Mechanistic Insights into the Oxygen Evolution Reaction on Nickel-Doped Barium Titanate via Machine Learning-Accelerated Simulations

TL;DR

This study uses machine learning-accelerated simulations to investigate how nickel doping improves the oxygen evolution reaction activity of barium titanate, a promising cost-effective catalyst for water splitting.

Contribution

It introduces a neural network-based interatomic potential enabling large-scale simulations of OER on Ni-doped BaTiO3, revealing mechanistic insights into catalytic enhancement.

Findings

Ni-doping increases OER activity of BaTiO3

Machine learning potential achieves near-DFT accuracy in simulations

Simulations align with experimental observations of enhanced catalysis

Abstract

Electrocatalytic water splitting, which produces hydrogen and oxygen through water electrolysis, is a promising method for generating renewable, carbon-free alternative fuels. However, its widespread adoption is hindered by the high costs of Pt cathodes and IrO$_{x}$/RuO$_{x}$ anode catalysts. In the search for cost-effective alternatives, barium titanate (BaTiO$_{3}$) has emerged as a compelling candidate. This inexpensive, non-toxic perovskite oxide can be synthesized from earth-abundant precursors and has shown potential for catalyzing the oxygen evolution reaction (OER) in recent studies. In this work, we explore the OER activity of pristine and Ni-doped BaTiO$_{3}$ at explicit water interfaces using metadynamics (MetaD) simulations. To enable efficient and practical MetaD for OER, we developed a machine learning interatomic potential based on artificial neural networks (ANN), achieving large-scale and long-time simulations with near-DFT accuracy. Our simulations reveal that Ni-doping enhances the catalytic activity of BaTiO$_{3}$ for OER, consistent with experimental observations, while providing mechanistic insights into this enhancement.