Machine Learning Force Fields in Electrochemistry: From Fundamentals to Applications

This article reviews the foundations and applications of machine learning force fields (MLFFs) in electrochemistry, highlighting their role as a transformative tool in materials science. We first provide an overview of MLFFs, then discuss their applications in ionics and electrochemical reactions, and finally outline future directions. Most MLFF approaches use invariant or equivariant descriptors derived from body-order expansions to represent many-body atomic interactions. These descriptors feed into linear regression models, kernel methods, or neural networks to construct potential energy surfaces for gases, liquids, solids, and interfaces involving inorganic and organic materials. MLFFs have enabled a wide range of advances, including all-atom molecular dynamics (MD), data extraction from MD, and accelerated materials discovery. In MD simulations, MLFFs allow accurate evaluation of ionic conductivity via the fluctuation-dissipation theorem and nonequilibrium MD under electric fields, applied to both solid and polymer electrolytes. For electrochemical reactions, MLFFs and Δ-ML models have been used to predict redox potentials in homogeneous and interfacial systems through thermodynamic integration. MLFFs also enable the extraction of key thermodynamic and kinetic information-such as free energy landscapes and local transport coefficients-from atomic trajectories, facilitating coarse-grained modeling of mass transport and reactions in complex electrolytes. In materials discovery, MLFFs have allowed high-throughput screening of 10⁷ to 10⁸ crystal structures, leading to the identification of promising Li-ion and Na-ion solid electrolytes. MLFFs are expected to continue evolving as a core technology in computational materials science, spanning a wide range from high-precision calculations to large-scale materials exploration.