Main

Proton exchange membrane water electrolysis (PEMWE) is an ideal technology for sustainable hydrogen production using renewable energy1,2. However, the high cost of the catalyst hampers its large-scale application, necessitating the development of anode catalysts with reduced precious iridium (Ir) loading. However, at present, the corrosion of low-content Ir in PEMWE urgently needs to be solved, but the fundamental nature of the evolution of corrosion remains under intensive debate3,4. Previous studies have shown that the formation of highly active Ir6+ species is a major cause of corrosion owing to its dissolution3. Therefore, the developed protection mechanisms mainly focus on steady-state material engineering strategies, such as accelerating charge transfer, applying armour protection and passivation layers, and so on5,6,7. These works have elucidated the reaction mechanisms and proposed catalyst design from an energetic perspective.

However, in the electrocatalysis of oxygen evolution reactions (OER), the transient mechanisms involving reaction intermediates and protons/electrons evolving over time have not yet been explored8,9. The anodic transformations triggered by start–stop events are typically transient and elusive, unlike steady-state operation, posing the challenges for direct observation10,11,12. In contrast to the well-studied evolution of stable intermediates such as *OH, *O and *OOH in the OER, the spatiotemporal evolution of critical Ir6⁺ species in transient proton-coupled electron transfer (PCET) steps remains poorly understood13,14. In particular, the microscopic kinetics of proton and electron evolution in PCET has evaded clear probing and precise control. The absence of microscopic insights regarding the initial state makes the prediction of the subsequent evolution of the macroscopic material structure challenging, thereby hindering the establishment of effective anti-corrosive design principles15,16.

Currently, the identification of steady-state catalytic active sites mainly relies on spectroscopic techniques with temporal resolution on the millisecond (10−3 s) to microsecond (10−6 s) level, such as electron energy loss spectroscopy, Fourier transform infrared spectroscopy and X-ray absorption spectroscopy17,18,19,20,21,22,23,24,25. However, the elementary steps of chemical bond breaking/formation in PCET are essentially ultrafast processes at the picosecond (10−12 s) to femtosecond (10−15 s) level. Owing to this timescale difference of six or more orders of magnitude between the elementary reactions and the resolution of current spectroscopic techniques, existing analytical methods face challenges in capturing these ultrafast processes to understand the catalytic mechanism26,27,28,29,30. The lack of cross-scale correlations between macroscopic steady-state characterization and microscopic transient dynamics prevents existing theories from resolving the spatiotemporal evolution sequence of active sites and reaction intermediates. This gap in the temporal dimension severely restricts the manipulation of catalytic reactions owing to ultrafast time-resolved perspectives.

In this Article, we used femtosecond electrochemical transient absorption spectroscopy (fs-ECTAS) to identify and decouple the temporal sequence steps in multistep PCET reactions on a typical IrO2 anode. By bridging isotope labelled macroscopic proton transport kinetics with femtosecond-resolved microscopic dynamics, we determined that the corrosion of commercial IrO2 originates from the synchronous presence of protons and electrons before 100 fs. On the basis of this, a Lewis acid, CeO2, which acts as a proton transfer channel, was introduced to regulate the temporal asynchrony between proton and electron transfer. Such modulation enabled electron transfer within the first 100 fs, while proton transfer from water oxidation occurred subsequently, within the 100–300 fs timescale. This spatiotemporal asynchrony enhanced the kinetics of proton transfer and stabilized transient Ir6+ species for improved anode corrosion resistance. This work represents an investigation of electrocatalytic processes and mechanisms at the femtosecond timescale, offering a new perspective to overcome theoretical and mechanistic bottlenecks in electrocatalysis.

Origin of commercial IrO2 corrosion

We report the development of fs-ECTAS that merged femtosecond pump–probe spectroscopy with an in situ electrochemical platform (Supplementary Fig. 1). During measurement, the pump beam acted as a photonic internal standard to precisely define the reaction time zero. Under the influence of external voltage, the selective excitation by pump photons enabled the labelling of distinct electronic states within the catalyst. These electronic states were influenced by interfacial charge transfer processes driven by the external field, giving rise to transient optical features associated with chemical bond reorganization on the femtosecond to picosecond timescale. We then captured the interfacial proton–electron evolution dynamics on a commercial IrO2 catalyst under working potentials (Fig. 1a–f and Supplementary Figs. 2–4). As shown in Fig. 1a, the excited-state absorption of IrO2 (790 nm) can all be observed under the steady-state condition of open circuit potential (OCP). With increasing voltage, two absorption bands attributed to Ir6+ species and H+ in the hydrogen bond network emerged at 830 nm and 963 nm, respectively (Fig. 1b–f). The Ir6+ species was identified on the basis of the correlation between the oxidation process and the potential-dependent absorption evolution observed in the ECTAS spectra, which was further confirmed by density functional theory (DFT) calculations to simulate their excited-state energy (Supplementary Fig. 5). The calculated excitation energy differences correspond to the characteristic absorption wavelengths observed in the transient absorption spectra. The results showed that the excited-state absorption peaks of IrO2, Ir6+ and H+ were located at 782.6 nm, 838.8 nm and 971.2 nm, respectively. The formation of Ir6+ at a high potential (1.9 V) and the adsorption of surface protons occurred simultaneously within 100 fs, suggesting that this was an electron–proton coupling process at the sub-picosecond timescale.

Fig. 1: Corrosion issues of commercial IrO2.
Fig. 1: Corrosion issues of commercial IrO2.
Full size image

a–f, fs-ECTAS of IrO2 under OCP (a) and 1.1 V (b), 1.3 V (c), 1.5 V (d), 1.7 V (e) and 1.9 V (f) potentials in 0.5 M H2SO4. Arrows indicate the characteristic signal peaks assigned to IrO2, Ir6+ and H+species, respectively. Milli-optical density (mOD) denotes the magnitude of the differential absorbance intensity.

Source data

We conducted DFT to reveal the influence of protons and electrons on the dissolution behaviour of Ir (Supplementary Figs. 6–8). As shown in Supplementary Fig. 9, the calculated cationic vacancy formation energy was 7.75 eV when only electron accumulation occurred at the Ir site, and 5.37 eV when only proton adsorption was considered. In contrast, the formation energy decreased to 5.26 eV when both proton and electron effects were present. These results indicate that the simultaneous presence of protons and electrons lowered the cationic vacancy formation energy, thereby promoting the dissolution corrosion of Ir. Proton transfer typically lags behind electron transfer, resulting in proton accumulation. This undesirable proton–electron synchronization is the key factor for Ir corrosion. Therefore, it was essential to accelerate proton transport and prevent its accumulation to mitigate Ir corrosion during OER.

Promotion of proton transport by CeO2–IrO2

On the basis of the results of the fs-ECTAS characterization and DFT computation, we synthesized a series of Lewis acidic metal oxide and IrO2 composite materials (MO–IrO2, where M is Ce, Ta, Nb, V, W, Mo or Cr) to accelerate proton transfer. As shown in Fig. 2a and Supplementary Figs. 10 and 11, the relatively weak Lewis acidity and low lattice oxygen solubility of Ce provided a soft proton conduction channel, facilitating rapid proton migration. Performance evaluation revealed that CeO2–IrO2 with 10% CeO2 exhibited the best catalytic performance (Fig. 2b,c and Supplementary Fig. 12). The physicochemical characterization of CeO2–IrO2 confirmed that CeO2 was supported on IrO2 (Fig. 2d,e and Supplementary Figs. 13–18). This revealed that the enhanced catalytic performance of the system was probably because of the rapid proton transfer across the CeO2–IrO2 interface without accumulation, rather than the conventional modulation of the electronic structure at metal active sites (Fig. 2f and Supplementary Figs. 19 and 20).

Fig. 2: Electrochemical properties of CeO2–IrO2.
Fig. 2: Electrochemical properties of CeO2–IrO2.
Full size image

a,b, The lattice oxygen solubility (a) and LSV (b) of MO–IrO2 (where M is Ce, Ta, Nb, V, W, Mo or Cr). c, A performance and Ir dissolution comparison of CeO2–IrO2 with different Ce ratios. J denotes the current density, E represents the applied potential and C indicates the concentration. d,e, A high angle annular dark field scanning transmission electron microscope image (d) and X-ray absorption spectroscopy plot (e) of CeO2–IrO2. f, The surface pH (bottom) and J (top) of CeO2–IrO2 and IrO2 detected by rotating ring disk electrode.

Source data

As a result, the CeO2–IrO2 catalyst exhibited no noticeable degradation under the accelerated corrosion tests with frequent start–stop cycling, while commercial IrO2 underwent severe degradation after approximately 1,500 cycles, leading to a rapid increase in cell voltage (Supplementary Fig. 21). This enhanced anti-corrosive ability of CeO2–IrO2 was further confirmed by the inductively coupled plasma measurements of Ir dissolution under constant current electrolysis (Supplementary Fig. 22). These findings indicate that CeO2–IrO2 underwent much less Ir dissolution, indicating that CeO2 effectively suppresses Ir corrosion and leaching during operation, which was further supported by more in situ and ex situ structural characterizations (Supplementary Figs. 23–25).

The in situ electrochemical ultraviolet–visible (UV–vis) absorption technique was applied to detect Ir oxidation states under various potentials31. The differential features observed in the UV–vis absorption spectra primarily reflected surface-related absorption changes. On the basis of these signals, the changes in catalyst valence states were precisely monitored32,33. In addition, it showed a linear relationship between the absorption intensity in UV–vis spectra and the number of oxidation state species. Thus, the charge-dependent bleaching behaviour of the spectra enabled accurate quantification of spectral changes induced by transient oxidation states15,34,35. As shown in Fig. 3a and Supplementary Fig. 26, potential-dependent UV–vis absorption spectroscopy revealed that the CeO2–IrO2 system exhibited stronger absorption characteristics of oxidation intermediates than the IrO2 system. Time-resolved UV–vis spectroscopy tracked the absorption changes of specific oxidation states under pulsed potentials. When the potential triggered an oxidation state change, the electronic structure shifted and altered the transition behaviour (Fig. 3b and Supplementary Fig. 27). As a result, the absorption increased when the potential rose, and returned to its initial value when the potential dropped (Fig. 3c). The adsorption signals were correlated with the extracted charge to calculate the molar extinction coefficients of the active species (Fig. 3d). The results showed that the Ir6+ active charge density of CeO2–IrO2 was an order of magnitude higher than that of the original IrO2 under the stepped potential of 1.6 V versus reversible hydrogen electrode (RHE) (Fig. 3e). This increased accumulation of active surface charges indicates that CeO2 plays a critical role in mitigating surface charge dissipation, which was otherwise induced by the dissolution of Ir6+ species. A charge transfer rate model was further constructed, enabling the extraction of time-dependent surface charge accumulation upon pulsed potential excitation (Fig. 3f). We then determined the electron transfer rate by differentiating Ir charge accumulation with respect to time. The charge transfer rate of CeO2–IrO2 (4.3 mC cm−2 s−1) was 28 times higher than that of commercial IrO2 (0.15 mC cm−2 s−1), demonstrating that Ce incorporation during catalysis stabilized the active Ir6+ species, thereby mitigating non-catalytic electron dissipation and suppressing corrosion-related side reactions.

Fig. 3: Evidence of electron transfer.
Fig. 3: Evidence of electron transfer.
Full size image

a, In situ UV–vis absorption spectra of CeO2–IrO2 with Ir species in three sequential redox transitions15. b–d, The pulse–current (b), pulse–absorbance (c) and charge–absorption (d) relationships in time-resolved UV–vis absorption spectra of CeO2–IrO2. e, The surface charge density as a function of potential for CeO2–IrO2 and IrO2. f, The evolution of the surface charge density with the pulse duration for CeO2–IrO2 and IrO2.

Source data

Kinetics of proton transfer

To further elucidate the relationship between the oxidation state transition of Ir6+ and its proton transfer behaviour, millisecond-level (10−3 s) time-resolved electrochemical quartz crystal microbalance measurements under pulsed potential control were performed. We further correlated the obtained charge with the differential mass change. As shown in Supplementary Fig. 28, integration of the current–time response at the end of the pulse revealed that the charge-to-mass (dM/Q) response coefficient of CeO2–IrO2 was higher than that of IrO2. This indicated that CeO2–IrO2 induced a more pronounced mass change associated with proton transfer for the same charge transfer. We then employed deuterium isotope tracing to investigate how CeO2 accelerates proton transfer, which was enabled by a customized membrane electrode assembly (MEA) cell integrated with differential electrochemical mass spectrometry. As shown in Supplementary Fig. 29, potential-dependent measurements revealed a marked increase in the D2/H2 ratio for the CeO2–IrO2 catalyst, reaching 2.05% at 1.8 V, which was 5.4 times higher than that of IrO2 (0.38%) under identical conditions. This suggests that CeO2 substantially lowered the activation energy barrier for proton transfer, thereby facilitating the efficient transport of heavier D+ ions across the membrane–electrode interface.

Time-decoupled proton kinetics

We revealed time-decoupled kinetic behaviour and its connection to the kinetics of proton evolution in the CeO2–IrO2 composite catalyst by fs-ECTAS (Fig. 4a–f). On the basis of the femtosecond timescale kinetic profile of proton evolution, we successfully established a quantitative correlation model between microscopic and macroscopic proton transfer kinetics. Compared with IrO2, CeO2–IrO2 exhibited a pronounced shift of the H+ kinetic peak towards a longer timescale during polarization (Fig. 4g,h). This indicated that the introduction of CeO2 effectively promoted the kinetics of proton transfer, thereby preventing the accumulation of protons. By extracting the differential absorption signals at 150 fs, we constructed a voltage-dependent proton transfer kinetic profile (Fig. 4i). The data revealed that the CeO2–IrO2 system displayed a higher sensitivity of differential absorption to potential variation compared with IrO2, confirming its superior proton transfer dynamics.

Fig. 4: Proton transfer kinetics.
Fig. 4: Proton transfer kinetics.
Full size image

a–f, fs-ECTAS of CeO2–IrO2 under OCP (a) and 1.1 V (b), 1.3 V (c), 1.5 V (d), 1.7 V (e) and 1.9 V (f) potentials in 0.5 M H2SO4. Arrows indicate the characteristic signal peaks assigned to IrO2, Ir6+ and H+ species, respectively. g,h, Transient absorption spectra of CeO2–IrO2 (g) and IrO2 (h) with proton dynamics curves. i, The microscopic proton dynamics of CeO2–IrO2 and IrO2.

Source data

On the basis of the above analysis, leveraging fs-ECTAS to decouple the kinetic features of key multistep anodic processes (proton and electron transfer) was crucial for elucidating the corrosion resistance mechanism of CeO2–IrO2 and guiding its rational optimization. As revealed by the time–wavelength–absorbance fs-ECTAS maps in Fig. 5a,b, the CeO2–IrO2 system exhibited distinctly different kinetics compared with IrO2. The CeO2–IrO2 system exhibited distinct charge transfer kinetics under OER conditions at 1.9 V versus RHE. Time-resolved absorption spectroscopy revealed that the Ir6+ absorption feature vanished within 100 fs, whereas the corresponding proton signal emerged at 100–300 fs. Kinetic analysis of the correlated absorbance signals within 500 fs further revealed that the incorporation of CeO2 established a spatiotemporal asynchronous mechanism. This temporal decoupling effectively disrupted the synchronization between Ir6+ and protons, thereby stabilizing Ir6+. This was attributed to the incorporation of CeO2, which introduced surface hydroxyl groups with strong oxophilicity, creating a low-energy pathway for proton transfer in the post electron transfer regime (100–300 fs). This temporal decoupling of proton–electron transfer markedly suppresses interfacial charge recombination and provides experimental insight into the dynamic regulation of PCET in anodic OER.

Fig. 5: Ultrafast reaction mechanism of CeO2–IrO2.
Fig. 5: Ultrafast reaction mechanism of CeO2–IrO2.
Full size image

a, An fs-ECTAS of CeO2–IrO2 at 1.9 V versus RHE. b, An fs-ECTAS of IrO2 at 1.9 V versus RHE. Arrows indicate the characteristic signal peaks assigned to IrO2, Ir6+ and H+ species, respectively. c, A schematic illustration of the time-resolved catalytic mechanism for CeO2–IrO2 and IrO2. IrVo denotes the lattice vacancy formed after Ir dissolution.

Source data

Time-resolved precise spectral analyses uncover a fundamental difference in proton transfer kinetics. In the CeO2–IrO2 system, proton-related signals emerged within 100 fs and decayed within 300 fs, indicating accelerated kinetics. This stands in sharp contrast to the longer proton characteristic duration (>500 fs) observed in the IrO2 system. This kinetic enhancement correlated directly with a distinct proton generation pathway. fs-ECTAS captured within 150 fs revealed the synchronous formation of Ir6+ intermediates from CeO2–IrO2, with their concurrent decay strongly correlated with the rise of the H+ signal after 100 fs. These findings indicate that protons in the CeO2–IrO2 system originated from the in situ oxidation of initial adsorption water (<50 fs) during OER, rather than from the residual protons accumulated in the previous OER cycle. As shown in Fig. 5a, the absorption peaks shifted markedly with time except for the H+ signal. This behaviour was closely associated with electron transfer at the Ir active centres and the evolution of surface-adsorbed species during the reaction. Voltage-dependent measurements further validated the regulatory role of CeO2. CeO2–IrO2 exhibited distinct Ir6+ spectral characteristics, indicating that CeO2 was conducive to the oxidation of water to produce protons (Supplementary Figs. 30 and 31). Temporal decoupling between H+ formation and Ir oxidation became evident after 1.7 V versus RHE, highlighting a voltage-driven mechanism in which CeO2 lowers the activation barrier for proton transfer. This enabled rapid proton conduction along the hydroxyl rich CeO2 surface, facilitating efficient charge separation during water oxidation.

Therefore, we propose a spatiotemporally decoupled anti-corrosion mechanism model as illustrated in Fig. 5c. During the initial adsorption stage (<50 fs), the CeO2 component preferentially adsorbs water molecules owing to its strong oxophilicity. This facilitates ultrafast electron transfer mediated by oxygen vacancies, driving the oxidation of Ir centres to the +6 oxidation state. In the subsequent charge transfer stage (50–150 fs), water dissociation was triggered to generate *OH intermediates and protons. Concurrently, electron transfer between the catalyst and intermediate led to the reduction of Ir6+ to Ir4+, as evidenced by the rapid decay of the Ir6+ signal within 100 fs. In the subsequent reaction stage (100–300 fs), the absence of coexisting Ir6+ species prevented the formation of cationic vacancies, although protons generated via water oxidation persist on the catalyst surface. This temporal mismatch between proton and electron dynamics ensured a consistently high formation energy for Ir vacancies, thereby suppressing Ir dissolution. In the final desorption stage (>300 fs), protons rapidly depart from the surface without accumulating, and the high vacancy formation energy was maintained. This spatiotemporal decoupling effect minimized the overlap of proton and electron populations in both time and space, effectively mitigating anodic dissolution. In contrast, commercial IrO2 exhibited residual protons in the early reaction stage (<150 fs) that were not efficiently removed during the previous oxidative cycle. The temporal overlap of these protons with electrons reduced the cationic vacancy formation energy, thereby accelerating Ir dissolution.

Performance testing of PEMWE under fluctuating power

The microscopic temporal decoupling of proton and electron dynamics effectively suppresses both the reduction in cationic vacancy formation energy and Ir dissolution, leading to enhanced stability of CeO2–IrO2 in PEMWE. The CeO2–IrO2 membrane electrode delivered outstanding performance even at reduced iridium loading (0.5 mg cm−2) (Fig. 6a). At a high current density of 3 A cm−2, the cell required only 1.93 V, substantially lower than the voltage needed by commercial IrO2 under the same conditions. The electrochemical impedance spectroscopy distribution of relaxation times analysis revealed a pronounced decrease in impedance at high frequencies, suggesting a substantial enhancement in proton-transfer kinetics (Fig. 6b). The incorporation of CeO2 in the membrane electrode under operating conditions effectively suppressed the red shift of OH vibrational modes, indicating enhanced proton transfer and mitigation of local acidification due to proton accumulation (Supplementary Figs. 32 and 33). Long-term durability testing demonstrated the catalyst’s exceptional operational stability, maintaining continuous electrolysis at 1 A cm−2 for about 1,400 h without performance degradation (Fig. 6c).

Fig. 6: Performance under fluctuating conditions.
Fig. 6: Performance under fluctuating conditions.
Full size image

a, The LSVs of CeO2–IrO2 and IrO2 in PEMWE. b, The electrochemical impedance spectroscopy distribution of relaxation times of CeO2–IrO2 and IrO2. c, A chronopotentiometric stability test of CeO2–IrO2 at a current density of 1 A cm−2. d, A techno-economic analysis of CeO2–IrO2 in proton exchange membrane electrolyser applications. e, An evaluation of MEA performance under simulated PV fluctuation conditions. f, The proton exchange membrane electrolysis stability under fluctuating power input simulating real-world PV operation.

Source data

Techno-economic analysis indicated that after coupling with low-cost renewable electricity, CeO2–IrO2 offered the advantages of high current density and low-Ir loading, enabling a hydrogen production cost less than US$2.4 kg−1 (Fig. 6d and Supplementary Figs. 34 and 35). With its rapid proton transport kinetics, the system effectively maintains dynamic stability under the fluctuating power conditions typical of renewable energy sources. Electrochemical start–stop tests showed that the CeO2-enhanced catalyst maintained a rapid voltage response on the timescale of seconds (Supplementary Fig. 36). In accelerated ageing experiments, the catalyst remained operational for 37,385 start–stop cycles at 500 mA cm−2. Simulated photovoltaic (PV)-driven operating conditions further validated the catalyst’s stability (Fig. 6e). Under a regime of daily stepped power loading (20% to 100% to 20% state of charge) alternating with nighttime shutdown, the CeO2–IrO2 system maintained stable performance over 14 days of continuous cycling, confirming its practical suitability for renewable-integrated hydrogen production (Fig. 6f).

Conclusion

We introduced Lewis acidic CeO2 onto IrO2 to regulate the sequence and kinetics of proton–electron transfer on the femtosecond timescale, enabling stable catalysis with reduced Ir loading. Experimental results revealed that CeO2 accelerated proton transfer and decoupled proton and electron transfer into a temporally asynchronous pathway (electrons <100 fs and protons >100 fs) to effectively suppress Ir oxidation and corrosion. The CeO2–IrO2 catalyst achieved an industrially relevant performance of 3 A cm−2 at 1.93 V with a low Ir loading of 0.5 mg cm−2, maintaining stability for about 1,400 h in PEMWE. This work offers a time-resolved design paradigm for developing low Ir PEMWE catalysts that meet industrial demands. Furthermore, a femtosecond timescale mechanistic perspective is provided for investigating complex multiphase electrochemical systems, including CO2 reduction and liquid phase batteries.

Methods

fs-ECTAS measurements

fs-ECTAS measurements were conducted using a custom designed in situ electrochemical optical platform. The working electrode was prepared by uniformly spray coating the target sample onto a fluorine-doped tin oxide conductive glass substrate. The electrolyte was continuously circulated through the electrochemical cell using a peristaltic pump to ensure stable operation. Counter and reference electrodes were symmetrically positioned on the non-optical regions flanking the working electrode, thereby minimizing interference with optical measurements. A femtosecond laser source was split into pump and probe beams using a beam splitter. The pump beam was directed through a tunable optical delay line and focused onto the surface of the working electrode to excite the sample. Simultaneously, the probe beam monitored transient changes in optical absorption in real time. A high-precision potentiostat applied a constant bias voltage to the system, and the transient absorption signals were recorded as a function of pump–probe delay. This setup enabled the extraction of excited-state kinetic information under steady-state electrochemical conditions.

In situ UV–vis measurements

The in situ UV–vis absorption spectroscopy was conducted using a custom transmission type absorption cell, which was compatible with the Shimadzu UV-2600 UV–vis absorption spectrometer in use. The sample was sprayed onto the fluorine-doped tin oxide as the working electrode and placed on the optical path. The working electrode was made of Pt wire, and the reference electrode was Ag/AgCl. During the scanning test of the linear scanning curve (LSV), the ultraviolet spectra were simultaneously collected in real time. The time-resolved test adopted the kinetic test method, and the real-time measurement was carried out at the position of maximum absorption. The electrochemical signal was collected by multiple acquisitions of 10 s pulse and 10 s stop.

PEMWE measurement

MEAs for PEMWE were prepared using Nafion 115 membranes and the catalyst prepared in our laboratory. Catalyst-coated membranes were fabricated by sequential spray coating five uniform layers with each containing 0.1 mgIr cm−2 catalyst loading. The mass of each sprayed layer was adjusted according to the Ir proportion. It resulted in a total Ir loading of 0.5 mg cm−2 on the anode. The cathode employed a commercial Pt/C catalyst (0.1 mg cm−2). Bare carbon paper and Pt-coated titanium mesh were used as the gas diffusion layers on the cathode and anode sides, respectively. The assembled MEAs were tested in a custom proton exchange membrane electrolyser using a continuous flow of deionized water without added electrolyte.