Explainable machine learning unveils a critical trade-off in SOFCs: The role of cathode-to-anode reaction site ratio

Solid oxide fuel cell (SOFC) research recognizes the critical importance of electrode reaction sites for cell performance. This study, for the first time, systematically investigates the influence of the cathode to anode reaction site ratio, denoted as λ, on SOFC performance and reliability. The ratio λ is defined as the thickness scaled ratio of cathode double phase boundary (DPB) area to anode triple phase boundary (TPB) length. Using integrated experiments, multiphysical modeling, and explainable AI (XAI), we characterize the effects of λ on maximum power density and failure probability under various operating conditions. Results demonstrate that λ is the most significant factor affecting power density and substantially influences failure probability. A strong nonlinear relationship exists between λ and both metrics, with optimal λ shifting dynamically with temperature and gas flow. At higher temperatures (T≥1023.15 K) and sufficient flow (Q≥50 SCCM), λ of approximately 3000 nm enables high power density (>1.1 W/cm²) and low failure probability (<0.01). At T=973.15 K, maintaining λ near 1750 nm improves power output by over 15 %. This work reveals a fundamental performance reliability trade off governed by λ and identifies key microstructural parameters for both electrode optimization.