Screening Lithium-Ion Conductors Based on Local and Global Topology

Abstract The discovery of high-performance solid-state electrolytes remains a critical challenge for all-solid-state Li-ion batteries. To accelerate the identification of novel Li-ion conductors, we developed a hierarchical screening framework combining local structure order parameters (LSOPs) and persistent homology (PH). By employing a LightGBM-based model trained on 171 Li-containing compounds with reported ionic conductivities, we achieved over 70% accuracy on the test set in distinguishing ionic conductivity. SHapley Additive exPlanations (SHAP) analysis, combined with Haven ratio and migration alignment analysis, revealed that local quasi-linear migration pathways (150° bend in the Li–Li–Li arrangement) favor high ionic conductivity, whereas sharp-angle geometries (L-shaped 90° bend in Li–Li–Li) hinder Li-ion transport, providing important guidance for designing superionic conductors. Subsequently, the trained model screened 3,664 unlabeled compounds and identified 778 promising candidates, which were further screened by PH-based clustering to incorporate long-range Li-ion transport characteristics. After MD evaluations of 72 candidates, we selected Li5SbS3I2 for experimental validation. The crystalline phase of Li5SbS3I2 showed limited conductivity (8.6 × 10–8 S cm–1) due to its highly ordered structure. Since aliovalent doping attempts to introduce defects failed due to its limited solid solubility, mechanical amorphization was employed to introduce defects and successfully increased the conductivity to 8.1 × 10–5 S cm–1 at 25 °C. The potential of amorphous Li5SbS3I2 as a solid-state electrolyte was further validated through charge–discharge tests of all-solid-state batteries. Our findings provide insights into the local and global Li-migration pathways to facilitate the development of next-generation all-solid-state battery technologies.