Lab-based X-ray micro-computed tomography coupled with machine-learning segmentation to investigate phosphoric acid leaching in high-temperature polymer electrolyte fuel cells.

A machine-learning approach is used to segment 14 X-ray computed-tomography datasets acquired by lab-based scanning of laser-milled, high-temperature polymer electrolyte fuel cell samples mounted in a 3D-printed sample holder. Two modes of operation, one with constant current load and the other with current cycling, are explored and their impact on microstructural change is correlated with electrochemical performance degradation. Constant-current testing shows the overall quantity of phosphoric acid in the gas diffusion layers is effectively unchanged between 50 and 100 h of operation but that inter-electrode distribution becomes less uniform. Current-cycling tests reveal similar quantities of phosphoric acid but a different intra-electrode distribution. Membrane swelling appears more pronounced after current-cycling tests and in both cases, significant catalyst layer migration is observed. The present analysis provides a lab-based approach to monitoring microstructural degradation in high-temperature polymer electrolyte fuel cells and provides a more accessible and more statistically robust platform for assessing the impact of phosphoric acid mitigation strategies. • Lab-based X-ray micro-computed tomography to investigate phosphoric acid leaching. • Machine learning segmentation yields quantification of microstructural changes. • Electrochemical performance loss linked to microstructure evolution. • Redistribution of phosphoric acid depends on fuel cell operation mode. • Accelerated stress tests give more membrane swelling and less uniform Pt migration. [ABSTRACT FROM AUTHOR]