Computational insight into the grain boundary structure and atomic mobility in metallic lithium

Whisker growth during lithium electrodeposition is one of the most critical factors limiting the safe operation of high-capacity lithium-metal batteries. According to the recent findings, the underlying fundamental mechanism includes solid-state mass transport occurring presumably along the grain boundaries (GBs) and feeding the whiskers. Here, we perform atomistic simulations of lithium for both (100) and (110) twist GBs using a high-accuracy machine learning interatomic potential . We find that at room temperature the GBs under study are liquified and possess a high self-diffusion coefficient comparable to that of the liquid phase. The grain growth via absorption of additionally deposited atoms at the GBs to build a new atomic layer of the crystal lattice (i.e., layer-by-layer growth) is explicitly modeled, which enables uncovering various atomistic configurations of the grain boundaries that may take place in the course of Li electrodeposition. Additionally, in order to assist further atomistic simulations of polycrystalline Li, we analyzed the capability of the existing conventional EAM-based force fields to model the GB behavior.