Machine Learning-Driven Phase Field Numerical Analysis for a Comprehensive Theoretical Framework of Local Electrochemical Environment and Operating Conditions Effects on Dendrite Evolution Mechanisms

Abstract Unraveling the intricate interplay between the local electrochemical environment and operating conditions (LEEO) at the mesoscale is fundamental for the rational design of dendrite-free electrolytes. Herein, we quantitatively decouple the effects of diffusion and electromigration on dendrite evolution by using a phase-field model. Elevated voltages or current densities destabilize ion migration, exacerbating dendritic branching, while unexpectedly optimizing stripping efficiency. Conversely, high diffusion coefficients facilitate lateral ion replenishment, whereas superior ionic conductivity ensures rapid and uniform electromigration; their synergy effectively suppresses tip growth and regulates the lateral deposition. Moreover, Gaussian process regression establishes a robust predictive mapping between the LEEO and Coulombic efficiency (CE). This integrated mechanistic framework offers a reliable platform for screening electrolyte materials toward stable Li metal anodes.