Towards longevity in solid oxide electrolysis cells: multi-scale modeling and machine learning for degradation diagnosis and mitigation

Abstract

Solid oxide electrolysis cells (SOECs) have become a key technology for the conversion of renewable electrical energy to green hydrogen due to their high efficiency. However, rapid degradation during long-term operation has limited their industrial application. The degradation of SOECs during long-term operation is dominated by a cross-scale coupling mechanism, the scientific essence of which involves atomic-level interface reconstruction, mesoscale microstructural evolution, and the synergistic interaction of macroscopic multi-physical fields. In response to the complexity of the mechanisms, multiscale modeling serves as a progressive tool for understanding the detailed mechanisms of SOECs from the microscopic to the macroscopic scale. Artificial intelligence (AI) methods have demonstrated significant potential in the research of SOECs at different levels, reducing the computational resource constraints on traditional modeling to varying degrees. This paper summarizes the latest progress in the application of multiscale modeling and AI methods in SOECs research, with a focus on the microscale degradation phenomena and mechanisms during SOECs operation. It aims to enhance the understanding of the degradation mechanisms of SOECs, integrate the application of multiscale modeling and AI methods in SOECs research, and improve the long-term stability of SOECs.