The electrochemical reaction in silicon (Si) electrode, accompanying with tremendous volume expansion, causes rapid capacity fade of Li-ion batteries. The Li-ion concentration gradient and structural distribution uniformity influence the inhomogeneous expansion, and the kinetic mechanism of lithiation and interfacial morphology evolvement remains debated. The present study focuses on the dynamics of Li-Si interdiffusion at Si/Li interfaces with various Si-facet orientations and phases using a machine-learning potential. We find that the Si flux from bulk Si to Li-Si interface regions controls the length of Li-Si interdiffusion region. The lithiation length in different Si/Li interface systems exhibits the order of amorphous-Si > crystalline-Si(110) > crystalline-Si(100) > crystalline-Si(111), which agrees with the experimental trend. Our atomic simulations further reveal that the key factor determining the Li-Si interdiffusion is the difference of on-site Si atomic energies between the bulk Si and the Li-Si interface regions. We propose that the large interdiffusion extent is due to a low thermodynamics barrier. Our findings provide insights for the development of high-performance Si anode materials.

硅（Si）电极中的电化学反应伴随着巨大的体积膨胀，导致锂离子电池的容量快速衰减。锂离子浓度梯度和结构分布均匀性影响不均匀膨胀，锂化和界面形貌演化的动力学机制仍存在争议。本研究利用机器学习潜力，重点研究具有各种 Si 面方向和相位的 Si/Li 界面处的 Li-Si 相互扩散动力学。我们发现从块体硅到锂硅界面区域的硅通量控制了锂硅相互扩散区​​域的长度。不同Si/Li界面体系中的锂化长度呈现出非晶-Si > 晶态-Si(110) > 晶态-Si(100) > 晶态-Si(111)的顺序，这与实验趋势一致。我们的原子模拟进一步表明，决定Li-Si相互扩散的关键因素是块体Si和Li-Si界面区域之间的现场Si原子能量的差异。我们认为大的相互扩散程度是由于热力学势垒低所致。我们的研究结果为高性能硅负极材料的开发提供了见解。