Performance prediction of high-entropy perovskites La0.8Sr0.2MnxCoyFezO3 with automated high-throughput characterization of combinatorial libraries and machine learning

Perovskite oxides (ABO3) represent a large family of materials with wide application in many fields due to their celebrated structural and chemical flexibility. Such a vast space of compositions requires efficient exploration strategies now possible with automated high-throughput experiments combined with machine learning prediction algorithms. In this study, we investigate the composition performance relationships of high-entropy La0.8Sr0.2MnxCoyFezO3 perovskite oxidesfor application as oxygen electrodes in Solid Oxide Cells. After deposition of a continuous compositional map using thin film combinatorial pulsed laser deposition, we obtain experimental data of structural, composition and functional propertiesfor the whole material family under study through a combination of six advanced characterization methodologies with mapping capabilities. We prove that supervised machine learning methods, particularly random forests, effectively capture the complex relationships between composition, structural features, and electrochemical performance including oxygen transport properties. Using these predictive methods, we create an accurate continuous map of performance for the whole compositional space under study and we open it to the community. Moreover, our model yields an unambiguousstatistical correlation between the distortion of the oxygen sublattice (obtained from spectral analysis of their Raman-active modes) and the highest performances. Finally, the study consistently identifies Fe-rich high-entropy oxides as the optimal compounds with the lowest area-specific resistance values for oxygen electrodes at 700°C. Overall, the work proves the potential of a detailed exploration of relevant chemical spaces by coupling high throughput experiments and machine learning models to gain new insights and optimize relevant families of materials.