Rapid Identification of Spent Lithium-ion Batteries Black Powder Types Using Handheld LIBS and Interpretable MobileNet Models

Abstract

This work demonstrates, for the first time, the direct application of handheld laser-induced breakdown spectroscopy (LIBS) combined with a lightweight deep learning model for the rapid classification of spent lithium-ion battery black powders. It addresses a critical industrial need for on-site, real-time analysis in battery recycling. By developing an embedded system-compatible approach, this study provides JAAS readers with a practical and efficient solution that moves elemental analysis from the laboratory directly to the field. The integration of SHAP analysis further enhances the interpretability of the LIBS-based model, increasing trust in its decisions and advancing the application of intelligent LIBS instrumentation in sustainable resource recovery.