In 2009, Wang and coworkers reported a novel metal-free polymeric semiconductor photocatalyst, graphitic carbon nitride (g-C₃N₄), for H₂ evolution [1]. Since then, many studies have been published concerning the use of g-C₃N₄ for solar energy conversion and environmental applications, mainly because of its facile synthesis, appealing bandgap (E_g = 2.7 eV), suitable electronic structure, and high physicochemical stability. Nevertheless, bulk g-C₃N₄ has a low surface area and the photogenerated charges recombine rapidly, which reduces the photocatalytic performance [2]. Hence, many efforts have been made to improve the performance of g-C₃N₄, for example, by doping with nonmetals and metals [3-9], protonation by strong acids [10], the introduction of porosity [11-13], and the fabrication of heterojunction composites [14-16]. Among the various strategies, the loading of moderate amounts of cocatalysts onto the surface of g-C₃N₄ is believed to be an ideal strategy to increase the separation of photogenerated electron–hole pairs and provide active sites for photocatalytic H₂ production via water splitting [17].

More recently, considering the practical applications, attention has focused on low-cost and earth-abundant materials as replacements for noble metal cocatalysts for photocatalytic H₂ evolution systems [18]. These non-noble-metal catalysts include Co [19], CoP [17], Ni₂P [20], NiS [21], Ni(OH)₂ [22], Cu₃P [23], Fe₂P [24], MoP [25], and NiS [21]. Meanwhile, transition metal nitrides have received considerable attention because of their good electrical conductivity and flexible (electro)catalytic properties [6, 26, 27]. For example, Xu et al. [28] first reported metallic Ni₃N nanosheets as efficient oxygen evolution reaction (OER) electrocatalysts. Shalom et al. [26] fabricated Ni₃N on Ni foam for electrocatalytic applications, which exhibited an extremely low overpotential, high current density, and excellent stability for the hydrogen evolution reaction (HER) in alkaline solution. Our group reported that the use of Ni₃N as a cocatalyst on CdS nanorods for the photocatalytic HER can enhance the rate by more than 10 times [6]. However, from a literature survey, the use of a noble-metal-free Ni₃N as a cocatalyst on g-C₃N₄ for photocatalytic H₂ production has received little attention.

Motivated by the above results, we present the use of low-cost, noble-metal-free Ni₃N as an active cocatalyst for g-C₃N₄ via modification of the semiconductor surface. These Ni₃N/g-C₃N₄ hybrids present an enhanced visible-light-driven H₂ production rate of 305.4 μmol g^-1 h^-1, which is about three-times higher than that of bulk g-C₃N₄, and the apparent quantum yield (AQY) was ~0.45% at 420 nm. In addition, the photoluminescence (PL) spectra and photoelectrochemical characterization revealed that Ni₃N is an efficient cocatalyst for photocatalytic H₂ evolution. Furthermore, a photocatalytic hydrogen evolution mechanism based on Ni₃N/g-C₃N₄ is proposed and discussed in detail.

All reagents (analytical grade) were purchased from Aladdin Chemical Regent Co., Ltd. (Shanghai, China) and used directly without further purification.

The g-C₃N₄ was prepared by thermal pyrolysis based on a previously reported method [29].

In a typical process, calculated amounts of Ni(NO₃)₂·6H₂O (29.079, 58.158, 145.395, and 203.553 mg) and hexamethylenetetramine (HMT, 8.038, 56.076, 140.19, and 196.266 mg) were dissolved in 40 mL H₂O with vigorous magnetic stirring for 0.5 h. Then, 0.5 g of g-C₃N₄ was added to the solution with vigorous stirring. The resulting suspension was transferred to a 50-mL Teflon-lined, stainless-steel autoclave and maintained at 120 ℃ for 12 h. After cooling to room temperature, the precipitates were collected by centrifugation and washed with ethanol and distilled water three times each and dried under vacuum overnight. After that, the precipitates were annealed at 380 ℃ for 3 h under a flow of NH₃ gas. The obtained samples were labeled Ni₃N/g-C₃N₄#1, Ni₃N/g-C₃N₄#2, Ni₃N/g-C₃N₄#3, and Ni₃N/g-C₃N₄#4. For comparison, pristine Ni₃N nanoparticles were also prepared using the same method in the absence of g-C₃N₄.

All the prepared photocatalyst samples were systematically investigated using powder X-ray diffraction (PXRD, D/max-TTR Ⅲ, 5° min^-1 from 10° to 70° in 2θ), scanning electron microscopy (SEM, SIRION200 equipped with electron diffraction), transmission electron microscopy (TEM, JEM-2010, acceleration voltage of 200 kV), UV-Vis spectrometry (UV-Vis, SOLID 3700), and X-ray photoelectron spectroscopy (XPS, ESCALAB 250).

The photocatalytic activity reactions were performed in a 50-mL flask with magnetic stirring using A 300-W xenon lamp as the irradiation source. The lamp was equipped with a UV cut-off filter (λ > 420 nm). Photocatalytic H₂ production was quantified by gas chromatography (GC, SP6890, thermal conductivity detector (TCD) detector, high purity nitrogen as a carrier gas, and 5 Å molecular sieve column). To investigate the long-term photocatalytic H₂ production stability under visible-light irradiation, 5.0 mg of the sample was ultrasonically dispersed in a 20 vol% aqueous solution of triethanolamine (TEOA) in a 250-mL flask. The apparent quantum efficiency was calculated using a 300-W Xe lamp with a band-pass filter (λ = 420 nm). The AQY was calculated using the following equation.

The crystalline phases of pristine g-C₃N₄ and the pristine Ni₃N nanoparticles, together with those of the Ni₃N/g-C₃N₄ hybrid composites, were analyzed by XRD. As shown in Fig. 1, the characteristic peaks at 13.08° and 27.4° correspond to the (100) and (002) crystalline planes, respectively, for g-C₃N₄ with a graphitic structure (JCPDS#87-1526) [30]. The characteristic peaks at 38.9°, 42.5°, 44.5°, and 58.6° can be indexed to the (110), (002), (111), and (112) planes, respectively, for hexagonal Ni₃N (JCPDS#89-5144) [27]. Meanwhile, the Ni₃N/g-C₃N₄ hybrid composites samples display similar XRD patterns to both pristine g-C₃N₄ and pristine Ni₃N nanoparticles. With increasing Ni₃N content, the peak intensities of Ni₃N gradually become stronger, and no other impurities were detected, indicating that there are no obvious changes to g-C₃N₄ after modification with Ni₃N nanoparticles under a flow of NH₃ gas.

The morphologies and material compositions of pristine g-C₃N₄ and the Ni₃N/g-C₃N₄#3 hybrid composite were further analyzed using SEM, TEM, and energy-dispersive X-ray (EDX) measurements. As shown in Fig. 2(a), g-C₃N₄ has a typical lamellar structure of multilayer g-C₃N₄ nanosheets, and the Ni₃N/g-C₃N₄#3 sample has a structure where the nanoparticles are closely anchored to the surface of the g-C₃N₄ nanosheets (Fig. 2(b)). As shown in the TEM image of the Ni₃N/g-C₃N₄#3 hybrid composite (Fig. 2(c)), the Ni₃N nanoparticles were deposited on g-C₃N₄ surface. In addition, the elemental mapping and EDS spectra (Fig. 3) clearly reveal the existence of Ni and N, as well as Cu from the sample base, suggesting that Ni₃N nanoparticles had been successfully loaded onto the g-C₃N₄ surface. Thus, these characterization results further show that the Ni₃N nanoparticles were successfully loaded onto surface of the g-C₃N₄.

The chemical states and surface composition of the N₃N/g-C₃N₄#3 hybrid composite photocatalyst were further investigated using XPS. The survey XPS scan spectrum shown in Fig. 4(a) clearly reveals the presence of Ni, C, O, and N, as well as C, which was used as the reference, and O from the absorbed gaseous molecules and oxidized Ni species [6]. In the high-resolution C 1s XPS spectrum (Fig. 4(b)), two peaks were deconvoluted into peaks at 284.8, 286.1, and 288.3 eV; these correspond to graphitic C-C bonds, C-O bonds, and sp²-hybridized carbon in N-containing aromatic ring (N-C=N), respectively, thus confirming the presence of g-C₃N₄. In the Ni 2p region, four peaks are seen at 856.2, 874.3, 861.3, and 880.3 eV (Fig. 4(d)), which are ascribed to the Ni 2p_3/2, and Ni 2p_1/2 peaks and oxidized Ni species (NiO) [26, 28]. In addition, N 1s peaks are located at 398.8, 399.9, and 401.1 eV (Fig. 4(c)); these Ni and N peaks are consistent with the characteristics of Ni₃N and are characteristic of C-N-C, tertiary nitrogen N-(C)₃ groups, and the tertiary nitrogen N-(C)₃ groups, respectively [31].

The light harvesting properties of the pristine g-C₃N₄, pristine Ni₃N, and Ni₃N/g-C₃N₄ #3 composites were measured by UV-vis diffuse reflectance spectroscopy. As depicted in Fig. 5(a), no obvious bandgap absorption structure was observed for Ni₃N, indicating its typical metallic character, and there is no apparent difference in the bandgap absorption edge between the g-C₃N₄ and Ni₃N/g-C₃N₄ #3 hybrid composites, suggesting Ni₃N was not doped into the g-C₃N₄ crystal lattice, and, thus, there was no change in its bandgap, which can be estimated based on the diffuse reflection spectral data. As shown in Fig. 5(b), the bandgaps of pristine g-C₃N₄ and the Ni₃N/g-C₃N₄ #3 composite were estimated to be 2.55 eV, which is slightly smaller than the reported values [32].

PL spectral analysis was used to determine the recombination rate of the photogenerated electron-hole pairs. A lower PL emission intensity is an indication of a lower recombination rate of the photogenerated electron-hole pairs [33]. Fig. 6(a) shows the PL spectra of the pristine g-C₃N₄ and Ni₃N/g-C₃N₄ composite samples. As shown, the PL spectra of all samples have the same wide emission peak at about 450 nm under an excitation wavelength of 385 nm, which is ascribed to the bandgap recombination of photoexcited electron-hole pairs in pristine g-C₃N₄. Moreover, the PL intensity of the samples decreases in order of pristine g-C₃N₄ > g-C₃N₄(NH₃ treated) > Ni₃N/g-C₃N₄#1 > Ni₃N/g-C₃N₄#2 > Ni₃N/g-C₃N₄#4 > Ni₃N/g-C₃N₄#3, which demonstrates that the recombination rate of photogenerated electron–holes first decreased with increasing amount of Ni₃N and, then, increased when an excess amount of Ni₃N was incorporated. The excess Ni₃N may act as recombination centers covering the active sites on the g-C₃N₄ surface, thus lowering the charge separation efficiency.

The transient photocurrent response curves of the electrodes coated with pristine g-C₃N₄ and Ni₃N/g-C₃N₄#3 hybrid composite were recorded for several on-off cycles in 0.5 mol L^-1 Na₂SO₄ aqueous solution at 0.0 V vs. Ag/AgCl (I-t curve, Fig. 6(b)) [34]. Both samples yield relatively low photocurrents without light irradiation. Interestingly, an apparent increase in the photocurrent appears when the visible-light source was turned on. The working electrode coated with the Ni₃N/g-C₃N₄#3 hybrid composite exhibited a much higher photocurrent than that coated with g-C₃N₄. This can be ascribed to the Ni₃N nanoparticles on the g-C₃N₄ interface, which efficiently separated the photogenerated charge carriers and resulted in decreased photoinduced electron-hole recombination and an enhanced photocurrent.

Electrochemical impedance spectroscopy (EIS) measurements of the electrodes coated with pristine g-C₃N₄ and Ni₃N/g-C₃N₄#3 hybrid composite were obtained in 0.5 mol L^-1 Na₂SO₄ aqueous solution to investigate the electron transfer and combination of the semiconductor in solution. As shown in Fig. 6(c), the EIS results indicate that the impedance arc radius of the Ni₃N/g-C₃N₄#3 hybrid composite is smaller than that of the pristine g-C₃N₄, indicating that the Ni₃N/g-C₃N₄#3 composite shows enhanced photoexcited charge carrier separation efficiency compared to that of g-C₃N₄. This result indicates the Ni₃N efficiently facilitates the transport of photogenerated charge carriers and promotes hydrogen production activity [35].

Photocatalytic H₂ production experiments were carried out under visible-light irradiation in aqueous TEOA solution using pristine g-C₃N_4, g-C₃N₄ treated with NH₃, and pristine g-C₃N₄ loaded with different amounts of Ni₃N. As shown in Fig. 7(a), the hydrogen production rate increased initially and, subsequently, decreased with increasing ratio of loaded Ni₃N, indicating that Ni₃N is an efficient cocatalyst. However, excess Ni₃N may shield the incident light and may also block the active sites responsible for hydrogen production. Pristine Ni₃N is not active for hydrogen evolution. In addition, the mechanically mixed sample of pristine g-C₃N₄ and Ni₃N exhibited a lower rate of hydrogen production than Ni₃N/g-C₃N₄#3, highlighting the importance of the close contact between pristine g-C₃N₄ and the Ni₃N cocatalyst. Meanwhile, the effect of the type and concentration of sacrificial electron donors was further investigated. As shown in Fig. 7(b), when Ni₃N/g-C₃N₄#3 and TEOA were combined, the photocatalytic activity for hydrogen production increased sharply, and a maximum H₂ production rate of ~305.4 μmol h^-1 g^-1 was obtained. The results are also shown in Fig. 7(c), which reveal that the rate of H₂ production slightly decreased when the TEOA concentration further increased, indicating that very high concentrations of the TEOA electron donor does not improve the photocatalytic activity beyond the optimal concentration.

The ability of the photocatalyst to remain active over multiple cycles is vital for practical applications [36]. As shown in Fig. 8(a), the Ni₃N/g-C₃N₄#3 photocatalyst showed no obvious decrease in the hydrogen production rate after five cycles under visible-light irradiation, which indicate that the Ni₃N/g-C₃N₄ photocatalyst have good photocatalytic durability and stability. Furthermore, the hydrogen production rate reached ~329.6 μmol^-1 g^-1 upon irradiation with 420 nm monochromatic light for 7 h (Fig. 8(b)). After 4 h, the AQY was maintained at an average value of ~0.45%.

Based on the above characterization, a possible photocatalytic mechanism for the enhanced photocatalytic activity of the Ni₃N/g-C₃N₄ hybrid composites has been proposed, as shown in Scheme 1. When pristine g-C₃N₄ was irradiated with visible light, the electrons in the valence band (VB) of g-C₃N₄ are excited to the conduction band (CB). The photogenerated electrons will either recombine with the holes or transfer to the surface for photochemical hydrogen evolution reactions. As a metallic compound, Ni₃N nanoparticles will form a typical metal-semiconductor interface with g-C₃N₄, and the photogenerated electrons are able to transfer from the semiconductor photocatalyst g-C₃N₄ to the metallic cocatalyst, Ni₃N [6]. Thus, loading moderate amounts of Ni₃N onto g-C₃N₄ facilitates the separation of the photogenerated electron-hole pairs in g-C₃N₄, resulting in the improved photocatalytic activity.

In summary, we have successfully developed a novel Ni₃N/g-C₃N₄ hybrid composite photocatalyst by a facile thermal ammonolysis method. The photocatalytic activity for hydrogen production is enhanced after the loading of Ni₃N onto g-C₃N₄ under visible-light irradiation. The hydrogen evolution rate reached ~305.4 μmol·h^-1·g^-1, which is about three times higher than that of pristine g-C₃N₄, and the AQY was ~0.45% at λ = 420 nm. The characterization results indicated that Ni₃N, a noble-metal-free cocatalyst, can efficiently promote the separation of the photogenerated electron-hole pairs in g-C₃N₄. This work demonstrates the potential of noble-metal-free Ni₃N as a cocatalyst comprising earth-abundant elements (Co and Ni) for photocatalysis.