固态电池的微观结构分辨阻抗建模
Microstructure-Resolved Impedance Modeling of Solid-State Batteries
- The Ohio State University（俄亥俄州立大学）
机构由 AI 辅助整理，请以论文原文为准。
AI总结：
本文提出一种微观结构分辨的阻抗建模框架，预测固态电池Li/Li6PS5Cl/Li对称电池的阻抗谱，揭示孔隙率、接触和SEI相的影响，并展示AI分割实验图像直接预测阻抗的路径。
AI中文摘要：
采用固体电解质（SE）的固态电池（SSB）正吸引汽车行业的大量投资，以实现快速充电和更安全的下一代电动汽车。SE的微观结构对电池性能有至关重要的影响。电化学阻抗谱（EIS）是一种强大的、非破坏性的探针，用于探测SE内部以及电极/SE界面上的电荷转移和传输过程，但将测量的阻抗谱定量地与潜在的微观结构特征联系起来仍然是一个开放的建模挑战。在此，我们提出了一个建模框架，用于预测Li/Li6PS5Cl/Li对称电池的微观结构分辨阻抗响应。利用相场烧结模拟得到的Li6PS5Cl SE微观结构，我们发现增加孔隙率会使Nyquist半圆向更高阻抗方向移动，而改善颗粒间接触则会降低阻抗，并且随着烧结的进行，阻抗单调减小。我们还分离了固体电解质界面相（SEI）形成的影响：晶界仅略微提高阻抗，但SEI体积分数以及每个SEI相（Li2S、Li3P、LiCl）的电导率都对其有强烈影响。与仅含晶界的网络相比，真实的多相SEI使界面阻抗增加约两个数量级。我们进一步证明，相同的流程可以直接操作于实验显微图像，通过AI辅助图像分割提取像素级相信息，并利用该信息预测阻抗谱，这展示了一条基于真实成像的SE微观结构实现定量微观结构-阻抗关联的路径。
英文摘要：
Solid-state batteries (SSBs) with solid electrolytes (SEs) are attracting substantial investment from the automotive industry to enable fast-charging and safer next-generation electric vehicles. The microstructure of the SE critically affects the battery performance. Electrochemical impedance spectroscopy (EIS) is a powerful, non-destructive probe of charge-transfer and transport processes within SEs and across electrode/SE interfaces but quantitatively connecting measured impedance spectra to the underlying microstructural features remains an open modeling challenge. Here we present a modeling framework that predicts the microstructure-resolved impedance response of a Li/Li6PS5Cl/Li symmetric cell. Taking the Li6PS5Cl SE microstructures from phase-field sintering simulations, we find that increasing porosity shifts the Nyquist semicircle to higher impedance while improved inter-particle contact reduces it, and that impedance decreases monotonically as sintering proceeds. We also isolate the effect of solid electrolyte interphase (SEI) formation: grain boundaries raise impedance only slightly, but SEI volume fraction, and the conductivity of each SEI phase (Li2S, Li3P, LiCl), both strongly affect it. A realistic multi-phase SEI increases interfacial impedance by about two orders of magnitude compared with a grain boundary only network. We further demonstrate that the same pipeline can operate directly on experimental micrographs by extracting pixel-level phase information with AI-assisted image segmentation and using it to predict an impedance spectrum, illustrating a path toward quantitative microstructure-impedance correlations grounded in real, imaged SE microstructures.
