In situ constructed oxygen-vacancy-rich MoO3−x/porous g-C3N4 heterojunction for synergistically enhanced photocatalytic H2 evolution

A simple method was developed for enhanced synergistic photocatalytic hydrogen evolution by in situ constructing of oxygen-vacancy-rich MoO3−x/porous g-C3N4 heterojunctions. Introduction of a MoO3−x precursor (Mo(OH)6) solution into g-C3N4 nanosheets helped to form a porous structure, and nano-sized oxygen-vacancy-rich MoO3−xin situ grew and formed a heterojunction with g-C3N4, favorable for charge separation and photocatalytic hydrogen evolution (HER). Optimizing the content of the MoO3−x precursor in the composite leads to a maximum photocatalytic H2 evolution rate of 4694.3 μmol g−1 h−1, which is approximately 4 times higher of that of pure g-C3N4 (1220.1 μmol g−1 h−1). The presence of oxygen vacancies (OVs) could give rise to electron-rich metal sites. High porosity induced more active sites on the pores' edges. Both synergistically enhanced the photocatalytic HER performance. Our study not only presented a facile method to form nano-sized heterojunctions, but also to introduce more active sites by high porosity and efficient charge separation from OVs.


Introduction
Photocatalytic hydrogen evolution (HER) is considered to be one of the most promising ways to alleviate the environment and energy crisis. [1][2][3][4][5][6] However, exploring highly efficient and environmentally-friendly catalysts remains a challenge for the current photocatalysis eld. 7 The efficiency of photocatalytic HER lies in three main aspects: (i) the generation and separation of photo-generated charges, (ii) the migration distance of carriers, and (iii) the oxidation-reduction reaction on the surface of photocatalysts. 8 Various strategies have been proposed to improve the photocatalytic efficiency, with a broad range of photocatalysts. 7 The prevailing photocatalysts include metal oxides, 9 metal suldes, 10 metal nitrides, 11 organometallic complexes, 12 and metal-free semiconductors. [13][14][15][16][17] Among the various photocatalysts, g-C 3 N 4 , a metal-free semiconductor, has attracted extensive attention in photocatalytic HER due to its outstanding characteristics, such as appropriate band edge, environmental friendliness, high thermal and chemical stability, facile fabrication and cost-effectiveness. 18,19 However, the photocatalytic HER efficiency of g-C 3 N 4 is still far from satisfactory, due to the low specic surface area, limited active sites and fast recombination rate of photogenerated electron-hole (e À -h + ) charges. 20 g-C 3 N 4 nanosheets were reported previously with better performance than bulk g-C 3 N 4 with increased surface area, higher photogenerated electron reduction potential, better electron transport capacity and longer lifetime etc. Our strategy has designed a more efficient nano-junction based on g-C 3 N 4 nanosheets.
Based on the review, various strategies have been proposed based on g-C 3 N 4 , including electronic structure modulation, crystal structure engineering, nanostructure and heterostructure construction. 8,[21][22][23][24][25] The construction of g-C 3 N 4 -based heterojunction can help the separation of charges at the interface, as a variety of g-C 3 N 4 -based heterojunctions has been reported, such as TiO 2 /g-C 3 N 4 , 26,27 ZnO/g-C 3 N 4 , 28,29 WO 3 /g-C 3 N 4 , 30 WS 2 /g-C 3 N 4 , 31 NiO/g-C 3 N 4 , 32 ZnIn 2 S 4 /g-C 3 N 4 , 33 Zn x Cd 1Àx In 2 S 4 /g-C 3 N 4 , 34 Bi 2 Se 3 /g-C 3 N 4 , 35 MoO 3 /1T-MoS 2 /g-C 3 N 4 , 36 etc. MoO 3 has a large band gap of 3 eV and a high dielectric constant of 6-18, suitable for constructing heterojunction photocatalysis with g-C 3 N 4 . 37-39 Oxygen-vacancy-rich MoO 3Àx was even more favourable. The synthesis of porous g-C 3 N 4 and few-layered MoO 3 has been widely reported, however, the combination of porous g-C 3 N 4 and oxygen-vacancy-rich MoO 3Àx to construct MoO 3Àx /g-C 3 N 4 photocatalytic system remains a great challenge. The introduction of porous structure in g-C 3 N 4 usually produces many photocatalytic active sites. 40 Xiao et al. introduced a bottom-up method for preparing porous few-layer g-C 3 N 4 by a sequential molecule self-assembly, alcohol molecules intercalation, thermal-induced exfoliation and polycondensation process. 40 Moreover, conventional silicon dioxide (SiO 2 ) template or ammonium bicarbonate (NH 4 HCO 3 ) has also been used as pore forming agent to prepare porous g-C 3 N 4 in the process of polycondensation. 41,42 For the construction of oxygen vacancies, metal oxides with oxygen vacancies are usually obtained by using hydrogen thermal treatment, 43 high energy particle bombardment, 44 and heating metal oxides under vacuum or oxygen depleted conditions. 45 Although these methods have their own advantages, they are inevitably limited by high-energy input and complex post-processing. According to the literature, there are electrostatic composite and in situ construction method. 41,46 The former is simple to implement, but it is greatly affected by the pH of photocatalytic reaction. Therefore, it is urgent to develop a new method to integrate porous structure and oxygen vacancy into a heterojunction system.
Herein, we propose a salt assisted in situ growth method to construct nano-sized oxygen-vacancy-rich MoO 3Àx /porous g-C 3 N 4 heterojunction. The introduction of MoO 3Àx precursor (Mo(OH) 6 ) into g-C 3 N 4 not only helps to prepare porous nanosheets, but also in situ grown oxygen-vacancy-rich MoO 3Àx can form an atomic-scale compact heterointerface with g-C 3 N 4 . In additional, the heterojunction between MoO 3Àx and g-C 3 N 4 was established, which synergically improved photocatalytic HER performance, including higher charge separation efficiency, shorter charge transport path, and more catalytic active sites. By HR-TEM, HAADF-STEM and AFM, the morphology of the nanosized MoO 3Àx /porous g-C 3 N 4 heterojunction was conrmed. XPS proved the oxygen-vacancy in MoO 3Àx . As a result, the optimized MoO 3Àx /g-C 3 N 4 heterojunction produces H 2 at 4694.3 mmol g À1 h À1 , approximately 4 times higher of that of pure g-C 3 N 4 (1220.1 mmol g À1 h À1 ). Our strategy can be universal for constructing diverse heterojunction with OVs and porous characteristics.

Materials and reagents
All chemical reagents and materials were purchased and used without further purication. Molybdenum powder (Mo, 98%), urea ((NH 2 ) 2 CO, $98%) were purchased from Sigma. Sodium chloride (NaCl, AR), absolute ethanol (CH 3 CH 2 OH, AR), and triethanolamine (TEOA, AR) were purchased from Sinopharm Chemical Reagent Co., Ltd. All chemicals were analytical grade and used without further purication.
2.2. Preparation of g-C 3 N 4 g-C 3 N 4 was synthesized by thermal condensation of urea directly. In detail, 10 g urea powder was placed in an alumina crucible and heated to 550 C for 2 h in static air with a heating rate of 5 C min À1 . The resulting yellow agglomerates were then collected and milled into powder for further synthesis and measurements.

Preparation of Mo(OH) 6 precursor
Typically, 0.1 g molybdenum (Mo) powder was dispersed in 10 mL ethanol with stirring for several minutes. Then 0.35 mL H 2 O 2 (30%) solution was added into the Mo power suspension solution. Aer 18 h, the Mo oxide solution turned from grey to yellow and nally turned to blue.

Preparation of MoO 3
The 10 mL Mo(OH) 6 precursor was dried in a 60 C vacuum oven for 30 min, then transferred to a tubular furnace and calcined at 400 C for 2 h (5 C min À1 , under nitrogen condition).

Preparation of MoO 3Àx
The Mo(OH) 6 precursor was added into a certain amount of NaCl, stirred well, and dried on the heating mantle at 80 C. Then, transferred to a quartz tube furnace, to anneal for 2 h at 280 C (5 C min À1 , under nitrogen condition). The sample was dissolved in deionized water, and then ltered, washed and dried to obtain solid powder.

Preparation of MoO 3Àx /porous g-C 3 N 4 nanosheets
A typical fabrication process as follows: 50 mg g-C 3 N 4 was dissolved in 200 mL of anhydrous ethanol and dispersed it evenly by ultrasonic treatment. 200 g NaCl prepared beforehand was added to the solution, followed by stirring, and the ethanol was removed by placing it on a heating mantle at 80 C, the obtained sample was annealed at 300 C for 2 h in a quartz tube furnace (5 C min À1 , under nitrogen condition). The Mo(OH) 6 precursor was added into the above g-C 3 N 4 /NaCl mixture, stirred well, and dried on the heating table at 80 C. The dried powder was transferred to a quartz tube furnace, to anneal for 2 h at 280 C (5 C min À1 , under nitrogen condition). The obtained sample was dissolved in deionized water, ltered, washed and dried to obtain solid powder. 1%, 5%, 10%, 50% MoO 3Àx /g-C 3 N 4 samples were prepared by adjusting the mass ratio of Mo powder to g-C 3 N 4 . NaCl acts as a removable template, provides a place for the construction of heterojunction in situ, and plays a good role in dispersing g-C 3 N 4 precursor, so that Mo(OH) 6 can better contact with g-C 3 N 4 . As an antisolvent, ethanol does not dissolve NaCl and can be quickly removed from the system under 80 C heating, so it acts as a medium for mixing the two materials.

Characterization
The phase structures of samples were recorded by X-ray powder diffraction (XRD, BrokerAXS Germany). Chemical state analysis was performed by using X-ray photoelectron spectroscopy (XPS, ESCALab 250). FTIR spectrometer (Nicolet iS50) was applied to investigate the functional groups and chemical structure of samples. Scanning electron microscope (SEM, HITACHI SU8010 Hitachi Japan), transmission electron microscope (TEM, Tecnai G20) and atomic force microscope (AFM, BRUKER MultiMode 8) were conducted to conrm the morphology and thickness of samples. Ultraviolet-visible (UV-vis) spectrophotometer (Shimadzu instruments Ltd., Suzhou, UV-2600) and Spectrouorometer (Edinburgh Instruments Ltd., FS5) measurements were performed to study optical properties of samples. Photocurrent response, EIS Nyquist plots and Mott-Schottky measurements were conducted to comprehend electrochemical properties with the help of CHI 760E electrochemical system (Chenhua, Shanghai, China).

Photoelectrochemical measurement
Photoelectrochemical experiments were carried out on the CHI 760E electrochemical workstation using a traditional threeelectrode system. The Ag/AgCl electrode was used as the reference electrode, Pt electrode was used as the counter electrode, and photocatalyst/ITO was used as working electrode. The detailed preparation of working electrode is as follows: rstly, 10 mg catalyst was dispersed in 2 mL DI water, followed by another ultra-sonication for half an hour to form a uniform suspension. Secondly, the suspension was dropped onto the surface of the ITO glass. Finally, the ITO glass was placed under an infrared lamp for drying. 0.1 mol L À1 sodium sulfate solution and a 300 W xenon lamp were used as electrolyte solution and light source respectively, in the photocurrent responses and electrochemical impedance spectra experiments. In Mott-Schottky measurement, the frequency is 1000 Hz, the potential range is À1 to 1 V, and the electrolyte solution is 0.5 mol L À1 sodium sulfate solution.

Photocatalytic H 2 evolution test
The photocatalytic H 2 evolution experiments were carried out in a 200 mL Pyrex reaction cell connected to a glass closed gas circulation system with vacuum (Labsolar-III (AG), Perfectlight Technology Co. Ltd, Beijing, P. R. China). In a photocatalytic reaction, 50 mg as-prepared photocatalyst powder was suspended in a glass cylindrical reactor, then, added to 90 mL deionized water, and stirred for a while. 10 mL triethanolamine (TEOA) was added in the above suspension as a sacricial agent. The photocatalytic reaction was performed using a 300 W xenon lamp (l ¼ 320-780 nm). Before irradiation, the whole system maintains a vacuum to avoid contact with air. H 2 PtCl 6 aqueous solution was added as the precursor for the co-catalyst Pt, which was in situ photo-reduced during the photocatalytic reaction (2 wt% Pt). The amount of H 2 produced during the photocatalytic reaction was determined by a gas chromatography (Techcomp, GC7900) (TCD detector, 5Å molecular sieve column) for every one hour.

Results and discussion
3.1. Synthesis of nano-sized oxygen-vacancy-rich MoO 3Àx / porous g-C 3 N 4 heterojunction by salt assisted in situ growth method We schematically illustrated the method in Fig. 1: rstly, g-C 3 N 4 nanosheets were synthesized as the method reported and dispersed in anhydrous ethanol by ultrasonication, followed by adding NaCl powder. Aer the removal of ethanol by annealing at 80 C, g-C 3 N 4 nanosheets/NaCl was obtained (step 1). Secondly, the ethanol solution of Mo(OH) 6 as precursor was mixed with the prepared g-C 3 N 4 nanosheets/NaCl, annealing at 280 C in a CVD furnace to obtain MoO 3Àx /g-C 3 N 4 /NaCl (step 2). Washing with deionized water several times to remove NaCl, the MoO 3Àx /g-C 3 N 4 heterojunction material was obtained (step 3). In addition, by changing the mass ratio of Mo powder to g-C 3 N 4 (1%, 5%, 10%, and 50%), MoO 3Àx /g-C 3 N 4 samples with different Mo percentage were also prepared. The amorphous structure of prepared MoO 3Àx /g-C 3 N 4 nanosheets was conrmed by X-ray diffraction (XRD) analysis as shown in Fig. S1. † Fig. S2 † shows the XRD pattern of pure a-MoO 3 fabricated from Mo(OH) 6 precursor.

The formation of porous g-C 3 N 4 nanosheets
FT-IR was applied to analyse the chemical structures of g-C 3 N 4 nanosheets and 5% MoO 3Àx /g-C 3 N 4 (Fig. 3a). A new vibration band at 2178 cm À1 in MoO 3Àx /g-C 3 N 4 nanosheets, attributed to C^N bond, indicated the deprotonation of the amino group (-NH 2 ) in g-C 3 N 4 nanosheets. 8 On the other side, NH 3 was released in the deprotonation process, and led to the formation of pores.
To further investigate the chemical states and elemental composition of the samples, XPS and elemental analysis measurements were performed, respectively. As shown Fig. 3b, the XPS spectrum of 5% MoO 3Àx /g-C 3 N 4 nanosheets exhibited C 1s and N 1s signals at the position same as that of g-C 3 N 4 nanosheets. In Fig. 3c, the C 1s peak of g-C 3 N 4 was tted into two peaks at 284.8, and 288.0 eV, assigned to graphitic C-C, and sp 2 -hybridized carbon in N containing aromatic ring (N-C]N), respectively. 47 The C 1s peak of MoO 3Àx /g-C 3 N 4 (5%) nanosheets slightly shied from 288.0 eV to 288.2 eV, attributed to the interaction between -NH 2 and HO-Mo(OH) 5 . 8 In Fig. 3d, the N 1s peak for g-C 3 N 4 could be deconvoluted into four peaks at 398.7, 399.9, 401.0 and 404.4 eV, attributed to the sp 2 -bonded N in the tri-s-triazine units (C]N-C), bridging nitrogen atoms in N-(C) 3 , nitrogen atoms bonded with hydrogen atoms (-NH x ) and the charging effects, respectively. 47 Compared with g-C 3 N 4 nanosheets, MoO 3Àx /g-C 3 N 4 (5%) nanosheets showed obvious shi of N-(C) 3 peaks (from 399.9 eV to 399.6 eV), suggesting cyano-groups and consistent with the FTIR results. 48 In addition, the introduction of N defects can be supported by elemental analysis measurements as showed in Table S1, † supporting NH 3 releasing. The N/C atomic ratio for g-C 3 N 4 is 1.32, close to the ideal g-C 3 N 4 composition (N/C ¼ 1.33), 49 while the N/C atomic ratios for MoO 3Àx /g-C 3 N 4 (5%) decrease to 1.02, suggesting the loss of lattice nitrogen in MoO 3Àx /g-C 3 N 4 (5%) due to the production of NH 3 .

The formation of oxygen vacancies on MoO 3Àx
As a reference, we synthesized MoO 3 by the reported method. We compared the XPS spectra of the MoO 3 and 5% MoO 3Àx /g-C 3 N 4 in Fig. 4a and b. For MoO 3 , the O 1s peaks at 530.5, 532.2, and 533.0 eV, are due to Mo-O, oxygen vacancy, and -OH, respectively. 50 For MoO 3Àx /g-C 3 N 4 (5%) nanosheets, the O 1s peaks from Mo-O and -OH positively shied 0.6 and 1.6 eV, respectively, attributed to the interaction between the -OH of Mo(OH) 6 and the -NH 2 of g-C 3 N 4 . In order to further obtain more details information of oxygen vacancy, the ratios of different O species of MoO 3 and MoO 3Àx /g-C 3 N 4 (5%) calculated by XPS, the analysis results are shown in Tables S2 and S3. † The concentration of oxygen vacancy increased from 21.84% to 48.67% in MoO 3 and MoO 3Àx /g-C 3 N 4 (5%), respectively. By the O 1s spectra comparison, the ratio of Mo-O decreases and the ratio of -OH increases, suggesting the unsaturation of Mo coordination. In MoO 3Àx /g-C 3 N 4 (5%) composite, the broadened doublets at 231.7 and 234.9 eV are ascribed to the electron binding energies of the Mo 3d 3/2 and Mo 3d 5/2 of Mo 5+ (Fig. 3b). The existence of the Mo 5+ may come from the oxygen vacancies and unsaturated coordination of Mo. 51,52

Photocatalytic H 2 evolution
Fig. 5a exhibits HER rates of the as-prepared samples all with 2 wt% Pt as co-catalyst and TEOA as hole-sacricial agent. The 5% MoO 3Àx /g-C 3 N 4 heterojunction exhibits the highest HER activity as the optimized condition and lower or higher MoO 3Àx percentage induced less HER activity. With the further   increasing MoO 3Àx , excess MoO 3Àx has a tendency to selfaggregate, reducing effective interfacial area between MoO 3Àx and g-C 3 N 4 and the active sites. So, the overmounted MoO 3Àx led to lower performance. Fig. 5b shows that the high HER stability of 5% MoO 3Àx /g-C 3 N 4 nanosheets, with no obvious decrease during 3 cycles. Fig. 5c compared the HER rate with different percentage of MoO 3Àx in MoO 3Àx /g-C 3 N 4 nanosheets. The 5% MoO 3Àx /g-C 3 N 4 nanosheets exhibits the highest HER rate (4694.3 mmol g À1 h À1 ), about 4 times higher than pure g-C 3 N 4 (1220.1 mmol g À1 h À1 ). Fig. 5d compared the HER rates of the pure g-C 3 N 4 and 5% MoO 3Àx /g-C 3 N 4 nanosheets using lactic acid and TEOA as hole-sacricial agents, respectively. MoO 3Àx / g-C 3 N 4 (5%) nanosheets exhibited higher HER rates in both systems of lactic acid and TEOA as hole-sacricial agents. Photocatalytic hydrogen evolution rates of g-C 3 N 4 , 5% MoO 3 /g-C 3 N 4 and 5% MoO 3Àx /g-C 3 N 4 as shown in Fig. S6. † 5% MoO 3Àx / g-C 3 N 4 showed the best HER performance, which further proved the promoting effect of OVs in MoO 3Àx /g-C 3 N 4 photocatalyst on HER. We also compared the HER performance of similar photocatalysts reported in some recent references as shown in Table S4. †

The electronic structure and photocatalytic mechanism
The optical absorption properties and band gap of the heterojunction were studied by UV-vis absorption spectroscopy. The g-C 3 N 4 shows an absorption edge of approximately 435 nm (Fig. 6a), consistent with the ref. 53. The 5% MoO 3Àx /g-C 3 N 4 heterojunction exhibits a red-shi of absorption edge, given to the interfacial charge transfer (Fig. 6a). UV-vis absorption spectra of g-C 3 N 4 and MoO 3Àx is shown in Fig. S7, † the illustrations are their corresponding Tauc's plot. In Fig. 6b, the pure g-C 3 N 4 nanosheets in ethanol exhibited a strong PL peak at 450 nm, while 5% MoO 3Àx /g-C 3 N 4 nanosheets with the same concentration of g-C 3 N 4 nanosheets in ethanol showed much smaller PL peak at 450 nm. The PL peak at 350 nm was attributed to MoO 3Àx (excitation wavelength: 300 nm). These results suggest that, in the heterojunction, the MoO 3Àx absorbed the excitation energy and transferred some part to g-C 3 N 4 . In Fig. 6c, the 5% MoO 3Àx /g-C 3 N 4 exhibit higher photocurrent density than g-C 3 N 4 . Compared with pure g-C 3 N 4 , 5% MoO 3Àx / g-C 3 N 4 nanosheets have stronger photocurrent density as well as smaller arc radius (Fig. 6d), leading to lower resistance and favourable for charge separation. The corresponding optical band-gaps (E g ) were calculated to be 2.69 eV for g-C 3 N 4 and 2.98 eV for MoO 3Àx (Fig. S7a and b †). In Mott-Schottky plots (Fig. 6e), the extrapolated conduction band edge (E CB ) positions of g-C 3 N 4 and MoO 3Àx are À1.23 and À0.20 V (vs. Ag/AgCl, pH ¼ 7) or À0.59 and +0.44 V (vs. NHE, pH ¼ 0), respectively. Regarding E g of g-C 3 N 4 (2.69 eV) and MoO 3Àx (2.98 eV) nanosheets, the valence band energy (E VB ) was calculated as 2.10 and 3.42 eV (vs. NHE, pH ¼ 0) as shown in Fig. 6f. According to the step-scheme photocatalytic system, 30 the e À on the CB of MoO 3Àx would recombine with the h + of VB g-C 3 N 4 at the interface in the MoO 3Àx /g-C 3 N 4 . Consequently, e À accumulated in the CB of g-C 3 N 4 and h + accumulated in the VB of MoO 3Àx . H + The photocatalytic H 2 evolution of MoO 3Àx /g-C 3 N 4 samples with 2 wt% Pt as co-catalyst. (b) Photocatalytic cycle stability test of 5% MoO 3Àx /g-C 3 N 4 with 2 wt% Pt as co-catalyst. (c) Optimum photocatalytic H 2 evolution efficiency of MoO 3Àx /g-C 3 N 4 photocatalysts with different contents of MoO 3Àx . (d) The photocatalytic hydrogen evolution performance for g-C 3 N 4 and 5% MoO 3Àx /g-C 3 N 4 with 2 wt% Pt as co-catalyst in different sacrificial agent systems (lactic acid and TEOA). was reduced to hydrogen by accumulated electrons at the CB of g-C 3 N 4 , while the TEOA was oxidized by the h + accumulated at the VB of MoO 3Àx .

Conclusions
In this work, we have constructed oxygen-vacancy-rich MoO 3Àx / porous g-C 3 N 4 nanosheets heterojunction by in situ growth. The heterojunction photocatalysts under light (l ¼ 320-780 nm) accelerated photogenerated e À -h + separation and hydrogen evolution reaction. This method is simple and feasible, which endows g-C 3 N 4 nanosheets with highly porous structure and more active sites, and endows MoO 3Àx with oxygen-vacancy-rich MoO 3Àx and more efficient charge separation, synergistically enhancing the photocatalytic performance. It could be a general method to inorganic semiconductor heterojunction or even multi-junction for photocatalysts.

Conflicts of interest
There are no conicts to declare.