期刊
ACS APPLIED MATERIALS & INTERFACES
卷 15, 期 43, 页码 50469-50478出版社
AMER CHEMICAL SOC
DOI: 10.1021/acsami.3c09643
关键词
lithium metal anode; all-solid-state battery; ToF-SIMS; machine-learning-assisted analysis; datascience
In this study, machine learning techniques, especially logistic regression, were used to identify characteristic secondary ions of 5 different pure lithium compounds and applied to mixture and lithium metal anode samples, enabling identification of their compositions. The machine-learning-based analysis approach showed good performance and simplified the analysis of lithium compound samples.
Detailed knowledge about contamination and passivation compounds on the surface of lithium metal anodes (LMAs) is essential to enable their use in all-solid-state batteries (ASSBs). Time-of-flight secondary ion mass spectrometry (ToF-SIMS), a highly surface-sensitive technique, can be used to reliably characterize the surface status of LMAs. However, as ToF-SIMS data are usually highly complex, manual data analysis can be difficult and time-consuming. In this study, machine learning techniques, especially logistic regression (LR), are used to identify the characteristic secondary ions of 5 different pure lithium compounds. Furthermore, these models are applied to the mixture and LMA samples to enable identification of their compositions based on the measured ToF-SIMS spectra. This machine-learning-based analysis approach shows good performance in identifying characteristic ions of the analyzed compounds that fit well with their chemical nature. Moreover, satisfying accuracy in identifying the compositions of unseen new samples is achieved. In addition, the scope and limitations of such a strategy in practical applications are discussed. This work presents a robust analytical method that can assist researchers in simplifying the analysis of the studied lithium compound samples, offering the potential for broader applications in other material systems.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据