Triggering charge redistribution–induced oxygen spillover mechanism in VO_x-supported IrSn alloy toward industrial water electrolysis

Abstract

Developing advanced iridium (Ir)–based oxygen evolution reaction (OER) catalysts is critical for proton exchange membrane water electrolyzers (PEMWEs). Unfortunately, conventional adsorbate evolution mechanism (AEM) and lattice oxygen mechanism (LOM) pathways suffer from an activity-stability trade-off, posing substantial challenges for catalyst design. Here, we report a charge-redistribution-induced oxygen (O) spillover strategy by designing amorphous VO_x-supported iridium-tin (IrSn) alloy OER catalysts, which can effectively transfer the poisoning oxygenated intermediates and maintain Ir valence stability (+2.5) during the dynamic OER to enhance activity and stability. In particular, the IrSn-VO_x–based PEMWE anode can deliver a current density of 3.0 amperes per square centimeter @ 1.798 volts (0.4 milligrams of platinum and Ir per square centimeter), surpassing the US Department of Energy (DOE) 2026 targets. A 25–square centimeter PEMWE operates stably for 5000 hours at industrial currents (≥25 amperes) with an exceptionally low degradation rate of 5.6 microvolts per hour, representing one of the best OER catalysts reported for practical PEMWEs. Theoretical calculations predict that the charge redistribution within IrSn-VO_x could reduce the kinetic energy barrier for *O spillover (from Ir to VO_x) by 69% relative to O-O coupling, thus triggering the O spillover against the Ir overoxidation/dissolution. Consequently, IrSn–VO_x follows a support-involved LOM pathway with a reduced rate-determining-step barrier (0.37 electron volts) relative to IrSn following AEM (0.66 electron volts). In addition, Sn doping in IrSn-VO_x can further promote the regeneration of VO_x lattice O and improve stability.

Charge redistribution triggers oxygen spillover of IrSn-VO_x for efficient water electrolysis.

INTRODUCTION

The proton exchange membrane water electrolyzers (PEMWEs) attract great interest due to their unique merits, featuring high current density, H₂ purity (>99.99%), and fast dynamic response (1, 2). However, the sluggish reaction kinetics and harsh reaction conditions of the anodic oxygen evolution reaction (OER) in PEMWEs would limit the operational efficiency and pose serious durability concerns during long-term operations, which requires efficient catalysts to address these challenges (3, 4). Given the relatively high corrosion-resistant properties under acidic media and high voltage conditions (pH < 3, >1.6 V), iridium oxide (IrO₂) remains the commercially viable anodic catalyst for PEMWEs (5, 6). However, high loading (>1.0 mg_Ir cm⁻¹) of Ir is usually demanded to maintain the OER rate and mitigate catalyst degradation under the operation at industrial-level current densities (≥1 A cm⁻²) (7). Unfortunately, the scarcity and high cost (∼$138,000 kg⁻¹) of the Ir would lead to a substantial increase in the manufacturing cost of PEMWE stacks. Therefore, effective strategies to improve the OER activity and stability of Ir-based catalysts and simultaneously reduce the Ir loading in PEMWEs are highly demanded.

Several strategies have been proposed to push the performance of Ir-based catalysts in PEMWEs (8). Alloying Ir with low-electronegative metals (e.g., Sn and Ga) represents a promising strategy to boost the intrinsic OER catalytic activity, where the electron transfer from the low-electronegative metals to Ir would downshift Ir’s d-band center and thus weaken excessively strong adsorption of oxygenated intermediates on Ir sites (9–11). In addition to the alloying strategy, introducing suitable catalyst supports, such as CeO₂ (12), TiO₂ (13), and TaB_x (14), has been identified as another effective approach to increase the Ir utilization efficiency and stabilize the Ir nanoparticles (NPs) via strengthening metal-support interaction (MSI), holding potential for PEMWE applications. Nevertheless, these approaches still face some limitations (15, 16). Under the harsh operating conditions of PEMWE, Ir-based catalysts inevitably suffer from metal dissolution into Ir³⁺ or IrO₄²⁻ when the voltage exceeds 1.4 V (17). The alloys will also undergo frequent reconstructions involving atomic migration/rearrangement and oxidative dissolution, leading to the collapse of the alloy framework and destruction of the active sites (18–20). In addition, the optimized MSI may diminish under harsh oxidative conditions (21, 22). Most of these Ir-based alloy catalysts follow the traditional adsorbate evolution mechanism (AEM) or lattice oxygen mechanism (LOM) during the OER (15). The scaling relationship of oxygenated intermediates in the AEM will result in a theoretical overpotential of 0.37 V and limit the OER activity. The LOM can break the scaling relationship to accelerate OER kinetics, but the participation of lattice oxygen in the LOM would lead to the loss of structural integrity and subsequently severe Ir dissolution. Thus, it is highly challenging to break the trade-off between activity and durability at a low Ir loading.

The key to breaking the activity-stability trade-off may lie in how to ingeniously switching the reaction mechanism to prevent the overoxidation of Ir and reduce the OER barrier. In this case, oxygen spillover refers to a dynamic process in which the adsorbed oxygenated species (e.g., O*) migrate rapidly from a metal-based component (e.g., Ir) to an adjacent heterogeneous site with a relatively stronger oxygen binding, offering a potential strategy to address the challenging trade-off. Effectively triggering oxygen spillover prevents excess accumulation of oxygenated intermediates on Ir active sites, thereby inhibiting overoxidation and mitigating irreversible catalyst deactivation or dissolution. On the other hand, it may reduce the energy barrier for the rate-determining step (RDS) or even alter the RDS to accelerate OER kinetics. Such strategy could also mitigate structural damage and support corrosion caused by the support-involved LOM as oxygen vacancies generated during the LOM may serve as buffers to adsorb migrating oxygenated intermediates, thus realizing dynamic lattice oxygen regeneration and stabilizing structure (23–25). However, how to rationally design such dual active site architectures (Ir-MO_x) and trigger efficient oxygen spillover during the OER to achieve remarkable PEMWE performance remains a largely uncharted domain.

Herein, we propose a previously unidentified design concept on a metal-support charge-redistribution-induced oxygen spillover mechanism to enhance the OER activity and stability for PEMWEs, using IrSn alloy on a VO_x support as a proof of concept. In this configuration, Sn doping and VO_x support could induce electron accumulation on Ir, rendering oxygenated intermediates easily desorbable to achieve oxygen spillover and these intermediates can be captured by VO_x with abundant oxygen vacancies. Such oxygen spillover could also rapidly remove the excess *O on Ir and considerably reduce the Ir valence state to a low value (<+2.5), thereby accelerating OER kinetics and mitigating Ir oxidative dissolution. As a result, the optimized hybrid IrSn-VO_x catalyst achieves a high OER activity with an ultralow overpotential of 207 mV at 10 mA cm⁻² and an exceptional catalytic stability (1000 hours) in 0.5 M H₂SO₄. Moreover, a practical PEMWE assembled with the IrSn-VO_x anode delivers a high current density of 3.0 A cm⁻² at a low cell voltage of 1.798 V (total loading: 0.4 mg_Pt+Ir cm⁻²), surpassing the US Department of Energy (DOE) 2026 target (3.0 A cm⁻² @ 1.800 V with <0.5 mg_Pt+Ir cm⁻²) and commercial IrO₂ catalysts (0.5 mg_Pt+Ir cm⁻² and 3.0 A cm⁻² @ 2.265 V). Moreover, the IrSn-VO_x catalyst was further fabricated into a 25-cm² membrane electrode assembly (MEA), which exhibits remarkable stability at the industrial current (25, 50, 100, and back to 25 A) for over 5000 hours with a low degradation rate of only 5.6 μV hour⁻¹. To the best of our knowledge, the performance of the IrSn-VO_x catalyst outperforms most of the state-of-the-art Ir-based catalysts. Theoretical calculations demonstrate that the IrSn NPs follow the traditional AEM pathway, with a high RDS (*OH to *O) barrier of 0.66 eV. Notably, the unique Sn doping and VO_x architecture would lead to a charge redistribution within IrSn-VO_x, enabling a 69% reduction in the kinetic energy barrier for oxygen spillover compared to that of O-O coupling and potentially triggering the oxygen spillover mechanism, i.e., *O migration from the Ir site to the VO_x site. As a result, the OER mechanism of the IrSn-VO_x transits to the support-involved LOM, where the regeneration of lattice oxygen in VO_x via the supplement from the water deprotonation becomes the RDS, with a lower energy barrier of 0.37 eV for activity enhancement. Sn doping in IrSn-VO_x can further promote oxygen spillover compared to Ir-VO_x, evidenced by an ∼35% reduction in the kinetic energy barrier, which, in turn, accelerates the regeneration of lattice oxygen in the VO_x support and improves stability.

RESULTS

Synthesis and structural characterizations

Sub–3 nm IrSn alloy NPs supported on vanadium oxide were prepared by a molten-salt-assisted reduction approach. In this protocol, the dispersed precursor was obtained by the decomposition of H₂IrCl₆ and C₁₈H₃₂O₄Sn on V₂O₃ through the impregnation procedure in the presence of NaNO₃. Subsequently, the above mixture was heat treated at 250°C for 2 hours under Ar + 5% H₂ to obtain IrSn-VO_x realizing oxygen spillover (Fig. 1A). The IrSn alloy and Ir-VO_x control samples were also synthesized via a similar procedure without the addition of V₂O₃ and Sn, respectively. The atomic-resolution high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image and corresponding fast Fourier transform (FFT) patterns were obtained from a single NP (Fig. 1B and fig. S1, A and B). The structure of the IrSn alloy was visualized by atomic arrangement with Z-contrasts (brighter: Ir, darker: Sn). The (200) and (111) lattice fringe spacings of IrSn-VO_x (0.190 and 0.219 nm) are similar to those of standard metallic Ir (0.191 and 0.221 nm), further implying the structural maintenance after the incorporation of Sn (fig. S1C) (26, 27). STEM–energy-dispersive spectroscopy (EDS) mapping images confirm the element molar/mass ratio and the homogeneous distribution of Ir, V, Sn, and O elements throughout the catalyst (figs. S1D and S2). According to the HAADF-STEM and transmission electron microscopy (TEM) images, the average size of IrSn NPs of the as-synthesized IrSn-VO_x was determined to be 2.2 ± 0.5 nm (fig. S3). The distribution of Ir NPs in IrSn (2.8 ± 0.8 nm) and Ir-VO_x (2.4 ± 1.4 nm) was also characterized (fig. S4). The Ir/Sn/V ratio in the IrSn-VO_x catalyst is about 10.6/4.4/85.0 (tables S1 and S2). X-ray diffraction (XRD) patterns (Fig. 1C) reveal that IrSn maintains a face-centered cubic (fcc) lattice with diffraction peaks slightly shifted to higher angles compared to Ir-VO_x with pure metallic fcc Ir phase, substantiating the formation of an alloy (28). The XRD pattern of VO_x exhibits no characteristic peaks, indicating its amorphous nature. Such a synthesis strategy can be readily extended to the preparation of other Ir-based alloy (e.g. IrGa, IrMn, and IrSb) catalysts (fig. S5).

Fig. 1. Structural characterizations of the studied samples.

(A) Illustration of the salt melting synthesis method and oxygen spillover mechanism of IrSn-VO_x. (B) Atomic-resolution HAADF-STEM image of IrSn-VO_x NPs. (C) XRD patterns of IrSn-VO_x, IrSn, and Ir-VO_x. a.u., arbitrary units. (D) High-resolution Ir 4f XPS spectra of IrSn-VO_x, IrSn, and Ir-VO_x, respectively. (E) Ir L₃-edge XANES spectra of Ir foil, IrO₂, and the studied samples. (F) Ir L₃-edge white line position as a function of the oxidation states of Ir for the studied samples, with Ir foil (0) and IrO₂ (+4) as standards. (G) Sn K-edge XANES spectra of IrSn-VO_x, IrSn, Sn foil, and SnO₂.

The electronic structures of the studied samples were analyzed by x-ray photoelectron spectroscopy (XPS) (fig. S6). As depicted in Fig. 1D, the high-resolution Ir 4f spectra can be deconvoluted into three distinct peaks. Notably, Ir⁰ and Ir²⁺ peaks account for the high proportion in IrSn and IrSn-VO_x, which clearly demonstrates the dominance of low-valence Ir (29). In contrast, the proportion of Ir⁴⁺ in Ir-VO_x increases markedly, indicating a higher oxidation degree. Obviously, the binding energy of the Ir 4f_7/2 peak in IrSn-VO_x (60.95 eV) shows a negative shift relative to that of Ir-VO_x (61.34 eV) (30). This is mainly attributed to the lower electronegativity of Sn (1.9) compared to Ir (2.2), resulting in the electron transfer from Sn to Ir. Moreover, the effects of electron transfer and MSI between IrSn NPs and the VO_x support lead to a greater Ir 4f_7/2 binding energy shift in IrSn-VO_x compared to that of pure IrSn (61.20 eV). To further analyze the chemical states and local coordination of Ir among the studied catalysts, x-ray absorption spectroscopy (XAS) was used. More electrons occupying the 5d orbitals mean a lower oxidized valence state of Ir, that is, a lower intensity of the white line in Ir L₃-edge x-ray absorption near-edge structures (XANES) spectra (31, 32). Figure 1E shows that the white line intensities decrease in the order of the Ir-VO_x, IrSn, and IrSn-VO_x, suggesting that Ir species in IrSn-VO_x have a lower average valence state (2.04) compared to that of IrSn (2.42) and Ir-VO_x (3.28), which is consistent with the XPS results and linear polynomial fit (Fig. 1F). As depicted in Fig. 1G, the absorption edge positions of IrSn and IrSn-VO_x of Sn K-edge are extremely close to that of SnO₂. This indicates a substantial electron transfer from Sn to Ir, causing a high oxidation state of Sn (+3.82). The interatomic distances characteristic of Ir─Ir/Sn and Ir─O bonds in R space of all the studied samples can be detected in the Fourier-transformed Ir L₃-edge extended x-ray absorption fine structure (EXAFS) spectra (fig. S7 and table S3). The fitting results demonstrate that there is scarcely any notable difference in the Ir─Ir bond lengths between IrSn, Ir-VO_x, and IrSn-VO_x, implying that these samples still have similar crystal frameworks.

Electrochemical measurements

The electrochemical active surface areas (ECSAs) of the studied catalysts were measured by cyclic voltammetry (CV) in a N₂-saturated 0.5 M H₂SO₄ electrolyte. Compared with Ir-VO_x and IrSn, the ECSA of IrSn-VO_x is ∼2 and 8 times larger, respectively (fig. S8) (33). The OER performance of the catalysts was evaluated via linear sweep voltammetry (LSV) in a 0.5 M H₂SO₄ electrolyte. As presented in Fig. 2 (A and B), the overpotential of IrSn-VO_x at a current density of 10 mA cm⁻² is merely 206 mV, which is substantially lower than that of commercial Ir-VO_x (249 mV), IrSn (254 mV), and commercial IrO₂ (333 mV). Furthermore, IrSn-VO_x requires a low overpotential of 261 mV to achieve 100 mA cm⁻². Besides, the Tafel slope of IrSn-VO_x (57 mV dec⁻¹) was notably smaller than those of IrSn (71 mV dec⁻¹) and Ir-VO_x (68 mV dec⁻¹; Fig. 2C). These results suggest that alloying with Sn and incorporating the VO_x support can substantially enhance the OER catalytic activity of Ir, representing one of the best performances among the reported studies (table S4) (34–40).

Fig. 2. Electrochemical measurements of different catalysts.

(A) Polarization curves of IrSn-VO_x, IrSn, and Ir-VO_x with 90% iR-drop compensation in 0.5 M H₂SO₄ at 5 mV s⁻¹. (B) Overpotential at 10 and 100 mA cm⁻² of the studied samples. (C) Tafel slopes of different catalysts. (D) EIS of the corresponding catalysts at 1.5 V_RHE. (E) MA at the overpotential of 250 mV of IrSn-VO_x, IrSn, and Ir-VO_x. (F) CP curves at 100 mA cm⁻² for IrSn-VO_x, IrSn, Ir-VO_x, and IrO₂. The inset shows the CP curve at 10 mA cm⁻² for IrSn-VO_x.

Electrochemical impedance spectroscopy (EIS) analysis at 1.50 V_RHE (Fig. 2D) reveals that IrSn-VO_x shows low charge-transfer resistance and rapid electron transfer during the OER process. In particular, their OER activities were further measured in terms of mass activity (MA) at 1.48 V versus reversible hydrogen electrode (RHE) (Fig. 2E). The MA of IrSn-VO_x reaches 1.33 A mg_Ir⁻¹, ∼6.3, 7.8, and 66.5 times higher than those of Ir-VO_x (0.21 A mg_Ir⁻¹), IrSn (0.17 A mg_Ir⁻¹), and IrO₂ (0.02 A mg_Ir⁻¹). The calculated turnover frequency (TOF) values indicate the higher intrinsic activity of IrSn-VO_x (0.105 s⁻¹), which is higher than those of Ir-VO_x (0.031 s⁻¹), IrSn (0.020 s⁻¹), and IrO₂ (0.003 s⁻¹) (fig. S9).

Chronopotentiometry (CP) tests at 100 mA cm⁻² were conducted to evaluate catalyst stability (Fig. 2F). Impressively, IrSn-VO_x shows no marked potential variation over 100 hours of continuous operation, whereas IrSn, IrO₂, and Ir-VO_x exhibit rapid potential increases within 30 hours. In addition, IrSn-VO_x undergoes only a slight overpotential increase after 1000 hours at 10 mA cm⁻², displaying the outstanding stability. Moreover, the XRD pattern after the long-term CP test confirms that the crystal structure of IrSn-VO_x (metallic fcc Ir core) can be well maintained (fig. S10). The TEM image of the IrSn-VO_x after a 100-hour stability test indicates that the average particle size of the IrSn NPs remains unchanged (fig. S11), implying the robust structural stability of IrSn-VO_x. In addition, this synthesis strategy of oxide-supported IrM alloy can be generalized for doping various low electronegative metals, such as IrGa-VO_x, IrSb-VO_x, and IrMn-VO_x (overpotentials of 236, 226, and 209 mV at 10 mA cm⁻², respectively), to improve the OER performance (fig. S12).

Dissolution mechanism of IrSn-VO_x

To elucidate the origin of the OER performance of IrSn-VO_x, it is essential to probe the alterations in the valence state of Ir and catalyst structure throughout the OER process. CP tests at different current densities were conducted to explore the ability of active sites to maintain high efficiency after the continuous OER process (Fig. 3A and fig. S13). The results reveal that IrSn-VO_x presents remarkable efficiency maintenance among the studied catalysts, where the IrSn-VO_x catalyst exhibits a low overpotential increase (4 mV) after CP tests at various current densities. In contrast, the IrSn (35 mV) and Ir-VO_x (26 mV) catalysts show apparent overpotential increase under the same conditions. Furthermore, CV curves with adjusted upper potential limits (UPLs) from 0.45 to 1.5 V_RHE were recorded under forward and backward scanning modes to dynamically monitor the surface structure evolution of IrSn-VO_x, IrSn, and Ir-VO_x to assess their reconstruction reversibility, i.e., the reversibility between metallic Ir and IrO_x (41). As presented in Fig. 3B, the hydrogen underpotential deposition (H_upd) peaks of IrSn-VO_x within the range of 0.05 to 0.3 V, characteristic of low-valence Ir (<+4), gradually become weakened as UPLs increase during the forward scan. Notably, this peak remains discernible even at the UPL of 1.45 V for IrSn-VO_x, indicating the maintenance of a stable metallic Ir valence state. Concurrently, a redox peak corresponding to the Ir³⁺/Ir⁴⁺ couple emerges between 0.8 and 1.2 V, strongly suggesting the formation of an IrO_x layer derived from moderate and reversible oxidation of Ir on the surface. Besides, the IrSn-VO_x catalyst consistently exhibits a pronounced H_upd peak upon backward scanning of the UPLs (Fig. 3C), and a similar behavior was also observed in the IrSn (fig. S14). Comparably, the Ir-VO_x and IrO₂ catalyst displays a rapidly diminished H_upd peak and pronounced Ir³⁺/Ir⁴⁺ redox peaks, indicating irreversible transformation to severe oxidation of the IrO_x layer with increasing applied UPLs (figs. S15 and S16). These electrochemical results underscore that the electron donation from Sn to Ir in IrSn-VO_x contributes to the formation of a more stable metallic Ir core, facilitating the reversible formation of a thin IrO_x layer. Furthermore, operando XANES analysis (Fig. 3D) was conducted to investigate the average chemical state changes of the studied samples after applying a potential and long-term electrolysis. For the IrSn-VO_x catalyst, although electrochemical oxidation of metallic Ir to thin IrO_x occurs throughout the OER process [open-circuit potential (OCP) to 1.55 V], the Ir valence state remains relatively stable (+2.5) even at higher potentials (Fig. 3E). After testing at high potential and holding back to the OCP, the Ir valence state can be recovered to near-initial levels, which is attributed to electronic donation from both alloying and MSI. Conversely, the Ir valence states of the IrSn and Ir-VO_x catalysts increase considerably at high potential and cannot fully recover upon returning to OCP (fig. S17). According to the EXAFS, the structural evolution of IrSn-VO_x during the OER (from the OCP to 1.55 V versus RHE) exhibits distinct potential-dependent trends in Ir─O coordination and Ir─Ir bond length (Fig. 3F, figs. S18 and S19, and table S5). Specifically, as the potential increases to 1.45 V, the coordination number (CN) of Ir─O and Ir─Ir bond length gradually increases, indicating that Ir actively adsorbs oxygenated species at the surface and leads to a slight rise in Ir oxidation. Furthermore, when the potential further increases beyond 1.55 V, the CN of Ir─O and Ir─Ir bond length decreases noticeably. Such an inverse variation in CN of Ir─O and Ir─Ir bond length may suggest the occurrence of oxygen spillover, where the originally adsorbed oxygenated species on Ir sites migrate to VO_x. The migration of oxygen species from Ir to VO_x can maintain a relatively low-valence state of Ir, which is consistent with the insights derived from in situ XANES. In addition, the valence state of Sn remains largely unchanged, maintaining a high oxidation state, thereby continuously supplying electrons to Ir under prolonged CP testing (fig. S20).

Fig. 3. Dissolution mechanism of IrSn-VO_x during the dynamic OER.

(A) CP curves at 10, 100, 200, and 10 (after 200) mA cm⁻² for IrSn-VO_x. CV curves of IrSn-VO_x recorded with (B) forward potential scans and (C) backward potential scans of 50 mV s⁻¹ in 0.5 M H₂SO₄. (D) Ir L₃-edge XANES spectra of IrSn-VO_x at applied voltages from the OCP to 1.55 V (inset: the enlarged white line region of Ir L₃-edge). (E) Average oxidation state of Ir in IrSn-VO_x, IrSn, and Ir-VO_x at applied voltages from the OCP to 1.55 V. (F) Values of the Ir─O CN and Ir─Ir bond length of IrSn-VO_x at applied voltages from the OCP to 1.55 V. (G) pH dependence on the OER potential at different current densities for IrSn-VO_x. (H) Molar dissolution rate for Ir, Sn, and V of the studied samples during CP and CV processes. (I) In situ Raman spectra of IrSn-VO_x under the applied potential range from the OCP to 1.5 V in 0.5 M H₂SO₄.

Subsequently, the intrinsic OER mechanisms of various studied samples were elucidated via the pH-dependent tests. IrSn and IrO₂ show pH-independent OER kinetics, typical for the AEM (figs. S21 and S22). In contrast, IrSn-VO_x and Ir-VO_x display distinct pH-dependent behavior with a slope of ∼50 mV dec⁻¹, approaching the previously reported 60 mV per pH shift on the RHE scale (Fig. 3G and fig. S23) (42). This divergence strongly implies the existence of nonconcerted proton-electron transfer steps for IrSn-VO_x and Ir-VO_x, thereby supporting the involvement of the LOM in driving the OER. This result could also serve as evidence that the oxygenated intermediate is transferred from IrSn to VO_x, thereby triggering the support-involved LOM. The dissolution rates of different elements were gathered under OER-relevant test conditions (e.g., 50 cycles CV between 1.0 and 1.5 V versus RHE and CP tests at 100 mA cm⁻² for 100 hours) and subsequently used as a crucial basis for analyzing the OER stability of the studied catalysts. As shown in Fig. 3H, the dissolution rates of Ir and Sn are relatively low across all catalysts. In Ir-VO_x, V exhibits a high dissolution amount both during CV activation and CP testing, reflecting structural instability caused by lattice oxygen participation in the reaction. By contrast, the V dissolution rate in IrSn-VO_x is approximately halved as Sn triggers efficient oxygen spillover, which facilitates the regeneration of VO_x lattice oxygen to enhance structural stability. In this system, although V undergoes mild reconstruction and initial dissolution during CV activation, its dissolution decreases substantially in the CP test, confirming that VO_x maintains structural integrity and sustains OER performance throughout the entire process. In addition, the in situ Raman spectra at selected applied potentials were also collected to explore the structure evolution of IrSn-VO_x throughout the electrochemical process (Fig. 3I). The peaks at 429 and 893 cm⁻¹ are assigned to the Sn-O vibrations and O-V-O bending modes (43, 44). With the gradual increase in applied voltage, the position of the peak at 593 cm⁻¹ (corresponding to the Ir-O stretching vibration for IrO_x) shifts slightly toward higher wave numbers, also indicating the shortening of the Ir─O bond length and the enhanced bond strength (45). Notably, the intensities of these peaks remained almost unchanged, implying that IrSn-VO_x retains its structural integrity without collapse or damage during the OER. As reported previously, the proportion of deconvoluted free H₂O peak (3625 cm⁻¹) increases to 6.3% during the OER process, manifesting enhanced connectivity of the hydrogen bond network, which could further enhance OER activity (fig. S24) (46). In addition, this dynamic evolution from hydrogen-bonded water to free water can not only ensure the timely replenishment of oxygenated species for IrSn-VO_x but also facilitate the regeneration of VO_x lattice oxygen during the OER.

PEMWE performance

The IrSn-VO_x catalyst was further evaluated in an MEA as the anode for PEMWEs (Fig. 4A). Even at 90% reduced Ir loading, the IrSn-VO_x-based MEA (5 cm², 0.25 mg_Ir cm⁻²) delivers an industrial current density of 1.0 A cm⁻² at a low voltage of 1.630 V (Fig. 4B), far surpassing commercial IrO₂ MEA (2.0 mg_Ir cm⁻², 1.0 A cm⁻² @ 1.950 V). It also shows a cell performance of 3.0 A cm⁻² @ 1.798 V, meeting the DOE 2026 target (3.0 A cm⁻² @ 1.800 V). The performance of IrSn-VO_x and IrO₂-based PEMWEs using the Nafion 115 membrane have also been collected, which also demonstrates that the IrSn-VO_x anode has better performance compared to IrO₂ (fig. S25). In addition, the IrSn-VO_x–based MEA could be stably operated more than 1200 hours at 1.0 A cm⁻² with a voltage degradation rate of only 23 μV hour⁻¹ (Fig. 4C and fig. S26). In contrast, IrSn, Ir-VO_x, and IrO₂ all suffer from rapid decay after operating (<450 hours). To verify the substantial promise regarding the industrial applications of PEMWEs, IrSn-VO_x was further used in a scale-up MEA of 25 cm² (Fig. 4D and fig. S27). The cell voltage maintained a stable value of 1.723, 1.969, and 2.168 V at currents of 25, 50, and 100 A for at least 400 hours, respectively. After 5000 hours of continuous operation under such high currents (≥25 A), the cell exhibits an negligible voltage decay rate of only 5.6 μV hour⁻¹, further validating the credibility of the oxygen spillover effect in enhancing the MEA durability. The XRD and TEM results confirm that the crystal structure of IrSn-VO_x and particle size can be well maintained after PEMWE operation (fig. S28). On the basis of the excellent MEA performance, the total hydrogen levelized cost is estimated at $1.98 per kg of H₂, meeting the DOE 2026 target of $2 per kg of H₂ (47). In addition, a high energy efficiency of 71.8% at 1.0 A cm⁻² can be obtained, exceeding the efficiency of commercial PEMWE (∼65%; details in Supplementary Notes). Such IrSn-VO_x catalyst represents one of the best PEMWE performances ever reported in terms of operation time and degradation rate, rendering it attractive as a promising OER electrocatalyst for practical applications (Fig. 4E and table S6) (7, 29, 48–67).

Fig. 4. PEMWE performance of IrSn-VO_x.

(A) Illustration of a single-cell PEMWE. (B) Polarization curves obtained in the PEMWE at 80°C. Cathode: 0.15 mg_Pt⁻¹ commercial Pt-C. Anode: IrO₂/IrSn-VO_x. (C) CP curve of a 5-cm² PEMWE with IrSn-VO_x and IrO₂ at 1 A cm⁻². (D) CP curves of a 25-cm² PEMWE with IrSn-VO_x at a sequential current of 25, 50, 100, and 25 A. (E) Comparison of the stability (operation time and degradation rate) of IrSn-VO_x systems with the advanced systems available for PEMWEs recently reported. The details of all the referenced catalysts can be found at table S6 (Supplementary Materials).

Oxygen spillover mechanism of IrSn-VO_x

In situ ¹⁸O isotope-labeling differential electrochemical mass spectrometry (DEMS) measurements are carried out to validate the OER mechanism. The electrochemical tests were carried out in ¹⁸O-labeled catalysts (0.5 M H₂SO₄). The signals of mass/charge ratio (m/z) = 32 and m/z = 34 are detected in IrSn-VO_x, which correspond to the generation of ³²O₂ (¹⁶O¹⁶O) and ³⁴O₂ (¹⁸O¹⁶O), respectively (Fig. 5A). In addition, the signal intensity ratio of ³⁶O₂ (0.1%) is almost undetectable, whereas that of ³⁴O is ∼83.5%. This result clearly demonstrates the involvement of lattice oxygen in the OER process of IrSn-VO_x, thereby confirming the activation of the support-involved LOM. The appearance of the ³²O₂ signal (16.4%) likely results from the replenishment of VO_x lattice oxygen by the ¹⁶O derived from water after the lattice oxygen participates in the OER, followed by oxygen release. Specifically, this provides a plausible elucidation for the good OER performance of IrSn-VO_x. In contrast, the majority signal of m/z = 32 can be found in IrSn and IrO₂, accounting for the AEM (fig. S29). This indicates that the VO_x support is crucial as the dual site to trigger the oxygen spillover. To offer compelling evidence, density functional theory (DFT) calculations were performed to evaluate the viability of the proposed oxygen spillover mechanism and the enhanced OER performance for IrSn-VO_x. Because Ir always maintains a low oxidation state in IrSn-VO_x during the OER, a slab model consisting of five atomic layers was used, featuring exposed IrSn (111) and amorphous VO_x (110; derived from V₂O₃) surfaces (fig. S30). Figure 5B and fig. S31 show the planar-averaged electron density difference curve of IrSn-VO_x and Ir-VO_x, in which more electron transfers from Sn atoms and VO_x support to Ir within IrSn-VO_x. Thus, the d-band centers of VO_x and Ir are optimized to synergistically regulate their adsorption of oxygenated species, thereby facilitating the triggering of oxygen spillover. To probe the electronic structures, the density of states (DOS) of Ir and V in IrSn-VO_x was computed. The results unveil that the d-band centers are found at −2.528 eV for Ir and −1.948 eV for V (Fig. 5C). Notably, the d-band center of V is positioned closer to the Fermi level, suggesting a stronger binding between V and the adsorbed oxygenated species (V-*O) compared to that of Ir-*O. This result indicates that lattice oxygen sites in VO_x are more favorable than that of IrO_x for *OO intermediate generation, which provides the prerequisite for the occurrence of oxygen spillover and the prevention of the structural collapse of IrSn.

Fig. 5. Catalytic mechanism investigation of the oxygen spillover process.

(A) DEMS of the ¹⁶O¹⁸O and ¹⁸O¹⁸O signals of IrSn-VO_x measured in the ¹⁸O-labeled catalysts. (B) Planar-averaged electron density difference curve of IrSn-VO_x (the inset is the corresponding charge density difference map). (C) DOS plots of Ir d and V d states in IrSn-VO_x. The dashed lines mean the d-band centers. PDOS, projected DOS. (D) Calculated energy barrier diagram of IrSn under U = 0 and 1.23 V with AEMs. (E) Comparisons of the energy barrier between IrSn-VO_x and Ir-VO_x under U = 0 and 1.23 V with oxygen spillover mechanisms. (F) Schematic illustration of the proposed oxygen spillover in the IrSn-VO_x catalysts for the OER in acid. Purple, blue, red, and white balls represent Ir, V, O, and H atoms, respectively. (G) Calculated activation barriers for the oxygen spillover step (red line) and O-O coupling on Ir sites (yellow line) of IrSn-VO_x. TS, transition state. (H) Comparison and promotion rate of the calculated interface cohesive energies between IrSn-VO_x and Ir-VO_x.

According to the predominant signal of ³⁶O₂ in IrSn, the Gibbs free energy diagrams along the OER coordinate of IrSn were obtained under the classic AEM pathway (fig. S32). Given the overly intense binding of *OOH, individual IrSn faces an excessive energy barrier of the RDS of *OH deprotonation (*OH → *O + H⁺ + e⁻) under both U = 0 V versus RHE (0.66 eV) and U = 1.23 V versus RHE (1.89 eV; Fig. 5D). Notably, the synergistic effect of Sn and VO_x in the IrSn-VO_x heterostructure plays a pivotal role in changing the OER mechanism, i.e., the support-involved LOM pathway (figs. S33 and S34). We calculated several critical parameters to validate the oxygen spillover. The *OH adsorption energy on the VO_x sites (0.41 eV) is higher than that at Ir sites (0.34 eV) in the IrSn-VO_x heterostructure (fig. S35), consistent with the d-band center analysis. Furthermore, the kinetic activation barriers for the oxygen spillover and O-O coupling steps on Ir of IrSn-VO_x were obtained (Fig. 5G and fig. S30). The kinetic barrier for oxygen spillover is ∼3 times lower than that for the O-O coupling, suggesting the favorable oxygen spillover process. Therefore, the *OH is initially formed at the IrSn site after the first deprotonation of H₂O (* + H₂O → *OH + H⁺ + e⁻). Subsequently, *O obtained by *OH deprotonation can spillover to the adjunct VO_x and couple with the lattice oxygen of VO_x to generate *OO intermediate, followed by O₂ desorption, accounting for the results of the valence maintenance of Ir. In addition, Sn doping can reduce the kinetic activation barrier of oxygen spillover on IrSn-VO_x by ∼35% relative to Ir-VO_x, which can promote the oxygen spillover process, facilitate the regeneration of VO_x lattice oxygen, and maintain good structural stability of VO_x for the IrSn-VO_x (fig. S36).

We further calculated the OER Gibbs free energy diagram according to the oxygen spillover and support-involved LOM (Fig. 5, E and F). The RDS of Ir-VO_x is the energy required for the spillover of *O adsorbed on Ir to VO_x (0.58 eV). After Sn doping, filling oxygen vacancy (O_v) in the VO_x with the *OH deprotonation and formation of fresh lattice oxygen in VO_x (*H → * + H⁺ + e⁻) becomes the RDS of IrSn-VO_x, with a lower energy barrier of 0.37 eV. Furthermore, the stability of IrSn-VO_x was evaluated by examining the adhesion energy (Fig. 5H). The adhesion energy of IrSn-VO_x (−2.36 eV) is ∼23.6 times lower than that of Ir-VO_x (−0.10 eV). This indicates that the addition of Sn to Ir notably reinforces the interfacial adhesion between IrSn and VO_x, thereby improving thermodynamic stability. The resistant of IrSn NPs to detachment and dissolution is ascribed to the strong MSI between IrSn and VO_x. It also explains that the V dissolution rate remains at a low level after the long-time CP test under a high current density.

Overall, Sn doping and the VO_x support in IrSn-VO_x induce charge redistribution, triggering oxygen spillover and enabling rapid migration of oxygenated intermediates away from Ir sites as part of the support-involved LOM. In this process, the highly active O-O coupling step shifts to the stable VO_x support, avoiding direct lattice oxygen supply from Ir. This prevents Ir overoxidation and dissolution by transferring oxygenated intermediates while also facilitating lattice oxygen regeneration on VO_x to maintain its structural integrity. As a result, the energy barrier of RDS (the regeneration of lattice oxygen) for the OER on IrSn-VO_x is lowered by 0.21 eV compared to that on Ir-VO_x, and both activity and stability of the IrSn-VO_x catalyst for the acidic OER have been enhanced.

DISCUSSION

In summary, we propose a previously unidentified design concept on a metal-support charge-redistribution-induced oxygen spillover mechanism, as embodied in the IrSn-VO_x catalyst, to enhance the OER activity and stability for PEMWE applications. The presence of Sn and VO_x is critical for triggering oxygen spillover and maintaining Ir valence stability during the dynamic OER, thereby improving the OER activity and stability of IrSn-VO_x. The optimized IrSn-VO_x catalyst exhibits a low overpotential of 207 mV at 10 mA cm⁻² and maintains stable performance for 1000 hours in 0.5 M H₂SO₄. A PEM electrolyzer with the IrSn-VO_x catalyst achieved 3 A cm⁻² at 1.798 V at a low catalyst loading of 0.4 mg_Ir+Pt cm⁻², outperforming the DOE 2026 target (1.800 V @ 3 A cm⁻², 0.5 mg_Ir+Pt cm⁻²). Moreover, the IrSn-VO_x exhibits a low degradation rate of only 5.6 μV hour⁻¹ after 5000-hour operation at the industrial current (≥25 A) in a 25-cm² MEA, ranking among the best reported PEMWE performance. Theoretical calculations predict that the Sn doping and VO_x support of IrSn-VO_x can effectively induce charge redistribution, which subsequently lowers the kinetic barrier for the oxygen spillover process by 69% compared to that for O-O coupling and promotes the support-involved LOM. The new OER mechanism of IrSn-VO_x can reduce the RDS barrier (0.37 eV) compared to the IrSn (AEM, 0.66 eV) and Ir-VO_x (support-involved LOM, 0.58 eV), rendering a better OER activity. Meanwhile, the spillover mechanism can inhibit the overoxidation/dissolution of Ir by removing oxygenated intermediates and maintain the structural integrity of VO_x by facilitating regeneration of lattice oxygen, accounting for the good stability of the IrSn-VO_x catalyst. This innovative method successfully adjusts the OER mechanism and overcomes the challenge of balancing activity and durability for Ir-based electrocatalysts in PEMWEs.

MATERIALS AND METHODS

Chemicals and materials

VOSO₄∙2H₂O, C₁₈H₃₂O₄Sn, and NaNO₃ were purchased from Aladdin. Chloroiridic acid hydrate (H₂IrCl₆·6H₂O; 99.9%) and citric acid anhydrous were purchased from Innochem. Ethanol (AR) and isopropanol were purchased from Sinopharm Group Co. Ltd. Nafion (5% in a mixture of lower aliphatic alcohols and water) were purchased from Alfa Aesar.

Synthesis of V₂O₃

A specific ratio of VOSO₄∙2H₂O was dissolved in 30 ml of deionized water. Citric acid anhydrous (1:1 molar ratio with metal ions) was dissolved in 30 ml of deionized water. After stirring the both solution for half an hour, the above solutions were mixed together and stirred continuously at 80°C until the solvent was evaporated to form a sol-gel. The obtained powder was treated under air at 350°C for 1 hour and then under Ar + 5% H₂ at 700°C for 2 hours to obtain the final sample V₂O₃.

Synthesis of IrSn-VO_x

Twenty milligrams of above-prepared V₂O₃ was dispersed in 3 ml of ethanol and sonicated for 30 min to ensure uniform dispersion. The Ir precursor (H₂IrCl₆·6H₂O; 1 g) was dissolved in distilled water (100 ml) to obtain a chloroiridic acid solution (10 mg ml⁻¹). Then, 300 mg of NaNO₃, 8 μl of C₁₈H₃₂O₄Sn, and 1 ml of the chloroiridic acid solution (10 mg ml⁻¹) were added to the V₂O₃ suspension and sonicated for 120 min. Then, the solution was stirred continuously at 80°C until dry. After getting the powder, the sample was annealed at 150°C for 2 hours and 375°C for 1 hour in air. The sample obtained above was dispersed in deionized water and sonicated for 2 hours to remove NaNO₃. The product was redispersed in ethanol and centrifugated (9000 rpm, 10 min). Then, the product was heated at 250°C for 2 hours under air and sequentially under Ar + 5% H₂ to obtain the final sample IrSn-V₂O₃.

Synthesis of IrSn

This sample was prepared using the same procedure as described above, except that V₂O₃ was not added as a support.

Synthesis of Ir-VO_x

This sample was prepared using the same procedure as described above, except that no C₁₈H₃₂O₄Sn precursor was used.

Material characterizations

The actual Ir loadings of IrSn-VO_x can be determined by inductively coupled plasma mass spectroscopy (ICP-MS) (7500CX, Agilent 720ES) and confirmed to be 15.9%. The morphology of the obtained catalysts was observed with TEM (Tecnai G2 20, 200 KV). The high-resolution TEM, HAADF-STEM, and the corresponding energy-dispersive x-ray spectroscopy (EDX) images were obtained on field-emission transmission electron microscope (Talos F200X). The crystalline phase information was obtained by XRD (MiniFlex-600, Rigaku) with Cu Kα radiation (0.15406 nm, 40 kV, 15 mA). The bonding of platinum with different supports was analyzed by XPS (Thermo Fisher Scientific K-Alpha). EXAFS at the Ir L₃-edge and Sn K-edge was performed at BL11B of the Shanghai Synchrotron Radiation Facility. The electron-storage ring was operated at 8 GeV with a current of 100 mA. A Si (111) double-crystal monochromator was used. All XAS spectra were recorded at room temperature in transmission mode.

Electrochemical measurements

Electrochemical measurements were performed with a three-electrode system on CHI760e electrochemical workstation (Shanghai Chenhua Instrument Corporation, China). A glassy carbon rotating disk electrode was used as a working electrode (5 mm in diameter), Ag/AgCl (saturated KCl) was used as a reference electrode, and platinum plate was used as a counter electrode. All potentials were experimentally converted to values with reference to an RHE (versus RHE). The catalysts were dispersed in a mixture containing 990 μl of isopropanol and 10 μl of Nafion (5%) to form an ink (2 mg ml⁻¹). Sixty microliters of ink was cast on the carbon paper and dried at room temperature. The cyclic voltammograms were used to determine the electrochemical surface areas, which were obtained by scanning the working electrode potential between 0.9 and 1.0 V (versus RHE) in N₂-saturated 0.5 M H₂SO₄ solutions at a scan rate range of 10 to 100 mV s⁻¹. The electrochemical double-layer capacitance was determined by recording the CV within non-Faradaic potentials at different scan rates (10, 20, 30, 40, 50, and 100 mV s⁻¹). The slope of linear regression of current density difference (j_anode − j_cathode) against scan rate is twice that of the double-layer capacitance, which is proportional to the ECSA of the catalysts. LSV was used to determine the OER activity performance of the catalysts by scanning the working electrode potential between 1.2 and 1.7 V (versus RHE) in 0.5 M H₂SO₄ with 90% iR drop compensation at a sweep rate of 5 mV s⁻¹. The CP test was used to assess the stability of electrocatalysts at a constant current density of 10 and 100 mA cm⁻² in the 0.5 M H₂SO₄ solution. The dissolution of metal elements from the catalyst into the electrolyte solution during the CV and CP tests was quantified by collecting 2 ml of the electrolyte at regular intervals for the ICP-MS test.

In situ ¹⁸O isotope-labeling DEMS experimental details

The in situ DEMS experiments of the IrSn-VO_x electrocatalyst were performed on an in situ differential electrochemical mass spectrometer provided by Linglu Instruments (QAS100, Shanghai) Co. Ltd. First, pristine samples were isotopically labeled with ¹⁸O in the ¹⁸O-labeled 0.5 M H₂SO₄ solution at 1.5 V for 10 min. The resultant electrodes were then rinsed several times with H₂¹⁶O and dried in an oven to remove residual H₂¹⁸O. After these processes, the samples were subjected to chronoamperometry at 1.5 V for in situ DEMS measurements.

Electrochemical measurements in a PEM electrolyzer

Anode catalysts (IrSn-VO_x and commercial IrO₂) were mixed with Nafion, isopropanol, and water by ultrasonicating for 1 hour to form homogeneous ink. The ink was then sprayed onto Nafion 212 and 115 membranes at 70°C with an active area of 5 and 25 cm². The anode Ir loading of IrSn-VO_x and commercial IrO₂ were controlled as 0.25 mg_Ir cm⁻². The cathode Pt loading of Pt-C was controlled as 0.15 mg_Pt cm⁻². Electrolysis tests were conducted using a single-cell PEM electrolyzer. The PEM electrolyzer was operated at 80°C with distilled water as a reactant under a flow rate of 100 ml min⁻¹. The MEA was sandwiched between two Ti plates with single serpentine flow channels.

X-ray absorption spectroscopy

XAS data were processed and analyzed by ATHENA and ARTEMIS, respectively. Energy scale calibration was carried out using the reference foil data, with the absorption edge energy (E₀) then set to the first derivative of the sample data. Fits were carried out using a k range of 2 to 14 Å⁻¹ and an R range of 1 to 4 Å, with multiple k weightings of 1, 2, and 3. The spectra were processed using standard procedures for subtracting the background and normalizing to the magnitude of the L₃-edge and K-absorption edge jump using the ATHENA. The first shell was used to fit the EXAFS spectra. With the information of the IrSn-VO_x structure, path fitting was carried out by fixing the Ir-Ir, Ir-Sn, and Ir-O path. The four parameters—CN, bond length (R), Debye-Waller factor (σ²), and E⁰ shift (ΔE⁰)—were fitted without any being fixed, constrained, or correlated.

First-principles calculations

All spin-polarized calculations were performed by using the DFT calculations in the Vienna ab initio simulation package (VASP) (68, 69). The projector augmented wave (PAW) method was applied to treat electron-ion interactions, and the Perdew-Burke-Ernzerhof (PBE) exchange-correlation functional within the generalized gradient approximation (GGA) was used to describe the electron interactions (70, 71). For the dispersion interaction, the van der Waals correction was performed in Grimme’s D3 scheme (72). A plane-wave cutoff energy of 400 eV was used in all simulations, and the Monkhorst-Pack k-point grid was adopted to sample the Brillouin zone for geometry optimizations (73). The adsorption energies were calculated using the following expressions. These adsorption energies are defined as

$$\mathrm{\Delta }{G}_{\ast \text{OH}}=G_{\ast \text{OH}}-G_{\ast }-(G_{\mathrm{H}2\mathrm{O}}-1/2G_{\text{H2}})$$

$$\mathrm{\Delta }{G}_{\ast \mathrm{O}}=G_{\ast \mathrm{O}}-G_{\ast }-(G_{\mathrm{H}2\mathrm{O}}-G_{\text{H2}})$$

$$\mathrm{\Delta }{G}_{\ast \text{OOH}}=G_{\ast \text{OOH}}-G_{\ast }-(\mathit{2}{G}_{\mathrm{H}2\mathrm{O}}-3/2G_{\text{H2}})$$

where G_*, G_*OH, G_*O, and G_*OOH are the free energies of *, *OH, *O, and *OOH, respectively. G_H2O and G_H2 are the free energies of H₂O and H₂ molecules, respectively. The computational hydrogen electrode (CHE) model proposed by Nørskov et al. (74) has been applied to calculate the Gibbs free energy (ΔG) change for each elementary step of the oxygen reduction reaction (ORR). The reaction free energy of each elementary step was calculated as follows

$$\mathrm{\Delta }G=\mathrm{\Delta }E+\text{ΔZPE}-T\mathrm{\Delta }S+\mathrm{\Delta }U$$

where ΔE, ΔE_ZPE, and TΔS are the changes in reaction energy, zero-point energy, and entropy, respectively. T is the temperature at 298.15 K. Zero-point energy and entropy can be calculated from the vibrational frequency calculation; the frequencies of gas-phase molecules are obtained from the National Institute of Standards and Technology (NIST) database. ΔU = −eU, where n_e denotes the number of electrons and U is the electrode potential.

Supplementary Materials

This PDF file includes:

Figs. S1 to S30

Tables S1 to S6

Supplementary Notes