Winn, J. N., Johnson, J. A., et al. 2012, ApJ, 757, 18 Batalha, N. M., Rowe, J. F., Bryson, S. T., et al. 2013, ApJS, 204, 24 Batygin, K. 2012, Nature, 491, 418 – 21 – Beaug´e, C., & Nesvorn´y, D. 2012, ApJ, 751, 119 Bitsch, B., Crida, A., Libert, A.-S., & Lega, E. 2013, ArXiv e-prints Boley, A. C., Payne, M. J., Corder, S., et al. 2012, ApJ, 750, L21 Brown, D. J. A., Cameron, A. C., Anderson, D. R., et al. 2012, MNRAS, 423, 1503 Chatterjee, S., Ford, E. B., Matsumura, S., & Rasio, F. A. 2008, ApJ, 686, 580 Chen, Y.-Y., Liu, H.-G., Zhao, G., & Zhou, J.-L. 2013, ApJ, 769, 26 Correia, A. C. M., Laskar, J., Farago, F., & Bou´e, G. 2011, Celestial Mechanics and Dynamical Astronomy, 111, 105 Crepp, J. R., Johnson, J. A., Howard, A. W., et al. 2012, ApJ, 761, 39 Crida, A., Masset, F., & Morbidelli, A. 2009, ApJ, 705, L148 Durisen, R. H., Boss, A. P., Mayer, L., et al. 2007, Protostars and Planets V, 607 Fabrycky, D., & Tremaine, S. 2007, ApJ, 669, 1298 Frebel, A., Christlieb, N., Norris, J. E., et al. 2007, ApJ, 660, L117 Fressin, F., Torres, G., Charbonneau, D., et al. 2013, ApJ, 766, 81 Gaudi, B. S., & Winn, J. N. 2007, ApJ, 655, 550 Holman, M., Touma, J., & Tremaine, S. 1997, Nature, 386, 254 Innanen, K. A., Zheng, J. Q., Mikkola, S., & Valtonen, M. J. 1997, AJ, 113, 1915 Katz, B., Dong, S., & Malhotra, R. 2011, Physical Review Letters, 107, 181101 Kozai, Y. 1962, AJ, 67, 591 Kratter, K. M., & Perets, H. B. 2012, ApJ, 753, 91 Kuzuhara, M., Tamura, M., Kudo, T., et al. 2013, ApJ, 774, 11 Lafreni`ere, D., Jayawardhana, R., & van Kerkwijk, M. H. 2008, ApJ, 689, L153 -. 2010, ApJ, 719, 497 Lai, D., Foucart, F., & Lin, D. N. C. 2011, MNRAS, 412, 2790 Lidov, M. L. 1962, Planetary Space Science, 9, 719 Lidov, M. L., & Ziglin, S. L. 1974, Celestial Mechanics, 9, 151 – 22 – Lin, D. N. C., & Papaloizou, J. 1986, ApJ, 309, 846 Lithwick, Y., & Naoz, S. 2011, ApJ, 742, 94 Macintosh, B., Graham, J., Palmer, D., et al. 2006, in Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, Vol. 6272, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series Mardling, R. A., & Aarseth, S. J. 2001, MNRAS, 321, 398 Marois, C., Macintosh, B., Barman, T., et al. 2008, Science, 322, 1348 Marois, C., Zuckerman, B., Konopacky, Q. M., Macintosh, B., & Barman, T. 2010, Nature, 468, 1080 Masset, F. S., & Papaloizou, J. C. B. 2003, ApJ, 588, 494 Mayor, M., & Queloz, D. 1995, Nature, 378, 355 McLaughlin, D. B. 1924, ApJ, 60, 22 Morton, T. D. 2012, ApJ, 761, 6 Nagasawa, M., & Ida, S. 2011, ApJ, 742, 72 Nagasawa, M., Ida, S., & Bessho, T. 2008, ApJ, 678, 498 Naoz, S., Farr, W. M., Lithwick, Y., Rasio, F. A., & Teyssandier, J. 2011, Nature, 473, 187 -. 2013a, MNRAS, 431, 2155 Naoz, S., Farr, W. M., & Rasio, F. A. 2012, ApJ, 754, L36 Naoz, S., Kocsis, B., Loeb, A., & Yunes, N. 2013b, ApJ, 773, 187 Papaloizou, J. C. B., & Terquem, C. 2001, MNRAS, 325, 221 Rameau, J., Chauvin, G., Lagrange, A.-M., et al. 2013, ApJ, 772, L15 Rasio, F. A., & Ford, E. B. 1996, Science, 274, 954 Rossiter, R. A. 1924, ApJ, 60, 15 Skemer, A. J., Hinz, P. M., Esposito, S., et al. 2012, ApJ, 753, 14 Sozzetti, A., Giacobbe, P., Lattanzi, M. G., et al. 2013, ArXiv e-prints Steffen, J. H., Ragozzine, D., Fabrycky, D. C., et al. 2012, Proceedings of the National Academy of Science, 109, 7982 Takeda, G., Kita, R., & Rasio, F. A. 2008, ApJ, 683, 1063 – 23 – Takeda, G., & Rasio, F. A. 2005, ApJ, 627, 1001 Terquem, C., & Papaloizou, J. C. B. 2002, MNRAS, 332, L39 Teyssandier, J., Terquem, C., & Papaloizou, J. C. B. 2013, MNRAS, 428, 658 Thies, I., Kroupa, P., Goodwin, S. P., Stamatellos, D., & Whitworth, A. P. 2011, MNRAS, 417, 1817 Triaud, A. H. M. J., Collier Cameron, A., Queloz, D., et al. 2010, A&A, 524, A25+ Wu, Y., & Lithwick, Y. 2011, ApJ, 735, 109 Wu, Y., & Murray, N. 2003, ApJ, 589, 605 Wu, Y., Murray, N. W., & Ramsahai, J. M. 2007, ApJ, 670, 820 Xiang-Gruess, M., & Papaloizou, J. C. B. 2013, MNRAS, 431, 1320 A. Appendix A.1. Maximum inclination Another variable well suited to map the dynamics of the system is the maximum mutual inclination imax reached by this system over one integration. Mapping the maximum inclination gives information about the dynamics: systems in which the EKL takes place should reach a maximum inclination of about 140◦, whereas in systems where it does not happen, the maximum inclination should remain close to the initial one. In order to proceed to the complete mapping of the parameter space through imax, we successively vary the initial mutual inclination itot, eccentricity e2, SMA a2 and mass m2, in an iterative way, like in Section 3.1. We plot the maximum mutual inclination between the two orbits as a function of two of these variables. The results displayed here give additional information from the one presented in §3.1. Mainly, when an orbit flips, it will always reach the critical Kozai angle of 140◦. On the other hand, systems for which the EKL is negligible will have a maximum inclination close to its initial one. There is an intermediate zone, where the EKL is triggered but the exchange of angular momentum is not enough to flip the orbit. In this case, the maximum inclination reached by the system is located at 90◦ (see, for instance, the light blue transition zone in Figure 22). This preprint was prepared with the AAS LATEX macros v5.2. ) J M ( 2 m 25 20 15 10 5 160 140 120 100 80 60 40 – 24 – n o i t a n i l c n i l i a m x a m ) U A ( 2 a 240 220 200 180 160 140 120 100 80 60 40 20 160 140 120 100 80 60 40 n o i t a n i l c n i l i a m x a m 1 2 3 4 5 6 7 8 9 2 4 6 8 m1 (MJ) 10 12 14 16 18 20 a1 (AU) Fig. 16.- Constant parameters are e2 = 0.5, itot = 65◦, a1 = 5 AU and a2 = 61 AU. The black dashed line represents the m2 = 2m1 function. Fig. 17.- Constant parameters are m1 = 1 MJ and m2 = 6 MJ, e2 = 0.5 and itot = 65◦. The lower-right white-dashed region is where orbits are likely to be unstable (see §2). 200 180 160 140 120 ) U A ( 2 a 100 80 60 n o i t a n i l c n i l a m x a m i 160 140 120 100 80 60 40 0.7 0.6 0.5 2 e 0.4 0.3 0.2 0.1 n o i t a n i l c n i l a m x a m i 160 140 120 100 80 60 40 1 2 3 4 5 6 m2 (MJ) 7 8 9 10 1 2 3 4 7 8 9 10 5 6 m2 (MJ) Fig. 18.- Constant parameters are m1 = 1 MJ, a1 = 5 AU, e2 = 0.6 and itot = 65◦. Fig. 19.- Constant parameters m1 = 1 MJ, a1 = 5 AU, a2 = 61 AU and itot = 65◦. A.2. Convergence We discuss the validity of choosing an integration time of 8 billion years. First, the age of the planetary systems provides an obvious physical limit. The oldest known star is 13.2 billion years old (Frebel et al. 2007). Conducting a large set of simulations is computationally expensive, which gives another limitation. We define a system as convergent for a given time of integration if f has reached a constant value over this time. In Figure 24 we show that an integration time of 8 billion years is sufficient for most systems to reach a steady state in f . Depending on the initial conditions (given in Table A.2), f takes a different time to reach a steady-state value. In the better case this value is achieved within less than a billion years, whereas when 2 e 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 160 140 120 100 80 60 40 – 25 – n o i t a n i l c n i l i a m x a m ) J M ( 2 m 10 9 8 7 6 5 4 3 2 1 60 80 100 120 140 160 180 200 40 a2 (AU) 60 50 initial mutual inclination 70 160 140 120 100 80 60 40 80 90 n o i t a n i l c n i l i a m x a m Fig. 20.- Constant parameters are a1 = 5 AU, m1 = 1 MJ and m2 = 6 MJ, with itot = 65◦. Fig. 21.- Constant parameters are m1 = 1 MJ, a1 = 5 AU, a2 = 61 AU and e2 = 0.5. 200 180 160 140 120 ) U A ( 2 a 100 80 60 40 n o i t a n i l c n i l a m x a m i 160 140 120 100 80 60 40 0.7 0.6 0.5 2 e 0.4 0.3 0.2 0.1 80 90 40 60 50 initial mutual inclination 70 n o i t a n i l c n i l a m x a m i 160 140 120 100 80 60 40 80 90 60 50 initial mutual inclination 70 Fig. 22.- Constant parameters are a1 = 5 AU, m1 = 1 MJ and m2 = 6 MJ with e2 = 0.5. Fig. 23.- Constant parameters are m1 = 1 MJ, m2 = 6 MJ, a1 = 5 AU, a2 = 61 AU. the perturbation is weak, it takes several billion years to converge. In order to use a timescale more relevant to each system, we use the period of Kozai oscillations, as given by Equation (1), where P1 is the period of the inner planet. In Table A.2 we give the period of the Kozai oscillations for each system that we study. In all the runs, we indicate by a vertical line the time t = 1000 × tquad. In some runs this timescale seems relevant to achieve convergence, whereas it fails in some others (see e.g., runs 1, 6 or 7). More precisely, the convergence is slower in the case of extreme initial conditions, such as low initial inclinations or large mass or semi-major axis ratios. An integration time between 5000 × tquad and 10000 × tquad appears safer in order to achieve convergence in all cases. It would, however, be numerically demanding (see ,e.g., run 4, where 5000 × tquad ≃ 20 billion years), so we choose to restrain ourselves to 8 billion years. – 26 – Table 1. Initial conditions of Figures 3–10 Figure m1 (MJ) m2 (MJ) 3 4 5 6 7 8 9 10 1-9 1 1 1 1 1 1 1 1-30 6 1-10 1-10 6 1-10 6 6 a1 (AU) 5 2-20 5 5 5 5 5 5 a2 (AU) 61 10-250 51-201 61 51-201 61 51-201 61 e1 e2 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.5 0.5 0.5 0.1-0.7 0.1-0.8 0.5 0.5 0.1-0.7 itot (deg) 65 65 65 65 65 35-90 35-90 35-90 Note. - Initial conditions for Figures 3 to 10. For all these runs, we took the arguments of pericenters to be initially g1 = g2 = 0◦. Table 2. Initial conditions of Figure 24 Run m2 (MJ) a2 (AU) 1 2 3 4 5 6 7 8 6 6 6 6 6 6 2 12 61 61 61 121 61 61 61 61 e2 0.5 0.3 0.7 0.5 0.5 0.5 0.5 0.5 ffinal itot (deg) 65 65 65 65 80 55 75 65 0.576 0.440 0.551 0.436 0.339 0.574 0.368 0.507 tquad (×106 yr) 0.466 0.623 0.261 3.639 0.466 0.466 1.399 0.233 Note. - Initial conditions of Figure 24. For all these runs, we ffinal is the took m0 = 1 M⊙ fraction of time spent on a retrograde orbit after 8 billion years, and tquad is the period of Kozai oscillations calculated from equation (1). , m1 = 1 MJ; a1 = 5 AU and e1 = 0.01. – 27 – 1 3 5 7 1 0.5 0 1 0.5 0 1 0.5 0 1 0.5 2 4 6 8 2e+09 4e+09 6e+09 8e+09 0 2e+09 4e+09 6e+09 8e+09 0 integration time (years) integration time (years) integration time (years) integration time (years) integration time (years) integration time (years) integration time (years) integration time (years) l l l l l l l l a a a a a a a a t t t t t t t t o o o o o o o o t t t t t t t t / / / / / / / / p p p p p p p p i i i i i i i i l l l l l l l l f f f f f f f f o o o o o o o o i i i i i i i i t t t t t t t t a a a a a a a a r r r r r r r r = = = = = = = = f f f f f f f f 1 0.5 0 1 0.5 0 1 0.5 0 1 0.5 0 0 Fig. 24.- Value of f as a function of the time of integration. The initial parameters of each run are given in Table A.2, with the label of each run given in the top right corner of each panel. The horizontal dashed line indicates f = 0.5 and the vertical dashed line indicated t = 1000 × tquad.
1202.1760	1	1202	2012-02-08T16:39:15	A massive exoplanet candidate around KOI-13: Independent confirmation by ellipsoidal variations	[ "astro-ph.EP" ]	We present an analysis of the KOI-13.01 candidate exoplanet system included in the September 2011 Kepler data release. The host star is a known and relatively bright $(m_{\rm KP} = 9.95)$ visual binary with a separation significantly smaller (0.8 arcsec) than the size of a Kepler pixel (4 arcsec per pixel). The Kepler light curve shows both primary and secondary eclipses, as well as significant out-of-eclipse light curve variations. We confirm that the transit occurs round the brighter of the two stars. We model the relative contributions from (i) thermal emission from the companion, (ii) planetary reflected light, (iii) Doppler beaming, and (iv) ellipsoidal variations in the host-star arising from the tidal distortion of the host star by its companion. Our analysis, based on the light curve alone, enables us to constrain the mass of the KOI-13.01 companion to be $M_{\rm C} = 8.3 \pm 1.25M_{\rm J}$ and thus demonstrates that the transiting companion is a planet (rather than a brown dwarf which was recently proposed by \cite{b7}). The high temperature of the host star (Spectral Type A5-7V, $T_{\rm eff} = 8511-8020$ K), combined with the proximity of its companion KOI-13.01, may make it one of the hottest exoplanets known, with a detectable thermal contribution to the light curve even in the Kepler optical passband. However, the single passband of the Kepler light curve does not enable us to unambiguously distinguish between the thermal and reflected components of the planetary emission. Infrared observations may help to break the degeneracy, while radial velocity follow-up with $\sigma \sim$ 100 m s$^{-1}$ precision should confirm the mass of the planet.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 7 (2002) Printed 16 August 2018 (MN LATEX style file v2.2) A massive exoplanet candidate around KOI-13: Independent confirmation by ellipsoidal variations D. Mislis1⋆, S. Hodgkin1 1Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK Accepted : 2012 February 08 ABSTRACT We present an analysis of the KOI-13.01 candidate exoplanet system included in the Septem- ber 2011 Kepler data release. The host star is a known and relatively bright (mKP = 9.95) vi- sual binary with a separation significantly smaller (0.8 arcsec) than the size of a Kepler pixel (4 arcsec per pixel). The Kepler light curve shows both primary and secondary eclipses, as well as significant out-of-eclipse light curve variations. We confirm that the transit occurs round the brighter of the two stars. We model the relative contributions from (i) thermal emission from the companion, (ii) planetary reflected light, (iii) doppler beaming, and (iv) ellipsoidal variations in the host-star arising from the tidal distortion of the host star by its companion. Our analysis, based on the light curve alone, enables us to constrain the mass of the KOI-13.01 companion to be MC = 8.3 ± 1.25MJ and thus demonstrates that the transiting companion is a planet (rather than a brown dwarf which was recently proposed by Szabo (2011)). The high temperature of the host star (Spectral Type A5-7V, Teff = 8511 − 8020 K), combined with the proximity of its companion KOI-13.01, may make it one of the hottest exoplanets known, with a detectable thermal contribution to the light curve even in the Kepler optical passband. However, the single passband of the Kepler light curve does not enable us to unambiguously distinguish between the thermal and reflected components of the planetary emission. Infrared observations may help to break the degeneracy, while radial velocity follow-up with σ ∼ 100 m s−1 precision should confirm the mass of the planet. Key words: KOI13 - Kepler mission - Extrasolar planets -- transits -- thermal emission -- ellipsoidal variations -- reflected light. 1 INTRODUCTION In Mislis et al. (2011) we demonstrated the potential for estimating planetary masses solely from light curve (LC) data for a restricted sample of systems. The distinguishing characteristic of these sys- tems is that the planet must be both massive enough and close enough to the host star to induce significant tidal distortions in the stellar ellipsoid. Such non-sphericity in the plane of the sky for a rotating star will lead to a periodic photometric signal which is detectable given sufficient signal-to-noise. In Mislis et al. (2011) we applied the technique to the transiting exoplanet HAT-P-7 ob- served with Kepler, and estimated the mass of the planet to be Mp(LC) = 1.27 ± 0.12MJ which is very close to the RV solution, Mp(RV) = 1.20 ± 0.05 (Welsh et al. 2010). The second exoplanet candidate list released by the Kepler team in February 2011, and published in (Borucki 2011), contains 1235 exoplanet candidates. The light curves span measurements taken between May 2009 and September 2011. Motivated by our previous work, we searched the light curves of all 1235 planetary candidates for evidence of ellipsoidal variation. The brightest and ⋆ E-mail: [email protected] best candidate for a star showing strong out-of-eclipse ellipsoidal variations is KOI-13.011 candidate, which is identified with an A5- 7V type star (with a close companion 0.8 arcseconds to the west2). The light curve magnitude is mKP = 9.95 (Kepler passband), and shows primary and secondary eclipses with a published period of 1.7635892 days. The shape of the eclipses are indicative of either a small transiting companion, or a blended system, whereby a deeper eclipse is diluted by contaminating light from nearby stars in the same Kepler pixels. KOI-13.01 is already the subject of a through analysis of the source characteristics and the transit shape by Szabo et al. (2011). They conclude that the system comprises a double A-star wide binary system, the more massive component of which is being eclipsed by a brown-dwarf or very-low-mass star on the basis of a radius determination RC = 2.2 ± 0.3RJ (RC: companion radius). In this paper we independently confirm that the brighter compo- nent of the visual binary is the host star for the eclipses (Section 2). 1 Shortly before submission of this manuscript we became aware of an analysis of the same system, drawing similar conclusions, submitted by Mazeh et al. (2011arXiv1110.3512). 2 Szabo et al. 2011 2 D. Mislis Barnes et al. (2011) use a more sophisticated analysis to measure the radius of the host star and transiting companion, including the effects of a gravity-darkened rapidly-rotating host star. The updated ) are significantly values (RC = 1.4 ± 0.016RJ, R⋆ = 1.77 ± 0.014R smaller than those found by Szabo et al. 2011, and lead Barnes et al. (2011) to conclude that KOI-13.01 may be a planet. Finally, Rowe et al. (2011) presented the system at the 2011 AAS meeting and also suggested the companion is a planet. ⊙ In Section 3 we perform modeling of the out-of-eclipse pho- tometric variation, which puts very strong constraints on the mass of the eclipsing/transiting companion, and lends support to the ar- gument that KOI13.01 has a planetary mass. 2 KEPLER DATA We have used the Kepler public light curve and centroid curve data available from the Kepler archive hosted by the Multimission Archive at STScI (MAST)3 for our analysis. The observations com- prised four long-cadence and four short-cadence datasets (spanning Kepler Q1 -- Q3). The data were processed by the Pre-Search Data Conditioning (PDC) module (Kepler Data Processing Handbook)4. For objects brighter than mKP = 11.3 some saturation, and then bleeding, of the CCD pixels will occur. This charge is not lost, and using an appropriately shaped mask will preserve the flux, and maintain precise photometry even for extremely bright sources (Gilliland et al. 2010, 2011). Both short-cadence and long-cadence data are equally affected given that they use the same 6 second ex- posure. Figure 1 we show the Kepler light curve folded on the pe- riod of 1.7635892 days found by Borucki et al. (2011), with phase zero set to be the center of the primary minimum. 2.1 Identification of the companion host star Borucki et al. (2011) suggest that the source associated with KOI- 13 is actually a double star (separation 0.8 arcsec, ∆mag = 0.4). Szabo et al. (2011) show that the Kepler source is coincident with the double star (BD+46 2629A and B, hereafter KOI-13A & KOI- 13B), two stars of spectral class A, separated by 1.18±0.02 arcsec- onds with a magnitude difference of ∆mag = 0.20 ± 0.04 magni- tudes. Howell et al. (2011) use speckle imaging to find a separation of 1.16 arcseconds and ∆mag = 0.85 at 562nm (∆mag = 1.03 at 692nm) between the two components. However, this photometric difference between the two stars is in serious disagreement with the Szabo et al. (2011) photometry. Having inspected the Szabo et al. (2011) image, we have opted to adopt their value for the mag- nitude difference between the two stars in the Kepler bandpass (as do Barnes et al. 2011). We note that this is a V-band measurement, which is not identical to the Kepler passband, but for an A-type star, the differences are small. In our analysis (Section 5) we con- sider the effects of changing the contribution of this companion star to the Kepler light curve. Based on an unresolved spectrum, and a combination of resolved and unresolved photometry, Szabo et al. use model isochrones to find effective temperatures for the A and B components of T A eff = 8222K (7852 -- 8610K) respectively. eff = 8511K (8128 -- 8913K) and T B The centroid curves for KOI-13 are shown in the middle panel 3 http://archive.stsci.edu 4 available via http://keplergo.arc.nasa.gov/Documentation.shtml Figure 1. Phase folded light curve and the secondary eclipse of KOI-13 system. The primary transit has been removed to enhance the visibility of the out-of-transit variability. of Figure 2. Batalha et al. (2010) have demonstrated that the very tiny shifts in centroid, expected for blended occulting sources, can be measured accurately from the Kepler data. Thus they can be used to show where the largest variations in flux are occurring with respect to the center-of-light. The strong periodic offset in the Ke- pler centroid curve data agrees extremely well with the time of pri- mary transit. Figure 2 (bottom) also shows an illustration of the direction (and magnified amplitude) of the source centroid during primary eclipse. This can be explained simply, if KOI-13A be- comes fainter during every transit and the center-of-light moves towards the west (towards KOI-13B). This interpretation indepen- dently confirms the conclusions from Szabo et al. (2011) who used both the Kepler pixel light curves, and additional ground-based lucky imaging. Thus we confirm that KOI-13A is the host to the companion. 2.2 Removal of the blended light For our analysis, the Q1, Q2 and Q3 datasets were used, corre- sponding to a light curve spanning 418 days in total. The light curve was first corrected to remove the flux from the secondary which we assume to be constant. A magnitude difference of ∆mag = 0.2 was used, and 45 % of the total flux is subtracted from the Kepler light curve (Szabo 2011). The light curve was phase-folded on the best-fit Kepler period of P=1.7635892 (Borucki 2011), rebinned by a factor 1770, and phase=0 set to be the midpoint of the primary transit. We use the updated values for RC, R⋆, α (semi-major axis of the orbit) and inclination (i) from Barnes et al. 2011 (R⋆ = 1.76± 0.014 R , RC = 1.4 ± 0.014 RJ, α = 0.0367 AU & i=85.9). The phase-folded light curve, including the secondary eclipse (but excluding the primary transit to emphasize the out-of-eclipse variation), is shown in Fig. 1. Finally we found no evidence for additional periods in the Kepler corrected light curve, such as Mazeh et al. 2011 suggested (P = 1.0595 days). ⊙ 3 MODELING In our modeling, we consider four main components which con- tribute to the out-of-eclipse phase dependence of the light curve of KOI-13, summarised in Equation 1. A massive exoplanet candidate around KOI-13 3 Figure 3. Top: Phase folded light curve and the best fit model (model − 3) including all three components ( fth, fref, fell, fdop). The blue solid line refers to the fth+ fref component. The additional flux from ellipsoidal variations at phase a = 0.5 is zero, that's why the total flux at the same phase becomes only from the fth+ fref component (ignoring the secondary eclipse). Bot- tom: The ellipsoidal variation residuals and the best fit (red solid line). The mass of the body which causes these variations must be MC ∼ 8.3 MJ. The blue solid line refers to a brown dwarf (BD) ellipsoidal model with mass MBD = 13.0MJ. stellar spectral energy distribution with respect to the bandpass of the instrument (Loeb & Gaudi 2003, Mazeh & Faigler 2010). This final component contains no additional parameters. The first four components are shown in equations 2 -- 5 (for their derivation, and a detailed discussion, see Mislis et al. (2011)). In Mislis et al. (2011), we show that if we can disentangle the ellipsoidal variations from the close companion's thermal and reflected light components, we can solve for the mass of the companion. fth(θ) f⋆ R⋆!2 = A(θ) RC · Const fref(θ) f⋆ 2 = Φ(θ)(cid:18) RC α (cid:19) · ag fell(θ) f⋆ = β 3 MC M⋆ (cid:18) R⋆ α (cid:19) sin3(i) fdop(θ) f⋆ = (3 − ρ) K c (2) (3) (4) (5) Figure 2. Top: The transit phase folded light curve. Middle: X & Y axis offset of the centroid. The offset is periodic (same period as the transit) in the same phase. Offset variations show that the primary star becomes fainter during each transit event. Bottom: RA & Dec coordinates of the KOI-13A (right) and KOI-13B (left) plus the RA & Dec offset vectors. The (0,0) point refers to the KOI-13A coordinates (RA = 286.971o & Dec = 46.8684o). The vectors have been magnified by a factor of 100. ftot f⋆ = 1 + fth,day + fth,night + fre f + fell + fdop f⋆ (1) In our simple model we include (1) thermal emission from the com- panion (day and night side - fth,day, fth,night ), (2) reflected light from the surface of the companion, (3) ellipsoidal variations due to tidal forces between the unseeen companion and the star, and (4) flux variations arising from the Doppler shift (Doppler boosting) of the 4 D. Mislis where fth/ f⋆ is the relative flux from the thermal emission (for the day side, fth,day/ f⋆, A(θ) = Φ(θ) and for the night side, fth,night/ f⋆, A(θ) = 1− Φ(θ)), fref/ f⋆ is the relative flux from the reflected light, fell/ f⋆ is the relative flux from the ellipsoidal variations and fdop/ f⋆ is the Doppler Boosting relative flux. Φ(θ) is the phase function, α is the semi-major axis, Const(Teff, λ, abol, ǫ) is a constant function of the effective temperature of the star and the temperature of the planet (Mislis et al. 2011), and β is the gravity darkening. Finally, K is the radial velocity amplitude, ρ is a function of ρ(λ, Te f f ). ehc/kλTe f f (3 − hc/kλTe f f ) − 3 ρ = ehc/kλTe f f − 1 (6) Our model for the flux changes arising from the ellipsoidal distor- tion (Eq. 4) is a simple approximation, and assumes that the system is in tidal equilibrium (Rowe et al. 2010, Carter et al. 2011, Mis- lis et al. 2011). The abol and ǫ parameters refer to the bolometric albedo and the energy circulation respectively. Both equations 2 & 3 are a function of Φ(θ), but the mass of the secondary body is a de- pendent parameter only in eq. 4 & 5. We note that in our equations the terms R⋆/α and RC/α both appear. These are measured directly from the depth and duration of the primary transit and the period of the system. For consistancy these values are taken from Barnes et al. 2011. If we were to solve for and remove the first two components (thermal emission & reflected light), the residual light curve will contains the mass information (Fig. 3 - bottom). An important re- sult of the model is that the ellipsoidal component contributes zero to the total light curve flux (assuming e=0) at phase=0.5 (centre of the secondary eclipse), and that the phase function depends only very slightly on the balance between the day- and night-side prop- erties of any companion's emission (i.e. it is extremely difficult to measure this with single passband optical data). Thus the thermal emission and the reflected light components will be very difficult to distinguish in a single filter. Nevertheless, one of the goals of mod- eling the full Kepler light curve for KOI-13 is to try to constrain the brightness temperature of KOI-13.01. The main different between our model and other algorithms (such as BEER - Faigler & Mazeh 2011) are that (i) we investigate two phase functions (Lambert and geometrical spheres - Mislis et al. 2011) and (ii) we include distinct reflected and thermal components (including day & night side, abol, ǫ). The choice of phase functions can alter the mass of the planet by 36% (see Section 5). ⊙ ∗ = 1.694 ± 0.013 R KOI-13A is so hot (Teff = 8511 ± 400 K), and the candi- date planet so close (α = 0.0367 AU, ∼ 4.5R⋆), that the thermal flux of the candidate is likely to be non-negligible, even at opti- cal wavelengths. The equilibrium temperature for any zero-albedo companion to KOI-13.01 is Teq = Te f f √R ∗/2α = 2864K (for (R , Barnes et al. 2011), assuming that it does not have its own internal source of heating (as one would expect for a brown dwarf). The hottest known planet discovered to date is probably WASP-33b, with an equilibrium temperature of 2750K, and a brightness temperature of TB = 3620+200 −250K (Smith (2011)), measured from ground-based infrared (λ = 0.9µm) observations of the secondary eclipse (depth of 0.109 per cent). The relative ther- mal emission for WASP-33b in the optical (Kepler passband) would be 10−4 × f (assuming a blackbody). We note that the secondary eclipse of KOI-13.01 is 1.2 · 10−4 × f (Szabo et al. 2011) in the Kepler passband. ∗ ∗ As a starting point, we consider three simple scenarios for the phase light curve data alone, i.e. excluding all datapoints within the primary transit or the secondary eclipse. In case Ctherm we assume that the bolometric albedo of the companion is fixed at abol = 1.0, which means that the planet is perfectly reflective, that the ther- mal emission of the planet is zero, and therefore the model con- tains only reflected light and ellipsoidal variations. In case Cre f l, we assume that the geometric albedo is zero (ag = 0.0), so the reflected light is zero and the model contains only thermal emis- sion ellipsoidal variations, and Doppler boosting ( fdop is a function of ρ(λ, Te f f ) and K - Eq. 6). Finally, in case Chybr we consider a hybrid case, and allow ag and abol to be free parameters, i.e. both thermal emission and reflected light are present in the light curve in addition to any ellipsoidal variations. So in Cre f l we fit for ag and MC, in Ctherm we fit abol, ǫ (energy circulation) and MC, and in Chybr we fit all the parameters above. Eccentricity (e) and the periastron angle (ω) are free parameters in all three scenarios. As a final step, we add the primary transit and secondary eclipse data back into the light-curve, and re-assess Chybr (and name it Cb hybr), primarily to see if the additional information on the depth of the secondary eclipse can provide useful constraints on the ther- mal emission form the planet (as has been suggested by Mazeh et al. 2011). An important and unresolved factor is how we treat the phase function of the reflected light. We consider two approaches: (1) modelling the planet as a Lambert sphere (Russell 1916), assum- ing that the intensity of the reflected light is fre f = 1/π at at phase z = π/2 (as in the case of Venus); (2) An alternative choice would be to assume that the reflected light is fre f = 0.5 at phase z = π/2. In Mislis et al. (2011) we demonstrate that this choice leads to a significantly different phase-contribution for the reflected light, and therefore has a significant impact on the derived mass of the com- panion. The companion mass is significantly larger (by 36 per cent) if we use the Lambert sphere compared to the geometrical sphere. In our analysis we are unable to distinguish between the two cases. Thus, in the next section we present the results from the Lambert sphere case, which gives the larger companion mass. 4 RESULTS With all three cases we find that the orbit is circular and that the eccentricity value is e = 0.0±0.05. All three scenarios require strong ellipsoidal variations to explain the phase light curve. The hybrid model Chybr is significantly preferred with a confidence of 99 per cent confidence (2.55σ) compared to Ctherm, and 99.6 per cent (2.85σ) compared to Cre f l. Scenarios Ctherm and Cre f l are in- distinguishable at the 1σ level. Table 4 shows the results of all three cases (excluding primary transit and secondary eclipse) and Fig. 3 (top) shows the Kepler light curve and the components for the Chybr scenario. In all three cases, we agree on the mass of the unseen companion to within a few per cent, suggesting that the technique is largely independent of the precise nature (be it thermal or reflected) of the emission arising from the companion. Although note that the mass constraint improves as the model becomes more complex. Both Ctherm and Cre f l are significantly less good than the hybrid model, suggesting that components of both thermal and reflected light are contributing to the phase light curve. Note that we do not give errors on ǫ, abol and ag in Chybr because they are essentially unconstrained at the 1-σ level, with large degeneracies between ag, abol, and ǫ. The mass derived from the best fitting model of the candidate is MC = 8.3± 0.9MJ, rather more massive than a typical hot Jupiter, but significantly below the brown dwarf or M dwarf mass proposed by Szabo et al. (2011). This error on MC is the error from fitting the model, and does not allow for the uncertainty in the mass of Cre f l [Reflected] Ctherm [Thermal] Chybr [Mixed] MC [MJ] ǫ abol ag e [deg] χ2 ν 8.0±1.1 0.28±0.1 0.0±0.05 1.042 8.3±1.0 0.17±0.17 0.01+0.1 0.0±0.05 1.032 8.3±0.9 0.10 0.80 0.28±0.28 0.0±0.05 1.03 Table 1. Model results for the three main scenarios discussed in the text. From left-to-right the models increase in complexity. We list the fitted pa- rameters and their 1-σ errors. In each case we are fitting 90 datapoints in the phase-folded light curve. For scenario Chybr, the parameters shown without errors are essentially unconstrained by the data. A massive exoplanet candidate around KOI-13 5 This then is our motivation for scenario Cb hybr introduced in the previous Section. We added back into the light curve an additional 36 (in eclipse) data points, while adding no additional parameters to the model, i.e. the shape of the secondary eclipse is completely de- scribed by Chybr. A little surprisingly, the additional data do nothing to improve our constraints on the nature of the secondary as param- eterized by ag, abol and ǫ. Perhaps this should not be so surprising, the depth of the secondary eclipse in a single filter is unlikely to tell us very much about the ratio of thermal to reflected emission from the planets atmosphere. Infrared measurements could perhaps make a significant impact here, as they could be combined with the optical passband to shed light on the colour of the planet. We also note that our solution to the mass of KOI-13.01 is not affected by the additional data points (see Figure 4). 5 DISCUSSION The very small amplitude of the ellipsoidal variations (∼ 7.2· 10−5) for this period, can only be explained by a body of planetary mass. Increasing the mass of the companion would significantly affect the amplitude of the ellipsoidal variations (Fig. 3 - bottom), there- fore we are confident that the mass of the companion KOI-13.01 is 8.3±1.25 MJ (i.e a 1σ upper limit of 9.55MJ), assuming that there are no other contributions to the Kepler light curve that have so far been ignored in our analysis. It is worth noting that the ellipsoidal component of the light curve is visibly non-symmetric about phase 0.5 due to Doppler Beaming (Figure 3). The difference in heights is ∼ 4.3·10−6, i.e. 3 per cent of the total phase function signal (assum- ing a simple model which neglects reddening). The mass we derive for the unseen companion to KOI-13A, places the object below the minimum mass for deuterium burning (Spiegel et al. 2011). One possible source of error in our estimation of the compan- ion's mass, is the uncertainty in the contribution of the companion A-star to the total Kepler light curve. If we increase the contribu- tion from the nearby stellar component, and the flux we attribute to KOI-13A itself decreases, the mass of the planet becomes cor- respondingly larger, as the relative contribution of the ellipsoidal variations goes up. However, the effect is rather small when com- pared to our other sources of error. Specifically, a factor of ±0.2 magnitudes, leads to a change in the mass of ±0.05MJ for the com- panion. If the Howell et al. (2011) delta-magnitude is used, then the mass of KOI-13b will decrease by ∼ 20 per cent; this implies the body is even more planet-like. Another possible source of error in our treatment of the system is our assumption that the companion behaves as a Lambert sphere. If the correct phase function should be represented by a geomet- rical sphere, then the best-fit mass of the system will actually be significantly reduced by a systematic factor of 36 per cent, down to MC = 5.3MJ. See Mislis et al. (2011) for a more thorough discus- sion of this issue. For this paper, we are using the Lambert sphere phase function, which gives the maximum MC. The only real way to increase the mass of the unseen compan- ion into the brown-dwarf or stellar regime, is to add yet another un- resolved (in the Szabo et al. 2011 lucky imaging), and comparably bright, companion to the Kepler light curve. This seems unlikely, given that the spectral analysis of Szabo et al. (2011) is consis- tent with the photometry and the model isochrones. Ultimately, the mass of KOI-13.01 can only be fully determined with high spa- tial resolution (to resolve the components A and B), time-resolved spectroscopy. We estimate that the radial-velocity amplitude for the KOI-13A system will be K ∼ 870 m s−1 for our best-fit mass. Figure 4. χ2 (Chybr, Cb lustrates the 1-σ error level for each model. ν vs MC for model fits with and without the secondary eclipse hybr) (solid and dashed curves respectively. The horizontal line il- the primary, rather this should be seen as a 20 per cent error on the mass ratio, MC/M ∗=0.0039. We take the mass from Szabo et al. and assume an error of 10 per cent. We also (2011) to be 2.05M ⊙ consider a 0.1 per cent error in the radius ratio RC/R (Barnes et al. 2011). Accounting for both of these leads to a final constraint on the mass of KOI-13.01, MC = 8.3 ± 1.25MJ. ∗ 4.1 How hot is KOI-13.01? Any attempt to constrain the brightness temperature of the compan- ion is severely hampered by the problem of distinguishing between the reflected and thermal components of our model, exacerbated by the single passband. In principle, if there were no reflected light, nor ellipsoidal variations, we could use the difference between the depth of the secondary eclipse (star only) and the point just before ingress to (or after egress from) the primary transit (i.e. star plus planet night-side) for an orbit with zero eccentricity (e.g. TrES-2, O'Donovan (2010)). In practice, this is rather more complicated, because very small contributions from the ellipsoidal variations, and the marginally visible reflected light components are contribut- ing to the total light of the system at this point in phase. And they need to be included in the modelling to constrain the night-side contribution to the system flux. Mazeh et al. (2011) have ignored these effects, and estimate the temperature of the companion to be 2600K. Our rather more complete treatment of the light curve includes both thermal and reflected components, and includes the secondary eclipse in a self-consistent manner. 6 D. Mislis Figure 5. Teq vs RC for planets found on exoplanet.org. KOI-13.01 is in the middle right part of the plot. The solid line is the RC ∝ T 1.4 eq model from Laughlin et al. (2011). For the plot we have used exoplanets.org (http://exoplanets.org/). Although measuring precise radial-velocities for a rapidly rotating and spatially-blended A-star is not simple, the formula in Battaglia et al. (2008) suggest that it should be eminently achievable, even with a small telescope, given sufficient signal-to-noise. Ignoring the complications arising from disentangling the two spectra, we find that a signal-to-noise ratio of 10, 000 per resolution element, should enable us to reach a velocity precision of around 100 me- tre/second. The low amplitude of the ellipsoidal variations rules out the suggestion from Szabo (2011) that the companion could be an M star (Fig. 3). Therefore KOI-13.01 is an exoplanet, and it is now il- lustrative to consider the object in the light of other exoplanets, es- pecially perhaps WASP-33b which also orbits an A-star. In the Teq vs RC diagram (Fig. 5), we see that the current radius (Barnes et al. 2011) is in good agreement with the Teq. Laughlin et al. (2011) examine the radius anomaly of exoplanets around their best fit R ∝ T γ eq ( with γ = 1.4) relation, and we note that KOI-13.01, as a candidate for one the hottest planets known, may provide a useful constraint on our understanding of planetary structures and atmospheres. Mazeh et al. (2011) find a mass of 4±2 and 6±3 MJ based on ellipsoidal variations and Doppler beaming effect using RC, R⋆ from Szabo et al. 2011. These values have since been updated by Barnes et al. (2011). In our analysis we found that the compan- ion is significantly heavier, smaller and closer to the star, than the Mazeh et al. 2011 measurement. Also, Shporer et al. (2011) found MC sin(i) = 9.2±1.1 MJ based solely on the Doppler beaming effect (also using the Szabo et al. 2011 mass). The Doppler beaming sig- nature in the light curve is roughly an order of magnitude smaller than the ellipsoidal variation, as Shporer et al. (2011) show in their paper. Assuming an inclination of i=85.9, the planetary mass from Shporer et al. is 8.1 6 MC 6 10.3. This mass value is significantly heavier than the Mazeh et al. value (both teams are using the same algorithm and the same system parameters), but agrees with our mass value. 6 CONCLUSIONS We present modeling of the KOI-13.01 Kepler light curve, confirm- ing the planetary nature of the companion. Our analysis illustrates the wealth of information that can be extracted from high precision light curves, in the absence of spectroscopy. We find that the out- of-eclipse light curve of KOI-13.01 is dominated by two effects. The light contributed by the planet is at the same amplitude as the ellipsoidal variations induced in the star. We are unable to solve for the relative contributions of ther- mal and reflected emission from the planet, however we expect the planet to be one of the hottest known, given its close proximity to an A-star, and its large radius. The equilibrium temperature for any zero-albedo companion to KOI-13.01 is Teq = Te f f √R ∗/2α = 2864K, which is rather hotter than calculated for WASP-33b and WASP-12b. We suggest that infrared observations would help to disentangle the thermal and reflected components of the light curve. By modelling the light curve, we find that the mass of the planet is MC = 8.3 ± 1.25MJ, robust against any assumptions about the albedo of the planet. We also find that the planet is orbiting in a circular orbit (e = 0.0±0.05). Our results suggest that the density of the planet is ∼3 times larger than Jupiter's density. This value is not surprising. There are much more dense exoplanets, such as XO-3b (density ∼6.5 times more dense than Jupiter - exoplanet.org) We have studied the case that the system is an M-dwarf or brown-dwarf eclipsing companion, but neither of these solutions can explain the light curve we observe. The expected radial veloc- ity amplitude for KOI-13 system is K ∼ 870 m s−1, which will be relatively easy to measure despite the nature of the primary (a rapidly rotating A star) and the contamination from the close A-star companion. It would be an interesting exercise to search all of the Kepler light curves to look for the signature of ellipsoidal variations as a planet detection method, even in the absence of transits. ACKNOWLEDGMENTS This research has made use of NASA's Astrophysics Data Sys- tem Bibliographic Services. DM is supported by RoPACS, a Marie Curie Initial Training Network funded by the European Commis- sions Seventh Framework Programme. REFERENCES Barnes, W., Linscott, E., and Shporer, A., 2011, ApJ 197, 10B Battaglia G.,Irwin M., Tolstoy E., Hill V., Helmi A., Letarte B. & Jablonka P., 2008, MNRAS, 383, 183 Batalha N., Rowe J., Gilliland R. et al. 2010, ApJ 713L 103B Borucki W., Koch D., Basri G., et al. 2011, arcXiv 1102, 541B Carter J., Rappaport S., & Fabrycky D., 2011, ApJ, 728, 139C Cowan N., Agol E., 2011, ApJ 729, 54C Croll B., Jayawardhana R., Fortney J., et al. 2010 ApJ 718, 920C Gilliland R., Jenkins J., Borucki W., et al. 2010, ApJ, 713, 160G Gilliland R., Chaplin W., Dunham E., et al. 2011, ApJ, 197, 6G Hansen B.M.S. 2008, ApJS, 179, 484H Howell S., Everett M., Sherry W., et al. AJ, 142, 19H Laughlin, G., Srismani, M., & Adams, F., ApJL, 729, 7 Loeb A., Gaudi S., 2003, ApJ 588, 117 Mislis D., Heller R., Schmitd J, et. al. 2012, A&A, 538A, 4M O'Donovan F. T., Charbonneau D., Harrington J., et al. 2010, ApJ, 710, 1551O Rowe J., Borucki W., Koch, D., et al. 2010, ApJ, 713L, 150R Rowe J., Borucki W., Howell, S., et al. 2011, AAS, 21710304R Smith A., Anderson D., Skillen I., et al. 2011, MNRAS 416, 2096S145W A massive exoplanet candidate around KOI-13 7 Shporer A., Jenkins J., Rowe J., et al. 2011, AJ, 142, 195S Spiegel D., Burrows A., Milsom J., 2011, ApJ 727, 57S Szabo M., Szabo R., Benko M, et al. 2011, arXiv:1105.2524v1 Welsh W., Orosz J., Seager S., et al. 2010, ApJ 713, 145W This paper has been typeset from a TEX/ LATEX file prepared by the author.
1805.00956	1	1805	2018-05-02T18:04:58	An Estimate of the Yield of Single-Transit Planetary Events from the Transiting Exoplanet Survey Satellite	[ "astro-ph.EP" ]	We present a semi-analytic estimate of the expected yield of single-transit planets from the Transiting Exoplanet Survey Satellite (TESS). We use the TESS Candidate Target List 6 (CTL-6) as an input catalog of over 4 million sources. We predict that from the 200,000 stars selected to be observed with the high-cadence postage stamps with the highest CTL-6 priority, there will be 241 single-transit events caused by planets detectable at a signal-to-noise ratio of SNR$\ge7.3$. We find a lower limit of an additional 977 events caused by single-transit planets in the full frame images (FFI); this is a lower limit because the CTL-6 is incomplete below a TESS magnitude of $T>12$. Of the single-transit events from the postage stamps and FFIs, 1091/1218 will have transit depths deeper than 0.1%, and will thus be amenable for photometric follow-up from the ground, and 1195/1218 will have radial velocity signals greater than 1 m/s. We estimate that the periods of 146 single transits will be constrained to better than 10% using the TESS photometry assuming circular orbits. We find that the number of planets detected by TESS in the postage stamps with periods $P>25$ days will be doubled by including single-transiting planets, while the number of planets with $P>250$ days will be increased by an order of magnitude. We predict 79 habitable zone planets from single-transits, with 18 orbiting FGK stars.	astro-ph.EP	astro-ph	Draft version May 4, 2018 Typeset using LATEX twocolumn style in AASTeX61 AN ESTIMATE OF THE YIELD OF SINGLE-TRANSIT PLANETARY EVENTS FROM THE TRANSITING EXOPLANET SURVEY SATELLITE Steven Villanueva Jr.,1 Diana Dragomir,2, 3 and B. Scott Gaudi1 1Department of Astronomy, The Ohio State University, 140 West 18th Av., Columbus, OH 43210, USA 2Massachusetts Institute of Technology, Cambridge, MA 02139 USA 3Hubble Fellow Submitted to AAS Journals ABSTRACT We present a semi-analytic estimate of the expected yield of single-transit planets from the Transiting Exoplanet Survey Satellite (TESS). We use the TESS Candidate Target List 6 (CTL-6) as an input catalog of over 4 million sources. We predict that from the 200,000 stars selected to be observed with the high-cadence postage stamps with the highest CTL-6 priority, there will be 241 single-transit events caused by planets detectable at a signal-to-noise ratio of SNR≥ 7.3. We find a lower limit of an additional 977 events caused by single-transit planets in the full frame images (FFI); this is a lower limit because the CTL-6 is incomplete below a TESS magnitude of T > 12. Of the single-transit events from the postage stamps and FFIs, 1091/1218 will have transit depths deeper than 0.1%, and will thus be amenable for photometric follow-up from the ground, and 1195/1218 will have radial velocity signals greater than 1 m/s. We estimate that the periods of 146 single transits will be constrained to better than 10% using the TESS photometry assuming circular orbits. We find that the number of planets detected by TESS in the postage stamps with periods P > 25 days will be doubled by including single-transiting planets, while the number of planets with P > 250 days will be increased by an order of magnitude. We predict 79 habitable zone planets from single-transits, with 18 orbiting FGK stars. Keywords: catalogs - planetary systems - surveys 8 1 0 2 y a M 2 . ] P E h p - o r t s a [ 1 v 6 5 9 0 0 . 5 0 8 1 : v i X r a [email protected] 2 Villanueva Jr., Dragomir, & Gaudi 1. INTRODUCTION The Transiting Exoplanet Survey Satellite (TESS), launched in Spring 2018, will discover thousands of tran- siting exoplanets that exhibit two or more transits dur- ing the mission. TESS will have a number of advan- tages over previous transiting planet surveys, including ground-based surveys such as the Hungarian Automated Telescope (Bakos et al. 2004) survey, the Wide Angle Search for Planets (Pollacco et al. 2006) survey, and the Kilodegree Extremely Little Telescope (Siverd et al. 2012) survey, as well as space-based missions like Corot (Baglin 2003), Kepler (Borucki et al. 2010), and K2 (Howell et al. 2014). Ground-based surveys are essen- tially limited to planets whose transits have depths above ∼ 0.1%, but do so around bright stars that are amenable to follow-up observations from ground-based telescopes and radial velocity measurements. Kepler has exquisite photometric precision down to several tens of parts-per-million (ppm) (Koch et al. 2010), but the ma- jority of the planets found by Kepler are orbiting stars that are too faint to be confirmed via radial velocity us- ing the current generation of telescopes and instruments. Both the original Kepler campaign and the extended K2 mission campaigns have relatively long baselines of al- most 4 years and 80 days respectively, but both are also limited in their sky coverage. By virtue of TESS's observing strategy and de- sign, it will observe 85% of the entire sky, monitoring and discovering planets transiting bright stars, which are amenable to both photometric and radial velocity follow-up, as well as detailed characterization of their at- mospheres via ground and space-based telescopes. The trade-off of achieving this nearly all-sky coverage is that 63% of the sky, or 74% of the mission's total sky cover- age, will only be observed for 27 days (as compared to 80 days for each K2 campaign and nearly 4 years for the pri- mary Kepler campaign). In this regime of many millions of stars monitored for a relatively short amount of time, the number of single-transit planetary events found by TESS will be much larger than that expected or found by Kepler (Yee & Gaudi 2008; Foreman-Mackey et al. 2016). Single transit events require significantly more resources to confirm than planets that exhibit two or more transits, but nevertheless can be quite scientifically valuable. With the planned survey strategy, and a requirement of at least two transits to confirm a planet, the major- ity of these planets will have periods of less than 10 days. The primary mission of the survey is to mea- sure masses and radii of 50 terrestrial planets. This leaves open the opportunity to discover planets outside of this regime, including giant planets and planets on long orbits that transit only once. However, recovering, confirming and studying planets that transit only once pose difficulties. Some of these difficulties include the fact that their ephemerides are difficult to constrain for the purpose of scheduling of future observations, and they are easily confused with false-positives. Previous studies have investigated single-transit events in Kepler (Yee & Gaudi 2008). Multiple sim- ulations have been performed to estimate the (two or more transit) yield of TESS (Sullivan et al. 2015; Bouma et al. 2017; Ballard 2018; Barclay et al. 2018). In each of these simulations, only systems that exhibit two or more transits are reported, and to date there has not been an estimate of the expected yield of single- transits in TESS, or their properties. Given the number of stars and observing strategy of TESS, we expect a 100 fold increase in the number of single-transit events in TESS relative to Kepler. It is both worthwhile and possible to follow up these longer-period transiting planets, as they represent an op- portunity to investigate a number of questions related to planet formation, such as the migration mechanism for Hot Jupiters, and the physical mechanisms that lead to inflated radii of close-in giant planets. Using the defi- nition of habitable zones described by Kopparapu et al. (2013), transiting planets of main sequence stars of spec- tral type earlier than roughly M5 (Teff ≈ 2800 K) will have the inner edge of the habitable zone at periods of ≈ 11 days. For the majority of the TESS survey, any habitable zone planets around M4 or earlier stars are expected to only display single transits. 2. EXPECTED NUMBER OF SINGLE-TRANSIT PLANETS The expected total number of planets detectable by TESS with exactly one or more transits is the integral over all periods and all planetary radii of the geometric probability of detecting a transit around a star ℘tr, the probability of observing the transit(s) during the finite baseline of observations ℘B, and the planet occurrence rates f (P ) with a Heaviside step function cut on the signal-to-noise ratio Θ(∆SN R), multiplied by the total number of stars observed by TESS N⋆ Ndet = N⋆Z ℘tr℘Bf (P )Θ(∆SN R)dP (1) where ∆SN R = SN R − SN Rmin, where each term is for a fixed period and planetary radius. In reality, each of the terms [N⋆, ℘tr, ℘B, f (P ), Θ(∆SN R)] depend on more than just the period P and planet radius, but also depend on other variables such as the stellar mass, stellar radius, apparent stellar magnitude, and intrinsic Single Transits in TESS 3 stellar variability. All of these variables are considered in the final analysis. We will also assume circular orbits and that rp ≪ R⋆. With this assumption, the geometric transit probability is then ℘tr = R⋆ a (2) for non-grazing geometries, where R⋆ is the host star It is possible to evaluate Equation 1 in units radius. of the semi-major axis a, but it is more convenient to use the period P as this is the direct observable in both transit and radial velocity detections of exoplanets. We use Kepler's third law assuming that the planet's mass is much smaller than the stellar mass to convert semi- major axis to period P 2 = a3 4π2 G(M⋆ + mp) ≈ a3 4π2 GM⋆ (3) where M⋆ is the stellar mass. The geometric transit probability then becomes a function of stellar mass, stel- lar radius, and orbital period: ℘tr = (cid:18) 4π2 G (cid:19)1/3 R⋆M −1/3 ⋆ P −2/3 (4) The geometric probability decreases as ℘tr ∝ P −2/3 and leads to a decreased probability of detection at long pe- riods. We also consider the probability of a transit occurring during the finite baseline of observation B of the TESS mission ℘B. During this paper, we will evaluate cases where two or more transits occur, or exactly one tran- sit occurs. In the case where both two or more transits are observed with planets on periods shorter than B/2, the probability of observing two or more transits occur- ring during the observing baseline is unity. However for planets on periods longer than B/2, the probability de- creases until only one, or no transits occur during the observing baseline. For a finite observing baseline B and ignoring the finite duration of the transits, the proba- bility of exactly one ℘B,1, or two or more ℘B,2+ transits occurring is given by: ℘B,1 = B P = 2P = 1 B − 1 , B B , B ℘B,2+ = 2 − 2P , P ≥ B 2 ≤ P ≤ B 2 ≤ P ≤ B , P ≤ B 2 (5) These are analogous to Equations 3 and 4 from Yee & Gaudi (2008), but we include the case of two or more transits, instead of exactly two transits. Days Square Degrees Sky Fraction Mission Fraction 0 27.4 54.8 82.2 110 6023 25989 6270 1238 231 14.6 63.0 15.2 3.0 0.56 0 73.8 17.8 3.5 0.66 137...301 578(total) 1.4(total) 1.6(total) 329 356 215 701 0.52 1.7 0.6 2.0 Table 1. Fraction of sky covered by various observing baselines. Yee & Gaudi (2008) incorrectly chose the lower limit for the two transit case to be P/4, instead of P/3, however, this only biases their yields for the two-transit cases and their single-transit yields should be unaffected. When investigating single-transit events with periods longer than the observing baseline, the total probability is ℘tot ≡ ℘tr℘B = 0.026(cid:18) R⋆ R⊙(cid:19)(cid:18) M⋆ M⊙(cid:19)−1/3(cid:18) P 27.4d(cid:19)−5/3(cid:18) B 27.4d(cid:19)(6) and scales as ℘tr℘B ∝ P −5/3. The product of these two terms can be seen in Fig- ure 1, for representative host stars and a observing base- line of B = 27.4 days. To detect two transits, the prob- ability is the geometric transit probability R⋆/a until periods of B/2, while the probability of detecting a sin- gle transit peaks at B, with the transition for observing one versus two transits occurring between B 2 ≤ P ≤ B. Each star observed by TESS will have an observing baseline of an integer multiple of N × 27.4 days where 1 ≤ N ≤ 13. The amount of sky covered in each observ- ing baseline is summarized in Table 1. The dominant baselines are 73.8% of the mission covered for 27.4 days, 17.8% for 54.8 days, and 3.5% for 82.2 days. There is an uptick at the ecliptic poles, which cover 2% of the mis- sion for 356 days. Each remaining observing baselines cover less than 1% of the mission. For the total number of stars observed by TESS, we use the TESS Candidate Target List-6 (CTL-6) pro- vided online by Stassun et al. (2017). The catalog has ∼4 million sources, with estimated host star masses, radii, and TESS magnitudes, the host star's coordi- nates, and and estimate of the blended flux from nearby stars that is expected to dilute the depth of transits. This method deviates from those used in simulations 4 Villanueva Jr., Dragomir, & Gaudi Figure 1. Top Left: Probability of observing a single transit (solid) or two or more transits (dashed) for the 27.4 day baseline as compared to the geometric transit probability (dotted). Colors correspond to a 1.0, 0.6, and 0.25 M⊙ host star. Top Right: Mission-weighted probability of observing a single transit (solid black line) or two or more transits (dashed black line) over all observing baselines as compared to the geometric transit probability (red dotted line) for a 1.0 M⊙ host star. All 13 individual single transit probability curves, corresponding to the 13 possible baselines, are shown as grey solid lines for reference. Bottom Left: Planet occurrence rates for Teff ≥ 4000 K stars (blue) and Teff < 4000 K stars (red). Dark lines are the fraction of stars expected to host at least one planet in each period bin, while the dotted lines represent our extrapolation to long periods. Bottom Right: Combining the first three panels with the total number of stars, we estimate the period distribution of single transit events expected from TESS postage stamps (black) and in the FFIs (gray). Events around stars with Teff ≥ 4000K are in blue, and stars Teff < 4000 K in red. The darker shades are for the 2-minute cadence, while the lighter shades are for the 30-minute cadence. The total number of planets exhibiting a single transit event expected from the TESS mission is over 1000. There are 241 planets expected in the 2-minute cadence data, and lower limit of 977 planets in the FFIs. Single Transits in TESS 5 by Sullivan et al. (2015) and Bouma et al. (2017) as we do not use a Galactic model, but instead calculate our yields directly from the CTL-6. For each of the 4 million stars in the CTL-6, we use the estimated mass, radius, effective temperature, magnitude, and ecliptic latitude of the star. As the the longitude of the first sector was not yet known, we use the ecliptic latitude to assign the number of sectors in which the star will be observed by TESS, and determines the total observing baseline. We do this by taking all stars above a given ecliptic latitude, such that the total area on sky is the same as described in Table 1. All observed stars are sorted by the CTL- 6 priority, where the top 200,000 stars are classified as postage stamp (PS) stars, and the remaining stars are placed in the FFI sample. Barclay et al. (2018) have show that using only the CTL priority may result in an over-selection of target stars in the ecliptic poles rela- tive to the true mission, however we remain agnostic as to the final selection strategy of the targeted stars and default to CTL-6 priority over speculation as to what the final selected stars will be. We use the planet occurrence rates of Fressin et al. (2013) for stars with Teff ≥ 4000K, and of Dressing & Charbonneau (2015) for stars with Teff < 4000K. The planet occur- rence rates are only complete to periods of ∼ 100 days, but we extrapolate these rates to periods of > 1000 days to explore the probability of finding planets at longer periods. The assumed planet occurrence rates can be seen in Figure 1. For each period and radius bin of the respective occurrence rates, we draw a radius and period from a random uniform logarithmic distribution in that bin. From the period of the planet, host star mass, and host star radius we calculate the geometric probabil- ity using Equation 4. From the ecliptic latitude of the star and the period of the planet, we can calculate the probability of the object being observed for one, two, or more transits from Equation 5. We note that, because the probability of detecting a single transit drops pre- cipitously with period (∝ P −5/3), changing our assumed form for the extrapolation in period (within reason) is unlikely to change our results substantially. We only calculate detections if the SNR is above 7.3. To calculate the SNR, we follow the formula used in Bouma et al. (2017) SN R = pNtr δD 1hr (cid:16) σ2 T + σ2 v(cid:17)1/2 , (7) using the number of transits Ntr = 1 for single-transits, the transit depth δ, the dilution from background stars and contamination D bounded from 0–1, the total noise per hour σ2 1hr from CCD read noise, photon-counting noise, zodiacal noise, and a systematic 60 ppm hr1/2 instrumental noise floor, the transit duration T , and an intrinsic variability term σv. We use the host star's Teff to assign an intrinsic stel- lar variability based on Basri et al. (2013) following the procedures of Sullivan et al. (2015) and Bouma et al. (2017). Using the planet's period and radius, along with the host star's variability and magnitude, we calculate the SNR of each planet, in each period and radius bin, around every star in the sample. For those that have a SN R > 7.3, we define the planet as being detected. We then sum the product of the geometric probabil- ity, probability of being observed, and planet occurrence rate, over all detected planets around all stars. The re- sults are shown in Figure 1. The total number of single- transit events is 1218, with 241 of the detections being found in the postage stamps. 201/241 have periods > 25 days, and 19/241 have periods > 250 days. Finally, we recover an estimate of the total integrated number of planets detected. As we integrate fractional probability over all stars, we only recover the total number of plan- ets detected, and not the total number of host stars. As such we and cannot make any quantitative state- ments on the expected multiplicity of the systems, but one could assume that each star hosts only one planet. 2.1. Demographics of Detected Single-Transits We present the demographics of the detected planets. In Figure 2 we show the distribution of detections in host star magnitude, host star effective temperature, planet radius, and stellar insulation relative to the Earth. Of the 1218 expected single transits, 173 are around stars brighter than T = 10 with 74 around postage stamp stars and 99 in the FFIs. In the postage stamps, the detected planets are split equally 118/124 among cool (Teff < 4000 K) and warm (Teff ≥ 4000 K) stars, but the FFIs favor the warm stars with a 93/883 split be- tween the cool/warm stars. We also find that 196 sub- Neptunes with rp < 4R⊕ will be detected as single- transits in the postage stamps, with an additional 230 detected in the FFIs of CTL-6 stars. All of the planets detected around cool stars have rp < 4R⊕ as there are no planets above 4R⊕ in the Dressing & Charbonneau (2015) occurrence rates. Clanton & Gaudi (2016) show that giant planets do indeed exist around cool stars at long periods, but are uncommon relative to small plan- ets occurrence rates. Fifty-six planets will be detected in the postage stamps with a stellar insulation within a factor of two of the Earth (0.5 ≤ S/S⊙ ≤ 2), and another 73 in the FFIs. Using the conservative habitable zone defined by Kopparapu et al. (2013), we expect 34 habitable zone planets from the postage stamps, 29 are around cool 6 Villanueva Jr., Dragomir, & Gaudi Figure 2. Number of expected single transit events by magnitude, Teff , planet radius, stellar insulation relative to the Sun. Dotted gray line is the demographics of the CTL-6 normalized to fit on this scale. The solid black and gray histograms are the total yield from the postage stamps and FFIs respectively. These are subdivided into Teff ≥ 4000 K (blue) and Teff < 4000 K (red) star samples with the darker shades for postage stamps and the lighter shades for the FFIs. stars, and another 45 habitable zone planets from the FFIs with 32 coming from cool stars. If we limit the hab- itable zone planets to terrestrial planets (R ≤ 1.5R⊕) then we expect only 1 planet to be detected, which hap- pens around a postage stamp star with Teff ≤ 4000 K. In a recent simulation of the TESS yield from Barclay et al. (2018), there were no planets found beyond ≈ 85 days which resulted in no planets being found in the habit- able zone of FGK stars, where we find 5 in the postage stamps and 13 in the FFIs. It is worth noting that we differ from Barclay et al. (2018) not only in the number of habitable zone planets around stars earlier than M, but also in our target star selection criteria, extrapola- tion of planet occurrence rates, and in the definition of habitable zone. We adopt the same SNR threshold SN R = 7.3 as used by both Sullivan et al. (2015) and Bouma et al. (2017) for multiple-transiting events. Given the added uncer- tainty of single-transit events (e.g. false positives) we follow Barclay et al. (2018) and also look at the dis- tribution of SNR for all of the detections in Figure 3 for those wishing to adopt a more stringent SNR cut. Single Transits in TESS where √δ is the transit depth to arrive at: √δ (cid:19)3/2 ρ∗(cid:18) Tdurτ Gπ2 P = 3 7 (11) This leads to a degeneracy between the period and the host star density. Seager & Malle'n-Ornelas (2003) showed that it is possible to estimate the period when the mass and radius of the host star is known. Yee & Gaudi (2008) show that the fractional uncer- tainty in the period (σP /P ) will come from the fractional uncertainty in the density (σρ/ρ) and the fractional un- certainty on the period due to the TESS photometry (σP /P )TESS added in quadrature: (cid:16) σP P (cid:17)2 ρ (cid:19)2 = (cid:18) σρ +(cid:16) σP P (cid:17)2 TESS (12) 3.1. Uncertainty on the Period Due to the Stellar Density In order to place a constraint on the period, one also needs a constraint on the density of the host star. There are a few avenues that allow for an independent con- straint on the density to better than 10%, such that the photometry is the limiting factor of estimating the period of single-transit planets in TESS. One method to constrain the density will be to first estimate the stellar radius R⋆ of the host star by combin- ing a spectral energy distribution (SED) compiled from broadband photometric measurements or spectropho- tometry (when available) with an estimate of the ef- fective temperature Teff of the star. The effective tem- perature can be obtained from the SED itself or from high-resolution spectra of the host star. By fitting the SED to a model stellar atmosphere, one can estimate the dereddened bolometric flux F⋆ and the extinction (adopting an extinction law). With an estimate of F⋆ and Teff, one can then estimate the stellar angular diam- eter and thus physical radius of the star using a parallax π from Gaia. The exact precision on R⋆ will depend on the quality of the parallax, SED, and spectra. With the radius and the surface gravity of the host star log g, the mass and therefore the density of the host star can be estimated. There are a variety of ways one can estimate log g. One can measure this quan- tity using gravity-sensitive lines in high-resolution spec- tra, although such spectroscopic estimates of log g can be relatively imprecise, and more importantly, inaccu- rate, particularly in some regions of parameter space. In some cases, granulation-based 'flicker' measurements can be used to obtain a more precise estimate of log g (Bastien et al. 2013), however this requires both high quality and relatively long-baseline photometry. Figure 3. The number of single transit detections by SNR with a SNR threshold of SN R = 7.3. Colors are the same as in Figure 2. 162/241 postage stamp star detections have SN R ≥ 10, and 14/241 have SN R ≥ 100. 695/977 detec- tions around FFI stars have SN R ≥ 10, and 90/977 have SN R ≥ 100. We find that of the 241 postage stamp detections with SN R ≥ 7.3, 162 have SN R ≥ 10, and 14 have robust detections at SN R ≥ 100. Among the 977 detections around FFI stars with SN R ≥ 7.3, 695 have SN R ≥ 10, and 90 have SN R ≥ 100. 3. ESTIMATING THE PERIOD An important aspect of identifying the single-transit candidates is the ability to predict the time of future transits to confirm their ephemerides and schedule fu- ture observations. One can relate the observed proper- ties of the light curve, the velocity of the planet assum- ing a circular orbit and zero impact parameter (b = 0), and Kepler's third law (Eq. 3) to relate the period, stellar density, and the light curve observable quanti- ties (Seager & Malle'n-Ornelas 2003; Yee & Gaudi 2008; Winn 2010). Beginning with the velocity of the planet vp = 2πa P = 2R⋆ Tdur,0 , (8) where Tdur,0 is the duration of the transit at b = 0. One can relate Tdur,0 to the observable quantities Tdur and τ , the measured duration of the transit and ingress/egress time, with Tdur = Tdur,0p1 − b2 τ = √δTdur,0 √1 − b2 , (9) (10) 8 Villanueva Jr., Dragomir, & Gaudi In some cases, it may also be possible to constrain the density from astroseismology. Kjeldsen & Bedding (1995) show that the density scales with the measurable average large-frequency spacing h∆νi as ρ⋆ ∝ h∆νi2. In this case, only a 5% measurement of h∆νi is required to place a constraint on the density to 10%. As with flicker measurements, this requires high quality and relatively long baseline photometry. For most cases, however, we expect that one will fit the radius determined as above, along with metallic- ity [Fe/H], log g and effective temperature from high- resolution spectra, to stellar isochrones, to determine precise (albeit model-dependent) estimates of the age, mass, and density of the star. The precision with which the density can be estimated for the star will ultimately depend on which method is used, and the quality of the data being used. We will simply adopt a fiducial value of a 10% precision on ρ⋆, but note that this may be optimistic in some cases. In the event that the single-transit planet is part of a multi-planet system with the host star hosting addi- tional interior planets, it will be possible to obtain the density of the host star from transits of the inner plan- ets if any of the inner planets have multiple transits detected and if there is an estimate of their eccentricity. From the period of the inner planets the density can be taken directly from Equation 11 and applied as a con- straint on the single-transit planet. Ballard (2018) has shown that it is likely that multiple planet systems will be common in TESS, although many will be detected as single-planet systems because the additional planets may go undetected due to lack of SNR or because they only exhibit a single transit. In the end, the exact precision will be determined by which observables are available, and their relative pre- cision. The density constraints of individual stars will likely vary by orders of magnitude, but a 10% precision is expected for many of the brighter, well characterized systems. 3.2. Uncertainty on the Period Due to the Photometry The fractional uncertainty in the period due to the photometry (σP /P )TESS is dominated by the ability to measure the ingress/egress time τ . From Equation 9 in Yee & Gaudi (2008), we get that the fractional uncer- tainty in the period due to the photometric precision is (cid:16) σP P (cid:17)2 TESS ≈ 9 4 (cid:16) στ τ (cid:17)2 ≈ 1 Q2 (cid:18) 27Tdur 2τ (cid:19) , (13) Figure 4. Number of expected single transit events by pho- tometric uncertainty on the period. Colors are the same as in Figure 2. The separate bin to the far right are the objects where the ingress/egress time was shorter than the exposure time and therefore only a upper limit can be place on the ingress/egress time. the ingress/egress time τ , assuming τ ≪ Tdur. A de- tailed investigation in to the details of the uncertainties in the observables can be found in Carter et al. (2008) and the Appendix of Yee & Gaudi (2008). In Figure 4 we show the fractional uncertainty ex- pected from single transits based on their photometry. 146/1218 planets will have a fractional uncertainty on the period to better than 10%, with 16 coming from the postage stamps and 130 of those coming from the FFIs. It is worth noting that even in the event of a 1% constraint on the period from photometry, the uncer- tainty on the density will likely dominate and limit the constraint on the inferred period. Another 72 planets will have a fractional uncertainty on the period of 10- 15%, with 5 around postage stamp stars and 72 coming from the FFIs. For cases where the ingress/egress time is shorter than the exposure time, we can only place an upper-limit on the ingress/egress time, and therefore lose the ability to constrain the period. This happens with 373 planets, all identified in the 30-minute cadence FFIs. The remaining 627 planets all have fractional un- certainties of greater than 15%, where the approxima- tion in Equation 13 breaks down and follow-up becomes difficult. and can be related to Q, the approximate total SNR of the transit, and the ratio of the transit duration Tdur to 3.3. Uncertainty on the Period Due to Eccentricity Single Transits in TESS 9 Up until now, we have assumed circular orbits for all of the estimates. In reality a number of these planets will likely have non-circular orbits. This will change the duration of the transit and will lead to an incorrect estimation of the period of the planet. Yee & Gaudi (2008) show that the maximum and minimum deviation from the true period is given as: (cid:18) ∆P P (cid:19)min/max = (cid:18) 1 + e 1 − e(cid:19)±3/2 (14) We show this in Figure 5 where the minimum and maximum range of true periods of a planet with e = 0.1 would be in the range of a factor of 0.74–1.35 of the assumed circular period. Assuming a median eccentric- ity of e = 0.17 from the Beta distribution described in Kipping (2013), we get a typical range of periods from 0.59–1.69, or a ≈ 50% uncertainty in the period. This would imply that for non-circular systems, the e 6= 0 uncertainties will limit over our ability to place a con- straint using either density or the photometry. How- ever, many of these systems will be in multiple planet systems (Ballard 2018). Zhu et al. (2018) showed that Kepler planet systems become dynamically cooler as the number of planets in the system increases. This follows the results from Xie et al. (2016) that found the mean eccentricity of Kepler multi-planet systems of e = 0.04 to be much lower than that of the single-planet systems e = 0.3. For an eccentricity of e = 0.04, the range of possible periods relative to circular is only 0.89–0.12, which is in line with the expected uncertainty from the density and photometry. 4. PROSPECTS FOR FOLLOW-UP 4.1. Recovery with Additional Photometry or Precovery in Archival Data After estimating the timing of a future transit, we also need to consider which of the single transit candidates will be observable from a typical ground based facility. An additional resource is to look for signals present in existing data sets given a known depth and approximate period. We present the distribution of the undiluted transit depths of single-transits in Figure 6. 197/241 planets will be detectable at δ ≥ 0.1% in the postage stamps, with 894/976 of planets detectable around stars in the FFIs. Of these, 40 planets around postage stamp stars will have deep δ ≥ 1% transit depths, and 253 planets around stars in the FFIs. 4.2. Expected Radial Velocity Signal To estimate the expected radial velocity semi- amplitude K, we first assign a mass to each planet based Figure 5. The maximum and minimum deviation from the true period under the assumption of a circular orbit. The spread is due to the change in planet velocity and transit duration when the planet transits at periastron versus apas- tron. Figure 6. Expected single transit transit depths. Colors are the same as in Figure 2. Of planets detected, 90% and 24% of the planets will have transit depths deeper than 0.1% and 1% respectively. on its radius. For planets with radii < 4.0 R⊕ we use the planetary mass-radius relations from Weiss & Marcy (2014) and for planets with radii ≥ 4.0 R⊕ we use the planetary mass-radius relations from Mordasini et al. 10 Villanueva Jr., Dragomir, & Gaudi Figure 7. Expected single transit transit radial velocity signals. Colors are the same as in Figure 2. Of planets detected, 98% and 46% of the planets will have RV signals greater than 1 and 10 m/s respectively. (2012). After assigning a planetary mass mp, we use the following equation to assign the radial velocity semi- amplitude K: K = mp (mp + M⋆)2/3 (cid:18) P 2πG(cid:19)−1/3 (15) The distribution of expected RV signals from the single transits is shown in Figure 7. The majority of single- transit planets 1195/1218, will have RV semi-amplitudes detectable by modern RV instruments K ≥ 1 m/s. 556/1218 will have K ≥ 10 m/s and 20 will have K ≥ 100 m/s. 5. COMPARISON TO OTHER SIMULATIONS As no one has published yields from single transits from TESS simulations, we present the expected yield of the TESS mission proper (i.e. detected two or more transits) for our analysis as a way to compare and scale our results to previous studies. We perform the same analysis described in Section 2 to provide an updated estimate of the yield of the primary TESS mission while only considering the number of planets detected with two or more transits at a SN R ≥ 7.3. We find 2114 planets detected in the postage stamps and another 5130 in the FFIs around stars in the CTL-6. These can be seen in Figure 8. Again, these numbers are incomplete fainter than T > 12 and are lower limits for the FFIs. 255/2114 have periods > 25 days, and only 2/2114 have periods > 250 days in the postage stamps, while and Figure 8. Expected yield from the TESS mission. Colors are the same as in Figure 2. Single transit distribution from Figure 1 are shown in dashed lines. Single transits in the postage stamps match the TESS mission yield for planets with 25 ≤ P ≤ 250 days, and dominate relative to the ex- pected yield in the FFIs. Nearly all planets detected with P ≥ 250 come from the single-transits in both the postage stamps and FFIs. 211/5130 have periods > 25 days and < 1/5130 have periods > 250 days in the FFIs. We also show the yield from single transits as the dashed lines in Figure8 for the postage stamps (black) and FFIs (gray). We find that within postage stamps the single transits match the TESS mission yield beyond ≈ 25 days, and domi- nate relative to the expected yield in the FFIs. Beyond ≈ 250 days nearly all detections come from the single- transits in both the postage stamps and FFIs. We find that our predicted yield from the postage stamps of 2114 represents a 20-25% increase over the yield of both the Sullivan et al. (2015) (1734) and Bouma et al. (2017) (1670) estimates using a galac- tic model, and a 70% increases over the more re- alistic Barclay et al. (2018) (1250) simulation. The Sullivan et al. (2015) yield predicts >20,000 planets de- tected in the FFIs, where we only detect 5130 planets in the FFIs, noting that we are incomplete and this is only a lower limit. Bouma et al. (2017) provides a lower limit of 3342 for the FFIs, which is consistent with our estimate. Barclay et al. (2018) found 1250 planets in the postage stamps, with another 3200 planets in the FFIs using the CTL-6 and TIC-6, with another 10,000 planets around stars faint stars not included in the CTL. We find that in general we over-estimate the number of Single Transits in TESS 11 planets in the postage stamps relative to other simula- tions, but are more consistent for the FFIs. It is worth noting that we have extrapolated our planet occurrence rates to much longer periods than all of the above simu- lations, and have a more simplistic target star selection criteria. We were able to obtain rough numbers for the esti- mated number of single transit events from individual trials of other simulations via private communication. Again, we find 977 single-transits in the FFIs and 241 in the postage stamps in our work. From one trial from the Sullivan et al. (2015) simulation, we estimate ≈ 1300 single-transits in the FFIs and ≈ 100 single-transits in the postage stamps (P. Sullivan, private communica- tion). From one trial from the Bouma et al. (2017) sim- ulation, we estimate ≈ 850 single-transits in the FFIs and ≈ 150 single-transits in the postage stamps (L. Bouman, private communication). Although unpub- lished, we find that the total number of single-transit events from each simulation, including our work, varies from 1000–1400 with each group disagreeing on the rel- ative fraction found in the postage stamps versus the FFIs. 6. RECOMMENDATIONS FOR OBSERVATIONS Given that nearly all of the 1218 single-transit events detectable in TESS will have either photometric (90%) or radial velocity (98%) signals measurable from current ground-based observatories, there will be more planets detected than could possibly be followed-up. Follow- up observers should coordinate to prioritize which plan- ets will be targeted for follow-up observations, either by their ability to constrain the period, signal-to-noise ra- tio, or scientific merit. With 98% of planets detectable in RV, that the RV measurements are required to de- termine the mass and planetary nature of the plan- ets, and that RV measurements can help constrain the period, RV resources should be immediately allocated towards confirming single-transit events. RV will be crucial to the single transit detections in single-planet systems with poorly constrained eccentricities. Addi- tionally, for those with photometric signals, searches in archival data and planned observations around the pre- dicted next transit can be used to determine the period and constrain the timing of future transits. 7. CONCLUSION The number of single-transit planets from TESS is ex- pected to be an order of magnitude greater than those found in Kepler, with 241 single-transit planets detected in the postage stamps, and another 977 detected from the FFIs around stars brighter than T = 12. Single tran- sits require greater follow-up resources than the typical TESS planet, and there will be more single-transit plan- ets signals than follow-up resources will be able to ob- serve or confirm. This is despite the fact that 90% and 98% of all such planets detected will have photometric and RV signals respectively that will be observable from current ground-based observatories. It is possible to predict future transits of single- transits by placing constraints on the light curve observ- ables and on the density of the stellar host. The uncer- tainties from the density of the host star will be ≈ 10% in many cases, however only 10% of the planets (146) will have photometry sufficient to provide constraints on the period to better than a 10% due to uncertainties on the photometry assuming circular orbits. The uncer- tainty due to eccentric orbits will make constraining the true period difficult, but multi-planet systems represent the best systems to place constraints on both the stellar density and eccentricity. Our single-transit yields predict a 80% increase in the number of planets detected beyond 25 days compared to the TESS mission, and a factor of 12 increase in the yield for planets beyond 250 days. This includes 79 habitable zone planets and ∼ 1 terrestrial planet in the habitable zone. This opportunity to substantially augment the yield of the TESS mission should not be overlooked. However, given the"abundance of riches" represented by these single-transit events, we recommend community collaboration to make the most of these opportunities. 8. ACKNOWLEDGEMENTS We would like to thank Chelsea X. Huang for her feedback, helpful advice, and for sharing her yields with our group. We would like to thank Tom Barclay, Luke Bouma and Peter Sullivan for sharing their yields with our group. We would like to thank Daniel Stevens for his insight and feedback. Work by S.V.Jr. is supported by the David G. Price Fellowship for Astronomical Instrumentation and by the National Science Foundation Graduate Research Fellow- ship under Grant No. DGE-1343012. D. Dragomir ac- knowledges support provided by NASA through Hubble Fellowship grant HST-HF2-51372.001-A awarded by the Space Telescope Science Institute, which is operated by the Association of Universities for Research in Astron- omy, Inc., for NASA, under contract NAS5-26555. Work by B.S.G. is supported by National Science Foundation CAREER Grant AST-1056524. 12 Villanueva Jr., Dragomir, & Gaudi REFERENCES Baglin, A. 2003, AdSpR, 31, 345B Bakos, G., Noyes, R. W., Kov´acs, G., et al. 2004, PASP, Koch, D. G., Borucki, W. J., Basri, G., et al. 2010, ApJL, 713, L79-L86 116, 266B Kopparapu, R. K., Ramirez, R., Kasting, J. F., et al. 2013, Barclay, T., Pepper, J., & Quintana, E. V. 2018, ApJ, 765, 131K arXiv:1804.05050 Mordasini, C., Alibert, Y., Georgy, C., et al. 2012, A&A, Ballard, S., 2018, arXiv:1801.04949 Basri, G., Walkowicz, L. M., Reiners, A., 2013, ApJ, 769, 37B 547A, 112M Pollacco, D. L., Skillen, I., Collier Cameron, A., et al. 2006, PASP, 118, 1407P Bastien, F. A., Stassun, K. G., Basri, G., & Pepper, J. Ricker, G. R., Winn, J. N., Vanderspek, R., et al. 2015, 2013, Nature, 500, 427 Borucki, W. J., Koch, D., Basri, G., et al. 2010, Sci, 327, 977B JATIS, 1, a4003R Seager, S. & Malle'n-Ornelas, G., 2003, ApJ, 585, 1038S Siverd, R. J., Beatty, T. G., Pepper, J., et al. 2012, ApJ, Bouma, L. G., Winn, J. N., Kosiarek, J., & McCullough, P. 61, 123S R., 2017, arXiv:1705.08891 Sullivan, P. W., Winn, J. N., Berta-Thompson, Z. K., et al. Carter, J. A., Yee, J. C., Eastman, J., et al. 2008, ApJ, 689, 2015, ApJ, 809, 77S 499C Clanton, C. & Gaudi, B. S. 2016, ApJ, 819, 125C Dressing, C. D. & Charbonneau, D., 2015, ApJ, 807, 45D Foreman-Mackey, D., Morton, T. D., Hogg, D. W., et al. 2016, AJ, 152, 206F Fressin, F., Torres, G., Charbonneau, D., 2013, ApJ, 766, 81F Howell, S. B., Sobeck, C., Haas, M., et al. 2014, PASP, 126, 398H Kipping, D. 2013, MNRAS, 434L, 51K Kjeldsen H. & Bedding T. R. 1995, A&A, 293, 87K Stassun, K. G., Oelkers, R. J., Pepper, J., et al. 2017, arxiv:1706.00495 Torres, G., Andersen, J., & Gim´enez, A. 2010, A&ARv, 18, 67T Weiss, L. M. & Marcy, G. W. 2014, ApJ, 783L, 6W Winn, J., 2010, arXiv:1001.2010 Xie, J., Dong, S., Zhu, Z., et al. 2016, PNAS, 11311431X Yee, J. C. & Gaudi, B. S. 2008, ApJ, 688, 616Y Zhu, W., Petrovich, C., Wu, Y., et al. 2018, arXiv:1802.09526
1903.03041	1	1903	2019-03-07T16:58:35	Studying the Evolution of Warm Dust Encircling BD +20 307 Using SOFIA	[ "astro-ph.EP" ]	The small class of known stars with unusually warm, dusty debris disks is a key sample to probe in order to understand cascade models and extreme collisions that likely lead to the final configurations of planetary systems. Because of its extreme dustiness and small radius, the disk of BD +20 307 has a short predicted collision time and is therefore an interesting target in which to look for changes in dust quantity and composition over time. To compare with previous ground and Spitzer Space Telescope data, SOFIA photometry and spectroscopy were obtained. The system's 8.8-12.5 $\mu$m infrared emission increased by $10 \pm 2 \%$ over nine years between the SOFIA and earlier Spitzer measurements. In addition to an overall increase in infrared excess, there is a suggestion of a greater increase in flux at shorter wavelengths (less than 10.6 $\mu$m) compared to longer wavelengths (greater than 10.6 $\mu$m). Steady-state collisional cascade models cannot explain the increase in BD +20 307's disk flux over such short timescales. A catastrophic collision between planetary-scale bodies is still the most likely origin for the system's extreme dust; however, the cause for its recent variation requires further investigation.	astro-ph.EP	astro-ph	Draft version March 8, 2019 Typeset using LATEX default style in AASTeX61 STUDYING THE EVOLUTION OF WARM DUST ENCIRCLING BD +20 307 USING SOFIA Maggie A. Thompson,1 Alycia J. Weinberger,2 Luke Keller,3 Jessica A. Arnold,2 and Christopher Stark4 1Department of Astronomy and Astrophysics, University of California, Santa Cruz, CA 95064 2Department of Terrestrial Magnetism, Carnegie Institution of Washington, Washington, D.C., 20008 3Department of Physics and Astronomy, Ithaca College, Ithaca, N.Y. 4Space Telescope Science Institute, Baltimore, M.D. ABSTRACT The small class of known stars with unusually warm, dusty debris disks is a key sample to probe in order to understand cascade models and extreme collisions that likely lead to the final configurations of planetary systems. Because of its extreme dustiness and small radius, the disk of BD +20 307 has a short predicted collision time and is therefore an interesting target in which to look for changes in dust quantity and composition over time. To compare with previous ground and Spitzer Space Telescope data, SOFIA photometry and spectroscopy were obtained. The system's 8.8-12.5 µm infrared emission increased by 10 ± 2% over nine years between the SOFIA and earlier Spitzer measurements. In addition to an overall increase in infrared excess, there is a suggestion of a greater increase in flux at shorter wavelengths (less than 10.6 µm) compared to longer wavelengths (greater than 10.6 µm). Steady-state collisional cascade models cannot explain the increase in BD +20 307's disk flux over such short timescales. A catastrophic collision between planetary-scale bodies is still the most likely origin for the system's extreme dust; however, the cause for its recent variation requires further investigation. 9 1 0 2 r a M 7 . ] P E h p - o r t s a [ 1 v 1 4 0 3 0 . 3 0 9 1 : v i X r a [email protected], [email protected] 2 1. INTRODUCTION Typical debris disks include leftover planetesimals and are thought to evolve through collisions or evaporation of solid bodies, ranging from small planetesimal to protoplanet/planet-sized (Krivov 2010; Wyatt 2008). Just like the Solar System's Kuiper Belt located beyond the orbit of Neptune, most debris disks contain low-temperature dust (≤ 100 K) orbiting far from the host star. However, there exists a small class of known stars with unusually warm, dusty debris disks that serve as a key sample to probe in order to understand cascade models and extreme collisions that likely lead to the final configurations of planetary systems. Collisional cascade, a process in which larger planetesimals in the disk collide and are continually broken up into smaller objects can explain most debris disks. In a collisional cascade, small debris disks with warm dust do not last for very long because once the dust has reached a small enough size, removal mechanisms operate quickly, such as radiation pressure that blows the dust out of the system or Poynting-Robertson drag that causes dust particles to fall into the star (Wyatt 2008). Compositional changes in warm, dusty disks may be observable on extremely short timescales of years. As a case in point, observations of TYC 8241 2652 1, a young, Sun-like star with a warm, dusty disk, saw a decrease in the disk's dust emission by a factor of 30 in less than two years, and there is currently no physical model that can explain what would cause such rapid dust depletion (Melis et al. 2012). Similarly, Meng et al. (2015) used Spitzer to study five debris disks with unusually high fractional luminosities and found major variations on timescales shorter than a year. BD +20 307 can perhaps provide insight into extreme collisional events occurring beyond our Solar System because its warm dust makes it the dustiest star known for its age of ≥1 Gyr. The system is a tidally-locked spectroscopic binary composed of nearly identical late-F-type dwarf stars orbiting with a short 3.4-day period, located at 120.0 ± 0.7 pc from Earth (Weinberger 2008; Fekel et al. 2012; Gaia Collaboration et al. 2018). Two previous instruments obtained infrared spectra for BD +20 307: the first in 2004-2005 using Keck and Gemini at 8 -- 13 µm (Song et al. 2005); the second in 2005-2007 using Spitzer at 5 -- 37 µm (Weinberger et al. 2011). In addition, long wavelength photometry was collected with Spitzer and in 2011-2012 using Herschel (Vican et al. 2016). About a decade later in 2015, we obtained new 8 -- 13 µm infrared spectra using the Stratospheric Observatory for Infrared Astronomy (SOFIA), which is the focus of this paper. For BD +20 307's disk, assuming its LIR/L(cid:63) = 0.032 represents the surface density of the dust grains, we expect a collision time of only 2.4 years. In addition, since small grains (≤1 µm) are likely created by collisions, radiation pressure should blow them away on an orbital timescale. Given the old age of the system, collisional cascade cannot explain the large amount of observed dust over the lifetime of BD +20 307 (Wyatt et al. 2007). Dust at this flux level cannot last over the system's age, so the dust must be transient. Therefore, another explanation is needed for the cause of the large amount of dust around such a mature system. In addition, it is important to note that, based on the spectral energy distribution (SED), Weinberger et al. (2011) and Vican et al. (2016) conclude that there is no cold dust emitting at far-infrared wavelengths and that the mean temperature of the dust is ∼420 K, which suggests an extreme planetary-scale collision within 1 AU to explain BD +20 307's debris disk. In this paper, we report on the changes in BD +20 307's debris dust by analyzing its spectrum from SOFIA and comparing it to the previous spectra from Keck/Gemini and Spitzer. In Section 2, we summarize previous observations. In Section 3, we describe our SOFIA observations of BD +20 307 and the analysis of its new spectra. We present our results in Section 4, which suggest that we detect significant differences between the SOFIA spectrum and those earlier spectra from Spitzer and the ground-based instruments. In Section 5, we discuss possible mechanisms to explain the observed changes in this extremely warm and dusty debris disk over a decadal timescale. Finally, in Section 6 we summarize our work and propose additional avenues for further analysis of BD +20 307's unusual debris disk system. 2. PREVIOUS OBSERVATIONS The first detailed set of observations to study the debris surrounding BD +20 307 were obtained using two ground- based telescopes: the W.M. Keck Observatory and the Gemini-North Telescope, both located on Mauna Kea in Hawaii. Using the Keck Long Wavelength Spectrometer (LWS) and the Gemini Michelle instrument, Song et al. (2005) obtained spectra of BD +20 307 taken over the course of three months from August to October of 2004. The resolving powers of the Keck LWS and the Gemini Michelle instrument are R≈150 and ≈900, respectively. Song et al. (2005) assumed that there was no change in flux over the time period of their observations. A year later, the Spitzer Space Telescope gathered the second detailed set of spectroscopy and photometry observa- tions on BD +20 307 (Weinberger et al. 2011). On August 20, 2005, Spitzer's Infrared Array Camera (IRAC) gathered short wavelength photometry, and on January 15, 2006, the Infrared Spectrometer (IRS) obtained spectra of BD +20 3 307. About a year later, on January 21, 2007, the Multiband Imaging Photometer for SIRTF (MIPS) gathered long wavelength photometry of BD +20 307. The IRS spectral resolution from 8 to 13 µm was R≈120-600. Weinberger et al. (2011) also made the assumption that there were no changes in the flux or spectrum despite the fact that the observations spanned August 2005 to January 2007. We re-normalized the Spitzer IRS spectrum to the IRAC 5.7 µm filter instead of to the MIPS 24 µm filter as in Weinberger et al. (2011) because the IRAC photometry was gathered closer in time to the IRS spectrum. There are slight differences in the Spitzer IRS spectrum normalized to the IRAC flux compared to the MIPS flux (see Section 2.3 of Weinberger et al. (2011)). By averaging them over SOFIA's 11.1 µm filter transmission curve, we found that they differ by 6.8%. In the ∼ 1.5 years between the Keck/Gemini and Spitzer observations, there is indication that the flux increased slightly over time, particularly at shorter wavelengths. The ratio of the weighted average flux values between the Spitzer and Keck/Gemini spectra for wavelengths less than 10.6 µm is 1.14 ± 0.002. It is important to note that we have taken the as-published Keck/Gemini data, and there may be additional calibration uncertainties that were accounted for. The uncertainty we have quoted should be considered as a lower bound. Figure 1. WISE photometry observations of BD +20 307 taken from January 2010 to January 2011. Observations in W3 and W4 only span about 200 days whereas those in W1 and W2 cover the entire year of observations. In January and July 2010, during the cryogenic mission, WISE measured BD+20 307 at all four of its bands (3.5 - 22 µm), and in January 2011, it did a scan in the two short wavelength bands. Meng et al. (2012) noted that BD +20 307's disk may vary by ∼ 7% between the two cryogenic scans taken ∼188 days apart but only in the W3 and W4 bands (12 and 22 µm, respectively). We show the WISE photometry at each epoch in Figure 1, and find, in agreement with Meng et al., that all four bands show the same trend, i.e. a small brightening over time, although W1 and W2 are constant within their uncertainties. In July 2011 and January 2012, the Herschel Space Observatory used its Photodetector Array Camera and Spec- trometer (PACS) to measure BD +20 307 at 70 and 100 µm. The reported 70 µm flux density of 47 ± 3 mJy (Vican et al. 2016) is substantially larger than that reported with MIPS from 2007 of 28.6 ± 1.9 mJy (Weinberger et al. 2011). Most recently, from 2012 to 2013, Meng et al. (2015) made near-infrared time-series observations of five extreme debris disks including BD +20 307 using Spitzer's IRAC at 3.6 and 4.5 µm. Although they did not detect a significant trend at 3.6 µm, at 4.5 µm they found that BD +20 307's disk flux increased over their period of observations with an average increase rate of 2.5 ± 0.7 mJy per year. Such an increase rate coincides within a few percent of the disk flux 4 every year (Meng et al. 2015). Table 1 summarizes the main sets of previous infrared observations on BD +20 307's dust. Observatory Instrument Wavelengths (µm) Dates Observation Type Keck LWS Gemini-North Michelle Spitzer Spitzer Spitzer WISE Herschel Spitzer IRAC IRS MIPS PACS IRAC 3.9-24.5 7.7-18.1 3.5-7.9 5.2-37.2 24, 70, 160 08/29/2004 Photometry & Spectroscopy 09-10/2004 Photometry & Spectroscopy 08/20/2005 01/15/2006 02/04/2007 Photometry Spectroscopy Photometry 3.4, 4.6, 12, 22 01/19/2010-01/20/2011 Photometry 70, 100, 160 07/2011, 01/2012 3.6, 4.5 2012-2013 Photometry Photometry Table 1. Summary of previous observations of BD +20 307's dust. 3. METHODS 3.1. SOFIA Observations For the SOFIA Cycle 2 Program 02 0050 (PI: Weinberger), we used SOFIA's Faint Object Infrared Camera (FOR- CAST), a dual-channel mid-infrared camera and spectrograph sensitive to 5-40 µm. Observations of BD +20 307 were obtained in 2015 flying at an average altitude of 40,000 feet. Tables 2 and 3 below summarize the SOFIA observations for gathering spectroscopy and imaging data of BD +20 307. Using the 'Nod-Match-Chop' grism observing mode with a chopper throw of 60(cid:48)(cid:48), data were taken with the 'G111' grism that has a nominal resolving power of R=130 for the 4.7" x 191" slit and covers 8.4-13.7 µm. Instrument Date Altitude Range (ft) Integration Time (s) FORCAST 'G111' 2015 Feb 04 39,953-41,014 2015 Feb 06 39,032-39,031 486 754 Table 2. Spectroscopy observations with SOFIA's FORCAST instrument using the 'G111' Grism in the Short Wavelength Channel. A total of 55 individual spectra were taken over two nights, 23 on February 4 and 32 on February 6. We analyzed the products produced by the standard 'FORCAST REDUX' pipeline version 1.2.0. The general pipeline process is as follows: load the data and calculate variance; clean bad pixels; correct for applied channel suppression (i.e., droop); correct for image non-linearity; perform background subtraction; remove jailbars; perform spectral extraction using FSpextool algorithm (for spectra only); combine multiple observations (i.e., stacking); calibrate flux. It is important to note that there is a large amount of noise in the data between 9.5-9.9 µm caused by telluric ozone absorption. The spectrum from each night is available from the SOFIA Data Cycle System as a FITS file that contains the combined spectrum corrected for atmospheric transmission and instrumental response. To combine the spectra taken on two different nights, we calculated the mean flux value between 10 and 13 µm for each of the spectra and normalized the first spectrum to the second spectrum's average flux value in that wavelength region. We then combined the two spectra via a weighted average in which we weighted each spectral channel by the reciprocal of the squared uncertainty in the flux. In order to determine the variation in the total flux for the two spectra and to ensure that our method of normalization was sound, we averaged the flux over the 11.1 µm filter ('FOR F111') transmission curve to determine the total flux density from each of the spectra and the weighted average spectrum across the 11.1 µm filter bandpass. We found that the individual spectra taken two days apart differed from each other by 8.3%. In addition to the spectroscopy, target acquisition images were taken in the F111 filter (centered at 11.1 µm). We used four images out of the 20 taken on February 4. As with the spectral data, all of the images were processed using 5 Instrument Date Altitude (ft) Integration Time (s) FORCAST (Imaging SWC, 'F111') 2015 Feb 04 39955 2015 Feb 04 39953 2015 Feb 04 41013 2015 Feb 04 41010 15.61 15.61 15.73 15.73 Table 3. Photometry observations with SOFIA. Figure 2. Image of BD +20 307 taken by SOFIA with the red circle representing the 5-pixel aperture radius and the blue circles representing the 13-17 pixel sky annulus we used to perform the aperture photometry. 'FORCAST REDUX.' For the images, the pipeline includes the same steps mentioned above for spectroscopy but there are several additional steps, including correcting for distortion, merging the images (by shifting and rotating), and finally registering. The pixel scale for the images is 1"=1.3 pixels (or 1 pixel = 0.8"). To average the four images, we determined the centroid of each of the four images and chose to shift all of the other images to the image having the brightest centroid (i.e., highest signal-to-noise). We then averaged all four aligned images via a weighted average, once again weighting by the reciprocal of the squared uncertainty of the image (i.e., the reciprocal of the variance associated with each pixel in the image). 3.2. Photometric Calibration and Normalization of SOFIA Spectrum We photometrically calibrated the SOFIA spectrum by comparing the flux of BD +20 307 from the spectrum to that from the imaging photometry. We created several curve of growth plots of flux as a function of pixel radius from the center of the source in the image to determine the proper aperture radius and sky annulus that should be used for our aperture photometry calculation. We determined that an aperture radius of 5 pixels and a sky annulus from 13 to 17 pixels was appropriate (Figure 2). Performing aperture photometry, we found that our source had an average flux of 0.41±0.013 Me/s which is equivalent to 1.07±0.03 Jy. To determine exactly how much flux is missing in our 5-pixel aperture, we did aperture photometry of Alpha Boo, the calibration standard star used for our source. We compared photometry with the standard aperture radius and sky annulus used in the FORCAST pipeline (i.e., 12 pixel aperture radius, 15-25 pixel sky background region) (FORCAST Science Instrument Team 2016) and ours (5 pixel radius, 13-17 pixel sky annulus) and determined that our choice returns 83% of the total flux from a source. Therefore, to determine the total average flux of our source, we divided 0.41 Me/s by 0.83, which gives the total flux of our source, 0.497 Me/s. To convert this flux to units of Jansky (Jy), we multiply by the calibration factor determined by the FORCAST pipeline, 0.383 Me/s/Jy, which results in an average flux of 1.30 ± 0.052 Jy. 6 There are four values contributing to the uncertainty of the photometric calibration: first, the photon counting statistics (0.0127 Me/s); second, the fact that we are choosing a specific sky annulus for the aperture photometry (0.01 Me/s); third, the standard deviation of the ratios of flux from the eleven calibration images (0.0118 Me/s); and fourth, the calibration factor uncertainty which is given in the FITS header (6.5 × 10−5M e/s/Jy). We combined these in quadruture to get a total uncertainty of 0.052 Jy. Finally, we normalized the SOFIA spectrum to the photometry by computing its flux density over the 11.1 µm filter transmission curve and then making this equal to our photometric measurement. This resulted in multiplying the SOFIA spectrum by 1.24. To determine the total uncertainty of BD +20 307's SOFIA flux we combined in quadrature the uncertainty of its photometric flux described above and the uncertainty of its spectral flux (i.e., the uncertainty of each flux value for BD +20 307's SOFIA spectrum). 3.3. Photosphere Subtraction of the Average SOFIA Spectrum To model the stellar flux, we fit a Kurucz model atmosphere of effective temperature of 6000 K and surface gravity (log(g)) of 4.5 to the visible and near-infrared photometry of the star and then extrapolated it to the SOFIA wavelength range. Our photosphere subtraction routine assumes that BD +20 307's stellar spectrum can be treated as that of a single star. 3.4. Normalization and Photosphere Subtraction of Ground-Based Spectrum Before comparing the two earlier epochs of BD +20 307's infrared spectra to its more recent SOFIA spectrum, we first normalized and photosphere subtracted the ground-based spectrum obtained in 2004-2005 using Keck and Gemini. We normalized this spectrum to Keck's Long Wavelength Spectrometer SiC filter whose bandpass is 10.64-12.96 µm. From Song et al. (2005), BD +20 307's flux at the LWS SiC filter is 843 mJy, we also normalized the spectrum to this flux value. Then, we photosphere subtracted in the same way as for the SOFIA data, above. We also found a new uncertainty for the ground-based spectrum integrated over the SiC filter, taking into account the uncertainty in the photometry reported in Song et al. (2005), 0.039 Jy. Figure 3. Spectra of BD +20 307 from SOFIA (blue), Spitzer (green), and Keck/Gemini (red) (each have been normalized and photosphere-subtracted). We now analyze how the infrared spectrum of BD +20 307 has evolved over the course of ∼10 years. Figure 3 shows the spectra from earlier observations with Keck/Gemini and Spitzer along with the spectrum from SOFIA. It is important to note that the region between ∼9.4-9.9 µm is where telluric ozone absorption adds noise to the spectrum; 4. RESULTS 7 therefore, we exclude this region in our analysis. When comparing the SOFIA to the ground-based spectra, we exclude a slightly broader ozone region (∼9.4-10.0 µm) because the ozone region is less well-calibrated for the ground-based data. Figure 4 illustrates the ratio between SOFIA's flux and the Keck/Gemini flux (magenta) and SOFIA's flux and the Spitzer flux (green). This plot suggests that the flux from BD +20 307's dust has increased from 2004 to 2015. In addition to examining the average ratios between the SOFIA and Spitzer spectra we also compare the two by integrating under their spectra. By comparing the integral under SOFIA's spectrum to that of Spitzer's spectrum (over the same wavelength range), we find that BD +20 307's flux has increased by 10 ± 2% in the 8 years between their observations. To determine if there is a change in the shape of the spectrum over the last 10 years, we compared the weighted average flux ratio for both SOFIA/Spitzer and SOFIA/Keck-Gemini (green and magenta curves in Figure 4, respectively) in two wavelength regimes: below 10.6 µm and above 10.6 µm. When calculating these ratios, we only consider flux values with signal/noise ratios greater than 5. We chose 10.6 µm as the dividing wavelength because 10.6 µm is the wavelength that most clearly distinguishes between the spectral peaks due to crystalline and amorphous grains (see Fig 2 in Weinberger et al. (2011)). For wavelengths less than 10.6 µm, we find that the average ratio between the SOFIA and Spitzer spectra is 1.10±0.02, while for wavelengths greater than 10.6 µm, the average ratio is 1.06±0.02. Figure 5 shows the ratio between the SOFIA and Spitzer flux values (blue curve) and the average ratio values for shorter (magenta line) and longer wavelengths (orange line). In addition, by comparing the ratio of the integral under the spectrum in the two )SOF IA = 1.77± 0.52 wavelength ranges (less than and greater than 10.6 µm) for the SOFIA and Spitzer data: ( FShort FLong )Spitzer = 1.67± 0.01, we find that, while the data are suggestive, the uncertainties in the SOFIA spectrum vs. ( FShort FLong are too large to allow us to say if there has been a larger increase in flux at shorter wavelengths. Figure 4. Ratio between the SOFIA spectrum and the ground-based and Spitzer spectra. The Y = 1 line (blue) is what we would expect if there was no difference between SOFIA and the previous spectra. The average ratio from 8.8 to 12.5 µm (and excluding the ozone region 9.4-9.9 µm, which is slightly broader (∼9.4-10.0 µm) when comparing SOFIA and the ground-based data due to the higher uncertainty in the ground-based ozone calibration) between the SOFIA and Spitzer spectra is 1.08±0.01 and the average ratio between SOFIA and the ground-based data is 1.15±0.01, suggesting that the flux has increased over time. It is important to note that the uncertainty in these plots also includes the calibration uncertainty. By comparing the integral under SOFIA's spectrum to that of the ground-based spectrum (over the same wavelength range), we determine that BD +20 307's flux has increased by 29 ± 6% over ∼10 years. While this result suggests an even greater increase in total flux between SOFIA and the Keck/Gemini spectra compared to SOFIA and Spitzer, the ground-based spectra calibration is less reliable compared to the Spitzer calibration. For the ratio between the SOFIA and Keck/Gemini spectra observed 10 years apart, the same general trends are observed, and in fact the differences 8 Figure 5. Ratio of SOFIA and Spitzer spectra (blue), only including flux values with signal/noise ratios greater than 5. The magenta and orange solid lines represent the weighted average ratio for wavelengths less than 10.6 µm (and ignoring the ozone region from 9.4 to 9.9 µm) and wavelengths greater than 10.6 µm, respectively. The turquoise line (Y=1) is what we would expect if there was no change between the SOFIA and Spitzer spectra. By comparing the ratios of the SOFIA-to-Spitzer spectra in these two regions, it is suggestive of a greater increase in flux over time at shorter wavelengths. between the two wavelength regimes is even more pronounced. For wavelengths less than 10.6 µm, we found the average ratio is 1.25±0.02, and for wavelengths greater than 10.6 µm, the average ratio is 1.08±0.02. The reported uncertainties on these values are dominated by the uncertainties in the SOFIA data, and there are likely additional calibration uncertainties for the Keck/Gemini data that exceed the quoted errorbars. We recomputed BD +20 307's disk parameters using the updated stellar luminosity from Gaia (Gaia Collaboration et al. 2018). A fit to B and V photometry from Hipparcos, and J, H, and Ks photometry from 2MASS and using the new Gaia parallax of 8.334 ± 0.046 mas yields a luminosity of 2.9 L(cid:12). In computing the stellar parameters, we ignore interstellar extinction effects given that BD +20 307 is nearby and not located in the Galactic plane (Galactic latitude = -39o). In addition, according to Schlafly & Finkbeiner, the extinction in the direction of BD +20 307 is small (Schlafly & Finkbeiner 2011). We assume each star contributed half of this luminosity, has a Tef f of 5900 K and [Fe/H]=-0.43 (Zuckerman et al. 2008), and an age of ∼1 Gyr (Weinberger 2008). These constraints correspond to a mass of ∼0.9 - 1 M(cid:12) on the Dartmouth evolutionary tracks (Dotter et al. 2008). We take the total mass to be 1.9 M(cid:12). At the updated luminosity, the dust is located at 1.04 AU. Otherwise, we use the dust model from Weinberger et al. (2011) to consider how the disk may have changed over time. That model assumes the dust to be at a single distance from the stars and made up of four compositions: amorphous olivine, crystalline olivine, amorphous pyroxene and large blackbody grains (see Section 3.2 in Weinberger et al. (2011) for more details). 5. DISCUSSION Our analysis reveals that between ∼8.8-12.5 µm BD +20 307's disk dust flux increased by ∼ 10 ± 2% over the eight years between the Spitzer and SOFIA measurements and ∼ 29 ± 6% over the ∼10 years between the Keck/Gemini and SOFIA observations. If we assume the disk is optically thin and all of the dust grains have the same size (0.5 µm) and opacity, then to get a 10% increase in luminosity between the SOFIA and Spitzer observations by just increasing the number of dust grains, we would need to introduce ∼ 9 ∗ 1032 more grains. If we imagine combining all of the dust grains into one spherical object, then the additional dust grains required to increase the luminosity by 10% would make up a 48 km radius body compared to the ∼110 km radius object that would contain enough mass for the grains required to get the LIR/L(cid:63) = 0.032 (Weinberger et al. 2011). It is important to acknowledge that this number of additional dust grains is a lower limit because we assume an optically thin disk. If, however, the disk were optically thick as some slightly younger debris disks can be, we would need to add even more dust grains to explain the change in flux. 9 Understanding how BD +20 307's dust flux can be increasing on such short timescales requires that we first discuss why this system is so unusual and does not conform to typical debris disk evolution models. Given the mature age (≥ 1 Gyr) of BD +20 307, it is extremely unusual for the system to have such copious amounts of warm dust (LIR/L(cid:63) = 0.032) within ∼1 AU. The most likely explanation for the origin of extremely dusty debris disks is an extreme collision between planetary-scale bodies. After such a catastrophic collision, the general assumption for dust evolution is a steady-state collisional cascade, a process by which planetesimals collide and break into smaller objects which in turn collide and continuously grind down to smaller and smaller fragments (e.g., Thebault et al. (2003), Dominik & Decin (2003), Wyatt et al. (2007)). Once objects break apart to small enough sizes, they will be removed either by radiation pressure pushing the particles out of the system or Poynting-Robertson drag pulling them into the star (Wyatt 2008). There is an important caveat to this which is that in some cases, depending on the ratio of the radiation pressure to the stellar gravitational force as a function of grain size, very small grains will not get removed by either process; and in other cases, grains will never get blown out by radiation pressure (see Figure 6 and further discussion below). Nevertheless, the collisional timescale for a system as dusty as BD +20 307 is still too short (several thousand years) to sustain so much warm debris over the age of the binary stars (Wyatt et al. 2007). Furthermore, it seems unlikely that we just happened to catch the initial start of the collisional cascade when we began observing the debris disk system only thirty years ago. Even if we did serendipitously start observing BD +20 307 at the beginning of the cascade, a purely steady-state collisional cascade model cannot explain why we see variations in the dust flux, and especially an increase in the infrared excess, over such short yearly timescales. Therefore, in order to explain the variations we observe, we must relax some of the assumptions made in steady-state collisional cascade. Under the model for steady-state cascade evolution, the size distribution of dust grains follows a power-law (usually f (a) ∝ aγ, where γ = -3.5) (Dominik & Decin 2003). This power-law assumption is usually assumed to work down to an abrupt small-size cutoff in the distribution caused by the characteristic blowout size for small dust grains (as shown in Figure 6) which introduces waves into the dust grain size distribution (Bagatin et al. (1994), Thebault et al. (2003)). As Figure 6 illustrates, for a given dust grain composition, there is a characteristic blowout size (denoted by where a given curve intersects the β=0.5 line) below which grains will be ejected from the system determined by β, the ratio between the radiation pressure pushing grains out and gravity from the host stars pulling them in. We consider grains made of either pure astro-silicate or an astro-silicate amorphous carbon mixture, and for each of these grain compositions we consider three types of grains: compact spheres, porous spheres and agglomerated debris particles (see Arnold et al. (2019)). In the case of this binary system, very small grains can persist in the disk, and these can be very effective at breaking up larger grains and launching avalanches that may happen stochastically over time (Krivov (2010), Grigorieva et al. (2007)). The persistence of small grains also helps explain why the silicate feature is so prominent even though the canonical blowout size is >1 micron (see Weinberger et al. (2011)). One could also imagine adding even more complexity to collisional cascade models, in which, for example, there are planetesimals of different strengths because of different compositions or thermal histories rather than just size. In addition, planetesimal strength can vary depending on their size distribution (O'Brien & Greenberg 2003). Relaxing assumptions like these in the steady-state cascade model may help explain why we see some variation in the dust flux over short timescales. There are two possible causes for a flux increase: an increase in temperature of the dust (since L ∝ T 4) or an increase in the surface area of the disk that is visible to the observer. In our model of Weinberger et al. (2011), the dust grains sit in a ring and have temperatures that vary according to their absorptivities, from 357 K for blackbody grains to 507 K for crystalline silicates. There are two ways the dust temperature could increase: increasing the stellar flux or moving the dust closer to the binary stars. The dust temperature would increase if the stellar flux at wavelengths where the dust absorbs strongly (UV-visible) increased. To increase the flux by 10% takes a temperature increase of ∼3.5% or a luminosity increase of 12%. Figure 7 shows the impact of increasing the stellar luminosity on the temperature of varying compositions of dust grains. We note that this would also produce a small change in the radiation pressure blowout size (β ∝ L∗; see below and Figure 6). Increasing the stellar luminosity by 12% would increase the grain temperatures to 475, 478 and 524 K (from 459, 461 and 507 K) for amorphous olivine, amorphous pyroxene and crystalline olivine, respectively (Figure 7, Weinberger et al. (2011)). These temperatures are well below the melting temperatures, and thus would not change the composition or grain size distribution of the dust. Although typical single F-type stars are not very variable at old ages, F-type binary stars can be variable, despite the fact that their activity is believed to decrease with age (Donahue 1998). For instance, HD 212280 is a chromospherically active spectroscopic binary of similar age and spectral types as BD +20 307 and was discovered to be variable through photometric monitoring, with several starspots being the assumed cause for the variability (Fekel et al. 1993). BD 10 Figure 6. β-ratio vs. dust grain radius for pure astro-silicate (teal) and an astro-silicate amorphous carbon mixture (indigo). For each of these two compositions, β-ratios for compact Mie spheres (solid lines) are compared with 76.4% porous Mie spheres modified using the Bruggeman mixing rule (dashed lines) and 76.4% porous irregular agglomerated debris particles. The method for generating agglomerated debris particles is described in Zubko et al. (2005). This porosity was chosen based on the average for agglomerated debris particles as calculated by Zubko et al. (2015). For pure astro-silicate, an extremely porous Mie sphere example (97.5%) is also shown (dash-dotted line). Scattering properties for the irregular agglomerated debris particles are averaged over at least 500 randomly generated and oriented particles, with particles added as necessary until the variation is less than 1%. +20 307 is not known for certain to be variable at visible wavelengths; however, once Gaia gathers all its data, it will be a good source to check for stellar activity. Zuckerman et al. (2008) measured BD +20 307's X-ray flux in the 0.5-2.0 keV band to be 1.1 × 10−13erg/cm2/s, corresponding to a luminosity of 4.7 × 10−5 L(cid:12) and a fractional X-ray L∗ ) of 1.6× 10−5, which alone cannot cause an ∼10% increase in the stellar luminosity. One possible way luminosity ( LX to get an increase in flux is if the disk is misaligned with respect to the orbital plane of the stars and is precessing on short timescales (∼ years). Assuming hotspots on the stars are generating the X-ray emission, then in the case of a precessing disk, the disk will see the spots periodically and heat up during those times when it is exposed to the spots. Unfortunately, the inclination of BD + 20 307's debris disk is not known, but Zuckerman et al. (2008) suggest that mid-infrared interferometric measurements might be able to prove whether or not the disk is inclined with respect to the orbital plane of the stars. The second way to increase the temperature of the dust grains is by moving them closer to the binary stars (∝ √ d). Most debris disks are thought to be in a radiation pressure dominated regime, where collisions grind dust to smaller sizes until the grains reach the blowout size and are ejected on an orbital timescale (e.g., Dominik & Decin (2003), Wyatt et al. (2007), Wyatt (2008)). This mechanism can explain the decline in the incidence of debris disks over time. However, the changes in BD +20 307 are in the direction of increasing the disk flux. At the new combined stellar luminosity of 2.9 L(cid:12), the ring is at 1.04 AU. For the dust flux to increase 10% requires moving the dust in ∼5% to 0.99 AU, which increases the temperatures of the dust grains by ∼15 K. The change in temperatures is similar amongst the different grain compositions, so that the shape of the spectrum is not much changed by such an increase 0.11.010.0100.0Dust Grain Radius (mm)1.0e-031.0e-021.0e-011.0e+001.0e+01bb = 0.5Si (Draine 2003)Si + 76.4% porositySi agglomerate (76.4% porosity)Si + 97.5% porosity95% Si + 5% C compact95% Si + 5% C + 76.4% porosity95% Si + 5% C agglomerate (76.4% porosity)11 Figure 7. Grain temperature as a function of stellar luminosity for four grain compositions. For each composition, the grain temperature increases as input luminosity from the star increases. in temperature. Comparing the ratio of the integral under the spectrum in the short and long wavelength regimes (less than and greater than 10.6 µm, respectively) for model spectra with the dust at 0.98 AU and 1.04 AU, we find there would be a ∼1.4% increase at the short wavelengths relative to the long wavelengths. The Poynting-Robertson drag timescale is given by 400d2/β yr where d is the distance of the grain from the star (Burns et al. 1979). As shown in Figure 6, we calculated β for compact spheres (Mie theory), spheres with porosity (Mie theory plus a Bruggeman mixing rule), and agglomerated debris particles (Zubko et al. 2005) assuming a combined stellar mass of 1.9 M(cid:12) and luminosity of 2.9 L(cid:12) (Arnold et al. 2019). For grains just larger than the blowout level of β=0.5, i.e., ∼2µm, at ∼1 AU, Poynting-Robertson drag can bring a grain all the way to the star in 800 yr, or a few tenths of an AU in ∼100 yr. This is not fast enough to account for the change we see. In addition, the short collisional timescale would favor radiation pressure blowout. Stellar wind drag can operate efficiently on small grains, even below the nominal blowout size because small grains have a greater coupling efficiency to the wind. The wind drag timescale can be considered a multiple of the Poynting- L∗ Robertson timescale, i.e. tsw = tP R M c2 (Plavchan et al. 2005). Here QP R and Qsw are the efficiencies of the two processes that depend on grain properties and whose ratio can be assumed to be ∼1, and M is the wind mass loss rate in Solar masses per year. For the Sun-like stars, X-ray activity and winds are likely correlated (Feigelson 1982), and flaring X-ray activity is correlated with stellar coronal mass ejections (Aarnio et al. 2012). The enhanced X-ray activity of the tidally locked spectroscopic binary in this system might also produce a significant stellar wind. A wind rate that is 10× Solar would reduce the timescale to drag in dust a few tenths of an AU by an order of magnitude, to ∼10 yr, a scale comparable to the time between our observations. QP R Qsw To account for the increased disk flux, it is also possible that the disk's surface area visible to the observer increased either by the production of more dust grains or a change in the optical depth which exposes more dust to the stars and the observer. The total number of grains in the disk would have to increase by about the same fraction as the increase in flux. This is similar to the "avalanche" type events considered by Grigorieva et al. (2007) in which the breakup of a large planetesimal happens in the background of a pre-existing disk of small particles, thus triggering yet more collisions that significantly increase the amount of small dust grains. The "parent" bodies that break up in sequential avalanches in this scenario could be fragments of a larger differentiated body that was involved in the original giant 12 collision. The largest body involved in the collision would have to have a timescale for destruction by smaller grains of ∼8 years, i.e. the difference in the observations, along with removal of the previously produced small grains on the same timescale. The amount of dust, ∼1021g, is comparable to that used in the calculations of Grigorieva et al. (2007), but the disk is closer to the star and smaller radially, which helps to cause a faster collisional cascade, although exact models have not been done for this regime. However, more recent work on collisional avalanches by Thebault & Kral find that the photometric excess due to an avalanche, even in the case of extreme debris disks like BD +20 307, is less than 10%. They claim that while a larger initial planetesimal mass could result in greater photometric excesses, it is less likely for a larger mass object to get broken apart (Thebault & Kral 2018). That being said, determining the frequency at which planetesimals break apart depends strongly on the specific debris disk system such as the size distribution of large objects and their dynamical excitation (Thebault & Kral 2018). Therefore, it would be beneficial to apply Thebault & Kral's updated collisional avalanche models to a BD +20 307-like system under various assumptions of planetesimal size distribution. We also found that the flux at shorter wavelengths has increased more than the flux at longer wavelengths over the last 10 years. This possible increase in dust flux, particularly pronounced at wavelengths less than 10.6 µm, suggests that the dust's silicate composition has possibly evolved during this short timescale. If future data can definitively find a change in the shape of the spectrum, there are possible explanations for greater increase in flux at shorter wavelengths. The process of converting from the crystalline to amorphous phase via irradiation has been previously studied in detail (e.g, Brucato et al. (2004), Christofferson & Keller (2011)), however the conversion from crystalline to amorphous via collisions has not been as well studied (see Gleason et al. (2017) as an example of recent experimental work on this conversion). A catastrophic collision is known to subject bodies to extremely high temperatures and pressures. If the temperatures are high enough that the grains reach their melting temperatures and the objects cool at a fast enough rate, which seems plausible in a gas-less disk like BD +20 307's, then they can be quenched into a glass (e.g., Birnie & Dyar (1986) studied the cooling rates required to form the silicate glasses on the Moon). The cooling rate must be faster than the tens of minutes it takes for grains to anneal from amorphous to crystalline as determined by Hallenbeck et al. (2000). This seems likely given the very short amount of time (nanoseconds) Gleason et al. (2017) predict that it takes for crystallized silica to convert to an amorphous phase as a result of shock loading induced by an extreme collision. Therefore, if such a collision can indeed induce the conversion from objects' crystalline to amorphous phases, this could provide a possible explanation for the possible greater change in flux at shorter wavelengths. 6. CONCLUSIONS We have detected significant variation in BD +20 307's dust flux over the course of ∼10 years, from the earlier sets of observations (Keck/Gemini in 2004-2005 and Spitzer in 2006-2007) to our recent SOFIA observations in 2015. BD +20 307's dust flux has increased by 10% between ∼8.8-12.5 µm. In addition, while there is indication that the shape of the spectrum has changed, with the dust flux potentially increasing more at shorter wavelengths (less than 10.6 µm) compared to longer wavelengths (greater than 10.6 µm), we cannot conclude this definitively due to the high uncertainties in the SOFIA spectra. New SOFIA observations covering a wider wavelength range out to 20 µm along with fitting composition models of the dust grains to the wider wavelength-range of data (as in Weinberger et al. (2011)) will allow us to draw more concrete conclusions regarding the most likely cause of this dust flux increase over such short timescales. Nevertheless, it is clear that steady-state collisional cascade alone cannot explain these variations, and particularly the increase in dust flux, over such short timescales. Therefore, we must relax some of the assumptions of the steady-state model, such as the size distribution of dust grains following a power-law all the way down to a single minimum grain size, or that all planetesimals have the same strength. We investigated several mechanisms that could cause the observed changes in the disk flux, including making the dust grains hotter, either through an increase in stellar luminosity or moving the dust grains closer to the stars, or increasing the number of dust grains in the system. If the origin of the copious amount of warm dust orbiting BD +20 307 is an extreme collision between planetary-sized bodies, then this system may help unlock clues into planetary systems around binary stars, along with providing a glimpse into catastrophic collisions occurring late in a planetary system's history. Understanding BD +20 307 and other systems like it with extremely dusty debris disks could advance our knowledge of catastrophic collisions, the effects of binary stars on debris disks and the evolution of planetary systems. This publication is based on observations made with the NASA/DLR Stratospheric Observatory for Infrared Astron- omy (SOFIA). SOFIA is jointly operated by the Universities Space Research Association, Inc. (USRA), under NASA contract NNA17BF53C, and the Deutsches SOFIA Institut (DSI) under DLR contract 50 OK 0901 to the University 13 of Stuttgart. Financial support for this work was provided by NASA through award #02-0050 issued by USRA. This publication makes use of data products from the Wide-field Infrared Survey Explorer, which is a joint project of the University of California, Los Angeles, and the Jet Propulsion Laboratory/California Institute of Technology, funded by the National Aeronautics and Space Administration. This research has made use of the NASA/ IPAC Infrared Science Archive, which is operated by the Jet Propulsion Laboratory, California Institute of Technology, under contract with the National Aeronautics and Space Administration. This work has made use of data from the European Space Agency (ESA) mission Gaia (https://www.cosmos.esa.int/gaia), processed by the Gaia Data Processing and Anal- ysis Consortium (DPAC, https://www.cosmos.esa.int/web/gaia/dpac/consortium). Funding for the DPAC has been provided by national institutions, in particular the institutions participating in the Gaia Multilateral Agreement. We would like to thank Inseok Song for useful preparatory discussions. Facility: SOFIA, Spitzer, Keck, WISE, Herschel, IRSA REFERENCES Aarnio, A. N., Matt, S. P., & Stassun, K. G. 2012, The Krivov, A. V. 2010, Research in Astronomy and Astrophysical Journal, 760, 11 Astrophysics, 10, 383 Arnold, J. A., Weinberger, A. J., Videen, G., & Zubko, Melis, C., Zuckerman, B., Rhee, J. H., et al. 2012, Nature, E. S. 2019, in prep 1207.1162 Bagatin, A. C., Cellino, A., Davis, D., Farinella, P., & Meng, H. Y. A., Rieke, G. H., Su, K. Y. L., et al. 2012, The Paolicchi, P. 1994, Planetary and Space Science, 42, 13 Birnie, D. P., & Dyar, M. D. 1986, Journal of Geophysical Astrophysical Journal Letters, 751, 5 Meng, H. Y. A., Su, K. Y. L., Rieke, G. H., et al. 2015, The Research, 91, 5 Brucato, J., Strazzulla, G., Baratta, G., & Colangeli, L. 2004, Astronomy & Astrophysics, 413, 6 Burns, J. A., Lamy, P. L., & Soter, S. 1979, Icarus, 40, 48 Christofferson, R., & Keller, L. P. 2011, Meteoritics & Planetary Science, 46, 19 Astrophysical Journal, 805, 1503.05610 O'Brien, D. P., & Greenberg, R. 2003, Icarus, 164, 11 Plavchan, P., Jura, M., & Lipscy, S. J. 2005, The Astrophysical Journal, 631, 8 Schlafly, E. F., & Finkbeiner, D. P. 2011, The Astrophysical Journal, 737, 13 Dominik, C., & Decin, G. 2003, The Astrophysical Journal, Song, I., Zuckerman, B., Weinberger, A. J., & Becklin, 598, 9 E. E. 2005, Nature, 436, 363 Donahue, R. A. 1998, Cool Stars, Stellar Systems and the Thebault, P., Augereau, J., & Beust, H. 2003, Astronomy Sun, ASP Conference Series, 154, 8 Dotter, A., Chaboyer, B., Jevremovi, D., et al. 2008, The Astrophysical Journal Supplement Series, 178, 12 Feigelson, E. D. 1982, Icarus, 51, 8 Fekel, F., Cordero, M., Galicher, R., et al. 2012, The Astrophysical Journal, 749, 11 Fekel, F. C., Browning, J. C., Henry, G. W., Morton, M. D., & Hall, D. S. 1993, The Astronomical Journal, 105, 11 FORCAST Science Instrument Team. 2016, Guest Investigator Handbook for FORCAST Data Products, revision b edn., SOFIA Gaia Collaboration, Brown, A. G. A., Vallenari, A., et al. 2018, ArXiv e-prints, arXiv:1804.09365 Gleason, A., Bolme, C., Lee, H., et al. 2017, Nature Communications, 8, 6 Grigorieva, A., Artymowicz, P., & Th´ebault, P. 2007, Astronomy and Astrophysics, 461, 537 Hallenbeck, S. L., III, J. A. N., & Nelson, R. N. 2000, The Astrophysical Journal, 535, 9 & Astrophysics, 408, 13 Thebault, P., & Kral, Q. 2018, Astronomy & Astrophysics, 609, 13 Vican, L., Schneider, A., Bryden, G., et al. 2016, The Astrophysical Journal, 833 Weinberger, A. J. 2008, The Astrophysical Journal Letters, 679, L41 Weinberger, A. J., Becklin, E. E., Song, I., & Zuckerman, B. 2011, The Astrophysical Journal, 726, 72 Wyatt, M. C. 2008, Annual Review of Astronomy & Astrophysics, 46, 339 Wyatt, M. C., Smith, R., Greaves, J. S., et al. 2007, The Astrophysical Journal, 658, 569 Zubko, E., Petrov, D., Shkuratov, Y., & Videen, G. 2005, Applied Optics IP, 44, 6 Zubko, E., Videen, G., Hines, D. C., et al. 2015, Planetary and Space Science, 118, 25 Zuckerman, B., Fekel, F., Williamson, M., Henry, G., & Muno, M. 2008, The Astrophysical Journal, 688, 7
1812.05053	2	1812	2019-01-23T15:29:53	The ARCiS framework for Exoplanet Atmospheres: The Cloud Transport Model	[ "astro-ph.EP" ]	Understanding of clouds is instrumental in interpreting current and future spectroscopic observations of exoplanets. Modelling clouds consistently is complex, since it involves many facets of chemistry, nucleation theory, condensation physics, coagulation, and particle transport. We develop a simple physical model for cloud formation and transport, efficient and versatile enough that it can be used in modular fashion for parameter optimization searches of exoplanet atmosphere spectra. The transport equations are formulated in 1D, accounting for sedimentation and diffusion. The grain size is obtained through a moment method. For simplicity, only one cloud species is considered and the nucleation rate is parametrized. From the resulting physical profiles we simulate transmission spectra covering the visual to mid-IR wavelength range. We apply our models towards KCl clouds in the atmosphere of GJ1214 b and towards MgSiO3 clouds of a canonical hot-Jupiter. We find that larger cloud diffusivity $K_{zz}$ increases the thickness of the cloud, pushing the $\tau=1$ surface to a lower pressure layer higher in the atmosphere. A larger nucleation rate also increases the cloud thickness while it suppresses the grain size. Coagulation is most important at high nuclei injection rates ($\dot\Sigma_n$) and low $K_{zz}$. We find that the investigated combinations of $K_{zz}$ and $\dot\Sigma_n$ greatly affect the transmission spectra in terms of the slope at near-IR wavelength (a proxy for grain size), the molecular features seen at ~1\micr (which disappear for thick clouds, high in the atmosphere), and the 10\micr silicate feature, which becomes prominent for small grains high in the atmosphere. The result of our hybrid approach -- aimed to provide a good balance between physical consistency and computational efficiency -- is ideal towards interpreting (future) spectroscopic observations of exoplanets.	astro-ph.EP	astro-ph	Astronomy & Astrophysics manuscript no. rain January 24, 2019 c(cid:13) ESO 2019 9 1 0 2 n a J 3 2 . ] P E h p - o r t s a [ 2 v 3 5 0 5 0 . 2 1 8 1 : v i X r a ARCiS framework for exoplanet atmospheres cloud transport model Chris W. Ormel1 and Michiel Min2 1 Anton Pannekoek Institute (API), University of Amsterdam, Science Park 904,1090 GE Amsterdam, The Netherlands 2 Netherlands Institute for Space Research (SRON), Sorbonnelaan 2, 3584 CA Utrecht, The Netherlands e-mail: [[email protected],m.min@sron] January 24, 2019 ABSTRACT Context. Understanding of clouds is instrumental in interpreting current and future spectroscopic observations of exoplanets. model- ing clouds consistently is complex, since it involves many facets of chemistry, nucleation theory, condensation physics, coagulation, and particle transport. Aims. We aim to develop a simple physical model for cloud formation and transport, efficient and versatile enough that it can be used, in modular fashion for parameter optimization searches of exoplanet atmosphere spectra. In this work we present the cloud model and investigate the dependence of key parameters as the cloud diffusivity K and the nuclei injection rate Σn on the planet's observational characteristics. Methods. The transport equations are formulated in 1D, accounting for sedimentation and diffusion. The grain size is obtained through a moment method. For simplicity, only one cloud species is considered and the nucleation rate is parametrized. From the resulting physical profiles we simulate transmission spectra covering the visual to mid-IR wavelength range. Results. We apply our models toward KCl clouds in the atmosphere of GJ1214 b and toward MgSiO3 clouds of a canonical hot- Jupiter. We find that larger K increases the thickness of the cloud, pushing the τ = 1 surface to a lower pressure layer higher in the atmosphere. A larger nucleation rate also increases the cloud thickness while it suppresses the grain size. Coagulation is most important at high Σn and low K. We find that the investigated combinations of K and Σn greatly affect the transmission spectra in terms of the slope at near-IR wavelength (a proxy for grain size), the molecular features seen at ∼µm (which disappear for thick clouds, high in the atmosphere), and the 10 µm silicate feature, which becomes prominent for small grains high in the atmosphere. Conclusions. Clouds have a major impact on the atmospheric characteristics of hot-Jupiters, and models as those presented here are necessary to reveal the underlying properties of exoplanet atmospheres. The result of our hybrid approach -- aimed to provide a good balance between physical consistency and computational efficiency -- is ideal toward interpreting (future) spectroscopic observations of exoplanets. Key words. Planets and satellites: atmospheres -- Planets and satellites: composition -- Methods: numerical 1. Introduction The composition of exoplanet atmospheres contains very im- portant clues to their formation and evolution. Different forma- tion scenarios predict different abundances of key elements like C, O, N, and Si (e.g., Oberg et al. 2011; Helling et al. 2014; Mordasini et al. 2016; Madhusudhan et al. 2017). Measuring the abundances of these elements is one of the major goals of performing exoplanet atmosphere spectroscopy (see e.g., Brewer et al. 2017). With the launch of the James Webb Space Telescope (JWST) scheduled in 2021, a new wavelength window, the near- to mid-IR, will open up for compositional analysis of exoplanet atmospheres. With the recently selected ARIEL mission on the 2028 horizon, performing spectroscopy of a statistically signifi- cant sample of exoplanets, the future for atmosphere character- ization looks particularly bright (Turrini et al. 2018). This new spectroscopic window presents us with many opportunities, but at the same time provides challenges in proper interpretation. One of the major hurdles in atmospheric characterization is the presence of clouds obscuring the gaseous content of the atmosphere. Besides shielding the gaseous atmosphere from detection, clouds also alter the chemical composition of the gaseous atmosphere. By removing elements from the gas phase and raining them down to deeper layers, cloud processes alter the chemical composition of the atmosphere. For the interpre- tation of the atmosphere spectrum, this can lead to an incorrect assessment of the atomic composition of the bulk planet. The difficulty of modeling cloud formation has led to a rich variety of different treatments of clouds. For models that re- trieve key atmospheric parameters (temperature, pressure, and chemical profiles) directly from the observations, so-called re- trieval models, it is very important that the simulations can be performed in the most computationally efficient manner. These methods often simply apply an atmospheric pressure be- low which the opacity of the atmosphere is gray (or infinite) with the possible addition of Rayleigh scattering haze (see e.g., Kreidberg et al. 2015; Barstow et al. 2017). This assumption might be acceptable for the narrow wavelength range considered in most studies right now. However, when the wavelength range extends, it becomes crucial to take into account the wavelength dependence of the optical properties of the cloud particles. In forward models the complexity of the cloud formation varies. The approximate cloud formation model by Ackerman & Marley (2001) is probably one of the most widely used cloud for- mation frameworks. In this model the physical properties of the 1 Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model cloud particles are parameterized in terms of a single parameter, fsed, the ratio between the particle sedimentation and the turbu- lent eddy velocities. It can be regarded as a proxy of the cloud particle size, although for constant fsed the size will vary with height. While the assumption of a constant fsed is not a priori evident, the advantage of this approach is that it avoids an elab- orate grain microphysical prescription. At the other extreme are full self-consistent models that follow the microphysics of grain nucleation, condensation, transport and chemistry (Helling et al. 2008a; Gao et al. 2018). Nevertheless, enhanced model com- plexity also introduces drawbacks. First, these models tend to be computationally demanding and are therefore not well suited for implementation in retrieval codes. In addition, increased model complexity often implies a great number of free parameters, which either need to be justified or else need to be explored, in- creasing the computational demand. Most crucially in this regard is the formation of condensation seeds (nucleation), which under the extreme conditions in exoplanet atmospheres is poorly un- derstood. These considerations might argue in favor of building a retrieval framework that contains no cloud formation physics and, by fitting the spectrum, have the observations tell us what is going on (see e.g., Barstow et al. 2017; Tsiaras et al. 2018). While this is a widely-used approach, a drawback of this ap- proach is that it comes with a plethora of free parameters, which physical consistency is not a priori guaranteed (e.g., the feed- back of cloud formation on the atmospheric composition is not necessarily accounted for). Here, we aim for an intermediate approach, in which the cloud structure is computed in a simplified but consistent for- ward model. We envision that such a hybrid model has the benefits of both worlds: it should include the most elementary cloud physics (e.g., condensation and cloud transport) consis- tently, but yet be be computationally fast and flexible enough to allow for parameter studies and incorporation in retrieval algo- rithms. Recent examples of this approach are the semi-analytical model by Charnay et al. (2018), applicable for Brown Dwarfs and young exoplanets, 1D dust coagulation models of atmo- spheres of planets embedded in their natal gas disk (Movshovitz et al. 2010; Mordasini 2014; Ormel 2014), and 1D cloud trans- port models for exoplanets (Ohno & Okuzumi 2017, 2018; Kawashima & Ikoma 2018). A common characteristics of these approaches is that they are one dimensional and consider a sin- gle, representative particle size that varies with height. In this paper we follow these leads to efficiently compute the forma- tion of clouds for hot Jupiters. We use a diffusion/condensation framework to compute the growth of cloud particles and include particle coagulation. On the other hand the nucleation rate is pa- rameterized to accommodate the large uncertainty in nucleation efficiency. The cloud model that we present in this paper will be- come part of a general framework for analysis and retrieval of exoplanet spectra1. In this context we are developing a code for computation of atmospheric properties, radiative transfer, and retrieval named ARCiS (ARtful modeling Code for exo- planet Science). The overarching aim of ARCiS is to develop an approach that is well-balanced between physical consistency, model complexity and computationally efficiency. The physical consistency allows for direct physical interpretation of observa- tions. The modest model complexity allows for in depth under- standing of the effects going on. The computational efficiency ensures that the model can be efficiently used in spectral re- 1 The cloud model, written in python, is publicly available at http://www.exoclouds.com/. 2 trieval analysis of observations. In this paper we focus on the cloud model; a validation of the entire ARCiS framework and subsequent fitting of real spectra will be deferred to upcoming studies. In Sect. 2 the cloud formation model is explained. In Sect. 3 we present the resulting cloud structures and transmission spec- tra for a sub-Neptune (GJ1214 b) and for a typical hot-Jupiter planet, while varying the diffusivity and nucleation rate. In Sect. 4, we present the synthetic transmission spectra in the near- to mid-IR for the hot-Jupiter configuration. An assessment of the cloud model is proved in Sect. 5. In Sect. 6 we summarize the results and discuss extensions to this modeling framework. 2. Model Our cloud particle model entails solving for the 1D steady state solutions to the transport equations involving vapor, conden- sates, and nuclei. Cloud particles are initiated through nucleation at prescribed rates. Vapor can condense on these seeds and the particles may further growth by coagulation. Particles are trans- ported by gravitational settling and turbulent (eddy) diffusion, until they reach the bottom of the cloud, hot enough to result in their evaporation. For simplicity a single species -- KCl in case of GJ1214 b and MgSiO3 in case of the generic hot Jupiter -- is considered. The choice for the species in question is arbitrary, although for the cloud to be observed, it must lie high in the at- mosphere. Hence, the temperature of the upper atmosphere must be lower -- but not much lower -- than the condensation temper- ature. It is also straightforward to extend the model to include other chemical compounds. 2.1. Atmosphere model We consider the atmosphere typical to a hot Jupiter planet. To obtain its physical structure -- the temperature T (z), pressure P (z) and density ρ(z) profiles -- we adopt the atmosphere model of Guillot (2010) to obtain a relation between temperature and depth: (cid:19) (cid:18) 2 (cid:20) 2 + τ T 4 = + 3T 4 int 3 4 3T 4 irrfirr 4 (cid:18) γ√ 3 √ − 1 γ 3 (cid:19) (cid:21) √ 3 e−γτ (1) + √ 1 3 γ + 3 where the internal temperature Tint is a measure of the planet's internal heat flux σT 4 int -- the rate at which the planet cools -- Tirr a measure of the heat flux received from the star, τ the optical depth at IR wavelengths, and γ = κvis/κIR the ratio between the opacity at visual (irradiated) and IR (outgoing) wavelengths. The parameter firr specifies the distribution of the incoming flux over the planet (firr = 1/4 for an equal distribution over the en- tire planet is used here). The irradiation temperature is defined Tirr = T(cid:63)(Rp/rp)2 where T(cid:63) is the stellar (effective tempera- ture), Rp the planet radius and rp the distance to the host star. See Guillot (2010) for further discussion. Employing this relation between T and τ, the hydrostatic balance, and the ideal gas law we obtain the temperature and density as function of pressure. (In particular, for constant κIR, as we will use here, we have P = gzτ /κIR). Figure 1 pro- vides the P-T and ρ-T structures resulting from the atmosphere model for our generic hot Jupiter model and GJ1214 b (pa- rameters are discussed in Sect. 3 and listed in Table 1). From Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Fig. 1. Left: the physical structure of a generic hot-Jupiter atmosphere. The temperature T (P ) and gas density profile ρgas(P ) are obtained from the Guillot (2010) atmosphere model. The equilibrium density ρeq of MgSiO3 is obtained from Eq. (18). Below the cloud deck (horizontal dashed line), ρv = ρ(MgSiO3 ) < ρeq; the species are present at constant abundance, such that xv = ρv/ρgas ≡ xv,bot is constant. Above the cloud base (ρv > ρeq; light blue) cloud formation (not included in this figure) will reduce the vapor density at the expense of condensates. Right: same for KCl in the GJ1214 b atmosphere. the temperature a saturation vapor density (Psat) can be cal- culated. Vapor is expected to condense out when the partial pressure will exceeds Psat. Here, we re-express this condition in terms of a density: condensation occurs when ρeq < ρv, where ρeq = mvPsat/kBT , ρv = xvρ the vapor density, kB Boltzmann constant, xv the mass concentration, and mv the molecular weight of the vapor species. Let us denote the vapor concentration below the cloud deck by xv,bot. Below the cloud deck we can expect that xv,bot is constant. The vapor density below the cloud deck is then simply ρv = xv,botρgas (blue line in Fig. 1). The height where ρv = ρeq (vertical dashed line) can be taken as the point where cloud for- mation starts -- the base of the cloud. Because of transport effects it is conceivable that the cloud will extend below this height, for example, heavy rain particles take time to evaporate. Conversely, the cloud deck could be located at a higher layer if cloud forma- tion requires strong, super-saturated conditions. These consider- ations are automatically accounted for in our numerical model. Above the cloud base the vapor concentration is expected to become less than xv,bot (light blue line) due to cloud formation. The surface (top) of the cloud is defined where the concentration of condensates xc = ρc/ρgas is (close to) zero. The height where this occurs is not a priori known. 2.2. Cloud transport model We model cloud transport of particles and vapor as advection- diffusion processes, solving equations like + ∇ · Mi = Si (2) ∂ρi ∂t where ρi is the mass density of a certain species i, t is time, and Mi the mass flux Mi ≡ ρivsed,i − Kρgas∇xi (3) xi the mass concentration of species i, vsed,i the particle sed- imentation velocity, K the diffusion tensor, and ρgas the gas density. In this work, we consider only vertical (z) transport, implying that only one velocity component and one diffusion element (Kzz) remain (vsed = 0 for a vapor species). The RHS of Eq. (2), Si, specifies source terms arising from deposi- tion (condensation), sublimation (evaporation) or nucleation, de- pending on the species i. The sedimentation velocity is obtained by equating the aerodynamic drag force with the planet's grav- ity, vsed,p = gztstop where tstop encapsulates the aerodynamic properties of the cloud particle. In general, the gas drag force is non-linear in particle-gas velocity vsed (see e.g., Whipple 1972) and tstop must be found by iteration. However, for small par- ticles tstop becomes independent of velocity. In particular, for the parameters of our model cloud, the gas drag law obeys the Epstein (1924) regime (free molecular flow) for which tstop−Epstein = apρ• vthρgas (cid:112)8kBT /πmgas the thermal velocity of the gas. The Epstein where ap is the radius of the grain, ρ• its internal density, vth = √ drag law applies in the free molecular flow regime, ap < 9 where lmfp = mgas/( 2ρgasσmol) is the mean free path. 4 lmfp, We employ the following assumptions: (4) 1. The medium consist of three components -- nuclei (n), con- densates (c), and vapor (v). Only a single cloud species is considered. Any gas-gas or gas-grain chemistry is not ac- counted for. 2. The model is plane parallel; only the vertical dimension (z) is modeled and the only relevant diffusion coefficient is Kzz. 3. The cloud model is in steady state, ∂/∂t = 0. This implies that Mv = −Mc (5) 4. Nucleation is parametrized in the form of a log-normal pro- at any location. file with height (pressure) (cid:34) (cid:18) (cid:19)2(cid:35) Sn = ρgasgz ΣN √ σ∗P 2π exp − 1 2σ2∗ log P P∗ (6) 3 110012001300140015001600170018001900temperature[K]10−510−410−310−210−1pressure[bar]10−1510−1310−1110−910−7density[gcm−3]generichotJupiterxv,botTρgasρeq,MgSiO3ρMgSiO35006007008009001000temperature[K]10−510−410−310−210−1100pressure[bar]10−1810−1610−1410−1210−1010−810−6density[gcm−3]GJ1214bxv,botTρgasρeq,KClρKClChris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model where ΣN , P∗ and σ∗ respectively indicate the integrated nu- clei production rate, the characteristic height where the nu- clei are deposited, and the width of the distribution. 5. At any layer, the characteristic particle mass mp is obtained taking the ratio of the total solid density of the particles (the density of condensates ρc = xcρgas plus the density of nu- clei xnρgas) to the particle number density np. In our model, the particle number density follows from the nuclei num- ber density. In the case without coagulation any particle will contain only one nuclei, np = nn. (Below, in Sect. 2.3 this assumption will be relaxed, accounting for coagulation ef- fects.) Hence mp = (xc + xn)ρgas np ≈ xcρgas nn = xcmn0 xn (7) where mn0 is the mass of a single nuclei. In Eq. (7) the sec- ond step assumes that the condensates dominate the mass and the last step employs the single nuclei per cloud parti- cle assumption: nnmn0 = xnρgas. From the characteristic particle mass mp a characteristic grain radius ap follows, as- suming that the grains are spherical. A grain size distribution is not accounted for, but it may be reconstructed from ap. In addition, ap changes with height through nucleation, con- densation, evaporation, and coagulation. The grain radius ap in turn determines the sedimentation velocity vsed,p of the particles. 6. We take the diffusivity (Kzz) equal for vapor and particles and independent of height. (These assumptions are easily re- laxed, though). We then obtain the following set of ordinary equations specify- ing the evolution of the condensate, nuclei, and vapor: = Sc ∂Mc ∂z ∂Mn ∂z = Sn − Mc − Mn Kzzρgas Kzzρgas xcvsed,p Kzz xnvsed,p ∂xc ∂z ∂xn ∂z = = Kzz = − Mv ∂xv ∂z (8a) (8b) (8c) (8d) (8e) where Mv is given by Eq. (5), Sn by Eq. (6), and the particle properties follow from xc and xn as described above. Kzzρgas The condensation rate Sc is given by Sc = fstick(xvρgas−ρeq)×min(cid:2)πa2 (cid:3) . (9) pvth,vnp; 4πapDinp where Di the diffusivity and fstick a sticking probability, here taken unity, and ρeq = mvPsat/kBT the equilibrium (or satura- tion) density. Equation (9) combines the free molecular flow (va- por molecules travel on ballistic trajectories on the scale of the particle) and the diffusion-limited regimes (Woitke & Helling 2003; Yau & Rogers 1996). In Eq. (9) we have not accounted for the (liberated) latent heat of condensation. For the diffusivity we follow Woitke & Helling (2003), after Jeans (1967), and write Di = kBT 3Pgasσcom vth,red. (10) 4 This equation describes diffusion of a quantity in a two compo- nent medium of vapor and hydrogen gas. The reduced thermal velocity vth,red is taken equal to the mean gas thermal velocity (vth,red = vth) and σcom is the combined cross section. We take, somewhat arbitrarily, σcom = 8 × 10−15 cm2. 2.3. Adding coagulation Coagulation among the cloud particles will decrease their num- ber density np, relaxing the identity np = nn, while leaving unaffected the mass concentration of nuclei. That is, coagulation will result in np < xnρgas/mn0. Within the above framework, it is possible to include coagulation among the cloud particles by adding two additional equations, describing np: ∂Np ∂z Sn mn0 = − np tcoag (11a) ∂cp ∂z = cpvsed,p/Kp − Np/Kpngas (11b) where Np is the particle number flux and tcoag is the coagulation timescale, cp = np/ngas the particle concentration (by number), and ngas the gas number density. The coagulation time includes contributions from differential settling (∆v) and Brownian mo- tion. In terms of the coagulation rate (dnp/dt = −np/tcoag) these can be added: 1 2 1 2 (12) t−1 coag = where vBM = (cid:112)16kBT /πmp for equal mass particles, Dp = 4π min(vBMap, Dp)apnp npπ(2ap)2∆v + kBT /6πηap (StokesEinstein equation), η = νmolρgas the dy- namic viscosity, νmol = 0.5lmfpvth the molecular viscosity (Chapman & Cowling 1970), and ∆v is the relative velocity be- tween the cloud particles. The factor 1 2 prevents double counting. Following Krijt et al. (2016) and Sato et al. (2016) it is appro- priate to take ∆v = 0.5vsed when the coagulation is driven by sedimentation. For identical particles having the same aerody- namical properties ∆v = 0, but in reality a distribution in aero- dynamical properties always ensures that ∆v ∼ vsed (Okuzumi et al. 2011). Adding these equations would bring the total number of equations to solve to seven. However, when we assume (cor- rectly) that the nuclei mass is insignificant, xn (cid:28) xc, there is no need to follow the nuclei mass density xn. Equation (11a) and Eq. (11b) then replace Eq. (8b) and Eq. (8d). To keep the expres- sions in units of mass concentrations (like x) and mass flux (like M), we transform Eq. (11) by defining: Mn = mn0Np xn = npmn0/ρgas (13a) (13b) In terms of these new variables, Eq. (11) read = Sn − xnρgas tcoag ∂ Mn ∂z ∂ xn ∂z = xnvsed,p/Kp − Mn/Kpρgas (14a) (14b) These are identical to Eq. (8b) and Eq. (8d), except for the term involving tcoag. In runs including coagulation we simply use these equations to follow the number density of nuclei (nn or xn). The nuclei mass density (xn) is not followed, but this is justified since it is in any case negligible compared to the mass density of the condensate (xc) and therefore bears no influence on the physical properties of the cloud particles. Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Table 1. Cloud and atmospheric parameters for the generic hot-Jupiter and GJ1214 b. (default) Value Generic HJ GJ1214 ba KCl Unit Description Symbol species Σn γ κIR ρ• σ∗ σcom σmol Kg, Kp Mplanet P∗ Rpl R(cid:63) T(cid:63) Tint an firr fstick gz mgas mv rp xv,bot MgSiO3 10−19, 10−15, 10−11 0.158 0.3 2.8 0.2 8 × 10−15 2 × 10−15 106, 108, 1010 1 60 1.087 1 5778 500 0.001 0.25 1.0 2192 2.34 34.67 0.05 3 × 10−3 g cm−2 s−1 cm2 g−1 g cm−3 cm2 cm2 cm2 s−1 dyn cm−2 0.038 0.03 2.8 108 104 0.0206 MJ 0.244 RJ 0.2064 R(cid:12) 3026 K 60 K µm 893 cm s−2 mH 74.45 mH 0.0143 au 3 × 10−4 cloud species nucleation rate opacity ratio visual and IR wavelengths (Eq. (1)) IR opacity particle internal density width of nucleation profile (Eq. (6)). combined (vapor and gas) molecular cross section molecular cross section (gas) particle and gas diffusivity planet mass reference height for the nucleation profile (Eq. (6)) planet radius stellar radius stellar effective temperature internal temperature particle nucleation radius heat distribution factor vapor sticking probability gravitational acceleration mean molecular weight (gas) mass vapor species distance to star vapor mass concentration at/below cloud base Notes. (a) Empty entries indicate the value listed in the generic hot Jupiter column is used. 2.4. Boundary conditions and solution technique Equations (8a) -- (8e) constitute a system of five first or- der, ordinary differential and five unknowns (xc, xn, xv,Mc,Mn). Therefore, five boundary conditions are necessary. We specify boundary conditions at the bottom and the top of the domain. At the top of the cloud (z = ztop) we demand that the condensate flux vanishes: equations and that the nuclei flux equals Mn(ztop) = − Mc(ztop) = 0 (cid:90) ∞ (15) (16) Sndz ztop while at the base of the cloud we put constraints on the mass concentrations: xn(zbot) = xc(zbot) = 0; xv(zbot) = xv,bot. (17) The condition xc = 0 reflects that at the base of the cloud the temperature has become high enough for all the condensates to be evaporated. The vapor concentration at the cloud base (xv,bot) is an input. The nuclei boundary condition xn = 0 strictly only holds when the nuclei are also made of MgSiO3, such that they would also evaporate. But this is not necessarily the case. Formally, we should extend the systems of equations describing evaporation of the nuclei species, which is a rather cumbersome extension of the model. Alternatively, we could introduce a free parameter for nn(zbot). But we found its effects rather insignif- icant as long as it is not too large. Hence, we considered the simple choice of a zero concentration nuclei boundary condition is preferable above an (arbitrary) specification of the nuclei seed. Since conditions are placed on both the upper and the lower boundary, Eqs. (8a) -- (8e) represent a boundary value problem (BVP). This BVP is solved using the solve bvp function from python's SciPy module (Ascher et al. 1994; Kierzenka & Shampine 2001; Shampine et al. 2006). These codes requires an initial "guess" for the solution, which must be sufficiently close to the actual solution. Otherwise, convergence is not guaranteed. This represents a problem since the actual solution is of course unknown. We therefore resort to an iterative approach, introducing a parameter that is added to the condensation rate Sc and the nucleation rate Sn. Hence, = 0 corresponds to the cloud-free solution (xv = xv,bot, xc = 0, Mc = 0). Then, a very small , for example = 10−8, will give a solution that will be close to the known (cloud-free) solution that solve bvp is able to solve. This new solution (with = 10−8) then provides the guess for the next iteration, where is larger. We progressively increase until = 1, with which the desired cloud profiles are obtained. A similar iterative approach can be designed for the bound- aries of the domain. Although the bottom boundary is given by the ρv,bot constraint,2 the upper boundary is in principle open, as diffusion always allows some particles to be transported to the very upper regions. As a final step, we therefore adjust the boundaries of the domain, searching for a solution where xc stays positive in the entire domain, while xc near the boundary is a very tiny fraction (e.g., 10−8) of its peak value. With these incremental approach of "relaxing" to the solu- tion, solve bvp is still computational efficient. The 24 runs listed in Table 2 took an average of 17 seconds to complete on a modern desktop PC, with the slowest one requiring 25 seconds. 2 The lower boundary may deviated from the ρv,bot = ρeq condition when transport timescales are shorter than evaporation times, e.g., when the particles have become large and settle quickly. 5 Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Table 2. Table of output quantities. coagulation Kzz planet generic HJ GJ1214 b x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) x (cid:88) (cid:88) x [cm2 s−1] 1010 1010 1010 1010 1010 1010 108 108 108 108 108 108 106 106 106 106 106 106 108 108 108 108 108 108 Σn [g cm−2 s−1] 10−11 10−11 10−15 10−15 10−19 10−19 10−11 10−11 10−15 10−15 10−19 10−19 10−11 10−11 10−15 10−15 10−19 10−19 10−11 10−11 10−15 10−15 10−19 10−19 Mc,max [g cm−2 s−1] −1.8 × 10−6 −1.7 × 10−6 −1.6 × 10−6 −1.8 × 10−6 −1.3 × 10−6 −1.3 × 10−6 −1.9 × 10−8 −1.8 × 10−8 −1.9 × 10−8 −1.8 × 10−8 −1.8 × 10−8 −1.8 × 10−8 −1.9 × 10−10 −1.9 × 10−10 −1.9 × 10−10 −1.9 × 10−10 −1.9 × 10−10 −1.9 × 10−10 −3.2 × 10−8 −3.0 × 10−8 −3.1 × 10−8 −3.0 × 10−8 −2.7 × 10−8 −2.9 × 10−8 Pτ =1 [bar] 4.3 × 10−7 9.5 × 10−7 8.9 × 10−6 1.0 × 10−5 2.6 × 10−4 2.7 × 10−4 3.5 × 10−6 1.9 × 10−5 3.6 × 10−5 4.6 × 10−5 2.1 × 10−3 2.1 × 10−3 1.0 × 10−4 6.6 × 10−4 2.2 × 10−4 3.8 × 10−3 5.6 × 10−3 5.8 × 10−3 5.6 × 10−6 2.0 × 10−4 1.2 × 10−4 4.1 × 10−4 3.0 × 10−3 2.5 × 10−3 τz,tot 3.0 × 103 740 140 73 5.0 3.9 1.3 × 104 29 460 22 4.6 2.5 6.2 × 104 1.7 450 1.6 1.3 0.86 2.7 × 104 810 1.2 × 103 250 16 44 amax [µm] 0.051 0.24 1.0 2.4 20 26 0.012 3.9 0.25 4.1 5.2 8.7 2.7 × 10−3 1.4 0.058 1.3 1.2 1.7 0.015 1.3 0.30 2.5 17 6.0 Notes. The first four columns list the input parameters: the planet (see Table 1 for parameters), whether coagulation is included or not, the diffusivity Kzz, and the nuclei production rate Σn. The model calculates a steady state cloud and we list: the peak mass flux (intensity of the rain) Mc,max, the pressure level where the geometrical transmission optical depth reaches unity Pτ =1, the total geometrical vertical optical depth of the cloud τz,tot, and the maximum grain radius amax. 3. Physical structure In this section we present the outcome of the cloud model in terms of its physical structure: the concentration and properties of the cloud particles. In Sect. 3.1 we consider a generic hot Jupiter planet, whereas in Sect. 3.2 we apply our model toward GJ1214 b to compare our results to previous findings. 3.1. Hot-Jupiter MgSiO3 clouds We consider a generic hot Jupiter planet situated at a distance of 0.05 au around a solar-like star. We consider MgSiO3 as our cloud species, for which we use the saturation pressure of Ackerman & Marley (2001) Psat = 1.04 × 1017 exp − 58 663 dyn cm −2. (18) (cid:20) (cid:21) T Because MgSiO3 does not exist in vapor phase, it would be erro- neous to consider taking the molecular weight of MgSiO3 (100.4 mH) for mv. Instead, we consider an effective vapor mass, which is given by the constituents from which MgSiO3 forms. Typically, MgSiO3 falls apart into three molecules (Helling et al. 2008a). We therefore simply take mv = mMgSiO3/3 = 33.5mH. For the atmospheric parameters, we adopt parameters similar values as Line et al. (2013), see Table 1. An internal tempera- ture of Tint = 500 K is used and an atmosphere IR-opacity of 0.3 cm2 g−1. The higher IR-opacity crudely reflects the appear- ance of clouds; the model does presently not treat (thermal) feed- back of the clouds on the profiles consistently. We have verified that changing these parameters does not affect the conclusions 6 of this work. The corresponding atmospheric physical structure was shown in Fig. 1. The outcome of the cloud model for the parameters listed in Table 1 is presented in Fig. 2 for the default model. In Fig. 3 we take eight other parameter combinations of Kzz and Σn, crudely corresponding what has been used in literature studies (e.g., Kawashima & Ikoma 2018). Several output quantities of the runs are further listed in Table 2. The intensity of the rain is characterized by the mass flux parameter Mc whose peak value is listed. A higher Mc,max reflects more vigorous mass trans- port; this parameter hence correlates with the diffusivity Kzz. For reference, a value of Mc = 10−7 g cm−2 s−1 amounts to a MgSiO3 precipitation of 11 mm yr−1. We calculate both the vertical optical depth (cid:90) ztop z τz(z) = (cid:90) ztop z np(z(cid:48))πa2 p(z(cid:48))dz(cid:48) (cid:115) p(z(cid:48)) 2R (z(cid:48) − z) as well as the transmission optical depth in the geometrical limit τtrans(z) = np(z(cid:48))πa2 dz(cid:48) (20) (19) that is, the optical depth corresponding from the line perpendic- ular to height z. In Table 2 the total geometrical optical depth refers to τz as measured from the base of the cloud whereas the pressure level where τ reaches unity (Pτ =1) refers to the trans- mission optical depth τtrans. The latter quantity is more mean- ingful in the context of transmission spectra. These geometrical values only serve as a crude guide as opacities are not often close Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Fig. 2. Results of the standard model (K = 108 cm2 s−1; Σn = 10−15 g cm−2 s−1). (a) Nucleation rate Sn, which follows a log- normal distribution around a reference pressure P∗. The dotted horizontal lines indicate the bottom and top of our computational domain. (b) The concentrations of nuclei (xn), condensates (xc), and vapor (xv). The equilibrium concentration corresponding to the saturation pressure is also plotted (xeq) but it virtually coincides with xv. (c) The characteristic grain radius ap. The dashed lines correspond to a model where the nuclei are inserted at a higher layer than the standard. The open circles correspond to the depth where the geometrical transmission optical depth τtrans equals 1. to their geometrical limit (especially for small particles). Proper simulated spectra are calculated in Sect. 4. Finally, we list the peak radius of the condensate particles, amax. Figure 2 presents profiles of nucleation rate, concentrations of vapor condensates and nuclei and grain size for the standard model (Kzz = 108 cm2 s−1, Σn = 10−15 g cm−2 s−1; the cen- tral panel of Fig. 3 corresponds to Fig. 2). Coagulation is in- cluded. Note the steep but continuous transition from cloudy to cloud-free near the bottom of the cloud. This is caused by the steep increase in the equilibrium density (Fig. 1). Several factors regulate the extent of the cloud. The first is the location where the nuclei form, which is given in Fig. 2a. Recall that the nuclei production profile Sn (Eq. (6)) is characterized by three param- eters: P∗, σ∗ and Σn. In Fig. 2 we also present a case where the nuclei are released at a higher height (P∗ is decreased by a factor three; dashed curves). Increasing the height where the nu- clei are released does not much affect the profiles deeper in the atmosphere. In both cases cloud particles readily consume the vapor locally, whereas transport and coagulation act on larger (time)scales. However, there may be some observational conse- quences as the grain size around τtrans = 1 is affected. Comparing Fig. 2a and b, it can be seen that the height where nuclei are injected is also the height where the concentration of nuclei (xn) peaks. Below this height xn decreases because par- ticles' velocity speeds up due to their growth by condensation. The ratio of xc and xn determines the size of the particles, which increases for our standard model to 4 µm just above the cloud base (Fig. 2c). In the upper regions, the particle radius levels off at ≈0.005 µm, several factors larger than the nucleation ra- dius. Grains tend to be somewhat smaller and more abundant in the model where the nuclei are injected at a larger height, be- cause they accrete less vapor before settling down (Fig. 2b and c). Although at these heights the density of MgSiO3 is rather low, the larger grain size may be of some observational importance for the transmission spectra, especially concerning the Rayleigh scattering at optical wavelengths. More important in regulating the cloud thickness is the eddy diffusivity Kzz. A larger Kzz implies that more vapor is trans- ported upwards and that more (small) particles can be found above the nuclei injection height. This is illustrated in Fig. 3 where we vary the diffusivity (rows) and the total nucleation rate (columns). Clearly, larger diffusivity results in denser and thicker clouds; particles are uplifted to higher regions and more vapor is being transported from below the cloud deck. In the limit where the transport becomes dominated by diffusion, we can expect the concentration of condensates xc to be identi- cal to the concentration of the vapor at the base of the cloud, xv,bot = 3 × 10−3. This explains the boxy cloud profile of the Σn = 10−11, Kzz = 1010 model (bottom right panel of Fig. 3). Finally, we find that the cloud thickness increases with the nucleation rate Σn (Fig. 3). A higher Σn tends to reduce the grain size, since the total amount of vapor on a given grain is smaller when there are more of them. Since the grain size is a directly observational property, the nucleation model is therefore an essential part to any cloud model. We also conducted runs without coagulation, in order to isolate its effects. These are presented by the dashed lines in Fig. 3. Clearly, coagulation does not affect the low Σn runs. The growth of particles in these runs is entirely due to condensation. Coagulation is more important at higher nucleation rates and for lower Kzz; the former because there is a larger surface available and the latter because the grains are more concentrated. The dif- ferences between the no-coagulation and coagulation runs are the greatest in the ( Σn, Kzz) = (10−11, 106) run (top right). The no-coagulation run is characterized by extremely small particles (similar to the nucleation size) and the total geometrical opti- cal depth of the cloud reaches values above 104 (see Table 2). Including coagulation, however, greatly increases the grain size, 7 10−2510−2410−2310−22nucleationrate[gcm−3s−1]10−510−410−310−2pressure[bar](a)P∗=2.0×10−5P∗=6.0×10−510−1010−810−610−4concentration(b)xcxnxvxeq10−310−210−1100grainradius[µm](c)Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Fig. 3. Cloud profiles. The concentration of cloud particles (black) and the characteristic particle size (dark red; shared x-axis) against pressure, plotted for combinations of diffusivities Kzz and nucleation rates Σn (panels). The grain radius of particles in models without coagulation is shown by the dashed dark curve. The height where the transition optical depth reached unity (τtrans = 1) is indicated by the circle. reducing the cloud vertical geometrical optical depth by over a factor of 104! 3.2. GJ1214 b KCl clouds the physical structure (particle sizes and concentrations) against these works in the broadest sense. A detailed comparison, let alone a calibration, is rather meaningless as these works employ vastly different cloud microphysical and atmospheric models. GJ1214 b is a super-Earth or sub-Neptune planet of radius Rp = 2.7±0.1 R⊕ and mass Mp = 6.5±1.0 M⊕ orbiting an M4.5 star at a distance of 0.015 au (Charbonneau et al. 2009). With these bulk properties GJ1214 b could both a "water world" or a more standard terrestrial planet with a H/He envelope. Interestingly, GJ1214 b transmission spectra is virtually featureless (Kreidberg et al. 2014), indicative of clouds. Cloud models have recently been applied to GJ1214 b (e.g., Gao & Benneke 2018; Ohno & Okuzumi 2018). Here we apply our cloud model toward GJ1214 b with the aim of comparing We consider KCl as our cloud species and use the Psat pro- file presented in Morley et al. (2012). The concentration of KCl at the bottom of the atmosphere is taken to be xv,bot = 3×10−4. We largely follow Ohno & Okuzumi (2018) in choosing our at- mospheric parameters (see Table 1). However we keep κIR fixed; with κIR = 0.03 cm2 g−1 we obtain a P-T profile (Fig. 1) that resembles theirs. The diffusivity is fixed at Kzz = 108 cm2 s−1 while we consider the same three values for the nuclei produc- tion rate Σn. Nuclei are injected at a height corresponding to a pressure of 0.01 bar. 8 10−710−610−510−410−310−2Σn=10−19,Kzz=106Σn=10−19,Kzz=106Σn=10−15,Kzz=106Σn=10−15,Kzz=106Σn=10−11,Kzz=106Σn=10−11,Kzz=10610−710−610−510−410−310−2pressure[bar]Σn=10−19,Kzz=108Σn=10−19,Kzz=108Σn=10−15,Kzz=108Σn=10−15,Kzz=108ap104xcΣn=10−11,Kzz=108Σn=10−11,Kzz=10810−310−210−110010110−710−610−510−410−310−2Σn=10−19,Kzz=1010Σn=10−19,Kzz=101010−310−210−1100101grainradius[µm],104xcΣn=10−15,Kzz=1010Σn=10−15,Kzz=101010−310−210−1100101Σn=10−11,Kzz=1010Σn=10−11,Kzz=1010Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model the generic hot-Jupiter model shown in Fig. 3 the transmission spectra. These are shown in Fig. 5. To compute the spectra we have developed a radiative trans- fer tool for simulating exoplanet spectra. This code uses molec- ular line lists from the ExoMol project and the HITEMP and HITRAN databases to compute the molecular opacities with the method by Min (2017). A validation of this ARCiS module is given in App. A. Even though for the cloud condensation equa- tions we use pure MgSiO3 as a condensate, we add 10% of metallic iron to the particles when computing the optical prop- erties. The implicit assumption is that the physical properties of the cloud particles (their sizes and concentrations) are well de- scribed by modeling the dominant condensate, in other words by our cloud model. However, this assumption cannot be made for the optical properties, which exhibit a strongly non-linear de- pendence on composition. MgSiO3, for example, is completely transparent in the near-IR, while only a small fraction of iron in the silicate lattice, or condensed inclusions, like metallic iron, will suffice to give a significant near-IR opacity. Hence, lacking a multi species cloud model, we account for this by adding a small amount of continuum opacity in the form of metallic iron. The 10% metallic iron we take here is rather arbitrary and could be up to 50% given the cosmic abundance of iron. The true iron fraction in planetary atmospheres is a parameter that we have to derive from observations or constrain from planet formation theory. The cloud opacities are computed using refractive index data from Jaeger et al. (1998) and Henning & Stognienko (1996) where we mix the iron and MgSiO3 together using effective medium theory. We use the distribution of hollow spheres (DHS) method from Min et al. (2005) to convert the refractive in- dex into particle optical properties. The gas phase chemistry is computed assuming thermochemical equilibrium using the code from Molli`ere et al. (2017). The atomic abundances that go into the chemical computations are assumed to be solar with deple- tions in Si, O, and Mg according to the computed value of xv. This causes the C/O ratio to change in the cloud forming region, affecting the chemistry there. Below the cloud deck the C/O ra- tio is solar, C/O = 0.55, while in the cloud forming region C/O ≈ 0.7. In Fig. 5 the mid-IR transmission spectra are plotted for the same combination of diffusivities and nucleation rate as in Fig. 3. The near- to mid-IR spectral region will become available with the MIRI instrument onboard JWST and further into the future with the recently selected ARIEL mission. Several inferences can be made. First, increasing the cloud thickness (either by in- creasing ΣN or increasing Kzz) suppresses the molecular fea- tures of, for example, H2O in the 1 -- 3 µm range. The reason is that, the τ = 1 height now resides much higher in the atmo- sphere to shield the molecular emission. A striking result is the spectral appearance of MgSiO3 around 10 micron. The 10 micron silicate resonance is very sen- sitive to particle size. Small particles give a strong resonance signature, while increasing the particle size, the signature is flattened (see e.g., Min et al. 2005). In addition, the solid fea- ture stands out stronger against the (molecular) background for thicker clouds. Therefore, the resonance around 10 micron is most clearly seen in the case with high nucleation rate and dif- fusion strength (lower right panel), that is, a thick cloud of small particles. Only for the lowest diffusion strength (upper panels) does the silicate signature become unobservable around 10 mi- cron. Finally, Fig. 5 displays a very interesting evolution of the slope of the near-IR signature. For the low nucleation rate mod- els, the effect of increasing the cloud thickness (i.e., the diffusion 9 Fig. 4. Obtained particle size as function of pressure for the cloud models applied to GJ1214b. The intensity of the rain in terms of the volume density of condensates (ρc = xcρgas) is in- dicated by the color bar. Three values of the nuclei production rate are considered (as indicated by color) and results are plotted with and without accounting for coagulation. The diffusivity pa- rameter is fixed at Kzz = 108 cm2 s−1. Particle radii are larger in runs that include coagulation. Results are shown in Fig. 4 where the particle radius is plot- ted against height for the three nucleation rates and for either the coagulation mode and the no-coagulation mode. The intensity of the rain in terms of the condensate volume density ρc is in- dicated by the color. The rain reaches its highest intensity near the cloud base. Clearly, the nuclear production rate -- a free pa- rameter in our model -- has a key influence on the grain size. Also, it can be seen that clouds with the smallest grains are also the most extended, since these grains tend to diffuse, rather than settle. Finally, grains are larger in runs where coagulation is in- cluded. These findings reflect the discussion of the hot generic hot-Jupiter clouds in Sect. 3.1. Comparing these curves to the Kzz = 108 cm2 s−1, 1x solar metallicity panel of Fig. 5 of Gao & Benneke (2018), we see that their typical sizes of 1 -- 10 µm correspond well to our results with the low Σn. (In their model the nucleation rate is given by a full microphysical model) The gradient in grain size with height seems to be a bit shallower in our models, however, considering that (Gao & Benneke 2018) did not include coagulation. Ohno & Okuzumi (2018) also modeled GJ1214 b and, like us, used a characteristic size approach. In addition, they too pre- scribed the nucleation. However, they fixed the nuclei number density at the cloud base. Compared to our choice of prescribing the entire profile, this has the advantage of only introducing a single free parameter. On the other hand, it results in the largest grains residing in the top of the cloud, which seems somewhat spurious. Their typical grain size of 1 -- 2 µm nevertheless corre- sponds well to our results (they too account for coagulation) and their volume mixing ratios approach xc = 10−4 -- the same as in our case. 4. Transmission spectra (hot Jupiter) From Fig. 3 we see that the cloud thickness and particle size are heavily influenced by the diffusion strength and the nucleation rate. To investigate their effect on the spectral appearance of the transit signal of the planet, we computed for the nine cases of 10−310−210−1100101particleradius[µm]10−310−210−1pressure[bar]Σn=10−19Σn=10−15Σn=10−11GJ1214b−12.0−11.5−11.0−10.5−10.0−9.5−9.0−8.5condensatedensity,log10ρcChris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Fig. 5. ARCiS-Simulated transmission spectra corresponding to the physical profiles presented in Fig. 3 as function of wavelength. Gray curves give the spectra without accounting for coagulation, black curves include coagulation. The scaling of the y-axis is different for the different diffusivities (higher Kzz results in a larger Rp). strength) results in a gray near-IR spectrum. On the other hand, for the high nucleation rate, the near-IR spectrum is character- ized by a much steeper slope. The reason behind this diverging trend with cloud thickness is the dependence of particle size with nucleation rate. Higher nucleation rates result in smaller grains whose opacity has a much steeper wavelength dependence in the near-IR region. Conversely, the 1 -- 10 µm grains that are pro- duced in the low Σn, high K run (bottom left panel) result in a gray opacity and a transmission spectra insensitive to wave- length. The spectra we computed are sensitive to the effects of par- ticle coagulation. The effects are twofold. One is that coagula- tion causes the grains to grow and settle deeper into the atmo- sphere. Second the opacity of the larger particles produced by coagulation is different. It can be seen that when we switch off the coagulation the cloud deck in the upper right four panels of Fig. 5 is much higher and thus mutes the molecular features more. In addition, the spectral appearance typical for small par- ticles, the silicate feature at 10 µm and the Rayleigh scattering slope at optical wavelengths, are reduced significantly by the effects of particle coagulation. While the case with low diffu- sion and high nucleation rate displays a strong cloud deck and silicate feature without coagulation, the spectral appearance is dominated by molecular features when coagulation is switched on. These considerations emphasize that cloud features can only be properly interpreted by models that include coagulation. 5. Model assessment We reflect on the achievement of our cloud model in the light of recent similar approaches. The key idea of our approach is to extend the simplicity and usability of the Ackerman & Marley (2001) model with a more physical justified cloud model, while preserving its simplicity. The Ackerman & Marley (2001) 10 1001010.01270.01280.01290.0130Σn=10−19,Kzz=106Σn=10−19,Kzz=106100101Σn=10−15,Kzz=106Σn=10−15,Kzz=106100101Σn=10−11,Kzz=106Σn=10−11,Kzz=1061001010.01270.01280.01290.01300.0131transmissionspectra,(Rp/R?)2Σn=10−19,Kzz=108Σn=10−19,Kzz=108100101Σn=10−15,Kzz=108Σn=10−15,Kzz=108100101Σn=10−11,Kzz=108Σn=10−11,Kzz=1081001010.01290.01300.01310.0132Σn=10−19,Kzz=1010Σn=10−19,Kzz=1010100101wavelength[µm]Σn=10−15,Kzz=1010Σn=10−15,Kzz=1010w/ocoagulationwithcoagulation100101Σn=10−11,Kzz=1010Σn=10−11,Kzz=1010Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model model already contained particle and vapor transport; however, it does not compute the size of the cloud particles. To pro- ceed, a nucleation prescription is required. This we have done very crudely, simply by imposing it through ad-hoc prescrip- tions. Alternatively, nucleation can be treated from first princi- ples. Photochemistry is a possible avenue for the formation of seed nuclei, which is thought to be the source of the haze as, for example, observed in Titan (Tomasko et al. 2005). Another nucleation pathway is that of homogeneous nucleation, where the nuclei seed directly form out of the vapor. The hot interiors of exoplanets characterized by thick envelopes will guarantee evaporation of any condensate at some depth. For these planets homogeneous condensation may be considered the natural way to form clouds. These additions to the nucleation model can, in principle, render the model more physically rigorous. However they also come at a drawback. A well-known issue with the classical nu- cleation theory is that it mispredicts nucleation rates by many orders of magnitude (e.g., Feder et al. 1966; Tanaka et al. 2005; Horsch et al. 2008; Diemand et al. 2013). Similarly, codes that model haze formation necessarily rely on a large chemi- cal network, with hundreds of reactions, and sophisticated radi- ation transport (Lavvas et al. 2008; Lavvas & Koskinen 2017; Kawashima & Ikoma 2018). Obviously, parametrizing nucle- ation implies that the size of the typical cloud particle no longer follows from first principles. But the transport model still ad- dresses variations of particle concentration and size with height, which act as independent model constraints. Another major simplification we have adopted is the charac- teristic particle approach (as in Ohno & Okuzumi 2017). A brief discussion on the validity and limitations of this approach can be found in Kawashima & Ikoma (2018). Recently, several stud- ies have used CARMA3 toward modeling clouds on exoplanets (Gao et al. 2018; Gao & Benneke 2018; Powell et al. 2018). An output of this code is the particle size distribution at any height. A possible approach is to reconstruct the entire grain size dis- tribution from the characteristic size (cf. Helling et al. 2008b or Birnstiel et al. 2012 for disks). Nevertheless, within the ARCiS framework, solving for the particle size distribution is too com- putationally intensive, since we intend it to be used in future MCMC parameter searches. Altogether, we make no claim to have invented the "best" cloud model in terms of physical rigor, but one that is minimalistic, physical consistent and above all useful. Its modular approach can easily be extended to include more physical processes and its results can guide sophisticated, computationally intensive models in a complementary fashion. 6. Summary In this paper we have studied the effects of diffusion strength and nucleation efficiency on the characteristics of clouds in exoplanet atmospheres. We have presented a relatively simple framework of cloud formation where these effects can be studied efficiently. Both the nucleation rate and the diffusion strength are key parameters in determining the properties of the cloud parti- cles and the extent of the cloud. Since both these parameters are highly uncertain, it is important to understand their effects. We have presented simulated infrared transmission spectra for different combinations of these two parameters in a typical hot Jupiter atmosphere. For the physical structure of the clouds we conclude that: 3 Community Aerosol and Radiation Model for Atmospheres. -- Increasing the nucleation rate results in thicker clouds of smaller particles. The high number of nuclei facilitate con- densation. At the same time, the condensed mass is dis- tributed over a larger number of particles, resulting in on average smaller particles. -- Increasing the diffusion strength results in thicker clouds. In this case, more vapor is mixed up and can condense on the nuclei. This causes simply more cloud material at each alti- tude and thus thicker clouds. For the transmission spectra resulting from these structures we conclude: -- For increasing diffusion strength and to a lesser degree in- creasing nucleation rate the molecular features weaken. This is caused by increasingly thicker clouds shielding more of the gaseous atmosphere. -- For high values of the diffusion strength and nucleation rate, the solid state 10 µm silicate feature appears. This feature of the cloud particles is visible in almost all parameter set- tings we consider here, but is most prominent for the highest values of diffusion and nucleation because they create the thickest clouds with small particles. -- For increasing nucleation rates, that is, smaller particles, the slope of the Near-IR steepens. -- Coagulation has a significant influence on the spectral ap- pearance of the clouds, especially in the case of high nucle- ation rates. The above observational features can be used to charac- terize cloud particles in exoplanet atmospheres. The modeling framework we present in this paper is computationally not very demanding. We can see two very important extensions of the present model. First, the opacities obtained from the cloud model can be fed back to the physical structure, such that for exam- ple the temperature profile is obtained self-consistently (recall that we used a fixed κIR in calculating the P -T profile). Second, and maybe more important, we can use this modeling framework to include a physically motivated cloud formation model in re- trieval methods. This way we can simulate the effect that clouds have on the atmospheric composition and observational features. In addition, it allows us to put observational constraints on physi- cal parameters like the nucleation rate and the diffusion strength. This will provide a significant step forward in understanding the physical processes in exoplanet atmospheres. Acknowledgements. C.W.O. and M.M. are grateful to Paul Molli`ere for pro- viding his petitCODE to calculate the gas composition. The authors also ac- knowledge fruitful discussion with Christiane Helling, Paul Molli`ere, and Peter Woitke. C.W.O. is supported by the Netherlands Organization for Scientific Research (NWO; VIDI project 639.042.422). The research leading to these re- sults has received funding from the European Unions Horizon 2020 Research and Innovation Programme, under Grant Agreement 776403. References Ackerman, A. S. & Marley, M. S. 2001, ApJ, 556, 872 Ascher, U. M., Mattheij, R. M. M., & Russell, R. D. 1994, Numerical solution of boundary value problems for ordinary differential equations, Vol. 13 (Siam) Barstow, J. K., Aigrain, S., Irwin, P. G. J., & Sing, D. K. 2017, ApJ, 834, 50 Baudino, J.-L., Molli`ere, P., Venot, O., et al. 2017, ApJ, 850, 150 Birnstiel, T., Andrews, S. M., & Ercolano, B. 2012, A&A, 544, A79 Brewer, J. M., Fischer, D. A., & Madhusudhan, N. 2017, AJ, 153, 83 Chapman, S. & Cowling, T. G. 1970, The mathematical theory of non-uniform gases. an account of the kinetic theory of viscosity, thermal conduction and diffusion in gases Charbonneau, D., Berta, Z. K., Irwin, J., et al. 2009, Nature, 462, 891 Charnay, B., B´ezard, B., Baudino, J.-L., et al. 2018, ApJ, 854, 172 11 Chris W. Ormel and Michiel Min: Exoplanet atmosphere cloud transport model Fig. A.1. Transmission spectra for the standard model without clouds computed using petitCODE and ARCiS. exceptionally well at almost all wavelengths. There are some small differences in the optical part of the spectrum which can be attributed to a different Rayleigh scattering law and different opacities for TiO and VO used in both codes. These differences are irrelevant for the purpose of this paper. The petitCODE spectrum is computed at higher spectral resolution and thus shows small high frequency variations which are smoothed in the lower resolution ARCiS spectrum. We conclude that the match is excellent. Diemand, J., Ang´elil, R., Tanaka, K. K., & Tanaka, H. 2013, J. Chem. Phys., Epstein, P. S. 1924, Physical Review, 23, 710 Feder, J., Russell, K. C., Lothe, J., & Pound, G. M. 1966, Advances in Physics, 139, 074309 15, 111 Gao, P. & Benneke, B. 2018, ApJ, 863, 165 Gao, P., Marley, M. S., & Ackerman, A. S. 2018, ApJ, 855, 86 Guillot, T. 2010, A&A, 520, A27 Helling, C., Woitke, P., Rimmer, P. B., et al. 2014, Life, 4 Helling, C., Woitke, P., & Thi, W.-F. 2008a, A&A, 485, 547 Helling, C., Woitke, P., & Thi, W.-F. 2008b, A&A, 485, 547 Henning, T. & Stognienko, R. 1996, A&A, 311, 291 Horsch, M., Vrabec, J., & Hasse, H. 2008, Phys. Rev. E, 78, 011603 Jaeger, C., Molster, F. J., Dorschner, J., et al. 1998, A&A, 339, 904 Jeans, J. 1967, An introduction to the kinetic theory of gases (Cambridge University Press) Kawashima, Y. & Ikoma, M. 2018, ApJ, 853, 7 Kierzenka, J. & Shampine, L. F. 2001, ACM Transactions on Mathematical Software (TOMS), 27, 299 Kreidberg, L., Bean, J. L., D´esert, J.-M., et al. 2014, Nature, 505, 69 Kreidberg, L., Line, M. R., Bean, J. L., et al. 2015, ApJ, 814, 66 Krijt, S., Ormel, C. W., Dominik, C., & Tielens, A. G. G. M. 2016, A&A, 586, Lavvas, P. & Koskinen, T. 2017, ApJ, 847, 32 Lavvas, P. P., Coustenis, A., & Vardavas, I. M. 2008, Planet. Space Sci., 56, 27 Line, M. R., Wolf, A. S., Zhang, X., et al. 2013, ApJ, 775, 137 Madhusudhan, N., Bitsch, B., Johansen, A., & Eriksson, L. 2017, MNRAS, 469, A20 4102 ApJ, 813, 47 ApJ, 832, 41 209, 616 Min, M. 2017, A&A, 607, A9 Min, M., Hovenier, J. W., & de Koter, A. 2005, A&A, 432, 909 Molli`ere, P., van Boekel, R., Bouwman, J., et al. 2017, A&A, 600, A10 Molli`ere, P., van Boekel, R., Dullemond, C., Henning, T., & Mordasini, C. 2015, Mordasini, C. 2014, A&A, 572, A118 Mordasini, C., van Boekel, R., Molli`ere, P., Henning, T., & Benneke, B. 2016, Morley, C. V., Fortney, J. J., Marley, M. S., et al. 2012, ApJ, 756, 172 Movshovitz, N., Bodenheimer, P., Podolak, M., & Lissauer, J. J. 2010, Icarus, Oberg, K. I., Murray-Clay, R., & Bergin, E. A. 2011, ApJ, 743, L16 Ohno, K. & Okuzumi, S. 2017, ApJ, 835, 261 Ohno, K. & Okuzumi, S. 2018, ApJ, 859, 34 Okuzumi, S., Tanaka, H., Takeuchi, T., & Sakagami, M.-a. 2011, ApJ, 731, 95 Ormel, C. W. 2014, ApJ, 789, L18 Powell, D., Zhang, X., Gao, P., & Parmentier, V. 2018, ApJ, 860, 18 Sato, T., Okuzumi, S., & Ida, S. 2016, A&A, 589, A15 Shampine, L. F., Muir, P. H., & Xu, H. 2006, J. Numer. Anal. Ind. Appl. Math, Tanaka, K. K., Kawamura, K., Tanaka, H., & Nakazawa, K. 2005, J. Chem. Phys., 122, 184514 Tomasko, M. G., Archinal, B., Becker, T., et al. 2005, Nature, 438, 765 Tsiaras, A., Waldmann, I. P., Zingales, T., et al. 2018, AJ, 155, 156 Turrini, D., Miguel, Y., Zingales, T., et al. 2018, Experimental Astronomy, 46, 1, 201 45 Whipple, F. L. 1972, in From Plasma to Planet, ed. A. Elvius, 211 Woitke, P. & Helling, C. 2003, A&A, 399, 297 Yau, M. K. & Rogers, R. 1996, A short course in cloud physics (Elsevier) Appendix A: Validation with petitCODE We have validated the computations performed with ARCiS with the exoplanet simulation code petitCODE (Molli`ere et al. 2015, 2017). The petitCODE has been extensively bench- marked in Baudino et al. (2017). We compute the transmission spectrum of the atmospheric setup from the model used here without any cloud formation. The chemistry, hydrostatic struc- ture, molecular opacities and resulting transmission spectrum are computed both by ARCiS and petitCODE independently. The chemical equilibrium module used in ARCiS is the same as the one used in petitCODE. This module is benchmarked in Baudino et al. (2017), so here we only check the proper imple- mentation of the module in ARCiS. Figure A.1 shows the com- parison of the resulting transmission spectra. The spectra match 12
1209.4358	1	1209	2012-09-19T20:02:43	Planet-Disk interaction in 3D: the importance of buoyancy waves	[ "astro-ph.EP" ]	We carry out local three dimensional (3D) hydrodynamic simulations of planet-disk interaction in stratified disks with varied thermodynamic properties. We find that whenever the Brunt-Vaisala frequency (N) in the disk is nonzero, the planet exerts a strong torque on the disk in the vicinity of the planet, with a reduction in the traditional "torque cutoff". In particular, this is true for adiabatic perturbations in disks with isothermal density structure, as should be typical for centrally irradiated protoplanetary disks. We identify this torque with buoyancy waves, which are excited (when N is non-zero) close to the planet, within one disk scale height from its orbit. These waves give rise to density perturbations with a characteristic 3D spatial pattern which is in close agreement with the linear dispersion relation for buoyancy waves. The torque due to these waves can amount to as much as several tens of per cent of the total planetary torque, which is not expected based on analytical calculations limited to axisymmetric or low-m modes. Buoyancy waves should be ubiquitous around planets in the inner, dense regions of protoplanetary disks, where they might possibly affect planet migration.	astro-ph.EP	astro-ph	Draft version June 21, 2018 Preprint typeset using LATEX style emulateapj v. 08/22/09 PLANET-DISK INTERACTION IN 3D: THE IMPORTANCE OF BUOYANCY WAVES Zhaohuan Zhu 1, James M. Stone 1, and Roman R. Rafikov 1 Draft version June 21, 2018 ABSTRACT We carry out local three dimensional (3D) hydrodynamic simulations of planet-disk interaction in stratified disks with varied thermodynamic properties. We find that whenever the Brunt-Vaisala frequency (N ) in the disk is nonzero, the planet exerts a strong torque on the disk in the vicinity of the planet, with a reduction in the traditional"torque cutoff". In particular, this is true for adiabatic perturbations in disks with isothermal density structure, as should be typical for centrally irradiated protoplanetary disks. We identify this torque with buoyancy waves, which are excited (when N is non-zero) close to the planet, within one disk scale height from its orbit. These waves give rise to density perturbations with a characteristic 3D spatial pattern which is in close agreement with the linear dispersion relation for buoyancy waves. The torque due to these waves can amount to as much as several tens of per cent of the total planetary torque, which is not expected based on analytical calculations limited to axisymmetric or low-m modes. Buoyancy waves should be ubiquitous around planets in the inner, dense regions of protoplanetary disks, where they might possibly affect planet migration. Subject headings: hydrodynamics, waves, stars: formation, stars: pre-main sequence, planet-disk in- teractions 1. INTRODUCTION The gravitational potential of a planet surrounded by a protoplanetary disk is known to give rise to non- axisymmetric density waves, which propagate away from the planet carrying angular momentum. Gravitational coupling with these density perturbations exerts a torque on the planet, which might lead to its migration if the one-sided torques produced by the inner and outer disks do not cancel. Pioneering linear calculations by Goldre- ich & Tremaine (1980) have shown that, neglecting this possible small asymmetry, the one-sided torque acting on a planet moving on a circular orbit in a razor-thin (2D) disk is T = q2D(GMp)2 Σ0RpΩp c3 s , q2D ≈ 0.93, (1) where Mp, Rp and Ωp are the planetary mass, semi- major axis and angular frequency respectively, cs ≡ (∂P/∂ρS)1/2 is the adiabatic sound speed, and Σ0 is the unperturbed disk surface density. Most of this torque is excited at Lindblad resonances corresponding to the m ∼ Rp/h ≫ 1 azimuthal harmonic of the planetary po- tential and located at a distance ∼ h = cs/Ωp away from the planetary orbit. Closer to the planet, at separations . h the excitation torque density decreases exponentially -- the so-called "torque cut off" phenomenon. In three-dimensional (3D) stratified disks a variety of waves can be excited (Lubow & Pringle 1993, Korycan- sky & Pringle 1995, Lubow & Ogilvie 1998), including buoyancy waves whenever the Brunt-Vaisala frequency N ≡(cid:20) g γ ∂ ∂z ln(cid:18) P0 ργ 0(cid:19)(cid:21)1/2 (2) js- Electronic [email protected], [email protected] [email protected], address: 1 Department of Astrophysical Sciences, Princeton University, Princeton, NJ, 08544 is non-zero. Global analytical calculation of the wave properties for all modes are available only for one spe- cial case, in which both the vertical disk structure and the equation of state (EOS) are isothermal, and N = 0. In this case, the result (1) still holds but with q2D low- ered to q3D ≈ 0.25q2D = 0.37 as a result of averaging the planetary potential over the vertical disk structure (Takeuchi & Miyama 1998). For other combinations of the vertical disk structure and EOS, in particular those characterized by non-zero N , analytical results are lim- ited only to either axisymmetric or low-m modes excited far from the planet. In that case Lubow & Ogilvie (1998) found that although buoyancy waves are responsible for a fraction of the total torque excited by the planet, the total torque from all modes is the same as that given by traditional Lindblad torque formulae. In this paper we carry out 3D simulations of stratified disks characterized by non-zero N with embedded low- mass planets to explore global excitation of buoyancy waves and their role in angular momentum exchange with the planet. 2. NUMERICAL METHOD The numerical tool we used is Athena (Stone et al. 2008), a grid-based code with a higher-order Godunov scheme, piecewise parabolic method (PPM) for spatial reconstruction, and the corner transport upwind (CTU) method for multidimensional integration. We use the stratified shearing-box set-up implemented in Athena (Stone & Gardiner 2010). Since the potential vortic- ity is zero in the shearing-box, the traditional corotation torque is zero (Goldreich & Tremaine, 1979). A planet is placed at the origin of a Cartesian box in Its smoothed potential is approxi- x, y, z coordinates. mated as Φp(d) = −GMp d2 + 3r2 (d2 + r2 s/2 s )3/2 , (3) 2 where rs is the smoothing length and d =px2 + y2 + z2. This potential converges to the point mass potential as (rs/d)4 for d ≫ rs. In most runs rs = 0.125h, which is resolved by 4 grids cells with our normal resolution of 32/h. To reduce the computation time we take advantage of the symmetry of the problem and only simulate the disk for x = [−X, 0], y = [−Y, Y ], and z = [0, Z], see Figure 1. We take X = 5h, Y = 20h so that the planetary wake located at y ∼ −3x2/4h fits well inside the box. We vary Z in different runs (see Table 1) in such a way that the density at z = Z is ∼5 orders of magnitude lower than the midplane density. We use different boundary conditions (BCs) at differ- ent box faces: outflow BC at the y = Y face, and "fixed state" BC (meaning that all physical variables in the ghost zones are fixed at their unperturbed Keplerian val- ues) at x = −X and y = −Y faces (for one model we also use periodic BC in y-direction, see Table 1). At the x = 0 face we employ a "symmetric" BC: the x = 0 face has been divided into two portions -- y > 0 and y < 0, and the variables in the last 4 active zones of one portion are copied into the ghost zones in the other portion symmet- ric with respect to x = y = 0; velocities (or momenta) in x and y directions are copied with the opposite sign. We use a reflecting BC at z = 0 since both the planetary po- tential and the disk structure are symmetric with respect to z. At z = Z we use an outflow BC but the densities are extrapolated assuming hydrostatic equilibrium into the ghost zones, and to prevent inflow if vz < 0, then we set vz to 0. 2.1. Physical setup In our calculations we use both an isothermal (γ=1 in p ∝ ργ), and an adiabatic EOS (γ = 5/3 or 7/5). The unit of length in our simulations is defined via the adiabatic scale height h = cs/Ωp = √γciso/Ω, where ciso ≡ (kT /µ)1/2 and cs is the adiabatic sound speed calculated using the midplane temperature. Obviously, cs = ciso for an isothermal EOS (γ = 1). This choice keeps sound waves traveling the same distance in a given amount of time in different models so that the wake po- sition is always the same. The vertical structure of a protoplanetary disk is typ- ically determined by irradiation from the central star so that is vertically isothermal ρ0(z) ρ00 = exp(cid:18)− z2Ω2 2c2 iso(cid:19) , (4) where ρ00 is the midplane density. The value of ciso is determined by the central irradiation flux (Kenyon & Hartmann 1987; Chiang & Goldreich 1997). We also consider thermally stratified disks in which the disk midplane is hotter than its atmosphere, which is expected if viscous heating is important. Including the fact that such disks will eventually become isothermal at hight z, the vertical structure of the disk is better represented as p0(z) = c2 s,sρ0(z)"(cid:18) Γ − 1 Γ (cid:19)(cid:18) ρ0(z) ρs (cid:19)Γ−1 + 1# , (5) where cs,s and ρs ( set as 0.01 ρc) represent the sound speed and density at the transition where the disk be- comes isothermal (Lin et al. 1990; Bate et al. 2002). How- ever, unlike Bate et al. (2002), we do not necessarily set Γ in Eq. (5) the same as γ in the EOS. We derive ρ0(z) by solving the equation of vertical hydrostatic equilibrium with Eq. (5) using the Newton-Raphson method. When Γ → 1 the isothermal disk structure is recovered. We run a set of five models up to 10 orbits with dif- ferent disk structure and EOS, summarized below and in Table 1. (a) Isothermal disk structure (4) with isothermal EOS (model II). (b) Disk structure (5) with adiabatic EOS and Γ = γ = 5/3 (model AA). (c) Isothermal disk structure with adiabatic EOS hav- ing γ = 5/3 (model IA). This setup is typical for a cen- trally irradiated disk in which density perturbations are adiabatic. (d) Analogous to (4) but with γ = 7/5 (model IA2), designed to illustrate the effect of adiabatic index on the buoyancy wave coupling. (e) Disk structure (5) with Γ = 5/3 and adiabatic EOS with γ = 7/5 (model PA), to illustrate viscously heated accretion disks. In the first two models N = 0 so that buoyancy waves are absent, while the last three models feature non-zero N and support buoyancy waves. All models were run at a resolution of 32/h, but we also run model (3) at higher resolution (64/h) to verify numerical convergence. The planet mass is Mp = 0.0058Mth, where Mth is the thermal mass Mth ≡ c3 s GΩp . (6) inside the linear Thus, disk-planet coupling is well regime. Density wave generated by such a planet in a 2D disk would shock at x = 6h (Goodman & Rafikov 2001), which is outside our box. 3. RESULTS 3.1. Without Buoyancy Waves First we show the results for models with no buoy- ancy waves (II and AA). The spatial distributions of the excitation torque density dT /dx (the amount of torque excited per unit radial distance) for these two models are shown in Fig. 2(a)(b) as green and orange curves. One can see that dT /dx rapidly goes to zero for x . h, clearly exhibiting the "torque cut-off" phenomenon (Gol- dreich & Tremaine 1980) in both models. For model II, our simulation agrees with the semi- analytical calculation by Takeuchi & Miyama (1998) re- markably well, and we confirm that Eq. (1) holds in this case with q3D(II) = 0.25q2D. Although wave structures are different in model AA (Lubow & Pringle 1993), the distribution of dT /dx for this model is quite similar in shape to that of model II, but the amplitude is slightly different. 3.2. With Buoyancy waves We next look at the behavior of dT /dx in models with non-zero N . The most important feature of these models, evident in Figure 2, is the weak torque cutoff near the planet -- all show a significant torque contribution near planet-disk 3 x = 0, including model IA2, which provides the best description for an irradiated protoplanetary disk. Since higher-m modes are excited at Lindblad resonances closer to the planet, the strong torque close to the planet also suggests the traditional torque cutoff in Fourier space (Ward 1997) is weaker when buoyancy waves are present. 2(c)(d)) in models IA and IA2 is characterized by q3D(IA) = 0.48 and q3D(IA2) = 0.41 respectively, which should be compared with q3D(II) = 0.34 for model II having the same verti- cal structure as IA and IA2 but different EOS. The one-sided torque T (Fig. In order to identify the origin of this excess torque near the planet, we integrate the volume density of the torque along y direction d2T dxdz dyδρ(x, y, z) ∂φ ∂y (7) =Z ∞ −∞ . This quantity is shown in Fig. 3(a) for models II and IA. We see that the excess torque in model IA comes from the region z ∼ 0.4 h and x < h. In Fig. 3(d) we plot the density contours in the xy plane at z = 0.4h. In clear contrast with model II (Fig. 3(c)), the density distri- bution in model IA exhibits density fluctuations/ridges close to the planet which are different from the usual wake structure. They extend to y > 0 and look like rays emanating from the origin. It is these density fluctua- tions that contribute to the excess torque. These fluctuations also have vertical structure, as shown in Fig. 4 where we plot δρ = ρ − ρ0 and vz in the yz plane at x = 0.5h for both models II and IA. Again, significant density fluctuations close to the planet exist only in case IA. Since the only difference between case IA and case II is that the case IA has non-zero Brunt-Vaisala frequency N , it is natural to relate these density fluctuations to buoyancy waves. To confirm this hypothesis, we note that a mode with wavenumber ky has the buoyancy fre- quency N whenever the condition 2Akyx = N (8) is fulfilled. For comparison, the Lindblad resonance con- dition is 2Akyx = ±κ. In the disk with isothermal struc- ture and adiabatic EOS (models IA and IA2), we have N = Ωz h r γ − 1 γ . (9) With γ = 5/3, and λy = 2π/ky, Eqs. 8 and 9 can be combined to derive λy h = 11.543 x z . (10) If we assume the phase of buoyancy waves is 0 at x=0, the geometric location of the constant phase 2nπ (n is integer) is given by y h = 11.543 x z n and n = 1, 2, 3, ... . (11) Curves corresponding to Eq. 11 are drawn in the lower right panels of Figs. 3 and 4. They agree with the pattern of the density perturbation and vz quite well, strongly suggesting that these fluctuations and the excess torque are related to buoyancy waves (their dissipation or exci- tation). The torque density and integrated torque for model IA2 (isothermal disk with adiabatic EOS and γ = 1.4) plotted as the cyan curves in Fig. 2 also exhibit the ex- cess torque density due to buoyancy waves close to the planet. However, the integrated torque in this case is characterized by q3D(IA2) ≈ 0.41, which is lower than q3D(IA), and the buoyancy wave coupling to the plane- tary potential is weaker for smaller γ as seen from dT /dx curve. Blue curves in the right panels of Fig. 2 show dT /dx and T (x) for model PA. There is again an apparent torque excess near the planet, which is not surprising since N 6= 0 in this model allowing buoyancy waves to be excited. 4. DISCUSSION The phenomenon of torque cut-off in 2D disks is due to the fact that high-ky modes (ky & h−1, excited close to the planet) of rotation-modified sound waves cou- ple poorly to the planetary potential (even though it is strongest there). However, the dispersion relation for buoyancy waves is very different from that of the waves in 2D disks. Based on Eq. (10), the pattern in our simu- lations should have ky ≈ h−1 at x ∼ 0.05h and z = 0.1h, resulting in strong coupling between buoyancy waves and the planetary potential. This may explain the lack of torque cutoff in models with non-zero N . A further in- vestigation will be presented in Zhu et al. (in prep.). Since the buoyancy torque is present very close to the planet, its strength could be sensitive to the potential softening length rs in the simulation. Thus, we re-ran model IA with a smaller smoothing length (rs = 1/16h) but obtained essentially the same result (purple curves in Fig. 2), suggesting that our numerical results are robust. 4.1. Comparison with previous studies Previous analytical and semi-analytical studies of buoyancy waves (Lubow & Pringle 1993, Korycansky & Pringle 1995, Lubow & Ogilvie 1998) have been limited to axisymmetric waves or low-m modes excited far away from the planet, at x ≫ h. These studies have demon- strated such modes to play rather insignificant role in planet-disk angular momentum exchange. The main result of this work is that coupling of the planetary potential to the locally-excited (x . h) buoy- ancy waves channels a considerable amount of angular momentum into these modes. As a result, the buoyancy torque strongly contributes to the planetary torque, at the level of tens of per cent. This result is different from analytical expectations extrapolated from low-m buoy- ancy modes (Lubow & Ogilvie 1998). Most previous 3D simulations of planet-disk interac- tion explored isothermal disk structure with an isother- mal EOS (e.g. Bate et al. 2003, D'Angelo & Lubow 2010). An adiabatic EOS has been explored in Paardekooper & Mellema (2006). However, because of the global geometry used in their work, the buoyancy torque may have been hard to distinguish from the coro- tation torque, see §4.3. 4.2. Existence and Signatures of Buoyancy Waves The vertical structure of protoplanetary disks around T Tauri stars should be close to isothermal given the radial gradients of specific vortensity and entropy across the horseshoe region (Paardekooper & Mellema 2006; Baruteau & Masset 2008; Paardekooper & Papaloizou 2008). By properly setting the disk properties to nullify these gradients one can eliminate the corotation torque in global simulations, thus isolating the contribution due to the buoyancy torque. 4.4. Wave dissipation and gap opening Gap opening by massive planets depends not only on the wave excitation but also on wave damping as a means of transferring angular momentum to the disk fluid (Lu- nine & Stevenson 1982; Rafikov 2002). The global den- sity wake excited by planet is thought to dissipate pri- marily via shock damping (Goodman & Rafikov 2001), but buoyancy waves are very distinct from rotation- modified sound waves and should dissipate differently (Lubow & Ogilvie 1998; Bate et al. 2002). Channeling of the wave action (Lubow & Ogilvie 1998) may con- siderably speed up their nonlinear evolution resulting in more efficient damping. Since we showed that buoyancy waves carry good fraction of the total angular momen- tum flux, understanding their damping mechanism and spatial pattern of dissipation may be important for clar- ifying the issue of gap opening by planets (Zhu et al. in prep). Our work suggests buoyancy waves can play an im- portant role in planet migration and gap opening which demands further studies. Authors are indebted to Jeremy Goodman, Steve Lubow, and Gordon Ogilvie for helpful comments and suggestions. This work was supported by NSF grant AST-0908269 and Princeton University. This research was supported in part by the NSF through TeraGrid resources provided by the Texas Advanced Computing Center and the National Institute for Computational Sci- ence under grant number TG-AST090106. 4 dominant role of stellar irradiation in their thermal bal- ance. Existence of buoyancy waves in such disks de- pends on the thermodynamics of the fluid perturbations on timescales ∼ N −1. Density perturbations induced by planets result in temperature fluctuations which tend to be erased by radiative diffusion. If the cooling time tc of such perturbations is longer than N −1 then modes are adiabatic with γ > 1 (as in models IA and IA2) and buoyancy waves should be present. In the opposite case of tcN . 1 fluid perturbations behave as isothermal (γ ≈ 1) and buoyancy waves are not excited, as in model II. We expect that the separation between the two regimes should occur between several to several tens of AU, de- pending on the disk properties, and that buoyancy waves should be efficiently excited by relatively close-in planets. Further out tcN . 1 and excitation of buoyancy waves should be suppressed. Note, that both tc and N are in general functions of z, see equation (9). This makes cool- ing considerations dependent not only on the radius but also on the height in the disk. Whenever buoyancy waves are efficiently excited by planets they can reveal themselves via associated ver- tical motions (see Fig. 1) which should give rise to cor- rugations of the disk surface aligned with density ridges seen in Fig. 3. Near-IR imaging of stellar light scat- tered by such perturbations at very high resolution (on scales ∼ h) can reveal even low-mass planets embedded in protoplanetary disks. 4.3. Migration The rate at which a planet migrates is determined by the imbalance of one-sided torques exerted on it by the inner and outer parts of the disk, caused by the global ra- dial gradients of the disk density and temperature (Ward 1986). The net torque can also be caused by the local asymmetries in the disk properties caused by the pres- ence of the planet itself, such as the temperature pertur- bations due to shadowing near the planet (Jang-Condell & Sasselov 2005), or the planet's own heat output caused by e.g. the accretion of planetesimals. Such asymmetries generated very close to the planet may not strongly affect the net standard Lindblad torque because of the torque cutoff at such separations. However, their effect on the net buoyancy torque excited primarily at small x may be disproportionately large and considerably affect the planetary migration. This statement is only true if the buoyancy waves ob- served in this work can propagate and deposit their angu- lar momentum far from the corotation region. Otherwise their angular momentum accumulates in the narrow an- nulus around the planetary orbit and the corresponding torque on the planet may be subject to saturation, like the standard corotation torque (Balmforth & Korycan- sky 2001; Masset 2001, 2002). This would eliminate the effect of the buoyancy waves on the planetary migration. We will investigate these possibilities in Zhu et al. (in prep). Direct measurement of the buoyancy torque effect on the migration speed would require global disk-planet cal- culations, in which corotation torque appears as well. It may then be difficult to separate the effects of the buoy- ancy and corotation torques in global simulations. How- ever, the strength of the corotation torque depends on the planet-disk TABLE 1 Models 5 Case II AA AAper IA IAper IA (64/h) IAss (64/h, rs=1/16 h) IA2 PA structure EOS isothermal poly. Γ = 5/3 poly. Γ = 5/3 isothermal isothermal isothermal isothermal isothermal poly. Γ = 7/5 γ 1 5/3 5/3 5/3 5/3 5/3 5/3 7/5 5/3 y-direc. boundary in/outflow in/outflow periodic in/outflow periodic in/outflow in/outflow in/outflow in/outflow Brunt-Vaisala N 0 0 0 6=0 6=0 6=0 6=0 6=0 6=0 Torque coefficient (q)a Z domain size 0.34 0.44 0.44 0.48 0.46 0.48 0.48 0.41 0.45 [0, 5 h] [0, 2 h] [0,2 h] [0, 3.873 h] [0, 3.873h] [0, 3.873h] [0, 3.873h] [0, 3.873h] [0, 2.5h] Resolution X×Y×Z 160×1280×160 160×1280×64 160×1280×64 160×1280×160 160×1280×160 320×2560×320 320×2560×320 160×1280×160 160×1280×80 a q as in T = q(GMp)2Σ0RpΩ/c3 s,adi REFERENCES Artymowicz, P. 1993, ApJ, 419, 166 Balmforth, N. J., & Korycansky, D. G. 2001, MNRAS, 326, 833 Baruteau, C., & Masset, F. 2008, ApJ, 672, 1054 Bate, M. R., Ogilvie, G. I., Lubow, S. H., & Pringle, J. E. 2002, MNRAS, 332, 575 Bate, M. R., Lubow, S. H., Ogilvie, G. I., & Miller, K. A. 2003, MNRAS, 341, 213 D'Angelo, G., & Lubow, S. H. 2010, ApJ, 724, 730 Dong, R., Rafikov, R. R., Stone, J. M., & Petrovich, C. 2011, ApJ, 741, 56 Chiang, E. I., & Goldreich, P. 1997, ApJ, 490, 368 Goodman, J., & Rafikov, R. R. 2001, ApJ, 552, 793 Goldreich, P., & Tremaine, S. 1979, ApJ, 233, 857 Goldreich, P., & Tremaine, S. 1980, ApJ, 241, 425 Jang-Condell, H., & Sasselov, D. D. 2005, ApJ, 619, 1123 Kenyon, S. J., & Hartmann, L. 1987, ApJ, 323, 714 Korycansky, D. G., & Pringle, J. E. 1995, MNRAS, 272, 618 Lin, D. N. C., & Papaloizou, J. C. B. 1993, Protostars and Planets III, 749 Lin, D. N. C., Papaloizou, J. C. B., & Savonije, G. J. 1990, ApJ, 364, 326 Lubow, S. H., & Ogilvie, G. I. 1998, ApJ, 504, 983 Lubow, S. H., & Pringle, J. E. 1993, ApJ, 409, 360 Lunine, J. I. & Stevenson, D. J. 1982, Icarus, 52, 14 Masset, F. S. 2001, ApJ, 558, 453 Masset, F. S. 2002, A&A, 387, 605 Ogilvie, G. I., & Lubow, S. H. 2002, MNRAS, 330, 950 Paardekooper, S.-J., & Mellema, G. 2006, A&A, 459, L17 Paardekooper, S.-J., & Papaloizou, J. C. B. 2008, A&A, 485, 877 Rafikov, R. R. 2002, ApJ, 572, 566 Rafikov, R. R. & Petrovich, C. 2012, ApJ, 747, 24 Stone, J. M., Gardiner, T. A., Teuben, P., Hawley, J. F., & Simon, J. B. 2008, ApJS, 178, 137 Stone, J. M., & Gardiner, T. A. 2010, ApJS, 189, 142 Takeuchi, T., & Miyama, S. M. 1998, PASJ, 50, 141 Tanaka, H., Takeuchi, T., & Ward, W. R. 2002, ApJ, 565, 1257 Ward, W. R. 1986, Icarus, 67, 164 Ward, W. R. 1997, Icarus, 126, 261 6 Fig. 1. -- The iso-density contour of ρ = 0.8ρ00 for an IA disk model. Planetary mass is increased to Mp = 0.3Mth to visually enhance the effect of buoyancy waves. Buoyancy waves cause ray-like density disturbances close to x = 0 (right side of the box), which are also shown in Fig. 3. Two streamlines passing through (-0.5, 0, 1)(blue) and (-3, 0, 1)(red) demonstrate vertical oscillation due to buoyancy waves and the small velocity perturbation of sound waves. planet-disk 7 Fig. 2. -- Torque densities (dT /dx, upper panels) and integrated torques (lower panels) for disks with isothermal disk structure (left panels) and polytropic disk structure (right panels) with different EOS. Models II (green curves) and AA (orange curves) not supporting buoyancy waves exhibit a clear torque cut-off at x . h. All other models support buoyancy waves and show significant buoyancy torque near the planet. 8 Fig. 3. -- Upper panels: (volume) density of the planet-induced torque integrated over y-coordinate for models II (left panel) and IA (right panel). Strong excess torque due to buoyancy waves close to the planet is clearly visible at z ∼ h in model IA, which features non-zero Brunt-Vaisla frequency. Lower panels: density contours in the xy plane at z where ρ0 equal to 0.8755 ρ00 (indicated by the dashed line at z ∼ 0.5h in (a) and z ∼ 0.4h in (b), h are different in these two cases). Geometric locations corresponding to the resonance condition for buoyancy waves (11) are shown by dotted lines in model IA, and agree quite well with the pattern of density fluctuations derived from simulations. planet-disk 9 Fig. 4. -- Upper panels: density fluctuations in the yz plane at x = 0.5h for models II (left panel) and IA (right panel). Lower panels: vertical velocity vz at x = 0.5h for models II and IA. The buoyancy resonance positions (11) are again plotted as dotted curves. The fluctuations of vz at z > 3h are due to the lack of exact hydrostatic equilibrium at the upper boundary.
1208.0598	2	1208	2012-10-09T21:48:11	Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1	[ "astro-ph.EP" ]	Discovered in October 2010 by the LINEAR survey, P/2010 TO20 LINEAR-Grauer (LG) was initially classified as an inert Jupiter Trojan. Subsequent observations obtained in October 2011 revealed LG to be a Jupiter-family comet. LG has one of the largest perihelia (q=5.1 AU) and lowest eccentricities (e=0.09) of the known JFCs. We report on observations of LG taken on 29 October 2011 and numerical simulations of its orbital evolution. Analysis of our data reveals that LG has a small nucleus (<3 km in radius) with broadband colours (B-R=0.99+/-0.06 mag, V-R=0.47+/-0.06 mag) typical of JFCs. We find a model dependent mass-loss rate close to 100 kg/s, most likely powered by water-ice sublimation. Our numerical simulation indicate that the orbit of LG is unstable on very short (10 to 100 yr) timescales and suggest this object has recently evolved into its current location from a more distant, Centaur-type orbit. The orbit, dynamics and activity of LG share similarities with the well known case of comet 29P/Schwassmann-Wachmann 1.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1 -- 10 (2012) Printed 8 June 2021 (MN LATEX style file v2.2) Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1 Pedro Lacerda1⋆† 1Astrophysics Research Centre, School of Mathematics and Physics, Queen's University Belfast, Belfast BT7 1NN Accepted for publication on MNRAS. Accepted 2012 October 4. Received 2012 October 4; in original form 2012 July 24 ABSTRACT Discovered in October 2010 by the LINEAR survey, P/2010 TO20 LINEAR- Grauer (P/LG) was initially classified as an inert Jupiter Trojan. Subsequent ob- servations obtained in October 2011 revealed P/LG to be a Jupiter-family comet. P/LG has one of the largest perihelia (q = 5.1 au) and lowest eccentricities (e = 0.09) of the known JFCs. We report on observations of P/LG taken on 29 October 2011 and numerical simulations of its orbital evolution. Analysis of our data reveals that P/LG has a small nucleus (< 3 km in radius) with broadband colours (B − R = 0.99 ± 0.06 mag, V − R = 0.47 ± 0.06 mag) typical of JFCs. We find a model dependent mass-loss rate close to 100 kg s−1, most likely powered by water-ice sublimation. Our numerical simulation indicate that the orbit of P/LG is unstable on very short (10 to 100 yr) timescales and suggest this object has recently evolved into its current location from a more distant, Centaur-type orbit. The orbit, dynamics and activity of P/LG share similarities with the well known case of comet 29P/Schwassmann-Wachmann 1. Key words: comets: individual (P/2010 TO20 LINEAR-Grauer) -- methods: data analysis -- minor planets, asteroids -- solar system: general -- techniques: photomet- ric -- methods: n-body simulations. 1 INTRODUCTION Jupiter family comets (JFCs) originate in the transneptu- nian region of the solar system known as the Kuiper belt. Kuiper belt objects (KBOs) preserve key information about the epoch of planetesimal formation. The low ambient tem- perature (∼ 40 K) allows KBOs to retain most if not all of the ices present in their formation environment. This makes them valuable time capsules with which to study an epoch long gone. JFCs represent the small end of the steep size distribution of KBOs. They are survivors of a dynamically intermediate population, the Centaurs, that were neither ejected from the solar system nor collided with one of the giant planets (Jupiter to Neptune) on their journey from the Kuiper belt into the inner solar system. The surfaces of JFCs are heavily processed when com- pared to KBOs, mainly by sublimation of surface ice. Indeed, the optical broadband colours of the two populations dif- fer substantially: JFCs have typically solar (neutral) colours (B − R ∼ 1.3, g − r ∼ 0.6), with a spread of about 0.2 mag (Lamy & Toth 2009; Solontoi et al. 2012) while KBOs span ⋆ E-mail: [email protected] † Some of the presented data were obtained at the European Southern Observatory (ESO) facilities at La Silla under pro- gramme 088.C-0634A. c(cid:13) 2012 RAS a broad range of optical colours, from neutral (B−R ∼ 1) to very red (B−R ∼ 2.5, Jewitt & Luu 2001; Jewitt 2002). The neutral surfaces are usually attributed to a fresh ice coating while very red surfaces are thought to be the end product of irradiation of initially neutral material (Thompson et al. 1987; Moroz et al. 1998). The intermediate Centaurs present a peculiar, also intermediate distribution of colours which includes neutral objects (blue group) and very red objects (red group) but very few cases in between (Peixinho et al. 2003). This Centaur colour bimodality may be related to their dynamical history (Melita & Licandro 2012) or simply to the fact that they are small (Peixinho et al. 2012). Some objects straddle the line between Centaurs and JFCs. They are active, like JFCs, but have Centaur-like or- bits. A famous example is 29P/Schwassmann-Wachmann 1 (hereafter 29P). The orbit of 29P is nearly circular (e = 0.04) with perihelion and semimajor axis beyond Jupiter (q = 5.7 au and a = 6.0 au), and obeys the dynamical definition of a Centaur. 29P is constantly active and displays spo- radic photometric outbursts (Roemer 1958; Jewitt 1990; Trigo-Rodr´ıguez et al. 2008). The activity of 29P is mainly driven by carbon monoxide sublimation (Senay & Jewitt 1994; Crovisier et al. 1995). Comet 29P has a radius r = 23 ± 3 km (Yan Fern´andez, private comm.). In this paper we present a study of comet P/2010 TO20 LINEAR-Grauer (hereafter P/LG) which is in many re- 2 Pedro Lacerda 10 5 0 -5 s t i n U l a c i m o n o r t s A 2010-10-03 q 10 5 0 -5 q 2011-10-29 2 OBSERVATIONS -10 -10 0 -5 5 Astronomical Units 10 -10 -10 0 -5 5 Astronomical Units 10 Figure 1. Osculating orbit of P/LG on 3 October 2010 (near the time it was discovered by LINEAR) and on 29 October 2011, when the observations reported here were taken. Due to the proximity to Jupiter the osculating orbit of P/LG changed significantly in a period of only one year. An arrow marks the direction to the last perihelion, the orbits of Earth, Mars, Jupiter and Saturn are plotted as dashed lines and the axes are labelled in astronomical units. Table 1. Orbital properties of LINEAR-Grauer on 2011 Oct 29. Property Semimajor axis, a Eccentricity, e Inclination, i Argument of perihelion, ω Longitude of ascending node, Ω Mean anomaly, M True anomaly, ν Last perihelion passage Perihelion distance, q Aphelion distance, Q Value 5.610 au 0.087 2.◦628 252.◦9 44.◦03 84.◦22 94.◦19 2008 Sep 5 5.122 au 6.097 au spects similar to 29P. Comet P/LG has a low eccentricity (e = 0.09), high perihelion (q = 5.1 au) orbit, very close to meeting the Centaur definition (see Table 1, Fig. 1). When P/LG was discovered by the Lincoln Near Earth Asteroid Research (LINEAR) survey, on 1 October 2010, it was mis- classified as a Trojan and its weak activity went unnoticed. We present optical broadband measurements of P/LG and numerical simulations of its orbital evolution. The former show that P/LG is weakly active, at a level comparable to the active Centaurs (Jewitt 2009) while the latter indicate that P/LG has in all likelihood recently evolved into its cur- rent location from a Centaur orbit. The implication is that P/LG may offer a glimpse of a relatively small and fresh Centaur object. Table 2. Observing Geometry. Property 2010 October 03 2011 October 29 Solar phase angle (α) Heliocentric distance (R) Geocentric distance (∆) 1′′ at distance ∆ 2.07◦ 5.295 au 4.309 au 3135 km 0.89◦ 5.603 au 4.613 au 3355 km P/LG was observed on 29 October 2011 at the ESO New Technology Telescope (NTT) located at the La Silla Ob- servatory, Chile. The night was photometric and the see- ing varied between 0.8 and 1.0′′. At the NTT we used the EFOSC2 instrument (Buzzoni et al. 1984; Snodgrass et al. 2008) which is installed at the f/11 Nasmyth focus and is equipped with a LORAL 2048×2048 CCD. We used the 2×2 binning mode to bring the effective pixel scale to 0.24′′/pixel. Our observations were taken through Bessel B, V, R filters (ESO #639, #641, #642, respectively). The images of P/LG were collected in a relatively regu- lar fashion, in sets of three consecutive exposures per filter, with exposure times of 240 s for B and 120 s for V and R. In total, we collected 12 images in the B band, 12 in V , and 18 in R. Throughout the observations of P/LG, the telescope was set to track the non-sidereal motion of P/LG at an approximate rate of −18′′/hour in right ascension and −6′′/hour in declination. Bias calibration frames and dithered twilight flats through all three filters were collected on the same night as the science data. The reduction of the science images, consisting of standard bias subtraction and flat fielding, was done using the IRAF ccdproc routines. The R band images suffered from fringing which was removed using an IRAF package optimised for EFOSC2 that was kindly supplied by Colin Snodgrass. The P/LG data were absolutely calibrated using observations of Landolt (1992) stars taken throughout the night. P/LG was observed as part of the Pan-STARRS PS1 all-sky survey very near the time of its discovery by LIN- EAR. Located on Haleakala, Maui, the 1.8-m PS1 telescope is equipped with a 1.4 gigapixel camera covering 3.2◦ × 3.2◦ on the sky. The survey repeatedly covers the 3π steradians of sky visible from Haleakala. PS1 uses a photometric system that approaches the SDSS filter system with the addition of a wide (w) band filter which roughly corresponds to the com- bined pass band of the gri filters (Tonry et al. 2012). The PS1 gigapixel camera has a pixel scale 0.25′′/pixel. P/LG was imaged in four consecutive 45 s exposures taken through the w filter on 3 October 2010. All images were processed au- tomatically by the Pan-STARRS Image Processing Pipeline. 3 COMET MORPHOLOGY Figure 2 shows R- and B-band summed stacks of P/LG and Fig. 3 shows enlarged sections of the region surrounding the nucleus of P/LG, to highlight details. A description of how the stacks were assembled can be found in §4.1 and §4.3; here we focus solely on inspecting the overall appearance of the comet. The projected comet tail is clearly visible and extends for more than 1′ in the direction opposite the or- bital motion. Due to the very small solar phase angle, the antisolar direction points almost radially away from the ob- server. As a result, the projected, antisolar component of the cometary tail is tiny, and more easily viewed in the enlarged, median-filtered version shown in Fig. 3. The bulk of the tail particles trails the comet as a result of Keplerian shear. The Laplacian-filtered image (Fig. 3) displays a darker ridge at the centre of the tail which originates in the nucleus and is c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1 3 Figure 2. Summed stacks of 9 R-band (left) and 6 B-band (right) images of comet P/LG. The projected tail is more than 1′ long and points in the direction opposite the orbital motion (−v). Because of the very small solar phase angle (Table 2), the antisolar direction lies nearly perpendicularly to the plane of the sky. For this reason, the near-nucleus tail has only a small component in the projected antisolar direction (−⊙; see Fig. 3) which rapidly bends towards the direction opposite the orbital motion. Figure 3. Enlarged versions of the near-nucleus tail structure from the R-band stack shown in Fig. 2. The original image (left) is shown together with versions processed using Laplacian (centre) and median (right) filters. A trail of larger particles leaving the nucleus and trailing the comet's orbital motion is visible in the Laplacian-filtered image. The median-filtered image highlights the near-nucleus part of the tail which points in the projected antisolar direction. aligned with the orbital motion. The ridge is probably due to large particles leaving the nucleus. Larger particles attain lower ejection velocities as they couple more weakly to the sublimating gas and hence tend to remain concentrated in the plane of the orbit where solar radiation acts to push them radially outwards and Keplerian shear forces them to trail the nucleus. Figure 4 shows a stack of four consecutive w-band im- ages of comet P/LG taken by the PS1 survey. The PS1 im- ages are taken only 2 days after discovery by LINEAR on 1 October 2010. The comet appears active, with a tail that ex- tends about 12′′ on the sky. Here, as in the case of our more recent observations, the solar phase angle is small, just over 2◦, probably causing the tail to extend mainly into the plane of the sky along the line of sight. As above, the near-nucleus portion of the sky-projected tail seems to be antisolar, but further from the nucleus the tail seems to trail the nucleus along the orbital direction. The three sets of images show that P/LG was active at the time of discovery (October 2010) and remains active a year later (October 2011). 4 PHOTOMETRY We performed aperture photometry of comet P/LG and nearby field stars using the IRAF apphot task with four main goals: 1) to estimate the size of the comet nucleus, 2) to estimate the mass-loss rate from the comet, 3) to mea- sure the colour of the nucleus and coma dust, and 4) to search for brightness variations in the nucleus and coma over time. With those goals in mind, we extracted photometric data from the science images using four synthetic apertures c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 -Ÿ N -v E 3 October 2010, PS1, w band 1.6' ´ 0.9' Figure 4. Stack of four consecutive PanSTARRS w-band im- ages of P/LG taken on 3 October 2010, very near the time of its discovery by the LINEAR survey. The comet is clearly active in these images. defining four regions of interest (see Table 3). The central region, R4, 4 pixels (0.96′′) in radius, was selected to rep- resent the flux from the nucleus, but includes an unknown fraction of light due to the coma. The intermediate annu- lus, R4,8, between 4 and 8 pixels (0.96′′ and 1.92′′) from the nucleus includes contributions from both the nucleus point spread function (PSF) and the coma. The annular region R8,13, located between 8 and 13 pixels (1.92′′ and 3.12′′) from the nucleus, is dominated by the coma. Finally, the outer annulus R13,20, located between 13 and 20 pixels from the nucleus (3.12′′ and 4.8′′), includes coma flux and significant noise contamination. Details of how these partic- ular apertures were chosen can be found in §4.1, §4.2 and 4 Pedro Lacerda Table 3. Photometry Regions Region Label Region Shape R4 R4,8 R8,13 R13,20 Circle Annulus Annulus Annulus Aperture radius Aperture radius Projected radius [pixels] r = 4 [′′] [km] φ = 0.96 d = 3, 220 r = 4 to 8 r = 8 to 13 r = 13 to 20 φ = 0.96 to 1.92 φ = 1.92 to 3.12 φ = 3.12 to 4.80 d = 3, 220 to 6,440 d = 6, 440 to 10,465 d = 10, 465 to 16,100 Magnitudes in Region Dominant Source m4 m4,8 m8,13 m13,20 Nucleus (+ Coma) Nucleus + Coma Coma Coma Table 4. Photometry Region measured B band [mag] V band [mag] R band [mag] m4 m4,8 m8,13 m13,20 21.94 ± 0.05 22.31 ± 0.08 22.56 ± 0.16 22.72 ± 0.30 21.42 ± 0.05 21.66 ± 0.08 21.80 ± 0.13 22.04 ± 0.27 20.95 ± 0.03 21.21 ± 0.05 21.33 ± 0.08 21.47 ± 0.15 §4.3. Magnitudes measured in these regions are labeled m4, m4,8, m8,13, and m13,20. Magnitudes within the annuli, la- beled min,out, were calculated from the magnitudes within the inner and outer apertures, min and mout, through min,out = −2.5 log10 (cid:0)10−0.4mout − 10−0.4min(cid:1) . We list the B, V and R apparent magnitudes in each region in Table 4. 4.1 Nucleus Size To estimate the radius of the nucleus of comet P/LG we began by generating a summed stack of 9 R-band frames that were shifted and aligned at the comet nucleus position. The stack, displayed in Fig. 2, has an equivalent integration time of 1080 s and offers an enhanced signal-to-noise ratio (SNR) detection of the comet. We used the R-band images because they are sharper (better seeing) and have higher SNR than those in B and V . The particular frames that were stacked have seeing between 0.8′′ and 0.9′′. An analo- gous stack that aligns the field stars instead of the comet was used to investigate the best choice of aperture for the pho- tometry. We found that a 4-pixel radius aperture, containing ∼ 85% of the flux of a point source, offers the best com- promise between maximising the SNR within the aperture and minimising the contribution from surrounding sources. We labeled this region R4 and used the magnitude within it, m4, as best representing the flux from the nucleus. In the case of comet P/LG we obtain a SNRR ∼ 75 within the central region R4 and measure an R-band magnitude m4 = 20.95 ± 0.03 mag. Assuming a spherical cometary nucleus with cross- section πr2 e, we calculated the equivalent radius re using the relation (Russell 1916): πr2 e pR 10−0.4βα = 2.25 × 1022 π R2 ∆2 10−0.4(m4 −m⊙) (1) where pR is the R-band geometric albedo and β is the linear phase coefficient of the nucleus. Table 2 lists the values of the solar phase angle, α, heliocentric distance, R, and geocentric distance, ∆, of P/LG on 29 October 2011. The apparent red magnitude of the Sun at Earth is m⊙ = −27.1 mag. The values of pR and β are unknown for P/LG. The uncertainty introduced by β is negligible when compared to that in pR (uncertain by a factor 2 or more) because P/LG was ob- served at very low phase angle (α < 1◦). We used β = 0.02 mag/◦(Millis, Ahearn & Thompson 1982; Meech & Jewitt 1987) and pR = 0.1 (Kolokolova et al. 2004) and found an equivalent radius re < 3 km. This figure is an upper limit due to the unknown contribution of dust coma to the flux within the central aperture. The apparent magnitude of P/LG may be converted to a nucleus absolute magnitude, m(1, 1, 0) (the theoretical magnitude of the nucleus when placed at 1 au from the Sun and the Earth, and seen at 0◦ phase angle), using: (2) m(1, 1, 0) = m4 − 5 log10(R∆) − βα. Substituting the quantities R, ∆ and α from Table 2 and again using β = 0.02 mag/◦ we obtain mR(1, 1, 0) = 13.04± 0.03 mag. The formal error in m(1, 1, 0) is the same as that in the apparent red magnitude, but the true uncertainty is higher due to the unknown value of β. We also note that this is a lower limit to the absolute magnitude of the P/LG nucleus because of the unknown contribution of near-nucleus coma to the magnitude m4. 4.2 Mass-loss rate An order of magnitude estimate of the mass-loss rate from P/LG can be calculated by dividing the total dust mass present within a coma-dominated annulus surrounding the nucleus by the time it takes the dust to cross the annu- lus (Jewitt & Luu 1989). As in §4.1 we used the 9-frame, R-band stack to measure the amount of sunlight-reflecting dust within the annular region. The annulus was set be- tween 8 and 13 pixels from the nucleus photocentre and labeled R8,13 (Table 3). The annulus dimensions were found by experimentation (using the profiles of field stars in the star-centred stack) to ensure that the flux within the annu- lus is dominated by the coma and sufficient to overcome the noise due to the sky background. In our stacked data, only 3% of the flux of a point source is present in the wings of the PSF beyond an aperture of 8 pixels. The R-band magnitude within the annulus R8,13 is m8,13 = 20.95 ± 0.03 mag (see Table 4). The magnitude m8,13 can be converted to a total dust cross-section, Ad, using Eq. 1 if we use the latter instead of the nucleus cross- section πr2 e. In this case, pR and β are the R-band albedo and linear phase coefficient of the dust particles, also unknown for P/LG. We assumed that these values are the same as for the nucleus (β = 0.02 mag/◦ and pR = 0.1) and found a total dust cross-section area Ad ∼ 2 × 107 m2. This fig- c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1 5 R ¯ B ¯ V ¯ ae à ì à ¬ D m8,13 ® ì ae ae à ì ae ae ì à ae ae à ì ae à ì ¬ D m4,8 ® ae à ì ae ì ae ae à ì ae à ì ¬ D m4 ® ae ì ae ae ì à à à ae ae 7.0 Projected Distance from Nucleus pixels 0 4 8 13 20 e d u t i n g a M e v i t a l e R 2.0 1.5 1.0 0.5 0.0 r u o l o C R -- B 1.5 1.4 1.3 1.2 1.1 1.0 0.9 0 1 2 3 4 Projected Distance from Nucleus arcsec 3.5 4.0 5.0 4.5 6.5 UT Hours on 29 October 2012 5.5 6.0 Figure 5. Radial variation of the B − R colour (thick horizontal lines) and uncertainty (grey boxes). The average colours (thick horizontal lines) are shown in each of the four concentric regions described in Table 3. The coma-dominated annuli (apertures > 8 pixels) appear redder than the central region despite the large uncertainties. ure is uncertain by at least a factor of 2 mainly due to the uncertainty in the albedo. To convert the dust cross-section, Ad, to a dust mass, Md, we assume that the dust consists of equal-sized spheres with an equivalent radius, rd, representative of their true size distribution (e.g., Li et al. 2011). The mass is then given by Md = (4/3)ρdrdAd (3) where ρd is the bulk density of the dust particles. We take an equivalent dust particle radius of rd = (0.1 µm × 1 cm)1/2 ≈ 32 µm which is based on a power-law dust size distribu- tion proportional to (dust grain radius)−3.5 with minimum and maximum grain radii of 0.1 µm and 1 cm (Jewitt 2009; Li et al. 2011). We further assume a dust bulk den- sity ρd = 1000 kg m−3 and find a total dust mass within the annulus Md ∼ 8.6 × 105 kg. The annulus crossing time, tcross, is obtained by divid- ing the annulus projected width, wcross = 4 × 106 m (see Table 3), by the dust velocity, vd. The velocity vd is highly uncertain and depends on grain size (Crifo et al. 2004) but estimates based on macroscopic fragment ejection from 17P/Holmes (Stevenson, Kleyna & Jewitt 2010) and on the coma expansion velocities of 17P/Holmes (Montalto et al. 2008; Hsieh et al. 2010) and C/Hale-Bopp (Biver et al. 2002) vary from a few 100 m s−1 to 1000 m s−1. We take an intermediate vd = 500 m s−1 and obtain an annulus crossing time tcross = wcross vd ∼ 0.8 × 104 s. (4) Finally, the mass loss rate is given by dMd/dt = Md/tcross ∼ 1.1 × 102 kg/s. At this rate, a nucleus of radius re = 3 km (see §4.1) and bulk density ρnuc = 1000 kg m−3 would have a sublimation lifetime tsub = 3× 104 yr, significantly shorter than the dynamical lifetime of P/LG (see §5.2). c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 Figure 6. Lightcurves of comet P/LG sampling the central re- gion R4 (∆m4, measured within a 4 pixel or 0.96′′ radius aper- ture), the intermediate region R4,8 (∆m4,8, from 0.96 to 1.92′′), and the coma-dominated region R8,13 (∆m8,13, from 1.92 to 3.12′′). For each region, we plot relative magnitude lightcurves in bands B (blue squares), V (green diamonds) and R (red circles). The lightcurves in different bands were aligned simply by median subtraction, and the lightcurves in each region (∆m4, ∆m4,8, ∆m8,13) were shifted vertically for clarity of presentation. 4.3 Nucleus and Coma Colour We investigated the possibility that the nucleus and coma dust of P/LG have different colours by comparing the fluxes through different filters within the regions detailed in Table 3. To do so, we generated B- and V -band stacks as described in §4.1 using the frames with best image quality (6 × B- band, seeing 1.0′′ to 1.1′′; 6 × V -band, seeing 0.8′′ to 0.9′′) to obtain single images with equivalent exposure times 1440 s in B (see Fig. 2) and 720 s in V . The resulting stacked images have SNRB ∼ 32 and SNRV ∼ 34 within the central aperture. Figure 5 plots the B − R colour of P/LG versus projected distance from the nucleus, and Table 4 lists the B, V and R apparent magnitudes in the regions of interest. The comet nucleus region, R4, has colours B − R = 0.99 ± 0.06 mag and V − R = 0.47 ± 0.06 mag, typical of Jupiter family comet nuclei (Jewitt 2002; Snodgrass, Lowry & Fitzsimmons 2006). Further from the nucleus, as the flux becomes dominated by the coma, the B − R colour becomes slightly redder. Despite the large un- certainties the effect is noticeable and suggests that the coma dust may have a different colour from the nucleus. 4.4 Lightcurves We searched for photometric variability in the P/LG data using the individual frames taken in B, V and R over the course of the night. Periodic photometric variations from the central region can signal a rotating, elongated nucleus, while temporal variations in the outer coma regions may indicate variable activity from the comet. To look for pho- tometric variations we performed photometry within the re- gions listed in Table 3 for each frame. As part of our observ- ing strategy we acquired sets of three consecutive images per 6 Pedro Lacerda filter. That allowed to take the median magnitude of each set and divide the median uncertainty by √3. Figure 6 shows the resulting B, V and R lightcurves in regions R4, R4,8 and R8,13. Since we are only interested in the relative variability, the lightcurves for each band were median subtracted for alignment purposes. The lightcurves within each region were then shifted vertically for clarity of presentation. The small variability seen in B and R within the central region R4 is probably not significant: the magnitudes m4 in B and R display a positive correlation with atmospheric see- ing, at a level that has a probability of only 4% of occurring by chance. The large scale variation observed in m8,13, in all three bands, does not correlate with seeing nor airmass and is probably real. Although the formal photometric un- certainties are large, the variation is regular. However, our time base is short rendering the interpretation difficult. 5 DYNAMICAL PAST AND FUTURE Discovered by the LINEAR survey in October 2010, P/LG was initially classified as a Jupiter Trojan. The object re- tained its Trojan status within the JPL Horizons system until October 2011, when cometary activity was detected (Grauer et al. 2011), at which point it was reclassified as a JFC. Currently, P/LG has a low inclination (i = 2.◦6), nearly circular orbit (e = 0.09) with a semimajor axis a = 5.6 au. The difference between the distances to perihelion and aphe- lion is a mere 1 au, leading to a surface temperature varia- tion of ∆T = 15 K (blackbody in equilibrium with sunlight). Such a temperature difference seems too small to cause peri- odic activity close to perihelion as is the case for most JFCs. This issue is discussed in more detail in §6. To investigate the past and future dynamical evolution of P/LG, we numerically integrate its current orbit back- wards and forwards in time. For this purpose, we use the N- body integration package Mercury version 6.2 (Chambers 1999). We integrate the orbits of 371 objects, P/LG plus 370 clones with normally distributed orbital elements centered on P/LG's current orbit and with a 1-σ dispersion equal to 10−4 of each parametre (Fig. 7). The Sun and 8 major planets are included as massive bodies and the 371 P/LG clones as massless test particles. We use Mercury's hy- brid algorithm mode which combines a second order mixed- variable symplectic algorithm with a Bulirsch-Stoer integra- tor to handle close encounters. We select an initial timestep of 8 days and remove clones that go beyond 200 au from the Sun at any point during the dynamical evolution. We note that our simulations of the orbital evolution of P/LG neglect non-gravitational forces. Collimated sublima- tion jets accelerate the nucleus and may alter the evolution presented here. An upper limit to this effect can be calcu- lated assuming that the non-gravitational acceleration, T , is due to a single sublimation jet, always aligned tangentially along the orbit of the comet, through which all mass loss occurs. In that case, the rate of change of the semimajor axis, da/dt, can be written as da/dt = 2 V a2 T / (cid:0)G M⊙(cid:1) (5) where V is the orbital velocity, G is the gravitational con- 8.696 ae 8.694 ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae øø ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae 2 - 0 1 y t i c i r t n e c c E 8.692 8.690 g e d n o i t a n i l c n I 2.6285 2.6280 2.6275 2.6270 ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae øø ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae aeae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae ae 5.608 5.609 5.610 5.611 Semimajor Axis au 8.690 8.694 8.692 Eccentricity 10-2 8.696 Figure 7. Initial orbital parametres of 370 clones of P/LG used in our numerical simulations. The clones are drawn from Gaussian distributions with centers on the actual orbital elements of P/LG (star symbol) and standard deviations equal to one thousandth of the central value. stant, M⊙ is the mass of the Sun, and the acceleration due to the jet is T = (dMd/dt) (vd/mnuc) . (6) Substituting dMd/dt = 102 kg, vd = 500 m s−1, mnuc = 1.1 × 1014 kg, and taking V as being the circular velocity at a = 5.6 au we obtain da/dt = 1.4 × 10−5 AU yr−1. In the sublimation lifetime (tsub = 3 × 104 yr) of comet P/LG its orbit can decay by (da/dt) × tsub = 0.42 au. This value is a strong upper limit, as it assumes a jet geometry contrived to achieve maximum orbital decay. 5.1 Orbit Stability Our simulations show that P/LG is dynamically unstable on very short timescales. The 1-σ spread of the main orbital parametres (semimajor axis, eccentricity, inclination) of the cloud of P/LG clone particles expands by a factor e in only 12 to 14 yr into the past and about 100 yr into the future. These times provide a reasonable estimate of the Lyapunov timescale for P/LG. Due to the strongly chaotic nature of the orbit of P/LG, we can expect at best to obtain a statistical assessment of the past and future evolution of this comet. The chaotic evolu- tion also implies that long term integrations into the past or future are statistically equivalent (Levison & Duncan 1994; Horner, Evans & Bailey 2004). However, current close en- counters between P/LG and Jupiter (see §5.3) generate an asymmetry in the dynamical evolution making it interesting to investigate the short term past and future cases sepa- rately. Those same close encounters lead to further strong divergence of the clone orbits reinforcing the need for a more statistical analysis beyond a few kyr from the current time. 5.2 Orbit History When integrating into the past we find that 50% of the clones are removed from the simulation by t = −182 kyr. The longest lived clone reaches t = −22.5 Myr. The frac- tion of surviving clones is plotted versus time in Fig. 8 (left panel). Each step in the figure corresponds to the removal of 1 clone, so, e.g. only 3 clones survive to times earlier than t = −10 Myr. The Figure also shows the median semima- jor axis and perihelion distance of the surviving clones. The c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1 7 s e n o l C g n i v i v r u S 1 N=371 0.1 0.01 0.001 -25 -20 -15 -10 Time Myr -5 Now U A n o i t u b i r t s i d a 100 80 60 40 20 0 aeae ae aeaeaeae aeae aeaeaeae aeae aeae ae aeae ae aeae ae aeae ae aeae ae -25 -20 -15 -10 Time Myr -5 Now U A n o i t u b i r t s i d q 30 25 20 15 10 5 0 aeaeaeaeaeaeaeae aeaeaeae aeaeaeae aeae aeae ae aeae ae aeae ae aeae ae -25 -20 -15 -10 Time Myr -5 Now Figure 8. Past orbital evolution of the 371 clones of P/LG. The fraction of surviving clones (left) decreases steeply as their orbits are integrated back in time. Half the clones do not survive the orbital integration beyond t = −182 kyr. The longest surviving clone reaches −22.5 Myrs. The surviving clones evolve into Centaur/Scattered Kuiper Belt Object orbits with increasing median semimajor axis (black points in center plot) and median perihelion distance (black points in plot on the right). The gray boxes contain the central 50% of the distribution in each time bin. 1.0 0.8 C F J Halley Centaur y t i c i r t n e c c E 0.6 0.4 0.2 0.0 0 Transneptunian 20 40 60 Semimajor Axis AU 80 Figure 9. Heat map of the evolution of P/LG clones during the backwards integration described in §5. Darker patches have been occupied for a longer time by clones. The a − e space is divided into 4 regions, representative of Jupiter family comets: a < aS/(1 + e), Centaurs: aS /(1 + e) < a < aN and e < 0.8, Hal- ley comets: Centaurs with e > 0.8, and Transneptunian objects: a > aN , where aS and aN are the semimajor axes of Saturn and Neptune. Dashed lines mark the perihelia of Jupiter, Sat- urn, Uranus and Neptune. The clones spend most of the time as Centaurs, although pathways exist that transport clones to the transneptunian region. former increases with time into the past, albeit with large scatter (indicated by the gray boxes which contain the cen- tral 50% of the points). The latter increases to about q ∼ 17 au and then remains approximately constant with a tail to- wards larger q. The past evolution suggests that P/LG has recently evolved into its current orbit from a Centaur-type orbit. Figure 9 shows a heat map of the evolution of the P/LG clones in the semimajor axis vs. eccentricity plane. Darker/redder regions are more often visited by clones as their orbits evolve backwards in time. The plane is divided into four regions, taken roughly to represent four types of orbits: JFC-type, defined as having a < aS/(1 + e), Centaur- type, bound by aS/(1 + e) < a < aN and e < 0.8, Halley- type, similar to Centaur-type except with e > 0.8, and transneptunian-type, with a > aN , where aS and aN are the semimajor axes of Saturn and Neptune. These regions do not aim at accurately representing the dynamical behaviour of clone particles; they are designed simply as a means to c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 JFC Halley Centaur Transneptunian 100% s e n o l c g n i v i v r u s f o n o i t c a r F -20 Myr Time now Figure 10. Fraction of surviving clones within each of the regions (orbit types) described in Fig. 9 plotted against time. Bins are 1 Myr wide. The leftmost three bins (−17 Myr to −20 Myr) contain a single clone (see Fig. 8). classify the type of orbit occupied by the clones as a func- tion of time. The Figure shows that clones evolve mainly along constant-perihelion lines, as their semimajor axis and eccentricity increase. Figure 10 traces the history of clones across the a-e plane as a function of time. The Figure shows the fraction of surviving clones within each of the orbit-type regions described above in bins of 1 Myr, 20 Myrs into the past. As already seen in Fig. 8, the surviving clones tend to quickly move away from the JFC region into Centaur-type orbits. Beyond just a few Myrs into the past the semimajor axis of the surviving clones moves even further away from the sun, into the transneptunian region. The last surviving clone possesses a semimajor axis and eccentricity typical of a scattered transneptunian object (a ∼ 80 au, e ∼ 0.65). In Figure 11 we show the history of the Tisserand parametre of P/LG clones with respect to Jupiter and Nep- tune in our numerical simulation. The Tisserand parameter is a useful dynamical quantity that is conserved in the re- stricted three body problem. Here, the three bodies are the Sun, a planet and P/LG. In the solar system, the Tisserand parameter is not exactly conserved, but when calculated for a given pair of small body and planet it can be used to quan- tify the dynamical influence the latter has on the former. For instance, the JFCs have Tisserand parametres 2 < TJ < 3 8 Pedro Lacerda 2<TJ <3 2<TN <3 Both None 100% s e n o l c g n i v i v r u s f o n o i t c a r F -20 Myr Time now Figure 11. Evolution of the Tisserand parametre of clones. Myr-wide bins show the fractions of clones that have Tisserand parametre 2 < T ≤ 3 with respect to Jupiter, Neptune, both planets, or none. A value 2 < T ≤ 3 with respect to a given planet indicates that that planet dominates the orbit of the ob- ject. The Figure shows that, as time evolves into the past, Jupiter hands control to Neptune as main influence to the clone orbits. with respect to Jupiter, to which their orbits are strongly dynamically coupled. Main belt asteroids have TJ > 3 and are relatively stable with respect to the Jovian gas giant. The Tisserand parametre of P/LG, calculated with respect to planet P , is given by TP = aP /a + 2p(1 − e2)(a/aP ) cos i where aP is the semimajor axis of the planet and a, e and i are the semimajor axis, eccentricity and inclination of P/LG. As shown in Figure 11, within the first 1 Myr into the past the clones of P/LG are controlled partly by Jupiter (2 < TJ < 3), partly by Neptune (2 < TN < 3) and partly by neither of those two planets. The latter indicates orbits between Jupiter and Neptune. Further into the past, Jupiter becomes less important dynamically while Neptune becomes more dominant. (7) In summary, our simulation results conspire to suggest that P/LG may be a recent arrival in the inner solar system, possibly from a Centaur or transneptunian type orbit. That makes this object an interesting target of study as it may offer a glimpse of a relatively unprocessed Centaur. 5.3 Close Encounters with Jupiter and the Trojan Clouds We performed a high time-resolution simulation (0.1 yr ini- tial timestep) going back only 3000 yr to investigate recent close encounters between the P/LG clones and the major planets. We focus here on Jupiter, due to its current close proximity to P/LG. In the very near past, we find that of the 371 P/LG clones, 331 (89%) had close encounters with Jupiter (< 100 planet radii) between 48 and 51 yr ago. Dur- ing that time interval, 34 clone particles came within 20 Jupiter radii (rJ ) of the giant planet and the closest ap- proach distance for any particle was 13 rJ . Two of the 331 clone particles had very close approaches to Jupiter in con- secutive orbits: one particle came within 14.0 rJ just 48 yr ago and within 8.7 rJ during the previous orbit (60 yr ago); another particle passed at 16.0 rJ and 14.6 rJ about 48 and 57 yr ago. Finally, if we consider the full 3000 yr time span of the simulation, 7 particles came within the Roche limit of Jupiter (≈ 2.7rJ assuming a fluid-like P/LG nucleus with bulk density ρnuc = 1000 kg m−3; Asphaug & Benz 1996). In a similar integration to the one just described but into the future we find that 11 out of the 371 clone particles come within 10 rJ and 3 of those come within 2 rJ roughly 145 yr into the future. A total of 41 clone particles approach Jupiter within 100 rJ at that time. We also searched for possible crossings of one of the Jo- vian Trojan clouds by P/LG clones. The Trojan clouds have an irregular shape, roughly 4 × 2 × 0.5 au (FWHM) in size (Jewitt, Trujillo & Luu 2000; Nakamura & Yoshida 2008). To identify potential crossings and estimate the time spent by clones inside the Trojan clouds we used Jupiter's posi- tion to calculate the positions of its L3 and L4 Lagrangian points in each timestep. Finally, for each clone we added the timesteps it spent within 0.3 au of either L3 or L4. We chose 0.3 au to ensure the clone is well within the densest region of the Trojan clouds. We found that clones spend an average of 10.1 of the last 3000 yr within the regions just described. 6 DISCUSSION Figure 4 shows P/LG as seen by the PanSTARRS PS1 sur- vey (see §2) close to the time the comet was discovered by LINEAR. The PS1 images show that P/LG was active at the time of discovery and yet the LINEAR pipeline first classified this object as a Jovian Trojan. This suggests that LINEAR may have misclassified several active objects as in- ert asteroids, which is an important point to consider at a time when the study of active asteroids is receiving increas- ing interest (Jewitt 2012, and references therein). Our data reveal that P/LG was active in October 2010, roughly 60◦ past perihelion, and remains active a year later, at 90◦ past perihelion. P/LG's nearly circular orbit, atypi- cal for a JFC, leads us to consider a number of explanations for its activity. One possibility is that P/LG has been acti- vated by a recent collision with a smaller object. We find this unlikely: our simulations show that P/LG spent only a neg- ligible fraction of its recent past within the Trojan clouds (but see below), and even there the chance of a collision onto its re < 3 km nucleus would be low (intrinsic colli- sion probability Pi = 6.5× 10−18 km−2 yr−1; dell'Oro et al. 1998). Another possibility is that P/LG was exposed to sig- nificant tidal stress due to close approaches to Jupiter. This scenario would require encounters with Jupiter closer than the Roche limit (mass shedding begins at d < 0.69Rroche for a comet on a parabolic orbit; Sridhar & Tremaine 1992; Asphaug & Benz 1996) which our simulations show to be improbable. It is also possible that P/LG became active due to rotational mass shedding but since the rotation period of P/LG is not known this possibility is untestable. One sce- nario that our simulations can not rule out is that P/LG originated in the Trojan population and was disloged from the 1:1 resonance with Jupiter through the action of non- gravitational effects (see §5). Conceivably, P/LG could have been collisionally actived by a smaller Trojan and led to drift from the stable region, accelerated by mass-loss jets; we note c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 Comet P/2010 TO20 LINEAR-Grauer as a Mini-29P/SW1 9 Table 5. Comparison between P/LG and 29P Comet P/LG 29P a [au] 5.6 6.0 e 0.09 0.04 i [◦] 2.6 9.4 B − R [mag] V − R [mag] re [km] M [kg] dM/dt [kg s−1] dm/dt [kg m−2 s−1] 0.99 ± 0.06 1.28 ± 0.04 0.47 ± 0.06 0.50 ± 0.03 < 3 23 ± 3 1.1 × 1014 5.1 × 1016 1.1 × 102 5.1 × 103 7.7 × 10−7 9.7 × 10−7 Columns are (1) comet ID, (2) semimajor axis, (3) eccentricity and (4) inclination, (5) and (6) broadband colours, (7) equivalent radius, (8) model dependent mass-loss rate, (9) specific mass-loss rate. Mass assumes nucleus/dust bulk density ρnuc = 1000 kg m−3. Specific mass-loss rate assumes a spherical nucleus of radius re. Comet 29P colours from Jewitt (2009). that P/LG's highly unstable orbit implies a recent departure from the Trojan region. Modelling this possibility is a com- plex task involving a number of unknown parametres and is beyond the scope of this paper but future observations of this comet may shed light on a possible link with the Trojan population. The activity of P/LG could simply be due to ice subli- mation. At first sight, the small difference between the P/LG perihelion and aphelion distances (Table 1), and the conse- quently small orbital variation of the subsolar equilibrium temperature (∆T ≈ 15 K) could seem insufficient to power periodic sublimation activity. However, between the perihe- lion and aphelion distances of P/LG, the specific mass loss rate due to water ice sublimation varies by more than an order of magnitude (from 10−5 to 10−4 kg m−2 s−1) while CO/CO2 sublimation driven mass loss varies by a factor of 2 and is around 10−2 kg m−2 s−1 (Jewitt 2009). We find that P/LG loses mass into the coma at a rate ∼ 102 kg s−1 implying that an area of 106 to 107 m2 would need to be active on the surface of P/LG if water ice sublimation is the source of the activity. Those areas are small compared to the maximum surface area of the nucleus, 4πr2 e ∼ 108 m2. Sublimation due to CO/CO2 would require only an area ∼ 104 m2 to be active, corresponding to a tiny active vent on the surface of the 3 km nucleus. We conclude that given the recent perihelion passage and the particular range of heliocentric distances traversed by P/LG since then our es- timated current mass loss rate is consistent with the activity being driven by ice sublimation. P/LG is in many ways comparable to comet 29P (see Table 5 and Fig. 12). Both comets have high perihelion (q > 5 au), low inclination (i < 10◦), nearly circular or- bits (e < 0.1) more typical of Centaurs than JFCs. Both orbits are unstable on very short timescales (a few hundred years; Horner, Evans & Bailey 2004). The two comets show similar levels of activity, with specific mass-loss rates close to dm/dt = 10−6 kg m−2 s−1, despite 29P being nearly an or- der of magnitude larger (>2 orders of magnitude more mas- sive) than P/LG. Both objects display continued activity, although the coverage of P/LG reported here is not enough to establish a pattern of activity. It will be interesting to see if P/LG's activity is significantly modulated by its or- bital motion or if it remains constant as is the case for comet 29P. It will also of interest to monitor P/LG and see if it too will display the sporadic outbursts seen in 29P. Significant differences include the size of the comets and the source of activity: 29P is dominated by CO sublimation, while P/LG is more likely active due to water-ice sublimation. c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10 1013 1014 Mass kg 1015 1016 1017 29P 1 - s g k e t a R s s o L s s a M 104 103 102 101 10- 5 10- 6 1 PLG 10- 7 10 Radius km Figure 12. Comparison between comets P/LG and 29P. Mass- loss rates are plotted against nucleus radius (and mass, assuming nucleus/dust bulk density ρnuc = 1000 kg m−3). The arrow indi- cates that we possess only an upper limit on the nucleus radius of P/LG. The uncertainty in the radius of 29P is marked by a hori- zontal error bar. The size of the P/LG and 29P points is linearly proportional to their geometric cross-section for visual compari- son. Lines of constant specific mass-loss rate are labeled in units of kg m−2 s−1. 7 CONCLUSIONS We report photometric observations and numerical simula- tions of the orbital evolution of the unusual comet P/2010 TO20 LINEAR-Grauer (P/LG). Our findings can be sum- marised as follows: (i) Comet P/LG was active at the time of discovery (Oc- tober 2010) by LINEAR, and remains active in October 2011. LINEAR did not detect the activity and initially mis- classified P/LG as a Trojan, which suggests that several ac- tive objects may have gone undetected by the survey. (ii) The nucleus of P/LG has equivalent radius re < 3 km, and colours B − R = 0.99 ± 0.06 mag and V − R = 0.47 ± 0.06 mag, values typical of Jupiter family comets. The data suggest a slight reddening of the dust colour with 10 Pedro Lacerda distance from the nucleus but the uncertainties are large. We find no significant rotational photometric variability from the nucleus region. (iii) We obtain a model-dependent estimate of the mass- loss rate from P/LG of ∼100 kg s−1. We favour water-ice sublimation as the simplest and most likely cause for activity in comet P/LG. (iv) Our numerical simulations show that the orbit of P/LG is unstable on very short timescales and suggest that it may be a Centaur that recently arrived in the inner so- lar system, although other possibilities exist involving non- gravitational effects. (v) Comet P/LG is in a number of ways reminiscent of the well-known 29P/Schwassmann-Wachmann 1. 29P is a comet/active Centaur that shows sporadic outbursts super- imposed on a background of constant activity. Comet P/LG is an order of magnitude smaller (three orders of magnitude less massive) than 29P and yet displays similar activity per unit area. Comets 29P and P/LG are interesting as possible examples of relatively unprocessed Centaurs. ACKNOWLEDGMENTS I thank the referee, Mario Melita, as well as David Jewitt and Ivo Labb´e for their helpful comments on the manuscript. I am grateful to Larry Denneau and Ken Smith for help- ful assistance with the PS1 survey data. I also acknowledge Colin Snodgrass for kindly supplying an IRAF fringe re- moval routine optimised for EFOSC2. The PS1 Surveys have been made possible through con- tributions of the Institute for Astronomy, the University of Hawaii, the Pan-STARRS Project Office, the Max-Planck Society and its participating institutes, the Max Planck In- stitute for Astronomy, Heidelberg and the Max Planck Insti- tute for Extraterrestrial Physics, Garching, The Johns Hop- kins University, Durham University, the University of Edin- burgh, Queen's University Belfast, the Harvard-Smithsonian Center for Astrophysics, and the Las Cumbres Observatory Global Telescope Network, Incorporated, the National Cen- tral University of Taiwan, and the National Aeronautics and Space Administration under Grant No. NNX08AR22G is- sued through the Planetary Science Division of the NASA Science Mission Directorate. REFERENCES Asphaug E., Benz W., 1996, Icarus, 121, 225 Biver N. et al., 2002, Earth Moon and Planets, 90, 5 Buzzoni B. et al., 1984, The Messenger, 38, 9 Chambers J. E., 1999, MNRAS, 304, 793 Crifo J. F., Fulle M., Komle N. I., Szego K., 2004, Nucleus- coma structural relationships: lessons from physical mod- els, Festou M. C., Keller H. U., Weaver H. A., eds., pp. 471 -- 503 Crovisier J., Biver N., Bockelee-Morvan D., Colom P., Jorda L., Lellouch E., Paubert G., Despois D., 1995, Icarus, 115, 213 dell'Oro A., PAolicchi F., Marzari P., Dotto E., Vanzani V., 1998, A&A, 339, 272 Grauer A. D., Sostero G., Melville I., Kasprzyk A., Howes N., Guido E., Spahr T., Williams G. V., 2011, IAUC, 9235, 1 Horner J., Evans N. W., Bailey M. E., 2004, MNRAS, 354, 798 Hsieh H. H., Fitzsimmons A., Joshi Y., Christian D., Pol- lacco D. L., 2010, MNRAS, 407, 1784 Jewitt D., 1990, ApJ, 351, 277 Jewitt D., 2009, AJ, 137, 4296 Jewitt D., 2012, AJ, 143, 66 Jewitt D., Luu J., 1989, AJ, 97, 1766 Jewitt D. C., 2002, AJ, 123, 1039 Jewitt D. C., Luu J. X., 2001, AJ, 122, 2099 Jewitt D. C., Trujillo C. A., Luu J. X., 2000, AJ, 120, 1140 Kolokolova L., Hanner M. S., Levasseur-Regourd A.-C., Gustafson B. A. S., 2004, Physical properties of cometary dust from light scattering and thermal emission, Festou M. C., Keller H. U., Weaver H. A., eds., pp. 577 -- 604 Lamy P., Toth I., 2009, Icarus, 201, 674 Landolt A. U., 1992, AJ, 104, 340 Levison H. F., Duncan M. J., 1994, Icarus, 108, 18 Li J., Jewitt D., Clover J. M., Jackson B. V., 2011, ApJ, 728, 31 Meech K. J., Jewitt D. C., 1987, A&A, 187, 585 Melita M. D., Licandro J., 2012, A&A, 539, A144 Millis R. L., Ahearn M. F., Thompson D. T., 1982, AJ, 87, 1310 Montalto M., Riffeser A., Hopp U., Wilke S., Carraro G., 2008, A&A, 479, L45 Moroz L. V., Arnold G., Korochantsev A. V., Wasch R., 1998, Icarus, 134, 253 Nakamura T., Yoshida F., 2008, PASJ, 60, 293 Peixinho N., Delsanti A., Guilbert-Lepoutre A., Gafeira R., Lacerda P., 2012, A&A (in press) Peixinho N., Doressoundiram A., Delsanti A., Boehnhardt H., Barucci M. A., Belskaya I., 2003, A&A, 410, L29 Roemer E., 1958, PASP, 70, 272 Russell H. N., 1916, ApJ, 43, 173 Senay M. C., Jewitt D., 1994, Natur, 371, 229 Snodgrass C., Lowry S. C., Fitzsimmons A., 2006, MNRAS, 373, 1590 Snodgrass C., Saviane I., Monaco L., Sinclaire P., 2008, The Messenger, 132, 18 Solontoi M. et al., 2012, Icarus, 218, 571 Sridhar S., Tremaine S., 1992, Icarus, 95, 86 Stevenson R., Kleyna J., Jewitt D., 2010, AJ, 139, 2230 Thompson W. R., Murray B. G. J. P. T., Khare B. N., Sagan C., 1987, JGR, 92, 14933 Tonry J. L. et al., 2012, ApJ, 750, 99 Trigo-Rodr´ıguez J. M., Garc´ıa-Melendo E., Davidsson B. J. R., S´anchez A., Rodr´ıguez D., Lacruz J., de Los Reyes J. A., Pastor S., 2008, A&A, 485, 599 This paper has been typeset from a TEX/ LATEX file prepared by the author. c(cid:13) 2012 RAS, MNRAS 000, 1 -- 10
1706.09986	1	1706	2017-06-30T00:42:50	The Effect of Adsorbed Liquid and Material Density on Saltation Threshold: Insight from Laboratory and Wind Tunnel Experiments	[ "astro-ph.EP" ]	Saltation threshold, the minimum wind speed for sediment transport, is a fundamental parameter in aeolian processes. The presence of liquid, such as water on Earth or methane on Titan, may affect the threshold values to a great extent. Sediment density is also crucial for determining threshold values. Here we provide quantitative data on density and water content of common wind tunnel materials that have been used to study conditions on Earth, Titan, Mars, and Venus. The measured density values for low density materials are higher compared to literature values, whereas for the high density materials, there is no such discrepancy. We also find that low density materials have much higher water content and longer atmospheric equilibration timescales compared to high density sediments. In the Titan Wind Tunnel, we performed threshold experiments with the standard walnut shells (125-150 \mu m, 7.2% water by mass) and dried walnut shells (1.7% water by mass). The threshold results for the two scenarios are almost the same, which indicates that humidity had a negligible effect on threshold for walnut shells in this experimental regime. When the water content is lower than 11.0%, the interparticle forces are dominated by adsorption forces, whereas at higher values the interparticle forces are dominated by much larger capillary forces. For materials with low equilibrium water content, like quartz sand, capillary forces dominate. When the interparticle forces are dominated by adsorption forces, the threshold does not increase with increasing relative humidity (RH). Only when the interparticle forces are dominated by capillary forces does the threshold start to increase with increasing RH/water content. Since tholins have a low methane content (0.3% at saturation, Curtis et al., 2008), we believe tholins would behave similarly to quartz sand when subjected to methane moisture. [abridged abstract]	astro-ph.EP	astro-ph	The Effect of Adsorbed Liquid and Material Density on Saltation Threshold: Insight from Laboratory and Wind Tunnel Experiments Xinting Yua,∗, Sarah M. Horsta, Chao Hea, Nathan T. Bridgesb, Devon M. Burrc, Joshua A. Sebreed, James K. Smithe aDepartment of Earth and Planetary Sciences, Johns Hopkins University ([email protected]), Baltimore, Maryland 21218, USA bApplied Physics Laboratory, Johns Hopkins University, Laurel, Maryland 20723, USA cDepartment of Earth and Planetary Sciences, University of Tennessee-Knoxville, 306 EPS Building, 1412 Circle Drive, Knoxville, Tennessee 37996, USA dDepartment of Chemistry and Biochemistry, University of Northern Iowa, Cedar Falls, eArizona State University, Tempe, AZ 85287-1404, USA Iowa 50614, USA Abstract Saltation threshold, the minimum wind speed for sediment transport, is a fundamental parameter in aeolian processes. Measuring this threshold using boundary layer wind tunnels, in which particles are mobilized by flowing air, for a subset of different planetary conditions can inform our understanding of physical processes of sediment transport. The presence of liquid, such as wa- ter on Earth or methane on Titan, may affect the threshold values to a great extent. Sediment density is also crucial for determining threshold values. Here we provide quantitative data on density and water content of common wind tunnel materials (including chromite, basalt, quartz sand, beach sand, glass beads, gas chromatograph packing materials, walnut shells, iced tea powder, activated charcoal, instant coffee, and glass bubbles) that have been used to study conditions on Earth, Titan, Mars, and Venus. The measured density values for low density materials are higher compared to literature values (e.g., ∼30% for walnut shells), whereas for the high density materials, ∗Corresponding author at: Department of Earth and Planetary Sciences, Johns Hop- kins University, Baltimore, Maryland 21218, USA Email address: [email protected] (Xinting Yu) Preprint submitted to Icarus October 6, 2018 there is no such discrepancy. We also find that low density materials have much higher water content and longer atmospheric equilibration timescales compared to high density sediments. We used thermogravimetric analysis (TGA) to quantify surface and internal water and found that over 80% of the total water content is surface water for low density materials. In the Titan Wind Tunnel (TWT), where Reynolds number conditions similar to those on Titan can be achieved, we performed threshold experiments with the standard walnut shells (125 -- 150 µm, 7.2% water by mass) and dried walnut shells, in which the water content was reduced to 1.7%. The thresh- old results for the two scenarios are almost the same, which indicates that humidity had a negligible effect on threshold for walnut shells in this experi- mental regime. When the water content is lower than 11.0%, the interparticle forces are dominated by adsorption forces, whereas at higher values the inter- particle forces are dominated by much larger capillary forces. For materials with low equilibrium water content, like quartz sand, capillary forces domi- nate. When the interparticle forces are dominated by adsorption forces, the threshold does not increase with increasing relative humidity (RH) or water content. Only when the interparticle forces are dominated by capillary forces does the threshold start to increase with increasing RH/water content. Since tholins have a low methane content (0.3% at saturation, Curtis et al., 2008), we believe tholins would behave similarly to quartz sand when subjected to methane moisture. Keywords: Aeolian processes, Titan, surface, Experimental techniques 1. Introduction Aeolian processes are fundamental in modifying the surfaces of all solid bodies in the Solar System with permanent or ephemeral atmospheres, in- cluding Earth, Venus, Mars, Saturn's moon Titan (Greeley and Iversen, 1985), Neptune's moon Triton (Smith et al., 1989), Pluto (Stern et al., 2015), and the comet 67P/Churyumov-Gerasimenko (Thomas et al., 2015). Study- ing aeolian features on planetary bodies enhances our understanding of near- surface winds, including the minimum wind speed to initiate saltation, wind direction, sediment flux, dune migration rates, and landscape modification. This information also provides input data and tests for global circulation predictions, leading to more powerful and accurate models. These models can then be run for different conditions providing insight into past or future 2 climates. Threshold wind speed is a fundamental parameter for understanding how and under what conditions wind detaches particles from the surface. Bound- ary layer wind tunnels serve as powerful laboratories for the study of aeolian processes, including threshold wind speed. Boundary layer tunnels were first used by Bagnold (1941), who pioneered the study of the minimum wind speed needed to initiate saltation on Earth. To test whether the parame- ters for quantifying threshold wind speed on Earth were also appropriate for Venusian and Martian conditions, the Martian Surface Wind Tunnel (MAR- SWIT) and Venus Wind Tunnel (VWT, now refurbished to the Titan Wind Tunnel, TWT) were built. The MARSWIT simulates the atmospheric pres- sure on Mars (4.0 -- 8.7 mb) with both martian atmosphere (CO2) and dry air (Greeley et al., 1976, 1977, 1980). To simulate the weight of the grains under the lower gravity of Mars, low density materials like walnut shells have been used (Greeley et al., 1976). The VWT achieved the same atmospheric den- sity as on Venus using CO2, and employed quartz sand as sediment (Greeley et al., 1984a, b). Cassini spacecraft data show extensive linear dunes covering 35% of equa- torial regions (±30◦) of Titan (Lorenz et al., 2006; Radebaugh et al., 2008). The dune materials are likely dominated by radar-dark (wavelength 2.17 cm) organic materials deposited from the atmosphere, with some minor wa- ter ice (McCord et al., 2006; Soderblom et al., 2007; Barnes et al., 2008; Clark et al., 2010; Le Gall et al., 2011; Hirtzig et al., 2013; Rodriguez et al., 2014). Global circulation models and measurements from the Huygens Doppler Wind Experiment show the dominant surface transporting winds are weak east to west winds (see e.g., Bird et al., 2005; Tokano, 2010). Con- versely, the streamlined appearance of the dunes is consistent with west to east winds (see e.g., Lorenz et al., 2006). In order to address this mystery, as well as to quantify the threshold wind speed and to study other aeolian processes on Titan, the TWT was built (Burr et al., 2015a,b). The robustness of wind tunnel experiments depends both on the degree of control of environmental conditions (e.g., pressure, relative humidity) and an understanding of experimental materials. The TWT simulates certain properties of Titan's near-surface atmosphere by using high pressure air (12.5 bar) to achieve the same Reynolds number (Re) as on Titan. The Reynolds number is the ratio of inertial to viscous forces (Re=uDp/ν, where u is the threshold friction wind speed, Dp is particle size, and ν is kinematic viscosity); this dimensionless number characterizes whether flow is laminar 3 (Re∗ (cid:28) 1) or turbulent (Re∗ (cid:29) 1). Titan has a surface temperature of 94 K, a surface pressure of 1.5 bar (Lindal et al., 1983, Fulchignoni et al., 2005), and an estimated atmospheric kinematic viscosity that is only about 1/12th of Earth (Burr et al., 2015a, Extended Data Table 1). The TWT at 12.5 bars and room temperature achieves the same value for kinematic viscosity (6.25×10−6 Pa · s) as on the surface of Titan. Low density materials have been used in threshold experiments in the TWT to compensate for Titan's low gravity, which is about 1/7th that of Earth (∼ 1.4 m/s2). In the case of TWT experiments, a range of particle sizes and densities have been utilized to measure a wide span of potential conditions over which threshold can occur, thereby allowing extrapolation to the very low weight materials on Titan (Burr et al., 2015a). The threshold wind speed is a function of the force balance of gravity (Fg), aerodynamic lift (Fl), aerodynamic drag (Fd), and interparticle forces (Fi), as shown in Fig. 1. Fg is dependent on mass which, in turn, is proportional to material density. For relatively heavy materials like quartz, the density determination is straight forward. However, for low density materials, most of which are porous and irregular, density values can vary considerably de- pending on the density definition used (see Section 3.1). The density values used in literature for low density materials (Greeley et al., 1976, 1977, 1980, Burr et al., 2015a) are usually taken from manufacturer labels, which usually do not specify a specific density definition. Thus it is necessary to reevaluate the data from these manufacturer labels. Interparticle forces consist of van der Waals, cohesion, and electrostatic forces. Van der Waals forces describe the dipole-dipole interactions between neutral molecules. Cohesion forces are the attraction forces between particles with condensed liquid on them. Electrostatic forces are the attraction or repulsion forces between charged particles or particles with different surface potential. On Earth, where water is abundant, cohesion forces are much larger than van der Waals and electrostatic forces (McKenna Neuman et al., 2003). On Titan, electrostatic forces likely dominate the interparticle forces (see discussion in Lorenz 2014, Burr et al., 2015a), and the same may be true on other planetary bodies where liquid water is not abundant, such as Mars, Venus, Triton, Pluto, and Comet 67P. On Titan, cohesion between liquid ethane or methane could also be important (Lorenz 2014). The surface tensions (γs) of liquid methane and ethane at Titan's surface temperature and pressure are only about 15 -- 20 mN/m (Baidakov et al. 2013) and are lower than water at 25◦C on Earth (γs=72 mN/m). Thus the cohesion force 4 for liquid methane and ethane on Titan should be lower than the cohesion for water on Earth given the same relative humidity for their respective surface temperatures. Using the TWT, Burr et al. (2015a) found that the experimental saltation threshold wind speeds are 50% higher than the model predictions of Iversen and White (1982) and Shao and Lu (2000). The inclusion in these models of a density ratio term (Iversen et al., 1976) with the very low density ratio (sediment density over atmospheric density) for Titan conditions caused the models to fit the TWT experimental data. Ongoing work includes using higher pressures in the TWT (15 and 20 bar) to further study the effect of the density ratio on threshold (Nield et al., 2016). Lower pressures are also used (1, 3, and 8 bar) to simulate possible past Titan conditions (possible past pressure as low as 0.7 bar, Charnay et al., 2014, Bridges et al., 2015). The models used in the previous TWT work and many other wind tunnel experiments did not specify interparticle forces for different materials. The experiments also did not dry the materials, which were exposed to ambient atmosphere with relative humidity (RH) ranging from 45 to 65%. During the TWT runs, the air used in the wind tunnel had an RH of 20 to 35% in the current experimental regime. Thus the interparticle forces were likely dominated by cohesion forces because the electrostatic charges dissipate very quickly when RH is greater than 5% (Bunker et al., 2007). In order to cor- rectly simulate aeolian processes on Titan, where electrostatic forces are pre- dicted to dominate the interparticle forces (Lorenz 2014, Burr et al., 2015a), a quantitative understanding of interparticle forces is therefore necessary. To accurately translate the TWT results to Titan conditions where liquid wa- ter is absent, we need to assess the effect of water present in Earth-based experiments on interparticle forces. Because of their low densities, which are used to provide some compensation for low extraterrestrial gravitational accelerations, the materials commonly used as analog sediments in Martian or Titan wind tunnel simulations are particularly susceptible to interparticle forces, highlighting the importance of this issue for understanding planetary aeolian processes. Previous studies of the effect of relative humidity on threshold which fo- cused on Earth are reviewed below (Section 2). In Section 3.1, we summarize the common materials used in planetary wind tunnels and their basic prop- erties according to the literature. The experimental methods are introduced in Section 3.2 -- 3.4. We measured the density of materials in use in planetary wind tunnels (Section 4.1) and their gravimetric water content and Earth 5 Figure 1: Forces acting on a particle stacked on two particles in airstream of density ρa (after Shao and Lu, 2000 and Kok et al., 2012). The particle density, ρp, and the particle size, Dp are the same for all three particles. The forces include gravity (Fg), the aerodynamic lift (Fl), the aerodynamic drag (Fd), and interparticle forces (Fi). The moment arm lengths ag, al, ad, and ai correspond to Fg, Fl, Fd, and Fi, respectively. 6 atmospheric equilibration timescales (Section 4.2). To further understand the effect of liquid on threshold, we measured the surface water content of the materials (Section 4.3). In Section 4.4, the threshold results of TWT experiments for wet and dry low density materials are shown. The implica- tions for the threshold wind speed and entrainment of particles on Titan are discussed in Section 5. Table 1: Summary of variables. Variable Symbols Description u* Dp ρa ρp µ ν Re* g RH w w' threshold friction wind speed particle size air density particle density dynamic viscosity kinematic viscosity, µ/ρa Reynolds number, uDp/ν gravity relative humidity water content initiation water content Unit m/s m kg/m3 kg/m3 kg/(m·s) m2/s - m/s2 % by pressure % by mass % by mass 2. Previous studies of the effect of water on threshold Bagnold (1941) used the balance of gravity and aerodynamic drag to derive the threshold friction wind speed for dry sand: (cid:114) ρp − ρa ρa u∗ b = A gd (1) where A is a function of Reynolds number Re and interparticle forces. When Re>3.5 (when particles are beyond the viscous sublayer and are more sus- ceptible to fluid drag), A is found to be a constant, with A=0.1 in air and A=0.2 in water, ρp and ρa are the density of the particle and atmosphere, respectively, and d is the mean aerodynamic particle diameter. This function is only appropriate for dry sand particles over 200 µm; for smaller sediments, interparticle forces become more significant compared to the weight of the particles. 7 Belly (1964) conducted the first wind tunnel experiments on the effect of humidity on threshold using 400 µm sand, and found that, b(1 + 1 2 RH 100 ), (2) w = u∗ u∗ b(1.8 + 0.6 log w) = u∗ w stands for threshold for wet sand, u∗ where u∗ b is the expression in Equation 1, w is the water content in percent by mass, and RH is the relative humidity in percent by pressure. When RH or water content increases, the threshold will increase accordingly. The results of Belly (1964) are shown in Fig. 2 (threshold RH) and Fig. 3 (threshold water content). Iversen et al. (1976), Iversen and White (1982), and Greeley and Iversen (1985) added interparticle forces and aerodynamic lift into the force balance and expanded the threshold model to small grains <200 µm. Their model is a piecewise function in three Reynolds number regimes. On the basis of a more explicit expression of interparticle forces, Shao and Lu (2000) simplified the model of Greeley and Iversen (1985) to a single equation: u∗ sl = f (Re∗)( ρp − ρa ρa gd + γ ρad ), (3) with f(Re∗) approximately equal to 0.0123, and γ between 1.65×10−4 N m−1 and 5×10−4 N m−1. This threshold model is for loosely packed dry materials. McKenna Neuman (2003) and McKenna Neuman and Sanderson (2008) slightly modified Shao and Lu's (2000) model for potentially humid or high/low temperature environments: (cid:114) (cid:115) u∗ w = where and f (Re∗)( ρp − ρa ρa gd + γ(cid:48) ρad2 ), γ(cid:48) = 6 π ai al Fi, Fi = βcd + ΨAc. (4) (5) (6) The term ai/al is the ratio of the moment arm lengths of interparticle and lift forces (see Fig. 1). The term βcd expresses the electrostatic and van der Waals forces. The term ΨAc describes the effect of cohesion. Ψ is the matric potential (also called Laplace pressure, ∆p) that describes the 8 pressure difference caused by surface tension, and it can be expressed by the Kelvin equation, )ln( Ψ = ( (7) where R=8.314 J mol−1 K−1 is the ideal gas constant for dry air, T is temper- ature in K, and Vl is the molar volume of the liquid (for water, Vl=1.8 × 10−5 m3 mol−1). Ac is the total contact area of adsorbed water films between particles and is approximated in McKenna Neuman and Sanderson (2008) as, RT Vl RH 100 ), Ac = δπkd(δ/δ0)n. (8) δ0 is the thickness for a monolayer of adsorbed water, and δ is the water film thickness, δ = ( 1/3 , H 6πΨ ) (9) and H is the Hamaker constant, an interaction parameter for adhesive sur- faces (−1.9 × 10−19 J, Iwamatsu and Horii, 1996; Tuller and Or, 2005). k is a dimensionless number describing the surface roughness (∼ 10−4 -- 10−5), and the power n varies between 6 -- 8, depending on the surface roughness and particle packing arrangement. Both k and n are determined by fitting the experimental data to the model. Thus threshold wind speed is a function of matric potential Ψ. The introduction of matric potential is useful for both humid coastal areas and cold regions (McKenna Neuman and Nicklings, 1989, data shown in Fig. 3) since it incorporates two variables, temperature and relative humidity, into one single variable. However, its applicability to low density materials has never been tested. Based on Equation (4), the results for the threshold RH variation are shown in Fig. 2 for 125 µm (k=2.1× 10−4, n=6.1) and 210 µm quartz sand (k=2.1× 10−4, n=5.0). Note that compared to slope of the threshold RH variation for 400 µm sand (Belly, 1964), the slope for 125 µm and 210 µm is lower, while it should be higher for smaller sediments according to Equation 4. This discrepancy may be due to the use of different threshold definitions (Fecan et al., 1999); Belly (1964) de- fined threshold as the point when bed movement is fully sustained, while McKenna Neuman and Sanderson (2008) defined it as the initiation of bed movement. Fecan et al. (1999) and Ravi et al. (2004, 2006) further investigated the effect of humidity on soils with different amounts of clay. They argued that with the clay component in soil, the matric potential of McKenna Neuman 9 Figure 2: Threshold friction wind speed (u) variation with relative humidity (RH). The symbols represent experimental data for: (1) 400 µm sand (Belly, 1964); (2) 125 µm sand (McKenna Neuman and Sanderson, 2008); (3) 210 µm sand (McKenna Neuman and Sanderson, 2008); (4) loamy sand with 8% soil clay content (Ravi et al., 2006); (5) clay loam with 31% soil clay content (Ravi et al., 2006). The lines show model fits to the data: (1) 400 µm sand using Equation (2); (2) 125 µm sand using Equation (4); (3) 210 µm sand using Equation (4). 10 020406080RH (%)0.250.30.350.40.450.50.550.60.650.7u* (m/s)Experimental dataloamy sand (Ravi et al., 2006)clay loam (Ravi et al., 2006)125 µm sand (McKenna Neuman & Sanderson, 2008)210 µm sand (McKenna Neuman & Sanderson, 2008)400 µm sand (Belly, 1964)Lines indicate model fit to dataand Nicklings (1989) for sand was no longer applicable because clay has much stronger adsorption forces to bond a layer of water film than quartz sand. Fecan et al. (1999) combined previous studies (Belly, 1964; Bisal and Hsieh, 1966; McKenna Neuman and Nickling, 1989; Saleh and Fryrear, 1995; Chen et al., 1996) and found an empirical formula for threshold as a function of gravimetric water content: w d w = [1 + 1.21(w − w(cid:48))0.68]0.5 when w > w(cid:48) = 1 when w < w(cid:48) u∗ u∗ u∗ u∗ w(cid:48) = 0.0014(%clay)2 + 0.17(%clay) w and u∗ d (10) where w is the water content per mass, and u∗ d respectively correspond to wet and dry threshold wind speeds. The result is shown in Fig. 3, and can be compared with the data and fitting of Belly (1964). Fecan et al. (1999) defined an initiation water content w', where they showed that once the water content of soil exceeds w', the threshold increases with increasing water content in soil. However, wind tunnel runs with clay and sandy loam under a range of humidities found different results. Ravi et al. (2006) found threshold increases with increasing RH only when RH is less than 40% or greater than 65%. When RH is between 40% and 65%, threshold decreased with increasing RH. The wind tunnel results data from Ravi et al. (2006) for two kinds of soil (different clay content) are shown in Fig. 2. They explain the results as follows: 1) for low RH (RH<40%), an adsorption layer covers the particle (which happens only for soil with a clay component) and the cohesion forces are dominated by the adsorption forces; 2) for high RH (RH>65%), water condenses and forms liquid bridges between particles and the cohesion forces are mainly the capillary forces between liquid bridges; 3) for RH in between 40% and 65%, a transition between the adsorption forces and capillary forces occurs, resulting in lower interparticle forces (see Equation 11 below). They thus modified the interparticle forces Fi (Equation 6) in McKenna-Neuman (2003) by modifying the total contact area Ac to describe the transition region (45%<RH<65%): Ac = π( w(cid:48) ρwσ − y 2 )d (11) where w' is the soil moisture content, ρw is the water density, and y is the distance between the two contacting sphere particles. Because w' varies as 11 Figure 3: Shown here is the threshold friction wind speed (u) variation with water content. The symbols represent experimental data for: (1) 400 µm sand (Belly, 1964); (2) 510 µm sand (McKenna Neuman and Nickling, 1989); (3) sandy loam with 12.2 % soil clay content (Saleh and Fryrear, 1995); (4) clay loam with 31.7 % soil clay content (Saleh and Fryrear, 1995); (5) clay with 49.2% soil clay content (Saleh and Fryrear, 1995). The lines show model fit to the data: (1) 400 µm sand using Equation (2); (2) 510 µm sand using Equation (10); (3) sandy loam using Equation (10); (4) clay loam using Equation (10); (5) clay using Equation (10). 12 024681012141618Water Content (%)0.30.350.40.450.50.550.60.650.70.75u (m/s)Experimental data400 µm Sand (Belly, 1964)510 µm sand (McKenna Neumann & Nickling, 1989)sandy loam (Saleh and Fryrear, 1995)clay loam (Saleh and Fryrear, 1995)clay (Saleh and Fryrear, 1995)Lines indicate model fit to datacΨ−b, and b < 1, the total cohesion ΨAc is proportional to Ψ1−b, and thus when RH increases, the cohesion forces decrease, leading to decreasing threshold wind speed. Overall, previous studies of the effect of water on the threshold wind speeds agree that for coarse-grained materials, the more water the materials have, the higher the threshold. That is, when RH or water content of sand increases, the interparticle cohesion between the particles increases, leading to a higher threshold. However, for sand with clay components, the threshold decreases with an increase in RH when RH is between 45% and 60%, although it is still larger than threshold in dry conditions with dry materials (Ravi et al., 2006, see Fig. 2). Thus, we should expect a higher threshold for low density materials when RH is high or when materials are not dried and thus have a high water content. 3. Methods 3.1. Materials Special care must be taken for planetary wind tunnels to reproduce rel- evant environmental conditions. As described in the previous section and shown in Table 2, there are several major planetary conditions that affect aeolian transportation: 1) transporting materials, 2) gravity, 3) atmospheric density, 4) atmospheric viscosity, and 5) density ratio. The materials transported by wind on inner Solar System terrestrial plan- ets (Earth, Venus, and Mars) are mainly from silicate rock, with Earth sedi- ments dominated by quartz sand and Venus and Mars sediments are mainly mafic basaltic sand. For materials that have been used in the TWT, both quartz sand (including white silica sand and beach sand from Cemexusa) and basaltic sand (acquired from Pisgah Crater) are easy to acquire and resemble the real aeolian sediments for Earth, Mars, and Venus. On the other hand, for icy worlds in the outer solar system, including Titan, Triton, and Pluto, the sediments are mainly organics. Analogs to those organic materials can be made in the laboratory ('tholins', Sagan et al., 1979 and Cable et al., 2012), but low yields and toxic composition means that they are not ideal for wind tunnel experiments, for which larger quantities are required (∼ 3000 cm3 for TWT). Tholins may also behave differently room temperature than under Titan conditions. Laboratory experiments indicate that Titan tholins have an effective density (ρeff) of 500 -- 1100 kg/m3 (Horst and Tolbert, 2013) and material density (ρm) of 1300 -- 1400 kg/m3 (Imanaka et al., 2012). Effective 13 density and material density are related by a shape and porosity factor (S). When the particles are perfect spheres without pores (S=1), the effective density and material density are equal; irregularities and porosity both de- crease S (Horst and Tolbert, 2013). Here we use these two measurements to estimate the maximum and minimum values of the material density on icy bodies, including Titan. To investigate aeolian planetary processes in an Earth laboratory, we have to use materials with lower densities to compensate for the higher gravity on Earth. For example, on Mars the material transported is basaltic sand with density of 3000 kg/m3, but since Martian gravity is only about 3/8ths that of Earth, previous experiments have used lower density material (1100 kg/m3) to simulate the weight of the materials as transported on Mars. Table 2 shows the density for equivalent weight aeolian materials on other planetary bodies. Low density materials that have been used in previous wind tunnel experi- ments (Greeley et al., 1980; Burr et al., 2015a) and which are investigated here are walnut shells (from Eco-shell, Inc), gas chromatograph packing ma- terials made from flux calcined diatomite (GC tan, Johns-Manville), iced tea powder (4C Totally Light), instant coffee (Foodhold U.S.A., LLC), activated charcoal (Sigma-Aldrich), and glass bubbles (3M). To extend the previous work into threshold conditions on Titan (Burr et al., 2015a), we include additional materials with different densities. These additional materials include non-acid washed and acid washed glass beads (Mo-Sci Corporation, Sigma-Aldrich, 2500 kg/m3), gas chromatograph pack- ing materials made from calcined diatomite (GC pink, Johns-Manville, 2150 kg/m3), and chromite (Reade Advanced Materials, 4000 kg/m3). The mate- rials investigated in this work include all the previously and currently used TWT materials (Greeley et al., 1980; Burr et al., 2015a). The materials are summarized in Table 3 in order of decreasing literature density values. The materials investigated in this study are the same batches (except iced tea powder and instant coffee) as the ones at the TWT, thus having the same size range as well as composition. 14 Table 2: Summary of planetary conditions. Values for Venus, Earth, Mars, and Titan are adopted from Burr et al. (2015b). For Triton and Pluto, atmospheric density values are derived using the ideal gas law, and surface temperature and pres- sure are adopted from Smith et al. (1989) and Gladstone et al. (2016), respectively. The atmospheric viscosity for Triton and Pluto is calculated by using gas type and temperature at http://www.lmnoeng.com/Flow/GasViscosity.htm. Planetary Density of Gravity (m/s2) Body Material ρp (kg/m3) Venus Earth Mars Titan Triton Pluto 3000 basalt 2650 quartz 3000 basalt 500 -- 1400 organics 500 -- 1400 organics 500 -- 1400 organics 8.9 9.8 3.7 1.4 0.62 0.78 Density of Equivalent Weight Material on Earth (kg/m3) Atmospheric Atmospheric Density ρa (kg/m3) Viscosity (Pa·s) Density Ratio (ρp/ρa) 2724 2650 1132 71 -- 200 31 -- 89 40 -- 111 65 1.2 0.015 5.1 ∼ 9 × 10−5 ∼ 9 × 10−5 3.27 × 10−2 1.85 × 10−5 1.30 × 10−5 6.25 × 10−6 ∼ 2 × 10−6 ∼ 2 × 10−6 46 2.2 × 103 2 × 105 78 -- 294 3.1 -- 12.0×106 3.1 -- 12.0×106 3.2. Density Measurements Measuring particle density requires a series of careful measurements. Mass is straightforward to obtain with an analytical balance. For this work, the mass of the materials was measured by an analytical balance (Satorius Entris 224-1S), with standard deviation of 0.1 mg. The particle volume is more difficult to measure, because of a number of different definitions of density. Fig. 4 compares three densities: the bulk density (ρb), particle density (ρp), and material density (ρm). Bulk density has the smallest value, as the bulk volume includes: 1) volume of the solid material, 2) closed internal voids, 3) open pores of particles, and 4) interpar- ticle voids. The volume defined in particle density (ρp) includes the volume of the solid material and the volume of the internal closed pores, whereas the 15 Table 3: Summary of material properties. GC indicates Gas Chromatograph packing materials. GC tan is calcined diatomite: according to Burr et al. (2015a), it a has different color compared to GC pink. For the literature density values, chromite, basalt, quartz sand, beach sand, and glass beads are standard values. Density of the GC pink, GC tan, activated charcoal, and glass bubbles were provided by the manufacturer. Density of walnut shells is originated from Greeley et al. 1980. Density of iced tea and instant coffee comes from FAO/INFOODS Density Database. Material Name Density in Literature (kg/m3) Size Range (µm) Chromite Basalt Quartz Sand Beach Sand Glass Beads GC pink GC tan Walnut Shells Iced Tea Powder Activated Charcoal Instant Coffee Glass Bubbles 4000 3000 2650 2650 2500 2150 1300 1100 1030 400 250 100 -- 140 212 -- 250; 250 -- 300 150 -- 250; 250 -- 500; 707 -- 1000 106 -- 125; 125 -- 150; 150 -- 175; 175 -- 212; 212 -- 250 500 -- 600; 600 -- 700; 707 -- 833; 833 -- 1000 150 -- 180 (non-acid washed); 150 -- 180 (acid washed); 180 -- 212; 500 -- 600 125 -- 150 150 -- 175 125 -- 150; 150 -- 175; 175 -- 250; 500 -- 600; 707 -- 833; 833 -- 1000 N/A 250 -- 300; 425 -- 500; 600 -- 707 N/A 30 -- 115 material volume (ρm) only includes the solid material volume. The particle density can be smaller or equal to the material density, depending on poros- ity. When the particles have internal pores, the particle density is always smaller than the material density (Fig. 4). Conversely, when particles have no pores, the particle density is equal to the material density (Webb, 2001, also see Fig. 4). Here we used an AccPyc II 1340 Automatic Gas (Helium) Pycnometer to measure the volume of the materials. The principle of the pycnometer is the gas displacement method and is illustrated in Fig. 5. Helium gas is first admitted into an empty compartment with calibrated volume Vempty, until 16 Figure 4: Comparison of different densities: bulk density (ρb), particle density (ρp), and material density (ρm), adapted from Webb (2001) and Horst and Tolbert (2013). 17 it equilibrates with a certain pressure (Fig. 5(b)). The samples are sealed in a second calibrated cup with volume Vcup. After the pressure is stable in the empty compartment (P0), the helium gas is discharged from the empty compartment to the cup with the samples. The helium gas fills the spaces within the sample as small as ∼3 A rapidly, and the final equlibrated pressure in the system is recorded as Pfinal (Fig. 5). Using the ideal gas law: P0Vempty = Pfinal(Vempty + Vcup − Vsample), (12) the pycnometer calculates the volume of the sample, Vsample. The volume of the materials (V) and the standard deviation are given automatically by the pycnometer after 10 purges and 20 runs with the materials. 3.3. Gravimetric Water Content Measurements We determine the water content of the materials by gravimetric measure- ments. The Relative Humidity (RH) and temperature in laboratory were recorded using a digital hygrometer (Dwyer Instrument) with 0 -- 100% RH range (accuracy of ±2%) and −30 -- 85◦C (accuracy ± 0.5◦C). The materials were put in an aluminum foil boat during the measurement. To eliminate the water adsorption of the aluminum foil, we put the foil in a 120◦C oven (Lab Safety Supply Model No.32EZ28, temperature accuracy ±1◦C at 100◦C) for (cid:48) 24 hours, and weighed it immediately after removal (m dry). The weight of the aluminum foil boat increases over time until equilibrating with water (cid:48) moisture in the atmosphere (m wet), usually in about 10 minutes. Thus, the amount of water adsorption on the aluminum foil is: (cid:48) ∆m (cid:48) = m wet − m (cid:48) dry (13) After the equilibration of the aluminum foil boat, we laid a thin layer of materials on the bottom of the boat. The materials and the boat were then dried together in the 120◦C oven for 24 hours. A lower temperature (105 ◦C) was tried to bake the materials, but it didn't change the overall results. After drying, they were weighed again immediately (mdry). Then we left the materials in air to let them equilibrate, weighing them every 0.5 -- 5 minutes. When the weight of the materials no longer changed with time, we recorded this final weight (mwet), ambient RH, temperature, and the time the materials took to equilibrate (teq). The final water content of the materials after they equilibrate (at a given RH and temperature) is given by: u(%) = mwet − mdry − ∆m mdry − m (cid:48) dry (cid:48) . (14) 18 Figure 5: Pycnometer work flow and the ideal gas law used to calculate the volume of the sample (Vsample). 19 3.4. Thermogravimetric (TGA) Measurements Since many of the low density materials we used are porous, water is both adsorbed on the surface and absorbed in the interior of the particles. The sur- face water affects threshold by increasing interparticle cohesion, whereas the interior water changes the density of the materials. Therefore it is important to differentiate between surface and internal water. Surface and internal wa- ter are released at different temperatures and can be separated using thermo- gravimetric analysis (TGA). The samples, weighing 10 -- 50 mg, were placed in an aluminum crucible and then loaded into a Mettler TGA/SDTA851e purged with nitrogen. The samples were heated from 25.0 to 600.0◦C at a rate of 10.00◦C/min. A slower heating rate (5.00◦C/min) was tested on a walnut shell sample, but it did not change the overall results. The RH during the experiment was measured by the RH probe described in Section 2.3. 3.5. Titan Wind Tunnel Experiments Using 'Wet' and 'Dry' Sediments To experimentally investigate the effect of water adsorption on threshold, we ran a set of experiments in the TWT at a range of pressures as a com- parison study of walnut shells (size 125 -- 150 µm) that were either subject to drying ('dry') or were in equilibrium with the ambient humidity at 1 bar ('wet'). The small end of the sediment size (125 -- 150 µm) was chosen for the 'wet' and 'dry' runs because smaller particles are more sensitive to interpar- ticle force change than larger particles with greater gravitational forces. To measure their water content, the 'wet' walnut shells were analyzed after the TWT run following the method described in Section 3.3, using a different an- alytical balance (A&D HR-120 with standard deviation of 0.1 mg) and oven (VWR Economy Vacuum Oven Model 1400E, temperature accuracy ±3.5◦C at 100◦C). To prepare the 'dry' walnut shells, we put the materials needed for a TWT run (approximately 3000 cm3) in a 120◦C oven for 24 hrs. Then we trans- ferred all the materials into a desiccator (Lab Safety Supply, I.D. 300mm) with desiccant (Carolina, Silica Gel, Indicating Beads, Laboratory Grade) in preparation for the TWT experiment. While we set the bed for the TWT experiment (see Extended Data Figure 2 in Burr et al. 2015a), the materials were exposed to ambient air for 40 minutes. We chose the walnut shells for this experiment because their equilibration timescale, as discussed in Section 4.2, is longer than the time required to set the bed. The procedure for con- ducting experiments in the TWT can be found in Burr et al. (2015a). For the experiments presented here, the pressures in the TWT were 1, 3, 8, 12.5, 20 15, and 20 bars. The freestream wind speed was converted from dynamic pressure collected by pitot tubes in the TWT (for details, see Methods in Burr et al., 2015a; the only change is that the current TWT has the fixed pitot tube in the test section to collect dynamic pressure, instead of at the back of wind tunnel as described in Burr et al., 2015a). 4. Results and Discussion 4.1. Particle density measurement of wind tunnel materials The particle density measurements show that for materials with densities over 2000 kg/m3 (chromite, basalt, quartz sand, beach sand, glass beads, and GC pink), the measured densities are very close to the densities reported in the literature (see Table 4). However, for material densities less than 2000 kg/m3 (GC tan, walnut shells, instant coffee, activated charcoal, iced tea powder, and glass bubbles), the measured densities differ from those in the literature or as provided by the manufacturer (see Table 5). Thus here we can divide the materials into two groups, high density and low density materials, where the division between low density and high density materials is 2000 kg/m3. The discrepancy between the particle density measured by the helium gas pycnomter and the density reported in the literature or by the manufacturer could be attributed to the different density definitions. The helium gas in the pycnometer can rapidly fill the open pores of the materials, thus the pycnometer measures the particle density (ρp<ρm if the particles have closed internal pores, or ρp = ρm if the particles have no internal pores). The density reported in the literature may be bulk density given by the manufacturer, as it is with activated charcoal, iced tea powder, instant coffee, and GC tan (calcined diatomite). The density used for walnut shells in the literature is 1100 kg/m3 (Greeley et al., 1980; Burr et al., 2015a), whereas the density measured by pycnometer gives 1400 kg/m3. One possible explanation is that the densities given by the manufacturer are defined in other ways or are not measured precisely. The density measured for the high density materials are likely closer to the literature value because those materials generally have no internal voids. However, the 'density' used in the TWT data analysis depends also on the porosity, surface area, size, and shape of the particles. Therefore, this value should fall between the bulk density and the material density, but probably closer to the material density because the wind can penetrate the interparticle voids. 21 Table 4: Summary of densities of high density wind tunnel ma- terials (literature density greater than 2000 kg/m3) in literature and measured by the pycnometer, with standard deviation in the measurements. Density in Updated Standard Literature Density Deviation (kg/m3) (kg/m3) Material Name Chromite Basalt Quartz Sand Beach Sand Glass Beads GC pink Size Range (µm) 212 -- 250 250 -- 300 150 -- 250 250 -- 500 707 -- 1000 175 -- 212 500 -- 600 600 -- 707 707 -- 833 833 -- 1000 150 -- 180 (non-acid washed) 150 -- 180 (acid-washed) 180 -- 212 500 -- 600 125 -- 150 (kg/m3) 4524.7 4518.8 2841.4 2882.8 2919.0 2656.4 2631.2 2639.0 2634.3 2632.3 2481.1 2420.2 2634.3 2632.3 2364.6 4000 3000 2650 2650 2500 2150 1.1 1.1 1.1 2.0 1.0 1.2 2.2 0.8 1.2 1.1 0.4 0.6 1.2 1.1 2.2 4.2. Water content and equilibration timescales of wind tunnel materials The gravimetric measurements allow us to classify the materials by water content. As shown in Fig. 6, the materials can be divided into 2 groups: 1) materials with low water content (<1%), including all materials with liter- ature densities over 2000 kg/m3 (high density materials) and glass bubbles and 2) materials with high water content (>6%), including materials with literature densities less than 2000 kg/m3 (low density materials), except glass bubbles. For the equilibration timescales shown in Fig. 7, we can classify the materials in the same way: low water content materials have a short equilibration time, usually less than 1 hr, while high water content materials have a long equilibration time, over 6 hrs. For the same kind of material, both the water content and equilibration timescales show no apparent size dependence (Fig. 6 and Fig. 7). 22 Table 5: Summary of the densities of low density wind tunnel materials (literature density less than 2000 kg/m3) in the literature and measured by the pycnometer, with standard deviation in the measurements. Density in Updated Standard Literature Density Deviation (kg/m3) (kg/m3) Material GC tan Walnut Shells Iced Tea Powder Activated Charcoal Instant Coffee Glass Bubbles Size Range µm 150 -- 175 125 -- 150 150 -- 175 175 -- 250 500 -- 600 707 -- 833 833 -- 1000 N/A 250 -- 300 425 -- 500 600 -- 707 N/A 30 -- 115 1300 1100 1030 400 250 100 -- 140 (kg/m3) 2006.1 1426.5 1419.1 1418.8 1415.4 1407.2 1411.8 1423.5 1932.4 1896.7 1881.8 1473.9 140.1 5.0 1.0 0.9 1.0 2.6 2.7 1.7 1.3 6.9 15.9 12.5 1.2 0.6 One possible explanation for the high water content of low density mate- rials is the combination of being hydrophilic and having large surface area- to-volume ratios, whereas the high density materials generally have smaller surface area-to-volume ratios and are hydrophobic. Glass bubbles are the ex- ception; they are low density but they also have low water content (∼0.05%) and short equilibration time (∼15 min) like high density materials. This is because they are designed to have low surface-area-to-volume ratio and are hydrophobic. In Fig. 8 we present the equilibration curves of two typical wind tunnel materials, quartz sand (low water content, high density) and walnut shells (high water content, low density). The walnut shells have much higher water content and equilibrate more slowly than quartz sand. In the first 10 minutes when quartz sand is approaching equilibrium, the walnut shells adsorb much more water by weight compared to quartz sand in the same time period. This correlation indicates water content and equilibration timescales are related. The long equilibration time for the low density materials also indicates that these materials cannot be dried quickly by circulating dry air in the wind 23 tunnel. However, a short exposure time to air for the low density materials will not increase the water content to the equilibrium state. We used the natural variation of humidity in the laboratory (15 -- 60%) to see how water content varies as a function of RH. There is a linear relationship between RH and water content for some of the materials shown in Table 6, including basalt, beach sand, walnut shells, activated charcoal, GC tan, iced tea powder, and instant coffee. They all have R2 values greater than 0.8 for a linear relationship. These linear relationships could be used to estimate water content of materials when only RH is recorded. For chromite, glass beads, quartz sand, GC pink, and glass bubbles, the coefficients of determination R2 for the linear relationships are only between 0.3 -- 0.6. There may be a linear relationship between RH and water content for these materials as well, but the relationship is difficult to measure without a more precise analytical balance because of the low water content of these materials. 4.3. Surface and internal water of wind tunnel materials As discussed in Section 3.4, surface water can change the interparticle cohesion and affect the threshold. The surface water measurements from TGA are listed in Table 7. For materials with less than 0.2% water content, the mass loss was below the limit of detection for the TGA. Thus for high density materials like basalt, quartz sand, beach sand, and chromite, the surface and internal water content cannot be detected using the TGA. However, the high density materials are not porous or hydrophilic, thus the surface water content should equal the total water content, which we measured by gravimetric analysis. For the low density materials like walnut shells, iced tea powder, and instant coffee, we can only get partial information from TGA, because ther- mal reactions will occur for these materials at high temperature. Generally, the surface water of a material releases from about 50◦C to 150◦C, then its internal water starts to release from about 200◦C. Walnut shells start to re- lease their internal water from about 175 -- 200◦C, but thermal destruction begins around 202◦C (Findor´ak et al., 2016), so we cannot get the internal water content of walnut shells directly from TGA measurement. For iced tea powder, thermal destruction happens at the lowest temperature of all the materials investigated, which is about 150◦C. For instant coffee, thermal destruction happens at 175◦C. Activated charcoal is stable during the entire heating process until 600◦C. From Table 7, we can find that surface water occupies over 80% of the total water content for activated charcoal. Even 24 Figure 6: Water content of wind tunnel materials of different sizes at the same RH (RH=52.8%). Density values for the materials are adopted from the pycnometer mea- surements in Section 4.1. 25 Figure 7: Equilibration Timescales of wind tunnel materials of different sizes. For the materials marked with * (GC tan, activated charcoal, instant coffee, walnut shells, and iced tea powder), the equilibrium timescales were long, so that the minimum equilibration timescales are plotted. Density values for the materials are adopted from the pycnometer measurements in Section 4.1. 26 Figure 8: Comparison of water content and equilibration process of one low density (walnut shells 150 -- 175 µm) and one high density (quartz sand 150 -- 175 µm) wind tunnel material up to 200 minutes. The inset graph magnifies the comparison in the first 10 minutes. 27 Table 6: Summary of the RH and water content linear relationship of the wind tunnel materials, with R2>0.8. The linear relationship is y = ax + b, where y is the water content by mass and x is the RH in %. R2 is the coefficient of determination for each linear relationship. Quartz sand (all sizes), GC pink (125 -- 150 µm), and glass bubbles have R2 values that vary between 0.3 -- 0.6, because their water content is very small (< 1%) compared to other materials and the measured water content values have large deviations. a b 1.60 × 10−1 1.06 × 10−1 6.58 × 10−2 8.62 × 10−2 8.87 × 10−2 8.34 × 10−2 7.13 × 10−2 1.93 × 10−3 1.38 × 10−3 8.70 × 10−4 1.22 × 10−3 1.17 × 10−3 1.15 × 10−3 1.06 × 10−3 9.87 × 10−2 8.28 × 10−2 8.24 × 10−2 8.02 × 10−2 7.90 × 10−2 8.04 × 10−2 2.30 × 10−1 −3.05 × 10−1 1.02 × 10−3 7.88 × 10−2 1.13 × 10−1 1.13 × 10−1 2.07 3.26 3.41 3.28 3.58 3.51 3.79 4.32 R2 0.944 0.954 0.941 0.930 0.935 0.964 0.953 0.898 0.850 0.835 0.794 0.903 0.815 0.946 0.927 0.929 0.875 Material Size Range (µm) Basalt Beach Sand Walnut Shells Activated Charcoal GC tan Iced Tea Powder Instant Coffee 150 -- 250 250 -- 500 707 -- 1000 500 -- 600 600 -- 700 707 -- 833 833 -- 1000 125 -- 150 150 -- 175 175 -- 250 500 -- 600 707 -- 833 833 -- 1000 400 -- 841 150 -- 175 N/A N/A 28 for activated charcoal with extremely high porosity, the surface water still dominates. The estimated total water content using the linear relationship in Table 6 should equal the sum of surface water and internal water. However, it seems clear from this analysis that most of the water measured, if not all, by gravimetric analysis is surface water. Table 7: Separation of surface and internal water from TGA analysis for some of wind tunnel materials. We calculated the estimated water content values using the linear rela- tionship of RH and water content from Table 6. The n/a* for basalt, quartz sand, beach sand, and chromite indicates no water was detected for those materials. The n/a† for walnut shells of all sizes, iced tea, and instant coffee indicated other chemical processes take place instead of the water loss process to high temperature, thus we cannot measure the internal water. For iced tea powder, neither the surface nor the internal water can be measured because chemical processes happen at lower temperature. The n/a‡ indicates the estimated water content of quartz sand and chromite at the specific RH are acquired from direct measurements rather than the linear relationships in Table 6. Size Range RH Estimated Total Surface Internal Material Name Basalt Quartz Sand Beach Sand Chromite Walnut Shells Activated Charcoal Iced Tea Powder Instant Coffee µm 150 -- 250 212 -- 250 500 -- 600 212 -- 250 125 -- 150 150 -- 175 175 -- 250 500 -- 600 707 -- 833 833 -- 1000 400 -- 841 N/A N/A (%) Water Content (%) Water (%) Water (%) 40 40 40 40 30 40 40 40 40 40 30 40 40 0.2 0.1 ‡ 0.1 0.1‡ 5.0 6.6 6.7 6.5 6.7 6.7 6.6 8.3 8.8 n/a* n/a* n/a* n/a* 5.7 7.3 7.1 7.2 8.2 7.9 4.0 n/a† 6.2 n/a* n/a* n/a* n/a* n/a† n/a† n/a† n/a† n/a† n/a† 1.0 n/a† n/a† To understand the equilibration process for low density materials, we exposed dry walnut shells (150 -- 175 µm) to ambient air for different lengths of time, and then measured their surface water content through TGA. The results are shown in Table 8. The surface water shows the same value for walnut shells exposed for 2 hours and walnut shells exposed for 4 hours, indicating the surface water equilibrates in 2 hrs or less. The walnut shells 29 exposed for 22 hours have a lower surface water content value, which may have been caused by an RH change during the longer time period. Table 8: The measured equilibration process of walnut shells 150 -- 175 µm. The four walnut shells samples were baked for 24 hrs in a 120◦C oven and then exposed to air (RH∼40%) for 0, 2, 4, and 22 hrs. The surface water was then separated by TGA analysis. Material Size Range RH Exposed Time Surface µm Walnut Shells 150 -- 175 (%) 40 40 40 40 (hr) Water (%) 0 2 4 22 0.3 5.3 5.3 4.8 Overall, we found that water in low density materials is dominated by surface water, while interior water occupies less than 20% of the total water content. Also, surface water is adsorbed first when dry materials are exposed to ambient air. Because only surface water would affect the interparticle cohesion, we believe for low density materials, a change of the water content of the materials would change their interparticle cohesion, and may affect threshold wind speed. 4.4. The effect of water adsorption on threshold wind speed For the 'wet' walnut shells, the materials sat and were in equilibrium with ambient air (RH varies between 50 -- 60%). The water content of the 'wet' walnut shells before the TWT run was 8.14% and after the TWT was 7.20%, which suggests that the materials were dried by the mixture of air in the TWT due to lower concentration of water vapor (see Section 1). During the TWT run, the RH outside the wind tunnel varied between 50.1% to 51.8%, while the RH inside the wind tunnel varied between 16.7% to 36.5%. For the 'dry' walnut shells run, after drying and cooling down the sedi- ments, the measured water content was 1.29%. After the bed was prepared and all the TWT runs finished, the water content of the 'dry' walnut shells increased to 1.67%. During this TWT run, the RH outside the wind tunnel varied between 51.6% to 53.7%, while the RH inside the wind tunnel varied between 3.7% to 11.9%. Note that the RH inside the wind tunnel for the 'wet' scenario is larger than the 'dry' scenario. So moisture may come out from the 'wet' walnut shells, thus increasing the RH inside the TWT. 30 Table 9 shows the threshold freestream wind speed for 'wet' and 'dry' TWT runs at different pressures. The 'wet' thresholds for different pressures are consistently a couple percent larger than the 'dry' thresholds. However, the differences are smaller than the standard deviations of the threshold wind speed for both 'wet' and 'dry' runs. The reason for the similar 'wet' and 'dry' thresholds may be due to the similar water adsorption behavior of walnut shells to clay minerals. Accord- ing to the measurements done by Pirayesh et al. (2012), walnut shells consist of 46.6% of holocellulose, 49.1% lignin, and 3.6% ash. Holocelluose is rich in hydroxyl groups, similar to clay minerals (e.g., kaolinite). The free hydroxyl groups of holocelluose can thus adsorb water through hydrogen bonding, cre- ating an adsorption layer covering the particles (Gwon et al., 2010), while lignin cannot adsorb such a layer of water (Nourbakhsh et al., 2011). Thus walnut shells behave like a mixture of 'clay' (hollucellulose) and 'quartz sand' (lignin). Note that clay also has long equilibration timescales and high water content similar to walnut shells (Ravi et al., 2006). Thus we would expect that walnut shells behave similarly to a clay/quartz mixture when subjected to water. According to Fecan et al., (1999), with clay mixed into quartz sand, threshold does not change when the water content of the materials is be- low the initiation water content (w'). Threshold will only start to increase with increasing water content when the initiation water content is reached. And with increasing clay content in quartz sand, this initiation water content value is higher. This is caused by the different interparticle cohesion schemes of clay and quartz sand. For quartz sand, the interparticle forces are domi- nated by capillary forces, which are of similar magnitude to the gravity and wind drag forces. Thus with increasing water content, the threshold will in- crease accordingly for quartz sand. While for clay minerals, when the water content is lower than the initiation water content value (w<w'), the interpar- ticle forces are dominated by adsorption forces due to the molecular bonding between the hydroxyl groups in the clay minerals and water. Adsorption forces are much weaker than capillary forces, thus when the water content increases, even though adsorption forces are increasing, the threshold doesn't change significantly. When the initiation water content is reached (w>w'), the cohesion forces start to be dominated by capillary forces and the thresh- old wind speed begins to increase with increasing water content. Simply substituting the 46.6% holocellulose content as the clay content for walnut 31 shells in Equation 10, we get the initiation water content value, w', w(cid:48) = 0.0014 ∗ 46.62 + 0.17 ∗ 46.6 = 11.0 (15) When water content of walnut shells is lower than 11.0%, interparticle forces are dominated by adsorption forces, which is much weaker than capillary forces. It is only when it exceeds 11.0%, that the interparticle forces are dominated by capillary forces and the threshold begins to increase with in- creasing water content. This comparison could explain the similar threshold results for the 'wet' and 'dry' walnut shells with water contents of 1.67% and 7.20%, respectively, as both values are lower than the initiation water content value. Table 9: The threshold freestream wind speed for 'wet' and 'dry' TWT runs at different pressures. The standard deviations were calculated using the procedure in Burr et al. (2015a). Pressure 'Wet' Threshold (bar) Freestream Standard Deviation 'Dry' Threshold Freestream Standard Difference Deviation (%) Wind Speed (m/s) 20 15 12.5 8 3 1 1.62 1.76 1.93 2.37 3.48 6.00 (%) 5.46 5.08 3.16 3.55 3.85 2.80 Wind Speed (m/s) 1.55 1.71 1.89 2.36 3.36 5.92 (%) 2.89 5.15 5.23 3.42 1.88 0.17 4.16 2.89 2.10 0.38 3.46 1.36 The low density materials are usually chosen to match the weight for the relevant planetary body. However, the low density materials may not have the same interparticle forces compared to the real transporting materials. For example, walnut shells used in both the MARSWIT and TWT are more similar to a clay/quartz mixture in terms of interparticle cohesion forces. GC tan (mixture of clay and other minerals), iced tea powder, and instant coffee are similar in that they all adsorb a layer of molecular bonded water (hygro- scopic water) and have high water content and low density such that their interparticle forces should behave like walnut shells. They all have an initia- tion water content, after which the threshold starts to change with increasing water content (see the left column in Fig 9). On Mars, the transporting ma- terial is mostly basaltic sand, a high density material, and its interparticle 32 Figure 9: Lists of clay/quartz sand mixture-like materials and pure quartz sand-like ma- terials and comparison of their behavior when subjected to water. The dark blue layer is the adsorption/hygroscopic water, while the light blue layer is the capillary water. The w is the water content of the material by mass, and w' is the initiation water content. cohesion should be closer to quartz sand. Quartz sand is hydrophobic, so it doesn't form the molecular bonded water layer. Rather, the water on its surface directly contributes to capillary water, as shown on the right column in Fig. 9. Thus with increasing water content of quartz sand, threshold increases accordingly. For materials with similar interparticle forces to quartz sand, we can use the model of McKenna Neuman and Sanderson (2008) to translate the TWT results to Titan conditions for different RH in the TWT. These materials have low water content and short equilibration timescales, including all high density materials and one low density material, glass bubbles (see Section 4.2). The conversion ratios to convert u∗TWT to u∗Titan for these materials 33 Figure 10: Modeled threshold wind speed ratios used for converting the u∗TWT to u∗Titan with variation of RH in the TWT between 0 to 80%, assuming the methane humidity on Titan is 0. (a) For high density materials of different sizes. (b) For part of low density materials, since iced tea powder and instant coffee have unknown size range, we cannot make the theoretical estimation. are shown in Fig. 10(a) and Fig. 10(b, blue line). Materials that are similar to a clay/quartz sand mixtures need a certain water content to alter the interparticle forces from adsorption forces to capillary forces. For walnut shells, this initiation water content is predicted in Section 4.4, 11.0%. Using the RH-water content relationship for walnut shells in Table 6, we find that the corresponding initiation RH is about 90%. Since we have never observed such high RH in the TWT, here we only include the density correction (1400 kg/m3 instead of 1100 kg/m3), for the conversion ratios in Fig. 10(b). For GC tan, the density correction is 2000 kg/m3 instead of 1300 kg/m3. Since no size range is provided for iced tea powder and instant coffee, we cannot make a prediction for them. Activated charcoal is hydrophobic, its high water content is mainly attributed to its large surface area, and currently we cannot conclude which group it belongs to. To correctly simulate different interparticle force regimes, determination of the water content of the materials is very important. If the water content of the material is high (>6%), as is usual for low density materials, the 34 01020304050607080RH (%)0.30.40.50.60.7utitan/utwt(a)212-250 µm chromite (4500 kg/m3)150-250 µm basalt (3000 kg/m3)212-250 µm quartz sand (2650 kg/m3)106-125 µm quartz sand (2650 kg/m3)180-212 µm glass beads (2500 kg/m3)125-150 µm GC pink (2350 kg/m3)01020304050607080RH (%)0.40.60.811.21.41.61.8utitan/utwt(b)75 µm glass bubbles (140 kg/m3)125-150 µm walnut shells (1400 kg/m3)150-175 µm GC tan (2000 kg/m3)material likely behaves like clay when exposed to water. If the water content of the material is low (<1%), it may behave as quartz sand when exposed to water. Thus the determination of water content could not only distinguish the density of the materials, but more importantly, this information may provide insight on the effect of water RH on threshold. This information makes the only exception, glass bubbles (a low density material with a low water content), very useful in simulating both the gravity and interparticle forces of real transporting Titan particles. Although electrostatic forces can make these glass bubbles stick together (thus make them hard to sieve) and further experiments should consider this effect before using them. 4.5. The effect of methane humidity on tholins On Titan, the transported material is dark organic sand (Barnes et al., 2008), with possible methane and ethane moisture affecting its cohesion force (Lorenz 2014). Laboratory studies of the adsorption of methane and ethane on tholins show that at saturation, tholins can adsorb only 0.3% of methane by mass (approximately a monolayer) or a monolayer of ethane (Curtis et al., 2008). As the molecular weights of water (18 g/mol) and methane (16 g/mol) are similar, the methane content of tholins may be close to the water content of quartz (and other high density materials). Thus it is possible that the interparticle cohesion of tholins (subjected to methane moisture) is similar to quartz sand (subjected to water vapor); that is, methane acts as capillary liquid instead of an adsorption liquid. With the increasing relative humidity of methane, or increasing methane content of tholins, the threshold wind speed for tholins or Titan's organic sand will increase accordingly, with no initiation liquid content like clay/quartz mixture or walnut shells (to water vapor). In the definition of the matric potential Ψ (Equation 7), when RH ap- proaches 0, Ψ → −∞, and when RH is 100%, Ψ = 0. Thus when including the matric potential in calculating the thickness of the liquid film (Equa- tion 9), the values become extreme when RH is very small or very large (see the solid blue curve in Fig. 11). These extreme RH values would also lead to extreme values for the threshold wind speed u. To avoid this issue, we developed a second model incorporating measurements of methane film thickness on tholins in Curtis et al. (2008), shown as the dash-dot blue curve in Fig. 11, with a Langmuir adsorption isotherm fit. This thickness fit has no extreme values for the whole RH range and is more realistic compared to McKenna Neuman and Sanderson (2008). To calculate the total interparticle 35 Table 10: Modeling parameters for Titan. Hamaker constant for methane is adopted from Iwamatsu and Horii (1996) and Israelachvili (2011). Modeling Parameter Temperature (T) Particle Size (Dp) Air Density (ρa) Particle Diameter (ρp) Molar Volume of Methane (VCH4) Hamaker Constant for Methane (H) Roughness Dimensionless Number (k) Value 94 K 125 µm 5.1 kg/m3 950 kg/m3 1.6 ×10−5 m3/mol −0.5 × 10−19 J 2.1 × 10−4 Roughness Power (n) f(Re) 4.5 0.024 cohesion force, instead of using Ψ (Equation 6) we use another expression of the Laplace pressure ∆p to avoid the extremes at RH=0 and RH=100% (Christenson, 1988): ∆p = , (16) γs rm ≈ γs 2δcosθ where γs is the surface tension of methane, 15 mN/m (Miquet et al., 2000), δ is thickness of the methane film, and θ is the contact angle between methane and tholins. Here we use cos θ = 0.97 (Lavvas et al., 2011). The results for the computed threshold variation with changing the rel- ative humidity of methane are shown in Fig. 11. The solid red curve shows the modeling result of threshold wind speed variation with RH of methane under Titan conditions using the model of McKenna Neuman and Sanderson (2008), and the dash-dot red curve shows the modeling result incorporating the Langmuir model and data of Curtis et al. (2008). The solid red curve dis- plays a more dramatic change with increasing methane RH than the dash-dot red curve; however, for both models, extreme methane humidity causes the threshold wind speed to change by less than 20%, compared to dry conditions (RH=0). This minimal change could be attributed to the lower surface ten- sion of methane compared to water and Titan's low temperature. However, this explanation assumes the geometry, contacting mechanics, and electro- static forces of Titan's organic sand is similar to Earth quartz, which is not known. Thus further research on these properties of tholins is necessary. 36 Figure 11: Two modeling results are shown of methane film thickness for 'tholins' and the threshold wind speed variation with changing relative humidity of methane. The blue curves shows the result for the methane film thickness and the red curves are the threshold wind speed variation with methane humidity. Model 1 refers to the revised McKenna Neuman and Sanderson (2008) model. As RH approaches 100%, the calculated thickness and threshold approach infinity, so here we only show RH between 0 to 90%. Model 2 shows the thickness and threshold wind speed variation with methane humidity using the data and Langmuir model of Curtis et al. (2008). 37 0102030405060708090100Methane RH (%)00.10.20.30.40.50.60.70.8Methane Film Thickness (nm)0.0430.0440.0450.0460.0470.0480.0490.050.051u (m/s)Model 1 ThicknessModel 2 ThicknessModel 1 ThresholdModel 2 Threshold5. Conclusion We measured various properties for low density materials used in plan- etary wind tunnels that have been missing or incomplete in the literature. The literature-given density of walnut shells, in use since the 1970s, 1100 kg/m3, is lower than our measurement of 1400 kg/m3, a difference of 30%. The effect of moisture on low density materials is also very distinct compared to high density materials. Low density materials generally have high water content (>6%) and long equilibration timescales (>6 hrs), while high density materials have low water content (<1%) and short equilibration timescales (<1 hr). The determination of the water content of the material provides insight into the sensitivity of threshold wind speed to RH based on our 'wet' and 'dry' walnut shell TWT runs. The results indicate that threshold is not very sensitive to 'wet' vs 'dry' walnut shells. The materials with high water content tend to behave like a clay/quartz mixture (where adsorption forces dominate below the initiation water content, and then the capillary forces dominate), whereas the materials with low water content are more likely to behave similarly to quartz sand (where capillary forces always dominate). When the interparticle forces are dominated by capillary forces, the threshold increases with increasing water content. Because tholins have a low methane content, we hypothesize that when the real transporting materials on Titan are subjected to methane moisture, they would behave similarly to quartz sand subjected to water. 6. Acknowledgements Partial support to Yu, Bridges, and Burr was provided by NASA grant NNX14AR23G/118460 that was selected under the Outer Planets Research Program (Bridges PI). We thank C. Chavez and J. Moore of NASA Ames Research Center for providing the oven to use in the wind tunnel experiments. 7. References References Bagnold, R. A., 1941. The physics of wind blown sand and desert dunes. Methuen, London 265(10). 38 Baidakov, V. G., Kaverin, A. M., and Khotienkova, M. N., 2013. Sur- face tension of ethane methane solutions: 1. Experiment and ther- modynamic analysis of the results. Fluid Phase Equilibria, 356, 90. http://dx.doi.org/10.1016/j.fluid.2013.07.008. Barnes, J. W., Brown, R. H., Soderblom, L., et al. 2008. Spectroscopy, mor- phometry, and photoclinometry of Titan's dunefields from Cassini/VIMS. Icarus, 195, 400. http://dx.doi.org/10.1016/j.icarus.2007.12.006. Belly, P. Y., 1964. Sand movement by wind. US Army Corps of Engineering (USACE), 1964. Bird, M. K., Allison, M., Asmar, S. W., et al., 2005. The vertical profile of winds on Titan. Nature, 438, 800. http://dx.doi.org/10.1038/nature04060. Bisal F., and Hsieh, J., 1966. Influence of moisture on erodibility of soil by wind. Soil Science, 102(3), 143. Bridges, N., Burr, D. M., Marshall, J., et al. 2015, New Titan Saltation Threshold Experiments: Investigating Current and Past Climates. AGU Fall Meeting Abstracts, P12B-05. Bunker, M. J., Davies, M. C., James, M. B., and Roberts, C. J, 2007. Direct observation of charging by atomic force microscopy. Pharmaceutical research, 24(6), 1165. http://dx.doi.org/10.1007/s11095-006-9230-z. single particle electrostatic Burr, D. M., Bridges, N. T., Marshall, J. R., et al., 2015a. Higher-than- predicted saltation threshold wind speeds on Titan. Nature, 517, 60. http://dx.doi.org/10.1038/nature14088. Burr, D. M., Bridges, N. T., Smith, J. K., et al., 2015b. The Titan Wind Tunnel: A new tool for investigating extraterrestrial aeolian environments. Aeolian Research, 18, 205. http://dx.doi.org/10.1016/j.aeolia.2015.07.008. Cable, M. L., Horst, S. M., Hodyss, R., Beauchamp, P. M., Smith, M. A., and Willis, P. A., 2012. Titan tholins: simulating Titan organic chemistry in the Cassini-Huygens era. Chemical Reviews 112(3), 1882. http://dx.doi.org/10.1021/cr200221x. 39 Charnay, B., Forget, F., Tobie, G., Sotin, C., and Wordsworth, 3D modeling of a pure ni- 269. R., trogen atmosphere and geological http://dx.doi.org/10.1016/j.icarus.2014.07.009. 2014. Titan's past and future: implications. Icarus, 241, Chen, W., Zhibao, D., and Zhenshan, L. et al., 1996. Wind tun- nel test of the influence of moisture on the erodibility of loessial sandy loam soils by wind. Journal of Arid Environments, 34(4), 391. http://dx.doi.org/10.1006/jare.1996.0119. Christenson, H. K, 1988. Adhesion between surfaces in undersaturated vapors: a reexamination of the influence of meniscus curvature and surface forces. Journal of Colloid and Interface Science, 121(1), 170. http://dx.doi.org/10.1016/0021-9797(88)90420-1. Clark, R. N., Curchin, J. M., Barnes, J. W., et al. 2010. Detection and map- ping of hydrocarbon deposits on Titan. Journal of Geophysical Research (Planets), 115, E10005. http://dx.doi.org/10.1029/2009JE003369. Curtis, D. B., Hatch, C. D., Hasenkopf, C. A., et al., 2008. Labo- ratory studies of methane and ethane adsorption and nucleation onto organic particles: Application to Titan's clouds. Icarus, 195, 792. http://dx.doi.org/10.1016/j.icarus.2008.02.003. F´ecan, F., Marticorena, B., and Bergametti, G., 1999. Parametrization of the increase of the aeolian erosion threshold wind friction velocity due to soil moisture for arid and semi-arid areas. Annales Geophysicae, 17, 149. http://dx.doi.org/10.1007/s00585-999-0149-7. Findor´ak, R., Frohlichov´a, M., Legemza, J., and Findor´akov´a, L., 2016. Thermal degradation and kinetic study of sawdusts and walnut shells via thermal analysis. Journal of Thermal Analysis and Calorimetry, 1-6. http://dx.doi.org/10.1007/s10973-016-5264-6. Fulchignoni, M., Ferri, F., Angrilli, F., et al. 2005. In situ measurements of the physical characteristics of Titan's environment. Nature, 438, 785. http://dx.doi.org/10.1038/nature04314. Gladstone, G. R., Stern, S. A., Ennico, K., et al., 2016. The atmo- sphere of Pluto as observed by New Horizons. Science, 351, aad8866. http://dx.doi.org/10.1126/science.aad8866. 40 Greeley, R., White, B., Leach, R., Iversen, J., and Pollack, J. B., 1976. Mars - Wind friction speeds for particle movement. Geophys. Res. Lett., 3, 417. http://dx.doi.org/10.1029/GL003i008p00417. Greeley, R., White, B. R., Pollack, J. B., Iverson, J. D., and Leach, R. N., 1977. Dust storms on Mars: Considerations and simulations. NASA Tech. Memo., NASA-TM -- 78423, 30 p. Greeley, R., Leach, R., White, B., Iversen, J., and Pollack, J. B., 1980. Threshold windspeeds for sand on Mars - Wind tunnel simulations. Geo- phys. Res. Lett., 7, 121. http://dx.doi.org/10.1029/GL007i002p00121. Greeley, R., Iversen, J., Leach, R., et al. 1984, Windblown sand on Venus - Preliminary results of laboratory simulations. Icarus, 57, 112. Greeley, R., Marshall, J. R., and Leach, R. N. 1985, Microdunes and other aeolian bedforms on Venus: Wind tunnel simulations. Reports of Planetary Geology and Geophysics Program. Greeley, R., and Iversen, J. D. 1985. Wind as a geological process on Earth, Mars, Venus and Titan. Cambridge Planetary Science Series, Vol. 4. Cam- bridge University Press. Gwon, J. G., Lee, S. Y., Chun, S. J., Doh, G. H., and Kim, J. H., 2010. Effects of chemical treatments of hybrid fillers on the physical and thermal properties of wood plastic composites. Com- posites Part A: Applied Science and Manufacturing, 41(10), 1491. http://dx.doi.org/10.1016/j.compositesa.2010.06.011. Hirtzig, M., B´ezard, B., Lellouch, E., et al. 2013. Titan's surface and at- mosphere from Cassini/VIMS data with updated methane opacity. Icarus, 226, 470. http://dx.doi.org/10.1016/j.icarus.2013.05.033. Horst, S. M., and Tolbert, M. A., 2013. In Situ Measurements of the Size and Density of Titan Aerosol Analogs. ApJL, 770, L10. http://dx.doi.org/10.1088/2041-8205/770/1/L10. Imanaka, H., Cruikshank, D. P., Khare, B. N., and McKay, C. P., 2012. Optical constants of Titan tholins at mid-infrared wavelengths (2.5-25 µm) and the possible chemical nature of Titan's haze particles. Icarus, 218, 247. http://dx.doi.org/10.1016/j.icarus.2011.11.018. 41 Israelachvili, J. N., 2011. Intermolecular and surface forces. Academic press. Iversen, J. D., White, B. R., Pollack, J. B., and Greeley, R. 1976. Saltation threshold on Mars - The effect of interparticle force, surface roughness, and low atmospheric density. Icarus, 29, 381. http://dx.doi.org/10.1016/0019- 1035(76)90140-8. Iversen, J. D., and White, B. R., 1982. Saltation threshold on Earth, Mars and Venus. Sedimentology, 29, 111. http://dx.doi.org/10.1111/j.1365- 3091.1982.tb01713.x. Iwamatsu, M., and Horii, K., 1996. Capillary condensation and adhesion of two wetter surfaces. Journal of colloid and interface science, 182(2), 400-6. http://dx.doi.org/10.1006/jcis.1996.0480. Kok, J. F., Parteli, E. J. R., Michaels, T. I., and Karam, D. B., 2012. The physics of wind-blown sand and dust. Reports on Progress in Physics, 75, 106901. http://dx.doi.org/10.1111/10.1088/0034-4885/75/10/106901. Lavvas, P., Griffith, C. A., and Yelle, R. V., 2011. Condensation in Titan's atmosphere at the Huygens landing site. Icarus, 215, 732. http://dx.doi.org/10.1016/j.icarus.2011.06.040. Le Gall, A., Janssen, M. A., Wye, L. C., et al. 2011. Cassini SAR, radiometry, scatterometry and altimetry observations of Titan's dune fields. Icarus, 213, 608. http://dx.doi.org/10.1016/j.icarus.2011.03.026. Lindal, G. F., Wood, G. E., Hotz, H. B., et al., 1983. The atmosphere of Titan - an analysis of the Voyager 1 radio occultation measurements. Icarus, 53, 348. http://dx.doi.org/10.1016/0019-1035(83)90155-0. Lorenz, R. D., 2014. Physics of transport http://dx.doi.org/10.1016/j.icarus.2013.06.023. on Titan: A brief review. saltation Icarus, and 230, sand 162. Lorenz, R. D., Wall, S., Radebaugh, J., et al., 2006. The Sand Seas of Titan: Cassini RADAR Observations of Longitudinal Dunes. Science, 312, 724. http://dx.doi.org/10.1126/science.1123257. McCord, T. B., Hansen, G. B., Buratti, B. J., et al., 2006. Composi- tion of Titan's surface from Cassini VIMS. Planet. Space Sci., 54, 1524. http://dx.doi.org/10.1016/j.pss.2006.06.007. 42 McKenna-Neuman, C. M., and Nickling, W. G., 1989. A theoretical and wind tunnel investigation of the effect of capillary water on the entrain- ment of sediment by wind. Canadian Journal of Soil Science, 69(1), 79. http://dx.doi.org/10.1016/10.4141/cjss89-008. McKenna-Neuman, C. M., 2003. Effects of Temperature and Humidity upon the Entrainment of Sedimentary Particles by Wind. Boundary-Layer Me- teorology, 108(1), 61. http://dx.doi.org/10.1023/A:1023035201953. McKenna-Neuman, C. M., and Sanderson, S., 2008. Humidity control of particle emissions in aeolian systems. Journal of Geophysical Research: Earth Surface, 113(F2). http://dx.doi.org/10.1029/2007JF000780. Miqueu, C., Broseta, D., Satherley, J., Mendiboure, B., Lachaise, J., and Graciaa, A., 2000. An extended scaled equation for the tempera- ture dependence of the surface tension of pure compounds inferred from an analysis of experimental data. Fluid Phase Equilibria, 172(2), 169. http://dx.doi.org/10.1016/S0378-3812(00)00384-8. Nield, E. V., Burr, D. M, Bridges, N. T, James, J. K,and Emery, J. P, et al., 2016. A Wind Tunnel Study of the Effect of Pressure on Saltation Threshold Conditions. LPSC 2016, 47, 1028. Nourbakhsh, A., Baghlani, F. F, and Ashori, A., filled SiO2 mechanical properties. http://dx.doi.org/10.1016/j.indcrop.2010.10.010. husk/polypropylene rice 2011. Nano- Physico- Industrial Crops and Products, 33(1), 183. composites: Pirayesh, H., Khazaeian, A., and Tabarsa, T., 2012. The potential for using walnut (Juglans regia L.) shell as a raw material for wood-based parti- cleboard manufacturing. Composites Part B: Engineering, 43(8): 3276. http://dx.doi.org/10.1016/j.compositesb.2012.02.016. Radebaugh, J., Lorenz, R. D., Lunine, J. I., et al., Dunes on Titan observed by Cassini Radar. http://dx.doi.org/10.1016/j.icarus.2007.10.015. Icarus, 194, 2008. 690. Ravi, S., D'Odorico, P., Over, T. M., and Zobeck, T. M., 2004. On the effect of air humidity on soil susceptibility to wind ero- sion: The case of air-dry soils. Geophys. Res. Lett., 31, L09501. http://dx.doi.org/10.1029/2004GL019485. 43 Ravi, S., Zobeck, T. M., and Over, T. M., 2006. On the effect of mois- ture bonding forces in air-dry soils on threshold friction velocity of wind erosion. Sedimentology, 53(3), 597. http://dx.doi.org/10.1111/j.1365- 3091.2006.00775.x. Rodriguez, S., Garcia, A., Lucas, A., et al. 2014. Global mapping and characterization of Titan's dune fields with Cassini: Corre- lation between RADAR and VIMS observations. Icarus, 230, 168. http://dx.doi.org/10.1016/j.icarus.2013.11.017. Sagan, C., and Khare, B. N., 1979. Tholins - Organic chemistry of interstellar grains and gas. Nature, 277, 102. http://dx.doi.org/10.1038/277102a0. Selah, A., and Fryrear, D. W., 1995. Threshold wind velocities of wet soils as affected by wind blown sand. Soil Science, 160(4), 304. Shao, Y., and Lu, H., sion threshold friction velocity. J. Geophys. Res., http://dx.doi.org/10.1029/2000JD900304. 2000. A simple expression for wind ero- 22437. 105, Smith, B. A., Soderblom, L. A., Banfield, D., et al., 1989. Voy- Imaging Science Results. Science, 246, 1422. ager 2 at Neptune: http://dx.doi.org/10.1126/science.246.4936.1422. Soderblom, L. A., Kirk, R. L., Lunine, J. I., et al., 2007. Correlations between Cassini VIMS spectra and RADAR SAR images: Implications for Titan's surface composition and the character of the Huygens Probe Landing Site. Planet. Space Sci., 55, 2025. http://dx.doi.org/10.1016/j.pss.2007.04.014. Stern, S. A., Bagenal, F., Ennico, K., et al., 2015. The Pluto system: Ini- tial results from its exploration by New Horizons. Science, 350, aad1815. http://dx.doi.org/10.1126/science.aad1815. Thomas, N., Sierks, H., Barbieri, C., et al., 2015. The morphological di- versity of comet 67P/Churyumov-Gerasimenko. Science, 347, aaa0440. http://dx.doi.org/10.1126/science.aaa0440. Tokano, T., 2010. Relevance of fast westerlies at equinox for the eastward elongation of Titan's dunes. Aeolian Research, 2, 113. http://dx.doi.org/10.1016/j.aeolia.2010.04.003. 44 Tuller, M., and Or, D., 2005. Water films and scaling of soil characteristic curves at low water contents. Water Resources Research, 41, W09403. http://dx.doi.org/10.1029/2005WR004142. Webb, P. A., 2001. Volume and density determinations for particle technol- ogists. Micromeritics Instrument Corp, 2(16). 45
1508.05715	1	1508	2015-08-24T08:07:16	A wide binary trigger for white dwarf pollution	[ "astro-ph.EP", "astro-ph.SR" ]	Metal pollution in white dwarf atmospheres is likely to be a signature of remnant planetary systems. Most explanations for this pollution predict a sharp decrease in the number of polluted systems with white dwarf cooling age. Observations do not confirm this trend, and metal pollution in old (1-5 Gyr) white dwarfs is difficult to explain. We propose an alternative, time-independent mechanism to produce the white dwarf pollution. The orbit of a wide binary companion can be perturbed by Galactic tides, approaching close to the primary star for the first time after billions of years of evolution on the white dwarf branch. We show that such a close approach perturbs a planetary system orbiting the white dwarf, scattering planetesimals onto star-grazing orbits, in a manner that could pollute the white dwarf's atmosphere. Our estimates find that this mechanism is likely to contribute to metal pollution, alongside other mechanisms, in up to a few percent of an observed sample of white dwarfs with wide binary companions, independent of white dwarf age. This age independence is the key difference between this wide binary mechanism and others mechanisms suggested in the literature to explain white dwarf pollution. Current observational samples are not large enough to assess whether this mechanism makes a significant contribution to the population of polluted white dwarfs, for which better constraints on the wide binary population are required, such as those that will be obtained in the near future with Gaia.	astro-ph.EP	astro-ph	MNRAS 000, 1 -- 12 (2015) Preprint 12 March 2021 Compiled using MNRAS LATEX style file v3.0 A wide binary trigger for white dwarf pollution Amy Bonsor1,2⋆ and Dimitri Veras3 1Institute of Astronomy, University of Cambridge, Madingley Road, Cambridge, CB3 0HA, UK 2School of Physics, H.H. Wills Physics Laboratory, University of Bristol, Tyndall Avenue, Bristol BS8 1TL, UK 3Department of Physics, University of Warwick, Coventry CV4 7AL, UK Accepted XXX. Received YYY; in original form ZZZ ABSTRACT Metal pollution in white dwarf atmospheres is likely to be a signature of remnant planetary systems. Most explanations for this pollution predict a sharp decrease in the number of polluted systems with white dwarf cooling age. Observations do not confirm this trend, and metal pollution in old (1-5 Gyr) white dwarfs is difficult to explain. We propose an alternative, time-independent mechanism to produce the white dwarf pollution. The orbit of a wide binary companion can be perturbed by Galactic tides, approaching close to the primary star for the first time after billions of years of evolution on the white dwarf branch. We show that such a close approach perturbs a planetary system orbiting the white dwarf, scattering planetesimals onto star-grazing orbits, in a manner that could pollute the white dwarf's atmosphere. Our estimates find that this mechanism is likely to contribute to metal pollution, alongside other mechanisms, in up to a few percent of an observed sample of white dwarfs with wide binary companions, independent of white dwarf age. This age independence is the key difference between this wide binary mechanism and others mechanisms suggested in the literature to explain white dwarf pollution. Current observational samples are not large enough to assess whether this mechanism makes a significant contribution to the population of polluted white dwarfs, for which better constraints on the wide binary population are required, such as those that will be obtained in the near future with Gaia. Key words: planet-star interactions, planets and satellites: dynamical evolution and stability, stars: evolution, stars: AGB and post-AGB, Oort Cloud, stars:kinematics and dynamics 1 INTRODUCTION Elements heavier than helium sink rapidly in the atmo- spheres of white dwarfs, where sinking timescales are on the order of days to weeks (DA white dwarfs) and 104 − 106 years (DB white dwarfs) (Koester & Wilken 2006) or see Fig. 1 of Wyatt et al. (2014). Observations of emission lines from metallic species in the atmospheres of white dwarfs (e.g. Koester et al. 1997; Zuckerman et al. 2003; Melis et al. 2010), therefore, suggest the recent accretion of the ob- served material. Accretion from the interstellar medium was ruled out by Farihi et al. (2010); Aannestad et al. (1993); Jura (2006); Kilic & Redfield (2007); Barstow et al. (2014). These observations are thought to indicate the accretion of planetary material (e.g. Alcock et al. 1986; Debes & Sigurdsson 2002; Jura 2003, 2008; Kilic et al. 2006), in at least 25% (Zuckerman et al. 2003, 2010), but ⋆ E-mail: [email protected] c(cid:13) 2015 The Authors maybe up to 50% (Koester et al. 2014) of white dwarfs. Dusty and/or gaseous discs, very close (r < R⊙) to a subset of these polluted white dwarfs may be indicative of the ac- creting material (e.g Kilic et al. 2006; Gansicke et al. 2006; von Hippel et al. 2007; Jura et al. 2007; Gansicke et al. 2007). The composition of the accreted material, when anal- ysed in detail supports the planetary material hypothesis (e.g. Klein et al. 2010, 2011; Gansicke et al. 2012). Planetary systems are common on the main-sequence, and throughout the Milky Way they are the rule, not the exception (Cassan et al. 2012). Observations with HARPS suggest that at least 50% of solar-type stars har- bour at least one planet with a period less than 100 days (Mayor et al. 2011), whilst Herschel found that at least 20% of FGK stars have a detectable debris disc (Eiroa et al. 2013). Although close-in (< 1 − 5AU) plan- ets may be swallowed by the expanding giant's enve- lope (Villaver & Livio 2007, 2009; Mustill & Villaver 2012; Adams & Bloch 2013; Villaver et al. 2014), there is no ev- 2 A. Bonsor et al. idence that all outer planetary systems are destroyed (e.g. Jura 2004; Bonsor & Wyatt 2010). Dynamical instabilities in the outer planetary system, induced following stellar mass loss (e.g Debes & Sigurdsson 2002; Veras et al. 2011; Veras & Wyatt 2012; Veras & Tout 2012; Veras et al. 2013c; Voyatzis et al. 2013; Mustill et al. 2014; Veras & Gansicke 2015), may scatter asteroids or comets onto star-grazing or- bits (Bonsor et al. 2011; Bonsor & Wyatt 2012; Debes et al. 2012; Frewen & Hansen 2014), where they are tidally dis- rupted and accreted onto the star (e.g. Graham et al. 1990; Jura 2003, 2008; Bear & Soker 2013; Veras et al. 2014c, 2015c). Given a sufficiently massive planetesimal belt, it has been shown that such a theory can explain the ob- served metal pollution (Bonsor et al. 2011; Debes et al. 2012; Frewen & Hansen 2014). However, a steep decrease in the mass of asteroids/comets scattered onto star-grazing orbits with time after the formation of the white dwarf, is predicted (Bonsor et al. 2011; Debes et al. 2012). This find- ing would correspond to a steep decrease in the level of pol- lution (or fraction of systems with pollution), as a function of time. Although late time instabilities may be induced in multi-planet systems (Veras et al. 2013c; Mustill et al. 2014; Veras & Gansicke 2015), the frequency of these will always be less than those around young white dwarfs. Observationally there is no evidence that there are fewer, old polluted white dwarfs, nor that the level of pollu- tion decreases with white dwarf cooling age (Wyatt et al. 2014; Koester et al. 2014). Even the archetypal polluted white dwarf, van Maanen's star (van Maanen 1920), has an effective temperature of 6,220K, or cooling age of ∼ 3Gyr (Sion et al. 2009), and there are many further ob- servations of polluted, old, white dwarfs (Teff < 8, 000K, equivalent to cooling ages of up to 5 Gyrs Farihi et al. (2011); Koester et al. (2011)). Instabilities induced follow- ing stellar mass loss struggle to explain pollution for white dwarfs with cooling ages of Gyrs (Debes & Sigurdsson 2002; Bonsor et al. 2011; Debes et al. 2012). Alternative explana- tions for the pollution in these systems have been suggested, including stellar encounters (Farihi et al. 2011), or a rela- tionship with magnetic fields (Hollands et al. 2015). Others, such as exo-Oort cloud comet impacts (Veras et al. 2014d; Stone et al. 2015) and volatile sublimation of minor bodies in planet-less systems (Veras et al. 2015a), have largely been ruled out. Here, we suggest an alternative explanation. It has been shown for main-sequence planetary systems, that whilst in general wide (a > 1, 000AU) binary companions do not in- fluence the dynamics of the planetary system, the orbit of the binary varies due to Galactic tides. During periods of 'close' pericentre passage, the binary may excite the eccen- tricities of planets, even ejecting them (Kaib et al. 2013). The increased eccentricity of exoplanets in systems with wide binary companions has been confirmed by observations (Kaib et al. 2013). Here, we propose that the same scenario could be applied to white dwarf planetary systems. If the white dwarf is orbited by a wide binary companion, the plan- etary system may remain unperturbed for billions of years, before Galactic tides alter the orbit of the companion such that it induces dynamical instabilities in the planetary sys- tem. These instabilities can lead to material being scattered onto star-grazing orbits and accreted onto the white dwarf, such that we observe pollution in the white dwarf's atmo- sphere. This process has the advantage of being independent of the age of the white dwarf, although it depends critically on the population of wide binaries orbiting white dwarfs. The population of wide (> 1, 000 AU) binaries is difficult to characterise, in general. Common proper motions indicate that two stars may be related (e.g. L´epine & Bongiorno 2007; Makarov et al. 2008), but follow- up parallax and radial velocity observations are required to determine whether they are bound. Wide binaries are, therefore, likely to be significantly more common than determined observationally. Theoretical models predict a wide binary fraction of 1-30%, with 103 < a < 0.1pc (Kouwenhoven et al. 2010), whilst observations of solar-type stars (L´epine & Bongiorno 2007; Raghavan et al. 2010) find a fraction of 9.5% or 11.5%, and observations of field white dwarfs yield a wide binary fraction of 22% (Farihi et al. 2005). There are, however, very few white dwarfs with known wide binary companions that have been searched for pollu- tion, and many surveys for pollution have focussed on ap- parently single white dwarfs. Zuckerman (2014) present a sample of 17 white dwarfs with companions separated by more than 1,000AU, searched for Ca II with Keck/VLT and find that 5 are polluted. We note here that white dwarfs in close binaries may be polluted by interactions with the stellar wind of the companion, as discussed in (e.g. Zuckerman et al. 2003; Zuckerman 2014), and for this reason we focus on 'wide' binaries. Fig. 8 of Veras & Tout (2012) shows that the boundary between 'close' and 'wide' occurs at tens of AU. In this paper, we illustrate the manner in which a wide binary companion could lead to pollution in a white dwarf. This mechansim is likely to be just one of many that lead to pollution. The critical difference of this mechanism is the absence of any dependence on white dwarf cooling age. §2 shows how a wide binary's orbit can be altered by Galactic tides, such that a close approach occurs between the sec- ondary and a planetary system orbiting the primary (white dwarf). §3 presents simulations that show how this close ap- proach can lead to planetesimals scattered onto star-grazing orbits, a pre-requisite for pollution. In §4 we estimate the fraction of white dwarfs with wide binary companions where pollution might occur, which is compared to observations in §5. In §6 and §7 we discuss our results and present our con- clusions. 2 THE CHANGE IN ORBIT OF A WIDE BINARY DUE TO THE GALACTIC TIDE We firstly illustrate the manner in which the Galactic tide can change the orbit of a wide binary. Above a critical sep- aration, the Galactic field strength is different for the two components of the binary, such that their orbit evolves. An- alytically the effect of the Galactic tide on the binary's orbit can be understood through the equations of motion for the binary's orbital parameters. Although we are unaware of a complete analytic solution to these equations (1-4), numer- ical solutions can be used to explore the phase space and help us to understand the evolution of the binary's orbit. MNRAS 000, 1 -- 12 (2015) A wide binary trigger for white dwarf pollution 3 2.1 Equations of motion As the majority of observed polluted white dwarfs are nearby (e.g. within 0.1kpc Holberg et al. 2008), we restrict our- selves to the Solar neighbourhood, which we assume re- sides within 8kpc of the Galactic centre. This restriction allows us to neglect planar Galactic tides (see Fig. 2 of Veras & Evans 2013a), which represents a widely-used sim- plification (Heisler & Tremaine 1986; Matese & Whitman 1989, 1992; Matese et al. 1995; Breiter et al. 1996; Brasser 2001; Breiter & Ratajczak 2005). Further, we consider only bound binary systems. Boundedness effectively equates to adiabaticity in the Galactic tidal regime because no escape can occur in the adibatic regime (Veras & Evans 2013a) and the non-adibatic limit is close to the system's Hill ellipsoid (Veras et al. 2014a). For very wide binaries (ab >> 105AU) adiabicity is no longer a good approximation (shown by Fig. 3 of Veras & Evans (2013b)) and the full equations of mo- tion describing this regime are shown in Column 5 of Table 1 of Veras et al. (2014a). Here, we consider the adiabatic regime. Consequently we can, by considering vertical tides alone and using the perturbed two-body problem, derive the equations of mo- tion for the orbital elements of the binary (as shown in Veras & Evans (2013a)); semi-major axis, ab, mean motion, nb, eccentricity, eb, inclination of the binary orbital plane to the Galactic plane, ib, argument of pericentre, ωb : dnb dt nb nb deb dt dib dt nb dωb dt = 0 = − 5ebp1 − e2 b 2 (1) cos ωb sin ωb sin2 ibΥzz (2) = = 5e2 b sin 2ωb sin 2ib 8p1 − e2 b Υzz 5 sin2 ωb (cid:0)sin2 ib − e2 2p1 − e2 b b(cid:1) − (cid:0)1 − e2 b(cid:1) (3) Υzz (4) The variable Υzz is determined by the Galactic model used, and contains information about the matter density and gravitational potential. Here, we use the disc component of the three-component model of Veras et al. (2013a). Because we fix the distance from the Galactic centre at 8 kpc, we also fix Υzz = −5.352 × 10−30s−2. 2.2 An example binary Although we are unaware of a complete analytic solution to equations (1-4), we can obtain numerical solutions that en- able us to follow the evolution of example orbits. There are several key properties of the evolution of binary orbits under the influence of the Galactic tide. Of particular relevance to this work is the evolution of the binary's semi-major axis and eccentricity. The semi-major axis remains constant (Eq. 1), under the adiabatic approximation of these equations. The eccentricity evolves periodically, with a time period and am- plitude that depend on the initial conditions of the binary's orbit, as well as the strength of the Galactic tide (Υzz). If a binary's orbit is to perturb the primary's planetary system, we are interested in the maximum eccentricity that its orbit reaches and the times between eccentricity peaks. Thus, we focus on these two parameters in the following sections. MNRAS 000, 1 -- 12 (2015) Firstly, in order to illustrate our proposed scenario, let us consider, for example, a 1M⊙ binary companion orbiting a 1M⊙ primary, in the solar neighbourhood (R = 8kpc), with a semi-major axis of 3000AU, inclined to the Galac- tic plane by 80◦. We gave the binary an initial eccentricity of e0 = 0.2, and a range of initial arguments of pericen- tre (ω0 = 5, 40, 90, 160◦) and followed the evolution of the binary's eccentricity, shown in the top panel Fig. 1. The ec- centricity of this binary's orbit evolves on time periods that are significantly longer than the main-sequence lifetimes of the stars, or even the age of the Universe. Therefore, if the binary starts at low eccentricity, as we assume, it remains at low eccentricity throughout the primary's main-sequence lifetime. We then consider the fate of this binary system, as the primary becomes a white dwarf. So far, the primary's plan- etary system is relatively unperturbed by the binary com- panion. By using a typical mass loss prescription1, the pri- mary becomes a white dwarf of mass 0.52M⊙ (see caption of Fig.4 of Veras et al. (2013c)). At this semi-major axis mass loss is likely to be non-adiabatic (see for example Fig.3 of (Veras et al. 2014a)), thus the exact expansion of the bi- nary's orbit is difficult to predict (Veras et al. 2011). We, therefore, make a reasonable assumption that the binary's orbit expands by a factor of 3 to ab = 9, 000AU. Both the re- duction in mass of the binary and the increase in semi-major axis of the binary, increase the ability of Galactic tides to change the binary's orbit. As is shown in the bottom panel of Fig. 1, the binary's eccentricity now evolves on 'shorter' timescales, several rather than tens of Gyrs. Thus, with our assumption that the binary's orbit starts the white dwarf phase with an eccentricity of e0 = 0.2, depending on ω0, it evolves to its maximum eccentricity (minimum pericentre) within 3-7Gyrs. Approaching the primary at a distance of closest approach of around 150AU, the binary is likely to perturb the primary's planetary system. It is these pertur- bations that we invoke for their ability to produce polluted white dwarfs, particularly at late times. 2.2.1 Including mass loss Fig. 1 integrates Eqs. 1-4 for the main-sequence and the white dwarf phases separately. It is possible to integrate these equations self-consistently, if a formulation for stellar mass loss is included. However, it is computationally inten- sive and therefore, we only perform this calculation for one example system. Using the stellar evolution prescription of Hurley et al. (2000), coupled with N-body integrations, as in Veras et al. (2011), Fig. 2 shows the evolution of the peri- centre of a binary orbit that starts with a 2M⊙ and a 1M⊙ planet on an orbit with ab = 4, 000AU, but where the 2M⊙ primary evolves to become a white dwarf of mass 0.64M⊙. The same behaviour as in Fig. 1 is seen. The binary's orbit 1 The SSE code (Hurley et al. 2000) applies Reimer's mass loss (Kudritzki & Reimers 1978) at early evolutionary stages, but during the AGB it applies the semi-empirical mass-loss rate of Vassiliadis & Wood (1993), reaching a maximum during the su- perwind of 1.36 × 109(L∗/L⊙)M⊙yr−1, where L∗ is the stellar luminosity. 4 A. Bonsor et al. M1=0.8MŸ, M2=1MŸ, i0=80é, a=3000 au Ω0=160é Ω0=90é Ω0=40é Ω0=5é 1.0 0.8 0.6 0.4 0.2 e 0 600 1200 1800 2400 u a q u a q 6000 5000 4000 3000 2000 1000 0.0 0 2 4 6 TimeGyr 8 3000 10 0 0 1 2 3 5 4 6 TimeGyr 7 8 9 10 M1=0.8MŸ, M2=0.52MŸ, i0=80é, a=9000 au 0 1.0 e 0.8 0.6 0.4 0.2 0.0 0 Ω0=160é Ω0=90é Ω0=40é Ω0=5é 4 6 TimeGyr 2 1800 3600 5400 7200 u a q 8 9000 10 Figure 1. The change in orbital parameters due to Galactic tides and stellar evolution, for several example binaries. The top panel represents the primary on the main-sequence and the bot- tom panel, once the primary has evolved to become a white dwarf. Tides can produce extreme eccentricity oscillations during the white dwarf phase which do not occur along the main sequence, thereby triggering planetary scattering which can ultimately lead to white dwarf pollution. Assumed in these plots is that the bi- nary orbit expands non-adibatically by a factor of 3 due to mass loss from the secondary star M2. is not perturbed significantly during the main-sequence evo- lution, but undergoes long period oscillations once the star becomes a white dwarf. The mass loss is modelled to be isotropic, which rep- resents an excellent approximation to reality (Veras et al. 2013b). In this case, the pericentre can never decrease dur- ing giant branch mass loss (Eq. 21 of Veras et al. (2011)). 3 THE POLLUTION OF THE WHITE DWARF We have shown that Galactic tides can perturb orbits such that a secondary companion approaches close to any plane- tary system orbiting the primary (white dwarf). In this sec- tion we present simulations that illustrate how the close ap- proach of the secondary could perturb the planetary system orbiting the white dwarf in such a manner that planetesi- mals are scattered onto star-grazing orbits. For the purpose of this work we consider such scattering sufficient to produce white dwarf pollution, see §6 for a discussion of this assump- tion. These simulations are intended as a proof of concept, rather than a detailed analysis. Figure 2. A demonstration of how the first close approach in a binary system can occur during the white dwarf phase. Plotted is the self-consistent pericentre distance evolution of one partic- ular binary system through all phases of stellar evolution. The initial binary parameters are: M1 = 2M⊙, M2 = 1M⊙, eb = 0.1, ab = 4000 au, ib = 110◦, ωb = 0◦. The red, blue and purple curves represent the main sequence, giant branch, and white dwarf phases of evolution, respectively. Adiabatic Galactic tidal equa- tions are used for the main sequence and white dwarf phases; the full non-adiabatic equations for both tides and mass loss are used along giant branch phases, with f0 = 150◦. 3.1 Details of our simulations Clearly, the parameter space available to investigate, both in terms of the diversity of planetary system architectures and binary orbits, is enormous. We can only investigate ex- ample systems. We, therefore, consider a binary in which a 1M⊙ primary is orbited by a companion of mass 0.8M⊙. The 1M⊙ primary evolves to become a 0.52M⊙ white dwarf (Veras et al. 2013c). 0.8M⊙ represents a typical value for a main-sequence stellar mass (Parravano et al. 2011), and such stars will remain on the main-sequence for the current age of the Universe (∼ 14 Gyrs). The primary is orbited by a very simplistic planetary system with one planet and one planetesimal belt or de- bris disc, similar to Neptune and the Kuiper belt in our So- lar System. The presence of both the planet and planetes- imals means that perturbations to the orbit of the planet can lead to planetesimals being scattered onto star-grazing orbits (Bonsor et al. 2011; Debes et al. 2012). In addition, the binary companion can perturb planetesimals directly onto star-grazing orbits (e.g. Marzari et al. 2005). We fix the plane of the planetary and binary orbits to be the same. We fix the planet's semi-major axis at 30AU, which mim- ics Neptune's semi-major axis, and the planetesimal belt runs from 30-50AU. We assume that at such a small orbital distance stellar mass loss is adiabatic, such that the white dwarf is orbited by a planet at 60AU and belt running from 60-100AU. We neglect the potentially destructive effects of YORP spin up on the planetesimals (Veras et al. 2014b) and the potentially significant movement of these asteroids due to the Yarkovsky effect from giant branch stellar evolution (Veras et al. 2015b). These effects would have a greater in- fluence on an exo-asteroid belt located within 10AU than an exo-Kuiper belt located at several tens of AU. Our simulations are run using the Mercury N-body in- MNRAS 000, 1 -- 12 (2015) A wide binary trigger for white dwarf pollution 5 tegrator, RADAU (Chambers 1999). We track particles that are scattered close to the star. It is these particles that we assume are tidally disrupted and pollute the white dwarf. The RADAU integrator is used in order to track the evolu- tion of these particles to small pericentre, in a manner that the sympletic integrator cannot. However, in order to limit the runtime for our simulations, we only track particles as close in as 0.1AU. We consider the particles that are scat- tered interior to 0.1AU to be representative of the popula- tion scattered onto star-grazing orbits where they would be disrupted, more likely to occur when r <∼ R⊙ (Veras et al. 2014c). Given that our purpose is to illustrate the feasibility of this mechanism, we consider this assumption to be suf- ficient. Our choice merely means that we over-estimate the population scattered onto star-grazing orbits, but do so in a consistent manner in all our simulations, thereby removing comparison inconsistencies between simulations. We use our simulations as a tool to investigate the bi- nary orbits that perturb the planetary system and scatter particles onto star-grazing orbits. Therefore, we focus on short integrations (10-100 binary orbits) for which the bi- nary orbit is fixed and does not evolve due to the Galactic tide. These integrations can represent the evolution of the planetary system directly following the binary's evolution onto the orbit considered. The exact orbits of the planet and planetesimals will clearly evolve with time and the evolution of the binary's orbit. It is not possible to consider all possible evolutionary paths. Instead we investigate the increase in the scattering of planetesimals onto star-grazing orbits in comparison with a simulation in which the binary is not present. Although this comparison may not always mimic the potential effects of an excited planetesimal belt, or scattered disc, it illus- trates clearly the perturbative effect of the binary. We start our planetesimals with orbits that are initially unperturbed by the planet and have semi-major axes, eccentricities, in- clinations, longitudes of pericentre, longitudes of ascending nodes and mean anomalies, which are randomly drawn from uniform distributions with ranges of 60 < a < 100AU, 0 < e < 0.02, 0 < i < 1◦ and 0◦ < λ, Ω, M < 360◦. 3.2 An example system The aim of this section is to illustrate the manner in which the change in orbit of a binary due to Galactic tides can lead to the pollution of a white dwarf, even a white dwarf with a cooling age of Gyrs or longer. We consider the example binary system shown in Fig 1. The binary companion orbits at ab = 3, 000AU and whilst it is on the main-sequence has its eccentricity perturbed very little by the Galactic tide, up to an absolute maximum of eb = 0.6, from eb = 0.2 ini- tially, during a long main-sequence lifetime of 10Gyr. Even if the binary remains on an orbit with eb = 0.6 for 900Myr, a typical main-sequence lifetime, an outer belt survives or- biting the primary star. In fact, our test simulation finds that more than 60% of the original belt remain on stable or- bits. Given that Fig. 1 shows that the binary's orbit evolves and will only be on such an eccentric orbit for timescales significantly shorter than the primary's full main-sequence lifetime. Thus, in this example, we show that a planetary system can survive the star's main-sequence lifetime with- out being perturbed unduly by its binary companion. MNRAS 000, 1 -- 12 (2015) Figure 3. The semi-major axis, eccentricity distribution of plan- etesimals orbiting the white dwarf, prior and following the peri- centre passage of a wide binary companion with eb = 0.99. This is the same system as shown in the bottom panel of Fig. 1. This figure demonstrates that despite being dynamically excited, an exo-Kuiper belt largely survives stellar evolution of a primary star with a wide-orbit companion. Once the primary star in this example binary evolves to become a white dwarf, the orbits of the binary and the pri- mary's planetary system expand, for the planetary system adiabatically, but for the binary non-adiabatically. As above, we estimate that the binary's orbit moves to ab = 9, 000AU and consider the perturbations that it now makes on the pri- mary's planetary system, that has expanded adiabatically by a factor of 2 (belt is now at about 60-100AU). The binary's orbit is now perturbed by the Galactic tides, evolving up to an eccentricity of 0.978 (q ∼ 200AU) after several Gyrs of evolution on the white dwarf branch. As the binary's eccen- tricity increases, so does its influence on the planetary sys- tem, particularly during pericentre passages. Fig. 3 shows the effect on our example planetary system of a binary with eccentricity, eb = 0.99. Fig. 3 shows that prior to the binary's pericentre passage, the planetary system is relatively undis- turbed, but that following its pericentre passage the planet and many planetesimals have been scattered. As discussed above, the scattered planetesimals have the potential to pol- lute a white dwarf. Fig. 4 shows the number of planetesimals scattered onto star-grazing orbits following a pericentre pas- 6 A. Bonsor et al. Figure 4. Our example white dwarf planetary system, orbited by a binary companion with ab = 9, 000AU (shown in Fig. 1 and Fig. 3). The number of particles (N) that hit the star, as a function of time, in units of the binary's orbit period, with t = 0, occurring at pericentre. The high binary eccentricity induced by the Galactic tide increase the number of planetesimal collisions with the white dwarf. sage by a binary on orbits with various eccentricities. For binary eccentricities above 0.8, significantly more planetesi- mals are scattered onto star-grazing orbits than without the presence of the binary. If we refer back to Fig. 1, depending on the exact parameters of the binary's orbit, the binary's eccentricity could be increased from 0.2 to above 0.8 be- tween 2 and 6 Gyrs after the start of the white dwarf phase. In this manner we illustrate a mechanism that could lead to the pollution of old white dwarfs. 3.3 A brief investigation of the parameter space The above section illustrated one example system in which white dwarf pollution could occur in white dwarfs with cool- ing ages of Gyrs. Clearly, the parameter space of binary or- bits is huge and this mechanism will not work for all of them. A detailed investigation of the entire parameter space is be- yond the scope of this work, and not necessary to illustrate the potential of this mechanism. Instead, we perform a small additional suite of N-body simulations in order to show that there are many binary orbits for which this mechanism could work. We focus here on the white dwarf phase only. We consider binary companions with semi-major axes of ab = 2, 500AU, ab = 5, 000AU and ab = 9, 000AU. The eccentricity of the binary companion is varied, but fixed throughout each individual simulation. Fig. 5 shows the number of planetesimals scattered onto star-grazing orbits as a function of time. The pericentre passages of the binary are evident from the increase in the scattering rate, creating sawtooth-like profiles. As noted in §3 these numbers must be compared to the simulation with the binary removed, in order to remove some dependencies on the initial con- ditions used. For sufficiently high eccentricity companions, it is clear that more material hits the sun than without a binary companion. A conservative approximation to the ec- centricity at which this occurs can be obtained from the criterion described in Holman & Wiegert (1999) for the sta- Figure 5. The number of particles (N) scattered onto star- grazing orbits as a function of time, for ab = 2, 500AU (top), ab = 5, 000AU (middle) and ab = 9, 000AU (bottom). The planet is ejected for eb = 0.99, ab = 9, 000AU after 0.5Myr, hence the scattering stops. This figure demonstrates that as the binary's orbit evolves to high eccentricity, the rate at which particles are scattered inwards increases significantly. bility of planetary systems orbiting binary systems. Fixing acrit at 60AU in their Eq.1, although technically only valid for eb < 0.8, yields approximately the same critical eccen- tricities for the binary (eb > 0.8; ab = 2, 500AU; eb > 0.9, ab = 5, 000AU; eb > 0.95,ab = 9, 000AU) as found from our simulations (see Fig. 6). Alternative criteria exist for the stability of three body systems (Eggleton & Kiseleva 1995; Petrovich 2015), and although valid for higher eccentricity, have not been derived for two planets orbiting one star, MNRAS 000, 1 -- 12 (2015) A wide binary trigger for white dwarf pollution 7 pared to typical planetesimal belt masses thought to ex- ist on the main-sequence, even those below detection limits (Wyatt et al. 2007; Wyatt 2008), such as the Kuiper belt with its mass of around 0.1M⊕ (Gladman et al. 2001). Al- though these masses may be reduced by collisions, or dynam- ics as material is scattered during the star's main-sequence lifetime, we do not anticipate that in most systems these processes remove sufficient material to prevent detectable pollution being produced (Bonsor & Wyatt 2010), although of course in some planetary systems dynamical clearing may be highly efficient (e.g. Raymond et al. 2009, 2010). We choose to only simulate binaries with fixed eccentric- ity, in order to simplify our results and save computational time. From this choice we can clearly show an increased rate of scattering relative to simulations where the binary is not present. However, we miss any changes due to the exact evo- lution of the system as the binary's orbit increases smoothly in eccentricity, or due to particularities of the exact configu- ration of the system (e.g. planetesimals with excited eccen- tricities or inclinations or trapped in resonance etc ). These details may change the scattering rates. However, given our vague knowledge of the structure of any planetary system, we consider that the present simulations are sufficient to show that material is scattered and a white dwarf could be polluted in this manner. 4 ESTIMATING THE FREQUENCY OF POLLUTION CAUSED BY A WIDE BINARY COMPANION Having shown that for particular orbital parameters of the binary and planetary system, pollution of the white dwarf can occur, it would be useful to estimate the contribution of this mechanism to the population of polluted white dwarfs. This value firstly depends on the binary having an appropri- ate orbit that remains at low eccentricity for the primary's main-sequence evolution, but increases to sufficiently high eccentricity to perturb the primary's planetary system dur- ing the primary's white dwarf evolution, and secondly, on the primary having a planetary system with sufficient ma- terial that can be scattered onto star-grazing orbits in order to produce the pollution. 4.1 The binary orbit Firstly, we note that the binary orbits that are most likely to be interesting in terms of the proposed scenario are those at intermediate wide orbits (i.e. 3, 000 . ab . 10, 000AU). Here, the influence of Galactic tides is strong, but not suf- ficiently strong that the main-sequence planetary system would have been perturbed (as might occur for larger semi- major axes) and the binary does not orbit too close to the primary, so that the primary's planetary system is only per- turbed when the binary's orbit increases significantly in ec- centricity. Here, we estimate the fraction, fWB, of wide binaries where the eccentricity of the binary's orbit increases suf- ficiently (above a critical value), during the white dwarf phase, such that the primary's planetary system is per- turbed. The distribution of orbital parameters of wide bina- ries in the galaxy are not well known. We, therefore, con- Figure 6. The analytic criterion (solid line) for the stability of cir- cumprimary orbits from Holman & Wiegert (1999) (Eq. 1). The arrows show a comparison with the results of our simulations for the white dwarf phase, on short timescales. The shaded region in- dicates the limit of validity of this approximation, e > 0.8, which unfortunately corresponds to the region of interest. rather than a binary companion and a planet orbiting the central star. The critical point to take from these simulations is that planetesimals are scattered onto star-grazing orbits whilst a binary companion has high eccentricity, but remain on more stable orbits whilst the binary's eccentricity is lower. Thus, planetesimals may survive the star's main-sequence evolu- tion, as well as some proportion of its white dwarf evolution on stable orbits, before being scattered onto star-grazing or- bits as the binary's orbit evolves to high eccentricity. 3.4 Caveats and Assumptions The purpose of the above simulations is to illustrate that a wide binary companion could lead to white dwarf pollution at late times by scattering planetesimals onto star-grazing orbits. This requires that sufficient material survives to the white dwarf phase, as well sufficient material is scattered onto star-grazing orbits to produce detectable pollution. We assume here that planetesimals scattered onto star- grazing orbits lead to white dwarf pollution. This is a fairly robust assumption, as discussed in previous work, although gaps still exist in our knowledge of the exact processes involved (e.g. Jura 2003; Bonsor et al. 2011; Debes et al. 2012; Frewen & Hansen 2014). Planetesimals scattered close to the star are tidally disrupted, and ac- crete onto the star (Veras et al. 2014c, 2015c), potentially via dusty or gaseous accretion discs (Hartmann et al. 2011; Rafikov 2011a,b; Bochkarev & Rafikov 2011; Metzger et al. 2012; Rafikov & Garmilla 2012). Our simulations are good at showing that material is scattered onto star-grazing or- bits, but only indicate increased rates of scattering, rather than exact levels, and cannot, therefore, be used to predict accretion rates. However, the amount of material required to produce even the most polluted systems (e.g. a Ceres mass ∼ 10−4M⊕ for Dufour et al. (2010) or see Fig. 9 of Girven et al. (2012)) is low, and our simulations indicate high levels of scattering (Fig. 5). These masses are low com- MNRAS 000, 1 -- 12 (2015) 8 A. Bonsor et al. Figure 7. Estimating the fraction of systems where the wide bi- nary mechanism has the potential to pollute the primary's plan- etary system. Plotted is the cumulative fraction of systems that have a pericentre passage closer than a critical value during the white dwarf phase. The maximum eccentricity (minimum peri- centre) can be estimated from Fig. 5; see discussion in §4.1 for details. sider a uniform distribution of initial binary eccentricity, e0 , longitude of pericentre, ω0, sine of inclination to the Galactic plane, sin i0, and use a Monte-Carlo model to se- lect 300 binary orbits for each semi-major axis2. We fix the binary semi-major axis at ab = 9, 000AU, ab = 2, 500AU or ab = 5, 000AU and the masses of the pair fixed at MW D = 0.52M⊙ (evolved from m1 = 1M⊙ ) and M2 = 0.8M⊙), in line with Fig. 1 and Fig. 5. Fig. 7 shows the fraction of systems where a pericentre passage closer than a critical value (or equivalently, the eccentricity evolves to above a critical value), at any point during the white dwarf phase. A maximum time period of 14Gyr, the age of the Universe, is considered. From Fig. 5, an eccentricity of greater than approximately 0.8 (ab = 2, 500AU), 0.9(ab = 5, 000AU) or 0.95 (ab = 9, 000AU) is required to perturb a planetary sys- tem at 60-100AU. In other words, the binary must enter the region interior to ∼ 500AU to perturb the planetary system. Fig. 7 shows that 35%, 17% and 10% of systems evolve to this pericentre. Although clearly this value will vary signif- icantly with the binary's semi-major axis, we consider 20% to be a reasonable estimate for the fraction of wide bina- ries with orbital parameters in the correct range to produce polluted white dwarfs. Another constraint on the orbit of the binary results from the requirement that a planetary system survives, rel- atively unperturbed, orbiting the primary until late times. This requirement means that the binary's eccentricity must remain below a critical value during the primary's main- sequence evolution. This value can be roughly estimated by considering the criterion of Holman & Wiegert (1999) (Eq 1.) for stable circumprimary orbits. For example for our three example binaries with ab = 2, 500AU, 5,000AU and 9,000AU, the binary eccentricity must remain below eb < 0.4, 0.67 and 0.8 (assuming ab ∼ 800AU, 1,600AU and 2 Tests found no significant difference betweeen using N=100 and N=300, so N=300 is assumed to be sufficient. 3,000AU on the main-sequence), which is not a very strict constraint given the reduced perturbations due to the Galac- tic tide whilst the primary is on the main-sequence and the binary orbit tighter. Given the other uncertainities in our estimation of the fraction of white dwarfs polluted by the wide binary mechanism, we leave this constraint out of our estimation. 4.2 The planetary system If fWB ∼ 20% of wide binaries have orbital parameters such that a close approach occurs during the primary's white dwarf evolution, the next question regards what fraction, fP S, of these white dwarfs have planetary systems that can be scattered in such a manner as to produce white dwarf pollution. Our simulations (§3) necessarily consider a very specific architecture for the planetary system. Scattering, however, would occur in a wide range of planetary system ar- chitectures, including Oort cloud-like structures, rather than Kuiper-like belts. Clearly, the architecture of the planetary system has a great influence on the efficiency at which mate- rial is scattered inwards (e.g. Bonsor et al. 2012), however, for the purposes of this estimation we consider it sufficient to assume that all planetary systems where sufficient ma- terial survives have the potential to produce white dwarf pollution via the wide binary mechanism. An advantage of this mechanism is its lack of dependence on the exact structure of the planetary system. Outer planetary systems should survive the star's evolution to the white dwarf phase, although planetesimal belts may become collisionally de- pleted (Bonsor & Wyatt 2010), and dynamical instabilities may clear some systems of material (e.g. Raymond et al. 2009). Given the prevalence of planetary systems in the So- lar neighbourhood (e.g. Borucki et al. 2011; Mullally et al. 2015) and that most planets or planetesimal belts cannot be detected, we consider a reasonable estimate to the frac- tion of wide binaries, where the white dwarf primary has a planetary system suitable for producing pollution to be fP S ∼ 50%. We note that this estimate is vague, and is ac- tually something that observations of polluted white dwarfs may help to better constrain. 4.3 Conclusion We can estimate the fraction of white dwarfs with wide bi- nary companions likely to be polluted by this mechanism, at some point during their lives as white dwarfs, as fpoll ∼ fWB × fP S ∼ 20% × 50% ∼ 10% (5) with large uncertainties on our estimations of all of the pa- rameters. At any given instance in time pollution would be detectable in only a fraction, fT , of these systems. This frac- tion will depend on the timescale on which pollution is de- tectable following a close approach of the binary star. Many factors, of which are understanding is limited, need to be considered to produce a good estimate for this timescale, including the dynamical timescales in the planetary system, lifetime of any accretion disc formed, sinking timescale of metal pollutants, etc . We anticipate that the dynamical timescales in the planetary system will dominate, being sev- eral orders of magnitude longer than the other timescales MNRAS 000, 1 -- 12 (2015) A wide binary trigger for white dwarf pollution 9 involved here. The dynamical timescales on which the outer planetary system continues to feed the white dwarf pollution may vary significantly depending on the exact architecture of individual planetary systems. Rather than attempt to ap- proximate these, we, therefore, conclude that a maximum of a few percent of any sample of white dwarfs with wide binary companions would be observed to have metal pollution pro- duced via the wide binary mechanism presented in this work. The wide binary fraction of white dwarfs is not well known, but if we consider that at least (and probably more than) 10% of white dwarfs should have wide binary companions, based on the 10% of solar-type stars observed with wide binaries (Raghavan et al. 2010; L´epine & Bongiorno 2007) (see discussion in the Introduction), then the wide binary pollution mechanism contributes to pollution in a maximum of a percent of any sample of white dwarfs. 5 OBSERVATIONAL TESTS The mechanism presented here works only for a subset of the potential orbital parameters of the binary and plane- tary system. As discussed in §4, this fraction is likely to be of the order of a few percent of any random sample of white dwarfs with wide binary companions. Given that many observations find that more than 25%, and maybe up to 50%, of white dwarfs are polluted (Zuckerman et al. 2003, 2010; Koester et al. 2014), this wide binary mechanism is only ever going to make a minor contribution, and other mechanisms must be important in polluting white dwarfs. In fact, of roughly one hundred 'pre-Sloan' polluted white dwarfs, discovered by their proper motions, such that any companions is likely to have been found, only 5 have known common proper motion companions (Farihi et al, private communication). It is not clear whether this wide binary fraction is different from the general population, where, for example ∼ 10% of solar-type stars are found to have wide separation (> 1, 000AU) companions (Raghavan et al. 2010; L´epine & Bongiorno 2007). Zuckerman (2014) consider a sample of 17 white dwarfs with wide (> 2, 500AU) companions, all observed with Keck/VLT to search for Ca II, and find that 5/17 (30%) are polluted. With such small number statistics, and significant bias in sample selection, it is difficult to make a good com- parison with observations of apparently single white dwarfs, where, for example 25% of DA white dwarfs exhibit Ca pol- lution (Zuckerman et al. 2003). It remains plausible that the wide binary mechanism could contribute to the pollution in these white dwarfs on the percent level, but without larger, well defined, samples, it is impossible to make a detailed comparison. The key evidence for a time-independent mechanism for pollution, such as suggested here, comes from the lack of any decrease in the fraction of polluted systems with age, and the observations of pollution in old (Gyr) white dwarfs, which, for example dynamical instabilities following stel- lar mass loss struggle to explain (e.g. Bonsor et al. 2011; Debes et al. 2012; Frewen & Hansen 2014). There are cur- rently insufficient white dwarfs with known binary compan- ions to assess any dependence in the fraction of polluted sys- tems with white dwarf cooling age. However, WD 1009-184, with its companion at 6,870AU and an effective tempera- MNRAS 000, 1 -- 12 (2015) ture of 9,940K (equivalent to a cooling age of ∼700Myr) (Sion et al. 2009; Zuckerman 2014) stands out as a can- didate example of where the companion could potentially be responsible for the white dwarf pollution. Future obser- vational searches for companions to cool white dwarfs will provide critical evidence regarding the contribution of wide binaries to pollution. Gaia will play a crucial role in discov- ering companions to many nearby white dwarfs. 6 DISCUSSION In this work we outline a proof of concept that illustrates the manner by which a wide binary companion could lead to pollution in white dwarfs. In §2 we illustrate that dur- ing the evolution of the binary's orbit as it interacts with the Galactic tides, periods of increased eccentricity occur. These periods are most likely to occur during the white dwarf phase, due to the increase in separation of the bi- nary orbit following stellar mass loss. We claim that a wide binary companion whose eccentricity increases for the first time during the primary star's white dwarf evolution has the potential to perturb any planetary system orbiting the white dwarf. In §3, we illustrated how planetesimals can be scattered onto star-grazing orbits, a pre-requisite for pollu- tion of the white dwarf's atmosphere. This mechanism is independent of white dwarf cooling age, and could, there- fore, produce pollution around old (1-5Gyr) white dwarfs that other mechanisms struggle to explain. We estimate that this mechanism may contribute to the total fraction of pol- luted white dwarfs on the level of a few percent. There are no current observational samples that are sufficiently large to assess whether the population of polluted white dwarfs around single stars is different to those with companions. We consider that this mechanism is robust and will occur for white dwarfs with planetary systems and com- panions with the 'correct' parameters. The weakest part of this argument regards the link between planetesimals scat- tered inwards and the pollution of white dwarf atmospheres, which was not investigated in this work. The tidal disrup- tion of planetesimals and the acccretion of close-in dusty ma- terial onto white dwarfs have been previously investigated (e.g. Rafikov 2011a,b; Debes et al. 2012; Metzger et al. 2012; Veras et al. 2014c, 2015c), although some gaps still remain in our understanding. Our discussion of whether a detectable level of pollution is produced is vague (see §4), and it remains possible that no detectable pollution is ever produced by planetesimal scattering. However, our simula- tions do show a significant increase in the number of plan- etesimals scattered onto star-grazing orbits with a binary companion on an eccentric orbit, compared to without a binary companion (see Fig. 5), and given the low accretion masses required to produce detectable pollution3, it remains plausible that the mechanism presented works for some re- gion of the parameter space. In fact, we anticipate that this mechanism has a weak dependence on the exact architecture of the planetary system, and a much stronger dependence on the orbital parameters of the binary. 3 Even one of the most highly polluted white dwarfs only needs a Ceres mass (10−4M⊕) (Dufour et al. 2010), or see Girven et al. (2012) 10 A. Bonsor et al. The purpose of this work was to outline a potential mechanism for the pollution of white dwarfs. We do not consider our estimation of the fraction of white dwarfs with wide binary companions polluted by this mechanism to be robust, and it is even trickier to determine the fraction of an observational sample that would be caught whilst displaying detectable pollution. Our estimation does, however, indicate that the wide binary mechanism is not the dominant mecha- nism for producing white dwarf pollution, even at late times, and that statistics in current observational samples are in- sufficient to determine whether or not this mechanism is pro- ducing pollution. Gaia and future observational searches for companions to nearby white dwarfs, should enable a clearer determination of its contribution. We have ignored the potential for exchange interactions, or changes to the binary's orbit, which could have a big effect on the evolution of individual systems, but are not going to significantly affect the total population of white dwarfs with wide binary companions. We have also ignored any poten- tial evolution of the companion star. If it loses a significant fraction of its mass, this could increase the fraction of bi- nary orbits that meet our criteria for producing pollution, as in essence the system has double the chance of entering the correct regime. For example, one could imagine that the expansion in the orbit of the binary is insufficient to induce large perturbations from the Galactic tide as the primary loses mass to become a white dwarf, but becomes sufficiently wide once the secondary evolves to become a white dwarf. Perturbations could occur to planetary systems orbiting ei- ther the primary or the secondary. Our mechanism depends critically on the survival of a planetary system, containing sufficient material, to late times in the white dwarf phase. Although, theoreti- cally there are good arguments for the survival of outer planetary systems during post-main sequence evolution (Duncan & Lissauer 1998; Veras & Wyatt 2012), even if col- lisionally depleted (Bonsor & Wyatt 2010), there is a yet little direct observational evidence for the presence of plan- etary systems around white dwarfs (Mullally et al. 2008; Burleigh et al. 2008; Xu et al. 2015), although a few po- tential exceptions exist (Luhman et al. 2011; Marsh et al. 2014). Planets have been detected around giant stars (e.g. Johnson et al. 2007, 2008; Sato et al. 2008) and horizontal branch stars (Silvotti et al. 2007). Dusty, debris-like discs have been observed at the centre of several planetary nebu- lae (Su et al. 2007; Chu et al. 2011), very early in the white dwarf phase. If future observations show that this mecha- nism is a significant contributor to white dwarf pollution, it provides further evidence for the survival of outer planetary systems to the white dwarf phase. The most important contribution of the wide binary mechanism presented here is to the pollution of old (Gyr) white dwarfs, that are hard to explain by other mecha- nisms. The mechanism is independent of white dwarf cool- ing age, except for any variation in the planetary system itself. On the main-sequence material can be depleted, both by collisions and dynamics, with a steep time dependence. By the white dwarf phase, collisional timescales are long (Bonsor & Wyatt 2010), and the system is much more likely to have reached a dynamically stable state, such that any such evolution is likely to only occur at low levels. 7 CONCLUSIONS In this work we present a mechanism to explain the metal pollution observed in many white dwarfs, especially at late times. The orbits of wide binaries vary periodically with the Galactic tides. The influence of the Galactic tide in- creases following stellar mass loss. Thus, a wide binary may evolve for Gyrs at large separations, and yet, have a close approach for the first time during the primary's white dwarf phase. Such a close approach could perturb any planetary system orbiting the primary, potentially scattering planetes- imals onto star-grazing orbits. We present simulations that illustrate the change in or- bital parameters of the binary (§2) and the scattering of planetesimals onto star-grazing orbits (§3). These demon- strations should be considered a proof of concept, rather than a detailed analysis. We estimate that this mechanism could contribute to pollution in up to a few percent of any sample of white dwarfs with wide binary companions, and up to a percent of all white dwarfs (§4), independent of the age of the white dwarf. Current observational samples do not have sufficient statistics to indicate whether or not this mechanism provides an important contribution to pollution in white dwarfs. The calcium-polluted, WD 1009-184, with its companion at 6,870AU and an effective temperature of 9,940K (equivalent to a cooling age of ∼700Myr) (Sion et al. 2009; Zuckerman 2014) stands out as an example system where this mechanism could be acting. Wide binary pollution is unlikely to be the sole con- tributor to the population of metal-polluted white dwarfs. Instead, we envisage that it acts alongside other mecha- nisms, adding a time-independent pollution rate. Future ob- servational searches for wide binary companions to nearby white dwarfs (e.g. with Gaia) and an increased sample of metal-encriched old, white dwarfs are required to determine whether or not wide binary companions play an important role in white dwarf pollution. 8 ACKNOWLEDGEMENTS We thank Silvia Catal´an and Jay Farihi for useful dis- cussions. We thank the referee Dr. Cristobal Petrovich for his useful comments that improved the quality of the manuscript. AB was supported by ANR-2010 BLAN-0505- 01(EXOZODI), NERC Grant NE/K004778/1, and ERC grant number 279973 whilst conducting this research. DV benefitted from support by the European Union through ERC Grant Number 320964. REFERENCES Aannestad P. A., Kenyon S. J., Hammond G. L., Sion E. M., 1993, AJ, 105, 1033 Adams F. C., Bloch A. M., 2013, ApJ, 777, L30 Alcock C., Fristrom C. C., Siegelman R., 1986, ApJ, 302, 462 Barstow M. A., Barstow J. K., Casewell S. L., Holberg J. B., Hubeny I., 2014, MNRAS, 440, 1607 Bear E., Soker N., 2013, New Astronomy, 19, 56 Bochkarev K. V., Rafikov R. R., 2011, ApJ, 741, 36 Bonsor A., Augereau J.-C., Th´ebault P., 2012, A&A, 548, A104 Bonsor A., Mustill A. J., Wyatt M. C., 2011, MNRAS, 414, 930 Bonsor A., Wyatt M., 2010, MNRAS, 409, 1631 MNRAS 000, 1 -- 12 (2015) A wide binary trigger for white dwarf pollution 11 Bonsor A., Wyatt M. C., 2012, MNRAS, 420, 2990 Borucki W. J., Koch D. G., Basri G., Batalha N., Brown T. M., Bryson S. T., Caldwell D., Christensen-Dalsgaard J., Cochran W. D., DeVore E., Dunham E. W., Gautier III T. N., Geary J. C., Gilliland R., Gould A., Howell S. B., Jenkins J. M., Latham D. W., Lissauer J. J., Marcy G. W., Rowe J., Sasselov D., Boss A., Charbonneau D., Ciardi D., Doyle L., Dupree A. K., Ford E. B., Fortney J., Holman M. J., Seager S., Stef- fen J. H., Tarter J., Welsh W. F., Allen C., Buchhave L. A., Christiansen J. L., Clarke B. D., Das S., D´esert J.-M., Endl M., Fabrycky D., Fressin F., Haas M., Horch E., Howard A., Isaacson H., Kjeldsen H., Kolodziejczak J., Kulesa C., Li J., Lucas P. W., Machalek P., McCarthy D., MacQueen P., Mei- bom S., Miquel T., Prsa A., Quinn S. N., Quintana E. V., Ragozzine D., Sherry W., Shporer A., Tenenbaum P., Torres G., Twicken J. D., Van Cleve J., Walkowicz L., Witteborn F. C., Still M., 2011, ApJ, 736, 19 Brasser R., 2001, MNRAS, 324, 1109 Breiter S., Dybczynski P. A., Elipe A., 1996, A&A, 315, 618 Breiter S., Ratajczak R., 2005, MNRAS, 364, 1222 Burleigh M. R., Clarke F. J., Hogan E., Brinkworth C. S., Berg- eron P., Dufour P., Dobbie P. D., Levan A. J., Hodgkin S. T., Hoard D. W., Wachter S., 2008, MNRAS, 386, L5 Cassan A., Kubas D., Beaulieu J.-P., Dominik M., Horne K., Greenhill J., Wambsganss J., Menzies J., Williams A., Jør- gensen U. G., Udalski A., Bennett D. P., Albrow M. D., Batista V., Brillant S., Caldwell J. A. R., Cole A., Coutures C., Cook K. H., Dieters S., Prester D. D., Donatowicz J., Fouqu´e P., Hill K., Kains N., Kane S., Marquette J.-B., Martin R., Pollard K. R., Sahu K. C., Vinter C., Warren D., Watson B., Zub M., Sumi T., Szyma´nski M. K., Kubiak M., Poleski R., Soszynski I., Ulaczyk K., Pietrzy´nski G., Wyrzykowski L., 2012, Nature, 481, 167 Chambers J. E., 1999, MNRAS, 304, 793 Chu Y., Su K. Y. L., Bilikova J., Gruendl R. A., De Marco O., Guerrero M. A., Updike A. C., Volk K., Rauch T., 2011, ApJ Debes J. H., Sigurdsson S., 2002, ApJ, 572, 556 Debes J. H., Walsh K. J., Stark C., 2012, ApJ, 747, 148 Dufour P., Kilic M., Fontaine G., Bergeron P., Lachapelle F., Kleinman S. J., Leggett S. K., 2010, ApJ, 719, 803 Duncan M. J., Lissauer J. J., 1998, Icarus, 134, 303 Eggleton P., Kiseleva L., 1995, ApJ, 455, 640 Eiroa C., Marshall J. P., Mora A., Montesinos B., Absil O., Augereau J. C., Bayo A., Bryden G., Danchi W., del Burgo C., Ertel S., Fridlund M., Heras A. M., Krivov A. V., Launhardt R., Liseau R., Lohne T., Maldonado J., Pilbratt G. L., Roberge A., Rodmann J., Sanz-Forcada J., Solano E., Stapelfeldt K., Th´ebault P., Wolf S., Ardila D., Ar´evalo M., Beichmann C., Faramaz V., Gonz´alez-Garc´ıa B. M., Guti´errez R., Lebreton J., Mart´ınez-Arn´aiz R., Meeus G., Montes D., Olofsson G., Su K. Y. L., White G. J., Barrado D., Fukagawa M., Grun E., Kamp I., Lorente R., Morbidelli A., Muller S., Mutschke H., Nakagawa T., Ribas I., Walker H., 2013, A&A, 555, A11 Farihi J., Barstow M. A., Redfield S., Dufour P., Hambly N. C., 2010, MNRAS, 404, 2123 Farihi J., Becklin E. E., Zuckerman B., 2005, ApJS, 161, 394 Farihi J., Dufour P., Napiwotzki R., Koester D., 2011, MNRAS, Gladman B., Kavelaars J. J., Petit J.-M., Morbidelli A., Holman M. J., Loredo T., 2001, AJ, 122, 1051 Graham J. R., Matthews K., Neugebauer G., Soifer B. T., 1990, ApJ, 357, 216 Hartmann S., Nagel T., Rauch T., Werner K., 2011, A&A, 530, A7 Heisler J., Tremaine S., 1986, Icarus, 65, 13 Holberg J. B., Sion E. M., Oswalt T., McCook G. P., Foran S., Subasavage J. P., 2008, AJ, 135, 1225 Hollands M. A., Gansicke B. T., Koester D., 2015, MNRAS, 450, 681 Holman M. J., Wiegert P. A., 1999, AJ, 117, 621 Hurley J. R., Pols O. R., Tout C. A., 2000, MNRAS, 315, 543 Johnson J. A., Fischer D. A., Marcy G. W., Wright J. T., Driscoll P., Butler R. P., Hekker S., Reffert S., Vogt S. S., 2007, ApJ, 665, 785 Johnson J. A., Marcy G. W., Fischer D. A., Wright J. T., Reffert S., Kregenow J. M., Williams P. K. G., Peek K. M. G., 2008, ApJ, 675, 784 Jura M., 2003, ApJ, 584, L91 -- , 2004, ApJ, 603, 729 -- , 2006, ApJ, 653, 613 -- , 2008, AJ, 135, 1785 Jura M., Farihi J., Zuckerman B., Becklin E. E., 2007, AJ, 133, 1927 Kaib N. A., Raymond S. N., Duncan M., 2013, Nature, 493, 381 Kilic M., Redfield S., 2007, ApJ, 660, 641 Kilic M., von Hippel T., Leggett S. K., Winget D. E., 2006, ApJ, 646, 474 Klein B., Jura M., Koester D., Zuckerman B., 2011, ApJ, 741, 64 Klein B., Jura M., Koester D., Zuckerman B., Melis C., 2010, ApJ, 709, 950 Koester D., Gansicke B. T., Farihi J., 2014, A&A, 566, A34 Koester D., Girven J., Gansicke B. T., Dufour P., 2011, A&A, 530, A114+ Koester D., Provencal J., Shipman H. L., 1997, A&A, 320, L57 Koester D., Wilken D., 2006, A&A, 453, 1051 Kouwenhoven M. B. N., Goodwin S. P., Parker R. J., Davies M. B., Malmberg D., Kroupa P., 2010, MNRAS, 404, 1835 Kudritzki R. P., Reimers D., 1978, A&A, 70, 227 L´epine S., Bongiorno B., 2007, AJ, 133, 889 Luhman K. L., Burgasser A. J., Bochanski J. J., 2011, ApJ, 730, L9 Makarov V. V., Zacharias N., Hennessy G. S., 2008, ApJ, 687, 566 Marsh T. R., Parsons S. G., Bours M. C. P., Littlefair S. P., Copperwheat C. M., Dhillon V. S., Breedt E., Caceres C., Schreiber M. R., 2014, MNRAS, 437, 475 Marzari F., Weidenschilling S. J., Barbieri M., Granata V., 2005, ApJ, 618, 502 Matese J. J., Whitman P. G., 1989, Icarus, 82, 389 -- , 1992, Celestial Mechanics and Dynamical Astronomy, 54, 13 Matese J. J., Whitman P. G., Innanen K. A., Valtonen M. J., 1995, Icarus, 116, 255 Mayor M., Marmier M., Lovis C., Udry S., S´egransan D., Pepe F., Benz W., Bertaux J. ., Bouchy F., Dumusque X., Lo Curto G., Mordasini C., Queloz D., Santos N. C., 2011, ArXiv e-prints Melis C., Jura M., Albert L., Klein B., Zuckerman B., 2010, ApJ, 413, 2559 722, 1078 Frewen S. F. N., Hansen B. M. S., 2014, MNRAS, 439, 2442 Gansicke B. T., Koester D., Farihi J., Girven J., Parsons S. G., Breedt E., 2012, MNRAS, 424, 333 Gansicke B. T., Marsh T. R., Southworth J., 2007, MNRAS, 380, L35 Gansicke B. T., Marsh T. R., Southworth J., Rebassa-Mansergas A., 2006, Science, 314, 1908 Girven J., Brinkworth C. S., Farihi J., Gansicke B. T., Hoard D. W., Marsh T. R., Koester D., 2012, ApJ, 749, 154 Metzger B. D., Rafikov R. R., Bochkarev K. V., 2012, MNRAS, 423, 505 Mullally F., Coughlin J. L., Thompson S. E., Rowe J., Burke C., Latham D. W., Batalha N. M., Bryson S. T., Christiansen J., Henze C. E., Ofir A., Quarles B., Shporer A., Van Eylen V., Van Laerhoven C., Shah Y., Wolfgang A., Chaplin W. J., Xie J.-W., Akeson R., Argabright V., Bachtell E., Barclay T., Borucki W. J., Caldwell D. A., Campbell J. R., Catanzarite J. H., Cochran W. D., Duren R. M., Fleming S. W., Fraquelli MNRAS 000, 1 -- 12 (2015) 12 A. Bonsor et al. D., Girouard F. R., Haas M. R., He lminiak K. G., Howell S. B., Huber D., Larson K., Gautier III T. N., Jenkins J. M., Li J., Lissauer J. J., McArthur S., Miller C., Morris R. L., Patil- Sabale A., Plavchan P., Putnam D., Quintana E. V., Ramirez S., Silva Aguirre V., Seader S., Smith J. C., Steffen J. H., Stewart C., Stober J., Still M., Tenenbaum P., Troeltzsch J., Twicken J. D., Zamudio K. A., 2015, ApJS, 217, 31 Mullally F., Winget D. E., De Gennaro S., Jeffery E., Thompson S. E., Chandler D., Kepler S. O., 2008, ApJ, 676, 573 Mustill A. J., Veras D., Villaver E., 2014, MNRAS, 437, 1404 Mustill A. J., Villaver E., 2012, ApJ, 761, 121 Parravano A., McKee C. F., Hollenbach D. J., 2011, ApJ, 726, 27 Petrovich C., 2015, ApJ, 808, 120 Rafikov R. R., 2011a, ApJ, 732, L3+ -- , 2011b, MNRAS, L287+ Rafikov R. R., Garmilla J. A., 2012, ApJ, 760, 123 Raghavan D., McAlister H. A., Henry T. J., Latham D. W., Marcy G. W., Mason B. D., Gies D. R., White R. J., ten Brummelaar T. A., 2010, ApJS, 190, 1 Raymond S. N., Armitage P. J., Gorelick N., 2009, ApJ, 699, L88 -- , 2010, ApJ, 711, 772 Sato B., Toyota E., Omiya M., Izumiura H., Kambe E., Masuda S., Takeda Y., Itoh Y., Ando H., Yoshida M., Kokubo E., Ida S., 2008, PASJ, 60, 1317 Silvotti R., Schuh S., Janulis R., Solheim J.-E., Bernabei S., Østensen R., Oswalt T. D., Bruni I., Gualandi R., Bonanno A., Vauclair G., Reed M., Chen C.-W., Leibowitz E., Paparo M., Baran A., Charpinet S., Dolez N., Kawaler S., Kurtz D., Moskalik P., Riddle R., Zola S., 2007, Nature, 449, 189 Sion E. M., Holberg J. B., Oswalt T. D., McCook G. P., Wasatonic R., 2009, AJ, 138, 1681 Stone N., Metzger B. D., Loeb A., 2015, MNRAS, 448, 188 Su K. Y. L., Chu Y.-H., Rieke G. H., Huggins P. J., Gruendl R., Napiwotzki R., Rauch T., Latter W. B., Volk K., 2007, ApJ, 657, L41 van Maanen A., 1920, Contributions from the Mount Wilson Ob- servatory / Carnegie Institution of Washington, 182, 1 Vassiliadis E., Wood P. R., 1993, ApJ, 413, 641 Veras D., Eggl S., Gaensicke B. T., 2015a, ArXiv e-prints Veras D., Eggl S., Gansicke B. T., 2015b, MNRAS, 451, 2814 Veras D., Evans N. W., 2013a, MNRAS, 430, 403 -- , 2013b, Celestial Mechanics and Dynamical Astronomy, 115, 123 Veras D., Evans N. W., Wyatt M. C., Tout C. A., 2013a, MNRAS -- , 2014a, MNRAS, 437, 1127 Veras D., Gansicke B. T., 2015, MNRAS, 447, 1049 Veras D., Hadjidemetriou J. D., Tout C. A., 2013b, MNRAS, 435, 2416 Veras D., Jacobson S. A., Gansicke B. T., 2014b, MNRAS, 445, 2794 Veras D., Leinhardt Z. M., Bonsor A., Gansicke B. T., 2014c, MNRAS, 445, 2244 Veras D., Leinhardt Z. M., Eggl S., Gansicke B. T., 2015c, MN- RAS, 451, 3453 Veras D., Mustill A. J., Bonsor A., Wyatt M. C., 2013c, MNRAS, 431, 1686 Veras D., Shannon A., Gansicke B. T., 2014d, MNRAS, 445, 4175 Veras D., Tout C. A., 2012, MNRAS, 422, 1648 Veras D., Wyatt M. C., 2012, MNRAS, 421, 2969 Veras D., Wyatt M. C., Mustill A. J., Bonsor A., Eldridge J. J., 2011, MNRAS, 417, 2104 Villaver E., Livio M., 2007, ApJ, 661, 1192 -- , 2009, ApJ, 705, L81 Villaver E., Livio M., Mustill A. J., Siess L., 2014, ApJ, 794, 3 von Hippel T., Kuchner M. J., Kilic M., Mullally F., Reach W. T., 2007, ApJ, 662, 544 Voyatzis G., Hadjidemetriou J. D., Veras D., Varvoglis H., 2013, MNRAS, 430, 3383 Wyatt M. C., 2008, ARA&A, 46, 339 Wyatt M. C., Farihi J., Pringle J. E., Bonsor A., 2014, MNRAS, 439, 3371 Wyatt M. C., Smith R., Su K. Y. L., Rieke G. H., Greaves J. S., Beichman C. A., Bryden G., 2007, ApJ, 663, 365 Xu S., Ertel S., Wahhaj Z., Milli J., Scicluna P., Bertrang G. H.- M., 2015, ArXiv e-prints Zuckerman B., 2014, ApJ, 791, L27 Zuckerman B., Koester D., Reid I. N., Hunsch M., 2003, ApJ, 596, 477 Zuckerman B., Melis C., Klein B., Koester D., Jura M., 2010, ApJ, 722, 725 This paper has been typeset from a TEX/LATEX file prepared by the author. MNRAS 000, 1 -- 12 (2015)
1610.07669	1	1610	2016-10-24T22:22:19	Saturn Variable Thermosphere	[ "astro-ph.EP" ]	Our knowledge of Saturns neutral thermosphere is far superior to that of the other giant planets due to Cassini Ultraviolet Imaging Spectrograph (UVIS) observations of 15 solar occultations and 26 stellar occultations analyzed to date. These measurements yield H2 as the dominant species with an upper limit on the H mole fraction of 5 %. Inferred temperatures near the lower boundary are ~ 150 K, rising to an asymptotic value of ~ 400K at equatorial latitudes and increasing with latitude to polar values in the range of 550-600 K. The latter is consistent with a total estimated auroral power input of ~ 10TW generating Joule and energetic particle heating of ~ 5-6TW that is more than an order of magnitude greater than solar EUV/FUV heating. This auroral heating would be sufficient to solve the energy crisis of Saturns thermospheric heating, if it can be efficiently redistributed to low latitudes. The inferred structure of the thermosphere yields poleward directed pressure gradients on equipotential surfaces consistent with auroral heating and poleward increasing temperatures. A gradient wind balance aloft with these pressure gradients implies westward, retrograde winds ~ 500 m/s or Mach number ~ 0.3 at mid-latitudes. The occultations reveal an expansion of the thermosphere peaking at or slightly after equinox, anti-correlated with solar activity, and apparently driven by lower thermospheric heating of unknown cause. The He mole fraction remains unconstrained as no Cassini UVIS He 58.4 nm airglow measurements have been published.	astro-ph.EP	astro-ph	9 Saturn's Variable Thermosphere Departments of Earth & Planetary Sciences and Physics & Astronomy, Johns Hopkins University, Baltimore, MD 21218, U.S.A. Darrell F. Strobel Tommi Koskinen Lunar and Planetary Laboratory, University of Arizona, Tucson, AZ 85721-0092 Ingo Müller-Wodarg ABSTRACT Blackett Laboratory, Imperial College, London, UK Our knowledge of Saturn's neutral thermosphere is far superior to that of the other giant planets due to Cassini Ultraviolet Imaging Spectrograph (UVIS) observations of 15 solar occultations and 26 stellar occultations analyzed to date. These measurements yield H2 as the dominant species with an upper limit on the H mole fraction of 5%. Inferred temperatures near the lower boundary are ~ 150 K, rising to an asymptotic value of ~ 400 K at equatorial latitudes and increasing with latitude to polar values in the range of 550-600 K. The latter is consistent with a total estimated auroral power input of ~ 10 TW generating Joule and energetic particle heating of ~ 5-6 TW that is more than an order of magnitude greater than solar EUV/FUV heating. This auroral heating would be sufficient to solve the "energy crisis" of Saturn's thermospheric heating, if it can be efficiently redistributed to low latitudes. The inferred structure of the thermosphere yields poleward directed pressure gradients on equipotential surfaces consistent with auroral heating and poleward increasing temperatures. A gradient wind balance aloft with these pressure gradients implies westward, retrograde winds ~ 500 m s-1 or Mach number ~ 0.3 at mid-latitudes. The occultations reveal an expansion of the thermosphere peaking at or slightly after equinox, anti-correlated with solar activity, and apparently driven by lower thermospheric heating of unknown cause. The He mole fraction remains unconstrained as no Cassini UVIS He 58.4 nm airglow measurements have been published. 1 9.1. INTRODUCTION The thermosphere is typically located above the Traditionally the thermosphere is defined as the region characterized by a steep temperature gradient generated at its base by intense heating from absorption of short wavelength solar ultraviolet radiation (< 170 nm) in the dissociation and/or ionization of homonuclear molecules, e. g. H2, N2) and the downward transport of thermal energy by heat conduction due to the absence of infrared active molecules. The base of the thermosphere is the mesopause defined as the level where the temperature reaches a minimum and the temperature gradient vanishes due to infrared active molecules there radiating away the solar UV heating above. As with all solar system giant planets, however, solar UV heating is a minor energy source for Saturn's thermosphere. homopause, where atmospheric species undergo gravitational diffusive separation and assume scale heights in accordance with their atomic and molecular masses rather than the atmospheric mean. Methane density profiles as determined from occultations provide relatively precise locations of the CH4 homopause in contrast to the much larger uncertainty in using inferred temperature profiles to locate the base of thermosphere. An authoritative chapter on Saturn's thermosphere is challenging for one cannot replicate the continuous vertical structure achieved by the Galileo probe at Jupiter or the Huygens probe at Titan. Over some altitude regions, temperature, composition, and density profiles are retrieved as a function of pressure, whereas for others data are acquired as a function of radial distance or, in particular near the mesopause, are only marginally available. A definitive He/H2 mixing ratio, which would enable construction of continuous atmospheric profiles at individual locations, is also lacking at Saturn, for which we have the value of 0.034 inferred from Voyager 1 infrared and radio occultation measurements, universally regarded as incorrect and at best an extreme lower bound. Instead, Conrath and Gautier (2000) inferred a ratio of ~ 0.13 from reanalysis of Voyager IRIS data. The chapter is organized to establish what we know or at least what we think we know from density data (Section 2), temperature data (Section 3), airglow data (Section 4), and composition data (Section 5), followed by an examination of the density, pressure, and temperature structure derived from the available data and the inference 2 9.2. REVIEW OF RELEVANT DENSITY DATA of net heating rates from radial temperature profiles (Section 6). In Section 7, UVIS stellar occultation data are used to infer the location of the CH4 homopause. Energetics and potential sources are discussed and reviewed in Section 8, which sets the stage for Section 9 on global general circulations models for the thermosphere and a discussion of the observed state of Saturn's atmosphere as compared to the models. Auroral physics and ionospheric chemistry and physics, often considered part of thermospheric physics, are dealt with in Chapters 7 and 8. We treat them here only as necessary for our purposes. The neutral thermosphere is composed mainly of H2 and He with trace amounts of H and water group molecules (O, OH, H2O). Atomic H is released by photochemical reactions below the CH4 homopause (e.g., Moses et al. 2005) and its mixing ratio increases with altitude in the thermosphere due to molecular diffusion. Water group molecules are delivered to the upper atmosphere from Saturn's magnetosphere and rings (e.g., Connerney and Waite 1984; Feuchtgruber et al. 1997; Cassidy et al. 2010; O'Donoghue et al. 2013). Their net mixing ratio is expected to decrease with increasing pressure in the stratosphere due to condensation in the lower atmosphere and be roughly constant with altitude in the upper atmosphere (e.g., Moses of CH4 and other hydrocarbons, on the other hand, decrease rapidly with altitude above the homopause due to molecular diffusion and photolysis, and should be negligible in the thermosphere. The mixing ratio of He also decreases with altitude in the thermosphere, but it does so less rapidly than the mixing ratios of the hydrocarbons. Observational constraints on H2 and H can be obtained from extreme (EUV) and far-ultraviolet (FUV) occultations of bright stars and the Sun as well as airglow and auroral emissions (e.g., Broadfoot et al. 1981; Sandel et al. 1982; Smith et al. 1983; Shemansky and Ajello 1983; Yelle 1988; Gérard et al. 1995, 2009, 2013; Shemansky et al. 2009; Gustin et al. 2010). Constraints on He can be obtained from EUV airglow measurements (Sandel et al. 1982; Parkinson et al. 1998) while CH4 near the homopause can be probed by FUV occultations (Smith et al. 1983; Shemansky and Liu 2012; Koskinen et al. 2015; Vervack and Moses 2015) as well as limb emissions and occultations in the near-IR CH4 bands at 3.3 µm (Baines et al. 2005; Kim et al. 2012). et al. 2000; Müller-Wodarg et al. 2012). The mixing ratios 3 Unfortunately there are no direct observations of the water group molecules reported at present even though they may play a substantial role in controlling the electron densities in the ionosphere (e.g., Moore et al. 2010; Müller-Wodarg et al. 2012, cf. Chapter 8). In principle, stellar and solar occultations are the most reliable source of information about the density and temperature structure in the thermosphere. This is because the interpretation of airglow and auroral emissions depends on free parameters in complex non-LTE radiative transfer models (e.g., Liu and Dalgarno 1996; Hallett et al. 2005; Gustin et al. 2010; García-Comas et al. 2011; Adriani et al. 2011) and instrument calibration that is not always well characterized. The analysis of the occultations, on the other hand, is much simpler. The data do not need to be absolutely calibrated and the retrieved densities are simply based on the line of sight optical depths of the absorbers that can be obtained from the raw data with relative ease. The occultations also probe the atmosphere at a wide range of altitudes and latitudes. In this context, it may appear surprising that there are significant disagreements between different analyses of the Voyager/UVS (Broadfoot et al. 1981; Festou and Atreya 1982; Vervack and Moses 2015) and Cassini/UVIS occultations (Shemansky and Liu 2012; Koskinen et al. 2013, 2015). The recent re-analysis of the Voyager/UVS data (Vervack and Moses 2015) and new results based on 41 solar and stellar occultations from Cassini/UVIS (Koskinen et al. 2013, 2015) have resolved this dispute. Voyager 1 (V1) observed two solar occultations and one stellar occultation of ι Herculis while Voyager 2 (V2) observed one solar occultation and two stellar occultations of δ Scorpii. Density and temperature profiles retrieved from all six of the Voyager/UVS occultations were published only recently by Vervack and Moses (2015). Previous results were limited to exospheric temperatures and simple forward models of the atmosphere based on only three of the occultations (Broadfoot et al. 1981; Sandel et al. 1982, Festou and Atreya 1982; Smith et al. 1983; Shemansky and Liu 2012). There are significant differences in the retrieved exospheric temperatures between these results that we discuss further in Section 9.3.2. Here we concentrate only on the newly published density profiles of H2 and H (Vervack and Moses 2015). 9.2.1 Voyager/UVS occultation data 4 In order to better understand the results and their limitations, it is useful to develop a basic understanding of the instruments and the occultation data. The processed data consist of transmission spectra as a function of radial distance and latitude from the center of Saturn. Together these spectra make up light curves i.e., transmission as a function of radial distance in each wavelength band. The wavelength range of the Voyager/UVS instruments is 51 – 170 nm with a spectral resolution of 1.8 – 3 nm for point sources. The signal-to-noise (S/N) of the occultation data, however, was often so poor that in reality the data had to be binned to a much lower resolution. The observations also suffered from significant spacecraft pointing drifts and slews, interference from the rings, changes in detector gain and pixel-to-pixel sensitivity variations. All these effects significantly complicated the retrieval of density profiles from the Voyager/UVS occultations (Vervack and Moses 2015). The density profiles of H2 and H are retrieved by analyzing transmission at wavelengths of 51 – 115 nm. Hydrogen in the interstellar medium (ISM) absorbs all starlight at wavelengths shorter than 91 nm and thus the shorter wavelengths in this range are only available in the solar occultations. This is an important point because absorption at 51 – 80.4 nm is dominated by the ionization continuum of H2. The absorption cross section in the ionization continuum varies smoothly with wavelength and, even more importantly, does not depend on temperature (Samson and Haddad 1994). This is in contrast to the electronic Lyman and Werner bands of H2 (hereafter, the LW bands) at longer wavelengths where the cross section is much more complicated and depends on temperature. Thus solar occultations provide the most robust measurements of the density profiles and exospheric temperatures. The results, however, are limited to relatively high altitudes near the exobase above the 0.1 nbar level. An analysis of the LW bands is still required to probe the density and temperature profiles in the lower thermosphere and near the homopause (~ 0.01-0.1 µbar). The exobase is formally defined as the level in the atmosphere where the mean free path of the major species (H2) is equal to its scale height divided by √2 and regarded as the top of the atmosphere (Strobel, 2002). If the collisionless, neutral gas above the exobase is gravitationally bound, it is referred to as the exosphere. Figure 9.1 shows the H2 and H profiles based on the V1 and V2 data from Vervack and Moses (2015), together with 5 results b from the previous analysis of the V2 occultations by Smith et al. (1983) and ground-based observations of the occultation of the star 28 Sgr (Hubbard et al. 1997). Vervack and Moses (2015) retrieved the lowest pressure H2 densities (red line) from the ionization continuum in the V2 solar occultation while the rest of the H2 profiles below the 0.1 nbar level are based on the LW bands. They used the same canonical T-P profile to calculate the absorption cross sections for all occultations and did not account for the change in temperature along the line of sight in this calculation. Given the relatively large uncertainties in the Voyager/UVS light curves, however, these assumptions are unlikely to result in significant systematic errors in the retrieved density profiles. As illustrated by Figure 9.1, the retrieval of the H2 profiles is typically limited to altitudes above the 10 nbar level – a level where transmission in the LW bands vanishes. Owing to the sophisticated treatment of the instrument effects and updated retrieval techniques used by Vervack and Moses (2015), the density profiles based on the Voyager/UVS data are now remarkably consistent. They also agree well with the density profile retrieved by Hubbard et al. (1997) at all relevant altitudes. The abundance of H in Figure 9.1 is retrieved from the Lyman continuum at 80.4 – 91 nm in the V1 and V2 solar occultations. The analysis of the higher resolution solar occultations from Cassini/UVIS (see Section 2.2 below) suggests that separating absorption by H and H2 in this region may be more challenging than previously thought. Based on their model of the Cassini/UVIS data, Koskinen et al. (2013) found that absorption in the Lyman continuum region arises mostly from the LW bands of H2. As a result, they were only able to retrieve upper limits on the abundance of H. These upper limits, however, are generally less than 5 % below the exobase and thus consistent with the Voyager/UVS results from both Smith et al. (1983) and Vervack and Moses (2015). This means that the thermosphere of Saturn is dominated by H2 at all altitudes below the exobase (~ 0.5–1 picobar). improvement and constitute the most extensive and reliable dataset to probe the upper atmosphere of any giant planet. Compared to the Voyager/UVS occultations, the Cassini/UVIS occultations afford higher S/N and a better point source The Cassini/UVIS data represent a significant 9.2.2 Cassini UVIS occultation data 6 resolution of 0.2 – 0.3 nm. The pointing stability of the Cassini/UVIS occultations is also excellent, thereby avoiding many of the issues that affected the Voyager observations. Finally, progress in computing, inversion techniques, and retrieval algorithms have undergone substantial evolution since the Voyager era (e.g., Koskinen et al. 2011, 2013, 2015; Vervack et al., 2004; Vervack and Moses 2015). This helps to improve the accuracy of the results, and to explore the reasons for the discrepancies in past analyses. Since the orbit insertion in 2004, Cassini/UVIS has observed more than 40 stellar and solar occultations so far. Shemansky and Liu (2012) published results based on 3 of the stellar occultations and Koskinen et al. (2015) have recently analyzed 26 of the stellar occultations that were observed between 2005 and 2014 and provide useful information on the thermosphere. Koskinen et al. (2013) also analyzed 15 of the solar occultations to retrieve the density profiles and exospheric temperatures near the exobase of Saturn. The results of this study that were obtained by analyzing the ionization continuum of H2 above the 0.1 nbar level (see Section 2.1) indicate that the exobase of Saturn is generally located between 2700 km and 3000 km above the 1 bar level and the composition of the thermosphere is dominated by H2 with negligible dissociation and production of H at the probed latitudes. retrieved by Koskinen et al. (2013, 2015) from the solar and stellar occultations. The finite size of the Sun in the solar occultations means that the relationship between the observed transmission and altitude in the atmosphere is ambiguous. Thus the density profiles had to be parameterized and fitted to the data by using a forward model that accounts for the perceived size of the solar disk in the atmosphere (Koskinen et al. 2013). Stars are point sources and a combination of direct retrieval and forward modeling was used to analyze stellar occultations. In direct retrieval each spectrum is fitted separately to obtain the column densities of the absorbers as a function of tangent altitude. The column density profiles can then be inverted to retrieve number densities (Koskinen et al. 2011, 2015). At higher altitudes (z > 1800 km), however, transmission is close to unity and the retrieved column density profiles are too noisy for inversion. Thus the best fit forward model number density profiles are shown at high altitudes in Figure 9.2. Statistical simulations show Figure 9.2 shows the 41 density profiles of H2 7 that the density profiles are accurate to within about 5 – 20 %. A comparison of the Voyager/UVS and Cassini/UVIS density profiles can be used to further test the validity of the results and detect differences in atmospheric properties between the Voyager and Cassini eras. Figure 9.3 compares four Voyager/UVS density profiles with Cassini/UVIS stellar occultation profiles at equivalent planetocentric latitudes. In general, the Voyager/UVS results agree well with the Cassini/UVIS results. Interestingly, the Cassini/UVIS densities at 21.1N agree well with the nearly symmetric V2 stellar ingress occultation at 21.7S while the Cassini/UVIS densities at 28.5N also agree well with the V2 solar egress occultation at 29N. This comparison puts upper limits on hemispheric differences near the equator and their time evolution between the Voyager and Cassini observations. The uncertainties in the Voyager/UVS density profiles, however, are often significantly larger than the uncertainties in the Cassini/UVIS profiles and they can mask temporal and spatial variations that we discuss in the next section. As argued by Vervack and Moses (2015), these uncertainties in the density profiles, have more than likely, also contributed to the ambiguities in the temperature retrievals (see Section 3.2). The overall density structure in Figure 9.2 illustrates the fact that surfaces of constant pressure on Saturn can be approximated as deformed ellipsoids of revolution (e.g., Zharkov and Trubitsyn 1970). Thus the density profiles reach minimum radial distances at the poles and a maximum near the equator. In order to see if the isobars in the thermosphere are similar to the lower atmosphere, we plotted the radial distances of the 0.01 nbar level (slightly below the exobase of Saturn) as a function of planetocentric latitude based on the Cassini/UVIS and Voyager/UVS occultations in Figure 9.4. In both cases we retrieved the radial distances from the forward model density profiles (Koskinen et al. 2015; Vervack and Moses 2015). We note that the Voyager/UVS data are generally noisier and the associated uncertainty in the pressure levels is likely to be larger than the uncertainty in the Cassini/UVIS results. Thus we assigned an uncertainty of 50 km to the Cassini/UVIS pressure levels and an uncertainty of 100 km to the Voyager/UVS occultations. occultation results are broadly consistent with the The upper panel of Figure 9.4 indicates that the 8 expected shape of Saturn's atmosphere. A closer inspection, however, reveals that there are differences between the shape of the thermosphere and the lower atmosphere. The deviations of the thermosphere from the lower atmosphere are shown by the lower panel of Figure 9.4 that compares the 0.01 nbar level based on the occultations with the 100 mbar reference model of Anderson and Schubert (2007). Here we extrapolated the reference model to 0.01 nbar by adjusting the equatorial radius of the model to match three of the equatorial stellar occultations at planetocentric latitudes of 2N and 3S from late 2008 and early 2009. We note that there is a lively debate on the rotation rate of Saturn (see Chapter 5) that affects the wind-driven perturbations to the reference model. We chose the model of Anderson and Schubert (2007) for convenience because it minimizes the wind-driven perturbations while still matching the Voyager and Cassini gravity field parameters and the observed shape of the atmosphere (Lindal et al. 1985; Jacobson et al. 2006). The deviations of the isobars in the thermosphere from the predicted shape (hereafter, the normalized altitudes) in Figure 4 show two interesting trends. First, the solar occultations (purple diamonds) from Cassini/UVIS indicate that the normalized altitude along the terminator limb increases with latitude away from the equator. The solar occultations in the southern hemisphere were all obtained between late 2007 and early 2008 so that they should be relatively free of time-dependent trends. The same is true of the solar occultations in the northern hemisphere that were obtained in 2010 with only two exceptions (the high latitude data point from 2007 and one of the low latitude data points from 2008). The trend of increasing normalized altitude with latitude was noted by Koskinen et al. (2013) who explained it by arguing that the thermosphere extends to deeper pressure levels at higher latitudes. The second trend is a relatively large 600 – 700 km scatter of the data points at low to mid (northern) latitudes in the lower panel of Figure 9.4. To our surprise this scatter appears not to be random. Instead, Figure 9.5 indicates that the exobase on Saturn expanded by about 500 km between 2006 and 2011, apparently followed by the onset of contraction some time after 2011. This trend is thought to arise from changes in the energy balance near the homopause that have caused the thermosphere to warm by about 100--200 K during the same time period (Koskinen et al. 2015), probably followed by cooling during contraction. 9 The lower panel of Figure 9.4 shows that the V2 results agree well with the Cassini/UVIS results from 2008/2009. Within their uncertainty, they also agree with the time-dependent trend in Figure 9.5. The V1 data points in Figure 9.4, however, appear significantly more elevated than the V2 data points. This result may be compromised by the relatively large uncertainties in the V1 data and we do not give it much significance. Overall, then, the Voyager/UVS points are consistent with the elevated state of the atmosphere that we observe in the Cassini/UVIS occultations of 2008/2009. We note that the expansion and warming of the atmosphere in the Cassini/UVIS occultations anti-correlates with solar activity between 2006 and 2010. Also, the Voyager/UVS and Cassini/UVIS observations of 2008/2009 occurred at opposite solar activity levels, indicating that the changes in the atmosphere are not driven by changes in solar activity. Instead, the Voyager and Cassini observations coincided with the same season during the northern spring. This implies that the observed changes, that are likely to arise from changes in dynamics, could be seasonal in nature. 9.3. REVIEW of RELEVANT TEMPERATURE DATA 9.3.1. Molecular H3+ near-IR thermal emission H3+ thermal emission has only been detected repeatedly in hotter, auroral/polar regions, where O'Donoghue, et al. (2014) report average thermospheric temperatures: 527±18 K in northern spring and 583±13 K in southern autumn seasons, respectively (see Chapter 7). However, on different Saturnian days, the southern aurora has exhibited a much wider range of temperatures, varying between ~400-600 K (Melin et al., 2007; Stallard et al., 2012; Lamy et al., 2013). In contrast to Jupiter, H3+ emissions from the disk were detected only recently by O'Donoghue et al. (2013) and they are of intermittent nature as described in Chapter 8. They are also not of sufficient quality to allow for the retrieval of temperatures. Thus for low and mid latitudes one must rely on solar and stellar occultation measurements to infer temperatures in Saturn's thermosphere from the derived H2 density scale heights. Radio occultation data yield electron density profiles from which plasma scale heights can be derived, but there is no assurance that the electron, ion, and neutral gases are in thermal equilibrium so at most only the sum of electron and ion temperatures can be inferred (see Chapter 8 for plasma temperatures). 10 temperature of 850 ± 100 K. Only a year later, Sandel et 9.3.2. Inferred temperatures from H2 density profiles. In line with the H2 density profiles (see Figures 9.2 and 9.3), the retrieval of temperatures from the occultations is limited to pressures lower than 10 nbar. As we pointed out in Section 2.1, there have been significant disagreements over the exospheric temperatures retrieved from the occultations. For example, Broadfoot et al. (1981) used the V1 solar egress occultation at the planetocentric latitude of 30S to derive an exospheric al. (1982) analyzed the V2 stellar egress occultation and reported an exospheric temperature of only 400 K at the planetocentric latitude of 3.5N. This result was contradicted by Festou and Atreya (1982) who retrieved a temperature of 800 K from the same stellar occultation. Broadfoot et al. (1981) may have been biased towards larger scale height and temperature by the finite size of the Sun that was not taken into account in the analysis. The angular size of the Sun at Saturn is about 1 mrad that, depending on the distance of the spacecraft from Saturn during the occultations, can translate to an apparent diameter of the solar disk comparable to or larger than the scale height of about 150 – 200 km of the thermosphere. Ignoring this effect, however, only overestimates the temperature by about 70 K even if the apparent diameter of the solar disk is as large as 500 km (Koskinen et al. 2013). Therefore we consider it unlikely that ignoring the solar disk in the analysis can explain a discrepancy of 400 K in the temperatures. temperature retrieved by Broadfoot et al. (1981) was too large because the latter misinterpreted the effects of an instrument gain change during the occultation and had problems in dealing with severe pointing drifts. Vervack and Moses (2015), on the other hand, found that the high altitude light curves of the V1 solar egress occultation are actually consistent with a temperature of 800 K but also concluded that the high altitude results were corrupted by bad data. Thus both Smith et al. (1983) and Vervack and Moses (2015) support the conclusion that the relatively high temperature of 800 K is erroneous. This does not explain the disagreement between Sandel et al. (1982) and Festou and Atreya (1982) over the V2 stellar egress occultation. In our opinion, however, Smith Sandel et al. (1982) argued that the results of Instead, Smith et al. (1983) argued that the 11 et al. (1983) already convincingly demonstrated that an exospheric temperature of 800 K does not provide as good a fit to this occultation as a temperature of 400 K. They also derived an exospheric temperature of only 450 K at the planetocentric latitude of 29N from the ionization continuum of H2 in the V2 solar occultation, which provides a fundamentally more robust measurement of the temperature than the the LW bands of H2 that were used by Festou and Atreya (1982). In addition, the lower temperature of 400 – 500 K is supported by the recent re-analyses of the V2 stellar egress occultation (Shemansky and Liu 2012; Vervack and Moses 2015). The lower temperature is also supported by the Cassini/UVIS occultations. To show this, Figure 9.6 compares the exospheric temperatures from Cassini/UVIS (Koskinen et al. 2013, 2015) with the Voyager/UVS results (Smith et al. 1983; Vervack and Moses 2015). The temperatures from Cassini/UVIS range from 370 K to 590 K, and the solar occultations in particular also indicate that the temperature increases by 100 – 150 K with latitude from the equator towards the poles. In general, the Cassini/UVIS results are in good agreement with the Voyager/UVS data, with the exception of the V1 solar ingress occultation near the south pole at the planetocentric latitude of 84S. This occultation, however, suffered from spacecraft slewing and anomalous channel behavior and Vervack and Moses (2015) do not place much significance on the disagreement with the Cassini/UVIS results there. In our opinion the results from Cassini/UVIS, together with the re-analysis of the Voyager/UVS data, finally settle the debate on Saturn's exospheric temperatures. In addition to the exospheric temperatures, the occultations can be used to retrieve temperature profiles that are critical to our understanding of the energy balance in the thermosphere. In general, there are three methods that have been used to retrieve the temperature profiles in the past. First, a parameterized temperature profile can be fitted to the data by forward modeling the light curves or the retrieved column density profiles. Second, the retrieved density profiles can be integrated directly to obtain partial pressures of H2 that can be converted to temperatures by using the ideal gas law. Third, the transmission spectra can be used to infer the rotational temperature of the H2 molecules by analyzing the absorption bands. 12 Many of the past studies relied on forward modeling to estimate the temperatures (Festou and Atreya 1982; Smith et al. 1983). This approach can be dangerous, particularly if the uncertainty in the light curves and thus the density profiles is large (Vervack and Moses 2015). It is especially dangerous if the atmosphere models start from the 1 bar level because in that case the results depend on several free parameters that are not constrained by the data. Direct retrieval of temperatures has the advantage that it does not make any assumptions about the temperature profile and the uncertainties are tractable with Monte Carlo techniques (e.g., Koskinen et al. 2015). With large uncertainties in the density profiles, however, direct retrieval can introduce artificial waves to the temperature profiles and the exospheric temperature depends on the upper boundary pressure that is often not known a priori. Both forward modeling and direct retrieval depend on the assumption of hydrostatic equilibrium. This is in general an excellent approximation in the thermosphere, but in principle it is also possible to constrain the rotational temperature of H2 directly from the observed spectra. This approach was attempted by Shemansky and Liu (2012) who used a combination of forward modeling and spectral analysis to constrain the temperatures. They, however, concluded that the existing databases of H2 absorption probabilities are not sufficiently extensive to reliably measure the temperatures. We note that spectral measurements of the temperature are also compromised by the insufficient wavelength resolution and S/N of the data. In addition, the absorption bands are affected by changes in temperature and level populations of H2 along the line of sight that are not separable in the transmission spectra. Koskinen et al.(2015) used a combination of forward modeling and direct retrieval to obtain temperature and density profiles iteratively from the Cassini/UVIS occultations. For example, Figure 9.7 shows the temperature-pressure (T-P) profile based on an occultation of β Crucis from January 2009 that probes the atmosphere at the planetocentric latitude of 3S (hereafter, ST32). In this case the exospheric temperature is 427 ± 11 K and the uncertainty along the profile ranges from a few percent to about 15 %. Here the forward model profile agrees well with the direct retrieval. The uncertainty depends on the brightness of the star and altitude sampling rate of the occultations, and in this regard ST32 is one of the best datasets. In order to reduce the associated uncertainties, 13 Shemansky and Liu (2012) suggested that the scale Curiously, the temperatures retrieved by Koskinen et al.(2015) disagree significantly with Shemansky and Liu (2012) for two of the three stellar occultations that were analyzed by the latter, i.e., ST32 and an occultation of δ Orionis from April 2005 that probes the atmosphere at the planetocentric latitude of 42S (hereafter, ST1). For ST1 Shemansky and Liu (2012) obtained an exospheric temperature of 318 ± 5 K whereas Koskinen et al.(2015) obtained a temperature of 429 ± 28 K for the same occultation. In addition, Shemansky and Liu (2012) obtained an exospheric temperature of 612 K for ST32 (cf. Figure 9.7). height of H2 decreases with altitude above 1400 km (1 nbar) due to significant dissociation of H2. We note that the stellar occultations cannot be used to directly retrieve the abundance of H, and the idea that H2 is significantly dissociated contradicts the relatively low abundances of H below the exobase that have been retrieved from solar occultations (Koskinen et al., 2013; Vervack and Moses 2015). Furthermore, Koskinen et al. (2015) did not find evidence for the dissociation of H2 in the light curves from the stellar occultations that are actually consistent with the scale height increasing with altitude as expected. This suggests that dissociation of H2 is not particularly important below the exobase and the H2 density profiles are likely to be close to diffusive equilibrium above the homopause. As a result, the retrieval of temperatures in the thermosphere should not be significantly affected by uncertainties in the composition. measure of the temperature in the upper thermosphere, Shemansky and Liu (2012) derived the temperature for ST32 by 'using a polynomic fit to the scale height' near 1400 km in altitude. We note that this method is not accurate in regions where the temperature changes with altitude, and it is typically much less accurate than forward modeling the density profiles or direct retrieval of temperatures (see above). To highlight this point, Figure 9.7 shows a comparison between the temperature profiles retrieved by Koskinen et al. (2015) and Shemansky and Liu (2012) for ST32. It is difficult to believe that a heat conducting atmosphere can support the temperature profile retrieved by Shemansky and Liu (2012) where the temperature increases by about 500 K within practically a single pressure level. occultations by Koskinen et al. (2015) allows for a more systematic exploration of the temperature structure in the Assuming that the scale height of H2 is not a reliable Finally, the extended analysis of the stellar 14 thermosphere. A particularly fruitful method to probe thermal structure in the atmosphere is to combine the T-P profiles from the occultations with Cassini/CIRS data to create T-P profiles that extend from the 1 bar level to the thermosphere. For example, Figure 9.8 shows five temperature profiles based on three occultations from the spring of 2006 (hereinafter, ST5, ST10 and ST12) and two from December 2008 (hereinafter, ST30 and ST31). ST5, ST10, ST12 and ST31 probe the atmosphere near the planetographic latitude of 20N while ST30 probes the atmosphere near 2N. These occultations were chosen because of a close coincidence between the UVIS occultations with the CIRS measurements in the spring of 2006, and for the fact that most of the data points showing the expansion of the atmosphere between 2006 and 2011 lie in this region. The figure also shows the best fit forward model mixing ratios of CH4 for the occultations that are discussed further in Section 7. The temperatures in the lower atmosphere are retrieved from CH4 emissions in the Cassini/CIRS limb scans and the results are valid up to the 3 µbar level. As we pointed out before, the Cassini/UVIS retrievals are valid down to the 0.1 – 0.01 µbar level, depending on the occultations. The implied agreement between the CIRS and UVIS temperatures is relatively good and the data indicate that the location of the base of the thermosphere varies between 0.1 and 0.01 µbar. We note, however, that this region falls into a gap in coverage between the two instruments, and thus we are prevented from accurately locating the base of the thermosphere. This also introduces additional uncertainty to the hydrocarbon mixing ratio profiles that are derived from the occultations (see Section 7). Interestingly, the temperature profiles from December 2008 in Figure 9.8 are generally hotter than the corresponding profiles from the spring of 2006 in the lower thermosphere (~ 0.1-10 nbar) while there are no detectable differences in the exospheric temperatures. The base of the thermosphere may also be at a higher pressure level in the December 2008 occultations. This supports the argument by Koskinen et al. (2015) that warming and extension of the thermosphere to deeper pressures can explain the expansion of the atmosphere and, by inference, that the contraction of the atmosphere that may have started after 2011 is accompanied by cooling of the lower thermosphere. The origin of these changes in thermal structure, however, is currently poorly understood. A more comprehensive study that uses photochemical and radiative transfer models to 15 interpret the temperature profiles together with the hydrocarbon abundances can shed further light on these processes and may provide more detailed information on dynamics in the mesosphere and lower thermosphere. 9.4. REVIEW of AIRGLOW DATA In common with the H2/He atmospheres of Jupiter, Uranus, and Neptune, Saturn's airglow is dominated by H2 electronic bands, the He 58.4 nm line, the H Lyman line series, and H3+ near-IR bands. Because each atmosphere has a thermosphere significantly hotter than would be predicted by solar EUV and FUV heating, other energy sources must be considered to understand the mechanisms for airglow emission. A discussion of airglow is further challenged by the long-term calibration issues in the EUV/FUV for space- borne spectrometers, the low spectral resolution of the Voyager Ultraviolet Spectrometers (UVS), and the "no resolution" of the Pioneer 10 photometer. Thus one looks to the Hopkins Ultraviolet Telescope (HUT), Hubble Space Telescope (HST), and Cassini /UVIS for high spectral resolution, well-calibrated data. Without accurate absolutely calibrated data, a discussion of airglow is reduced to purely qualitative statements without any firm understanding. 9.4.1 H Lyman Alpha In principle, the H Lyman-α (121.6 nm) dayglow on Saturn should be quite straightforward to explain. The strong solar Lyman-α line, with a line width of ~ 0.1 nm characteristic of line formation in a region where the temperature is ~ 104−5 K in the solar atmosphere, is resonantly scattered by Saturn's atomic hydrogen above the CH4 absorbing region, whose upper bound is approximately the homopause. The thermospheric temperature, ~ 300 - 600 K, governs the intrinsic planetary line width and the H column density above the absorbing CH4 region governs the scattering optical depth at line center, and together they determine what fraction of the solar line can be resonantly scattered out of the atmosphere. While the thermospheric scattering optical depth at line center can be very large, up to 105, it may be optically thin in the wings of the solar line, due to the mismatch of line widths associated with the mismatch of line formation temperatures, ~ 400 K vs. ~ 30,000 K. In addition radiative transfer to properly 16 The two Voyager UVSs measured Saturn's H Lyman-α compute planetary line formation and the emergent intensity from the atmosphere must include angle dependent scattering with frequency redistribution (cf. Lee and Meier, 1980). Voyager UVS and Cassini UVIS observations yield a relatively flat center-to-limb variation (Ben-Jaffel et al., 1995; Gustin et al., 2010), which would suggest optically thin emission (e.g. Ben-Jaffel et al. 2007 and references therein), even though the Saturn line is optically thick at line center. If total emission were dominated by line center photons, then a conservatively scattering atmosphere would give a center-to-limb cosine- like variation. However as the limb is approached the optically thin wings of the Saturn line become a more important source of emission and produce a flatter center- to-limb variation. brightnesses at ~ 3.3 kR (V1) and 3.0 kR (V2), (Broadfoot et al., 1981; Sandel et al. 1982). The average Lyman-α disk brightness from 29 IUE observations was 1.1 ± 0.36 kR (McGrath and Clarke 1992). This discrepancy between UVS and IUE is perplexing in light of their agreement on the Jovian Lyman-α brightness. Gustin et al. (2010) give peak limb brightness values with adjustments for solar activity for V1: 1.9, 2.5 kR; V2: 1.8 kR; to be compared with their UVIS limb scans with peak brightness of only 0.8 kR and scan averages of 0.44 kR. But in Table 3 of Shemansky et al. (2009) the UVIS non-auroral Lyman-α brightness range near the limb is stated to be higher at ~ 1-1.2 kR, with reference to Shemansky and Ajello (1983) that the V1 brightness was larger, ~ 4.9 kR, at mid-latitudes in 1980. No UVIS center of the disk nor disk averaged values have been published to facilitate a better comparison, but the prudent conclusion would be that the Voyager values need downward adjustment. line should also be straightforward were it not for the requirement that the He/H2 ratio be accurately known. Planetary He absorbs solar He I 58.4 nm radiation and reemits/scatters it with a probability equal to 0.9989. In addition, knowledge of the thermospheric temperature for planetary line width, and location of the homopause for the He column density above the unit H2 absorption optical depth are necessary for accurate interpretation of He I 58.4 nm observations and all are uncertain to various degrees. Like Lyman-α, the interpretation of the He I 58.4 nm 9.4.2 He 58.4 nm 17 Parkinson (2002) performed the most recent analysis of The originally reported Voyager brightnesses were V1: 2.2 ± 0.3, and V2: 4.2 ± 0.5 R (Broadfoot et al., 1981; Sandel et al. 1982), whereas Parkinson et al. (1998) reported these measurements as disk center brightness values of V1: 3.1 ± 0.4 and V2: 4.2 ± 0.5 R, with no discussion for the increased V1 value. the Saturnian He 58.4 nm line brightness for Voyager UVS, for which some aspects were previously reported in Parkinson et al. (1998). Constrained by the Voyager IRIS He/H2 mixing ratio ~ 0.03 and UVS occultation data, Parkinson (2002) required an implausibly high homopause altitude and large vertical mixing of Kzz > 109 cm2 s−1, whereas if a solar He/H2 mixing ratio ~ 0.13 were appropriate as Conrath and Gautier (2000) inferred from reanalysis of IRIS data, then Kzz > 2 x107 cm2 s−1 for V1 and Kzz > 1 x108 cm2 s−1 for V2, with the latter still exceedingly large. It must be kept in mind that the Voyager He 58.4 nm brightnesses might need downward adjustment. 9.4.3 H2 Electronic Bands The surprisingly large H2 EUV/FUV dayglow intensities observed by Voyager for Jupiter, Saturn, and Uranus generated a lively debate about excitation mechanism(s), primarily because at the time there were no rigorous calculations available on strong solar line contributions to H2 fluorescence in the dayglow. Three principal mechanisms were advanced to explain the dayglow: 1) additional electron excitation (Shemansky 1985), 2) dynamo- plasma acceleration (Clarke et al. 1987), and 3) solar fluorescence (Yelle 1988), in addition to dayglow generated by photoelectrons (cf. Strobel et al. 1991). The "excess" dayglow was given a name "electroglow" (Broadfoot et al. 1986), yet the measured intensities exhibited a dependence on the incident solar EUV and FUV fluxes at each planet. Broadfoot et al. (1986) emphasized excitation by low-energy electrons as a necessary component of the phenomenon. However, the power requirements to energize these electrons exceeded substantially what the Sun could supply in the UV from known processes. It was the combination of the high resolution HUT spectra (0.3 nm) of Jupiter's dayglow (Feldman et al., 1993) and the definitive calculation performed by Liu and Dalgarno (1996), who demonstrated that solar-induced H2 fluorescence creates a spectrum distinctly different from photoelectron impact on H2 that explains Jupiter's dayglow. 18 Fortunately for Saturn Cassini UVIS data has a The strongest fluorescence, ~ 14% of the total, is due to the solar Lyman-β line at 102.572 nm (as proposed by Yelle, 1988), which is coincident with the P(1) line of the H2 Lyman 6-0 band at 102.593 nm. spectral resolution of ~ 0.55 nm, sufficient to separate the solar fluorescence contribution from electron impact generated H2 band emissions. Gustin et al. (2010) used UVIS limb scan data taken at low latitudes below the ring plane to derive volume emission rates for various components of the dayglow. With disk-averaged Jupiter dayglow contributions adopted from Liu and Dalgarno (1996) adjusted for solar activity and scaled to Saturn, Gustin et al. (2010) obtained 173 R for fluorescence generated dayglow and 131 R for electron impact produced dayglow; thus in the ratio of 0.57:0.43. From the UVIS limb data, Gustin et al. (2010) derived limb-averaged values of 460 and 1054 R, respectively, with a ratio of 0.3:0.7. They noted that this ratio reaches a minimum of 0.2:0.8 at a tangent altitude of 1400 km which suggests that solar fluorescence is relatively more important on disk and relatively unimportant on the limb and that electron impact becomes progressively more important at high altitudes. A detailed analysis of disk-center dayglow would be extremely enlightening to determine whether solar fluorescence plus photoelectron-generated H2 dayglow is sufficient to explain Saturn's dayglow as it is for Jupiter's dayglow. 9.4.4 H3+ Thermal Emission The H3+ ion plays a fundamental role as a thermospheric thermostat for the giant planets in a manner analogous to NO in the Earth's thermosphere. By near-IR thermal emissions in its ν2 band, between 3.4–4.1 microns (described in detail in Chapter 7) H3+ regulates Saturn's thermospheric temperature. Saturn's low and mid-latitude thermosphere is colder (~ 400-450 K) with fewer H3+ ions. Thus H3+ thermal emission has mostly been detected in hotter (> 500 K) auroral/polar regions. 9.5. REVIEW OF COMPOSITION Saturn's thermosphere is mostly H2, with an uncertain amount of He and a maximum volume mixing ratio of H atoms at the exobase of 0.05, (Koskinen et al., 2013) and proportionally decreasing with decreasing altitude given the 2:1 ratio in H:H2 scale heights above the homopause. We 19 To date there are no measurements of HD in the note that the UVIS occultation forward models (Section 9.7) are also more consistent with the higher mixing ratio of 0.13 from Conrath and Gautier (2000) while the lower bound of 0.03 leads to atmospheric structure that provides a worse fit to the H2 and CH4 density profiles retrieved from the occultations. From the He 58.4 nm line emission analysis by Parkinson et al. (1998) and Parkinson (2002), our inferred location of Saturn's homopause from UVIS occultation data, and downward revision of the Voyager 58.4 nm brightnesses, only a He/ H2 ratio close to Jupiter's ratio of 0.157 could yield 58.4 nm intensities in the revised Voyager range. thermosphere, but it may be possible to detect HD with the Cassini Ion Neutral Mass Spectrometer (INMS) during the Proximal Mission when the Cassini spacecraft flies through Saturn's thermosphere. In Saturn's well-mixed lower atmosphere, there are a number of measurements of the D/H ratio in molecular hydrogen and methane, i.e., of HD and CH3D. From the review of these measurements by Fouchet et al. (2009), one concludes that the HD/H2 ratio is ~ 3.5 x 10-5 with error bars of ~ ± 50%. With a D/H ratio (= ½ HD/H2) in the well-mixed atmosphere, H atoms with an upper limit of 5% at the exobase, and D with the same scale of height as H2, the atomic D mixing ratio will be in the range of 10-8 to 10-7 (Parkinson et al., 2006), and hence of limited interest in the thermosphere. importance of CH4, with a volume mixing ratio in the lower atmosphere of 0.0047 is to locate the homopause. Chapter 10 discusses CH4 and its photochemistry in depth. Likewise, H2O is another minor species in the thermosphere, which is of much greater interest for Saturn's ionosphere, in connection with a phenomenon known as "ring rain" and discussed in depth in Chapter 8. H2O molecules are heavier than H2 and have a large loss rate in the lower stratosphere due to chemical loss, if they survive condensation, as they diffuse downward through the atmosphere. If they diffuse at their maximum velocity, their volume mixing ratio, µ, is approximately the downward flux, φ(H2O), multiplied by the H2O scale height divided by the H2O-H2 binary collision coefficient and in cgs units µ(H2O) = φ(H2O)/1013, essentially the "inverse Hunten limiting diffusive flux" for heavy gases (Hunten, 1973, cf. his Eq. 15). Thus, for example, a flux of 1 x 106 H2O cm-2 s-1 estimated by Müller- Wodarg et al. (2012) near the planetocentric latitude of For the purposes of this chapter the only real 20 20N yields a thermospheric mixing ratio of 1 x 10-7. Figure 9.9 illustrates density profiles and volume mixing ratios representative of the above discussion. 9.6. Inferred Net Heating Rate from Radial Temperature Profile . κ= 252 T 0.751 HF∇ ⋅ Tκ= − ∇ , where In the thermosphere molecular heat conduction is an important process for redistribution of thermal energy. A temperature profile yields from its gradient the heat conduction flux, HF in ergs cm-1 s-1 K- 1 (Hanley et al., 1970) for a H2 dominated atmosphere, and from its curvature the heating/cooling rate Occultation data yield fundamentally the line of sight (los) column density. The local number density is derived by inverting the column density profile and the temperature is inferred from the retrieved density profile. The partial pressures are first obtained by integrating the equation of hydrostatic equilibrium downward, starting from an assumed temperature and thus pressure at the upper boundary of the observed density profile. The ideal gas law then yields the temperature at each altitude point based on the derived pressure and the observed densities. An alternate approach is to create a model atmosphere with the temperature lower boundary condition from CIRS and wavelength dependent light curves that match the observed UVIS light curves. A comparison of both approaches is shown in Fig. 9.8 for UVIS stellar occultations obtained in 2006 and 2008, with diamonds for the data-only method and solid lines for the forward model approach. radial distance, is: C(r) are the heating and cooling rates, respectively, and can be due to dynamical as well as radiative processes. If one transforms the heat equation from variable r to 1/u, an analytic solution is obtained in terms of Gaussian-like functions for Q(r) and C(r))(cf. Stevens et al., 1993). Derived heating and cooling rates are most valid if their sources are spatially well separated and the temperature profile being modeled is well constrained by data over the entire profile. Referring to Figure 9.8, CIRS data constrain temperature profiles reliably up to 3 µbar and can be extrapolated to 0.01 µbar. Only in exceptional circumstances such as the December 2008 do stellar The radial heat conduction equation, with r for , where Q(r) and T ∂ κ r ∂ Q r C r ( ) ( ) − ∂ r ∂ 1 2 r 2 r    =    − 21 occultations yield an adequate temperature profile in this critical region. In Figure 9.10 an illustrative solution to the above heat conduction equation is given for December 2008 stellar occulation derived temperature profiles shown in Figure 8. Solution of the heat conduction equation yields a net integrated heating rate of 0.072 ergs cm-2 s-1, with peak heating at 1450 km and 0.65 nbar, while the peak cooling is inferred at 870 km and 70 nbar. The 2008 occultation was at low latitude, 18 N, where the asymptotic temperatures are ~ 400 K, whereas at auroral/polar latitudes the temperature rises to values of 550-600 K. Thus a solution of the heat equation at high latitudes would yield a larger heating rate. If one performed a series of solutions at discrete latitudes and then globally averaged the rates, one would find a globally average heating rate of ~ 0.1 erg cm-2 s-1 or ~ 5 TW total for Saturn's thermosphere, in considerable excess of what solar EUV/FUV power can deliver (~ 0.15-0.3 TW). Note that this is the global heating rate and the required power input is the heating rate divided by the heating efficiency, which according to Waite et al. (1983) is ~ 0.5 for solar UV heating and auroral energy sources. Thus the power input required is ~ 10 TW, for which only Joule/ion- neutral heating can supply this amount as discussed in Section 8.3. 9.7. Inferred Homopause Location from CH4 Data Absorption by H2 in the occultations is negligible at wavelengths higher than about 120 nm and this allows for minor species such as CH4, C2H6, C2H4, and C2H2 to be detected. The CH4 profiles can then be used to constrain the eddy mixing coefficient Kzz and the location of the homopause. Unfortunately mixing ratios are required to properly pinpoint the location of the homopause and absorption by H2 is saturated at the level in the atmosphere where methane densities are retrievable (i.e., roughly below the 0.01 µbar level), making it difficult to determine the mixing ratio without interpolating between regions. Combining temperature measurements in the stratosphere with the temperature profiles from the UV occultations is therefore critical for creating atmosphere models that can be used to calculate the mixing ratios of CH4 and other hydrocarbons (cf. Figure 9.9). The wealth of observations from Cassini/CIRS makes this approach more reliable for the Cassini/UVIS data than for the Voyager/UVS occultations. 22 The results are still subject to uncertainties, however, because CIRS and UVIS do not observe the same location at the same time, and there is a gap in the temperature coverage of the two instruments between 0.01 µbar and 3 µbar (see Section 3.2). With these caveats in mind, Vervack and Moses (2015) draw two conclusions based on the Voyager/UVS occultations that can now be re-evaluated in light of the Cassini data. First, Saturn's upper atmosphere is subject to strong mixing with a relatively high altitude homopause and second, the location of the homopause may be highly variable. Based on their analysis of five Voyager/UVS occultations, Vervack and Moses (2015) found that the CH4 profiles could only be fitted by Kzz profiles that increase with altitude throughout the thermosphere. Thus the homopause pressure, where by definition Kzz is equal to the CH4 – H2 molecular diffusion coefficient, was typically very low. The Voyager 2 solar ingress occultation near the planetocentric latitude of 29N showed the lowest pressure homopause at 0.7 nbar with Kzz = 2 x 109 cm2s-1. We note that such high values of Kzz agree with the inference of strong mixing from the Voyager/UVS He 58.4 nm data (see Section 4.2). atmosphere models, Vervack and Moses (2015) found it more convenient to derive the pressure and Kzz at the level where the mixing ratio of CH4 is 5 x 10-5 (hereafter, the CH4 reference level), than locating the homopause, to facilitate comparison with other work and to look for variations. In the three Voyager/UVS occultations probing the southern hemisphere the CH4 reference level was located at 0.01 – 0.1 µbar with Kzz = 1 – 3 x 107 cm2s-1. In the two occultations probing the northern hemisphere, on the other hand, the CH4 reference level was closer to 0.01 µbar with a higher Kzz of (1–2) x 108 cm2s-1. Cassini/UVIS observations have only been developed for five occultations to date. The model temperature profiles and the resulting CH4 mixing ratios are shown in Figure 9.8. Koskinen et al.(2015) retrieved these CH4 profiles from the FUV channel of the Cassini/UVIS instrument and created the atmosphere models. A more comprehensive analysis of all of the Cassini/UVIS occultations is in progress and it will provide highly anticipated global constraints on the variability of the homopause and associated dynamics. generally more consistent than from the Voyager data, Meanwhile, the results from Cassini/UVIS so far are Given the behavior of the Kzz profiles in their Similar background atmosphere models based on the 23 particularly because they do not confirm the peculiarly low pressure homopause in the Voyager 2 solar ingress occultation. Four of the occultations in Figure 8 probe almost the same location as the Voyager 2 solar occultation near the planetographic latitude of 20N and they indicate that the homopause pressure is 0.01 – 0.1 µbar with Kzz = 106 – 107 cm2s-1. The CH4 reference level based on these occultations, on the other hand, is located closer to the 0.1 µbar level with similar values of Kzz as at the homopause. These results agree reasonably well with the Voyager/UVS results in the southern hemisphere but not in the northern hemisphere. We note that the Cassini fits to the CH4 profiles are in agreement with the Voyager/UVS results in that the Kzz profiles that are required to match the data often increase with altitude until relatively low pressures. This differs from the typical behavior of the Kzz profiles in many planetary atmosphere applications that are assumed to asymptote to a constant value at some point in the thermosphere. The curious behavior of the Kzz profiles could arise from photochemical processes that, contrary to expectations, affect the CH4 profile, and/or waves or other dynamical processes that are not captured by the form of the Kzz profile assumed in the current studies. 9.8. ENERGETICS OF THE THERMOSPHERE 9.8.1 Inadequacy of Solar EUV/FUV Heating Strobel and Smith (1973) reviewed the literature on calculations of the temperature of the Jovian thermosphere and performed new calculations for Jupiter, Saturn, and Titan. For Saturn, they estimated that solar EUV/FUV heating could raise the asymptotic isothermal thermospheric temperature by only ~ 10 K above the mesopause temperature. As noted in Section 6, the inferred heating rate from thermospheric temperature profiles far exceeds what the Sun can supply at EUV/FUV wavelengths. 9.8.2 Wave Heating The possibility of wave heating was evaluated in Strobel (2002) for the giant planets' thermospheres based on previous detailed calculations performed by Matcheva and Strobel (1999) for gravity waves in Jupiter's thermosphere. With appropriate values for the input parameters, dynamic viscosity, µ, gravitational acceleration, g, and gas 24 p constants, cp/R, the maximum gravity wave energy flux in g R µ = 0.13 erg isothermal regions for Saturn is just 3.22 c cm-2 s-1(corrected expression from Strobel, 2002), which when coupled with the estimated heating efficiency, ~ 0.41, reduces the maximum integrated heating rate to ~ 0.055 erg cm-2 s-1, too low by about a factor of 2, and less if wave heating were not globally distributed and continuously active. The latter conditions are extremely improbable. Another important class of vertically propagating internal waves is Rossby waves whose restoring force is the meridional variation of the Coriolis force and whose dynamics are based on conservation of potential vorticity. Generally the potential vorticity of the atmosphere is dominated by planetary vorticity (f = twice the rotation rate times the sin(latitude)) with a minor contribution v∇×  . As from the relative vorticity of the velocity field, Rossby waves propagate vertically they must extract potential vorticity from the mean flow, q0, in order for their wave potential vorticity, q', to grow exponentially 0ρ− , in the absence of dissipation. But in amplitude as the wave potential vorticity cannot exceed the basic state potential vorticity, i.e., q' < f, and this restricts wave amplitudes to two orders of magnitude lower than estimated by amplitude growth (Schoeberl and Lindzen, 1982). The last class of propagating waves is acoustic waves, which are generated by lightning and thunderstorms (Schubert et al., 2003). Their amplitudes and associated energy fluxes are poorly constrained. To reach the thermosphere, their horizontal phase speeds need to be supersonic relative to the local speed of sound or otherwise they will be refracted by the thermosphere's increasing index of refraction. This requires that the storms launching these waves must be moving at supersonic speeds in the troposphere (> 1.5 km s-1) and three times the speed of the equatorial tropospheric jet. 9.8.3 Joule (Ion-Neutral) Heating due to Magnetosphere- 0.5 Ionosphere-Atmosphere Coupling In the upper atmosphere, the presence of ionospheric plasma provides a medium which responds to the presence of magnetic and electric fields, and thereby to processes that occur in the magnetosphere. When a magnetospheric electric field maps along the magnetic field lines into the atmosphere, the ionospheric ions are accelerated and 25 collide with the ambient neutral gas particles. This collisional interaction leads to an acceleration of the neutral gases in the direction of the zonal ion drift, generating a region of large zonal wind velocities where magnetospheric electric fields are strongest, near the auroral emission regions. The acceleration is given by a = -νni(u-v) = (j x B)/ρ, νni being the neutral-ion collision frequency, u the neutral wind vector, v the ion velocity vector, B the planetary magnetic field and ρ the mass density of the neutral atmosphere. The electrical current density perpendicular to the magnetic field in the ionosphere is j = σ(E + u x B), with σ being a tensor with components for the Pedersen conductivity and the Hall conductivity, E being the sum of the magnetospheric electric field mapped into the upper atmosphere and any polarization field set up by divergence of j. Because the ionosphere is not perfectly conducting, resistive heating occurs, a process often referred to as Joule heating. The thermal heating of the atmosphere by electrical currents per unit mass can be written as qJoule = (j∙E)/ρ where the electrical current density j and the electric field E both include the effect of neutral winds via the dynamo field term, u x B (Vasyliunas and Song 2005, equation 43). We note that electrical currents also result in momentum change due to ion drag that affects the kinetic energy of the gas. Sometimes this latter effect is referred to as "ion drag heating". While the thermal heating by currents alone (without considering neutral winds) can only be a positive quantity, the ion drag heating, qIon, can also attain negative values, implying the loss of kinetic energy of the neutral atmosphere. As a result, the calculation of ion drag heating requires knowledge of the thermospheric winds. The Saturn Thermosphere Ionosphere Model (STIM) is a General Circulation Model (GCM) which numerically solves non-linear coupled Navier-Stokes equations of energy, momentum and continuity for both neutral gas particle and ions in Saturn's thermosphere and ionosphere (Müller-Wodarg et al., 2006; 2012). The model currently relies on provision of magnetospheric electric fields and electron energy fluxes as external boundary conditions but then calculates the magnetosphere-ionosphere-thermosphere interaction self-consistently. The model includes solar and electron impact ionization, using for the latter the parameterization of Galand et al. (2011). Once created, the ions undergo chemical reactions as described by Moore et al. (2004). 26 Figure 9.11 shows the magnetospheric electric field strength (color contours) as mapped into the southern polar region. Also shown are the locations of maximum field- aligned current (black symbols), which coincide with the regions of largest electron precipitation into Saturn's polar upper atmosphere. The values of Figure 9.11 are taken from the BATSrUS MHD model of Saturn's magnetosphere (Jia et al., 2012) for quiet solar wind conditions. We have multiplied the original field strength by a factor of 4 in order to better reproduce observed polar temperatures and winds. The electric field is directed primarily equatorward and thereby generates a westward acceleration of the ionospheric ions. Electron precipitation occurs along the ring-shaped region in Figure 9.11 and is local time dependent not only in terms of its latitude (as seen in the figure) but also in terms of the magnitude of the electron energy flux. We apply in STIM-GCM the local time shape of electron flux consistent with that inferred from auroral observations by Lamy et al. (2009) with a maximum flux in the dawn sector near 08:00 (Müller-Wodarg et al., 2012). Near midnight the electron flux is close to zero. In the simulation shown here we apply 10 keV electrons alone, but the model allows for implementation of other electron populations as well. We assume a longitudinally averaged auroral energy flux of 1.0 mW m-2, a value based on the findings of Lamy et al. (2009). The electron impact ionization causes enhanced Pedersen and Hall conductivities in the atmosphere, which closely follow the local time changes of electron precipitation. Figure 9.12 shows zonally averaged Pedersen conductances in Saturn's ionosphere as a function of latitude, assuming that magnetic field lines are aligned radially in the thermosphere. Since the auroral magnetospheric interaction is confined to polar latitudes, this assumption is acceptable in Saturn's almost perfect dipole field (see Chapter 4). At low latitudes the conductance results from solar radiation ionization, reaching around 2-3 mho, depending on the season (larger at equinox). These rapidly decrease towards the poles with increasing solar zenith angle. From 70-75 latitude, however, we see a strong enhancement to values of around 5- 7 mho which result from the 10 keV electron impact ionization. The figure shows no seasonal variation of conductances at auroral latitudes but a hemispheric asymmetry. Southern auroral conductances may attain 7 mho, while those in the north reach around 5 mho only. This 27 Figure 9.13 shows neutral temperature contours and the difference is a direct consequence of the magnetic field asymmetry between north and south. We assume the Saturn Pioneer Voyager (SPV) magnetic field model in our simulations (Davis and Smith, 1990). meridional circulation wind vectors, as simulated by STIM for equinoctial conditions. The longest arrow corresponds to around 350 m s-1. Also shown in the figure are two line plot panels with the normalized quantities: Joule heating rates (solid line), ion drag acceleration (dashed), and zonal wind velocities (dashed-dotted). The red dot in the temperature panel denotes the location of maximum Joule heating, the green dot denotes the region of maximum zonal ion drag. Both occur at 72S latitude but around 100 km apart vertically. The curves on the right panel are vertical profiles at this latitude while the curves on the top panel are latitudinal profiles at the height of peak Joule heating (solid line) and at the height of peak ion drag (dashed and dashed-dotted lines). Peaks of Joule heating and ion drag in our simulation are slightly below the region of peak H3+ emission (1155±25 km) observed by Stallard et al. (2012). Zonal winds in the upper panel are normalized to a value of 334 m s-1, and their largest values in the right panel reach 1500 m s-1, the local sound speed. The peak values of zonal ion drag and Joule heating in the two panels are 0.02 m s-2 and 1.9x10-8 W m-3, respectively. As expected, the 72S locations of maximum Joule heating and ion drag coincide with the region of peak electron precipitation and largest resulting conductance (see Figure 9.12). Interestingly, the largest zonal winds occur more poleward at 78S (see top panel), showing the influence of pressure gradients and Coriolis acceleration as additional factors affecting the winds. A westward zonal wind will experience poleward Coriolis acceleration. While ion drag is largest in the deeper ionosphere near 900 km (green dot in main panel of Figure 9.13 and dashed line in right panel), zonal winds are essentially in a gradient wind balance aloft driven by poleward directed pressure gradients on equipotential surfaces generated by auroral heating and resulting in increasing temperatures with latitude as illustrated in Figure 9.13. Ground based Doppler analyses of H3+ emissions have revealed zonal ion velocities at polar latitudes on Saturn reaching supersonic speeds of several km s-1 (Stallard et al., 2007), described in Chapter 7. As shown by Müller-Wodarg et al. (2012), plasma velocities for the conductances encountered in our simulations (Figure 9.12) can exceed neutral velocities by 28 The temperatures, like zonal winds, are not largest in around a factor of 2, so our simulations are broadly consistent with these observations. the region of strongest coupling with the magnetosphere (red and green dots) but instead peak in the polar cap region near 1300 km altitude, decrease again towards higher altitudes and reach their asymptotic values near 2000 km. Investigation of the energy equation terms in STIM (Müller- Wodarg et al., 2012) reveals this behavior to be a direct consequence of the transport terms, advection and adiabatic heating and cooling. As illustrated in Figure 9.13 (red arrows), the polar thermosphere hosts a complex circulation pattern in the meridional/altitude plane, with anti- clockwise and clockwise circulation cells, respectively, on the poleward and equatorward side of 72S below 1500 km. The poleward flow below 1500 km transports energy away from the region of peak Joule heating and downwelling over the polar region causes adiabatic compression and heating near 1100-1600 km. The red dotted line in Figure 9.13 indicates the boundary between upward and downward winds. At higher altitudes, the circulation is broadly from pole to equator, but broken into several clockwise circulation cells due to Coriolis forces. This cools the polar region above 1600 km adiabatically (causing the negative temperature gradient there, as seen in Figure 9.13) and transporting some of the energy equatorward (thereby causing a slight temperature increase with altitude equatorward of 80S). These simulations illustrate that Saturn's high latitude thermosphere and ionosphere are driven primarily by coupling to its magnetosphere. The strong westward winds, which reduce the degree of corotation of the thermosphere to only 25% near 78S, are a signature of angular momentum transfer from the upper atmosphere into the magnetosphere. The atmospheric response to this localized coupling to the magnetosphere spreads over the entire high latitude region poleward of 60 in both hemispheres. The zonal winds rapidly decrease towards more equatorial latitudes as a result of angular momentum conservation. Joule heating in our simulations provides about 6 TW of thermal energy into Saturn's upper atmosphere (summed over both hemispheres), around 20-40 times the total energy deposited by solar heating. This further emphasizes the importance of magnetosphere-atmosphere coupling on Saturn and giant planets. 9.8.4 Resistive Heating and Ion Drag by Wind-Driven Electrodynamics 29 The generation of ionospheric currents in general Outside the auroral regions, the neutral atmosphere can be affected by resistive heating and ion drag driven by electric fields that arise from the interaction of the ionospheric plasma with neutral winds, turbulence or waves. This interaction is known to be important in the E and F region of the Earth's ionosphere (e.g., Richmond et al. 1992; Richmond 1995; Richmond and Thayer 2000). It should be important also in Saturn's ionosphere, given the importance of electrodynamics in the high latitude thermosphere, and could play a role in explaining the remarkable variability in the electron density profiles retrieved from Cassini/RSS observations (Nagy et al. 2006; Kliore et al. 2009, see Chapter 8). It may also interfere with the circulation driven by auroral heating and ion drag, modify the ionospheric electric fields that are mapped down from the magnetosphere, and heat the non- auroral thermosphere. relies on plasma-neutral collisions that force the electrons and ions to move at different velocities across magnetic field lines, thus violating the frozen-in flux condition of ideal magnetohydrodynamics. Wind-driven electrodynamics or the ionospheric dynamo, on the other hand, is based on the generation of polarization electric fields by neutral winds that lie perpendicular to the magnetic field lines and, due to high field-aligned conductivity, remain approximately constant along magnetic field lines that traverse different layers of the atmosphere. On the Earth, for example, the dayside dynamo layer is in the E region, and the electric fields are mapped between the E and F regions. The result is an electric circuit that allows current to flow in the F region, leading to ion drag and resistive heating. At night when the E region electron density diminishes, however, the polarization electric fields that are set up in the F region significantly reduce the current density. One suggested system of thermospherically-driven currents at Saturn is a polar twin-cell vortex that has been evoked to drive oscillations in the planetary period, as described in Chapter 5. To further understand the ionospheric dynamo, it is convenient to divide Saturn's ionosphere into different "magnetization" regions M1, M2 and M3 (e.g., Koskinen et al., 2014). In the M1 region (> 10 µbar) both the electrons and ions are collisionally coupled to the neutrals and currents are generally negligible. In the M2 30 region (0.01-10 µbar), which is similar to the Earth's E region dynamo layer, the electron gyrofrequency is higher than the electron-neutral collision frequency while the ions remain collisionally coupled to the neutrals. In the M3 region (< 0.01 µbar) both the ion and electron gyrofrequencies are higher than the ion/electron-neutral collision frequencies. By definition, the Hall conductivity dominates in the M2 region while the Pedersen conductivity dominates in the M3 region. Recently, Smith (2013) suggested a new source of thermospheric heating on Jupiter based on electrodynamic coupling of the thermosphere and stratosphere that relies on wind-driven electric fields. His study shows that in principle the ionospheric dynamo can generate perpendicular electric fields on the order of 10 mV m-1 that result in resistive heating rates that are sufficiently large to explain the high temperatures in the Jovian thermosphere. For Saturn's equatorial thermosphere, the required peak heating rate to explain the low latitude temperature profiles retrieved from the Cassini/UVIS occultations is about 10-10 W m-3. Given that the peak Pedersen conductivity is about 10-5 S m-1 (Moore et al.2010), an electric field of about 3.2 mV m-1 in the neutral reference frame would produce the required heating rate. The maximum dynamo electric field strength can be estimated as ≈UB where U is the wind speed, which implies winds of ~ 150 m s-1 to generate the required electric field. While not impossible, these winds are needed in the M2 region (0.01- 10 µbar) where we have no data. Whether this mechanism actually turns out to be feasible on Saturn, however, depends on a number of assumptions. For example, Smith (2013) assumed zero winds in the thermosphere. This is problematic because the electric field in the neutral reference frame is given by En = E + u x B where u is the neutral wind and E is the dynamo electric field. The assumption of zero winds thus provides only a crude estimate of the current density that is likely to be an upper limit because the strength of the dynamo field also depends sensitively on the winds in both the M2 and M3 regions (Koskinen et al., 2014). For example, if the current initiated in the M3 region cannot close in the M2 region, a polarization electric field E is set up that cancels out the current in the M3 region. In addition, an electric field of 3.2 mV m-1 at low latitudes would lead to substantial ion drag on the neutral atmosphere that affects the heating rates. Resistive heating and ion drag must therefore be modeled self-consistently by a circulation 31 model that includes ionospheric electrodynamics. Lastly, electrodynamic coupling and the ionospheric dynamo rely on substantial conductivities in both the M2 and M3 regions. The M2 region on Saturn lies almost entirely in the hydrocarbon ion layer where recombination rates are much faster than in the M3 region. Nevertheless, recent calculations by Kim et al. (2014) indicate that electron densities ~ 103 cm-3 are possible in the M2 region, but the complex ion chemistry makes the identity of major heavy ions, effective dissociation recombination rates, and calculated electron densities uncertain. While radio occultations do not rule out significant electron densities in the M2 region, this region is at the limit for retrieving reliable electron densities. 9.9. Global General Circulation Models of the Thermosphere + emissions at auroral Apart from Doppler measurements of H3 latitudes, no observational evidence exists of the winds in Saturn's (or any giant planet's) mesosphere and thermosphere. We thereby rely on the use of numerical models to examine the general circulation of Saturn's mesosphere and thermosphere, for which only the STIM model has published results (Müller-Wodarg et al., 2006; 2012). But for Jupiter, however, several such models have been published (Achilleos et al, 1998; Bougher et al., 2005; Tao et al., 2014). These build on a heritage of thermosphere- ionosphere models for Earth which have been adapted to simulate the giant planet environment. GCMs numerically integrate the non-linear coupled Navier Stokes equations of momentum, energy and continuity on a global spherical grid. For giant planets the most common vertical coordinate used is pressure, based on the hydrostatic assumption for a high gravity environment. drivers, namely particle (mostly electron) precipitation and the magnetospheric convection electric fields. These appear as sources of ionization (alongside solar EUV), thermal energy and momentum in the codes. For Earth, detailed knowledge is available of the high latitude electric convection fields and exact locations of electron precipitation. For giant planets, locations of electron precipitation are thought to coincide with locations of peak auroral UV emissions and can be inferred geographically, along with the electron energy fluxes, from observations and modeling (cf. Chapter 7). The electric fields are more challenging to obtain as no in-situ The models require the inclusion of magnetosphere 32 The other important aspect relates to the electron measurements are available. Assuming the magnetosphere- ionosphere coupling processes via Birkeland currents as originally proposed for Jupiter by Hill (1979), observations of the degree of corotation in the magnetosphere plasma may in principle yield estimates of electric fields. For Saturn, the model of Müller-Wodarg et al. (2012) initially assumed high latitude electric fields based on magnetospheric plasma flow patterns calculated by Cowley et al. (2004) but most recently replaced these with electric fields calculated by the BATSrUS Saturn magnetosphere MHD model (Jia et al., 2012). precipitation. Energetic electrons from the magnetosphere are known to induce auroral emissions on Saturn in the EUV and FUV as well as the IR (Kurth et al., 2009). Electron precipitation occurs primarily along a narrow auroral oval region located between 70 and 85 latitude of width 1.5- 3.5 (Badman et al., 2006). The electrons have a mean energy of around 10 keV but observations have identified energies ranging from 400 eV to 30 keV (Sandel et al., 1982; Gérard et al., 2004, 2009; Gustin et al., 2009). Including the effects of this electron precipitation on the ionosphere requires calculation of the collisional interaction between the electrons and atmosphere, including secondary ionization. This is best done via numerical solution of the Boltzmann equations for suprathermal electrons, which for practical reasons is done in 1-D rather than 3-D. Several such models have been proposed for Jupiter and Saturn (Grodent et al., 2001; Gustin et al., 2009; Galand et al., 2011) and STIM relies on the parameterization for Saturn which was developed by Galand et al. (2011) on the basis of full 1-D calculations. This parameterization provides vertical profiles of ionization rates for different electron populations, scaled by the background neutral densities and initial electron energy flux. Calculations have shown electron precipitation controls ionospheric plasma densities, with solar ionization in auroral regions playing only a secondary role due to the large zenith angles (Galand et al., 2011). Therefore, any realistic calculations of the high-latitude regions with GCMs require explicit inclusion of electron precipitation. 9.10. Concluding Remarks One of the outstanding problems in the study of Saturn's atmosphere is the He/H2 ratio as the mean molecular 33 mass of the atmosphere is needed to correctly calculate the pressure levels as a function of radial distance. As noted the Cassini UVIS team has not reported to date any measurements of the He 58.4 nm line which could constrain this critical ratio. But this line is formed well above the He homopause shown in Figure 9. Thus one has to extrapolate the inferred ratio into the well mixed atmosphere with considerable uncertainty. At the end of the Cassini Mission, known as the Grand Finale Tour, the last five orbits will be flybys through Saturn's thermosphere penetrating down to the sub nanobar level, but above the homopause and before a final fatal plunge into the atmosphere. The motivation is to take advantage of the onboard INMS that is capable of measuring the neutrals: H2, He, for sure and possibly HD at these low pressures. With a spacecraft velocity ~ 30 km s-1, the kinetic energy of H2 molecules colliding with the spacecraft is ~ 8.8 eV, in excess of the 4.5 eV H2 dissociation energy. If one of the H2 proton nuclei is imparted more than 4.5 eV of kinetic energy in a collision with the instrument, the H2 bond would be broken and H2 could be undercounted relative to atomic He. The measurement of HD will be marginal if its actual mixing ratio were close to what is displayed in Figure 9.9. But if one steps down sequentially in altitude during the last four orbits after spacecraft safety is confirmed from the first orbit, one can improve the chances of measuring HD and get better density profiles of H2 and He. While performing the Grand Finale Tour, there will be many opportunities for UV stellar occultations to add more H2 density profiles at a variety of latitudes to complement the more than 40 occultations discussed in Section 9.2.2. The challenge will be to translate He/H2 and HD/H2 density ratio profiles above the homopause into extrapolation of asymptotic values deep in the well-mixed atmosphere. For Jupiter, the latter was only achieved by dropping the Galileo Probe into the atmosphere and making measurements down to ~ 20 bar. In comparison to Jupiter our knowledge of the structure of Saturn's thermosphere will be far superior due to the large number solar and stellar occultations. composition of Saturn's ionosphere and, hopefully, obtain clear evidence of water group ions and infer their effect on electron densities. The ion composition will provide a consistency check on the neutral composition measurements. One of the fundamental problems in understanding the thermospheres of Saturn and Jupiter is the heating Far more certain, is that INMS will measure the 34 mechanism(s) that accounts for their temperatures far exceeding what solar UV radiation can generate. Auroral heating is sufficient to solve this "energy crisis" for Saturn's thermospheric heating, if it can be efficiently redistributed to low latitudes. This solution has been rejected on the basis of thermospheric GCM calculations such as the STIM model. The net effect is considerably colder thermospheric temperatures than derived from occultations outside the polar regions. The fundamental cause of this under prediction by GCMs is that ion drag induces zonal winds in the retrograde, westward direction in the auroral regions, which when acted on by the Coriolis force generate poleward meridional winds that transport and confine auroral heating to polar latitudes rather than transport heat to the equator where it is most needed. Almost all thermospheric GCMs addressing this problem include the assumption of hydrostatic equilibrium, which is only applicable to low Mach number flows of less than 0.3 (Kundu, 1990). But the STIM model produces neutral winds up to the sound speed in some regions and the hydrostatic assumption is no longer valid. The Coriolis force has a term in the radial direction of 2Ω u cos(lat), which cannot be ignored for flows in excess of Mach 0.5. Also, the hydrostatic approximation filters out acoustic-gravity waves which can transmit energy out of the polar regions to lower latitudes. Auroral heating is a spatial and time dependent forcing capable of generating such waves, which have horizontal group velocities close to the sound speed. Thus auroral power pulses can be propagated to the equator in less than two Saturn days. Likewise, supersonic ions E x B convecting through the auroral thermosphere must generate shocks in the neutral atmosphere, which cannot be handled by hydrostatic GCMs. Thus the energy "crisis" may not be inadequate total power input but a problem in the global redistribution of power in models. Contributing to this is a further shortcoming of all GCM simulations published to-date, the neglect of dynamical coupling to regions below in the form of mean background winds and upward propagating waves, an aspect that is known to considerably affect the circulation in the Earth's lower thermosphere. Work to address this shortcoming is underway with STIM and may lead to more realistic lower thermosphere dynamics which favor polar energy redistribution towards the equator. Finally the thermal structure of the pressure region (~ 0.030-1 µbar) is not well characterized and the question of whether a mesopause is present is unanswered. A well 35 instrumented probe(s) could address this question as well as shed light on the He/H2 ratio and atmospheric structure from sub-nanobar to 10+ bars. Acknowledgments DFS was supported by the Cassini-Huygens Mission through JPL Contract Nos. 1408487 and NASA Grant NNX10AB84G. TTK was supported by the NASA CDAPS Grant NNX14AD51G. References Achilleos, N. S. Miller, J. Tennyson, A. D. Aylward, I. C. F. Müller-Wodarg, and D. Rees, 1998. JIM: A time- dependent, three-dimensional model of Jupiter's thermosphere and ionosphere. J. Geophys. Res., 103, doi: 10.1029/98JE00947 Adriani, A., Dinelli, B. M., López-Puertas, M., Gárcia- Comas, M., Moriconi, M. L., D'Aversa, E., Funke, B., Coradini, A., 2011. Distribution of HCN in Titan's upper atmosphere from Cassini/VIMS observations at 3 µm. Icarus, 214, 584–595, doi:10.1016/j.icarus.2011.04.016. Anderson, J. D., Schubert, G., 2007. Saturn's gravitational field, internal rotation, and interior structure. Science, 317, 1384–1387, doi:10.1126/science.1144835. Badman, S. V., S. W. H. Cowley, J.‐C. Gérard, and D. Grodent (2006), A statistical analysis of the location and width of Saturn's southern auroras, Ann. Geophys., 24, 3533–3545. Baines, K. H., et al., 2005. The atmospheres of Saturn and Titan in the near-infrared: First results of Cassini/VIMS. Earth, Moon, and Planets, 96, 119–147, doi:10.1007/s11038-005-9058-2. Ben-Jaffel, L. Ben, R. Prangé, B.R. Sandel, R.V. Yelle, C. Emerich, D. Feng, D.T. Hall, 1995. New Analysis of the Voyager UVS H Lyman-α Emission of Saturn. Icarus, 113, 91-102, doi:10.1006/icar.1995.1007. Ben-Jaffel, L. Ben, Y.J. Kim, and J. Clarke, 2007. The H Lyman-α emission line from the upper atmosphere of 36 Jupiter: Parametric radiative transfer study and comparison with data. Icarus, 190, 504-527, doi:10.1016/j.icarus.2007.03.013. Bougher, S.W., J. H. Waite, T. Majeed, and G. R. Gladstone, 2005. Jupiter Thermospheric General Circulation Model (JTGCM): Global structure and dynamics driven by auroral and Joule heating. J. Geophys. Res., 110, doi: 10.1029/2003JE002230 Broadfoot, A. L., et al., 1981. Extreme Ultraviolet Observations from Voyager 1 Encounter with Saturn, Science, 212, 206-211, doi:10.1126/science.212.4491.206. Broadfoot, et al., 1986. Ultraviolet Spectrometer Observations of Uranus, Science, 233, 74-79, doi:10.1126/science.233.4759.74. Cassidy, T. A., Johnson, R. E., 2010. Collisional spreading of Enceladus' neutral cloud. Icarus, 209, 696–703, doi:10.1016/j.icarus.2010.04.010. Clarke, J.T., M. Hudson, Y.L. Yung, 1987. The excitation of the far ultraviolet electroglow emissions on Uranus, Saturn, and Jupiter. J. Geophys. Res., 92, 15,139– 15,147, doi:10.1029/JA092iA13p15139. Connerney, J. E. P., Waite, J. H., 1984. New model of Saturn's ionosphere with an influx of water from the rings. Nature, 312, 136–138, doi:10.1038/312136a0. Conrath, B.J., and D. Gautier, 2000. Saturn Helium Abundance: A Reanalysis of Voyager Measurements. Icarus, 144, 124–134, doi:10.1006/icar.1999.6265. Cowley, S.W.H., Bunce, E.J. & O'Rourke, J.M., 2004. A simple quantitative model of plasma flows and currents in Saturn's polar ionosphere. J. Geophys. Res. Space Phys. , 109, doi: 10.1029/2003JA010375 Curdt, W., et al., 2001. The SUMER spectral atlas of solar- disk features. Astron. Astrophys., 375, 591-613. Davis Jr., L., Smith, E.J., 1990. A model of Saturn's magnetic field based on all available data. J. 37 Geophys. Res. 95, 15257–15261,doi: 10.1029/JA095iA09p15257. Esposito, L. W., et al., 2004. The Cassini ultraviolet imaging spectrograph investigation. Space Sci. Rev., 115, 299-361. Feldman, P. D., M. A. McGrath, H. W. Moos, S. T. Durrance, D. F. Strobel, and A. F. Davidsen, 1993. The spectrum of the Jovian dayglow observed at 3 A resolution with the Hopkins Ultraviolet Telescope, Astrophys. J., 406, 279-284. Festou, M. C., Atreya, S. K., 1982. Voyager ultraviolet stellar occultation measurements of the composition and thermal profiles of the Saturnian upper atmosphere. Geophys. Res. Lett., 9, 1147–1150, doi:10.1029/GL009i010p01147. Feuchtgruber, H., Lellouch, E., de Graauw, T., Bézard, B., Encrenaz, T., Griffin, M., 1997. External supply of oxygen to the atmospheres of the giant planets. Nature, 389, 159–162, doi:10.1038/38236. Fletcher, L. N., Swinyard, B., Salji, C., Polehampton, E., Fulton, T., Sidher, S., Lellouch, E., Moreno, R., Orton, G., Cavalie, T., Courtin, R., Rengel, M., Sagawa, H., Davis, G. R., Hartogh, P., Naylor, D., Walker, H., Lim, T., 2012. Sub-millimetre spectroscopy of Saturn's trace gases from Herschel/SPIRE. Astron. Astrophys., 539, A44, doi:10.1051/0004-6361/201118415. Fouchet, T., J. I. Moses, and B. J. Conrath, 2009. Saturn: Composition and Chemistry. In Saturn From Cassini- Huygens, eds. M. K. Dougherty, L. W. Esposito, and S. M. Krimigis, Springer, pp. 83-112, doi:10.1007/978-1- 4020-9217-6. Friedson, A. J., Moses, J. I., 2012. General circulation and transport in Saturn's upper troposphere and stratosphere. Icarus, 218, 861–875, doi:10.1016/j.icarus.2012.02.004. Galand, M., L. Moore, I. C. F. Müller‐Wodarg, M. Mendillo, and S. Miller (2011), Response of Saturn's auroral ionosphere to electron precipitation: Electron 38 density, electron temperature, and electrical conductivity, J. Geophys. Res., 116, A09306, doi:10.1029/2010JA016412 García-Comas, M., López-Puertas, M., Funke, B., Dinelli, B., Moriconi, M. L., Adriani, A., Molina, A., Coradini, A., 2011. Analysis of Titan CH4 3.3 µm upper atmospheric emission as measured by Cassini/VIMS. Icarus, 214, 571–583,doi:10.1016/j.icarus.2011.03.020. Gérard, J.-C., Dols, V., Grodent, D., Waite, J. H., Gladstone, G. R., Prange, R., 1995. Simultaneous observations of the Saturnian aurora and polar haze with the HST/FOC. Geophys. Res. Lett., 22, 2685–2688. Gérard, J.‐C., D. Grodent, J. Gustin, A. Saglam, J. T. Clarke, and J. T. Trauger (2004), Characteristics of Saturn's FUV aurora observed with the Space Telescope Imaging Spectrograph, J. Geophys. Res., 109, A09207, doi:10.1029/2004JA010513. Gérard, J.-C., Bonfond, B., Gustin, J., Grodent, D., Clarke, J. T., Bisikalo, D., Shematovich, V., 2009. Altitude of Saturn's aurora and its implications for the characteristic energy of precipitated electrons. Geophys. Res. Lett., 36, doi:10.1029/2008GL036554. Gérard, J.-C., Gustin, J., Pryor, W. R., Grodent, D., Bonfond, B., Radioti, A., Gladstone, G. R., Clarke, J. T., Nichols, J. D., 2013. Remote sensing of the energy of auroral electrons in Saturn's atmosphere: Hubble and Cassini spectral observations. Icarus, 223, 211–221, doi:10.1016/j.icarus.2012.11.033. Grodent, D., J. H. Waite Jr., and J.‐C. Gérard (2001), A self‐consistent model of Jovian auroral thermal structure, J. Geophys. Res., 106, 12,933–12,952, doi:10.1029/2000JA900129. Guerlet, S., Fouchet, T., Bézard, B., Simon-Miller, A. A., Flasar, M., 2009. Vertical and meridional distribution of ethane, acetylene and propane in Saturn's stratosphere from CIRS/Cassini limb observations. Icarus, 203, 214–232, doi:10.1016/j.icarus.2009.04.002. 39 Guerlet, S., Fouchet, T., Bézard, B., Flasar, F. M., Simon- Miller, A. A., 2011. Evolution of the equatorial oscillation in Saturn's stratosphere between 2005 and 2010 from Cassini/CIRS limb data analysis. J. Geophys. Res. Lett., 38, L09201, doi:10.1029/2011GL047192. Gustin, J., J.‐C. Gérard, W. Pryor, P. D. Feldman, D. Grodent, and G. Holsclaw (2009), Characteristics of Saturn's polar atmosphere and auroral electrons derived from HST/STIS, FUSE and Cassini/UVIS spectra, Icarus, 200, 176–187,doi:10.1016/j.icarus.2008.11.013. Gustin, J., Stewart, I., Gérard, J.-C., Esposito, L., 2010. Characteristics of Saturn's FUV airglow from limb- viewing spectra obtained by Cassini-UVIS. Icarus, 210, 270–283, doi:10.1016/j.icarus.2010.06.031. Gustin, J.-C. Gérard, J. Gustin, W.R. Pryor, D. Grodent, B. Bonfond, A. Radioti, G.R. Gladstone, J.T. Clarke, J.D. Nichols, 2013. Remote sensing of the energy of auroral electrons in Saturn's atmosphere: Hubble and Cassini spectral observations. Icarus, 223, 211–221, doi:10.1016/j.icarus.2010.06.031. Hallett, J. T., Shemansky, D., Liu, X., 2005. A rotational-level hydrogen physical chemistry model for general astrophysical application. Astrophys. J., 624, 448–461, doi:10.1086/428935. Hanley, H. J., R. D. McCarthy, and H. Interman 1970. The viscosity and thermal conductivity of dilute hydrogen from 150 to 5000 K. J. Res. Natl. Bur. Stds. A74, 331– 353. Hill, T.W., 1979. Inertial limit on corotation. J. Geophys. Res., 84, doi: 10.1029/JA084iA11p06554. Herbert, F., Sandel, B. R., 1991. CH4 and haze in Triton's lower atmosphere. J. Geophys. Res., 96, 19,241-19,252. Hubbard, W. B., Porco, C. C., Hunten, D. M., Rieke, G. H., Rieke, M. J., McCarthy, D. W., Haemmerle, V., Haller, J., McLeod, B., Lebofsky, L. A., Marcialis, R., Holberg, J. B, 1997. Structure of Saturn's mesosphere from the 28 Sgr occultations. Icarus, 130, 404–425, doi:10.1006/icar.1997.5839. 40 Hunten, D. M, 1973. The escape of light gases from planetary atmospheres. J. Atmos. Sci. 30, 1481-1494. Jacobson, R. A., Antreasian, P. G., Bordi, J. J., Criddle, K. E., Ionasescu, R., Jones, J. B., MacKenzie, R. A., Meek, M. C., Parcher, D., Pelletier, F. J., Owen, W. M. Jr., Roth, D. C., Roundhill, I. M., Stauch, J. R., 2006. The gravity field of the Saturnian system from satellite observations and spacecraft tracking data. Astron. J., 132, 2520–2526, doi:10.1086/508812. Jia, X., K. C. Hansen, T. I. Gombosi, M. G. Kivelson, G. Toth, D. L. DeZeeuw, and A. J. Ridley , 2012. Magnetospheric configuration and dynamics of Saturn's magnetosphere: A global MHD simulation. J. Geophys. Res., 117, doi: 10.1029/2012JA017575. Kim, S. J., Sim, C. K., Lee, D. W., Courtin, R., Moses, J. I., Minh, Y. C., 2012. The three-micron spectral feature of the Saturnian haze: Implications for the haze composition and formation process. Plan. Space Sci., 65, 122–129, doi:10.1016/j.pss.2012.02.013. Kim, Y. H., Fox, J. L., Black, J. H., Moses, J. I., 2014. Hydrocarbon ions in the lower ionosphere of Saturn. J. Geophys. Res. Space Phys., 119, 384-395, doi:10.1002/2013JA019022. Kliore, A. J., Nagy, A. F., Marouf, E. A., Anabtawi, A., Barbinis, E., Fleischman, D. U., Kahan, D. S., 2009. Midlatitude and high-latitude electron density profiles in the ionosphere of Saturn obtained by Cassini radio occultation observations. J. Geophys. Res., 114, A04315, doi:10.1029/2008JA013900. Koskinen, T. T., et al., 2011. The mesosphere and thermosphere of Titan revealed by Cassini/UVIS stellar occultations. Icarus, 216, 507-534. Koskinen, T.T., B.R. Sandel, R.V. Yelle, F.J. Capalbo, G.M. Holsclaw, W.E. McClintock, S. Edgington, 2013. The density and temperature structure near the exobase of Saturn from Cassini UVIS solar occultations. Icarus, 226, 1318–1330, doi:10.1016/j.icarus.2013.07.037. Koskinen, T. T., et al., 2014a. The variability of 41 Saturn's thermosphere from Cassini/UVIS occultations. 46th DPS Meeting, Tucson, AZ, 511.08. Koskinen, T. T., Yelle, R. V., Lavvas, P., Cho, J. Y-K., 2014b. Electrodynamics on extrasolar giant planets. Astrophys. J., 796, 16, doi:10.1088/0004- 637X/796/1/16. Koskinen, T.T., B.R. Sandel, R.V. Yelle, D. F. Strobel, I.C.F. Müller-Wodarg, J. Erwin, 2015. Saturn's variable thermosphere from Cassini/UVIS occultations, Icarus, doi:10.1016/j.icarus.2015.07.008. Kundu, P. K., 1990. Fluid Mechanics, San Diego, Academic Kurth, W. S., et al., 2009. Auroral processes, in: Saturn Press, 638 pp, pp.580-582. from Cassini‐Huygens, M. Dougherty, L. W. Esposito, and S. M. Krimigis, Eds, Chapter 12, pp. 333–374, Springer, New York, doi:10.1007/978–1–4020–9217–6. Lamy, L., B. Cecconi, R. Prangé, P. Zarka, J. D. Nichols, and J. T. Clarke, 2009. An auroral oval at the footprint of Saturn's kilometric radio sources, colocated with the UV aurorae. J. Geophys. Res., 114, doi: 10.1029/2009JA014401. Lamy, L., Prangé;, R., Pryor, W., Gustin, J., Badman, S. V., Melin, H., Stallard, T., Mitchell, D.-G., Brandt, P. C.., 2013. Multispectral simultaneous diagnosis of Saturn's aurorae throughout a planetary rotation, J. Geophys. Res., 118, 4817-4843, DOI:10.1002/jgra.50404. J. S. Lee, R. R. Meier, 1980. Angle-dependent frequency redistribution in a plane-parallel medium - External source case. Astrophys. J. 240, 185–195, doi:10.1086/158222. Lindal, G. F., Sweetnam, D. N., Eshleman, V. R., 1985. The atmosphere of Saturn: An analysis of the Voyager radio occultation measurements. Astron. J., 90, 1136 – 1146, doi:10.1086/113820. Liu, W. and A. Dalgarno, 1996. The Ultraviolet Spectrum of the Jovian Dayglow. Astrophys. J., 462, 502–518, doi:10.1086/177168. 42 Matcheva, K. I., and D. F. Strobel, 1999. Heating of Jupiter's thermosphere by dissipation of gravity waves due to molecular viscosity and heat conduction, Icarus, 140, 328-340, doi:10.1006/icar.1999.6151. McGrath, M.A., and J.T. Clarke, 1992. H I Lyman alpha emission from Saturn (1980-1990). J. Geophys. Res. 97, 13691–13703, doi:10.1029/92JA00143. Melin, H., Miller, S., Stallard, T., Trafton, L. M., Geballe, T. R., 2007. Variability in the H3+ emission of Saturn: Consequences for ionisation rates and temperature. Icarus, 186, 234 – 241, doi:10.1016/j.icarus.2006.08.014. Melin, H., Stallard, T., Miller, S., Gustin, J., Galand, M., Badman, S. V., Pryor, W. R., O'Donoghue, J., Brown, R. H., Baines, K. H., 2011. Simultaneous Cassini VIMS and UVIS observations of Saturn's southern aurora: Comparing emissions from H, H2, and H3+ at high spatial resolution. Geophys. Res. Lett., 38, L15203, doi:10.1029/2011GL048457. Moore, L.E., M. Mendillo, I. C. F. Müller-Wodarg, and D. L. Murr, 2004. Modeling of global variations and ring shadowing in Saturn's ionosphere. Icarus, 172, doi: 10.1016/j.icarus.2004.07.007. Moore, L., Müller-Wodarg, I. C. F., Galand, M., Kliore, A., Mendillo, M., et al., 2010. Latitudinal variations in Saturn's ionosphere: Cassini measurements and model comparisons. J. Geophys. Res., 115, A11317, doi:10.1029/2010JA015692. Moses, J. I., Bass, S. F., 2000. The effects of external material on the chemistry and structure of Saturn's ionosphere. J. Geophys. Res., 105, 7013 – 7052, doi:10.1029/1999JE001172. Moses, J. I., Fouchet, T., Bézard, B., Gladstone, G. R., Lellouch, E., Feuchtgruber, H., 2005. Photochemistry and diffusion in Jupiter's stratosphere: Constraints from ISO observations and comparisons with other giant planets. J. Geophys. Res., 110, E08001, doi:10.1029/2005JE002411. 43 Müller-Wodarg, I. C. F., Yelle, R. V., Mendillo, M. J., Aylward, A. D., 2003. On the global distribution of neutral gases in Titan's upper atmosphere and its e_ect on the thermal structure. J. Geophys. Res., 108, 1453, doi: 10.1029/2003JA010054 Müller-Wodarg, I.C.F., Mendillo, M., Yelle, R.V., Aylward, A.D., 2006. A global circulation model of Saturn's thermosphere. Icarus, 180, 147–160. http://dx.doi.org/10.1016/j.icarus.2005.09.002 Müller-Wodarg, I. C. F., Moore, L., Galand, M., Miller, S., Mendillo, M., 2012. Magnetosphere-atmosphere coupling at Saturn: 1 – Response of thermosphere and ionosphere to steady state polar forcing. Icarus, 221, 481 – 494, doi:10.1016/j.icarus.2012.08.034. Nagy, A. F., Kliore, A. J., Marouf, E., French, R., Flasar, M., Rappaport, N. J., Anabtawi, A., Asmar, S. W., Johnston, D., Barbinis, E., Goltz, G., Fleischman, D., 2006. First results from the ionospheric radio occultations of Saturn by the Cassini spacecraft. J. Geophys. Res., 111, A06310, doi:10.1029/2005JA011519. O'Donoghue, J., Stallard, T. S., Melin, H., Jones, G. H., Cowley, S. W. H., Miller, S., Baines, K. H., Blake, J. S. D., 2013. The domination of Saturn's low-latitude ionosphere by ring rain. Nature, 496, 193–195, doi:10.1038/nature12049. O'Donoghue, J., T. Stallard, H. Melin, S. W. H.Cowley, S. V. Badman, L. Moore, S. Miller, C. Tao, K. H. Baines, J. S. D. Blake, 2014. Conjugate observations of Saturn's northern and southern H3+ aurorae, Icarus, 229, 214-220, doi:10.1016/j.icarus.2013.11.009. Parkinson, C.D., E. Griffioen, J.C. McConnell, G.R. Gladstone, B.R. Sandel, 1998. He 584 Å Dayglow at Saturn: A Reassessment, Icarus, 133, 210–220, doi:10.1006/icar.1998.5926. Parkinson, C. D, 2002. Photochemistry and Radiative Transfer Studies in the Atmospheres of Jupiter and Saturn. Unpublished Ph. D. thesis, York University, North York, Ontario, Canada, 193 pp. Parkinson, C. D., J.C. McConnell, L. Ben Jaffel, A.Y.-T. 44 Lee, Y.L. Yung, and E. Griffioen, 2006. Deuterium chemistry and airglow in the jovian thermosphere, Icarus, 183, 451–470, doi:10.1016/j.icarus.2005.09.02. Read, P. L., Dowling, T. E., Schubert, G., 2009. Saturn's rotation period from its atmospheric planetary wave con_guration. Nature, 460, 608-610. Richmond, A. D., Ridley, E. C., Roble, R. G., 1992. A thermosphere/ionosphere general circulation model with coupled electrodynamics. Geophys. Res. Lett., 19, 601- 604, doi:10.1029/92GL00401. Richmond, A. D., 1995. Modeling equatorial ionospheric electric fields. J. Atmos. Terr. Phys., 57, 1103- 1115, doi:10.1016/0021-9169(94)00126-9. Richmond, A. D., Thayer, J. P., 2000. Ionospheric electrodynamics: A tutorial. Magnetospheric current systems, Geophysical monograph 118, 131-146 (Americal Geophysical Union), doi:10.1029/GM118p0131. Samson, J. A. R., Haddad, G. N., 1994. Total photoabsorption cross section of H2 from 18 to 113 eV. J. Opt. Soc. Am. B, 11, 277–279, doi:10.1364/JOSAB.11.000277. Sandel, B. R., et al., 1982. Extreme Ultraviolet Observations from Voyager 2 Encounter with Saturn, Science, 215, 548-553, doi:10.1126/science.215.4532.548. Schoeberl, M. R., and R. S. Lindzen, 1982. A note on the limits of Rossby wave amplitudes, J. Atmos. Sci., 39, 1171-1174. Schubert,G., M. P. Hickey, and R. L. Walterscheid, 2003. Heating of Jupiter's thermosphere by the dissipation of upward propagating acoustic waves, Icarus, 163, 398–413, doi:10.1016/S0019-1035(03)00078-2. Shemansky, D.E., 1985. An explanation for the H Ly α longitudinal asymmetry in the equatorial spectrum of Jupiter: An outcrop of paradoxical energy deposition in the exosphere. J. Geophys. Res. 90, 2673–2694, doi:10.1029/JA090iA03p02673. 45 Shemansky, D. E., Ajello, J. M., 1983. The Saturn spectrum in the EUV: Electron excited hydrogen. J. Geophys. Res., 88, 459–464, doi:10.1029/JA088iA01p00459. Shemansky, D.E., X. Liu, and H. Melin, 2009. The Saturn hydrogen plume. Planetary Space Sci., 57, 1659-1670, DOI: 10.1016/j.pss.2009.05.002. Shemansky, D. E., Liu, X., 2012. Saturn upper atmosphere structure from Cassini EUV and FUV occultations. Can. J. Phys., 90, 817–831, doi:10.1139/p2012-036. Sinclair, J. A., et al., 2013. Seasonal variations of temperature, acetylene and ethane in Saturn's atmosphere from 2005 to 2010, as observed by Cassini/CIRS. Icarus, 225, 257 – 271, doi:10.1016/j.icarus.2013.03.011. Smith, C. G. A., 2013. Electrodynamic coupling of Jupiter's thermosphere and stratosphere: A new source of thermospheric heating. Icarus, 226, 923-944, doi:10.1016/j.icarus.2013.07.001. Smith, G. R., Shemansky, D. E., Holberg, J. B., Broadfoot, A. L., Sandel, B. R., 1983. Saturn's upper atmosphere from the Voyager 2 EUV solar and stellar occultations. J. Geophys. Res., 88, 8667 – 8678, doi:10.1029/JA088iA11p08667. Snowden, D. S., et al., 2013. The thermal structure of Titan's upper atmosphere, I: Temperature pro_les from Cassini INMS observations. Icarus, 226, 552-582. Stallard, T. et al., 2007. Saturn's auroral/polar H+3 infrared emission, I. General morphology and ion velocity structure. Icarus, 189, 1-13, doi:10.1016/j.icarus.2006.12.027 Stallard, T. S., Melin, H., Miller, S., Badman, S. V., Brown, R. H., Baines, K. H., 2012. Peak emission altitude of Saturn's H3+ aurora. Geophys. Res. Lett., 39, L15103, doi:10.1029/2012GL052806. Stevens, M. H., D. F. Strobel, and F. Herbert, 1993. An analysis of the Voyager 2 ultraviolet spectrometer occultation data at Uranus: Inferring heat sources and model atmospheres, Icarus, 100, 45-63, 46 doi:10.1006/icar.1993.1005. Strobel, D. F., 2002. Aeronomic Systems on Planets, Moons, and Comets. In Atmospheres in the Solar System: Comparative Aeronomy, eds. M. Mendillo, A. Nagy, and H. Waite. Washington, D. C.: American Geophysical Union, Geophysical Monograph Series, pp. 7-22. Strobel, D. F., and G. R. Smith, 1973. On the Temperature of the Jovian Thermosphere, J. Atmos. Sci., 30, 718- 725. Strobel, D. F., R. V. Yelle, D. E. Shemansky, and S. K. Atreya, 1991. The Upper Atmosphere. In Uranus, eds. J. Bergstrahl and M. S. Matthews, Tucson, AZ: University of Arizona Press, pp. 65-109. Tao, C., Y. Miyoshi, N. Achilleos, and H. Kita, 2014. Response of the Jovian thermosphere to variations in solar EUV flux. J. Geophys. Res. Space Phys., 119, doi: 10.1002/2013JA019411. Vervack, R. J., Jr., B. R. Sandel, and D. F. Strobel, 2004. New perspectives on Titan's upper atmosphere from a reanalysis of the Voyager 1 UVS solar occultations, Icarus, 170, 91-112, doi:10.1016/j.icarus.2004.03.005 Vervack, R. J. Jr., Moses, J. I., 2015. Saturn's upper atmosphere during the Voyager era: Reanalysis and modeling of the UVS occultations. Icarus, 258,135– 163, doi: 10.1016/j.icarus.2015.06.007. Waite, J. H., Cravens, T. E., Kozyra, J., Nagy, A. F., Atreya, S. K., Chen, R. H., 1983. Electron precipitation and related aeronomy of the Jovian thermosphere and ionosphere, J. Geophys. Res., 88, 2156-2202, doi: 10.1029/JA088iA08p06143. Yelle, R. V., 1988. H2 emissions from the outer planets. Geophys. Res. Lett., 15, 1145-1148, doi:10.1029/GL015i010p01145. Yelle, R. V., et al., 1996. Structure of Jupiter's upper atmosphere: Predictions for Galileo. J. Geophys. Res., 101, 2149-2161. Yelle, R. V., Cui, J., Müller-Wodarg, I. C. F., 2008. 47 Joule heating. Journal of Geophysical Research, 110(A), doi: 10.1029/2004JA010615. Zhang, X., et al., 2013. Radiative forcing of the Methane escape from Titan's atmosphere. J. Geophys. Res., 113, E10003. Vasyliūnas, V.M. and P. Song, 2005. Meaning of ionospheric stratosphere of Jupiter, Part I: Atmospheric cooling rates from Voyager to Cassini. Planet. Space Sci., 88, 3-25. Zharkov, V. N., Trubitsyn, V. P., 1970. Theory of the figure of rotating planets in hydrostatic equilibrium -- a third approximation. Sov. Phys. Astron., 13, 981–988. 48 Figures Figure 9.1. Density profiles of H2 and H retrieved by Vervack and Moses (2015) from five low to mid latitude Voyager/UVS occultations. The results are compared with previous retrieval by Smith et al. (1983) and the density profile retrieved by Hubbard et al. (1997) from a ground- based stellar occultation. The figure is from Vervack and Moses (2015) and reprinted from Icarus with permission by Elsevier. 49 Figure 9.2. H2 density as a function of radius (from the center of Saturn) based on the Cassini/UVIS stellar (solid lines and diamonds, Koskinen et al., 2015) and solar occultations (dashed lines, Koskinen et al., 2013). The color scale is based on latitude, which decreases from lower radial distances (black) to higher radial distances (purple). No distinction between northern and southern latitude is made. The data points (diamonds) show inverted densities while the solid and dashed lines show forward model density profiles. The figure is based on Koskinen et al. (2015). 50 Figure 9.3. H2 density profiles based on Cassini/UVIS stellar occultations (purple diamonds and solid lines, Koskinen et al., 2015) and Voyager/UVS solar (29N) and stellar occultations (green triangles, Vervack and Moses, 2015) at roughly equivalent planetocentric latitudes. The uncertainties in the Cassini profiles are mostly too small to be visible while the uncertainties in the Voyager profiles are larger. Note that the Cassini (2009) and Voyager V1 and V2 occultations were obtained during the equinox season. 51 Figure 9.4. Upper panel: Radial distances to the 0.01 nbar pressure level based on Cassini/UVIS solar (red squares), Koskinen et al., 2013) and stellar (black diamonds, Koskinen et al., 2015) occultations as well as Voyager 1 and Voyager 2 occultations (green stars and filled circles, respectively, Vervack and Moses, 2015). The solid line is the 100 mbar reference level (Anderson and Schubert, 2007) and the dashed line is an extrapolation of this reference model to 0.01 nbar that matches the equatorial occultations 52 from 2008/2009. Lower panel: altitude of the data points relative to the 0.01 nbar reference level (dashed line in the upper panel). The square indicates data points included in Figure 9.5. The dashed-dotted line shows normalized altitudes predicted by the new results from the GCM of Müller-Wodarg et al. (2012) (see Section 9.8.3). The figure was taken from Koskinen et al. (2015). Figure 9.5. Normalized altitude above and below the 0.01 nbar reference model as a function of time, including the data points inside the square in the lower panel of Figure 9.4. The Voyager/UVS data points (based on Vervack and Moses, 2015) are shown at the equivalent season and time after the equinox. The Cassini occultations that fall either into the ring shadow or probe the latitudes of the ring shadow are indicated by open circles. The sunspot number is shown by the dotted line. The figure is based on Koskinen et al. (2015). 53 Figure 9.6. Exospheric temperatures retrieved from Cassini/UVIS solar (purple diamonds, Koskinen et al., 2013) and stellar (black triangles, Koskinen et al., 2015) occultations, and Voyager/UVS occultations by Smith et al. (1983) (green squares) and Vervack and Moses (2015) (green circles). The dashed line shows the exospheric temperatures based on new results from the GCM of Müller- Wodarg et al. (2012) (see Section 9.8.3). The figure was taken from Koskinen et al. (2015). 54 Figure 9.7. Temperature-pressure (T-P) profile based on the occultation of β Crucis from January 2009 that probes the atmosphere at the planetocentric latitude of 3°S. The diamonds and solid line show the direct retrieval and forward model profiles from Koskinen et al.(2015) while the crosses show the temperature profile from Shemansky and Liu (2012) (see text). 55 Figure 9.8. Forward model temperature profiles (solid lines) and CH4 mixing ratio profiles (dashed lines) for spring of 2006 (black) and December 2008 (red). The occultations probe the atmosphere at planetographic latitudes of 2 – 20°N (see text). Diamonds show the direct retrieval temperatures for two of the occultations. The figure was taken from Koskinen et al. (2015). 56 Figure 9.9. Calculated volume mixing ratio profiles that correspond to the atmospheric structure model (temperature- pressure profile) based on the occultation of ε Eridani near the planetographic latitude of 25°N from spring of 2006 and nearly coinciding CIRS limb observations (see Figure 9.8). Diamonds show the CH4 mixing ratios retrieved from the occultation to constrain the Kzz profile. The assumed mole fractions at the 1 bar level are 0.1355 for He, 4.7 x 10-3 for CH4 and 3.5 x 10-5 for HD. For water we assumed an influx of 106 cm-2 s-1 and fixed the mixing ratio to 3 x 10-9 at 0.5 mbar based on recent Herschel observations (Fletcher et al., 2012). The mixing ratio of H was set to match the upper limit of 5 % in the thermosphere (Koskinen et al., 2013). 57 Figure 9.10. Inferred heating, Q, and cooling, C, rates from December 2008 stellar occultation derived temperature profiles illustrated in Figure 9.8 (red diamonds, there; here dashed line) and from forward model (solid red lines, there; here dash dot line). Solving the heat conduction equation with the inferred Q and C profiles yields the solid line temperature. The inferred net integrated heating rate is 0.072 ergs cm-2 s-1, with peak heating at 1450 km and 0.65 nbar, while the peak cooling is inferred at 870 km and 70 nbar. 58 Figure 9.11. The magnetospheric electric field strength (color contours) as mapped into the southern polar region with the locations of maximum field-aligned current (black symbols), which coincide with the regions of largest electron precipitation into Saturn's polar upper atmosphere. The figure axes indicate local times. 59 Figure 9.12. Zonally averaged Pedersen conductances in Saturn's ionosphere as a function of latitude. Solid lines denote the southern hemisphere values, dashed values are for the northern hemisphere. The blue lines are for an equinox simulation of STIM and the red lines for southern hemisphere summer conditions. 60 Figure 9.13. Neutral temperatures contours and meridional/vertical circulation wind vectors (grey arrows) as simulated by STIM for equinoctial conditions. The longest grey arrow corresponds to around 350 m s-1. Red arrows illustrate the broad circulation pattern and the red dotted line separates regions of upward and downward vertical winds. Two line plot panels with normalized quantities show Joule heating rates (solid line, QJoule/QJoule,max), ion drag acceleration (dashed, aidrag/aidrag,max) and zonal wind velocities (dashed-dotted, uzonal/uzonal,max). The right panel shows vertical profiles at 72S latitude while the curves in the top panel are latitudinal profiles at the height of peak Joule heating (solid line) and at the height of peak ion drag (dashed and dashed-dotted lines). The red dot in the temperature panel denotes the location of maximum Joule heating shown in the line plot, also 61 labeled there with a red dot. The green dots on the temperature panel and line plot denote the region of maximum zonal ion drag. Peak Joule heating and ion drag both occur at 72S latitude but shifted vertically by ~100 km from one another. Zonal winds in the upper panel are normalized to a value of 334 m s-1; their largest values in the right panel reach 1500 m s-1. The peak values of zonal ion drag and Joule heating in the two panels are 0.02 m s-2 and 1.9x10-8 W m-3, respectively. 62
1602.07649	1	1602	2016-02-24T19:56:50	Thermal properties of Rhea's Poles: Evidence for a Meter-Deep Unconsolidated Subsurface Layer	[ "astro-ph.EP", "physics.geo-ph" ]	Cassini's Composite Infrared Spectrometer (CIRS) observed both of Rhea's polar regions during two flybys on 2013/03/09 and 2015/02/10. The results show Rhea's southern winter pole is one of the coldest places directly observed in our solar system: temperatures of 25.4+/-7.4 K and 24.7+/-6.8 K are inferred. The surface temperature of the northern summer pole is warmer: 66.6+/-0.6 K. Assuming the surface thermophysical properties of both polar regions are comparable then these temperatures can be considered a summer and winter seasonal temperature constraint for the polar region. These observations provide solar longitude coverage at 133 deg and 313 deg for the summer and winter poles respectively, with additional winter temperature constraint at 337 deg. Seasonal models with bolometric albedos of 0.70-0.74 and thermal inertias of 1-46 MKS can provide adequate fits to these temperature constraints. Both these albedo and thermal inertia values agree (within error) with those previously observed on both Rhea's leading and trailing hemispheres. Investigating the seasonal temperature change of Rhea's surface is particularly important, as the seasonal wave is sensitive to deeper surface temperatures (~10cm to m) than the more commonly reported diurnal wave (<1cm). The low thermal inertia derived here implies that Rhea's polar surfaces are highly porous even at great depths. Analysis of a CIRS 10 to 600 cm-1 stare observation, taken between 16:22:33 and 16:23:26 UT on 2013/03/09 centered on 71.7 W, 58.7 S provides the first analysis of a thermal emissivity spectrum on Rhea. The results show a flat emissivity spectrum with negligible emissivity features. A few possible explanations exist for this flat emissivity spectrum, but the most likely for Rhea is that the surface is both highly porous and composed of small particles (less than approximately 50 um).	astro-ph.EP	astro-ph	Thermal properties of Rhea's Poles: Evidence for a Meter-Deep Unconsolidated Subsurface Layer C.J.A. Howett1, J.R. Spencer1, T. Hurford2, A. Verbiscer3, M. Segura2. 1 - Southwest Research Institute, Colorado, USA. 2 - Goddard Space Flight Center, Maryland, USA. 3 – University of Virginia, Charlottesville, Virginia, USA. Corresponding Author and their Contact Details: C.J.A. Howett Email: [email protected] Telephone Number: +1 720 240 0120 Fax Number: +1 303-546-9687 Address: 1050 Walnut Street, Suite 300 Boulder, Colorado 80302 USA Abstract Cassini's Composite Infrared Spectrometer (CIRS) observed both of Rhea's polar regions during a close (2,000 km) flyby on 9th March 2013 during orbit 183. Rhea's southern pole was again observed during a more distant (51,000 km) flyby on 10th February 2015 during orbit 212. The results show Rhea's southern winter pole is one of the coldest places directly observed in our solar system: surface temperatures of 25.4±7.4 K and 24.7±6.8 K are inferred from orbit 183 and 212 data respectively. The surface temperature of the northern summer pole inferred from orbit 183 data is warmer: 66.6±0.6 K. Assuming the surface thermophysical properties of the two polar regions are comparable then these temperatures can be considered a summer and winter seasonal temperature constraint for the polar region. Orbit 183 will provide solar longitude (Ls) coverage at 133° and 313° for the summer and winter poles respectively, whilst orbit 212 provides an additional winter temperature constraint at Ls 337°. Seasonal models with bolometric albedo values between 0.70 and 0.74 and thermal inertia values between 1 and 46 J m-2 K-1 s-1/2 (otherwise known as MKS units) can provide adequate fits to these temperature constraints (assuming the winter temperature is an upper limit). Both these albedo and thermal inertia values agree within the uncertainties with those previously observed on both Rhea's leading and trailing hemispheres. Investigating the seasonal temperature change of Rhea's surface is particularly important, as the seasonal wave is sensitive to deeper surface temperatures (~tens of centimeters to meter depths) than the more commonly reported diurnal wave (typically less than a centimeter), the exact depth difference dependent upon the assumed surface properties. For example, if a surface porosity of 0.5 and thermal inertia of 25 MKS is assumed then the depth of the seasonal thermal wave is 76 cm, which is much deeper than the ~0.5 cm probed by diurnal studies of Rhea (Howett et al., 2010). The low thermal inertia derived here implies that Rhea's polar surfaces are highly porous even at great depths. Analysis of a CIRS focal plane 1 (10 to 600 cm-1) stare observation, taken during the orbit 183 encounter between 16:22:33 and 16:23:26 UT centered on 71.7° W, 58.7° S provides the first analysis of a thermal emissivity spectrum on Rhea. The results show a flat emissivity spectrum with negligible emissivity features. A few possible explanations exist for this flat emissivity spectrum, but the most likely for Rhea is that the surface is both highly porous and composed of small particles (<~50 µm). 1 Introduction On 9th March 2013 the Cassini spacecraft had a close (2,000 km) encounter with Saturn's mid-sized icy satellite Rhea. The remote sensing instruments onboard Cassini viewed Rhea's southern winter hemisphere on approach and Rhea's northern summer hemisphere during departure at approximately nadir geometry. Thus, during one encounter high- spatial resolution observations were obtained of both of Rhea's poles. Nearly two years later CIRS caught another glimpse of Rhea's southern pole, this time from further away (51,000 km) and at a high emission angle (80° to 90°). To date these data sets provide the best coverage of Rhea's polar regions by CIRS, as they were taken at high-spatial resolution and at mostly low emission angles. The Composite Infrared Spectrometer (CIRS) is one of Cassini's remote sensing instruments and was taking data during both of these Rhea encounters. It is from these data that the surface temperatures of Rhea's polar regions can be inferred, where a polar region is loosely defined as lying between the pole and 60° N/S. CIRS has three focal planes covering 10 to 1400 cm-1 (c.f. Flasar et al., 2004). Focal plane 1 (FP1) covers 10 to 600 cm-1 (16.7 to 1000 µm), enabling the temperatures of even very cold surfaces to be determined (<40 K). However, FP1's drawback is that it's made from a single circular detector, which has the lowest spatial resolution of CIRS' three focal planes (3.9 mrad/pixel). The other focal planes (focal planes 3 and 4, known as FP3 and FP4) cover 600 to 1100 cm-1 (9.1 to 16.7 µm) and 1100 to 1400 cm-1 (7.1 to 9.1 µm) respectively, and are both 1x10 arrays of 0.273 mrad/pixel detectors. These wavelength ranges make FP3 and FP4 sensitive to surfaces warmer than 65 K and 110 K respectively, thus making them only suitable for looking at the warmer daytime temperatures of satellites and Enceladus' active south polar terrain. In 2010 Cassini's Ion Neutral Mass Spectrometer (INMS) discovered traces of molecular oxygen and carbon dioxide around Rhea (Teolis et al., 2010). Then evidence for a tenuous atmosphere around Dione was discovered using data collected in 2005 and 2010 by Cassini's Dual Technique Magnetometer (MAG) (Simon et al., 2011). The composition of Dione's tenuous atmosphere was later shown (using 2010 Cassini Plasma Spectrometer data) to include molecular oxygen (Tokar et al, 2012). Follow up observations showed that the gas concentrations were higher over the northern hemispheres of both satellites (Teolis et al., 2012), possibly due to seasonal variability. Since spring equinox in 2009 Rhea's (and Dione's) northern hemisphere has been warming, potentially causing volatiles previously trapped on its surface to sublimate. A surface capable of trapping the quantity of volatiles needed to produce the observed density of the exosphere must be both porous and cold (< 50 K) (Teolis et al., 2010). Both of these requirements may be met in Rhea's polar regions. The presence of volatiles on the surface of Rhea, depending on their particle size and composition, could introduce observable emissivity variations into the CIRS spectrum. The previous study by Carvano et al. (2007) found no evidence of emissivity variations in the CIRS spectrum of Phoebe, Iapetus, Enceladus, Tethys or Hyperion, but did not consider Rhea or localized emissivity variations. 2 Data and Analysis 2.1 Thermophysical Property Determination CIRS' FP1 captures the majority of the blackbody emission thus providing the most robust surface temperature determination for both colder and warmer surfaces. Table 1 shows the CIRS FP1 data obtained both during the March 2013 and February 2015 flyby that are analyzed in this work. The low altitude of the March 2013 Cassini flyby provides high spatial resolution observations even with FP1 (16 km/pixel to 359 km/pixel, see Table 1 for more information). Thus, during these two encounters FP1 has sufficient spatial resolution to determine polar temperatures and therefore only results from this focal plane will be discussed. During these observations the sub-solar latitude was low (18° N throughout the observation) so the northern sunlit pole was observed at high phase (72°). The more distant February 2015 observation had an FP1 spatial resolution between 194 to 202 km/pixel and viewed the southern pole at a high emission angle (>80°). Temperatures inferred from observations taken at high phase and high emission angles should be treated with a certain degree of caution because when rough terrain is viewed at these geometries non-representative surface regions are observed. For example during high phase observations a rough surface will create shadows that decrease the surface temperature observed by CIRS, which in turn would require a higher albedo and a lower thermal inertia to fit the observation. Whereas observations at high emission angles would preferentially observe high elevation regions (e.g. in this geometry equator-facing crater walls), the temperature of which may not be representative of the bulk surface temperature. These high elevation polar regions are probably warmer than their bulk surroundings since they will experience longer summer heating than the lower-lying and hence more shadowed neighboring terrain. Similar effects have already been observed on other icy satellites, for example Spencer (1987) analyzed Voyager 1 and 2's Infrared Interferometer Spectrometer and Radiometer (IRIS) observations of Callisto and found large variations of spectrum slope with emission angle at low solar elevation, which were attributed to the effect of surface topography. Such warmer temperatures require a lower albedo and higher thermal inertia to fit the observations. For each CIRS spectrum, the best-fitting blackbody temperature emission curve is found using a downhill simplex method based on the work of Nelder and Mead, 1965, as implemented in the IDL "amoeba" routine. Where observations overlap we calculate the mean spectral radiance, find the blackbody spectrum that fits this mean spectrum, and assume the corresponding blackbody temperature for the overlapping region. The spectral noise is converted to temperature noise using a two-step Monte Carlo technique (as detailed below). These surface temperatures of Rhea are then mapped, and the results are shown in Figures 1, 2 (South and North polar regions as observed in orbit 183) and Figure 3 (South polar region as observed in orbit 212). If we assume that surface thermophysical properties of both of Rhea's poles are the same, and that the temperatures of the fields of view that lie closest to the poles are comparable with those at the actual poles, then it is possible to use these observations to constrain Rhea's polar seasonal temperature variations. This is because both winter and summer temperature constraints are provided by these observations. There are of course potential problems with these assumptions. For example what if the thermophysical surface properties of each polar region greatly varies? That would mean that the surface temperatures close to the pole are in fact very different to those at the actual pole, and that the inferred temperatures are very dependent upon the size of the field of view (a problem since the FP1 field of view has a very different spatial resolution between the different polar observations). Another possible problem is if the two poles have very different thermophysical properties, which would undermine the conclusions drawn from our modeling. It is difficult to test the validity of these assumptions because there is little in the published literature comparing the surface of Rhea's poles, primarily because of the lack of Cassini coverage. However, the recently produced PIA18438 global map of Rhea shows no significant color difference between the poles at extended optical wavelengths (0.930 µm, 0.568 µm and 0.338 µm). Additionally, Howett et al. (2014) showed that Rhea's thermal inertias (outside of the anomalous Inktomi crater region) are uniform across low- and mid-latitudes. The PIA18438 map reveals no Inktomi-like crater at high- latitudes on Rhea. Therefore there is nothing in the available color or surface property maps to imply that the thermophysical properties of Rhea's polar regions are likely to vary significantly, either across each polar region or between the poles. So in the absence of better temporal and spatial coverage it is reasonable to use these observations to provide the first opportunity to probe Rhea's seasonal temperature variation. Seasonal temperatures are a uniquely useful quantity as it has a deeper skin-depth than diurnal temperature variations (i.e. they are sensitive to the deeper properties of the regolith). Thus, a combination of diurnal and seasonally derived surface thermophysical properties provides a strong constraint on how a surface regolith varies with depth. The properties of the surface. Skin depth is defined as depth the seasonal wave penetrates into the surface depends upon the thermophysical !!" !, where 𝜌 is the density, c is the specific heat, 𝜔 is the angular velocity of rotation and I is the thermal inertia (described as 𝑘𝜌𝑐, or as 𝑘𝜌!(1−𝑝)𝑐), where 𝜌! is the zero porosity density, p is the porosity and k is the thermal conductivity). Thermal inertia and albedo are the two quantities this work aims to constrain, since they are the prime parameters controlling surface temperature. In essence thermal inertia describes how well a surface is able to store and release thermal energy, whilst the bolometric albedo A describes the wavelength- integrated fraction of incident solar radiation reflected in all directions by a body's surface. (1-A) is thus the fraction of incident radiation that is absorbed and is available to heat the surface. To constrain these two thermophysical properties we compare the observed temperatures to the results of a 1-D diurnal thermophysical model (c.f. Spencer et al., 1989) that has been modified to predict seasonal polar temperatures. The model solves for the one-dimensional heat flow conducted to and from the surface in order to calculate the temperature as a function of depth. The upper boundary is set so that thermal and incident solar radiation are balanced with the heat conducted to and from the surface, whereas the lower boundary is set deep enough that diurnal and seasonal temperature variations result in negligible temperature change at this level. The surface emissivity is set to unity, no additional geothermal heat flow is assumed. Solar insolation is calculated using Rhea's sub-solar latitude and heliocentric distance variations. The seasonal surface temperature variations, as a function of bolometric albedo and thermal inertia, are pre-calculated and stored as look-up tables. Figures 4 and 5 show how the night and daytime temperatures observed by CIRS vary with latitude. The figures show that Rhea's South pole has a temperature of 25.4±7.4 K, and that the observation taken closest to Rhea's North pole has a temperature of 66.6 ±0.6 K. Note, a two-step Monte Carlo technique is used to convert the spectral noise estimate into a temperature noise. The process is fully described in Howett et al. (2011) but an overview is provided here for completeness. Firstly synthetic noise with a comparable magnitude to the observed noise is created and added to the previously determined best fitting blackbody curve. In the second step each spectrum (created in the first step) is fitted by a blackbody emission spectrum. These two steps are repeated numerous times and the temperature error estimate is given by the standard deviation of the temperatures whose blackbody emission spectra are best able to fit the created spectra. Figure 5 also shows that the South pole temperature as inferred from the orbit 212 data is 24.7±6.8 K, which is slightly colder but agrees (within error) with the south polar temperature derived from orbit 183 data. These blackbody curves polar temperature are above, but close to, the limit CIRS can detect (compare the noise and signal in Figures 5b and 5c). So whilst the quoted errors are robustly derived these temperatures can conservatively be considered an upper-limit, which is how they're considered when used to constrain the seasonal model temperatures. Saturn's season is described by the solar longitude LS, which is defined here to be 90° at the northern vernal equinox and 270° at the northern autumnal equinox. Thus, the orbit 183 north polar observations were taken at LS = 113° while the orbit 183 and 212 southern polar observations were taken at the equivalent LS = 313° and LS = 337° respectively (i.e. northern summer/ southern winter). If it is assumed that the observed southern pole (winter) temperatures are a proxy for those the north pole would experience at the same solar longitude then any credible seasonal model will have to predict winter and summer temperatures within the errors, at these LS. The seasonal model was run for albedo values between 0.60 and 0.80 in 0.01 increments and thermal inertias between 1 and 60 J m-2 K-1 s-1/2 (henceforth referred to as MKS) in 1 MKS increments to produce the predicted seasonal temperature variations shown in Figure 6. As the figure shows, only albedo values of 0.72±0.02 and thermal inertias values of 24!!"!!! MKS are able to predict accurately the summer and winter seasonal polar temperatures observed on Rhea. Additional seasonal model runs using an increasing thermal inertia profile with depth were also tested over a limited thermal inertia and albedo range (shown in Figure 7a). The depth profile used a nominal surface value for the top 1 cm, increasing to three times this nominal value at 3 cm, and five times its value by 10 cm (Figure 7b). Nominal surface thermal inertias between 1 and 50 MKS in 5 MKS increments, and albedo values between 0.69 and 0.75 in 0.02 increments were used. The only models able to fit the data had an albedo of 0.73 and a nominal surface thermal inertia of 1 or 5 MKS, increasing to a maximum thermal inertia of 10 MKS and 50 MKS respectively (Figure 7c). The albedo and thermal inertia of these model fits is in good agreement with those of the single thermal inertia model fits: the albedos agree within error, and only the highest thermal inertia observed at large skin depths (50 MKS below 10 cm) are just outside the range previously derived. 2.2 Emissivity Determination Since emissivity variations are subtle, high signal-to-noise observations are required to constrain them. Co-adding observations taken whilst the CIRS field of view remains at a single location (known as stare observations) increases the signal-to-noise. Volatile cold- traps are most likely to occur on the coldest surfaces of Rhea: this is the winter South Pole at the time the data was acquired. CIRS obtained a separate set of 12 stare observations near this region (centered at 71.7º W, 58.7º S) during orbit 183 on March 9th 2013 between 16:22:33 and 16:23:26 UTC. Table 1 shows full details of the geometry of the observation during this time. The location of the FP1 field of view is given in Figure 8(a). The spectra obtained during this stare are shown in Figure 8(b), along with the mean value and best fitting multi-component blackbody spectrum. In order to obtain the best- fitting spectrum we allow a combination of two blackbody spectra assuming a unity bolometric emissivity. The best fit to the observed spectrum is obtained assuming a surface temperature of 54.9±13.6 K over 24.8±17.7 % of the field of view, with the remainder at 31.4±10.5 K. This result implies large temperature contrasts in the south- polar region, with much of the surface being extremely cold. The stare observations were taken at high solar incidence angles: 61º (sun is low on the horizon) to 100º (sun is below the horizon). If the surface is not smooth then under these illumination conditions (up to incidence angles of 90º) large shadows would be expected, which may somewhat explain the observed temperature contrasts. Whether shadowing effects alone can explain such large temperature contrasts warrants further attention, being beyond the scope of this work. The standard deviation of the stare observations is also given in Figure 8(b), providing an estimate of the random noise in the data. Figure 8(c) shows the deep space spectrum taken close in time to the observation and can be used to provide an estimate of potential CIRS FP1 systematic errors. As the figure shows each of the five deep space spectrum have a very low (~negligible) magnitude that averages to zero within the errors and shows no significant errors except for wavenumbers smaller than 60 cm-1. The ratio between the observed spectrum and the best fit gives an estimate of the spectral emissivity shown in Figure 8(d) with error bars showing the standard error of the mean for both the random noise and systematic errors. Finally, Figure 8(e) shows the brightness temperature of the mean radiance spectra given in Figure 8(b). If all of the observed surface were radiating at a single temperature then the brightness temperature would be constant with respect to wavelength. Instead the brightness temperature increases with wavenumber, which instead implies a range of temperatures were observed (e.g. Bandfield et al., 2015). It's possible that these temperature variations are due to topographic/roughness variations within the CIRS field of view, similar to those observed by IRIS on Callisto (Spencer, 1987). 3 Discussion The temperature observed at Rhea's winter pole matches the coldest surface temperature ever directly measured in our solar system, within the Hermite Crater at the lunar pole (Paige et al., 2010). Figure 9 compares this temperature to those of other cold Solar System bodies. Bolometric albedo values between 0.70 and 0.74 are able to provide seasonal model fits to the observed winter and summertime polar temperatures on Rhea (Figure 6). These albedos are at the upper end but in keeping with those previously determined for Rhea's trailing and leading hemispheres (Howett et al., 2010, Figure 9), especially since they represent somewhat of an upper-limit (see the earlier discussion on shadowing effects). The thermal inertias able to fit the observations lie between 4 and 46 MKS, which encompasses the range of thermal inertias derived for Rhea's trailing and leading hemispheres using diurnal temperature observations: 8!!!!" and 9!!!! (Howett et al., 2010). The skin depth probed by the seasonal (and diurnal) waves, as previously described, is directly proportional to thermal inertia and the square root of the angular frequency (i.e. the diurnal or seasonal cycle). Assuming a surface density of non-porous ice at 93 K of 0.934 g cm-3, a specific heat for water ice at 90K of 0.8 K J-1 g-1 (Spencer and Moore, 1992), a 0.5 porosity, a diurnal cycle of 4.518 Earth days and a thermal inertia of 8 MKS (Howett et al., 2010) then Rhea's diurnal wave skin depth is ~0.6 cm. Assuming the same water ice density, specific heat and porosity but a seasonal cycle of 10832 Earth days and the range of seasonal thermal inertias derived here then the seasonal skin depth is between 12 and 140 cm. Thus, assuming the same porosity the seasonal wave is probing Rhea's surface 20 to >200 times deeper than the diurnal wave. However, Rhea's surface porosity is not well constrained. Laboratory studies estimate any icy satellite body with a diameter approximately less than or equal to Rhea's should have a residual porosity of >0.3 (Yasui and Arakawa, 2009). So instead the seasonal wave skindepth is better expressed as 1.53I/(1-p) cm, where p is the porosity and I is the thermal inertia in MKS. For comparison this is ~49 times larger than the diurnal skindepth: 0.031I/(1-p) cm, due to the large difference in the diurnal and seasonal timescales. To further illustrate this point Figure 11 shows how the seasonal skin depth varies with porosity for the mean polar thermal inertia and its limits. The figure shows that the skindepth of the seasonal wave is between 52 and 92 cm for a 24 MKS surface (for porosities between 0.3 and 0.6), and between ~2 and 175 cm over the full thermal inertia range (1 to 46 MKS). These depths probed by the seasonal wave are much deeper than previously probed by CIRS diurnal observations of Rhea and all of Saturn's classical icy satellites (see Figure 12). The upper boundary of these thermal inertia values is surprisingly low. It might be expected that the ice surface would compact quickly with depth, increasing the thermal inertia. However, the upper limit of the polar thermal inertia value is much smaller than seen on the upper surface of the Galilean satellites (50 to 70 MKS, see references inside Howett et al., 2010) and inside the thermal anomaly on Mimas (66 MKS, Howett et al., 2011). It is also consistent with the seasonal thermal inertia upper-limit set for Enceladus North polar region: < 100 MKS (Spencer et al., 2006). In short, these results are consistent with Rhea's surface being fluffy to depths of tens of centimeters. Figure 8(d) shows the emissivity variation observed for the FP1 stare observation, the location of which is shown in Figure 8(a). Figure 8(d) shows that the CIRS emissivity is unity at the 1σ level across the highest signal-to-noise part of the spectrum (~20 to 200 cm-1; 50 to 500 µm). Outside of this region the CIRS emissivity is much noisier, but does not deviate from unity within 1σ, with the exception of the ~230 to 240 cm-1 wavenumber region. However, it is not believed that this wavenumber region contains a real emissivity feature, as its >1 value is non-physical and its deviation from unity is still <2σ. Additional stare observations taken over a longer duration (therefore providing a higher signal-to-noise) are necessary to unambiguously resolve this issue, and detect any potentially weak emissivity signal that is currently lost in the noise. Thus, we conclude that no significant deviations from unity are observed in the emissivity spectrum, either as a slope or discrete feature, and it can be considered uniform. Carvano et al., (2007) modeled the emissivity of various grain sizes of water ice (normalized at 200 cm-1) and showed that large grain sizes (500 and 1000 µm) have approximately unity emissivity over ~60 to 300 cm-1, whilst that of smaller grain sizes (50 to 100 µm) have less than unity emissivity both at wavenumbers less than ~150 cm-1 and greater than ~250 cm-1. The same study also showed small grains (50 µm) of tholin material have ~10% decreasing emissivity between ~60 to 300 cm-1, while larger grains (>500 µm) have a unity emissivity over the same wavenumber range. Spilker et al. (2005) showed Saturn's A, B and C rings display a ~30% increase in emissivity between 20 and 60 cm-1, leveling off at higher wavenumbers at emissivities between 0.75 and 0.85. The CIRS emissivity spectrum does not display any such trends in emissivity. Emissivity variations occur when the observed surface does not radiate as a blackbody, which may occur for many reasons and can be manifest as slopes or discrete features. For example if a rough surface is remotely observed at high solar incidence angles then it will view both shadowed/non-shadowed regions, which will likely have vastly different surface temperatures. This will cause a range of surface emission, which when folded together may produce a spectrum that deviates from a single blackbody, producing a slope in the emissivity (Davidsson et al., 2015; Bandfield et al., 2015). Similarly spatial variations in the composition or thermal inertia of the surface can also lead to emissivity variations. Conversely surfaces made up of large particles (>500 𝜇𝑚) and surfaces made of small particles (i.e. sizes smaller than the observing wavelength, which in the high signal-to-noise region of FP1 is <50 µm) with large porosities (>90%) can produce a featureless emissivity spectra over the FP1 wavenumber range (Carvano et al., 2007). A final, but less likely option is that a spectrally featureless contaminant can suppress any volatile emissivity features. Carvano et al. (2007) showed such feature suppression requires the surface to consist of >40% contaminant, which is incompatible with observations made by Cassini's Visual and Infrared Mapping Spectrometer (VIMS). Ciarniello et al. (2011) showed Rhea's water ice is very pure, with models only requiring <0.4% Triton tholin-like contaminants to fit the data. Additionally, this study found Rhea's polar regions to have a high albedo, between 0.70 and 0.74, which implies there is no (or at least very little) dark material contamination. Analysis of observations made across Rhea by Cassini's Visual and Infrared Mapping Spectrometer (VIMS) shows grain size across Rhea to be of the order of tens and not hundred of microns (e.g. Filacchione et al., 2012, Ciarniello et al., 2011; Scipioni et al., 2014). Hapke modeling has already shown that icy satellite surfaces have high porosity, for example porosity values between 65 and 95% have been calculated for Rhea (Domingue et al., 1995; Ciarniello et al., 2011). Furthermore, the low thermal inertia limit determined in this work (≤46 MKS) is consistent with a porous surface, even at seasonal wave depths. Thus, a surface made up of small particles with large porosities is the most likely explanation for the flat emissivity spectrum observed. However, other explanations are possible, for example surface roughness effects maybe masking actual slopes in the emissivity (e.g. 50 µm tholins, as described by Carvano et al., 2007). The original motivation to investigate this particular observation was to search for evidence of volatiles located in cold-traps in Rhea's polar regions. However, no spectral signatures were observed in the small region investigated close Rhea's north pole. If the bland emissivity spectrum were caused by a highly porous surface composed of small grains then this region would allow effective adsorption (Ayotte et al., 2001). However, since no volatile spectral signature was observed this implies that either the volatiles are located at depths CIRS is insensitive to (Figure 11 and 12), or that the volatile concentration (and hence signature) is too low in this region for CIRS to detect. Both options are feasible as CIRS would not be sensitive to any volatiles condensing on the top few mm of the surface and we know from observations of Rhea's atmosphere that the available volatile concentration is low. In fact Rhea's exosphere is very tenuous: it is two orders of magnitude lower than Europa's or Callisto's. A mean atmospheric O2 column density of 3.4 ±0.7 ×10 16 m-2 was inferred from Cassini/INMS measurements (the CO2 density is lower still), which is well below the 1018 m-2 lower detection limits of both Cassini's Magnetospheric Imaging Instrument (MIMI) and VIMS (Teolis et al., 2010). 4 Conclusion The derivation of thermophysical properties of Rhea's polar region over seasonal timescales shows that Rhea's polar surfaces do not compact quickly with depth, but rather they remain porous to depths of possibly several meters. One implication is that Rhea's polar surfaces (and perhaps by extension much of Rhea) are more efficient than expected at adsorbing volatiles. However, no evidence for surface volatiles was found in the emissivity spectrum of the south polar region. This could be explained by a combination of a polar surface that has a low porosity and made up of small particles. In this study CIRS data was used to determine the emissivity of only one specific region at the southern polar region. Thus, this result does not rule out the occurrence of further emissivity variations in other CIRS observations of Rhea. Tables Date Observation Start Time End Time Sub- Sub- Range Sub-Solar Spatial Details (UT) (UT) spacecraft spacecraft (km) longitude Resolution longitude latitude (° W) of FP1 (° W) (°) (km/pixel) 9th FP1 Map of 15:58:14 16:22:23 126° W to 64° S 63,203 11.0° W 246 to 301 March Rhea's South 2013 Polar Region 128° W to to 77,329 12.4° W 9th FP1 stare centered 16:22:33 16:23:26 128° W 64° S 62,621 12.4° W 244 to 246 March at 71.7° W and 2013 58.7 ° S to to 63,110 12.5° W 9th FP1 Scan across 18:20:11 18:25:14 197° W to 56° N to 4,168 to 18.8° W 16 to 18 March Rhea's North 251° W 73° N 4,678 to 2013 10th Polar Region 19.1 ° W FP1 Scan from 07:38:00 07:55:00 304° W to 24° S to 49,744 183 ° W 194 to 202 Februar equatorial y 2015 latitudes to Rhea's South Pole 310° W 27° S to to 184° W 51,773 Table 1 – Details of the encounter geometry during various significant periods of the March 9th 2013 and February 10th 2015 Cassini Rhea flybys. The sub-solar latitude was 18° N and 23.6° N during the entire observing period of the two flybys respectively. Figures Figure 1 – Orthographic projection map centered on Rhea's South Pole, showing the orbit 183 observed surface temperature. The sub-solar point changes from 11° W, 18° N to 12° W, 18° N during this observation period. The base-map is Rhea ISS map PIA08343. Figure 2 – Orthographic projection map centered on Rhea's North Pole, showing the surface temperatures derived from orbit 183 observations. The white cross indicates the sub-solar point; the base-map is Rhea ISS map PIA08343. Figure 3 – Orthographic projection map centered on Rhea's South Pole, showing the orbit 212 observed surface temperature. The sub-solar point changes from 183° W, 24° N to 184° W, 24° N during this observation period. The base-map is Rhea ISS map PIA08343. Figure 4 – Temperature variations with latitude for Rhea's orbit 183 South polar observation (left) and North polar observation (right). Daytime (sunlit) observations are in red; nighttime observations are in blue. Local time is defined as 90° at dawn and 270° at dusk. Figure 5 – CIRS FP1 radiance measurements of surfaces nearest to Rhea's North pole from orbit 183 (a) and South pole from orbit 183 (b) and orbit 212 (c). The red solid line indicates the best blackbody fit to the observed radiances, with the uncertainty indicated by the red-dotted line. The best fit temperature for the North pole observation is 66.6±0.6 K, whilst the South pole temperatures from orbit 183 and 212 are 25.4±7.4 K and 24.7±6.8 K respectively. The blue lines show the deep-space spectrum obtained closest to the observation time, providing a measure of the uncertainty. If the instrument and calibration were perfect the deep-space spectrum would be zero. The south pole detection matches the coldest directly detected temperature on a Solar System body. Figure 6 – The grey and colored crosses on the left show the range of albedo and thermal inertia combinations used by the model to predict the seasonal polar temperatures variations, which are shown on the right. The black crosses on the seasonal temperature curves show the observed summer and winter polar temperatures, as deduced in this work. The seasonal curves able to predict both the summer and winter temperatures appear in color, and the combination of albedo and thermal inertia values used to produce them are shown in the corresponding color in the left-hand figure. Assuming that the thermophysical properties of the two poles are similar the southern pole temperatures are used as the winter temperature constraint (Ls=313°), whilst the northern pole temperatures are shown at the actual encounter solar longitude (Ls=133°). Figure 7 – Seasonal model fits to the observed polar temperature of Rhea using a thermal inertia profile that increases with depth. (a) The range of albedo and surface thermal inertia values tested by the model are shown by the crosses. Only those shown by the bold turquoise and red crosses were able to fit the observed temperatures. (b) The thermal inertia with depth profile for the two models able to fit the observed temperatures, shown in the corresponding color. Note the modeled thermal inertia remains constant below 3 cm extending to ~5 m in depth, but only the top 15 cm are show for clarity. (c) All of the model seasonal temperatures are shown (grey and colored curves). Rhea's observed polar temperatures and their errors are shown by the thick black crosses, using the same definitions as previously described in Figure 6. The two models able to fit the observed temperatures are shown in corresponding colors to their model parameters given in (a). Figure 8 – (a) The location of the CIRS south polar stare at Rhea between 16:22:33 and 16:23:26 UTC. (b) The 12 spectra obtained during the stare observation (grey), the mean spectrum and the uncertainty in the stare observations are shown by the black solid and dotted lines respectively. Overplotted in red is the best fitting multi-component blackbody spectrum to the mean radiance (54.9±13.6 K over 24.8±17.7 % of the field of view, 31.4±10.5 K over the remainder) (c) The five deep space spectra taken before and after the stare observations are shown in grey, they are plotted on the same radiance scale as the stare observations shown in (b). The mean of these deep space spectra and its standard error are given by the red solid and dotted lines respectively. The deep space observations provide a measure of the systematic errors in the CIRS data, and show no significant errors except for wavenumbers smaller than 60 cm-1. (d) The emissivity variation with wavenumber of the stare observation. The solid and dotted black error bars give the 1- and 3-sigma random noise error estimate determined from the variation in the stare observations respectively, whilst the light grey error bars provide an estimate of systematic error provided by the observed variation in the nearby deep space spectra. No believed emissivity variations above the noise levels are observed from unity, see main text for details. (e) The brightness temperature of the mean spectra, shown by the red colored line in Figure 7(b). Figure 9 – Comparison of the coldest of Rhea's winter polar temperatures observed in this work: 24.7±6.8 K from orbit 212 data, which is considered an upper-limit in modeling work (indicated by the downward facing arrow), to that of directly observed surface temperatures of other cold bodies in our solar system. The coldest Lunar temperature was observed by Lunar Diviner on Lunar Reconnaissance Orbiter in Hermite Crater by Paige et al., (2010). This Eris surface temperature is an upper-limit, it was determined using 300 to 2480 nm ESO-Very Large Telescope observations by Alvarez- Candal et al. (2011). Pluto's surface temperature was determined from 1.4 mm Spitzer observations, from which the brightness temperature was converted to a physical (kinetic) temperature by assuming an emissivity of 0.9 (Lellouch et al., 2011). Please note the given surface temperature for Europa is actually a brightness temperature, it was observed at 70° S / 240° W by the photopolarimeter-radiometer onboard Galileo using 0.35 to ~100 µm data (Spencer et al., 1999) and thus the physical temperature at this location is likely to be higher. It is predicted that Europa's polar temperatures get as low at 50 K (Prockter and Pappalardo, 2006). Figure 10 – The range of thermal inertia and bolometric albedo values able to fit the observed seasonal temperatures of Rhea's poles are shown by the grey shading. These thermal inertias and albedos are compared to those seen on other icy Saturnian satellites (black lines) (Howett et al., 2010), as well as those inside (red) and outside (blue) the thermally anomalous regions observed on some of these bodies (Howett et al., 2011, 2012, 2014). For context, these values are also compared to the Jovian Galilean satellites (Rathbun et al., 2010, Rathbun et al., 2004, Simonelli et al. (2001), Spencer et al., 1987). Figure 11 – The variation in seasonal wave skindepth with porosity for the mean (24 MKS) and lower/upper thermal inertia limits determined for Rhea's polar regions. A zero porosity surface density of 1.0 g cm-3, a specific heat for water ice at 90 K of 0.8 J K-1 g-1 (Spencer and Moore, 1992) is assumed. Figure 12 – The variation in skindepth across the major icy Saturian satellites, determined using the thermal inertias provided in Figure 8. For Mimas, Tethys and Dione the blue and red lines indicate the thermal inertias outside and inside the thermally anomalous region respectively. The thermal inertia for Enceladus North Polar region as derived by Spencer et al. (2006) was an upper limit and therefore two bars are shown: the upper 100 MKS limit (deeper skindepth) and lower 1 MKS limit (shallower skindepth). For all targets the skindepths are given for a range of porosities (0.1 to 0.9), with diamonds, crosses and triangles symbols corresponding to a porosity of 0.25, 0.50 and 0.75 respectively. The same specific heat of water ice and zero porosity density as in Figure 9 is assumed, the specific heat of a basalt flow for Io's surface of 1.5 J K-1 g-1 (Davies, 1996). The skindepth of Rhea's polar region, as determined in this work, is shown in purple. 5 Acknowledgements The authors would like to thank the Cassini Project and the Cassini Data Analysis Program (NNX12AC23G and NNX13AH84G) for funding this work. 6 References Alvarez-Candal, A., N. Pinilla-Alonso, J. Licandro, J. Cook, E. Mason, T. Roush, D. Cruikshank, F. Gourgeot, E. Dotto and D. Perna. The spectrum of (136199) Eris between 350 and 2350 nm: results with X-Shooter (2011). Astronomy and Astrophysics 532, A130, doi: 10.1051/0004-6361/201117069. Ayotte, P., S.R. Smith, K.P. Stevenson, Z. Dohnálek, G.A. Kimmel and B.D. Kay. Effect of porosity on the adsorption, desorption, trapping and release of volatile gases by amorphous solid water (2001). Journal of Geophysical Research 106, 33387-33392, doi: 10.1029/2000JE001362. Bandfield, J.L., P.O. Hayne, J-P Williams, B.T. Greenhagen and D.A. Paige. Lunar surface roughness derived from LRO Diviner Radiometer, Icarus 248, 357-372, doi:10.1016/.j.icarus.2014.11.009. Carvano, J.M., A. Miglorini, A. Barucci, M. Segur and the CIRS Team. Constraining the surface properties of Saturn's icy moons, using Cassini/CIRS emissivity spectra (2007). Icarus 187, 574-583, doi: 10.1016/j.icarus.2006.09.008. Ciarniello, M., F. Capaccioni, G. Filacchione, R.N. Clark, D.P. Cruikshank, P. Cerroni, A. Coradini, R.H. Brown, B.J. Buratti, F. Tosi and K. Stephan. Hapke modeling of Rhea surface properties through Cassini-VIMS spectra (2011). Icarus 214, 541-555, doi: 10.1016/j.icarus.2011.05.010. Davies, A.G., Io's Volcanism: Thermo-Physical Models of Silicate Lava Compared with Observations of Thermal Emission (1996). Icarus 124, 45-61, doi: 10.1006/icar.1996.0189. Davidsson, B.J.R., H. Rickman, J.L. Bandfield, O. Groussin, P.J. Gutiérrez, M. Wilska, M.T. Capria, J.P. Emery, J. Helbert, L. Jorda. A. Maturilli and T.G. Mueller (2015). Icarus 252, 1-21, doi:10.1016/j.icarus.2014.12.029. Domingue, D.L., G.W. Lockwood and D.T. Thompson. Surface textural properties of icy satellites a comparison between Europa and Rhea (1995). Icarus 115, 228-249, doi: 10.1006/icar.1995.1094. Filacchione, G., F. Capaccioni, M. Ciarniello, R.N. Clark, J.N. Cuzzi, P.D. Nicholson, D.P. Cruikshank, M.M. Hedman, B.J. Buratti, J.I. Lunine, L.A. Soderblom, F. Tosi, P. Cerroni, R.H. Brown, T.B. McCord, R. Jaumann, K. Stephan, K.H. Baines and E. Flamini. Saturn's icy satellites and rings investigated by Cassini-VIMS: III – Radial compositional variability (2012). Icarus 220, 1064-1096, doi: 10.1016/j.icarus.2012.06.040. Flasar, F.M., V.G. Kunde, M.M. Abbas, R.K. Achterberg, P. Ade, A. Barucci, B. Bézard, G.L. Bjoraker, J.C. Brasunas, S.B. Calcutt, R. Carlson, C.J. Césarsky, B.J. Conrath, A. Coradini, R. Courtin, A. Coustenis, S. Edberg, S. Edgington, C. Ferrari, T. Fouchet, D. Gautier, P.J. Gierasch, K. Grossman, P. Irwin, D.E. Jennings, E. Lellouch, A.A. Mamoutkine, A. Marten, J.P. Meyer, C.A. Nixon, G.S. Orton, T.C. Owen, J.C. Pearl, R. Prangé, F. Raulin, P.L. Read, P.N. Romani, R.E. Samuelson, M.E. Segura, M.R. Showalter, A.A. Simon-Miller, M.D. Smith, J.R. Spencer, L.J. Spilker and F.W. Taylor. Exploring the Saturn System in the Thermal Infrared: The Composite Infrared Spectrometer (2004). Space Science Review 115, 169-297, doi: 10.1007/s11214-004- 1454-9. Howett, C.J.A., J.R. Spencer, J. Pearl and M. Segura. Thermal inertia and bolometric albedo values for Mimas, Enceladus, Tethys, Dione, Rhea and Iapetus as derived from Cassini/CIRS measurements, (2010). Icarus 206, 573-593, doi: 10.1016/j.icarus.2009.07.016. Howett, C.J.A, J.R Spencer, P. Schenk, R.E. Johnson, C. Paranicas, T.A. Hurford, A. Verbiscer and M. Segura. A high-amplitude thermal inertia anomaly of probable magnetospheric origin on Saturn's moon Mimas, (2011). Icarus, 216, 221-226, doi: 10.1016/j.icarus.2011.09.007. Howett, C.J.A, J.R Spencer, T. Hurford, A. Verbiscer and M. Segura. PacMan returns: An electron-generated thermal anomaly on Tethys (2012). Icarus 221, 1084-1088, doi: 10.1016/j.icarus.2012.10.013. Howett, C.J.A, J.R Spencer, T. Hurford, A. Verbiscer and M. Segura. Thermophysical Property Variations across Dione and Rhea (2014). Icarus 241, 239-247, doi: 10.1016/j.icarus.2014.05.047. Lellouch, E., J. Stansberry, J. Emery, W. Grundy and D.P. Cruikshank. Thermal properties of Pluto's and Charon's surfaces from Spitzer observations, (2011). Icarus 214, 701-716, doi: 10.1016/j.icarus.2011.05.035. Nelder, J. and R. Mead. The Simplex Method for Function Minimization (1965). The Computer Journal 7, 308-313, doi: 10.1093/comjnl/7.4.308. Paige, D.A., M.A. Siegler, J.A. Zhang, P.O. Hayne, E.J. Foote, K.A. Bennett, A.R. Vasavada, B.T. Greenhagen, J.T. Schofield, D.J. McCleese, M.C. Foote, E. DeJong, B.B. Bills, W. Hartford, B.C. Murray, C.C. Allen, K. Snook, L.A. Soderblom, S. Calcutt, F.W. Taylor, N.E. Bowles, J.L. Bandfield, R. Elphic, R. Ghent, T.D. Glotch, M.B. Wyatt and P.G. Lucey. Diviner Lunar Radiometer Observations of Cold Traps in the Moon's South Polar Region (2010). Science 330, 479-482, doi: 10.1126/science.1187726. Prockter, L.M. and R.T. Pappalardo, Europa Chapter of Encyclopedia of the Solar System (2nd edition) Editors: L. McFadden, P. Weissman and T. Johnson, pp 432 (2006). Rathnun, J., J.R. Spencer, L.K. Tamppari, T.Z. Martin, L. Barnard and L.D. Travis. Mapping of Io's thermal radiation by the Galileo photopolarimeter-radiometer (PPR) instrument (2004). Icarus 169, 129-139, doi: 10.1016/j.icarus.2003.12.021. Rathbun, J., N. Rodriguez and J. Spencer. Galileo PPR observations of Europa: Hotspot detection limits and surface thermal properties, (2010). Icarus 210, 763–769 doi: 10.1016/j.icarus.2010.07.017. Scipioni, F., F. Tosi, K. Stephan, G. Filacchione, M. Ciarniello, F. Capaccioni, P. Cerroni, the VIMS Team. Spectroscopic classification of icy satellites of Saturn II: Identification of terrain units on Rhea, (2014). Icarus 234, 1-16 doi: 10.1016/j.icarus.2014.02.010 . Simon, S., J. Saur, F.M. Neubauer, A. Wenmacher and M.K. Dougherty. Magnetic signatures of a tenuous atmosphere at Dione (2011). Geophysical Research Letters 38, L15102, doi: 10.1029/2011GL048454. Simonelli, D., C. Dodd and J. Veverka. Regolith variations on Io: Implications for bolometric albedos (2001). Journal of Geophysical Research 106, 33241–33252, doi: 10.1029/2000JE001350. Spencer, J.R. The Surfaces of Europa, Ganymede, and Callisto: An Investigation using Voyager IRIS Thermal Infrared Spectra., Ph.D. Thesis, University of Arizona, 1987. Spencer, J.R. A rough-surface thermophysical model for airless planets (1989), Icarus 83, 27-38, doi: 10.1016/0019-1035(90)90004-S. Spencer, J. and J. Moore. The influence of thermal inertia on temperatures and frost stability on Triton (1992). Icarus 99, 261–272, doi: 10.1016/0019-1035(92)90145-W. Spencer, J.R., L.K. Tamppari, T.Z. Martin and L.D. Travis. Temperatures on Europa from Galileo Photopolatimeter-Radiometer: Nighttime Thermal Anomalies (1999). Science 284, 1514, doi: 10.1126/science.284.5419.1514. Spencer, J.R., J.C. Pearl, M. Segura, F.M. Flasar, A. Mamoutkine, P. Romani, B.J. Buratti, A. Hendrix, L.J. Spilker and R.M.C. Lopes. Cassini Encounters Enceladus: Background and the Discovery of a South Polar Hot Spot (2006). Science 311, 1401, doi:10.1126/science.1121661. Spilker, L.J., S.H. Pilorz, S.G. Edgington, B.D. Wallis, S.M Brooks, J.C. Pearl and F.M. Flasar. Cassini CIRS observations of a roll-off in Saturn ring spectra at submilillimeter wavelengths (2005). Earth, Moon, and Planets 96, 149-163, doi: 10.1007/s11038-005- 9060-8. Teolis, B. D., G.H. Jones, P.F. Miles, R.L. Tokar, B.A. Magee, J.H. Waite, E. Roussos, D.T. Young, F.J. Crary, A.J. Coates, R.E. Johnson, W.-L.Tseng and R.A Baragiola. Cassini Finds an Oxygen-Carbon Dioxide Atmosphere at Saturn's Icy Moon Rhea (2010). Science 330, 1813-1815, doi: 10.1126/science.1198366. Teolis, B. D. and J.H. Waite. Evidence for polar regolith cold trapping of exospheric O2 and CO2 on icy moons Rhea and Dione: A low--‐temperature anomalous to inner solar system volatiles? (2012). iPLEX Ices and Organics in the Inner Solar System Conference. Tokar, R.L., R.E. Johnson, M.F. Thomsen, E.C. Sittler, A.J. Coates, R.J. Wilson, F.J. Crary, D.T. Young and G.H. Jones. Detection of exospheric O2+ at Saturn's moon Dione (2012). Geophysical Research Letters 39, L03105, doi: 10.1029/2011GL050452. Yasui, M. and M. Arakawa. Compaction experiments on ice-silica particle mixtures: Implication for residual porosity of small icy bodies (2009). Journal of Geophysical Research 114, E09004, doi: 10.1029/2009JE003374.
1111.6176	1	1111	2011-11-26T17:02:13	Granular physics in low-gravity environments using DEM	[ "astro-ph.EP", "cond-mat.soft" ]	Granular materials of different sizes are present on the surface of several atmosphere-less Solar System bodies. The phenomena related to granular materials have been studied in the framework of the discipline called Granular Physics; that has been studied experimentally in the laboratory and, in the last decades, by performing numerical simulations. The Discrete Element Method simulates the mechanical behavior of a media formed by a set of particles which interact through their contact points. The difficulty in reproducing vacuum and low-gravity environments makes numerical simulations the most promising technique in the study of granular media under these conditions. In this work, relevant processes in minor bodies of the Solar System are studied using the Discrete Element Method. Results of simulations of size segregation in low-gravity environments in the cases of the asteroids Eros and Itokawa are presented. The segregation of particles with different densities was analysed, in particular, the case of comet P/Hartley 2. The surface shaking in these different gravity environments could produce the ejection of particles from the surface at very low relative velocities. The shaking causing the above processes is due to: impacts, explosions like the release of energy by the liberation of internal stresses or the re accommodation of material. Simulations of the passage of impact-induced seismic waves through a granular medium were also performed. We present several applications of the Discrete Element Methods for the study of the physical evolution of agglomerates of rocks under low-gravity environments.	astro-ph.EP	astro-ph	Mon. Not. R. Astron. Soc. 000, 1–15 (2011) Printed 11 June 2018 (MN LATEX style file v2.2) Granular physics in low-gravity environments using DEM G. Tancredi 1,2⋆, A. Maciel 1, L. Heredia 3, P. Richeri 3, S. Nesmachnow 3 1Departamento de Astronom´ıa, Facultad de Ciencias, Igu´a 4225, 11400 Montevideo, URUGUAY 2Observatorio Astron´omico Los Molinos, Ministerio de Educaci´on y Cultura, Montevideo, URUGUAY 3Centro de C´alculo, Instituto de Computaci´on, Facultad de Ingenier´ıa, Montevideo, URUGUAY Accepted 2011 November 24. Received 2011 November 21; in original form 2011 August 12 ABSTRACT Granular materials of different sizes are present on the surface of several atmosphere- less Solar System bodies. The phenomena related to granular materials have been studied in the framework of the discipline called Granular Physics; that has been studied experimentally in the laboratory and, in the last decades, by performing nu- merical simulations. The Discrete Element Method simulates the mechanical behavior of a media formed by a set of particles which interact through their contact points. The difficulty in reproducing vacuum and low-gravity environments makes numer- ical simulations the most promising technique in the study of granular media under these conditions. In this work, relevant processes in minor bodies of the Solar System are studied using the Discrete Element Method. Results of simulations of size segregation in low- gravity environments in the cases of the asteroids Eros and Itokawa are presented. The segregation of particles with different densities was analysed, in particular, the case of comet P/Hartley 2. The surface shaking in these different gravity environments could produce the ejection of particles from the surface at very low relative velocities. The shaking causing the above processes is due to: impacts, explosions like the release of energy by the liberation of internal stresses or the re accommodation of material. Sim- ulations of the passage of impact-induced seismic waves through a granular medium were also performed. We present several applications of the Discrete Element Methods for the study of the physical evolution of agglomerates of rocks under low-gravity environments. Key words: minor planets, asteroids: general – comets: general – methods: numerical 1 INTRODUCTION Granular materials of different sizes are present on the sur- face of several atmosphere-less Solar System bodies. The presence of very fine particles on the surface of the Moon, the so-called regolith, was confirmed by the Apollo astronauts. From polarimetric observations and phase angle curves, it is possible to indirectly infer the presence of fine particles on the surface of asteroids and planetary satellites. More re- cently, the visit of spacecraft to several asteroids and comets has provided us with close pictures of the surface, where par- ticles of a wide size range from cm to hundreds of meters have been directly observed. The presence of even finer par- ticles on the visited bodies can also be inferred from image analysis. It has been proposed that several typical processes of granular materials, such as the size segregation of boulders ⋆ E-mail: [email protected] on Itokawa, the displacement of boulders on Eros, among others (see e.g. Asphaug (2007) and references therein), can explain some features observed on the surfaces of these bod- ies. The conditions at the surface and the interior of these small Solar System bodies are very different compared to the conditions on the Earth's surface. Below we point out some of these differences: • while on the Earth's surface the acceleration of gravity is 9.8 m/s2 with minor variations, on the surface of elon- gated km-size asteroid is on the order of 10−2 to 10−4 m/s2, with typical variations of a factor of 2 • the presence of an atmosphere or any other fluid media plays an important role on the evolution of grains, partic- ularly in the small ones (Pak et al. (1995)). Under vacuum conditions in space, this effect does not occur. • In vacuum and low gravity conditions, other forces might play a role comparable to that of gravity, e.g. van der 2 G. Tancredi et al. Waals forces (Scheeres (2010)), although these forces are not considered in our present approach. the movements of a large amount of particles which are sub- jected to certain physical interactions. The phenomena related to granular material have been studied in the discipline called Granular Physics. Granular media are formed by a set of macroscopic objects (grains) which interact through temporal or permanent contacts. The range of materials studied by Granular Physics is very broad: rocks, sands, talc, natural and artificial powders, pills, etc. Granular materials show a variety of behaviours under different circumstances: when excited (fluidised), they often resemble a liquid, as is the case of grains flowing through pipes; or they may behave like a solid, like in a dune or a heap of sand. These processes have been studied experimentally in the laboratory, and, in the last decades, by numerical anal- ysis. The numerical simulation of the evolution of granular materials has been done recently with the Discrete Element Method (DEM). DEM is a family of numerical methods for computing the motion of a large number of particles such as molecules or grains under given physical laws. DEMs simu- late the mechanical behavior in a media formed by a set of particles which interact through their contact points. Low-gravity environments in space are difficult to re- produce in a ground-based laboratory; especially if one is interested in keeping a stable value of the acceleration of gravity on the order of 10−2 to 10−4 m/s2 for several hours, since under these low-gravity conditions the dynamical pro- cesses are much slower than on Earth. Parabolic flights are not suitable for these experiments, since it is not possible to attain a stable value during the free-fall flight. For labo- ratory experiments, we are then left with experiences to be carried on board space stations. Therefore, numerical simulation is the most promising technique to study the phenomena affecting granular mate- rial in vacuum and low-gravity environments. The rest of the article is organised as follows. In Section 2 we describe the implementation of the Discrete Element Methods used in our simulations. In Section 3 we present the results of simulations of the process of size segregation in low-gravity environments, the so-called Brazil nut effect, in the cases of Eros, Itokawa and P/Hartley 2. In Section 4, the segregation of particles with different densities is anal- ysed, with the application to the case of P/Hartley 2. The surface shaking in these different gravity environments could produce the ejection of particles from the surface at very low relative velocities; this issue is discussed in Section 5. The shaking that causes the above processes is due to impacts or explosions like the release of energy by the liberation of internal stresses or the re accommodation of material. Al- though DEM methods are not suitable to reproduce the im- pact event, we are able to make simulations of the passage of impact-induced seismic waves through a granular media; these experiments are shown in Section 6. The conclusions and the applications of these results are discussed in Section 7. DEMs present the following basic properties that gen- erally define this class of numerical algorithms: terial. DEM is a case of a Lagrangian numerical method. • The quantities are calculated at points fixed to the ma- • The particles can move independently or they can have bounds, and they interact in the contact zones through dif- ferent types of physical laws. • Each particle is considered a rigid body, subject to the laws of rigid body mechanics. The forces acting on a particle are calculated from the in- teraction of this particle with its nearest neighbors, i.e. the particles it touches. Several types of forces are usually con- sidered in the literature; e.g free elastic forces, bonded elastic forces, frictional forces, viscoelastic forces, interaction of the particles with other objects, such as walls and mesh objects acting as boundary conditions, global force fields (i.e. grav- ity), velocity dependent damping, etc. The main drawback of the method is the computational cost of computing the interacting forces for each particle at each time step. A simple all-to-all approach would require to perform O(N (N − 1)/2) operations per time step, where N is the number of particles in the simulation. Several efficient methods to reduce the number of pairs to compute have been implemented; e.g. the Verlet lists method, the link cells algorithm, and the lattice algorithm. Another problem for the simulation is the length of the time step, which should be much less than the duration of the collisions, typically 1/10 to 1/20 of collisions duration. Based on the Hertzian elastic contact theory, the duration of contact (τ ) can be expressed as: τ = 5.84(cid:18) ρ(1 − ν 2) E (cid:19)0.4 rv−0.2 (1) (Wada et al. (2006), after Timoshenko and Goodier (1970)), where ρ is the grain density, nu is the Poisson ratio, E is the material strength, r is the radius of the particle, and v is the collisional velocity. In Figure 1 we plot the previous estimate of the dura- tion of collision as a function of the collisional velocity for particles with r = 0.1, 1 and 10 m. The other parameters are assumed as follows: ρ = 2000 gr/cm3, ν = 0.17, and E = 100 Gpa. We are interested in the processes that oc- cur on the surface of the small Solar System bodies, where the interactions among the boulders occur at velocities com- parable to the escape velocity on their surface. The upper x-axis indicates the radius of the body, while the lower one shows the corresponding escape velocity (assuming a con- stant density of ρ = 3000 gr/cm3). For km-size asteroids and m-size boulders, the collisions typically last for a few 10−3s, therefore, the time step required to correctly simu- late the collision would be ∼ 10−4s. 2.1 Viscolelastic spheres with friction 2 DISCRETE ELEMENT METHODS DEM are a set of numerical calculations based on statisti- cal mechanic methods. This technique is used to describe The contact force between two spherical particles can be de- composed in two vectors (Figure 2): the normal force, along the direction that joins the centres of the interacting par- ticles; and the tangential force, perpendicular to this line. 10−1 100 100 Body radius (km) 102 101 103 r = 10 m r = 1 m r = 0.1 m ) s ( n o s i i l l o c f o n o i t a r u D 10−1 10−2 10−3 10−4 10−1 100 100 10−1 10−2 10−3 101 102 Collisional velocity (m/s) 103 10−4 104 Figure 1. Estimate of the duration of collision as a function of the collisional velocity for particles of r = 0.1, 1 and 10 m . Figure 2. Scheme of the contact forces between two spherical particles Naming i and j the two interacting particles, the total force −→Fij can then be expressed as: −→Fij =(cid:26) −→F n ij + −→F t ij 0 if ψij > 0 otherwise (2) where −→F n ij is the normal force and −→F t ij the tangential one. ψij is the deformation given by: ψij = Ri + Rj − −→ri − −→rj (3) where Ri and Rj are the radius of the particle i and j, respectively, and −→ri and −→rj are the position vectors. Several models have been used for the normal and tan- gential forces in the literature. Among the most used ones is the damped dash-pot, also known as the Kelvin-Voigt model. Instead of using this model in our simulation, we use an extension of an elastic-spheres one developed by Hertz (1882), since it is a more realistic representation of two col- liding particles. The normal interaction force between two elastic ij was inferred by Hertz as a function of the spheres F n;el deformation ψ: F n;el = 2YpRef f 3(1 − ν 2) ψ3/2 (4) where Y is the Young modulus and ν is the Poisson ratio. The effective radius Ref f is given by the expression: 1 Ref f = 1 Ri + 1 Rj (5) Granular physics in low-gravity environments 3 A viscoelastic interaction between the particles can be modelled by including a dissipation factor in eq. 4. The vis- coelastic normal forces F n;ve then become: (6) F n;ve = 2YpRef f 3(1 − ν 2) (cid:16)ψ3/2 + Apψ dψ dt(cid:17) where A is a dissipative constant and dψ/dt is the time derivative of the deformation. Considering the previous expression for the normal force could lead to unrealistic results, since it does not take into account the fact that the particles do not overlap, but they become deformed (Poschel and Schwager (2005)). During the compression phase and most of the decompression phase, the term (cid:0)ψ3/2 + A√ψ dψ dt(cid:1) in eq. 6 is positive, leading to a repulsive (positive) normal force. However, at a certain stage of the decompression, the deformation ψ could still be pos- itive, but the second term could be negative, which would lead to a negative (attractive) force. This is an unrealistic situation, since there are no attractive forces during the col- lision of two particles. The problem arises when the centres of the particles separate too fast from one another to allow their surfaces to keep in touch while recovering their shape. In order to overcome this problem, for the condition ψ > 0 in eq. 2, we use the following expression for the normal force: F n;ve = max(0, 2YpRef f 3(1 − ν 2) (cid:16)ψ3/2 + Apψ dψ dt(cid:17)) (7) Following the model by Cundall and Strak (1979) for the tangential force (F t), when two particles first touch, a shear spring is created at the contact point. The static friction is then modelled as a spring acting in a direction tangential to the contact plane. The particles start sliding with the shear spring resisting the motion. When the shear force exceeds the normal force multiplied by the friction co- efficient, dynamic sliding starts. We limit the shear force by Coulomb's friction law; i.e. F t 6 µF n;ve. The expression for the tangential force then becomes: F t = −sign(vt rel) min{kκςk , µkF n;vek} (8) where the first term inside the curly brackets corre- sponds to the static friction, and the second one is the dy- namic friction. κ is a constant, ς is the elongation of the spring, and µ is the dynamic friction parameter. 2.2 ESyS-particle package (Abe et al. For the DEM simulations we developed a version of the ESyS-particle (2004); https://launchpad.net/esys-particle) adapted to our needs. ESyS-particle is an Open Source software for particle-based numerical modeling, designed for execution on parallel supercomputers, clusters or multi-core computers running a Linux-based operating system. The C++ simulation engine implements a spatial domain decomposition for parallel programming via the Message Passing Interface (MPI). A Python wrapper API provides flexibility in the design of numerical models, specification of modeling parameters and contact logic, and analysis of simulation data. The separation of the pre-processing, simulation and post-processing tools facilitates the ESyS-particle develop- ment and maintenance. The setup of the model geometry is 4 G. Tancredi et al. given by scripts, since the whole package is script driven (no interactive GUI is provided by ESyS). The particles can be either rotational or non-rotational spheres. The material properties of the simulated solids can be elastic, viscoelastic, brittle or frictional. Particles can be bonded to other particles in order to simulate break- able material. It is possible to implement triangular meshes for specifying boundary conditions and walls. The package also includes a variety of particle-particle and particle-wall interaction laws; such as linear elastic repulsion between un- bounded contacting particles, linear elastic bonds between bonded particle pairs, non-rotational and rotational fric- tional interactions between unbounded particles, rotational bonds implementing torsion and bending stiffness and nor- mal and shear stiffness. Boundary conditions and walls can move according to pre-defined laws. The DEM implementation in ESyS-particle employs the explicit integration approach, i.e. the calculation of the state of the model at a given time only considers data from the state of the model at earlier times. Although the explicit approach requires shorter time steps, it is easier to develop a parallel version for execution in high performance computing infrastructures. ESyS-particle has shown good scaling performance when using additional computing elements (processor cores), if at least ∼ 5000 particles are processed by each core. Oth- erwise, when a lower number of particles is handled by each core, the impact of the overhead by the communications be- tween processes reduces the computational efficiency of the application. As long as the problem size is scaled with the number of cores, the scalability is close to linear. Therefore, large amounts of particles, typically a few million, are pos- sible to model. For the analysis of the results, ESyS-particle can for- mat the output to be used in 3D visualisation platforms like VTK and POV-Ray. In particular, we use the software Par- aview, based on VTK and developed by Kitware Inc. and Sandia National Labs (EEUU), which offers good quality in 3D graphics and allows us to implement several visualisation filters to the data. ESyS-particle has been employed to simulate earth- quake nucleation, comminution in shear cells, silo flow, rock fragmentation, and fault gouge evolution, to name but a few applications. Just to give a few references, we mention exam- ples in fracture mechanics (Schopfer et al. (2009)), fault me- chanics (Abe and Mair (2005), Mair and Abe (2008)), and fault rupture propagation (Abe et al. (2006)). For our simulations, we have implemented the Hertzian viscoelastic interaction model with and without friction into the ESyS-particle package, according to the eqs. 7 and 8. Several modifications were necessary to implement ei- ther in the C++ code as well as in the Python interface (Heredia and Richeri (2009)). 2.3 Tests In order to test the code and to set the values of the rele- vant physical and simulation parameters, we choose a few problems for which there exists an analytical solution or we can compare the output with experiments. 2.3.1 Test case 1: a direct collision of two equal spheres We consider the case of two equal viscoelastic spheres: one starts at rest and the other one approaches from the nega- tive x-direction at a given speed along the line joining the particle centres. Friction is not considered, since the colli- sion between the particles is normal. The aim of this test is to study the viscoelastic collision. The coefficient of restitution can be used to characterise the change of relative velocity of inelastically colliding parti- cles. Let us note −→v1 and −→v2 the velocities before the collision of particles 1 and 2, respectively; and −→v′ 2 the veloc- ities immediately after. When the relative velocity is along 1 and −→v′ the line joining the particle centres, we note v = −→v2 − −→v1 and v′ = −→v′ 1. The coefficient of restitution ǫ is then 2 − −→v′ calculated as: ǫ = v′ v (9) In general, this coefficient depends not only on the im- pact velocity, but also on material properties. Because of their deformation, particles lose contact slightly before the distance of the centres between the spheres reaches the sum of the radii. Schwager and Poschel (2008) present an analytical estimate of the coefficient of restitution which takes into account this fact. The computa- tion of ǫ is then presented as a divergent series of the dimen- sionless parameter βv1/5, where β = γκ−3/5; γ = 3 ; 2 + 1 Rj 3(1−ν 2)√(Ref f . The material parameters Y , ν and A mef f are defined above. R1, R2, m1, m2 are the radius and mass of particle 1 and 2, respectively. ρA mef f ; and mef f = 1 mi = 1 Ri + 1 mj ; 1 Ref f 2Y κ = δ ; δ = 1 We run simulations of two colliding particles with the following combination of parameters: Y = {109, 1010} P a, A = {10−4, 10−3} s−1, ν = 0.3; for a set of initial relative velocities v = {0.1, 0.5, 1, 5, 10} m/s. The particles have a radii of 1m and a density of 3000 kg/m3. The time step of the integration is 10−5 s. The coefficient of restitution for the numerical simula- tions is presented in Figure 3a as a function of βv1/5. The symbols correspond to different combinations of parameters: circle – Y = 109, A = 10−4; down triangle – Y = 109, A = 10−3; square – Y = 1010, A = 10−4; up triangle – Y = 1010, A = 10−3 (Y in P a and A in s−1). The analyt- ical estimates are computed with Maple's codes presented in Schwager and Poschel (2008), where the expansions are up to 40th order. The ratio between the numerical and an- alytical estimate is presented in Figure 3b. We find a good agreement between the two estimates up to values of βv1/5 closer to 1. The discrepancy is due to the cut-off of the higher terms. Several laboratory experiments have been conducted to estimate the coefficient of restitution of rock materials (see e.g. Imre et al. (2008), Durda et al. (2011)). For impact ve- locities in the range 1 − 2 m/s, values of ǫ ∼ 0.8 − 0.9 have been obtained. Looking back at Figure 3a, we observe that this range of values of ǫ are obtained for the following set of material parameters: Y = 1010 P a, A = 10−3 s−1, ν = 0.3. Therefore, we will choose these parameter values for our numerical simulations of colliding rocky spheres. a . t s e R . f e o C 1 0.8 0.6 0.4 0.2 0 −10 10 −5 10 β5 Velocity b . r o e T / . l i u m S o 1.5 1 i t a R 0 10 0.5 −10 10 −5 10 β5 Velocity Figure 3. a) The coefficient of restitution for the numerical sim- ulations of the collision of two viscoelastic spheres as a function of βv1/5. The symbols correspond to different combinations of pa- rameters: circle – Y = 109, A = 10−4; down triangle – Y = 109, A = 10−3; square – Y = 1010, A = 10−4; up triangle – Y = 1010, A = 10−3 (Y in P a and A in s−1). b) The ratio between the numerical and analytical estimate. 2.3.2 Test case 2: a grazing collision between two spheres In this case we consider two equal viscoelastic spheres: one starts at rest and the other one approaches from the negative x-direction at a given speed; but, in contrast to the previ- ous case, the distance between the y-values of the particles centres is slightly less than the sum of the radius. We then have a grazing collision. The aim of this test is to compare the results of the viscoelastic interaction with and without friction. We run simulations of two colliding particles with the following set of parameters: Y = 1010 P a, A = 10−3 s−1, ν = 0.3, R1 = R2 = 1 m, and a density of ρ = 3000 kg/m3. The friction parameters of eq. 8 are chosen as: κ = 0.4, µ = 0.6. The distance between the centres in the y-direction is (0.999R1 + R2). We run simulations where particle 2 has initial velocities of v = {10−3, 0.01, 0.1, 1, 10} m/s. In Figure 4 we plot the ratio between the modulus of the particles relative velocity after exiting and before the interaction, as a function of the initial velocity. The star symbols correspond to the simulations without the friction interaction and the cross symbols to the ones with it. Due to the fact that the collision is almost grazing, the ratios are almost 1 for the simulations without friction, regardless of the initial velocity. For simulations with friction, as it is expected, the ratio decreases as the initial velocity decreases, because the friction interaction becomes more relevant for low velocities. 2.3.3 Test case 3: a bouncing ball In ESyS-particle the interaction between a particle and a mesh wall can be linear, elastic or a linear elastic bond. Viscoelastic and frictional interactions of particles and walls are not yet implemented. Therefore, in order to simulate a frictional viscoelastic collision of a ball against a fixed wall, we have to glue balls to the wall with a linear elastic bond. The following test case consists on a free-falling ball impacting on an equal size ball that is bonded to the floor. The objective of this experiment is to test different time steps for the simulations. We use the following set of parameters: Y = 1010 P a, A = 10−3 s−1, ν = 0.3, R1 = R2 = 1 m, and a density of ρ = 3000 kg/m3. The bonded particle has an elastic bond with a modulus K = 109 P a. Particle 2 falls from a height of Granular physics in low-gravity environments 5 . l e V l a i t i n i I / l a n F o i t a R 1.1 1.05 1 0.95 0.9 10−4 0 10 10−2 100 Initial Velocity (m/s) Figure 4. The ratio between the modulus of the relative velocity between the particles after exiting and before the interaction as a function of the initial velocity. The symbols correspond to differ- ent contact force models: square – Hertzian viscoelastic spheres; circle – Hertzian viscoelastic spheres with friction. a ) m ( t h g i e H 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 0 b 6´ 10−4 5´ 10−4 10−4 10−5 10−6 5´ 10−7 7 − 0 1 ´ 5 = t d r o f t h g i e H / t i h g e H o 6´ 10−4 5´ 10−4 10−4 10−5 10−6 1.1 1.05 1 0.95 5´ 10−4 10−4 10−5 10−6 6´ 10−4 10−6 10−5 10−4 5´ 10−4 6´ 10−4 The rest 5´ 10−4 6´ 10−4 i t a R 0.9 0.2 0.4 0.6 Time (s) 0.8 1 0.85 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 Time (s) Figure 5. a) The distance of the falling particle respect to the edge of the resting one (centre height minus 3R) as a function of time. b) The ratio of the previous values to the one for the smallest time step at each time. 2.75m. In the first set of simulations we assume the Earth's surface gravity (g = 9.81 m/s2). For this set, we use the following time steps in the simulations: dt = {6 × 10−4, 5 × 10−4, 10−4, 10−5, 10−6, 5 × 10−7} s. The duration of the collisions is computed from the sim- ulations as the interval of time while the deformation param- eter defined in eq. 3 is greater than 0. As mentioned above this interval is slightly larger than the time the balls are in contact, but it is good enough for the purpose of having an order of magnitude estimate of it. For the previous set of parameters, the duration of the collision is ∼ 0.003 s. In Figure 5a, we plot the distance of the falling par- ticle respect to the edge of the resting one (centre height minus 3R) as a function of time. In Figure 5b, we plot the ratio of the previous values to the one for the smallest time step at each time. We find that for time steps dt 6 10−5 s, there is a very good agreement between the simulations. For longer time steps the bouncing ball presents an implausible behavior. The coefficient of restitution can be computed as the ratio between the velocity at the iteration step just after the collision and at the step just before the collision (just after and before the deformation defined in eq. 3 is ψ < 0). For time steps dt 6 10−5 s there is a good agreement among the different estimates. We obtained a value of 0.593. 6 G. Tancredi et al. Therefore, for the previous set of parameters, we will use a time step of dt = 10−5 s for the simulations with Earth's gravity, since the collision is covered with ∼30 time steps and it is a good compromise between quality of the results and a longer time step. In another set of simulations we use very low surface gravity, similar to the one found on the surface of aster- oid Itokawa and comet P/Hartley 2; i.e. a rocky object of ∼ 500 m in diameter or an icy object of ∼ 1 km in di- ameter. For this set, we use the following time steps in the simulations: dt = {5 × 10−4, 10−4, 10−5, 10−6} s. For the previous set of parameters, the duration of the collision is ∼ 0.01 s. For time steps dt 6 10−4 s there is a good agree- ment among the different runs. The coefficient of restitution in these simulations is 0.721. For the simulations in this low- gravity environments we will use a time step dt = 10−4 s, which corresponds to a collision lasting ∼ 100 time steps. 2.3.4 Test case 4: Newton's cradle A Newton's cradle is a device used to demonstrate the con- servation of linear momentum and energy via a series of swinging hard spheres. When one ball at the end is lifted and released, it knocks a second ball and this one the next until the last ball in the line is pushed upward. A typical Newton's cradle consists of a series of identically sized metal balls hanging by equal length strings from a metal frame so that they are just touching each other at rest. We simulate the Newton's cradle with four spheres aligned in the x-axis. We number the particles from right to left: #1 being the particle at the right extreme and #4 the one at the left extreme. The x-axis increases to the right. Particle #1 has a negative initial velocity vx = −10 m/s. Two types of simulation are run: Hertzian elastic and vis- coelastic spheres. We use the following set of parameters: Y = 1010 P a, A = 10−3 s−1, ν = 0.3 (for the viscoelastic simulation). The radius of the spheres are R = 1 m, and a density of ρ = 3000 kg/m3. ii) x-velocity for each particle; In Figure 6 we present the time evolution of the fol- lowing parameters for each simulation: i) x-position of each particle (#1 to #4); iii) relative change of x-total momentum: (M omentum(t) − M omentum(t = 0))/M omentum(t = 0); iv) relative change of total kinetic energy: (K.E(t)−K.E.(t = 0))/K.E.(t = 0). Figure 6 a) corresponds to the Herztian elastic (HE) simu- lation, and Figure 6 b) to the Herztian viscoelastic (HVE) one. Note that for the HE simulation particle #4 acquires almost the velocity of the initial impacting particle and lit- tle rebound is observed in the particles #1 to #3. The linear momentum is conserved after the collision up to a relative precision < 10−12, and the kinetic energy after the rebound is conserved up to a relative precision of 10−11. In the HVE simulation, the particle #4 acquires 70% of the velocity of the initial impacting particle, and particle #3 acquires 25%. No rebound is observed and all the particles move to the left. The final velocities increase from right to left. The lin- ear momentum is also conserved after the collision up to a relative precision < 10−12 (down to the last output digit). The kinetic energy after the rebound is not conserved ∼ 50% of the initial kinetic energy is spent on the damping of the viscosity interaction. i ) m ( n o i t i s o p − x 4 2 0 −2 −4 0 #4 #3 #2 #1 #1 #2 #3 #4 ii ) s / m ( y t i c o l e v − x 0 −5 −10 #1 #2 #3 #4 a #4 #3 #2 #1 0.05 0.1 0.15 0 0.05 0.1 0.15 iii Time (s) −12 x 10 2 1 0 0.05 0.1 Time (s) m u t n e m o m − x e g n a h c . l e R i ) m ( n o i t i s o p − x −1 −2 0 4 2 0 −2 −4 0 #4 #3 #2 #1 m u t n e m o m − x e g n a h c . l e R 1 0.5 0 −0.5 −1 0 Time (s) −11 x 10 4 2 0 −2 −4 0 0.05 0.1 0.15 Time (s) iv y g r e n e c i t e n i k e g n a h c . l e R ii b #4 #3 #2 #1 #1 #2 #3 #4 ) s / m ( y t i c o l e v − x 0 −5 −10 #1 #2 #3 #4 Time (s) iv y g r e n e c i t e n i k e g n a h c . l e R 1 0.5 0 −0.5 −1 0.05 0.1 0.15 0 0.05 0.1 0.15 iii −12 x 10 Time (s) 0.05 0.1 Time (s) 0 0.05 0.1 0.15 Time (s) Figure 6. a) Results of the Herztian elastic (HE) simulations of the Newton's cradle: i) x-position of each particle (#1 to #4); ii) x-velocity for each particle; iii) relative change of x-total momen- tum; iv) relative change of the total kinetic energy. b) Similar set of plots for the Herztian viscoelastic (HVE) simulations of the Newton's cradle. 3 SIZE SEGREGATION IN LOW-GRAVITY ENVIRONMENTS: THE BRAZIL NUT EFFECT 3.1 The shaking or knocking procedure Consider a recipient with one large ball on the bottom and a number of smaller ones on top of it. All the balls have similar densities. After shaking the recipient for a while, the large ball rises to the top and the small ones sink to the bottom (Rosato et al. (1987), Knight et al. (1993), Kudrolli (2004)). This is the so called Brazil nut effect (BNE), because it can be easily seen when one mixes nuts of different sizes in a can; the large Brazil nuts rise to the top of the can. Unless there is a large difference in the density of the balls, a mixture of different particles will segregate by size when shaken. Granular physics in low-gravity environments 7 lations of the BNE, we run a set of simulations where the particles start at a certain height over the surface and they free fall. The floor is slightly shaken at the beginning of these preliminary simulations in order to obtain a random settling of the particles. After finishing the shaking and let- ting the particles settle down, we use the positions at the end of the runs as the initial conditions for the set of BNE simulations. We must run different preliminary simulations for each gravity environment. In the BNE simulations, the floor's velocity is linearly increased from 0 up to the final value vf loor, which is reached after 20 jumps. We note that the shaking procedure is pa- rameterized with the floor's velocity. The 3D box is constructed with elastic mesh walls. The box has a base of 6 × 6 m and a height of 150 m. A set of 12× 12 m small balls of radius R1 = 0.25 m are glued to the floor. The big ball has a radius R2 = 0.75 m, and on top of it, there are 1000 small balls with a normal distribution of radii (mean radius R1 = 0.25 m, standard deviation σ = 0.01 m). We use the same box for all the simulations. The size range of the balls are selected in correspon- dence with the boulders size observed on the surface of as- teroid Itokawa and Eros. 3.2 Earth We run simulations with the following set of floor veloci- ties: vf loor = {0.3, 1, 3, 5, 10} m/s. Snapshots at start and after 100 shakes (100 sec. of simulated time) are presented in Figure 8. The snapshots correspond to the simulation with floor's velocity vf loor = 5 m/s. In the supplementary material we include movies with the complete simulation (movie1 with all the spheres drawn and movie2 with the small spheres erased). In Figure 9 we present the evolution of the big ball's height as a function of the number of shakes for the differ- ent floor velocities. The thick black line marks the height of a box enclosing the 1000 small particles with a random close packing. Random close packing has a maximum poros- ity of P = 0.64 (Jaeger and Nagel (1992)). The volume of the enclosing box is calculated as the sum of the volume of the 1000 small particles divided by the porosity; i.e.: V = 1000( 4/3 1)/P = 102 m3. For a box with a 6 × 6 m base, we obtain a height of the enclosing box of 2.84 m. The thick black line is drawn at this height. p iR3 For the two lowest velocities (vf loor = {0.3, 1} m/s) the big ball stays at the bottom, for the two largest ones (vf loor = {5, 10} m/s) it rises to the top, and for the inter- mediate one (vf loor = 3 m/s) it starts rising but does not reach the top at the end of the simulation. When the floor's displacement velocity is below ∼ 3 m/s, the Brazil nut effect does not occur. Above this threshold, the time required by the big ball to reach the top decreases for increasing floor velocities. Note that there is a sharp decrease in the rising time for small changes in the floor's velocity (from 3 to 5 m/s). For large displacement velocities, the balls on the top, including the big one that is 27 times more massive than the small ones, can be lifted at considerable heights, as it is seen in the large excursions made by the big ball for vf loor = 10 m/s. Figure 7. The floor is vertically displaced at a certain speed (vf loor) for a short interval (dtshake), according to a staircase- like function. The BNE has been attributed to the following processes (Hong et al. (2001)): i) the percolation effect, where the smaller ones pass through the holes created by the larger ones (Jullien and Meakin (1992)); ii) geometrical reorgani- sation, through which small particles readily fill small open- ings below the large particles (Rosato et al. (1987)); iii) global convection which brings the large particles up but does not allow for reentry in the downstream (Knight et al. (1993)); iv) due to its larger kinetic energy, the large par- ticle still follows a ballistic upraise, penetrating by inertia into the bed (Nahmad-Molinari et al. (2003)). While size ratio is a dominant factor, particle-specific properties such as density, inelasticity and friction can also play important roles. Williams (1963) performed a model experiment with a single large particle (intruder) and a set of smaller beads inside a rectangular container. When the container was vi- brated appropriately, the intruder would always rise and reach a height in the bed that depends on vibration strength. In order to simulate this effect under different gravity conditions, we run simulations of a 3D box with many small particles and one big particle at the bottom, the so-called intruder model system (Williams (1963), Kudrolli (2004)). On the floor we glue one row of small particles with a linear elastic bond. The box is subjected to a given surface gravity. We run simulations under several gravity conditions: the surface of the Earth, Moon, Ceres, Eros and a very-low gravity environment like the surface of asteroid Itokawa or comet P/Hartley 2. The parameters for the simulations are summarised in Table 1. The physical and elastic parameters of the particles are similar to the ones used in the previous tests: Y = 1010 P a, A = 10−3 s−1, ν = 0.3, κ = 0.4, µ = 0.6, K = 109 P a, ρ = 3000 kg/m3. The floor is vertically displaced at a certain speed (vf loor) for a short interval (dtshake), according to a staircase-like function like the one presented in Fig 7. The process is repeated every given number of seconds (∆trep), depending on the settling time given by the surface gravity. We have chosen this vibration scheme instead of the fre- quently used sinusoidal oscillation of the floor, because we are interested in the effects of a sudden shock coming from below. This shock could arise from the translation of the im- pulse generated by an impact in a far region. We refer this vibration scheme as a shaking or knocking procedure. In order to prepare the initial conditions for the simu- 8 G. Tancredi et al. Table 1. Parameters for the simulations of the BNE under different gravity environments. Parameter Earth Moon Ceres Eros Surface gravity g (m/s2) Escape velocity vesc (m/s) Floor's velocity vf loor (m/s) Duration of displacement dtshake (s) Time between displacements ∆rep (s) 9.81 11.2 × 103 0.3 - 10 0.1 2 1.62 2.38 × 103 0.1 - 3 0.1 5 0.27 510 0.03 - 1 0.1 5 5.9 × 10−3 10 0.01 - 0.3 0.1 15 a) Moon 10 ) m ( t h g i e H 9 8 7 6 5 4 3 2 1 0 0 3 1 c) Eros 10 vel (m/s) 0.1 0.3 0.5 0.8 1 3 0.8 0.5 0.3 0.1 50 100 150 200 Number of shakes vel (m/s) 0.01 0.03 0.05 0.1 0.3 0.3 0.1 0.05 0.03 0.01 ) m ( t h g i e H 9 8 7 6 5 4 3 2 1 0 0 Low-gravity Itokawa & P/Hartley 2 10−4 0.17 0.003 - 0.1 0.1 15 b) Ceres ) m ( t h g i e H 10 9 8 7 6 5 4 3 2 1 0 0 vel (m/s) 0.03 0.1 0.3 0.5 1 1 0.5 0.3 0.1 0.03 50 100 150 200 250 300 350 400 Number of shakes d) Itokawa 8 7 6 5 4 3 2 1 0 ) m ( t h g i e H vel (m/s) 0.003 0.01 0.02 0.03 0.05 0.1 0.1 0.05 0.03 0.02 0.01 0.003 Figure 8. Snapshots at start and after 50 shakes (100 sec. of sim- ulated time) for the simulation under Earth's gravity. The snap- shots correspond to the simulation with floor's velocity vf loor = 5 m/s. See the movies in the supplementary material. ) m ( t h g e H i 10 9 8 7 6 5 4 3 2 1 0 0 0.3 1 3 5 10 10 10 1 0.3 40 50 5 3 20 30 Number of shakes Figure 9. The evolution of the big ball's height as a function of the number of shakes for different floor velocities (vf loor = {0.3, 1, 3, 5, 10} m/s) under Earth's gravity. Note that the floor's velocity is used as the varying parameter in the shaking process. A black line is drawn at a height of 2.84 m, which is the height of a compact enclosing box (see text). 100 200 300 400 500 Number of shakes 0 100 200 300 400 500 600 700 Number of shakes Figure 10. The evolution of the big ball's height as a function of the number of shakes for different floor velocities under different gravity environments: a) Moon; b) Ceres; c) Eros; d) Itokawa. The legends correspond to the floor velocities (vf loor). 3.3 Comparison with other gravity environments Similar simulations were run for other gravity environments, like the surface of the Moon, Ceres, Eros and a very-low gravity environment like the surface of asteroid Itokawa or comet P/Hartley 2. The simulation parameters are pre- sented in Table 1. For the simulation under the very-low gravity environ- ment, we present a movie of 4500 sec. of simulated time (300 shakes) in the supplementary material. The movie cor- responds to the simulation with floor's velocity vf loor = 0.05 m/s (movie3 with all the spheres drawn and movie4 with the small spheres erased). Figure 10 presents the evolution of the big ball's height as a function of the number of shakes for the different floor velocities and the different gravity environments: a) Moon, b) Ceres, c) Eros, d) Itokawa. As in the cases of the simulations in Earth's gravity, in all the different gravity environments we can find a thresh- old for the floor's velocity, below which the Brazil nut effect does not occur. From the previous plots, we get a rough es- timate of these thresholds. In Figure 11 we plot the velocity thresholds as a function of the surface gravity in a log-log 101 100 10−1 10−2 ) s / m l ( d o h s e r h t y t i c o l e v g n i s i R Granular physics in low-gravity environments 9 Rising velocity threshold Surface escape velocity log10 vthre = 0.42 log10 g + 0.05 10−4 10−2 Gravity (m/s2) 100 Figure 12. Snapshots of the initial and final state (after 1300 shakes) for a simulation under the low-gravity environment and a floor velocity of vf loor = 0.05 m/s. (see movie5 and movie6 in the supplementary material. Figure 11. Comparison of the floor's velocity threshold for the different gravity environments. The Brazil nut effect does not oc- cur if the floor's velocity is below the threshold. The floor's ve- locity thresholds are presented as small circles. Note that the thresholds are not precisely estimate, because they are computed from the plots in Figure 9 and 10 a-d. A straight line in the log-log space is fitted to the data points. The up triangles represent the escape velocity for the given surface gravity. The escape velocity for the largest objects are out of the plots. scale. A straight line in the log-log space is a good fit to the data points: log10 vthre [ m/s] = 0.42 log10 g(cid:2) m/s2(cid:3) + 0.05 We conclude that the Brazil nut effect is effective in a wide range of gravity environments, expanding 5 orders of magnitude on surface gravity. (10) In Figure 11, we plot the escape velocity for the given surface gravity. Note that the floor's velocity thresholds ap- proach the escape velocity for the low gravity environments. For example, in the case of Itokawa, the escape velocity is vesc = 0.17 m/s, while the estimated floor's velocity thresh- old is vf loor = 0.015 m/s. This point is revisited in Section 5. 4 DENSITY SEGREGATION IN LOW-GRAVITY ENVIRONMENTS As mentioned above, other particle-specific properties can affect the segregation process. In particular, the effects of density have been studied the most. For ratios of the den- sity of the large to the small particles much larger than 1 (denser large particles), the segregation effect could be re- versed, and the large particles would sinks to the bottom, producing the so-called Reverse Brazil Nut Effect (RBNE) (Shinbrot and Muzzio (1998), Hong et al. (2001)). However, for particles of similar sizes but different densities, both laboratory (Mobius et al. (2001), Shi et al. (2007)) or numerical (Lim (2010)) experiments have shown that the lighter particles tend to rise and form a pure layer on the top of the system, while the heavier particles and some of the lighter ones stay at the bottom and form a mixed layer. In the Solar System, we might encounter bodies with such a mixture of heavy and light particles. Cometary nuclei are believed to be formed of a mix of icy and rocky material. However, the intimacy of this mixture is still unknown, with two possible scenarios: 1) every particle is made of a mix- ture of ice and dust, and 2) there exist some particles mainly formed by icy material and some others mainly formed by rocky constituents that are mixed together. We shall investigate the behavior of a mixture of light and heavy particles under different gravity environments. For the simulations we create a 3D box similar to the previous one, with a 6 × 6 m base and a height of 150 m. The box is constructed with elastic mesh walls. On the floor we glue a set of 12 × 12 small balls of radius R1 = 0.25 m and density ρ = 2000 kg/m3. There are 500 light balls with a normal distribution of radii (mean radius R1 = 0.25 m, standard deviation σ = 0.01 m) and density ρ = 500 kg/m3. On top of them, there are 500 heavy balls with the same dis- tribution of radii and density ρ = 2000 kg/m3. At the be- ginning of the simulations the balls are placed sparsely, the light balls at the bottom and the heavy ones on top. They free fall and settle down before starting the floor shaking. Elastic parameters of the particles are the same for both types of particles and similar to the ones used in the previous tests for all the particles: Y = 1010 P a, A = 10−3 s−1, ν = 0.3, κ = 0.4, µ = 0.6, K = 109 P a. The floor is displaced with a staircase function in a sim- ilar way as in the previous set of simulations. Two gravity environments were tested: the Earth's sur- face gravity and a very-low gravity environment like the sur- face of comet P/Hartley 2. Figure 12 presents snapshots of the initial and final state (after 1300 shakes) for a simulation under the low-gravity environment and a floor velocity of vf loor = 0.05 m/s. In the supplementary material we include movies with the com- plete simulations (movie5 corresponds to the simulation un- der Earth's gravity and vf loor = 3 m/s; movie6 corresponds to the simulation under low gravity and vf loor = 0.05 m/s. Note that in these movies the camera moves with the floor, therefore it seems that the floor is always located in the same position, but it really is moving with the staircase function described above). At every snapshot, we compute the median height of the light and heavy particles, respectively. These median heights are plotted as a function of the number of shakes in Figure 13 a) for the Earth's gravity simulations, and b) for the low- gravity ones. For each simulation there are two lines: the one that starts on top corresponds to the heavy particles and the one that starts at the bottom to the light ones. In the Earth environment simulations, the lines do not cross for the two 10 G. Tancredi et al. a ) m ( t i h g e H 6 5 4 3 2 1 0 1 3 5 1h 3h 5h 5l 3l 1l 50 100 150 200 250 300 350 400 450 Number of shakes b ) m ( t i h g e H 6 5 4 3 2 1 0 0.03 0.05 0.1 0.1l 0.05l 0.03l 0.1h 0.03h 0.05h 200 400 600 800 1000 1200 Number of shakes Figure 13. The median height of the light and heavy particles are plotted as a function of the number of shakes for different floor velocities and under different gravity environments: a) Earth, b) low-gravity like P/Hartley 2. The legends correspond to the floor velocities (vf loor). For each simulation there are two lines: the one that starts on top corresponds to the medium height of the heavy particles (labeled with h) and the one that starts at the bottom to the medium height of the light ones (labeled with l) lowest floor velocities: vf loor = {1, 3} m/s; therefore, the particles do not overturn the initial segregation. Though, for vf loor = 3 m/s, the lines start to approach. However, for the highest floor velocities, i.e. vf loor = 5 m/s, the lines cross at an early stage of the simulation after which they remain almost parallel. Most of the light particles move to the top and most of the heavy ones sink to the bottom; the end state is similar to the one seen in Figure 12 for the low- gravity simulations. Due to the strong shakes, the particles suffer large displacements, but, in a statistical sense, the two set of particles are segregated. A density segregation is then observed, although it is not complete. The results of the simulations under a low-gravity envi- ronment are presented in Figure 13b. The lines for the light and heavy particles median height do cross for the three studied floor velocities (v f loor = {0.03, 0.05, 0.1} m/s), although for the lowest velocity the simulations do not last long enough to reach the stable stage where the median heights reach almost a stable value. Note that in both gravity environments, the density seg- regation is effective for floor's velocity over a threshold sim- ilar to the ones of the size segregation effect of Section 3. 5 PARTICLE LIFTING AND EJECTION Let us consider the following simple experiment: we have a layer of material that is uniformly shocked from the bot- tom. The motivation of this experiment is to consider what would happen if a seismic wave, generated somewhere in a body and propagating through it, reaches another region of the body from below. What would happen with material deposit on the surface? Let us take into account three dif- ferent materials: a solid block, a compressible fluid and a set of grains. The outcome of the experiment will be different depending on the material. When the seismic wave knocks the solid block, the block is pushed upward. It starts to move upward, forming a gap between the layer's bottom and the floor. In the case of a layer of compressible fluid, an elas- tic p-wave is transmitted through it, producing compression and rarefaction of the material. But, what happens in the case of a layer of grains? Be- fore presenting the results of some simulations, let us re- consider the simulations of Newton's cradle with Hertzian viscoelastic spheres. We have seen that after the first par- ticle knocks the second one from the right, all the particles move to the left. Particle #4, the last one on the row, moves faster, the next one to the right moves slower and so forth. Therefore, the whole set of particles move in the same direc- tion, but they do not do it as a compact set, the particles separate from each other. We perform a first set of simulations with a homoge- neous set of particles. A 3D box with a base of 7.5 × 7.5 m is filled with 15 × 15 = 225 particles glued to the bottom, with a radius R = 0.25 m. We create 2744 particles with a mean radius R1 = 0.25 m, standard deviation σ = 0.01 m and density ρ = 3000 kg/m3. To generate the initial condi- tions for the simulations, these particles are located a few cm from the bottom and they free fall under the different gravity environments until they settle down. Elastic parameters of all the particles are similar to the ones used in the previous tests for all the particles: Y = 1010 P a, A = 10−3 s−1, ν = 0.3, κ = 0.4, µ = 0.6, K = 109 P a. With the initial conditions generated above, we run the following experiment: after a given time (tsep), the floor is vertically displaced at a certain speed (vf loor) for a short interval (dtshake), only one time. Two gravity environments are used for the simulations: Earth's surface and the low- gravity environment of Itokawa. For the Earth's simulations we use the following set of parameters: tsep = 1 s, vf loor = {1, 3, 10} m/s, dtshake = 0.1 s. At every snapshot, we sort the particles by their height respect to the floor, and we compute the height of the particles at the 10% (h10) and 90% (h90) percentile. In Figure 14 we plot the difference of these two quantities (h90 − h10) for the different floor velocities. We observe that these differences increase with time up to a certain instant when the particles fall back. Therefore, the particles are not moving as a compact set, rather, the upper particles are moving faster and the particles separate from each other. The upper particles can reach velocities larger than the floor's velocity; e.g. in the case of vf loor = 10 m/s, the 10% fastest particles reach velocities of ∼ 17 m/s just after the end of the floor's displacement. We observe that the upper particles are lifted at considerable heights before they fall back. Similar results are obtained in low-gravity simula- tions, using the following set of parameters: tsep = 10 s, vf loor = {0.01, 0.03, 0.1} m/s, dtshake = 0.1 s. The up- per particles move faster and they can reach velocities up to ∼ 0.02, 0.05, 0.2 m/s with respect to the floor veloci- ties. Note that the escape velocity in this environment is vesc = 0.17 m/s, therefore the fastest ejection velocities of the lifted particles are higher than vesc. We run another experiment: on top of the layer of particles with mean ra- dius R1 ∼ 0.25 m, we deposit a layer of 2700 smaller parti- cles, with mean radius R2 = 0.1 m and standard deviation σ = 0.01 m. The rest of the physical parameters are the same as for the bigger particles. The aim of this experiment is to check whether the small particles are ejected with higher velocities than the big ones. As in the previous simulations, we order the particles in increasing height. We compute the height of the 90% percentile of the big (hb,90) and small par- ticles (hs,90). Although the small particles on top of the big ones tend to separate, the differences in the velocities are Granular physics in low-gravity environments 11 10 1 3 10 velocity could be comparable to the escape velocity at the surface. The particles could enter in sub-orbital or orbital flights, creating a cloud of gravitational weakly bounded par- ticles around the object. ) m ( t h g i e H . f f i D 20 18 16 14 12 10 8 6 4 0 3 1 1 2 3 Time (s) 4 5 6 Figure 14. Lifting of particles under Earth gravity. At every snapshot, we sort the particles by their height respect to the floor, and we compute the height of the 10% (h10) and 90% (h90) per- centile. We plot the difference of these two quantities (h90 − h10) for the different floor velocities. ) m ( t h g e H i 30 25 20 15 10 5 0 0 0.003 0.01 0.02 0.03 0.05 0.1 100 0.1 0.05 0.03 0.02 0.003 0.01 200 300 400 500 Number of shakes 600 700 Figure 15. The maximum height of the particles as a function of the simulated time for the case of the low-gravity environment and different floor velocities. relatively small. There is no significant ejection of the small particles. Another relevant result regarding the lifting and ejec- tion of particles from the surface due to an incoming shock from below, can be obtained from the Brazil nut effect sim- ulations presented in Section 3. In the animations produced with a sequence of snapshots for the simulations where the segregation process was effective, we observe many particles lifted at considerable heights. In Figure 15 we plot the max- imum height of the particles as a function of the simulated time for the case of the low-gravity environment and dif- ferent floor velocities. Note that the ejection velocities the fastest particles can acquire are comparable to the floor's displacement velocities, and even, a little bit higher. For a floor velocity of 0.1 m/s, the particles can reach an ejection velocity higher than the escape velocity at the surface. Taking into consideration the previous results, we con- clude that a layer shocked from below would produce the lift- ing of particles at the surface if the displacement of the bot- tom exceeds a certain velocity threshold. Particles can ac- quire vertical velocities comparable to the displacement ve- locity of the bottom. For very low-gravity environments, this 6 GLOBAL SHAKING DUE TO IMPACTS AND EXPLOSIONS In the previous sections we have shown that several physi- cal processes can occur in a layer of granular media when it is shocked from below: size and density segregation, lifting and ejection of particles. A big quake in a distant point could produce such a shock. The quake could be produced by an- other small object impacting the body or by the release of some internal stress. Interplanetary impacts typically occur at velocities of several km/s. These are hypervelocity im- pacts, i.e. impacts with velocities that are above the sound speed in the target material, which give rise to physical de- formation of the target, heating and shock waves spreading out from the impact point. The DEM algorithms described above can not successfully reproduce these set of phenom- ena. Therefore, we have to implement a different approach if we are interested in understanding the effect of an impact induced shock wave passing through a granular media. Let's consider a km-size agglomerated body, formed by many m-size boulders. We raise the following question: what happens if a small projectile impacts in such an object at distances far from the impact point? Or alternatively, what happens if a large amount of kinetic energy is released in a small volume close to the surface of such an object? In order to answer these questions we run the following set of simulations. We fill a sphere of radius 250 and 1000 m with small spheres of a given size range, using the configurations, number of moving particles, and total mass listed in Table 2. For each sphere, we fill the volume with two different dis- tributions of small spheres: one with ∼ 90, 000 particles and another one with a larger number of particles ∼ 700, 000. We try to make the total mass of the moving particles similar for each of the studied radii. A time step of dt = 10−4 s is used in all the simulations. The simulations are run in a cluster with Intel Xeon multi- core processors (Model E5410, at 2.33 GHz, with 12MB Cache). For cases B and D we use up to 8 cores. In these cases, a simulation of 10 s takes ∼ 20 hr of CPU-time in each core. Since we can not successfully simulate the physics of a hypervelocity impact during the very short initial stages, we implemented another approach. At a given point on the sur- face we select a certain number of particles of the body that are close to this place. Each particle has at the beginning of the simulation a velocity along the radial vector toward the centre. We substitute the impact by a near-surface un- derground explosion, where several particles are released at a given speed. For each set of configurations listed in Table 2, we run simulations with initial particle velocities of 100 m/s and 500 m/s. These velocities are well below the sound speed in the target material. Since these initial conditions would correspond to a stage after the impact where some energy has already been spent in the compression, fracturing and heating of the tar- get material, we cannot equal the sum of the kinetic en- 12 G. Tancredi et al. Table 2. Parameters for the simulations of underground explosions Case Parameter Size range of spheres (m) Number of particles Porosity Total Mass (1012kg) Escape velocity at surface (m/s) Number of initially moving particles Mass of moving particles (106kg) Energy-equivalent projectile radius (m) for v = 100m/s Energy-equivalent projectile radius (m) for v = 500m/s Momentum-equivalent projectile radius (m) for v = 100m/s Momentum-equivalent projectile radius (m) for v = 500m/s Ratio Kinetic Energy / Potential Energy for v = 100 m/s Ratio Kinetic Energy / Potential Energy for v = 500 m/s Specific energy Q∗ (J/kg) for v = 100 m/s Specific energy Q∗ (J/kg) for v = 500 m/s A B C D Radius Radius 250 m Radius 250 m 1000 m 1000 m Radius 2.5 - 12.5 88570 0.31 0.135 0.269 10 21 1 - 10 783552 0.22 0.152 0.285 140 21 9.88 2.57 3.26 5.53 36 907 0.79 20 0.88 2.56 3.23 5.52 29 710 0.69 17 10 - 50 89144 0.31 8.66 1.075 10 599 2.67 7.82 9.84 16.83 1 25 0.35 8.6 5 - 25 688443 0.31 8.63 1.073 200 608 2.68 7.85 9.89 16.92 1 26 0.35 8.8 ergy of the moving particles with the kinetic energy of the impactor. However, we can provide a lower limit to the ki- netic energy of the impactor by assuming efficiency factor ǫKE = 1, or a corresponding lower limit of the impactor size for a given impact velocity. In Table 2, we also present the radius of the equivalent projectile for the two set of ini- tial particle velocities, assuming an energy efficiency factor ǫKE = 1 and an impact velocity of 5 km/s. For lower values of the efficiency factor, the projectile radius would scale with ǫ−1/3 KE . As we have seen in the simulations of the Newton's cradle with viscoelastic interactions, there is a considerable loss of kinetic energy after a series of collisions, although the total linear momentum is conserved. As far as we know, there is very limited data on the transfer of momentum in hypervelocity impacts, and we do not know the efficiency factor of this transfer (ǫLM ). A similar estimate of the lower limit for the impactor size can be done by assuming a mo- mentum efficiency factor ǫLM = 1 and an impact velocity of 5 km/s. In Table 2, we present the radius of the equivalent projectile for the two set of initial particle velocities. The projectile radius would scale with ǫ−1/3 LM . The location of the explosion is always at the sur- face and with angular coordinates (latitude = 45 deg , longitude = 45 deg). In Figure 16 we present snapshots showing the propagation of the wave into the interior, by using slices passing through the centre of the sphere, the explosion point and the poles. Figure 16 a and b correspond to the simulations with body radius of 250 m, the largest number of particles (N = 783552) and particles velocities of 100 m/s (case B-100). Snapshot a is at 0.4 s after the explosion and b is at 2 s. The particles are coloured using a colour bar that scales with the modulus of the velocity. On the other hand, Figure 16c and d correspond to the simu- lations with body radius of 1000 m, the largest number of particles (N = 688443) and particles velocities of 500 m/s (case D-500). Snapshot c is at 3 s after the explosion and d is at 6 s. In the supplementary material we present movies of these simulations. (movie7 corresponds to the case B-100 Figure 16. Snapshots of the sphere explosions simulations. These are slices passing through the centre of the sphere, the explosion point and the poles. Snapshots a) and b) correspond to the simu- lations with body radius of 250 m, the largest number of particles (N = 783552) and particles velocities of 100 m/s (case B-100). Snapshot a is at 0.4 s after the explosion and b is at 2 s. The parti- cles are coloured using a colour bar that scales with the modulus of the velocity. Snapshots c) and d) correspond to the simula- tions with body radius of 1000 m, the largest number of particles (N = 688443) and particles velocities of 500 m/s (case D-500). Snapshot c is at 3 s after the explosion and d is at 6 s. (see movies in the supplementary material) m/s, movie8 to case B-500 m/s, movie9 to case D-100 m/s, and movie10 to case D-500 m/s. In the movies we observed the variation of the velocity of the particles in a slice pass- ing through the centre of the sphere, the explosion point and the poles. The particles are coloured using a colour bar that scales with the modulus of the velocity. We note that a shock front with a spherical shape prop- agates to the interior from the explosion point. On the sur- Granular physics in low-gravity environments 13 face, there appears a layer of fast moving particles that ex- tends until it intersects with the spherical front, creating in- side the volume limited by the surface layer and the spherical front, a cavity of slow moving particles. The velocity of the propagation front has a weak dependence on the velocity of the initial particles. For example, in the simulations of the smaller body (case B-100), the propagation shock requires 1.8 s to reach the antipodes of the explosion point, implying a velocity of 278 m/s. In the case B-500, the required time is 1.2 s, and the velocity 416 m/s. For the largest body, the figures are: case D-100: time 9.6 s, velocity 208 m/s; case D-500: time 5.8 s, velocity 435 m/s. Although there is an increase in the initial velocity of the moving particles of a factor of 5 among the cases, the velocity of the propagation shock has an in increase of ∼ 2. The velocity of the propaga- tion shock is quite constant while the shock travels through the interior. We are interested in the effects of the explosion at large distances from the explosion point. The body is divided in 8 quadrants. The explosion occurs on the surface at the cen- tre of the first quadrant (in Cartesian coordinates the first quadrant is: x > 0 & y > 0 & z > 0; and the explosion point is at: x = y = z = R/√3, R - radius). We analyse the distribution of ejection velocities of the particles close to the surface (r > 0.8R) on the other 7 quadrants. Histograms of these distributions are presented in Figure 17 a and b. In Figure 17a there are two overlapping histograms which cor- respond to the cases B-100 and B-500, while in Figure 17b, they correspond to the cases D-100 and D-500. A vertical line marking the escape velocity for each body is included in the plots. Note that for the smallest object and for both initial velocities, there is a significant fraction of particles that acquire ejection velocities over the escape limit. Con- sidering the total fraction of particles with velocities over this threshold (not only the ones near the surface), we ob- tain values of 18% in the case B-100, and 81% in the case B-500. In the case of the largest body, there is a significant fraction of escaping particles only for the largest initial ve- locity. The total fraction of escaping particles are 0.6% in the case D-100, and 100% in the case D-500. For the sim- ulations with initial velocities of 500 m/s, there is a total disruption of both bodies (> 50% of the mass is ejected at velocities over the escape one). It is out of the scope of this paper to derive the disruption laws for this type of ex- periments; we just mention that with a set of experiments like the previous ones, we could obtain the kinetic energy threshold over which the explosions lead to a total disrup- tion of the body as a function of size. In Table 2 we also include the ratio between the kinetic energy of the moving particles over the potential energy of the body and the spe- cific energy (defined as the deposited energy per unit mass). Housen and Holsapple (1990) have defined the critical spe- cific energy (Q∗) as the energy per unit mass necessary to catastrophically disrupt a body. Ryan (2000) presents a plot comparing different estimates of Q∗ by several authors as a function of the target radius. Let us note the fact that the largest body (R = 1000 m) is more disrupted than the small- est body (R = 250 m), although the specific energy is lower, it is in agreement with the dip in the Q∗ vs R plot (Ryan (2000)) in this radius range. Except for the case of low velocity explosions for the large body, in all the other simulations, a fraction of the 100 a n o i t c a r F 0.14 0.12 0.1 0.08 0.06 0.04 0.02 500 −3 x 10 100 b n o i t c a r F 7 6 5 4 3 2 1 500 0 0 0.5 1 Velocity (m/s) 1.5 2 2.5 0 0 2 4 Velocity (m/s) 6 8 10 Figure 17. The distribution of the ejection velocity of the parti- cles for the simulated explosion. a) Simulations with body radius of 250 m and the largest number of particles (N = 783552) (case B). Two histograms are presented for initial velocities of 100 and 500 m/s. b) Simulations with body radius of 1000 m and the largest number of particles (N = 688443) (case D). Two his- tograms for initial velocities of 100 and 500 m/s. In each plot, a vertical dashed line is drawn at the value of the escape velocity at the surface. near surface particles far from the explosion point acquire velocities over the escape one (see Figure 17). Therefore, an explosion would induce the ejection of particles from the sur- face at low velocities. These particles could either enter into orbit around the body or slowly escape from it, producing a cloud of fine particles that may take many days before disap- pearing. This result is complementary to the one obtained in Section 5 regarding the lifting and ejection of particles produced by a shake coming from below the surface. In the case of the smallest body, even the low velocity explosions would induce displacement velocities over several tenths of m/s on many near surface particles far from the explosion point (see Figure 17). This displacement would produce a shake coming from below, similar to the shakes simulated in Section 3. The surface gravity of the smallest body is similar to the surface gravity used in the low-gravity simulations of Section 3, and for the largest body the con- ditions are similar to the simulation of Eros. Looking back to Figure 16, we conclude that explosion events like the one produced in our simulations would be enough to induce the shaking required to produce size and density segregation on the surface of these bodies. This process of shaking the entire object after an im- pact is suitable for small bodies where the escape velocity is comparable to the impact induced displacement velocity at large distance from the impact point. Further work should study up to which body sizes the shaking process is expected to occur. 7 CONCLUSIONS AND APPLICATIONS OF THE RESULTS The main objective of this paper is to present the applica- tions of Discrete Element Methods for the study of the phys- ical evolution of agglomerates of rocks under low-gravity en- vironments. We have presented some initial results regard- ing process like size and density segregation due to repeated shakings or knocks, the lifting and ejections of particles from the surface due to an incoming shock and the effect of a sur- face explosion on a spherical agglomerated body. We recall that our shaking process is due to repeated set of knocks. Heredia, L., Richeri, P. 2009, Paralelismo aplicado al estu- dio de medios granulares, Proyecto de Grado, Inst. Com- putaci´on, Fac. Ingenier´ıa, UdelaR, Uruguay Hertz, H. 1882, J. f. reine u. angewandte Math., 92, 156-171 Hong, D., Quinn, P., Luding, S. 2001, Phys. Rev. Let., 86, 3423-3426 Housen, K., Holsapple, K. 1990, Icarus, 84, 226-253 Imre, B., Rabsamen, S., Springman, S. 2008, Computers & Geosciences, 34, 339350 Jaeger, H., Nagel, S. 1992, Science 255, 1523 Jullien, R., Meakin, P. 1992, Phys. Rev. Lett., 69, 640-643 Knight, J., Jaeger, H., Nagel, S. 1993, Phys. Rev. Lett., 70, 372831 Kudrolli, A. 2004, Rep. Prog. Phys., 67, 209-247 Lim, E. 2010, American Institute of Chemical Engineers Journal, 56, 2588-2597 Mair, K., Abe, S. 2008, Earth and Planetary Science Let- ters, 274, 7281 Mobius, M., Lauderdale, B., Nagel, S., Jaeger, H. 2001, Nature, 414, 270 Nahmad-Molinari, Y., Canul-Chay, G., Ruiz-Surez, J.C. 2003, Phys. Rev. E, 68, 041301 Pak, H., Van Doom, E., Behringer R. 1995, Phys. Rev. Let., 74, 4643-4646 Poschel, T., Schwager, T. 2005, Computational Granular Dynamics (Springer-Verlag, Berlin Heidelberg) Richardson, Jr. J., Melosh, H., Greenberg, R., O'Brien, D. 2005, Icarus, 179, 325-349. Rosato, A., Strandburg, K., Prinz, F., Swendsen, R. 1987, Phys. Rev. Let., 58, 1038-1040 Ryan, E. 2000, Annu. Rev. Earth Planet. Sci., 28, 367389 Scheeres, D. 2010, Icarus, 210, 968984 Schopfer, M., Abe, S., Childs, C., Walsh, J. 2009, Interna- tional Journal of Rock Mechanics and Mining Sciences, 46, 250–261 Schwager, T., Poschel, T. 2008, Phys. Rev. E, 78, 51304, 1-12 Shinbrot, T., Muzzio, F. 1998, Phys. Rev. Let., 81, 4365- 4368 Shi, Q., Sun, G., Hou, M., Lu, K. 2007, Phys. Rev. E, 75, 61302, 1-4 Timoshenko, S., Goodier, J. 1970, Theory of Elasticity, third ed. (McGraw Hill, New York) Wada, K., Senshu, H., Matsui, T. 2006, Icarus, 180, 528545 Williams, J. 1963, Fuel Soc. J., 14, 2934 14 G. Tancredi et al. The main conclusions of these preliminary results are: • A shaking induced size segregation –the so-called Brazil nut effect–does occur even in the low-gravity environments of the surface of small Solar Systems bodies, like km-size asteroids and comets. in these environments, although it is not complete. • A shaking induced density segregation is also observed • A particle layer shocked from below would produce the lifting of particles at the surface, which can acquire vertical velocities comparable to the surface escape velocity in very low-gravity environments. • A surface explosion, like the one produced by an impact or the release of energy by the liberation of internal stresses or by the re accommodation of material, would induce a shock transmitted through the entire body, and the ejection of surface particles at low velocities at distances far from the explosion point. This process is only suitable for small bodies. The application of these results to real cases will be the subject of further papers, but we foresee some situations where the results presented here will be relevant: • The internal structure of asteroid Itokawa and similar small asteroids formed as an agglomerate of m-size parti- cles, and the relevance of the Brazil nut effect produced by repeated impacts. • The non-uniform distribution of active zones in comets, like P/Hartley 2, and the internal density segregation of icy and rocky boulders produced by shakes caused by explosions and impacts. • The formation of dust clouds at low escaping velocities after an impact onto a km-size asteroid. The supplement online material can be accesed at: http://www.astronomia.edu.uy/Publications/ Tancredi/Granular_Physics/ ACKNOWLEDGEMENTS We would like to thank Dion Weatherley, Steffen Abe and the ESyS-particle users community for helpful suggestions about the package. We thank Mariana Mart´ınez Carlevaro for a careful reading of the text and many linguistic sugges- tions. REFERENCES Abe, S., Mair, K. 2005, Geophysical Research Letters, 32, L05305, doi:10.1029/2004GL022123 Abe, S., Place, D., Mora, P. 2004, Pure Appl. Geophys., 161, 2265-2277 Abe, S., Latham, S., Mora, P. 2006, Pure Appl. Geophys. 163, 18811892 Asphaug, E. 2007, Science 316, 993-994 Cundall, P., Stark, P. 1979, Geotechnique, 29, 47-65 Durda, D., Movshovitz, N., Richardson, D., Asphaug, E., Morgan, A. Rawlings, A., Vest, C. 2011, Icarus, 211, 849855 SUPPLEMENTARY ONLINE MATERIAL FOR "GRANULAR PHYSICS IN LOW-GRAVITY ENVIRONMENTS USING DEM" In movie5 the density segregation is not reached; while in movie6, most of the light particles move to the top and most of the heavy ones sink to the bottom. Granular physics in low-gravity environments 15 GLOBAL SHAKING DUE TO IMPACTS AND EXPLOSIONS We consider a km-size agglomerated body, formed by many small size boulders. We fill a sphere of radius 250 and 1000 m with ∼ 700, 000 small spheres of a given size range (1- 10 m-size boulders in the case of the small body, and 5-25 m-size boulders for the big body). At a given point on the surface we select a certain num- ber of particles of the body that are close to this place. Each particle has at the beginning of the simulation a veloc- ity along the radial vector toward the centre. The location of the explosion is always at the surface and with angular coordinates (latitude = 45 deg , longitude = 45 deg). We run simulations with initial particle velocities of 100 m/s and 500 m/s. In the movies we present snapshots showing the propa- gation of the wave into the interior. These are slices passing through the centre of the sphere, the explosion point and the poles. The particles are coloured using a colour bar that scales with the modulus of the velocity. movie7.avi and movie8.avi correspond to the simulation with a body of radius 250 m, N = 783552 small particles and 140 particles with initial velocities of 100 m/s (case B-100) and 500 m/s (case B-500), respectively. movie9.avi and movie10.avi correspond to the simula- tion with a body of radius 1000 m, N = 688443 small parti- cles and 200 particles with velocities of 100 m/s (case D-100) and 500 m/s (case D-500). All the movies correspond to 10 seconds of simulated time. Hereby you will find a set of movies included in the article "Granular physics in low-gravity environments using DEM" by Tancredi et al. (MNRAS, 2011). The supplement online material can be accesed at: http://www.astronomia.edu.uy/Publications/ Tancredi/Granular_Physics/ SIZE SEGREGATION (THE BRAZIL NUT EFFECT) SIMULATIONS A 3D box is constructed with elastic mesh walls. The box has a base of 6×6 m and a height of 150 m. A set of 12×12 m small balls of radius R1 = 0.25 m are glued to the floor. The big ball has a radius R2 = 0.75 m, and on top of it, there are 1000 small balls with radii R1 ∼ 0.25 m. scribed in the paper with different velocities. The floor is displaced with a staircase function as de- We present movies for two set of simulations: a) under Earth's gravity (surface gravity g = 9.81 m/s2) and a floor's velocity (vf loor = 5 m/s), b) in a low-gravity environment (g = 10−4 m/s2) and (vf loor = 0.05 m/s). movie1.avi is a movie with all the spheres drawn and movie2.avi with the small spheres erased for the first simu- lation. The movies correspond to 100 seconds of simulated time and 50 shakes. While movie3.avi and movie4.avi correspond to the sec- ond one. The movies correspond to 10000 seconds of simu- lated time and 667 shakes. DENSITY SEGREGATION SIMULATIONS A 3D box similar to the previous one is created, with a 6 × 6 m base and a height of 150 m. The box is con- structed with elastic mesh walls. On the floor we glue a set of 12 × 12 small balls of radius R1 = 0.25 m and den- sity ρ = 2000 kg/m3. There are 500 light balls with radii R1 ∼ 0.25 m and density ρ = 500 kg/m3. On top of them, there are 500 heavy balls with similar radii and density ρ = 2000 kg/m3. At the beginning of the simulations the balls are placed sparsely, the light balls at the bottom and the heavy ones on top. They free fall and settle down before starting the floor shaking. The floor is displaced with a staircase function in a sim- ilar way as in the previous set of simulations. We present movies for two set of simulations: a) under Earth's gravity (surface gravity g = 9.81 m/s2) and a floor's velocity (vf loor = 3 m/s), b) in a low-gravity environment (g = 10−4 m/s2) and (vf loor = 0.05 m/s). movie5.avi is a movie of the first simulation, while movie6.avi corresponds to the second one. The movie5 cor- responds to 1000 seconds of simulated time and 500 shakes, while the movie6 corresponds to 20000 seconds of simulated time and 1333 shakes. Note that in these movies the camera moves with the floor, therefore it seems that the floor is always located in the same position, but it really is moving with the staircase function described above.
1701.05564	3	1701	2018-10-24T18:23:53	A Theory of Exoplanet Transits with Light Scattering	[ "astro-ph.EP" ]	Exoplanet transit spectroscopy enables the characterization of distant worlds, and will yield key results for NASA's James Webb Space Telescope. However, transit spectra models are often simplified, omitting potentially important processes like refraction and multiple scattering. While the former process has seen recent development, the effects of light multiple scattering on exoplanet transit spectra has received little attention. Here, we develop a detailed theory of exoplanet transit spectroscopy that extends to the full refracting and multiple scattering case. We explore the importance of scattering for planet-wide cloud layers, where the relevant parameters are the slant scattering optical depth, the scattering asymmetry parameter, and the angular size of the host star. The latter determines the size of the "target" for a photon that is back-mapped from an observer. We provide results that straightforwardly indicate the potential importance of multiple scattering for transit spectra. When the orbital distance is smaller than 10-20 times the stellar radius, multiple scattering effects for aerosols with asymmetry parameters larger than 0.8-0.9 can become significant. We provide examples of the impacts of cloud/haze multiple scattering on transit spectra of a hot Jupiter-like exoplanet. For cases with a forward and conservatively scattering cloud/haze, differences due to multiple scattering effects can exceed 200 ppm, but shrink to zero at wavelength ranges corresponding to strong gas absorption or when the slant optical depth of the cloud exceeds several tens. We conclude with a discussion of types of aerosols for which multiple scattering in transit spectra may be important.	astro-ph.EP	astro-ph	A Theory of Exoplanet Transits with Light Scattering Tyler D. Robinson1,2,3 Department of Astronomy and Astrophysics, University of California, Santa Cruz, CA 95064, USA [email protected] ABSTRACT Exoplanet transit spectroscopy enables the characterization of distant worlds, and will yield key results for NASA's James Webb Space Telescope. However, transit spectra models are often simplified, omitting potentially important pro- cesses like refraction and multiple scattering. While the former process has seen recent development, the effects of light multiple scattering on exoplanet tran- sit spectra has received little attention. Here, we develop a detailed theory of exoplanet transit spectroscopy that extends to the full refracting and multiple scattering case. We explore the importance of scattering for planet-wide cloud layers, where the relevant parameters are the slant scattering optical depth, the scattering asymmetry parameter, and the angular size of the host star. The lat- ter determines the size of the "target" for a photon that is back-mapped from an observer. We provide results that straightforwardly indicate the potential im- portance of multiple scattering for transit spectra. When the orbital distance is smaller than 10 -- 20 times the stellar radius, multiple scattering effects for aerosols with asymmetry parameters larger than 0.8 -- 0.9 can become significant. We pro- vide examples of the impacts of cloud/haze multiple scattering on transit spectra of a hot Jupiter-like exoplanet. For cases with a forward and conservatively scat- tering cloud/haze, differences due to multiple scattering effects can exceed 200 ppm, but shrink to zero at wavelength ranges corresponding to strong gas ab- sorption or when the slant optical depth of the cloud exceeds several tens. We conclude with a discussion of types of aerosols for which multiple scattering in transit spectra may be important. 1Sagan Fellow 2University of California, Santa Cruz, Other Worlds Laboratory 3NASA Astrobiology Institute's Virtual Planetary Laboratory -- 2 -- 1. Introduction Transit spectroscopy (Seager & Sasselov 2000; Brown 2001; Hubbard et al. 2001) is cur- rently the leading technique for studying exoplanet atmospheric composition. Following the discovery of the first exoplanet atmosphere (Charbonneau et al. 2002), transit observations have enabled the characterization of a number of different exoplanets for atmospheric molec- ular species, clouds, and/or hazes (Pont et al. 2008; Swain et al. 2008; Sing et al. 2009; Bean et al. 2010; Fraine et al. 2014; Knutson et al. 2014b). Recent observational results have shown that hot Jupiter transit spectra demonstrate a complex continuum from clearsky to heavily clouded conditions (Sing et al. 2016; Stevenson 2016), and that cloudiness remains a key factor into the regime of lower mass planets (Line et al. 2013; Kreidberg et al. 2014; Knutson et al. 2014a). The field of exoplanet transit spectroscopy will be revolutionized with the anticipated launch of NASA's James Webb Space Telescope (JWST) in 2018 (Gardner et al. 2006). Over the course of the design five year mission for JWST, the observatory is expected to provide in-depth observations of many tens of transiting exoplanets (Beichman et al. 2014). Some of these observations will probe planets in the poorly understood 2 -- 4 Earth mass regime (Deming et al. 2009; Batalha et al. 2015). Excitingly, JWST may even be capable of characterizing temperate Earth-sized planets (Kaltenegger & Traub 2009; Cowan et al. 2015; Barstow et al. 2016), though the ability to conduct such studies will depend largely on how well systematic noise sources can be constrained (Greene et al. 2016). As the quality of transit spectrum observations continues to improve, so should models of exoplanet transits. Thus, certain processes initially thought to be of second-order impor- tance should be revisited and possibly added to modeling tools. For example, atmospheric refraction, which was initially shown to be unimportant for the case of certain hot Jupiters (Hubbard et al. 2001), has recently been shown to be critically important for understand- ing some terrestrial exoplanet spectra (Muñoz et al. 2012; Bétrémieux & Kaltenegger 2014; Misra et al. 2014; Bétrémieux & Kaltenegger 2015) and, possibly, gas giant transit spectra (Dalba et al. 2015; Bétrémieux 2016). Additionally, refraction can affect the shape of an exoplanet transit lightcurve (Hui & Seager 2002; Sidis & Sari 2010). Beyond refraction, another process that has seen little study with regards to exoplanet transits is light multiple scattering. Hubbard et al. (2001) used a plane-parallel Monte Carlo scattering model to determine the significance of a molecular Rayleigh scattering "glow" around the limb of a transiting exoplanet. For the case of HD 209458b, these authors found that the contribution of Rayleigh scattering glow to the transit lightcurve is negligible. Similar conclusions were reached by Brown (2001), who used analytic arguments to deduce that multiple isotropic scatterings could have only a small impact on transit depths. -- 3 -- Since the study of Hubbard et al. (2001), scattering opacity in exoplanet transits has largely been treated as being equivalent to absorption opacity (e.g., Irwin et al. 2008; Line et al. 2013; Waldmann et al. 2015). This equivalence cannot always hold. For example, a very strong forward and conservatively scattering aerosol would only weakly attenuate a beam following a straight-line path from a stellar disk, passing through an exoplanet atmosphere, and traveling towards a distant observer. This issue was plainly demonstrated by De Kok & Stam (2012), who used a Monte Carlo scattering model to study the transmission of a pencil beam through a cloud or haze layer with variable optical thickness and whose scatterers had asymmetry parameters of either 0, 0.9, or 0.98 (their Figure 3). In the work that follows, we do not seek to outline specific definitions for "hazes" or "clouds." In general, though, we use the former to refer to aerosols that form aloft and grow due to coagulation during the sedimentation process, thereby developing a distribution that may look something like Titan's tholin haze, which has a number density profile de- scribed (roughly) by an exponential with a constant scale height in Titan's upper atmosphere (Tomasko et al. 2008). For clouds, we envision an aerosol distribution formed by lifting a gas to its condensation point, thereby developing a distinct cloud base with an overlying cloud deck whose thickness is controlled by mixing processes (see, e.g., Ackerman & Marley 2001). Advances in exoplanet and brown dwarf cloud models have been reviewed recently by Mar- ley et al. (2013). The most popular tools include the Eddysed model (Ackerman & Marley 2001), where an upward diffusion of condensible vapor balances a downward sedimentation of aerosol particles, which has been successfully used to study both cloud and haze processes in transiting exoplanet atmospheres (Morley et al. 2013). Additionally, the sophisticated microphysical cloud treatments of Helling et al. (2001) continue to see development and application to transiting exoplanets (Lee et al. 2016). Thus, the question of the relative importance of scattering, refraction, and absorption remains largely unexplored for a wide portion of exoplanet parameter space. Such studies have been hindered by the lack of available simulation tools, and also by a lack of a single, coherent theory of exoplanet transit spectroscopy. Significant advances can be made by developing such a theory, and by building models that implement these ideas. In this paper, we construct a theory of exoplanet transit spectroscopy that spans key fundamental integral relations through to its efficient, vectorized implementation. Simpli- fications of the theory cover the geometric (i.e., straight-line) limit (that is currently used in computationally-demanding spectral retrieval models; e.g., Line et al. 2012; Benneke & Seager 2012; Barstow et al. 2012; Lee et al. 2014; Waldmann et al. 2015; Morley et al. 2016) and cases that add refraction. In general, though, we emphasize the "full physics" scenario, where a three-dimensional Monte Carlo model is used to incorporate scattering effects. -- 4 -- Below, we introduce the concept of the "path distribution", which effectively separates the paths of photons (or rays) through a transiting exoplanet atmosphere from gaseous and/or aerosol absorption effects. As the former tends to vary weakly in wavelength, while the latter can vary strongly with wavelength, our approach is computationally efficient. Following the presentation of these ideas, we validate our implementation of the theory against a number of trusted simulation tools. We then give examples of the path distribution for a variety of different model atmospheres. Finally, we use our tools to present the first simulated transit spectra of hot Jupiters with multiply-scattering clouds. 2. Theory and Model Description Transit spectra typically probe low density regions of exoplanetary atmospheres. Here, molecular absorption lines are relatively narrow, as the effects of pressure broadening are not dominant over Doppler broadening (as is the case in the deep atmosphere). Thus, the gas opacity can vary by orders of magnitude over narrow spectral ranges. By contrast, gaseous refractive indexes as well as gas and aerosol scattering properties tend to vary smoothly (and, sometimes, weakly) in wavelength. Since refraction and scattering, which are more compu- tationally intensive processes to simulate, influence the path of a photon (or ray) through an atmosphere, significant advances can be made by outlining a technique that separates the processes which influence photon trajectories from those which influence absorption. We introduce the concept of the ray (or photon) atmospheric path distribution, Pb(h), where b is the impact parameter of the ray and h is altitude in the planetary atmosphere. We define the path distribution such that Pb(h) dh is the linear distance traversed by the ray at altitudes between h and h + dh. While Pb is dimensionless, it can be thought of as having units of km linear distance per km vertical distance. The geometry of these parameters, for a simple case, is shown in Figure 2. The Appendix contains additional discussion on the use of the path distribution. In transit spectroscopy, the essential quantity is the attenuation of a ray along its path, as this distinguishes the opaque portions of the atmosphere (that block light from the stellar disk) from the transparent portions. In 1-D atmospheric structure models, which are the variety most commonly used for exoplanets, the atmospheric opacity is only a function of altitude (or pressure) and wavelength. Here, then, the optical depth along the path of a ray can be computed using the path distribution via, (cid:90) ∞ where αλ is the atmospheric extinction (in units of inverse distance), and the integral can 0 τλ,b = αλ(h)Pb(h) dh , (1) -- 5 -- be taken to infinity as the extinction is zero at very large altitudes. Since the layer vertical differential optical depth is defined as dτλ(h) = αλ(h)dh, this expression demonstrates that the path distribution can also be thought of as the linear optical depth encountered between τλ and τλ + dτλ. Similar extensions exist for quantities like the column mass and number densities. Note that Equation 1 is critical, as the strongly wavelength-dependent extinction is separated from the ray path. Using the standard definition of transmission, tλ(b) = e−τλ,b , (2) a simple transit spectrum in the pure absorbing limit (i.e., where all optical depth is taken as absorption optical depth) can then be computed by considering the light transmitted through concentric annuli on the planetary disk, each at their own impact parameter, [1 − tλ(b)] bdb , (3) (cid:18) Rp,λ (cid:19)2 Rs (cid:90) ∞ 0 = 2 R2 s where Rp,λ is the wavelength-dependent planetary radius, Rs is the stellar radius, and (Rp,λ/Rs)2 is the transit depth. In practice, model planetary atmospheres are defined on an altitude (or pressure) grid, and Equation 3 is computed using a sum over a collection of impact parameters (and as- suming that the planetary disk is opaque below some fiducial radius, Rp, which is either at the surface or deep in the atmosphere). In this case, the path distribution is a matrix, Pi,j, where Pi,j∆hj is the path traversed through atmospheric layer 'j' (whose thickness is ∆hj) by a ray whose impact parameter is bi. At any given wavelength, the transmission is now a sum over Nlay atmospheric layers, − Nlay(cid:88) −  = EXP Nlay(cid:88)  , tλ,i = EXP αλ,jPi,j∆hj ∆τλ,jPi,j (4) j=1 j=1 where we have used the definition of the layer vertical differential optical depth, ∆τλ,j = αλ,j∆hj. A similar expression was used by Bétrémieux (2016, their Equations 5 and 7), where these authors work in numerically-computed column number densities and molecular opacities, as compared to our dimensionless quantities. Note that Equation 4 can be easily written in matrix notation as, tλ = 1 − aλ = EXP(−∆τ λ · P) , (5) where we have defined the absorptivity vector, aλ. The pure absorption transit spectrum derives from a sum over Nr impact parameters, (cid:32) (cid:18) Rp,λ (cid:19)2 Rs = 1 R2 s Nr(cid:88) i=1 (cid:33) R2 p + 2 [1 − tλ,i] bi∆bi , (6) where ∆bi is the thickness of the impact parameter gridpoint. Continuing with the matrix notation, if we define a vector of annulus areas as Ai = 2πbi∆bi, then the transit spectrum can be written simply as, (cid:18) Rp,λ (cid:19)2 Rs = 1 R2 s R2 p + aλ · A 1 π . (7) -- 6 -- (cid:18) (cid:19) We briefly note that the transit spectra expressions given in Equations 3, 6, and 7 are in the pure absorption limit, and assume that all rays trace back to the stellar disk, and that the star has uniform surface brightness. While these are common assumptions when computing transit spectra for model atmospheres, we will now discuss the more general multiple scattering case. The Appendix contains brief details about computing the path distribution in the geometric limit or in the case that refraction is considered. 2.1. Multiple Scattering Path Distributions and Transmissions Cases that include clouds, especially strongly forward scattering clouds, require three- dimensional radiative transfer treatments to compute accurate transit spectra, and are well suited to Monte Carlo models. Here, the path distributions for a number of photons, Np, are used to derive an average transmission for a grid of impact parameters. The individual path distributions are determined using a Monte Carlo approach, and only consider the scattering optical depths (τλ,s = ω0τλ,e, where ω0 is the single-scattering albedo, and a sub-script 's' indicates "scattering" while 'e' indicates "extinction") since only τλ,s affects the path of rays through the atmosphere. Given the path distribution, P m, for photon 'm', the transmission for this photon is, (8) where ∆τ λ,a is a vector of the wavelength-dependent layer differential absorption optical depths, and the transmission averaged over all photons is simply, tλ,m = EXP(−∆τ λ,a · P m) , (cid:80)Np ¯tλ = m=1 tλ,m Np . (9) Execution of our Monte Carlo model follows a standard approach. For a given impact parameter, bi, a photon is directed into the atmosphere along the −x direction in Figure 2, and then tracked through the atmosphere. As this photon originated from the direction of the observer, we are performing a so-called "backward" Monte Carlo simulation. The optical distance the photon travels, either initially or following a scattering event, is determined by randomly sampling the scattering optical depth distribution, with, ξ = f (τ )dτ , (10) where ξ is a random number between zero and one, and f (τ )dτ is the probability that a photon scatters between τ and τ + dτ. As f (τ )dτ = e−τ dτ, we have, 0 τs = − ln(1 − ξ) . (11) For whichever layer (centered at hj) the photon is currently in, the pathlength corresponding to τs is, (12) where αs,j is the layer scattering extinction coefficient. Given a pathlength, a photon location (r = xx + y y + zz), and trajectory ( µ = µx x + µy y + µz z), the photon position is updated according to x → x + µxs, y → y + µys, and z → z + µzs. s = τs/αs,j , Depending on the size of s, the photon either experiences a scattering event within layer j, in which case s/∆hj is added to the path distribution for this photon (i.e., P m), or the photon exits the layer before a scattering event occurs. For the latter, s must be compared to the distance to the layer boundaries. Given the trajectory of the photon, and its current radial position, r =(cid:112)x2 + y2 + z2, the altitude of the photon along its path is, (cid:113) -- 7 -- (cid:90) τs h(s) + Rp = r(s) = (x + µxs)2 + (y + µys)2 + (z + µzs)2 . Thus, the quadratic equations (in s), h(sj) = hj ± ∆hj/2 , (13) (14) give the distances to the layer boundaries. When a photon travels to a layer boundary before a scattering event, sj/∆hj is added to P m, the cumulative optical depth experienced by the photon us updated as τ → τ + αs,jsj, and the photon is passed to the appropriate layer (either j − 1 or j + 1). This process of randomly generating τs and passing the photon through sequential layers is repeated until the photon either exits the atmosphere, or reaches the lower boundary of the model and is considered absorbed. (Note that, as with any transit spectrum model of gaseous worlds, the lower boundary must be set deep enough to not influence the simulated spectrum.) For photons that exit the atmosphere, the position and trajectory at exit determine whether or not the photon will intersect the stellar disk along a straight-line trajectory. -- 8 -- (cid:90) µ −1 ξ = P (µ(cid:48))dµ(cid:48) , (15) (16) (17) (18) When a scattering event occurs, a scattering angle must be sampled from the scattering phase function and the photon trajectory must then be updated. Assuming that the scat- tering phase function, P , is only a function of the cosine of a single angle, µ = cos(θ), then a randomly sampled scattering deflection angle is determined via, and an azimuthal scattering angle, φ, is sampled uniformly, with, φ = 2πξ . As the scattering angles are referenced from the propagation direction, a transformation must be done to convert the initial propagation direction, µ, and the scattering angles into a (cid:48). Standard expressions exist for completing this transformation new propagation direction, µ (e.g., Witt 1977, their Equations 22 and 23). Refraction can be included in the Monte Carlo simulation. As the photon moves along s, the path can be sub-divided, and a curvature applied to each smaller ∆s. At the photon height, the curvature, rc, is computed according to Equation C1, and a deflection angle for the photon is determined from the refractive portion of Equation C4 (i.e., ∆θr = ∆s/rc). The change in trajectory is referenced from µ and occurs in either the local upward or downward direction (depending on the sign of rc). A new set of direction cosines is found by requiring, = cos(∆θr) , µ · µ (cid:48) ( µ × r) · µ (cid:48) = 0 , and (19) where the first equation implies that the trajectory changes through ∆θr, the second equation forces the trajectory change to be either locally up or down, and the final equation ensures that the new direction of travel is a unit vector. µ (cid:48) = 1 , The computational efficiency of the Monte Carlo approach to multiple scattering in transit spectra outlined above can be increased in several ways. First, for a given impact parameter, the Monte Carlo routine need only be executed if the straight-line scattering optical depth is a substantial fraction of the absorption optical depth. (We choose a conser- vative cutoff at τλ,s < 10−3τλ,a.) Also, as suggested by De Kok & Stam (2012), the number of photons used in the Monte Carlo simulation at a given impact parameter should be in- fluenced by the straight-line transmission -- these authors recommend the use of 105 photons in transparent conditions and up to 108 photons in opaque conditions. However, De Kok & -- 9 -- Stam (2012) used a traditional Monte Carlo radiative transfer approach, where photons are lost to absorption while passing through the atmosphere, thus driving the need for very large numbers of photons in their simulations. As we have separated absorption from scattering in our model, (nearly) all of the photons we simulate in our Monte Carlo exit the atmosphere, so that we only require 104 -- 106 photons. Even more dramatic efficiency gains can be made by noting that, since our Monte Carlo model only considers scattering optical depths, these simulations need only be recomputed if either the layer differential scattering optical depths or asymmetry parameters change significantly. Thus, the individual path distribution ma- trices for each of the photons in our simulations can be saved and reused over a wide range of wavelengths in a high-resolution spectral grid. Of course, running Np transmission calculations (i.e., using Equation 8) at every wave- length in a high-resolution grid becomes computationally unfeasible if the grid is large. Here, runtime reductions of factors of 10 -- 100 can be straightforwardly made by computing the an- alytic Jacobian of Equation 9, ∂¯tλ/∂∆τ λ,a, with, (cid:18) ∂¯tλ (cid:19) ∂∆τ λ,a i,j Np(cid:88) m=1 Np(cid:88) m=1 = 1 Np ∂tλ,m(bi) ∂∆τλ,a,j = − 1 Np Pm,i,j · tλ,m(bi) . (20) This Jacobian can be used to rapidly adapt ¯tλ to changes in ∆τ λ,a, and our tests indicate that this approach is accurate with up to 25 -- 50% variations in level-dependent absorption optical depths. Since a set of path distributions, transmissions, and Jacobians apply to wavelengths where the atmospheric optical property profiles are similar, the Monte Carlo approach outlined above is best implemented within a spectral mapping model (West et al. 1990; Meadows & Crisp 1996). We note that the many efficiency gains outlined above stem from our separate treatments of scattering (which influences the path distributions) and absorption (which influences the path-derived transmissions and Jacobians). So far as we know, this treatment is a novel approach to Monte Carlo radiative transfer. Additionally, our approach to multiple scattering can be applied beyond exoplanet transits, and, for example, could be used to study scattering effects in occultation observations (i.e., an extension of De Kok & Stam 2012). 2.2. Generalized Transit Spectra The transit spectra expressions given in Equations 3, 6, and 7 are primarily useful in the geometric limit, since rays in refracting or scattering models are not guaranteed to trace back to the stellar disk. To efficiently model spectra in these latter cases, we generalize the -- 10 -- concept of the vector of annulus areas, A. Key variables used in the discussion below are visualized in Figure 3. Ultimately, a transit spectrum is determined by comparing intensities integrated over the range of solid angles influenced by the planet (Ωp) in the case when the planet is present versus when only the star is considered (with corresponding intensities Ip,λ and Is,λ, respec- tively). Thus, Ip,λdΩ , (21) (cid:18) Rp,λ (cid:19)2 Rs (cid:82) Is,λdΩ −(cid:82) (cid:82) Ωp Is,λdΩ Ωs Ωp = where Ωs is the range of solid angles corresponding to the stellar disk. Given the large distance D between Earth and any exoplanetary system, so that dΩ = dA/D2, and taking the stellar intensity to be normalized such that, then, (cid:18) Rp,λ (cid:19)2 Rs = (cid:82) , I0,λ = Ωs Is,λdΩ Ωs Is,λ/I0,λdA −(cid:82) (cid:82) Ap πR2 s (22) (23) . Ip,λ/I0,λdA Ap The first integral in the numerator of Equation 23 is simply a stellar limb darkening law integrated over the portion of the planetary disk that overlaps the stellar disk. Using a polar coordinate system centered on the planetary disk, and letting d be the distance from the center of the star to the center of the planet, this integral is then, d ≥ Rs + Rp + ht (µs)rdθdr , d < Rs + Rp + ht , Is,λ I0,λ (24) where µs = µs(r, θ) is the cosine of the angle of incidence on the stellar disk at coordinates r and θ, and (cid:90) Is,λ I0,λ dA = Ap (cid:40) (cid:82) Rp+ht 0 , r0 −θ0(r) (cid:82) θ0(r) (cid:40) r0 = (cid:40) d ≤ Rs 0 , d − Rs , d > Rs , θ0(r) = π , d2+r2−R2 s 2dr d ≤ Rs − Rp − ht , d > Rs − Rp − ht . (25) (26) However, as limb darkening is accounted for in standard transit observation data reduction procedures (e.g., Mandel & Agol 2002; Kreidberg 2015), one may wish to omit the limb darkening law from the first integral in the numerator of Equation 23, leaving an analytic integral that can be performed by considering the geometry of two overlapping disks. -- 11 -- The second integral in the numerator of Equation 23 contains the details associated with absorption, scattering, and refraction in the planetary atmosphere. We define I(r, θ) as the background surface brightness mapped onto by an area element on the planet at r and θ. This function is zero for rays that do not map back to the stellar disk, and, in the geometric limit, is equal to Is,λ(µs(r, θ)) for portions of the planetary disk that overlap the star. Given this definition, we then have, (cid:90) Ip,λ I0,λ dA = 1 I0,λ Ap (cid:90) Rp+ht (cid:90) π Rp −π tλ(r) I(r, θ)rdθdr , (27) where we have assumed that Rp is set sufficiently deep so that Ip,λ = 0 for r < Rp. Of course, it is straightforward to consider thermal emission from the planetary disk, which can be important at longer wavelengths (Kipping & Tinetti 2010), by including the planet-to-star flux ratio for the planetary nightside on the right-hand side of Equation 23. For efficient implementation, we define a grid on the atmospheric portion of the planetary disk in Nθ angular and Nr radial points. An Nr×Nθ area matrix, with Ai,j = ri∆ri∆θj, need only be computed once, and the Nr × Nθ background surface brightness mapping matrix, I, is computed alongside the path distributions and transmissions. Given these, we have (cid:90) Ip,λ I0,λ dA = 1 I0,λ Ap tλ ·(cid:104)(cid:16) I ◦ A (cid:17) · 1 (cid:105) , (28) where '◦' indicates the Hadamard product, 1 is a vector of ones with length Nθ, and, as before, tλ is computed using either Equation 5 or Equations 9 and 20. Critically, this approach still efficiently isolates the parameters that vary rapidly in wavelength, tλ, from the parameters influenced by geometry and ray paths -- the background surface brightness mapping matrix need only be computed once in the geometric limit, and 1 -- 2 times when refraction is included. For the multiple scattering case, full Monte Carlo simulations need only be run at wavelength scales over which the atmospheric scattering properties vary, and transmissions need only be recomputed at wavelength scales over which their linear corrections via the analytic Jacobians become inaccurate. Integration can also be made more efficient by noting the symmetry about the line connecting the planet center to the star center. The matrix I is straightforward to compute in the geometric limit, as rays trace straight lines back to either the stellar disk or not. When refraction is included, as in the Appendix, a ray tracing is performed once for each gridpoint in Nr ray impact parameters. Then, given the exit location and direction for these rays, and the angular location, θj, for each area element, simple geometry indicates which if the Nθ gridpoints have rays that map back to the stellar disk (and what the µs value is for each of these rays). In the multiple scattering -- 12 -- case, Np Monte Carlo simulations are run, and I is built up by averaging over the photons in each instance. For each photon that exits the atmosphere, its trajectory leaving the top of the atmosphere is investigated to see if the stellar disk is intersected. The geometry is then rotated through ∆θj, and the intersection is reinvestigated. Thus, only Np Monte Carlo simulations are needed to build up the average Nr × Nθ background mapping matrix. Figure 4 shows the surface brightness in a full Monte Carlo simulation prior to integra- tion over solid angle. A haze with vertical optical depth of τs = 0.01 is placed above the 1 µbar pressure level following a linear power-law in pressure. The atmosphere is isothermal at 1,500 K, the stellar and planetary size are appropriate for HD 189733b, and refraction and Rayleigh scattering are included (at 0.8 µm wavelength). A secondary figure enhances the surface brightness off the stellar disk by a factor of 10. For the portion of the planet that is not overlapping the stellar disk, the sub-figure with enhanced surface brightness shows a thin ring due to forward scattered light. Additional surface brightness structure in this ring is related to the scattering phase function -- the region of the planetary limb that is at the greatest angular separation from the stellar disk less strongly samples the haze forward scattering peak. 2.3. Model Summary The "path distribution" formalism represents a framework that allows for the computa- tion of transit spectra across a broad range of conditions. Most existing transit spectrum models assume that rays travel along straight-line trajectories, and are extinguished from the line-of-sight path according to the extinction optical depth (e.g. Line et al. 2012; Benneke & Seager 2012; Barstow et al. 2012). Equations B1, 4, and 6 apply in this simplified regime. Slightly more sophisticated transit spectrum tools incorporate the physics of refraction through a ray tracing scheme (e.g. Bétrémieux & Kaltenegger 2014; Misra et al. 2014). We outline a similar ray tracing approach to computing the path distribution for a refracting atmosphere in the Appendix. For planets centered on their host stellar disk, only a single ray tracing need be performed. This symmetry is broken as the planet moves away from the center of the disk, implying that different locations around the planetary disk may or may not refractively map back to the stellar disk. Here, Equation 4 gives the (radially dependent) transmission, and Equation 27 or 28 describes the angular integration required to obtain a transit spectrum. Extremely few models (or modeling approaches) exist that allow for the simulation of multiple scattering effects in transit spectra. Hubbard et al. (2001) describes a Monte Carlo -- 13 -- model that treats the stratified limb of a transiting exoplanet as a set of non-interacting slabs. Following these authors' investigation into Rayleigh scattering effects, this model has seen little use. More recently, (De Kok & Stam 2012) constructed a realistic, three-dimensional Monte Carlo for investigating forward scattering effects in occultation and transit obser- vations. This tool has not been used to predict exoplanet transit spectra that include the effects of forward scattering, though. Our own three-dimensional Monte Carlo approach (de- tailed above), when paired with the path distribution formalism and with its use of analytic Jacobians, enables the calculation of high-resolution, multiple scattering transit spectra with computational efficiency gains of factors of 102 -- 104 over previous three dimensional models. To best enable the implementation of the theory outlined above, a stand-alone model, which we call scaTran, has been made publicly available. The software can be downloaded from https://github.com/tdrobinson/scaTran, or can be found by searching GitHub for "scaTran." 3. Validation We validated our transit model, which we implement within the framework of the Spec- tral Mapping Atmospheric Radiative Transfer (SMART) model (developed by D. Crisp; Mead- ows & Crisp 1996), using a number of techniques and data sources. First, with refraction and scattering optical depth omitted, we verified that our ray tracing routine (discussed in the Appendix) and the Monte Carlo approach return the path distribution in the geometric limit. Then, with refraction included and scattering optical depth omitted, we showed that the ray tracing and Monte Carlo routines are in agreement. Finally, to check our Monte Carlo approach with scattering included, we made the simple switch to a plane-parallel geometry (where height is only measured relative to the z-axis, not radially) and validated the output radiances against results from the DISORT radiative transfer model (Stamnes et al. 1988) for a wide range of conditions. Figure 5 shows the results from two of these experiments, plotting the top-of-atmosphere intensity (scaled by the incident flux) as a function of observer zenith angle. Clouds with a given scattering optical depth and asymmetry parameter were placed over a perfectly absorbing surface, and a Henyey-Greenstein phase function was assumed. To further validate our model, we performed model inter-comparisons. First, using a standard Earth atmospheric model (McClatchey et al. 1972), we compared our transit spectrum for an Earth-Sun twin system to the refracting model of Misra et al. (2014). Note that the Misra et al. (2014) model has been extensively validated against solar occultation data for Earth (Gunson et al. 1996; Irion et al. 2002) and observations of transmission through Earth's atmosphere during a lunar eclipse (Pallé et al. 2009). Figure 6 shows the result of -- 14 -- this inter-comparison, where agreement (in effective transit height, equal to Rp,λ − Rp) is to within the atmospheric model grid spacing. Agreement improves if we use the coarser integration path lengths adopted by Misra et al. (i.e., 5 km). Second, and finally, we applied our transit spectrum tool to a standardized hot Jupiter- like atmospheric model widely used for inter-comparison purposes. This model has a plane- tary radius (at the 10 bar pressure level) of 1.16 RJ, a planetary mass of 1.14 MJ, a stellar radius of 0.78 R(cid:12), and atmospheric volume mixing ratios of H2, He, and H2O of 0.85, 0.15, and 4 × 10−4, respectively. The 126 model layers are placed evenly in log-pressure between 10 bar and 10−9 bar. The atmosphere is isothermal at 1500 K. Figure 7 compares our model, in the geometric limit, to output from the CHIMERA retrieval suite (Line et al. 2013) for this standard case. Agreement is within the 20 ppm scatter seen in comparisons between other (Irwin et al. 2008; Waldmann et al. 2015) transit spectrum tools (M. R. Line, personal com- munication). We note that the offset in the Rayleigh scattering slopes between our model and the CHIMERA calculation is due to differing approaches to computing Rayleigh scattering opacities which results in optical depths that differ at the 5% level or less. 4. Results We use the formalism outlined above to, first, compute the path distribution for Earth- like and hot Jupiter-like atmospheres for a range of conditions. Following these instructive examples, we explore how scattering influences the transit depth associated with a single aerosol layer over a wide range of parameter space. Finally, we demonstrate the effects of thin, scattering clouds on a typical hot Jupiter transit spectrum. 4.1. Path Distributions The aforementioned standard Earth and hot Jupiter atmospheric models can be used to demonstrate the path distribution for a range of conditions and assumptions. To show the effects of clouds on the path distribution for the hot Jupiter atmosphere, we use the "cloud" and "haze" differential optical depth profiles shown in Figure 8. The former is based on the shape of the aerosol mixing ratio profiles for a condensational cloud from the Ackerman & Marley (2001) model, while the latter follows a power-law in altitude with scale height equal to the pressure scale height. Both models are taken to have integrated vertical optical depth unity, and to be conservatively and moderately forward scattering (g = 0.9). A Henyey- Greenstein phase function is assumed for cloudy and hazy simulations. -- 15 -- Figure 9 demonstrates the path distribution in the geometric limit for the standard hot Jupiter-like model atmosphere. Altitudes are measured relative to the 10 bar radius (at 1.16RJ). This graph is interpreted by selecting an impact parameter, and using the color shading to interpret the path distribution (which, as stated earlier, can be thought of as an enhancement of the vertical optical depth). For example, taking an impact parameter "alti- tude" (b− Rp) of 3000 km, the path distribution is zero for all atmospheric layers below this height, since rays with this impact parameter never pass through these deeper atmospheric layers. The path distribution is then largest for layers near an altitude of 3000 km, as rays pass through these layers either horizontally or nearly horizontally. Finally, for atmospheric layers well above 3000 km, rays pass through on much less extreme slant paths, so the path distribution is smaller (typically 5 -- 15, for this particular model atmosphere). The path distribution considering refraction for our cloud-free Earth atmosphere is shown in Figure 10. For comparison, a case in the geometric limit is also presented. For elements of the path distribution where the impact parameter altitude is near the surface, the path distribution deviates substantially from the geometric case due to the refractive bending of ray paths. As an example, take an impact parameter altitude of 5 km. In the geometric case, rays with this impact parameter would never reach altitudes below 5 km. But, due to refraction, these rays are bent "downward" to probe deeper atmospheric layers. Thus, the path distribution for the refracting case is now non-zero for atmospheric layers below 5 km. Furthermore, the atmospheric layer that sees the most enhancement of optical depth (i.e., has the largest path distribution) is now slightly below 5 km, instead of being right at 5 km as in the geometric case. Figures 11 and 12 show an average path distribution from our multiply-scattering Monte Carlo approach, with, ¯P = 1 Np Np(cid:88) m=1 P m . (29) We note that such an average path distribution is not used in our calculation of transit spectra, but is simply meant to indicate the characteristic path that scattered photons take through the atmosphere. In the Figure 11 we only consider molecular Rayleigh scattering opacity (at 0.55 µm wavelength) in our hot Jupiter-like model atmosphere. For Figure 12, we show the effects of our nominal haze and cloud models for the hot Jupiter-like case. While both the haze and cloud have the same integrated optical depth, the haze path distribution has a less pronounced transition into the aerosol-affected atmospheric layers since the con- densate cloud is concentrated into a more narrow range of altitudes/pressures. Note the use of a logarithmic color contour scale to emphasize the impacts of scattering. The path distributions for the scattering cases are less straightforward to interpret. -- 16 -- For the Rayleigh scattering case (Figure 11), and taking an impact parameter altitude of 800 km, we see that the path distribution is non-zero for altitudes below 800 km (but would be zero in the geometric case). Here, a small fraction of photons have been scattered down to these deeper layers, and the small values of the path distribution at depth come at the (small) expense of the path distribution aloft. Selecting a deeper impact parameter altitude of 200 km, we see that the path distribution is now greatest at large altitudes (where the path distribution would have been roughly 10 in the geometric case), and is small at altitudes below about 400 -- 600 km. Here, photons with a 200 km impact parameter altitude are scattered at altitudes larger than about 400 -- 600 km, so that relatively few of these photons can probe much deeper than this. The haze case in Figure 12 can be interpreted similarly to the Rayleigh scattering case, although the range of impact parameters that are influenced by the haze is greater than that of the Rayleigh scattering case. The condensate cloud case in Figure 12 appears distinct due to the well-defined cloud top. Here, rays/photons with impact parameter altitudes larger than 1400 km never encounter the cloud, and the path distribution is the same as in the geometric case. At altitudes just below this, though, the photons encounter the cloud, and are scattered before reaching atmospheric layers below the cloud (whose base is at about 1200 km). For these photons, their path distribution is concentrated at altitudes above the cloud and in the cloud itself. 4.2. Exploration of the Importance of Scattering As most exoplanet transit spectrum models do not include multiple scattering, a key exercise is to explore the range of conditions for which scattering is expected to influence transit depth. For this exploration, we note that the transit depth attributed to a narrow s. annulus on the planetary disk (i.e., at a single ray impact parameter) scales with 2πb∆b/R2 Thus, in the absence of limb darkening, with the atmosphere transparent at radii larger than b + ∆b, and with the planetary disk centered on the star, only three parameters will influence the annulus transit depth for a scattering scenario -- the layer scattering optical depth, the aerosol scattering phase function, and the angular size of the stellar disk from the orbital location of the planet. Figure 13 show the relative difference between a transit spectrum model in the geometric and pure absorption limits (where scattering optical depth is treated as absorption optical depth) and a multiple scattering model for a single annulus on the planetary disk. Results are given for three different values of the host star angular size, which is determined by the ratio of the stellar radius to the planetary orbital distance (a). These values span a host star -- 17 -- angular size like that for Mercury and the Sun ( a = 0.39 au, so that Rs/a ∼ 10−2) to that of a hot Jupiter-like planet orbiting at 0.05 AU from a Sun-like star (Rs/a ∼ 10−1). Also, this range of host star angular sizes spans those for the Habitable Zones of M and K dwarf stars (Kopparapu et al. 2013). Color contours in Figure 13 are given as a function of layer scattering slant optical depth and the scattering asymmetry parameter. No other opacity source is added to the layer. For simplicity, a Henyey-Greenstein phase function is assumed (Henyey & Greenstein 1941), as using a multi-parameter phase function would introduce additional variables to our phase space exploration. Regardless, these results will still serve to indicate under which conditions scattering can become important. 4.3. Thin Clouds and Hazes in Hot Jupiter Transits The standard hot Jupiter-like atmospheric model outlined in Section 3 provides a useful test case for exploring some important aspects of scattering in exoplanet transit spectra, especially when considering that planets near to their host stars are those for which scattering can have the largest effects. The parameter space for such an exploration is extremely large, since the wavelength-dependent asymmetry parameter and scattering optical depth, as well as the cloud/haze vertical distribution, will all influence the transit spectrum. Exploring all of this phase space is certainly beyond the scope of this manuscript, so we adopt a straightforward set of conditions. Specifically, we investigate the impact of general, isolated aerosol layers, placed at different pressure levels in the model atmosphere. Rather than adopt a specific cloud or haze optical depth profile (e.g., Figure 8), we simply distribute the aerosol optical depth uniformly over a single pressure scale height. Scattering optical depths and asymmetry parameters are assumed gray, the single-scattering albedo is taken to be unity (i.e., pure scattering clouds), and, as before, a Henyey-Greenstein scattering phase function is adopted. Figure 14 shows transit spectra for strongly forward scattering cloud/haze particles (g = 0.95) over the 1 -- 2 µm wavelength range, which overlaps the accessible wavelength range for Hubble/WFC3 (Kimble et al. 2008). Figure 15 is similar, but for less strongly forward scattering cloud/haze particles (g = 0.90). Single-layer clouds were placed at pressures of 10−4, 10−3, 10−2, and 10−1 bar, and spanned a vertical range of d ln p = 1. The clearsky case is shown in black, while pure absorbing and scattering cases are shown in gray and purple, respectively. For each cloud pressure, three scenarios with different slant scattering optical depths are shown, with τs equal to 1, 10, and 100. This range of scattering optical depths spans the limits from a relatively optically thin aerosol layer to a relatively optically thick layer. -- 18 -- 5. Discussion The path distribution approach provides a coherent and computationally efficient method to computing transit spectra. Furthermore, analyzing the path distribution can provide an understanding of which atmospheric layers can contribute information to rays emerging from the planetary disk at a given impact parameter. For example, Figure 10 demonstrates how refraction bends rays to probe deeper atmospheric layers than would be encountered in the geometric limit. However, since the path distribution is intrinsic to just the planetary atmo- sphere, the other important effect of refraction -- bending rays such that they do not strike the stellar disk -- is not represented. This is a strength of the path distribution approach, since only a single distribution would need to be computed for identical atmospheric models for planets orbiting different host stars. Similarly, Figure 11 shows how, on average, Rayleigh scattering will scatter a small fraction of photons at large impact parameters "downward" to probe deeper parts of the at- mosphere. Conversely, rays at small impact parameters no longer probe the deep atmosphere, as they are scattered at larger altitudes. Of course, as Rayleigh scattering is not strongly forward scattering, it is unlikely that any significant fraction of the scattered photons will still intersect the stellar disk. Figure 12 shows similar scattering effects to those in Figure 11. However, the impact parameter where (roughly) deeper atmospheric layers become ineffectively probed depends on the vertical structure of the aerosols and, more specifically, where the slant scattering optical depth unity occurs. We note, again, that these figures display average path distributions for our Monte Carlo simulations, and that transit spectra that include scattering must be computed using using the formalism developed in Section 2.1. Ultimately, when considering scattering, the brightness of any given annulus (at impact parameter b) on the planetary disk will depend on the transparency of atmospheric layers at radii larger than b, the slant scattering optical depth of the layer at b, the scattering phase function for the aerosols in this layer, and the angular size of the host star as seen from the planet. For the situation where overlaying layers are transparent (implying that we can, thus, see down to a cloud), Figure 13 provides a rough guide to the circumstances where scattering can be important. Here, one can use simple details from a forward model -- the angular size of the host star, the scattering optical thickness of a cloud layer, and cloud scattering asymmetry parameter -- to determine if scattering can safely be ignored in a simulation. -- 19 -- In the case where the host star angular size is much smaller than would be at a distance of roughly 0.4 au from a Sun-like star (i.e., Rs/a ∼ 10−2), Figure 13 shows that essentially any amount of scattering will prevent the path of a ray from tracing back to the stellar disk (which is a very small "target"). With Rs/a at roughly 1 -- 5×10−2, which, for example, is appropriate for the Habitable Zones of late M dwarf stars, strongly forward scattering aerosols (g > 0.9) can have significant impacts on the transit depth due to an annulus for clouds whose slant optical depths are not very optically thick (i.e., τs less than 10). This effect becomes quite dramatic for hot Jupiter-like conditions, where the host star is a relatively large "target." Here, even modestly forward scattering aerosols (g ∼ 0.8) can cause substantial variations as compared to the commonly-used pure absorption assumption. This dependence on the asymmetry parameter explains why both Hubbard et al. (2001) and Brown (2001) found that multiple scattering was unimportant for their presented hot Jupiter transit spectra. As these authors considered only either Rayleigh scattering or isotropic phase functions (which both have g = 0), scattering could effectively be treated as absorption. Our adoption of the Henyey-Greenstein phase function is driven by the computational simplicity of this parameterization, and also the need to capture the process of forward scattering while minimizing the number of added parameters to our simulations. Neverthe- less, is has been shown that the Henyey-Greenstein phase function, when compared to Mie calculations, can underestimate the power in the forward scattering peak (Toublanc 1996; Boucher 1998), especially at wavelengths comparable to the particle size. Thus, depend- ing on the specifics of a given aerosol in an exoplanetary atmosphere, the results shown in Figure 13 may underestimate the increased transmission due to forward scattered light at a given asymmetry parameter (as compared to the pure absorption limit). Figures 14 and 15 further explore the importance of scattering in hot Jupiter transit spectra. For high-altitude clouds/hazes, these results show that differences between a pure absorption model versus a scattering model can approach (or exceed) 200 ppm, which is larger than the typical uncertainties achieved in current observations with Hubble (Kreidberg et al. 2014) or that are expected for JWST (Greene et al. 2016). Very optically thick clouds (i.e., with slant scattering optical depths approaching 100) approach the pure absorption limit, and the difference between a scattering model and a pure absorption model for a thin cloud (i.e., scattering optical depth less than about unity) is small since the overall impact of the cloud on the transit spectrum is weak. Similarly, the impact of scattering for a deep cloud or haze layer is small, as few wavelength regions are sensitive to the presence of this aerosol layer. In general, the impact of aerosol multiple scattering will depend on cloud configurations and optical properties, and this manuscript does not undertake the complex microphysical -- 20 -- calculations needed to attempt predictions of cloud/haze distributions and compositions in exoplanet atmospheres -- our goal is to highlight the conditions where scattering becomes an important consideration. Nevertheless, the results in Figures 14 and 15 indicate that forward scattering exoplanet clouds analogous to Earth's cirrus clouds (which are at higher altitudes and have optical depths less than roughly unity) may warrant a multiple scattering approach. Additionally, any forward scattering cloud with a scale height comparable to (or greater than) the pressure scale height could require a multiple scattering model to properly simulate a transit spectrum, as transit depth variations due to multiple scattering effects in this clouds will be comparable to those of molecular features (whose scale are typically set by the pressure scale height). Differences between the scattering and pure absorption models for the hot Jupiter cases will also depend on the asymmetry parameter and single-scattering albedo. For very strongly forward scattering aerosols (i.e., asymmetry parameters larger than our adopted value of 0.95), the transit spectrum will approach the clearsky limit (if the aerosols are weakly- or non- absorbing). In essence, such clouds would be invisible, having no effect on the transit depth. On the other hand, asymmetry parameters much smaller than roughly 0.8 -- 0.9 will push the scattering transit spectra towards the pure absorption limit, since these scattered photons are increasingly unlikely to connect to the observer. Similarly, as the single-scattering albedo is decreased (away from our adopted value of unity), the scattering spectra will approach the pure absorption case. The extent of the influence of the single-scattering albedo depends on the average number of scatterings the photons experience along their path. For example, for our τs = 10 case, the photons are scattered of order ten times, implying a single-scattering albedo of less than about 0.9 would reduce the transmission to roughly 50%. Perhaps the most important message from the scattering studies above is the influence of the scattering asymmetry parameter on transit spectra of close-in exoplanets. Aerosols whose scattering properties vary from strongly forward scattering to weakly forward scattering could impart features in a transit spectrum, as the cloud (or haze layer) will be more transparent where the particles have a larger asymmetry parameter. Such signatures would appear along with absorption bands caused by particle vibrational modes (Wakeford & Sing 2015) and cloud base features (Vahidinia et al. 2014). Additionally, as transit spectra of forward scattering clouds would be reproduced by relatively thinner clouds in the pure absorption limit, transit spectral retrievals using pure absorption models could return cloud thickness (or number densities) that are biased low, where retrieved cloud optical depths would, instead, represent an effective optical depth that incorporates information about the optical properties of the cloud particles and the stellar angular size. Whether or not aerosol multiple scattering will be important for any given exoplanet -- 21 -- transit spectrum will depend on a number of parameters, including the slant optical thickness of any haze/cloud structure and the scattering asymmetry parameter (which will both, in general, depend on wavelength). The planet must also be located near to its host star, which limits the range of relevant aerosols to those which form in warm and hot atmospheric conditions. For example, asymmetry parameters for water droplets in the visible wavelength range are 0.8 -- 0.9 (Kokhanovsky 2004), and span 0.74 -- 0.94 for ice crystals (where crystals with plate-like morphologies are the most strongly forward scattering; Macke et al. 1998). Thus, conditions may be appropriate for multiple scattering to influence the transit spectra of potentially habitable worlds around the coolest stars (e.g., TRAPPIST-1 Gillon et al. 2016). Looking beyond Earth in the Solar System, De Kok & Stam (2012) (their Figure 1) show that a variety of ices, dusts, droplets, and fractal hazes are both non-absorptive and forward scattering (with g (cid:38) 0.9) below 2 -- 3 µm. The wavelength range for these Solar System cases is especially relevant to transiting exoplanets around later-type stars that may be investigated by Hubble, JWST, the Fast Infrared Exoplanet Spectroscopy Survey Explorer concept (FINESSE; Deroo et al. 2012), and/or the Atmospheric Remote-Sensing Infrared Exoplanet Large-survey concept (ARIEL; Tinetti et al. 2016). The composition, size distributions, and optical properties of aerosols in warm and hot exoplanet atmospheres are largely unknown. For the hottest exoplanets, metal and silicate condensates may be expected to form (Lunine et al. 1989; Marley et al. 1999; Burrows et al. 2001). Budaj et al. (2015) investigated the optical properties of a large variety of metallic and silicon-bearing aerosols, and showed that many of these species can be strongly forward scattering over certain wavelength ranges. Especially relevant cases are aluminum oxide, forsterite, enstatite, and pyroxine. It is interesting to note that our results show that multiple scattering can be an impor- tant consideration for exoplanet transits when the angular size of the host star is relatively large, which is the opposite regime from where refraction has been shown to be important (Bétrémieux & Kaltenegger 2014; Misra et al. 2014; Bétrémieux & Kaltenegger 2015). For refraction, a host star with small angular size implies that relatively little refractive bending is required to deflect a ray off the stellar disk, thereby setting a floor in the transit spectrum which is not captured in the commonly-used geometric, pure absorption limit. By compar- ison, only a single scattering (with even a strong forward peak) is required to deflect a ray from the stellar disk, which is then in agreement with the geometric, pure absorption limit. Of course, the opposite of these statements is true for a star with large angular size. In future work we plan to further explore the importance of refraction in transit observations of gas giants. Finally, since this manuscript primarily emphasizes the development of a light scattering -- 22 -- theory for exoplanet transits and the impacts of vertically thin cloud layers, a number of future studies will be undertaken. These will include scattering effects in vertically more- extended cloud structures, and could also investigate the role of the assumed (or computed) particle scattering phase function. Finally, the formalism outlined above can be used to study light scattering effects in time-resolved transit lightcurves. 6. Conclusions We have detailed a new theory of exoplanet transit spectroscopy that includes the effects of light multiple scattering. By effectively separating the path that photons take through an exoplanet atmosphere from the absorption processes of that atmosphere, the technique discussed in this manuscript yields models that are both physically rigorous and computationally efficient. This approach, which relies on the so-called "path distribution" defined herein, can be extended to other areas of study, including (most straightforwardly) stellar occultations by planetary atmospheres. When applying our validated scattering model to isolated cloud layers, models show that multiple scattering is most important for cases where the exoplanet host star is large in angular size as seen from the world (i.e., when the orbital distance is less than 10 -- 20 times the stellar radius). In these cases, multiple scattering by aerosols with asymmetry parameters larger than 0.8 -- 0.9 can have substantial effects on the transmission of the cloud layer. For all cases, differences between a multiple scattering model and a geometric pure absorption diminish for clouds (or hazes) with slant scattering optical depths approaching 100. In an exploratory case of a conservatively and forward (g =0.90 -- 0.95) scattering cloud/haze layer in the atmosphere of a hot Jupiter, differences in the transit depth from a multiple scattering model and a model in the pure absorption limit can exceed 200 ppm. These dif- ferences are most pronounced when the cloud/haze is well above the level of gas absorption optical depth unity, and when the slant scattering optical depth is of order several to several tens. Future work will further explore the situations where scattering in more extended exoplanet cloud (or haze) layers is key to understanding transit observations. TR gratefully acknowledges support from NASA through the Sagan Fellowship Program executed by the NASA Exoplanet Science Institute. The results reported herein benefitted from collaborations and/or information exchange within NASA's Nexus for Exoplanet Sys- tem Science (NExSS) research coordination network sponsored by NASA's Science Mission Directorate. Certain essential tools used in this work were developed by the NASA Astro- -- 23 -- biology Institute's Virtual Planetary Laboratory, supported by NASA under Cooperative Agreement No. NNA13AA93A. The author thanks M. Line and J. Lustig-Yaeger for sharing model outputs for inter-comparison purposes, and J. Fortney and R. de Kok for providing helpful feedback. This manuscript is dedicated to the memory of Prince, who passed away during the completion of this project -- "thank you for a funky time." A. Appendix: Uses of the Path Distribution Given the ray atmospheric path distribution as defined in Section 2, a number of key quantities can be derived. First, the pathlength (in km, for example) traversed between any two altitudes, h1 and h2, by a ray incident with impact parameter b is simply, sb(h1, h2) = Pb(h) dh , (A1) where the absolute value forces all distances to be measured non-negative. Then, if h = 0 at the bottom of a terrestrial planetary atmosphere or at some reference pressure (e.g., 10 bar) for a gaseous world, and ht is the effective top of the atmosphere, the total linear distance traverse through the entire atmosphere by the ray is, (cid:90) h2 h1 (cid:90) ht sb(0, ht) = Pb(h) dh , (A2) where we note that this integral would diverge if taken to infinite altitude. 0 The pathlength integral can also be written in terms of two pressure coordinates, p1 and p2, assuming the ideal gas law and hydrostatic equilibrium, sb(p1, p2) = Pb(p) RgT (p) mg d ln p = Pb(p) H(p) d ln p , (A3) (cid:90) p1 p2 (cid:90) p1 p2 where Rg is the universal gas constant, T (p) is the atmospheric temperature profile, g is the acceleration due to gravity, m is the atmospheric mean molecular weight, and H(p) is the pressure scale height. Also, given the definition of the path distribution, and the differential definition of (path) column mass from the atmospheric mass density profile [ρ(h)], dMc = ρ(h)ds , (A4) the air mass encountered by a ray with impact parameter b between two altitudes, h1 and h2, is given by, Mc,b(h1, h2) = ρ(h)Pb(h) dh . (A5) (cid:90) h2 h1 Similarly, a column (path) number density is, -- 24 -- (cid:90) h2 Nc,b(h1, h2) = n(h)Pb(h) dh , (A6) where n(h) is the number density profile. h1 B. Appendix: Geometric Limit In the geometric limit, rays pass straight through the planetary atmosphere. Here, the path distribution can be determined analytically using geometric arguments. For an atmospheric layer centered at h with width ∆h, and for a ray incident on the atmosphere with impact parameter b, the path distribution is given by  0 , 2 ∆h 2 ∆h Pb(h) = (cid:113) (cid:20)(cid:113) (r + ∆h/2)2 − b2 −(cid:113) (r + ∆h/2)2 − b2 , b ≥ r + ∆h/2 r − ∆h/2 < b < r + ∆h/2 b ≤ r − ∆h/2 . (cid:21) , (r − ∆h/2)2 − b2 (B1) where we have defined r = Rp + h for conciseness. Note that this expression is only indepen- dent of ∆h in the limit that this value is small when compared to r. Performing a first-order expansion in ∆h yields, Pb(h) ≈ 2r√ r2 − b2 , (B2) which is in agreement with the linearized geometry discussed in Fortney (2005) (their Fig- ure 1). The geometric path distribution need only be computed once for a model atmosphere. Thus, when paired with Equation 7, the geometric approach can be executed with great com- putational efficiency. C. Appendix: Refraction-Only Cases When considering refraction, rays no longer travel on straight-line paths through the atmosphere. In this case, the path distribution must be computed numerically. Fortunately, as the refractive indexes for gases likely to be major atmospheric constituents vary weakly in wavelength, the path distribution need only be computed at a small number of spectral points when generating a transit spectrum. -- 25 -- Refraction will deflect the trajectory of a ray upward or downward, depending on the sign of the local atmospheric refractive index profile. Here, the local curvature experienced by the ray, rc, is, 1 rc = sin (θr) d ln nref dh , (C1) where θr is the zenith angle for the ray trajectory, and nref is the atmospheric refractive index. The zenith angle is determined via, cos(θr) = r · µ . (C2) Using pathlength as the integration variable, and according to the geometry shown in Fig- ure 2, we have, = cos (θr) , (cid:20)sin (θr) dh ds = − (cid:21) + 1 rc r dθr ds , (C3) (C4) (C5) (C6) and dω ds = dφ ds = sin (θr) , r 1 rc , where φ is the polar angle measured from the +x direction, and ω is the so-called refraction integral (which measures the deflection from the initial straight-line trajectory). For a ray with initial impact parameter bi, the path is determined by numerically inte- grating the expressions above using an algorithm like that described in van der Werf (2008). While the ray is passing through the atmospheric layer at hj, the increments of ∆s used in the path integration are divided by ∆hj and added to Pi,j. The path integration proceeds through all atmospheric layers until either the ray exits the atmosphere or the planetary surface is struck. When the ray exits the atmosphere, the radial coordinate and trajectory are stored for later use in determining which rays map back to the stellar disk. The straight- line trajectory the ray follows after exiting the atmosphere is defined by the total refraction angle at exit, which is equal to the direction cosine of the ray along the x-axis in Figure 2 (i.e., µ · x = µx = ω). Ackerman, A. S., & Marley, M. S. 2001, ApJ, 556, 872 REFERENCES -- 26 -- Barstow, J. K., Aigrain, S., Irwin, P. G. J., et al. 2012, ArXiv e-prints, arXiv:1212.5020 Barstow, J. K., Aigrain, S., Irwin, P. G. J., Kendrew, S., & Fletcher, L. N. 2016, MNRAS, 458, 2657 Batalha, N., Kalirai, J., Lunine, J., Clampin, M., & Lindler, D. 2015, ArXiv e-prints, arXiv:1507.02655 Bean, J. L., Kempton, E. M.-R., & Homeier, D. 2010, Nature, 468, 669 Beichman, C., Benneke, B., Knutson, H., et al. 2014, Publications of the Astronomical Society of the Pacific, 126, 1134 Benneke, B., & Seager, S. 2012, The Astrophysical Journal, 753, 100 Bétrémieux, Y. 2016, MNRAS, 456, 4051 Bétrémieux, Y., & Kaltenegger, L. 2014, ApJ, 791, 7 Bétrémieux, Y., & Kaltenegger, L. 2015, Monthly Notices of the Royal Astronomical Society, 451, 1268 Boucher, O. 1998, Journal of Atmospheric Sciences, 55, 128 Brown, T. M. 2001, The Astrophysical Journal, 553, 1006 Budaj, J., Kocifaj, M., Salmeron, R., & Hubeny, I. 2015, MNRAS, 454, 2 Burrows, A., Hubbard, W. B., Lunine, J. I., & Liebert, J. 2001, Reviews of Modern Physics, 73, 719 Charbonneau, D., Brown, T. M., Noyes, R. W., & Gilliland, R. L. 2002, ApJ, 568, 377 Cowan, N. B., Greene, T., Angerhausen, D., et al. 2015, PASP, 127, 311 Dalba, P. A., Muirhead, P. S., Fortney, J. J., et al. 2015, ApJ, 814, 154 De Kok, R., & Stam, D. 2012, Icarus, 221, 517 Deming, D., Seager, S., Winn, J., et al. 2009, PASP, 121, 952 Deroo, P., Swain, M. R., & Green, R. O. 2012, in Proc. SPIE, Vol. 8442, Space Telescopes and Instrumentation 2012: Optical, Infrared, and Millimeter Wave, 844241 Fortney, J. J. 2005, Monthly Notices of the Royal Astronomical Society, 364, 649 -- 27 -- Fraine, J., Deming, D., Benneke, B., et al. 2014, Nature, 513, 526 Gardner, J. P., Mather, J. C., Clampin, M., et al. 2006, Space Sci. Rev., 123, 485 Gillon, M., Jehin, E., Lederer, S. M., et al. 2016, Nature, 533, 221 Greene, T. P., Line, M. R., Montero, C., et al. 2016, The Astrophysical Journal, 817, 17 Gunson, M. R., Abbas, M., Abrams, M., et al. 1996, Geophysical Research Letters, 23, 2333 Helling, C., Oevermann, M., Lüttke, M. J. H., Klein, R., & Sedlmayr, E. 2001, A&A, 376, 194 Henyey, L. G., & Greenstein, J. L. 1941, ApJ, 93, 70 Hubbard, W., Fortney, J., Lunine, J., et al. 2001, The Astrophysical Journal, 560, 413 Hui, L., & Seager, S. 2002, The Astrophysical Journal, 572, 540 Irion, F. W., Gunson, M. R., Toon, G. C., et al. 2002, Appl. Opt., 41, 6968 Irwin, P. G. J., Teanby, N. A., de Kok, R., et al. 2008, J. Quant. Spec. Radiat. Transf., 109, 1136 Kaltenegger, L., & Traub, W. 2009, The Astrophysical Journal, 698, 519 Kimble, R. A., MacKenty, J. W., O'Connell, R. W., & Townsend, J. A. 2008, in Proc. SPIE, Vol. 7010, Space Telescopes and Instrumentation 2008: Optical, Infrared, and Mil- limeter, 70101E Kipping, D. M., & Tinetti, G. 2010, MNRAS, 407, 2589 Knutson, H. A., Benneke, B., Deming, D., & Homeier, D. 2014a, Nature, 505, 66 Knutson, H. A., Dragomir, D., Kreidberg, L., et al. 2014b, ApJ, 794, 155 Kokhanovsky, A. 2004, Earth Science Reviews, 64, 189 Kopparapu, R. K., Ramirez, R., Kasting, J. F., et al. 2013, ApJ, 765, 131 Kreidberg, L. 2015, PASP, 127, 1161 Kreidberg, L., Bean, J. L., Désert, J.-M., et al. 2014, Nature, 505, 69 Lee, G., Dobbs-Dixon, I., Helling, C., Bognar, K., & Woitke, P. 2016, A&A, 594, A48 -- 28 -- Lee, J.-M., Irwin, P. G. J., Fletcher, L. N., Heng, K., & Barstow, J. K. 2014, ApJ, 789, 14 Line, M. R., Knutson, H., Deming, D., Wilkins, A., & Desert, J.-M. 2013, ApJ, 778, 183 Line, M. R., Zhang, X., Vasisht, G., et al. 2012, ApJ, 749, 93 Line, M. R., Wolf, A. S., Zhang, X., et al. 2013, The Astrophysical Journal, 775, 137 Lunine, J. I., Hubbard, W. B., Burrows, A., Wang, Y.-P., & Garlow, K. 1989, ApJ, 338, 314 Macke, A., Francis, P. N., McFarquhar, G. M., & Kinne, S. 1998, Journal of Atmospheric Sciences, 55, 2874 Mandel, K., & Agol, E. 2002, ApJ, 580, L171 Marley, M. S., Ackerman, A. S., Cuzzi, J. N., & Kitzmann, D. 2013, in Comparative Clima- tology of Terrestrial Planets, ed. S. J. Mackwell, A. A. Simon-Miller, J. W. Harder, & M. A. Bullock (Tucson: University of Arizona Press), 367 -- 391 Marley, M. S., Gelino, C., Stephens, D., Lunine, J. I., & Freedman, R. 1999, ApJ, 513, 879 McClatchey, R. A., Fenn, R. W., Selby, J. E. A., Volz, F. E., & Garing, J. S. 1972, Opti- cal Properties of the Atmosphere (Third Edition), Tech. rep., Air Force Cambridge Research Labs Meadows, V. S., & Crisp, D. 1996, J. Geophys. Res., 101, 4595 Misra, A., Meadows, V., & Crisp, D. 2014, ApJ, 792, 61 Morley, C. V., Fortney, J. J., Kempton, E. M.-R., et al. 2013, The Astrophysical Journal, 775, 33 Morley, C. V., Knutson, H., Line, M., et al. 2016, ArXiv e-prints, arXiv:1610.07632 Muñoz, A. G., Osorio, M. Z., Barrena, R., et al. 2012, The Astrophysical Journal, 755, 103 Pallé, E., Osorio, M. R. Z., Barrena, R., Montañés-Rodríguez, P., & Martín, E. L. 2009, Nature, 459, 814 Pont, F., Knutson, H., Gilliland, R., Moutou, C., & Charbonneau, D. 2008, Monthly Notices of the Royal Astronomical Society, 385, 109 Seager, S., & Sasselov, D. 2000, The Astrophysical Journal, 537, 916 Sidis, O., & Sari, R. 2010, The Astrophysical Journal, 720, 904 -- 29 -- Sing, D., Désert, J.-M., Lecavelier Des Etangs, A., et al. 2009, Astronomy and Astrophysics, 505, 891 Sing, D. K., Fortney, J. J., Nikolov, N., et al. 2016, Nature, 529, 59 Stamnes, K., Tsay, S.-C., Jayaweera, K., Wiscombe, W., et al. 1988, Applied optics, 27, 2502 Stevenson, K. B. 2016, ApJ, 817, L16 Swain, M. R., Vasisht, G., & Tinetti, G. 2008, Nature, 452, 329 Tinetti, G., Drossart, P., Eccleston, P., et al. 2016, in Proc. SPIE, Vol. 9904, Society of Photo-Optical Instrumentation Engineers (SPIE) Conference Series, 99041X Tomasko, M., Doose, L., Engel, S., et al. 2008, Planetary and Space Science, 56, 669 Toublanc, D. 1996, Applied optics, 35, 3270 Vahidinia, S., Cuzzi, J. N., Marley, M., & Fortney, J. 2014, ApJ, 789, L11 van der Werf, S. 2008, Applied optics, 47, 153 Wakeford, H. R., & Sing, D. K. 2015, A&A, 573, A122 Waldmann, I. P., Tinetti, G., Rocchetto, M., et al. 2015, ApJ, 802, 107 West, R., Crisp, D., & Chen, L. 1990, J. Quant. Spec. Radiat. Transf., 43, 191 Witt, A. N. 1977, ApJS, 35, 1 This preprint was prepared with the AAS LATEX macros v5.0. -- 30 -- D. Tables and Figures -- 31 -- Table 1. Symbol Usage Description vector/array of area elements planet-star orbital distance absorptivity along a ray path atmospheric extinction coefficient impact parameter projected separation between planet center and star center scattering asymmetry parameter, acceleration due to gravity pressure scale height altitude, altitude at effective top of atmosphere stellar surface brightness surface brightness on planetary disk disk-averaged stellar surface brightness background to planetary disk surface brightness mapping atmospheric mean molecular weight cosine of scattering angle cosine of angle of incidence on stellar disk direction of photon/ray travel number of impact parameters in model grid number of photons in a Monte Carlo simulation number of layers in atmospheric model atmospheric refractive index scattering phase function atmospheric path distribution pressure polar angle when ray tracing, scattering azimuth angle Earth radius universal gas constant Jupiter radius fiducial planetary radius wavelength-dependent planetary radius stellar radius radial distance from planet center integration bounds for stellar brightness/area ray curvature due to refraction ray path transmission along ray path Monte Carlo average transmission vector layer vertical optical depth optical depth along ray path or slant optical depth local zenith angle for ray propagation direction scattering deflection angle, polar angle in disk integration solid angle, differential solid angle refraction integral/angle single-scattering albedo random number between zero and one ∂¯tλ/∂∆τ λ,a Monte Carlo average transmission Jacobian matrix Symbol A a aλ αλ b d g H = RgT /mg h, ht Is,λ Ip,λ I0,λ I m µ = cos(θ) µs µ Nr Np Nlay nref P (µ) Pb, P p φ RE Rg RJ Rp Rp,λ Rs r r0, θ0 rc sb tλ ¯tλ ∆τ τ θr θ Ω, dΩ ω ω0 ξ -- 32 -- Fig. 1. -- : Visualization of the path distribution for a ray incident exiting a planetary atmosphere with impact parameter b and passing through an atmospheric layer of width ∆h centered at altitude h. Rpto starto observerh+RpΔhbb(h)Δh -- 33 -- Fig. 2. -- : Visualization of key parameters in computing the path distribution in the non- geometric limit case. Note how ray direction is reversed, as paths are traced backwards to investigate whether they strike the stellar disk. to starto observerbr = Rp+ hxzyɸ-𝜔ϴr𝒓#𝝁% -- 34 -- Fig. 3. -- : Visualization of geometry for integrating planetary and stellar surface brightness when generating a transit spectrum. dRsRprhtϴ -- 35 -- Fig. 4. -- : Result of a full Monte Carlo transit spectrum simulation prior to solid angle integration. The wavelength is red-visible (0.8 µm), hot Jupiter-like conditions are assumed, refraction and Rayleigh scattering are included, and a forward scattering (g = 0.9) haze is placed above the 1 µbar pressure level. The right sub-figure has the surface brightness off the stellar disk enhanced by a factor of 10. -- 36 -- Fig. 5. -- : Comparison of the output radiances (scaled by the incident flux) from our Monte Carlo routine, in the plane-parallel limit, against those from the DISORT radiative transfer model (Stamnes et al. 1988). Clouds with a given scattering optical depth and asymmetry parameter are placed over a black surface, and a Henyey-Greenstein phase function is as- sumed. Results are shown as a function of the observer zenith angle for both the forward and backward scattering azimuths in the plane of the incident light source. -- 37 -- Fig. 6. -- : Comparison between our transit spectrum model and the refracting model of Misra et al. (2014) for a standard Earth atmospheric model (McClatchey et al. 1972). Differences between the models are shown in the lower sub-panel, and are typically within 60 ppb. -- 38 -- Fig. 7. -- : Comparison between our transit spectrum model, in the geometric limit, and output from the CHIMERA retrieval suite (Line et al. 2013) for a standard hot Jupiter-like model atmosphere described in the text. Differences between the models are shown in the lower sub-panel, and are typically within 30 ppm. -- 39 -- Fig. 8. -- : Layer differential extinction optical depths for a nominal haze-like model and a condensate cloud-like model (after Ackerman & Marley 2001) used when computing example path distributions for a hot Jupiter-like atmosphere. For comparison, the layer differential Rayleigh scattering optical depths (at 0.5 µm) for out hot Jupiter-like atmosphere are also shown. -- 40 -- Fig. 9. -- : The path distribution, Pb(h), in the geometric limit for a hot Jupiter-like at- mosphere. Altitudes are relative to the 10-bar planetary radius at 1.16RJ. Darker colors indicate larger path distribution values, which implies a larger enhancement of the vertical differential optical depth. The largest enhancements occur when a ray is passing near the horizontal for a layer, which is where b − Rp is roughly equal h. -- 41 -- Fig. 10. -- : The path distribution, Pb(h), for a cloud-free Earth atmosphere that includes the effects of refraction. A dashed line is given along the diagonal. The right sub-figure shows, for comparison, the path distribution without refraction. The path distribution for rays with impact parameter altitudes below 1.6 km are not shown as these rays strike the surface. Rays with impact parameter altitudes smaller than about 10 km are deflected downward due to refraction, and thus probe deeper atmospheric layers than in the geometric case. Here the peak of the path distribution is also shifted to slightly lower altitudes. -- 42 -- Fig. 11. -- : The average path distribution, ¯P, including molecular Rayleigh scattering at 0.55 µm wavelength for a hot Jupiter-like atmosphere. Photons (or rays) with impact pa- rameter altitudes above about 1000 km only probe low pressure regions of the atmosphere, and experience relatively little Rayleigh scattering. Photons with impact parameters below about 400 -- 600 km experience substantial Rayleigh scattering, and are typically scattered before reaching deeper atmospheric layers. Thus, rays/photons with low impact parameter altitudes have small path distributions at depth and relatively large path distributions aloft. -- 43 -- Fig. 12. -- : The average path distribution, ¯P, for a hot Jupiter-like atmosphere where multiple scattering in an upper atmospheric haze (left) and condensate cloud (right) are considered. Aerosol optical depth distributions are shown in Figure 8, the single-scattering albedo is unity, and the asymmetry parameter is taken as forward scattering (g = 0.9). The haze case resembles the Rayleigh scattering case in Figure 11 due to the power-law distribu- tion of haze scattering optical depth. The condensate cloud case is identical to the geometric case for rays/photons with impact parameters above the cloud. For impact parameters below (or in) the cloud, scattering prevents photons from reaching deeper atmospheric layers, so that the path distribution is distrinctly different above versus below the cloud. -- 44 -- Fig. 13. -- : Relative difference between the transit depth due to a single annulus in the geometric limit versus a model that includes multiple scatterings. Sub-figures are for different angular sizes of the host star as seen from the planet, and results are given as a function of scattering slant optical depth within the annulus and the scattering asymmetry parameter. -- 45 -- Fig. 14. -- : Near-infrared transit spectra for a hot Jupiter-like planet. Different panels are for cases where a vertically thin cloud is placed at the indicated pressure level. Different line styles indicate cases with different slant scattering optical depths for the cloud. Gray lines assume that all optical depth is absorption optical depth, while purple lines include realistic scattering. The clearsky limit is shown in black. Aerosol optical properties are assumed to be gray, a Henyey-Greenstein scattering phase function is used, and the aerosols are taken to be forward scattering, with g = 0.95. -- 46 -- Fig. 15. -- : The same as Figure 14 except with g = 0.90.
1805.09353	1	1805	2018-05-23T18:02:02	The Reactivation and Nucleus Characterization of Main-Belt Comet 358P/PANSTARRS (P/2012 T1)	[ "astro-ph.EP" ]	We present observations of main-belt comet 358P/PANSTARRS (P/2012 T1) obtained using the Gemini South telescope from 2017 July to 2017 December, as the object approached perihelion for the first time since its discovery. We find best-fit IAU phase function parameters of H_R=19.5+/-0.2 mag and G_R=-0.22+/-0.13 for the nucleus, corresponding to an effective radius of r_N=0.32+/-0.03 km (assuming an albedo of p_R=0.05). The object appears significantly brighter (by >1 mag) than expected starting in 2017 November, while a faint dust tail oriented approximately in the antisolar direction is also observed on 2017 December 18. We conclude that 358P has become active again for the first time since its previously observed active period in 2012-2013. These observations make 358P the seventh main-belt comet candidate confirmed to exhibit recurrent activity near perihelion with intervening inactivity away from perihelion, strongly indicating that its activity is sublimation-driven. Fitting a linear function to the ejected dust masses inferred for 358P in 2017 when it is apparently active, we find an average net dust production rate of 2.0+/-0.6 kg/s (assuming a mean effective particle radius of 1 mm) and an estimated activity start date of 2017 November 8+/-4 when the object was at a true anomaly of 316+/-1 deg and a heliocentric distance of R=2.54 AU. Insufficient data is currently available to ascertain whether activity strength has changed between the object's 2012-2013 and 2017 active periods. Further observations are therefore highly encouraged during the object's upcoming observing window (2018 August through 2019 May).	astro-ph.EP	astro-ph	Draft version 2018-09-12 Typeset using LATEX twocolumn style in AASTeX62 The Reactivation and Nucleus Characterization of Main-Belt Comet 358P/PANSTARRS (P/2012 T1) Henry H. Hsieh,1, 2 Masateru Ishiguro,3 Matthew M. Knight,4 Marco Micheli,5, 6 Nicholas A. Moskovitz,7 Scott S. Sheppard,8 and Chadwick A. Trujillo9 1Planetary Science Institute, 1700 East Fort Lowell Rd., Suite 106, Tucson, AZ 85719, USA 2Institute of Astronomy and Astrophysics, Academia Sinica, P.O. Box 23-141, Taipei 10617, Taiwan 3Department of Physics and Astronomy, Seoul National University, Gwanak, Seoul 151-742, Korea 4Department of Astronomy, University of Maryland, 1113 Physical Sciences Complex, Building 415, College Park, MD 20742, USA 5ESA SSA-NEO Coordination Centre, Largo Galileo Galilei, 1, 00044 Frascati (RM), Italy 6INAF - Osservatorio Astronomico di Roma, Via Frascati, 33, 00040 Monte Porzio Catone (RM), Italy 7Lowell Observatory, 1400 W. Mars Hill Rd, Flagstaff, AZ 86001, USA 8Department of Terrestrial Magnetism, Carnegie Institution for Science, 5241 Broad Branch Road NW, Washington, DC 20015, USA 9Department of Physics and Astronomy, Northern Arizona University, Flagstaff, AZ 86001, USA (Received May 2, 2018; Revised May 21, 2018; Accepted May 23, 2018) Submitted to AJ ABSTRACT We present observations of main-belt comet 358P/PANSTARRS (P/2012 T1) obtained using the Gemini South telescope from 2017 July to 2017 December, as the object approached perihelion for the first time since its discovery. We find best-fit IAU phase function parameters of HR = 19.5 ± 0.2 mag and GR = −0.22 ± 0.13 for the nucleus, corresponding to an effective radius of rN = 0.32 ± 0.03 km (assuming an albedo of pR = 0.05). The object appears significantly brighter (by ≥ 1 mag) than expected starting in 2017 November, while a faint dust tail oriented approximately in the antisolar direction is also observed on 2017 December 18. We conclude that 358P has become active again for the first time since its previously observed active period in 2012-2013. These observations make 358P the seventh main-belt comet candidate confirmed to exhibit recurrent activity near perihelion with intervening inactivity away from perihelion, strongly indicating that its activity is sublimation-driven. Fitting a linear function to the ejected dust masses inferred for 358P in 2017 when it is apparently M = 2.0± 0.6 kg s−1 (assuming a mean effective active, we find an average net dust production rate of particle radius of ¯ad = 1 mm) and an estimated activity start date of 2017 November 8 ± 4 when the object was at a true anomaly of ν = 316◦ ± 1◦ and a heliocentric distance of R = 2.54 AU. Insufficient data is currently available to ascertain whether activity strength has changed between the object's 2012-2013 and 2017 active periods. Further observations are therefore highly encouraged during the object's upcoming observing window (2018 August through 2019 May). Keywords: comets: general -- comets: individual (358P/PANSTARRS) -- minor planets, asteroids: general 1. INTRODUCTION 1.1. Background Comet formerly designated P/2012 T1 (PANSTARRS), was discovered on UT 2012 358P/PANSTARRS, Corresponding author: Henry H. Hsieh [email protected] October 6 by the Pan-STARRS1 (PS1) survey telescope on Haleakala (Wainscoat et al. 2012). At the time of its discovery, it was the seventh object to be identified as a likely main-belt comet (MBC) (Hsieh et al. 2013; Moreno et al. 2013), where about a dozen such objects are known to date. MBCs exhibit activity in the form of comet-like dust emission that has been determined to be at least par- tially due to the sublimation of volatile ice, yet occupy 2 Hsieh et al. stable orbits in the main asteroid belt (Hsieh & Jewitt 2006). They comprise a subset of the population of ac- tive asteroids (Jewitt et al. 2015a), which include all objects that exhibit comet-like activity due to a variety of mechanisms or combination of mechanisms, including sublimation, impact disruption, and rotational destabi- lization, yet have dynamically asteroidal orbits. Small solar system bodies are commonly considered dynami- cally asteroidal if they have Tisserand parameter values of TJ > 3 (Kres´ak 1979), although in practice, the dy- namical transition zone between asteroids and comets actually appears to lie roughly between TJ = 3.05 and TJ = 3.10 (Tancredi 2014; Jewitt et al. 2015a; Hsieh & Haghighipour 2016). While some MBCs may be Jupiter-family comets (JFCs) that have recently evolved onto main-belt-like orbits, others reside in regions of orbital element space that are largely unreachable by interloping JFCs, and could potentially have formed in situ (Hsieh & Haghigh- ipour 2016). This raises the intriguing possibility that they might be able to help constrain the temperature and composition of objects in this region in the early solar system, and also provide a means to explore the possibility that icy material originally from the main- belt region of the solar system, or at least icy material similar in composition to objects currently occupying the main asteroid belt, could have been a significant primordial source of terrestrial water (Morbidelli et al. 2000; Raymond et al. 2004; Raymond & Izidoro 2017; O'Brien et al. 2006, 2018, and references within). 358P is one of four MBCs that were found to have or- bital elements (specifically relatively high eccentricities and inclinations) similar to those taken on by test parti- cles that were found by Hsieh & Haghighipour (2016) to temporarily transition from initially JFC-like orbits to main-belt-like orbits in numerical integrations. In that work, however, most test particles exhibiting that type of dynamical behavior were not found to remain on their adopted main-belt orbits for very long (<20 Myr). Nu- merical integrations specifically focused on 358P found that the object is largely dynamically stable over a timescale of 100 Myr (Hsieh et al. 2013), suggesting that the likelihood of a JFC-like origin, while non-zero, is low. The direct detection of sublimation products (i.e., gas or vapor) from MBCs has proven to be extremely challenging to achieve using currently available obser- vational facilities (Snodgrass et al. 2017b). Attempts to specifically search for outgassing from 358P using the Keck Observatory, Very Large Telescope, and Her- schel Space Telescope yielded only upper limit pro- duction rates of Q(CN) < 1.5 × 1023 molecules s−1, Q(H2O) < 7.6 × 1025 molecules s−1, and Q(OH) < 6× 1025 molecules s−1, respectively, all of which roughly correspond to similar upper limit water production rates of ∼ 8× 1025 molecules s−1 (Hsieh et al. 2013; O'Rourke et al. 2013; Snodgrass et al. 2017a). As such, for now, the identification of the driver of an active asteroid's ac- tivity must be accomplished by indirect methods such as numerical dust modeling or the identification of re- current activity after intervening periods of inactivity. Dust modeling can allow for the determination of the duration of the period of active dust emission for an ob- ject, which can then indicate what mechanism or com- bination of mechanisms is likely to be responsible for that emission event. A short-duration (i.e., impulsive) dust emission event is most likely to be caused by an impact, while sublimation and rotational disruption are more plausible explanations for longer lasting dust emis- sion events. In addition to ambiguity as to whether a long-lasting emission event could have been caused by sublimation or rotational disruption, though, dust mod- eling is also susceptible to parameter degeneracies which could lead to misleading results depending on parameter spaces and emission scenarios being explored (cf. Hsieh et al. 2012). Meanwhile, recurrent activity interspersed with peri- ods of inactivity, particularly if it is periodic and occurs near perihelion, is considered to be an extremely strong indication that an object's activity is sublimation- driven, given that such behavior cannot be plausibly explained by any other proposed mechanism (e.g., Je- witt et al. 2015a). As such, the search for and identifica- tion of recurrent activity for MBC candidates has been a key component of the study of active asteroids over the last several years (e.g., Hsieh et al. 2010, 2014, 2016; Hsieh & Sheppard 2015; Agarwal et al. 2016). Iden- tification of activity is often achieved by simple visual detection of an extended coma or tail, or analysis of an object's point-spread function (PSF) compared to those of nearby background stars. If the precise absolute mag- nitude (as well as, ideally, basic lightcurve properties) of an object's inactive nucleus is known, however, much more sensitive searches can be conducted via searches for photometric enhancements indicative of unresolved dust emission (e.g., Tholen et al. 1988; Hartmann et al. 1989; Hsieh & Sheppard 2015). Knowledge of the absolute magnitude of an object's nucleus furthermore facilitates detailed analyses of ac- tivity strength during active periods by enabling the pre- cise determination of the flux contribution from emitted dust via the subtraction of the nucleus contribution from the total flux measured for a MBC while it is active. As such, physical characterization of MBC nuclei, while also being valuable for improving our understanding of the The Reactivation and Nucleus Characterization of 358P/PANSTARRS 3 physical characteristics of the overall MBC population and providing constraints for thermal models, also com- prises an important complement to activity searches and characterization efforts. In this work, we present observations obtained to physically characterize the nucleus of MBC 358P/PAN- STARRS, as well as report on the discovery of its reac- tivation in late 2017. 2. OBSERVATIONS Figure 1. Orbit position plot with the Sun (black dot) at the center, and the orbits of Mercury, Venus, Earth, Mars, 358P, and Jupiter shown as black lines. Perihelion (P) and aphelion (A) are marked with crosses. Green diamonds mark positions of observations when 358P was active in 2012-2013, open circles mark positions of observations when 358P was apparently inactive in 2017, and blue circles mark positions of observations when 358P was active in 2017. Observations of 358P presented here were obtained with the 8.1 m Gemini South (Gemini-S) telescope at Cerro Pachon (Gemini Program IDs GS-2017A-LP-11 and GS-2017B-LP-11). We employed the Gemini Multi- Object Spectrograph (GMOS; Hook et al. 2004; Gimeno et al. 2016) in imaging mode, a Sloan r(cid:48)-band filter, non- sidereal tracking, and 300 s individual exposure times for all observations. All observations were conducted at airmasses of < 1.8, and random dither offsets of up to 10(cid:48)(cid:48) east or west, and north or south were applied to each individual exposure. Standard bias subtraction, flatfield correction, and cosmic ray removal were per- formed for all images using Python 3 code utilizing the ccdproc package in Astropy1 (The Astropy Collabora- tion et al. 2018) and the L.A.Cosmic python module2 (van Dokkum 2001). 1 http://www.astropy.org 2 Written for python by Maltes Tewes (https://obswww.unige. ch/∼tewes/cosmics dot py/) Photometry measurements of the target object and at least five background reference stars were performed using IRAF software (Tody 1986, 1993), with abso- lute photometric calibration performed using field star magnitudes from the PS1 field star catalogs (Schlafly et al. 2012; Tonry et al. 2012; Magnier et al. 2013). Conversion of r(cid:48)-band Gemini and PS1 photometry to R-band was accomplished using transformations derived by (Tonry et al. 2012) and by R. Lupton (http://www.sdss.org/). Target photometry was per- formed using circular apertures with sizes chosen using curve-of-growth analyses of each night of data, where background statistics were measured in nearby but non- adjacent regions of blank sky to avoid potential dust contamination from the object or nearby field stars. To maximize signal-to-noise ratios (S/N), we construct composite images of the object for each night of data by shifting and aligning individual images on the object's photocenter using linear interpolation and then adding them together. Details of our observations of 358P are listed in Ta- ble 1, where we also mark the orbit positions of both the observations reported here and observations previously reported in Hsieh et al. (2013) in Figure 1. Composite images of the object during each night of observations reported in this work are shown in Figure 2. 3. RESULTS AND ANALYSIS 3.1. Phase Function Determination In order to determine the phase function of 358P's nucleus, we first normalize the measured apparent mag- nitudes, m(R, ∆, α) to unit heliocentric and geocentric distances, R and ∆, respectively (i.e., R = ∆ = 1 AU), where α is the solar phase angle, using m(1, 1, α) = m(R, ∆, α) − 5 log(R∆) (1) The resulting reduced magnitude, m(1, 1, α), remains dependent on the solar phase angle via the solar phase function, as well as the rotational phase of the nucleus at the time of the observations in question. At this time, the rotational properties of 358P are unknown, and since all of our observations were short-duration "snap- shot" observations, we are unable to constrain any of these rotational properties from our available data. Our data also all have relatively low S/N, meaning that any brightness variations within observing sequences can- not be reliably used to construct even partial rotational lightcurves. Therefore, for phase function fitting, we treat the mean brightness of the object on each night as a single instantaneous photometric point at an arbi- trary rotational phase. Given enough sparsely sampled 4 Hsieh et al. Figure 2. Composite r(cid:48)-band images of 358P (at the center of each panel) constructed from data listed in Table 1, where no evidence for activity is found for the object at the time of the observations in panels (a) through (f), while we have determined that the object is likely to be active at the time of the observations in panels (g) and (h). All panels are 30(cid:48)(cid:48) × 30(cid:48)(cid:48) in size, with north (N), east (E), the antisolar direction (−(cid:12)), and the negative heliocentric velocity vector (−v), as projected on the sky, marked. Observation dates (in YYYY-MM-DD format) are listed in the upper right corner of each panel. Table 1. 358P Observations νe 284.9 289.3 302.9 303.1 304.1 312.7 318.8 327.0 Rf 2.801 2.756 2.633 2.631 2.623 2.560 2.522 2.479 ∆g 2.005 1.816 1.805 1.811 1.839 2.154 2.404 2.722 αh 15.4 10.0 15.0 15.3 16.4 22.3 23.0 21.1 N b 3 3 2 1 3 1 3 5 tc 900 900 600 300 900 300 900 1500 UT Date Tel.a Airmassd Filter r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) r(cid:48) 1.03 1.04 1.00 1.00 1.10 1.01 1.22 1.67 2017 Jul 01 Gemini-S 2017 Jul 21 Gemini-S 2017 Sep 17 Gemini-S 2017 Sep 18 Gemini-S 2017 Sep 22 Gemini-S 2017 Oct 26 Gemini-S 2017 Nov 18 Gemini-S 2017 Dec 18 Gemini-S a Telescope used. b Number of exposures. c Total integration time, in seconds. d Average airmass of observations. e True anomaly, in degrees. f Heliocentric distance, in AU. g Geocentric distance, in AU. h Solar phase angle (Sun-object-Earth), in degrees. i Equivalent mean apparent R-band nucleus magnitude. j Equivalent mean reduced R-band nucleus magnitude. k Is activity detected? mR,n 24.7±0.1 24.0±0.1 24.3±0.1 24.4±0.2 24.2±0.1 25.0±0.4 24.2±0.1 23.2±0.1 20.8±0.1 20.5±0.1 20.9±0.1 21.0±0.2 20.8±0.1 21.3±0.4 20.3±0.1 19.1±0.1 i mR,n(1, 1, α)j Active?k no no no no no no yes yes photometric points, we assume that rotational bright- ness variations will average to zero, yielding a phase function and absolute magnitude reflecting the object's average brightness over its entire rotational lightcurve. For the purposes of determining best-fit phase function parameters, however, we adopt a reasonable lightcurve amplitude of A = 0.30 mag (corresponding to a peak- to-trough photometric range of ∆mR = 0.60 mag, and an axis ratio for the body as projected on the sky of (a/b)N ∼ 1.7; cf. Equation 7) as the photometric uncer- tainty of each data point due to the object's unknown rotational phase at the time of observation, and add this in quadrature with measured photometric uncertainty to derive the final uncertainty used in performing our fitting analysis. Omitting the two photometric points from 2017 November and 2017 December from our fitting analysis (due to likely activity at those times; cf. Section 3.2), we find best-fit parameters of HR = 19.5 ± 0.2 mag and GR = −0.22 ± 0.13 for 358P's inactive nucleus, using the standard IAU H, G formalism (Bowell et al. 1989). In this formalism, H is the absolute magnitude of an object (i.e., the magnitude at R = ∆ = 1 AU and α = 0◦) and G is sometimes referred to as the slope parameter. A newer three-parameter formalism (the H, G1, G2 system) for phase functions (Muinonen et al. 2010) is increasingly used in the community in addition The Reactivation and Nucleus Characterization of 358P/PANSTARRS 5 For reference, we also plot in Figure 3 the best-fit H, G function, for which we find HR = 20.01 ± 0.06 mag, as- suming the commonly assumed default value of G = 0.15 for small solar system bodies when a measured G value is not otherwise available. We see from the plot that this phase function is still formally consistent with all of the relevant photometric data given the range of pos- sible brightness variations relative to the object's mean brightness due to rotation that we assume in this anal- ysis. For reference, we also compute best-fit values for a linear phase function for our data, finding an absolute magnitude of mR(1, 1, 0) = 19.9 ± 0.2 mag and a phase- darkening coefficient of β = 0.061 ± 0.015 mag deg−1. The effective nucleus radius (in km), rN , of an object with an absolute magnitude of HR is given by N = (2.24 × 1016) × 100.4[m(cid:12)−HR] pRr2 (2) where pR is the object's geometric R-band albedo, and m(cid:12) = −27.07 mag is the absolute R-band magnitude of the Sun (Hardorp 1980; Hartmann et al. 1982, 1990). Assuming a geometric R-band albedo of pR = 0.05, similar to that measured for other MBCs (Hsieh et al. 2009a), we estimate an effective nucleus radius for 358P of rN = 0.32 ± 0.03 km. 3.2. Activity Analysis 3.2.1. Activity Detection and Confirmation During the course of computing the best-fit phase function of 358P using data acquired in 2017 when the object was apparently inactive (Section 3.1), we noted that photometric points from 2017 November 18 and 2017 December 18 appeared to be significantly brighter than expected (by ≥1 mag) relative to the best-fit func- tion found using only the other data points from 2017 July 1 through 2017 October 26. A possible faint, short dust tail extending ∼1 -- 2 arcsec roughly eastward, close to the antisolar direction, is also barely visible in the composite image from December (Figure 2h). Care- ful examination of individual and composite image data did not show any indications of this photometry or ob- served morphology being contaminated by underlying faint field stars or nearby bright field stars. The ob- served photometric deviations are well outside the range of expected rotational variation for the object assuming a physically plausible axis ratio for the nucleus (cf. Fig- ure 3). Measurements of the full width at half-maximum (FWHM) of the point spread functions (PSFs) of the ob- ject and nearby field stars did not give any indications of the presence of resolved coma. To verify the visual detection of activity from 358P on 2017 December 18, we analyze the surface brightness profile of the object on that date. To do so, we create a Figure 3. Plot of best-fit IAU phase functions for 358P with photometric points measured in 2017 overplotted (with plotted uncertainties incorporating both measured photo- metric uncertainties and estimated uncertainties due to the unknown rotational phases at which observations were ob- tained), where the best-fit IAU phase function with HR and GR as free parameters is marked with a solid black line and the best-fit IAU phase function assuming GR = 0.15 is marked with a solid red line. The blue-shaded region in- dicates the 1-sigma range of uncertainty due to phase func- tion parameter uncertainties for the best-fit phase function with GR as a free parameter, while the gray-shaded and light-red-shaded regions (bounded by gray dashed lines and red dashed lines, respectively) indicate the possible photo- metric ranges for the nucleus due to rotational brightness variations, assuming a peak-to-trough photometric range of ∆m = 0.6 mag, for the best-fit phase functions with GR as a free parameter and assuming GR = 0.15, respectively. Open circles and dates mark photometry measured from 2017 July to 2017 October where the comet is presumed to be inac- tive, while blue-filled circles mark photometry obtained in this work where we determine that the comet is likely to be active. to or instead of the H, G system, but in this case, we lack photometric data of sufficient precision over a suf- ficient phase angle range to achieve a meaningful fit to this more complex function. We plot our best-fit H, G phase function and the data used to fit it in Figure 3. As can be seen in the plot, much of our data is clustered in phase angle space (near α ∼ 15◦). This clustering means that the resulting val- ues of the slope parameter and, in turn, the absolute magnitude of the best-fit phase function are strongly de- pendent on individual photometric points at α = 10.0◦ and α = 22.3◦ that were obtained at unknown rotational phases. 6 Hsieh et al. Table 2. Analysis of 2012-2013 and 2017 Photometry b HR,t c d Ad e Md UT Date mR,t νa 7.4 19.6±0.1 15.5±0.3 8.0 19.9±0.1 15.9±0.3 9.1 19.6±0.0 15.7±0.2 9.8 19.5±0.1 15.6±0.2 10.0 19.4±0.1 15.5±0.2 10.0 19.4±0.1 15.6±0.2 10.9 19.0±0.0 15.3±0.2 11.2 19.1±0.0 15.4±0.2 12.0 19.0±0.0 15.4±0.2 12.9 19.1±0.0 15.5±0.2 17.0 19.0±0.0 15.6±0.1 17.2 18.7±0.0 15.3±0.1 18.4 18.6±0.0 15.1±0.1 18.7 18.8±0.0 15.3±0.1 21.0 19.1±0.0 15.4±0.2 21.3 18.9±0.0 15.2±0.2 28.3 19.5±0.0 14.9±0.3 28.6 19.5±0.0 14.9±0.3 28.9 19.8±0.0 15.2±0.3 34.2 20.4±0.0 15.3±0.4 41.4 21.4±0.1 15.9±0.4 284.9 24.7±0.1 19.7±0.3 289.3 24.0±0.1 19.5±0.2 302.9 24.3±0.1 19.7±0.3 303.1 24.4±0.2 19.7±0.3 304.1 24.2±0.1 19.4±0.3 312.7 25.0±0.4 19.6±0.6 318.8 24.2±0.1 18.6±0.4 327.0 23.2±0.1 17.5±0.4 45 ± 10 130 ± 30 32 ± 7 100 ± 20 38 ± 7 120 ± 20 39 ± 7 120 ± 20 43 ± 7 130 ± 20 42 ± 7 130 ± 20 54 ± 8 160 ± 20 49 ± 7 150 ± 20 49 ± 6 150 ± 20 44 ± 5 130 ± 20 40 ± 4 120 ± 10 53 ± 5 160 ± 10 65 ± 6 190 ± 20 55 ± 6 170 ± 20 50 ± 7 150 ± 20 60 ± 8 180 ± 30 80 ± 20 240 ± 60 80 ± 20 240 ± 60 60 ± 20 180 ± 50 60 ± 20 170 ± 50 30 ± 10 100 ± 30 −0.3± 0.9 −0.1± 0.3 0.0± 0.8 0.0± 0.3 −0.3± 0.9 −0.1± 0.3 −0.5± 1.0 −0.2± 0.3 0.4± 1.1 0.1± 0.4 −0.1± 1.7 0.0± 0.6 2 ± 1 5 ± 3 20 ± 7 7 ± 2 2012 Oct 06g 2012 Oct 08 2012 Oct 12 2012 Oct 14 2012 Oct 15 2012 Oct 15 2012 Oct 18 2012 Oct 19 2012 Oct 22 2012 Oct 25 2012 Nov 08 2012 Nov 09 2012 Nov 13 2012 Nov 14 2012 Nov 22 2012 Nov 23 2012 Dec 18 2012 Dec 19 2012 Dec 20 2013 Jan 08 2013 Feb 04 2017 Jul 01h 2017 Jul 21 2017 Sep 17 2017 Sep 18 2017 Sep 22 2017 Oct 26 2017 Nov 18 2017 Dec 18 a True anomaly, in degrees. b Equivalent total apparent R-band magnitude. c Equivalent total absolute R-band magnitude. d Estimated total scattering surface area of visible ejected dust, in 105 m2. e Estimated total dust mass, in 106 kg, assuming ρd = 2500 kg m−3. f Af ρ values, in cm. g All 2012-2013 photometry from Hsieh et al. (2013). h All 2017 photometry from this work. Af ρf 15 ± 3 12 ± 3 13 ± 2 16 ± 3 19 ± 3 16 ± 3 17 ± 2 17 ± 2 15 ± 2 15 ± 2 16 ± 1 15 ± 1 16 ± 2 16 ± 2 18 ± 2 21 ± 3 18 ± 4 18 ± 5 16 ± 4 15 ± 5 11 ± 4 0.0± 0.3 0.0± 0.3 0.0± 0.3 0.0± 0.3 0.1± 0.3 0.0± 0.3 0.9± 0.5 2.2± 0.8 composite image constructed by shifting and aligning in- dividual images on the photocenter of a reference star in each frame, where the reference star is chosen based on its proximity to our target object, the fact that it is rel- atively bright without being saturated, and the absence of other nearby background sources. This star-aligned composite image and the object-aligned composite im- age from the same date (Section 2; Figure 2) are rotated by the same angle such that star trails are horizontal in each image frame. We then construct one-dimensional surface-brightness profiles oriented along an axis (marked by blue and red arrows superimposed on unrotated composite images of the object and reference star in Figures 4a and 4b) per- pendicular to the direction of the non-sidereal motion of the object. We measure profiles along this direction because it allows us to directly compare results from im- ages of the object and reference star in a manner that cannot be done along different axes (e.g., the one aligned with the apparent dust tail) due to the trailing of field stars caused by non-sidereal tracking of the telescope to follow the target during the acquisition of these images. A series of horizontal rectangular apertures are placed along the vertical axes of the rotated images of the ob- ject and reference star, where each aperture is one pixel high and the width of the apertures is chosen such that ∼90% of the flux from the source along the horizontal row passing through each source's photocenter is encom- passed by the aperture along that central row. We mea- sure average fluxes within these rectangular apertures, and subtract sky background sampled from nearby areas of blank sky. The Reactivation and Nucleus Characterization of 358P/PANSTARRS 7 regarded as a lower limit to the true amount of excess flux present in the object's PSF compared to that of an inactive, point-source-like object. The S/N of the object detection in our 2017 Novem- ber 18 data is too low to obtain a useful surface bright- ness profile and perform a similar analysis as described above for our 2017 December 18 data. As such, we can- not independently corroborate our photometric detec- tion of activity on that date described above. Nonethe- less, based on the photometric indication of activity for our 2017 November 18 data and confirmation of activ- ity via surface brightness profile analysis for our 2017 December 18 data, we conclude that 358P has become active again. Subtracting the expected brightness of the nucleus from the actual total measured brightness of the object at the time of observations and averaging over a photometry aperture encompassing the visible flux from the object and its apparent dust tail, we estimate the average surface brightness of the tail and any coma that may be present to be Σd ∼ 26.5 mag arcsec2. 3.2.2. Detailed Activity Characterization Using the phase function derived earlier, we can esti- mate the amounts of excess dust present in observations from 2012 and 2013 reported by Hsieh et al. (2013), and in 2017 reported in this work. Following Hsieh (2014), we estimate the total scattering surface area, Ad, of vis- ible ejected dust using (cid:19) 100.4(HR,t−HR) (cid:18) 1 − 100.4(HR,t−HR) (cid:18) 1 − 100.4(HR,t−HR) 100.4(HR,t−HR) (3) (4) (cid:19) and the corresponding total mass, Md, using Ad = πr2 N Md = 4 3 πr2 N ¯aρd where HR,tot is the equivalent total absolute magnitude of the active nucleus at R = ∆ = 1 AU and α = 0◦ computed using the H, G phase function and the best-fit G parameter determined above (Section 3.1; assuming that the dust exhibits the same phase darkening behav- ior as the nucleus). We assume dust grain densities of ρd = 2500 kg m−3, consistent with CI and CM carbona- ceous chondrites, which are associated with primitive C-type objects like the MBCs (Britt et al. 2002). Moreno et al. (2013) found dust grain radii for 358P's observed dust emission in 2012 ranging from ad,min = 0.5 µm (although they report that this lower limit is not well-constrained) to ad,max = 1−10 cm, assuming a power law size distribution with an index of q = 3.5. Following Jewitt et al. (2014), we can compute a mean effective particle radius (by mass), ¯ad, weighted by the Figure 4. (a) Cropped image of the unrotated (i.e., North up, East left) composite image of 358P (center of the panel) from 2017 December 17 showing the position angle and di- rection of the axis along which the one-dimensional surface brightness profile showing in panel (c) is measured (solid blue arrow). (b) Cropped image of the unrotated composite im- age of a field star (center of the panel) in our 358P data from 2017 December 17, showing the position angle and direction of the axis along which the one-dimensional surface bright- ness profile showing in panel (c) is measured (dashed red arrow). (c) One-dimensional normalized surface brightness profile for a composite image of 358P (solid blue line and open circles) on 2017 December 18 overplotted on the one- dimensional normalized surface brightness profile of a corre- sponding composite image of a reference field star (dashed red line) constructed from the same data set. We normalize the profiles of 358P and the reference star to unity at their peaks and plot them together (Fig- ure 4). Significant excess flux in the direction of the sus- pected faint dust tail can be clearly seen in the wings of the object profile relative to the stellar profile (while we also see, as noted above, that the FWHMs of the object and stellar profiles are nearly identical). As can be seen in Figure 4, the requirement that our one-dimensional surface brightness profiles be measured along the axis perpendicular to the non-sidereal velocity vector of the object means that we do not measure the profile along the axis of the apparent dust tail, along which the maximum deviation from a stellar PSF is ex- pected. As such, the excess flux in the object PSF rel- ative to the stellar PSF shown in Figure 4 should be 8 Hsieh et al. Figure 5. (a) Total absolute R-band magnitude of 358P during its 2012-2013 active period (green diamonds), 2017 inactive period (open circles), and 2017 active period (blue circles) plotted versus true anomaly (ν). The expected magnitude of the inactive nucleus is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (b) Total estimated dust masses measured for 358P during the same periods of observations as in (a) plotted as a function of ν. The excess dust mass expected for the inactive nucleus (i.e., zero) is marked with a horizontal dashed black line, while perihelion is marked with a dotted vertical line. (c) Estimated total dust masses measured for 358P during 2012-2013 and 2017 plotted versus time from perihelion (where negative values denote time before perihelion and positive values denote time after perihelion). A diagonal dashed blue line shows a linear fit to data obtained on 2017 November 18 and 2017 December 18 (−41.2◦ < ν < −33.0◦), reflecting an estimate of the average net dust production rate over this period and allowing us to estimate the onset time of activity, while the shaded blue region shows the range of uncertainty of the linear fit. size distribution, scattering cross-section, and residence time, and assuming ad,max (cid:29) ad,min, using ¯ad ∼ ad,max ln(ad,max/ad,min) (5) Using the particle size distribution determined by Moreno et al. (2013), we obtain ¯ad = 1−10 mm, where for the purposes of our following analysis, we will use ¯ad = 1 mm. For reference, we also compute A(α = 0◦)f ρ values (hereafter, Af ρ; A'Hearn et al. 1995), given by 100.4[m(cid:12)−mR,d(R,∆,0)] (2R∆)2 Af ρ = ρ (6) where R is in AU, ∆ is in cm, ρ is the physical radius in cm of the photometry aperture used to measure the magnitude of the comet at the distance of the comet, and mR,d(R, ∆, 0) is the phase-angle-corrected (to α = 0◦, assuming the same H, G phase function behavior for the dust as for the nucleus) R-band magnitude of the excess dust mass of the comet (i.e., with the flux contribution of the nucleus subtracted from the measured total mag- nitude). We note, however, that this parameter is not always a reliable measurement of the dust contribution to comet photometry in cases of non-spherically sym- metric comae (e.g., Fink & Rubin 2012). The results of all calculations described above are shown in Table 2 for observations of 358P from Hsieh et al. (2013) and this work. We also plot total absolute magnitudes and computed excess dust masses as func- tions of true anomaly, ν, and days relative to perihelion in Figure 5. We fit a linear function to the two data points from 2017 when the object appears to be active, aiming to estimate the average dust production rate of the ob- ject over this period (represented by the slope of this function) and the start time of its activity. For ref- erence, the object's heliocentric distance changes from R = 2.522 AU to R = 2.479 AU between the two dates at which those data points were acquired. This heliocen- tric distance change corresponds to a ∼5-20% increase in the water sublimation rate on the object's surface, de- pending on whether the isothermal or subsolar approx- imation is assumed (following the calculations detailed by Hsieh et al. 2015a). The resulting dust production rate and corresponding activity start date we find are of course subject to numerous sources of uncertainty in- cluding the non-linearity of the actual dust production rate as a function of heliocentric distance, ordinary pho- tometric calibration uncertainties, uncertainties specifi- cally associated with measuring extended objects (e.g., The Reactivation and Nucleus Characterization of 358P/PANSTARRS 9 selection of optimal photometry apertures), and the un- known rotational phases of the object at the times when each photometric point was obtained. Although the rotational phases of the object at the time of our 2017 November and December observations are unknown, we can estimate the uncertainty that the nucleus's rotational lightcurve imparts on our photom- etry given the amount of unresolved coma dust (which will tend to damp the amplitude of observed rotational brightness variations) that we have calculated to be present at those times. The implied minimum axis ra- tio (neglecting possible projection effects), (a/b)N , for an object with a rotational lightcurve with a measured or assumed peak-to-trough photometric range, ∆mR, is given by = 100.4∆mR (7) (cid:16) a (cid:17) b N The underlying axis ratio of the nucleus of an active object including coma dust within a given photometry aperture can be computed using (cid:0)1 − 100.4∆mobs(cid:1)(cid:16) a (cid:17)1/2 (cid:16) a (cid:17) b b N N + Fd FN − 100.4∆mobs = 0 (8) where ∆mobs is the observed peak-to-trough photomet- ric range, and Fd/FN is the computed dust-to-nucleus flux ratio. This equation can be solved for (a/b)1/2 N us- ing standard techniques for solving quadratic polynomi- als (Hsieh et al. 2011a), where Fd/FN = Ad/AN and N = 3 × 105 m2 for 358P. Rearranging Equa- AN = πr2 tion 8, we can solve for the expected observed photomet- ric range for an object with a known or assumed nucleus axis ratio and measured dust-to-nucleus flux ratio using (cid:16) Fd (cid:0) a N +(cid:0) a (cid:1) (cid:1)1/2 (cid:17) (cid:16) Fd (cid:1)1/2 1 +(cid:0) a N FN b b b N FN (cid:17)  (9) ∆mobs = 2.5 log Assuming the same lightcurve amplitude of AN = 0.30 mag (corresponding to ∆mR = 0.60 mag) for 358P's nucleus that we used earlier (Section 3.1) and the estimated total scattering surface areas of dust in Table 2, we then find ∆mobs = 0.24 mag (cor- responding to an expected observed amplitude of Aobs = 0.12 mag), for the nucleus on 2017 Novem- ber 18 and ∆mobs = 0.09 mag (corresponding to Aobs = 0.05 mag) on 2017 December 18. We there- fore find that for a reasonable assumed axis ratio for 358P's nucleus, lightcurve amplitude damping by coma material should reduce potential rotational brightness variations to comparable or negligible levels relative to measured photometric uncertainties and uncertainties on our best-fit values for HR and GR that are used to compute Md via Equation 4. We also conclude that any photometric fluctuations observed for 358P in observa- tions obtained in 2012 and 2013, when inferred excess dust masses were much larger than in 2017, are unlikely to be due to nucleus rotation and more likely to be due to observational effects such as seeing fluctuations causing fluctuations in the amount of dust contained within fixed photometry apertures from one image to the next. Given the slow dust ejection speeds found for most MBCs (e.g., Hsieh et al. 2004, 2009b, 2011a; Moreno et al. 2011, 2016), we do not expect rotational variations in dust production rates to cause significant fluctuations in measured photometry from ground-based data. Using the uncertainties we originally calculated for the dust masses measured for 2017 November 18 and 2017 December 18, we estimate an average dust production M = 2.0 ± rate shortly after 358P becomes active of 0.6 kg s−1 (roughly consistent with the average dust production rate found for the object in 2012 by Moreno et al. 2013) and an estimated start date of 155 ± 4 days prior to perihelion. This start date corresponds to 2017 November 8±4 when the object was at ν = 316.1◦±0.9◦, or equivalently, ν = −43.9◦±0.9◦, and at R = 2.538 AU. For reference, we perform a simple dust modeling anal- ysis to determine whether the start date we estimate from the object's dust mass evolution is consistent with the activity that we visually detect in our 2017 Decem- ber 18 observations. Using an online tool3 developed by J. B. Vincent for plotting syndyne and synchrone grids (Finson & Probstein 1968), we find that ad ∼ 1 mm dust grains ejected with zero initial velocity and evolving under the influence of solar gravity and solar radiation pressure would be expected to travel ∼1 arcsec from the nucleus in 40 days (the length of time elapsed between our estimate for the start date of the activity and the observations in question). Slightly smaller dust grains (ad ∼ 100 µm) would be expected to travel about 7 arc- sec from the nucleus over the same period of time. Given that the dust ejected by 358P likely contains a range of dust particle sizes, and not just the mean particle size of ¯ad = 1 mm that we use above for our dust mass calcula- tions, we find that these dust modeling results are fully consistent with our observations of a ∼1-2 arcsec dust tail on 2017 December 18 and our estimated start date of 2017 November 8 ± 4 for the current active period. More detailed dust modeling is not justified at this time due to the minimal amount of data presently available for 358P's current active period, although it is planned in the upcoming year once more data can be obtained 3 http://www.comet-toolbox.com/FP.html 10 Hsieh et al. during the object's next available observing window (see below). Observations when the object is active in 2012-2013 and 2017 do not overlap in true anomaly, and so it is not possible at this time to compare activity levels from the different active epochs to ascertain how much activ- ity attenuation, if any, has taken place. From Figure 5, it appears possible that if the initial dust production rate estimated for 358P in 2017 is assumed (likely in- correctly) to be maintained at a constant level at later times, a similar amount of total dust (within uncertain- ties) could be ejected by the object during its current ac- tive period as was measured during its 2012-2013 active period. Detailed dust modeling or comparison of ob- servations from both epochs covering overlapping orbit arcs (or ideally both) are needed, however, to robustly compare the strength of the activity observed during each epoch. Both of these tasks will be the focus of fu- ture work, subject to the acquisition of new observations during 358P's next upcoming observing window. 358P is next observable from 2018 August until 2019 May, during which it will cover a true anomaly range of 30◦ (cid:46) ν (cid:46) 100◦, where observability will be best at Northern Hemisphere sites. This observing window should allow for the acquisition of observations that will overlap the latter portion of the orbital arc covered by previously reported 2012-2013 data and that should also help constrain the rate of fading of residual activity from the object as it approaches aphelion. Observations of the object are highly encouraged during this period. 4. DISCUSSION The effective nucleus radius found here for 358P (rN = 0.32 ± 0.03 km) places it among the smallest MBC nu- clei measured to date, along with 238P (rN ≈ 0.4 km; Hsieh et al. 2011b) and 259P (rN = 0.30 ± 0.02 km; MacLennan & Hsieh 2012). For comparison, the largest measured MBC nuclei have effective radii of rN ∼ 2 km (133P and 176P; Hsieh et al. 2009a). Meanwhile, the initial dust production rate estimated here for 358P is generally comparable (within an order of magnitude) to other dust production rates computed for other MBCs (e.g., Hsieh et al. 2009b; Moreno et al. 2011, 2016, 2017; Licandro et al. 2013; Jewitt et al. 2014, 2015b), though we note that these estimated production rates can vary somewhat (within a factor of a few) even for the same active episode for the same object depending on assump- tions made about grain densities and mean grain sizes. We thus find 358P to be similar both in terms of nu- cleus size and dust production rate with at least some previously characterized MBCs. Observations of the reactivation of 358P in 2017 re- ported in this work make the object the seventh active asteroid in the main asteroid belt to be confirmed to ex- hibit recurrent activity (i.e., whose activity is very likely to be sublimation-driven, making the object a MBC), after 133P/Elst-Pizarro, 238P/Read, 259P/Garradd, 288P/(2006) VW139, 313P/Gibbs, and 324P/La Sagra (Hsieh et al. 2004, 2011b, 2015b; Hsieh & Sheppard 2015; Hsieh & Chavez 2017; Agarwal et al. 2016). The identification of these objects as likely ice-bearing bod- ies is significant as it increases the number of objects that can potentially be used to help constrain the dis- tribution of icy material in the inner solar system. In addition to the activity turn-on point determined here for 358P (ν ∼ 315◦; R ∼ 2.55 AU), well-constrained turn-on points (where observations of inactivity are fol- lowed after a relatively short period of time, e.g., a few months or less, by observations of activity) are now available for a handful of MBCs, including 133P (ν ∼ 350◦; R ∼ 2.65 AU; Hsieh et al. 2010) and 238P (ν ∼ 305◦; R ∼ 2.60 AU; Hsieh et al. 2011b), where 324P has also been detected to be active as early as ν ∼ 300◦ (R ∼ 2.80 AU; Hsieh & Sheppard 2015). Compared to these objects, the starting point of 358P's activity in 2017 is unremarkable in terms of ν or R. As more well-constrained turn-on points for MBCs are determined, it will be useful to search for correlations between true anomalies and heliocentric distances of turn-on points with measured dust production rates and other metrics related to activity strength, and to com- bine these analyses with thermal modeling to see if these properties can be used to place constraints on other pa- rameters of interest, such as ice depth and quantity (e.g., Schorghofer 2016; Schorghofer & Hsieh 2018). The observation of recurrent activity also offers the opportunity to monitor the evolution of activity strength for an individual object from orbit to orbit, provided that data covering overlapping orbit arcs from differ- ent active epochs can be acquired, allowing direct com- parison of morphology, total scattering surface area of dust, and other indicators of activity strength, or suffi- cient data during each active epoch can be acquired to allow for average or peak dust production rates to be computed using numerical dust modeling, even if orbit arcs covered by the data in question are not overlapping. Real-world characterization of the attenuation of MBC activity over time would help to evaluate the applicabil- ity of theoretical models of mantle growth and activity attenuation on MBCs (e.g., Kossacki & Szutowicz 2012) and also provide additional constraints on estimated ac- tivation rate calculations based on analyses of detection rates of MBCs found in surveys (e.g., Hsieh 2009). This The Reactivation and Nucleus Characterization of 358P/PANSTARRS 11 type of analysis is beyond the scope of the current work, but does represent an area of growing potential as the amount of data available for characterizing multiple ac- tive epochs for individual MBCs increases. 5. SUMMARY In this work, we present the following key findings: 1. A photometric analysis of observations obtained in 2017 of 358P using Gemini-S while the object appeared inactive yields best-fit IAU phase func- tion parameters of HR = 19.5±0.2 mag and GR = −0.22 ± 0.13 for the inactive nucleus, correspond- ing to an effective radius of rN = 0.32 ± 0.03 km (assuming a R-band albedo of pR = 0.05). 2. In the course of our photometric analysis to deter- mine best-fit phase function parameters for the nu- cleus of 358P, we find that the object appeared sig- nificantly brighter than expected in 2017 Novem- ber and 2017 December given the phase function of the object computed from data obtained prior to that time. A faint dust tail extending ∼1 -- 2 arc- sec roughly eastward, approximately in the antiso- lar direction, is also observed in data obtained on 2017 December 18. The presence of this dust tail is confirmed by surface brightness profile analysis, where we estimate an average surface brightness of the tail and any coma that may be present to be Σd ∼ 26.5 mag arcsec2. We conclude that 358P has become active again for the first time since its previously observed active period in 2012-2013, making it now the seventh MBC candidate con- firmed to exhibit recurrent activity with at least one intervening period of inactivity, a strong indi- cation that its activity is sublimation-driven. 3. Fitting a linear function to the ejected dust masses inferred from the excess fluxes measured for 358P in 2017 when it is apparently active, we find an estimated average net dust production rate of M = 2.0 ± 0.6 kg s−1 and an estimated activity start date of 2017 November 8± 4 when the object was at ν = 316◦ ± 1◦ and at R = 2.54 AU. 4. Insufficient data are available at the present time to ascertain whether the strength of 358P's activ- ity has changed between its 2012-2013 and 2017 active periods. Further observations are highly en- couraged during 358P's upcoming observing win- dow (2018 August through 2019 May), during which the orbit arc covered by the object will par- tially overlap the arc covered by the object during its 2012-2013 apparition, allowing for direct com- parison of the strength of the activity observed during the two active epochs. HHH, MMK, NAM, and SSS acknowledge support from the NASA Solar System Observations program (Grant NNX16AD68G). We are grateful to J. Chavez, J. Fuentes, G. Gimeno, M. Gomez, A. Lopez, L. Magill, S. Margheim, P. Prado, M. Schwamb, A. Shugart, and E. Wenderoth for assistance in obtaining observations, and to an anonymous reviewer who helped to improve this manuscript. This research made use of Astropy, a community-developed core Python package for astron- omy, uncertainties (version 3.0.2), a python package for calculations with uncertainties by Eric O. Lebigot, and scientific software at www.comet-toolbox.com (Vin- cent 2014). This work also benefited from support by the International Space Science Institute in Bern, Switzer- land, through the hosting and provision of financial sup- port for an international team, which was led by C. Snodgrass and included HHH and MMK, to discuss the science of MBCs. This work is based on observations obtained at the Gemini Observatory, which is operated by the Association of Universities for Research in As- tronomy, Inc., under a cooperative agreement with the NSF on behalf of the Gemini partnership: the National Science Foundation (United States), the National Re- search Council (Canada), CONICYT (Chile), Ministe- rio de Ciencia, Tecnolog´ıa e Innovaci´on Productiva (Ar- gentina), and Minist´erio da Ciencia, Tecnologia e In- ova¸cao (Brazil). Facilities: Gemini South (GMOS-S) Software: Astropy (The Astropy Collaboration et al. 2018),IRAF(Tody1986,1993),L.A.Cosmic(vanDokkum 2001, written for python by M. Tewes), uncertainties (E. O. Lebigot), Comet-Toolbox (Vincent 2014) Agarwal, A., Jewitt, D., Weaver, H., Mutchler, M., & A'Hearn, M. F., Millis, R. C., Schleicher, D. O., Osip, D. J., Larson, S. 2016, Central Bureau Electronic Telegrams, 4306 & Birch, P. V. 1995, Icarus, 118, 223 REFERENCES 12 Hsieh et al. Bowell, E., Hapke, B., Domingue, D., et al. 1989, in Jewitt, D., Ishiguro, M., Weaver, H., et al. 2014, AJ, 147, Asteroids II, 524 -- 556 117 Britt, D. T., Yeomans, D., Housen, K., & Consolmagno, G. 2002, Asteroids III (Tucson, University of Arizona Press), 485 -- 500 Jewitt, D., Li, J., Agarwal, J., et al. 2015b, AJ, 150, 76 Kossacki, K. J., & Szutowicz, S. 2012, Icarus, 217, 66 Kres´ak, L. 1979, Asteroids (Tucson, University of Arizona Fink, U., & Rubin, M. 2012, Icarus, 221, 721 Finson, M. J., & Probstein, R. F. 1968, ApJ, 154, 327 Gimeno, G., Roth, K., Chiboucas, K., et al. 2016, in Proc. SPIE, Vol. 9908, Ground-based and Airborne Instrumentation for Astronomy VI, 99082S Hardorp, J. 1980, A&A, 88, 334 Hartmann, W. K., Cruikshank, D. P., & Degewij, J. 1982, Icarus, 52, 377 Hartmann, W. K., Tholen, D. J., Meech, K. J., & Cruikshank, D. P. 1989, Meteoritics, 24, 274 -- . 1990, Icarus, 83, 1 Hook, I. M., Jørgensen, I., Allington-Smith, J. R., et al. 2004, PASP, 116, 425 Hsieh, H. H. 2009, A&A, 505, 1297 -- . 2014, Icarus, 243, 16 Hsieh, H. H., & Chavez, J. 2017, Central Bureau Electronic Press), 289 -- 309 Licandro, J., Moreno, F., de Le´on, J., et al. 2013, A&A, 550, A17 MacLennan, E. M., & Hsieh, H. H. 2012, ApJL, 758, L3 Magnier, E. A., Schlafly, E., Finkbeiner, D., et al. 2013, ApJS, 205, 20 Morbidelli, A., Chambers, J., Lunine, J. I., et al. 2000, Meteoritics and Planetary Science, 35, 1309 Moreno, F., Cabrera-Lavers, A., Vaduvescu, O., Licandro, J., & Pozuelos, F. 2013, ApJL, 770, L30 Moreno, F., Lara, L. M., Licandro, J., et al. 2011, ApJL, 738, L16 Moreno, F., Licandro, J., Cabrera-Lavers, A., & Pozuelos, F. J. 2016, ApJ, 826, 137 Moreno, F., Pozuelos, F. J., Novakovi´c, B., et al. 2017, Telegrams, 4388 ApJL, 837, L3 Hsieh, H. H., Forshay, P., & Schwamb, M. 2016, Central Muinonen, K., Belskaya, I. N., Cellino, A., et al. 2010, Bureau Electronic Telegrams, 4307 Icarus, 209, 542 Hsieh, H. H., & Haghighipour, N. 2016, Icarus, 277, 19 Hsieh, H. H., Ishiguro, M., Lacerda, P., & Jewitt, D. 2011a, O'Brien, D. P., Izidoro, A., Jacobson, S. A., Raymond, S. N., & Rubie, D. C. 2018, SSRv, 214, 47 AJ, 142, 29 Hsieh, H. H., & Jewitt, D. 2006, Science, 312, 561 Hsieh, H. H., Jewitt, D., & Fern´andez, Y. R. 2009a, ApJL, 694, L111 Hsieh, H. H., Jewitt, D., & Ishiguro, M. 2009b, AJ, 137, 157 Hsieh, H. H., Jewitt, D., Lacerda, P., Lowry, S. C., & Snodgrass, C. 2010, MNRAS, 403, 363 Hsieh, H. H., Jewitt, D. C., & Fern´andez, Y. R. 2004, AJ, 127, 2997 Hsieh, H. H., Meech, K. J., & Pittichov´a, J. 2011b, ApJL, 736, L18 Hsieh, H. H., & Sheppard, S. S. 2015, MNRAS, 454, L81 Hsieh, H. H., Yang, B., & Haghighipour, N. 2012, ApJ, 744, 9 Hsieh, H. H., Kaluna, H. M., Novakovi´c, B., et al. 2013, ApJL, 771, L1 Hsieh, H. H., Denneau, L., Fitzsimmons, A., et al. 2014, AJ, 147, 89 O'Brien, D. P., Morbidelli, A., & Levison, H. F. 2006, Icarus, 184, 39 O'Rourke, L., Snodgrass, C., de Val-Borro, M., et al. 2013, ApJL, 774, L13 Raymond, S. N., & Izidoro, A. 2017, Icarus, 297, 134 Raymond, S. N., Quinn, T., & Lunine, J. I. 2004, Icarus, 168, 1 Schlafly, E. F., Finkbeiner, D. P., Juri´c, M., et al. 2012, ApJ, 756, 158 Schorghofer, N. 2016, Icarus, 276, 88 Schorghofer, N., & Hsieh, H. H. 2018, ArXiv e-prints, arXiv:1802.01293 Snodgrass, C., Yang, B., & Fitzsimmons, A. 2017a, A&A, 605, A56 Snodgrass, C., Agarwal, J., Combi, M., et al. 2017b, A&A Rv, 25, 5 Tancredi, G. 2014, Icarus, 234, 66 Hsieh, H. H., Denneau, L., Wainscoat, R. J., et al. 2015a, Icarus, 248, 289 The Astropy Collaboration, Price-Whelan, A. M., Sipocz, B. M., et al. 2018, ArXiv e-prints, arXiv:1801.02634 Hsieh, H. H., Hainaut, O., Novakovi´c, B., et al. 2015b, Tholen, D. J., Hartmann, W. K., Cruikshank, D. P., et al. ApJL, 800, L16 1988, IAUC, 4554, 2 Jewitt, D., Hsieh, H., & Agarwal, J. 2015a, Asteroids IV Tody, D. 1986, in Proc. SPIE, Vol. 627, Instrumentation in (Tucson, University of Arizona Press), 221 -- 241 Astronomy VI, 733 The Reactivation and Nucleus Characterization of 358P/PANSTARRS 13 Tody, D. 1993, in Astronomical Society of the Pacific Vincent, J. 2014, in Asteroids, Comets, Meteors 2014, ed. Conference Series, Vol. 52, Astronomical Data Analysis Software and Systems II, 173 Tonry, J. L., Stubbs, C. W., Lykke, K. R., et al. 2012, ApJ, 750, 99 van Dokkum, P. G. 2001, PASP, 113, 1420 K. Muinonen, A. Penttila, M. Granvik, A. Virkki, G. Fedorets, O. Wilkman, & T. Kohout Wainscoat, R., Hsieh, H., Denneau, L., et al. 2012, Central Bureau Electronic Telegrams, 3252, 1

Subsets and Splits