Self-healing mechanism of lithium metal

Li metal as an ideal anode material meets the requirements of state-of-the-art secondary batteries. However, the dendrite growth of Li causes safety concerns and results in a low coulombic efficiency, which has significantly restricted the commercial applications of Li secondary batteries. Owing to the intrinsic limitations of even the most advanced experimental and computational techniques, a mechanistic understanding of Li deposition (growth) on the atomic scale is lacking. Here, we construct a Li potential model by machine learning with an accuracy of quantum mechanical computations. Our molecular dynamics simulations based on this potential model reveal two self-healing mechanisms in a large Li metal system, surface self-healing and bulk self-healing, and identify three Li dendrite morphologies in different conditions, "needle," "mushroom," and "hemisphere." We finally propose the concepts of local current density and variance of local current density as a supplement to critical current density to determine the probability of self-healing triggered.