Electronic charge density as a fast approach for predicting Li-ion migration pathways in superionic conductors with first-principles level precision

Abstract

Finding Li-ion migration pathways in solid state electrolytes is an important prerequisite for disruptive development of all-solid-state battery materials. Previous studies either used empirical method such as bond valence (BV) or relied on on-the-fly first-principles molecular dynamics (FPMD) simulations, which are very time and resource consuming. In this paper, we propose an approach of using spatially dependent electronic charge density to predict Li-ion migration pathways in superionic conductors with first-principles level precision. Since the electronic charge density can be simultaneously calculated along with the structure optimization, this method saves tremendous computing time in finding the migration pathways, as is the case for the currently widely used FPMD method. Its accuracy and feasibility are validated by reproducing 3D diffusion channel of six representative Li-ion structures [LiFePO4, Li2S, Li5PS4Cl2, Li10GeP2S12, Li4GeS4, and Li3Y(PS4)2]. Our approach is expected to accelerate high-throughput screening of superionic conductors, such as using the most likely migration paths replaces global search of migration paths for first-principles method. The direct relationship between ion transport pathways and electronic charge densities constructed here could serve as an efficient descriptor for training machine learning models. Due to the inherent relationship between bottom-level electronic charge density and macroscopic properties, we expect that this method can be also extended to designing materials with other target physical or chemical properties such as thermoelectrics, photovoltaics, and fuel cells.