Data-driven machine learning model for the prediction of oxygen vacancy formation energy of metal oxide materials

Abstract

Metal oxides are widely used in the fields of chemistry, physics and materials. Oxygen vacancy formation energy is a key parameter to describe the chemical, mechanical, and thermodynamic properties of...

Structure–Activity Relationships in Lewis Acid–Base Heterogeneous Catalysis

“…Although DFT is instrumental in studying heterogeneous catalysis, an accurate description of MOs electronic structural properties is challenging . A key hurdle is the direct dependency of the calculated electronic properties (e.g., OVFE) on the DFT exchange-correlation functional, noting that a real functional is still unknown. − Exchange-correlation functionals may cause erroneous self-interactions of electrons, especially in 3d and 4f metals, which contain highly localized d- and f-orbitals. One common approach to treat the self-interaction between correlated electrons is the Hubbard U model (i.e., DFT+U), , which applies an onsite Coulomb correction potential term U and an exchange term J on the localized orbitals.…”

“…Such databases can boost the application of ML algorithms in catalysis by avoiding expensive DFT calculations and offering a platform for model benchmarking. In addition, extraction of data from the literature ,, and in-house , data generation, have been also employed . For instance, Xu et al selected suitable geometric and energetic descriptors by percentile-LASSO to improve the BEP relationship of methane activation on rutile-type TMOs, based on DFT-calculated activation energies for radical and surface-stabilized mechanisms .…”

Universal machine learning framework for defect predictions in zinc blende semiconductors

“… 37 , 38 Indeed, ML has been instrumental in accelerating the prediction of properties related to point defects and dopants in materials. This includes predicting vacancy formation and substitutional energies of oxides using regression algorithms applied on DFT data, 39 , 40 , 41 , 42 ML formation energies, transition levels, and the migration energies of point defects in known semiconductors and alloys, 43 , 44 predicting the dopability of semiconductors, 45 and improving high-fidelity predictions of point defect properties using previously unknown correlations. 46 Recent work from our group involved performing high-throughput DFT computations to study the formation energies and charge transition levels of impurities in halide perovskites 3 and Cd-chalcogenides, 12 following which ML models were trained for the prediction and screening of impurity atoms that can shift the equilibrium Fermi level as determined by dominant native defects.…”

Graph neural network modeling of vacancy formation enthalpy for materials discovery and its application in solar thermochemical water splitting

“…Previous efforts to predict vacancy formation enthalpies span a Sandia National Laboratories, Livermore, California 94551, United States; E-mail: mwitman@sandia.gov, amcdani@sandia.gov various methods and material classes within which the models are applicable. Notable examples include modeling vacancy formation enthalpy using a simple hand-derived or machine learning (ML) model based on hypothesized important features [17][18][19][20][21] and descriptor derived properties to train ML regression models for defect property prediction in semiconductors 22,23 . For 2D materials consisting of TMDs, hexagonal boron nitride, and other selected wide band gap 2D materials, similar utilization of handengineered features and random forests predicted vacancy defect formation enthalpies.…”