As the dramatic increase in the development of industry proceeds with overuse of fossil fuels, environmental problems and energy crises may become ever more serious. To help alleviate this situation, water splitting using semiconducting photocatalysts has been shown, since the discovery of the Honda-Fujishima effect, to be a promising method to generate H2, as an environmentally friendly energy source, directly from the sun [1].
A new type of metal-free semiconductor material, g-C3N4, has received attention because of its non-toxicity, visible light response, and thermal-chemical stability [2-5]. Its low cost and simple synthesis make g-C3N4 easily obtainable from calcining melamine, dicyandiamide, urea, or other raw materials [6-8]. However, because of the high binding energy of g-C3N4 and its low crystallinity, it is difficult to separate the photogenerated electron-hole pair and it suffers from fast recombination of photoexcited charge carriers [9, 10]. Researchers have been working on modifications to g-C3N4, and several metal-containing carbon nitride compounds have been reported to greatly improve the performance of g-C3N4 under visible light [2, 4, 5, 11]. However, such doping modification cannot solve the problem of recombination. To solve that, coupling g-C3N4 with other materials to form a composite structure shows some promise [12]. Examples include CdS/g-C3N4 [13], TiO2/g-C3N4 [14], WO3/g-C3N4 [15], g-C3N4/Ag3PO4 [16], Cd0.5Zn0.5S/g-C3N4 [17], and CoTiO3/g-C3N4 [18], but all of these works focused only on enhancing H2 evolution and no O2 was observed.
So called "Z-scheme" principled photocatalyst has stirred the interest of scientists because of its charge transport mechanism, similar to the natural green plant photosynthesis, and its Z-scheme charge transfer path. In this system, two different photocatalysts are combined using an appropriate shuttle redox mediator [19]. Its unique energy gap, structure, and charge transfer mode make the electrons and holes separately spaced [20]. The Z-scheme method also makes it possible to combine photocatalysts for either water reduction or oxidation potential to the system [19]. For example, BiVO4, whose conduction band is 0.3 V versus NHE, and which has no ability to produce H2, is capable of oxidizing O2– from a solution containing appropriate electron acceptors, as will be discussed vide infra. Some work to date has focused on using BiVO4 to reinforce the oxidation capacity of g-C3N4 [21-25], whose conduction band (CB) is –1.42 V and valence band (VB) is 1.25 V versus NHE. However, most of these studies only formed a heterojunction that results in a more negative VB and more positive CB, which is undesirable for water splitting.
Herein, we report on the overall water splitting using Zn-doped g-C3N4 for H2 evolution and BiVO4 photocatalyst for O2, with Fe3+/Fe2+ redox couple as an electron mediator. Our photocatalytic experiments indicate that g-C3N4 can be used in a typical Z-scheme water splitting system with both H2 and O2 gases evolved in a stoichiometric ratio (H2/O2 ≈ 2).
Urea (> 99.0%), Ethanol (99.7%), NH4VO3 (> 99%), ZnCl2 (> 98.0%), HCl (> 36.0% to aqueous solution), dicyandiamide (> 99%), Bi(NO3)3·5H2O (> 98%), HNO3 (65%–68%), NH3·H2O (28% to aqueous solution), and H2PtCl6·6H2O were purchased from Sinopharm Chemical Reagent Co., Ltd. (China). FeCl2 (> 98%) was purchased from Shanghai Aladdin Biochemical Technology Co., Ltd. (China). All reagents were used without further purification.
The photocatalyst of g-C3N4 was synthesized according to the procedure presented in Ref. [26]. Typically, 16 g urea was heated to 873 K at a heating rate of 2.5 K/min in an alumina crucible with aluminum foil as a cover. After naturally cooling to room temperature, the light-yellow-colored g-C3N4 was obtained without further treatment.
The as-prepared g-C3N4 (1 g) was then mixed with 10 mL of ethanol and different proportions of ZnCl2 (e.g. 0.1 g ZnCl2 for CN-10-U, where U represents use of urea as the precursor) were added to the mixed solution. A few drops of HCl (36%) were added to avoid hydrolysis of ZnCl2. This solution was heated and stirred at 353 K to remove ethanol. It was then heated to 673 K at a rate of 5 K/min under nitrogen flow and kept at this temperature for 4 h. Zinc-containing g-C3N4 was then produced after naturally cooling in a nitrogen atmosphere.
Different methods have been used to realize Zn doping. Wang's group [4] used dicyandiamide and ZnCl2 as precursors, combined with a polycondensation process at elevated temperatures. Meanwhile, Ye's group [22] mixed as-prepared g-C3N4 with ZnCl2 and then dried this under nitrogen flow. In the Wang and Ye work, the materials used are the same and the only difference is the synthetic method employed. Our research uses the same synthetic method as Ye's work except that we use urea instead of dicyandiamide. Three different types of g-C3N4 doped with zinc (10%) were synthesized and denoted as CN-10-D2, CN-10-D1, and CN-10-U.
BiVO4 catalyst was synthesized using a hydrothermal method [27] as follows. First, 3 mmol Bi(NO3)3·5H2O and 3 mmol NH4VO3 were dissolved in 40 mL HNO3 (2 mol/L) with vigorous stirring. The pH was then adjusted to 2.0 using ammonia and dilute nitric acid to form a yellowish suspension that was kept under stirring for 1 h. Then, the resulting suspension was transferred into a Teflon recipient, which was performed at 473 K for 24 h.
X-ray diffraction (XRD) patterns of the as-prepared samples were measured on a D8 DA VINCE (Bruker) X-ray diffractometer using Cu Kα radiation under 40 kV and 40 mA. The scanning speed was 6°/min and 2θ = 10°–80°. The morphologies of the samples were revealed by scanning electron microscopy (SEM, JEOL JSM-6380LV). The XPS patterns were measured on an AXIS UltraDLD electronic energy spectrometer (Kratos group) at 300 W using Mg Kα X-rays as the excitation source.
The photocatalytic activities were evaluated in a Pyrex reaction cell connected to a closed gas circulation and evacuation system. Particulate Pt was photodeposited onto the photocatalyst samples as a water reduction cocatalyst as the reaction progressed. Specifically, 1.8 g FeCl2 was dissolved in 80 mL deionized water and 1 wt% Pt (1 mg Pt for 0.1g CN-x-U, for example) was added into this solution. 0.1 g of CN-x-U (x = 0, 5, 10, or 20) and 0.05 g of BiVO4 was then suspended in the solution. The suspension was thoroughly degassed and irradiated by a 300 W Xe-arc lamp. The amount of H2 and O2 was analyzed every hour using online gas chromatography.
The photodeposition method was also used to load Pt co-catalyst onto g-C3N4 doped with Zinc. An aqueous solution of H2PtCl6 (containing 1 wt% Pt) was added to 80 mL aqueous solution suspended in the CN-10-U powder, and then the solution was irradiated using a Xe lamp (300 W). The Pt cocatalyst was deposited in metallic form according to previous descriptions [28].
The diffraction patterns for the Zn-doped g-C3N4 with different amounts of zinc, as can be seen in Fig. 1(a), shows two peaks at 13.4° and 27.4°, corresponding to the (100) and (002) diffraction planes, respectively [3, 21, 29]. The main peak at 27.4° can be indexed as the distance of 0.681 nm between nitride pores in C3N4, which indicates a two-dimensional carbon nitride structure both in the modified and unmodified samples [2, 4, 22]. Its intensity gradually decreases as the Zn content increases, and the same trend can be observed in the peak at 13.4°. This peak can be ascribed to the interplanar distance of 0.326 nm, and this may indicate that there is a deterioration of crystallinity upon Zn doping and it may strongly impact the distance between g-C3N4 hosts [2, 23]. No heterogeneous phase, such as zinc, its oxide or chloride, was observed from the XRD data [5]. The phase and purity of the synthesized BiVO4 can be observed from Fig. 1(b), which were consistent with JCPDS card No. 14-0688 (pure phase monoclinic point group).
For the purpose of studying the optical absorption of the as-prepared g-C3N4 based samples, the UV-Vis absorption spectra are provided in Fig. 2. The light absorption edges of the CN-x-U with different Zn contents gave a small change to the short wavelength range compared to the pristine g-C3N4 (the maximum light absorption edge was obtained with the CN-10-U samples).
Fig. 3 shows the changes in photocatalytic activity as various amounts of Zn was doped into g-C3N4. It has been revealed previously that the hydrogen yield is a non-monotonic function of the Zn content [11]; therefore, the overall water splitting was only obtained when 10% of Zn was loaded. This can be explained from when there is too much doped Zn, they may unite as reaction sites for the recombination of photoelectric electron-hole pairs, resulting in a decrease in the redox ability of g-C3N4 [20].
To make sure that the CN-10-U was combined with BiVO4 through a shuttle redox mediator, Fig. 4 shows the photocatalytic activity of the different components in aqueous FeCl2 solution. Overall water splitting (H2 and O2 gases generated in a stoichiometric ratio) was only observed when both CN-10-U and BiVO4 was added to the aqueous FeCl2 solution with a certain amount of H2PtCl4.
However, there is another theory which purports that Pt nanoparticles would be formed and play a key role in overall water splitting when H2PtCl6 is added to the original solution and deposited as the reaction progresses [30]. To ascertain whether the overall water splitting actually contributed to g-C3N4 doped with zinc and BiVO4, pre-photodeposed g-C3N4 was tested and the results can be seen in Fig. 5. As the picture shows, stoichiometric H2 and O2 gases can be observed in the Z-scheme system built by both pre-photodeposed (CN-10(Pt)/BiVO4) and when it is photodeposed during the reaction (CN-10/Pt/BiVO4). The formation of a complex is supported by TEM images (Fig. 6(b)) of the CN-10/Pt example.
These results indicate that CN-10-U can be applied to a typical Z-scheme water splitting system as H2-evolving photocatalysts via a shuttle redox mediator. A typical SEM image of the suspension of CN-10-U and BiVO4 (2:1; 12 h), as illustrated in Fig. 6(a), also supports this. CN-10-U and BiVO4 are separated in this mixture without any obvious form of heterojunction.
The stability of the as-prepared Z-scheme system can be observed in Fig. 7. We can see that the system retains a stable stoichiometric H2 and O2 production over a long time period. Under visible light the H2 production was very low, and this may be due to the poor activity of g-C3N4 decomposed by urea, which was also shown by the UV DRS data (Fig. 2). There is no obvious difference in O2 production over the initial two hours, which can be attributed to the high capability of BiVO4 under visible light. However, after six hours of radiation, the production of O2 using aqueous NaNO2 as a light filter slowly decreased. This can be explained by the low production of H2, which influences the interchange of the Fe3+/Fe2+ redox couple. The low production of H2 might also lead to an accumulation of ferrous ion, a potential competitor to the oxidation reaction.
Thus, we obtained a stable Z-scheme system for overall water splitting by combining Zn-doped (10%) g-C3N4 and BiVO4 though a Fe3+/Fe2+ redox couple. Pt was used as a cocatalyst for H2 revolution.
According to previous research [4, 11], there are two different ways to realize doping with Zn. Specifically, Zn-doped g-C3N4 can be synthesized by an impregnation method (CN-10-U and CN-10-D1) or by in-situ synthesis (CN-10-D2). These three types of Z-scheme were constructed and tested, with the results shown in Fig. 8. Overall water splitting can be observed in the Z-scheme using both CN-10-U and CN-10-D1. The difference in the production between CN-10-U and CN-10-D1 may contribute to the introduction of pores in g-C3N4 by using urea as the precursor. The high surface area and continuous porosity, which can be active centers, are important requirements for catalysis [12].
The chemical state and surface chemical composition were characterized by XPS (Fig. 9). There are significant differences in the C 1s and N 1s peaks for Zn-doped g-C3N4 synthesized by the impregnation method (CN-10-D1) relative to the in-situ synthesized CN-10-D2, as discussed vide infra. A Zn 2p3/2 binding energy (BE) peak at 1020.3 eV in both Zn-doped samples is revealed in Fig. 9(d). This is lower than the 1021.9 eV value measured for Zn(Ⅱ) in ZnCl2, and there are no Cl peaks detected in either sample. Hence, it can be speculated that Zn(Ⅱ) was successfully linked to the g-C3N4 framework in both samples through a Zn-N bond without Cl– ions [11]. The reason for the different catalytic ability between the two types of Zn-doped g-C3N4 may be ascertained from the C 1s (Fig. 9(a)) and N 1s (Fig. 9(b)) peaks.
The C 1s and N 1s peaks of the two samples in Fig. 10 are both clear, but the ways in which they were modified by zinc differ. In Fig. 10(a) and (b), the C 1s peaks can be convoluted as several binding energies. The main peak at 287.7 eV (C1) corresponds to the triazine ring of the N=C–N group [4, 30, 31]. The peak at 285.4 eV (C3) can be assigned to the sp3 C–N bond, and the peak at 284.2 eV (C2) corresponds to the sp2 C=C bond. The weak peak at 289.0 eV (C4) can be attributed to C–O groups caused by inevitable oxidation [11]. Zn-doped g-C3N4 obtained by in-situ synthesis shows a significantly weaker C2 peak than that obtained via impregnation. For the N 1s signal, a similar trend is observed, as shown in Fig. 8(c) and (d). The N 1s XPS peak can also be deconvoluted into three typical peaks. The dominant peak at 398.2 eV (N1) can be attributed to the C–N=C group, which is a manifestation of the triazine ring. The peak at 399.3 (N2) and 400.5 eV (N3) can be assigned, respectively, to the –N– and ≡N bonds [32]. Compared to CN-10-D1, the integral strength of the N 1s signal of the CN-10-D2 sample is reduced. From these observations, we conclude that Zn might be doped by forming Zn–N bonds, which promote carrier transport from Zn to the g-C3N4 bodies [4, 31, 33]. Also, by using the impregnation method, Zn can be combined more homogeneously without distorting the triazine ring. This method could be a way to better improve the redox ability of g-C3N4 to dope zinc and make it suitable for forming a Z-scheme water splitting system with bismuth vanadate.
A typical Z-scheme system composed of a Fe3+/Fe2+ redox mediator that splits water into H2 and O2 by using zinc-doped (10%) g-C3N4 for H2 production and BiVO4 as O2-photocatalyst was reported. The Z-scheme system showed a stoichiometric ratio (2:1) in the generation of H2 and O2 over a long time period when the zinc was doped by an impregnation method. This provides a homogeneous formation of Zn-N bonds without distorting the triazine ring.
2018,
Vol. 39 Issue (3): 472-478 
PDF
2018,
Vol. 39 Issue (3): 472-478 PDF