Neutron Imaging of Isothermal Sub-Zero Degree Celsius Cold-Starts of a Polymer Electrolyte Fuel Cell (PEFC)

As a major candidate for main power supply in future automotive applications, the Polymer Electrolyte Fuel Cell (PEFC) must be able to start up and operate at subzero degree Celsius temperatures. Advantageously, its polymer electrolyte membrane offers sufficient proton conductivity at these low temperatures, so that a reasonable current can be drawn. Main drawback is that the water produced can freeze and accumulate in the catalyst layer by covering the surface area and filling the pores. In the worst case, this can completely block the access of oxygen to the active site so that the voltage drops and that no more power can be extracted. In practice, technical stack thermal management is so that electrochemically produced heat power is sufficient to reach the melting point before the voltage drops, this without using any auxiliary heating device. Mechanisms explaining the clogging of the catalyst layer are though not fully understood yet. Models developed to that issue must absolutely be verified by experimental data, especially from visualization studies which have been quite scarce up to now. Neutron imaging has emerged in the past decade as a powerful diagnostic tool for investigating in-situ spatiotemporal distribution of water in PEFCs, due to its good spatial and temporal resolution as well as the strong contrast of water it provides, compared to other constitutive materials of the cell. Taking benefit of the last improvements made at ICON beam line [1] of the Paul Scherrer Institut (PSI) that permit to use a spatial resolution of 20 μm (FWMH value) with 10s exposure time [2] by keeping sufficient quality, in-plane imaging of isothermal cold-starts were performed in this study. Isothermal mode is preferred as it allows being independent of any thermal design implemented in practice. In Figure 1 are displayed the voltage evolution along with neutron radiograms during a sub-zero start-up at -10°C and 0.l A/cm. The striking feature is the apparition of a condensed phase of water out of the membrane-electrodeassembly (MEA) in the diffusion layer (GDL) and even in the channels which is rarely mentioned in the literature. Far from a case where the MEA water storage capacity only would determine the working time, the existence of water removal flows towards other bigger storage volumes of the cell allows achieving higher working times, around 100 min in this case. These results strongly suggest that a capillary transport must exist to evacuate the water from the catalyst layer to the different accumulation locations, meaning that liquid water is present at this temperature. After having reviewed and excluded the different assumptions that could support the presence of liquid state of water in this context [3], the presence of super-cooled water is found to be the most probable explanation. This meta-stable state of water is well known to appear under specific conditions, amongst whose PEFC sub-zero start-up has already been reported [4]. Super-cooled water freezing is known as a probabilistic and fast phenomenon than can be induced by various external conditions. Further cold-starts realized without neutron imaging (approx. 400 start-ups) exhibit a stochastic distribution of working times under various conditions or voltage drops induced by mechanical shocks on the system, as it was four times applied in our experimental campaign. Therefore, we attribute that the voltage drop is the consequence of the sudden freezing of super-cooled water in most cases studied here.