Unconventional direct synthesis of Ni₃N/Ni with N-vacancies for efficient and stable hydrogen evolution

Abstract

Transition metal nitrides are a fascinating class of catalyst materials due to their superior catalytic activity, low electrical resistance, good corrosion resistance and earth abundance; however, their conventional synthesis relies on high-temperature nitridation processes in hazardous environments. Here, we report a direct synthesis of Ni₃N/Ni enriched with N-vacancies using one-step magnetron sputtering. The surface state of Ni₃N(001) with 75% N-vacancies is hydrogen-terminated and exhibits four inequivalent Ni₃-hollow sites. This leads to stronger H* binding compared to Ni(111), and is affirmed as the most stable surface termination under the electrochemical working conditions (pH ≈ 13.8 and E = −0.1 V) from the Pourbaix diagram. The Ni₃N/Ni catalyst shows low crystallinity and good wettability and exhibits a low overpotential of 89 mV vs. RHE at 10 mA cm⁻² in 1.0 M KOH with excellent stability over 3 days. This performance closely matches that of the Pt catalyst synthesized under the same conditions and surpasses that of other reported earth-abundant catalysts on planar substrates. The application of Ni₃N/Ni as a cocatalyst on Si photocathodes produces an excellent ABPE of 9.3% and over 50 h stability. Moreover, its feasibility for practical application was confirmed with excellent performance on porous substrates and robustness at high operating currents in zero-gap alkaline electrolysis cells. Our work demonstrates a general approach for the feasible synthesis of other transition metal nitride catalysts for electrochemical and photoelectrochemical energy conversion applications.

---

Broader context

High purity hydrogen promises to be a zero-carbon fuel that can be widely adopted as a renewable energy carrier in the near future. The development of high performance catalysts that are non-precious, scalable, and robust is a crucial step for achieving a low levelized cost of hydrogen. Earth-abundant Ni exhibits hydrogen adsorption energies and exchange current densities close to that of Pt for the hydrogen evolution reaction. Among the derivatives of Ni, Ni-nitrides and their heterostructures are particularly promising due to their high binding capabilities for the adsorbates and relatively low electrical resistance. This work reports the direct synthesis of Ni₃N enriched with N-vacancies using one-step reactive magnetron sputtering from earth-abundant Ni, which differs from the high-temperature nitridation process in the presence of hazardous chemicals. The important role of the N-vacancies for achieving high activity and stability from the low surface free energy (γ) and positive Bader charge values, and surface Pourbaix diagram under working conditions is rationalized by density functional theory calculations. Moreover, the Ni₃N/Ni catalyst deposited on different (porous and planar) substrates produces remarkable performance and stability even at high current densities in zero-gap electrolysers, demonstrating its feasibility for practical application.

Introduction

Converting intermittent renewable energy to value-added chemicals (e.g., hydrogen, ethanol, and methanol) is a feasible strategy to address the issues of sustainable energy storage, transportation, and consumption.¹ In particular, research on the efficient production of green hydrogen through electrochemical water splitting using renewable electricity or photoelectrochemical (PEC) water splitting using sunlight has received tremendous attention in recent years. Central to these efforts is the development of high-performing earth-abundant catalysts with good operational stability for hydrogen evolution reactions (HERs) and oxygen evolution reactions (OERs) using scalable synthesis methods, which could allow the replacement of noble-metal catalysts such as Pt and Ir/Ru-based compounds in practical systems.^2,3

Ni-based compounds including oxides, sulphides, carbides, selenides and nitrides are highly attractive as HER catalysts due to their unique d-orbital electron configuration, earth-abundance and potential for low-cost synthesis.^4–8 Computations by Trasatti and Nørskov et al. have verified that the hydrogen adsorption energies and exchange current densities of metallic Ni are close to those of Pt for HERs.^9,10 Among the derivatives of Ni, Ni-nitrides and their heterostructures are particularly promising due to their high binding capabilities for adsorbates (atomic hydrogen, protons, or water molecules) and relatively low electrical resistance.^11–13 Nevertheless, nitrides are usually synthesized by self-propagating high-temperature synthesis with nitridation at >400 °C in the presence of hazardous chemicals (such as azides, hydrazine, cyanamide, and ammonia), as the high strength of the N–N bonds hinders their direct synthesis.^12,14,15

Reactive magnetron sputtering has been widely used as an industrial technique in preparing large-area uniform thin films with high adhesion to the underlying substrates.¹⁶ Monometallic and multimetallic electrocatalysts can be directly deposited onto target substrates with controlled loading amounts and desirable composition.^17,18 Similarly, cocatalyst layers can be deposited onto semiconductor photoelectrodes in a clean vacuum environment without the risk of corrosion of semiconductor materials as in solution-based synthesis. The structural properties of sputter-deposited films can be controlled by adjusting the process parameters with good reproducibility. Moreover, catalysts with structural disorders and defects are shown to exhibit superior catalytic performance compared to their highly crystalline counterparts due to their high intrinsic activity from the abundant supply of active sites.^3,19 For example, magnetron sputtering has been shown to generate oxygen vacancies due to the bombardment of energetic particles, and the concentration of such defects plays an important role in achieving high activity.²⁰ Despite these advantages, there are no reports on the preparation of sputtered Ni-nitrides for electrochemical or PEC water splitting.

Crystalline-Si has revolutionized the photovoltaic industry due to its suitably narrow bandgap (∼1.1 eV), high earth-abundance and availability of mature manufacturing technologies, and it has become a subject of intense research for PEC water splitting.²¹ Various photoelectrodes have been successfully fabricated achieving high saturation photocurrent densities of up to 40 mA cm⁻².^21–23 Cocatalyst on Si photoelectrodes play a crucial role in achieving high photovoltage as well as good operational stability. Given the chemical sensitivity of Si in alkaline solutions, compact metal-based cocatalyst could be employed both as catalyst and protection layers. These cocatalyst layers could effectively isolate the Si surface from corrosive electrolytes and avoid it being etched away by the solution. Therefore, it is of significance to develop sputtering deposition of cocatalyst for Si photocathodes, as it produces compact films, which could protect Si from corrosion and is performed via a vacuum process compatible with Si fabrication.

Herein, we report for the first time the one-step synthesis of Ni₃N/Ni with triangular pyramid nanostructures using magnetron sputtering for the electrochemical and PEC HER in alkaline solutions. In particular, the sputtered catalyst shows low crystallinity and N-vacancies with rich inequivalent N₃-hollow active sites, strong H* binding strength and superior electrochemical surface area. As a result, the Ni₃N/Ni electrode presents an HER activity comparable to that of Pt electrodes with an overpotential of 89 mV at 10 mA cm⁻². Density functional theory (DFT) calculations rationalize the important role of N-vacancies for this high activity and stability from the low surface free energy (γ) and positive Bader charge values, and surface Pourbaix diagram under working conditions. A Si photocathode with Ni₃N/Ni deposited as cocatalyst produces a current density of 38.8 mA cm⁻² at 0 V vs. reversible hydrogen electrode (RHE) with 0.59 V onset potential, an applied bias photon-to-current efficiency (ABPE) of 9.3%, and with excellent durability of over 50 h in strong alkaline media. In addition, the Ni₃N/Ni catalyst deposited on porous substrates produces remarkable performance and stability even at high current densities in zero-gap electrolyser cells demonstrating its feasibility for practical applications.

Results and discussion

Sputtering deposition of Ni₃N/Ni and Ni₃N/Ni/Si photocathode preparation

To verify the synthesis of Ni₃N by magnetron sputtering and further optimize its HER activity, we first deposited Ni₃N films on fluorine-doped tin oxide (FTO) glass substrates using a Ni sputter target and nitrogen plasma. Nitrogen plasma includes molecular, atomic, and ionic nitrogen (N¹⁺and N²⁺) species, which lead to the formation of N-vacancies.^24–26 Therefore, the formation of Ni₃N can be described using the following reactions:

Nitrogen dissociation: e⁻ + N₂ → N + N + e⁻ (1)

Nitrogen ionization: e⁻ + N₂ → N₂⁺ + 2e⁻ (2)

Nitrogen dissociative ionization: e⁻ + N₂ → N⁺ + N + 2e⁻ (3)

Ni₃N formation: xNi + yN₂ → Ni_xN_2y (4)

Considering that Ni₃N is an interstitial compound, nitrogen atoms occupy the octahedral interstitial sites in the Ni lattice in order to minimize the repulsive N–N interactions.^11,27 The nitrogen species from dissociation and ionization of nitrogen atoms permeate into the Ni lattice, resulting in the formation of Ni₃N with N-vacancies as illustrated in Fig. 1a. To evaluate the PEC performance of Ni₃N/Ni films as cocatalyst, Si photocathodes were fabricated with p⁺nn⁺ buried-junction structure, as illustrated in Fig. 1b. Surface texturing was introduced with micropyramids on the front side and microgrooves on the rear side to minimize reflection losses and enhance the catalytic active area, respectively. The front surface of the photocathodes was optimized for light-harvesting and hole collection by depositing a pinhole-free Al₂O₃ (∼10 nm) passivation layer, Si₃N₄ anti-reflection coating and Cr/Pd/Ag electrode grid. The rear surface was coated with the Ni₃N/Ni cocatalyst film for hydrogen evolution in direct contact with an alkaline electrolyte. More details of the fabrication process are provided in the experimental section in the ESI.†

Fig. 1 (a) Schematic illustration of sputter-deposition of Ni₃N films with N-vacancies. (b) Schematic illustration of the structure of Si photocathode consisting of a buried-junction Si coated with sputter-deposited Ni₃N/Ni cocatalyst.

Synthesis and characterization of Ni₃N/Ni films

Following a systematic study (described in the next section) of various Ni, Ni₃N and Ni₃N/Ni films on FTO glass substrates, the bilayer film consisting of ∼140 nm of Ni₃N and ∼70 nm of Ni deposited at 150 W direct current (DC) power presented the best HER activity and was chosen for in-depth material characterizations. Because the surface roughness of the FTO glass is relatively high, thin films were deposited on mirror polished Si wafers coated with thick SiO_x for the purpose of material characterizations. The individual Ni film shows a nanoparticle morphology with a feature size of ∼20 nm with visible grain boundaries and cracks (Fig. 2a). The Ni₃N film shows a triangular pyramid structure morphology (Fig. 2b), which is different from the Ni film and similar to that of the bilayer film (Fig. S1, ESI†). No obvious cracks appeared in the Ni₃N film, which suggests that the film is stress-free. The cross-sectional images (Fig. 2c and d) of the bilayer film verify that it has a hierarchical structure with ∼140 nm-thick Ni₃N and ∼70 nm-thick Ni.

Fig. 2 Scanning electron microscopy images of (a) Ni and (b) Ni₃N from top-view, and (c) Ni₃N/Ni from a cross-sectional view. (d) Scanning transmission electron microscopy image of Ni₃N/Ni layers with Ni₃N marked as block 1 and Ni marked as block 2. The Pt overlayer seen in (d) was added to protect the sample during the lamella preparation using a focus ion beam (FIB). (e and f) High-resolution transmission electron microscopy (HRTEM) images collected from blocks 1 and 2, respectively. (g) Transmission electron microscopy (TEM) image and the corresponding elemental mappings of Ni₃N/Ni. Pt observed in the images was deposited as a protective layer during the milling process for the preparation of cross-section samples.

The high-resolution transmission electron microscopy (HRTEM) image of block 1 (Fig. 2e) shows disordered lattice fringes with inter-planar spacings of 0.204 and 0.214 nm, which are assigned to the (111) and (002) crystal planes of hexagonal Ni₃N with an intersection angle of 62°, respectively.^28,29 The related selected area electron diffraction (SAED) pattern (Fig. S2a, ESI†) with broad halos and sporadically located diffraction dots indicate low crystallinity of Ni₃N. The HRTEM and SAED of block 2 in Fig. 2f and Fig. S2b (ESI†) confirm the presence of (111) and (200) crystal planes of cubic Ni with inter-planar spacings of 0.203 and 0.178 nm, respectively, with low crystallinity indicated by the fuzzy and random lattice fringes.³⁰ In addition, the elemental mapping images of Ni₃N/Ni (Fig. 2g) show that Ni is homogeneously distributed in both the layers, while N is uniformly located in the Ni₃N layer. The energy dispersive X-ray (EDX) spectra in Fig. S3 (ESI†) confirm the presence of Ni and N as the major constituents.

X-Ray photoelectron spectroscopy (XPS) was performed to examine the composition of the Ni₃N film. From Ni 2p XPS spectra (Fig. 3a), two peaks at 852.4 and 853.3 eV are observed, which are assigned to metallic Ni and Ni(I) species of Ni₃N, respectively.^31–34 The peak at 855.4 eV is attributed to Ni²⁺ from surface oxidation, and the peak at 873.1 eV is assigned to metallic Ni.³⁰ The peak at 869.5 eV belongs to 2p_1/2 of Ni⁺, which is attributed to the presence of valence Ni¹⁺ (Ni < ¹⁺).¹¹ The predominance of Ni < ¹⁺ in Ni₃N suggests that the electron density of a considerable fraction of Ni atoms is affected due to the existence of N-vacancies. The “shake up” satellites at the binding energies of 859.8 eV and 879.4 eV are also seen as significant Ni 2p peaks.¹⁸ Furthermore, the N 1s XPS spectrum in Fig. 3b presents peaks at 397.5 and 399.9 eV, which are ascribed to N species of Ni₃N and N-vacancies, respectively. The latter peak at 399.9 eV agrees with a similar observation made on oxide nanomaterials with oxygen vacancies.³⁵ In addition, detailed compositional analysis of XPS spectra reveals that the atomic ratio of Ni : N in Ni₃N is ∼3.5 : 1, which is slightly nonstoichiometric due to the presence of N-vacancies in Ni₃N.

Fig. 3 (a) XPS spectra of (a) Ni 2p and (b) N 1s of Ni₃N. (c) X-Ray diffraction patterns of Ni₃N/Ni on Si wafer and the inset shows the amplified XRD patterns of Ni and Ni₃N on Si wafer. (d) Water contact angle measurements for Ni/FTO and Ni₃N/Ni/FTO measured after resting water droplets on the surface for 4 s.

The X-ray diffraction (XRD) pattern for Ni₃N/Ni/Si in the 2θ scan range of (10–80°) shows no discernible peaks of Ni and Ni₃N (Fig. 3c). Upon magnification in the range 10–60°, the Ni₃N spectrum shows weak and broad diffraction peaks attributable to hexagonal Ni₃N (JCPDS: 10-0280) accompanied by some noisy peaks, indicating poor crystallinity or defective structure of the material prepared by magnetron sputtering. There also exist peaks corresponding to Si (ICSD#96-901-1057).³⁶ Meanwhile, Ni prepared under the same conditions shows a cubic structure with low crystallinity. These results agree well with HRTEM observations (Fig. 2e and f) and suggest that Ni₃N/Ni prepared by magnetron sputtering exhibits low crystallinity with disordered structure and N-vacancies.

We further measured the water contact angles on Ni/FTO and Ni₃N/FTO samples. The surface wettability of the electrocatalyst indicates surface adsorption of water molecules and the interaction of the catalyst with electrolyte, thereby playing an important role in the electrocatalytic activity. The much smaller contact angle on Ni₃N/FTO as illustrated in Fig. 3d suggests a higher wettability for Ni₃N/Ni/FTO due to hydrophilic Ni–N bonds on the surface, which could be beneficial for enhanced HER kinetics.³⁷

Electrochemical performance of Ni₃N/Ni films

The electrocatalytic performance of Ni₃N-based electrodes towards the HER was evaluated in a 1.0 M KOH electrolyte (pH = 14) saturated with N₂ gas. The sputtering power and the thickness of Ni₃N were first optimized based on the HER overpotential of the deposited films on FTO glass substrates at 10 mA cm⁻² current. As shown in Fig. S4 in the ESI,† Ni₃N/Ni synthesized at 150 W presents the best activity with the lowest overpotential. Similarly, Ni₃N presents the lowest overpotential when the sputtering duration was fixed at 1000 s (Fig. S5, ESI†), which suggests that a further increase in Ni₃N film thickness hinders its catalytic kinetics.

To understand the relationship between the synthesis conditions and N-vacancies, we varied the N₂: Ar gas ratio in the sputtering chamber while keeping the flow rate of Ar carrier gas constant. We also assessed their respective performances for the HER using linear sweep voltammetry (LSV) measurements as shown in Fig. 4a. The LSV data confirms that the Ni₃N sputtered under 50% N₂ : Ar gas ratio outperforms the other electrodes. As confirmed by the overpotentials at 10 mA cm⁻² current as shown in the inset in Fig. 4a, the HER performance worsened when the ratio is either increased or decreased. Fig. S6 (ESI†) shows XPS spectra of N 1s of Ni₃N catalysts synthesized under different gas ratios. When the N₂ : Ar ratio was 70%, stoichiometric Ni₃N was obtained with an atomic ratio of 3 : 1. With the decreasing gas ratio, the atomic ratio departed gradually from the stoichiometric ratio. When the ratio was below 10%, no N signal was detected, indicating the deposition of pure Ni with Ni-termination. However, N-vacancies first increased with the decreasing gas ratio until 40% as indicated by the area of the peak at 399.9 eV,¹¹ and then decreased with a further decrease in the gas ratio due to the appearance of metallic Ni. Accordingly, the Ni–N peak position shifted to a lower position first and then to a higher position due to the change in the position of the d-band centre of Ni₃N in the presence of N-vacancies.^38–40 The superior performance of Ni₃N deposited under 50% N₂ : Ar ratio is likely due to an optimal synthesis condition, which could, in turn, result in an optimal N-vacancy concentration as further explained in the later section using DFT calculations.

Fig. 4 (a) LSV curves (with 90% iR-compensation) of Ni₃N/FTO electrodes sputtered under different N₂:Ar flows. The flow of Ar carrier gas was fixed at 20 sccm. The inset summarizes the overpotential at 10 mA cm⁻² for different N₂:Ar flows. (b) LSV curves (with 90% iR-compensation) of Ni₃N/Ni, Ni₃N, and Pt on FTO substrates and the corresponding Nyquist plots (inset in Fig. 4b). (c) H₂ (red dots) evolved from the Ni₃N/Ni cathode detected by gas chromatography and the calculated amounts of H₂ (black dots). The measurement was performed at j_HER = 10 mA cm⁻². (d) Chronoamperometry measurement of the Ni₃N/Ni for the HER at an overpotential of 89 mV in 1.0 M KOH. (e) Comparison of the overpotential and stability of Ni₃N/Ni for the HER with other reported earth-abundant catalysts on planar substrates at a current density of 10 mA cm⁻².

As Ni is often used as a substrate material to prepare high-performance heterostructured catalysts, we also investigated the effect of Ni inclusion by preparing Ni₃N/Ni bilayer electrodes. Pt, known as the best HER catalyst, was also sputtered on FTO for comparison. The LSV curves obtained for different electrodes for the HER in 1.0 M KOH are shown in Fig. 4b. The Ni₃N/Ni/FTO electrode exhibits a commendable HER performance presenting low overpotentials of only 89 and 155 mV at 10 and 40 mA cm⁻², respectively. In comparison, the Pt/FTO electrode requires corresponding overpotentials of 57 and 169 mV to deliver the same HER current densities. It is noted that low overpotential at 40 mA cm⁻² current density is particularly beneficial for PEC hydrogen generation on Si photocathodes. Interestingly, the addition of Ni film significantly enhances the performance of sputtered Ni₃N films. Compared to the Ni₃N/FTO electrode, the Ni₃N/Ni/FTO electrode shows a reduction in overpotential at 40 mA cm⁻² by ∼30 mV (see Fig. S7, ESI†). This drastic improvement in the HER activity could be attributed to the synergistic effect of Ni₃N and Ni, which could enhance the electronic conductivity and relieve the film stress, thereby improving the HER performance. For example, a similar observation was reported for hematite deposited on FTO substrates, which resulted in lattice strain between hematite and the substrate. The lattice strain was addressed by adding a thin layer of other materials at the interface.⁴¹

The Tafel plots of the electrodes (see Fig. S8, ESI†) derived from the polarization curves in Fig. 4b confirm that Ni₃N/Ni/FTO (38 mV dec⁻¹) has a much lower slope compared to Ni₃N/FTO (60 mV dec⁻¹), but only slightly higher than that of Pt/FTO (28 mV dec⁻¹). Since the Tafel slope of an electrocatalyst is an indication of its HER reaction kinetics,^42,43 the lower Tafel slope of Ni₃N/Ni/FTO illustrates that the bilayer catalyst possesses faster kinetics and superior catalytic activity.

To further analyze the HER performance of different electrodes, electrochemical impedance spectroscopy (EIS) was conducted. As shown by the Nyquist plots inset in Fig. 4b, Ni₃N/Ni/FTO presents a much smaller charge transfer resistance (R_ct) at the electrode–electrolyte interface compared to Ni₃N/FTO (16.19 vs. 33.47 Ω cm⁻², Table S1, ESI†), illustrating an enhanced charge transport upon the addition of the Ni layer. In order to assess the specific intrinsic activities of Ni₃N/Ni/FTO and Ni₃N/FTO, their electrocatalytic activities were normalized by the electrochemically active surface area (ECSA) measured using the double layer capacitance method on the basis of cyclic voltammetry in 1.0 M KOH electrolyte (Fig. S9, ESI†). The higher double-layer capacitance (C_dl) of Ni₃N/Ni/FTO (1.10 mF cm⁻²) compared to that of Ni₃N/FTO (0.65 mF cm⁻²) indicates that Ni₃N/Ni/FTO possesses a higher ECSA.^3,44,45 This further confirms that the inclusion of the Ni layer plays a key role in boosting the catalytic activity of Ni₃N films.

The faradaic efficiency measurements (Fig. 4c) of the Ni₃N/Ni cathode based on the measured and calculated amounts of the evolved H₂ gas prove that there were no undesirable side reactions. To further examine the durability of Ni₃N/Ni/FTO as a cathode, a fixed overpotential of 89 mV was applied. As shown in Fig. 4d, a stable cathodic current density of ∼10 mA cm⁻² was maintained for over 3 days with negligible degradation, demonstrating its remarkable stability in an alkaline medium. In addition, as summarized in Fig. 4e and Table S2 (ESI†), the hierarchical Ni₃N/Ni/FTO exhibits the smallest overpotential and superior stability at 10 mA cm⁻² among other earth-abundant alkaline catalysts reported thus far on planar substrates. These results confirm that the sputtered Ni₃N/Ni films exhibit not only exceptionally high HER catalytic activity, which closely matches that of Pt but also robust electrochemical stability for practical applications.

Density functional theory calculations

DFT calculations were performed to understand the surface structure and the origin of the high activity of Ni₃N/Ni. Ni₃N(001) is a flat metallic surface and can be either N- or Ni-terminated. The Ni₃N(001) facet is chosen for the calculation, which is the same surface as the (002) facet observed in both TEM images (Fig. 2e) and XRD patterns (Fig. 3c). To simulate the effect of surface N-vacancies, the N-terminated and Ni-terminated surfaces are defined as 0% and 100% N-vacancies, respectively, while the surface also depleted of sub-surface N is defined as 200% N-vacancy (see also Fig. S10, ESI†). Based on this, a series of surface structures with gradual concentrations of N-vacancies from 0% to 200% were constructed, which also include NH-terminated surfaces in the aqueous conditions. The corresponding surface free energies (γ) were calculated to identify the most stable surface structures of Ni₃N(001) under both the synthesis and operating conditions. As shown in Fig. 5a, γ is related to the chemical potential for N (μ_N). With a decrease in μ_N to be more negative, the concentration of N-vacancies of the most stable surface structure with lowest γ (marked in colored regions) increases gradually from 0% to 200%. In the N₂ atmosphere as representative of the synthesis process, the surface with 75% N-vacancies is the most stable. To further relate the surface structure to the experimental HER working condition, a Pourbaix diagram of the most stable surface terminations among the whole range of 0–200% N-vacancy concentrations with possible *N or *NH surface terminations as a function of the working potential (E) and the pH value was calculated as shown in Fig. 5b. As marked by black dashed square, at pH ≈ 13.8 and E = −0.1 V in aqueous solution, the surface with 75% N-vacancies is the most stable, but it is hydrogen-terminated. N-vacancies are thus found to be strongly enriched at the surface, which agrees with the XPS characterization results in Fig. 3.

Fig. 5 (a) Surface free energy (γ) of Ni₃N(001) with varying amounts of N-vacancies as a function of surface chemical potential difference to gaseous N₂ (Δμ_N). The N-vacancy concentration of 0–100% corresponds to the vacancies of the surface layer, and 100–200% to the sub-surface layer. The coloured areas show the region of most stable structures under the lines with the lowest γ. Black, pink, green, yellow and red lines correspond to combined N-vacancy concentrations in the surface and sub-surface layer of 200% (Ni), 175%, 75%, 50% and 0% (Ni₃N), respectively (atomic structures are shown in Fig. S10, ESI†). 75% N-vacancies concentration is most stable at gaseous N₂ conditions (horizontal black dashed line). (b) Surface Pourbaix diagram of Ni_xN_yH_z with N with reference to dissolved N₂. The HER working condition is 1.0 M KOH (pH ≈ 13.8) and E = −0.1 eV, as approximately marked in the black dashed box. (c) Adsorption free energies of H* (G_H) on 4 inequivalent Ni₃-hollow sites exhibited by a Ni₃N(001) surface with varying degrees of surface N-vacancies. Multiple data points of the same colour correspond to the non-equivalent sites with different distances to surface N. The G_H on the inequivalent 2 Ni₃-hollow sites offered by Ni(111) are marked as horizontal dashed lines for comparison. The high HER activity region (G_H within 0–0.2 eV) is marked in light blue, see the text. (d) Bader charge of 4 inequivalent Ni₃-hollow sites exhibited by a Ni₃N(001) surface with varying degrees of surface N-vacancies. Positive Bader charge values represent atoms with a positive charge. The corresponding values for the Ni₃-hollow sites at Ni(111) are shown as dashed lines for comparison (the two inequivalent sites share the same value). The average values of the surface Ni atoms for each termination are marked as circles.

The Ni₃N(001) surface with or without N-vacancies is metallic, as evidenced by the finite projected density of states (pDOS) around the Fermi level as shown in Fig. S11 (ESI†). As judged from the pDOS shape, the 75% N-vacancy Ni₃N(001) surface resembles a regular Ni(111) surface. This shows that the surface is electrically conducting and the standard descriptor to assess the HER activity on transition metal surfaces, the H adsorption free energy (G_H), could also be employed here. From a thermodynamic point of view, the optimum HER activity is achieved at G_H ∼ 0 eV.^46,47 According to a recently proposed model of overpotential influence on the volcano line,⁴⁸ the working overpotential (η) of +0.1 V shifts the high HER activity region to ∼0–0.2 eV. DFT calculations were conducted to determine G_H on a variety of adsorption sites of Ni₃N(001). Especially, Ni₃ sites (hollow sites formed by three surface Ni atoms) offered by the Ni₃N(001) surface is found to be similar to those on Ni(111) and are thus suggested as potentially active sites for the HER. There are 4 such inequivalent Ni₃-hollow sites (see inset in Fig. 5c): one above a lower-layer N (H_N(l)), one above an upper-layer N(H_N(u)), one above a Ni atom (H_Ni) and one on the pore without sub-surface atoms (H_v). Each site type can have two or three different configurations when N-vacancies are present at the surface.

In order to avoid the risk from the insufficient representation of the selected site structure, a systematic computational screening of all active-site motives was performed to achieve a general trend to present the performance of the catalyst.⁴⁹ As shown in Fig. 5c, the adsorption free energies of the intermediate H*(G_H) exhibit a general decreasing trend with increasing N-vacancy concentration, indicating an increase in the H* binding strength at the Ni₃N(001) surface. In the limit of a pure Ni-termination (200% vacancies), the binding then essentially resembles one of the regular Ni(111) surface with G_H around −0.25 eV. Intriguingly, at around 75–100% N-vacancies concentration, i.e., at the surface termination identified as most stable under HER working conditions, significantly weaker binding is possible at the H_N(u) site that then falls exactly into the high HER activity window. On this basis, the higher HER activity than that of regular Ni(111) would be expected, consistent with the measured higher HER performance as shown in Fig. 4b.

In order to further understand the origin of the different binding strength of H* on surfaces with N-vacancies, the d-band center position (Fig. S12, ESI†) and Bader charge (Fig. 5d) of the four inequivalent Ni₃-hollow sites were calculated. While there is a considerable spread over the four sites for each termination, there are trends in the average values of the d-band center position and Bader charge of the surface Ni atoms. With an increase in N-vacancy concentration, the d-band center position first increases and then decreases, whereas the Bader charge decreases monotonously. Therefore, compared to the d-band center, the Bader charge value is more appropriate to explain the influence of the surface N-vacancy concentration on the catalytic performance of the surface Ni atoms. For the most stable 75–100% N-vacancies termination with optimum HER activity, the d-band position is Ni-like, while the average Bader charge at the active sites is still positive at about 0.2 e, i.e., the Ni atoms transfer charge to N atoms to keep the surface positively charged. Compared with the Ni(111) surface, where Ni atoms have almost zero charges, the electron-deficient sites on Ni₃N allow a weakened H bonding with higher HER activity. Instructively, for Ni-based catalysts, due to the strong binding between Ni and H atoms, adjusting the content of non-metallic atoms appropriately to make the surface atoms have an average positive charge of about 0.2 e that could achieve an optimal HER catalytic activity. This could be employed as a general strategy to design Ni-based catalysts with high activity.

Photoelectrochemical characterization of Ni₃N/Ni/Si photocathodes

To evaluate the performance of sputter-deposited Ni₃N/Ni as a cocatalyst for PEC water splitting, LSV measurements of Si, Pt/Si and Ni₃N/Ni/Si photocathodes were conducted in a three-electrode system, illuminated from the front p⁺ side. As shown in Fig. 6a, the Si photocathode without any cocatalyst presents a low onset potential below the thermodynamic HER potential of 0 V vs. RHE, rendering it ineffective for water splitting. In contrast, the photocathodes modified with Pt and Ni₃N/Ni catalysts show much-improved performance. The Ni₃N/Ni/Si photocathode exhibits an onset potential at 0.59 V and a photocurrent density of 38.8 mA cm⁻² at 0 V vs. RHE. In comparison, Pt/Si delivers a slightly lower photocurrent density of 36.5 mA cm⁻² with a similar onset potential. Fig. 6b shows that Ni₃N/Ni/Si displays an ABPE of 9.3%, which is comparable to that of the Pt/Si photocathode (9.9%), suggesting that Ni₃N/Ni co-catalyst is a superior choice to construct low-cost photocathodes.

Fig. 6 (a) Linear sweep voltammetry of Si, Pt/Si and Ni₃N/Ni/Si photocathodes and (b) the corresponding applied bias photon-to-current efficiency, and (c) open-circuit potential plots. (d) Chronoamperometry measurements for the Ni₃N/Ni/Si photocathode at 0 V vs. RHE. All measurements were performed at 0 V vs. RHE in 1.0 M KOH electrolyte (pH = 14) under AM1.5G illumination (100 mW cm⁻²).

To further understand the positive shift of onset potential for the as-prepared electrodes, open circuit potential (OCP) were measured in the dark and under illumination as shown in Fig. 6c. The OCPs for the samples move towards higher potentials upon exposure to the incident light, which is caused by the formation of a built-in electric field. The OCP differences before and after illumination are 0.1, 0.60, and 0.64 and for Si, Ni₃N/Ni/Si and Pt/Si photocathodes, respectively, which corresponds well with their onset potential values. Evidently, the integration of the Ni₃N/Ni and Pt cocatalyst greatly enhances the transfer of photogenerated holes toward the substrate, thereby delivering high photovoltage. The wavelength-dependent incident photon-to-current conversion efficiency (IPCE) for Ni₃N/Ni/Si photocathode (Fig. S13, ESI†) measured at 0 V vs. RHE in the wavelength range of 300–1200 nm reveals good light-harvesting ability of the photocathode at energies above the bandgap of Si. The stability measurement (Fig. 6d) over 50 h under simulated one sun illumination indicates that the Ni₃N/Ni catalyst film effectively stabilizes Si photocathodes in alkaline solutions. On the other hand, the Si photocathode without any cocatalysts was etched by the solution immediately (Fig. S14, ESI†). These results confirm that the sputtered Ni₃N/Ni film serves not only as a high-performance cocatalyst for Si photocathodes but also as an excellent protection layer in alkaline media.

Ni₃N/Ni on porous substrates and application in zero-gap cells

In order to evaluate the performance of the Ni₃N/Ni catalyst on porous substrates and at high current densities relevant to practical applications, we fabricated Ni₃N/Ni catalysts on different porous substrates, including Ni foam (NF), Ti mesh and stainless steel fiber (SS fiber), and evaluated their performance in zero-gap alkaline water electrolysis cells. The substrates were cleaned following the procedure as described in the experimental section in (ESI†). Ni₃N/Ni was sputtered on different porous substrates using the synthesis conditions optimized on FTO substrates. This is expected to yield a similar thickness, except for small changes due to the variation in the surface structure of different porous substrates. From SEM images of the blank substrates (Fig. S15a–c, ESI†), it is clear that different porous substrates present different surface characteristics. Among them, SS fiber presents a flatter and smoother surface except for the rubbed surface, which is comparable to that of planar FTO substrate. On the other hand, Ti mesh and NF present rough and uneven surfaces. Upon the catalyst coating, there is no obvious change in the surface morphology observed in SEM images collected at low magnification (Fig. 7a and Fig. S15d, e, ESI†). This indicates that sputter deposition results in a compact coating of catalyst layers on porous substrates with no aggregation. From SEM images collected at higher magnification (Fig. 7b–d), obvious surface features can be noticed. Ni₃N/Ni shows a similar triangular pyramid structure on all porous substrates. However, the size of the structures varies among different substrates, which is due to the difference in their surface properties. Noticeably, the morphology of Ni₃N/Ni on SS fiber is almost identical to that on FTO substrates due to its flatter and smoother surface.

Fig. 7 (a–d) SEM images of Ni₃N/Ni on different porous substrates. (e) LSV curves (with 90% iR-compensation) of Ni₃N/Ni on Ni foam (NF), Ti mesh, and stainless steel (SS) fiber substrates in 1.0 M KOH. (f) Schematic illustration of the configuration of the zero-gap electrolyser cell employed to test the performance at high current densities. (g) Current vs. voltage plots for the zero-gap alkaline water electrolyser with different Ni₃N/Ni porous electrodes as cathodes and NiFe₂O₄ as anode separated by Fumapem FA-3-50 anionic membrane in 1.0 M KOH at 55 °C. Inset is the long-term stability of the Ni₃N/Ni/Ti mesh electrode in the zero-gap cell.

The electrochemical performance of different porous electrodes was examined using LSV measurements in a three-electrode system in 1.0 M KOH. As shown in Fig. 7e, the Ni₃N/Ni/NF electrode outperforms the other porous electrodes presenting ∼66 mV overpotential at 10 mA cm⁻². In comparison, the Ni₃N/Ni on SS fiber and Ti mesh presented 132 and 118 mV overpotentials, respectively. It is noted that the overpotential of NF is even lower than that observed on FTO, which is due to its large interfacial surface area. In addition, Ni₃N/Ni/NF shows excellent electrochemical stability, as indicated by nearly constant overpotential at 15 mA cm⁻² for 139 hours (Fig. S16a, ESI†). The LSV data of Ni₃N/Ni/NF collected before and after the stability test shows negligible change (Fig. S16b, ESI†). These results confirm that our Ni₃N/Ni catalyst with N-vacancies directly synthesized using reactive magnetron sputtering presents promising HER performance even on porous substrates, which also compare favorably among other porous Ni nitrides reported in the literature (Table S3, ESI†). The Nyquist plots from the EIS measurements (Fig. S17 and Table S4, ESI†) confirm that the Ni₃N/Ni SS fiber electrode presents the largest semicircle and the highest charge transfer resistance (26.28 Ω cm²) among all electrodes, which results in a low HER performance.

To observe the performance of Ni₃N/Ni at high current densities for overall water splitting, we designed zero-gap cells as illustrated by the schematic in Fig. 7f. Further details on the zero-gap cell configuration are provided in the experimental section. Commercially-available NiCoFe on Ni fiber (cathode) and NiFe₂O₄ on stainless steel fiber (anode) were obtained from Dioxide Materials Inc. and used as a ref. 50. The performances of the commercial electrodes for the HER and OER in the three-electrode configuration are presented in Fig. S18 (ESI†). In all zero-gap cell measurements, the NiFe₂O₄ electrode was used as an anode and the cell was stabilized at 0.25 A cm⁻² for 1 h, prior to recording the response.

It can be seen that water electrolysis commences at a cell voltage >1.65 V and very high current densities are achieved with increasing cell voltages. The cell voltages for current densities of 0.5 A cm⁻² for Ni₃N/Ni on Ti mesh, SS fiber and NF are ∼1.88 V, 1.90 V and 1.92 V, respectively, which is approximate to the performance of commercial NiCoFe electrode with 1.85 V. It is interesting to note that Ni₃N/Ni on NF does not show any superiority in the performance in the zero-gap cell different to its performance in three-electrode measurements at low current densities. We attribute this to the relatively low through-plane conductivity of the NF substrate compared to the other porous substrates resulting in voltage losses at high current densities as confirmed from the EIS data (Fig. S19 and Table S5, ESI†). On the other hand, the Ni₃N/Ni on Ti mesh shows comparable performance with that of the commercial HER electrode. Ti mesh presents high through-plane conductivity and supports efficient bubble dissipation through its 2D porous structure, thereby outperforming the other porous electrodes in zero-gap cells. The performance also compares favorably to the other reported Ni-based electrodes in a single-cell water electrolysis cell (Table S6, ESI†). The long-term performance of the Ti mesh electrode in the zero-gap cell is shown as an inset in Fig. 7g, confirming that it maintains stable activity even at very high operating currents. In summary, the performances on various porous substrates and in zero-gap alkaline water electrolysis cells confirm the feasibility of sputtered Ni₃N/Ni with N-vacancies as a highly versatile HER electrocatalyst with potential for practical applications.

Conclusions

A hierarchical catalyst comprising pyramidal structured Ni₃N/Ni with N-vacancies was directly synthesized by reactive magnetron sputtering for alkaline hydrogen evolution reaction. Due to a synergistic enhancement of N-vacancies, high wettability, and high electrochemical surface area, the Ni₃N/Ni bilayer catalyst presented an excellent activity together with remarkable stability. The DFT calculations demonstrated that 75% N-vacancies concentration on the surface leading to superior activity and stability under strong alkaline media. Different from Ni(111), such Ni₃N(001) surfaces with N-vacancies exhibit metallic properties, lower surface free energy, positive charge sites and stronger H* binding strength under electrochemical working conditions. The Ni₃N/Ni decorated as a cocatalyst on Si photocathodes enhances its onset potential to 0.59 V, achieving 9.3% applied bias photon-to-current efficiency with high operational stability. Moreover, commendable performances and robustness were observed on porous substrates and at high current densities in the zero-gap alkaline electrolysis cell. The demonstration of earth-abundant Ni₃N/Ni catalysts with superior HER performance and one-step synthesized using reactive magnetron sputtering, i.e., without the need of a high-temperature nitridation process, provides a general approach for the fabrication of transition metal nitrides for electrochemical and photoelectrochemical hydrogen production.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

The financial support from the Australian government through the Australian Research Council (ARC) and Australian Renewable Energy Agency (ARENA) is gratefully acknowledged. Access to fabrication and characterization facilities of the Australian National Fabrication Facility (ANFF) and Centre for Advanced Microscopy is also gratefully acknowledged. H. L. gratefully acknowledges funding from the Alexander-von-Humboldt foundation. K. R. acknowledges funding by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) under Germany's Excellence Strategy – EXC 2089/1 – 390776260. The authors gratefully acknowledge the Gauss Centre for Supercomputing e.V. (http://www.gauss-centre.eu) for funding this project by providing computing time through the John von Neumann Institute for Computing (NIC) on the GCS Supercomputer JUWELS at Jülich Supercomputing Centre (JSC).

References

1. N. S. Lewis and D. G. Nocera, Proc. Natl. Acad. Sci. U. S. A., 2006, 103, 15729–15735.
2. I. Roger, M. A. Shipman and M. D. Symes, Nat. Rev. Chem., 2017, 1, 1–13.
3. D. Zhang, J. Z. Soo, H. H. Tan, C. Jagadish, K. Catchpole and S. K. Karuturi, Advanced Energy and Sustainability Research, 2021, 2, 2000071.
4. V. Vij, S. Sultan, A. M. Harzandi, A. Meena, J. N. Tiwari, W.-G. Lee, T. Yoon and K. S. Kim, ACS Catal., 2017, 7, 7196–7225.
5. Y. Li, X. Bao, D. Chen, Z. Wang, N. Dewangan, M. Li, Z. Xu, J. Wang, S. Kawi and Q. Zhong, ChemCatChem, 2019, 11, 5913–5928.
6. Z. Wang, J. Fan, B. Cheng, J. Yu and J. Xu, Mater. Today Phys., 2020, 100279.
7. X. Zheng, X. Han, Y. Zhang, J. Wang, C. Zhong, Y. Deng and W. Hu, Nanoscale, 2019, 11, 5646–5654.
8. Y. Yan, B. Y. Xia, B. Zhao and X. Wang, J. Mater. Chem. A, 2016, 4, 17587–17603.
9. S. Trasatti, J. Electroanal. Chem. Interfacial Electrochem., 1972, 39, 163–184.
10. J. Greeley, J. K. Nørskov, L. A. Kibler, A. M. El-Aziz and D. M. Kolb, Chem. Phys. Chem., 2006, 7, 1032–1035.
11. B. Liu, B. He, H. Q. Peng, Y. Zhao, J. Cheng, J. Xia, J. Shen, T. W. Ng, X. Meng and C. S. Lee, Adv. Sci., 2018, 5, 1800406.
12. F. Song, W. Li, J. Yang, G. Han, P. Liao and Y. Sun, Nat. Commun., 2018, 9, 1–10.
13. W. Hua, H. Sun, H. Liu, Y. Li and J.-G. Wang, Appl. Surf. Sci., 2021, 540, 148407.
14. W. F. Chen, K. Sasaki, C. Ma, A. I. Frenkel, N. Marinkovic, J. T. Muckerman, Y. Zhu and R. R. Adzic, Angew. Chem., Int. Ed., 2012, 51, 6131–6135.
15. M. Shalom, D. Ressnig, X. Yang, G. Clavel, T. P. Fellinger and M. Antonietti, J. Mater. Chem. A, 2015, 3, 8171–8177.
16. P. J. Kelly and R. D. Arnell, Vacuum, 2000, 56, 159–172.
17. J. Liang, Q. Liu, T. Li, Y. Luo, S. Lu, X. Shi, F. Zhang, A. M. Asiri and X. Sun, Green Chem., 2021, 23, 2834–2867.
18. D. Zhang, L. Meng, J. Shi, N. Wang, S. Liu and C. Li, Electrochim. Acta, 2015, 169, 402–408.
19. L. Chen and Q. Xu, Science, 2020, 367, 737.
20. J. Li, C. Shu, C. Liu, X. Chen, A. Hu and J. Long, Small, 2020, 16, 2001812.
21. Z. Luo, T. Wang and J. Gong, Chem. Soc. Rev., 2019, 48, 2158–2181.
22. D. Zhang, J. Shi, W. Zi, P. Wang and S. Liu, ChemSusChem, 2017, 10, 4324–4341.
23. S. K. Karuturi, H. Shen, A. Sharma, F. J. Beck, P. Varadhan, T. Duong, P. R. Narangari, D. Zhang, Y. Wan and J. H. He, Adv. Energy Mater., 2020, 10, 2000772.
24. B. Ouyang, Y. Zhang, Z. Zhang, H. J. Fan and R. S. Rawat, Small, 2017, 13, 1604265.
25. J. S. Kang, M.-A. Park, J.-Y. Kim, S. H. Park, D. Y. Chung, S.-H. Yu, J. Kim, J. Park, J.-W. Choi and K. J. Lee, Sci. Rep., 2015, 5, 1–11.
26. D. Kuznetsov, G. Ugodnikov and I. Filatov, Tech. Phys. Lett., 2008, 34, 87–89.
27. N. E. Brese and M. O'Keeffe, Complexes, Clusters and Crystal Chemistry, 1992, pp. 307–378.
28. K. Xu, P. Chen, X. Li, Y. Tong, H. Ding, X. Wu, W. Chu, Z. Peng, C. Wu and Y. Xie, J. Am. Chem. Soc., 2015, 137, 4119–4125.
29. M.-S. Balogun, Y. Zeng, W. Qiu, Y. Luo, A. Onasanya, T. K. Olaniyi and Y. Tong, J. Mater. Chem. A, 2016, 4, 9844–9849.
30. D. Zhang, J. Shi, Y. Qi, X. Wang, H. Wang, M. Li, S. Liu and C. Li, Adv. Sci., 2018, 5, 1801216.
31. T. Liu, M. Li, C. Jiao, M. Hassan, X. Bo, M. Zhou and H.-L. Wang, J. Mater. Chem. A, 2017, 5, 9377–9390.
32. D. Gao, J. Zhang, T. Wang, W. Xiao, K. Tao, D. Xue and J. Ding, J. Mater. Chem. A, 2016, 4, 17363–17369.
33. S. H. Park, Y.-H. Cho, M. Choi, H. Choi, J. S. Kang, J. H. Um, J.-W. Choi, H. Choe and Y.-E. Sung, Surf. Coat. Technol., 2014, 259, 560–569.
34. Y. Wang, Z.-W. Fu, X.-L. Yue and Q.-Z. Qin, J. Electrochem. Soc., 2004, 151, E162.
35. L. Xu, Q. Jiang, Z. Xiao, X. Li, J. Huo, S. Wang and L. Dai, Angew. Chem., 2016, 128, 5363–5367.
36. E. Rohaeti and I. Hikmawati, J. Mater. Sci. Technol., 2010, 265–272.
37. H. Ang, H. T. Tan, Z. M. Luo, Y. Zhang, Y. Y. Guo, G. Guo, H. Zhang and Q. Yan, Small, 2015, 11, 6278–6284.
38. S. Geng, F. Tian, M. Li, X. Guo, Y. Yu, W. Yang and Y. Hou, J. Mater. Chem. A, 2021, 9, 8561–8567.
39. B. S. Mun, M. Watanabe, M. Rossi, V. Stamenkovic, N. M. Markovic and P. N. Ross Jr, J. Chem. Phys., 2005, 123, 204717.
40. Y. Pan, K. Sun, Y. Lin, X. Cao, Y. Cheng, S. Liu, L. Zeng, W.-C. Cheong, D. Zhao and K. Wu, Nano Energy, 2019, 56, 411–419.
41. F. Le Formal, M. Grätzel and K. Sivula, Adv. Funct. Mater., 2010, 20, 1099–1107.
42. Y. Zheng, Y. Jiao, Y. Zhu, L. H. Li, Y. Han, Y. Chen, A. Du, M. Jaroniec and S. Z. Qiao, Nat. Commun., 2014, 5, 1–8.
43. N. Krstajić, M. Popović, B. Grgur, M. Vojnović and D. Šepa, J. Electroanal. Chem., 2001, 512, 16–26.
44. W.-F. Chen, J. T. Muckerman and E. Fujita, Chem. Commun., 2013, 49, 8896–8909.
45. Y. Yoon, B. Yan and Y. Surendranath, J. Am. Chem. Soc., 2018, 140, 2397–2400.
46. J. K. Nørskov, T. Bligaard, A. Logadottir, J. Kitchin, J. G. Chen, S. Pandelov and U. Stimming, J. Electrochem. Soc., 2005, 152, J23.
47. J. Greeley, T. F. Jaramillo, J. Bonde, I. Chorkendorff and J. K. Nørskov, Nat. Mater., 2006, 5, 909–913.
48. K. S. Exner, Angew. Chem., Int. Ed., 2020, 59, 10236–10240.
49. H. Li and K. Reuter, ACS Catal., 2020, 10, 11814–11821.
50. Z. Liu, S. D. Sajjad, Y. Gao, H. Yang, J. J. Kaczur and R. I. Masel, Int. J. Hydrogen Energy, 2017, 42, 29661–29665.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1ee02013g

---