【New Publication】Orbital-Free DFT-Assisted Machine-Learned Molecular Dynamics for Electric-Field-Driven Ionic Transport

https://arxiv.org/abs/2609.03518

Quemix and Idemitsu Kosan Co., Ltd. have published a new study proposing an efficient molecular dynamics (MD) method for simulating ionic transport under electric fields.

The proposed approach combines a machine-learned interatomic potential for efficiently evaluating field-independent interatomic forces with orbital-free density functional theory (OFDFT) to determine atomic charges that vary with the local atomic environment. This enables field-driven ionic transport and accompanying changes in charge states to be simulated at a lower computational cost than Kohn–Sham DFT-based molecular dynamics, without requiring an additional material-specific machine-learning model for charge prediction.

As a proof of concept, the method was applied to an S₈/Li₃PS₄ interface representing a lithium-battery material. The simulations reproduced the migration of Li ions from the Li₃PS₄ region into the S-rich region under an applied electric field, together with the accompanying negative charging of S atoms. Comparison with Kohn–Sham DFT for a representative interface structure also qualitatively reproduced the Li charges and the negative charging of S atoms.

The KSDFT and OFDFT calculations in this study were performed using Quemix's materials computation platform, Quloud.