Machine Learning Tailored Anodes for Efficient Hydrogen Energy Generation in Proton-Conducting Solid Oxide Electrolysis Cells.

In the global trend of vigorously developing hydrogen energy, proton-conducting solid oxide electrolysis cells (P-SOECs) have attracted significant attention due to their advantages of high efficiency and not requiring precious metals. However, the application of P-SOECs faces challenges, particularly in developing high-performance anodes possessing both high catalytic activity and ionic conductivity. In this study, LaBaCoNiO (LBCN9173) and LaCaCoNiO (LCCN9173) oxides are tailored as promising anodes by machine learning model, achieving the synergistic enhancement of water oxidation reaction kinetics and proton conduction, which is confirmed by comprehensively analyzing experiment and density functional theory calculation results. Furthermore, the anodic reaction mechanisms for P-SOECs with these anodes are elucidated by analyzing distribution of relaxation time spectra and Gibbs energy of water oxidation reaction, manifesting that the dissociation of HO is facilitated on LBCN9173 anode. As a result, P-SOEC with LBCN9173 anode demonstrates a top-rank current density of 2.45 A cm at 1.3 V and an extremely low polarization resistance of 0.05 Ω cm at 650 °C. This multi-scale, multi-faceted research approach not only discovered a high-performance anode but also proved the robust framework for the machine learning-assisted design of anodes for P-SOECs.