Nowadays, the energy crisis and global warming problems caused by fossil fuel combustion are becoming more and more serious issues, and are two major challenges hindering the long-term and sustainable development of human civilizations [1, 2]. In view of the above problems, the development of environmentally friendly and renewable technologies for green energy production is an urgent task. Ever since the landmark event of photoelectrochemical water splitting on TiO₂ electrodes was reported by Fujishima et al. in 1972 [3], photocatalytic technology has been considered as an effective and sustainable technology to solve the global energy issue [3, 4]. Especially, photocatalytic H₂ production has gained considerable interdisciplinary attention due to high-specific-energy-efficient, non-polluting, and storable hydrogen energy [5]. In the photocatalytic hydrogen production process, not only is clean and inexhaustible solar energy required as a driving force, but a suitable semiconductor as the photocatalyst is required as well. Until now, many traditional materials, like transition-metal-based oxides [6], sulfides [7], selenides [8], etc., have been studied for photocatalytic hydrogen production. In addition to these traditional materials, polymeric semiconductor-based materials like graphitic carbon nitrides (g-C₃N₄) have also received increasing attention as a new family of promising photocatalysts for photocatalytic H₂ generation.

The state-of-the-art g-C₃N₄ was found to perform photocatalytic water splitting under visible light in 2009 [9], potentially shifting the search hotspot for photocatalysts from inorganic semiconductors to more abundant organic/polymeric ones. As a metal-free polymeric material, g-C₃N₄ possesses a two-dimensional (2D) planar structure with a π-conjugated electronic system, a moderate optical band gap (≈ 2.7 eV), and good physical and chemical stability [10]. These allow its direct use as a heterogeneous catalyst in the activation of carbon dioxide (CO₂) and other organic reactions [11, 12], and as a photocatalyst for the splitting of water into hydrogen gas [13-15]. In spite of the advantages mentioned above, the photocatalytic hydrogen evolution activity of pure g-C₃N₄ is not always noticeable due to the fast recombination of photogenerated electron-hole pairs in the bulk or on its surface. Thus, the transfer and separation of the charge carriers would be key factors in improving the photocatalytic performance of g-C₃N₄. To date, many attempts have been made to modify the physical and electron structures of g-C₃N₄, for example, integrating with other semiconductors [16-23], doping with metal or nonmetal elements [24-26], or surface modification [27-30].

Among the various strategies, modification with organic molecules has been identified as an effective and feasible route to promote the separation of photoinduced electron-hole pairs [31, 32]. From this viewpoint, we have previously designed organic molecule-g-C₃N₄ photocatalysts for H₂ production. We have found that hydrogen bonds or π-π interactions can form between the molecule and planar aromatic structure of g-C₃N₄ [33, 34]. The driving force among them could facilitate the transfer and separation of photoinduced electron and hole (e^-/h⁺) pairs, leading to higher photocatalytic hydrogen production. In addition, functionalizing g-C₃N₄ through means of π-π non-covalent interactions can preserve its electronic structure. Therefore, we can design more efficient g-C₃N₄-based photocatalysts through π-π interactions for enhancing photogenerated charge transfer and even expanding the photoresponse range of graphitic carbon nitride.

Phthalocyanines (Pc) make up a class of macrocyclic molecules with an intrinsic planar 18-π conjugated network [35, 36]. It has excellent optical and electrochemical properties, especially its intrinsic absorption in the 600–800 nm range, which is usually referred to as the Q-band and is responsible for the characteristic intense blue color of the materials [37]. Therefore, it is beneficial to immobilize phthalocyanine molecules on g-C₃N₄ through π-π stacking interactions for photocatalysis due to the conjugated two-dimensional electron system of g-C₃N₄. Recently, a series of phthalocyanines, including magnesium phthalocyanine (MgPc) and zinc phthalocyanine (ZnPc) derivatives, has been used to sensitize g-C₃N₄ and has showed enhanced photocatalytic hydrogen production [38, 39]. However, the utilization of vanadyl phthalocyanine (VOPc) is rarely reported. According to the literature, the LUMO and HOMO positions of VOPc are -1.29 and +0.71 eV, respectively, which are well matched with the energy bands of g-C₃N₄ [9, 40].

Hence, we report that g-C₃N₄ decorated with VOPc molecules via a simple ultrasonic method facilitates H₂ production from water under visible-light irradiation. The obtained VOPc/CN photocatalyst showed a broader spectral responsive range (460–800 nm) compared to that of pure g-C₃N₄ (< 460 nm). More importantly, the π-π interaction between VOPc and the planar aromatic structure of g-C₃N₄ could speed up the interfacial charge transfer rate. As a result, the VOPc/CN photocatalyst exhibited significantly enhanced photocatalytic activity for H₂ production under visible-light irradiation in comparison to pure g-C₃N₄. This work demonstrates that phthalocyanine has a promising application for photocatalytic hydrogen production systems that can utilize solar radiation more efficiently.

Urea (≥99%) and triethanolamine (TEOA, ≥78%) were purchased from Sinopharm Chemical Reagent. Vanadyl phthalocyanine (VOPc, ≥98%) was bought from J & K Chemical Reagents Ltd. Hexahydrate (H₂PtCl₆·6H₂O, ≥37% Pt basis) was bought from Aldrich. All chemical reagents were used without further purification. Double-distilled water used in the whole experimental process was purified through an SZ-93A auto-double distillation apparatus (Ya Rong Corp., Shanghai, China).

In a typical synthesis, 4.0 g of urea was added to an alumina crucible with the cover and then heated in a muffle furnace under an air atmosphere at a rate of 5 ℃ min^-1 to 550 ℃, and kept at this temperature for 4 h. After cooling to room temperature naturally, the resulting yellow powders were milled and collected for further use.

The crystal structures of the samples were investigated using X-ray diffraction (XRD, MXPAHF, Japan) at room temperature with Cu K_α radiation (λ = 1.541 Å ). The acceleration voltage and applied current were 40 kV and 20 mA, respectively. Transmission electron microscopy (TEM) images, high-resolution transmission electron microscopy (HRTEM) images, and energy-dispersive X-ray spectroscopy (EDX) elemental mapping analyses were obtained on a JEOL JEMARF200F atomic resolution analytical microscope with a spherical aberration corrector. Fourier transform infrared (FT-IR) spectra were recorded on a Nicolet Magana-IR 750 spectrometer in the 400 to 4000 cm^-1 range. X-ray photoelectron spectroscopy (XPS) patterns were acquired with a Thermo ESCALAB 250 system. The UV-vis diffuse reflectance spectra over a range of 200–800 nm were recorded by a Shimadzu spectrophotometer (Model 2501 PC). The UV-vis absorption spectra of aqueous solutions were obtained using a Shimadzu UV-2510 spectrophotometer. The steady-state photoluminescence (PL) of the photocatalyst was measured on a fluorescence spectrophotometer (JY Fluorolog-3-Tau) at room temperature. The time-resolved photoluminescence (TRPL) measurements were performed using a fluorescence detector (QM400-TM, Photo Technology International, USA). N₂ adsorption-desorption isotherms were measured at 77 K by an adsorption instrument (Micromeritics ASAP 2010 system) and the Brunauer-Emmett-Teller (BET) specific surface area (S_BET) was calculated using a multi-point BET method.

Photoelectrochemical tests were done at a CHI 760E electrochemical workstation (Chenhua Instrument Company, Shanghai, China) based on a standard three-electrode system composed of Ag/AgCl as the reference electrode, Pt wire as the counter electrode, and indium-tin oxide (ITO) glass as the working electrode. Na₂SO₄ solution (0.5 mol/L) was used as the electrolyte. The g-C₃N₄ and VOPc/CN electrodes were prepared by depositing suspensions made of g-C₃N₄ or VOPc/CN and water (the concentration of g-C₃N₄ and VOPc/CN was 2 mg mL^-1) onto ITO glass, respectively.

The experiments on H₂ evolution from water splitting were carried out in an outer top-irradiation gas-closed Pyrex glass system (500 mL) with a 300-W xenon lamp (Perfect Light, PLS-SXE300C, Beijing) equipped with a cut-off filter (λ ≥ 420 nm). Typically, 50 mg of g-C₃N₄ powder and 2 mg of VOPc were dispersed in 100 mL of a water solution containing 10% triethanolamine (TEOA) as the sacrificial electron donor. Then, 1 wt% Pt, as a co-catalyst to boost H₂ generation, was loaded onto the surface of the catalyst by in situ photodeposition of H₂PtCl₆·6H₂O. Before the photocatalytic reaction, the solution was evacuated for 1 h in order to remove air completely. To eliminate any thermal effects, the temperature of the reaction solution was maintained at 10 ℃ by a flow of cooling ethylene glycol. The amount of hydrogen evolved from photocatalytic water splitting was measured by an on-line gas chromatograph (GC1120, Shanghai Sunny Hengping Limited, HTCD), and N₂ was used as the carrier gas. To get an accurate amount of the generated H₂, an average value from three measurements was adopted.

The apparent quantum yields (AQYs) for H₂ evolution were measured with irradiation under different monochromatic light (irradiated by a 300-W Xe lamp using a band-pass filter of λ ± 5 nm for 420, 450, 500, 550, and 600 nm) under the same photocatalytic reaction conditions. The AQY was calculated according to the following equation:

The photocatalyst of g-C₃N₄ was prepared by directly heating urea at 550 ℃ for 4 h in an air flow. Vanadyl phthalocyanine modified g-C₃N₄ hybrid photocatalysts (VOPc/CN) were obtained via an ultrasonic method in solution. The crystalline phases of VOPc, g-C₃N₄, and the 4 wt%-VOPc/CN photocatalysts were analyzed by XRD and shown in Fig. S1. The VOPc/CN hybrid photocatalyst exhibits similar peaks to those of g-C₃N₄, indicating that the introduction of VOPc does not change the crystal phase or electronic structure of g-C₃N₄. There was no characteristic peak of VOPc molecules from the VOPc/CN photocatalyst. This fact might be due to the loading amount of VOPc being relatively low for detection and the highly dispersed VOPc molecules on the surface of g-C₃N₄. In addition, the morphology of g-C₃N₄ nanosheets does not change after modification by VOPc (Fig. S2). The successful hybridization of VOPc molecules with g-C₃N₄ was proved through EDX (Fig. S3), elemental mapping (Fig. S3), FT-IR spectra (Fig. S4), and XPS patterns (Fig. S5). In Fig. S3, we can see that C, N, V, and O elements are distributed homogeneously all over the whole sample; as depicted in Figs. S4 and S5, there were obvious peaks of VOPc from the VOPc/CN composite, indicating the successful introduction of VOPc molecules into g-C₃N₄.

As we all know, the optical absorption properties of materials have a great influence on their photocatalytic performance, and more strong light absorption usually leads to higher photocatalytic H₂ evolution activity. To investigate the optical properties of VOPc, g-C₃N₄, and the as-prepared 4 wt%-VOPc/CN photocatalysts, the UV-vis diffuse reflectance spectra (DRS) of the samples were recorded using a Shimadzu UV-2510 spectrophotometer. It can be seen from Fig. 1(a) that pristine g-C₃N₄ has an absorption edge of 460 nm, which can be assigned to a band gap of 2.70 eV [9]. Compared to pristine g-C₃N₄, the VOPc/CN hybrid photocatalyst exhibits a much broader absorption band throughout the visible/near-IR light region from 400 to 800 nm due to the introduction of VOPc molecules. The excellent UV-visible light absorption properties of the VOPc/CN hybrid photocatalyst are extremely beneficial to photocatalytic H₂ production. Moreover, it is worth noting that the characteristic absorption band of VOPc in VOPc/CN shows a red shift of about 37 nm relative to that of pure VOPc molecules. The apparent bathochromic shift reflects the efficient interaction between the VOPc and g-C₃N₄, which is known as a π-π stacking interaction [41, 42]. A schematic of the π-π interaction between g-C₃N₄ and VOPc is illustrated in Fig. 1(b). It is well known that VOPc molecules, which have a delocalized π-electron conjugation system, could interact with g-C₃N₄ aromatic frameworks through π-π interactions. On one hand, this non-covalent interaction would not disturb the physical properties of either material; on the other hand, it could promote the separation of the photogenerated hole-electron pairs.

The charge carrier dynamics, especially the charge carrier separation and recombination rates of the photoexcited electron-hole pairs, have a direct influence on the photocatalytic performances of materials [43]. To study the charge carrier dynamics of the photocatalysts, we measured the steady-state PL emission spectra of g-C₃N₄ and the as-prepared VOPc/CN hybrid composite with an excitation wavelength of 315 nm and they are depicted in Fig. 2(a). The PL intensity of the VOPc/CN composite shows an obvious decrease at approximately 450 nm compared to that of pure g-C₃N₄, suggesting that the recombination of photogenerated charge carriers is efficiently suppressed after the incorporation of VOPc, and thus, the non-radiative decay process from the excitation status of g-C₃N₄ is promoted [44]. In addition, time-resolved PL spectroscopy provided additional insight into the charge recombination and the transfer process between VOPc molecules and g-C₃N₄. As shown in Fig. 2(b), the average lifetime of an electron for pure g-C₃N₄ is about 0.31 ns, while the lifetime for the VOPc/CN hybrid photocatalyst is longer (1.11 ns). Generally speaking, the longer lifetime of the electrons means some photoinduced charge carriers undergo rapid transfer between g-C₃N₄ and VOPc, and suppression of electron-hole recombination [45]. Both the steady-state PL quenching and time-resolved PL spectroscopy results confirmed that the introduction of VOPc molecules can efficiently decrease electron-hole recombination; this demonstrates that the VOPc/CN hybrid may be an appealing photocatalyst for H₂ generation.

The transfer and generation of the photoexcited charge carriers were investigated by transient photocurrent responses on a photoelectrochemical test device with visible-light irradiation (λ ≥ 420 nm). As shown in Fig. 2(c), the photocurrent responses of g-C₃N₄ and the VOPc/CN composite were examined under several on/off visible-light irradiation cycles. It is clearly seen that the VOPc/CN composite exhibits a much higher photocurrent than a bare g-C₃N₄ sample, indicating that the VOPc/CN composite possesses enhanced charge separation under visible-light irradiation, resulting in higher photocatalytic performance in water splitting [7]. In addition, EIS of the samples was employed to study the charge transfer resistance. As displayed in Fig. 2(d), the VOPc/CN composite exhibits a smaller semicircle at the intermediate frequency compared to that of pure g-C₃N₄. In general, a semicircle at a high frequency characterizes the process of charge transfer, and a smaller arc radius implies efficient separation of photogenerated electron-hole pairs and a fast interface charge migration process [46]. Obviously, the VOPc/CN composite exhibits a smaller semicircle than that of the pure g-C₃N₄ electrode, implying that introduction of VOPc could make the separation and migration of photogenerated electron-hole pairs more efficient. Taken together, the photoelectrochemical test result is in good accordance with the results of steady-state PL and time-resolved PL, demonstrating that the introduction of VOPc molecules into g-C₃N₄ is favorable for boosting photocatalytic H₂ production.

The photocatalytic activities of the VOPc/CN catalysts were studied through the water reduction reaction for hydrogen evolution under visible-light (λ ≥ 420 nm) illumination using TEOA as the scavenger agent. The H₂ evolution rates of each sample were an average value from three measurements. The results plotted in Fig. 3(a) show that the photocatalytic activity of g-C₃N₄ alone is relatively low (11.21 μmol h^-1), which is probably the result of intrinsically rapid charge recombination [47]. As expected, upon introducing VOPc molecules into g-C₃N₄, an obvious increase in photocatalytic H₂ generation activity was observed, suggesting a positive effect of VOPc molecules on g-C₃N₄ for photocatalytic H₂ production. By increasing the percentage of VOPc, the photocatalytic H₂ evolution rate could be increased gradually. The highest H₂ production rate was 65.52 μmol h^-1 when the content of VOPc molecules was 4 wt%, and this value exceeds that of pure g-C₃N₄ by a factor of 6. The VOPc content has a significant influence on the photocatalytic activity of g-C₃N₄; the hydrogen evolution rate decreases sharply when further increasing the amount of VOPc molecules. For example, when the amount of VOPc was 6 wt%, the average H₂ evolution rate dramatically decreased to 27.08 μmol h^-1. This is likely due to the following. (1) A suitable amount of VOPc could cause strong coupling that facilitates charge transfer, and then promotes the separation of photogenerated electron-hole pairs; excess CBV²⁺ could cover the surface active sites of g-C₃N₄, subsequently lowering interfacial charge transfer and photocatalytic activity [48]. (2) The phenomenon could be ascribed to the "shielding effect" [49]; the existence of the excess VOPc in the VOPc/CN composite will partially block the light absorption of g-C₃N₄ and the hole scavengers, and thus lead to reduced photocatalytic activity for hydrogen evolution. A similar phenomenon has also been found in other previous research for photocatalytic systems [33, 50]. From the BET data in Fig. S6, we can see that the S_BET surface area of the 4 wt%-VOPc/CN composite is 81 m² g^-1, which is smaller than that of g-C₃N₄ (101 m² g^-1). The corresponding pore size distribution of the sample showed almost no change after introducing VOPc molecules. The results imply that the introduction of VOPc molecules has little effect on the corresponding pore size distribution of g-C₃N₄, but negatively affects the specific surface area of g-C₃N₄. Thus, introducing a suitable content of VOPc is very important to optimize the photocatalytic hydrogen production activity of VOPc/CN composites.

AQYs of the 4 wt%-VOPc/CN hybrid for photocatalytic hydrogen evolution were measured under irradiation with different monochromatic light (λ = 420, 450, 500, 550, and 600 ± 5 nm), which is often used to distinguish whether the present H₂ generation process is driven by photoexcited charge carriers or not [51]. From Fig. 3(b), we can observe that the highest AQY of 6.29% was obtained for the 4 wt%-VOPc/CN hybrid at 420 nm. In addition, AQY decreases with an increase in the incident light wavelength, which is consistent with the UV-vis absorption spectrum of VOPc/CN. This revealed that the harvested visible photons dominated the driving force for the present reaction [52]. In addition, the photocatalytic activity of the 4 wt%-VOPc/CN hybrid was compared with those from previous reported works (Table 1) [20, 21, 53]. Notably, the as-prepared VOPc/CN exhibits high photocatalytic activity for H₂ evolution under visible-light irradiation.

The stability as well as the recycling behavior are very important issues for photocatalytic materials in terms of practical application. In order to confirm the persistence of VOPc/CN, we performed cycling tests on 4 wt%-VOPc/CN under the same experimental conditions (Fig. 4). The results show that there is no significant deactivation in terms of H₂ evolution activity after 20 h of visible-light illumination, suggesting that the VOPc/CN hybrid material has favorable stability and can be a promising photocatalyst towards H₂ generation processes.

Based on the above-mentioned experimental results and discussion, a possible mechanism for visible-light-induced H₂ production over the VOPc/CN photocatalyst was proposed and is shown in Scheme 1. The VOPc (Eg = 2 eV) can absorb the photons with energies above 2 eV, which could enhance the light absorption at λ < 460 nm, in addition to harvesting the light at longer wavelengths (λ > 460 nm) [40]. According to previous literature, the LUMO and HOMO positions of VOPc are -1.29 and +0.71 eV, respectively, and the conduction band (CB) and valence band (VB) energy levels of g-C₃N₄ are -1.1 and +1.6 eV, respectively [9, 40, 54, 55]. In the VOPc/CN structure, the LUMO position of VOPc is lower than the CB of g-C₃N₄, while its HOMO value is higher than the VB of g-C₃N₄; therefore, the photoinduced electrons on the LUMO of VOPc can be directly transferred to the CB of g-C₃N₄ and the photogenerated holes in the VB of g-C₃N₄ can spontaneously migrate to the HOMO of VOPc. Therefore, the probability of electron-hole recombination can be reduced because of the well-matched band potentials.

The photogenerated electrons from the intrinsic excitation of g-C₃N₄ and the injected electrons from the excited VOPc molecules are trapped by the loaded Pt co-catalyst for photocatalytic H₂ production. At the same time, the separated holes remaining at VOPc are scavenged by accepted electrons from TEOA, so that the VOPc and g-C₃N₄ can be regenerated for cyclic utilization. As a result, the photogenerated electrons and holes are spatially separated between VOPc and g-C₃N₄, reducing the recombination probability significantly and resulting in the enhanced photocatalytic hydrogen production activity of g-C₃N₄. The VOPc molecules not only serve as a dye sensitizer to enhance the visible light utilization efficiency but also enhance the separation and transfer of photogenerated charges in g-C₃N₄, leading to greatly improved photocatalytic activity for g-C₃N₄.

In summary, we presented a simple method to prepare vanadyl phthalocyanine/graphitic carbon nitride (VOPc/CN) hybrid photocatalysts for H₂ generation in aqueous solution with TEOA as a sacrificial electron donor. The VOPc molecules can easily hybridize with g-C₃N₄ through strong π-π stacking interactions, which do not have much influence on the electronic properties of g-C₃N₄. The introduction of VOPc molecules can extend the spectral response region to longer than 460 nm; at the same time, it can retard charge recombination, thus resulting in significantly improved photoactivity for H₂ production. When the amount of VOPc molecules was 4 wt%, the developed VOPc/CN hybrid photocatalyst exhibited the highest photocatalytic activity, with a H₂ evolution rate of 65.52 μmol h^-1, which is nearly 6 times higher than that of bare g-C₃N₄ (11.20 μmol h^-1), and a high AQY of 6.29% at 420 nm. This study offers a way to develop a new class of visible-light-driven hybrid materials with high photocatalytic performance.