A Non-Ideal Transport Model of Bipolar Membranes to Understand and Manage Multi-Ion Interactions in Water Electrolysis

Water electrolysis provides a means of producing H2 gas without carbon emissions. However, H2 production through water electrolysis is more costly compared to methods using fossil fuels. A key element to the cost of water electrolysis is the need to purify the water fed to the electrolyzer. Impure feeds can lead to corroding electrodes (e.g. chloride oxidation) and fouling of commonly-used monopolar membranes (i.e., cation and anion exchange membranes). Bipolar Membranes (BPMs), consisting of a cation and anion exchange membranes with a water dissociation catalyst at the interface, pose unique advantages. Because each layer can be tuned independently, BPMs enable higher control of co-ion transport and impurities, unlike monopolar membranes. However, challenges arise in rationally-designing BPMs for impure water electrolysis due to a lack of fundamental understanding of the coupled interactions between water, mobile ions, and fixed charge groups. In this study, we develop a continuum model to systematically investigate the use of BPMs for water electrolysis with impure feedstocks (e.g. seawater). The model defines the flux of each species using a modified Nernst-Planck-Poisson framework while explicitly considering non-ideal contributions to chemical potential from ion-ion, ion-membrane, swelling, electrostatic, steric, and solvation interactions to the electrochemical potential. The simulations quantify the effect of these interactions and membrane properties on ion transport, and how they could be modulated to mitigate the deleterious effects of impurities and co-ions. Our analysis shows that simple dilute-solution theory is not capable of predicting the effect of impurities on BPM water electrolysis and identifies the dominant interactions that dictate transport in BPMs. This study provides guidelines for modelling water electrolysis and materials design strategies for water electrolysis in non-ideal conditions.