Synchrotron imaging of pore formation in Li metal solid-state batteries aided by machine learning

High-rate capable, reversible lithium metal anodes are necessary for next generation energy storage systems. In situ tomography of Li|LLZO|Li cells is carried out to track morphological transformations in Li metal electrodes. Machine learning enables tracking the lithium metal morphology during galvanostatic cycling. Nonuniform lithium electrode kinetics are observed at both electrodes during cycling. Hot spots in lithium metal are correlated with microstructural anisotropy in LLZO. Mesoscale modeling reveals that regions with lower effective properties (transport and mechanical) are nuclei for failure. Advanced visualization combined with electrochemistry represents an important pathway toward resolving non-equilibrium effects that limit rate capabilities of solid-state batteries.