3.1 Structural, componential, and morphological analysis
The crystalline structures of u-CNB, O-CNS, MoS2 and O-CNS/MoS2 composites were detected through X-ray powder diffraction (XRD). The u-CNB catalyst exhibited two prominent peaks around 2θ of 13.2° and 27.8°, individually assigned to be (100) and (002) crystal planes, as is illustrated in Fig. 1. Several characteristic peaks at 14.1°, 32.9°, 39.5°and 58.7°, respectively assigned to be (002), (100), (103), and (110) of MoS2 (JCPDS No. 75-1539), are clearly observable in Fig. 1. No characteristic peak associated with MoS2 crystal phase was detectable for O-CNS/0.5MoS2, O-CNS/1MoS2 and O-CNS/3MoS2 samples mainly owing to its lower loading amount, as illustrated in Fig. S1. After the loading amount of MoS2 was beyond the critical value, we can find that the intensities of the characteristic peaks related with MoS2 increased perceptibly with the augmentation of MoS2 content in the O-CNS/MoS2 composites, whereas the counterparts for O-CNS gradually decreased.
From Fig. 2a, the O-CNS sample is comprised of abundant distorted nanosheets with average size of approximately 1–2 µm. Furthermore, the distorted nanosheets of O-CN were provided with thin thickness and coarse surface, which was suitable to act as a substrate for loading cocatalysts. It is very distinct from Fig. 2b-c that the MoS2 sample displayed 3D flower-like structure assembled with 2D nanoflakes. Figure 2d-i illustrates the SEM images of O-CNS/xMoS2 (x = 0.5, 1, 3, 5, 7, 9). As illustrated from the red ellipse in Fig. 2d-g, MoS2 microflowers were randomly anchored on the surface of O-CNS, indicative of the formation of O-CNS/MoS2 hybrid composites. Contrarily, with further increasing MoS2 loading amount, MoS2 microflowers couldn’t be evenly dispersed on the surface of O-CNS under ultrasound treatment (Fig. 2h-i). The aggregation of MoS2 microflowers failed to form sufficient contact between MoS2 nanosheets and O-doped C3N4 porous nanosheets, when the loading amount of MoS2 was more than 5 wt%. Such aggregated hybrid structure was insufficient for the effective transfer of photo-induced carrier between the two components of MoS2 and O-doped C3N4 porous nanosheets.
As inspected from Fig. 3a, the morphology of O-CNS/5MoS2 located in the upper-left section was quite different from that in green oval circled region. The former (upper-left) indicated the porous characteristics of O-doped C3N4 disordered nanosheets, while the latter (green oval) was closely related with MoS2 microflowers. Figure 3b-g illustrates the HAADF-STEM image of O-CNS/5MoS2 with corresponding EDS mapping images. It is evident from Fig. 3c-e that four elements containing C, N, as well as O were homogeneously spread out in the O-CNS/5MoS2 sample, while the Mo and S elements were distributed throughout the bright region (Fig. 3f-g). The observation of HAADF-STEM further demonstrated that 3D MoS2 microflowers were evenly distributed on the surface of 2D O-CN porous nanosheets. From Fig. 3h, the adjacent lattice spacing of 0.626 nm could be designated to MoS2 (002) crystal plane. The results of energy dispersive spectrometer (EDS) spectrum in Fig. 3i ulteriorly confirmed the coexistence of O, C, N, Mo together with S elements in O-CNS/5MoS2 composites. Consulting to previously reported research, we can rationally conjecture that the hybrid structure formed by O-CNS porous nanosheets and MoS2 nanosheets could facilitate efficient carrier migration, presumably resulting in remarkable improvement in PHP efficiency.
3.2 Surface composition and elemental chemical status
The XPS spectrum of O-CNS (Fig. 4a) indicates that the O-CNS catalyst is constituted of C, N, as well as O elements. Besides above, elements of C, N, O, Mo and S can be simultaneously found in O-CNS/5MoS2 composites from Fig. 4a, which is almost concordant with the EDS results. In the C1s spectrum (Fig. 4b), two peaks of the O-CNS/5MoS2 sample at 288.3 and 284.8 eV could be respectively designated as sp2-hybridized carbon (N = CN) and adventitious carbon [23, 24]. In the N1s spectrum for the O-CNS/5MoS2 sample (Fig. 4c), two peaks situated around 398.7, and 400.7 eV could be assigned to the sp2-hybridized N atoms (CN = C) and N-(C)3 [14]. In addition, the peak situated around 404.4 eV was attributable to incomplete bonding defects caused by π excitation [17]. Interestingly, compared with those of O-CNS, the C1s and N1s peaks in the O-CNS/5MoS2 composites exhibited a slight blueshift, indicating the presence of firm interreaction between O-CNS and MoS2. The peak of O1s for O-CNS and O-CNS/5MoS2 (Fig. 4d) appeared around 531.8 eV, which was originated from the N-C-O species [25]. From Fig. 4e, the doublet peaks at 228.2 and 232.0eV were respectively assigned to Mo 3d5/2 and 3d3/2. The weak peak split into double peaks located at 161.8 and 163.0eV could be respectively designated as S 2p3/2 and S 2p1/2 (Fig. 4f). Based above, the results of XPS authenticated that MoS2 microflowers were successfully incorporated into O-CNS to form composites with firm interaction.
3.3 Photocatalytic performance
The photocatalytic behavior of the as-obtained u-CNB, O-CNS, and O-CNS/MoS2 products was assessed by PHP exposed to visible light illumination. As illustrated in Fig. 5a-b, the u-CNB exhibited poor photocatalytic behavior with a hydrogen production rate (HPR) of 157.3µmol·g− 1·h− 1. The poor hydrogen production behavior of the u-CNB was probably originated from insufficient active sites, relatively narrow visible-light absorption scope and rapid photogenerated carrier recombination rate. In comparison, the O-CNS samples took on enhanced hydrogen evolution behavior with a HPR of 2012.9 µmol·g− 1·h− 1. This enhancement in PHP rate could be result from the increment of reaction active sites combining with the unique porous structure. As expected, the PHP rate was further significantly boosted after coupling MoS2 with O-CNS. Interestingly, the PHP rate of O-CNS/MoS2 composites first scaled up with the augment of the MoS2 loading amount, and tended to lessen thereafter. Most notably, the O-CNS/5MoS2 hybrid catalyst owned the highest HPR of 4570.3µmol·g− 1·h− 1 among all tested photocatalysts, which was 29.1 and 2.3 times larger than that of u-CNB and O-CNS, respectively. When the MoS2 loading amount exceeded the critical value (5 wt%), the HPR for O-CNS/MoS2 hybrid composites decreased instead. This phenomenon was presumably owing to the fact that the excess MoS2 aggregated in composites inhibited the incident light absorption of O-CNS, and then hindered the photo-activated electron transport pathway. Besides above, in order to further illustrate the superior HPR over O-CNS/5MoS2 hybrid catalyst, some previously reported results of g-C3N4 based photocatalysts are presented in Table S1. By comparison, it is evident that the HPR of O-CNS/5MoS2 hybrid catalyst was higher or almost equal to that of other g-C3N4 based photocatalysts. By considering the potential application, the recyclability of catalyst is a crucial factor. Thus, cycling experiments of PHP were executed to investigate the stability of the O-CNS/5MoS2 catalyst. The HPR for cycling experiments is presented in Fig. 5c. Resultantly, a mild decline was detectable in the HPR from the first to the third cycles, demonstrating the superior recyclability.
3.4 Possible enhancement mechanism
It is broadly accepted that the optical absorption and charge transfer characteristics of semiconductor catalyst are tightly associated with the PHP behavior. To gain a deep insight into the enhancement mechanism, the light absorption and photoelectrochemical spectra intimately correlated with photocatalytic hydrogen evolution were explored in details. Figure 6a-b displays the UV-Vis diffraction reflectance spectra (DRS) and the homologous Tauc curves of the u-CNB, O-CNS, O-CNS/5MoS2 and MoS2 samples. As indicated in Fig. 6a, the u-CNB catalyst displayed an evident absorption edge situated around 460 nm which was closely associated with the band-gap excitation. After being modified by O doping, the absorption edge of O-CNS was red shift in comparison with that of u-CNB. As for MoS2, it possessed a broad light responsive range together with exceptional light harvesting in the entire region, demonstrating that MoS2 could be suitably taken as co-catalyst to form O-CNS/MoS2 hybrid composites. As expected, the O-CNS/5MoS2 composites showed a red shift and better visible-light absorption characteristics contrasted with u-CNB and O-CNS, originating from the interaction between MoS2 nanosheets and O-CNS porous nanosheets. The corresponding Tauc’s energy bandgap plots are shown in Fig. 6b and c. The energy band gaps (Eg) of u-CNB, O-CNS, O-CNS/5MoS2 and MoS2 were estimated to be approximate to 2.61, 2.51, 2.19 and 1.87eV, respectively. The above results suggested that the introduction of MoS2 expanded the scope of visible-light absorption and improved the visible-light absorption capacity, which was beneficial for generating more photoactivated electron-hole pairs to involve the reaction of photocatalytic splitting water. The existence of unpaired electron was studied through room temperature EPR spectroscopy. As shown in Fig. 6d, the O-CNS/5MoS2 sample exhibited the highest intensity of EPR signal amongst u-CNB, O-CNS, and O-CNS/5MoS2, revealing that photoinduced carriers in the O-CNS/5MoS2 composites had strongest separation capability, facilitating the significant improvement of PHP behavior.
Photoelectrochemical measurements including I-T and EIS plots are presented to deeply investigate the variation of the photogenerated carriers’ separation and migration behavior of u-CNB, O-CNS, and O-CNS/5MoS2 photocatalysts. As illustrated in Fig. 7a, all samples exhibited significantly positive and reproducible photocurrent during light irradiation in the on-off cycles, meaning the photosensitivity and stability of the obtained photocatalysts [26]. Obviously, the O-CNS exhibits higher photocurrent density than that of u-CNB, confirming that the O doping indeed promoted the charge separation efficiency. The O-CNS/5MoS2 composites showed the substantial enhancement of photocurrent density, confirming that the O-CNS/5MoS2 hybrid composites possessed the highest charge carriers’ separation and transfer rate. As previously confirmed in relative studies [27–29], the larger the radius of a semicircle is, the lower the separation efficiency of photoactivated electron-hole pairs is. From the EIS Nyquist curve in Fig. 7b, the O-CNS/5MoS2 exhibited the smallest radius amongst u-CNB, O-CNS, and O-CNS/5MoS2 photocatalysts, implying its larger separation rate of carriers along the interface. These above results validated that the construction of O-CNS/5MoS2 hybrid composites enormously accelerated the carrier migration rate, further promoted the PHP performance.
Mott-Schottky (M-S) plots are depicted to estimate flat-band (Efb) potentials of semiconducting catalysts to further investigate the influence of energy band structure on photocatalytic performance. From Fig. 8a, it is easily observable that both O-CNS and MoS2 had a positive slope of the linear partial, suggesting the typical n-type semiconducting material characteristics [30]. Based on the tangents intercept of the curves, the Efb potentials of O-CNS and MoS2 were individually determined as -0.69 V and 0.09 V vs. NHE. In generally, the Efb potential is about 0.2 V positive than the conduction band (CB) potential in the n-type semiconductor [31, 32]. Therefore, the CB potential of O-CNS and MoS2 could be respectively determined to be -0.89eV and − 0.11eV vs. NHE. Combined with the bandgap obtained from UV-Vis DRS results, the valence band (VB) potential of O-CNS and MoS2 were individually determined as 1.62 and 1.76 eV vs. NHE. Additionally, the VB potential of MoS2 was also estimated by employing valence band XPS (VB-XPS). From Fig. S2, the VB maximum of MoS2 was approximate 1.76 eV, agreeing well with the above M-S results.
In the view of above demonstration, a Z-scheme carrier transfer mechanism is proposed in the O-CNS/5MoS2 composites for H2 generation reaction. The energy band structure of O-CNS and MoS2 is illustrated in Fig. 9a. Since the level of Fermi energy (FE) is approximate to the CB potential for the n-type semiconductors [33, 34], MoS2 has a lower FE level compared with that of O-CNS. After contact, the electrons (e−) will spontaneously flow from O-CNS to MoS2 until the two FE levels achieve equilibrium. Due to the changes of interface charge, a negative charge electron accumulation layer is shaped on the MoS2 and the band bends downward. While a positive charge electron depletion layer is shaped on the O-CNS and the band bends upward. Accordingly, an internal electric field (IEF) is formed in the interface between O-CNS and MoS2. This above theoretical inference is well consistent with the XPS results presented above, in which the binding energies of C and N in O-CNS/5MoS2 hybrid composites shift positively. Upon light irradiation, electrons are activated and are transferred from VB to CB in the O-CNS and MoS2, leading to the fact that photogenerated electrons accumulate in CB, whereas the holes still retain in the VB. Under the function of the IEF and energy band bending, the photoactivated electrons from the CB of MoS2 recombine with the holes from the VB of O-CNS. Subsequently, the photoactivated holes in the VB of MoS2 and the electrons in the CB of O-CNS are retained by experiencing the above recombination process. This mechanism of interfacial charge transfer in Z-scheme hybrid composites can greatly promote the spatial charge separation and reserve photoinduced e−-h+ pairs with stronger redox capabilities. Therefore, the photocatalytic hydrogen production performance of Z-scheme O-CNS/MoS2 hybrid composites was significantly improved.