Machine learning-assisted advances and perspectives for electrolytes of protonic solid oxide fuel cells

Protonic solid oxide fuel cells (P-SOFCs), as a promising power generation technology, have garnered increasing attention due to their advantages of cleanliness, high efficiency, and high reliability. As a critical component of P-SOFCs, proton-conducting electrolytes exhibit high ionic conductivity, enabling high chemical-to-electrical energy conversion efficiency at intermediate temperatures. However, there are still many challenges in further enhancing the proton conductivity and stability of the currently widely used Ba(Zr, Ce)O3 electrolytes through traditional experimental methods. Herein, this review firstly summarized the current research status of proton-conducting oxides, including ABO3 perovskite-type oxides and other structural oxides, and highlighted the challenges faced by electrolyte development in terms of proton conductivity, compatibility with other components, and long-term durability. Then, the relevant progress of machine learning (ML) in the research of P-SOFC electrolytes was meticulously discussed and the promising applications of ML in proton-conducting electrolyte performance screening, stability prediction, and morphology analysis were pointed out. More importantly, the challenges and solutions of proton-conducting electrolytes designed by ML were uncovered by considering the reliable database, feature engineering, accurate model, and experimental validation. Overall, this review concluded the advances of ML-assisted P-SOFC electrolytes and addressed the future research directions in the synergy of ML and electrolytes.