id
stringlengths 24
24
| idx
int64 0
402
| paragraph
stringlengths 106
17.2k
|
|---|---|---|
662155c691aefa6ce1d61c7b | 14 | When discussing alkene addition reactions, there is a strong focus on the formation of 375 the carbocation to account for the account. In fact, some textbooks only mention the formation of the carbocation as a resource and depict the truncated RCD (ending at the intermediate) to explain the outcome of the chemical reactions. The limited use of resources and limited representation provides an incomplete picture. |
662155c691aefa6ce1d61c7b | 15 | The relationship between stability and transition state was mentioned:P7: I think product C is going to have both the higher transition and … … final product state. However, a detailed representation for the transition state was never provided. Additionally, students never provided a thorough argument for stability (electronics) versus sterics to account for why the 405 activation energy of the major product is lower than the activation energy of the minor product. Similar to the Markovnikov addition, students appeared to apply a heuristic that the major product is both kinetically and thermodynamically favored. |
662155c691aefa6ce1d61c7b | 16 | The third emerging theme is linked to applying the appropriate resources, but they were not applied productively. Participants 2 and 4 defined the reaction as E2 and concerted, but they constructed stepwise RCDs. This observation reiterates the significance between the resources and representational competence frameworks. Constructing RCDs requires 415 activating key resources with productive applications in competently transitioning between different representations. |
662155c691aefa6ce1d61c7b | 17 | The fourth theme was observed with participant 5, who selected an E1 stepwise pathway because of the tertiary alkyl halide. Tertiary alkyl halides do undergo E1 eliminations but in the presence of weak bases. The problem provided used a strong base, which promoted E2 elimination. In learning substitution and elimination reactions, analyzing the alkyl halide structure is generally the first step in the problem-solving strategy; therefore, it is not surprising that this is the first resource that is activated. |
662155c691aefa6ce1d61c7b | 18 | given that the starting material and products are the same. Therefore, the heuristic using the 450 extent of alkyl substitution to gauge stability should apply. However, while some addressed this resource, the predominant resource that gained the most attention was the major and minor product labels, which guided the construction of the RCD. This finding reiterates the importance of metacognition during problem-solving. During the interview, students were not coached to rethink their solutions. |
662155c691aefa6ce1d61c7b | 19 | The heuristic that the major product is kinetically and thermodynamically favored is a theme that appeared during each of the interview prompts, and is one that needs to be carefully addressed. With Zaitsev and Hofmann elimination, instructors and textbooks focus on the structure of the alkene as a guiding point. A key is resource is the Zaitsev product being more substituted alkene and more stable. Therefore, this is a resource that students The second theme is the realization of kinetic control from the steric hindrance of the bulky tert-butoxide base. For example, P9: Okay, so this is highly branched, as far as the 470 base, which means that it will tend more toward disobeying Zaitsev's rule. So, I would say. I think it would have maybe a higher energy. I'm not sure. Okay. So go with like B here and C here, because I think this one might have a higher transition state….B might have a higher |
662155c691aefa6ce1d61c7b | 20 | [transition] energy state. Because the base is highly branched or the Yeah, the base is highly branched. So in order to reach the beta hydrogen, it would be more difficult. The student did 475 not elaborate on electronic factors but did note that the base would lead the products to disobey Zaitsev's rule. Detailed transition states were not constructed. As resourses, transition states were only verbally described and not elaborately represented. The correct RCDs were constructed using steric resources, but no one accounted for the role of electronic resources in determining the overall transition state energies. In fact, while used extensively 480 as a resource to predict thermodynamic stability, no one used the extent of substitution as a resource for gauging transition state stability for either the Zaitsev or Hofmann elimination problems. |
662155c691aefa6ce1d61c7b | 21 | The second research question was designed to explore how resource activation and productive use evolved when students were provided with a set of RCDs. Figure illustrates the trajectory of the participants across the three interviews. Namely, the information summarizes whether students correctly constructed the RCD, changed their answers when 490 provided with the multiple-choice prompts, and whether additional resources were activated. |
662155c691aefa6ce1d61c7b | 22 | Figures summarize the specific resources activated when the multiple-choice items were provided. Three themes provided insight into how scaffolding assignments could be designed in future iterations. productively applied these resources to identify the correct RCD. Likewise, a similar response was observed for P7 (Markovnikov Addition): I just remembered that I think HBr addition to an alkene is not a concerted reaction that has a carbocation intermediate… So because the difference between the minor and major product would be in the cation formation step and the more stable carbocation would be the lower energy. Although most prevalent with the 530 Markovnikov addition, similar observations were noted for the Zaitsev elimination. When provided with the four options, P5 (Zaitsev Elimination): "Okay, um, so I guess it was E2." The RCDs were concerted, which allowed P5 to rule out the E1 as a possible mechanism. |
662155c691aefa6ce1d61c7b | 23 | In comparing the RCDs constructed by P3 in Figures The RCDs constructed by P4 were almost identical; however, different resources, such as stepwise versus concerted mechanisms, were activated in the individual interview prompts. Therefore, while key resources were activated, productive use was hindered by RCin both the Zaitsev and Hofmann, they constructed a stepwise RCD, yet selected a concerted 545 multiple-choice option. They never noted the disconnect in their representations and the multiple-choice options. Finally, students were not asked to identify the axes, but for the xaxis, reaction rate, reaction progress, time, and extent of reaction were observed, consistent with existing findings by Parobek and Popova regarding challenges students have with interpreting the meaning of the x-axis. |
662155c691aefa6ce1d61c7b | 24 | We analyzed how resource activation and application to constructing RCDs varied depending on the reaction context. For the stepwise Markovnikov addition reactions, students activated resources related to the formation of the carbocation, the rate-determining 555 step, and the stepwise nature of the reaction. Fewer resources were activated focusing on the structures and differences in stabilities of the major and minor products. The opposite was observed with the concerted elimination reactions in which resources focusing on the structure of the product stability were activated to a greater extent than resources focusing on transition state energies. The strategies observed stem from the strategies used 560 successfully in lectures. On exams, the major products can be successfully predicted by considering the most substituted carbocation for Markovnikov additions or the impact on alkene substitution as the bulkiness of the base increases in an E2 elimination. |
662155c691aefa6ce1d61c7b | 25 | The Qualtrics survey was assigned as a homework assignment and was not given as a proctored assignment in the classroom. Therefore, it is possible that students reviewed textbooks and other materials to answer the problems. However, the grades were based on 580 participation, and it was emphasized that the responses would only be reviewed at the end of the term for research purposes. A multiple-choice format was used that provided a partial list of options. For the Zaitsev and Hofmann eliminations, the RCD options were all concerted, limiting the total responses landscape by removing E1 as a possible mechanism. |
662155c691aefa6ce1d61c7b | 26 | The interviews were conducted with students at the end of organic chemistry I. The For coding purposes, we only coded explicit statements. Data saturation is not guaranteed in the study, although data saturation of metathemes can be attained by as little as six 590 interviews. Since the sample may not encompass a representative distribution of undergraduate students, the study cannot make generalizable claims about the relative frequencies of observed beliefs. However, the aim of this research does not rely on identifying the prevalence of different concepts in a population but rather on how students work through the 595 relationships between chemical reactions, RCDs, and problem-solving resources. |
662155c691aefa6ce1d61c7b | 27 | RCDs are the foundation for the EPF mechanisms we introduce to students. In introducing reactions, we can use limited resources to account for the outcome of chemical reactions without discussing thermodynamics and kinetics. Carbocation stability accounts 600 for the major and minor products formed in addition reactions. Thermodynamic stability explains why the Zaitsev product is formed with less sterically hindered, strong bases. Using limited resources provides students with a narrow understanding of all energy changes observed in the reaction. As instructors, we aim to provide a more complete picture outlining both kinetic and thermodynamic factors to enhance mechanistic reasoning skills. The 605 interplay of kinetics and thermodynamics appears throughout the curriculum, and by using a complete approach from the beginning, students will be more primed to analyze the kinetics and thermodynamics of more complex reactions such as the Diels-Alder reaction, the Michael addition, and the Wittig reaction-all of which require discussions of kinetic and thermodynamic control to explain why specific products are formed. |
662155c691aefa6ce1d61c7b | 28 | A future strategy will include developing scaffolded problem-solving activities designed to activate critical resources. Future research will focus on developing these scaffolds, using a flow chart like Figure , and implementing these within large-scale assessments via Qualtrics and individually in interviews. Although the interactive interview format provides the opportunity to ask clarification questions for students' reasoning, would enable researchers to pinpoint the specific resources students apply while solving problems. |
673ca21b7be152b1d09f8f66 | 0 | Metal-based core-shell catalysts consist of a thin mono-or multilayer metal film covering a different metal core. Selectively changing the shell thickness, the particle size, or the composition of the core (alloying) or the shell results in changes in the electronic properties of the surface atoms, which in turn has a significant impact on their catalytic activity. Prominent examples are Pt-modified Ru-core nanoparticle catalysts, which were intensively studied over the last decade for the oxygen reduction reaction (ORR), the hydrogen oxidation reaction (HOR) including also the CO tolerance in that reaction, the hydrogen evolution reaction (HER), the formic acid oxidation reaction (FAOR), the ethanol oxidation reaction (EOR), the carbon monoxide oxidation reaction (COOR), or the methanol oxidation reaction (MOR). An important aspect for possible applications of such electrodes is their stability during operation. For Pt-Ru nanoparticles, this includes, in particular, the dissolution of the individual components into the electrolyte, where Ru dissolution is more facile than Pt, but also a possible surface restructuring. For Ru-core Ptshell nanoparticles, it was shown that the Pt shell slows down or even inhibits the dissolution of Ru, and the effect is usually stronger with increasing Pt layer thickness. Nevertheless, Ru dissolution often occurred also for these materials during long-term stability tests, which was attributed to the formation of pinholes in the Pt shell. Fundamental studies aiming at a more detailed understanding of the reactivity, adsorption properties, and stability of Ru-Pt coreshell electrodes often use Pt-modified Ru(0001) electrodes, which are perfect, well-defined model systems and therefore well suited for fundamental studies, including comparison of theory and experiment. Model studies using such electrodes in the ORR, the hydrogen underpotential deposition (HUPD), or the MOR revealed a distinct influence of the Pt layer thickness on Ru(0001). The stability of such model electrodes has been inves-tigated so far only for oxidation reactions on (sub-) monolayer Pt film-covered surfaces, employing ultrahigh vacuum (UHV) scanning tunnelling microscopy (STM) measurements before and after the electrochemical/-catalytic measurements. The results indicated that Pt protects the low-coordinated Ru atoms at step edges against dissolution, shifting the onset potential for dissolution to more positive values. For potentials larger than 1.05 V, Ru dissolution on the Pt-modified Ru(0001) electrodes was mainly observed at the Ru steps (step flow corrosion). Furthermore, Pt was found to restructure, forming Pt clusters on the surface in the regions of the original Pt islands. |
673ca21b7be152b1d09f8f66 | 1 | In this work, we focus on the stability of Pt-modified Ru(0001) electrodes with Pt thicknesses of up to three layers during the electrooxidation of methanol (MeOH) and CO in acid at potentials larger than 1.05 V. This potential had been chosen as upper potential limit in previous studies to prevent possible restructuring of the electrodes during the experiment. Here, we first studied the stability and structural modifications of an electrode with three Pt layers in pure supporting H 2 SO 4 electrolyte and during the MOR by surface X-ray diffraction (SXRD) at the synchrotron. The measurements indicated that these electrodes are surprisingly stable up to 1.4 V, in contrast to the submonolayer Pt modified Ru(0001) electrodes studied previously, or to pure Pt(111) electrodes. Inspired by this finding, we then explored the stability of Pt-modified Ru(0001) electrodes with a Pt thickness of one and three layers for the COOR. For these measurements, we chose the COOR instead of the MOR since |
673ca21b7be152b1d09f8f66 | 2 | In the following, we will first present and discuss our results on the stability of Ru(0001) electrodes modified with one and three Pt layers during the COOR, including STM images recorded before and after the COOR. These results are then used to substantiate the findings in operando SXRD-MOR measurements on similar electrodes. We will demonstrate that the electro-oxidation of organic species can indeed improve the stability of the electrodes. |
673ca21b7be152b1d09f8f66 | 3 | Ru(0001) electrodes modified by a single (Pt-1) or 3.5 monolayers (ML) of Pt (Pt-3) were prepared and characterized by STM imaging under UHV conditions, as described in the Experimental Section. Schematic illustrations of the surface cross-section and representative STM images of the as-prepared Pt-1 and Pt-3 electrodes are shown in Figs. and, respectively. As illustrated by these images, the Pt film does not grow in a perfect layer-by-layer fashion due to kinetic limitations. On the Pt-1 electrode, the film consists of monoand bilayer islands, exposing also the Ru(0001) substrate. The Pt-3 electrode consists of regions with two to four Pt layers and a few small five-layer islands. Figs. and show STM images of the electrode surfaces recorded under UHV conditions after exposing the electrodes to potential cycles up to 1.35 V during the COOR (see Fig. ). |
673ca21b7be152b1d09f8f66 | 4 | After the potential cycling up to 1.35 V in CO saturated 0.5 M H 2 SO 4 (see electrocatalytic measurements below), the Pt-1 surface shows a distinct restructuring (Fig. ), in agreement with results in previous studies. This includes holes in the monolayer film and the Ru(0001) substrate at the former Ru step (marked by an arrow). Interestingly, the restructuring of the bilayer Pt film areas (marked by dashed circles) seems much more pronounced than that of the monolayer areas, which will be discussed below. One possibility for hole formation is the dissolution of material in the electrolyte as ions or oxidized species, which is known to occur for both pure Pt and Ru during the positive-and/or negativegoing scan in a complete potential cycle, in particular in the potential region for surface oxide formation and reduction. This is reflected in Pourbaix diagrams, and dissolution of Pt and Ru was directly demonstrated also by inductively-coupled plasma mass spectrometry (ICP-MS) measurements. For Ru, we suggest that dissolution is the main process behind the observed structural changes. For Pt, it is not as clear, and we will show later that under the present conditions, dissolution is probably unlikely. |
673ca21b7be152b1d09f8f66 | 5 | The Pt-3 electrode (Fig. ), in turn, remains almost unchanged during / after the COOR, i.e., it is highly stable under reaction conditions. This result is somewhat unexpected, considering that exposure of bare Pt(111) electrodes to potential cycles in CO-free electrolytes with a potential limit larger than 1.17 V leads to significant irreversible restructuring of the electrode after a few potential cycles. Additional STM measurements for Pt-3 electrodes restructured in a controlled way by consecutive potential cycles in pure 0.5 M H 2 SO 4 are shown in Fig. in the supporting information (SI). Under these conditions, the electrochemical data suggested that the surface restructured at around 1.3 V. STM images revealed that thicker Pt layers (three atomic layers or more) remained unchanged, while the remaining regions with mono-or bilayer Pt films were entirely restructured. Reasons for the layerdependent stability (first to third layer) are discussed below. |
673ca21b7be152b1d09f8f66 | 6 | The COOR on the Pt-modified Ru(0001) electrodes was characterized by cyclic voltammetry at a scan rate of 50 mV s -1 in CO-saturated 0.5 M H 2 SO 4 . The measurements were performed in a dual thin-layer electrochemical flow cell coupled with a DEMS set-up, allowing for operando CO 2 (m/z=44) product detection. 65 During the COOR measurements, the upper potential limit (UPL) was gradually increased in steps of 0.05 V, recording two potential cycles for each UPL. For better identification of differences, we only show a single potential cycle for UPLs of 1.05 V and 1.35 V in Fig. , respectively, for the Pt-1 electrode (left column) and the Pt-3 electrode (right column). Details of the experimental procedure and the respective parameters can be found in the Experimental section and the SI in Section S2, which also includes data for additional UPLs not shown here. The COOR-CVs are shown in black in Figs. and. The mass spectrometric ion current of the CO 2 product formation is converted into a Faraday current and normalized to the Faraday current recorded in the COOR-CV. A detailed description of the conversion and normalization procedure and the data evaluation is provided in the SI in Section S3. The resulting curves are denoted as mass spectrometric cyclic voltammograms (MS-CV), and are depicted in red in Figs. and. For direct comparison, the COOR-CVs and MS-CVs are plotted on top of each other in Figs. and. Since only CO 2 is formed during the COOR, subtracting the MS-CV from the COOR-CV (black minus red curve in Figs. and), yields a CV that represents other surface redox processes, as they are usually found in CVs recorded in CO-free elec-trolytes, plus contributions from CO 2 formation by reaction with preadsorbed OH/O species, which release less than the two electrons per CO 2 assumed in the calibration of the K-factor (see Section S3). We denote this curve as redox-CV, shown in green in Figs. and (plotted along with the COOR-CV for comparison). |
673ca21b7be152b1d09f8f66 | 7 | Overall, the shape of the COOR-CVs is rather similar to those reported for Pt electrodes. Based on these studies and previous ones on Pt-modified Ru(0001) electrodes, the processes taking place in different potential regions can be qualitatively described as follows. Starting with the positive-going scan from the low-potential limit up to the onset of the COOR (0.5 V for Pt-1 and 0.6 V for Pt-3), the surface is poisoned by CO, inhibiting any surface redox processes, as has been demonstrated for pure Pt 62,66-71 and Ptmodified Ru(0001) electrodes. With the onset of OH/O formation on the surface (from water splitting) in the CO adlayer, the COOR starts, which had been described in detail for the COOR in CO-saturated electrolytes on pure Pt electrodes. Note that the increase of the current density around the onset potential changes in the first potential cycles, which is illustrated in Fig. . A similar phenomenon has been reported for Pt(111) during the COOR in CO saturated 0.1 M HClO 4 . In this potential region, up to the onset of the reaction, the MS-CV closely follows the COOR-CV. |
673ca21b7be152b1d09f8f66 | 8 | COOR with an UPL up to 1.05 V From the COOR onset potential, the current density in the COOR-CV rises sharply, both for the Pt-1 and Pt-3 electrodes, resulting in a distinct peak A1. A similar behaviour is commonly observed for bulk CO electro-oxidation on Pt electrodes. Note that the MS-CV does not show a peak in the peak A1 region, in contrast to earlier observations for the COOR on supported Pt/C catalyst electrodes and rough Pt ATR films. The peak is also absent in further measurements on various Pt electrodes presented in Section S4. The additional peak A1 in the CV must result from the rapid formation / uptake of OH/O species on the surface, which does not show up in the MS-CV signal. This OH/O formation is apparent in a peak A1 in the positive going scan of the redox-CV in Figs. and. |
673ca21b7be152b1d09f8f66 | 9 | In the negative-going scan, the COOR shows a higher activity in the potential region around the onset potential for COOR in the positivegoing scan. Such a hysteresis is commonly observed for the COOR on Pt-modified Ru and also on pure Pt electrodes. More important, the current density in the MS-CV seems to be slightly larger than that in the COOR-CV, which must be due to surface reduction processes taking place simultaneously in this potential region. This is discussed further in the next section. Once the current density decayed to zero in the COOR-CV, a reduction peak / region C1 appears at more negative potentials, which is rather broad for Pt-1 and more peak-like for Pt-3 (see insets in Figs. and and Fig. ). The appearance of this peak C1 indicates that under present conditions, part of the surface OH/O adspecies are reduced to water instead of reacting with adsorbed CO. Nevertheless, the removal via COOR is also possible, as indicated by a small current in the MS-CV in the potential region of peak C1. This also means that the observed current peak C1 is a sum of currents related to the direct removal of OH/O from Pt to form water (negative current) and OH removed by the COOR (positive current). The amount of removed, adsorbed OH/O, either by reduction to water or by reaction with CO, is presented in the redox-CVs in Figs. and. These curves indicate the reductive or reactive OH/O removal in a peak / region for potentials below about 1.0 V for both electrodes. The oscillatory behaviour of the reduction peak in the redox-CV is possibly an artefact caused by problems in the temporal correlation of the MS-CV and the COOR-CV. |
673ca21b7be152b1d09f8f66 | 10 | For the Pt-1 electrode, recording CVs with an UPL higher than 1.05 V leads to a shift of the onset potential for the COOR to lower potentials, apparent in a pre-peak or shoulder A2 in the CV (Fig. and Fig. ). It appears for UPLs larger than 1.1 V. The assignment of this peak / shoulder A2 to an ear-lier onset of the COOR is confirmed by an increase in the CO 2 current in the MS-CV in (Fig. ). The extent of the current increase depends on the number of potential cy-cles at a given UPL and the applied UPL (see Figs. and). A similar phenomenon has been reported previously and has been attributed to a potential induced restructuring of the electrode surface, which is also supported by the STM image in Fig. , recorded after the electrochemical measurements. It was suggested that newly formed low-coordinated sites on the restructured electrode enhance the COOR. Apparently, continuing surface restructuring of the Pt-1 electrode reduces the formation of OH/O species on the Pt-1 surface in the peak A1, leading to a decrease in size of that peak. On the other hand, this restructuring also leads to an enhanced COOR activity in the region of peak A2. During restructuring, the newly formed sites with lower coordination will bind OH/O stronger than sites on the flat Pt layers, thus allowing OH/O formation on the surface already at lower potentials. Hence, this also leads to the peak A2 in the redox-CV in Fig. (see inset), indicating OH/O adsorption in this potential region. Consequently, the restructuring reduces the number of sites on the flat Pt areas, leading to a decrease of Peak A1 in the redox-CV. This is also evident for the Pt-2 electrode described in Section S5, where with an increasing number of potential cycles (with increasing UPL -Fig. or fixed UPL -Fig. ), the COOR activity in the peak A1 decreases even further. Finally, the current density decreases in the entire potential region where the Pt-2 electrode is active for the COOR. STM images recorded after the COOR for the fully restructured electrode indicate that the Pt film formed large agglomerates on the surface, exposing the severely restricted Ru(0001) surface. |
673ca21b7be152b1d09f8f66 | 11 | For the Pt-3 electrode, such kind of restructuring induced increase of the COOR at around the onset potential is hardly visible in the COOR-CV upon extending the UPL up to 1.35 V (see inset of Fig. ). The remaining subtle changes in the current density (one order of magnitude lower than for Pt-1) are tentatively associated with a slight restructuring of the electrode, possibly in the small fraction of mono-and bilayer Pt areas (see Fig. ) present on the Pt-3 electrode, although this could not be resolved by STM imaging. The magnitude of peak A1 remains almost unchanged during potential cycling for the Pt-3 electrode, different from the decrease for the Pt-1 electrode. |
673ca21b7be152b1d09f8f66 | 12 | For potentials more positive than peak A1, the COOR current density declines with increasing potential, which is more pronounced for the Pt-1 than for the Pt-3 electrode. This potential region is often denoted as the mass transport-limited region. In the present case, however, under enforced mass transport, kinetic limitations seem to be active as well, leading to the observed decay in COOR rate with increasing potential. The presence of kinetic limitations is also consistent with the lower COOR current on the Pt-3 electrode in this potential region as compared to the Pt-1 electrode (for similar flow rates). For each of the two electrodes, the current densities exhibit the same potential dependence for different UPLs. This is also true for the Pt-1 electrode, which tends to restructure under these conditions (see Fig. ). Apparently, the processes responsible for the decay in COOR rate with increasing potential are fully reversible between subsequent cycles. |
673ca21b7be152b1d09f8f66 | 13 | In the negative-going scan, the COOR current increases again in this potential region for the Pt-1 electrode, while for the Pt-3 electrode, it remains about constant. The constant COOR current on the Pt-3 electrode indicates that the slight decline of the COOR rate due to the further accumulation of OH/O species on the surface in the positive-going scan, in addition to the amount formed in the peak A1, is not reversed, but also not continued. Thus, during the reaction, the coverage of these species remains constant in this potential range. Their amount can only be reduced at lower potential. In this case, the kinetic limitations remain present in the negative-going scan in this potential range. This also means that the difference in COOR rates between the positive-going and the negative-going scan increases with increasing UPL. For the Pt-1 electrode, the reasons for the pronounced decay in COOR rate at potentials positive of peak A1 are largely, though not completely, removed in this potential range, leading to only small differences in the COOR current between the positive-going and negative-going scans at the potential directly above the peak A1. Furthermore, different from the Pt-3 electrode, this removal of the kinetic limitations always leads to the same current at about 0.95 V, independent of the UPL. Therefore, for this electrode, most of the more pronounced COOR current decay in the positive-going scan in this potential range must be due to reversible processes such as reversible accumulation and replenishment of additional adsorbed OH/O species or the reversible buckling of Pt surface atoms due to insertion of oxygen atoms underneath, which was described as reversible extraction during the oxidation of Pt(111). Only a small part of this decay in activity is caused by processes that cannot be reversed in this potential range. |
673ca21b7be152b1d09f8f66 | 14 | In the potential range of the peak A1 and below in the negative-going scan, increasing the UPL results in a shift of the steep decline in COOR current to lower potentials, both for the Pt-1 and the Pt-3 electrode. Since the steep increase of the COOR current in the positivegoing scan does not shift significantly, the hysteresis becomes broader with increasing UPLs. Starting with the Pt-3 electrode, the shift of the steeply declining COOR current to lower potentials with increasing UPL must be due to an increasing amount of OH/O species on the surface, which lowers the COOR rate at potentials positive of 1.0 V. These species can support the COOR for a longer time and thus to lower potential in the scan. Once most of these species are consumed reactively, rapid blocking of the surface by adsorbed CO sets in. This explanation closely agrees with previous modelling results for CO bulk oxidation on Pt, where such shifts had been related to the larger degree of OH/O formation on the sample at high potentials. For the Pt-1 electrode, where the COOR current in the potential range of the peak A1 does not seem to vary with the UPL, the explanation must be more complex. Clearly, one would expect a similar physical reason for the observed down-shift in the COOR current decline for potentials more negative than peak A1 as for the Pt-3 electrode. But in the case of the Pt-1 elec-trode, restructuring effects must also be considered, which may further affect the COOR rate at potentials lower than peak A1. |
673ca21b7be152b1d09f8f66 | 15 | Once the COOR rate has decreased to zero, we now observe a distinct reduction peak C1 in the COOR-CVs (instead of a reduction region observed for lower UPLs, see above), which must be related to the electrochemical removal of OH/O species by reduction to water. This peak increases with increasing UPL for the Pt-1 electrode, while for the Pt-3 electrode, it seems to decrease (see insets and Fig. ). For the Pt-3 electrode, we suggest that this apparent decrease of the peak C1 (and shift to lower potential) is caused by an increasing overlap between the steeply declining COOR current and the peak C1 reduction peak. For the Pt-1 electrode, the reduction peak is located at more negative potentials than the declining COOR curves. Here, the increase in peak C1 can be rationalized by increasing amounts of OH/O species formed on the exposed Ru(0001) substrate. In the redox-CV in Figs. and, the reduction peak for OH/O removal (see inset) has a similar size for both electrodes and seems to be mostly independent from the applied UPL. The shape varies slightly, and the peak maximum is at slightly different potentials. In principle, if our suggestion from above is correct, that the amount of OH increases at higher UPLs, then the charge within that peak should increase. Such information can, however, not be confirmed from the present data. |
673ca21b7be152b1d09f8f66 | 16 | Next, we present the results of combined CV and SXRD measurements performed on a Pt-3 electrode in pure and 0.1 M MeOH containing 0.5 M H 2 SO 4 . In this case, the potential was cycled between 0.05 V and 1.4 V, while the surface sensitive (1 0 2.35) reflection was monitored (Fig. ). The Experimental Section and Section S6 contain further information about the synchrotron measurements. For the sake of completeness, additional CVs and SXRD curves illustrating the effect of changing the UPLs from 1.05 to 1.40 V, are shown in Section S6, including also information on the noise in the CV data. CVs recorded in 0.5 M H 2 SO 4 on similar electrodes in the DEMS set-up are shown in (Figs. and). Despite the considerable noise, the CV (blue curve in Fig. ) resolves a slight increase in the current density in the positive-going scan above 0.8 V, which coincides with the onset of a steep decrease in the X-ray intensity. Previously, we had attributed a current increase (or peak) in this potential region to the desorption of bisulfate and the simultaneous adsorption of OH. Note that a possible exchange of bisulfate with OH in a 1:1 ratio would not show up in the CV since the net current density will be zero. Therefore, the actual onset for surface OH formation cannot be deduced directly from the CV and might even occur at lower potentials than the onset of the surface OH/O formation current. In general, the adsorption of OH/O will lead to a change in the interlayer spacing of the topmost Pt atoms, which is re-flected in a change in the X-ray intensity. In recent SXRD studies on the electro-oxidation of single-crystal Pt surfaces, a pronounced reversible decrease (positive-going scan) and increase (negative-going scan) of the SXRD intensity in the (0 1 1.5) and (1 1 1.5) 82 refelections was attributed to a reversible buckling of the Pt surface, involving a partial extraction of Pt surface atoms by reversible incorporation and removal of subsurface oxygen. On Pt(111), the onset for this process is at about 1.1 V. Earlier onsets for this process were reported for Pt(100) surfaces or carbon-supported Pt particles. Considering also the significant change in XRD intensity, we assume that the decrease observed here is primarily related to the reversible extraction of Pt atoms. From the present data we cannot distinguish whether this extraction occurs on the terraces or predominantly on defect sites (step edges and regions on the electrode that might not be as perfectly structured as those presented in the STM images in Fig. ). Changes in X-ray intensity due to desorption of bisulfate and adsorption of OH/O on the terrace sites are expected to be much smaller (see below). In the reverse, negative-going scan, the X-ray intensity starts to increase steeply at around 0.8 V, coinciding with the onset of a small reduction peak C2 in the CV. We associate this increase with the reversible removal of O atoms from subsurface sites and the de-buckling of the related Pt surface atoms, accompanied by reductive OH/O desorption and possibly bisulfate adsorption, which, however, does not fully compensate for the OH/O removal charge. After a complete potential cycle, the X-ray intensity reaches its original value. Further CVs recorded with increasing UPLs do not change significantly, suggesting that the surface did not irreversibly restructure substantially. Subtle changes in the CVs recorded for different UPLs, both in the combined SXRD-CV measurements as well as in CVs recorded in a dedicated electrochemical setup, are described and discussed in the SI. MOR measurements performed on these electrodes after the cyclic voltammetry measurements in the supporting electrolyte (red curves in Fig. ) resemble those recorded in the MOR on Pt electrodes. Adsorbed CO blocks the surface at low potentials, resulting in a negligible reaction current. The current starts to increase in the positive-going scan at around 0.6 V. Interestingly, while a detailed picture of the product composition is still missing, the formation of CO 2 close to the onset potential of the MOR is negligible. Hence, in this potential range, formaldehyde or formate / formic acid are the most likely products formed on Pt-3. After passing through a maximum at about 0.8 V, the current density decreases again and reaches negligible values at around 1.1 V. Formerly, it was suggested that this current density decrease is associated with OH/O adlayer formation on the surface, with adsorbed OH/O blocking the surface for the MOR. Also, in the negative-going scan, the electrode becomes only active for the MOR once the OH/O adlayer starts to be reductively removed at around 0.8 V (see below). |
673ca21b7be152b1d09f8f66 | 17 | The evolution of the X-ray intensity at the (1 0 2.35) position during the MOR (red curve) is very similar to that obtained in the pure supporting electrolyte, except that the onset of the sharp decrease in the positive-going scan shifts by 300 mV to more positive potentials, from 0.8 V to 1.1 V. As mentioned above, the reversible insertion of adsorbed O atoms together with the partial extraction of the associated Pt surface atoms was suggested to be the dominant process in this potential region in the pure supporting electrolyte, which accounts for this change. Still, there is a slow decay in X-ray intensity in the potential range between the MOR peak maximum at about 0.8 V and the end of this peak at about 1.1 V, which is possibly related to the adsorption of OH/O on the surface, inhibiting the MOR. The up-shift to 1.1 V of the steep decrease of the X-ray intensity indicates that the associated process is mostly inhibited in this potential range during the MOR, as compared to measurements in the pure supporting electrolyte on Pt-3. Most likely, this up-shift is caused by a reaction of OH/O (possibly at defect sites as described above) with methanol or with reaction intermediates formed during the MOR, which lowers the steady-state OH/O coverage at a given potential and thus shifts the coverage-potential curve to higher potential, in agreement with the up-shift of the onset of the steep decline of the X-ray intensity observed experimentally. Surface-enhanced infrared reflective adsorption spectroscopy (SEIRAS) measurements on pure polycrystalline Pt electrodes indicated that the adsorbed CO blocking the surface before the MOR onset is entirely removed at the MOR peak maximum. Hence, also adsorbed CO could block the OH/O adsorption and consequently also O insertion / Pt extraction in the potential range up to the X-ray intensity decrease. Furthermore, with the onset of the MOR, adsorbed formate (HCOO ad ) was found to form on the surface, with the SEIRAS intensity following the MOR current density in both scan directions, which can contribute to the slow decay in X-ray intensity in this potential region. |
673ca21b7be152b1d09f8f66 | 18 | Finally, based on the reversibility of the Xray intensity, we can rule out that the decline in X-ray intensity at higher potentials is related to any irreversible surface restructuring. This is also supported by additional STM measurements performed after the operando SXRD measurements under UHV conditions, which indicated that the Pt-3 electrodes are indeed very stable under present MOR reaction conditions (see Section S6). |
673ca21b7be152b1d09f8f66 | 19 | The results presented above for the Pt-1 and Pt-3 electrodes, as well as those for Pt-2 (Section S5), have shown that in the COOR during potential cycles up to 1.35 V, the monolayer Pt film is more stable against irreversible restructuring than bilayer islands on the monolayer film and that three to four-layer films remain almost unchanged. Bilayer Pt films also seem less stable in CO-free electrolytes, as shown in Fig. . For the Pt-3 electrode, a high stability was also shown for the MOR up to 1.4 V. In the following, we will briefly discuss possible reasons for these differences in the layer-dependent stability and draw comparisons with Pt(111). |
673ca21b7be152b1d09f8f66 | 20 | Traditionally, surface restructuring of Pt(111) electrodes during oxidation and reduction cycles in pure acidic electrolytes (with an UPL larger than 1.2 V) was rationalized by an irreversible place exchange process between adsorbed atomic oxygen and surface Pt atoms. More recently, this picture was supplemented by a reversible O insertion / partial Pt extraction process, which occurs before irreversible Pt and O exchange. In a simple picture, one would expect that both reversible and irreversible restructuring of the Pt surface layer(s) set in once a critical coverage of adsorbed OH/O species is reached. For Pt layers on Ru(0001), it is known that the binding energy of OH/O is lowest on the monolayer and increases with increasing Pt layer thickness, until reaching the binding energy of Pt(111) for thick layers. In that case, the onset of restructuring should shift to lower potentials with increasing Pt layer thickness, and the critical potential for the onset of a possible irreversible place exchange process would then always be more positive than that for Pt(111), i.e., 1.2 V. Hence, the Pt films on Ru(0001) should, in principle, be intrinsically more stable than Pt(111) with respect to reversible Pt extraction and irreversible place exchange, with monolayer Pt films presenting the most stable structure. Since the Pt-3 electrodes are stable under present reaction conditions, the potential for irreversible place exchange processes must be beyond the explored potential limit of 1.35 V for the COOR or 1.4 V for the MOR. Furthermore, considering that both irreversible place exchange and dissolution lead to a restructuring of the surface, this also means that the Pt dissolution potential should shifts to more positive potentials. Nevertheless, once the critical potential for the formation of a dense O adlayer is reached, we expect that the Pt films on Ru(0001) restructure similarly to Pt(111) electrodes. Indeed, we found similar structures as for restructured Pt(111) on Pt-3 electrodes, induced by potential cycles to high potentials in pure supporting electrolyte shown (in Section S1). |
673ca21b7be152b1d09f8f66 | 21 | In contrast to the above argumentation, however, we observed that very thin Pt films (incomplete Pt monolayer or bilayer films) are less stable than Pt(111). We suggest that the enhanced restructuring tendency of the (incomplete) monolayer film is induced by the strong interaction of OH/O species with the Ru(0001) substrate, which must be considered. Apparently, the gain in energy provided by the strong interaction of OH/O species with Ru(0001) is sufficient to displace Pt from monolayer sites into next layer sites, providing space for an additional uptake of OH/O species on the Ru(0001) substrate. Computational works showed that Pt atoms in the second layer bind less strongly on the first Pt layer than Pt atoms directly attached to the Ru(0001) substrate. However, this loss in energy is apparently overcompensated by the energy gain of the OH/O adspecies, if they are strongly adsorbed on Ru(0001) rather than weakly adsorbed on a Pt monolayer film. The energy gain of the entire system upon OH/O-induced displacement of Pt atoms into the second or higher layers provides the energetic driving force for such kind of restructuring. For the Pt-2 electrode, the COOR data indicate that these electrodes seem relatively stable when the layer is closed (up to 1.3 V, see Section S5 in the SI), which does not seem to be the case in the CO-free supporting electrolyte (Section S4 in the SI). Based on the explanations above, we would expect that Pt bilayers are less stable than the monolayer but also more stable than the third layer. The physical origin of this discrepancy is so far open. |
673ca21b7be152b1d09f8f66 | 22 | During the MOR, we observed a shift of the decrease in X-ray intensity to more positive potentials compared to the measurements in the methanol-free electrolyte. This effect was rationalized by a potential dependent lowering of the OH/O adsorbate coverage and thus by a similar up-shift of the O insertion and Pt extraction in the presence of methanol or reaction intermediates formed during the MOR. We suggest that a similar shift of the decrease in X-ray intensity to more positive potentials, as observed in the SXRD measurements in the MOR, will occur also during the COOR. In that case, the effect might be even more pronounced since CO oxidation also occurs up to the OER. CO can readily adsorb and react on the electrode surface, even by chemical reaction with Pt surface oxides. In contrast, active interme-diates can only be formed in the MOR by splitting MeOH, which is increasingly hindered at high potentials. Therefore, we assume that CO in the CO-saturated solution can efficiently remove OH/O species from the surface even at a potential where the MOR is inhibited and thus shift the potential for partial Pt extraction and reversible surface restructuring to even more positive potentials. Accordingly, also the potential for irreversible restructuring may be up-shifted. |
673ca21b7be152b1d09f8f66 | 23 | Finally, we address the relevance of our results for realistic Ru-core Pt-shell particles used as highly efficient catalysts in various reactions, as presented in the introduction. The main requirement for these catalysts is that Pt always covers the Ru substrate to prevent Ru dissolution or Pt restructuring induced by the strong interaction of OH/O species with Ru. For closed Pt layers and reactions such as the ORR, HER and HOR, a possible restructuring by an irreversible place exchange process is presumably negligible, since these reactions normally proceed at potentials well below the onset potential of restructuring, which for these systems is higher than that for pure Pt. In the presence of oxidizable species in the feed, one could expect an even higher stability since their reaction with OH/O species would decrease the steadystate coverage of these species at a given potential and thus shift the onset potential for surface restructuring to higher potentials. The detailed understanding of the layer-dependent stability of such systems is, however, still incomplete. Earlier studies indicated that Ru-core Pt-shell systems with two layers of Pt are rather stable, which is not the case for well-defined Pt layers on Ru(0001). These discrepancies are most likely due to the different structural properties of the core-shell particles as compared to the present model systems. |
673ca21b7be152b1d09f8f66 | 24 | Using Ru(0001) electrodes covered by a monolayer and three layers of Pt, we demonstrate by DEMS measurements during the COOR and by operando SXRD measurements during the MOR, as well as by STM imaging before and after the electrocatalytic investigation, that submonolayer Pt films start to restructure for potentials more positive than 1.1 V (as shown previously), in contrast to films with three layers (and more), which are stable up to about 1.4 V during the COOR and MOR. |
673ca21b7be152b1d09f8f66 | 25 | • The higher stability of three-layer Pt films towards restructuring during the MOR and COOR in acid electrolyte, also in comparison to bare Pt(111) electrodes, suggests that the potential for reversible surface restructuring shifts to more positive potentials. This shift is attributed to a weaker bonding of OH/O species on the three-layer Pt film than for the adsorption on Pt(111). |
673ca21b7be152b1d09f8f66 | 26 | • From the SXRD measurements, we conclude that, in general, the electrooxidation of organic molecules reactively reduces the potential dependent steadystate coverage of surface OH/O species, which needs to reach a critical value before it can induce structural changes. This increases the stability range of the respective Pt electrode, as compared to the situation in pure supporting electrolyte. |
673ca21b7be152b1d09f8f66 | 27 | The Pt-modified Ru(0001) single-crystal electrodes used in this work were prepared and characterized by STM under UHV conditions at the Institute of Surface Chemistry and Catalysis at Ulm University (Ulm, Germany). The electrochemical DEMS characterization was performed in Ulm and the combined electrochemical and SXRD measurements at the ID03 beamline of the European Synchrotron Radiation Facility (ESRF, Grenoble, France). Where necessary, we indicate facility-specific experimental details. |
673ca21b7be152b1d09f8f66 | 28 | For the measurements performed in Ulm, we used H 2 SO 4 Merck Suprapure 98% and methanol Merck EMSURE ® . The electrolytes were purged with N 2 (Westfalen 6.0). For the measurements performed at the ESRF, the solutions were prepared from H 2 SO 4 93-98%, VWR ARISTAR ® ULTRA and methanol (Fluka Analytical, LC-MS Ultra Chromasolv ® ). The electrolytes were purged with Ar (Air Liquide, BIP grade). The Ru(0001) single crystal electrodes were purchased from MaTecK GmbH (purity 99.99%, orientation accuracy <0.1°). The glassware was cleaned before each experiment by storage in highly concentrated KOH (Ulm) or Piranha etch (ESRF) overnight. Subsequently, it was thoroughly boiled and rinsed with hot MilliQ water. |
673ca21b7be152b1d09f8f66 | 29 | Sample preparation: The Ru(0001) electrodes were prepared under UHV conditions in an experimental set-up described elsewhere. Clean and atomically flat surfaces were obtained by several cycles of Ar ion sputtering (p Ar = 3 × 10 -5 mbar, I = 4 µA cm -2 , t = 30 min) and flash annealing to 1600 K, followed by seven cycles of flash annealing to 1600 K and adsorption of 10 L of O 2 at T< 600 K during sample cooling to remove carbon impurities, finished by three flash annealing cycles to 1600 K without O 2 adsorption. Pt was deposited by means of electron beam physical vapour deposition with an Omicron EFM-3 evaporator at a rate of ca. 0.1 ML min -1 . During the deposition process, the sample was held at 670 K to improve the Pt layer-by-layer growth. Higher deposition temperatures must be avoided to prevent surface alloy formation. The deposition rate was frequently tested by depositing submonolayer amounts of Pt on a freshly prepared Ru(0001) surface. |
673ca21b7be152b1d09f8f66 | 30 | Electrochemical characterisation: In Ulm, the electrochemical characterisation of the Ptmodified Ru(0001) electrodes was performed in a dual thin-layer flow cell in a DEMS configuration which is described elsewhere. Before the experiments, the flow cell, including all inlet capillaries, was thoroughly rinsed with hot MilliQ water. For this work, we used a homebuilt RHE as a reference electrode (RE) and a Au wire as a counter electrode (CE). The potential was controlled with a Pine AFRDE5 bipotentiostat, and the data were acquired with an in-house programmed software.The different measurement procedures are described in Section. S2. A detailed description of the calibration of the DEMS signals is provided in Section S3. |
673ca21b7be152b1d09f8f66 | 31 | At the ESRF the electrochemical measurements were performed in a cell that allows for operando SXRD measurements, which is similar to the one described in Ref. For this work, we employed a Ag|AgCl|3.5M KCl (eDaq company) RE and a glassy carbon CE. The potential was controlled with a Biologic SP300 potentiostat. All potentials were converted to the RHE scale. A more detailed description of the electrochemical procedure at the ESRF is provided in the Section. S6. |
673ca21b7be152b1d09f8f66 | 32 | The X-ray measurements were performed at the ID03 beamline of ESRF in Grenoble. The 24 keV X-ray beam was focused to a size of 300 µm × 50 µm (horizontal×vertical relative to the plane of the sample surface) at the sample position. The incidence angle of the beam was set to 0.3°relative to the surface plane. The data were collected with a 2D Maxipix detector and analyzed as described in Ref. 95 |
678a0fce81d2151a027f0f2c | 0 | One major class of photochemical reactions are photoinduced energy transfer (PEnT) reactions, which typically involves an electronically excited photocatalyst (PC*) transferring its energy to a substrate (sub). The substrate in its electronic excited state (sub*) then undergoes a photochemical transformation. PEnT reactions can be classified into two main types: Förster Resonance Energy Transfer reactions (FRET) and Dexter Energy Transfer reactions (DET). In FRET, the energy transfer proceeds through a dipole-dipole interaction between PC* and sub and occurs over distances on the order of 1-10 nm. The majority of PEnT FRET reactions involves an energy transfer from the S1 state of the PC* and generates sub* in its S1 state. This requires a spectral overlap between the emission of the PC and the absorption of the sub. Most of the documented cases of FRET occur from the S1 state of the PC* to the sub, which is then excited to its S1 state; there are some reports of FRET occurring from the T1 of the PC to generate sub* in the singlet state. PEnT photochemical reactions proceeding by FRET, and more generally photochemical reactions involving substrates in their singlet excited state, are not nearly as common as those proceeding by DET to excite the substrate to its triplet excited state. DET proceeds via a simultaneous double electron exchange mechanism between the PC* and the sub to generate a PC in its S0 state and a sub*. This can take place from the singlet state from the 1 PC* to generate 1 sub* or more commonly from the triplet state of 3 PC* to generate sub* (Figure ) and does not involve a change of the overall multiplicity. DET processes occur over shorter distances, typically less than 10 Å, given the exponential dependence on the rate of DET with PC-sub distance. DET requires not only spectral overlap, like FRET, but also orbital overlap between the PC and the sub, which implies a collisional interaction is needed for DET to occur intermolecularly (Eq. 1). |
678a0fce81d2151a027f0f2c | 1 | In the case where DET occurs from the 3 PC* to sub, it is the spectral overlap of the phosphorescence of the PC* and the spin-forbidden absorption of the substrate that is relevant to generate sub*. Notably, as the spin-forbidden absorption spectrum is of negligible intensity in most organic substrates, the low energy onset, which is used as a surrogate estimation of the energy at which spectral overlap no longer occurs, is estimated from the onset of the low temperature phosphorescence spectrum, which is none other than the energy of the T1 state (ET) of the sub. It is generally accepted that DET will occur when ET(PC) > ET(sub), and the closer these two values are the more likely is the DET. Given that the sub* has biradical-like character, this enables a diversity of reactions such as E/Z isomerization, cycloadditions like [2+2], sensitization of metal complexes, and homolytic bond cleavages such as N2 release from benzoyl azides, scission of S -S bonds, and N -O dissociation of oxime esters to instigate the formation of carbon-centered and nitrogen centered radicals in concert with the loss of CO2. Figure . Simplified mechanism of the Förster Resonance Energy Transfer (FRET) and Dexter |
678a0fce81d2151a027f0f2c | 2 | Of the PCs typically employed in PEnT reactions, organic carbonyl-based photosensitizers have the highest ET. For instance, xanthone and acetophenone (Figure ) have similar ET values of 3.22 eV (74.2 kcal mol -1 , 310.5 kJ mol -1 ) and 3.21 eV (74.0 kcal mol -1 , 309.6 kJ mol -1 ). Their very high ET comes at the cost that they do not absorb light in the visible region (absorption maxima, at λabs of 337 and 275 nm for xanthone and acetophenone, respectively), which means that their absorption spectra may likely overlap with the absorption spectra of the substrate or other additives and thus the PC may not be selectively excited. Organometallic PCs are particularly popular for PEnT reactions in part because the presence of the heavy metal centre ensures an essentially quantitative triplet yield as a result of fast intersystem crossing (ISC) mediated by its large spin-orbit coupling (SOC). Some of the most commonly used organometallic PCs for PEnT reactions include [Ir(dF(CF3)ppy)2(dtbbpy)](PF6) (ET = 2.68 eV) for reactions requiring a relatively high ET and [Ru(bpy)3](PF6)2 (ET = 2.12 eV) for those that do not (Figure ). 1 MeOH/EtOH glass. ET for [Ir(dF(CF3)ppy)2(dtbbpy)]PF6 taken from the room temperature emission maximum in MeCN. For the organic PCs ET is measured from the phosphorescence spectra at 77 K. Over the last decade or so organic thermally activated delayed fluorescence (TADF) |
678a0fce81d2151a027f0f2c | 3 | compounds have been increasingly used in the academic community as alternatives to organometallic complexes, initially as emissive materials for organic light-emitting diodes (OLEDs), and later, as PCs. This is because TADF compounds have accessible triplet states owing to the small energy gap (∆EST) between the lowest excited singlet (S1) and triplet states (T1), which allows for relatively fast ISC and reverse ISC (RISC) to take place, despite the small SOC between these states in the absence of heavy atoms. Molecules with small ∆EST are those where the donor (D) and acceptor moieties (A) are sufficiently electronically decoupled such that there is minimal orbital overlap between the highest occupied molecular orbital (HOMO) and the lowest unoccupied molecular orbital (LUMO), i.e., the exchange energy is small. This spatial separation also produces an emissive excited state having charge transfer (CT) character. The most widely employed molecular design is a strongly twisted D-A structure, exemplified in 4CzIPN and 4DPAIPN (Figure ). Owing to the distance between D and A units, the excited state character is termed long-range charge transfer (LRCT). D-A TADF PCs based on carbazoyl dicyanobenzenes (CDCBs) have been extensively employed as PCs in myriad photocatalytic reactions. This family of TADF PC covers a broad range of ET. For example, the ET values of 4DPAIPN, 4CzTPN, 2CzIPN, 4CzIPN, and 3,5-2CzBN increase from 2.30 eV to 2.34, 2.72, 2.73, and 3.03 eV, respectively. These values overlap with those of, for example, [Ir(dF(CF3)ppy)2(dtbbpy)]PF6 |
678a0fce81d2151a027f0f2c | 4 | (ET = 2.64 eV ) and benzophenone (ET = 3.00 eV ). Beyond CDCBs, other D-A TADF PEnT PCs include pDTCz-DPmS and DI-PF (Figure ). pDTCz-DPmS (ET = 2.93 eV) possesses a suitable ET to photocatalyze the E/Z isomerization of stilbene (ET = 2.2 eV), producing 63% of the Z-isomer, while a similar yield of 66% was reported using DI-PF (ET = 2.38 eV), despite its lower triplet energy. For pDTCz-DPmS, the moderate yield may be due to its ET being higher than that of Z-stilbene (ET = 2.5 eV), thus permitting an unproductive back isomerization. 4CzIPN produces an 87% yield of Z-stilbene and hence provides the highest yield compared to the other D-A TADF PCs in this analysis. Using a substrate with a higher ET like diisopropyl fumarate (ET = 2.7 eV), results in a more efficient isomerization, with 81% formation of the diisopropyl maleate. 4CzIPN, by contrast, produced only 6% of the isopropyl fumarate ostensibly due to its too low ET of 2.68 eV. A second class of TADF compounds are so-called multi-resonant TADF (MR-TADF) emitters. |
678a0fce81d2151a027f0f2c | 5 | These are rigid polycyclic aromatic hydrocarbons that are typically doped with both electronrich and electron-deficient groups. Suitable regiochemistry of these p-and n-dopants produces an alternating pattern of increasing and decreasing electron density in the excited state as compared to the ground state, leading to the necessary small exchange energy that turns on TADF. Given the short distance between D and A motifs, the emissive excited state is of short range CT (SRCT) character. Compared to D-A TADF compounds, MR-TADF emitters have more intense low energy absorption bands, narrower emission spectra, smaller Stokes shifts, and their emission spectra show only minimal positive solvatochromism, all potentially attractive properties for their use as PCs. Our group has demonstrated the broad utility of two families of MR-TADF PCs in a range of photoinduced electron transfer (PET) and PEnT reactions. These two families contained nitrogen donor atoms and either boron or carbonyl groups as n-dopants within the polycyclic aromatic hydrocarbon framework. |
678a0fce81d2151a027f0f2c | 6 | We, as well as Lee and co-workers, each recently showed that the acceptor moiety is not required in MR-TADF emitter design for OLEDs. Indeed, when the nitrogen atoms in diindolocarbazoles are para disposed, there is very weak TADF as the ∆EST is moderately large, resulting in slow kRISC. Literature reported compounds DiICztBu4 and DiICzMes4 have similar ET of 2.55 (in DCM) and 2.57 eV (in toluene), |
678a0fce81d2151a027f0f2c | 7 | The The S1 and T1 energies were determined from the onsets of the steady-state PL and delayed emission (gate time: 1-9 ms) spectra in 2-MeTHF glass at 77 K, respectively (Figure ); the EST value is the difference in energy between these two (Figure , Table ). DiICztBu4, DiICzMes4 and DiICztBuCz4 have very similar triplet energies (ET = 2.58, 2.57, and 2.55 eV, respectively) and as their emission profiles are similar, so too will be their spectral overlap with the sub and thus the kDET should be comparable across these three PCs. ). DiICztBu4 and DiICzMes4 have previously been reported as MR-TADF emitters, and given the similar photophysical properties of DiICztBuCz4 and DiICztBuDPA4 to these two, the high molar absorptivity of the low-energy absorption band, the minimal positive PL solvatochromism, the narrowband emission and the small ∆EST, they can also be classified as MR-TADF emitters ( λexc = 400 nm for DiICztBuDPA4, λexc = 380 nm for 4CzIPN. Room temperature ET were estimated following our previously reported methodology, where ET(RT) = ES(RT -∆EST(LT), with ES(RT) being the onset of the steady-state PL in the respective solvents and ∆EST(LT) being measured in 2-MeTHF at 77 K. |
678a0fce81d2151a027f0f2c | 8 | While the low temperature (LT) measurements permit a robust estimation of ∆EST, noting that in the vast majority of TADF compounds the S1 state has greater CT character than T1, this measurement will not accurately capture ES at room temperature. This is because CT states are stabilized as a function of solvent polarity and the greater the CT character, the stronger the stabilization of the state. As triplet states tend to show lesser CT character than singlets, the estimated ET = ES(RT -∆EST(LT) represents the outer bound value for what ET may be at room temperature. Thus, given the LRCT character of the S1 state of 4CzIPN the change in ET at room temperature is more pronounced than for the four DICz MR-TADF compounds (Table ). This implies that in DET reactions the spectral overlap between the PC* and the sub may change significantly, especially when the T1 state possesses some CT character. For MR-TADF compounds this stabilization effect is small given the SRCT character of the S1 and T1 states. MR-TADF emitters maintain their ET in polar solvents, while for 4CzIPN, the ET is much more stabilized. ). Given the moderately large EST, it is not uncommon for MR-TADF emitters to not show delayed emission in solution as non-radiative decay competes with RISC, while in the solid state this is largely suppressed and delayed emission becomes apparent. DiICztBu4 and DiICzMes4 have reported delayed fluorescence lifetimes, τDF, of 12.5 ms as 1 wt% doped films in mCP:TSPO 1:1 and τDF = 433 µs as 3 wt% doped films in mCP, respectively, while there was no observed delayed emission in solution. |
678a0fce81d2151a027f0f2c | 9 | We began by studying the E/Z isomerization of an alkene, E-cinnamyl acetate, with a T1 energy similar to those of the MR-TADF emitters; notably, the structurally similar E-methyl cinnamate has a ET = 2.38 eV (Table ). The geometric isomerization does not take place in the absence of an irradiated PC ( Figure shows the reaction progression for the E/Z isomerization of cinnamyl acetate in air at short times, while for DiICztBuDPA4 the reaction under N2 is shown. The conversion saturates after 30 min for DiICztBu4, DiICzMes4 and DiICztBuCz4, with these PCs having similar reaction rates. There is no correlation observed between ∆EST and either the E/Z ratio or reaction rate for this transformation. The absence of any substantive change in the post reaction absorption spectra of DiICztBu4, DiICzMes4, and DiICztBuCz4 revealed that these compounds are photochemically stable to O2 under the reaction conditions (Figure ). |
678a0fce81d2151a027f0f2c | 10 | Given that O2 does not affect the E/Z ratio, we investigated the effect on the reaction kinetics as a function of the presence/absence of O2 by comparing E/Z ratios after 5 min. O2 acts as an ineffective yet competitive quencher of the triplet excited state of these PCs as the change in E/Z ratio at early time is marginal. |
678a0fce81d2151a027f0f2c | 11 | The reaction progression with DiICztBuDPA4 was assessed only under N2 as this reaction is completely quenched when oxygen is present. As a result of the ostensibly poorer spectral overlap due to its lower ET DiICztBuDPA4 photocatalyzes the reaction significantly slower than the other three PCs of this family, where after 60 min the E/Z ratio is 60/40. The slower rate of isomerization may explain why, for this PC, oxygen acts as a more effective competitive quencher. Given that there is no significant difference in the PL under N2 or in air, we can rule out this parameter as governing the reaction rates (Table ). Photostability studies of DiICztBuDPA4 reveal a more significant change after the reaction both in air and under N2 than using the other PCs (Figure ). This implies that there may be an O2-induced photochemical degradation of DiICztBuDPA4, which would explain why the E/Z isomerization reaction does not proceed in air with this PC. Under N2, there is an observed change in the absorption spectrum of DiICztBuDPA4 after 24 h of reaction; however, this spectral change is different to the one observed in air. Under N2, the SRCT and at λabs = 485 nm remains, yet there is a slight red-shift in its onset compared to efore irradiation (new λabs = 491 nm). It is not clear what effect, if any, this spectral change has on the performance of the PC in the E/Z isomerization of cinnamyl acetate (Figure ) nor were we able to detect or isolate any photodegradation products. |
678a0fce81d2151a027f0f2c | 12 | The isomerization reaction with 4CzIPN is significantly slower than with DiICztBu4, DiICzMes4, and DiICztBuCz4, only producing a 39/61 E/Z ratio after 60 min (Figure ). Thus, despite similar thermodynamic driving forces (i.e., similar ET), the origin of the higher yields of the Z isomer and faster reaction kinetics could be due to either the higher molar absorptivity of these three MR-TADF compounds at the photoexcitation wavelength and/or an increased spectral overlap of the phosphorescence of the MR-TADF emitters with the spin-forbidden absorption of the sub (Eq. 1). 4CzIPN is mostly photostable; however, a slight decrease in the absorbance of the CT band is observed in air, suggesting that some degradation is taking place if the reaction is carried out when oxygen is present (Figure ). Further, there is a significant dependency of the kinetics of the reaction with respect to the presence/absence of O2. After 5 min, the E/Z ratios are 80/20 and 94/6 under N2 and air, respectively. These results reveal that 4CzIPN photocatalyzes this reaction more slowly than DiICztBu4, DiICzMes4, and DiICztBuCz4 and that its triplet excited state is also more effectively quenched by O2. We next investigated the E/Z isomerization of diisopropyl fumarate, which has a higher ET of 2.7 eV, to produce the corresponding maleate (ET = 3.1 eV) (Table ). Despite the ETs of the PCs being lower than 2.7 eV ( To understand the divergence in behavior of these two isomerization reactions in the presence of O2, we interrogated the reaction rates of diisopropyl fumarate under N2 (Figure ). If the rates are slower compared to those with cinnamyl acetate, then the lower E/Z ratios may be explained by competitive O2 quenching of the T1 state to that of the fumarate. Unlike the reaction with cinnamyl acetate, the reaction rates under N2 differ between DiICzMes4, DiICztBu4, and DiICztBuCz4 but still are all faster than 4CzIPN. The fastest conversion to the Z isomer uses DiICzMes4, yielding an E/Z ratio of 22/79 ± 3 after 60 min (13:87 after 24 h, Table , Entry 6), while DiICztBu4 only produces an E/Z ratio of 60/41 ± 1 after 60 min, yet affords the same E/Z ratio after 24 h as DiICzMes4 (13:87 after 24 h, Table , Entry 4). The reaction reaches its steady state in E/Z ratio after 2 h for DiICzMes4 (E/Z ratio of 12/88), while for DiICztBu4 it takes 4 h to reach the 13/87 E/Z ratio. These rates are slower than those with cinnamyl acetate, where the reaction was completed after 30 min. Given that both DiICzMes4 and DiICztBu4 are stable during the reaction when conducted under N2, photodegradation of the PC can be excluded as the origin of the slower reaction kinetics observed for DiICztBu4. As the phosphorescence of both DiICztBu4 and DiICzMes4 are the same, the spectral overlap between the phosphorescence spectra of these two PCs and the spin-forbidden absorption spectrum of the sub will therefore be similar. A plausible conclusion is that the tert-butyl groups are bulkier than mesityl groups and impede the collisional interaction more, thus adversely affecting the reaction kinetics with DiICztBu4. We observed that these two PCs do photodegrade when the reaction is carried out under air. Thus, there is a divergence in the photochemical outcome wherein the kinetics of the E/Z isomerization of cinnamyl acetate outcompetes photodegradation while this is not the case with the E/Z isomerization of diisopropyl fumarate (Figures S45a&b). Surprisingly, while the initial rate of isomerization using DiICztBuCz4 is comparatively as fast as using DiICzMes4, there is a significant off-cycle photodegradation both in air and N2 that effectively caps the E/Z ratio at only 66/34 ± 4 (Figures S45c). |
678a0fce81d2151a027f0f2c | 13 | DiICztBuDPA4 is photostable when the reaction is conducted under N2 while there are changes in the absorption spectrum when the reaction is conducted in air (Figure ). Hence, the fact that the reaction does not proceed is a consequence of there being no spectral overlap between the phosphorescence of DiICztBuDPA4 and the spin-forbidden absorption of the substrate. |
678a0fce81d2151a027f0f2c | 14 | 4CzIPN is significantly slower and less efficient in this isomerization reaction, producing a steady-state E/Z ratio of only 69/31 ± 4 after 24 h (Table , Entry 12); indeed, after 4 hours there is only an E/Z ratio of 95/5 ± 1 (Table ). Under these conditions, 4CzIPN is photostable under N2, though in air photodegradation of 4CzIPN is observed (Figure ). Thus, the poor performance of 4CzIPN can be explained by the too-small spectral overlap with the fumarate substrate. |
678a0fce81d2151a027f0f2c | 15 | The E/Z isomerization of diisopropyl fumarate studies demonstrate that three of these PCs can catalyze PEnT reactions with substrates having ET of 2.7 eV. Thus, we next explored the intramolecular [2+2] cycloaddition of norbornadiene (ET = 2.7 eV ) to quadricyclane (Table ). Schmid et al. reported a 99% yield after one hour using an anionic Ir complex having an ET of 2.99 eV (Table , Entry 14). The reaction does not take place in the absence of a PC (Table , Entries 1 and 2). Initial tests using DiICztBu4 showed that quadricyclane is obtained in an 87% yield after two hours under N2 (Table , Entry 4). Running this reaction on a shorter time scale revealed that the reaction is already completed after 30 min. In contrast to the E/Z isomerization of fumarate, DiICzMes4 and DiICztBuCz4 perform as well as DiICztBu4, yielding 85, 88, and 89%, respectively, after 30 min (Table , Entries 5, 7, and 9). For all three PCs, the reaction is slightly quenched in the presence of oxygen, producing yields of 64, 71, and 75% after 30 min when using DiICztBu4, DiICzMes4, and DiICztBuCz4, respectively ( ). These results not only demonstrate the value of three of these MR-TADF PCs to photocatalyze a demanding PEnT reaction but also that ET as a parameter is too coarse when cross-comparing different reactions using substrates with similar ET. a Norbornadiene (0.2 mmol) and PC (1 mol%) in DCM (2 mL). Yields were determined by 1 H NMR spectroscopy using 1,3,5-trimethoxybenzene as the internal standard. b from ref. in CD3CN exc = 415 nm after 1 h, norbornadiene (0.05 mmol), PC (0.3 mol%), CD3CN (0.6 mL). |
678a0fce81d2151a027f0f2c | 16 | The photostability of the PCs was tested under the reaction conditions under N2 and in air (Figure ). The absorption spectra of DiICztBu4, DiICzMes4, and DiICztBuCz4 show only a minimal decrease in intensity under N2, while there are greater changes with oxygen present that suggests more photodegradation, which is consistent with the less efficient performance when the reaction is carried out under air (Figure ). Interestingly, the small degree of photodegradation of the MR-TADF PCs in the [2+2] cycloaddition in air contrasts with the significant photodegradation in the E/Z isomerization with diisopropyl fumarate in air and the associated absence of product formation, despite both substrates having the same reported ET. |
678a0fce81d2151a027f0f2c | 17 | For DiICztBuDPA4 no photodegradation is observed, and this is because no reaction takes place either with norbornadiene or O2 (Figure ). Instead, the intensity of the absorption spectrum after the reaction is slightly higher than before irradiation, which can be attributed to the poor solubility of this PC in the solution and where the concentration of dissolved PC slightly increases after stirring for 30 min. There is significant photodegradation observed for 4CzIPN under N2, and the profile is similar to that in air (Figure ). The absorption band at λabs = 377 nm and the shoulder at λabs = 450 nm disappear while a more red-shifted, less intense band at λabs = 520 nm appears. These results imply that the substrate reacts with 4CzIPN and that the [2+2] cycloaddition is likely radical stepwise as opposed to a concerted mechanism (Table , Entry 12 and 13). |
678a0fce81d2151a027f0f2c | 18 | We then explored a PEnT reaction with a substrate of higher ET, the sigmatropic shift of (S)verbenone (ET = 3.0 eV) to chrysanthenone (Table ). This rearrangement has been used in the synthesis of the natural product xishacorene B where verbenone was directly irradiated in the first step with UV light (λexc = 365 nm) yielding 67% chrysanthenone. Schmid et al. |
678a0fce81d2151a027f0f2c | 19 | ). There is essentially no change in the absorption profiles of DiICztBu4, DiICzMes4, and DiICztBuCz4 before and after irradiation, suggesting that any photodegradation of the PCs is not the cause for the low yields in this reaction but rather the too low triplet energy (Figure ). 4CzIPN and DiICztBuDPA4 are stable under the reaction conditions (Figure and e) and the 5% yield for 4CzIPN and the 0% yield for DiICztBuDPA4 can be attributed to each having a T1 energy that is effectively too low to enable the DET with any degree of efficiency. a (S)-verbenone (0.2 mmol) in DCM (0.1 M) with given PC loadings. Yields were determined by 1 H NMR spectroscopy using 1,3,5-trimethoxybenzene as the internal standard. b from ref. in CD3CN with hv = 415 nm after 3 h. |
678a0fce81d2151a027f0f2c | 20 | Thus far, we have explored four intramolecular PEnT reactions. To expand the portfolio of available reactions we next investigated an example of a bimolecular cross-coupling reaction. This is frequently done by combining the photocatalytic cycle with a Ni-mediated crosscoupling reaction. In such dual-catalyzed reactions, the PC can either be involved in a PET mechanism, a so-called metallaphotoredox reaction, or in a PEnT reaction to sensitize the Ni co-catalyst by DET after it has undergone oxidative addition and transmetallation steps. This photoactivation accelerates the reductive elimination of the product. One such crosscoupling reaction where the PC is purported to engage in DET is an esterification involving aryl halides being cross-coupled with carboxylic acids where Welin et al. used fac-Ir(ppy)3 as the PC (ET(fac-Ir(ppy)3) = 2.58 eV). In this cross-coupling reaction, an undesired side product is lactone 2 (Table ). Several groups have employed organic PCs like 4DPAPN or SACR-IPTZ for this reactions (Figure ). 4DPAPN, after optimization of the reaction conditions, yielded 91% of 4-(trifluoromethyl)phenyl benzoate in the cross-coupling of benzoic acid with 4-bromobenzotrifluoride as substrates. With the same substrates but under slightly different conditions, Welin et al. reported an 86% yield using fac-Ir(ppy)3 as the PC. The reaction photocatalyzed using SACR-IPTZ yielded 99% of the coupled product between 5bromophthalide and benzoic acid, and no protodehalogenated product was observed. a 5-Bromophthalide (0.188 mmol, 1.0 equiv.), benzoic acid (0.301 mmol, 1.6 equiv.), 2,2,6,6tetramethylpiperidine (TMP) (0.375 mmol, 2.0 equiv.), Ni source (0.011 mmol, 6 mol%), 4,4'di-tert-butyl-2,2'-bipyridine (dtbbpy) (0.013 mmol, 7 mol%) and PC (0.004 mmol, 2 mol%) in DCM (0.09 M). Yields were determined by 1 H NMR spectroscopy using 1,3,5-trimethoxy benzene as the internal standard. |
678a0fce81d2151a027f0f2c | 21 | We first investigated the use of different Ni precursors in the presence of dtbbpy as an ancillary ligand under conditions similar to those reported by Hojo et al. We changed the solvent from DMF to DCM as the DiICz PCs are more soluble in the later. We observe a generally higher product yield in DCM when using Ni(COD)2 than with NiBr2•glyme, regardless of choice of PC. The use of DiICzMes4 yielded 65% in combination with Ni(COD)2 while using NiBr2•glyme only results in a 41% yield ( was obtained in a 74% yield when using a 1:1 mixture of DMSO:toluene (Table , Entry 4), while the control experiment with fac-Ir(ppy)3 yielded a comparable yield of 77% (Table , Entry 6). The 1:1 DMSO:toluene solvent system had previously been shown to be optimal with SACR-IPTZ as the PC, yielding 99% of 1 (Table , Entry 8). The reaction does not readily proceed in toluene, yielding only 10% product, while in DMSO an increase in the yield (64% of 1) compared to in DCM is observed, ostensibly due to a partial suppression of the formation of 2 (Table , Entry 1-3). Using 1:1 DMSO:toluene in combination with the preformed Ni complex [Ni(dtbbpy)(OH2)4]Cl2 and DiICzMes4 as the PC resulted in 81% yield of 1 and only 13% of 2 (Table , Entry 7). Reducing the reaction time from 24 to 3 h resulted in a lower yield of 1 (54%) but not of 2 (13%) (Table , Entry 5), while with fac-Ir(ppy)3 as the PC the product was obtained in 68% yield after 3 h (Table , Entry 7). While the reaction with fac-Ir(ppy)3 is slightly faster than with DiICzMes4, effectively similar yields of 74 and 77% were obtained after 24 h using DiICzMes4 and fac-Ir(ppy)3, respectively. |
6764032b6dde43c9089c9ea6 | 0 | All domains of life depend on redox chemistry to generate energy through respiration or fermentation. Electrons flow from electron donors to various terminal electron acceptors (TEAs) through redox active biomolecules. These molecules include proteins (e.g., cytochromes, oxidoreductases) and small molecules (e.g., quinones, flavins, phenazines) and all have the ability to change oxidation states under biological conditions. Oxygen is the most common TEA and is used by all aerobic organisms; however, many microorganisms have evolved to utilize insoluble TEAs, such as metal oxides. Utilization of insoluble TEAs requires passing electrons through cellular membranes and is commonly referred to as extracellular electron transfer (EET). There are three main mechanisms to accomplish EET: 1) direct contact of redox proteins located in the cell envelope with the TEA, 2) secretion of cytochrome containing nanowires or outer membrane vesicles, or 3) use of small molecule redox mediators, also known as extracellular electron shuttles. Of all the EET mechanisms, mediated EET is the least understood with little known about the physicochemical properties of the mediators that drive EET activity. |
6764032b6dde43c9089c9ea6 | 1 | In addition to utilization of natural TEAs, many organisms that perform EET can also use an anode within a bioelectrical system (BES) as a TEA, thereby generating a current. The detectable current is an electrical feature that has been leveraged for biosensing applications as it can report on a microorganism's local environment. For example, engineered bioelectronic signaling systems have successfully been designed to sense biologically-relevant small molecules, including pyocyanin and riboflavin. An engineered two microbe system has been designed to detect and degrade organophosphate pesticides, where modified Shewanella oneidensis detects degradation products of the pesticides that are produced by Escherichia coli. However, the readout is slow (requires translation), insensitive, and lacks selectivity. Alternatively, E. coli has been engineered so that an analyte triggers a conformational change in one of the proteins in a synthetic EET pathway, allowing an endocrine disruptor to be detected in minutes. However, these sensors rely on bespoke engineered electron transfer pathways and proteins that must be reengineered for each analyte. Advancing our understanding of mediated EET has the potential to overcome some of these challenges. |
6764032b6dde43c9089c9ea6 | 2 | Lactiplantibacillus plantarum, a commensal lactic acid bacterium, is capable of performing mediated EET using flavins and quinones as small molecule redox mediators, together with its highly conserved FLEET gene locus encoding a type-II NADH dehydrogenase (Ndh2), flavin transport proteins, membrane demethylmenaquinone (DMK) synthesis proteins (DmkA, EetA, and EetB), and flavin mononucleotide (FMN) cofactor (PlpA). Recent studies by our group showed that Ndh2 is required for utilization of quinone mediators. Although L. plantarum has defined pathways that utilize quinones, it is incapable of producing them, and instead, must acquire quinones exogenously from its environment. Quinones are known to be abundant in many environmental niches, with high concentrations present in human gut. L. plantarum can convert exogenously acquired DHNA (1; 1,4-dihydroxy-2-naphthoic acid), an intermediate in the biosynthesis of menaquinone (vitamin K), to demethylmenaquinone through prenylation by DmkA. L. plantarum uses a variety of exogenous quinone mediators, though very little is known about their specificity for Ndh2. A small study that evaluated the Ndh2-dependent EET of six commercially available mediators indicated that reduction potential (E o ) and predicted free energy of binding (ΔGcomp) impacted EET activity. However, that study was limited in scope and only evaluated commercially available, structurally related 1,4-naphthoquinones, and one benzoquinone analog as mediators. For a mediator to be viable for EET, it must possess certain physicochemical properties and favorable biological interactions. In addition to binding favorably to Ndh2, mediators must have a favorable E o to accept an electron from Ndh2 and pass it to a given TEA. A mediator must also be lipophilic enough to diffuse through a lipid bilayer, yet have a diffusion coefficient (Dwater) that supports diffusion through aqueous environments to an insoluble TEA (Fig. ). However, a mediator's EET activity is unlikely to be equally dependent on each of these parameters as the timescale of each varies. Of the steps that a mediator must accomplish, the electron transfer is known to occur on the microto milli-second timescale and just requires an acceptable E o to facilitate the transfer of electrons. Meaning, the mediator must simply have a E o between that of Ndh2 and the TEA. We hypothesize that other physicochemical properties of the mediator, such as Dwater, lipophilicity (LogD), and topological polarization surface area (tPSA), have a larger impact on EET activity as they drive slower, potentially rate-limiting steps. Passive transport through a membrane and diffusion through an aqueous environment occur on the timescale of milliseconds for compounds with favorable properties but can be much slower for larger, charged compounds. This is up to an order of magnitude slower than the redox chemistry, thus we hypothesize these properties are likely very important for EET activity. Herein, we aim to better understand the chemical properties most important to mediating Ndh2-dependent EET in L. plantarum. A library of quinones, including naphtho-, anthra-, and benzoquinones, was designed to probe diverse chemical properties with a focus on polarity and lipophilicity. The mediator library was inspired by natural products that have previously exhibited shuttling properties. Our library was assembled using semi-synthesis to derivatize three scaffolds, isolated from natural sources, or acquired commercially. Examining the relationship between Ndh2-dependent EET activity with physicochemical and biochemical properties of the mediators revealed that LogD and ΔGcomp significantly correlates to increased EET activity (p < 0.0001 and p = 0.0115, respectively). Analysis of a subset of the mediators in a BES confirmed that select mediators can perform Ndh2-dependent EET using a carbon felt anode as the TEA, producing a stable current for duration of the experiment (up to 5 days). |
6764032b6dde43c9089c9ea6 | 3 | To probe the importance of physicochemical properties and biochemical interactions on Ndh2-dependent EET in L. plantarum, a library of quinones, consisting of naphthoquinones (1 -29), anthraquinones (30 -34), and benzoquinones (35 -40), was designed by drawing inspiration from natural small molecules with known electron shuttling properties. A subset of the quinones were obtained through commercial or natural sources (Fig. ). Chimaphilin (6) and hydroxy-chimaphilin (7) were isolated from lyophilized Chimaphila umbellata bark (Fig. ). Compounds 8 -11 are produced by the Pau D'Arco (Tabebuia impetiginosa) tree, but the purification proved challenging, so they were purchased for this study. Quinones 16 -29 were obtained through semi-synthesis as part of this study (Schemes 1 and 2; Fig. ). Our synthetic efforts were focused on diversifying 1,4-naphthoquinones, specifically adding functionality to the unoccupied 3-position of menadione ( ) or the unoccupied 2-and 3-positions of naphthoquinone (15), as these positions are in conjugation with the quinone functionality and are therefore hypothesized to have the largest impact on EET activity. In addition to the naphthoquinones, structurally diverse anthraquinones (30 -34) and benzoquinones (35 -40) were acquired through commercial sources. |
6764032b6dde43c9089c9ea6 | 4 | Our synthetic efforts were focused on installing modifications that are commonly found in nature, e.g., halogenations, hydroxylation, and animation. Using 14 and 15 as a scaffold, halogenated and hydroxylated analogs were synthesized using one or two step reaction sequences (Schemes 1 and 2). First, the unoccupied 3-position of 14 was chlorinated by treatment with N-chlorosuccinimide and copper(II) chloride to yield chloro-menadione ( ) in decent yields (53%). Bromination of the same position involved treatment of 14 with bromine, sodium acetate, and acetic acid to yield bromomenadione (17) in moderate yields (71%). Similar halogenation reactions were also conducted on 15. Dichlorination of 15 was accomplished through treatment with thionyl chloride and pyridine to yield dichloro-naphthoquinone (18) in relatively poor yields (21%). Attempts to increase overall yields by converting 3 to 18 using similar reaction conditions led to 2-chloro-naphthoquinone (19), in good yields (77%). Dibromination of 15 was accomplished using the same conditions as the bromination of 14 but the reaction needed to be heated to produce 20 in decent yields (45%). Overall, the majority of the halogenation reactions proceeded very slowly and starting material was still observed when the reactions were quenched. |
6764032b6dde43c9089c9ea6 | 5 | Next, we focused on hydroxylation, nitration, and amination reactions with 1, 14, and 15. Hydroxylation of 14 was accomplished in two steps. First, treatment of 14 with hydrogen peroxide (30%) in sodium carbonate generated 2,3-epoxidemenadione (21i), which was subsequently treated with sulfuric acid to yield hydroxy-menadione (21) in an overall yield of 62%. Several attempts were made to synthesize methoxy-menadione (13) from 21 using methyl iodide and silver(I) oxide, however it yielded a mixture of inseparable products -13 and 4methoxynaphthalene-1,2-dione. Therefore, 13 was obtained from a commercial source and no other attempts were made to derivatize the hydroxyl in 21. Introduction of the nitro-functionality into 14 involved treatment with a mixture of concentrated nitric acid and sulfuric acid. The orange crude product was recrystallized in ethanol to yield nitro-menadione (22) in low yields (24%). Uncrystallized product was observed in the filtrate but was discarded rather than conducting a large-scale biphasic extraction involving concentrated nitric acid. Pure 22 was subsequently reduced with 10% Pd/C and H2 (g) in quantitative yields to produce amine-menatione (23). In addition, ACNQ (5) was obtained through treating 1 with ammonium chloride following previously reported methods. Following literature precedent, primary (e.g., methyl and ethyl amine) and secondary (e.g., piperidine and morpholine) amines were added to 14 through a perchloric acid catalyzed reaction, yielding mediators 24 -27 in moderate yields (20 -44%). Lastly, similar conditions on 15 yielded methylamine-naphthoquinone (28) and morpholinenaphthoquinone (29) in moderate yields (38 -57%). Many chemical properties of the mediators are easily obtained using available software or online tools, such as ChemDraw and MolGpka, which were used to calculate LogD, and to determine tPSA (Table and). However, calculating the E o and Ndh2 binding affinity (Ki) of a compound is considerably more challenging. For E o , previous literature established that there is a linear relationship between the energy difference of the ground state quinone and radical anion, and experimentally determined one electron (1e -) E°. Therefore, we used Density Functional Theory (DFT) to calculate this energy on all mediators (Table ) and the E o was experimentally measured for a diverse subset (twenty mediators) of the library (Fig. ). Graphing the theoretical molecular energy difference versus E o yielded a line of best fit (R 2 = 0.85) that was used to calculate the E o for the remaining mediators (Fig. ). All E o were determined to fall between -1.44-(-0.16) V (Table ). This method allowed us to calculate the E o of 1, which posed a challenge experimentally due to the difficulty to obtain the molecule commercially in the quinone form. Finally, we used molecular docking tools, specifically AutoDock Vina, to approximate each mediator's Ndh2 binding affinity. Although this predicted free energy of binding (ΔGcomp) is not the same as binding affinity, it correlates with the mediator's ability to bind to Ndh2. Experimental determination of Ki values is not trivial as Ndh2 is a membrane protein, thus significantly complicating protein expression. Ndh2-dependent EET activity in L. plantarum was quantified using an established iron(III) oxide nanoparticle reduction assay, which uses ferrozine to quantify the amount of Fe 3+ that is converted to Fe 2+ by L. plantarum (Fig. ). Absent of any exogenous mediator, no reduction of Fe 3+ to Fe 2+ is observed. Although L. plantarum is incapable of producing quinones, the bacterium does have the ability to metabolize these compounds. Specifically, DmkA has the ability to prenylate unoccupied 2-or 3positions of naphthoquinone analogs to produce demethylmenaquinone analogs. To ensure the observed EET is mediated by Ndh2, we compared the EET activity between two previously developed mutant strains of L. plantarum (L. plantarum ∆dmkA∆ndh1 and L. plantarum ∆dmkA∆ndh1/2). Removal of the genes dmkA and ndh1 eliminates EET side-reactions that may modify, or consume the mediator, ensuring EET depends primarily on the interaction between Ndh2 and the given mediator. All mediators were evaluated for EET activity at 5 μM, counter screened for toxicity at 5 and 50 μM, and evaluated for chelation properties (Fig. ). As additional confirmation that observed EET activity is Ndh2-dependent, we conducted additional iron(III) oxide nanoparticle assays with all quinones and two additional strains of L. plantarum ∆dmkA∆ndh1/2: L. plantarum expressing a functional Ndh2 (∆dmkA∆ndh1/2 + pSL93) and with a point mutant (∆dmkA∆ndh1/2 + pRA01). This point mutation reduces activity in the homologous Ndh2 of C. thermarum, strongly suggesting that the activity in an analogous mutation in L. plantarum Ndh2 will alter EET activity. Overall, just about half of the quinone library was able to perform Ndh2-dependent EET (Fig. ). Both naphthoquinones and benzoquinones were capable of mediating EET, while none of the anthraquinones were selective for Ndh2-dependent EET. These patterns of Ndh2-dependent EET were also observed in the assay comparing the point mutant Ndh2 to functional Ndh2 (Fig. ). In addition, abiotic controls of the iron(III) oxide nanoparticle assays showed no significant background activity (Fig. ). The top ten performing mediators in Fig. included seven naphthoquinones (3, 5, 7, 8, 19, 21, and 23) and three benzoquinones (36, 37, and 39). All seven of the naphthoquinones and two of the benzoquinones contain an electron-donating group (hydroxyl, halogen, methoxy, or amine) that is conjugated to the quinone moiety. In addition, nine of the top ten mediators possess a hydrogen bond donor (e.g., methoxy, hydroxyl, cyano, carboxylic acid, amine) in either the 2-or 3- As predicted by pKa calculations; All E o listed were calculated using the linear relationship between the energy difference of the ground state quinone and radical anion compared to select experimentally determined one electron E o using an Ag/Ag + reference electrode; Predicted free energy of binding which was calculated using pharmacophore modeling; position. Closer analysis of the molecular docking poses for EET active compounds with hydrogen bond donating substituents (3, 21, 23) revealed the hydrogen bond donor substituent is positioned within 2.7 -3.0 Å of the sidechain amide carbonyl on asparagine-387 (Fig. ). Conversely, for alkylated analogs (13, 24, 26), the lowest energy poses have the mediator rotated and not within hydrogen bonding distance of asparagine-387 (Fig. ). Therefore, indicating a hydrogen bond donor on the mediator might promote favorable interactions within the Ndh2 active site and could be stabilizing the electron transfer. |
6764032b6dde43c9089c9ea6 | 6 | Although much of these findings are in support of Li et al. 2024, one discrepancy is that 1 was determined to be relatively inactive under our conditions but was previously shown perform Ndh2 mediated EET in the iron(III) oxide nanoparticle assay. The driving factor behind this discrepancy is unknown but we hypothesize this could be a result of stability of 1, where degradation products are more active than 1. Although stable in powder, 1 has shown to be inherently unstable in solution and likely degraded during the 24 h timescale of the previous iron(III) oxide nanoparticle reduction assay. Evaluation of the stability of 1 under iron(III) oxide nanoparticle assay conditions confirmed stability issues, while 94.5 ± 3.8% of 1 is remaining at 3 h, only 26 ± 4.2% is remaining at 24 h (Fig. ), with numerous more polar metabolites detected by HPLC (Fig. and). Based on the data presented here, metabolites of 1 that are more polar would be expected to show higher EET activity, aligned with what was previously observed. |
6764032b6dde43c9089c9ea6 | 7 | Comparison of Ndh2-dependent EET to the different physicochemical properties that were hypothesized to drive activity, including LogD, ΔGcomp, E o , tPSA, and Dwater, revealed that activity correlated linearly to ΔGcomp (p = 0.0115, Fig. ) and LogD (p < 0.0001, Fig. ). The LogD analysis does omit 34 as this mediator possesses a positive ΔGcomp, thus indicating an unfavorable interaction with Ndh2. No relationship was observed between EET activity and E o , tPSA, or Dwater (Fig. ). Additionally, the E o of the top ten EET active compounds (-1.30 to -0.16 V, Table ) was not distinguishable from the range of E o in the library Therefore, a successful mediator should be relatively polar (~LogD < 2), have favorable predicted free energy of binding (< -2.5 kcal/mol). Since screening mediators via iron(III) oxide reduction is much higher throughput than BES experiments (Fig. ), we used the high throughput assay data to select compounds for further study. We selected five structurally distinct mediators for a more thorough analysis in a BES. This included three amine analogs (23, 24, and 27; Fig. ), and two hydroxyl analogs (7 and 21; Fig. ). To characterize each mediator's electrochemical properties, we conducted cyclic voltammetry (CV) at varying scan rates (Fig. ). The Dwater of each quinone was calculated from the CV data by plotting the peak current (ip) against the square root of the scan rate ( 1/2 ). According to the Randles-Sevcik equation, the resulting linear curves confirmed that the electron transfer process to the electrode is diffusion-controlled, as expected. The diffusion coefficients ranged from 1.1210 -6 to 4.3510 -7 cm 2 /s, consistent with values reported for quinone diffusion in Table . To further evaluate each quinone's potential use in biosensing applications, we performed BES experiments following previously published protocols, evaluating the current generation from each mediator in the presence of the same two L. plantarum strains previously described in triplicate independent reactor experiments (Fig. ). All five mediators enabled Ndh2-dependent EET; however, the hydroxyl analogs (7 and 21) produced limited signal above the background current generated by the Ndh2-deficient strain (Fig. ). In contrast, the amine analogs (23, 24, and 27) produced steady-state current in the Ndh2-competent strain that was significantly greater than the background signal. The background signal from the Ndh2deficient strain was very low, settling to a current density value close to 0 μA/cm 2 within 8 to 12 h (Fig. ). Moreover, the temporal evolution of current across the three analogs was qualitatively similar, reaching the peak within ~4 h and remaining near that peak level for the duration of the experiment. These data show that these amine analogs stably mediate Ndh2-dependent EET over 24 h, strongly suggesting they could effectively mediate signals for bioelectronic sensing. The difference in relative current magnitude compared to iron(II) generation for each mediator is likely due to inherent differences in their interactions with the different TEA (iron(III) oxide vs electrode). |
6764032b6dde43c9089c9ea6 | 8 | Since 23 produced the strongest signal in the Ndh2competent strain, we conducted additional experiments to investigate this mediator's performance in a BES over longer time scales. Indeed, 23 stably mediated EET with Ndh2 for 5 days (120 Current density for all data is given as the change in current density from the time of quinone injection at t = 0 h. h), yielding incredibly stable, predictable current output over the full experiment (Fig. ). We conducted these experiments in comparison against the EET performance of 14, previously characterized to be a strong natural substrate for Ndh2 in BES experiments. While 14 produced a much greater current magnitude than 23, 14 also produced a high background signal in the Ndh2-deficient strain. Additionally, current generation for 14 reaches its maximum within a short period of time after mediator addition and current output continually drops over the course of the experiment under these experimental conditions. Neither 14 nor 23 generate current under abiotic conditions or in the presence of heat-killed cells, further confirming that EET signal is dependent on the presence of living cells (Fig. ). Comparison of normalized chronoamperometry traces for 14 and 23 (see experimental details), clearly illustrates the differences in current profile between the two compounds, specifically the stability of 23 (Fig. ). Taken together, these results highlight the potential of 23, 24, and 27 for use in biosensing applications, where stable current responses with low background are ideal properties. As such, these compounds have the potential to be used in wholecell L. plantarum bioelectronic devices, expanding the molecules available for EET-based biosensing. |
6764032b6dde43c9089c9ea6 | 9 | In summary, we generated a library of 40 structurally diverse quinone mediators to gain a better understanding of the important physicochemical properties and biochemical interactions driving Nhd2-dependent EET in L. plantarum, a commensal gut bacterium. The library consists of twenty-nine naphthoquinones, six benzoquinones, and five anthraquinones that were obtained through semi-synthesis, natural product isolation and commercial sources. A particular focus of the library was to assemble mediators with diverse LogD, E o , and ΔGcomp. Correlation of these properties to the Ndh2-dependent EET activity in an iron(III) oxide nanoparticle reduction assay revealed a significant relationship between Ndh2-dependent EET activity and both LogD (p < 0.0001) and ΔGcomp (p = 0.0115). Analysis of some of the mediators in a BES confirmed that they are each capable of generating a detectable current, and the current of one of the mediators generated was stable over 5 d. This study has significantly increased our library of known mediators, provided insights into the properties that make a mediator successful, and has led to a mediator (23) that generates stable current outputs for potential use in a variety of biosensing applications. |
63bbe84ba5c354823a8c1f54 | 0 | Inverse molecular design proposes molecules with predefined properties and is a reverse process of traditional direct design processes. In the reverse design process, molecules are optimized for the properties. One approach is to explore chemical space, a set of candidate molecules, and functionality space mapped onto the chemical space. In contrast to early efforts toward inverse design, exploration leads directly to chemical structures. Only the subspace of the chemical space is explored by using the information of the functionality space, and the exhaustive enumeration of the properties of all the candidate molecules is avoided. This efficiency is appealing because the chemical space combinatorially expands with molecular size. Thus, inverse design is a promising approach for accelerating the development of functional molecules and materials. Importantly, the choice of representation of molecules and their properties dramatically impacts the performance of molecular design. Electronic structure theories can map molecular structures onto various physical properties without experimental records and are a natural choice for inverse design. The inverse molecular design approaches based on electronic structure theories involve the variational particle approach, linear combination of atomic potentials (LCAP), and quantum algorithm-based alchemical optimization. Considering the geometric stability of molecules in the design process is essential for designing functional molecules available in experiments under ambient conditions. Recently, we proposed an inverse molecular design method based on quantum alchemy to efficiently explore chemical space composed of molecules with equilibrium geometries. Hereafter, this method is referred to as MDM. Quantum alchemy models the change in composition ("alchemical" changes) from a reference molecule to a target molecule at the level of quantum mechanics. MDM simultaneously predicts the molecular species, target properties, and equilibrium geometry of the functional molecule without empirical data. In the design process, MDM gradually varies the molecular species and equilibrium geometry using the gradient of the target property with respect to the molecular species that accounts for the change in equilibrium structure. MDM was applied to various chemical spaces, one of which contains 3.1 × 10 5 BN-doped phenanthrene derivatives, and its capabilities in the design of energy-related functionalities were demonstrated. MDM can target various molecular properties calculated by the quantum alchemy method. However, evaluating the property gradient other than for energy-related properties (e.g., energy, atomization energy, and reaction energy) is computationally expensive. This greatly limits the scope of applications. To overcome this problem, MDM applicable to the general properties with a small computational cost is needed. |
63bbe84ba5c354823a8c1f54 | 1 | In this work, we have extended MDM for general functionalities by improving the optimization algorithm. The present MDM separates two coupled optimization problems for the molecular species and equilibrium geometry; MDM gradually and iteratively changes the molecular species and equilibrium geometry during the exploration of chemical space but avoids evaluating the contribution of the geometry change in the gradient. This allows for a dramatic reduction in the computational cost. This type of approach has been demonstrated in discrete optimization using semiempirical AM1-based LCAP. Moreover, we achieved rapid design by improving the constraint optimization method for updating molecular species. The proposed MDM was successfully applied to molecular design for electric dipole moment and atomization energy with small chemical spaces. The electric dipole moment is important for optical properties and intermolecular interactions. The atomization energy is related to the stability of molecules. We adopted alchemical perturbation density functional theory (APDFT) as the quantum alchemy method. The electric dipole moment is less investigated in APDFT, and it is worth examining its accuracy. We have assessed the applicability of APDFT for estimating the electric dipole moment. |
63bbe84ba5c354823a8c1f54 | 2 | We describe the MDM algorithm introduced for general molecular properties. A conceptual illustration is shown in Figure (a). The present MDM searches for a functional molecule in the equilibrium geometry by optimizing the molecular species, as introduced in our previous MDM paper. However, the previous search algorithm is designed for energy-associated properties and is not necessarily appropriate for general properties. In this section, a common strategy of our MDM is first shown. Then, the algorithm of the present MDM algorithm is compared with that of Ref. 10. MDM designs the molecule with the desired property by optimizing a using the property gradient while suitably changing the equilibrium geometry. The calculations of and geometry optimization are performed using quantum alchemy during the design. The MDM algorithm introduced for general target properties is described in detail in the following. |
63bbe84ba5c354823a8c1f54 | 3 | These constraints are necessary to limit the value of p. In addition, the equilibrium geometry is imposed on the molecule during the exploration of the chemical space. This criterion for geometric stability is employed in the gradient-driven molecular construction approach. In the equilibrium geometry, the derivatives of energy with respect to the nuclear coordinates are zero: |
63bbe84ba5c354823a8c1f54 | 4 | This can be solved by using the (quasi-)Newton method with respect to R. For the full optimization problem consisting of Eqs. ( )-( ), we have two coupled optimization problems with respect to participation coefficients a and nuclear coordinates R. The former seeks a that maximizes the functionality p for a given fixed R. The latter searches for the equilibrium geometry with local energy minimization for a given a. |
63bbe84ba5c354823a8c1f54 | 5 | dp da change in the equilibrium geometry. Assuming the equilibrium geometry, can be expressed as (5) The first term on the right-hand side represents the gradient in the iso-geometric chemical space and contains the elements of . The second term takes into account the displacement of the equilibrium geometry. in the second term has the elements of . is the Hessian matrix of energy with respect to the nuclear coordinates. contains the elements of . While quantum alchemy can efficiently compute these derivatives as described later, the computational costs of and are still high. When p is the energy-related functionality, Eq. ( ) becomes , and their evaluations are avoided. For the other functionalities, must be evaluated. That is, the computational cost of MDM becomes significantly higher when targeting general properties. Therefore, in our previous study, MDM was only applied for optimizing energetic functionalities. For the design of general properties, the present MDM avoids This procedure is repeated until the variation in a becomes sufficiently small (Figure (b)). The full optimization problem is solved under that condition since the equilibrium geometry change is |
63bbe84ba5c354823a8c1f54 | 6 | caused by the update of a. When the full optimization problem is separated, the substantial update of a results in a large error due to the large equilibrium geometry change. Therefore, MDM gradually updates a in the continuous chemical space. It is expected that this also leads to the small displacement of the equilibrium geometry at each design step. Therefore, in practice, the geometry optimization that suitably changes R and approximately solves Eq. ( ) is not performed at all design steps. As the update of a depends on the current equilibrium geometry R, their coupling persists through the iterations. The LCAP discrete optimization performs geometry optimization of nearby candidate molecules in discrete chemical space and subsequently computes the gradient by finite difference. This strategy is costly in our MDM since all the candidate molecules are located at equal distances in the chemical space. Moreover, the error in the discrete update of a should be large. |
63bbe84ba5c354823a8c1f54 | 7 | a are updated using the optimality criteria method 26 used in topology optimization for macroscopic material design. The similarities between this type of molecular design and topology optimization have already been pointed out in the context of LCAP. The use of the optimality criteria method in MDM is expected to lead to rapid optimization convergence. To obtain the update formula, we introduce the Lagrangian function with respect to constraint Eq. ( ): (6) where λ is the Lagrange multiplier. The optimality condition with respect to a is expressed as (7) Based on this equation, we obtain the scale factors B for updating a: |
63bbe84ba5c354823a8c1f54 | 8 | Here, are positive. The optimality condition is satisfied when Bi = 1. Increasing ai is effective in making p smaller when Bi > 1. When Bi < 1, the opposite is true. Multiplying a by B, a are gradually updated while satisfying Eq. (3). A heuristic updating scheme is as follows: |
63bbe84ba5c354823a8c1f54 | 9 | Δa is a parameter for limiting each update. To gradually update the molecular species, Δa was set to 0.01 at each design step. λ is obtained by a bisection search to satisfy constraint Eq. ( ). The effectiveness of the algorithm comes from the fact that each ai is updated independently of the others, except for the scaling that must take place to satisfy Eq. (2). To obtain the candidate molecular species ( ) as the numerical procedure, optimized a are rounded off and converted into integers. From Eq. ( ), it is found that the candidate molecules are stationary points. Therefore, the candidate molecule can be obtained by MDM. When starting the exploration from the candidate molecule, a must be varied slightly. |
63bbe84ba5c354823a8c1f54 | 10 | of molecules from a combination of derivatives of the electron density of a reference molecule with respect to nuclear charges based on perturbation expansions. While the change in the number of electrons is allowed, this work considers isoelectronic chemical space. In the MDM procedure, and for all the candidate molecules with equivalent nuclear coordinates are calculated using APDFT. The accuracy of APDFT energy has been intensively investigated for the perturbation order and basis sets. In the iso-geometric chemical space, APDFT gives accurate predictions. An efficient analytical energy derivative has been proposed within the restricted Hartree-Fock theory. Because the accuracy of APDFT depends on the reference electronic structure theory, the direct derivative of the APDFT energy with respect to the nuclear coordinates was employed for MDM. The convergence of the perturbation expansion was confirmed for several atoms and molecules. Quantum alchemy has been widely applied to investigate catalysis, covalent bond energies, 18, 37 non-covalent interactions, mixtures, protonation and deprotonation energies, 33, 39-40 and chemical reactions. Although APDFT can calculate various response properties, to the best of our knowledge, few studies have applied APDFT to the prediction of non-energetic properties. In this work, we investigate the accuracy of APDFT for the electric dipole moment for molecular design. |
63bbe84ba5c354823a8c1f54 | 11 | This means that the efficiency and results of MDM depend on the chemical space. To examine this issue, the chemical space containing two candidate molecules was linearly interpolated, and we investigated the corresponding atomization energy and electric dipole strength (EDS) (Supporting Information). EDS is the norm of the electric dipole moment. The adopted chemical spaces are (N2, CO)CS, (BF, CO)CS, and (N2, BF)CS for the atomization energy and (N2, CO)CS for EDS, where |
63bbe84ba5c354823a8c1f54 | 12 | The energies of the candidate molecules and their derivatives with respect to the nuclear coordinates were calculated using APDFT with the second-order perturbation expansion (APDFT2). The reference electronic structure theory is coupled cluster singles and doubles (CCSD) or Kohn-Sham DFT with the PBE0 functional. We employed def2-TZVP as the basis set. All the molecules are assumed to be in the spin singlet state. APDFT/def2-TZVP accurately predicts energies and energy derivatives. The energy and equilibrium geometries at the adopted calculation level are reasonably accurate, as shown in our previous study. The electric dipole moment was calculated from the electron density derivatives for the APDFT2 energy according to the Hellmann-Feynman theorem (Supporting Information). Details of the atomization energy calculations are given in Ref. 10. MDM was performed using our original code. |
63bbe84ba5c354823a8c1f54 | 13 | The MDM results depend on the adopted quantum alchemy method. We adopted APDFT in this work. Here, we have investigated the accuracy of APDFT EDS with respect to (N2, CO)CS and (BN-doped benzene derivatives)CS. MDM for EDS will be demonstrated with these chemical spaces in the next section. APDFT is an approximation of the reference electronic structure theory. |
63bbe84ba5c354823a8c1f54 | 14 | The APDFT calculations of electronic EDS of CO were performed using reference N2 (Supporting Information). Although CCSD-based APDFT EDM for CO has already been calculated, we focus on the investigation of the dependence of EDS on the reference electronic structure theory. We adopted CCSD, Hartree-Fock, and PBE0 theories as the reference electronic structure theory. Regardless of the reference theory and perturbation expansion order, APDFT EDS is reasonably accurate (Table ). As the perturbation order of APDFT increases, the error from the reference theory decreases. In particular, the first-order perturbation expansion considerably improves EDS. |
63bbe84ba5c354823a8c1f54 | 15 | The calculation results for EDSs of BN-doped benzene derivatives 1-18 (Figure ) at the APDFT and PBE0 levels are shown in Figure . PBE0-based APDFT is used. The reference molecule is benzene 1. The lower-order APDFT up to the third-order perturbation expansion fails to reproduce the PBE0 trend (see also , S9, and S11). Moreover, the change in the reference molecule from 1 to 17 or 18 worsens the overall trend (Figure ). At the PBE0 level, the difference in EDSs between 5 and 16 is 1.07 |
63bbe84ba5c354823a8c1f54 | 16 | Debye. Accurate quantum alchemy predictions are required to search for more desirable molecules. EDS was calculated in the equilibrium geometries of the derivatives. EDSs in ascending EDS order can be found in Figure . We discuss possible improvements for precise molecular design. To improve the accuracy of APDFT, assuming convergence, the perturbation expansion order must be increased. Analytic derivatives of energy and electron density 33, 50-52 allow numerically stable yet computationally efficient evaluations of higher-order APDFT. Finite differences are also useful for systematically increasing the expansion order. A hybrid approach of high-and low-level reference theories for the derivatives of the electron density is effective for reducing the computational cost of higherorder APDFT calculations. A very recently developed alchemical integral transformation (AIT) transforms the integral in APDFT, allowing the calculation of the target system from only the electron density of the reference system. We believe that advances in quantum alchemy will allow more precise molecular design by MDM. |
63bbe84ba5c354823a8c1f54 | 17 | We apply MDM to search for functional molecules with equilibrium geometries in various chemical spaces. First, the search for a molecule with high atomization energy in simple (BF, CO)CS is performed to verify the performance of the optimality criteria method. The results are compared with those in Ref. 10. Second, we search for a molecule with high electric dipole strength (EDS) in simple (N2, CO)CS and realistic chemical space (BN-doped benzene derivatives BnNnC6-2nH6, n = 1-3)CS. This work focuses on the proof of concept and is limited to applications to small chemical spaces. Future work will address larger chemical space and candidate molecules with considerably different equilibrium geometries. For the APDFT reference molecules, we used N2, CO, and benzene for (N2, CO)CS, (BF, CO)CS, and (BN-doped benzene derivatives)CS, respectively. The reference electronic structure theory is CCSD for (N2, CO)CS and (BF, CO)CS and Kohn-Sham DFT with the PBE0 functional for (BN-doped benzene derivatives)CS. |
63bbe84ba5c354823a8c1f54 | 18 | We used MDM to search for a molecule with high atomization energy in (BF, CO)CS. The results are compared with our previous work 10 (Figure ). MDM started with BF and designed CO with higher atomization energy than BF. The atomization energy, participation coefficients representing the molecular species, and equilibrium bond length vary monotonically in the MDM procedure. The update of the molecular species accounts for the change in the equilibrium geometry in the optimization of the atomization energy (see Methods section). This design was also implemented in our previous study Ref. 10; the difference between this work and Ref. 10 is the algorithm for updating the molecular species. In Ref. 10, the constraints on the domain of definition for a (Eqs. ( ) and ( )) are removed by the variable transformation, and a are updated according to the steepest descent formula. However, the present MDM uses the optimality criteria method (Eq. ( )). Both the steepest descent and optimality criteria methods require a few parameters. In addition, the convergence conditions for are different. Nonetheless, we compared the design efficiencies. We obtained identical design results regardless of the optimization algorithm. On the other hand, there is an obvious difference in design efficiency. |
63bbe84ba5c354823a8c1f54 | 19 | A molecular design method (MDM) that explores chemical space accounting for geometric stability is demonstrated for general functionalities. We improved the optimization algorithm of MDM introduced in our previous work, which enables computationally efficient design targeting general properties. MDM simultaneously predicts the molecular species, desired property, and equilibrium geometry of functional molecules without experimental records. MDM is based on quantum alchemy and is effective for large chemical space. APDFT was adopted as a quantum alchemy method. The electric dipole strength (EDS) and atomization energy are set to target properties for molecular design as examples. Few studies have applied APDFT to the electric dipole moment. Therefore, we investigated the accuracy with (N2, CO)CS and (BN-doped benzene derivatives)CS prior to molecular design. The accurate APDFT EDS of CO was obtained with reference N2. In (BN-doped benzene derivatives)CS, we showed that APDFT up to the third-order perturbation expansion cannot correctly reproduce the EDS trend of derivatives at the reference PBE0 level of theory, regardless of the reference electronic structure theory, perturbation order, and basis set. Higher-order APDFT calculations based on analytical and numerical differentiation or integral transformation are expected to enable more accurate molecular design by MDM. We applied MDM to search for molecules with high EDS and atomization energy in the chemical spaces (BF, CO)CS, (N2, CO)CS, and (BN-doped benzene derivatives)CS. In (BF, CO)CS, BF with high atomization energy was obtained. This result was compared with Ref. 10. It was found that the update of the molecular species by the optimality criteria method leads to rapid convergence of the optimization with small computational cost. In (N2, CO)CS, CO with high EDS was designed. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.