The solid electrolyte interphase (SEI) is regarded as the most important factor affecting the durability of lithium-metal anode in all-solid-state batteries (ASSBs). Despite its significance, the nucleation and growth mechanism of SEI is not yet well understood. Here, we elucidate the thermodynamics and kinetics governing SEI formation at the Li|β-Li₃PS₄ interface at the atomic scale via thermodynamic phase equilibrium analysis and machine-learning-potential-assisted molecular dynamics (MD) simulations. An accurate moment tensor potential using the machine-learning method is developed for a reactive model of Li|β-Li₃PS₄. This potential enabled us to perform large-scale MD simulations with the model size expanded to the experimental dimensions (∼40 nm) while maintaining the same level of accuracy as density functional theory calculations. The results reveal a four-stage evolution process at the Li|β-Li₃PS₄ interface, namely (i) fast ion diffusion, (ii) nucleation, (iii) Li₂S growth, and (iv) stabilization. Notably, we demonstrate that the SEI can be categorized into crystalline and amorphous regions. The simulated SEI thickness, structure, and composition closely match experimental findings, validating the accuracy of the MD simulations. We further disclose the significant impact of ion diffusion kinetic limitations on the phase formation and crystallization of interfacial products. Furthermore, we shed light on the detailed potential energy (PE) distribution of lithium along the direction perpendicular to the interface. This information is crucial for better understanding interfacial ion mobility.

中文翻译：

---

可视化锂金属和固态电解质之间的 SEI 形成

固体电解质界面（SEI）被认为是影响全固态电池（ASSB）中锂金属负极耐久性的最重要因素。尽管具有重要意义，但 SEI 的成核和生长机制尚不清楚。在这里，我们通过热力学相平衡分析和机器学习势辅助分子动力学（MD）模拟，在原子尺度上阐明了控制Li|β-Li ₃ PS _{4界面SEI形成的热力学和动力学。}使用机器学习方法为Li|β-Li ₃ PS ₄的反应模型开发了精确的矩张量势。这种潜力使我们能够进行大规模 MD 模拟，模型尺寸扩展到实验尺寸（~40 nm），同时保持与密度泛函理论计算相同的精度水平。_{结果揭示了Li|β-Li 3} PS ₄界面处的四阶段演化过程，即(i)快离子扩散、(ii)成核、(iii)Li ₂ S生长和(iv)稳定化。值得注意的是，我们证明SEI可以分为结晶区和非晶区。模拟的 SEI 厚度、结构和成分与实验结果非常吻合，验证了 MD 模拟的准确性。我们进一步揭示了离子扩散动力学限制对界面产物的相形成和结晶的显着影响。此外，我们还揭示了锂沿垂直于界面方向的详细势能（PE）分布。这些信息对于更好地了解界面离子淌度至关重要。