Communication—Contribution of Catalyst Layer Proton Transport Resistance to Voltage Loss in Polymer Electrolyte Water Electrolyzers

Experiments were performed using an in-house developed test-bench equipped with a SP-150 potentiostat and a HCP-803 80A booster from Biologic. Water was circulated at 400 mL min⁻¹ only in the anode cell compartment, while the electro-osmotically dragged water was discarded periodically from the cathodic gas-water separator. The PEWE cell with an active area of 25 cm² was assembled using a Nafion 117 based E400 catalyst coated membrane (CCM) from Greenerity, with an Ir-based catalyst for the anode and the Pt-based catalyst for the cathode. Sintered-Ti PTLs T5, T10 and T20 from GKN were varied on the anode and cathode side. Data describing the cells with different PTLs will be designated with T5/T10/T20. The previous study from Suermann et al.⁵ included X-ray tomography characterization of dry samples, showing that the samples had similar porosity between 30–35%. T5 and T10 have d^particle₅₀ of 68 μm, with T5 having smaller pores, and the "smoothest" surface. The T20 sample had both the largest particles (d^particle₅₀ = 128 μm) and a higher fraction of larger pores. For details we refer to Ref. 3.

CCMs were conditioned for 12 hours prior to measurements by cycling the current density between 1 and 2 A cm⁻² in 5 min intervals at 60°C. The polarization curves were recorded galvanostatically at 60°C, and the high frequency resistance (HFR) was measured at 10 kHz for the iR-correction.

For the determination of R^{H +}_CLa, a N₂-stream of 500 Nml min⁻¹ was injected into the working electrode (WE) water loop, while the reference/counter electrode (RE/CE) was supplied with humidified H₂ at 400 Nml min⁻¹. At the applied DC-bias of the cell, hydrogen evolution takes place at the RE/CE, and oxidation of hydrogen that has permeated through the membrane from the H₂ to the N₂ side at the WE. The DC current in this case is the limiting hydrogen crossover current. The PEWE cell was operated potentiostatically at 1.0, 1.2 and 1.4 V, and the impedance was measured from 10 kHz to 100 mHz. The measurements were performed in the potential region with higher capacitance to eliminate the effects of the inductive elements.⁸ No contribution of the faradaic current was observed at these potentials in the H₂/N₂ regime (cf. Supplementary Material, Figure S3). In this study, we analyze the WE AC response using a one-dimensional transmission-line model consisting of differential elements describing charge transfer and proton transport in the CL. The general model is widely used in fuel cell CL characterization^8–10 and the governing equations are given in detail in the Supplementary Material. The proton transport losses in the CL_a appear as R^{H +}_CLa/3 in the polarization curve, under the assumption of uniform potential distribution in the CL.^8–10 Due to a good agreement between the measured and the modeled impedance response (cf. Supplementary Material, Figure S2), we extrapolated the low frequency, capacitive response to the real axis obtaining the R^{H +}_CLa/3 projections. This procedure was repeated for different PTLs to determine the impact of the CL/PTL interface on R^{H +}_CLa. The R^{H +}_CLc contribution is assumed to be negligible as a result of the low charge transfer resistance of the hydrogen evolution reaction (HER).¹¹ Furthermore, the fast HER takes place at the catalyst/membrane interface, resulting in a negligible effective R^{H +}_CLc.

So far, there have not been any attempts to experimentally determine the R^{H +}_CLa contribution to η_mtx. The values of R^{H +}_CLa/3 are extrapolated from the impedance spectra from the H₂/N₂ operation (Figure 1) (8.1 ± 0.9/11.2 ± 1.8/20.6 ± 2.1 mΩcm² for T5/T10/T20, respectively). The R^{H +}_CLa appears to strongly depend on the interface between the PTL and the CL_a, as the PTL with the coarsest and finest surfaces result in highest and lowest R^{H +}_CLa, respectively. To better comprehend the effect of the PTL/CL_a interface on the cell characteristics, we have conducted an overpotential analysis on the polarization curves.

Zoom In Zoom Out Reset image size

The PEWE polarization curves were analyzed according to the Tafel model, with the procedure and equations given in detail in References 5–7. PEWE cell voltage is a sum of the thermodynamic, reversible cell voltage E_rev(p, T), given by the Nernst equation, and the overpotentials, which can be broken down to kinetic η_act, ohmic η_ohm, and mass-transport η_mtx contributions.⁷ Correcting the cell voltage, E_cell, using the HFR yields the iR-free cell voltage (E_{IR − free}) and η_ohm. The Tafel slope was fitted in the low current density region (0.01–0.08 A cm⁻²) and subsequent subtraction of E_rev(p, T) yielded η_act. The difference between E_{IR − free} and the extrapolated Tafel line yields residual losses, considered to be the results of mass transport losses.^5–7

Polarization curves and the impedance spectra (Figure 2) reveal significant differences in cell performance with different PTLs. Although cells with T5 showed the lowest HFR, they are outperformed by the cells with T10 PTL. The cells with T20 exhibited both the highest HFR and different impedance response at low frequencies (Figure 2b), indicating higher mass transport losses. The difference in HFR most likely stems from a higher ohmic interfacial contact resistance in the cell when using coarser T20 compared to T5. Interestingly, Tafel slopes are somewhat different for the three cells; 71 ± 1, 68 ± 1, 78 ± 2 mV/dec for T5, T10, T20, respectively. PTL-induced structural inhomogeneity in the CL_a might result in the different apparent Tafel slopes. Overall, calculated overpotentials with different commercial PTLs are in line with the previous study from Suermann et al.⁵

Zoom In Zoom Out Reset image size

Since R^{H +}_CLa/3 appear as part of η_mtx, η_mtx was corrected for the R^{H +}_CLa contribution from the H₂/N₂ measurements to extract η_{mtx − rest}. R^{H +}_CLa contributes 11, 18 and 21% to η_mtx at 3 A cm⁻² for the cells with T5, T10 and T20 PTLs, respectively. The η_{mtx − rest} still accounts for the major contribution to η_mtx at high current densities and differs between the three configurations, being highest for T20 and lowest for T10 (Figure 3). A coarse PTL_a surface appears to be detrimental to the cell performance indicators. We offer the following tentative explanation: as the CCM deforms between the rigid Ti-PTLs, the CL_a area under the solid PTL particles is compressed and its porosity is thereby reduced. The surface of T20 consists of larger particles, leading to more pronounced local reductions in porosity. We assume that proton transport takes place in the compacted CL_a structure, as it is the area under the PTL particle in contact with the CL_a that contributes to the OER.^12,13 The water reaches the CL_a through the voids on the PTL surface and relies on lateral diffusion through the ionomer binder to reach the active sites under the PTL solid particles (Figure 4). The T20 PTL surface consists of larger particles and pores,⁵ and would in turn result in larger diffusion distances for the water in the CL_a ionomer, and thus higher η_{mtx − rest}. The different η_{mtx − rest} for the T5 and T10 might stem from the more constrained T5 structure, with a higher fraction of small, potentially bottlenecking pores.⁵ The compaction of the CL_a under the PTL particle leads to higher degree of ionomer binder confinement and inhomogeneous wetting of the CL_a active sites, resulting in increased R^{H +}_CLa.^14,15 SEM images of the CL_a surface and CCM cross-sections are presented in the Supplementary Material to give more insight into the impact of the interface morphology on the CCM physical properties.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size