Identification of Lithium Compounds on Surfaces of Lithium Metal Anode with Machine-Learning-Assisted Analysis of...

Area of Science:

• Materials Science
• Electrochemistry
• Analytical Chemistry
• Data Science

Background:

• Understanding surface contamination and passivation on lithium metal anodes (LMAs) is critical for developing stable all-solid-state batteries (ASSBs).
• Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) is a powerful surface-sensitive technique for characterizing LMAs, but its complex data requires advanced analysis.
• Manual analysis of ToF-SIMS spectra from LMAs is often laborious and time-consuming, hindering rapid material characterization.

Purpose of the Study:

• To develop and validate a machine learning approach for automated analysis of ToF-SIMS data from lithium compounds and LMAs.
• To identify characteristic secondary ions specific to different lithium compounds using logistic regression.
• To assess the accuracy and applicability of the machine learning method for compositional analysis of LMA surfaces.

Main Methods:

• Application of logistic regression (LR), a machine learning technique, to identify characteristic secondary ions from pure lithium compounds.
• Training and testing of LR models using ToF-SIMS spectral data from 5 distinct pure lithium compounds.
• Extension of the validated LR models to analyze complex mixtures and actual lithium metal anode samples.

Main Results:

• The machine learning approach successfully identified characteristic ions for pure lithium compounds, aligning with their chemical properties.
• High accuracy was achieved in identifying the compositions of unknown mixture and LMA samples using the trained models.
• The developed method demonstrated robustness in analyzing ToF-SIMS data, simplifying the characterization process.

Conclusions:

• Machine learning, specifically logistic regression, offers an efficient and accurate method for analyzing complex ToF-SIMS data from lithium metal anodes.
• This automated analysis accelerates the understanding of surface chemistry in LMAs, crucial for ASSB development.
• The study presents a valuable analytical tool with potential for broader application in materials characterization beyond lithium systems.