Ramchandra Gawas,
Douglas I. Kushner,
Xiong Peng* and
Rangachary Mukundan*
Energy Storage and Distributed Resources Division, Lawrence Berkeley National Laboratory, Berkeley, California, USA. E-mail: xiongp@lbl.gov; rmukundan@lbl.gov
First published on 15th April 2025
Understanding hydrogen permeation in proton exchange membrane water electrolyzers (PEMWEs) operating at high differential pressures (>25 bar) is critical towards developing effective gas recombination strategies that enable safe operation and high efficiency. Developing this understanding relies on accurate quantification of hydrogen crossover rates in water electrolyzers operating under such conditions. In this work, we show that PEMWEs operating at high differential pressures exhibit noticeable hydrogen oxidation reaction (HOR) currents. As the HOR consumes part of the permeated hydrogen at the anode, neglecting HOR currents leads to severe underestimation of the hydrogen crossover rate. We implemented a new method combining hydrogen oxidation current with online gas chromatography measurements to accurately quantify hydrogen crossover rates as a function of operating current density in PEMWEs operating at high differential pressures (10–30 barg).
Broader contextThe proton exchange membrane water electrolyzer (PEMWE) remains the emergent technology for clean hydrogen production due to its ability to dynamically follow loads, enabling facile integration with renewable energy sources. Moreover, operation at high current densities and high differential pressure further make this technology very attractive for low-cost hydrogen production. While the use of thin membranes improves the efficiency of PEMWE systems by reducing ohmic losses, it also exacerbates safety concerns associated with hydrogen crossover. This study investigates the hydrogen crossover phenomenon in PEM water electrolyzers operating at high differential pressure using electrochemical techniques combined with online gas chromatography. Our results show that PEMWE cells exhibit hydrogen oxidation reaction (HOR) currents under differential pressure. This HOR current, only observed at differential pressure, is diffusion limited and consumes a significant portion of the permeated hydrogen. Overlooking these HOR currents could lead to severe underestimation of the hydrogen crossover rates in PEMWEs operating at high differential pressure. This work demonstrates the importance of HOR current and proposes a new method that integrates HOR current measurement with online gas chromatography for accurate quantification of hydrogen crossover rates at various operating current densities and differential pressures. |
Existing studies to investigate hydrogen permeation in PEMWEs operating at differential pressure employ online gas chromatography or mass spectrometry measurements to quantify hydrogen concentration at the anode, reported as %H2 in O2.10,15,16,19,20 This hydrogen concentration is then converted to hydrogen permeation flux as a function of current and differential pressure. Trinke et al. used this methodology to measure the hydrogen permeation rate in PEMWEs utilizing a 250 μm PFSA membrane as a function of differential pressure, current density, and temperature.20 They found that the hydrogen permeation rate in PEMWEs increases linearly with current density and attributed this increase to supersaturation of hydrogen in the ionomer phase. The hydrogen permeation rate measured by Wrubel et al.15 as well as Martin et al.16 in 50 μm PEMs also exhibited a similar increase with current density. However, most of these measurements overlook the possibility of hydrogen oxidation at the anode catalyst layer. If some of the permeated hydrogen is oxidized at the anode, hydrogen concentration measurement at the anode outlet would be insufficient to quantify hydrogen permeation rates through the PEM. Although previous studies have shown facile hydrogen oxidation reaction (HOR) kinetics on metallic Ir surfaces,21–23 the hydrogen oxidation reaction on commercial oxygen evolution reaction (OER) electrocatalysts i.e., iridium oxide surfaces, remains unexplored. The existing PEMWE literature assumes that iridium (Ir) exists as iridium oxide (IrOx) under the operating potential of the anode in an electrolyzer, with negligible HOR occurring on this IrOx surface.18,19 However, the permeated hydrogen can potentially reduce the iridium oxide catalyst in the anode catalyst layer and form surface metallic Ir sites that can facilitate the HOR.24–26 In this work, we show conclusive evidence of the hydrogen oxidation reaction (HOR) at the anode catalyst layer in PEMWEs operating at high differential pressure (10–30 barg). Our careful investigation reveals that the HOR, only observed at differential pressures, is diffusion-limited and consumes a significant portion of permeated hydrogen. These hydrogen oxidation current measurements were combined with online gas chromatography measurements to accurately quantify hydrogen permeation rates in PEMWEs at various operating current densities and differential pressures (0–30 barg).
Measuring hydrogen crossover at high differential pressures is significantly more challenging than simply evaluating the electrolyzer cell performance.27 The presence of any small leak in the system can lead to the loss of hydrogen and severely underestimate hydrogen concentration measured at the anode. Taking this into consideration, we carefully modified the electrolyzer components and the testing setup to ensure leak-free and precise measurements at high differential pressures. The ESI,† contains a detailed description of the catalyst-coated membrane fabrication process, modified electrolyzer cell components (Fig. S1, ESI†), test station with a custom backpressure unit (Fig. S2, ESI†), and measurement protocols used for electrochemical testing as well as online gas chromatography measurements (Fig. S3, ESI†). Each membrane electrode assembly (MEA) was identically conditioned prior to performance and hydrogen crossover evaluation at ambient and high differential pressure (10–30 barg) to ensure consistency.
Fig. 1(a) and (b) show the PEMWE cell performance and the high-frequency resistance (HFR) at 0–30 barg cathode backpressures. The high-frequency resistance (HFR) was extracted from the galvanostatic electrochemical impedance spectra (GEIS) performed at each current density. The ambient cell performance of 1.88 V at 3 A cm−2 shows good agreement for MEAs at comparable Ir loadings (0.4 mgIr cm−2) and membrane thickness (127 μm).28,29 The reversible cell voltages of PEMWE operating at high cathode backpressures are expected to increase due to elevated hydrogen partial pressures (Table S1, ESI†). This effect has been discussed in detail in Section 1(d) of the ESI.† HFR-corrected polarization curves demonstrate this expected Nernstian shift at high differential pressures (Fig. 1(c)). However, cell voltages in Fig. 1(a) do not exhibit a notable difference under different cathode backpressures at currents above 1 A cm−2. The HFR-free overpotentials, i.e., polarization curves corrected for both the iR-drop and the Nernstian shift (Fig. 1(d)), demonstrate the impact of differential pressure on electrode kinetics, with ∼ 60 mV improvement at 4 A cm−2 and 30 barg. This improvement offsets the Nernstian penalty at high cathode pressures (62 mV at 30 barg), resulting in identical overall cell performance in Fig. 1(a). A full voltage breakdown analysis needs to be performed in order to attribute these voltage gains to specific electrochemical phenomena.
The HFR-free overpotentials can be further deconvoluted into kinetic and residual (transport) contributions.30,31 Kinetic contributions are extrapolated using the Tafel equation with kinetic parameters extracted from HFR-free overpotentials at low currents (<80 mA cm−2). Fig. S4 (ESI†) shows the voltage breakdown analysis at ambient backpressure showing ohmic, kinetic and residual overpotentials. However, this analysis wasn’t feasible at high differential pressures i.e., 10–30 barg, as we observed that the cell voltages dropped below 1.0 V at low currents (Fig. S5, ESI†). Furthermore, cyclic voltammograms (CVs) of the anode measured at 10–30 barg cathode backpressures showed only positive currents in both the forward and reverse directions. CVs of PEMFC cathodes also show such positive currents due to the background hydrogen oxidation current (Fig. S6, ESI†). These two observations pointed towards the oxidation of permeated H2 at the anode in electrolyzers operating at differential pressure even when utilizing IrOx catalysts. Although the hydrogen oxidation reaction (HOR) is relatively well characterized on metallic Ir surfaces,21–23 HOR on iridium oxide surfaces remains poorly understood. Fig. S7 (ESI†) shows rotating disk electrode (RDE) measurements in H2-saturated acidic electrolytes confirming the presence of HOR on the TKK amorphous IrOx surface. These measurements, discussed in detail in Section S2a of the ESI,† reveal that IrOx thin film electrodes exhibit similar HOR activity to metallic Ir or Pt/C.
We systematically analyzed the hydrogen oxidation reaction (HOR) current in PEMWE single cells at 10–30 barg cathode backpressures using linear sweep voltammetry at 1 mV s−1 from open circuit voltage (OCV) until the oxygen evolution reaction (OER) onset potential. The slow scan rate, 1 mV−1, was chosen to minimize contributions from the double layer capacitance and other faradaic reactions on the surface. Fig. 2(a) shows HOR currents for three commercial anode catalysts as a function of anode potential at 10, 20, and 30 barg cathode pressures. These three OER electrocatalysts were chosen to represent three kinds of iridium oxide catalysts – amorphous iridium oxyhydroxide (Tanaka Kikinzoku (TKK) amorphous IrOx), rutile iridium dioxide (TKK rutile IrO2), and iridium oxide with some metallic Ir (Alfa Aesar (AA) IrOx). The HOR polarization curves, irrespective of the catalyst and differential pressure, exhibited three distinct regions – (i) from OCV to 0.1 V showing very facile HOR kinetics, (ii) from 0.1 to 1.4 V showing diffusion limited HOR current and (iii) beyond 1.4 V showing OER current. The diffusion-limited HOR current is proportional to the differential pressure as H2 permeation without any water oxidation current is largely driven by diffusion. As the most ubiquitous PTL coating layer, Pt, is also an active HOR catalyst, we used Au-coated PTL at the anode with an IrOx anode catalyst layer to eliminate any HOR contributions from Pt present as the coating.32 The PEMWE single cells consisting of the HOR inactive Au-coated PTL showed similar HOR currents as Pt- and Ir-coated PTLs. Fig. 2(b) further shows that HOR current is independent of the PGM-coating material (Pt, Ir, Au) confirming that IrOx catalyzes hydrogen oxidation at the anode. We would like to clarify that HOR current was observed only at differential pressures (Fig. S8, ESI†); HOR current was negligible under ambient pressure due to the lack of hydrogen at the anode.
Furthermore, we coupled HOR measurements with online gas chromatography (GC) to assess whether the mass transport-limited HOR consumes all permeated H2. These GC measurements were taken using Ar as the sweep gas, as shown in Fig. S2(b) (ESI†) at various operating potentials and cathode backpressures and converted to H2 flux at the anode or H2 permeation rate (the ESI† contains further details of this measurement in Section S1e). Fig. 2(c) shows zero H2 flux at the anode outlet in the diffusion limited region at all (10–30 barg) cathode backpressures, indicating that the HOR consumes all permeated hydrogen. Thus, the diffusion-limited HOR current represents an electrochemical H2 permeation measurement in PEMs at differential pressures. These H2 permeation rates or HOR currents concur with H2 permeation rates reported by Schalenbach et al. for a 127 μm PFSA membrane (N115) at the corresponding differential pressures.33 Note that the diffusion limited HOR current measures H2 permeation rate at zero operating current (no water oxidation or OER). Fig. 2(c) also shows cathode backpressures (dotted line) recorded during HOR measurements. Interestingly, we did not observe a decrease in the cathode backpressure despite H2 permeation from the cathode to the anode. The HOR at the anode generates protons that come back to the cathode and regenerate H2 by the hydrogen evolution reaction. These observations suggest that the PEMWE functions as a hydrogen pump at differential pressures and voltages below 1.4 V (Fig. S5, ESI†). Moreover, cathode backpressure retention during the hydrogen pump operation confirms that our system does not have any leaks. Hydrogen pump operation was also observed during low-current galvanostatic measurements at high differential pressure when the applied current was below the diffusion limited HOR current at the corresponding cathode backpressure (Fig. S5, ESI†). H2 flux at the anode at OCV, i.e., without any HOR current, depicting H2 permeation rate, showed reasonable agreement with the HOR currents at 10–30 barg cathode backpressures.
The existing PEMWE literature widely neglects HOR currents while analyzing polarization curves and reporting faradaic efficiencies, which is valid only for ambient pressure operation. However, our careful investigations reveal that PEMWEs operating at high differential pressure exhibit noticeable HOR currents. As more laboratories acquire the capability to operate under differential pressure, the procedures reported in this manuscript can be utilized to quantify the HOR currents. Since the HOR consumes some of the permeated H2, it needs to be accurately measured and accounted for while estimating H2 permeation rates through the PEM (Fig. 3). In the next section, we will discuss the implications of the HOR at the anode on H2 crossover rates in PEMWEs under practical operating currents and 0–30 barg cathode backpressures. We implemented online gas chromatography to quantify the hydrogen volume fraction at the anode. Gas chromatography offers the highest accuracy among techniques used for on-line quantification of hydrogen concentration at the anode outlet.
![]() | ||
Fig. 3 Schematic of the experimental setup showing PEMWE equipped with manual backpressure at the cathode and online gas chromatography to measure hydrogen content at the anode. |
Fig. 4(a) shows the hydrogen volume fraction at the anode (%H2 in O2, dry basis) as a function of current density (0.25 to 3.5 A cm−2) at 0 to 30 barg cathode backpressures with error bars obtained from on-line GC measurements on three independent MEAs. As expected, the hydrogen concentration at the anode increases with cathode backpressure due to an increase in H2 permeation rate and decreases with current density as oxygen from the OER dilutes the hydrogen concentration. The volume fraction of hydrogen in oxygen was converted to H2 flux at the anode using eqn (S1) (ESI†). Fig. 4(b) shows the H2 flux at the anode outlet as a function of current density in comparison with the H2 permeation rate measured at zero OER current, i.e., HOR current at 0–30 barg cathode backpressures. Previously, this H2 flux at the anode outlet measured using online gas chromatography was solely projected as the H2 permeation rate. However, Fig. 4(b) clearly shows that GC measurements alone severely underestimate the H2 permeation rate at high differential pressures. The H2 flux at the anode outlet at low current densities is significantly smaller than the ex situ H2 permeation rate at high differential pressures (indicated by crosses at zero current at 10–30 barg). This discrepancy cannot be explained by a recombination reaction on Pt in the PTL without considering the hydrogen oxidation reaction at the anode. The MEAs with an Ir-coated PTL also show significantly lower H2 flux at the anode outlet than ex situ permeation rate measurements (Fig. S9, ESI†). In fact, identical H2 fluxes at the anode outlet for MEAs with Pt and Ir-coated PTLs suggest that Pt present in the PTL is not an effective recombination catalyst. Since, some of the permeated H2 is oxidized at the anode catalyst layer, the HOR current needs to be considered when computing H2 crossover rates in PEMWEs. Assuming that the diffusion-limited HOR current would not change beyond 1.5 V, we computed the H2 permeation rate by adding the HOR current to the H2 flux at the anode calculated using in operando GC measurements. Fig. 4(c) shows H2 permeation rate as a function of current density for 0–30 barg cathode backpressures illustrating the effect of differential pressure as well as the current density on hydrogen crossover.
Trends in H2 permeation rates at 0–30 barg cathode backpressure (Fig. 4(c)) point towards two distinct driving forces for H2 crossover through PFSA membranes in PEMWEs operating at high differential pressures. The hydrogen permeation under differential pressure without any operating current, measured as the mass transport-limited HOR current, is directly proportional to the cathode backpressure and represents diffusion or a ‘pressure-driven permeation’ through the PFSA membrane. The H2 flux at the anode i.e., hydrogen permeation rate excluding the HOR, increases significantly with operating current density, showing a weak dependence on differential pressure. This increase in H2 permeation rate with current density suggests H2 supersaturation in ionomer or water phases, representing a ‘concentration-driven permeation’ through PEMs.5,19,20,34 This analysis further emphasizes the importance of hydrogen oxidation current in understanding the hydrogen crossover phenomenon in PEMWE systems operating under high differential pressure. These findings also underscore the significance of detailed electroanalytical studies at the single-cell level to gain insights into complex phenomena occurring in large scale systems. Combining this methodology with H2 and O2 mole balance experiments could further advance our understanding of H2 crossover in these systems and aid in maximizing the efficiency of electrolyzer systems.35,36
Footnote |
† Electronic supplementary information (ESI) available: Detailed descriptions of experimental methods, supporting results and discussion, and supporting figures (Fig. S1–S9). See DOI: https://doi.org/10.1039/d5ee00048c |
This journal is © The Royal Society of Chemistry 2025 |