Accelerated discovery of solid-state battery properties enabled by active learning approaches

• • Li4.5TiO3.25 and Li2VCl5 exhibit ionic conductivities exceeding 1 mS/cm at 25°C, matching the performance of liquid electrolytes in conventional lithium-ion batteries, which enables higher power density and faster charging in solid-state cells. • • The machine learning screening identified these compounds from a pool of over 10,000 lithium-containing candidates, reducing experimental validation time by a factor of 10 compared to traditional trial-and-error approaches, as evidenced by the rapid transition from prediction to AIMD and experimental confirmation. • • The design methodology increased the ionic conductivity of Li4.5TiO3.25 by 50% (from 1.0 to 1.5 mS/cm) through targeted compositional tuning, demonstrating a pathway to meet the >1 mS/cm threshold required for practical solid-state batteries operating at room temperature. • • Experimental validation confirmed that Li2VCl5 maintains electrochemical stability up to 4 V vs. Li/Li+, with no significant interfacial degradation after 100 cycles, addressing the critical challenge of electrolyte-electrode compatibility that has stalled sulfide-based systems.