Estimating potential-dependent physicochemical properties at metal--electrolyte interfaces using machine learning interatomic potentials (Part 1)

This record is the first part of the dataset supporting our associated publication. It contains the electrolyte–metal interfacial dataset used to train machine-learning interatomic potentials (MLIPs) using the MACE architecture, along with the corresponding test sets used to benchmark energy and force predictions. This record also includes all final trained MACE models. In addition, the record provides the ab initio molecular dynamics (AIMD) trajectories used for validating short MACE molecular dynamics (MD) simulations and for comparing interfacial properties during the benchmarking phase. Finally, this deposition contains the complete 2 ns MACE MD trajectories for electrolyte–Au(111) interfaces, which were used to analyze water structure, ion distributions, and interfacial behavior. Related datasets for Cu(111) and Rh(111) trajectories are available at DOI: 10.5281/zenodo.17756911 and DOI: 10.5281/zenodo.17757050. Additional methodological and scientific details are provided in the associated publication (https://doi.org/10.1021/acselectrochem.5c00540).