Machine learning-assisted design of oxygen-containing inorganic coating materials on a separator for lithium metal anodes

Abstract

The growth of lithium dendrites and its associated challenges pose significant obstacles to the widespread adoption of lithium metal anodes. Although numerous inorganic materials offer the potential for stabilizing lithium metal anodes, trial-and-error experiments are time-consuming and cost-intensive. In this work, first, a high-throughput screening workflow integrated with machine learning and calculations has been used to identify possible materials, which incorporates several key indicators encompassing electronic conductivity, phase stability, mechanical properties, chemical stability, and lithium-ion transport performance. Four materials were used in experiments, and the results from both characterization and electrochemical testing show that HfO₂@PP exhibits the best performance, which includes having the highest Young's modulus. Furthermore, an Li||Li symmetric cell assembled using HfO₂@PP operating at 1 mA cm⁻² and 1 mA h cm⁻² exhibited stable cycling for over 1000 h, while an Li||LFP cell assembled using HfO₂@PP has a capacity retention rate of more than 90% and an average coulombic efficiency of 99.7% after 200 cycles at 1 C. This work provides a design method and ideas for inorganic coating materials on separators for lithium metal anodes.