Presentation Information
[18a-PA1-4]Large-Scale Simulation of Solid Electrolyte Interfaces in All-Solid-State Batteries Using Neural Network Potentials
〇Naoki Matsumura1, Kazutaka Nishiguchi1, Yuta Yoshimoto1, Meguru Yamazaki1, Yasufumi Sakai1 (1.Fujitsu)
Keywords:
neural network potential,all-solid-state battery,molecular dynamics simulation
All-solid-state batteries are attracting attention for their high safety and energy density, but the stability of the solid electrolyte interphase (SEI) at the Li metal anode/sulfide electrolyte interface critically affects performance. In this study, we employed knowledge distillation in automated NNP generation to achieve DFT-level accuracy for large-scale simulations (127,296 atoms, 10 ns). This enabled atomic-level analysis of SEI formation and Li diffusion behavior, allowing quantitative evaluation of interfacial reactions and ion transport properties.