Hierarchical Sheet-on-Sheet ZnIn2S4/g-C3N4 Heterostructure with Highly Efficient Photocatalytic H2 production Based on Photoinduced Interfacial Charge Transfer

We have realized in-situ growth of ultrathin ZnIn2S4 nanosheets on the sheet-like g-C3N4 surfaces to construct a “sheet-on-sheet” hierarchical heterostructure. The as-synthesized ZnIn2S4/g-C3N4 heterojunction nanosheets exhibit remarkably enhancement on the photocatalytic activity for H2 production. This enhanced photoactivity is mainly attributed to the efficient interfacial transfer of photoinduced electrons and holes from g-C3N4 to ZnIn2S4 nanosheets, resulting in the decreased charge recombination on g-C3N4 nanosheets and the increased amount of photoinduced charge carriers in ZnIn2S4 nanosheets. Meanwhile, the increased surface-active-sites and extended light absorption of g-C3N4 nanosheets after the decoration of ZnIn2S4 nanosheets may also play a certain role for the enhancement of photocatalytic activity. Further investigations by the surface photovoltage spectroscopy and transient photoluminescence spectroscopy demonstrate that ZnIn2S4/g-C3N4 heterojunction nanosheets considerable boost the charge transfer efficiency, therefore improve the probability of photoinduced charge carriers to reach the photocatalysts surfaces for highly efficient H2 production.

Scientific RepoRts | 6:19221 | DOI: 10.1038/srep19221 behavior can be also promoted through tailoring the secondary nanostructures of heterostructural photocatalysts [25][26][27] . Therefore, design and architecture of g-C 3 N 4 -based heterostructural photocatalysts with the matchable bandgap, desired component, and hierarchical nanostructures is still the hot topics in the field of solar-to-fuels conversion.
As an important ternary chalcogenide semiconductor, hexagonal phase ZnIn 2 S 4 with 2D layered structure and narrow bandgap has been extensively investigated in photocatalysis, especially serving as the photosynergistic components in heterojunction photocatalysts to enhance the photocatalytic efficiency for H 2 production [28][29][30][31][32] . In the viewpoint of band structure, the bandgap of ZnIn 2 S 4 (E g ZnIn S 2 4 = ∼ 2.6 eV) is smaller than that of g-C 3 N 4 ( E g C N 3 4 ~2.8 eV) while the conduction band (CB) of ZnIn 2 S 4 (E CB ZnIn S 2 4 = − 1.0 V) is higher than that of g-C 3 N 4 (E CB C N 3 4 15,17,28,33 . Accordingly, when integrating of ZnIn 2 S 4 nanostructure with g-C 3 N 4 nanosheets, a "type I" heterojunction would be formed in their interface, meaning that the CB and valence band (VB) positions of g-C 3 N 4 straddle those of ZnIn 2 S 4 . As such, the ZnIn 2 S 4 can be seemed as a "charge sink" to accept the photoinduced charge carriers from adjacent g-C 3 N 4 , leading to the improvement of charge separation on g-C 3 N 4 and thereby enhancing its photocatalytic activity. On the other hand, the 2D sheet-like ZnIn 2 S 4 nanostructures could be easily anchored onto the active or flexible 2D substrates, such as F-doped SnO 2 (FTO) thin film and reduced graphene oxide (RGO) nanosheets, to form the "sheet-on-sheet" type heterostructure. This kind of hierarchical nanostructure usually exhibits a high surface area, strong light harvesting, and efficient charge mobility due to its unique structure advantages 25,26,29,31,33,34 . The above analysis implies that once the ZnIn 2 S 4 nanosheets combine with g-C 3 N 4 nanosheets, a significant enhancement on photocatalytic H 2 production may be realized through synergistic promotion on the inner charge carriers and outer hierarchical structures. However, little effort has been donated to the synthesis of ZnIn 2 S 4 /g-C 3 N 4 heterostructure toward the highly efficient photocatalytic H 2 production. Herein, we report a novel kind of "sheet-on-sheet" heterostructure synthesized through in-situ growth of ultrathin ZnIn 2 S 4 nanosheets onto g-C 3 N 4 nanosheets surfaces. After introducing the ZnIn 2 S 4 nanosheets, the specific surface area of g-C 3 N 4 nanosheets is obviously promoted, resulting in providing the more active sites for the photoreaction. Furthermore, the intimate contacted interface between the ZnIn 2 S 4 and g-C 3 N 4 nanosheets facilitates the photoinduced charge-carriers transfer from g-C 3 N 4 to ZnIn 2 S 4 based on the heterojunction effect. By taking of the above features, the as-synthesized ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets exhibit a significantly enhanced visible-light photocatlaytic H 2 production performance as compared to the single component of ZnIn 2 S 4 or g-C 3 N 4 nanosheets.

Results
X-ray diffraction (XRD) patterns of the as-synthesized samples are shown in Fig. 1. Two pronounced diffraction peaks appear at 13.1° and 27.4° for g-C 3 N 4 nanosheets, reflecting to the periodic structure of intra-planar tri-s-triazine packing as the (100) peak, and the interlayer stacking of conjugated aromatic structures as the (002) peak for graphitic materials, respectively 11,15,17 . The diffraction peaks of ZnIn 2 S 4 nanosheets can be perfectly indexed as a pure hexagonal phase of ZnIn 2 S 4 (JCPDS No. 65-2023) 28,29,31 . In the case of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets, the XRD pattern shows diffraction peaks of both ZnIn 2 S 4 and g-C 3 N 4 nanosheets, while the feature peaks of ZnIn 2 S 4 (27.7°) and g-C 3 N 4 (27.4°) are very close and overlap with each other. Note that the diffraction peaks of ZnIn 2 S 4 nanosheets are very weak. This phenomenon may be ascribed to two reasons: (1) the ultrathin 2D nanostructure of ZnIn 2 S 4 , leading to the ultra-small size in its c-axis orientation 31 ; (2) the low content of ZnIn 2 S 4 component in the heterostructure 25 . Besides, after introducing the ZnIn 2 S 4 by hydrothermal method, the (100) diffraction peak of g-C 3 N 4 becomes more intense, and its position shifts toward the lower diffraction angle (10.8°) ( Figure S1). This reveals that some metal ions from ZnIn 2 S 4 surfaces may be connected with the g-C 3 N 4 through the lone-pair electrons of nitrogen in the "nitrogen pots", thus leading to enlarging the intra-planar separation of ordered tri-s-triazine packing 21,22,35 . Moreover, some weak peaks corresponded to the intermediates of thermal polymerized g-C 3 N 4 are also detected on the XRD pattern of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets, implying that a small part of g-C 3 N 4 nanosheets might be further exfoliated (or reduced) into the structural units of g-C 3 N 4 , such as melamine, ammeline, or tri-s-triazine units, due to the longer reaction time for hydrothermal growth of ZnIn 2 S 4 onto g-C 3 N 4 nanosheets 16,36 .
Scanning electron microscopy (SEM) and transmission electron microscopy (TEM) images were performed to directly observe the morphologies and structures of the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets in comparison with the single component of ZnIn 2 S 4 and g-C 3 N 4 nanosheets, as displayed in Fig. 2. The thermal polymerized g-C 3 N 4 shows a bulk structure with smooth surfaces ( Fig. 2A), which can be easily exfoliated into the wrinkled sheet-like nanostructures by the ultrasonic treatment in methanol solution (Fig. 2D). Figure 2B reveals significant aggregation of the nanosheets into microspheres with the average diameter of ~1.5 μ m for the hydrothermally synthesized ZnIn 2 S 4 sample. Interestingly, when introducing the as-fabricated g-C 3 N 4 nanosheets as the substrates during the hydrothermal process, the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets could be achieved in the form of "sheet-on-sheet" structure ( Figure S2 and Figure S3). As observed in Fig. 2C, the layered surfaces of g-C 3 N 4 nanosheet are covered with the high density of secondary ZnIn 2 S 4 nanosheets. These nanosheets with uniformly ultrathin 2D-structure are connected and even across to each other, finally forming the sheet-like networks vertically aligned on the g-C 3 N 4 nanosheet surface. This kind of unique hierarchical heterostructure (65.9 m 2 g −1 ) shows much higher specific-surface-area than the general sheet-like structure for pure g-C 3 N 4 (12.8 m 2 g −1 ), thereby providing more active sites for the photocatalytic reaction. (Figure S4) Moreover, the interspaces among interweaved ZnIn 2 S 4 nanosheets on the g-C 3 N 4 nanosheets may also boost the light-harvesting behavior of this hierarchical heterostructure though the multi-reflection processes on the incident electromagnetic waves 25,26,29 . TEM image of the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets further confirms that the ultrathin ZnIn 2 S 4 nanosheets with thickness of 4∼ 9 nm are vertically grown onto the g-C 3 N 4 nanosheets surface ( Fig. 2E and Figure  S3). Figure 2F presents the high-resolution (HR) TEM of an individual ZnIn 2 S 4 nanosheet on g-C 3 N 4 surface, in which the lattice-fringe spacing of 0.41 nm appeared on the side view (perpendicular sheet) can be assigned to the (006) crystal plane of hexagonal ZnIn 2 S 4 , while the top view (planar sheet) image shows the interplanar distances of 0.32 nm, belonging to the d-spacing of (102) planes of hexagonal ZnIn 2 S 4 . In short, intimate contacted heterojunctions between ZnIn 2 S 4 and g-C 3 N 4 nanosheets are indeed constructed by the in-situ growth process, which may be beneficial for the photoinduced interfacial charge-transfer from g-C 3 N 4 to ZnIn 2 S 4 . Figure 3A presents the Fourier transform infrared (FT-IR) spectra of the as-synthesized ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets along with the single ZnIn 2 S 4 and g-C 3 N 4 nanosheets for the purpose of structure comparison. The stretching vibration bands on the spectra of g-C 3 N 4 nanosheets show characteristics similar to those of the reported results 13,16,21,23 . Accordingly, the peaks appeared between 1200 cm −1 and 1650 cm −1 are attributed to the stretching vibration modes of CN heterocycles. The peak located at 3200 cm −1 is originated to the NH stretching vibration mode, while the 811 cm −1 to the feature vibration mode of s-triazine ring unit. For the ZnIn 2 S 4 sample, only two peaks at 1396 cm −1 and 1610 cm −1 , belonging to the surface hydroxyl groups and absorbed water molecules, can be observed on the FT-IR spectrum 25 . After the growth of ZnIn 2 S 4 nanosheets onto g-C 3 N 4 surfaces, the heterojunction nanosheets show the typical stretching vibration modes of both ZnIn 2 S 4 and g-C 3 N 4 nanosheets. Besides, a series of new vibration bands can be detected simultaneously on the spectrum of ZnIn 2 S 4 /g-C 3 N 4 nanosheets, which are in agreement with the vibration bands of melamine and/or ammeline 16,36 . This further suggests that during the long-time hydrothermal process, a few number of g-C 3 N 4 nanosheets were exfoliated (or reduced) into the sub-structures of g-C 3 N 4 . To study in-depth the chemical configurations of the as-synthesized samples, the X-ray photoelectron spectroscopy (XPS) analyses were performed. As observed in Fig. 3B, three main peaks with the binding energies at 284.6 eV, 285.7 eV, and 287.9 eV can be found on the C 1s core-level spectrum of g-C 3 N 4 nanosheets, which are assigned to sp 2 C-C bonds of graphitic carbon, sp 3 -coordinated carbon bonds, and sp 2 -bonded carbon (N-C = N) of the s-triazine rings, respectively 17,21 . The binding energy for the C 1s peak at 284.6 eV can be attributed to the adventitious carbon species on the samples and the carbon-containing contaminants, which was used as the reference for calibration. The N 1s signal of g-C 3 N 4 nanosheets also shows three feature peaks, corresponding to the sp 2 -bonded N (C-N = C) (398.1 eV), tertiary nitrogen N-(C) 3 groups (399.2 eV), and amino groups (C-N-H) (400.7 eV) 17,21 . Investigations found that the relative intensity of the peaks relating to the N-C = N and C-N = C groups of g-C 3 N 4 nanosheets are decreased after hydrothermal treatment for a long time, indicating that some of the tri-s-triazine units of g-C 3 N 4 were distorted during this process. Meanwhile, when introducing the ZnIn 2 S 4 nanosheets onto g-C 3 N 4 nanosheets to form the heterojunction, both the C 1s and N 1s characteristic signals of g-C 3 N 4 nanosheets shift slightly toward the higher binding energy side. On the contrary, the binding energies of Zn 2p (1022.0 eV for 2p 3/2 and 1045.5 eV for 2p 1/2 ), In 3d (444.9 eV for 3d 5/2 and 452.5 eV for 3d 3/2 ), and S 2p (161.5 eV for 2p 3/2 and 162.6 eV for 2p 1/2 ) for the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets are a little lower than the corresponding values for the pure ZnIn 2 S 4 nanosheets 25,29,31 , as shown in Fig. 3D-F. The binding energy shifts λ for the heterojunction components could be explained by a strong interaction between ZnIn 2 S 4 and g-C 3 N 4 nanosheets 21,37-39 . Theoretically, the enhancement (B) C 1s core-level spectra; (C) N 1s core-level spectra: (a) g-C 3 N 4 nanosheets, (b) 15 wt% ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets; (D) Zn 2p core-level spectra; (E) In 3d core-level spectra; (F) S 2p core-level spectra: (b) 15 wt% ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets, and (c) ZnIn 2 S 4 nanosheets.
Scientific RepoRts | 6:19221 | DOI: 10.1038/srep19221 of binding energy means the weakened electron screening effect caused by the decreased electron concentration, while the increase in electron concentration leads to the decrease of binding energy due to the promoted electron screening effect. Thus, in this case, it is reasonable to conclude that the higher and lower binding energy shifts are ascribed to the decreased electron concentration of g-C 3 N 4 nanosheets and increased electron concentration of ZnIn 2 S 4 nanosheets due to the strong interaction between the g-C 3 N 4 to ZnIn 2 S 4 nanosheets based on the interfacial charge transfer.
The optical properties of the as-synthesized samples were investigated through UV-vis absorption spectra which converted from the corresponding diffuse reflectance (DR) spectra by means of the Kubelka-Munk function. As shown in Fig. 4, the absorption edges of g-C 3 N 4 and ZnIn 2 S 4 nanosheets appears at ~443 and ~476 nm, corresponding to the band energies of ~2.8 and ~2.6 eV, respectively. These values are consistent with the reported values of g-C 3 N 4 and ZnIn 2 S 4 nanosheets 15,[17][18][19]28,33,34 . In the case of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets, two obvious absorption bands ascribed to the characteristic absorption of g-C 3 N 4 and ZnIn 2 S 4 nanosheets can be found on the absorption curve of Fig. 4b. Moreover, the absorption peaks of ZnIn 2 S 4 nanosheets become more intense with increase of ZnIn 2 S 4 content in the heterojunction nanosheets ( Figure S5), further confirming that the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets with controllable component contents were obtained.

Discussion
Photocatalytic H 2 production activities of the as-synthesized samples were evaluated under visible light (λ > 400 nm) irradiation by using triethanolamine (TEOA) as the sacrificial reagent to quench the photoinduced holes. The H 2 production rates of pure g-C 3 N 4 nanosheets, pure ZnIn 2 S 4 nanosheets, and ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets with various ZnIn 2 S 4 ratios are summarized in Fig. 5A, in which the heterojunction nanosheets show the enhanced H 2 production rates as compared to the single heterojunction components. It implies that in-situ growth of ZnIn 2 S 4 nanosheets onto g-C 3 N 4 nanosheets could noticeably improve the photocatalytic activities on H 2 production. Even with only 2.5 wt% ZnIn 2 S 4 nanosheets, the heterojunction nanosheet displays a H 2 production rate of 5.2 μ mol h −1 , which is more than 6 times higher than that of pure g-C 3 N 4 nanosheets (0.8 μ mol h −1 ). The poor photoactivity of g-C 3 N 4 nanosheets for the H 2 production is mainly ascribed to its limited light-harvesting efficiency and fast recombination of photoinduced charge carriers 18,19 . The optimal photocatalytic activity was achieved on 15 wt% ZnIn 2 S 4 with a H 2 production rate of 14.1 μ mol h −1 . This value is ~17.6 times higher than that of pure g-C 3 N 4 nanosheets and even nearly 4 times higher that of the pure ZnIn 2 S 4 nanosheets. Accordingly, the apparent quantum efficiency of this optimal sample is estimated as 0.28% under irradiation at 420 nm. Note that only 5 mg of 15 wt% ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets was used for H 2 production in our work. However, when the ZnIn 2 S 4 content is higher than 15 wt%, a further increase in ZnIn 2 S 4 content (20 wt%) leads to a rapid decrease in the photocatalytic activity for H 2 production. This photoactivity reduction can be attributed to the increased opacity (so-called shield effect), resulting in a decrease of irradiation passing through the suspension photoreaction solution 21,25,29,31,40,41 . As observed in Fig. 5B, the significant enhancement of photocatalytic activity for the heterojunction sample can be further confirmed by the time-dependent H 2 production behaviors. It could be found that the H 2 production amounts linearly increases with the irradiation time. After visible light irradiation for 2h, the H 2 production yield of 15 wt% ZnIn 2 S 4 -decorated g-C 3 N 4 nanosheet could reach 28.2 μ mol, which is greatly superior to the pure g-C 3 N 4 and ZnIn 2 S 4 nanosheets. The enhanced photoactivity on H 2 production could be explained by two main reasons: (1) the reduced recombination process of photoinduced charge carriers on g-C 3 N 4 and increased amount of charge carriers on ZnIn 2 S 4 based on the interfacial charge transfer; (2) the higher specific-surface-area and enhanced light absorption for the unique "sheet-on-sheet" heterostructure as aforementioned. However, it should be point out that in comparison with the pure ZnIn 2 S 4 nanosheets (132.0 m 2 g −1 ), the heterojunction nanosheets (65.9 m 2 g −1 ) shows the lower specific-surface-area, but the higher photoactivity. Meanwhile, the g-C 3 N 4 nanosheets treated by the hydrothermal process in the absence of ZnIn 2 S 4 precursor show a lower photocatalytic H 2 production rate (0.41 μ mol/h) as compared to the pure (untreated) g-C 3 N 4 nanosheets ( Figure S6), because of the poor photoactivities of the exfoliation or reduction of g-C 3 N 4 with the ultra-small structures, such as melamine, ammeline, or tri-s-triazine units 16,36 . These observations also the indirect evidence that the dynamics process of photoinduced charge transfer occurring on the interface between the ZnIn 2 S 4 and g-C 3 N 4 nanosheets may be crucial to the photocatalytic H 2 production activities of heterojunction nanosheets. At low ZnIn 2 S 4 content, a well dispersion of ZnIn 2 S 4 nanosheets on the g-C 3 N 4 nanosheets could be formed. In this way, the increase of ZnIn 2 S 4 content would induce more ZnIn 2 S 4 nanosheets assembled onto the g-C 3 N 4 surface, which generates larger contact area between the ZnIn 2 S 4 and g-C 3 N 4 nanosheets, thereby allowing more efficient interfacial charge transfer. However, excess ZnIn 2 S 4 loading causes considerable reduction of light absorption for the covered g-C 3 N 4 , and decreases its excitation process for interfacial charge transfer. Therefore, a balance would be built between ZnIn 2 S 4 loading amount and ZnIn 2 S 4 /g-C 3 N 4 contact area, which can maximize the photocatalytic H 2 production for the heterojunction nanosheets. Through the sequential remediation of heterojunction components, we concluded an optimal loading of 15 wt% ZnIn 2 S 4 onto g-C 3 N 4 , which exhibited the highest photocatalytic activity in our work.
of H 2 evolution were detected during the three-run test, powerfully verifying the good stability of 15 wt% ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets for using as the photocatalysts.
To provide more evidence on the photoinduced charge transfer occurring in the heterojunction interface, the surface photovoltage spectroscopy (SPS) and photoluminescence (PL) spectroscopy of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets were investigated in comparison with those of pure g-C 3 N 4 nanosheets. As shown in the inset of Fig. 6A, a poor response appears on the SPS curve of pure g-C 3 N 4 nanosheets, indicating very low efficiency on the photovoltaic conversion. However, the photon absorption and conversion processes have been proven by the photoelectrochemistry in our previous work 42 . Thus, the low photovoltage of g-C 3 N 4 nanosheets can be attributed to the Schottky interface contact between the g-C 3 N 4 and ITO glass in the absence of electrolyte, which leads to the limited electron transfer process from the g-C 3 N 4 to ITO glass electrode. After decoration of ZnIn 2 S 4 onto g-C 3 N 4 nanosheets, the SPS signal in the region from 300 to 450 nm is remarkably enlarged, suggesting the promoted charge generation and separation based on the semiconductor heterojunction effect. This phenomenon can be further understood through the steady-state and transient PL spectroscopy. In Figre 6B, the pure g-C 3 N 4 nanosheet shows a strong emission peak with the center at ~450 nm. However, as compared to the g-C 3 N 4 nanosheets, the emission process of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets suppresses significantly, revealing either the faster migration process with the shorter lifetime or the slower recombination process with the longer lifetime for the photoinduced charge carriers. To shed more light on this issue, we tried to fit the time-resolved transient PL spectroscopy based on the multi-exponential kinetics function expressed as follow 43 : where τ 1 and τ 2 are the fluorescent lifetime, and A 1 , and A 2 are the corresponding amplitudes. As listed in the insets of Fig. 6C,D, the short lifetime component for τ 1 is originated from the nonradiative recombination of charge-carriers in the defect states of g-C 3 N 4 , while the longer lifetime component for τ 2 is caused by the free excitons recombination in the g-C 3 N 4 body 12,21,34,42 . In the case of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets, the emission lifetime for the component τ 1 (3.3 ns) is longer than the corresponding lifetime of g-C 3 N 4 nanosheets (3.1 ns), while its component τ 2 (12.3 ns) is shorter than the component for g-C 3 N 4 nanosheets (14.7 ns). To gain further understanding on this phenomenon, the average emission lifetimes, relating to the overall emission decay behaviors of the samples, were also evaluated through the following equation 44 : It is clearly that after loading the ZnIn 2 S 4 nanosheets, the average lifetime of g-C 3 N 4 nanosheets is shorten from 10.45 ns to 8.97 ns. The combination of decreased emission lifetime and quenched PL indicates the emergence of a nonradiative pathway from the electron transfer between ZnIn 2 S 4 and g-C 3 N 4 nanosheets 45 . According to the energy band structures of these semiconductors (Fig. 6E), it could be deduced that the photoinduced electrons transfer from the conduction band (CB) of g-C 3 N 4 to the CB of ZnIn 2 S 4 nanosheets. This assumption is in agreement with the emission quenching phenomenon of g-C 3 N 4 nanosheets after decorating the ZnIn 2 S 4 nanosheets. The rate constant for the interfacial electron-transfer ( ) k et can be estimated by the expression 46 : Obviously, the k et value is approximately . × − 1 6 10 S 7 1 . As illustrated in Fig. 6E, the CB and valence band (VB) positions of g-C 3 N 4 straddle those of ZnIn 2 S 4 , forming the "type I" heterojunction interface. When this heterojunction is excited by visible light with the photon energy higher or equal to the band gaps of both ZnIn 2 S 4 and g-C 3 N 4 nanosheets, the photoinduced electrons and holes of g-C 3 N 4 nanosheets would move to the CB and VB of ZnIn 2 S 4 nanosheets, respectively. As such, the recombination process on the photoinduced charge carriers of g-C 3 N 4 could be suppressed effectively by the photosynergistic effect of ZnIn 2 S 4 /g-C 3 N 4 heterojunction. Accordingly, the amount of photoinduced charge carriers on ZnIn 2 S 4 is remarkably increased based on the photoinduced interfacial charge transfer. During the photocatalytic H 2 production process, the photoinduced electrons accumulated on the CB of ZnIn 2 S 4 could initiate the catalytic proton reduction to H 2 . Accordingly, the photoinduced holes transfer from the VB of g-C 3 N 4 to the VB of ZnIn 2 S 4 were quenched by the sacrificial reagent of TEOA (or directly quenched by the sacrificial reagent on the VB of g-C 3 N 4 ). In this way, the effective charge transfer at the interface between the ZnIn 2 S 4 and g-C 3 N 4 nanosheets results in the enhanced photocatalytic activity on H 2 production.
In summary, a series of ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets with various contents of ZnIn 2 S 4 have been successfully synthesized through in-situ growth of ultrathin ZnIn 2 S 4 nanosheets onto g-C 3 N 4 nanosheets fabricated by a traditional thermal polymerization and followed ultrasonic dispersion method. The unique "sheet-on-sheet" heterostructure obtained by vertically loading ZnIn 2 S 4 nanosheets onto the g-C 3 N 4 nanosheets surfaces leads to the enlarged reactive sites and enhanced light absorption ability. More importantly, the formation of "type I" heterojunction can effectively suppress the photoinduced charge recombination of g-C 3 N 4 through the interfacial charge transfer, as evidenced by the electron microscopic analyses, steady-state and time-resolved transient photoluminescence decay investigations. As a result, the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets exhibited considerable enhancement on the photocatalytic activity for H 2 production as compared the single component nanosheets. It is believed that our study provides a promising strategy to develop the new generation of hierarchical heterostructure photocatalysts for highly efficient solar-to-fuels conversion and environmental remediation.

Methods
Materials synthesis. The graphitic carbon nitride (g-C 3 N 4 ) was obtained by a traditional thermal polymerization method. 10 g of melamine powder was grinded for 60 min in a mortar and then transferred to an alumina crucible with a cover. Afterward, the crucible was heated to 550 °C with a rising rate of 20 °C min −1 and kept for 2 h at the required temperature under semiclosed environment, resulting in the bulk g-C 3 N 4 with faint-yellow color. ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets were synthesized by in-situ growth of ultrathin ZnIn 2 S 4 nanosheets onto g-C 3 N 4 nanosheets through a facile hydrothermal method. In a typical procedure, 600 mg of as-synthesized bulk g-C 3 N 4 was grinded to fine powder and then added into 20 ml of methanol. After ultrasonic treatment for 2 h, the bulk g-C 3 N 4 was exfoliated into thin nanosheets which was then collected and washed by using centrifugation-redispersion with deionized water. Subsequently, these exfoliated g-C 3 N 4 nanosheets was resuspended into 20 ml of premade aqueous solution consisting of 0.2125 mmol of Zn(CH 3 COO) 2 · 2H 2 O, 0.425 mmol of In(NO 3 ) 3 · 6H 2 O, and 1.7 mmol of L-cysteine. After being ultrasonically treated for 30 min, this mixture was transferred into a Teflon-lined stainless steel autoclave with a capacity of 25 mL. Afterward, the autoclave was sealed and maintained at 180 °C for 12 h in an electric oven. When natural cooling the autoclave to room temperature, the yellow-green suspension was collected, washed with ethanol and deionized water for several times, and finally dried in an electric oven at 60 °C for a night. Thus, the 15 wt% ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets were synthesized. The pure ZnIn 2 S 4 nanosheets were fabricated by the same hydrothermal conditions in the absence of the g-C 3 N 4 nanosheets substrates. Meanwhile, to achieve the optimal photocatalytic activity, the ZnIn 2 S 4 /g-C 3 N 4 heterojunction nanosheets with different ZnIn 2 S 4 loading amount were also synthesized using the similar route by adjusting the concentrations of hydrothermal precursor solution in the same component ratios. In order to further prove that the enhanced photocatalytic activity of ZnIn 2 S 4 /g-C 3 N 4 nanosheets is due to the heterojunction effect, another control sample were fabricated through hydrothermal treatment of pure g-C 3 N 4 nanosheets in the absence of the above ZnIn 2 S 4 precursors. spectroscopy being attached to scanning electron microscopy (SEM) was used to analyze the composition of products. Fourier transform infrared (FT-IR) spectra were recorded on a Magna 560 FT-IR spectrometer with a resolution of 1 cm −1 . X-ray photoelectron spectroscopy (XPS) was carried out on a VG-ESCALAB LKII instrument with a Mg Kα ADES (hν = 1253.6 eV) source at a residual gas pressure below 10 −8 Pa. UV-vis diffuse reflectance spectra (DRS) were taken with a Lambda 750 UV/Vis/NIR spectrophotometer (Perkin Elmer, USA). The specific surface areas of the products were measured with a Micromeritics ASAP-2020 instrument and analyzed by the Brunauer-Emmett-Teller (BET) method. Decay curves of the as-fabricated products were obtained on a FLS920 fluorescence lifetime spectrophotometer (Edinburgh Instruments, UK) under the excitation of a hydrogen flash lamp with the wavelength at 325 nm (nF900; Edinburgh Instruments). The surface photovoltage spectroscopy (SPS) was performed on PL-SPS1000 instrument (Beijing Perfectlight Technology Co., Ltd). During the process, the sample was put between the indium tin oxide (ITO) glass and stainless steel electrodes to form a sandwich structured photovoltage cell.
Photocatalytic H 2 production. The photocatalytic H 2 production tests were performed in a 35-mL quartz reactor. Typically, 5 mg of the as-synthesized samples were suspended in 10-mL triethanolamine (TEOA, 15 vol.%) aqueous solution. This suspension was sealed in the quartz reactor by a rubber plug, and then purged with argon gas for half an hour to drive away the residual air. Subsequently, the reactor was exposed under a 300-W Xe lamp (PLS-SXE300UV) coupled with a 400 nm cut-off filter. The gas product composition from the upper space above the liquid in the quartz reactor was periodically analyzed by a gas chromatograph (GC) equipped with a thermal conductivity detector (TCD) (Beifen-Ruili Analytical Instrument, SP-3420A). The apparent quantum efficiency (QE) was estimated by using the following equation.