Synthesis of Cd_1-xZn_xS/K₄Nb₆O₁₇ Composite and its Photocatalytic Activity for Hydrogen Production

Abstract

$Cd_{1-x}Zn_xS$-sensitized $K_4Nb_6O_{17}$ composite photocatalysts (designated $Cd_{1-x}Zn_xS/K_4Nb_6O_{17}$) were prepared via a simple deposition-precipitation method. The samples were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM), energy dispersive X-ray spectrometry (EDS), $N_2$ sorption, ultraviolet-visible light diffuse reflectance spectroscopy (UV-Vis DRS), photoluminescence measurements (PL), and X-ray photoelectron spectroscopy (XPS). The $Cd_{0.8}Zn_{0.2}S$ particles were scattered on the surface of $K_4Nb_6O_{17}$, and had a relatively uniform size distribution around 50 nm. The absorption edge of $K_4Nb_6O_{17}$ was shifted to the visible light region and the recombination of photo-generated electrons and holes suppressed after $Cd_{0.8}Zn_{0.2}S$ loading. The $Cd_{0.8}Zn_{0.2}S$(25 wt %)/$K_4Nb_6O_{17}$ composite possessed the highest photocatalytic activity for hydrogen production under visible light irradiation, evolving 8.278 mmol/g in 3 h. Recyclability tests were performed, and the composite photocatalysts were found to be fairly stable. The mechanism of charge separation between the photogenerated electrons and holes at the $Cd_{0.8}Zn_{0.2}S/K_4Nb_6O_{17}$ composite was discussed.

Introduction

Hydrogen evolution from photocatalytic water splitting under solar irradiation is a prospective sustainable solution to global environmental and energy problems.1-3 In light of this, the development of visible-light-driven photocatalysts remains a crucial issue to address for the efficient and effec-tive production of hydrogen using solar energy.4

Layered semiconductors, such as K4Nb6O17,5 K2Ti4O9,6 K2La2Ti3O10 and La2Ti2O7,7-9 have been reported in literature and shown to possess high photocatalytic activities for photocatalytic water splitting. Particularly, K4Nb6O17 exhibits high activity for hydrogen production from water under UV irradiation due to its unique layered structure.10 K4Nb6O17 is composed of layers of niobium oxide sheets, in which potassium ions locate in two different kinds of interlayers. One type of interlayer is easily hydrated and the K+ ions can be exchanged with multivalent cations (Ca2+, Ni2+, Li+, Na+ ions, etc), and the other cannot take up water molecules and can only be exchanged with univalent cations.11,12 Compared with TiO2, one of the most widely used photocatalysts, the layered structure of K4Nb6O17 possesses more reaction active sites, the interlamination can be modified and the photo-induced electrons can more easily migrate to the surface of catalyst to prevent the recombination of photo-generated holes and electrons, which make a contribution to the improvement of photocatalytic activities. However, the wide band gap of this material (~3.1 eV) leads to low efficiency under visible light irradiation,13 motivating recent research in the development of visible-light-driven K4Nb6O17 composite photocatalysts. Such composites have been prepared through coupling narrow band gap guest semiconductors, such as CdS,14 PbS,15 etc., into the host layered compounds to improve the visible light absorption for the composite layered semi-conductors. However, the practical applications of layered structure semiconductors are still limited due to low photo-catalytic efficiencies, so further work is necessary to improve the visible light response and charge separation mechanisms in these materials.

Cd1-xZnxS, a solid solution of CdS and ZnS, possesses tunable composition and band gap,16 which can be changed by adjusting the stoichiometric ratio between Zn and Cd, resulting in an observed band gap energy between that of CdS (2.4 eV) and of ZnS (3.6 eV). Therefore, the photo-catalytic activity of K4Nb6O17 can be improved by utilizing the sensitization effect of this Cd1-xZnxS to enhance the visible light response of K4Nb6O17 and by promoting the separation of photo-induced carriers. To the best of the authors’ knowledge, the photocatalytic performance over Cd1-xZnxS surface-sensitized K4Nb6O17 has not yet been reported.

Herein, we report a novel nano-composite Cd1-xZnxS/K4Nb6O17 photocatalyst with high visible light activity, which was synthesized via a simple deposition-precipitation method by depositing the adjustable band-gap Cd1-xZnxS particles on the surface of K4Nb6O17. The photocatalytic activity of the catalyst was investigated by splitting water into hydrogen under visible light irradiation, and a possible mechanism for the reaction was proposed.

 

Experimental

Photocatalyst Synthesis. All of the reagents were ana-lytical grade and used without further purification. The K4Nb6O17 materials were obtained via a high temperature solid state method,17 as reported in our previous research.18 The Cd1-xZnxS sample was prepared via a deposition-precipitation method. In a typical synthesis, a white inter-mediate was precipitated by adding a mixed Cd2+/Zn2+ aqueous solution containing appropriate molar ratios of Cd(CH3COO)2·2H2O and Zn(CH3COO)2·2H2O dropwise into an NaOH aqueous solution under stirring at 85 °C. The intermediate obtained was then converted to CdS/ZnS by dropwise addition of an Na2S·9H2O aqueous solution, con-taining a double excess of S2− relative to the amount of metal ions present, followed by continuous stirring for 1 h. The as-produced precipitate was collected by filtration and washed with deionized water and ethanol respectively. The product was then dried at 80 °C for 12 h and subsequently calcined at 400 °C for 1 h. The CdS/ZnS samples prepared, which contained various concentrations of CdS and ZnS, were labeled as Cd1-xZnxS (where x is the molar concentration of ZnS: 0, 0.2, 0.4, 0.5, 0.6, 0.8, 1, respectively).

Cd1-xZnxS/K4Nb6O17 composites were synthesized simi-larly by a deposition-precipitation method using Cd(CH3-COO)2·2H2O, Zn(CH3COO)2·2H2O, NaOH, Na2S·9H2O, and K4Nb6O17.

Characterization of the Photocatalysts. The crystal structures and the phases of the samples were studied by X-ray diffractometry (XRD) using a Rigaku D/MAX2500 PC diffractometer with Cu Kα radiation, with an operating volt-age of 40 kV and an operating current of 100 mA. The morphologies of the samples were probed using scanning electron microscopy (SEM) (Hitachi, s-4800) and trans-mission electron microscopy (TEM) (Jeol ltd, JEM-2010). The chemical compositions of the sample were tested using an energy dispersive X-ray detector (EDS, Thermo Noran 7). UV-visible light (UV-vis) diffuse reflectance spectra were recorded on a UV-vis spectrometer (Puxi, UV1901), and the luminescence of the powdered samples was measured on a spectrofluorometer (Hitachi, f7000). The chemical states of the photocatalysts were analyzed by XSAM800 X-ray photoelectron spectroscopy (XPS). The Brunauer-Emmett-Teller (BET) sorption measurements were performed on a JW-BK nitrogen adsorption apparatus at 77.3 K.

The photocatalytic activities of hydrogen production for the photocatalyst samples were examined in an overhead irradiation system. Typically, 0.3 g photocatalyst powder was dispersed in a Pyrex reaction cell containing 100 mL of aqueous solution of 0.1 M Na2S, 0.5 M Na2SO3 and 1 M KOH. Cooling was provided by an external cooling jacket, and the temperature of the reaction was controlled to 25 ± 2 °C. The light source used was a 300 W xenon lamp (with the light with wavelengths λ < 400 nm filtered out by a filter). The suspensions were deairated with Ar gas for 30 min prior to irradiation to prevent uptake of photo-generated electrons by dissolved oxygen. The produced hydrogen gas was detect-ed using an online gas chromatography system (SHIMADZU-GC-2014C, molecular sieve 5 A column, TCD detector, Ar carrier). Ar gas was used as a carrier for the products.

Figure 1.XRD patterns of Cd1-xZnxS solid solutions containing various molar ratios of Cd and Zn.

 

Results and Discussion

Catalyst Characterization. XRD patterns for Cd1-xZnxS materials prepared at various Zn molar concentrations (x-values) are shown in Figure 1. The obtained ZnS exhibited diffraction peaks at 2θ values of 28.608°, 47.589° and 56.471°, originating from the (111), (220) and (311) crystal planes of cubic zincblende ZnS (JCPDS 65-0309), respec-tively. The characteristic diffraction peaks for CdS were observed at 2θ of 24.807°, 26.507°, 28.182°, 36.620°, 43.681°, 47.879° and 51.880°, which were attributed to the (100), (002), (101), (102), (110), (103) and (112) crystal planes of the CdS crystal, respectively. This was indexed to hexagonal CdS (JCPDS 65-3414). The XRD spectra obtain-ed for the Cd1-xZnxS material varied from those of CdS and ZnS. The main strong peaks in Cd1-xZnxS were shifted com-pared to those of pure CdS and ZnS, and the intensities were reduced. The diffraction peaks of the photocatalysts were also observed to be shifted to higher angles as the Zn molar concentrations increased. This continuous shift indicated that the resulting crystals obtained were not simple physical mixtures of CdS and ZnS particles, but were Cd1-xZnxS solid solutions. The Zn2+ was considered to be incorporated into the CdS lattice or to have entered its interstitial sites, since the radii of Zn2+ ions (0.74 Å) were smaller than those of Cd2+ (0.97 Å).19-21 Moreover, the electronegativities of Cd (1.69) and Zn (1.65) were very close, which increased their likelihood of forming a solid solution.22 According to Scherrer’s equation,23 the average crystallite size of Cd0.8Zn0.2S, Cd0.5Zn0.5S, and Cd0.2Zn0.8S were calculated to be 15.8 nm, 10.5 nm, 8.6 nm, respectively. The size of the Cd1-xZnxS solid solution became smaller with increasing Zn concent-ration, which may have been due to the small radii of the Zn2+ ion compared to that of Cd2+, which led to a decrease in the crystal lattice parameters.

The XRD patterns for K4Nb6O17, Cd0.8Zn0.2S, and Cd0.8Zn0.2S/K4Nb6O17 are given in Figure 2, respectively. The characteristic diffraction peaks for Cd0.8Zn0.2S were observed at 2θ values of 24.835°, 26.526°, 28.203°, 36.648°, 43.737°, 47.869° and 51.871°, and were attributed to the (100), (002), (101), (102), (110), (103), and (112) crystal planes of the Cd0.8Zn0.2S crystal, respectively, and the spectra was indexed to hexagonal Cd0.8Zn0.2S (JCPDS 49-1302). The XRD patterns of the prepared potassium niobate were also indexed to pure phase orthorhombic K4Nb6O17, according to JCPDS (31-1064). The diffraction peaks of the Cd0.8Zn0.2S (100), (002), (101), (110) and (103) crystal planes also appeared in the XRD patterns of Cd0.8Zn0.2S/K4Nb6O17 composites, suggesting that Cd0.8Zn0.2S and K4Nb6O17 were successfully synthesized and mixed together. The diffraction peaks obtained in the spectra for the Cd0.8Zn0.2S/K4Nb6O17 composite were weaker than those observed in the pure K4Nb6O17 pattern, indicating that the surface of K4Nb6O17 was covered by Cd0.8Zn0.2S particles.

Figure 2.XRD patterns of Cd0.8Zn0.2S, K4Nb6O17 and Cd0.8Zn0.2S/K4Nb6O17.

The morphologies of the samples were observed by SEM and TEM, HRTEM and SAED, as shown in Figure 3. From the SEM images shown in Figure 3(a)-(g), the Cd1-xZnxS particles prepared with x values ranging from 0 to 1 were well developed, and exhibited a size distribution between 20 nm to 100 nm. As shown in Figure 3(a), CdS exhibited nearly spherical, well-crystallized morphology and a relative-ly uniform size distribution of approximately 100 nm. From Figure 3(b)-(f), the particle size was found to gradually decrease with increasing Zn content, while the morphology became increasingly irregular. The particle size of ZnS was found to be approximately 20 nm, as seen in Figure 3(g). The smaller primary ZnS nanocrystals were also observed to form huge granular aggregates. From Figure 3(b)-(g), the Cd1-xZnxS particles displayed a sheet-like morphology. This varied from that of CdS, which may have been due to an associated change in the crystal structure from hexagonal to cubic.24 From the SEM image shown in Figure 3(h), the K4Nb6O17 particles were found to exhibit a relatively uni-form size distribution of approximately 1-3 μm. The morpho-logy of the prepared Cd0.8Zn0.2S/K4Nb6O17, shown in Figure 3(i), indicated that nano-Cd0.8Zn0.2S particles were uniform-ly dispersed on the surface of K4Nb6O17 with a size distri-bution distribution within 50 nm. In addition, at a higher resolution, the TEM image shown in (Figure 3(j)) also indicated that each of these nanoparticles was in fact composed of even smaller primary nanocrystals with crystallite size of 10-20 nm, which matched well with the average crystallite size derived from the XRD data. The synthesized Cd0.8Zn0.2S/K4Nb6O17 composite was also observed by HRTEM (Figure 3(j)), and the fringe spacing was found to be about 3.36 Å, which was in good agreement with the interplanar distance of the (002) plane of Cd0.8Zn0.2S, further confirming the existence of Cd0.8Zn0.2S in the composite. The selected area electron diffraction (SAED) pattern taken from an individual particle is shown in Figure 3(m), with the presence of the ring pattern evidencing that the nanoparticles obtained were polycrystal-line. Three diffraction rings corresponding to the (100), (002), (101) planes of Cd0.8Zn0.2S were observed, respectively. The EDS spectrum obtained from the Cd0.8Zn0.2S (25 wt %)/K4Nb6O17 composite is given in Figure 3e. In this spectrum, peaks associated with O, Zn, K, Nb, S and Cd were observed, corresponding to K4Nb6O17 (K, Nb, O peaks, respectively) and Cd1-xZnxS (Cd, Zn, S peaks, respectively). The EDS results confirmed that the obtained product was comprised of K4Nb6O17 loaded with Cd1-xZnxS. To further explore the chemical compositions of the composites, XRF analysis was employed to determine the as-prepared Cd1-xZnxS and Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 samples, as given in Table 1. It can be seen that the metal contents of Cd and Zn in each Cd1-xZnxS sample were slightly reduced compared to the theoretical ratio, which was similar with the reported literature.25 For x=0(CdS catalyst) the molar ratio of Cd and S was found to be 47.68(mass %):51.15(mass %), indicating that the sample was slightly S-rich. The Cd2+ and Zn2+ ions that were not used for the formation of solid solution were thought to be washed off when XRF samples were collected. The Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 sample was also mea-sured by XRF, the contents of elements Cd, Zn and S were found to be 16.55(mass %), 2.34(mass %) and 6.11(mass %), respectively, yielding a molar ratio of Cd:Zn:S of 0.77:0.19:1, and the calculated value of Cd0.8Zn0.2S content in the composite was 23.87%. The results confirmed that the actual loading content of Cd0.8Zn0.2S onto the layered host material was in good agreement with the theoretical values.

The optical absorptions of the as-prepared Cd1-xZnxS samples were investigated using UV-vis diffuse reflectance spectra, and the results obtained are given in Figure 4. The absorption edge of ZnS was observed at approximately 410 nm, and the absorption edge of the Cd1-xZnxS solid solution was gradually red shifted with increasing content of Cd, with the pure CdS exhibiting an absorption edge at about 550 nm. This shift may have been caused by a band transition of the Cd1-xZnxS solid solution, which also further proved that CdS and ZnS successfully formed a solid solution. The band gaps of these samples were estimated from the onset of the absorption edges, and found to to be between 2.26 eV and 3.03 eV (for x values between 0 and 1, respectively). There-fore, the band gap position of the Cd1-xZnxS solid solution could be adjusted by adjusting the ratio of the composition of CdS and ZnS particles. The sharp absorption curve also implied that the absorption was almost only attributed to the band gap transition of electrons.26

Figure 3.SEM images of prepared photocatalysts: (a) CdS, (b) Cd0.8Zn0.2S, (c) Cd0.6Zn0.4S, (d) Cd0.5Zn0.5S, (e) Cd0.4Zn0.6S, (f) Cd0.2Zn0.8S, (g) ZnS, (h) K4Nb6O17, (i) Cd0.8Zn0.2S(25 wt %)/K4Nb6O17; TEM and HRTEM images of Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 (j and k) and corresponding SAED pattern and EDS pattern (l and m).

The UV-vis diffuse reflectance spectra for various com-posite photocatalysts prepared using various weight percents of Cd1-xZnxS on K4Nb6O17 are given in Figure 5. The ab-sorption edge of K4Nb6O17 was observed at approximately 400 nm, corresponding to a band gap of 3.1 eV. Therefore, most of the light absorption from the layered host material was in UV range. In contrast, the Cd0.8Zn0.2S sample showed strong absorption in the range of λ < 530 nm, corresponding to a band gap of approximately 2.36 eV. Upon introduction of the Cd0.8Zn0.2S into the composite, the absorption edge in the resulting materials was noticeably red-shifted to approxi-mately 530 nm. The absorption spectra obtained for all of the prepared Cd0.8Zn0.2S/K4Nb6O17 composites possessed this visible-light absorption, and the absorption intensity gradually increased with increasing amount of Cd0.8Zn0.2S. This indicated that the visible light absorption in the com-posite samples was due to distinct specimens of Cd0.8Zn0.2S, and that the external surface of K4Nb6O17 was gradually covered with Cd0.8Zn0.2S. When the Cd0.8Zn0.2S loading content reached 30 wt %, the absorption inflection due to K4Nb6O17 was less noticeable, indicating that most of the K4Nb6O17 surface was covered by Cd0.8Zn0.2S nanoparticles.

Table 1.Composition of solid solutions, as measured by XRF

Figure 4.UV-vis diffuse reflectance spectra of prepared photo-catalysts: (a) ZnS, (b) Cd0.2Zn0.8S, (c) Cd0.4Zn0.6S, (d) Cd0.5Zn0.5S, (e) Cd0.6Zn0.4S, (f) Cd0.8Zn0.2S, (g) CdS.

Molecular fluorescence spectroscopy is a kind of emission spectrum caused by electron-hole recombination, which can reflect the migration and capture of photo-induced carriers.27 The results obtained for the prepared Cd1-xZnxS samples are given in Figure 6(a). The emission peak exhibited more or less sharp peaks at 434, 468, 488, 514, 548 nm, respectively. It can be seen that, for each sample, the PL peak energy was consistently lower than its corresponding band gap due to the direct transition between valence band and conduction band, which indicated that the radiative transition occurred from the surface states rather than the excitation transition.28 Moreover, the broad PL peaks obtained shifted toward higher energies with increasing Cd content, due to the formation of nanocrystalline Cd1-xZnxS solid solutions. The shift of emission peak was thought to be attributable to the changes in surface structure as the Zn substituted Cd in the CdS lattice, influencing the emission from surface defects.

Figure 5.UV-vis diffuse reflectance spectra of prepared photo-catalysts: (a: K4Nb6O17, b: Cd0.8Zn0.2S(5 wt%)/K4Nb6O17, c: Cd0.8Zn0.2S (10 wt%)/K4Nb6O17, d: Cd0.8Zn0.2S(15 wt %)/K4Nb6O17, e: Cd0.8Zn0.2S (20 wt%)/K4Nb6O17, f: Cd0.8Zn0.2S(25 wt %)/K4Nb6O17, g: Cd0.8Zn0.2S (30 wt %)/K4Nb6O17, h: pure Cd0.8Zn0.2S).

Figure 6.Fluorescence spectra of photocatalysts (λex = 250 nm): a: Cd1-xZnxS; b: (a) K4Nb6O17; (b) Cd0.8Zn0.2S (5 wt %)/K4Nb6O17; (c) Cd0.8Zn0.2S (15 wt %)/K4Nb6O17; (d) Cd0.8Zn0.2S (25 wt %)/K4Nb6O17.

As shown in Figure 6(b), the pure K4Nb6O17 exhibited an excitation peak around 370 nm. Compared to pure K4Nb6O17, the positions of the emission peaks of Cd0.8Zn0.2S/K4Nb6O17 composites were hardly shifted, indicating that the emission was due to K4Nb6O17 in the composite. However, the emission intensity was drastically decreased. This may be have been due to the transition of photo-induced electrons from the CB to VB in the bulk K4Nb6O17 being inhibited and the recombination of electron-hole pairs suppressed due to the Cd0.8Zn0.2S loading. This also suggested that the energy bands of K4Nb6O17 and Cd0.8Zn0.2S were coupled upon Cd0.8Zn0.2S loading, and that the K4Nb6O17 and Cd0.8Zn0.2S particles were not physically mixed. Moreover, the intensity of the emission peak decreased as the loading amount of Cd0.8Zn0.2S increased, which may have been because an increased Cd0.8Zn0.2S amount would have increased the probability of transfer for the excited electrons from the K4Nb6O17.29,30

X-ray photoelectron spectroscopy (XPS) analysis was carried out to study the chemical states of the surface ele-ments of the synthesized samples. The XPS spectra of Nb 3d, Cd 3d, Zn 2p, and S 2p are given in Figures 7-10. As given in Figure 7, the binding energy of Nb 3d5/2 and Nb 3d 3/2 for K4Nb6O17 appeared at 206.35 eV and 208.75 eV, respectively. After Cd0.8Zn0.2S loading onto K4Nb6O17, the binding energy of Nb 3d5/2 and Nb 3d3/2 for K4Nb6O17 increased to 207.05 eV and 209.5 eV, respectively. However, the binding energies of Cd 3d, Zn 2p and S 2p for Cd0.8Zn0.2S/K4Nb6O17 were all lower than those observed for pure Cd0.8Zn0.2S. Therefore, chemical bonds were formed between the Cd0.8Zn0.2S and K4Nb6O17, leading to the variation of these binding energies of Nb, Cd, Zn and S, respectively.31 As the electronegativities of Cd and Zn were higher than that of Nb, the surrounding electronic density of Nb atoms in Cd0.8Zn0.2S/K4Nb6O17 decreased, and the shielding effect became weaker, which led to the increase in the binding energy of Nb 3d.32 Hence, with the increase of the surrounding electronic density of the Cd, Zn and S atoms of Cd0.8Zn0.2S in the composite, the binding energy became lower. This also indicated that the ions diffused between Cd0.8Zn0.2S and K4Nb6O17 in the Cd0.8Zn0.2S/K4Nb6O17 com-posite, and a solid solution was formed.

Figure 7.Nb XPS spectra for various samples.

Figure 8.Cd XPS spectra for various samples.

Figure 9.Zn XPS spectra for various samples.

Figure 10.S XPS spectra for various samples.

Figure 11.Photocatalytic activity for hydrogen evolution under visible light irradiation using Cd1-xZnxS samples with various values of x (zinc concentration).

Photocatalytic Activities for Hydrogen Production. The photocatalytic activities of Cd1-xZnxS samples prepared with various zinc concentrations (x-values) were investigated for hydrogen evolution under visible light irradiation, and the results are shown in Figure 11. After 3 h of visible light irradiation, hydrogen generation was not observed in the presence of pure ZnS, indicating that ZnS was not active for photocatalytic hydrogen evolution under visible light, in accordance with its UV-vis diffuse reflectance spectra. The activity of CdS was also low, and the hydrogen production was about 1.031 mmol/g in three hours. In contrast, the Cd0.8Zn0.2S showed high photocatalytic activity, with a photo-catalytic hydrogen evolution of 3.074 mmol/g under visible light irradiation in three hours. As shown in Figure 11, the final hydrogen evolution observed in the presence of the Cd0.2Zn0.8S solid solution was 0.935 mmol/g, which was much lower than that of Cd0.8Zn0.2S, suggesting that dissolu-tion of CdS and ZnS may have occurred, and that the composition of the solid solution prepared with an x-value of 0.2 was significant for the high activity of photocatalytic hydrogen evolution observed. The photocatalytic activity of the solid solutions gradually decreased as the Cd/Zn molar ratio decreased, and it was thought that the addition of wide bandgap catalyst, such as ZnS, to that of CdS decreased the extent of light absorbed. This was considered disadvantage-ous in terms of light absorption of the photocatalyst. Addi-tionally, for the Cd1-xZnxS solid solution, the conduction band potentials were thought to be more negative as the value of x increased.33

Figure 12.Photocatalytic activity for hydrogen production under visible light using Cd0.8Zn0.2S/K4Nb6O17 photocatalysts prepared at various loadings.

The photocatalytic performance of Cd0.8Zn0.2S/K4Nb6O17 samples prepared at various Cd0.8Zn0.2S loadings were evaluated by measuring photocatalytic hydrogen evolution under visible light irradiation for 3 h. As shown in Figure 12, no H2 was detected using pure K4Nb6O17 under visible light irradiation, while all of the Cd0.8Zn0.2S/K4Nb6O17 samples exhibited higher photocatalytic activities than that of pure K4Nb6O17, due to the enhancement of activity upon Cd0.8Zn0.2S loading. This improved activity was thought to be attribut-able to various factors. Firstly, loading Cd0.8Zn0.2S onto K4Nb6O17 was found to enlarge the catalyst surface area according to the results of BET measurements, where the pure K4Nb6O17 and Cd0.8Zn0.2S/K4Nb6O17 possessed BET surface areas of 0.91 m2/g and 8.693 m2/g, respectively. Secondly, the band gaps of K4Nb6O17 and Cd0.8Zn0.2S were thought to be coupled in the composite, which may have improved the separation of electron-hole pairs, as discussed in subsequent sections. The Cd0.8Zn0.2S loading amount also influenced the photocatalytic performance, and a suitable content of Cd0.8Zn0.2S particles was necessary to obtain a fine particle dispersion on the surface of K4Nb6O17. How-ever, too little Cd0.8Zn0.2S on the surface would lead to limited visible light absorption capability of the composite, while excessive loading would form overlapping particle agglomerates and shade the active sites on the surface of K4Nb6O17. Based on these considerations, and from the results given in Figure 12, the Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 composite was found to exhibit the highest activities for photocatalytic hydrogen evolution, producing about 8.278 mmol/g hydrogen in 3 h under visible light irradiation. Therefore, 25 wt % was thought to represent an optimum Cd0.8Zn0.2S loading onto the layered K4Nb6O17.

Stability is an important parameter for photocatalysts. To evaluate the catalyst stability, we performed repeated runs for the photocatalytic activity over 0.3 g of Cd0.8Zn0.2S(25 wt %)/K4Nb6O17, recovering the photocatalyst between runs. As shown in Figure 13, after five cycles, the photocatalytic activity for hydrogen evolution was maintained, suggesting that the Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 composite possess-ed good stability for repeated use in photocatalytic reactions.

Figure 13.Repeated runs of photocatalysis over recycled sample 0.3 g of Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 in 0.1 M Na2S, 0.5 M Na2SO3, and 1 M KOH under visible light.

Some Cd0.8Zn0.2S composite have already been studied before, MWCNTs/Cd0.8Zn0.2S was prepared by Liu, etc,34 15 wt % MWCNTs/Cd0.8Zn0.2S showed the highest photocata-lytic hydrogen production rate and producing 1.17 mmol·g−1·h−1 H2 under visible light irradiation. Macias-Sanchez, etc35 synthesized Cd1-xZnxS/SBA-16 and studied the visible light photocatalytic activity for water splitting, the result showed that the Cd0.8Zn0.2S/S16 possessed the highest hydrogen evolution rate and generating 0.98 mmol·g−1·h−1 H2. Li etc36 prepared Cd0.5Zn0.5S/NiS photocatalyst, with the H2 evolution rate reaching 2.32 mmol·g−1·h−1. However, the photocataytic for hydrogen evolution for the present work of Cd0.8Zn0.2S (25 wt %)/K4Nb6O17 was 2.76 mmol·g−1·h−1, which exhibit-ed higher photocatalytic activity than the materials mention-ed above, indicating that the band gaps of Cd0.8Zn0.2S and K4Nb6O17 coupled well and offered an increased probability of separation of electron and hole pairs.

Mechanistic Analysis. The Cd0.8Zn0.2S/K4Nb6O17 com-posites prepared possessed excellent activity for hydrogen evolution under visible light irradiation, and the photocata-lytic mechanisms were analyzed considering the transition of the photo-induced charge carriers in the composite. Speci-fically, using two semiconductors in contact with different redox energy levels of their CB and VB enhanced the the efficiency of separation of photo generated carriers and also enhanced the interfacial charge transfer.

As shown in Scheme 1, according to the relative positions of the valence and conduction bands of K4Nb6O17 and Cd0.8Zn0.2S, respectively, when irradiated by visible light, the Cd0.8Zn0.2S solid solution absorbed photons and pro-moted electrons from its valence band to its conduction band to form the electron-hole pairs. The electric field at the Cd0.8Zn0.2S/K4Nb6O17 interface was thought to then push the photo-generated electrons toward the conduction band of K4Nb6O17, which caused the electrons to further migrate into the inner surface of the layered structure of K4Nb6O17,37 while the photo-generated holes remained on the valence band of Cd0.8Zn0.2S. As a result, the photogenerated electron- hole pairs were effectively separated and the probability of elec-tron-hole recombination was reduced.

Scheme 1.Schematic illustration of electron-hole transfer under visible light.

In the absence of sacrificial reagents, hydrogen was gene-rated in the photocatalytic process and the catalyst consumed by photoexcited holes, as shown in Eqs. (1)-(3). However, when S2− and SO32− were utilized, various reactions occurred for the photoexcited holes, as detailed in Eqs. (4)-(7).38

In this case, photoexcited holes reacted with S2− and SO32−, respectively. The SO22− ions produced could then act as an optical filter and electron acceptor, capturing the photogenerated electrons. However, this pathway was efficiently sup-pressed by mixing with SO32−ions, causing ionic S2O32− and S2− to be produced (6). Therefore, the overall process was represented by the main reaction shown in Eq. (7). As the major reaction product, S2O32− was assumed to have little negative impact on the reaction. Based on this analysis, the sacrificial reagents were thought to be transformed into S2O32− as the photocatalytic reaction proceeded.

 

Conclusion

Cd1-xZnxS/K4Nb6O17 composite photocatalysts were success-fully prepared by a facile deposition-precipitation method. The Cd1-xZnxS/K4Nb6O17 composites exhibited strong visible light absorption and displayed enhanced photocatalytic performance for hydrogen evolution under visible light irradiation. The Cd0.8Zn0.2S(25 wt %)/K4Nb6O17 photocata-lyst exhibited the highest photocatalytic activity under visible light irradiation, and the hydrogen evolution was 8.278 mmol/g after 3 h. This enhanced activity was thought to be due to the increase in surface area of the composite relative to the pure layered material alone. Additionally, the band gaps of K4Nb6O17 and Cd1-xZnxS were coupled in the composite, which improved the separation of electron-hole pairs.