Data-Driven Optimization of Polymer Electrolytes for High-Performance Solid-State Li-Ion Batteries

High-performance polymer electrolytes are of paramount importance for advancing solid-state batteries due to the advantages such as high chemical stability and enhanced safety. In addition, the desirable flexural strength of polymer electrolyte offers the effective strategy to mitigate the effect of vast volume variation of Si anodes during electrochemical operations. However, the ionic conductivity of most polymer electrolytes is at the level of 10^-4 S/cm, about one to two orders magnitudes lower than that of typical liquid electrolytes and even ceramic solid electrolytes. The discovery of novel polymer electrolytes with improved conductivities is crucial for their implementation in battery applications. In this talk, we present a data-driven optimization approach to enable the fast screening and identification of polymer electrolytes with desirable electrochemical properties. Our method combines the molecular dynamics (MD) workflow and multi-objective Bayesian optimization to streamline the candidate selection and property evaluation process. Through the screening in the vastly large chemical space of potential polymer electrolytes, the data-driven approach discovers a series of promising candidates with the ionic conductivity reaching to the level of 10^-3 S/cm and above. The discovery of novel polymer electrolytes through data-driven optimization paves the road for enabling the realization of high-performance solid-state Li-ion batteries and propelling the clean energy revolution forward.