Discovery of unconventional proton-conducting inorganic solids via defect-chemistry-trained, interpretable machine learning

High-throughput computational screening and machine learning hold significant promise for exploring diverse chemical compositions and discovering novel inorganic solids. However, the complexity of point defects, which occur in all inorganic solids and are often central to their functionality and synthesizability, presents research challenges. Here, we introduce defect-chemistry-trained, interpretable machine learning for the efficient exploration and discovery of unconventional proton-conducting inorganic solid electrolytes. By considering dopant dissolution and hydration, our machine learning models provide quantitative predictions and physical interpretations for synthesizable host–dopant combinations with hydration capabilities across various structures. Utilizing these insights, the unconventional proton conductors Pb-doped Bi12SiO20 sillenite and eulytite-type Sr-doped Bi4Ge3O12 are discovered in the first two synthesis trials. Notably, Pb-doped Bi12SiO20 represents an unprecedented class of proton-conducting electrolyte composed exclusively of groups 14 and 15 cations and featuring a sillenite structure, exhibiting fast and unique three-dimensional proton conduction along a softly bonded BiO5 network.