Machine Learning-Enabled Microstructure Design of Solid-State Battery Cathodes

Abstract

Electrochemical performance of battery electrodes is dictated by the composition and physiochemical interactions underlying their porous microstructure. Coupled kinetic and transport processes in these electrodes are correlated to their internal resistive pathways, which directly influence cell performance. However, the intrinsically stochastic and heterogenous nature of these electrodes makes the characterization of their effective transport properties and performance very complex. In this regard, machine learning can enable a data-driven route towards faster prediction of electrode properties, such as tortuosity, electronic and ionic conductivities and active surface area. In this work, we aim to analyze porous solid-state battery cathodes using physics-based modeling and machine learning-augmented data-driven analytics towards achieving optimal electrochemical performance under operational extremes via microstructural modulation.