Mathematical Modelling and Experimental Analysis of Thin, Low-Loading Fuel Cell Electrodes

The use of expensive platinum in PEFC remains a major barrier to achieve a low cost fuel cell stack. To reduce stack cost, thin, low loading electrodes have been studied at various laboratories [1]. Experimental data showed that the cell performance is severely limited for cathode electrode loadings below 0.1 mg/cm². This decreased performance has been attributed to several physical processes [2]. First, the oxygen reduction reaction (ORR) is a multi-step reaction that does not follow Tafel kinetics. Low loading electrodes might operate following a different ORR pathway. Second, due to the higher local mass flux to the Pt sites, mass transport resistances in the ionomer thin film, presumably covering the catalytic sites, might limit oxygen transport to the reaction sites. Finally, water management issues are exacerbated due to the reduced thickness and thereby increased water production rate per unit volume.

Experimental and theoretical investigations in thin, low loading fuel cell electrodes are being performed in our laboratory. Thin, low loading fuel cell electrodes of various ionomer and Pt loadings, and Pt/C content were fabricated and tested under a variety of operating conditions. In order to accurately control catalyst/ionomer loading and layer uniformity, the electrodes were deposited on a membrane using a material inkjet printer. Results show that cells with cathode Pt loadings of 0.025 mg/cm² are severely limited in the kinetic region with cell voltages as low as 0.6V at 200 mA/cm² under hydrogen/air conditions at ambient pressure. At cell voltages of 0.2 V however, mass activities as high as 50 A/mg_Pt and current densities above 1 A/cm²were observed [3].

Macro-scale mathematical modelling studies of low loading electrodes have either been based on simplified one-dimensional models, or have relied on simple Tafel kinetics for the ORR. In this work, meso- and macro-scale simulations of thin, low loading electrodes are performed using OpenFCST, an open-source fuel cell simulation software developed by our research group [4]. A multi-dimensional, non-isothermal, two-phase membrane electrode assembly (MEA) model was developed. The Butler-Volmer/Tafel representations of the HOR and ORR were replaced by multi-step reaction kinetic models, i.e., the dual path and double trap models proposed by Wang et al. and modified by our group [5]. To account for thin film mass transport resistances, a pseudo-agglomerate model that accounts only for the ionomer thin film was implemented. The model is justified by microscopy observations and our previous mathematical modelling results that suggest that, based on the expected agglomerate size, the ionomer thin film is the most important micro-scale parameter [6]. The use of multi-step kinetic and ionomer thin film models improves fuel cell predictions for both conventional and low loading electrodes in the kinetic, ohmic and mass transport regions.

Water build-up in thin, low loading electrodes is likely to be exacerbated. Several researchers have proposed a variety of pore-size distribution (PSD) based models to account for the layers' micro-structures [7]. These models however have not yet been implemented in a multi-dimensional MEA model. A dual wettability PSD based model was implemented in OpenFCST. The PSD based model has been compared to the common saturation-based two-phase flow model in porous media, e.g., [8]. Preliminary results show that the PSD model is able to identify discontinuities in the saturation profiles inside the MEA as a result of the MPL hydrophobic nature and small PSD. The saturation-based model on the other hand, provides a continuous saturation profile.

Focused ion beam-scanning electron microscopy (FIB-SEM) and nano-computed tomography (nanoCT) have been performed on our low loading electrodes. A statistical analysis tool was developed and used to compute multiple correlation functions to assess porosity, active area and pore connectivity. Using the statistical correlation functions and a novel different phase neighbors based simulated annealing strategy, meso-scale reconstructions of the electrode have been created [9]. The electrode reconstructions can be used to estimate the effective transport properties in each phase of the electrode using OpenFCST.

[1] J. Wee et al., J. Power Sources, 2007, 165(2) 667.

[2] J. Owejan et al., JES, 2013, 160(8), F824.

[3] S. Shukla et al., Electrochimica Acta, 2015, 156:289.

[4] M. Secanell et al., ECSTrans., 2014, 64(3): 655.

[5] J. Wang et al., JES., 2006, 153(9), A1732; J. Wang et al., J. Phys. Chem. A, 2007, 111, 12702; Moore et al., JES, 2013, 160(6), F670.

[6] M. Moore et al., JES., 2014, 161(8):E3125.

[7] A. Weber. JES., 2004, 151(10), A1715.

[8] D. Natarajan, JES, 2001, 148(12), A1324.

[9] L. Pant et al., Phys.Rev.E, 2014, 90, 023306.