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

Metal-support interactions in metal oxide-supported atomic, cluster, and nanoparticle catalysis

Supported metal catalysts are essential to a plethora of processes in the chemical industry. The overall performance of these catalysts depends strongly on the interaction of adsorbates at the atomic level, which can be manipulated and controlled by the different constituents of the active material (i.e., support and active metal). The description of catalyst activity and the relationship between active constituent and the support, or metal-support interactions (MSI), in heterogeneous (thermo)catalysts is a complex phenomenon with multivariate (dependent and independent) contributions that are difficult to disentangle, both experimentally and theoretically. So-called "strong metal-support interactions" have been reported for several decades and summarized in excellent review articles. However, in recent years, there has been a proliferation of new findings related to atomically dispersed metal sites, metal oxide defects, and, for example, the generation and evolution of MSI under reaction conditions, which has led to the designation of (sub)classifications of MSI deserving to be critically and systematically evaluated. These include dynamic restructuring under alternating redox and reaction conditions, adsorbate-induced MSI, and evidence of strong interactions in oxide-supported metal oxide catalysts. Here, we review recent literature on MSI in oxide-supported metal particles to provide an up-to-date understanding of the underlying physicochemical principles that dominate the observed effects in supported metal atomic, cluster, and nanoparticle catalysts. Critical evaluation of different subclassifications of MSI is provided, along with discussions on the formation mechanisms, theoretical and characterization advances, and tuning strategies to manipulate catalytic reaction performance. We also provide a perspective on the future of the field, and we discuss the analysis of different MSI effects on catalysis quantitatively.

Defect graph neural networks for materials discovery in high-temperature clean-energy applications

We present a graph neural network approach that fully automates the prediction of defect formation enthalpies for any crystallographic site from the ideal crystal structure, without the need to create defected atomic structure models as input. Here we used density functional theory reference data for vacancy defects in oxides, to train a defect graph neural network (dGNN) model that replaces the density functional theory supercell relaxations otherwise required for each symmetrically unique crystal site. Interfaced with thermodynamic calculations of reduction entropies and associated free energies, the dGNN model is applied to the screening of oxides in the Materials Project database, connecting the zero-kelvin defect enthalpies to high-temperature process conditions relevant for solar thermochemical hydrogen production and other energy applications. The dGNN approach is applicable to arbitrary structures with an accuracy limited principally by the amount and diversity of the training data, and it is generalizable to other defect types and advanced graph convolution architectures. It will help to tackle future materials discovery problems in clean energy and beyond.

Universal machine learning framework for defect predictions in zinc blende semiconductors

We develop a framework powered by machine learning (ML) and high-throughput density functional theory (DFT) computations for the prediction and screening of functional impurities in groups IV, III–V, and II–VI zinc blende semiconductors. Elements spanning the length and breadth of the periodic table are considered as impurity atoms at the cation, anion, or interstitial sites in supercells of 34 candidate semiconductors, leading to a chemical space of approximately 12,000 points, 10% of which are used to generate a DFT dataset of charge dependent defect formation energies. Descriptors based on tabulated elemental properties, defect coordination environment, and relevant semiconductor properties are used to train ML regression models for the DFT computed neutral state formation energies and charge transition levels of impurities. Optimized kernel ridge, Gaussian process, random forest, and neural network regression models are applied to screen impurities with lower formation energy than dominant native defects in all compounds.

•

Large computational dataset of defect properties in semiconductors is developed

•

Regression algorithms are used to train predictive models for defect properties

•

Best models are used for high-throughput prediction and screening

•

Lists of low energy “dominating” impurities are generated

Our article introduces a universal predictive framework for point defect formation energies and charge transition levels in a wide chemical space of zinc blende semiconductors and possible impurity atoms selected from across the periodic table. This framework was developed by leveraging high-throughput quantum mechanical simulations benchmarked using some experimental data from the literature, as well as machine learning (ML)-based regressions techniques that map unique materials descriptors to computed defect properties and yield optimized and generalizable models. The power and utility of these models is revealed through quick predictions for thousands of new defects and screening of low-energy impurities, which may tune the equilibrium conductivity in the semiconductor. This work presents, to our knowledge, the largest density functional theory (DFT) dataset of defect properties in semiconductors and the largest DFT+ML-based screening of point defects in semiconductors to date.

Machine Learning Prediction of the Exfoliation Energies of Two-Dimension Materials via Data-Driven Approach

Exfoliation energy is one of the fundamental parameters in the science and engineering of two-dimensional (2D) materials. Traditionally, it was obtained via indirect experimental measurement or first-principles calculations, which are very time- and resource-consuming. Herein, we provide an efficient machine learning (ML) method to accurately predict the exfoliation energies for 2D materials. Toward this end, a series of simple descriptors with explicit physical meanings are defined. Regression trees (RT), support vector machines (SVM), multiple linear regression (MLR), and ensemble trees (ET) are compared to develop the most suitable model for the prediction of exfoliation energies. It is shown that the ET model can efficiently predict the exfoliation energies through extensive validations and stability analysis. The influence of the defined features on the exfoliation energies is analyzed by sensitivity analysis to provide novel physical insight into the affecting factors of the exfoliation energies.