Predicting MEA Catalyst Layer Performance at 120 °C Using a Pressurized GDE Testing Platform and Machine Learning

The performance of Pt-based oxygen reduction reaction (ORR) catalysts in proton exchange membrane fuel cells (PEMFCs) arises from complex interactions among Pt nanoparticles, carbon supports, and proton-conductive ionomer. Membrane electrode assembly (MEA) testing is the industrial benchmark for catalyst evaluation but is time- and resource-intensive, limiting systematic development. Here, we utilize our pressurized gas diffusion electrode (GDE) platform that enables medium-throughput exploration of catalyst layer fabrication parameters for PEMFC operation at elevated temperature conditions (120 °C) relevant for heavy-duty applications. By combining a novel type of catalyst layer fabrication procedure with machine learning methods, we systematically optimized five critical fabrication parameters, achieving a tenfold performance improvement of a commercial benchmark Pt/C catalyst (TEC10E30E) as compared to initial, non-optimized procedures adopted from lower temperatures. Even more important, the parameter–performance relationships established in the GDE setup translated well to MEA testing, where our optimized catalyst layer outperformed the commercial benchmark at 120 °C, demonstrating the potential of our optimization strategy, while the less optimized layers followed the expected performance order. These results establish medium-throughput studies using the pressurized GDE in combination with machine learning as a predictive and cost-effective tool for accelerating the development of next-generation fuel cell catalysts.