This paper presents a machine learning workflow combining quantum calculations and graph convolutional neural networks to discover ionic liquids (ILs) suitable for ionic polymer electrolytes (IPEs) in lithium metal batteries (LMBs). The workflow screens a large pool of IL candidates based on ionic conductivity and electrochemical window, leading to the experimental investigation of selected ILs in IPE membranes. These membranes demonstrate high critical current density, excellent capacity retention over numerous cycles, fast charge/discharge capabilities, and high efficiency, exceeding the performance of many other single-layer polymer electrolytes.

Kai Li, Jifeng Wang, Yuanyuan Song, Ying Wang