Machine Learning in Computational Surface Science and Catalysis: Case Studies on Water and Metal-Oxide Interfaces.

Xiaoke Li,Wolfgang Paier,Joachim Paier

doi:10.3389/fchem.2020.601029

Abstract

The goal of many computational physicists and chemists is the ability to bridge the gap between atomistic length scales of about a few multiples of an Ångström (Å), i. e., 10−10 m, and meso- or macroscopic length scales by virtue of simulations. The same applies to timescales. Machine learning techniques appear to bring this goal into reach. This work applies the recently published on-the-fly machine-learned force field techniques using a variant of the Gaussian approximation potentials combined with Bayesian regression and molecular dynamics as efficiently implemented in the Vienna ab initio simulation package, VASP. The generation of these force fields follows active-learning schemes. We apply these force fields to simple oxides such as MgO and more complex reducible oxides such as iron oxide, examine their generalizability, and further increase complexity by studying water adsorption on these metal oxide surfaces. We successfully examined surface properties of pristine and reconstructed MgO and Fe3O4 surfaces. However, the accurate description of water–oxide interfaces by machine-learned force fields, especially for iron oxides, remains a field offering plenty of research opportunities.

Highlights

Machine learning (ML) is currently attracting the interest of a broad community
This work explores practical aspects and transferability or generalizability of machine-learned FFs (MLFFs) following the Gaussian approximation potentials (GAPs) ansatz, which was efficiently implemented in the Vienna ab initio simulation package, VASP, recently
The generalized-gradient approximation to Kohn–Sham density functional theory (DFT) after PBE serves as an electronic structure reference method for assessing these MLFFs

Summary

INTRODUCTION

Machine learning (ML) is currently attracting the interest of a broad community. Searching the keywords machine learning and chemistry combined by the Boolean AND yields about 4,300 hits solely for the year 2019. The present work systematically explores capabilities of so-called “on-the-fly” machine-learned FFs (MLFFs) (Jinnouchi et al, 2019b) using a variant of the GAP approach together with molecular dynamics (MD) runs, as recently implemented in the Vienna ab initio simulation package (VASP) We apply these FFs to relevant problems in surface and interface science valuable for catalysis research. For the plane wave expansion of Bloch waves, the minimal (default) kinetic energy cutoff was employed except for MD runs on the H2O/Fe3O4(111) surfaces to train the MLFF Open-shell (spin-polarized) FeOx-related calculations use an energy break criterion of 10−5 eV Input structures for these MLFF-training MD runs are described in detail .

RESULTS AND DISCUSSION

E Order MSE MUE

CONCLUSIONS

DATA AVAILABILITY STATEMENT