(Invited) Oxide-Water Interfaces from Ab Initio Based Deep Neural Network Simulations

Abstract

Photo-electrocatalysis involving complex oxide-water interfaces is a highly promising technology for the sustainable production of fuels. However, probing these complex interfaces and gaining atomistic insights is still very challenging for current experimental methods, and is often only possible through accurate computational simulations. While first principles-based simulations provide the best accuracy, they are also limited to small system sizes and time-scales due to their high computational cost. Machine learned potentials that reproduce the accuracy of first-principles methods, while at the same time allowing the exploration of larger systems and timescales, are a promising approach to circumvent this issue. In this talk, I will discuss recent applications of deep neural network based molecular dynamics simulations to understand the structure and dynamics of interfacial water on photoelectrochemically relevant oxide surfaces. Focusing on TiO2 and IrO2 surfaces, we examine interfacial proton transfer and discuss how the water dissociation fraction and hydroxyl lifetimes depend on the surface atomic and electronic structures.