We present an automated, open source toolkit for the first-principles screening and discovery of new inorganic molecules and intermolecular complexes. Challenges remain in the automatic generation of candidate inorganic molecule structures due to the high variability in coordination and bonding, which we overcome through a divide-and-conquer tactic that flexibly combines force-field preoptimization of organic fragments with alignment to first-principles-trained metal-ligand distances. Exploration of chemical space is enabled through random generation of ligands and intermolecular complexes from large chemical databases. We validate the generated structures with the root mean squared (RMS) gradients evaluated from density functional theory (DFT), which are around 0.02 Ha/au across a large 150 molecule test set. Comparison of molSimplify results to full optimization with the universal force field reveals that RMS DFT gradients are improved by 40%. Seamless generation of input files, preparation and execution of electronic structure calculations, and post-processing for each generated structure aids interpretation of underlying chemical and energetic trends. © 2016 Wiley Periodicals, Inc.

Zeolites are porous, aluminosilicate materials with many industrial and “green” applications. Despite their industrial relevance, many aspects of zeolite synthesis remain poorly understood requiring costly trial and error synthesis. In this paper, we create natural language processing techniques and text markup parsing tools to automatically extract synthesis information and trends from zeolite journal articles. We further engineer a data set of germanium-containing zeolites to test the accuracy of the extracted data and to discover potential opportunities for zeolites containing germanium. We also create a regression model for a zeolite’s framework density from the synthesis conditions. This model has a cross-validated root mean squared error of 0.98 T/1000 Å 3 , and many of the model decision boundaries correspond to known synthesis heuristics in germanium-containing zeolites. We propose that this automatic data extraction can be applied to many different problems in zeolite synthesis and enable novel zeolite morphologies.

Computational high-throughput screening is an essential tool for catalyst design, limited primarily by the efficiency with which accurate predictions can be made. In bulk heterogeneous catalysis, linear free energy relationships (LFERs) have been extensively developed to relate elementary step activation energies, and thus overall catalytic activity, back to the adsorption energies of key intermediates, dramatically reducing the computational cost of screening. The applicability of these LFERs to single-site catalysts remains unclear, owing to the directional, covalent metal-ligand bonds and the broader chemical space of accessible ligand scaffolds. Through a computational screen of nearly 500 model Fe(II) complexes for CH 4 hydroxylation, we observe that 1) tuning ligand field strength yields LFERs by comparably shifting energetics of the metal 3d levels that govern stability of different intermediates and 2) distortion of the metal coordination geometry breaks these LFERs by increasing the splitting between the d xz /d yz and d z 2 metal states that govern reactivity. Thus, in single site catalysts, low Brønsted-Evans-Polanyi slopes for oxo formation, which would limit peak turnover frequency achievable through ligand field tuning alone, can be overcome through structural distortions achievable in experimentally characterized compounds. Observations from this screen also motivate the placement of strong HB donors in targeted positions as a scaffold-agnostic strategy for further activity improvement. More generally, our findings motivate broader variation of coordination geometries in reactivity studies with single-site catalysts.

Virtual high throughput screening, typically driven by first-principles, density functional theory calculations, has emerged as a powerful tool for the discovery of new materials. Although the computational materials science community has benefited from open source tools for the rapid structure generation, calculation, and analysis of crystalline inorganic materials, software and strategies to address the unique challenges of inorganic complex discovery have not been as widely available. We present a unified view of our recent developments in the open source molSimplify code for inorganic discovery. Building on our previous efforts in the automated generation of highly accurate inorganic molecular structures, first-principles simulation, and property analysis to accelerate high-throughput screening, we have recently incorporated a neural network that both improves structure generation and predicts electronic properties prior to first-principles calculation. We also provide an overview of how multi-million molecule organic libraries can be leveraged for inorganic discovery alongside cheminformatics concepts of molecular diversity in order to efficiently traverse chemical space. We demonstrate all of these tools on the discovery of design rules for octahedral Fe(II/III) redox couples with nitrogen ligands. Over a search of only approximately 40 new molecules, we obtain redox potentials relative to the Fc/Fc + couple ranging from-1 to 4.5 V in aqueous solution. Our new automated correlation analysis reveals heteroatom identity and the degree of structural branching to be key ligand descriptors in determining redox potential. This inorganic discovery toolkit provides a promising approach to advancing transition metal complex design.

Approximate density functional theory (DFT) suffers from many-electron selfinteraction error, otherwise known as delocalization error, that may be diagnosed and then corrected through elimination of the deviation from exact piecewise linear behavior between integer electron numbers. Although paths to correction of energetic delocalization error are wellestablished, the impact of these corrections on the electron density is less well-studied. Here, we compare the effect on density delocalization of DFT+U, global hybrid tuning, and rangeseparated hybrid tuning on a diverse test set of 32 transition metal complexes and observe the three methods to have qualitatively equivalent effects on the ground state density. Regardless of valence orbital diffuseness (i.e., from 2p to 5p), ligand electronegativity (i.e., from Al to O), basis set (i.e., plane wave versus localized basis set), metal (i.e., Ti, Fe, Ni) and spin state, or tuning method, we consistently observe substantial charge loss at the metal and gain at ligand atoms (ca. 0.3-0.5 e or more). This charge loss at the metal is preferentially from the minority spin, leading to increasing magnetic moment as well. Using accurate wavefunction theory references, we observe that a minimum error in partial charges and magnetic moments occur at higher tuning parameters than typically employed to eliminate energetic delocalization error. These observations motivate the need to develop multi-faceted approximate-DFT error correction approaches that separately treat density delocalization and energetic errors in order to recover both correct density and magnetization properties.

Accurate predictions of spin-state ordering, reaction energetics, and barrier heights are critical for the computational discovery of open-shell transition-metal (TM) catalysts. Semilocal approximations in density functional theory, such as the generalized gradient approximation (GGA), suffer from delocalization error that causes them to overstabilize strongly bonded states. Descriptions of energetics and bonding are often improved by introducing a fraction of exact exchange (e.g., erroneous low-spin GGA ground states are instead correctly predicted as high-spin with a hybrid functional). The degree of spin-splitting sensitivity to exchange can be understood based on the chemical composition of the complex, but the effect of exchange on reaction energetics within a single spin state is less well-established. Across a number of model iron complexes, we observe strong exchange sensitivities of reaction barriers and energies that are of the same magnitude as those for spin splitting energies. We rationalize trends in both reaction and spin energetics by introducing a measure of delocalization, the bond valence of the metal-ligand bonds in each complex. The bond valence thus represents a simple-to-compute property that unifies understanding of exchange sensitivity for catalytic properties and spin-state ordering in TM complexes. Close agreement of the resulting per-metal-organic-bond sensitivity estimates, together with failure of alternative descriptors demonstrates the utility of the bond valence as a robust descriptor of how differences in metal-ligand delocalization produce differing relative energetics with exchange tuning. Our unified description explains the overall effect of exact exchange tuning on the paradigmatic two-state FeO/CH reaction that combines challenges of spin-state and reactivity predictions. This new descriptor-sensitivity relationship provides a path to quantifying how predictions in transition-metal complex screening are sensitive to the method used.

<p>Computational high-throughput screening is an essential tool for catalyst design, limited primarily by the efficiency with which accurate predictions can be made. In bulk heterogeneous catalysis, linear free energy relationships (LFERs) have been extensively developed to relate elementary step activation energies, and thus overall catalytic activity, back to the adsorption energies of key intermediates, dramatically reducing the computational cost of screening. The applicability of these LFERs to single-site catalysts remains unclear, owing to the directional, covalent metal-ligand bonds and the broader chemical space of accessible ligand scaffolds. Through a computational screen of nearly 500 model Fe(II) complexes for CH₄ hydroxylation, we observe that 1) tuning ligand field strength yields LFERs by comparably shifting energetics of the metal 3<i>d</i> levels that govern stability of different intermediates and 2) distortion of the metal coordination geometry breaks these LFERs by increasing the splitting between the <i>d</i>_xz/<i>d</i>_yz and <i>d</i>_z² metal states that govern reactivity. Thus, in single site catalysts, low Brønsted-Evans-Polanyi slopes for oxo formation, which would limit peak turnover frequency achievable through ligand field tuning alone, can be overcome through structural distortions achievable in experimentally characterized compounds. Observations from this screen also motivate the placement of strong HB donors in targeted positions as a scaffold-agnostic strategy for further activity improvement. More generally, our findings motivate broader variation of coordination geometries in reactivity studies with single-site catalysts.</p>

been described as a key metric for understanding the effects of global warming due to its direct impact on climate change. [2] Extensive modeling with energy-economyenvironment scenarios or projections to keep the global temperature from rising above 1.5 °C by the year 2100 shows that such a positive outlook is only possible if the global energy system is completely decarbonized (i.e., net-zero global CO 2 emissions) by mid-century, followed by active CO 2 removal (i.e., carbon-negative) in the second half of the century. [3] The catalytic conversion of CO 2 to valuable energy-related products (e.g., syngas, CH 4 , CH 3 OH, and longer-chain hydrocarbons) [4] by thermal reforming and hydrogenation, [5][6][7][8][9][10][11] electrocatalysis, [12][13][14] or photocatalysis [15] could occupy an important position in a future carbon-negative economy by directly replacing nonrenewable sources of these molecules, provided that the energy inputs to these processes are themselves derived from renewable sources. [16,17] In this regard, photocatalytic strategies stand out by not requiring a secondary medium of energy storage, and being able to realize the CO 2 conversion into high value-added fuels directly via renewable solar energy without external energy input. [18] Indeed, photocatalytic CO 2 reduction has been considered to be a "kill two birds with one stone" approach for sustainable energy production and greenhouse gas reduction. [19] In contrast, thermocatalytic approachesThe solar-energy-driven photoreduction of CO 2 has recently emerged as a promising approach to directly transform CO 2 into valuable energy sources under mild conditions. As a clean-burning fuel and drop-in replacement for natural gas, CH 4 is an ideal product of CO 2 photoreduction, but the development of highly active and selective semiconductor-based photocatalysts for this important transformation remains challenging. Hence, significant efforts have been made in the search for active, selective, stable, and sustainable photocatalysts. In this review, recent applications of cutting-edge experimental and computational materials design strategies toward the discovery of novel catalysts for CO 2 photocatalytic conversion to CH 4 are systematically summarized. First, insights into effective experimental catalyst engineering strategies, including heterojunctions, defect engineering, cocatalysts, surface modification, facet engineering, and single atoms, are presented. Then, datadriven photocatalyst design spanning density functional theory (DFT) simulations, high-throughput computational screening, and machine learning (ML) is presented through a step-by-step introduction. The combination of DFT, ML, and experiments is emphasized as a powerful solution for accelerating the discovery of novel catalysts for photocatalytic reduction of CO 2 . Last, challenges and perspectives concerning the interplay between experiments and data-driven rational design strategies for the industrialization of large-scale CO 2 photoreduction technologies are described.