Ioannis
Daskalakis
a,
Ioannis
Vamvasakis
a,
Ioannis T.
Papadas
b,
Sotirios
Tsatsos
c,
Stelios A.
Choulis
b,
Stella
Kennou
c and
Gerasimos S.
Armatas
*a
aDepartment of Materials Science and Technology, University of Crete, Heraklion 70013, Greece. E-mail: garmatas@materials.uoc.gr
bMolecular Electronics and Photonics Research Unit, Department of Mechanical Engineering and Materials Science and Engineering, Cyprus University of Technology, Limassol 3041, Cyprus
cDepartment of Chemical Engineering, Surface Science Laboratory, University of Patras, Patra 26504, Greece
First published on 9th October 2020
Transition metal sulfides have been emerging as one of the most attractive and prospective catalysts for the direct conversion of solar energy into chemical fuels. Their intriguing compositional and electronic characteristics and their feasibility for integration in porous architectures endow metal sulfide materials with superior activity for photochemical catalysis. In the present work, high-surface-area Cu-doped ZnS nanocrystal (NC)-linked mesoporous frameworks are successfully synthesized for use as cost-effective catalysts for photochemical hydrogen evolution. Benefiting from the suitable band-edge alignment and enhanced visible light absorption resulting from the interfacial charge transfer between ZnS and Cu2S NCs, there is a spatial separation of charge carriers which leads to excellent activity for photocatalytic hydrogen production. Moreover, the results obtained here show that surface defect passivation through a wet-chemical sulfidation process effectively increases the photochemical performance of the composite catalysts by improving the transport efficiency of electrons at the Cu2S/ZnS interface and changing the Helmholtz layer potential drop at the ZnS/Cu2S/electrolyte junction. Thus, a remarkable improvement of 1 mmol h−1 gcat−1 for hydrogen evolution is observed with the sulfide-treated Cu2S/ZnS catalyst containing 5 mol% Cu, which is associated with a 17.6% apparent quantum yield under 410 nm irradiation. This work provides an interesting strategy for enhancing the interface charge transfer properties and hydrogen evolution activity of metal sulfides by surface defect engineering with sulfide ions.
Recently, II–VI compound semiconductors, especially zinc sulfide and selenide (ZnS/Se), have drawn intensive attention in energy-storage and photochemical devices, such as Li-ion batteries,5 solar cells6 and light-induced photocatalysts.7 Their main advantages are the tunability of electronic structures, high intrinsic electrical conductivity, rich redox activity, and strong reduction ability of photoinduced electrons for hydrogen evolution.8 Nevertheless, ZnS-based photocatalysts, such as S-defective ZnS microspheres9 and ZnS microparticles,10 usually suffer from limited solar light absorption and/or poor electron–hole separation yield, which severely impede their viable applications. To overcome these limitations and enhance the visible-light photocatalysis of ZnS, recently, various strategies such as synthesis of solid solutions (e.g., Zn1−xCuxS and (AgIn)xZn2(1−x)S2),11 construction of heterojunctions (e.g., Au-decorated ZnS and Ru-loaded ZnS),12 and optimization of the morphology (e.g., fabrication of 3D ZnS nanoarchitectures)13 have been employed in the synthesis of ZnS-based photocatalysts. Among them, construction of heterojunctions (electronic contacts) between ZnS and a narrow bandgap semiconductor provides one of the most efficient ways to realize high photochemical performance. ZnS-based heterostructures have been proved as efficient catalysts for chemical energy conversion due to the band bending at the junction interface, which provides effective transfer and separation of the photogenerated charge carriers.14 Furthermore, the multi-componential nature of these materials can enhance the overall catalytic efficiency due to the spatial separation of reduction and oxidation active sites on the surface of different components – the excited electrons can participate in reduction reactions on one surface while the holes can be consumed by the oxidation reaction at a separate surface. Therefore, the rational design of nanocomposite catalysts with desired morphologies and appropriate electronic band structures is critically important to achieve a more stable and enhanced photocatalytic performance.
In this work, high-surface-area mesoporous networks of Cu-doped ZnS nanocrystals (NCs) were synthesized and investigated as photocatalysts for visible light irradiated (λ ≥ 420 nm) reduction of water to hydrogen. Although ZnS is a wide bandgap semiconductor that absorbs only UV light (it has a 3.5–3.9 eV energy gap), Cu+ doping can extend its visible-light response by introducing tail states (electron donor levels) within the bandgap of ZnS.15 On the other hand, the Cu+ dopant will also induce the formation of sulfur vacancies which may act as electron and hole-trapping centers.16 For this purpose, we systematically study the effect of Cu+ doping on the electronic band structure and visible-light photochemical activity of the assembled Cu/ZnS NCs. To this end, porous Cu-doped ZnS nanoarchitectures with a large surface area (up to 313 m2 g−1) and various compositions (the Cu doping level varies from 2 to 10 mol%) were successfully prepared by a two-step chemical process that involves polymer-assisted aggregating assembly of colloidal ZnS NCs, followed by chemical transformation of constituent nanoparticles into Zn–S–Cu composites. Moreover, on a particular composite catalyst, we show that surface defect sites can be passivated to a large extent by a sulfidation process on the NC-linked structure, apparently leading to improved charge transfer at the Cu2S/ZnS interface and better photocatalytic activity. This finding, which is supported by optical absorption and electrochemical spectroscopic studies, is especially important as it may offer a potential solution to the poor photochemical performance of defective interfacial structures. Lastly, a mechanism for visible-light-driven photocatalytic hydrogen evolution in the present system is proposed according to the experimental results.
![]() | ||
Scheme 1 Schematic representation of the synthesis of mesoporous ZnS (ZS) and Cu-doped ZnS (CZS) NC-linked assemblies (NCAs). |
The incorporation of Cu+ ions into the ZnS lattice may lead to the formation of interface defect sites at the ZnS/Cu2S junction according to eqn (1), in which the exchange of Zn2+ with Cu+ (Cu′Zn) involves the production of S-atom vacancies in the Zn–S–Cu lattice for charge balance.
![]() | (1) |
Such S-related defect sites can induce the formation of mid-gap electronic states (deep or shallow trap states), which may serve as trap centers for charge carriers, causing electron–hole pair recombination. Therefore, controlling the quantity of defects in the ZnS lattice is critical in photocatalytic reactions. To improve the charge transport and catalytic efficiency, we post-treated the CZS-5 sample, which is the most active catalyst in this series, with sulfide (S2−) solution in order to replace missing sulfur atoms in the composite structure. The resultant sulfurated catalyst was denoted as CZS′-5.
Energy-dispersive X-ray spectroscopy (EDS) analysis showed that the CZS heterostructures are composed of Zn, Cu and S elements with different molar ratios, which correspond to a Cu loading from ∼2 to ∼10 mol% (see Table S1†). Of particular note, the Cu content, as determined by EDS analysis, is well consistent with the expected composition from the stoichiometry of reactions. This suggests the complete incorporation of Cu+ ions into the ZnS lattice.
The morphology and crystal structure of the as-prepared materials were examined by combining X-ray diffraction (XRD) and transmission electron microscopy (TEM). Fig. 1a shows the XRD patterns of the mesoporous ZS and CZS NCAs. All the diffraction peaks can be indexed to the zinc blende structure of ZnS with a lattice parameter a0 = 5.41 Å (space group F4m, JCPDS no. 05-0566). The broadening of the X-ray diffraction peaks signifies the small grain composition of the materials; in particular, the average diameter of the crystallites deduced from the peak broadening using the Scherrer equations is about 3 nm. After the sulfidation process, the crystal structure of pristine CZS-5 is well retained, as shown in Fig. 1a. XRD analysis, however, does not provide any insightful information about the presence of copper sulfide (CuxS) species, even in the XRD pattern of the 10 mol% Cu-doped sample. This is attributed to the high dispersion of Cu+ ions in the lattice of ZnS. Nevertheless, our further studies with electron diffraction and X-ray photoelectron spectroscopy confirmed the existence of the Cu2S phase, indicating that the Cu+ ions successfully replace the Zn2+ ions during the ion-exchange reaction (see below).
Fig. 1b and c display typical TEM images of the as-prepared CZS-5 sample. The images show that small-sized nanoparticles are interconnected with each other to form a network-like structure with high porosity. From the magnified TEM image in Fig. 1c, the size of the constituent nanoparticles is found to be around 5–6 nm, which is close to the size of starting nanoparticles (ca. 5 nm).19 Further analysis by high-resolution TEM (HRTEM) reveals that the assembled structure possesses well-defined lattice fringes throughout the entire nanoparticles with interplanar spacings of 2.7 and 3.1 Å, which can be ascribed to the (200) and (111) planes (along the [011] zone-axis) of cubic ZnS (Fig. 1c, inset), in agreement with XRD results. Moreover, the comparison of the calculated electron diffraction patterns and the experimentally recorded ones demonstrates the presence of the zinc blende structure of ZnS along with the cubic structure (space group Fmm) of Cu2S; in Fig. 1d, the selected area electron diffraction (SAED) pattern shows a series of Debye–Scherrer diffraction rings that could be indexed to the (111), (220), (311) and (422) facets of ZnS (marked with yellow curves) and the (200), (220) and (311) facets of Cu2S (marked with blue curves). Combined with the XRD patterns, these results clearly indicate that the pore walls consisted of connected cubic zinc blende ZnS and cubic chalcocite Cu2S NCs. This supports the notion that the transformation of ZnS into Zn–S–Cu is topotactic in nature, that is, the crystal structure of ZnS is preserved after transformation.
X-ray photoelectron spectroscopy (XPS) was employed to investigate the phase composition and chemical nature of the composite Zn–S–Cu framework. Fig. 2 shows the XPS spectra for the pristine and sulfurated CZS-5 samples. All binding energy values in the XPS spectra were calibrated according to the C 1s line at 284.6 eV. The carbon element mainly originates from atmosphere contamination during the handling of the samples. In the CZS-5 material, the Zn 2p3/2 photoelectron peak appears at a binding energy of 1021.6 ± 0.2 eV (Fig. 2a). However, this position does not allow for an unambiguous identification of Zn(II) compounds, since the literature values for Zn 2p3/2 binding energies are quite similar to ZnS (1021.6 eV) and ZnO (1022.1 eV).20 For this reason, the Auger α parameter, that is, the binding energy of the Zn 2p3/2 peak + the kinetic energy of the Zn L3M45M45 Auger peak, was evaluated and used to confirm the presence of ZnS. According to the literature, the Zn Auger parameter value varies from ∼2010.1 to ∼2011.3 eV for ZnS, while the corresponding values range from ∼2009.5 to ∼2010 eV for ZnO.21 For the CZS-5 NCAs, the Auger parameter is found to be 2010.3 ± 0.2 eV, which probably reflects the existing phase of ZnS (Fig. 2b). The Cu 2p3/2 core-level signal appears at a binding energy of 932.1 ± 0.2 eV, which can be ascribed to the monovalent copper ions (Fig. 2c), consistent with other reported results.22 Generally, the 2p3/2binding energy for the compounds containing Cu+ (e.g., Cu2O, Cu2S) locates at 932.2–932.8 eV,23 while that for the compounds containing Cu2+ ions locates around 933–935 eV,24 accompanying with a characteristic “shake-up” satellite peak at 940–945 eV. Taken together the SAED analysis, these results are consistent with the presence of Cu2S species in the ZnS lattice. This is further corroborated by the presence of a single peak in the core-level spectrum of the S 2p region at 162 ± 0.2 eV, which is attributed to the sulfide (S2−) species.
Interestingly, during the sulfidation process, some of the S2− ions are incorporated into the Zn–S–Cu host lattice, causing a partial oxidation of Cu+ to Cu2+ (eqn (2)).
Cu2S(s) + S2−(aq) → 2CuS(s) + 2e− | (2) |
The presence of paramagnetic Cu2+ (probably as CuS and/or surface Cu(OH)2 groups) in the sulfurated sample was confirmed by the Cu 2p XPS spectrum due to the appearance of the weak “shake-up” satellite peak in the higher binding energy range (∼944.3 eV), as seen in Fig. 2c.25 In addition, a small amount of sulfur species at certain higher oxidation states is, however, observed in the surface of CZS′-5, judging from the weak feature at 169 ± 0.2 eV in the S 2p region (Fig. 2d).
The porosity of the resulting materials was probed with nitrogen physisorption measurements at −196 °C. Fig. 3 and Fig. S1† display the N2 adsorption and desorption isotherms and the corresponding pore-size distribution plots of the mesoporous ZS, CZS and CZS′-5 NCAs. All the isotherms showed typical type-IV behavior associated with a H2-type hysteresis loop (according to IUPAC classification) at higher relative pressures, which are characteristic of mesoporous structures with interconnected pores.26 The mesoporous CZS samples possess a Brunauer–Emmett–Teller (BET) surface area of 150–235 m2 g−1 and a total pore volume of 0.14–0.18 cm3 g−1, which are slightly lower than those of the pristine ZS material (314 m2 g−1, 0.28 cm3 g−1). The decrease in the surface area and pore volume for the Cu-doped samples is presumably due to a partial cleavage of the NC-linked network during the ion-exchange process. Furthermore, the surface area of the sulfurated material becomes lower than that in the untreated sample (yet remains high enough to enable mass transport and diffusion between the nanoparticles), possibly due to an etching-like effect of S2− ions on the nanoparticle-assembled structure.27 The pore size in these materials was determined from the adsorption branch of isotherms using the non-local density functional theory (NLDFT) model. The results of ZS, CZS and CZS′-5 NCAs reveal a quite narrow pore size distribution with an average size of ∼4.4–4.8 nm (insets of Fig. 3 and Fig. S1†). All the textural parameters for the as-prepared materials are given in Table 1.
![]() | ||
Fig. 3 N2 adsorption and desorption isotherms at −196 °C and (inset) the corresponding pore-size distribution plots for the mesoporous ZS and CZS NCAs. |
The well-defined electronic structure of the title materials leads to strong optical absorption in the ultraviolet-visible (UV–vis) spectrum, ranging from ∼320 nm for ZS to ∼400 nm for the CZS NCAs (Fig. 4), which corresponds to the interband VB-to-CB electron transition. The electronic band structure of these materials is characterized by a large energy gap, which is estimated to be 3.88 eV for pristine ZS and 3.09 to 3.75 eV for the CZS NCAs (Table 1), using the Tauc method for a direct band gap semiconductor (Fig. 4, inset). The large hypsochromic (blue) shift in the energy gap of mesoporous ZS relative to the bandgap of bulk zinc blende ZnS (∼3.7 eV)28 is attributed to the quantum confinement in the ZnS nanoparticles due to the very small grain size (ca. 5–6 nm according to the TEM observation). Moreover, the decrease in the energy gap with increasing Cu content is attributed to the strong electron interactions between the electronic states in the ZnS and Cu2S NCs. This indicates that the Cu2S species are mainly incorporated into the structure of ZnS and do not grow on the ZnS surface. In addition, the incorporation of Cu2S into the ZnS lattice leads to an increase in the light absorption between 380 and 500 nm, which is helpful for improving the visible-light absorbance and thus the photocatalytic efficiency of the catalyst.
![]() | ||
Fig. 4 UV–vis diffuse reflectance spectra and Tauc plots (inset) for mesoporous ZS, CZS and CZS′-5 NCAs. |
Next, we investigated the optimized performance of the CZS-5 catalyst by measuring the hydrogen evolution with different concentrations of the catalyst and types of sacrificial agents. Control experiments showed that there is an increase of the H2 evolution rate with increasing catalyst addition until reaching a maximum at 1 g L−1 (Fig. S2†). We interpret the dependence of H2 evolution on the catalyst concentration as an enhancement of light absorption by the catalyst nanoparticles. However, at an excess concentration of the catalyst (more than 1.5 g L−1), a saturation level is reached due to the full absorption of the incident light and/or light scattering by the colloidal particles, so that the generation rate of hydrogen becomes slightly lower. In the following, we compared the hydrogen generation activity of the CZS-5 NCAs in the presence of Na2S/Na2SO3 pairs, methanol, and ethanol in neutral and alkaline (5 M NaOH) solutions as electron donors. All these photocatalytic reactions were conducted under visible-light irradiation (λ ≥ 420 nm) using a fixed mass of the catalyst (1 g L−1). The results showed that, for the present system, the S2−/SO32− pairs are the most efficient sacrificial electron donors, resulting in a significant reaction enhancement (Fig. S3†).
Furthermore, the photocatalytic hydrogen evolution efficiency was considerably improved after sulfidation of the CZS-5 catalyst. Catalytic results in Fig. 5a show that CZS′-5 outperforms all other samples, showing a hydrogen evolution rate of 30.0 μmol h−1 (∼1 mmol h−1 gcat−1) within 4 h, which is about 1.7 times higher than that of the untreated catalyst. The apparent quantum yield (AQY) for this reaction, assuming that all incident photons are absorbed by the catalyst suspension, is estimated to be 17.6% at 410 ± 10 nm. Combined with further electro- and photochemical studies (see below), we attribute the superior hydrogen evolution efficiency of CZS′-5 to the chemical passivation of defect states (mainly sulfur vacancies arising from the charge compensation of Cu+ substitution with Zn2+ sites). In particular, the light excitation of CZS′-5 generates more surface-reaching electrons for hydrogen evolution, instead of losing a portion to electron–hole recombination at the structure or surface defects. The AQY of the CZS′-5 NCAs is also higher than or comparable to those of other noble-metal-free ZnS-based photocatalysts, such as CdS/ZnS core–shell microparticles (9.3% AQY at 420 nm),29 Cd0.8Zn0.2S solid-solution particles (10.23% AQY at 420 nm),30 CdS QD-sensitized Zn1−xCdxS solid solution (Cd/Zn ratio = 5) (6.3% at 420 nm),31 (Zn0.95Cu0.05)0.67Cd0.33S solid solution NCs (15.7% AQY at 420 nm),32 CuS(5.9%)/Zn0.65Cd0.35S nanospheres (8.1% AQY at 420 nm),33 AgIn5S8/ZnS heterostructures (3.7% AQY at 420 nm),34 Cd0.1Cu0.01Zn0.89S particles (9.6% AQY at 420 nm),35 Cu(OH)2-loaded ZnO/ZnS nanobranches (11.5% AQY at 420 nm),36 CuS/ZnS porous nanosheets (20% AQY at 420 nm),37 (CuAgZnSnS4)0.9(ZnS)0.4 mixed crystals (0.25% AQY at 400 nm),38 Bi (0.3%)-doped ZnS hollow spheres (0.99% AQY at 420 nm)39 and Ga(0.1%), Cu(0.01%)-co-doped ZnS nanospheres (0.14% AQY at 425 nm).40 Interestingly, when sulfidation was performed on the starting ZnS NCs or mesoporous network of the cross-linked ZnS NCs to form the sulfurated products (denoted as C/ZS-S′-5 and C/ZS′-5 NCAs, respectively, details in the Experimental section), the resulting catalytic activity was significantly lower; namely, C/ZS-S′-5 and C/ZS′-5 catalysts gave a H2 production rate of 3.9 and 22.5 μmol h−1 over a 3 h reaction period, respectively (Fig. S4†). This further confirms that the photochemical enhancement achieved by the CZS′-5 catalyst for hydrogen evolution is due to the passivation of defect sites generated mainly at the Cu2S/ZnS interface (as a result of the incorporation of Cu+ into the ZnS mesostructure).
Results of the repeated photocatalytic hydrogen evolution tests showed that CZS′-5 remains stable during the photocatalytic process. The stability of the catalyst was tested by performing three consecutive 5 h photocatalytic cycles. After each reuse, the catalyst was isolated from the reaction mixture by centrifugation, washed with a polysulfide solution to remove the adsorbed sulfur species, and re-dispersed in a fresh reaction solution. Before each photocatalytic reaction, the reaction cell was de-aerated by purging with argon to remove dissolved oxygen. As shown in Fig. 5b, the CZS′-5 catalyst maintains its photocatalytic activity (within 5% experimental error) and an amount of ∼0.39 mmol of H2 was detected (ca. 8.8 mL STP) after 15 h of irradiation. The small decline of the H2 evolution rate after three 5 h cycles of reuse may be caused by the mass loss of the catalyst during the recovery process, although minor photocorrosion of the sulfide catalyst during irradiation is a possible explanation. Elemental analysis of the reused catalyst by EDS showed a distribution of Zn and Cu elements with a ∼49.1:
2.6 Zn/Cu molar ratio that corresponds to a Cu content of about 5 mol%, in agreement with the composition of the fresh material (Table S1†). From N2 physisorption data, the BET surface area of this sample is found to be 102 m2 g−1 and the pore size to be 4.6 nm (Fig. S5a†) which are very close to those of fresh CZS′-5. Also, the TEM study showed that the reused catalyst maintains its crystallinity and structural integrity during catalysis (Fig. S5b and S5c†). In addition, powder XRD and XPS data confirm that the zinc blende structure of ZnS and the chalcocite structure of Cu2S are well retained after cycle reactions (Fig. S6†), further evidencing the structural stability of the catalyst under the examined conditions.
Sample | Flat band potential, EFB (V vs. NHE) | VB energy, Ev (V vs. NHE) | Donor density, Nd (cm−3) |
---|---|---|---|
ZS | −0.88 | 3.00 | 9.32 × 1016 |
CZS-2 | −1.07 | 2.68 | 8.80 × 1016 |
CZS-5 | −1.12 | 2.20 | 5.56 × 1016 |
CZS-10 | −1.02 | 2.07 | 3.86 × 1016 |
CZS′-5 | −0.95 | 2.41 | 6.74 × 1016 |
On the basis of the EFB values and optical band gaps, in Fig. 6b we propose the band energy diagram for each catalyst. Here we assumed that the flat band lies very close to the CB edge position for heavily doped n-type ZnS. The VB energy (Ev) is thus calculated to be 2.07 to 3.00 V vs. NHE by subtracting the corresponding band gap values (based on the optical absorption spectra) from the measured EFB energies, as shown in Table 2. It is likely that the negative shift in the EFB potential upon Cu+ doping is due to the formation of mid-gap Cu 3d and sulfur vacancy states near the CB edge of ZnS (n-type dopants), which results in an upshift of the Fermi level.41 As for the decrease of Nd concentration in ZnS with increasing Cu+ doping level, it can be explained by the formation of the p–n junction created between ZnS and the Cu2S nanoparticles; Cu2S is a well-known semiconductor with p-type conductivity.42 In particular, given that the Fermi level of p-type Cu2S (ca. 5.1 eV)43 is positioned well below the Fermi level of n-type ZnS (ca. 4.6–4.8 eV),44 a built-in electrical potential should be established at the Cu2S/ZnS interface. This thus leads to an electron flow from the ZnS CB to Cu2S until the chemical potentials of two semiconductors reach equilibrium. The transfer of photoinduced electrons may lead to a deformation of the band structure, that is, a depletion layer in ZnS and an accumulation layer on the Cu2S surface at the p–n Cu2S/ZnS junction. The formation of a depletion region in the ZnS phase coincides with the observed decrease in the Nd concentration with increasing Cu content in the CZS NCAs.
On the other hand, sulfidation of the CZS-5 NCAs causes a positive shift in EFB level (−0.95 V vs. NHE) and an increase in the Nd concentration (6.74 × 1016 cm−3). Since surface (particularly low-coordinated face and edge atoms) termination with S2− ions should not affect the intrinsic doping properties of the CZS-5 material, it is likely that the observed down-shift in the flat-band position is due to the surface-state passivation of the composite structure by S2− ions that suppress the Fermi-level pinning. In particular, the charged surface-states (i.e., mid-gap localized electronic states on the semiconductor surface) can induce a large potential drop across the Helmholtz double layer (VH), which, however, can be reduced to a large extent by surface passivation.45 In this case, the Fermi level of the semiconductor becomes almost independent from the allowed energy levels of the bulk and is essentially “pinned” at the energy level of the surface-states. Thus, the Fermi level position and band bending at the surface of the semiconductor will be affected by changes in VH, as shown in eqn (3).
EFB = VH + ϕSC − 4.5 | (3) |
For a specific solution pH, the Helmholtz layer potential drop depends on the point of zero charge of the semiconductor surface (pHPZC) according to eqn (4).
VH = 0.059 × (pHPZC − pH) | (4) |
To elucidate this possibility, the effect of surface sulfide ions on the charge of CZS colloids was probed through zeta potential measurements using a 0.5 M Na2SO4 solution at pH 7 (similarly to the EIS experiments). The zeta potentials of the CZS-5 and CZS′-5 NCAs were measured as −12.7 and −10.2 mV, respectively. Clearly, the particles in these suspensions are all negatively charged and their corresponding pHPZC values are lower than 7, consistent with literature results (ZnS has a pHPZC value of 4.2).46 Moreover, the less negative zeta potential of the sulfurated sample (CZS′-5) implies a more positive VH value at the ZnS/Cu2S/solution junction, which, according to eqn (3), is in line with the observed positive (anodic) shift of EFB (see Table 2). As for the slight increase in the carrier density (Nd) of CZS′-5, this could be tentatively attributed to the lower recombination rate and faster electron transport at the Cu2S/ZnS interface. This assumption is also supported by further electrochemical and photoluminescence studies.
EIS Nyquist measurements of the synthesized electrodes (FTO substrates covered with ZS and the CZS NCAs samples) allowed us to derive information on the charge transfer properties of the assembled nanostructures (Fig. 6c). The charge-transport resistance (Rct) of drop-cast films can be determined from fitting the EIS data using an equivalent circuit model (the inset of Fig. 6c) and the values were obtained as 100.5 Ω for ZS and as 102.6, 102.9 and 105.4 Ω for CZS-2, CZS-5 and CZS-10, respectively. The descending trend of the charge transfer efficiency with increasing Cu content indicates explicitly that Cu+ doping contributes to a lower electron mobility in the mesoporous structures. In particular, the high level of Rct suggests a diminished charge transfer presumably along the composite structure, which correlates with the loss of surface-reaching electrons to interface recombination at the Cu2S/ZnS junction. In contrast, the sulfurated CZS′-5 catalyst exhibited the lowest Rct resistance (95.6 Ω) among the examined samples, which can be attributed to the smooth charge transfer within the composite Zn–S–Cu structure due to relatively low defect concentrations at the Cu2S/ZnS interface. Note that the EIS slope of the linear portion for CZS′-5 is comparable with that of CZS-5 in the low frequency region (Fig. 6c), indicating that the diffusion resistance (Rs) is similar in these materials. This implies that the diffusion rate of the electrolyte to the surface-active sites does not specify the effectiveness of the catalysts, and other inherent properties, such as electrical conductivity and charge mobility, may have a more profound influence on the underlying photocatalytic hydrogen evolution behavior. In agreement with this, further photocurrent measurements performed at an applied voltage of 1 V (see details in the Experimental section) showed that the sulfurated CZS′-5 catalyst generates a higher photocurrent density (13.5 mA cm−2) than does CZS-5 (10.3 mA cm−2) under 400–800 nm light irradiation. This means that the CZS′-5 catalyst effectively reduces the charge recombination within the composite structure and affords a higher photoinduced electron transfer and separation efficiency compared to CZS-5. In combination with the above electrochemical results, these findings provide an interesting strategy for enhancing free exciton dissociation and charge transfer properties at the interface of the sulfide treated Cu2S/ZnS junctions. The charge transfer efficiency of the CZS heterostructures is furthermore evident from photo-luminescence (PL) spectroscopy analysis. Fig. S7† shows the comparison of the PL spectra of mesoporous ZS with those of the CZS-5 and sulfurated CZS′-5 samples at an excitation wavelength of 330 nm. ZS shows a broad PL signal at ∼442 nm, which is related to the radiative relaxation of excitons through shallow in-gap levels.47 This suggests that defect-localized states such as sulfur vacancies (n-type dopants) are present near the surface of the ZnS NCs, which serve as recombination centers for electrons and holes. In contrast, the CZS-5 heterostructure shows a noticeable decrease of the PL emission, which is attributed to the efficient interfacial electron transfer along the Cu2S/ZnS junctions. Of note, after treatment with S2−, the corresponding defect-related emission (at 442 nm) of CZS′-5 almost vanished, again confirming that the defect states within the Zn–S–Cu composite structure are efficiently suppressed, leading to a higher delocalization of electron–hole pairs, in line with the above EIS and photocurrent experiments.
The above analyses directly demonstrate that Cu+ doping in the ZnS NCs prominently affects their optical response and charge transport properties. Overall the Cu2S/ZnS junctions facilitate the spatial distribution of charge carriers through the appropriate alignment in the band-edge positions of the composites. This effect definitely has a profound impact on the photocatalytic behavior of the CZS heterostructures. The mechanism for photocatalytic H2 evolution in the CZS NCAs can be depicted as in Fig. 6d. It is considered that, under visible light irradiation, Cu2S gets excited and generates electron–hole pairs; Cu2S has an energy band gap of about 1.2–1.8 eV.48 In addition, during the course of irradiation, a fraction of photoexcited electrons from the VB of ZnS may be transferred to the CuS clusters (formed by the partial oxidation of Cu+ from S2− ions according to eqn (2)), causing their partial reduction to Cu2S (E° = −0.91 V vs. NHE, pH = 7). Given that the transition energy from the VB of ZnS to Cu2S/CuS is approximately 2.9–3.0 eV (∼415–430 nm), this electron transfer appears feasible. Note that the absence of the XPS signal corresponding to Cu2+ in the CZS′-5 catalyst examined after catalysis is highly related to this reaction scheme (Fig. S6d†). In this respect, the Cu2S/CuS species can also function as cocatalysts facilitating the charge separation through the interfacial electron transfer from ZnS to Cu2S/CuS. Besides, photoexcited Cu2S may also react with protons and sulfide ions from the solution to form molecular hydrogen, and even CuS, according to the following equations:
Cu2S(s) (2e−) + 2H+(aq) → Cu2S(s) + H2(g) | (5) |
Cu2S(s) + S2−(aq) + 2H+(aq) → 2CuS(s) + H2(g) | (6) |
Meanwhile, driven by the internal electron field at the p–n junction interface, the photogenerated holes in the VB of ZnS can migrate to the VB of the Cu2S NCs where they effectively oxidize the sacrificial reagents (S2−/SO32−). Therefore, an efficient charge separation and transfer is successfully achieved through the p–n junctions of the Cu2S and ZnS NCs, leading to an enhanced photocatalytic hydrogen production activity.
For a comparative study, Cu-doped ZnS NC assemblies with 5 mol% Cu2+ content (denoted as C(II)ZS-5 NCAs) were also prepared following a similar procedure, using CuCl2 as a precursor.
Other sulfurated Cu-doped ZnS samples were also prepared as following: 1 mmol of colloidal ZnS NCs and 1 mmol of Na2S were dispersed in 10 mL of water, and the mixture was kept under stirring for 2 h at room temperature. Afterwards, the S2−-treated NCs were isolated by precipitation with the addition of 2-propanol, and dried at 60 °C for 24 hours. Next, a mesoporous network of 5% Cu-doped ZnS NCs (denoted as C/ZS-S′-5) was prepared following the method as described above, by using S2−-treated ZnS NCs as starting building blocks. In a different experiment, 1 mmol of mesoporous ZS NCAs was dispersed in 10 mL of water containing 1 mmol of Na2S and the resulting mixture was stirred for 2 h. The product was isolated by centrifugation, washed several times with water and ethanol, and dried at 60 °C for 12 h. As a next step, the S2−-treated ZnS mesoporous sample was doped with Cu+ (5 mol%) following the method as described above, giving the sulfide-treated 5% Cu/ZnS mesoporous sample (denoted as C/ZS′-5).
The working electrodes were fabricated as follows: 10 mg of each sample was ultrasonically mixed with 1 mL of deionized water to form a homogeneous suspension. Glass slides (1.5 × 2 cm2) coated with a fluorine-doped tin oxide (FTO, 9 Ω sq−1) substrate were cleaned thoroughly by sonication in acetone and then in isopropanol for 15 min and dried at 100 °C for 10 min. Next, the FTO substrates were further treated with a UV-ozone plasma for 5 min. The samples were drop-cast into the FTO substrates, which were masked with an epoxy resin to leave an exposure area of 1.0 cm2, and heated for 30 min at 60 °C.
The apparent quantum yield, AQY = (2 × NH2)/Nhv where NH2 and Nhv are the numbers of evolved H2 molecules and incident photons, respectively, was estimated by obtaining the amount of evolved hydrogen at a λ = 410 ± 10 nm irradiation wavelength. The incident photon number was determined with a StarLite power meter equipped with an FL400A-BB-50 thermal sensor (Ophir Optronics Ltd).
Footnote |
† Electronic supplementary information (ESI) available: EDS and EIS results, N2 physisorption isotherms, catalytic data, XRD patterns, and XPS and PL spectra of ZS, CZS-n and CZS′-5 NCAs. See DOI: 10.1039/d0qi01013h |
This journal is © the Partner Organisations 2020 |