Introduction

In recent years, bismuth compounds have attracted extensive concerns as a preferred photochemical catalyst in the visible range. Bismuth compounds mainly include bismuth oxide (Yan et al. 2014), halogen-free bismuth oxide (Zhang et al. 2008b), bismuth sulfide (Manna et al. 2014), bismuth vanadate (BiVO4) (Kudo et al. 1998), bismuth tungstate (Natarajan et al. 2014), carbonate bismuth oxide (Allured et al. 2014), bismuth titanate (Allured et al. 2014), and so on. Due to Bi 6s orbital and O 2p orbital hybridization, the semiconductor oxide has a small band-gap energy and shows great potential in the photochemical and photocatalytic field (Wang et al. 2014a). Among compounds such as bismuth oxyhalide (Br, Cl, and I) (Natarajan et al. 2016), Bi2O3 (Azman et al. 2016), Bi2S3 (Uddin et al. 2016), and Bi2O2(CO3) (Chang et al. 2016), BiVO4 is an inexpensive, stable, and non-toxic ternary compound semiconductor, with direct semiconductor band gap of about 2.4 eV and strong visible-light absorption capacity. However, pure BiVO4 is subjected to two disadvantages, of which one is the low absorption of incident light and the other is its non-porous structure. To eliminate the first disadvantage, researchers have made attempts to modify BiVO4 in the hope of improving its photocatalytic absorption efficiency and of inhibiting the recombination of electron carriers. Major modification methods include metal ion modification (Xu et al. 2008; Zhang and Zhang 2010), non-metallic particle modification (Tang et al. 2013; Yin et al. 2013), and semiconductor composite modification (Tang et al. 2013; Yin et al. 2013). Among these modification methods, metal ion modification is less recommended due to the higher cost of precious metal deposit, toxicity of some metals, inactivation of some catalysts, as well as the thermal instability of the load of metals serving as the center for electron–hole recombination. Therefore, this modification method can only improve photocatalytic activity in a very narrow range of load metal ions. As non-metallic ions can enter into a semiconductor and replace oxygen, resulting in stable chemical bonds with metal ions, the technique of non-metal ion doping has attracted worldwide attention. By doping such non-metallic ions as F, C, N, and S, the band structure of BiVO4 can be adjusted, thereby improving its photocatalytic properties (Geng et al. 2014; Huang et al. 2015). In response to the second disadvantage, it is mainly needed to improve the structure, morphology, and particle size of BiVO4, so as to make great improvement of its photocatalytic performance (Libor Kvítek et al. 2008; Wang et al. 2014a). Among all polymorphs of BiVO4, only monoclinic BiVO4 (m-BiVO4) has a band gap of 2.4 eV, allowing it to have visible absorption capacity (Strobel et al. 2008; Yu et al. 2009; Zhang et al. 2009a, b). Structure, morphology, and particle size are greatly influenced by the preparation methods and conditions. So far, the preparation methods of BiVO4 include solid-phase method, hydrothermal synthesis method, solvent co-precipitation method, micro-emulsion method, ultrasonic-assisted method, microwave-assisted method, metal-organic deposition, flame spraying method, etc. (Geng et al. 2014; Lim et al. 1998; Tang et al. 2013; Yin et al. 2013; Zhang and Zhang 2010). Conventional law of solid-phase synthesis (Lim et al. 1998) requires higher temperature and longer reaction duration, but results in large particles and small surface area, which are not conducive to the absorption of the incident light or to the improvement of photocatalytic activity. Although the traditional co-precipitation method is simple, it may lead to poor purity and uncontrollable morphology and size of the product. In recent years, the hydrothermal synthesis of BiVO4 photocatalyst has attracted increasing attention for its simple operation process, easiness in structure control, and good morphology of materials. Through adjusting the synthesis conditions such as pH value, temperature, reaction time, and the type of structure-directing agent, we can prepare BiVO4 in different morphologies and structures, such as particle-like BiVO4, spindle body-like BiVO4, tube-like BiVO4, leaf-like BiVO4, line-like BiVO4, nanoblock BiVO4, ellipsoid body BiVO4, star-like BiVO4, etc. Good morphology can improve the catalyst’s activity as well as absorption capacity of light (Kho et al. 2011; Liu et al. 2010; Su et al. 2009; Sun et al. 2009; Wang et al. 2011; Wei et al. 2010; Xi and Ye 2010). A lot of researches have been conducted to overcome the disadvantages of BiVO4. Nevertheless, few reports have looked into morphology control via carbon doping (C-doping) of non-metallic doping. In this paper, a carbon-doped three-dimensional (3D) hierarchical nanosheet BiVO4 was synthesized using a simple hydrothermal method, in the hope of overcoming both disadvantages in a simple one-step way.

In this paper, by using PVP as structure-directing agents and l-cysteine as carbon sources, a flower-like 3D C-doped m-BiVO4 with strong capacity of visible-light absorption was synthesized by one-step hydrothermal method. Using rhodamine B (RhB) as the simulated organic pollutant, the photocatalytic properties of C-doped 3D hierarchical BiVO4 were investigated, finding that the C-doped 3D m-BiVO4 made a better performance in catalyzing and degrading the simulated pollutant under visible-light irradiation as compared with the pure BiVO4 and the two-dimensional (2D) C-doped BiVO4 nanosheet structure. Further, we speculated the possible mechanism that may lead to crystal growth and photocatalytic performance enhancement, which may be due to the mechanism that the l-cysteine promoted the 2D growth of BiVO4 and C-doping process, while PVP promoted 2D BiVO4 nanosheets into 3D self-assembly according to contrast test and density functional theory (DFT). Carbon doping can reduce the band gap of BiVO4 and thus improve the absorption of visible light, while 3D hierarchical structure can enhance the separation of electrons and holes. The C-coped flower-like BiVO4 with 3D structure synthesized by one-step hydrothermal synthesis method is expected to be further applied and modified in practice.

Experimental

Experimental materials

In this experiment, we used analytically pure chemicals such as bismuth nitrate (Bi(NO3)3·5H2O; Chengdu Area of the Industrial Development Zone Xinde Mulan, China), sodium hydroxide powder (NaOH; Chongqing Chuandong Chemical Company, China), polyvinylpyrrolidone K-30 (PVP) in analytical grade (Chengdu Kelong Chemical Co. Ltd. China), l-cysteine (Aladdin Industrial Corporation), ammonium metavanadate (NH4VO3; Chongqing Chuandong Chemical Company, China), nitric acid (HNO3; Chengdu Area of the Industrial Development Zone Xinde Mulan, China), rhodamine B (RhB; Tianjin Guangfu Fine Chemical Research Institute, China), and ethylene glycol (Chongqing Chuandong Chemical Company, China).

Catalyst preparation

Synthesis of pure BiVO4

To synthesize pure BiVO4, 2 mmol Bi (NO3)·5H2O was dissolved in 4 mL of 4 mol/L HNO3 and 50 mL of deionized water, and then subjected to constant stirring for 30 min before resulting in solution called solution A. On the other hand, 2 mmol NH4VO3 was dissolved in 4 mL of 2 mol/L NaOH and subjected to 30 min of constant stirring, resulting in solution B. After that, solution A and B were mixed together and then transferred into 100 mL of Teflon-lined autoclave to be sealed and heated at 180 °C for 16 h. Subsequently, the system was naturally cooled down to room temperature. The final product was collected from the mixture system by centrifuging treatment, washed with distilled water and ethanol six times, and dried under vacuum overnight at 60 °C for 12 h.

Synthesis of PVP-assisted BiVO4 (P-BiVO4)

To synthesize PVP-assisted BiVO4 (P-BiVO4), 2 mL of 0.5% PVP was added into solution A, while the remaining procedures were the same as those for synthesizing pure BiVO4.

Synthesis of C-doped BiVO4 (L-BiVO4)

To synthesize C-doped BiVO4 (L-BiVO4), 0.6 mmol l-cysteine was added into solution A, while the remaining procedures were the same as those for synthesizing pure BiVO4.

Synthesis of C-doped 3D hierarchical BiVO4 (P-L-BiVO4)

To synthesize C-doped 3D hierarchical BiVO4 (P-L-BiVO4), 2 mL of 0.5% PVP and 0.6 mmol l-cysteine were added into solution A, while the remaining procedures were the same as those for synthesizing pure BiVO4, resulting in a sample called P-L-BiVO4.

Characterization

The crystal structures of all prepared samples were characterized by x-ray diffraction (XRD) using a Rigaku D/Max2500pc diffractometer under Cu K radiation. The scanning angle of 2θ ranged from 10 to 70°, and the scanning rate was set as 4°/min. Scanning electron microscopy (SEM) images were obtained using a Tescan FEG-SEM microscope (high voltage = 10 kV; TESCAN, MARI3, Czech Republic). Transmission electron microscopy (TEM) was conducted using JEM-3010 (JEOL, Japan) at an acceleration voltage of 300 kV. The photoluminescence (PL) spectra of the photocatalysts were obtained using a Hitachi F-7000 spectrometer with an excitation wavelength of 280 nm. Energy dispersive spectrometer (EDS) was carried out during the SEM measurements. Raman spectra were obtained using an HR evolution instrument under an Ar+ laser source of 532 nm. The surface chemical environments were analyzed according to x-ray photoelectron spectra (XPS) on a PHI5000 Versa Probe system with monochromatic Al K x-rays. UV-Vis diffuse-reflectance spectroscopy (UV-Vis DRS) was performed using a Hitachi U-3010 UV-Vis spectrophotometer. Through mixing samples in a 50 mM DMPO solution tank (aqueous dispersion for DMPO-•OH and methanol dispersion for DMPO-•O2 −), we prepared the samples for electron spin resonance (ESR) measurement. In this experiment, the sample was under the irradiation of visible light. Photoelectrochemical properties of samples were evaluated using a CHI Electrochemical Workstation (CHI 760E; Shanghai Chenhua Co. Ltd., China). All experiments were performed at room temperature.

First-principles study

Band structure calculations were performed using the First-principles theory based on density functional theory with projector-augmented wave approach (Blöchl 1994). The energy cutoff was 450 eV. A Hubbard U value of 2.7 eV (Zhao et al. 2016) was applied to d orbitals of V for accurate description of the electronic properties of BiVO4. An 8810 Monkhorst-Pack mesh was sampled on the Brillouin zone. The exchange correlation functional was treated within generalized gradient approximation with Perdew–Burke–Ernzerh parameterization (Zhang et al. 2008b). Energy convergence criterion of 10−6 eV was ensured in all calculations. The treatment of the doped system of interest was based on the virtual crystal approximation (Bellaiche and Vanderbilt 2000). The calculated model of monoclinic BiVO4 was fixed at the experimental structure. In the doped system, we considered the substitutional C at O sites in BiVO4, i.e., BiV(O1 − x C x )4, x = 0.01 and 0.05.

Photocatalytic activity

The photocatalytic activities of the samples were assessed by evaluating the photodegradation degree of RhB solution under visible-light irradiation at room temperature. In the experiment, 0.20 g of catalyst was first added to 200 mL of 5 mg/L RhB aqueous solution in a 250-mL beaker and then subjected to magnetic stirring for 30 min in the dark, reaching good dispersion and adsorption–desorption equilibrium between dye and catalyst. After that, the experimental solution was placed 350 mm away from a 500-W Xe lamp with a >400 nm UV cutoff filter as the visible light irradiation source. For every 1 h of irradiation, the solution was collected and then subjected to centrifugation at 10,000 rpm to remove all catalysts. After that, the concentration of the remaining dye was spectrophotometrically monitored by measuring the absorbance of the solutions at 552 nm. For comparison, the photocatalytic experiments were carried out with pure BiVO4, P-BiVO4, L-BiVO4, or P-L-BiVO4 or no catalyst under the same conditions.

Results and discussion

Pattern and morphology analysis by XRD and SEM

Figure 1a shows the XRD spectra of the prepared samples, from which we can see all the crystal planes and related 2θ values are consistent with m-BiVO4 standard powder diffraction card #PDF14-0688. According to the magnified peaks of the (121) and (040) planes in Fig. 1b, we can find that the characteristic diffraction peaks of (121) of L-BiVO4 and P-L-BiVO4 have smaller angle shifts than pure BiVO4 and P-BiVO4 after the addition of l-cysteine, which indicates that the spacing between crystal planes has become greater. This is probably due to the carbon atoms with relatively larger atomic radius (0.077 nm) that replace oxygen atoms with relatively smaller atomic radius (0.074 nm) in the BiVO4 (Jagadale et al. 2008). XRD results indicate that carbon has been well inserted into the BiVO4 lattice, which avoids the formation of any segregated impurity phase. As the implementation of non-metallic ion doping changes the band-gap energy, the photocatalytic activity of the sample becomes higher (Li et al. 2013; Wang et al. 2014b; Zhao et al. 2013). Therefore, the as-prepared L-BiVO4 and P-L-BiVO4 samples are expected to show enhanced photocatalytic performances.

Fig. 1
figure 1

a XRD patterns. b Magnified peaks of (121) and (040) planes. c Simulated XRD patterns of BiVO4 and carbon-doped BiVO4

Full size image

To explore the reason for the occurrence of a relative peak shift of C-doped BiVO4, XRD patterns have been modeled by considering carbon doping and volume expansion. Considering the change of volume, we simulated three cases for carbon-doped BiVO4, i.e., C-doped BiVO4 (including doping but not including volume expansion), C-doped BiVO4@in-plane (including doping and volume expansion along in-plane direction), and C-doped BiVO4@out-of-plane (including doping and volume expansion along out-of-plane direction). The simulated XRD patterns are shown in Fig. 1c, from which it can be found that the XRD pattern of C-doped BiVO4 that only contains carbon dopants does not introduce any peak shift as compared with pure BiVO4. However, the volume expansion along in-plane direction (C-doped BiVO4@in-plane) shifts the peak (121) significantly. In the case of C-doped BiVO4@out-of-plane, both peak (121) and peak (040) are shifted. By comparing the simulated XRD patterns with the experimental XRD patterns in Fig. 1b, we can conclude that the only one obvious shift (i.e., peak 121) is caused by the increase of in-plane lattice constants of carbon-doped BiVO4.

Figure 2 shows the SEM of the samples. Under the condition where no structure-directing agent or carbon sources are added, the BiVO4 are particles with diameters ranging from about 40 to 100 nm, in a large degree of agglomeration (see Fig. 2a–c). However, under the condition where structure-directing agent PVP is added, most of the products grew into much more bigger particles with a particle size of about 1 μm (see Fig. 2d–f). Under the condition where only carbon source l-cysteine is added, morphologies of samples are greatly changed as shown in Fig. 2g–i. The aggregations of a massive amount of nanosheets are in thicknesses ranging within 17.63 ± 8.17 nm (see D1–D6 in Fig. 2i). Under the condition where both structure-directing agent PVP and carbon sources l-cysteine are added, the morphologies of the products are changed into spherical structure, with the ball’s diameter of about 2 μm. Many of these ball-shaped objects are 2D nanosheets, which are connected together, resulting in a multilayer structure like a flower (see Fig. 2j–l). The thicknesses of the composite sheets were varying within 19.86 ± 8.48 nm (194 statistical samples in Fig. 3). In conclusion, with the change of the reaction conditions, the crystal morphology of bismuth vanadate is changed accordingly. l-Cysteine contributes to its 2D growth and C-doping process, which is beneficial for electrons to migrate onto the surface and helps BiVO4 to improve the range of visible light (Zhang et al. 2006), while PVP is promoting the 3D hierarchical sheet-assembled flower-like structure. This kind of structure is reported to have reflecting and scattering effects, which is conducive to the increase of the absorption capacity of light (Xiong et al. 2014) and the enhancement of contact with contaminants (Huang et al. 2011).

Fig. 2
figure 2

SEM images of BiVO4 (a–c), P-BiVO4 (d–f), L-BiVO4 (g–i), and P-L-BiVO4 (j–l)

Full size image
Fig. 3
figure 3

a Particle size distribution of P-L-BiVO4 obtained by measuring the thickness of 194 particles shown in SEM (b) images

Full size image

Composition and chemical states analysis through Raman scattering spectra, EDS, TEM, and XPS

The peaks of Raman spectrum are sensitive to the influence of short-range ordered structure of the photocatalyst (Zhang et al. 2008a). Raman spectrum can reflect certain material information such as crystallinity, local structure, and electronic properties (Liu et al. 2003). As shown in Fig. 4, the peaks at 827, 712, and 324 cm−1 are ascribed to the typical vibrations of m-BiVO4. The peak on 324 cm−1 is due to the asymmetric stretching vibration and bending vibration of VO4 3−. The Raman peaks at 712 and 827 cm−1 can be attributed to two V–O stretching vibration modes in different orders. In addition, we can observe that the characteristic peak at 712 cm−1 is weak, which may be because the samples were prepared by the hydrothermal method (Galembeck and Alves 2000). On the other hand, Raman features might also be affected by sample shape. The absence of band at 712 cm−1 (added l-cysteine) in the Raman spectra of BiVO4 thin films might be due to its low thickness (Yin et al. 2013). Owing to the addition of l-cysteine¸ the peak at 827 cm−1 is allowed to have a certain blue shift, which means the V–O band becomes longer. The blue shift is caused by the C-doping effect, i.e., carbon is inserted into the O sites of the BiVO4 lattice to form the V–C bond. The V–C bond has a longer length than the V–O bond (Huang et al. 2015).

Fig. 4
figure 4

Raman spectra of the samples

Full size image

The test results of HRTEM and EDS are shown in Fig. 5. We can see that C element has been successfully doped in BiVO4 and no second-phase crystal face occurs, which is consistent with XRD test results. However, the interplanar spacing of C-doped BiVO4 is larger than that of PDF standard card, which indicates that the lattice constant of BiVO4 has been changed due to C doping. This is mainly because as the doping C replaces the position of O, the charge imbalance occurs, thus leading to lattice distortion (Cui et al. 2010). Figure 6 shows the EDS line scan of P-L-BiVO4, from which we can see that the contents of C, O, Bi, and V are higher in the middle while lower at both ends due to the 3D spherical structure of the sample.

Fig. 5
figure 5

HRTEM (a–e) and EDS (f) of P-L-BiVO4

Full size image
Fig. 6
figure 6

EDS line scan of P-L-BiVO4

Full size image

Moreover, XPS analysis was conducted to further reveal the chemical states of samples. Figure 7a shows the fully scanned spectra in the range of 0–700 eV, and the survey spectrum shows that the composite is composed of elements Bi, O, V, and C. According to Fig. 7b, the peaks at binding energies of 523.9 (V 2p1/2) and 516.4 eV (V 2p3/2) corresponded to the split signal of V 2p while the V 2p peak is assigned to V5+ (Li et al. 2015). As can be seen in Fig. 7c, the samples show XPS signals of O1s at 529.7 and 532.7 eV, respectively, which is probably due to the O2 − anions in the BiVO4 crystallites. Figure 7d shows that the binding energies for Bi 4f7/2 and Bi 4f5/2 are 158.8 and 164.1 eV, respectively, which is significantly related to the Bi3+ peak in the monoclinic BiVO4 (Kudo et al. 1999). The peak shifts of elements such as Bi, V, and O are probably due to element C that is inserted into the O sites of the BiVO4 lattice, leading to the formation of C-doped BiVO4 in the l-cysteine, which is consistent with the XRD, EDS, and Raman results.

Fig. 7
figure 7

X-ray photoelectron spectra of the samples: a survey XPS spectrum, b V 2p3/2 and V 2p1/2 peaks, c O1s peak, and d Bi4f5/2 and Bi4f7/2 peaks

Full size image

Measurement of optical properties by UV-Vis DRS, DFT, ESR and transient photocurrent response, PL spectra, and IPCE

The UV-Vis diffuse reflectance spectra of samples are shown in Fig. 8a, from which we can see that L-BiVO4 and P-L-BiVO4 not only have strong absorption capacities in the ultraviolet region but also have one in the visible region. Their UV absorption capacities can be attributed to the band transition from O 2p to V 3d, while the process of visible-light absorption can be regarded as the transition from valence band (generated by hybrid orbitals of Bi3+ 6 s2 and O 2p) to V 3d conduction band (Xie et al. 2006). A step-shaped absorption spectrum indicates that the migration of electronics from band gap is not an impure state migration (Xiong et al. 2014). According to the UV-visible diffuse reflectance spectra, the photo energies of samples are respectively estimated as 2.4, 2.4, 2.39, and 2.38 eV according to the intercept of the tangents to plot depicting (Ahυ)2 versus hυ (see the inset of Fig. 8a). Therefore, the semiconductor is expected to be used as a photocatalyst under visible-light irradiation for wastewater treatment. Moreover, P-BiVO4 and P-L-BiVO4 have a much smaller band gap compared with pure BiVO4, P-BiVO4, which is mainly due to that C-doping decreases the energy of the band gap. The reason why P-L-BiVO4 has stronger absorption of visible light than L-BiVO4 might be due to that more C atoms enter in P-L-BiVO4 (see the DFT calculation) and the 3D hierarchical structure of P-L-BiVO4 has reflecting and scattering effects in some degree, thus increasing the response range and absorption capacity of light (Payne et al. 2011).

Fig. 8
figure 8

a Estimated band gaps of UV-Vis spectra of the samples. b Band structures of BiVO4, BiV(O0.99C0.01)4, and BiV(O0.95C0.05)4

Full size image

As is known to all, the DFT calculation can calculate the structure of the band gap. In this work, we conducted the DFT calculation to prove that the band gap was changed due to the C-doping. Figure 8b shows the band structures of BiVO4, BiV(O0.99C0.01)4, and BiV(O0.95C0.05)4. The band structure of monoclinic BiVO4 without doping is an indirect band-gap semiconductor with a wide band gap of 2.43 eV, which agrees well with the experimental band gap of 2.4 eV of pure BiVO4 (Zhang et al. 2010). As for the doped systems, the band gaps of BiV(O0.99C0.01)4 and BiV(O0.95C0.05)4 are estimated to be 2.37 and 2.18 eV, respectively. This suggests the substitutional C at the O sites can narrow the band gap, which is consistent with the results of our experiments.

In this research, electron spin resonance spectroscopy was adopted to detect the active radicals produced during the catalytic reaction. 5,5-Dimethyl-1-pyrroline-N-oxide (DMPO) was used to capture the •OH (DMPO-•OH) and •O2 − (DMPO-•O2 −) generated in the catalytic process. According to Fig. 9, it can be observed that the •O2 − and •OH− signal peaks appear after 10 min of visible-light irradiation. However, there are no signal peaks of the •O2 − and •OH under dark condition. The experiment results show that the •OH and •O2 − are generated under light irradiation (Xiong et al. 2014). The signals of P-L-BiVO4 appearing after 10 min of visible-light irradiation is the strongest, while that of BiVO4 is significantly weaker. This indicates that the production of active species is further enhanced by the photogenerated carriers produced due to C-doping process and good morphology.

Fig. 9
figure 9

DMPO spin-trapping ESR spectra under visible light: a DMPO-∙OH for BiVO4, b DMPO-∙OH for P-L-BiVO4, c DMPO-∙O2 − for BiVO4, d DMPO-∙O2 − for P-L-BiVO4

Full size image

Photoelectrochemical test is usually conducted to study the excitation, separation, and transfer of carriers in a catalyst. Figure 10a shows the transient photocurrent responses of the samples under visible-light irradiation. The photocurrent densities of the P-L-BiVO4 and L-BiVO4 are much higher than those of BiVO4 and P-BiVO4, which might be attributed to C-doping process that leads to the increased usage efficiency of visible light. The photocurrent density of P-L-BiVO4 is higher than that of L-BiVO4, which is mainly due to that more C atoms enter in P-L-BiVO4 (see the DFT calculation) and the 3D hierarchical structure of P-L-BiVO4 has reflecting and scattering effects in some degree, thus leading to the increase of response range and absorption capacity of light (Hu et al. 2015). Therefore, P-L-BiVO4 is expected to show excellent photocatalytic activity under visible-light irradiation.

Fig. 10
figure 10

a Transient photocurrent response. b PL spectra at room temperature. c IPCE

Full size image

PL spectrum is determined by the migration, transfer, and separation efficiency of the photogenerated charge carriers in semiconducting materials. PL study could effectively testify the improvement of the migration efficiency of photogenerated electrons and suppressing the recombination of electron–hole pairs. Figure 10b shows the comparisons of PL spectra of BiVO4, P-BiVO4, L-BiVO4, and P-L-BiVO4 under the excitation wavelength of 380 nm, from which we can see that the PL peak intensity of P-L-BiVO4 shows a significant decreasing trend. These results show that P-L-BiVO4 is conducive to the effective separation of electron–hole pair.

We have conducted an incident photon-to-current conversion efficiency (IPCE) test. The electrodes were prepared according to the method proposed by Iwase and Kudo (2010). Photoelectrochemical properties were evaluated with a three-phase electrode consisting of the working electrode (prepared electrode), counter electrode (Pt electrode), and reference electrode (saturated Ag/AgCl electrode), respectively. The working electrode was irradiated from the FTO side with visible-light through a cutoff filter. The IPCE was calculated as follows:

$$ \mathrm{IPCE}/\%=\left(1240\times {I}_{\mathrm{sc}}\right)/\left(\lambda \times {P}_{\mathrm{in}}\right) $$

where λ (nm) is the incident photon wavelength, I sc (μA/cm2) is the photocurrent of the device, and P in (W/m2) is the incident power. As shown in Fig. 10c, the monochromatic light can be obtained by grating control at 1.5 V bias voltage under Ag/AgCl condition. The IPCE of P-L-BiVO4 electrode can reach 4.38% at 1.5 V versus under Ag/AgCl condition, while that of BiVO4 is only 0.78% (at 410 nm). The enhancement of photoconversion efficiency indicates that P-L-BiVO4 may have the best photocatalytic activity.

Evaluation of photocatalytic activity through degradation of RhB

Figure 11 shows the photocatalytic activities of the samples under visible-light irradiation for 10 h. For comparison, the self-degradation of RhB was also estimated without any catalyst. Results showed that only about 0.45% of RhB was decomposed after 10 h of irradiation without any photocatalyst, while 73.56, 64.75, 44.38, and 38.26% were degraded with P-L-BiVO4, L-BiVO4, P-BiVO4, and BiVO4, respectively. P-L-BiVO4 achieved the highest catalytic activity, successively followed by L-BiVO4, P-BiVO4, and pure BiVO4. This is mainly due to that P-L-BiVO4 and L-BiVO4 have relatively larger BET surface area and pore volume (Table 1), which is conducive to enhancing the contact between samples and organic contaminants, so that the photocatalytic performance of samples can be improved after C-doping. On the other hand, the degradation rate of RhB is largely determined by the visible-light adsorption capacity. The band gap of BiVO4 was shortened by C-doping; therefore, BiVO4 enjoyed a stronger and broader absorption in the visible region. All these reasons explain why BiVO4 had an increased absorption of visible light and why the degradation rate of RhB was increased.

Fig. 11
figure 11

a Degradation of RhB using different catalysts under visible-light irradiation. b Photocatalytic reaction with linear fitting modes and the reaction rate constant k

Full size image
Table 1 Characteristics obtained from nitrogen desorption isotherms
Full size table

For comparison, the photodegradation of RhB was performed using photocatalyst Degussa P25 under visible light. The results are shown in Fig. 12. Blank test showed that no RhB was degraded under visible light in the absence of photocatalyst. The as-prepared P-L-BiVO4 exhibited better photocatalytic activity than Degussa P25 in the degradation of RhB under the same experimental conditions, which could degrade 83.24 and 30.05% of RhB in 10 h, respectively.

Fig. 12
figure 12

Photocatalytic degradation of RhB by the as-prepared P-L-BiVO4 catalyst and Degussa P25 under visible light

Full size image

Postulated formation and degradation mechanism of photocatalysts

In the hydrothermal synthesis process of solid material, the morphology of the product is affected by various factors such as structure-directing agent, alkaline source, nature of the metal precursor, the precursor solution pH, temperature, and hydrothermal time (Grassmann and Löbmann 2003; Wang et al. 2006; Xu et al. 2006; Yu and Golfen 2004). Directional coalescence mechanism can successfully account for the formation of specific crystal morphology (And and Yu 2007; Gong et al. 2006). In this mechanism, a highly ordered superstructure can be formed by the directional self-assembly of primary particles. In this paper, the BiVO4 particles of various morphologies were formed by following the mechanism shown in Fig. 13. The surface energy of the nanocrystals of BiVO4 could be reduced by adding l-cysteine, which was then absorbed on the surface of the nanoparticles of newly generated BiVO4. l-Cysteine molecules were selectively adsorbed on the surface of the BiVO4 particle (Xie et al. 2013). Due to the adsorption of l-cysteine, the growth rate of the crystal surface was reduced while the growths of other planes were promoted. Along with the formation of two-dimensional nanosheet of BiVO4, the carbon atoms were inserted into the BiVO4 crystal lattice. Under the effect of PVP, the generated two-dimensional particles were selectively assembled into a 3D flower-like C-doped BiVO4.

Fig. 13
figure 13

Possible growth mechanism and degradation mechanism of the P-L-BiVO4

Full size image

According to the mechanism of degradation in Fig. 13, upon visible-light excitation, the P-L-BiVO4 surface generates electron–hole pairs. Then, the photogenerated electrons react with dissolved oxygen molecules (O2) to yield superoxide radical anions (·O2 −). The holes can react with OH− to form ·OH, and through a series of reactions with H2O, the activated ·O2 − further forms ·OH and O2. The holes, ·O2 −, and ·OH are strong oxidizing agents for the decomposition of organic dyes (Du et al. 2016). The entire sequence is summarized as follows:

$$ \begin{array}{l}\mathrm{P}\hbox{-} \mathrm{L}\hbox{-} {\mathrm{BiVO}}_4+\mathrm{h}\upnu \left(\mathrm{visible}\ \mathrm{light}\right)\to {\mathrm{h}}^{+}+{\mathrm{e}}^{-}\kern0.9em \mathrm{R}1\hfill \\ {}\begin{array}{cc}\hfill {\mathrm{O}}_2+{\mathrm{e}}^{-}\to \bullet {{\mathrm{O}}_2}^{-}\hfill & \hfill \mathrm{R}2\hfill \end{array}\hfill \\ {}\begin{array}{cc}\hfill \bullet {{\mathrm{O}}_2}^{-}+{\mathrm{H}}_2\mathrm{O}\to \bullet \mathrm{OH}+{\mathrm{O}\mathrm{H}}^{-}+{\mathrm{O}}_2\hfill & \hfill \mathrm{R}3\hfill \end{array}\hfill \\ {}\begin{array}{cc}\hfill {\mathrm{h}}^{+}+{\mathrm{O}\mathrm{H}}^{-}\to \bullet \mathrm{OH}\hfill & \hfill \mathrm{R}4\hfill \end{array}\hfill \\ {}\begin{array}{cc}\hfill {\mathrm{h}}^{+}\mathrm{or}\bullet {{\mathrm{O}}_2}^{-}\;\mathrm{or}\bullet \mathrm{OH}+\mathrm{RhB}\to \mathrm{degraded}\ \mathrm{products}\hfill & \hfill \mathrm{R}5\hfill \end{array}\hfill \end{array} $$

Stability evaluation through recycling test

The stability of P-L-BiVO4 is a major factor affecting its application effect. The stability of P-L-BiVO4 was investigated through cyclic degradation of RhB under visible-light irradiation, as shown in Fig. 14a. The results show that the degradation rate of RhB tends to be nearly constant over five cycles. In the repeated tests, the degradation rate after each cycle is maintained at 73.24, 73.05, 72.45, 71.58, and 70.41%, respectively. The P-L-BiVO4 sample was easily recycled by simple filtration. In addition, the phase structures of the sample remain unchanged, which verifies that the components of the P-L-BiVO4 are difficult to be photodecomposed and its structure is stable during the photocatalytic process. This feature is very important for its practical application and modification.

Fig. 14
figure 14

a Cycling runs of photocatalytic degradation of RhB using photocatalyst P-L-BiVO4; b SEM image and XRD patterns of P-L-BiVO4 after five rounds of cycling

Full size image

Conclusions

In this paper, carbon-doped nanosheets m-BiVO4 with 3D hierarchical structure and high visible-light photocatalytic property was synthesized by one-step hydrothermal method. The as-synthesized samples were characterized using various analytical methods, and their morphologies and compositions were thoroughly investigated as well. l-Cysteine was used as the carbon source for C-doping and the simultaneous control of the crystal morphology of BiVO4. The photocatalytic activities of resulted samples were estimated by testing the photodegradation degree of RhB under visible-light irradiation. The results showed that the carbon-doped nanosheets BiVO4 in 3D hierarchical structure had a significantly higher photocatalytic activity as compared to the BiVO4 in sheet or amorphous form. In addition, C-doped 3D hierarchical nanosheets BiVO4 with high visible-light photocatalytic property can easily be recycled without decreasing the photocatalytic activity. This new material is expected to be applied in future applications.