3DReact: Geometric Deep Learning for Chemical Reactions.

Geometric deep learning models, which incorporate the relevant molecular symmetries within the neural network architecture, have considerably improved the accuracy and data efficiency of predictions of molecular properties. Building on this success, we introduce 3DReact, a geometric deep learning model to predict reaction properties from three-dimensional structures of reactants and products. We demonstrate that the invariant version of the model is sufficient for existing reaction data sets. We illustrate its competitive performance on the prediction of activation barriers on the GDB7-22-TS, Cyclo-23-TS, and Proparg-21-TS data sets in different atom-mapping regimes. We show that, compared to existing models for reaction property prediction, 3DReact offers a flexible framework that exploits atom-mapping information, if available, as well as geometries of reactants and products (in an invariant or equivariant fashion). Accordingly, it performs systematically well across different data sets, atom-mapping regimes, as well as both interpolation and extrapolation tasks.

Microkinetic Molecular Volcano Plots for Enhanced Catalyst Selectivity and Activity Predictions.

Molecular volcano plots, which facilitate the rapid prediction of the activity and selectivity of prospective catalysts, have emerged as powerful tools for computational catalysis. Here, we integrate microkinetic modeling into the volcano plot framework to develop "microkinetic molecular volcano plots". The resulting unified computational framework allows the influence of important reaction parameters, including temperature, reaction time, and concentration, to be quickly incorporated and more complex situations, such as off-cycle resting states and coupled catalytic cycles, to be tackled. Compared to previous generations of molecular volcanoes, these microkinetic counterparts offer a more comprehensive understanding of catalytic behavior, in which selectivity and product ratios can be explicitly determined by tracking the evolution of each product concentration over time. This is demonstrated by examining two case studies, rhodium-catalyzed hydroformylation and metal-catalyzed hydrosilylation, in which the unique insights provided by microkinetic modeling, as well as the ability to simultaneously screen catalysts and reaction conditions, are highlighted. To facilitate the construction of these plots/maps, we introduce mikimo, a Python program that seamlessly integrates with our previously developed automated volcano builder, volcanic.

Overcoming the Pitfalls of Computing Reaction Selectivity from Ensembles of Transition States.

The prediction of reaction selectivity is a challenging task for computational chemistry, not only because many molecules adopt multiple conformations but also due to the exponential relationship between effective activation energies and rate constants. To account for molecular flexibility, an increasing number of methods exist that generate conformational ensembles of transition state (TS) structures. Typically, these TS ensembles are Boltzmann weighted and used to compute selectivity assuming Curtin-Hammett conditions. This strategy, however, can lead to erroneous predictions if the appropriate filtering of the conformer ensembles is not conducted. Here, we demonstrate how any possible selectivity can be obtained by processing the same sets of TS ensembles for a model reaction. To address the burdensome filtering task in a consistent and automated way, we introduce marc, a tool for the modular analysis of representative conformers that aids in avoiding human errors while minimizing the number of reoptimization computations needed to obtain correct reaction selectivity.

Reaction-Agnostic Featurization of Bidentate Ligands for Bayesian Ridge Regression of Enantioselectivity.

Chiral ligands are important components in asymmetric homogeneous catalysis, but their synthesis and screening can be both time-consuming and resource-intensive. Data-driven approaches, in contrast to screening procedures based on intuition, have the potential to reduce the time and resources needed for reaction optimization by more rapidly identifying an ideal catalyst. These approaches, however, are often nontransferable and cannot be applied across different reactions. To overcome this drawback, we introduce a general featurization strategy for bidentate ligands that is coupled with an automated feature selection pipeline and Bayesian ridge regression to perform multivariate linear regression modeling. This approach, which is applicable to any reaction, incorporates electronic, steric, and topological features (rigidity/flexibility, branching, geometry, and constitution) and is well-suited for early stage ligand optimization. Using only small data sets, our workflow capably predicts the enantioselectivity of four metal-catalyzed asymmetric reactions. Uncertainty estimates provided by Bayesian ridge regression permit the use of Bayesian optimization to efficiently explore pools of prospective ligands. Finally, we constructed the BDL-Cu-2023 data set, composed of 312 bidentate ligands extracted from the Cambridge Structural Database, and screened it with this procedure to identify ligand candidates for a challenging asymmetric oxy-alkynylation reaction.

Engineering Frustrated Lewis Pair Active Sites in Porous Organic Scaffolds for Catalytic CO2 Hydrogenation.

Frustrated Lewis pairs (FLPs), featuring reactive combinations of Lewis acids and Lewis bases, have been utilized for myriad metal-free homogeneous catalytic processes. Immobilizing the active Lewis sites to a solid support, especially to porous scaffolds, has shown great potential to ameliorate FLP catalysis by circumventing some of its inherent drawbacks, such as poor product separation and catalyst recyclability. Nevertheless, designing immobilized Lewis pair active sites (LPASs) is challenging due to the requirement of placing the donor and acceptor centers in appropriate geometric arrangements while maintaining the necessary chemical environment to perform catalysis, and clear design rules have not yet been established. In this work, we formulate simple guidelines to build highly active LPASs for direct catalytic hydrogenation of CO2 through a large-scale screening of a diverse library of 25,000 immobilized FLPs. The library is built by introducing boron-containing acidic sites in the vicinity of the existing basic nitrogen sites of the organic linkers of metal-organic frameworks collected in a "top-down" fashion from the CoRE MOF 2019 database. The chemical and geometrical appropriateness of these LPASs for CO2 hydrogenation is determined by evaluating a series of simple descriptors representing the intrinsic strength (acidity and basicity) of the components and their spatial arrangement in the active sites. Analysis of the leading candidates enables the formulation of pragmatic and experimentally relevant design principles which constitute the starting point for further exploration of FLP-based catalysts for the reduction of CO2.

Benchmarking machine-readable vectors of chemical reactions on computed activation barriers.

In recent years, there has been a surge of interest in predicting computed activation barriers, to enable the acceleration of the automated exploration of reaction networks. Consequently, various predictive approaches have emerged, ranging from graph-based models to methods based on the three-dimensional structure of reactants and products. In tandem, many representations have been developed to predict experimental targets, which may hold promise for barrier prediction as well. Here, we bring together all of these efforts and benchmark various methods (Morgan fingerprints, the DRFP, the CGR representation-based Chemprop, SLATMd, B2Rl2, EquiReact and language model BERT + RXNFP) for the prediction of computed activation barriers on three diverse datasets.

A genetic optimization strategy with generality in asymmetric organocatalysis as a primary target.

A catalyst possessing a broad substrate scope, in terms of both turnover and enantioselectivity, is sometimes called "general". Despite their great utility in asymmetric synthesis, truly general catalysts are difficult or expensive to discover via traditional high-throughput screening and are, therefore, rare. Existing computational tools accelerate the evaluation of reaction conditions from a pre-defined set of experiments to identify the most general ones, but cannot generate entirely new catalysts with enhanced substrate breadth. For these reasons, we report an inverse design strategy based on the open-source genetic algorithm NaviCatGA and on the OSCAR database of organocatalysts to simultaneously probe the catalyst and substrate scope and optimize generality as a primary target. We apply this strategy to the Pictet-Spengler condensation, for which we curate a database of 820 reactions, used to train statistical models of selectivity and activity. Starting from OSCAR, we define a combinatorial space of millions of catalyst possibilities, and perform evolutionary experiments on a diverse substrate scope that is representative of the whole chemical space of tetrahydro-ß-carboline products. While privileged catalysts emerge, we show how genetic optimization can address the broader question of generality in asymmetric synthesis, extracting structure-performance relationships from the challenging areas of chemical space.

From Organic Fragments to Photoswitchable Catalysts: The OFF-ON Structural Repository for Transferable Kernel-Based Potentials.

Structurally and conformationally diverse databases are needed to train accurate neural networks or kernel-based potentials capable of exploring the complex free energy landscape of flexible functional organic molecules. Curating such databases for species beyond "simple" drug-like compounds or molecules composed of well-defined building blocks (e.g., peptides) is challenging as it requires thorough chemical space mapping and evaluation of both chemical and conformational diversities. Here, we introduce the OFF-ON (organic fragments from organocatalysts that are non-modular) database, a repository of 7869 equilibrium and 67,457 nonequilibrium geometries of organic compounds and dimers aimed at describing conformationally flexible functional organic molecules, with an emphasis on photoswitchable organocatalysts. The relevance of this database is then demonstrated by training a local kernel regression model on a low-cost semiempirical baseline and comparing it with a PBE0-D3 reference for several known catalysts, notably the free energy surfaces of exemplary photoswitchable organocatalysts. Our results demonstrate that the OFF-ON data set offers reliable predictions for simulating the conformational behavior of virtually any (photoswitchable) organocatalyst or organic compound composed of H, C, N, O, F, and S atoms, thereby opening a computationally feasible route to explore complex free energy surfaces in order to rationalize and predict catalytic behavior.

SPAHM(a,b): Encoding the Density Information from Guess Hamiltonian in Quantum Machine Learning Representations.

Recently, we introduced a class of molecular representations for kernel-based regression methods─the spectrum of approximated Hamiltonian matrices (SPAHM)─that takes advantage of lightweight one-electron Hamiltonians traditionally used as a self-consistent field initial guess. The original SPAHM variant is built from occupied-orbital energies (i.e., eigenvalues) and naturally contains all of the information about nuclear charges, atomic positions, and symmetry requirements. Its advantages were demonstrated on data sets featuring a wide variation of charge and spin, for which traditional structure-based representations commonly fail. SPAHM(a,b), as introduced here, expand the eigenvalue SPAHM into local and transferable representations. They rely upon one-electron density matrices to build fingerprints from atomic and bond density overlap contributions inspired from preceding state-of-the-art representations. The performance and efficiency of SPAHM(a,b) is assessed on the predictions for data sets of prototypical organic molecules (QM7) of different charges and azoheteroarene dyes in an excited state. Overall, both SPAHM(a) and SPAHM(b) outperform state-of-the-art representations on difficult prediction tasks such as the atomic properties of charged open-shell species and of π-conjugated systems.

Enhanced inverted singlet-triplet gaps in azaphenalenes and non-alternant hydrocarbons.

Inverted singlet-triplet gaps may lead to novel molecular emitters if a rational design approach can be achieved. We uncover a substituent strategy that enables tuning of the gap and succeed in inducing inversion in near-gapless molecules. Based on known inverted-gap emitters, we design substituted analogs with even more negative singlet-triplet gaps than in the parent systems. The inversion is lost if the reverse substituent-strategy is used. We thus demonstrate a definite set of conceptual design rules for inverted gap molecules.

Data-Driven Discovery of Organic Electronic Materials Enabled by Hybrid Top-Down/Bottom-Up Design.

The high-throughput exploration and screening of molecules for organic electronics involves either a 'top-down' curation and mining of existing repositories, or a 'bottom-up' assembly of user-defined fragments based on known synthetic templates. Both are time-consuming approaches requiring significant resources to compute electronic properties accurately. Here, 'top-down' is combined with 'bottom-up' through automatic assembly and statistical models, thus providing a platform for the fragment-based discovery of organic electronic materials. This study generates a top-down set of 117K synthesized molecules containing structures, electronic and topological properties and chemical composition, and uses them as building blocks for bottom-up design. A tool is developed to automate the coupling of these building blocks at their C(sp2/sp)-H bonds, providing a fundamental link between the two dataset construction philosophies. Statistical models are trained on this dataset and a subset of resulting top-down/bottom-up compounds, enabling on-the-fly prediction of ground and excited state properties with high accuracy across organic compound space. With access to ab initio-quality optical properties, this bottom-up pipeline may be applied to any materials design campaign using existing compounds as building blocks. To illustrate this, over a million molecules are screened for singlet fission. tThe leading candidates provide insight into the features promoting this multiexciton-generating process.

Toward in silico Catalyst Optimization.

In this minireview, we overview a computational pipeline developed within the framework of NCCR Catalysis that can be used to successfully reproduce the enantiomeric ratios of homogeneous catalytic reactions. At the core of this pipeline is the SCINE Molassembler module, a graph-based software that provides algorithms for molecular construction of all periodic table elements. With this pipeline, we are able to simultaneously functionalizenand generate ensembles of transition state conformers, which permits facile exploration of the influencenof various substituents on the overall enantiomeric ratio. This allows preconceived back-of-the-envelope designnmodels to be tested and subsequently refined by providing quick and reliable access to energetically low-lyingntransition states, which represents a key step in undertaking in silico catalyst optimization.

Genetic Algorithms for the Discovery of Homogeneous Catalysts.

In this account, we discuss the use of genetic algorithms in the inverse design process of homogeneous catalysts for chemical transformations. We describe the main components of evolutionary experiments, specifically the nature of the fitness function to optimize, the library of molecular fragments from which potential catalysts are assembled, and the settings of the genetic algorithm itself. While not exhaustive, this review summarizes the key challenges and characteristics of our own (i.e., NaviCatGA) and other GAs for the discovery of new catalysts.

Double-bond delocalization in non-alternant hydrocarbons induces inverted singlet-triplet gaps.

Molecules where the first excited singlet state is lower in energy than the first excited triplet state have the potential to revolutionize OLEDs. This inverted singlet-triplet gap violates Hund's rule and currently there are only a few molecules which are known to have this property. Here, we screen the complete set of non-alternant hydrocarbons consisting of 5-, 6-, 7-membered rings fused into two-, three- and four-ring polycyclic systems. We identify several molecules where the symmetry of the ground-state structure is broken due to bond-length alternation. Through symmetry-constrained optimizations we identify several molecular cores where the singlet-triplet gap is inverted when the structure is in a higher symmetry, pentalene being a known example. We uncover a strategy to stabilize the molecular cores into their higher-symmetry structures with electron donors or acceptors. We design several substituted pentalenes, s-indacenes, and indeno[1,2,3-ef]heptalenes with inverted gaps, among which there are several synthetically known examples. In contrast to known inverted gap emitters, we identify the double-bond delocalized structure of their conjugated cores as the necessary condition to achieve the inverted gap. This strategy enables chemical tuning and paves the way for the rational design of polycyclic hydrocarbons with inverted singlet-triplet gaps. These molecules are prospective emitters if their properties can be optimized for use in OLEDs.

The fundamental relation between electrohelicity and molecular optical activity.

Electrohelicity arises in molecules such as allene and spiropentadiene when their symmetry is reduced and helical frontier molecular orbitals (MOs) appear. Such molecules are optically active and electrohelicity has been suggested as a possible design principle for increasing the chiroptical response. Here we examine the fundamental link between electrohelicity and optical activity by studying the origin of the electric and magnetic transition dipole moments of the π-π* transitions. We show that the helical character of the MOs drives the optical activity in allene, and we use this knowledge to design allenic molecules with increased chiroptical response. We further examine longer carbyne-like molecules. While the MO helicity also contributes to the optical activity in non-planar butatriene, the simplest cumulene, we show there is no relation between the chiroptical response and the helical π-MOs of tolane, a simple polyyne. Finally, we demonstrate that the optical activity of spiropentadiene is inherently linked to mixing of its two π-systems rather than the helical shape of its occupied π-MOs. We thus find that the fundamental connection between electrohelicity and optical activity is very molecule dependent. Although electrohelicity is not the underlying principle, we show that the chiroptical response can be enhanced through insight into the helical nature of electronic transitions.

Theory-guided development of homogeneous catalysts for the reduction of CO2 to formate, formaldehyde, and methanol derivatives.

The stepwise catalytic reduction of carbon dioxide (CO2) to formic acid, formaldehyde, and methanol opens non-fossil pathways to important platform chemicals. The present article aims at identifying molecular control parameters to steer the selectivity to the three distinct reduction levels using organometallic catalysts of earth-abundant first-row metals. A linear scaling relationship was developed to map the intrinsic reactivity of 3d transition metal pincer complexes to their activity and selectivity in CO2 hydrosilylation. The hydride affinity of the catalysts was used as a descriptor to predict activity/selectivity trends in a composite volcano picture, and the outstanding properties of cobalt complexes bearing bis(phosphino)triazine PNP-type pincer ligands to reach the three reduction levels selectively under different reaction conditions could thus be rationalized. The implications of the composite volcano picture were successfully experimentally validated with selected catalysts, and the challenging intermediate level of formaldehyde could be accessed in over 80% yield with the cobalt complex 6. The results underpin the potential of tandem computational-experimental approaches to propel catalyst design for CO2-based chemical transformations.

Symmetry-Induced Singlet-Triplet Inversions in Non-Alternant Hydrocarbons.

Molecules with inversion of the singlet and triplet excited-state energies are highly promising for the development of organic light-emitting diodes (OLEDs). To date, azaphenalenes are the only class of molecules where these inversions have been identified. Here, we screen a curated database of organic crystal structures to identify existing compounds for violations of Hund's rule in the lowest excited states. We identify two further classes with this behavior. The first, a class of zwitterions, has limited relevance to molecular emitters as the singlet-triplet inversions occur in the third excited singlet state. The second class consists of two D2h -symmetry non-alternant hydrocarbons, a fused azulene dimer and a bicalicene, whose lowest excited singlet states violate Hund's rule. Due to the connectivity of the polycyclic structure, they achieve this symmetry through aromatic stabilization. These hydrocarbons show promise as the next generation of building blocks for OLED emitters.

OSCAR: an extensive repository of chemically and functionally diverse organocatalysts.

The automated construction of datasets has become increasingly relevant in computational chemistry. While transition-metal catalysis has greatly benefitted from bottom-up or top-down strategies for the curation of organometallic complexes libraries, the field of organocatalysis is mostly dominated by case-by-case studies, with a lack of transferable data-driven tools that facilitate both the exploration of a wider range of catalyst space and the optimization of reaction properties. For these reasons, we introduce OSCAR, a repository of 4000 experimentally derived organocatalysts along with their corresponding building blocks and combinatorially enriched structures. We outline the fragment-based approach used for database generation and showcase the chemical diversity, in terms of functions and molecular properties, covered in OSCAR. The structures and corresponding stereoelectronic properties are publicly available (https://archive.materialscloud.org/record/2022.106) and constitute the starting point to build generative and predictive models for organocatalyst performance.

Helical versus linear Jahn-Teller distortions in allene and spiropentadiene radical cations.

The allene radical cation can be stabilized both by Jahn-Teller distortion of the bond lengths and by torsion of the end-groups. However, only the latter happens and the allene radical cation relaxes into a twisted D2 symmetry structure with equal double-bond lengths. Here we revisit the Jahn-Teller distortion of allene and spiropentadiene by assessing the possible implications of their helical π-systems in the radical cations. We describe a general relation between the structure and the number of π-electrons in spiroconjugated and linearly conjugated systems. Through constrained optimizations we compare the stabilization achieved by bond-length alternation and axial torsion in the radical cations, which we explain with a simple frontier molecular orbital (MO) picture. While structurally different, allene and spiropentadiene have similar helical frontier MOs. Both cations relax through torsion because the stabilization of their helical frontier MOs is bigger than that which can be achieved by linear π-conjugation. Electrohelicity thus manifests in molecular systems with partial occupation as a helical π-conjugation effect, which evidently provides more stabilization than its linear counterpart in terms of the Jahn-Teller distortion. This mechanism may be a driving factor for the relaxation in a range of spiroconjugated and linearly conjugated cationic systems.

Design of Frustrated Lewis Pair Catalysts for Direct Hydrogenation of CO2.

Despite recent progress in the chemistry of frustrated Lewis pairs (FLPs), direct FLP-catalyzed hydrogenation of CO2 remains elusive. From a near-infinite array of plausible Lewis pairs, it is challenging to identify individual combinations that are appropriate for catalyzing this reaction. To this end, we propose a mapping of the chemical composition of FLPs to their activity towards direct catalytic hydrogenation of CO2 into formate. The maps, built upon linear scaling relationships, pinpoint specific FLP combinations with the proper complementary acidity and basicity to optimally balance the energetics of the catalytic cycle. One such combination was experimentally validated to achieve hitherto unreported catalytic turnover for this transformation.