Machine learning accelerated finite-field simulations for electrochemical interfaces

TL;DR

This paper introduces a machine learning-enhanced finite-field simulation method for electrochemical interfaces, enabling faster and more accurate modeling of electrochemical systems under realistic conditions without classical approximations.

Contribution

The authors develop a neural network-based finite-field approach trained on first-principles data, significantly accelerating simulations and enabling extrapolation beyond training potentials.

Findings

Accelerated simulations of electrochemical interfaces.

Successful extrapolation to higher cell potentials.

Revealed water molecule behavior changes at the anode.

Abstract

Electrochemical interfaces are of fundamental importance in electrocatalysis, batteries, and metal corrosion. Finite-field methods are one of most reliable approaches for modeling electrochemical interfaces in complete cells under realistic constant-potential conditions. However, previous finite-field studies have been limited to either expensive ab initio molecular dynamics or less accurate classical descriptions of electrodes and electrolytes. To overcome these limitations, we present a machine learning-based finite-field approach that combines two neural network models: one predicts atomic forces under applied electric fields, while the other describes the corresponding charge response. Both models are trained entirely on first-principles data without employing any classical approximations. As a proof-of-concept demonstration in a prototypical Au(100)/NaCl(aq) system, this approach not only dramatically accelerates fully first-principles finite-field simulations but also successfully extrapolates to cell potentials beyond the training range while accurately predicting key electrochemical properties. Interestingly, we reveal a turnover of both density and orientation distributions of interfacial water molecules at the anode, arising from competing interactions between the positively charged anode and adsorbed Cl$^-$ ions with water molecules as the applied potential increases. This novel computational scheme shows great promise in efficient first-principles modelling of large-scale electrochemical interfaces under potential control.