Enumeration of de novo inorganic complexes for chemical discovery and machine learning

Abstract

Enumerated, de novo transition metal complexes have unique spin state properties and accelerate machine learning model training.

Machine learning to tame divergent density functional approximations: a new path to consensus materials design principles

“…We employ two subsets of data, as curated in prior work 60 from five prior studies [33][34][35]107,115 that originally corresponded to a total of 2,828 mononuclear octahedral transition-metal complexes in equilibrium geometries obtained with gas-phase density functional theory (DFT). In comparison to the prior curation 60 (i.e., where the sets were referred to as MD1 and OHLDB), we refined the data further by de-duplicating structures with identical molecular graph, charge, and spin state across the two sets.…”

“…Details of all complexes are provided in the ESI. As in the original studies [33][34][35]107,115 For all DFT geometry optimizations carried out in the original work, TeraChem 116 , as automated by molSimplify 117,118 and molSimplify automatic design (mAD) 107 , was employed.…”

“…In our accelerated workflow, we used the previously converged [33][34][35]107,115 B3LYP/LACVP* TeraChem wavefunction molecular orbital coefficients as an initial guess for a Psi4 B3LYP/LACVP* SCF calculation. The converged Psi4 B3LYP/LACVP* wavefunction was used as an initial guess to obtain self-consistent single-point energies in Psi4 for all other functionals with LACVP* basis set using the recommended default grid size 120,121 with 99 radial points and 590 spherical points (ESI Fig.…”

Electron configuration-based neural network model to predict physicochemical properties of inorganic compounds

“…Table S13: electron conguration vector for MP dataset. structure properties (quantum-mechanically-derived properties), [12][13][14] crystallographic parameters, 15,16 or material synthesis parameters. 17,18 If prediction models for nanomaterials are excluded from the scope of discussion, no models for inorganic compounds were developed to predict endpoints signicant in a regulatory perspective except one which predicts inorganic toxicity of substances toward rats.…”

ChemSpaX: Exploration of Chemical Space by Automated Functionalization of Molecular Scaffold

“…TMCs are often used as bio-inspired homogeneous catalysts which account for over 15% of all industrial catalytic processes and enable key catalytic transformations such as for pharmaceuticals, fine chemicals, and energy applications [22,23,24,25,26,27,28,29]. Recent studies have revealed the promise of data-driven quantum chemical methods to understand structure-function relations in TMCs [30,31,32,33,34,35]. Availability of structure-property data from QM calculations and/or experiments is central to the success of data-driven chemical approaches.…”