Introduction

Hydrogen is regarded as a clean energy carrier due to its high energy density and zero carbon emission during combustion (Gangadharan et al. 2017). Among various hydrogen production methods, photocatalytic water splitting with semiconductors as photocatalysts and sustainable solar light as energy source is the most ideal way (Cao and Yu 2014; Matsuoka et al. 2007). However, the photo-driven separated hole and electron pairs could recombine easily on the pure semiconductors, leading to the poor activity and limited application into practice (Xiang et al. 2012). Nitroaromatic compounds are important organic pollutants in the water system, which are difficult to be degraded under natural environment for the strong withdrawing ability of the nitro group (Mohamed and Al-Sharif 2013; Peng et al. 2016). Photocatalytic reduction of nitroaromatic compounds into the corresponding amino-aromatics is an effective pre-treatment process for these pollutants, and the obtained amino containing organics are much easier to be degraded with amino group present as an electron donating group (Peng et al. 2016).

Both the photocatalytic hydrogen evolution and nitroaromatic reduction need effective photocatalysts. Cadmium sulfide (CdS) is a widely used semiconductor with a band gap of 2.42 eV, which is therefore a promising visible light-driven photocatalyst (Gao et al. 2013). However, the activity of bare CdS is still poor due to the rapid recombination of photogenerated electron-hole pairs, which also lead to its photo-corrosion and weak stability. Compositing with kinds of co-catalysts, such as noble metal nanoparticles (Huang et al. 2013; Singh and Pal 2013), the photo-activity of CdS can be improved obviously. However, these metals are rare and expensive to be applied.

Graphene, which consists of layered sp2-hybridized carbon structure, has remarkably high crystal quality and good electronic property. Due to the excellent charge carrier rate (200,000 cm2V−1 s−1) at room temperature, graphene could act as effective co-catalyst for many semiconductors (Fan et al. 2015). By using graphene as support and co-catalyst, photo-activity of the composite photocatalysts can be greatly increased, mainly owing to enhanced charge transfer ability and effective electron-hole separation. Doping with non-metal heteroatoms (B, N, S, etc.) could re-distribute the surface charge density and improve the electronic properties of graphene, thus could enhance its co-catalytic activities further (Choi et al. 2013; Han et al. 2017; Putri et al. 2015). Han et al. (2018) reported that CdS nanoparticles modified by N-doped rGO exhibited higher photocatalytic activity than pure CdS both for photocatalytic H2 evolution and 4-nitrophenol reduction. Furthermore, CdS NWs-rGO nanocomposites also showed enhanced performance for selective photocatalytic reduction of nitro aromatic compounds in water (Han et al. 2015). Yao’s group has done lots of researches about the graphene materials with abundant of structure defects, and proved that the five- or seven-ring structure defects in graphene were more active than the heteroatoms doped graphene in electrochemical water splitting (Jia et al. 2019; Ortiz-Medina et al. 2019; Yan et al. 2018). Therefore, the defected graphene should be potential candidate as an effective co-catalyst for semiconductor photocatalysts.

In this study, a series of CdS-based photocatalysts were synthesized, and several graphene materials were selected as co-catalysts for the activity enhancement of CdS. The photocatalytic behaviors of these nanocomposites were tested on photocatalytic 4-nitrophenol reduction and H2 generation under visible light irradiation (λ ≥ 420 nm). It was found that the defected graphene had the best co-catalytic activity compared with rGO and N-rGO.

Experiments

Materials and reagents

Cadmium acetate dehydrate (Cd(CH3COO)2·2H2O), 4-nitrophenol, sodium sulfide (Na2S), and ammonium nitrate (NH4NO3) were purchased from Aladdin Chemistry Co., Ltd., Shanghai, China. Graphite powder (> 99.99%) was purchased from HWRK Chem Co., Ltd., Beijing, China. Lactic acid was purchased from J&K Scientific Ltd., Beijing, China. All the reagents were used as received without further purification. The deionized water was used in all experiments.

Synthesis of N-rGO

Graphene oxide (GO) was prepared using the modified Hummers’ method (Hummers and Offeman 1958). N-doped reduced graphene oxide (N-rGO) was synthesized using our previous method (Duan et al. 2015). Typically, 1 g freeze-dried GO was uniformly dispersed in 100 mL of ethanol; 1 g NH4NO3 was then added into the suspension. Heating was kept under stirring to remove the solvent at 50 °C. The mixture was then put into the muffle furnace for calcination at 350 °C for 1 h to obtain the final N-rGO sample.

Synthesis of D-rGO

D-rGO was obtained from the N-graphene precursor. In a typical experiment, N-rGO was annealed at 1150 °C for 2 h with a ramp rate of 5 °C under nitrogen atmosphere. The doped nitrogen atoms will be then removed from the N-rGO for the production of topological defects.

Synthesis of CdS-based composites

A total of 4 mmol Cd(CH3COO)2·2H2O was dissolved in 40 mL DI water, and then 20 mL Na2S (0.2 M) solution was added under stirring. After stirring for 24 h, 150 mg pre-prepared CdS was mixed with graphene materials (rGO, N-rGO, or D-rGO) in 50 mL ethanol; the solvent was then evaporated at 60 °C. After that, the samples were annealed in argon flow at 400 °C for 2 h to obtain the final materials.

Characterization

The XRD data of each powder sample were recorded on a D8 Focus Diffractometer (Bruker-Nonius) equipped with a Cu Kα X-ray tube for phase identification. The morphologies of each sample were observed by a scanning electron microscopy (SEM) equipped with a secondary electron detector (Hitachi S-4800). The Raman spectra of powder samples were recorded on a HORIBA LABRAM HR Evolution having optical resolution of 0.03 cm−1, in the range from 500 to 2500 cm−1 in KBr disk medium (633 nm laser). The Fourier transform infrared spectroscopy (FT-IR) was recorded on a Thermo-Nicolet 380. The specific surface area (SSA) and pore size distribution of samples at different temperature were measured by Bjbuilder SSA-7000 surface area and pore size analyzer using the BET method after degassing by heating at 120 °C for 12 h with He-gas purging.

Photocatalytic activity test

The photocatalytic H2 production was conducted in a Pyrex reaction cell connected with an online photocatalytic hydrogen production system (CEL-SPH2N, CEAULight, Beijing). A 300-W Xe lamp (CEL-HXF300, CEAULight, Beijing) with a UV-cutoff filter (λ ≥ 420 nm) was used as light source. Fifty milligrams of the photocatalyst was dispersed in the lactic acid solution (10%). Before the reaction, the system was degassed 30 min to remove the dissolved air. The reaction temperature was controlled at 6–10 °C during the whole experiment. The generated H2 was analyzed by an online gas chromatograph (GC-7920, CEAULight, China) equipped with a TCD detector. Photocatalytic reduction of 4-NP was performed under the same light source. For a typical test, 20 mg photocatalyst and 50 mg ammonium formate (unless otherwise specified) were dispersed in 50 mL 4-NP solution (20 mg L–1). Before the reaction, the suspension was stirred in the dark condition for 2 h to allow the adsorption of reactant by the photocatalyst. During the reduction process, a N2 flow (80 mL min–1) was kept. At a 2-min interval, the samples were taken out and analyzed on a UV–vis spectrophotometer (UNICO UV-3802) at the maximum wavelength of 4-NP (380 nm). For the recycle test, the mixture solution was filtered after the reaction, and the obtained materials were then washed with water and ethanol. After the drying process at 60 °C, the regenerated materials are then used for the reuse. The PEC measurements were performed in a standard three-electrode quartz cell using an electrochemical workstation (CHI660E, Chenhua, Shanghai) under the same light source. The fluorine-doped SnO2 (FTO, < 15 Ω square−1) glass electrode coated with samples was used as working electrode with an active area of 1.0 cm2. The Pt plate and Ag/AgCl (saturated KCl) electrodes were used as counter electrode and reference electrode, respectively. And 0.5 M Na2SO4 aqueous solution was used as the electrolyte.

Results and discussions

Materials characterization

Figure 1 a shows the XRD patterns of the prepared carbon materials. The XRD pattern of GO shows a sharp (001) diffraction peak at 10.9° due to the presence of oxygen-containing functional groups. After the thermal reduction and nitrogen doping, the characteristic peak at 10.9° disappears, and the left broad diffraction peak of (002) is at ~ 22°. This value is earlier than the deeply reduced graphene, indicating the partial reduction of GO and the presence of left functional groups. The further treatment at high temperature (1150 °C) removes most of the doped nitrogen atoms and left functional groups, leading to the red-shift of peak position. p-CdS is synthesized using the deposition method without a further annealing process. XRD patterns of pure p-CdS, CdS, and CdS/D-rGO nanocomposite and the standard diffraction pattern of hexagonal CdS are shown in Fig. 1b. Compared with CdS, p-CdS has poorly crystalline structure and wide diffraction peaks because of the low preparation temperature (25 °C). No obvious difference is found between CdS and CdS/D-rGO due to weak reflection of the low percentage of D-rGO (Bai et al. 2017; Peng et al. 2016; Zhang et al. 2013).

Fig. 1
figure 1

a XRD patterns of GO, rGO, N-rGO, and D-rGO. b XRD patters of p-CdS, CdS, and CdS/D-rGO composites

Full size image

The FT-IR spectra of these samples are shown in Fig. 2a. Abundant of functional groups are present on GO, including 3430 cm−1 for O-H stretching vibration, 1733 cm−1 for C=O stretching vibration, 1623 cm−1 for C=C stretching vibration, 1426 cm−1 for O-H bending vibration, and 1259 and 1087 cm−1 for C-O stretching vibrations. After the annealing, most of the oxygenous groups are removed, and the presence of NH4NO3 could keep more groups anchored on the carbon sheets. The high-temperature treatment could then reduce the N-rGO further, and nearly no oxygenous groups are left for D-rGO.

Fig. 2
figure 2

FT-IR (a) and Raman spectra (b) of the carbon materials

Full size image

Raman spectra could be used to study the structural defect of GO, rGO, N-rGO, and D-rGO. The two characteristic peaks observed at about 1340 and 1590 cm−1 correspond to the D and G bands of rGO (Fig. 2b). The intensity ratio of D to G (ID/IG) could show the defect level of carbon crystals (Duan et al. 2015; Li et al. 2017). According to Fig. 2b, rGO has a higher ID/IG value (1.73) than that of GO (1.51). And the ID/IG value of N-rGO increased to 1.83 for the doping process could change the stable six-carbon ring structure. The ID/IG value of D-rGO is the largest of ~ 2.0, indicating the presence of abundant defects after the high-temperature treatment. It is reported that the high-temperature annealing could remove the doped nitrogen atoms, and form five- or seven-ring structures, thus creating more structure defects (Jia et al. 2019).

Figure 3 a and b show that the SEM images of GO and N-rGO are with wrinkle and transparent layers. After high-temperature calcination, more defects are generated on D-rGO, and some pores are observed clearly (Fig. 3c). The un-supported CdS nanoparticles are aggregated together with a diameter of ~ 50 nm (inset of Fig. 3d). As shown in Fig. 3d, the CdS nanoparticles with a diameter of ~ 25 nm are dispersed on the surface of D-rGO uniformly.

Fig. 3
figure 3

SEM images of rGO (a), N-rGO (b), D-rGO (c), and CdS/D-rGO (d) composites. Inset of d is the pure CdS nanoparticles without any carbon materials

Full size image

Figure 4 a illustrates the N2 adsorption-desorption isotherms of GO, rGO, N-rGO, and D-rGO. The isothermal curves of these materials are typical type IV with a capillary condensation step. The larger hysteresis loop of D-rGO indicates the formation of more mesopores attributing to the removal of heteroatoms. The rGO material has the largest N2 adsorption ability. Figure 4 b shows the BET surface areas of the samples, and GO has the smallest specific surface area (SSA) of 20.4 m2/g. However, the SSA value for rGO increases to ~ 170.8 m2/g due to the exfoliation of graphene layers by thermal annealing. Due to the addition of organic precursors into GO, the SSA decreases to ~ 67.5 m2/g for N-rGO (Duan et al. 2015). For D-rGO, the subsequent calcination at high temperature leads to the slight increase of SSA to ~ 73.8 m2/g by creating more structure defects and digging pores on the surface. The SSA for CdS is only ~ 30 m2/g due to the obvious aggregation. Small percentage of D-rGO and the composite of CdS/D-rGO have decreased SSA compared with the pure CdS.

Fig. 4
figure 4

N2 adsorption-desorption isotherms (a) and corresponding BET surface areas (b) of GO, rGO, N-rGO, and D-rGO, and the isotherms for CdS and CdS/D-rGO composites (c)

Full size image

Photocatalytic tests

The photocatalytic activities of these samples are then investigated for 4-nitrophenol reduction, photo-electric transformation, and hydrogen evolution. The adsorption abilities of the samples are first investigated on CdS and CdS/D-rGO. As shown in Fig. S1, nearly no 4-nitrophenol (< 1%) could be adsorbed on pure CdS. The SSA of CdS/D-rGO is much larger than that of CdS by the introduction of D-rGO as co-catalyst. While the adsorption capacity does not increase obviously, less than 2% of the 4-nitrophenol could be removed only by adsorption under dark condition. Although with weak adsorption, 30-min stirring is used to achieve the adsorption-desorption equilibrium before each photocatalytic reaction. As shown in Fig. 5a, the naked CdS shows weak photo-activity for 4-nitrophenol reduction, and ~ 57% could be reduced into 4-aminophenol within 14 min. With the addition of rGO, the reduction percentage could be increased into 90% within 14 min. By the nitrogen doping modification, the N-rGO exhibits better enhancement ability than rGO probably due to the charge redistribution of the surface. All the 4-nitrophenol could be reduced within 12 min. High-temperature annealing could remove most of the doped nitrogen atoms and generate amounts of structure defects, which can be used as active sites for 4-nitrophenol reduction. Therefore, with D-rGO as co-catalyst, the reduction time could be decreased obviously into 6 min. The reaction rate constants of these materials are then calculated using a zero-order reaction model. The rate constant of CdS is only 0.034 min−1, and this constant can be increased to 0.063 min−1 and 0.075 min−1 with rGO and N-rGO as co-catalysts, respectively. The CdS/D-rGO has the largest rate constant of 0.168 min−1, nearly 5 times faster than that of the pure CdS, which is much larger than lots of reported catalysts (Table S1). The stability test of CdS/D-rGO is also performed, and no obvious deactivation can be observed during the first three cycles. This might be due to the intimate contact between CdS and D-rGO obtained by annealing at higher temperature.

Fig. 5
figure 5

Photocatalytic reduction of 4-nitrophenol on the samples (a), the average hydrogen evolution rates of different photocatalysts (b), and the hydrogen evolution of CdS/D-rGO for a long-term test (c)

Full size image

The photocurrents are observed for all these samples under visible light irradiation, and the photocurrent rapidly dropped to zero once the light is cut off. The largest current of pure CdS is only ~ 2 mA/cm2, which will be decreased rapidly to ~ 1 mA/cm2 due to the photo-corrosion. CdS/D-rGO exhibited much higher photocurrent density (~ 8 mA/cm2), indicating that the presence of D-rGO could enhance the charge separation and transfer (Fig. S2). The photocatalytic performance of these composites is then investigated for the H2 evolution (Fig. 5b). The photocatalytic H2 evolution process is carried out under visible light irradiation (λ ≥ 420 nm) using lactic acid as the sacrificial agent. Although CdS is a visible light response semiconductor, the pure CdS NPs has a relatively low H2 evolution rate of 15.2 μmol h−1 g−1 due to the fast recombination of photogenerated electron-hole pairs. Loading rGO as co-catalyst could obtain increased photocatalytic activity, and the H2 evolution rate can be increased to 22.1 μmol h−1 g−1. By the combination with N-rGO, the photocatalytic H2 evolution rate is increased gradually to ~ 50.1 μmol h−1 g−1. The average rate of H2 evolution is significantly increased to 102.7 μmol h−1 g−1 when D-rGO is used as co-catalyst, larger than many reported catalysts (Table S1). In addition, the stability of CdS/D-rGO is also tested, and the hydrogen evolution rate can remain for 5 h without obvious deactivation (Fig. 5c).

Enhanced mechanism

During the above photocatalysis process, CdS is activated and hole-electron pairs are separated under irradiation. The electrons can then be used for reducing 4-nitrophenol into 4-aminophenol or reducing H+ to produce H2. On the pure CdS, the electron and hole will however recombine easily, thus leading to the weak photocatalytic activity (Ma et al. 2016). The absolute potential of D-rGO/D-rGO•− should be similar with that of rGO/rGO•− (− 0.08 V versus NHE) (Jia et al. 2014), which is more positive than the CB potential of CdS (− 0.52 eV versus NHE). Thus, the photogenerated electrons would transfer from the CB of CdS to the surface of D-rGO and be accepted by the defect sites for the reduction of 4-nitrophenol or H+. This electron transfer process could inhibit the recombination of hole-electron pairs, resulting in the activity enhancement (Han et al. 2015).

According to the Raman spectra (Fig. 2b) and SEM image (Fig. 3c), there are abundant of structure defects on D-rGO, thus could provide more active sites for the utilization of photogenerated electrons. Moreover, it is reported that the D-rGO has smaller Gibbs energy for H* absorption (|∆GH*|) than N-rGO (Jia et al. 2016) and rGO, thus should have better HER activity than N-rGO and rGO. Therefore, using D-rGO as co-catalyst, the photogenerated electrons can be used for H+ and 4-nitrophenol reduction more effectively (Fig. 6).

Fig. 6
figure 6

The photocatalytic mechanism on the CdS/D-rGO photocatalyst

Full size image

Conclusions

In this study, rGO, N-rGO, and D-rGO are used as co-catalysts for the modification of CdS, and the obtained composites are effective for photocatalytic 4-nitrophenol reduction and H2 evolution. The D-rGO has good charge mobility and decreased |∆GH*|, and the presence of D-rGO on CdS could increase the surface area and provide more active sites for solar energy utilization, thus leading to the enhanced photocatalytic activity of CdS/D-rGO. The novel CdS/D-rGO composite contains no noble metals, thus should have great potential in the photocatalysis field.