Enhanced Performance of Protonic Ceramic Electrolysis Cells through a Machine-Learning-Designed Anode and an Optimized Half-Cell Preparation Process

Protonic ceramic electrolysis cells (PCECs) are promising devices for hydrogen production, yet their performance is still limited by the anode materials and fabrication bottlenecks of the half cell. In this study, the electrochemical performance of PCEC is enhanced by integration of the machine-learning-assisted, screened anode and the optimized half-cell preparation process. La0.8Sr0.2Co0.8Ni0.2O3-δ (LSCN8282) oxide was screened with the KNN model as a promising anode owing to its high hydrated proton concentration, abundant surface oxygen vacancies, and mixed Co3+/Co4+ states that simultaneously boost hydration kinetics and oxygen evolution activity. The half-cell fabrication process was further systematically optimized by raising the sintering temperature and introducing a sieving step, yielding a dense electrolyte and porous cathode. The resulting PCEC with LSCN8282 anode reached 1.74 A cm–2 at 1.3 V and 650 °C with an ASR of only 0.04 Ω cm–2, outperforming most famous anodes. This work demonstrates the practical potential of a closed-loop machine-learning/experimental strategy for highly efficient energy conversion devices.