Neural network based molecular dynamics simulations for oxide ion transport in solid oxide electrolysis cell materials

Abstract

Solid oxide electrolysis cells (SOECs) based hydrogen production is regarded as one of the most efficient methods for sustainable energy conversion. Brownmillerite-type oxides, such as Ca₂Fe₂O₅, have recently gained significant interest due to their inherent oxygen vacancy channels that facilitate oxide ion transport through the structure. In this research work, neural network-based interatomic potentials (ML-IAPs) were employed in molecular dynamics (MD) simulations to study the oxide ion transport in Co-doped brownmillerites. The simulation results and experimental data aligned within 98 % accuracy. Parametric analysis revealed that temperature, Co doping, and oxygen vacancies enhances oxide ionic conductivity in these materials. Co doping leads to a 2–3 fold increase in diffusion coefficient compared to undoped Ca₂Fe₂O₅ structure. From the trajectory analysis, oxygen pathways, and anisotropic nature of ionic diffusion in brownmillerites is observed. The findings of this research provide a strong data-driven framework for accelerating material selection strategies, paving the way for next-generation high-performance electrolysis technologies.