Insights into the Atomic Mechanism of Lithium-Ion Diffusion in Li6PS5Cl via a Machine Learning Potential

JF Chen and MD Fang and Q Wu and S Tang and JH Zheng and CL Wei and XY Cao and Y Shi and N Xu and Y He, CHEMISTRY OF MATERIALS, 37, 591-599 (2025).

DOI: 10.1021/acs.chemmater.4c01152

Solid-state electrolytes are promising for next-generation lithium batteries but often suffer from low ionic conductivity. Gaining atomic- level insights into lithium diffusion mechanisms is crucial for rationally designing solid-state electrolytes with optimized ionic conductivity. The jump rate of lithium ions between sites is commonly used to evaluate lithium-ion diffusion. However, prior computational studies of solid-state electrolytes found inconsistent optimal anion disorder levels from jump rate analysis versus mean square displacement (MSD) calculations. Using Li6PS5Cl as a model solid-state electrolyte, this work demonstrates that using the effective jump rate, which excludes nondiffusive back-and-forth jumps, resolves this discrepancy. Through molecular dynamics simulations with an accurate machine learning potential, the optimal range of 37.5%-50% for the S/Cl anion disorder level was identified from the effective jump rate analysis, agreeing with the range for maximal ionic conductivity calculated from MSD. Further analyses illustrated how anion disorder impacts the connectivity of diffusion pathways and ionic conductivity. This combined machine learning and effective jump rate approach links bulk conductivity to microscopic mechanisms, delivering insights to guide the design of superior solid-state electrolytes.

Return to Publications page