Design of asymmetric catalysts generally involves time- and resource-intensive heuristic endeavors. In view of the steady increase in interest toward efficient catalytic asymmetric reactions and the rapid growth in the field of machine learning (ML) in recent years, we envisaged dovetailing these two important domains. We selected a set of quantum chemically derived molecular descriptors from five different asymmetric binaphthyl-derived catalyst families with the propensity to impact the enantioselectivity of asymmetric hydrogenation of alkenes and imines. The predictive power of the random forest (RF) built using the molecular parameters of a set of 368 substrate–catalyst combinations is found to be impressive, with a root-mean-square error (rmse) in the predicted enantiomeric excess (%ee) of about 8.4 ± 1.8 compared to the experimentally known values. The accuracy of RF is found to be superior to other ML methods such as convolutional neural network, decision tree, and eXtreme gradient boosting as well as stepwise linear regression. The proposed method is expected to provide a leap forward in the design of catalysts for asymmetric transformations.

Recently, nickel phosphides (Ni x P y ) have been reported to enable selective electrochemical formation of multicarbon products (C 3 and C 4 ) via the CO 2 reduction reaction (CO 2 RR); nevertheless, their activities remain low. In order to understand the roots of their high selectivity and low activity and to direct the design of more active Ni x P y -based CO 2 RR catalysts, we investigate the CO 2 RR mechanism on Ni 2 P using density functional theory (DFT) calculations. We reveal that the reaction proceeds through the formate pathway, followed by formaldehyde (H 2 CO*) formation and self-condensation. Moreover, we demonstrate that surface hydride transfer steps, along with surface-mediated C−C coupling, are essential in order to avoid C 1 product formation and boost selectivity toward multicarbon products. In addition, we find that the thermal surface hydride transfer from the surface to the physisorbed CO 2 is one of the key rate-limiting steps, and since it is not electroactive, it cannot be accelerated by applying an overpotential. Finally, our results also show that the hydrogen affinity of the surface and the dynamic surface reconstruction via H adsorption facilitate selective CO 2 reduction and C−C coupling on Ni 2 P. These findings provide an impetus for exploring materials design space to identify the physical principles that govern the thermodynamics of rate-limiting thermal steps in electrocatalytic processes.

Gaining predictable control over various forms of selectivities, such as enantio- and/or regio-selectivities, has been a long-standing goal in chemical catalysis. Although a number of factors such as the molecular features of the reactants and catalysts, as well as the reaction conditions, can influence the outcome of a reaction, it is not quite conspicuous as to what combinations of these parameters would offer a desired form of selectivity. We use machine learning tools, such as the neural network (NN), decision tree (DT), logistic regression (LR) and Random forest algorithms, to (a) analyze the outcome of an important catalytic regio-selective difluorination reaction of alkenes, and (b) decipher the complex interplay of various molecular parameters and their non-linear dependencies. The connection between what features of alkenes will yield 1,1-difluorination and how subtle changes would steer the reaction to 1,2-difluorination under identical conditions is enunciated. The NN was able to accurately predict whether a given alkene would yield a 1,1- or 1,2-difluorinated product. A combination of DT and the random forest classifier offered important chemical insights, which could be used in making a more rational choice of the reactant alkene for the desired regioisomeric product. The results could have far reaching implications in predicting which regioisomer is likely to be formed under a given set of conditions, and thus this technique is capable of expediting the development of catalytic transformations.

In this study, we propose that the curvature of graphene can be exploited to perform directional molecular motion and provide atomistic insights into the curvature-dependent molecular migration through density functional theory calculations. We first reveal the origin of the different migration trends observed experimentally for aromatic molecules with electron-donating and -withdrawing groups on p-doped functionalized graphene. Next, we show that the kinetic barrier for migration depends on the amount and nature of the curvature, that is, positive versus negative curvature. We find that the molecular migration on a wrinkled/rippled graphene sheet preferentially happens from the valley (positive curvature) to the mountain (negative curvature) regions. To understand the origin of such curvature-dependent molecular motion and migrational kinetic barrier trends, we develop a descriptor based on the frontier orbital orientation of graphene. Finally, based on these findings, we predict that time- and space-varying curvature can drive directional molecular motion on graphene and thus further propose that efforts should focus on exploring other two-dimensional materials as active platforms for performing controlled molecular motion.

The unique properties of hybrid organic–inorganic perovskites (HOIPs) promise to open doors to next-generation flexible optoelectronic devices. Before such advances are realized, a fundamental understanding of the mechanical properties of HOIPs is required. Here, we combine ab initio density functional theory (DFT) modeling with a diverse set of experiments to study the elastic properties of (quasi)­2D HOIPs. Specifically, we focus on (quasi)­2D single crystals of phenethylammonium methylammonium lead iodide, (PEA)2PbI4(MAPbI3) n−1, and their 3D counterpart, MAPbI3. We used nanoindentation (both Hertzian and Oliver–Pharr analyses) in combination with elastic buckling instability experiments to establish the out-of-plane and in-plane elastic moduli. The effect of Van der Waals (vdW) forces, different interlayer interactions, and finite temperature are combined with DFT calculations to accurately model the system. Our results reveal a nonmonotonic dependence of both the in-plane and out-of plane elastic moduli on the number of inorganic layers (n) rationalized by first-principles calculations. We discuss how the presence of defects in as-grown crystals and macroscopic interlayer deformations affect the mechanical response of (quasi)­2D HOIPs. Comparing the in- and out-of-plane experimental results with the theory reveals that perturbations to the covalent and ionic bonds (which hold a 2D layer together) is responsible for the relative out-of-plane stiffness of these materials. In contrast, we conjecture that the in-plane softness originates from macroscopic or mesoscopic motions between 2D layers during buckling experiments. Additionally, we learn how dispersion and π interactions in organic bilayers can have a determining role in the elastic response of the materials, especially in the out-of-plane direction. The understanding gained by comparing ab initio and experimental techniques paves the way for rational design of layered HOIPs with mechanical properties favorable for strain–intensive applications. Combined with filters for other favorable criteria, e.g., thermal or moisture stability, one can systematically screen viable (quasi)­2D HOIPs for a variety of flexible optoelectronic applications.

CO 2 activation is an integral component of thermocatalytic and electrocatalytic CO 2 conversion to liquid fuels and valueadded chemicals. However, the thermodynamic stability of CO 2 and the high kinetic barriers to activating CO 2 are significant bottlenecks. In this work, we propose that dual atom alloys (DAAs), homo-and heterodimer islands in a Cu matrix, can offer stronger covalent CO 2 binding than pristine Cu. The active site is designed to mimic the Ni− Fe anaerobic carbon monoxide dehydrogenase CO 2 activation environment in a heterogeneous catalyst. We find that combinations of early transition metals (TMs) and late TMs embedded in Cu are thermodynamically stable and can offer stronger covalent CO 2 binding than Cu. Additionally, we identify DAAs that have CO binding energies similar to Cu, both to avoid surface poisoning and to ensure attainable CO diffusion to Cu sites so that the C−C bond formation ability of Cu can be retained in conjunction with facile CO 2 activation at the DAA sites. Machine learning feature selection reveals that the more electropositive dopants are primarily responsible for attaining the strong CO 2 binding. We propose seven Cu-based DAAs and two single atom alloys (SAAs) with early TM late TM combinations, (Sc, Ag), (Y, Ag), (Y, Fe), (Y, Ru), (Y, Cd), (Y, Au), (V, Ag), (Sc), and (Y), for facile CO 2 activation.

Nickel phosphides (Ni x P y ) are a class of materials that are made out of earth abundant elements and have shown relatively high hydrogen evolution reaction (HER) activity. Here, we perform first-principles density functional theory (DFT) calculations to systematically investigate the stoichiometric and nonstoichiometric surface reconstructions of six different Ni x P y , i.e., Ni3P, Ni12P5, Ni2P, Ni5P4, NiP2, and NiP3, under electrochemical conditions and to illustrate the implications of such reconstructions for the catalytic activity toward HER. Our results can explain a broad range of experimental observations on the HER activity of Ni x P y in a unified framework. For the majority of cases, our protocol can closely reproduce the experimentally measured overpotential trends in the literature, which validates its usefulness in generating physical insight into the surface phenomena responsible for HER activity. We find that, among the Ni x P y studied here, Ni3P and Ni5P4 are the most active catalysts toward HER in acid, whereas Ni5P4 performs the best compared to other Ni x P y in base, in agreement with previous experimental reports. We show that P-vacancy formation in base renders the Ni-rich Ni x P y (Ni3P, Ni12P5, Ni2P, and Ni5P4) worse performers in base when compared to their activity in acid and hence propose that introducing nonmetals, which are less prone to dissolution, can improve their catalytic performance. In terms of active site design, we find Ni3 hollow sites bind H too strongly and surface P sites with P–Ni bonds bind H too weakly. On the other hand, we identify that surface P sites with P–P bonds offer the best catalytic performances, and therefore, we predict that active site engineering to maximize the abundance of such surface motifs can further improve the HER activity. Moreover, we unravel the nature of H binding across the material class for different binding motifs via electronic structure theory analysis. The chemical insight we provide in this work can help rationalize the search for materials composed of inexpensive earth abundant elements that provide improved HER catalytic activity.

In mechanochemistry, the application of controlled forces is key to altering reaction rates and pathways to direct product yields and selectivity. However, a fundamental knowledge gap exists between what is occurring on the atomic scale in mechanically driven reactions and the resulting macroscale outcomes. Two-dimensional (2D) materials, such as graphene, proffer a model system to study the impact of mechanical forces, such as strain, on chemical reactivity, as force distributions may be applied across a well-organized atomic-scale structure comprising a single layer of C atoms. Here, using Raman micro-spectroscopy and first-principles calculations, we have investigated the reaction of graphene, under varying degrees of strain, with 4-nitrobenzenediazonium tetrafluoroborate (4-NBD). We find that only with increased out-of-plane distortion (shifting the C atoms of graphene from sp 2 toward sp 3 electronic states) would the reactivity be increased, with larger out-of-plane distortions yielding greater reactivity. Density functional theory (DFT) calculations reveal that increasing the curvature of graphene decreases the activation barrier of 4-NBD functionalization and enhances the thermodynamic favorability of the reaction. Furthermore, we find that curvature affects the orientation of the graphene 2p z orbitals, and we then relate the thermodynamic feasibility of 4-NBD functionalization with the orbital orientation. These studies point to how the precise application of forces can be used to direct the functionalization of graphene for C−C bond forming reactions, which has significant implications for controlling its corresponding electronic structure in a well-defined fashion.