Accelerating the Search for New Solid Electrolytes: Exploring Vast Chemical Space with Machine Learning-Enabled Computational Calculations

Abstract

Discovering new solid electrolytes (SEs) is essential to achieving higher safety and better energy density for all-solid-state lithium batteries. In this work, we report machine learning (ML)-assisted high-throughput virtual screening (HTVS) results to identify new SE materials. This approach expands the chemical space to explore by substituting elements of prototype structures and accelerates an evaluation of properties by applying various ML models. The screening results in a few candidate materials, which are validated by density functional theory calculations and ab initio molecular dynamics simulations. The shortlisted oxysulfide materials satisfy key properties to be successful SEs. The advanced screening method presented in this work will accelerate the discovery of energy materials for related applications.