id
stringlengths 24
24
| idx
int64 0
402
| paragraph
stringlengths 106
17.2k
|
|---|---|---|
6450f5f207c3f02937167e86 | 18 | In this direction, we have employed in silico approach to understanding the epigenetic modification by the process of lactylation. Autodock Vina was used to perform molecular docking experiments due to its better accuracy in predicting binding patterns, less run time, higher reproducibility, and its ability to powerfully search for potential energy surfaces . The molecular binding patterns of lactyl-CoA and acetyl-CoA against HAT p300 showed similar attributes in terms of binding energy and interacting amino acid residues (Table ). DSV3 was used after molecular docking to find the binding residues and details of polar bonds including hydrogen bonds. Lactyl-CoA binds through six strong conventional hydrogen bonds to the binding residues ARG1312, GLU1423, LYS1426, LYS1427, GLU1477, ARG1478) of HAT p300 protein (Figure and Figure , Table ). The binding affinity (-9.6 kcal/mol) and specificity of lactyl-CoA within the active site of HAT p300 shared almost identical binding attributes such as docking energy (-10.3 kcal/mol) and key residues (ARG1305, ASP1306, ARG1312, GLU1416, GLU1423, GLU1477, ARG1478) responsible for hydrogen bonds for a stable acetyl-CoA and HAT p300 enzyme complex (Figure and Figure ). Molecular interactions suggested that lactyl-CoA may display equivalent binding affinity that occupies the similar binding pocket of acetyl-CoA in the substrate binding site of the HAT p300 enzyme. |
6450f5f207c3f02937167e86 | 19 | Cancer cells within the tumor microenvironment achieve metabolic reprogramming by concerted contributions from cellular and non-cellular factors . Indeed, the requirements of cancer cells are met through various metabolic networking including glucose metabolism and distinctive metabolic products including lactate . Given Warburg's effects on cancer cells, the production of lactate is suggested as a waste product. Currently, there is an emergence of understanding that cancer cells may export metabolic waste lactate to fuel the growth and metastasis by supporting various intracellular metabolic and non-metabolic epigenetic regulations of cancer-supporting cells such as macrophages within the tumor microenvironment . |
6450f5f207c3f02937167e86 | 20 | Epigenetic alterations such as methylation, acetylation, succinylation, and newly included lactylation of histones and other protein targets are known to alter the pro-tumor attributes of cancer and cancer-supporting cells including macrophages . Furthermore, accumulating shreds of evidence have shown the existence of an axis between epigenetic changes and metabolic adaptations . At the same time, however, the key insights on the nature of substrate, the biological abundance of a substrate, and associated enzymes are missing. |
6450f5f207c3f02937167e86 | 21 | Here, our data suggested a possible role of acetyl-CoA synthetase as a key enzyme that allows the binding of lactate to the same binding site as acetate. Interestingly, active site binding amino acid residues such as TRP413, TRP414, GLN415, ARG515, ASN521 are earlier reported in case of acetyl-CoA synthetase as in case of lactate vs. acetate . These findings hinted at the potential uses of acetyl-CoA synthetase by cancer cells to generate lactyl-CoA for the proposed lactylation process because of metabolic reprogramming. Then non-cancer cells including microbiotas with suitable enzymes may convert lactate to lactyl-CoA and then lactyl-CoA is released into the tumor microenvironment . The role of acetyl-CoA synthetase is depicted in the activation of macrophages mediated by pro-inflammatory bacteria (28-30). |
6450f5f207c3f02937167e86 | 22 | In recent, modulation of epigenetic writer enzymes such as HATs are implicated in the inflammatory landscape of tumor microenvironment specifically in the context of polarization of M1 macrophage to M2 macrophage . Among various HATs, the enzymatic role of HAT p300 to alter the acetylation mark on histone and other tumor suppressors is highlighted But the relevance of HAT p300 in contributing lactylation marks on histone and other proteins using lactyl-CoA as a substrate remains obscure. The possibility of metabolic strategies by cancer cells to use lactylation as a part of mitigating acetylation marks on histone proteins could be explored. This would be almost similar to shunting off HATs by cancer and cancer-associated cells as a part of the activation of oncoproteins during chromatin remodeling to achieve protumor attributes. In recent, non-histone proteins such as PKM2 and beta-catenin are suggested as protein targets that modulate the macrophage phenotype transition in cancer and other human disease conditions . The relevance of HAT p300 in modulating the transcriptional landscape of cancer cells and tumor-associated immune cells such as macrophages is considered as a link between the metabolic-epigenomic axis that may use metabolites such as acetyl-CoA and lactyl-CoA In a direction to the unresolved question of the nature of substrate and enzymes in lactylation, there is a hint of the potential enzymatic role by HAT p300 . Among various HATs, HAT p300 is suggested to modulate the transcription of genes that are linked to the development of various human diseases including cancer and immune-related diseases that involve macrophages in their tissue environment . HAT p300 is determined to comprise several domains including HAT domain (1285-1664) amino acid residues . HAT p300 catalytic domain is antagonized by various small molecules including A-485, I-CBP112, natural products, and bi-substrate analogs (Lys-CoA) . However, binding affinity and position by lactyl-CoA to p300 HAT is not known and may be potentially linked with the lactylation process. Although the active site on HAT p300 is reported in earlier works by showing key amino acid residues such as PHE1374, LEU1398, SER1400, ARG1410, THR1411, TYR1414, HIS1415, ARG1312, GLU1423, LYS1426, LYS1427, GLU1477, and ARG1478 . |
6450f5f207c3f02937167e86 | 23 | Importantly, molecular docking data on specific and similar binding by both lactyl-CoA and acetyl-CoA indicated a strong binding to the reported amino acid residues such as ARG1312, GLU1423, LYS1426, LYS1427, GLU1477, and ARG1478. These amino acid residues interact with several known natural and synthetic substrates including acetyl-CoA, lysyl-CoA, A-485, and I-CBP112. It is interesting to note that lactate did not have any binding affinity within the active site of HAT p300 |
6450f5f207c3f02937167e86 | 24 | The above observations are in coherence with a recent finding that the lactylation process modulates the transcriptional gene regulation in M1 macrophage and allows it to change into M2 macrophage . Zhang et al. 12 discovered the process of lactylation with experimental evidence at molecular and cellular levels. However, pertinent questions were not answered on the nature of the enzyme and biological source and relevance of lactyl-CoA that may potentially act as a substrate for a potential enzyme such as HAT p300. Our data is the first and novel proposition on the biological possibilities of lactyl-CoA within the tumor microenvironment and the mode of lactylation mediated by the HAT p300 enzyme. |
6450f5f207c3f02937167e86 | 25 | Since both lactyl-CoA and acetyl-CoA bind to the same active HAT domain of HAT p300, it would be interesting to evaluate whether lactylation epigenetic marks may inhibit acetylation marks on chromatin that could drive certain cells such as macrophages toward polarization from anti-tumor M1-macrophage to pro-tumor M2 macrophage in the tumor microenvironment. |
6450f5f207c3f02937167e86 | 26 | In addition, the authors make a proposition that the availability of lactyl-CoA within the tumor microenvironment may be linked to the nature of microbiotas that are equipped with an enzyme that may convert lactate into lactyl-CoA. Hence shuttling of lactate and lactyl-CoA is proposed between cancer cells, microbiotas, and immune cells such as macrophages within the tumor microenvironment. |
6450f5f207c3f02937167e86 | 27 | The impact of these findings will have a significant contribution to solving the unanswered questions on the molecular mechanisms of lactylation in the context of the tumor microenvironment. Herein, the authors propose a future model in that lactate is shuttled into the tumor niche harboring microbiotas during external agents-mediated cell death to cancer cells. |
6450f5f207c3f02937167e86 | 28 | In this way, microbiotas are equipped with the metabolic machinery including lactylcoenzyme A and acetyl-CoA enzyme to generate lactyl-CoA from lactate . In the future, research attempts would be challenging to understand and evaluate the shuttling of lactate and lactyl-CoA among cancer cells, cancer-supporting macrophages, and microbiotas that could contribute to the pro-tumor microenvironment. |
6450f5f207c3f02937167e86 | 29 | A very recent paper revealed the mechanism of lysine lactylation in E. coli and it would be fascinating to see the effects of microbiotas with the machinery of lactylation enzymes such as YiaC, CobB, and YdiF that can use extracellular lactate in the tumor microenvironment and providing signaling molecules to cancer cells and cancerassociated immune such as macrophages as pro-tumor phenotype. |
6450f5f207c3f02937167e86 | 30 | Additionally, lactyl-CoA was not detected in the extracellular compartment of breast cancer cells. Molecular docking and MD simulations predicted that lactate could serve as a substrate for acetyl-CoA synthetase besides its known substrate acetate. These projections helped to hypothesize that acetyl-CoA could have the potential to convert lactate into lactyl-CoA in the tumor microenvironment and the ensuing lactylation-mediated epigenetic process. Furthermore, molecular interaction studies helped to propose that HAT p300 may serve as a potential enzyme that can use lactyl-CoA to transfer the lactate group for the lactylation of histones and other target proteins. The nature of the data is based on in vitro detection of lactate and molecular docking and MD simulations, and these findings are novel and could be a meaningful incremental step to resolve gaps in the lactylation and tumor microenvironment. |
64382c5b1d262d40ea65d444 | 0 | The leaching efficiency of ion adsorption type rare earths is usually leached by 2wt% ammonium sulfate solution with 100% rare earth content by full drenching, and the amount of rare earths leached by other leaching agents compared to it, which is the leaching efficiency . The leaching efficiency is affected by the type of leaching agent, concentration, pH, leaching method, type of ore sample, solid-liquid ratio, etc. As far as leaching agents are concerned, different types of leaching agents leach different efficiencies under the same conditions. Sun Yuanyuan reported that the leaching efficiencies of different electrolytes were different at a cation concentration of 0.128 mol/L. The leaching efficiencies were in the order of (NH4)2SO4 > NH4Cl > K2SO4 > KNO3 > KCl > Na2SO4 > NaCl > NaNO3, and Yang Lifen reported that at an equivalent concentration of 0.128 N, the leaching efficiencies of the following leaching agents in the following order: Al2(SO4)3 > (NH4)2SO4 > AlCl3 > Al(NO3)3 > NH4NO3 > NH4Cl > MgSO4 > Mg(NO3)2 > CaCl2 > Ca(NO3)2 > MgCl2, and Xiao Yanfei et al reported that at a leaching agent concentration of 0.1 mol/L pH = 2, the leaching efficiency of Fe2(SO4)3 > FeSO4 > MgSO4 > ZnSO4 > Fe(NO3)2 > FeCl2 > MnSO4 > Mg(NO3)2 > MgCl2 > K2SO4 > (NH4)2SO4. It can be seen that the order of leaching efficiency is not the same when different treatments are adopted. Different concentrations of the same leaching agent leaching effect also varies, generally speaking the greater the concentration of leaching agent, the higher the leaching efficiency, but the high concentration of magnesium sulfate (5wt%) leaching effect is better than the same concentration of ammonium sulfate, while aluminum sulfate at a lower concentration has a good leaching efficiency. |
64382c5b1d262d40ea65d444 | 1 | Leaching method also has an impact on the leaching effect, the current common leaching methods are: static leaching, heap leaching, balanced leaching, column drenching. The leaching efficiency is generally: column drenching > equilibrium leaching > static leaching. This is due to the on-column drenching, leaching agent solution under the action of its own gravity, top-down contact with the solid phase several times for ion exchange, the process is a dynamic equilibrium, exchange after the separation of products, exchange more fully. On-column drenching usually uses a sample volume ranging from a few tens of grams to several kilograms, depending on the size of the filled column. On-column drenching is influenced by the degree of filling uniformity, the granularity of the mineral sample, the degree of filling compaction, the height of the filled column, etc. The degree of filling uniformity determines whether the leaching agent has gully flow, bias flow, etc., thus making some ore samples over-leached and some unleached. Tian Jun et al. reported that the ore particle size has a great influence on the leaching rate and the leaching rate of rare earths decreases with the increase of particle size. Large granularity has more pores, fast leaching agent flow rate, and insufficient leaching agent exchange; small granularity, sufficient leaching agent exchange, but slow flow rate, the existence of re-adsorption phenomenon, and serious tail dragging. The loading column height is similar to the column efficiency theory of chromatographic column, the column height determines the theoretical tower plate number, i.e., the number of cycles of adsorption-desorption, but the column drenching is significantly different from the larger the column height of chromatographic column when the injection volume is certain. When the solid-liquid ratio is fixed, the effective exchange of solutions is limited, and the first part of the solution can effectively participate in the exchange, after the solution exchange is limited, and the increase of the filling column is limited. The compaction of the filling and the particle size of the sample affect the height and porosity of the filling column, thus affecting the flow rate or the height of the leaching peak and the sequence of the peak time. Equilibrium leaching uses equilibrium oscillation to exchange to reach equilibrium, and is usually used to study the thermodynamic model of ion-adsorption rare earths, adsorption model, etc. The amount of ore samples used is within a few tens of grams, and most of them are screened clay minerals, and the control of ore sample type, temperature, leaching time, etc. is simple and easy to study. The equilibrium leaching method to study the efficiency of leaching needs to be further standardized. |
64382c5b1d262d40ea65d444 | 2 | In this paper, potentiometric titration is used to study the leaching performance of ion-adsorbed rare earths, which is different from the common oscillatory equilibrium leaching. Unlike acid-base potentiometric titration which has obvious mutation points, the potentiometric titration of leaching agent has no obvious mutation points except the initial point and the appearance of the color remains unchanged, so it is impossible to judge the end point of titration directly. The change of Zeta potential reflects the change of internal charge of clay minerals slip surface, which is the result of the joint action of anion and cation. The clay minerals generally have a negative skeleton charge, when the clay minerals adsorb more cations, the more negative charges are offset, and the negative value of Zeta potential decreases. When the rare earth content changes, there must be ions involved in the exchange, and the zeta potential generally changes. For this reason, by determining the content and concentration of the leaching agent added drop by drop and the leaching amount of rare earth ions, the correspondence will be able to evaluate the leaching performance of the leaching agent at different concentrations. In addition to the determination of the rare earth content, the ion-adsorbed rare earth minerals were titrated by potentiometric dip leaching to monitor the changes in zeta potential, pH and conductivity during the leaching process and to provide a basis for the interpretation of the ion exchange process. drenching pH Abrasion pH: Take 20.00 g of the screened mineral sample below 20 mesh and deionized water mixed in the ratio of 1:4 by mass, put it in a 250 ml beaker, put it on a multi-head magnetic stirrer and stirred for 30 min, set the speed to 1000 r/s, and then filtered, take the clarified filtrate, and determine its pH value as the abrasion pH value. |
64382c5b1d262d40ea65d444 | 3 | Exchange pH value: take 20.00g of screened ore sample below 20 mesh and 2wt% ammonium sulfate mixed according to mass ratio 1:4, put it in 250ml beaker, put it on multi-head magnetic stirrer and stirred, set the speed to 1000 r/s, stirred for 30 min, then filtered, take the clarified filtrate, measure its pH value is the exchange pH value with 2wt% ammonium sulfate. The exchange pH of different mineral samples and different components of the same mineral sample with different types and concentrations of leaching agents are different. |
64382c5b1d262d40ea65d444 | 4 | Leaching pH on the column: 300 g of ore samples below 20 mesh were uniformly packed into a Φ×H=60 mm×400 mm sand-core quartz column and 300 ml of 2 wt% (NH4)2SO4 solution was added at a solid-liquid ratio of 1:1. Each 10 ml of leachate was collected and the pH value of each leachate was measured. Experimental plots are shown in Figure 1 The clay samples were accurately weighed and mixed with 250 ml of deionized water, equilibrated in a constant temperature shaker at 25 ℃ for 30 min, and the Zeta potential during the leaching process was measured on an A80045-Zeta Probe at an adjusted speed of 200 rpm, using a certain concentration of leaching agent added drop by drop to the samples, and the Zeta potential, pH and conductivity were measured every 0.5 ml. The Zeta potential, pH and conductivity were measured every 0.5 ml, and for every 2 ml of leaching agent, 2 ml of sample was taken to determine the rare earth content, where the amount of leaching agent added drop by drop and the volume change were known, and the leaching agent concentration corresponding to the recorded data could be clearly calculated. When the drip addition of leaching agent was stopped, the tailings were filtered and leached with 250 ml of deionized water to determine their zeta potential during water leaching. During the determination of Zeta potential, the corresponding leaching and water leaching pH values were measured, and the equilibrium leaching filtrate pH was determined. Abrasion pH measures the pH of clay minerals in pure water, which reflects the hydration characteristics of the clay particles. The clay particles are negatively charged, and the cations inside the slip surface can only partially counteract the negative charge, and cations that maintain electrical neutrality will be present in the diffuse layer of the clay minerals. In pure water, the whole clay minerals resemble the hydrolysis of strong acid and weak base salts, the cations in the diffusion layer enter the solution, and the overall equivalent of a negatively charged group inside the slip surface, the group can combine with H + in the water, and the OH -in the solution increases, making the abrasion pH generally larger. Exchange pH measures the pH of clay minerals in the leaching and exchange process. The change of exchange pH is influenced by the hydration of ions and groups in the exchange process, but is mainly determined by the anion and cation of the leaching agent. |
64382c5b1d262d40ea65d444 | 5 | From Figure , the abrasion pH and exchange pH of the mine samples from Longnan, Anyuan and Dingnan can be seen that the abrasion pH of the mine samples from all three regions is higher than the exchange pH, which is due to the combination of negatively charged clay particles with H + in water, which makes the solution OH -increase and the abrasion pH rise. The higher the abrasion pH, the stronger the hydration process of clay minerals, the more sparse the structure of the original ore in aqueous solution, and the more likely to cause soil erosion, and the loss of fine particles makes the surviving ore bodies generally exist in the form of biased large particles, which is consistent with the large granularity and less fine particles in An Yun area. The larger pH value of Dingnan exchange is due to the fact that the same ore sample with less than 20 mesh, There are more fine particles of clay distribution in the Dingnan ore sample, and the clay mineral skeleton is negatively charged and easy to combine with H + . Under the same conditions of leaching agent solution, there are more fine particles of clay distribution and more combined H + , thus the exchange pH value of Dingnan is larger. The difference is that equilibrium leaching is a single equilibrium showing the final result of exchange, while column drenching is multiple dynamic equilibrium, which can reflect the changes of the whole exchange process on the column. From Fig. , the relationship between the content of rare earth ions and the concentration of leaching agent, it can be seen that the content of leached rare earth ions is the most in Anyuan ore sample, followed by Dingnan, and the lowest in Longnan, where the amount of rare earth leached from Anyuan ore sample is about 2.5 times that of Longnan ore sample, which is different from the column leaching results. The content of the ore samples below 800 mesh was less than that of the generally larger granularity of the An Yuan ore samples. Combined with Fig. , it can be seen that the leaching of rare earth ions existed in the process of dramatic change of pH and slow change of pH in the solution, which means that H + and OH -in the solution were involved in the ion exchange between clay minerals and leaching agent at the beginning, and the pH of the solution with leaching agent did not change much during the ion exchange process afterwards, and the balance of H + and OH -was basically stable. |
64382c5b1d262d40ea65d444 | 6 | Combined with Fig. , the relationship between leaching agent concentration and conductivity changes shows that the conductivity is the largest in Dinan, the second largest in Anyuan and the smallest in From Fig. Zeta potential versus rare earth ion content variation graph, it can be seen that during the titration of 0.15 mol/L MgSO4 leaching agent, its Zeta potential is linearly related to the leaching amount of rare earth ions, the larger the leaching amount of rare earth ions, the larger the absolute value of Zeta potential and the more negative the Zeta potential. The Zeta potential corresponds to the slip surface potential, where the rare earth ions are exchanged out of the slip surface by the leaching agent, and the positive charge inside the slip surface decreases and the negative charge increases due to the lower valence of the exchanged ions, the presence of anions or poor bonding and rediffusion, etc. The linear relationship between the Zeta potential and the rare earth ion content indicates that the overall Zeta potential is negative. The linear relationship between Zeta potential and ion content indicates that the internal structural changes of the slip surface of clay minerals caused by the leaching of rare-earth ions are the main reason for the change of Zeta potential throughout the exchange process. |
64382c5b1d262d40ea65d444 | 7 | By quantitatively adding the leaching agent to the clay minerals dropwise and continuously measuring their zeta potential, pH and conductivity as well as taking samples to determine the rare earth content in the solution, it was found that the variation of zeta potential was related to the leaching content of rare earth ions and was linearly correlated, and the more rare earth ions were leached, the more negative zeta potential was. The relationships between the changes of Zeta potential, pH and conductivity of the three leaching agents, ammonium sulfate, magnesium sulfate and aluminum sulfate, |
641cda7c62fecd2a835588e8 | 0 | Boron-rich solids are compounds with unusual structures. Almost all of them contain B 12 icosahedral units and various interstitial atoms (ranging from nonmetals to metals) . Such structures make boron-rich compounds extremely stable, resulting in high (up to 2700 K) melting temperatures, chemical inertness and excellent mechanical properties . These properties of icosahedral boron-rich solids suggest that they will find many new applications. The extensive study of existing materials and the search for new boron-rich compounds are therefore of great importance and receive considerable attention in experimental and theoretical science. |
641cda7c62fecd2a835588e8 | 1 | Recently, two new boron-rich chalcogenides, B 6 Se and B 6 S, have been synthesized at high pressures and high temperatures . The structure of materials produced under pressure is of great interest because it usually contains defects, which, in turn, affect the properties of the resulting material. The deformation that occurs during the pressure treatment introduces various defects into the material: dislocations, stacking faults, twins, etc. Localized deformation in the form of shear bands and areas of strong local flow leads to structural heterogeneity. Due to the presence of defects, the shear strength of real crystals is significantly reduced compared to an ideal crystal. The defects affect the electrical, diffusion, optical, emission, photoelectric, magnetic and other properties of the material . The aim of the present work was to investigate the real structure of new boron-rich chalcogenides, B 6 Se and B 6 S, using high-resolution transmission electron microscopy. |
641cda7c62fecd2a835588e8 | 2 | According to TEM data, both B 6 S and B 6 Se powders consist of flat polyhedral particles with dimensions ranging from 20 to several hundred nanometers (Fig. ), a significant proportion of which contain twins with coherent twin boundaries. Fig. shows a fragment of the B 6 S structure with ABAB stacking. In contrast to Fig. , where such a structure was formed during twinning as a result of the appearance of individual stacking faults, here the ABAB stacking is not presented fragmentarily, but throughout the entire section. As we can see, the ABAB structure is formed in two ways: as a layer within the initial phase with ABCDABCD stacking as a result of the appearance of the stacking fault, or it can be formed as an independent structure in a large fragment of the particle. Fig. shows a deformation band in B 6 S along the plane (011) (indicated by the arrow). Similar deformation bands are found in many materials after various types of deformation. Phase transformations, recrystallization, and transitions from a crystalline state to a disordered and/or amorphous state can occur in the deformation (or shear) bands. Shear bands are widely observed in all types of materials: metals, polycrystalline alloys, metallic glasses, rocks, ceramics, organic compounds, polymers and fullerenes. The band in Fig. shows amorphous fragments formed as a result of disorder in the crystalline lattice. Shear bands represent regions where inelastic shear deformation is significantly localized and exceeds the deformation in the surrounding material. They represent a mode of nonelastic response of the material to mechanical stress . Similar shear bands were previously observed in B 4 C , whose crystal lattice is formed by distorted B 11 C icosahedra connected by C-B-C three-atom chains. As in the present work, the cracks and amorphous bands in B 4 C are located along the same plane. The formation of amorphous bands during shear deformation has been explained as a result of the fracture of icosahedra . The formation of cracks is evidence of cleavage, a property of a crystal to split along certain crystallographic directions due to the structural peculiarities of its crystal lattice. The cleavage planes and the crystal planes characterized by the highest density of atoms usually coincide. The direction of crack propagation is assumed to be almost strictly perpendicular to the direction of load application. Crack propagation occurs by successive and recurrent breaking of atomic bonds along specific crystallographic planes. As can be seen from the corresponding FFT images, all the cracks are in the {011} planes, and while in B 6 S there are many cracks and they all correspond to (011) planes, in B 6 Se there are cracks that belong to both (011) and (0-11) planes. Thus, both shear bands and cracks present in B 6 S and B 6 Se lie in {011} planes. Fig. shows the twinning structure in B 6 Se. The twinning plane is the same as in B 6 S, i.e. (101). |
641cda7c62fecd2a835588e8 | 3 | The inset shows the corresponding diffraction pattern. Only one reflection system is indexed. and layer D as D 1 . The FFT image in Fig. shows additional reflections that cannot be labeled in the coordinates of the original lattice; the corresponding reflections are indicated by circles. As can be seen, this structure is characterized by the appearance of an additional reflection labeled (002). |
641cda7c62fecd2a835588e8 | 4 | Similar doubled structures have been found in B 12 P 2 and in B 6 O , and have been named ʻnanotwinsʼ. (Fig. ) and ABA stacking (Fig. ). The AB 1 CD 1 AB 1 CD 1 structure is the result of shifting the even planes B and D in opposite <110> directions in the a 1 ,b 1 ,c 1 coordinates. The FFT image corresponding to the AB 1 CD 1 AB 1 CD 1 structure (Fig. ) is identified in the a 1 ,b 1 ,c 1 coordinates. The ABAB structure shown in Fig. corresponds to a double-spaced orthorhombic lattice with unit cell parameters: a = 1.005 nm, b = 0.529 nm, c = 0.948 nm for B 6 S and a = 1.025 nm, b = 0.534 nm, c=0.968 nm for B 6 Se (Fig. ). |
641cda7c62fecd2a835588e8 | 5 | Thus, two types of polytypes were found: ABAB in both chalcogenides and AB 1 CD 1 AB 1 CD 1 in B 6 Se. Polytypism is defined as the ability of a compound to crystallize in several structural modifications, each of which can be considered as built up by stacking layers of (almost) identical structure; the modifications differing only in their stacking sequence. Polytypes have two common unit cell parameters, and the third one is a variable integer multiple of the same value equal to the spacing between the nearest neighboring layers . In this context, polymorphism is sometimes called one-dimensional polymorphism. |
641cda7c62fecd2a835588e8 | 6 | Structures with AB 1 CD 1 AB 1 CD 1 stacking in B 6 Se were often observed when the fragment under study contained two twin systems. It is known that when twins in two or more planes appeared in the crystal, they come into contact as the twin layers grow and then pass through each other. This can lead to two effects: one twin system absorbs another or a new phase is formed . As has been shown previously, the layers obtained by twinning can form a new structure . In fact, the twinning mechanism is related to the deformation of the unit cell under an external force, which leads to a change in the orientation of different parts of the crystal. In reality, the evolution of the deformation manifests itself as the appearance and successive propagation of layers of the twin system in the original crystal. For this reason, twinning can also be considered a structure forming process. Crossing of twinned interlayers leads to strengthening of crystalline solids. The twinned interlayers that form under mechanical stresses increase the resistance of crystals to plastic deformation . |
641cda7c62fecd2a835588e8 | 7 | Figs. and show the twins in B 6 S and B 6 Se, respectively. These twins are different from those shown in Figs. and. It is not possible to identify these structures in the a,b,c axes, but it is possible in the a 1 ,b 1 ,c 1 axes. The matrix and twin reflections shown in the insets are next to each other. In Fig. , arrows indicate the reflections corresponding to one of the crystal lattices. In Fig. , the reflections of one of the lattices are indicated by circles. In both cases, the twinning planes in the a 1 ,b 1 ,c 1 axes are {202}, which correspond to (001) planes in the a,b,c axes. Thus, it is possible to state that the twinning in B 6 S and B 6 Se can be realized in two planes: (101) and (001) in the a,b,c axes or, correspondingly, (001) and (202) in the a 1 ,b 1 ,c 1 axes. The existence of two twinning planes in the materials studied is not surprising. For example, many metals with hcp structure have two twinning planes each. There are materials where the number of twinning planes can reach four . |
641cda7c62fecd2a835588e8 | 8 | The appearance of twins and polytypes during high-pressure synthesis has been observed in several systems. It bas been found that even small differences in the p-T conditions or the degree of approach to equilibrium can lead to large differences in the periodicity of the resulting crystal(s) . For example, high-pressure synthesis of boron suboxide B 6 O results in the formation of two polytypes: α-B 6 O with ABCABC stacking of B 12 icosahedra and β-B 6 O with ABAB stacking of B 12 icosahedra ; and twinning of α-B 6 O can produce local β-B 6 O stackings. Thus, the polytypes of all boron-rich chalcogenides B 6 X (X = O, S, Se) are characterized by different stacking of B 12 icosahedra, making them cardinally different from the ʻpolytypesʼ of boron carbide based on varying carbon content and site-specific placement of carbon atoms in the icosahedra . |
641cda7c62fecd2a835588e8 | 9 | High-resolution transmission electron microscopy study of new orthorhombic boron-rich chalcogenides, B 6 S and B 6 Se, allowed us to identify microstructural features, typical for boron-rich solids synthesized at high pressures and high temperatures. It was found that twinning in both materials is realized in two planes: (101) and (002). Along with numerous twins on the (101) plane, two types of polytypes were found: ABAB stacking of B 12 icosahedra in both chalcogenides and AB 1 CD 1 AB 1 CD 1 stacking of B 12 icosahedra in B 6 Se. Shear bands and cracks have been found in {011} planes. |
6659978c418a5379b0bd75f1 | 0 | Historically, computing education in chemistry has gone through cycles of popularity. When desktop computers first became available in the 1980s, the adoption of teaching programming was common across the general chemical curriculum but was removed in the late 1990s. Since then, computing has been the primary tool of theoretical or computational chemists, focused on atomistic and electronic modelling. As a result, when undergraduate chemists are trained, the digital skills they are exposed to are often limited to this area. However, with growth areas such as machine learning, automation, and data science becoming increasingly present in chemical research , a higher level of digital literacy is becoming a necessity for many chemists. |
6659978c418a5379b0bd75f1 | 1 | The growing need for computational literacy in chemistry goes beyond the research community. Along with an international shift to a more digital economy, employers have a greater focus on the ability to understand and manipulate data at scale . Therefore, chemistry graduates' training must reflect this changing environment to keep them competitive. Chemists need to be equipped with logical thinking and data analysis skills, which are often taught in physics, engineering, and biology degrees . Improving digital skills training will retain the stock value of a chemistry graduate in the job market and equip chemical domain experts with the needed computational literacy for the digital future. |
6659978c418a5379b0bd75f1 | 2 | A recent comment in this journal's sibling publication argued computing skills could pave the way for a renewal of undergraduate physics education . We believe that the same is true for chemistry. For example, as is the case in physics, there have been substantial developments in pedagogical practice in chemistry education, including using digital technology, promoting active learning, blended learning environments, and problem-solving on authentic tasks . |
6659978c418a5379b0bd75f1 | 3 | Chemical industry and research are facing a phase shift, with a substantial drive towards automation and digitisation . Specialist skills like programming, statistics, data handling, and visualisation will soon become indispensable. New skills relating to artificial intelligence, such as prompt engineering, could quickly be added to the list. This shift poses a need for graduates with a digital skillset and the ability to apply it to their specialist domain. |
6659978c418a5379b0bd75f1 | 4 | We wish to insist that these specialist domains do not end at computational or physical chemistry. The increased automation of experimental chemistry means that the demand for "digital chemists" in all strands of chemistry will continue growing, and the employability of data-illiterate chemists will come under threat. Universities need to pre-empt this change for the sake of their graduates. |
6659978c418a5379b0bd75f1 | 5 | Accordingly, many institutions already include more computing and programming, often using Python , into their degree programs. There has been a growth in undergraduate modules, such as the "Data-driven Chemistry" unit from Edinburgh , and postgraduate taught programs in digital chemistry or similar areas across many institutions in Europe and the US. |
6659978c418a5379b0bd75f1 | 6 | Being able to count on digitally competent students unlocks opportunities for teaching chemistry . Naturally, programming allows students to surpass the "blackbox" approach to computational chemistry by giving them hands-on access to source code . It also allows students to directly interact with modern instrumentation and automated data acquisition as can be done in LabVIEW or Python interfacing with an Arduino or Raspberry Pi . The Jupyter Book Scientific Computing for Chemists with Python contains many examples of chemical applications to programming, such as producing calibration curves, NMR peak finding, and microscopic image analysis. However, digital skills go beyond just programming, and it is important that, as educators, we do not shy away from details of statistics in data analysis or the problems of non-standard file formats in working with experimental equipment . A data-trained student could use Jupyter Notebooks or Electronic Laboratory Notebooks in analytical or inorganic chemistry laboratory exercises for data analysis and visualisation . Traditional lab techniques may be complemented by exercises in automation . |
6659978c418a5379b0bd75f1 | 7 | However, adopting digital skills a few modules at a time poses challenges. Students struggle to remain motivated in learning skills unless proven relevant to their degree as a whole . Furthermore, digital skills are seldom used in pre-university chemistry or advertised as integral to chemistry degrees. As such, students may choose a degree in chemistry, hoping to avoid a more mathematical or computational science. This anxiety can limit the rigour with which important computing concepts are taught . |
6659978c418a5379b0bd75f1 | 8 | As mentioned above, low-level integration is the most effective way to introduce digital skills to the chemistry curriculum. By this, we mean introducing coding, data handling, and analysis skills early in the degree program and continuing to use these skills in applied settings throughout the degree. This approach ensures that students see the relevance of their learning, which is essential for interdisciplinary curricula. It allows universities to advertise their chemistry degree as integrally digital, changing applicants' predispositions to digital skills. Finally, it protects graduates from obsolescence due to automation, whatever area of chemistry they are interested in. |
6659978c418a5379b0bd75f1 | 9 | One difficulty in integrating digital skills at all levels of chemistry education is the requirement for equivalent levels of computational literacy in academic staff . The rate of technology change alongside exceptionally high staff workloads makes it challenging to stay at the technological forefront. Despite this, there are encouraging but anecdotal accounts of those new to programming learning Python to improve their teaching practice . |
6659978c418a5379b0bd75f1 | 10 | Recently, there has been a movement in the United such as Software Carpentry, the Software Sustainability Institute (SSI), Centre Européen de Calcul Atomique et Moléculaire (CECAM) and the Physical Sciences Data Institute (PDSI). Furthermore, the "Python in Chemistry" group aims to facilitate this with "Train the Trainers" material to help existing staff become comfortable and familiar with coding. Hopefully, this project to build a network of digital chemistry educators will have international appeal and welcome contributions from all. The next generation of chemists needs digital skills to remain competitive in academia and industry. This is driven by the growth of computation, data science, and machine learning across the chemical sciences. Chemistry training must evolve to provide this training holistically, and we hope to build an international network of academics interested in training chemists with these digital skills. |
62f3b3f38dba68454c1e480d | 0 | Machine learning applied to chemistry is a growing field of research. Many factors, such as advancements in graphics processing units, larger dataset collections and new algorithms have contributed to this renaissance. Baum et al. observe that this growth is not uniform and postulate the reason fields such as analytical chemistry have seen faster developments compared to others like organic synthesis is due to the availability of large training datasets in areas which are traditionally more data intensive. For example, in analytical chemistry, machine learning algorithms have been used to find chemical species at concentrations below the usual limit of detection by finding hidden patterns of signals within the noise. However, considerable progress has been made in applying machine learning to organic chemistry. For example, machine learning techniques have been used effectively in computer aided synthesis planning by training on reaction data from Reaxys or the United States Patent and Trademark Office dataset. Sensor data such as in-situ temperature and pH can be a good source for machine learning algorithms in time series modelling. Interest in sensor usage has grown with the development of the internet of things (IoT) -which is the exchange of data between internet-connected devices, and can be extended to include chemistry equipment and chemically relevant data. The concepts of cloud chemistry or telechemistry have been introduced and involve remotely monitoring reactions by uploading the results from analytical equipment to the cloud. IoT is part of the wider concept of industry 4.0, which has received attention in recent years and refers to the current trends of interconnectivity, data, and automation. The application of machine learning techniques to real-time data, including sensor data, for predictive maintenance is an example of industry 4.0 practice. Within process chemistry, these principles have been applied to enable predictive maintenance for pilot plants and selfoptimisation of reactions. The use of sensors and other in-line methods for process monitoring was part of the vision for process analytical techniques set out by the United States Food and Drug Administration in 2004. These methods are typically non-destructive and real-time, offering advantages over traditional sampling methods. The use of sensors in organic chemistry is an emerging area, fuelled by advances in flow and automated synthesis. Using sensors inside organic chemistry reactions generates data which could be valuable in the development of machine learning tools to augment the synthesis process, ultimately helping the chemist. In the field of reaction kinetics, hybrid models, which combine machine learning with traditional modelling methods, have found success at predicting the chemo-and regioselectivity of substitution reactions. Sensors in chemistry can be used to monitor the reaction or to control the processes involved in performing the reaction. Mettler Toledo's ReactIR™ is used for monitoring reactions in real time using infrared (IR) spectroscopy, while FlowIR™ is an adaptation of this designed for use in flow chemistry where additional sensors measure pressure and temperature to monitor the flow within the system. In automation, conductivity sensors have been used to detect the phase boundary between two immiscible solvents during extraction. There has been work to standardise the hardware and code used to run experiments to improve reproducibility of results and data sharing. Automation has been used effectively to explore unknown chemical space and predict the reactivity using the NMR spectra from before and after the reaction which has led to the discovery of novel transformations. Despite significant advances in automation, most synthetic chemistry in the lab is done manually in glassware as batch chemistry. In this study, we explore the utility of using sensor data gathered from hand-performed reactions. We use data collected by DeepMatter's DigitalGlassware platform and a mix of proprietary and original equipment manufacturer sensor devices. Specifically, these consist of a DeviceX™ reaction probe (temperature, ultraviolet (UV), pressure, stir rate and camera) and a Vernier thermometer, both suitable for a multi-necked flask, and an environmental sensor which would go adjacent to the reaction setup. The sensors used in this study recorded and more importantly saved the data in an open extensible markup language format. The proposed utility focuses on predicting current and future product formation to track the reaction progress. NMR and chromatography are two methods commonly used for monitoring reaction progress. To be quantitative, UV chromatography requires calibration of the UV detector with the analyte. This can be a challenge because authentic samples of the analyte may not be available. Alternative detection methods such as evaporative light scattering detection and chemiluminescent nitrogen detection are less accurate, but do not need calibration with the sample. NMR provides quantitative results with use of an internal standard but can be more challenging to interpret and may require more extensive sample preparation and interpretation. In hand-performed reactions, these methods take time for the chemist to perform and valuable machine time. Sample preparation means these methods are limited to being applied during chemists' worktime, whereas reactions often run overnight or over non-workdays. |
62f3b3f38dba68454c1e480d | 1 | Once enough data has been collected for a reaction, sensors could be employed to send data to a trained model which can monitor the product formation and predict the future product formation. This would inform the chemist when the reaction is nearing completion or if the reaction has stagnated. Sensor choice can be tailored to the specific chemistry. For example, pH sensors could be used in a pH-dependent reaction or colour data could be used if a colour change occurs in the reaction. This can be seen as a step towards automating, quantifying, and recording the data from some of the simpler tasks a human chemist does to understand their reaction. A litmus paper pH test is replaced by a quantitative pH sensor with a timeseries and the same for a colour change, qualitatively observable by eye but can be quantitatively defined by red, green, and blue (RGB) colour values. Sensors also offer the possibility for monitoring aspects which cannot be monitored easily in traditional ways. Data which can typically be harder to discern includes activity status of catalyst or reagents which may have poor ionisation, high reactivity, or not show under UV and common thin-layer chromatography stains. |
62f3b3f38dba68454c1e480d | 2 | Chemists can use a variety of methods and intuition to analyse the reaction mixture to predict progression at the current instant in time or the future progression. For example, it may be determined a reaction has plateaued if two measurements in succession show no change in the reaction profile. Alternatively, a chemist may have experience with the reaction and develop an empirically based intuition of when the reaction has ended. For example, a reaction may typically reach completion at a variable conversion but within the same timeframe. The research here aims to test the suitability of machine learning for this objective. |
62f3b3f38dba68454c1e480d | 3 | Machine learning for yield prediction using the reaction scheme and conditions is an approach which has been implemented by others. This research has been applied to palladium catalysed reactions, including Suzuki and Buchwald-Hartwig cross-coupling reactions. However, the methodology struggled when applied to patent data which had too much inconsistency for accurate yield prediction. Two issues with chemical reaction data that make predictive tasks challenging are the sparsity of the data within chemical space, and a lack of reported failed experiments. This approach is complementary to the one we developed in this research as both have different aims. Our strategy aims to tackle the variability in yield that can be seen for a specific reaction. This can be a non-trivial problem for reactions that suffer from poor reproducibility. The method we have developed treats each repeat of a reaction as a single instance that can be distinguished from other instances by differences in the time-series data. Our hypothesis is that patterns within these differences can be exploited by a suitably trained machine learning model, thereby allowing accurate predictions of product formation. |
62f3b3f38dba68454c1e480d | 4 | The reaction studied in this work is a Buchwald-Hartwig cross-coupling reaction between benzophenone hydrazone and 4-chlorotoluene. Two experienced chemists at a contract research organisation carried out 15 repeats of this reaction and used the hardware described earlier, provided by DeepMatter to generate time-series data for each repeat. The dataset will be processed to a suitable format for training and testing machine learning models to predict the current and future product formation. A series of machine learning algorithms will be used to evaluate the suitability of different models for the predictive task. Product formation is a continuous variable; hence, regression models are of interest including linear and polynomial regression models, decision tree models, and neural networks. The main limitation of the models being developed is they will only relate to this dataset, and therefore, only this specific reaction. However, extending this work to include larger datasets of multiple related reactions would be worth exploring in future work. |
62f3b3f38dba68454c1e480d | 5 | We begin by describing the experiments that were performed to generate the raw data that forms the basis of our study. A DeviceX™ reaction probe and a Vernier thermometer were inserted into the reaction vessel for each reaction (computer-rendered image shown in Figure ). The DeviceX™ contains a series of immersed sensors, exposed sensors, and a camera at the end for recording colour. Additionally, an environmental sensor, placed in the fumehood, measured the ambient temperature, humidity, and light levels. The data collected from this equipment was saved to a cloud storage and comprises several time-series. The experimental procedure was based on an optimised reaction obtained from the literature. This exact literature route was repeated four times. Then the modified version shown in Figure was used in the next 11 runs. The original route used tert-amyl alcohol (TAA) in place of isopropyl alcohol (IPA) and only ran for two hours. One difference between the solvents is that IPA's boiling point is 82.5 °C compared to 102 °C for TAA. Other variations between reactions included: the version of DeviceX™ probe which was used and what sensor data was recorded, the type of thermometer used and the rate of hydrazine addition. All of these are variations are displayed in Table in the Supplementary Information. The first four experiments (001 to 011) and 024 were excluded, because the reaction probe used in these did not collect colour data (Figure in the SI). This left ten reactions to use for modelling and evaluation of the models. The first eight runs ran for a duration of five hours and the last two were left for longer (eight hours) as there was evidence that the reaction was still progressing. |
62f3b3f38dba68454c1e480d | 6 | The sensor data were collected by four different instruments. The reaction probe measured UV A and UV B wavelengths, stir rate, temperature, and pressure. The system extracted the average RGB components of the images captured by the submersed camera. The environmental sensor measured light, humidity, temperature, and pressure. The Vernier thermometer provided a more accurate measure of temperature than the reaction probe (±1 °C versus ±3 °C). This temperature data was therefore used in preference to the temperature data from the probe. |
62f3b3f38dba68454c1e480d | 7 | The dataset also included liquid chromatography-mass spectrometry (LC-MS) data collected at 30-minute intervals to determine the conversion of the reaction. The peaks were assigned by their molecular weights. Specifically, these were benzophenone hydrazine (starting material), the product, and an impurity from hydrazine reacting with IPA which was identified by H NMR (Figure in the Supplementary Information). All the monitored reaction components contain the highly conjugated phenylhydrazone group. Therefore, it was expected their UV absorbance coefficients would be similar enough to allow the percentage area of the peak integrals to be directly compared and used for monitoring reaction progression. |
62f3b3f38dba68454c1e480d | 8 | The data was processed by changing the timestamps in coordinated universal time to time in seconds relative to the start of the reaction. Datapoints were kept if they fell in the window between the first and final LC-MS measurement. The sensor data was down-sampled to every ten seconds and the LC-MS outcome data was up-sampled by linear interpolation also to every ten seconds. |
62f3b3f38dba68454c1e480d | 9 | The problem was approached as a regression problem with a continuous spread of outputs representing product formation. Many widely used time-series specific approaches, such as autoregressive integrated moving average, were unsuitable for this problem as they assume a static output with only seasonal variations. This contrasts with the task described here that involves predicting an ever-increasing product amount. However, the approaches we use, such as recurrent neural networks, are suitable for application to time-series and do not assume a stationary output. |
62f3b3f38dba68454c1e480d | 10 | Due to the time-series nature of the data, cumulative and change values between measures were calculated. Cumulative values were calculated by successive addition of each time instance of a feature. Change values were calculated by measuring the difference between the previous and current time instance of a feature. Power transformations were applied to the features and to the cumulative features to increase the linearity of their relationship with product formation. Models were made using linear regression, cubic polynomial regression, random forest, gradient boosting regression, and long short-term memory (LSTM) neural networks. The linear regression, random forest, and gradient boosting regression models were all implemented in Python using conventional sci-kit learn libraries. The cubic polynomial regression model was constructed using a custom function in Python. The decision tree models each used 1000 trees and the same random state was used. The LSTM neural networks were made using the Python package Keras. Two LSTMs were constructed. The first LSTM predicts the current product. A second LSTM was constructed to predict future product formation. Both LSTMs used a sliding window of data to make predictions. |
62f3b3f38dba68454c1e480d | 11 | To evaluate the predictive performance of the models, the data was divided into a training and test set, according to run rather than aggregated samples. Due to the small number of runs within the dataset, the test set consisted of a single run and all datapoints from a run were kept together within the same partition to prevent leakage from the training set into the test set which would give an overly optimistic prediction. Uniform Manifold Approximation and Projection (UMAP) was used in conjunction with K-means clustering algorithm to visualise and group similar runs together. |
62f3b3f38dba68454c1e480d | 12 | The sensor data from the reaction data was averaged across the runs. From these averages the Pearson correlation coefficients, r, between the features and product formation were calculated (Figure ). The coefficients between a feature and product formation were used as a starting point to determine the utility of a feature in modelling. To contextualise the coefficients, most cumulative features were expected to show some correlation to product, as if the feature is always positive or negative, both product and the cumulative feature will continually increase in value until product formation has stopped. The features most correlated with product formation are the cumulative colour features, particularly green and red, and a more negative correlation for blue. From the average values across the runs of these three features for the first five hours of the reaction, it can be observed that green and red steeply decrease over the course of a reaction and blue shows a more shallow decrease (Figure ). A hypothesis for this is the reaction tends to black (where RGB would be 0,0,0) as the palladium catalyst is lost from the reaction and is oxidised to palladium (0) black from the activate palladium (II) in the palladium acetate and the palladium ligand complex. This fits with the chemists' observations of the initial mixture being described as cream coloured but turning black or darker brown later. The runs were kept whole during the 9:1 train/test split to prevent data leakage and more accurately replicate a real-life scenario where the model would not be exposed to any data from the unseen test run. Initially, a linear regression model was developed on the cumulative green values from nine runs to predict the product for the tenth run. Assessing the model using cross-fold validation provides a robust estimate, in data-limited situations, of the accuracy of the model and its ability to generalise by testing its predictions on all the datapoints. Models were evaluated by measuring the mean absolute error (MAE) of the predictions. All product yields are as percentage points Thus, despite MAE being reported with units of % it is an absolute and not a relative error. |
62f3b3f38dba68454c1e480d | 13 | Linear regression is a natural starting point. However, it was observed the relationship between the cumulative green and product formation was not linear. To remedy this different power transformations were applied, and they were evaluated by the improvement in accuracy of the transformed features in a linear regression model compared to the nontransformed feature. After evaluating the different power transformations, a Yeo-Johnson transformation was selected as the most promising. The purpose of a Yeo-Johnson transformation is to reduce the skewness and make the distribution of the data more Gaussian. Applying the Yeo-Johnson transformation to cumulative green yielded a more Gaussian-like distribution, as can be observed in Figure . Other models explored included polynomial regression. The polynomial used was a cubic function. This model was considered suitable due to the non-linear relationship between cumulative green and product. Random forest and gradient boosting regression, both based on decision tree models, were also explored. |
62f3b3f38dba68454c1e480d | 14 | Use of all features would allow an evaluation of predictions without prior manual identification of features. Green cumulative and Yeo-Johnson transformed green cumulative compare the effect of the power transform and to what extent a single feature can provide accurate predictions. The non-colour dataset investigates how predictive the non-colour data is. |
62f3b3f38dba68454c1e480d | 15 | As a baseline, we calculated the MAE of predictions obtained from just the mean product values of the training runs. This gave an MAE of 7.0%. The four datasets were combined with the four models to produce all possible combinations. The results of these are shown in Figure . The Yeo-Johnson transformation improved results when using a linear regression model (Untransformed: 4.2% and transformed 3.8% MAE). The results for each algorithm with the no colour dataset (MAE 7.0, 7.1, and 8.2%) were all comparable to or worse than the baseline results (7.0%). This suggests the Pearson correlation coefficients identified the useful features in this context, and there is little value in the non-colour data in isolation. The best predictions were obtained using polynomial regression with cumulative green (MAE 3.6%). However, this required the prior step of feature identification. A gradient boosting regression model made predictions which were nearly as accurate (MAE 3.9%) but with all the features. These predictions were more accurate than those obtained using a gradient boosting regression model with cumulative green as the sole feature in contrast to all other models, suggesting there is predictive value in the remaining data when using a suitable model. Using all features has the advantage of obviating the need for prior feature selection. This does not interfere with the interpretability of the model either, as feature importance values can be extracted from the decision tree models. Feature importance using all features in a gradient boosting regression model (Figure in the Supplementary Information) further confirms colour to be the most informative feature and suggests that it is likely all four models identify similar patterns in the relationship between cumulative green and product formation. in the associated datasets. Given that the input data is a time-series signal, this raises the question of whether a time-series model would be more accurate. |
62f3b3f38dba68454c1e480d | 16 | Recurrent neural networks, such as LSTMs, are well suited to time-series problems and other sequential tasks such as natural language processing and have been successfully applied in other areas of chemistry. The LSTM model constructed uses a sliding window approach, whereby a window of fixed size is formed over the data, and this window slides over the data to capture different portions of it. This method ensures the volume of data used in the model remains consistent. The method was employed to use the previous 20 minutes of sensor data. |
62f3b3f38dba68454c1e480d | 17 | Changing to an LSTM model using all features, improved prediction accuracy with a MAE of 1.2%; this is compared to a best of 3.6% for the non-neural-network models. These results can be contextualised by comparison to the yield range in addition to the earlier described baseline. The observed yield was between 7.5 and 38.1%, giving a range of 30.6%. Therefore, a predictive accuracy with a MAE of 1.2% was considered useful in this context. |
62f3b3f38dba68454c1e480d | 18 | Following on from the promising results obtained for the instantaneous predictions from the LSTM, a second LSTM was designed for the more ambitious aim for predicting the future product. The second model uses sensor data and predictions from the first model between times t1 -y and t1, where t1 is the current time and y is a fixed time interval, to predict product at time t1 + z where z is a variable time interval. |
62f3b3f38dba68454c1e480d | 19 | To balance ambition (i.e., the extent of the forward time interval), accuracy, and computational cost, y = 2 hours, and z = 0, 30, 60, or 120 minutes. These time interval values can be put into context by comparison to reaction duration which was between five and eight hours (300-480 minutes). The expected behaviour is the larger the value of z, the harder it is to predict product formation. A range of values were selected for z to allow an evaluation of the relationship between the MAE and z. Because of the greater challenge associated facing the second model, y was assigned a higher value than x. This meant sensor data over a longer duration of time was used in the second model. In this task, it is more important to be able to see the larger context of the reaction progress. Having two models was advantageous, as it allowed the assessment of what would have been the intermediate output when t3 = 0. |
62f3b3f38dba68454c1e480d | 20 | Since the LSTM gave a significant improvement, this model was chosen for the more ambitious target of predicting future product formation. Because current product can be predicted relatively accurately (MAE of 1.2%), a time-series of predicted product values was obtained. These predicted product values and the sensor data are then fed into a second model to predict the product conversion in the future. The results from the cross-validation of the two LSTM models are shown in Figure . There is a large (but unsurprising) increase in error when predicting just 30 minutes ahead. |
62f3b3f38dba68454c1e480d | 21 | To assess the model, a series of real-world scenarios were identified. One scenario was to split the data, so the last chronological run was the test data, and the previous runs were the training data. This corresponds to the training data that would exist whilst the final reaction was being performed. The trend of the results from this use case closely align with those from cross-validation (Figure in the Supplementary Information). Compared to the crossvalidation, the model performed slightly better for all values of z greater than zero. |
62f3b3f38dba68454c1e480d | 22 | Another example use-case would be for predicting when product formation has stagnated in a reaction at a lower conversion than expected. The goal in this scenario would be to detect a failed reaction at an earlier stage, and the live data in combination with predictions would give an indication of this to a chemist. This can be rationalised as, if the colour change occurs to indicate the catalyst has been consumed, no more product will be formed. Reactions 16 and 17 were both low yielding and the chemists performing these reactions observed a potential exposure of the hydrazine to atmospheric moisture. An 8:2 train: test split was used and reactions 17 and 25 were used as the test runs. The rationale for using this split is twofold. The first and primary aim is to demonstrate that if a failed reaction, 16, is included in the training set, the model can identify future failed runs, such as 17, early. The second aim is keep run 25 in the test set to ensure the model is still capable of generalising and accurately predicting successful runs. |
62f3b3f38dba68454c1e480d | 23 | A weakness identified in the model was the imbalanced dataset due to the training data containing only one failed reaction. To address this, oversampling was implemented. A condensed version of the reaction data was interpreted by a SMOGN (Synthetic Minority Over-Sampling Technique for Regression with Gaussian Noise) algorithm. The SMOGN algorithm can be particularly useful when the values in the interest of predicting are rare or uncommon, such as the failed reactions in this scenario. The algorithm identifies underrepresented reaction runs from the condensed dataframe and returns a new dataframe with oversampling of under-represented data and undersampling of over-represented data. The algorithm performed as expected, with the newly returned dataframe oversampling the only low yielding reaction in the training set, reaction 16. The results for this can be seen in Figure training size increases before decreasing again. The reason for this could be due to the differences between runs, with some being more challenging to predict than others. This is also seen in the results from cross-validation. Figure This suggests that these runs are more challenging to predict. Some runs perform better in the chronological results, despite having a smaller training dataset. This could be due to the runs in the dataset being more similar. This can be observed with run 17 as it is similar to run 16, since both had poor product formation. In a larger dataset, it could be envisaged that this could allow similar runs or reactions to be identified. This could allow the use of tailored models which would seek to place emphasis within the training data on runs identified as similar to the run of interest and this could create models which could give more accurate results. A small-scale example of this can be observed earlier in the accurate results of run 17 in the chronological use-case, which included also failed run 16 in the small volume of training data. |
62f3b3f38dba68454c1e480d | 24 | The reaction used here was suitable due to the reliable curve shape and profile of the reaction. Mild changes between runs, such as the rate of hydrazine addition, did not have a noticeable impact on prediction. However, large changes may; concentration or temperature could affect the rate of reaction. Future work will investigate how more significant changes affect the accuracy of the model. To assess further the potential and utility of using machine learning to predict product, more examples would need to be examined. This methodology could enable AI augmentation of reaction monitoring to assist synthetic chemists and facilitate a greater understanding of the reaction by identification of correlations between sensor features and reaction outcomes. Insights into the chemistry being performed could also be developed, for example, the correlation between cumulative green and product formation in this work, providing a quantitative description of the colour change in the reaction. |
644125f971383d0921033632 | 0 | In the last decade, metal-halide perovskites (MHPs) have grown to the most promising material classes for future photovoltaic (PV) devices. While their huge potential is out of question, recent experimental and computational research expands beyond the class of leadhalide perovskites exploring chemical and structural phase spaces of MHPs and perovskites derivatives. Several structural diverse systems are now under investigation, including, for example, 3D, 2D, and double perovskites, targeting applications beyond PV, as e.g. photodetection and photocatalysis. The expansion of the discovered material space of MHPs has recently moved towards the investigation of lead-free systems in order to overcome the concerns related to Pb-toxicity. |
644125f971383d0921033632 | 1 | While effective alternatives for PV applications rely most on tin-based compositions, several other phases containing different metals such as Bi, Sb, Cu, Ge, have been discovered and investigated. In many cases, such perovskites resulted to be not suitable for PV devices but possess very appealing optoelectronic properties which can be exploited in other applications. Among these systems, bismuth-and antimony-based perovskite derivatives of general formula Cs3M2X9 (M=Bi, Sb; X=Br, I) have shown a strong technological potential in particular in the area of photocatalysis (both for solar fuel generation and organic synthesis) and photodetection. Bismuth-based defect-ordered perovskites have been object of more intense studies with respect to the corresponding Sb systems, notwithstanding their appealing optoelectronic properties such as high carrier mobility, low trap density, and long diffusion length. As a matter of fact, Cs3Sb2Br9 millimeter-sized single crystals and nanoflakes have been applied in the fabrication of photodetectors with excellent performance in terms of responsivity and detectvity. Good photodetectors based on Cs3Sb2Br9 have been also prepared by means of chemical vapor deposition technique. Cs3Sb2Br9 and Cs3Sb2I9 revealed also to be suitable candidates for photocatalytic applications and in this respect have been, for example, applied in the photocatalytic carbon dioxide reduction, water splitting, and in aromatic C-H bond activation. Some efforts has been also applied in the possible use of these defect-ordered perovskites in the manufacturing of solar cells. Cs3Sb2I9 has been preferentially used in for PV due to its suitable band gap around 2 eV, providing good stability but relatively low efficiencies around 2-3%. While there is a clear relevant interest in Sb-based materials, many of their fundamental features still need to be explored, for example, in terms of synthetic procedures, band gap tuning by metal or halide alloying, and their electronic structure. Considering the structural and optoelectronic similarity with Bi-based analogues, which have been more deeply investigated, it appears that some tuning strategies of the photophysical properties of Sb materials should be clarified to enlarge the plethora of lead-free perovskite derivatives. For example, the preparation of mixed Sb/Bi phases of vacancy ordered perovskites has not been reported in the current literature but there is a strong motivation to investigate such a system. Indeed, a recent work on the Cs2Ag(SbxBi1-x)Br6 double perovskite solid solution has demonstrated a nonlinear trend of the band gap value as a function of x, providing mixed compositions with lower values with respect to Cs2AgBiBr6 and Cs2AgSbBr6. However, the possibility of successfully preparing a continuous solid solution within the vacancy-ordered Cs3(Sb1-xBix)2Br9 system has still to be verified. The choice of a suitable synthetic route for these Sb/Bi materials is critical due to the low solubility of halide precursors posing strong limitations for traditional synthesis procedures applied to MHPs. A viable choice could be the use of mechanochemistry which has been recently applied to Bi-based systems but never to Sb-containing perovskites or mixed compositions. Finally, the halide alloying strategy to modulate the optical properties of perovskites has still to be investigated in Sb vacancy-ordered materials. |
644125f971383d0921033632 | 2 | Based on the above considerations, in this work we carried out the mechanochemical synthesis of the Cs3(Sb1-xBix)2Br9 and Cs3Sb2(I1-xBrx)9 systems (0≤x≤1) to understand the role of Sb/Bi and I/Br alloying on the structural and optical properties. For both alloying routes, we employ mechanochemistry as the main preparation technique to verify its suitability to Sbbased systems and its ability in providing phase-pure mixed compositions. |
644125f971383d0921033632 | 3 | Powdered samples of the Cs3(Sb1-xBix)2Br9 system for x=0, 0.1, 0.2, 0.4, 0.6, 0.8, 0.9, and 1 have been prepared by using a planetary ball miller according to the experimental conditions reported in the Experimental Section (see the Supporting Information, SI). Figure shows the appearance of the eight samples prepared from Cs3Sb2Br9 (left) to Cs3Bi2Br9 (right). Figure reports the room temperature (RT) x-ray diffraction (XRD) patterns collected on all the samples reported in Figure . Samples have been measured both with Cu-K radiation (Figure ) and with Mo-K radiation (Figure ) to provide the most reliable structural results on these novel samples. Compositions (expressed as x) in Figure refer to the effective stoichiometries determined by microprobe analysis (see later in the text). All the samples are single phase and are in agreement with the hexagonal symmetry of the P-3m1 space group describing the two end members, namely Cs3Sb2Br9 and Cs3Bi2Br9. A superposition of the reference structure model for these last two samples and the experimental patterns are shown in Figures and, respectively. According to these results, the mechanochemical synthetic procedure was effective in preparing single-phase materials for both the two stoichiometric compositions Cs3Sb2Br9 and Cs3Bi2Br9, and for the mixed Sb/Bi samples. Such green and sustainable synthetic approach has never been applied to Sb-based defective perovskites nor to B-site mixed compositions, while a previous report on its effective use for the preparation of Cs3Bi2Br9 is reported in the current literature. The morphology and chemical composition of the samples has been determined by means of scanning electron microscopy (SEM) and microprobe analysis. Images of the sample morphology are reported in Figure . As a consequence of the mechanochemical synthesis, the samples do not possess a well-defined morphology, as occurs in solution chemistry synthesis, and grains with various shapes and dimensions (in the micron and sub-micron range) are observed. Elemental analysis by SEM confirmed a good agreement with the nominal compositions (see Table ). In the rest of the paper, we will make use of the effective compositions instead of the nominal ones when discussing the experimental results. |
644125f971383d0921033632 | 4 | The XRD patterns have been refined by Rietveld method. The lattice parameters determined are reported in Table , while Figures show their trend together with lattice volume as a function of x (Bi-content). The linear trend of the lattice volume reported in Figure agrees with the Vegard's law for solid solution, confirming the complete solubility of Sb/Bi in the lattice. This result could be anticipated based on the similar ionic radii between Sb 3+ and Bi 3+ ions and the same crystal structure of the two end-members. |
644125f971383d0921033632 | 5 | The Raman investigation has been performed focusing on the low energy part of the Raman activity, i.e. in the range 25-250 cm -1 . At a first glance, the single-phase of the endmembers' samples is confirmed, as reported in Figure . Indeed, the Raman spectrum of the Cs3Bi2Br9 sample clearly shows the characteristic modes at 165 and 190 cm - , due to the bond vibrations inside the octahedral cage, BiBr6, as already described in our previous work. In analogy with the isomorphic Cs3Bi2Br9, also for Cs3Sb2Br9 the Raman response qualitatively presents the same spectral fingerprints: a broadened structure centered at about 70 cm -1 clearly resulting from the overlapping of different modes and two sharp and symmetric modes peaked at 182 and 210 cm -1 , resulting from the Sb-Br bond vibrations in the octahedral unit SbBr6. The proper assignment of the mode symmetry seems to be controversial. The crossover between Bi-rich to Sb-rich system is also well evidenced in Figure where the intensity of the mode at about 165 cm -1 proper of Cs3Bi2Br9 and the one at 210 cm -1 of Cs3Sb2Br9 are plotted as a function of the Bi content: as the former increases, the latest decreases. In Figure , the spectrum for the x = 0.57 sample is reported together with the result from the best-fitting procedure using four Lorentzian curves according to the above mentioned two-mode behavior. observed and described also from a computational point of view. The values of the direct and indirect band gaps determined from the Tauc plots (reported in Figure ) are shown in Figure as a function of x. The values of direct and indirect band gaps for Cs3Sb2Br9 and Cs3Bi2Br9 are in agreement with those previously reported in the current literature. By looking at Figure , a significant bowing of the band gap for mixed compositions is observed. This result is unexpected since, in general, metal ion replacement in solid solution provides a scaling of the band gap obeying the Vegard's law. The reduction of the band gap (direct) for mixed compositions is relevant, reaching a minimum value of about 2.35 eV when x is around 0.2. |
644125f971383d0921033632 | 6 | The same trend in the band gap bowing is observed for the indirect band gap. Such a trend has never been observed before in any defective perovskite but has been reported for the double perovskite system Cs2AgSb(1-x)BixBr6 and attributed to chemical rather than structural effects. In our case we cannot exclude a synergistic effect due to coexistence of mass disorder and microscopic strain as expected in a two-mode scenario evidenced by Raman data. The present results corroborate the strategy of band gap tuning by Sb/Bi mixing which seems to be general considering the structural difference between defective perovskite and double perovskites and therefore related to the peculiar electronic properties of antimony and bismuth in mixed compositions. |
644125f971383d0921033632 | 7 | To understand the unexpected change in the band gap, density functional theory (DFT) calculations have been performed for the reference systems Cs3Sb2Br9 and Cs3Bi2Br9 as well as the mixed Cs3(Sb0.5Bi0.5)2Br9. Ionic positions of the three systems were obtained by PBE+D3 geometry optimization, followed by refined electronic structure calculations using the hybrid HSE06 functional including spin-orbit coupling (see SI for computational details). The indirect bandgap and first direct transition of Cs3Bi2Br9 (Cs3Sb2Br9) were calculated to be 2.97 (2.91) and 3.10 (2.96) eV, respectively (see Table ). Previous GW-BSE calculations showed large exciton binding energies of ~300 meV, which are not captured in our DFT calculations. Correcting our DFT results by the large exciton binding energies results in a fair agreement with the experimental band gap values. |
644125f971383d0921033632 | 8 | Cs3(Sb0.5Bi0.5)2Br9 Cs3Bi2Br9 For the pure Sb and Bi species, the contribution of valence band is primarily given from bromine 4p orbitals with slight metal contribution, while the conduction band is mainly due to the metal cation (for Sb 5p orbitals and Bi 6p orbitals) and its interaction with bromine 4p orbitals, see Figure . Moving to the mixed Bi/Sb material, we observe substantial differences depending on the arrangement of the metal ions. The fully mixed system, with alternating SbBr6 and BiBr6 octahedra, interestingly shows a negligible difference in band gaps, with indirect and direct bandgap of 2.90 and 3.00 eV, compared to the pure Sb and Bi species, see Figure . |
644125f971383d0921033632 | 9 | When we consider aggregates of Sb and Bi, see Figure , we observe a decrease to 2.59 and 2.71 eV of the indirect and direct band gaps, respectively, see Figure . In terms of stability, the two models show the same energy suggesting that both appear equally in the crystal structure. This clearly suggests that the decreasing of the band gap experimentally found for the mixed Sb/Bi material is associated to the presence of this kind of aggregates. In fact, considering the energy alignment in the density of states, Figure , we find an upshift of the VB and CB moving from Cs3Bi2Br9 to Cs3SbBiBr9 of 0.52 eV and of 0.15 eV, respectively. In The band gap bowing of mixed metal perovskites has been observed in Pb/Sn perovskites and attributed to chemical effects mismatch in energy between s and p atomic orbitals of the two metals. A similar interpretation holds also the present Cs3(Sb1-xBix)2Br9 b) |
644125f971383d0921033632 | 10 | solid solution where a mismatch in energy between s and p atomic orbitals of Sb and Bi is found. It is of relevance to report here this novel tuning strategy for defective perovskites which allows to achieve significant lower band gap values through metal alloying instead of the usual halide mixing which is known to be the main strategy to modulate the absorption edge in these systems. Cs3Sb2(Br1-xIx)9 System |
644125f971383d0921033632 | 11 | We further explored alloying strategies by tuning the halide content Br/I on the Cs3Sb2Br9 perovskite. While such strategy has been explored for the Bi analogue, no studies are reported for the Sb-based composition. For this purpose, samples of the Cs3Sb2Br9-xIx system for x= 0 (also reported above), 2.5, 4.5, 6.5, and 9 have been synthesized by means of mechanochemistry according to the experimental conditions reported in the SI. Photographs of the five samples are reported in Figure . The XRD patterns of the whole series of samples are reported in Figure . For the Cs3Sb2I9 perovskite two different polymorphs have been reported, namely the layered modification with s.g. (space group) P-3m1 (analogous to Cs3Sb2Br9, see above) and the dimer modification with space group P63/mmc, featuring SbI6 octahedra fused into Sb2I9 3-dimers through sharing of their triangular faces. According to literature, the dimer form is synthesized from solution using a polar solvent, whereas the layered form is obtained from a solid state reaction at low temperatures. However, the present synthetic approach has never reported before for Cs3Sb2I9 and additional information on the stability of the different polymorphs have been obtained. According to Figure , reporting the XRD patterns of Cs3Sb2I9 perovskite against the expected structure for the P-3m1 space group, we could obtain the layered modification through mechanochemistry. However, this was possible after optimizing the synthesis approach in terms of milling cycles and time. As a way of example, All the data follow a linear trend as a function of the iodide content, according to the Vegard's law, confirming the formation of a continuous solid solution in this system in analogy with the analogous Bi-based system. a) b) c) The observed expansion of the unit cell together with the progressive substitution of the Br atoms with a heavier one leads again to a clear redshift of the Raman modes, more pronounced with respect to what observed when Bi/Sb are replaced at the B sites, as above described. From Figure this behavior is particularly evident for the mode at 210 cm -1 proper of the Cs3Sb2Br9 structure: its center shifted till 185 cm -1 for x = 6.5 sample, then it completely disappears in the Cs3Sb2I9 compound. |
644125f971383d0921033632 | 12 | Moving from one endmember to the other gradual changes are observed but it is more difficult to recognize a two-mode behavior probably caused by the higher disorder in the chemical distribution related to the specific site involved in the substitution. Moreover, as a consequence of the bandgap redshift, later discussed, the Raman intensity of x = 4.5, 6.5 and 9 |
644125f971383d0921033632 | 13 | Finally, the Raman activity of Cs3Sb2I9 sample is consistent with the literature. The spectrum is dominated by two intense features at 148.9 and 166. literature. On the other hand, no previously data for Br/I mixed samples are available. Data in Figure show a relevant initial reduction of the (direct) band gap from about 2.5 to 2.1 eV (at 50% of Br and I), followed by a smoother reduction of the band gap up to about 2.0 eV, showing again a bowing of the Eg. A similar alloying strategy was investigated on the Cs3Bi2Br9-xIx defective perovskites reporting as well a degree of band gap bowing which is, however, less pronounced with respect to the present Sb-based perovskites. |
644125f971383d0921033632 | 14 | In For both Cs3(Sb1-xBix)2Br9 and Cs3Sb2Br9-xIx the Raman inspection allowed to confirm the single-phase structure of all the endmembers. Moreover, the reported results, confirmed by the XRD evidence, point out a redshift of the modes as a consequence of the cell expansion and the substitution of heavier ions that enlarged the reduced mass of the vibrational unit. The Raman features for the mixed samples are consistent with a two-mode behavior and the mixed phonon landscape which can in turn play an active role in the electronic response. |
644125f971383d0921033632 | 15 | The present results provide a novel sustainable synthetic route to prepare Sb-based defective perovskites and, more importantly, two alloying strategies which allow to tune the band gap in most of the visible spectrum providing also the first evidence of band gap reduction by metal mixing (Sb/Bi) in a defective perovskite structure. Such effect has been only previously observed in double perovskites (Cs2AgSbxBi(1-x)Br6) suggesting its possible universal character which is worth of being further explored in other related systems. |
6274cc87f053df3d36114962 | 0 | Organic room temperature phosphorescence (RTP) is the radiative emission of triplet excitons, which occurs with a phosphorescence lifetime, t ph , ranging from microseconds to milliseconds. In the past decade, the development and study of organic RTP materials have gained increasing attention due to their potential in applications across diverse fields such as optoelectronics, lasers, sensing, imaging, data storage, and X-ray scintillators. To date, a plethora of design motifs have been proposed, mainly based on the modulation of intramolecular and/or intermolecular interactions. These RTP design principles are primarily based on improving the intersystem crossing (ISC) efficiency and/or suppressing the non-radiative and oxygen quenching deactivation of the triplet excitons. For example, ISC efficiency has been shown to be enhanced via the incorporation of heavy atoms, the breaking of conjugation, and the introduction of a charge-transfer (CT) bridge motif and resonance linkage (intramolecular modulation). Also, matrix rigidification, molecular aggregation, and polymerization have each been demonstrated to suppress the non-radiative decay of triplet excitons (intermolecular interactions) for achieving efficient RTP systems. |
6274cc87f053df3d36114962 | 1 | Consequently, most reported RTP research has focused on enhancing RTP efficiency (i.e., photoluminescence quantum yield, F PL ) and t ph . To date, there are only a few documented examples where RTP originates from a higher-lying triplet excited state. For instance, Tang and co-workers reported several benzothiophene derivatives showing dual phosphorescence, which was assumed to originate from both the first (T 1 ) and second (T 2 ) triplet excited states (Figure ). Huang and co-workers designed materials based on carbazole that enabled dual phosphorescence from T 1 and an H-aggregation stabilized species (T 1 * , T 1 * < T 1 ) (Figure ). However, this dual phosphorescence was only observed to occur in crystals, making it difficult to identify the implication of any higher-lying triplet states in the emission processes. Recently, Zhang and co-workers developed a series of triphenylamine-sp 3 linker-acceptor motifs exhibiting phosphorescence that was attributed to a high-lying T 1 (T 1 H ) state associated with the electronically decoupled donor and acceptor groups in the compound (Figure ). Nevertheless, emission from the T 1 H state could only be observed at low temperature (< 250 K). It thus remains a formidable challenge to design RTP materials whose emission originates from higher-lying triplet excited states. |
6274cc87f053df3d36114962 | 2 | Emission from CT states in donor-acceptor compounds is sensitive to the solvent polarity and originates from the solvent-stabilized excited states of emissive conformers. We hypothesized that it may be possible to achieve RTP from higher-lying triplet states by accessing non-equilibrium conformers in the excited state in the solid state, where the interconversion barrier from one conformer to another must be overcome for excitonic coupling to occur. Based on this concept, we report two conjugated RTP molecules, PXZ-Nap and PTZ-Nap (Figure ) using phenoxazine (PXZ) and phenothiazine (PTZ) as donors, which have both been observed to show conformational dynamics under external stimuli, and a substituted naphthalene (Nap) as the acceptor. The decoration of a 10H-phenothiazine-5,5-dioxide unit on the Nap distal to the donor is anticipated to modulate the excited state dynamics and to accelerate the ISC process owing to the presence of the sulfur atom. We found that when the two compounds are doped in poly(methylmethacrylate) (PMMA) or 1,3-bis(N-carbazolyl)benzene (mCP) (1 wt%), RTP from higher-lying triplet states occurs due to a thermally activated excitonic coupling between different conformers. Moreover, the RTP quantum efficiency of the emitters doped in mCP (1 wt%) improved over 80-fold compared to that in PMMA. The photophysics at a higher doping of 10 wt% in PMMA films revealed that there exist aggregation-regulated intermolecular interactions that offer a route to modulating T 1 H and low-lying T 1 (T 1 L ). Finally, we have exploited the anomalous RTP behavior to produce a temperature sensor. |
6274cc87f053df3d36114962 | 3 | The synthesis of PXZ-Nap and PTZ-Nap is outlined in Scheme S1. The molecular structures and purity of the two compounds were confirmed by a combination of 1 H & 13 C nuclear magnetic resonance spectroscopy (NMR), high-resolution mass spectrometry (HRMS), elemental analysis (EA), melting point determination and high-performance liquid chromatography (HPLC) (Figures ). We first modelled the optoelectronic properties of PXZ-Nap and PTZ-Nap in the gas phase using density functional theory (DFT) with the PBE0 functional and the 6-31G (d,p) basis set (Figure ). PXZ-Nap and PTZ-Nap possess similar lowest unoccupied molecular orbitals (LUMOs) that are localized on the Nap and highest occupied molecular orbitals (HOMOs) that are localized on the donor PXZ or PTZ, respectively (Figure ). The destabilized HOMO of PXZ-Nap indicates that in these compounds PXZ is the stronger donor group. The dihedral angles of PXZ and PTZ with Nap in PXZ-Nap and PTZ-Nap at the optimized ground state (S 0 ) geometry are 79.92° and 99.27° (Figure ), respectively, revealing compounds with weakly electronically coupled donors to the Nap. Time-dependent DFT (TD-DFT) calculations at the same level of theory reveal that the lowest singlet excited state (S 1 ) of each compound has a chargetransfer (CT) character (Figures 2b and Figure ). Although PXZ-Nap and PTZ-Nap possess similar T 1 energies, the nature of the T 1 states in these compounds is distinct, being CT for PXZ-Nap and locally excited (LE) for PTZ-Nap, which are highly dependent on a specific excitedstate structure. However, triplet spin density distributions at the optimized T 1 geometry indicate the T 1 of both possess LE character (Figure ). At the relaxed S 1 geometry there is a larger S 1 -T 1 spin-orbit coupling (SOC) matrix element (0.12 cm -1 ) in PTZ-Nap due to the presence of the additional heavy sulfur atom than in PXZ-Nap (0.09 cm -1 ). The predicted phosphorescence of PXZ-Nap and PTZ-Nap at their respective relaxed T 1 geometries decrease to 2.42 eV and 2.02 eV, respectively, compared with the excitation T 1 energies obtained from the TD-DFT calculations at the S 0 geometry (Figure ). |
6274cc87f053df3d36114962 | 4 | Next, the energies of the frontier molecular orbitals were inferred from the electrochemical behavior of PXZ-Nap and PTZ-Nap, measured using cyclic voltammetry (CV) and differential pulse voltammetry (DPV), in deaerated DMF with 0.1 M tetra-n-butylammonium hexafluorophosphate as the supporting electrolyte. The oxidation/reduction potentials (E ox /E red ) of PXZ-Nap and PTZ-Nap determined from the DPV peaks are 0.79 eV/-2.09 eV and 0.78 eV/-2.09 eV vs SCE, respectively (Figure ). Thus, PXZ-Nap and PTZ-Nap show similar HOMO values of -5.14 eV and -5.13 eV, respectively, despite the different donors, and the same LUMO value of -2.26 eV (Table ), the result of the LUMO being localized on the Nap core. The results at first did not seem to align with the gas phase calculations (Figure ). However, a closer inspection of the DFT calculations of PTZ-Nap, now employing a DMF continuum model, indicates that a quasi-axial conformer is responsible for the destabilized HOMO (Figure ); indeed, accessible quasi-equatorial and axial conformations of PTZ have been previously reported. Okazaki et al. have also reported a series of PTZ-containing molecules that exists in different conformations, with corresponding HOMO values that vary from -5.59 eV to -5.20 eV. The UV-vis absorption spectra show a weak CT transition at around 400 nm for both compounds (Figure ). The optical gaps (E g ) of PXZ-Nap and PTZ-Nap estimated from the onset of absorption spectra in toluene are 2.92 eV and 3.10 eV, respectively, which are almost same values as those measured in DMF (~3.1 eV). We hypothesized that variation in E g is caused by the existence of different conformers. The CT character of the emissive excited state was evidenced from the positive solvatochromism of the photoluminescence (PL) spectra (Figure and) and the accompanied increased PL lifetimes (Figure , Table ). We next investigated the photophysical properties of PXZ-Nap and PTZ-Nap in PMMA at 1 wt% doping. At this low doping concentration intermolecular interactions will only negligibly affect triplet excited dynamics while the impregnation of the emitters in a solid-state matrix will serve to suppress non-radiative decay. The steady-state PL spectra of PXZ-Nap in air and under vacuum (at 298 K) are broad, unstructured, and centered at 470 nm, which are characteristic of emission from a CT state (Figures 3a). At 77 K, there is the emergence of a second emission band at ~510 nm that is assigned to phosphorescence. The PL lifetime of PXZ-Nap under vacuum is 7.55 ns (Figure , Table ). Time-gated PL measurements (10 ms delay) detected the RTP spectrum of PXZ-Nap (Figure ), which is centered also around 470 nm but is narrower and its decay has an associated t ph of 83.3 ms (Figure ). Such a long lifetime rules out thermally activated delay fluorescence (TADF) as the origin of this delayed emission. The low-temperature phosphorescence (LTP) emission at 77 K is centered at 520 nm and possesses a slightly structured character, implying emission from a state with dominant LE character, which coincides with the calculations (Figure ). The differences in energy of the RTP and LTP suggest that they originate from two distinct conformers. Indeed, the blue RTP (T 1 H ) and green LTP (T 1 L ) afterglows provide stark evidence of emission from two different T 1 states (Figure ). Temperature-dependent steady-state PL and phosphorescence studies show the gradually red-shifted and intensified emission band centered at 525 nm (Figures 4a) and red-shifted phosphorescence emission (Figure ), respectively, further demonstrating that temperature is the key to excitonic communication between T 1 H and T 1 L . The related Commission Internationale de l'Éclairage (CIE) diagrams also reflect the gradual emission color change with changing temperature (Figures and). |
6274cc87f053df3d36114962 | 5 | Temperature-dependent lifetime studies (Figures and) reveal that the delayed emission is not thermally activated, providing further evidence that the origin of the long-lived luminescence at 470 nm results from RTP and not from TADF; in fact, only at 77 K is there a slight decrease in the excited state lifetime (Figure ), which is attributed to the significantly suppressed T 1 H emission (Figure ). Phosphorescence emission at 200 K measured at different time-gated windows indicates variable excited-state dynamics (Figure ), also implying the existence of The picture for PTZ-Nap is different. The steady-state PL under air is blue-shifted and structured, centered at 400 nm, characteristic of emission from an LE state; notably, there is also a broad emission tail (Figure ). The emission lifetime is 5.19 ns (Figure ). When placed under vacuum at 298 K, the structured emission beyond 475 nm is strengthened, which is strongly enhanced at 77 K. RTP emission can be observed in the range of 450-650 nm (Figure ), with an associated t ph of 66.7 ms. The LTP at 77 K has a similar emission but narrower profile compared to the RTP; however, the t ph (628 ms) is much longer (Figure and Table ). Similar RTP and LTP afterglows indicate essentially two degenerate triplet states (Figure ). Temperature-dependent photophysical investigations also reveal essentially indistinguishable T 1 H and T 1 L (Figure ). |
6274cc87f053df3d36114962 | 6 | We then investigated the photophysics of PXZ-Nap and PTZ-Nap as 1 wt% doped films in mCP, a suitably high triplet energy host matrix. The photophysical data acquired under vacuum of PXZ-Nap and PTZ-Nap are summarized in Table . The steady-state PL behavior of PXZ-Nap and PTZ-Nap in mCP is similar to that observed in PMMA (Figure ). Notably, the lowtemperature PL is red-shifted compared to the room temperature PL, which we contend is due to the dominant T 1 L emission at 77 K as conformational dynamics are expected to be essentially arrested at this temperature. The RTP spectrum of PXZ-Nap, collected with a time-gated window of 10-100 ms, is red-shifted (l em = 490 nm) and the t ph is shortened to 10.04 ms compared to that in 1 wt% doped PMMA (Figure ). A much more structured LTP emission at 77 K is observed (Figure ) compared to that of PXZ-Nap in PMMA (Figure ). The enhanced LE character (~400 nm) results from host-guest interactions between PXZ-Nap and mCP. For PTZ-Nap, a new high-energy RTP emission band at around 425-500 nm can be observed, with an associated t ph of 20.76 ms (Figure ). At 77 K, this emissive triplet state is suppressed compared to the main emission band; however, the lifetime at 475 nm significantly increases to 1.5 s (Figure ). The F PL values of PXZ-Nap and PTZ-Nap are improved to 10.8% and 7.7% in air compared to those in 1 wt% doped PMMA (1.3% and 5.3%, Tables ), respectively, values that are enhanced to 26.9% and 15.9% under vacuum, respectively (Table ). Therefore, the F PL associated with just the RTP (F RTP ) of PXZ-Nap and PTZ-Nap in mCP are no smaller than 16.1% and 8.2%, respectively. Compared with the F PL values of PXZ-Nap and PTZ-Nap in doped PMMA films, those in doped mCP films are enhanced 80.5 and 27.3 times, respectively (Figure ). This enhancement benefits from the Dexter energy transfer process occurring between the T 1 states of mCP and the guest emitters. These results demonstrate that different host-guest interactions can regulate triplet excited dynamics. To exclude that the triplet emission at around 525 nm originates from an aggregate, the prompt and delayed emission spectra were investigated in dilute 2-MeTHF glass (1´10 -6 M) at 77 K (Figure ). Notably, the structured phosphorescence in 2-MeTHF at 77 K for both compounds are at about the same energy, and also of similar energy to those measured in PMMA and mCP, which demonstrates that the phosphorescence from T 1 L does not originate from an aggregate. The DE ST values of PXZ-Nap and PTZ-Nap between S 1 and T 1 L in 2-MeTHF glass are 0.40 eV and 0.43 eV, respectively. Despite the difference between the measured DE ST in 2-MeTHF compared with the predicted value (Figure ), we can clearly see that in doped PMMA and mCP films DE ST between S 1 and T 1 H is almost degenerate, it is these values that coincide with the theoretical predictions. |
6274cc87f053df3d36114962 | 7 | Notably, there is significant LE emission from PTZ at around 380 nm for PTZ-Nap (Figure ) revealing that electron transfer to the CT state is slow compared to radiative decay. The experimentally calculated DE ST at 77 K deviates from the computed values (0.05 eV and 0.24 eV, respectively, Figure ), which we attribute to different accessible relaxed conformers at room temperature compared to those at 77 K. Combining the above-discussed results in PMMA, mCP and 2-MeTHF glass, we rationally conclude that the T 1 emission can be regulated by the host-guest interactions that govern the conformational landscape of PTZ-Nap. |
6274cc87f053df3d36114962 | 8 | Considering aggregation can induce conformational changes, we next investigated the photophysics of the compounds at a higher concentration in PMMA films (10 wt%) to investigate the influence of intermolecular interactions on the phosphorescence behavior. Compared to the 1 wt% doped films of PXZ-Nap and PTZ-Nap in PMMA, a spectral red-shift of the steady-state PL occurs (Figures and). For PTZ-Nap, the CT emission at around 505 nm is enhanced, associated with a 13.06 ns lifetime, compared with the LE emission at 400 nm (tPL = 5.02 ns) (Figure , Table ). The 10 wt% film of PXZ-Nap in PMMA exhibits similar RTP and LTP behavior to the 1 wt% doped film in PMMA, but with a shorter associated RTP lifetime (tPh = 42.1 ms) (Figures and). Due to the similarity in RTP spectra with those of the 1 wt% doped films, it can be concluded that there is no additional contribution to the emission from aggregates. The distinct afterglows of the 10 wt% doped PMMA films of PXZ-Nap at 298 K and 77 K also indicate emissions from two different T 1 states (Figure ). Similarly, temperature-dependent, and timegated-dependent PL spectra also reveal the dual phosphorescence from T 1 H and T 1 L (Figures ). At room temperature, the 10 wt% PTZ-Nap doped PMMA film is more emissive than the 1 wt% PTZ-Nap doped PMMA film, which is reflected also in the much longer RTP lifetime (Figure ) and the increased F PL of the RTP (2.1% vs. 0.3%, Table ). The LTP recorded at 77 K is more structured and is 33 nm red-shifted compared with the RTP (Figure ). The temperaturedependent delayed emission decay at 500 nm shows two distinct regimes. There is a shorter component that is thermally activated and that we assign to TADF (Figure ), which likely results from RISC from T 1 H to S 1 as these two states are nearly degenerate, while the other longlived component (up to sub-second) originates from phosphorescence. Compared with PXZ-Nap, the improved FPL (Table ) of PTZ-Nap is ascribed to the suppression of nonradiative decay processes. Based on these photophysical results, the intermolecular interactions present at higher doping concentrations are responsible for the modulation of the T 1 H and T 1 L emissions by influencing the T 1 geometry. We next investigated how the conformational dynamics affects the T 1 energy. A relaxed potential energy surface scan modelling for T 1 conformations was conducted to rationalize the observed dual phosphorescence mechanism (Figure ). The dihedral angle between the donor and Nap was progressively modulated. From Figure , the interconversion barrier (transition state, TS) between the two conformers is sufficiently small and thus the population of the T 1 H conformer dominates. Thus, under photoexcitation, the populated singlet excitons intersystem cross and eventually relax to the lowest energy T 1 L state (Figure ). At ambient temperature, the T 1 L excitons can be thermally activated to access the T 1 H and from this state RTP is observed. This process is suppressed at low temperature, resulting in the observed T 1 L phosphorescence. Finally, benefitting from the significant difference in the spectral afterglow response to temperature of PXZ-Nap doped in PMMA at 1 wt% and 10 wt%, these films were envisioned to act as temperature sensors. As shown in Figure , the afterglow gradually varies from blue to green over the temperature range of 77-295 K. Considering most COVID-19 vaccines must be stored below room temperature to be stable (Pfizer-BioNTech COVID-19: £ -70 °C; Moderna COVID-19 vaccine: ≤ -20 °C; CoronaVac vaccine-SARS-CoV-2: 2-8 °C), we demonstrate how our RTP system could be employed as a cost-effective temperature sensor for monitoring the environment temperature in real time during the ultracold chain logistics and storage (Figure ). |
6274cc87f053df3d36114962 | 9 | In summary, two RTP compounds (PXZ-Nap and PTZ-Nap) have been synthesized. Both compounds possess complex conformational dynamics in the excited state, evidenced by the observed dual RTP. In doped PMMA films, we demonstrated that RTP emission from the T 1 H state originates from thermal activation between relaxed triplet excited state conformers. The photophysical studies in doped mCP films further support the observation that emitter-matrix interactions can modulate triplet-state emission. The F RTP in mCP from T 1 H was improved by a factor of 80.5 times to 16.1% compared to the that in PMMA. We ascribe this enhancement due to the improvement of RTP efficiency as a result of Dexter energy transfer from the T 1 of mCP to that of the guest emitters. Aggregation-regulated triplet emission was demonstrated because of enhanced intermolecular interactions as observed in the photophysical behavior at higher doping concentration in PMMA. This study provides a rare yet clear picture of the triplet exciton dynamics in conformationally flexible donor-acceptor molecules and shows how RTP can be enhanced as a function of host-emitter interactions. Given that we demonstrate that the relative populations of triplet excitons can be modulated as a function of the temperature, we demonstrated that these molecules could be used as functional temperature sensors for monitoring real-time temperature for COVID-19 vaccines during the ultracold chain logistics and storage. |
67ac29c881d2151a02169df6 | 0 | As the Earth's climate continues to shift and affects living conditions, human societies face the challenge of reducing their planetary impacts . Moreover, our high-standards of living heavily rely on limited resources . Our infrastructures are predominantly made from optimized engineered materials, primarily dependent on fossil sources and unevenly distributed elements . Both climate change and resource scarcity already contribute to human migrations, loss of life, economic burdens and geopolitical tensions . |
67ac29c881d2151a02169df6 | 1 | To mitigate these upcoming consequences, efforts have intensified in the past decades . But compromising standards of living raises additional challenges . Given the large environmental impact of our material production industries, reducing their footprint is particularly significant . However, improvements in sustainability are often accompanied with trade-offs on material performance, slowing down their implementation . Biological materials open new avenues to keep producing high-performance and sustainable solutions (Fig. ). |
67ac29c881d2151a02169df6 | 2 | Biological materials, such as wood or spider silk, are often considered more sustainable than most conventional synthetics . They display high mechanical properties, such as tensile strength, toughness, elasticity, adhesion, etc., while being made from globally available and renewable resources; notably transforming wastes into resources (Fig. ). Natural materials can even achieve a zero or negative carbon footprint and their biodegradability supports a circular economy . Their production involves little to no external energy, avoids toxic chemicals, and takes place in water-based conditions . Nature's biofabrication processes for creating these materials differ significantly from conventional manufacturing . Material-producing organs contain specialized cell types with complex molecular machinery that guides material selfassembly from the nano-to the macro-scopic scales , a process we name self-microfabrication (Fig. ). These refined hierarchical microstructures, optimized through evolution , are responsible for their outstanding properties . To produce them, one can use synthetic biomimicry or recombinant approaches but difficulties remain in replicating their full complexity (Fig. ). While living organs are complex biological systems that are difficult to replicate synthetically; they offer valuable insights for conventional manufacturing methods . |
67ac29c881d2151a02169df6 | 3 | Innovations in material engineering have consistently helped to improve everyday objects and infrastructures. However, the trade-off between performance and sustainability reveals limitations in current manufacturing approaches , specifically in (a) material microstructures, (b) production processes and (c) environmental impact: (a1) Engineering complexity: biological systems present high levels of complexity, hierarchical control and automation that are hard to replicate synthetically . (a2) Many engineered materials may be highly optimized for specific properties (e.g. stiffness) but are often less versatile . (b1) Conventional microfabrication has high infrastructure costs, illustrated by semiconductor foundries . (b2) Manufacturing tools and microstructures: although current microfabrication technologies can produce complex multi-scale materials, they cannot yet produce microstructures in large materials, at scale , as nature can . (c1) Raw materials scarcity and energy costs: dependency on rare earth elements and fossil-based materials . (c2) Sustainability: optimizing properties in engineered composites often introduces trade-offs on disassembly and recyclability , while biological materials are biodegradable. (c3) Significant environmental and health impacts of conventional industrial facilities . In contrast, the versatile and competitive properties of biological materials stem from their (a) refined microstructures, (b) unique manufacturing, and (c) ecological integration , all of which could help address these limitations if successfully replicated . These hierarchical materials have mechanical properties essential in various industries ; for example wood tensile strength stems from its self-organized composite microstructure . Understanding such materials to mimic them synthetically (Fig. ) implies to first characterize their properties, hierarchical structure and self-assembly mechanisms . Second, the material-producing organ must also be studied to understand its physiology and development . Biological materials' intricate micro-architectures (a1) have been optimized through evolution for high mechanical performance and versatility (a2) (Fig. ). Organs such as the spider silk gland regulate complex processes, notably the molecular assembly of fibers . These organ production units are themselves self-organized following a developmental program. Instead of designing such highly complex systems from scratch, one can recapitulate the conditions guiding organogenesis . Starting with precise initial conditions and cell sources, highly complex tissues can self-organize and produce materials autonomously . Organoids illustrate this principle and can grow hierarchical materials, such as hair follicle organoids . These tissues, made of heterogeneous cell types differentiated from stem cells, show physiologically relevant behaviors that can mimic material-producing organs . This approach could lead to living machines , potentially produced in large numbers by leveraging self-organization . Moreover, they can grow microstructured materials at macroscopic scales, as observed in nature (b2), using abundant raw materials . The final products can be highly complex while biodegradable . Organ production units can themselves be recycled, thus reducing the environmental impact . |
67ac29c881d2151a02169df6 | 4 | Collaborative efforts in tissue engineering and material science have driven significant progress, especially in creating 3D hydrogel scaffolds and biomaterials that interface with biological tissues . Material innovations have been instrumental in advancing organoid research . Conversely, this paper explores how organoids could improve material science . Section 2 shows biological materials applications and highlights how an organ can regulate the self-assembly of materials, using spider silk as a case study. Section 3 introduces the principles of material-producing organoids, current advances and an existing proof of concept. Section 4 discusses key challenges, including organoid growth in non-model organisms and scaling up production. Section 5 presents theoretical aspects of organoid development, including tissue homeostasis strategies, the variety of developmental trajectories, and multiscale representations for modeling purposes. Section 6 proposes future directions to overcome challenges and biological limitations. Finally, Section 7 addresses ethical, societal, and technological implications, concluding with key contributions in Section 8. |
67ac29c881d2151a02169df6 | 5 | Biomimicry of biological materials and potential applications Many biological materials have been characterized for their outstanding mechanical properties such as stiffness, elasticity, toughness, hardness or adhesion . These properties stem from diverse microstructures observed across species, including fibrous, layered, helical, gradual, tubular, and latticed arrangements . To broaden the perspective, Table presents a diversity of biological materials as candidates for selfmicrofabrication, with their main properties, biological origins and potential applications. |
67ac29c881d2151a02169df6 | 6 | Preprocessing materials for external self-assembly Some materials are formed within cells, notably the hair shaft , or in a controlled extracellular space, such as spider silk glands. However, other biological materials form without requiring an organ for their production. Self-assembly can occur upon secretion in external environments, which generally simplifies biomimicry approaches , without the need for selfmicrofabrication. Yet some materials still depend on an organ for internal preprocessing prior to their secretion. For instance, the cuticle of insects is initially soft but hardens rapidly in contact with air . This chitin and protein biomolecular composite has a preformed microstructure that hardens into a tough exoskeleton through sclerotization . Catecholamines, such as dopamine derivatives, oxidize with air into quinones and crosslink cuticular proteins but the microstructure is performed by the epidermis . In other cases, materials rely on environmental conditions such as temperature changes, humidity absorption, or dehydration. Some aquatic species notably rely on their saline aqueous environment for material processing, such as barnacle adhesive , sandcastle worm cement , coral skeleton , and mussel byssus . |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.