Oxygen vacancy-induced efficient hydrogen spillover in Ni₁₇W₃/WO_3−x/MoO_3−x for a superior pH-universal hydrogen evolution reaction

Abstract

Searching for a stable and efficient electrocatalyst for the hydrogen evolution reaction is still challenging, especially under a wider pH operation condition. In this study, a multicomponent Ni₁₇W₃/MoO_3−x/WO_3−x catalyst was designed and synthesized, in which the unique hierarchical structure of entangled nanorods confined in a polyhedral framework ensures the maximum utilization of active sites. Significantly, electrochemical performance can be regulated by adjusting the oxygen vacancy concentration of the metal support. Combined with various characterization techniques, we discovered that abundant oxygen vacancies in the MoO_3−x/WO_3−x support not only significantly enhanced the hydrogen insertion/extraction kinetics in the metal oxide but also increased the hydration capacity, resulting in an efficient hydrogen adsorption/transfer/desorption kinetics on the Ni₁₇W₃/MoO_3−x/WO_3−x surface and interface. As a result, the fabricated electrocatalyst exhibits an ultralow overpotential of 16, 42, and 14 mV at 10 mA cm⁻² in alkaline, neutral, and acid electrolytes, respectively. Our work proves the important role of metal oxide supports in the hydrogen spillover process.

---

Introduction

Water electrolysis systems driven via renewable energy sources to produce hydrogen offer a potential solution to the dual challenges of energy crisis and environmental pollution.^1–5 For the hydrogen evolution reaction (HER), in the half-reaction of water electrolysis, platinum is the best-performing catalyst owing to its optimized free energy of hydrogen bonding (ΔG_H).^6–9 However, its high cost makes it difficult to achieve large-scale applications. Furthermore, the catalyst inevitably experiences localized proton concentration changes during the reaction process; therefore, a well-qualified catalyst should be able to adapt to a wide pH range of operating environments to make device operation more stable and reliable.

The construction of metal/metal oxides has been proven to be an efficient strategy since a metal oxide support modulates an active metal phase's hydrogen intermediate (H*) adsorption/desorption behavior through strong metal–support interactions (SMSIs).^10–12 Among various metal oxide support candidates, WO_x has attracted widespread attention because of its excellent electronic conductivity and outstanding synergistic effect with active metals.¹³ Meanwhile, the hydrogen spillover phenomenon is observed on the metal/WO_x interface in two steps. First, H* transfers from the metal (ΔG_H-metal < 0) to the WO_x surface (ΔG_H-support > 0),¹⁴ causing the active metal surface to rapidly re-expose. Then, H* inserts/extracts into the WO_x lattice to provide additional active sites for the hydrogenation of WO_x. Previous studies have primarily focused on reducing the hydrogen spillover transfer barrier through active metal modification (EG–Pt/CoP and Pt SA/WO_x).^15,16 To the best of our knowledge, few studies have investigated the effect of support modification strategy on the HER performance.

In this work, we designed and synthesized Ni₁₇W₃/MoO_3−x/WO_3−x multicomponent catalysts via an in situ confinement method, in which Ni₁₇W₃ functions as the active metal phase and MoO_3−x/WO_3−x serves as the support. Moreover, the fabricated catalyst's oxygen vacancy concentration can be adjusted simply by changing the mole ratio between MoO₃ and WO₃. Next, by tracking phase evolution and mass/charge transfer kinetics with in situ Raman and in situ electrochemical impedance spectroscopy techniques, the superior performance of the catalyst results from the abundant oxygen vacancies of the metal oxide support, which accelerates phase transition from the metal oxide to hydrogen bronze, further resulting in a rapid H* adsorption/transfer/desorption kinetics on the Ni₁₇W₃/MoO_3−x/WO_3−x surface and interface. As a result, the fabricated Ni₁₇W₃/MoO_3−x/WO_3−x catalyst shows superior HER performance at all pH values, requiring only 14, 42, and 16 mV overpotential to reach a geometric current density of 10 mA cm⁻² in acid, neutral, and alkaline electrolytes, respectively. Our research confirms the practical potential of Ni₁₇W₃/MoO_3−x/WO_3−x and provides new design concepts for metal/metal oxide catalysts.

Experimental

Preparation of NW₁M₁R

Carbon cloth (CC) with an area of 1 × 1 cm² was first immersed in nitric acid for surface treatment. Then, the treated carbon cloth was ultrasonically cleaned with ethanol and DI consecutively for 20 minutes to remove surface impurities before synthesis. 0.68 mmol Ni(NO₃)₂·6H₂O, 0.0625 mmol (NH₄)₁₀H₂(W₂O₇)₆·xH₂O and 0.15 mmol (NH₄)₆Mo₇O₂₄·4H₂O were dispersed in 16 mL DI water and then heated to 60 °C for 1 h to obtain a uniform solution. Next, the solution was poured into a 25 mL Teflon-lined stainless autoclave with the CC immersed in it, followed by heating at 150 °C for 18 h. After hydrothermal treatment, Mo–Ni₄W₆O₂₁ was washed with water and methanol and dried at 60 °C for 6 h. Finally, it was annealed under an Ar/H₂ (5 v% H₂) atmosphere in a tube furnace at 500 °C for 2 h. The obtained sample was labelled as NW₁M₁R. The product without the Mo-doped process was named NWP, while the product without the W-doped process was named NMR.

Physical characterization

The microstructures of the catalysts, as well as the corresponding elemental distributions, were derived by SEM, TEM, HRTEM, and EDS mapping on FESEM, Quanta FEG 450 and FEI, TECNAI F20. The phase structure of each catalyst was characterized by X-ray diffraction at a scan rate of 5° min⁻¹ and a 2θ range of 10–80°. The chemical composition and electronic structure of the catalysts were analyzed by X-ray photoelectron spectroscopy. Oxygen vacancy concentration was determined by electron paramagnetic resonance (EPR, EMXnano, Bruker, USA).

Electrochemical measurement

The CV and LSV polarization curves were tested using an electrochemical workstation (GAMRY) with a standard three-electrode system. The prepared catalyst was used as the working electrode, Ag/AgCl as the reference electrode, and the carbon rod as the counter electrode. Moreover, the polarization curves were corrected by iR compensation (85%). Each catalyst's bilayer capacitance (C_dl) was tested by CV polarization curves in 1.0 M KOH electrolyte at different scan rates (20–100 mV s⁻¹) in the voltage range of 0.173–0.273 V vs. RHE. In situ electrochemical impedance spectra were obtained in the 0 to −0.065 V voltage range vs. RHE. All HER stability tests were performed without iR compensation.

In situ electrochemical Raman measurement

In situ Raman tests were performed on a LabRAM HR 800 using a custom-made Raman electrolytic cell with NW₁M₁R and NWP as the working electrodes, Ag/AgCl as the reference electrode, and Pt as the counter electrode, respectively. In this case, the individual potentials (−0.05 V to −0.60 V) were controlled by an electrochemical workstation, and the Raman signals were collected at 70 s intervals at each potential. All tests were performed without iR compensation.

Anion exchange membrane water electrolysis (AEMWE) performance

The AEMWE performance was evaluated by testing the CV polarization curves in 1.0 M KOH electrolyte using NW₁M₁R as the cathode and platinum-modified NiFe LDH as the anode over a voltage range of 1.0 to 2.0 V (5 mV s⁻¹). The stability test was conducted using chronopotentiometry at 1 A cm⁻² for 90 hours at room temperature. The performance and stability tests were performed with no iR compensation.

Theoretical calculation

The adsorption energy calculated by spin-polarized density-functional theory (DFT) was computed using the Vienna Ab initio Simulation Package (VASP) and the Perdew–Burke–Ernzerhof functional within generalized gradient approximation. During the calculation process, the electronic iterations convergence was 10⁻⁵ eV, and Hellmann–Feynman forces were 0.02 eV Å⁻¹. Surface isolation was performed using a 10 Å vacuum plate. The k-points mesh was set into 4 × 3 × 1 for structural optimization, and the corresponding vacuum layer was 20 Å.

Results and discussion

The Ni₁₇W₃/MoO_3−x/WO_3−x ternary heterostructure was synthesized via the in situ confinement method with the Ni₄W₆O₂₁ polyhedron as the soft template. As shown in Fig. 1a, the precursors were first prepared via a hydrothermal method, followed by heat treatment in a mixed Ar/H₂ atmosphere to obtain the final product (the detailed preparation process is shown in the Experimental section). It should be noted that MoO₃ was selected to modify the support because MoO₃ and WO₃ are similar; both can transfer to bronzes upon H insertion.¹⁷ Interestingly, when the precursor contains only Ni(NO₃)₂ and (NH₄)₁₀H₂(W₂O₇)₆, typical micrometer-sized particles with polyhedral morphology were fabricated (denoted as NWP). After introducing (NH₄)₆Mo₇O₂₄, twisted and intertwined nanorods were confined for in situ nucleation growth in the polyhedral framework (denoted as NW₁M_0.5R, NW₁M₁R and NW₁M_1.5R when the molar ratio W and Mo in the precursor is 1 : 0.5, 1 : 1 and 1 : 1.5). The scanning electron microscopy (SEM) images exhibit a precise morphology observation for the electrocatalyst. Regular particles with a size of ≈3 μm were obtained when only Ni and W elements were in the precursor, and tiny nanoparticles were randomly attached to the microparticle surface (Fig. 1b, c and S1†). Compared with Ni₄W₆O₂₁ precursors that have not been annealed at high temperatures, the surface is smooth, and no nanoparticles are formed, indicating that the formation of these nanoparticles is due to the phase separation (Fig. S2†). Then, after Mo introduction, the NWMR catalyst maintains a polyhedral shape but is assembled from tangled nanorods (Fig. 1d, e and S3†). However, the NMR catalyst without adding W shows nanorod morphology (Fig. S4†). Based on the transmission electron microscopy (TEM) image, the diameter of the nanorods is about 45 nm (Fig. 1f). The above characterization results suggest that the introduction of Mo successfully induced a morphological transition, with a significant increase in the surface area from 33.524 m² g⁻¹ to 53.52 m² g⁻¹ (Fig. S5†), which is conducive to the total exposure of the active stable sites in the electrolyte.

Fig. 1 (a) Schematic illustration of the synthesis of NWP and NW₁M₁R. (b and c) SEM images of NWP. (d and e) SEM images of NW₁M₁R. (f) TEM and (g) HRTEM images of NW₁M₁R. (h) EDS mapping of NW₁M₁R for Mo, W, and Ni elements.

Further detailed structural information is explored by high-resolution TEM (HRTEM). The characteristic lattice fringes of 0.211, 0.260, and 0.386 nm correspond well to the (111) plane of Ni₁₇W₃, (220) plane of WO₃ and (011) plane of MoO₃, respectively (Fig. 1g). In addition, the characteristic lattice fringes of 0.210, 0.260, and 0.340 nm in NWP catalysts correspond to the (111) plane of Ni₁₇W₃, (220) plane of WO_3, and (−110) plane of WO₂, respectively (Fig. S6†).

Combined with the crystal and phase information obtained through the X-ray diffraction (XRD) patterns (Fig. S7–S9†), the characteristic peaks of NWP located at 23.48°, 33.60°, 41.40° and 48.04° correspond to the (020), (202), (222) and (040) facets of WO₃ (JCPDS 46-1096), and the peaks at 25.79° and 37.07° are indexed to the (011) and (−211) planes of WO₂ (JCPDS 05-0431).¹⁸ Also, three peaks at 43.70°, 50.88°, and 74.82° correspond to the diffraction of (111), (200) and (220) planes of the Ni₁₇W₃ phase, respectively (JCPDS 65-4828).¹⁹ Thus, the crystal phase composition for NWP is Ni₁₇W₃/WO₂/WO₃. On the other hand, for the NW₁M_0.5R, NW₁M₁R, and NW₁M_1.5R samples, the characteristic peaks for WO₂ disappear totally, and new peaks located at 23.05°, 25.30°, 36.09°, and 47.28° appear, which can be assigned to MoO₃ (JCPDS 89-1554).²⁰ Thus, the crystal phase composition for these three samples is Ni₁₇W₃/MoO₃/WO₃. The characteristic NMR peak at 50.9° corresponds to the (310) plane of Ni₄Mo (JCPDS 03-1036). The energy dispersive X-ray (EDX) elemental mapping of NW₁M₁R (Fig. 1h) shows that Mo, Ni, and W are present and evenly distributed throughout the entire nanorod, indicating the successful preparation of Ni₁₇W₃/MoO₃/WO₃ ternary heterostructures.

The surface chemical composition and electronic structure of the obtained electrocatalysts are then assessed by X-ray photoelectron spectroscopy (XPS). Obviously, the obtained spectrum confirms the existence of Ni, Mo, W, and O in NW₁M_0.5R, NW₁M₁R, NW₁M_1.5R (Fig. S10†). The high-resolution Ni 2p spectrum in Fig. S11† shows two peaks at ∼852.8 and ∼869.2 eV attributed to metallic state Ni⁰, while the other four peaks are corresponding to Ni²⁺ (∼856.1 and ∼873.7 eV) and Ni³⁺ (∼857.6 and ∼875 eV) and their corresponding satellite peaks (∼861.7 and ∼880.1 eV).²¹ The high-resolution W 4f spectrum (Fig. S12†) of NW₁M₁R displays three doublets, which can be attributed to W⁰ (∼31.4 and ∼33.6 eV), W⁴⁺ (∼32.8 and ∼33.9 eV) and W⁶⁺ (∼35.5 and ∼37.5 eV), respectively.^22,23 In addition, the Mo 3d spectrum (Fig. S13†) exhibits three sets of peaks, corresponding to Mo⁴⁺ (∼229.4 and ∼231.7 eV), Mo⁵⁺ (∼230.6 and ∼234.3 eV) and Mo⁶⁺ (∼232.7 and ∼235.5 eV), respectively.^24,25 It is also noted that after Mo introduction, the Ni 2p and W 4f binding energies of NW₁M₁R are the largest compared with the NWP, blue-shifted by 0.3 and 0.6 eV. For the Mo 3d spectrum, no obvious binding energy shift is observed.^26,27 Moreover, the O 1s XPS spectrum shows three typical peaks centered at 530.4, 531.5, and 532.6 eV, corresponding to the lattice oxygen, oxygen vacancy, and absorbed oxygen, and NW₁M₁R shows the highest oxygen vacancy concentrations (Fig. S14†).^28,29 To further confirm the presence of oxygen vacancies, the prepared catalysts were subjected to electron paramagnetic resonance (EPR) spectroscopy. NW₁M_0.5R, NW₁M₁R and NW₁M_1.5R all show EPR signals at g = 2.002, while NWP shows a signal at g = 2.003.^30,31 The EPR results further support that NW₁M₁R exhibits a stronger EPR signal than NW₁M_0.5R, NW₁M_1.5R and NWP (Fig. S15†). When the W and Mo atomic ratio is 1 : 1, the sample exhibits the highest oxygen vacancy concentration, which can be attributed to the lattice mismatch between cubic phase α-WO₃ and monoclinic phase β-MoO₃. The above electronic structure analysis results indicate that the optimized metal oxide ratio can not only adjust the oxygen vacancy concentration but also enhance the interface contact among these material phases.³²

To shed light on the excellent electrochemical performance of the fabricated electrocatalyst, we first investigated the HER activity of each catalyst in 1.0 M KOH via linear scanning voltammetry (LSV). As shown in Fig. 2a and S14,† among all the tested samples, NW₁M₁R establishes the best HER catalytic performance, needing a lowest overpotential of only 16 mV to achieve the current density of 10 mA cm⁻², which is significantly superior than those of NWP (72 mV), NW₁M_1.5R (44 mV), NW₁M_0.5R (54 mV), NMR (51 mV) and Pt/C (29 mV), respectively. Then, the corresponding Tafel slope of NW₁M₁R is fitted to 34.9 mV dec⁻¹, superior to those of NWP (72.4 mV dec⁻¹), NW₁M_1.5R (54.8 mV dec⁻¹), NW₁M_0.5R (62 mV dec⁻¹), NMR (54.1 mV dec⁻¹) and commercial Pt/C (49.2 mV dec⁻¹) (Fig. 2b and S16†), indicating a fast HER kinetics.³³ Specifically, the Tafel slope value of NW₁M₁R is close to 30 mV dec⁻¹, suggesting that the HER process follows the Volmer–Tafel mechanism, where hydrogen is formed by the combination of neighboring adsorbed H* rather than by the combination of H* and water molecules in the electrolyte. In addition, electrochemical impedance spectroscopy (EIS) was further employed to analyze the HER kinetics at the interface between the electrode and electrolyte (Fig. 2d). The Nyquist plots show that NW₁M₁R has the smallest radius, while the equivalent circuit diagram (Fig. S17†) yields that NW₁M₁R also has the lowest charge transfer resistance (R_ct = 2.8 Ω). Notably, compared to the recently reported HER catalysts, the excellent HER performance of NW₁M₁R is almost the best among them (Fig. 2c and Table S1†).

Fig. 2 (a) LSV curves of NW₁M₁R, NW₁M_1.5R, NW₁M_0.5R, NWP, and Pt/C (inset: overpotentials at 10 mA cm⁻² for these electrocatalysts) in 1.0 M KOH. (b) Tafel slope of different electrodes. (c) HER performance comparison of recently reported electrocatalysts. (d) Nyquist plots of NW₁M₁R, NW₁M_1.5R, NW₁M_0.5R and NWP. (e) Multistep chronopotentiometry curves of NW₁M₁R at 50, 100, 150, and 200 mA cm⁻² for 160 h (without iR correction). (f) Polarization curves of the NiFe LDH (+)‖NW₁M₁R (−) and NiFe LDH (+)‖Pt/C (−) in the AEM flow cell. (g) Chronopotentiometry stability tests of NiFe LDH (+)‖NW₁M₁R (−) at 1 A cm⁻² and room temperature.

After that, double-layered capacitances (C_dl) are measured by cyclic voltammetry (CV) at different sweep rates from 20 to 100 mV s⁻¹ to assess the electrochemical active surface area (ECSA) of each catalyst (Fig. S18 and S19†). It is observed that the C_dl value of NW₁M₁R is 22.24 mF cm⁻², which is 4.82 times higher than that of NWP (4.61 mF cm⁻²). When the current densities are normalized with respect to ECSA, NW₁M₁R still exhibits a much larger current density than other samples during the entire voltage range, indicating that NW₁M₁R is indeed intrinsically more active than NWP, NW₁M_1.5R, and NW₁M_0.5R (Fig. S20†). In addition to the high catalytic performance demonstrated, the Faraday efficiency of the NW₁M₁R was determined to be 99.37%, nearly 100% (Fig. S21†). The long-term HER performance of NW₁M₁R at different current densities and 40 hours for each current density (160 hours in total) and at a high current density of 500 mA cm⁻² (100 hours) shows that the voltage has no apparent upward trend (Fig. 2e and S22†), indicating that this unique hierarchical structure improves the stability by preventing aggregation, which is usually one of the main factors leading to catalyst deactivation. The almost unchanged NW₁M₁R nanostructure, even after 160 h of stability testing, further supports the superiority of the hierarchical structure during the HER process (Fig. S23 and S24†). After long-term stability, the sample was also subjected to XPS characterization (Fig. S25–S27†). The comparison of the pre- and post-reaction states showed that Ni, Mo, and W promote the HER process in the form of high valence states. In addition, because of the superior HER performance under alkaline conditions, an AEM flow cell is assembled with the NW₁M₁R electrocatalyst, which serves as a cathode, and the homemade NiFe LDH serves as the anode (Fig. S28†). The polarization curves of the AEM are shown in Fig. 2f; commercial Pt/C is added for comparison. Notably, the NiFe LDH (+)‖NW₁M₁R (−) couple achieves a current density as high as 1 A cm⁻² at 2 V, which is superior to that of the commercial NiFe LDH (+)‖Pt/C (−) and most of the reported electrocatalyst (Fig. S29†). More importantly, the stability of the assembled AEM flow cell is investigated at a constant current density of 1 A cm⁻², which exhibits excellent durability for 90 hours, indicating the NW₁M₁R electrode to be a promising HER electrocatalyst in the AEM water electrolyzer (Fig. 2g).

To reveal the effect of each component on the performance enhancement, we first employed density functional theory (DFT) to gain insights into the H* absorption/desorption trend. As shown in Fig. S30,† the ΔG_H of Ni₁₇W₃, WO₃, and MoO₃ is −0.60, 1.57, and 0.86 eV, respectively. Thus, the NW₁M₁R constructed Ni₁₇W₃ with a negative value and WO₃/MoO₃ with a positive value, indicating the potential hydrogen spillover process between the metal and metal oxide interface.^34,35 Then, the in situ Raman technique was utilized to monitor the phase evolution during the HER process. For the NWP sample, it is clear that the full width at half maximum (FWHM) of the O–W–O vibrational peak located at 325 cm⁻¹ increases as the reaction proceeds, indicating a decrease in the crystallinity of WO_x (Fig. 3a).³⁶ Similarly, the characteristic peaks of W–O–W located at 820 and 843 cm⁻¹ decreased when the negative voltage was applied.³⁷ The spontaneous disappearance in an alkaline environment contributes to surface reconstruction. Importantly, a new Raman signal appeared at 1055 cm⁻¹ when the voltage reaches −0.35 V,³⁸ which belongs to the metal oxide hydration vibration. In this way, metal oxide combines with water to form metal hydroxides, which further transform into hydrogen bronze with the assistance of electrons (eqn (1) and (2)). It should be noted that the hydrogen atoms in hydrogen bronze will dissociate into highly mobile protons and electrons. The protons are localized near the O²⁻ in the lattice, leading to hydroxide character, whereas the electrons localize on the metal d-orbitals, reducing the M⁶⁺ to M⁵⁺. More importantly, the saturated concentration of protons will result in H₂ generation on the metal oxide surface (eqn (3)).

M^VIO₃ + H₂O ↔ M(OH)_x (1)

M(OH)_x + H₂O + e⁻ ↔ H_yM^VM^VIO_3−z (2)

H_yM^VM^VIO_3−z ↔ y/2H₂↑ + M^VIO₃ + e⁻ (3)

Fig. 3 (a and b) In situ Raman spectroscopy of NW₁M₁R and NWP. (c and d) In situ Bode phase plots of NW₁M₁R and NWP. (e) Plots of T₂vs. η of NW₁M₁R and NWP. (f) Plots of log(R₁) vs. η of NW₁M₁R and NWP. Schematic illustration of the HER processes occurring on (g) NW₁M₁R and (h) NWP.

For the NW₁M₁P sample, the characteristic peak of Mo–O stretching vibration appears at 887 cm⁻¹ (Fig. 3b).³⁷ Meanwhile, the location of the characteristic peak of W=O shows a slight redshift, indicating that MoO₃ induces a change in the lattice stress–strain behavior.³⁹ Unlike the NWP sample, the W–O–W signal located at 820 and 843 cm⁻¹ consistently has distinct characteristic peaks in the low voltage range (<−0.3 V), suggesting a stronger interaction between MoO₃ and WO₃, which is in agreement with the XPS results. Interestingly, the characteristic peaks of metal oxide hydration appear at a low voltage of −0.2 V, completing the phase transition from the metal oxide to the real active phase hydrogen bronze more quickly. The lower activation potential of the NW₁M₁R electrocatalyst is supposed to be due to the fact that the high oxygen vacancies concentration is beneficial for the hydrogen insertion/extraction process.

In addition, due to the different HER kinetics based on the Tafel slope analysis, the in situ EIS technique was used to characterize the detailed mass/charge transfer kinetic behavior. Both samples were subjected to EIS analysis at different voltages (Fig. 3c and d). Bode plots can be divided into two regions, high-frequency and low-frequency, and the signals in each of these regions can be used to analyze the conductivity of the catalysts at the inner layer and the interface as well as the charge transfer between the catalysts and the electrolyte.⁴⁰ It is clear that the kinetics of NW₁M₁R and NWP are significantly different. NW₁M₁R has characteristic peaks only at low frequencies, indicating that the kinetic process is restricted to the charge/mass transfer process at the electrode/electrolyte interface.⁴¹ Meanwhile, the almost negligible inner-layer/interface resistance can be attributed to the unique morphology and the SMSI effect. However, for NWP, the lower phase angle in the low-frequency area and the higher phase angle in the high-frequency area demonstrated that the kinetic process is mainly controlled by the inner layer/interface charge transfer resistance.³³ Then, the double-parallel equivalent circuit model is established based on the recorded Nyquist plots (Fig. S31 and S32†). The time constants of CPE₁ and R₁ can represent the hydrogen adsorption behavior, while CPE₂ and R₂ are related to the mass/charge transfer kinetics (Tables S2 and S3†).¹⁴ Firstly, the lower resistance (R₁) of NW₁M₁R across the entire voltage range suggests efficient mass/charge transfer kinetics, in line with Tafel slope analysis. Then, the detailed H* adsorption/desorption behavior was further investigated. The H* adsorption charge can be represented by integrating the CPE₂-T with the overpotential (Fig. 3e). The larger area of NW₁M₁R than that of NWP indicates that NW₁M₁R adsorbs a larger amount of H*, and the higher H* coverage on the surface is conducive to the Volmer–Tafel process. In addition, the H* adsorption behavior of these two samples exhibits the potential-dependent increase as soon as the cathodic potential is applied, indicating that the hydrogen spillover phenomenon occurs at the interface; only the transfer efficiency from metal to support is different. Then, the desorption process of H* is investigated by analyzing the EIS-derived Tafel slope (Fig. 3f). The smaller slope proves that NW₁M₁R has a larger H* desorption charge, demonstrating that NW₁M₁R simultaneously has faster hydrogen desorption kinetics. In conclusion, compared to NWP, the kinetics of NW₁M₁R are limited only by the H* charging/mass transfer process and exhibit superior H* adsorption/desorption kinetics.

Ni₁₇W₃ + H₂O + e⁻ ↔ Ni₁₇W₃–H* + OH⁻ (4)

Ni₁₇W₃–H* + Ni₁₇W₃–H* + e⁻ ↔ Ni₁₇W₃ + H₂ (5)

Ni₁₇W₃–H* + MO₃ ↔ Ni₁₇W₃ + H_yM^VM^VIO_3−z (6)

Ni₁₇W₃–H* + H₂O + e⁻ ↔ Ni₁₇W₃ + H₂↑ + OH⁻ (7)

Based on the analysis results, the HER process is detailed below. For the NM₁W₁R sample, the obvious hydrogen spillover occurs at the metal/metal oxide interface. First, the adjacent H* combine with each other on the metal surface to produce H₂ (eqn (4) and (5)). Then, due to the abundant oxygen vacancies of metal oxide support, the H* insertion/extraction proceeds easily in the metal oxide, resulting in a faster H* transfer from the metal to the metal oxide, leading to the metal oxide support surface, also producing significant amounts of hydrogen (eqn (3) and (6)). After increasing cathodic potential, the hydration capacity of metal oxides increased significantly, further promoting the phase transition process from metal oxides to hydrogen bronzes (eqn (2) and (3)). Thus, in turn, the whole H* adsorption, transfer, and desorption kinetic increased on the catalyst surface and interface, resulting in a superior HER performance. However, in the case of NMP, the lower H* coverage on the surface suggests that the transfer of H* from the metal to the metal oxide support is inefficient. The H₂ is produced primarily from the combination of H* on the surface and H₂O in the electrolyte, resulting in sluggish H* adsorption/desorption kinetics (eqn (4) and (7)). Meanwhile, the noticeable hydration of the metal oxide support requires a larger potential drive, and the kinetic behaviors of both hydrogen spillover and hydrogenation are relatively slow, resulting in unsatisfactory catalytic activity (Fig. 3g and h).

In order to verify whether the hydrogen spillover and metal oxide hydrogenation work at a wider pH range, the electrochemical performance of NM₁W₁R is investigated in 0.5 M H₂SO₄ (pH = 0.21) and 1.0 M PBS (pH = 7.2) electrolyte, respectively. Surprisingly, the same trend was observed under acidic and neutral conditions as under alkaline conditions. As shown in Fig. 4a and b, NW₁M₁R has the best HER performance (14 mV in 0.5 M H₂SO₄ and 42 mV in 1.0 M PBS) compared to NWP (61 mV in 0.5 M H₂SO₄ and 86 mV in 1.0 M PBS), NW₁M_1.5R (44 mV in 0.5 M H₂SO₄ and 50 mV in 1.0 M PBS), NW₁M_0.5R (42 mV in 0.5 M H₂SO₄ and 59 mV in 1.0 M PBS) and Pt/C (39 mV in 0.5 M H₂SO₄ and 46 mV in 1.0 M PBS). The Tafel slope corresponding to NW₁M₁R (32.6 mV dec⁻¹ in 0.5 M H₂SO₄ and 73.9 mV dec⁻¹ in 1.0 M PBS) is similarly lower than that of NWP (66.6 mV dec⁻¹ in 0.5 M H₂SO₄ and 76.3 mV dec⁻¹ in 1.0 M PBS), NW₁M_1.5R (41.9 dec⁻¹ in 0.5 M H₂SO₄ and 77.7 mV dec⁻¹ in 1.0 M PBS), NW₁M_0.5R (51.1 mV dec⁻¹ in 0.5 M H₂SO₄ and 77.3 mV dec⁻¹ in 1.0 M PBS) and Pt/C (37.03 mV dec⁻¹ in 0.5 M H₂SO₄ and 81.1 mV dec⁻¹ in 1.0 M PBS) (Fig. 4c and d). The long-term HER performance of NW₁M₁R at different current densities and in different electrolytes (300 h in 0.5 M H₂SO₄ and 140 h in 1.0 M PBS) shows no significant upward trend in voltage (Fig. 4d and e). These results demonstrate that NW₁M₁R could serve as a superior pH-universal electrocatalyst for HER.

Fig. 4 (a) LSV curves of NW₁M₁R, NW₁M_1.5R, NW₁M_0.5R, NWP, and Pt/C (inset: overpotentials at 10 mA cm⁻² for these electrocatalysts) in 0.5 M H₂SO₄. (b) LSV curves of NW₁M₁R, NW₁M_1.5R, NW₁M_0.5R, NWP, and Pt/C (inset: overpotentials at 10 mA cm⁻² for these electrocatalysts) in 1.0 M PBS. (c) Tafel slope of different electrodes in 0.5 M H₂SO₄. (d) Tafel slope of different electrodes in 1.0 M PBS. (e) Multistep chronopotentiometry curves of NW₁M₁R at 50, 100, and 150 mA cm⁻² for 300 h in 0.5 M H₂SO₄ (without iR correction). (f) Multistep chronopotentiometry curves of NW₁M₁R at 50, 100, 150, and 200 mA cm⁻² for 140 h in 1.0 M PBS (without iR correction).

Conclusions

In summary, we designed and prepared the Ni₁₇W₃/MoO_3−x/WO_3−x catalyst for HER reaction at all pH values. This catalyst prepared via the in situ confinement method provided a large specific surface area, ensuring the total exposure of active sites to the electrolyte. Importantly, the electrochemical performance can be regulated by adjusting the metal oxide support's oxygen vacancy concentration. Combined with DFT, in situ Raman, and in situ EIS results, we prove that the abundant oxygen vacancies in the WO_3−x/MoO_3−x support is beneficial for the improvement of hydrogen intermediate insertion/extraction kinetic and hydration capacity, resulting in a faster hydrogen intermediate absorption/transfer/desorption kinetics on the catalyst surface and interface, thus achieving a superior HER performance. As a result, the prepared catalysts exhibited overpotentials of only 16, 42, and 14 mV at current densities of 10 mA cm⁻² in the alkaline, neutral, and acid conditions, respectively. Our work not only shows the potential practical application of the fabricated electrocatalyst but also provides a new insight into the electrocatalyst design of metal/metal oxide.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is financially supported by the Shanghai Sailing Program (23YF1455000) and the City University of Hong Kong (Project 7020088 and 9229138).

References

1. D. Li, A. R. Motz, C. Bae, C. Fujimoto, G. Yang, F.-Y. Zhang, K. E. Ayers and Y. S. Kim, Energy Environ. Sci., 2021, 14, 3393–3419.
2. M. Chen, N. Kitiphatpiboon, C. Feng, A. Abudula, Y. Ma and G. Guan, eScience, 2023, 3, 100111–100127.
3. L. Dong, G.-R. Chang, Y. Feng, X.-Z. Yao and X.-Y. Yu, Rare Met., 2022, 41, 1583–1594.
4. Y. S. Park, F. Liu, D. Diercks, D. Braaten, B. Liu and C. Duan, Appl. Catal., B, 2022, 318, 121824–121833.
5. Q. Qian, X. He, Z. Li, Y. Chen, Y. Feng, M. Cheng, H. Zhang, W. Wang, C. Xiao, G. Zhang and Y. Xie, Adv. Mater., 2023, 35, 2300935–2300948.
6. Y. Shi, Z. R. Ma, Y. Y. Xiao, Y. C. Yin, W. M. Huang, Z. C. Huang, Y. Z. Zheng, F. Y. Mu, R. Huang, G. Y. Shi, Y. Y. Sun, X. H. Xia and W. Chen, Nat. Commun., 2021, 12, 3021–3032.
7. Y. Li, W. Wang, M. Cheng, Y. Feng, X. Han, Q. Qian, Y. Zhu and G. Zhang, Adv. Mater., 2023, 35, 2206351–2206361.
8. F. Li, G. F. Han, H. J. Noh, J. P. Jeon, I. Ahmad, S. S. Chen, C. Yang, Y. F. Bu, Z. P. Fu, Y. L. Lu and J. B. Baek, Nat. Commun., 2019, 10, 4060–4067.
9. Y. Feng, X. He, M. Cheng, Y. Zhu, W. Wang, Y. Zhang, H. Zhang and G. Zhang, Small, 2023, 19, 2301986–2301995.
10. X. Wang, X. Yang, G. Pei, J. Yang, J. Liu, F. Zhao, F. Jin, W. Jiang, H. Ben and L. Zhang, Carbon Energy, 2023, e391.
11. Z. Z. Jiang, Z. B. Wang, Y. Y. Chu, D. M. Gu and G. P. Yin, Energy Environ. Sci., 2011, 4, 2558–2566.
12. G. N. Vayssilov, Y. Lykhach, A. Migani, T. Staudt, G. P. Petrova, N. Tsud, T. Skála, A. Bruix, F. Illas, K. C. Prince, V. Matolín, K. M. Neyman and J. Libuda, Nat. Mater., 2011, 10, 310–315.
13. W. Zhang, B. Huang, K. Wang, W. Yang, F. Lv, N. Li, Y. Chao, P. Zhou, Y. Yang, Y. Li, J. Zhou, W. Zhang, Y. Du, D. Su and S. Guo, Adv. Energy Mater., 2021, 11, 2003192–2003199.
14. J. Li, J. Hu, M. Zhang, W. Gou, S. Zhang, Z. Chen, Y. Qu and Y. Ma, Nat. Commun., 2021, 12, 3502–3514.
15. J. Li, H.-X. Liu, W. Gou, M. Zhang, Z. Xia, S. Zhang, C.-R. Chang, Y. Ma and Y. Qu, Energy Environ. Sci., 2019, 12, 2298–2304.
16. J. Park, S. Lee, H.-E. Kim, A. Cho, S. Kim, Y. Ye, J. W. Han, H. Lee, J. H. Jang and J. Lee, Angew. Chem., Int. Ed., 2019, 58, 16038–16042.
17. C. Jian, Q. Cai, W. Hong, J. Li and W. Liu, Int. J. Hydrogen Energy, 2017, 42, 17030–17037.
18. C. Han, D. Wang, Q. Li, Z. Xing and X. Yang, ACS Appl. Energy Mater., 2019, 2, 2409–2413.
19. J. Liu, G. Qian, T. Yu, J. Chen, C. Zhu, Y. Li, J. He, L. Luo and S. Yin, Chem. Eng. J., 2022, 431, 134247–134256.
20. H.-S. Kim, J. B. Cook, H. Lin, J. S. Ko, S. H. Tolbert, V. Ozolins and B. Dunn, Nat. Mater., 2017, 16, 454–463.
21. J. Zhang, T. Wang, P. Liu, Z. Liao, S. Liu, X. Zhuang, M. Chen, E. Zschech and X. Feng, Nat. Commun., 2017, 8, 15437–15445.
22. Y. Tang, L. Dong, H. B. Wu and X.-Y. Yu, J. Mater. Chem. A, 2021, 9, 1456–1462.
23. W. Xu, B. Wang, X. Ni, H. Liu, W. Wang, L. Zhang, H. Zhang, Z. Peng and Z. Liu, ACS Appl. Mater. Interfaces, 2021, 13, 13838–13847.
24. T. Wang, X. Cao and L. Jiao, eScience, 2021, 1, 69–74.
25. Q. Qi, D. Shao, Y. Zhou, Q. Wang and X.-Y. Yu, J. Mater. Chem. A, 2023, 11, 15663–15669.
26. O. Karslioglu, X. Song, H. Kuhlenbeck and H.-J. Freund, Top. Catal., 2013, 56, 1389–1403.
27. F. Nacimiento, M. Cabello, R. Alcantara, C. Perez-Vicente, P. Lavela and J. L. Tirado, J. Electrochem. Soc., 2018, 165, A2994–A2999.
28. Z. Wang, X. Mao, P. Chen, M. Xiao, S. A. Monny, S. Wang, M. Konarova, A. Du and L. Wang, Angew. Chem., Int. Ed., 2019, 58, 1030–1034.
29. L. Xu, Q. Jiang, Z. Xiao, X. Li, J. Huo, S. Wang and L. Dai, Angew. Chem., Int. Ed., 2016, 55, 5277–5281.
30. X. Hu, J. Wang, J. Wang, Y. Deng, H. Zhang, T. Xu and W. Wang, Appl. Catal., B, 2022, 318, 121879–121894.
31. Y. Yang, Z. Wang, P. Yan and X.-Y. Yu, Appl. Phys. Lett., 2023, 123, 033903–033910.
32. G. Yang, Y. Jiao, H. Yan, Y. Xie, C. Tian, A. Wu, Y. Wang and H. Fu, Nat. Commun., 2022, 13, 3125–3137.
33. C. Xie, W. Chen, S. Du, D. Yan, Y. Zhang, J. Chen, B. Liu and S. Wang, Nano Energy, 2020, 71, 104653–104661.
34. R. Prins, Chem. Rev., 2012, 112, 2714–2738.
35. W. Karim, C. Spreafico, A. Kleibert, J. Gobrecht, J. VandeVondele, Y. Ekinci and J. A. van Bokhoven, Nature, 2017, 541, 68–71.
36. G. Poirier, Y. Messaddeq, S. J. L. Ribeiro and M. Poulain, J. Solid State Chem., 2005, 178, 1533–1538.
37. E. I. Ross-Medgaarden and I. E. Wachs, J. Phys. Chem. C, 2007, 111, 15089–15099.
38. C. R. Santiago-Ramirez, J. Vera-Iturriaga, P. del Angel, A. Manzo-Robledo, M. L. Hernandez-Pichardo and J. Soto-Hernandez, Appl. Catal., B, 2021, 282, 119545–119558.
39. J. Dai, Y. Zhu, Y. Chen, X. Wen, M. Long, X. Wu, Z. Hu, D. Guan, X. Wang, C. Zhou, Q. Lin, Y. Sun, S.-C. Weng, H. Wang, W. Zhou and Z. Shao, Nat. Commun., 2022, 13, 1189–1199.
40. J. Huang, H. Sheng, R. D. Ross, J. Han, X. Wang, B. Song and S. Jin, Nat. Commun., 2021, 12, 3036–3047.
41. Y.-Y. Chen, Y. Zhang, X. Zhang, T. Tang, H. Luo, S. Niu, Z.-H. Dai, L.-J. Wan and J.-S. Hu, Adv. Mater., 2017, 29, 1703311–1703318.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta00729h

---