Identifying Temperature Effects of Lithium Dendrites Formation from Phase-Field Simulations Using Machine Learning

Lithium metal anodes offer exceptionally high energy density but are hindered by dendrite growth, which compromises safety and stability. Temperature, a critical yet complex factor, influences both ion diffusion and interfacial reaction kinetics, but its overall effect remains undetermined. Herein, aphase-field model incorporating Arrhenius-type diffusion and kinetics was developed to investigate the influence of temperature on lithium deposition stability. To overcome the limitations of high computational cost, a multilayer perceptron was trained on phase-field simulation data to predict and interpolate the stable deposition capacity across a broad parameter space. The model identifies optimal temperature ranges corresponding to different combinations of diffusion and kinetics parameters, revealing a nonmonotonic temperature dependence of dendrite formation. This machine-learning-assisted framework provides quantitative insight into temperature-governed deposition behavior and offers guidance for electrolyte engineering and operational strategies in high-energy lithium–metal batteries.