4D Structural Characterization of Mechanical Degradation in Reinforced Fuel Cell Membranes Using in Situ Visualization

Abstract

Perfluorosulfonic acid (PFSA) ionomer membranes in fuel cells are susceptible to operational mechanical stresses resulting in fatigue and/or creep failures that compromises their durability and lifetime. The fatigue based mechanical degradation is typically a result of repeated wet/dry humidity cycles that cause micro crack initiation and /propagation within the membrane [1]. Fatigue induced membrane failure in the form of cracks/tears/pinholes leads to a gradual increase in gas crossover and ultimately to fuel cell failure. Membranes with less conductive mechanical reinforcements have been developed to alleviate mechanical degradation yielding demonstrated improvements in lifetime and durability. Nevertheless, development of membrane damage remains a critical failure mode and the fundamental understanding of membrane mechanical degradation is a subject of ongoing research. Scanning electron microscopy (SEM) based studies have been used to characterize the degradation-induced structural changes in fuel cell membranes; however, SEM imaging is inherently destructive and inhibits any tracking of structural changes at a particular location over time. Hence, membrane degradation evolution studies are limited to ex situ analysis at various stages of degradation and with different samples. Recently, laboratory-based X-ray computed tomography (XCT) was introduced as an alternative imaging technique, which has enabled three-dimensional (3D) failure analysis of fuel cell membranes revealing novel insights on membrane failure [2,3]. In the present work, the XCT-based 3D failure analysis approach is extended to an in situ investigation of pure mechanical membrane degradation by utilizing a custom designed fixture. This X-ray transparent fixture houses a gas diffusion electrode (GDE) based MEA with a reinforced membrane, which is subjected to wet/dry cycling of N2 gas flowing through both anode and cathode sides, thereby producing a pure mechanical fatigue type degradation within the membrane. XCT-based 3D identical location tracking of membrane morphology as a function of degradation time facilitates a novel four-dimensional (4D) in situ workflow [4], which enables the characterization of the damage growth or evolution. Preliminary results show that no through- thickness membrane cracks developed until 3000 wet/dry cycles. However, minor crazes initiated on the cathode side membrane surface (Fig. 1) between 2000 and 2500 cycles. Crack initiation and growth within the reinforced membrane are comprehensively examined from various perspectives by simultaneously studying the two-dimensional (2D) planar and cross-sectional views. A clear interaction of membrane cracks with defect features, such as delamination and catalyst layer cracks is observed. Furthermore, the centrally located reinforcement layer is found to restrict the through-thickness growth of membrane cracks at several locations in the early stages of damage initiation. Overall, the size and density of membrane crack formation at a given number of cycles is found to be considerably reduced with the use of a reinforced membrane when compared to a non-reinforced membrane. A detailed study to understand the variation in degradation mechanisms between reinforced and non-reinforced membranes is carried out. Overall, the work summarized here is a unique study on the evolution of reinforced membrane degradation with a 4D perspective. The new findings from this work demonstrate the distinct advantage of XCT technology in gaining an improved fundamental understanding of membrane degradation by capturing critical failure modes and mechanisms at their different developmental stages. Acknowledgement This research was funded by the Natural Sciences and Engineering Research Council of Canada, Canada Foundation for Innovation, British Columbia Knowledge Development Fund, and Ballard Power Systems through an Automotive Partnership Canada (APC) grant. This research was undertaken, in part, thanks to funding from the Canada Research Chairs program. The authors thank Kevin Dahl and Alex Boswell for technical support.