Machine Learning Assisted Design of High‐Entropy Alloy Interphase Layer for Lithium Metal Batteries

Abstract

Lithium dendrite growth and the resulting safety concerns hinder the application of lithium metal. Compared with single metal or medium entropy alloys, high‐entropy alloys (HEAs) are a promising solution to solve the challenges of lithium metal anodes due to their unique properties. However, designing HEA layer with appropriate elements and proportion has become obstacles. Herein, machine learning (ML), density functional theories (DFT) calculation and data analysis reveal the contribution of Zn in lithiophilicity, Al in hardness and lithiophilicity, Fe, Co, and Ni in providing magnetism. The magnetron sputtering is used to construct the HEA interphase layer, and three parameters (sputtering power, sputtering time, and substrate rotation speed) are optimized via particle swarm optimization (PSO) based on the logarithm of the average coulombic efficiency (CE) of Li||Cu half cells. While the HEA layer with high strength, compactness, and flatness is constructed, Li||Li symmetric cell assembled by HEA@Li at 1 mA cm⁻², 1 mAh cm⁻² can cycle stably for 2400 h, and discharge capacity retention rate of Li||LFP cell is >90% after 300 cycles at 1 C with average CE of 99.67%. Design of the HEA interphase layer assisted by ML provides a path for the potential application of lithium metal batteries.