| Abstract Scope |
LaCoO₃ perovskites are promising materials for energy and catalysis applications. To enhance their performance, researchers introduce dopants to improve structural stability and oxygen ionic conductivity. However, rational dopant design remains challenging due to the vast configurational and chemical space. Here, we develop a machine learning-driven framework to systematically explore the effects of 20 dopants—including A-site (Mg, Ca, Sr, Ba, Ce, Pr, Nd, Sm, Gd) and B-site (Sc, Ti, V, Cr, Mn, Fe, Ni, Cu, Zn, Al, Ga) substitutions—on the stability and ionic conductivity of LaCoO₃. Using Bayesian active learning with an Expected Improvement acquisition strategy, we efficiently distill composition–structure–energy relationships. These insights guide forward modeling to predict ionic transport behavior. To bridge research and application, we introduce LCO-DOPER, an interactive web application powered by machine learning models, offering a comprehensive tool for doped LaCoO₃ analysis and design. |