Deep learning-enabled large-scale analysis of particle geometry-lithiation correlations in battery cathode materials

Abstract

A deep learning model is employed to address the challenging problem of V${}_2$O${}_5$ nanoparticle segmentation and the correlation between the chemical composition and the geometrical features of lithiated V${}_2$O${}_5$ nanoparticles as an exemplar of a phase-transforming battery cathode material. First, the deep learning-enabled segmentation model is integrated with the singular value decomposition technique and a spectral database to generate accurate composition and phase maps capturing lithiation heterogeneities as imaged using scanning transmission X-ray microscopy. These phase maps act as the output properties for correlation analysis. Subsequently, the quantitative influences of the geometrical features of nanoparticles such as the particle size (i.e., projected perimeter and area), the aspect ratio, circularity, convexity, and orientation on the lithiation phase maps are revealed. These findings inform strategies to improve lithiation uniformity and reduce stress in phase-transforming lithium battery materials via optimized particle geometry.