Bridging the Gap Between Laboratory Catalyst Research and Practical Water Electrolyzer Commercialization

Abstract

Water electrolyzers are key technologies for enabling a carbon-neutral society by facilitating sustainable hydrogen production. Despite extensive efforts to develop highly active catalysts, a persistent gap remains between catalyst performance evaluated in rotating disk electrode (RDE) systems and that achieved in practical membrane electrode assembly (MEA) configurations. In this perspective, we highlight the origins of this discrepancy and emphasize that intrinsic catalytic activity alone is insufficient to predict electrolyzer performance. Using the acidic hydrogen evolution reaction (HER) as a model system, we demonstrate that apparent activity metrics such as mass activity and Tafel slope can be significantly influenced by catalyst loading, composition, and measurement conditions in RDE, often leading to misleading conclusions. In contrast, MEA performance is governed by complex interfacial phenomena including mass, electron, and charge transport. A similar behavior was also observed for alkaline HER and acidic oxygen evolution reaction (OER). We further present case studies illustrating the critical roles of electron transfer and durability in water electrolyzers. In addition, recent advances in proton-exchange-membrane water electrolyzer commercialization are discussed, highlighting the importance of system-level optimization, including ionomer distribution control and scalable manufacturing via slot-die coating, which enabled the deployment of a 2.5 MW system. Overall, bridging the gap between laboratory catalyst research and industrial water electrolyzer applications requires a shift from catalyst-centric design toward integrated system-level engineering.

Introduction

As climate change has received increasing attention, the chemical industry has been pushed to transition from solely fossil fuel-based technologies toward carbon-neutral technologies. Among the key technologies enabling a carbon-neutral society are water electrolyzers.

Water electrolyzers can enable the storage and transportation of intermittently produced electricity. While batteries generally provide only short-term energy storage within a day, water electrolyzers can enable long-term energy storage by converting surplus electricity into H₂ through water splitting. The produced H₂ can be used as a fuel (e.g., H₂ fuel cell vehicles), as a chemical feedstock in chemical refineries, or for electricity generation in more populated areas through fuel-cell power plants.

Practically used electrochemical reactors, such as fuel cells and water electrolyzers, mostly share the same membrane-electrode assembly (MEA) structure. In this configuration, the anode and cathode catalyst layers are located on both sides of a polymer electrolyte membrane. These MEAs can be stacked to enable large-scale production. The specific electrode reactions and the chemical species transported through the membrane vary depending on the type of electrolyzer. A single cell consisting of one MEA can easily show a current density of >1 A cm^–2.

Catalysts used in these electrodes are often initially screened in a half-cell rotating disk electrode (RDE) setup consisting of working, counter, and reference electrodes. Microgram-level amounts of powder catalysts are deposited on the tip of the RDE, which is designed to minimize mass-transfer limitations. The current density is typically measured in the range of ∼10 mA cm^–2. The electrochemically active surface area is also often measured in this setup to estimate the intrinsic activity of the catalyst. H-cell configurations with three electrodes are also commonly used to evaluate catalytic activity. In this case, a larger amount of catalyst can be deposited on electrodes with larger surface areas, such as carbon paper; however, these systems often suffer from mass-transfer limitations. While RDE and H-cell configurations are useful for initial catalyst screening, the overall energy efficiency of these three-electrode setups is typically very poor, largely due to high charge-transfer resistance.

It has increasingly been recognized that high intrinsic activity observed in an RDE setup does not necessarily guarantee high performance in an MEA configuration. , Newly developed catalysts often exhibit mass activities that are an order of magnitude higher than those of commercially available catalysts in RDE measurements, but such improvements are rarely translated into comparable increases in single-cell performance.

In this perspective, we discuss the performance discrepancy between RDE and MEA systems and present examples demonstrating that factors beyond intrinsic activity, such as electron transfer and catalyst phase transformation, play key roles in determining MEA performance. We also introduce a recent case study on the large-scale deployment of proton-exchange-membrane water electrolyzers (PEMWEs).

Gap Between RDE and MEA in Water Electrolyzer

PEMWEs and anion exchange membrane water electrolyzers (AEMWEs) are typically used for electrochemical H₂ production. The electrode reactions and the most widely used catalysts are listed below.

PEMWE

$$(\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{d}\mathrm{e}){2\mathrm{H}}^{+}+{2\mathrm{e}}^{-}\to {\mathrm{H}}_2\mathrm{P}\mathrm{t}/\mathrm{C}$$

$$(\mathrm{a}\mathrm{n}\mathrm{o}\mathrm{d}\mathrm{e}){2\mathrm{H}}_2\mathrm{O}\to {\mathrm{O}}_2+{4\mathrm{H}}^{+}+{4\mathrm{e}}^{-}{\mathrm{I}\mathrm{r}\mathrm{O}}_2$$

AEMWE

$$(\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{d}\mathrm{e}){2\mathrm{H}}_2\mathrm{O}+{2\mathrm{e}}^{-}\to {\mathrm{H}}_2+{2\mathrm{O}\mathrm{H}}^{-}\mathrm{P}\mathrm{t}/\mathrm{C},\mathrm{N}\mathrm{i}\mathrm{M}\mathrm{o}$$

$$(\mathrm{a}\mathrm{n}\mathrm{o}\mathrm{d}\mathrm{e}){4\mathrm{O}\mathrm{H}}^{-}\to {\mathrm{O}}_2+{2\mathrm{H}}_2\mathrm{O}+{4\mathrm{e}}^{-}\mathrm{N}\mathrm{i}\mathrm{F}\mathrm{e}$$

Many studies have focused on minimizing the use of precious metals or replacing them with nonprecious metals. − The high intrinsic activity of these newly developed catalysts is typically confirmed using an RDE setup. However, catalysts that show high activity in RDE measurements often fail to exhibit corresponding activity in MEAs. , This discrepancy between RDE and MEA performance has been a persistent challenge across different types of electrolyzers.

Figure demonstrates the different reaction environments of the RDE and MEA for the acidic hydrogen evolution reaction (HER). In an RDE setup, protons dissolved in the aqueous solution are reduced to form gaseous H₂, generating bubbles (Figure a). In an MEA, however, protons must be transported through the ionomer from the anode, and the catalyst layer must provide sufficient electronic conductivity (Figure b). When protons and electrons meet at the catalyst surface, H₂ is formed, and the resulting gas bubbles must be efficiently removed to allow for further reaction. Additionally, the MEA represents only one part of the overall electrolyzer system (Figure c). Additional interfaces exist between the MEA and the porous transport layer (PTL) or between the bipolar plate and the current collector, and these extra contact points may introduce further resistance or interfacial issues.

1.

Schemes of (a) RDE, (b) MEA, and (c) a full reactor setup of a single cell water electrolyzer. Acidic HER was exemplified.

Estimating intrinsic activity accurately is not straightforward, even for the seemingly simple acidic HER. Chorkendorff et al. showed that an ultralow Pt loading of 13 ng cm^–2 is sufficient to catalyze HER, yet even such a low Pt loading still suffers from mass transport limitations in acidic HER. El-Sayed et al. further reported that it is impossible to obtain reliable HER and hydrogen oxidation reaction (HOR) kinetic data from RDE measurements in acidic media. − Although RDE is designed to dissipate chemical species produced from surface reactions, H₂ is generated faster than it can be removed, resulting in supersaturation of H₂ near the electrode surface. While HER is the reaction intended to be observed, HOR inevitably occurs simultaneously, requiring a more negative potential for H₂ production. It is known that acidic HER on a Pt surface exhibits a Tafel slope of 30 mV dec^–1. Although experimentally observed Tafel slopes are frequently used to assign specific rate-determining steps, the Tafel slope is coverage-dependent and electron-transfer kinetics are often overlooked.

Figures and S1–S3 show RDE results for acidic HER using 1, 5, 10, and 20 wt % Pt/C catalysts. Upon Ar purging, HOR is suppressed, and the HER current can be estimated more accurately. However, the equilibrium potential of the H₂/H⁺ couple shifts, as shown in Figure a, resulting in an origin mismatch. The Tafel slope varies significantly from 20.0 to 44.5 mV dec^–1 depending on total catalyst loading for 1 wt % Pt/C, and this value further varies with different Pt contents. Some values were below the theoretical value of 30 mV dec^–1, which is consistent with the previous report showing that locally trapped hydrogen on the electrode surface results in an artificially smaller Tafel slope. This suggests that an accurate Tafel slope cannot be obtained even under Ar-purged conditions.

2.

Ambiguity of measuring mass activity for acidic HER. Linear sweep voltammetry (LSV) curves measured in 0.1 M HClO₄ solution using a 1 wt % Pt/C catalyst under (a) Ar purging or (b) H₂ purging in an RDE setup. The Tafel slope is estimated by linear fitting of the overpotential vs logarithm of the HER kinetic current in the insets of (a,b). Mass activity estimated at the potential of −15 mV_RHE for 1, 5, 10, and 20 wt % Pt/C catalysts obtained from (c) Ar purging and (d) H₂ purging. Total catalyst loading was varied in 10–100 μg cm^–2. (e) PEMWE single cell results when the total cathode catalyst amount was fixed at 0.5 mg cm^–2 for 1, 5, 10, and 20 wt % Pt/C. (f) PEMWE single cell results when the total catalyst amount was varied for 1 wt % Pt/C.

Upon H₂ purging, the equilibrium potential can be measured more accurately, as evidenced by the correct origin point (Figure b). However, if the Tafel slope is estimated using the same method as under Ar purging, assuming suppression of the reverse reaction, the resulting slope becomes much lower, in the range of 14.3–28.8 mV dec^–1. Under H₂ purging, the Tafel slope becomes even more inaccurate than under Ar purging due to the readily occurring HOR. Instead, Tafel plots incorporating both HER and HOR should be used to estimate the exchange current density and transfer coefficient (Figure S4). Consequently, direct comparison of Tafel slopes can be highly misleading.

When the mass activity was estimated for various total catalyst loadings, various Pt contents, and Ar/H₂ purging conditions, it varied significantly, as shown in Figure c,d. When the mass activity was evaluated at – 15 mV_RHE, Ar purging yielded higher mass activity than H₂ purging due to the suppressed reverse reaction. Although the Pt nanoparticles in these catalysts likely possess similar intrinsic activity, a lower Pt content (1 wt %) and a lower total catalyst loading (10 μg cm^–2) can readily increase the apparent mass activity by an order of magnitude. When ultralow-Pt catalysts are developed, their mass activity is often compared with that of a control Pt/C sample having different total catalyst loading and Pt content. The higher mass activity of a new ultralow-Pt catalyst may not reflect genuinely higher intrinsic activity, but rather simply reflect the lower Pt loading.

The 1, 5, 10, and 20 wt % Pt/C catalysts were used to fabricate MEAs with the same total cathode catalyst loading of 0.5 mg cm^–2 (Figure e). The corresponding Pt loadings were 5, 25, 50, and 100 μg_Pt cm^–2, respectively. When their PEMWE performance was compared at 1.6 V, the current densities were 0.47, 0.79, 0.87, and 1.06 A cm^–2, respectively. When the 1 wt % Pt/C and 20 wt % Pt/C samples are compared, the current density increases by only slightly more than 2-fold, whereas the Pt content increases by a factor of 20. However, this does not imply that 1 wt % Pt/C is a better catalyst than 20 wt % Pt/C, because increasing the catalyst loading of 1 wt % Pt/C would not lead to a proportional increase in current density due to mass transport limitations (Figure f).

A similar trend was observed for alkaline HER and the corresponding AEMWE performance. While the theoretically expected Tafel slope for alkaline HER is 120 mV dec^–1, the value varies significantly for different Pt contents and catalyst loadings (Figures a,b and S5). The mass activity also varies depending on Pt contents and catalyst loadings, while the extent of variation was much smaller than that for the acidic HER (Figure c,d). The single-cell performance was also compared with the total cathode catalyst loading at 1.0 mg cm^–2 (Figure e). The 1 wt % Pt/C presented poor performance, while 5, 10, and 20 wt % Pt/C catalysts were better with the current densities of 0.23, 0.29, and 0.32 A cm^–2 at 1.6 V, respectively. When the catalyst loading of the 5 wt % Pt/C increased to 2 and 3 mg cm^–2, the current density only marginally increased to 0.27 and 0.30 A cm^–2 at 1.6 V (Figure f).

3.

Measuring mass activity for alkaline HER. LSV curves measured in 1 M KOH solution using (a) 5 wt % Pt/C or (b) 20 wt % Pt/C catalyst in an RDE setup. The Tafel slope is estimated by linear fitting of the overpotential vs logarithm of the HER kinetic current in the insets of (a,b). Mass activity estimated at the potential of (c) −15 mV_RHE or (d) −50 mV_RHE for 1, 5, 10, and 20 wt % Pt/C catalysts. Total catalyst loading was varied in 10 ∼ 100 μg cm^–2. (e) AEMWE single cell results when the total cathode catalyst amount was fixed at 1.0 mg cm^–2 for 1, 5, 10, and 20 wt % Pt/C. (f) AEMWE single cell results when the total catalyst amount was varied for 5 wt % Pt/C.

Acidic OER catalysts were also evaluated in both RDE and MEA configurations. Commercial iridium oxide (Premion), comprising thick IrO _x shells with a metallic Ir core, was selected as a representative OER catalyst. Figure a,b shows linear sweep voltammetry (LSV) curves recorded at two different scan rates of 10 and 1 mV s^–1, whereas Figure c shows I–V curves obtained by chronoamperometry (CA) measurements. The corresponding Tafel slopes are shown in the insets, and the mass activities derived from each measurement are presented in Figure d–f. Notably, the Tafel slope and mass activity differ significantly for these three cases, indicating the effect of a large capacitive charge of Premion. , Compared with steady-state polarization analysis, LSV measurements produced higher apparent Tafel slopes and mass activities, and this discrepancy became more pronounced at higher scan rates. This result demonstrates that the commonly used LSV-based method can substantially overestimate the activity, particularly when capacitive currents are significant.

4.

Measuring mass activity for acidic OER using a Premion catalyst. LSV curves measured in 0.05 M H₂SO₄ solution in an RDE setup with the scan rate of (a) 10 mV s^–1 and (b) 1 mV s^–1. (c) I–V curve measured in 0.05 M H₂SO₄ solution in an RDE setup by chronoamperometry (CA). Tafel slope was estimated by linear fitting of the overpotential vs logarithm of the OER kinetic current in the insets of (a–c). Mass activity estimated at the potential of 1.55 V_RHE: (d) from LSV at 10 mV s^–1, (e) from LSV at 1 mV s^–1, and (f) from CA measurements. Total catalyst loading was varied from 50 ∼ 200 μg cm^–2.

Figure S6 shows the PEMWE cell performance as a function of catalyst loading at the anode. Increasing the catalyst loading improved the cell performance only up to a certain point, after which the cell performance was saturated. The rutile IrO₂ synthesized by the Adams fusion method presented a much lower Tafel slope and smaller mass activity, even when the LSV method was used with a high scan rate of 10 mV s^–1, indicating that it has much less capacitive contribution (Figure S7). Overall, these results show that steady-state analysis may be required in RDE measurements to minimize capacitive artifacts and obtain more reliable activity metrics. However, for practical electrolyzer evaluation, identifying the optimal catalyst loading and catalyst-layer structure at the MEA level may be more critical than further refining the half-cell activity measurement alone.

Considering that newly developed catalysts are ultimately intended for use in MEA configurations, their evaluation in MEA systems is essential rather than relying solely on extrapolation from RDE measurements. Although RDE measurements remain valuable for elucidating intrinsic catalyst properties and reaction mechanisms, they cannot fully reproduce the multicomponent environment of MEA. Because of the complex interfacial structures in MEAs, intrinsic catalytic activity may not be the primary factor governing practical electrolyzer performance; instead, other factors can become dominant. ,, In the following sections, we discuss cases in which electron transfer and durability play key roles in determining the performance of water electrolyzers. We then present a recent example demonstrating the commercialization of the PEMWE.

Case Study Demonstrating the Importance of Electron Transport

In PEMWEs, research has predominantly focused on the OER, particularly on minimizing or replacing expensive and scarce Ir-based catalysts. − Considerable efforts have been devoted to understanding the electronic structure of surface Ir atoms and neighboring oxygen species to elucidate OER mechanisms. In most cases, when these catalysts are implemented in MEAs, Pt-coated Ti PTLs are employed to facilitate electron transport to the current collector. However, the Pt loading on Ti PTLs is often excessively high (e.g., ∼ 5.8 mg_Pt cm^–2), and the adhesion between the Pt layer and the Ti PTL remains insufficient, causing the Pt overlayer to detach easily.

Recently, Lee et al. systematically investigated the effect of Ir loading for both crystalline and amorphous Ir-based catalysts. Crystalline IrO₂ with a rutile phase was synthesized via the Adams fusion method, while a commercial Premion catalyst was used as a representative amorphous catalyst. When a bare Ti PTL was employed, Adams IrO₂ exhibited comparable performance at loadings of 0.8 ∼ 0.4 mg_Ir cm^–2, whereas performance deteriorated at 0.2 mg_Ir cm^–2 (Figure a). In contrast, the Premion catalyst showed significantly poorer cell performance at 0.2 mg_Ir cm^–2 (Figure b).

5.

I–V curves of PEMWE at various Ir loadings when using (a) Adams IrO₂, (b) commercial Premion IrO _x , and (c) IrNiO _x platelet catalysts. Comparison of i–V curves at an Ir loading of 0.2 mg_Ir cm^–2 using two different PTLs for (d) Adams IrO₂, (e) Premion, and (f) IrNiO _x . Schemes showing the situation at low Ir loading for (g) Adams IrO₂, (h) Premion, and (i) IrNiO _x . In (g), isolated IrO₂ domains (gray parts) cannot participate in the reaction. In (h), amorphous IrO _x suffers from poor electron transfer due to the Schottky barrier. In (i), crystalline platelets form the active sites effectively even at low Ir loading. Reprinted with permission from ref . Copyright 2025 American Chemical Society.

When comparing i–V curves at 0.2 mg_Ir cm^–2 using bare versus Pt-coated Ti PTLs, Adams IrO₂ showed minimal performance difference, whereas Premion exhibited a dramatic improvement upon Pt coating (Figure d,e). This discrepancy arises from differences in morphology and interfacial electronic properties. Highly crystalline Adams IrO₂ fails to uniformly cover the membrane at low loadings, leading to isolated catalyst domains (gray particles in Figure g) that lack electronic connectivity with the PTL. In contrast, the amorphous Premion catalyst can form a continuous film even at low loading; however, a Schottky barrier forms between the amorphous IrO _x layer and the native TiO _x layer, hindering electron transport. Pt coating on the Ti PTL mitigates this barrier, significantly enhancing electron transfer and overall cell performance.

Lee et al. further synthesized IrNiO _x platelets by modifying the Adams fusion method. These catalysts exhibited nearly identical i–V performance across a range of Ir loadings (Figure c). Moreover, at 0.2 mg_Ir cm^–2, their performance was comparable regardless of whether a bare or Pt-coated Ti PTL was used (Figure f). Although the IrNiO _x platelets are highly crystalline, their platelet morphology enables uniform membrane coverage even at low loadings, thereby maintaining effective electronic pathways (Figure i).

This case demonstrates that catalyst evaluation for water electrolyzers should extend beyond intrinsic OER activity and include the ability of the catalyst to form an electronically connected catalyst layer in the MEA. In particular, low-Ir anodes should be assessed in terms of catalyst-layer continuity and electrical contact with the PTL, because these factors directly determine whether catalytic sites identified in RDE can be effectively utilized under practical cell operation. , To achieve high catalyst-layer continuity, morphologies favorable for interconnections, such as rods and platelets, might be more suitable than simple nanoparticles. Furthermore, for efficient operation under Ti-PTL conditions, catalysts with low work functions should be designed to minimize the Schottky barrier.

Similarly, Kreider et al. showed in AEMWE that the effect of catalyst loading is governed by the interplay among catalyst electronic conductivity, catalyst-layer uniformity and coverage, ionomer-mediated ion transport, and PTL pore structure, rather than by catalyst mass alone. Xu et al. further showed that the catalyst exhibiting the highest OER activity in half-cell measurements delivered the poorest AEMWE performance, because its low electronic conductivity and limited integration with ionomer-mediated transport pathways restricted catalyst utilization in the MEA. Therefore, future catalyst development should treat the catalyst not only as an active material but also as a structural component of the electrode, whose morphology and interfacial properties must be optimized.

Case Study Demonstrating the Importance of Durability under Intermittent Conditions

Water electrolyzers are considered a promising approach for storing surplus electricity generated from solar and wind power, both of which are inherently intermittent. While catalysts can exhibit high stability under steady-state operation, they often suffer from significant degradation under transient conditions. Therefore, the long-term stability of catalysts under intermittent operation must be carefully evaluated for the practical deployment of water electrolyzers in energy storage systems. Although this issue has only recently begun to receive attention, there is already evidence that exposure to open circuit voltage (OCV) induces various durability issues in water electrolyzers. −

AEMWEs have attracted considerable interest because they can utilize nonprecious metal catalysts, such as NiMo for the cathode and NiFe for the anode. Although NiMo catalysts have demonstrated high activity in RDE measurements, their application in MEAs has remained limited. Lee et al. showed that when NiMo catalysts are used in MEAs, Mo species readily dissolve, and the extent of Mo dissolution depends strongly on the activation procedure.

Furthermore, to evaluate durability under intermittent conditions, the NiMo catalyst was operated by applying 1 A cm^–2 for 30 min, followed by OCV exposure for 30 min in repeated cycles. The individual overpotentials were decoupled using a three-electrode MEA configuration, in which a reference electrode was connected via an extended membrane. This setup enabled separate estimation of thermodynamic, ohmic, anode kinetic, cathode kinetic, anode mass transport, and cathode mass transport overpotentials. At the beginning of life (BoL), the anode kinetic overpotential contributed the most to the total cell voltage (Figure a). However, after 48 h of intermittent operation, the cathode kinetic overpotential exhibited the largest increase (Figure b). Post-mortem analysis revealed the formation of platelet-like structures (Figure c), indicative of the formation of β-Ni­(OH)₂. This crystalline phase formed during OCV exposure and could not be reduced back to the metallic Ni phase, leading to a loss of HER activity.

6.

I–V curves of AEMWE at the (a) beginning of life (BoL) and (b) end of life (EoL) with the Ni₃Mo cathode catalyst. The intermittent protocol, consisting of repeated cycles of 30 min of chronopotentiometry at 1 A cm^–2 and 30 min of OCV, was applied for 48 h. The overpotentials were decoupled into thermodynamic, ohmic, kinetic (anode), kinetic (cathode), mass transport (anode), and mass transport (cathode) using a three-electrode MEA system. TEM images of (c) Ni₃Mo at EoL showing the formation of platelet Ni­(OH)₂ and (d) carbon shell-encapsulated Ni₃Mo. Reprinted with permission from ref . Copyright 2025 Wiley.

To improve durability, the NiMo catalyst was coated with a carbon layer (Figure d). This carbon layer was formed via the Boudouard reaction (2CO ↔ CO₂ + C) by flowing CO at 330 °C over the reduced NiMo catalyst. The same intermittent protocol was then applied for up to 8 days, and the MEA maintained stable operation. The NiMo@C catalyst demonstrates outstanding durability under OCV exposure.

Considering the inherently intermittent nature of renewable energy sources, the design of durable electrode catalysts must explicitly account for such operating conditions. The stability of the catalysts under intermittent operation cannot be evaluated solely from their initial activity or half-cell stability. Although RDE can provide useful information on transient catalyst dissolution and redox behavior, it cannot fully reproduce the coupled degradation processes that occur in an operating MEA. Under dynamic current profiles, performance losses can arise not only from catalyst dissolution or surface reconstruction, but also from changes in catalyst/ionomer integration, membrane contamination, and PTL degradation. − Therefore, catalysts for renewable energy-driven water electrolysis should be developed to withstand intermittent operation and evaluated at the MEA level under relevant intermittent protocols.

At the same time, durability evaluation under intermittent operation requires further standardization. − Recent studies have shown that degradation behavior can vary substantially depending on the applied stress protocol, including current or voltage amplitude, pulse duration, cycling frequency, and total testing time. Therefore, future catalyst evaluation should also clearly report the operating protocol. Establishing standardized protocols for intermittent operation will be essential for reliable benchmarking of catalyst durability under practical operating conditions.

Case Study Demonstrating PEMWE Commercialization

PEMWE has long been considered a cornerstone technology for large-scale green hydrogen production, offering a high current density, compact design, rapid dynamic response, and compatibility with intermittent renewable energy. However, despite these promising attributes, the transition from laboratory-scale demonstrations to commercial-scale deployment has been hindered by several challenges, including durability under high current and pressure as well as the complexities of manufacturing consistency and system-level integration (e.g., gas/water management, thermal control, and safety).

In response to these challenges, the research and commercialization efforts led by Tao et al. have been particularly noteworthy. Over the past five years, the group has successfully advanced PEMWE from promising academic prototypes to a commercially viable 2.5 MW industrial system, achieving real-world deployment and technology transfer. This success illustrates a holistic approach that extends beyond catalyst improvement to encompass full-chain integrationfrom materials development and stack design to manufacturing standardization, system integration, and pilot-scale validation. In particular, advancements in large-scale MEA production using a double-sided slot-die coating technique are described in more detail.

As one of the main components of MEA, the anode catalyst layer (ACL), which typically contains 5–30 wt % ionomer, involves a combination of electron, proton, and mass transport processes, rendering its catalysis significantly more complex. The high ionomer ratio within MEA establishes a continuous proton transport network and a solid catalyst network, which are essential for achieving high activity and durability in PEMWE systems. Consequently, optimizing the structure of MEA is pivotal to maintaining both high activity and long-term durability.

Tao et al. have proposed several strategies aimed at optimizing the structure of MEA and enhancing its long-term stability in PEMWE. One of the key strategies involves the development of gradient ionomer distribution within the MEA. By exploiting the characteristics of catalyst sedimentation of ink, MEAs with gradient ionomer distribution were fabricated (Figure a). A higher ionomer concentration near ACL/PEM interfaces facilitates proton transport, while the catalyst-rich side promotes electron transport across PTL/ACL interfaces. Such distribution not only enhances proton conductivity but also improves the mechanical integrity of the MEA by minimizing the excess ionomer, which can cause swelling and other detrimental effects. This resulted in a more efficient and stable MEA that exhibited a reduced decay rate by a factor of 3 at 2 A cm^–2 and 80 °C (Figure b).

7.

(a) Schematic illustrations of gradient ionomer/catalyst distribution of the catalyst ink and ACL. (b) Durability measurement of gradient ionomer distributed (GID) and inverse gradient ionomer distributed (IGID) ACL at 2 A cm^–2 and 80 °C. Reproduced with permission from ref . Copyright 2024 Wiley-VCH. (c) Schematic illustrations of thermal/water cycles for the formation of ionomer networks. (d) Long-term durability verification of 50 cm² ionomer network-based ACL in a PEMWE at 80 °C and 2 A cm^–2. Reproduced with permission from ref . Copyright 2024 American Chemical Society.

Another critical approach is the pretreatment of MEA with thermal and water cycles to remove randomly distributed ionomers that could block mass transfer channels (Figure c). These treatments promote the formation of robust, three-dimensional ionomer networks (3D INs) in the ACL, thereby enhancing the structural integrity and durability of the MEA. The formation of efficient and stable mass transfer channels contributes to a low degradation rate of 3.0 μV/h at 2.0 A/cm² on a 50 cm² water electrolyzer (Figure d), demonstrating the improved stability achieved through this method.

Despite the effective strategies for enhancing the long-term operational stability of MEA, transitioning from lab-scale experiments to full-scale industrial applications presents significant challenges. Scaling up the MEA production process requires ensuring uniformity and precision in coating techniques and implementing automation and process control to maintain manufacturing consistency across larger volumes. Here, manufacturing consistency refers to the reproducibility and uniformity of MEAs fabricated by using the scalable coating process. To meet the requirements, Tao et al. introduced an innovative double-face slot-die coating technique (Figure a,b) for high-quality MEA manufacturing. The process includes automated slot-die coating units, drying and curing sections, and real-time monitoring systems. By optimizing these production processes, a high production line speed of 5 m min^–1 and a yield rate over 99% could be achieved, significantly outperforming traditional coating-decal transfer techniques, which operate at <0.5 m min^–1 and have a yield rate of around 90%. Additionally, manufacturing consistency has been satisfied, where the fabricated MEAs exhibited a cell voltage range of less than 30 mV at 2 A cm^–2, indicating their uniform quality. In addition, the hydrogen in the oxygen level remained below 0.5% over the entire current-density range, demonstrating their reliable gas-separation performance.

8.

(a) Schematic diagram and (b) physical picture of the slot-die coating technique. Physical pictures of MEA with different sizes: (c) 4 cm², (d) 50 cm², and (e) 600 cm². (f) Polarization curves of PEMWEs equipped with MEAs of different sizes and quantities. (g) Physical pictures of cells or stacks of different powers.

Benefiting from double-face slot-die coating techniques, Tao et al. can achieve scalable production of MEAs, from small laboratory samples (4 cm² and 50 cm²) to large industrial sizes (600 cm²) (Figure c–e). Remarkably, MEAs of different sizes demonstrated acceptable performance parity, which refers to maintaining comparable cell voltage (<20 mV) at industrially relevant current density (e.g., 2 A cm^–2) as the area of MEAs is increased, confirming the consistency and quality of the fabrication process (Figure f).

Particularly, the electrolyzers could be classified into “cell-, stack-, and system-scale” based on their corresponding power levels and evaluation objects. Cell-scale PEMWE refers to an individual single-MEA cell, typically with an active area below 50 cm² and an operating power of approximately 10–200 W. This scale is used to evaluate the intrinsic MEA properties, including polarization behavior, high-frequency resistance, mass-transport characteristics, Ir-specific activity, and degradation rate. Stack-scale PEMWE consists of multiple stacked large-area MEAs, generally with an active area above 100 cm² and, in some cases, up to 2000 cm², giving a total power from the kW to MW level, depending on cell area and number of layers. The scale is used to assess whether the cell-level MEA design can be transferred to large-area fabrication and stack assembly while maintaining voltage and temperature uniformity, homogeneous water/gas distribution, low hydrogen crossover, and stack-level durability.

System-scale deployment refers to the integrated PEMWE system, including the stack, power electronics, water purification, gas–liquid separation, thermal management, control system, safety interlocks, and other balance-of-plant components. Such scale evaluates macroscopic engineering integration and coordinated operation, including power conversion, heat and water management, pressure and purification control, safety monitoring, and long-term operational stability. Notably, this double-face slot-die manufacturing technique supports the production of MEAs for cells or stacks ranging from small 10 W units to large 2.5 MW systems, which is the largest single electrolyzer in China (Figure g). This capability to scale up production while maintaining high performance makes the double-face slot-die coating technique a crucial technology for cost-effective and efficient PEMWE systems. Nevertheless, industrial-scale PEMWE production is not limited to a single fabrication strategy, and other high-throughput approaches, such as gravure coating and screen printing, can also be considered depending on production requirements.

Beyond the technical validation of the 2.5 MW PEMWE deployment, lifetime and comparison with competing electrolysis technologies are critical for evaluating its industrial significance. The stack lifetime and operating stability were evaluated under industrial conditions. The 2.5 MW PEMWE system has operated stably for more than one year, with a voltage degradation rate below 5 μV h^–1 and hydrogen in oxygen level below 0.5%. These results indicate that the stack can maintain stable electrochemical performance and gas purity under long-term industrial operation.

To further enhance the understanding of the potential and limitations of PEMWE, a comparison between PEMWE and the competing technologies, i.e., alkaline water electrolysis (ALKWE) and AEMWE, is also summarized. ALKWE is currently the most mature and cost-competitive technology for large-scale hydrogen production, but it generally operates at lower current density and has slower dynamic response, larger footprint, and additional challenges associated with concentrated alkaline electrolyte and gas crossover under low-load operation. AEMWE is attractive because it combines several potential advantages of alkaline and PEM systems, including a lower noble-metal dependence, compact configuration, and rapid response. However, its long-term membrane stability, stack lifetime, and large-area manufacturing consistency still require further validation. PEMWEs offer advantages in high-current-density operation, compact stack design, rapid dynamic response, broad load range, and high-pressure hydrogen production, which are particularly relevant for renewable-energy-coupled hydrogen production.

The full commercial deployment of PEMWEs is still obstructed by several engineering barriers. First, further reduction of Ir loading must be achieved without compromising catalyst layer continuity, proton accessibility, or interfacial durability. Second, long-term operation under fluctuating power input may accelerate catalyst dissolution, ionomer migration, catalyst layer reconstruction, membrane degradation, and interfacial resistance growth. Third, stack-level water/gas distribution, compression uniformity, sealing reliability, and thermal management must be optimized to prevent local performance loss or accelerated degradation. Beyond MEA manufacturing, the cost of stack components should be reduced further; for example, Ti-based bipolar plates and PTLs can be replaced with coated stainless steel components while maintaining high performance and durability. Therefore, the significance of the present 2.5 MW PEMWE deployment lies in demonstrating the transferability of the MEA and stack design from laboratory-scale validation to industrial-scale operation, while further cost reduction and extended lifetime validation remain necessary for large-scale commercialization. At a broader system level, such cost reduction should also consider factors beyond the electrolyzer itself, such as electricity supply configuration and constraints on Ir supply.

By combining innovative MEA design strategies with advanced manufacturing techniques, the high performance and long-term stability of PEMWE systems have been ensured. These advancements are essential in bridging the gap between laboratory-scale research and industrial-scale applications, supporting the widespread adoption of PEMWEs for sustainable hydrogen production.

Summary and Outlook

There has been extensive research on water electrolyzers, and various catalysts have been developed to minimize the use of precious metals or even replace them entirely. While the high intrinsic activity of these newly developed catalysts is typically confirmed through RDE measurements, such a high intrinsic activity does not necessarily guarantee high performance in MEAs. Whereas intrinsic activity is measured in the absence of transport limitations, cell performance in MEAs is often governed by various transport processes including mass, electron, and charge transport. In particular, water electrolyzers are considered a means of storing surplus electricity generated from solar or wind power, which is inherently intermittent. Under such transient operating conditions, catalysts often undergo severe degradation; therefore, durability under an intermittent power supply must be carefully addressed.

When exemplified with acidic HER on Pt/C, the mass activity and Tafel slope vary significantly depending on the total catalyst loading (10–100 μg cm^–2) and Pt content (1, 5, 10, and 20 wt %) in RDE measurements, although the intrinsic activity of Pt nanoparticles is expected to be similar. However, when these catalysts are applied in MEAs, the mass activity observed in RDE does not translate proportionally into MEA performance, highlighting the importance of transport-related effects. Similar discrepancies are also observed in alkaline HER on Pt/C and acidic OER on IrO _x .

We also demonstrated recent works highlighting the importance of electron transfer at the PEMWE anode when a Ti PTL is used without Pt coating, as well as the importance of the durability of NiMo cathode catalysts in AEMWE operating under intermittent conditions. We also presented an example of PEMWE commercialization enabled by a double-sided slot-die-coating technique with controlled ionomer distribution, successfully deploying a 2.5 MW PEMWE system.

The following points are particularly important for bridging the gap between laboratory catalyst research and practical electrolyzer commercialization. We suggest the following framework for catalyst evaluation.

• RDE measurements should be used for rapid screening and mechanistic analysis while carefully accounting for measurement-dependent effects.

• Promising catalysts should be promptly evaluated in MEAs under practically relevant conditions to determine whether their advantages are translated to device-level performance.

• The dominant sources of performance loss should be identified under the target operating conditions by distinguishing kinetic, ohmic, ionic/electronic transport, and mass transport limitations.

• Catalyst loading, ionomer content, catalyst layer structure, and interfacial properties should be properly optimized according to the identified dominant losses.

• The optimized MEAs should be evaluated for durability under relevant dynamic operating conditions, followed by validation of their performance, reproducibility, and manufacturability on larger scales.

These considerations further suggest principles for the design of practical electrolyzers. The optimal condition of a catalyst cannot be defined independently of surrounding components; the catalyst, ionomer, membrane, and PTL should be cooptimized as an integrated system. Durability should be incorporated from the initial design stage so that catalytic activity, catalyst layer structure, and interfacial integrity can be retained during long-term and fluctuating operation. Scalability should be considered from the beginning of catalyst development by accounting for material utilization, large-area uniformity, manufacturing reproducibility, productivity, and energy efficiency. Overall, advancing electrolyzer technologies will require a paradigm shift from catalyst-centric design toward system-level optimization, integrating materials, interfaces, and operating conditions to achieve practical and scalable performance.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/jacsau.6c01104.

• RDE LSV curves with Tafel plots and PEMWE single cell results (PDF)

∥.

Sujin Park and Tianshu Gao contributed equally to this study.

The authors declare no competing financial interest.