During fuel cell operation, proton-exchange membranes (PEMs) are subjected to aggressive chemical environment and mechanical fatigue that alter their properties and may lead to performance loss or, in the worst cases, fuel cell shutdown due to membrane failure. Membrane degradation due to chemical and mechanical stresses remains one of the main factors limiting PEM fuel cell lifetime [1]. Chemical damages are mostly due to radical attacks leading to the scission of the polymer chains (backbone or/and side chains), while mechanical stress comes from repeated swelling/shrinkage cycles caused by the membrane water-uptake. Despite quantities of works focusing independently on chemical (most often) or mechanical (more rarely) degradations, only a few studies were published on the effect of both combined [2]–[4]. Among them, Kusoglu et al. demonstrated that a static compression affects its microstructure and accelerates the chemical decomposition of the polymer chains in Nafion® membranes [2].In this study, we investigate the impact of both chemical and mechanical degradations on the morphology and physicochemical properties of Nafion® membranes: chemical structure, as well as water transport and sorption. The membrane degradation process was induced by a custom-made device that mimics the operating conditions of the fuel cell by exposing the membrane simultaneously to a free radical environment and cyclic compression. The generation of free radical environment is based on the circulation of a continuous flow of hydrogen peroxide or Fenton’s solution (i.e. hydrogen peroxide and ferrous ions), while mechanical fatigue is induced by a cyclic compression to reproduce the swelling/shrinkage sequences entailed by the membrane water-uptake during transient fuel cell operation.Preliminary studies were necessary to determine the appropriate experimental conditions, both from a chemical and a mechanical point of view. Firstly, an investigation similar to the one of Frensch et al. was performed to determine the effect of Fenton’s reagent concentrations on membrane degradation [5]: fluoride emission measurements showed that a higher degradation rate was obtained at a low iron ions concentration. Moreover, at low iron concentration, the degradation rate depended also significantly on the hydrogen peroxide concentration. On the other hand, high hydrogen peroxide concentration (≈ 30 %vol) entailed strong membrane morphology modification with the appearance of many bubbles - with a diameter varying from some micrometers to several millimeters- at the surface of both membranes [6]. Such morphology evolutions cannot be considered as representative of the degradations occurring during fuel cell operation.The first mechanical and chemical degradation tests were carried out using a 0.0503 cm3.s-1 flow of either pure water or 3%vol diluted H2O2 through two thermostated (80°C) half-cells. The half-cells and the membrane were inserted between the clamps of an electromechanical testing machine (MTS load frame model 312.21) and the mechanical stress consisted in a sinusoidal constraint between 0 and 5 MPa at 0.1 Hz. As expected, the application of this sinusoidal constraint increased significantly the degradation rate (Figure): the fluoride emissions were 5 to 7 times higher than under chemical stress only. Indeed, Nafion® XL’s degradation rate increases from 11 µg/h/gNafion to 68 µg/h/gNafion when mechanical stress was applied in addition to chemical stress. Likewise, the values increased from 11 to 54 µg/h/gNafion with NR-211. Moreover, with the combination of cyclic compression and H2O2 solution exposure, the degradation rates approached the values obtained with a Fenton’s reagents exposure.[1] R. Borup et al., Chem. Rev., 107 (10), 3904–3951, 2007.[2] A. Kusoglu et al., ECS Electrochem. Lett., 3 (5), F33–F36, 2014.[3] S. velan Venkatesan et al., J. Electrochem. Soc., 163 (7), F637–F643, 2016.[4] V. M. Ehlinger et al., 166 (7), F3255–F3267, 2019.[5] S. H. Frensch et al., J. Power Sources, 420, 54–62, 2019.[6] S. Mu et al., J. Appl. Polym. Sci., 129 (3), 1586–1592, 2013. Figure 1
Read full abstract