Comparing PEMFC state-of-health characteristics obtained by Galvanostatic Charge Method with and without nitrogen flush

Abstract

Onboard determination of state-of-health (SoH) parameters becomes increasingly important with tightened lifetime requirements for automotive polymer electrolyte membrane (PEM) fuel cell systems. One way to electrochemically characterize SoH parameters like active catalyst surface area (roughness factor, RF), double layer capacitance and membrane permeability is the nitrogen-flush-requiring Galvanostatic Charge Method (GCM). In this publication, a nitrogen-flush-free and thus, onboard-compatible adapted version of GCM (AGCM from Möller et al. (2023)), is directly compared to its original method, GCM. A 20-cell automotive shortstack was frequently characterized with both GCM and AGCM methods as well as standard Cyclic Voltammetry (CV) during a degradation test. Additionally, the evaluation model for AGCM was varied in two aspects: Firstly, modeling the hydrogen partial pressure difference (ΔpH2) responsible for hydrogen crossover and secondly, modeling hydrogen diffusion in the cathode volume with a 0-D transport model. The characterization results prove that all investigated model variations of AGCM can reproduce double layer capacitance results from GCM and CV, as well as the decreasing trend in roughness factor results. RF-values characterized by AGCM agree with either GCM or CV results, depending on the evaluation model. A thorough analysis revealed the neglected short circuit resistance in combination with the chosen modeling approach of ΔpH2 in AGCM as the reason. The two modeling approaches for ΔpH2 base on extreme assumptions about the homogeneity of hydrogen distribution in the anode volume. Neither of them could reproduce the membrane permeability of GCM. Instead, one overestimates, one underestimates it, indicating the truth to be in between the two extreme models. However, AGCM measurement conditions are identified as a lever to improve the AGCM permeability results and their robustness, even if the modeling approaches remain unchanged. Furthermore, this study showed that neglecting hydrogen diffusion in the cathode volume in the AGCM evaluation model leads to overestimation of the membrane permeability.