Fabrication of sandwich structure membrane and its performance for photocatalytic reduction of Cr(VI)

The regenerated cellulose membrane (RC) was synthesized by dissolving cotton cellulose in NaOH/CO(NH2)2 system. The Polydopamine/Bismuth tungstate/RC composite membrane (RCPB) with photocatalytic activity was synthesized by loading polydopamine-modified bismuth tungstate (PDA/BWO) composite onto the RC by blending method. The RCPB/PAN/RCPB sandwich structure membrane was synthesized by combining the polyacrylonitrile (PAN) nanofiber membrane and RCPB by scraping method, which can reduce aqueous Cr(VI) under visible light effectively. Characterization shows that the tensile strength, elongation at break, roughness and initial water contact angle of RCPB/PAN/RCPB are 32.1 MPa, 5.34%, 0.658 μm and 69.0°, respectively. The photoreduction percent of Cr(VI) by RCPB/PAN/RCPB can reach 99.7% within 120 min with a rate constant of 0.0869 min−1, and the photoreduction percent remains above 84.6% after four cycles. The introduction of PAN nanofiber membrane further improves the mechanical properties and recycling ability of RCPB. Meanwhile, the capture experiments reveal that the main active substance for photocatalytic reduction of Cr(VI) by RCPB/PAN/RCPB is photogenerated e−. This work provides a new idea for the treatment of Cr(VI)-containing wastewater.


Introduction
In recent decades, due to the massive emission of untreated wastewater, the health of all living organisms on Earth has been seriously threatened, and the shortage of drinking water has also emerged (Hu et al. 2020). In order to achieve sustainable development, it is urgent to solve the environmental pollution problem Yu et al. 2020). Recently, photocatalysis, as a green technology, plays a significant role in the field of wastewater treatment (Maity et al. 2019;Dong et al. 2019).
Cr(VI) is a common heavy metal ion in textile industrial wastewater, which possesses a great threat to water safety and human health. In recent years, bismuth tungstate has attracted wide attention due 1 3 Vol:. (1234567890) to its high stability and visible light response characteristics. However, the high recombination rate of electron-hole (e − −h + ) pairs severely restricts the photocatalytic activity of bismuth tungstate. In our previous study, Bi self-doped bismuth tungstate (Bi 2.15 WO 6 , denoted as BWO) was successfully synthesized by hydrothermal method (Ren et al. 2022). Composite PDA/BWO with core-shell structure was also synthesized by introducing dopamine (DA) with abundant functional groups, which has good photocatalytic activity. However, most of the powder photocatalysts are used only once, which not only increases the cost, but also causes secondary pollution. At the end of the aqueous photocatalytic reaction, the separation of photocatalyst must be accomplished by gravity or mechanical centrifugation, which is time-consuming and laborious (Yu et al. 2016). Therefore, from an economic point of view, it is a promising method to fix the powder photocatalyst on the substrates with large specific surface area (such as fibers, glass or membrane materials). Large specific surface area can provide more active sites in the reaction process, and the presence of carriers can improve the recycling ability of powder catalysts (Gude et al. 2017). Among them, cellulose has good biocompatibility, low cost, high stability and high recovery, which can be used as the carrier of photocatalyst (Orooji et al. 2021).
In this study, in order to solve the problem of poor reusability and easy agglomeration of PDA/ BWO powder catalysts, RC will be synthesized by dissolving cotton cellulose in NaOH/CO(NH 2 ) 2 system. Composite RCPB with photocatalytic activity will be synthesized by loading PDA/BWO onto the RC through blending. RCPB/PAN/RCPB sandwich structure membrane will be formed by scraping RCPB on PAN nanofiber membrane. The effects of pH, membrane thickness and dosage on the photocatalytic reduction of Cr(VI) by RCPB will be investigated. The main reactive substances for reducing Cr(VI) by RCPB/PAN/RCPB will be investigated by capture experiments. Furthermore, PAN nanofiber membrane will be introduced to improve the mechanical properties and recycling of RCPB. The synthesized samples will be thoroughly characterized with respect to composition, morphology, functional groups, distribution and content of elements, thermal stability, roughness and mechanical properties.

Synthesis of composite RCPB
In this study, PDA/BWO photocatalyst was synthesized by stirring method reported in the paper of Ren (Ren et al. 2022). Crush the cotton pulp board with a shredder and put it in an oven for 8 h at 60 °C to remove moisture. Cellulose was then dissolved in NaOH/CO(NH 2 ) 2 system, and transparent cellulose solution with mass fraction of 1 wt% was obtained.
Subsequently, 0.5 g PDA/BWO photocatalyst was mixed with 50 g cellulose solution by simple blending. After magnetic stirring for 1 h, 1 wt% PDA/ BWO/cellulose suspension was obtained. The cellulose suspension was evenly coated on the surface of the glass plate via an automatic coating machine, and then the glass plate was immersed in a gel bath (5% H 2 SO 4 ) for 5 min. The composite membrane was then transferred to deionized water for soaking to remove residual alkali and urea. The deionized water was replaced several times within 24 h, and the obtained composite membrane finally dried naturally. According to the thickness of the composite membrane (0.2 and 0.3 mm), they were named as RCPB-0.2 and RCPB-0.3, respectively. PDA/BWO powder was replaced by BWO powder, and the obtained composite membrane was named as RCB. Without adding photocatalyst, the obtained regenerated cellulose membrane was named as RC.
The synthesis of RCPB was repeated three times to ensure the accuracy of the experimental data.
Synthesis of RCPB/PAN/RCPB PAN nanofiber membrane was synthesized by the electrospinning method. The specific steps were as follows: 2.2 g PAN powder was dissolved in 20 mL DMF and stirred at room temperature for 12 h to obtain uniform solution. The electrospinning machine was set to 20.0 kV supply voltage, 15 cm spinning distance and 0.8 mL h −1 injection pump flow rate. Then, the obtained nanofiber membrane was dried in an oven at 70 °C.
The sandwich structure of RCPB/PAN/RCPB was synthesized by scraping a layer of RCPB-0.2 on the glass plate, and then the PAN nanofiber membrane was placed on the top of RCPB-0.2, followed by scraping a layer of RCPB-0.2 on the PAN nanofiber membrane.
The synthesis of RCPB/PAN/RCPB was repeated three times to ensure the accuracy of the experimental data.
Characterization of photocatalysts X-ray diffractometer (XRD, Rigaku D/max 2200/ PC) with CuKα radiation (λ = 1.5418 Å) was used to measure the composition of the synthesized samples. The surface element composition of the samples was determined by X-ray photoelectron spectroscopy (XPS, VG Multilab 2000). Fourier transform infrared (FTIR) spectroscopy was used to identify the functional groups of the samples (NICOLET iS10, Thermo Fisher Scientific, USA). Raman spectroscopy (Renishaw InVia Reflex) was studied at 532 nm excitation wavelength. UV-vis diffuse reflectance spectra (UV-vis DRS, Shimadzu UV-2700) of all samples were measured in the range of 200-800 nm with BaSO 4 as a standard reference. The photoluminescence (PL) spectra were obtained by fluorescence spectrophotometer (Hitachi F-2500) with an excitation wavelength of 490 nm. Scanning electron microscope (SEM, Hitachi Regulus 8100) was used to observe the morphology of the material, and the working voltage was 10 kV. Energy dispersive spectroscopy (EDS, Octane Super) was used to observe the distribution and content of elements on the surface of the sample. The thermal stability of the samples was investigated by thermogravimetric analyzer (TG, Rigaku). The samples were heated from 40 to 800 °C at a heating rate of 10 °C min −1 in nitrogen atmosphere (flow rate of 60 mL min −1 ). The roughness of the sample surface was measured by a true color confocal microscope (Zeiss CSM700). The mechanical properties were tested by electronic single fiber strength meter (YM-06 A). The tensile speed was 2 mm min −1 and the spacing was 20 mm. The membrane was made into a rectangle of 50 × 5 mm. Unit X(cN) of test values were converted to Y(MPa) (Zhao et al. 2019) by Eq. (1).
where A is the width of the membrane, and B is the thickness of the membrane. The thickness of the membrane is measured by a thickness gauge.
All characterizations of photocatalysts were repeated two times to ensure the accuracy of the results.
Photocatalytic performance test Cr(VI) was used to evaluate the photocatalytic activity of the catalyst. The photocatalytic reactor is mainly composed of light source (565 Mw cm −2 ), reactor, circulating cooling water system and magnetic stirrer. In each experiment, 0.1 M HCl or NaOH aqueous solution was used to adjust the initial pH of the reaction system. During the reaction, 0.7 mL suspension was sampled within a scheduled time. After centrifugation, the absorbance of Cr(VI) was determined by a UV-vis spectrophotometer at 540 nm.
The photocatalytic reduction of Cr(VI) was expressed via a pseudo-first-order kinetic equation, such as Eq. (2): where t is the reaction time (min), C 0 is the initial concentration (mg L −1 ) of Cr(VI) solution, C t is the concentration of Cr(VI) solution at a certain time; while k is the quasi-first-order rate constant (min −1 ). The effects of hole (h + ) and electron (e − ) on the photocatalytic reduction of Cr(VI) by RCPB were studied by adding EDTA and KBrO 3 to suspension, respectively.
All photocatalytic performance tests were repeated three times to ensure the accuracy of the experimental data. The obtained rate constants in the reduction of Cr(VI) under different experimental conditions are shown in Table 1.

Photoelectrochemical experiments
Photoelectrochemical measurements were conducted using a CHI660E electrochemical workstation. Specifically, the synthesized samples were dispersed in ethanol and sonicated for 30 min to form a 20 g L −1 mixture. The mixture dripped onto a clean FTO glass was thoroughly dried in a vacuum oven to obtain a working electrode. Pt as the counter electrode, Ag/ AgCl as the reference electrode and sample as the working electrode were tested in 0.5 M Na 2 SO 4 electrolyte solution. The transient photocurrent was measured by using a 565 mW cm −2 visible light source.
Photoelectrochemical experiments were repeated three times to ensure the accuracy of the experimental data.

Characterization analysis
The XRD patterns of RC, RCB, RCPB and RCPB/ PAN/RCPB membranes are shown in Fig. 1. The RC membrane gives rise to two typical diffraction peaks    . The strong intensity of these diffraction peaks and the absence of impurity peaks indicate that BWO and PDA/BWO were successfully loaded on RC. The (100) crystal plane of RCPB/PAN/RCPB appears at 2θ = 17°, which is due to the existence of PAN (Ren et al. 2019), indicating that RCPB and PAN nanofiber membrane were successfully composited. The morphologies of RC, RCPB and RCPB/PAN/ RCPB were observed by SEM. As shown in Fig. S1a, the RC membrane has a porous structure (Fan et al. 2019). This is due to the rapid phase separation and subsequent solidification caused by solvent evaporation during cellulose regeneration. After the introduction of PDA/BWO, these nanoparticles aggregate into irregular shapes on the surface of RCPB membrane (Fig.  S1b), making the membrane surface much rougher. In addition, RCPB also has abundant pore structure (Fig. S1a). As shown in Fig. 2a, porous structure and nanoparticles are observed on the surface of RCPB/PAN/ RCPB (Fan et al. 2019). However, PAN nanofiber membrane are not observed, because the middle PAN nanofiber membrane was covered by RCPB. Therefore, we also observed the cross section of RCPB/PAN/ RCPB (Fig. 2b), and can clearly see PAN nanofiber membrane wrapped by RCPB, demonstrating the successful synthesis of RCPB/PAN/RCPB composite. In order to confirm the distribution of PDA/BWO on the RC, EDS analysis was carried out. The surface element content of RCPB is shown in Fig. S1d. EDS mapping images (Fig. S1e) further show that there are C, N, O, Bi and W elements in the RCPB, and they are uniformly distributed on RCPB. The above results confirm that the RCPB was successfully synthesized in this study. On the other hand, it can be seen from Fig. 2c, d that C, N, O, Bi and W elements are uniformly distributed on RCPB/PAN/RCPB.
As shown in Fig. 3, the FTIR spectra of RC, RCB, RCPB and RCPB/PAN/RCPB indicate that the introduction of nanoparticles doesn't change the structure of RC. Previous studies showed that the CH 2 infrared absorption peak at 1429 cm −1 is cellulose I, and that at 1420 cm −1 is cellulose II (Yoon et al. 2022;Nelson et al. 1964). It can be seen from Fig. 3a that the CH 2 absorption peak of the RC appears at 1420 cm −1 , indicating that the NaOH/CO(NH 2 ) 2 system destroys the intermolecular hydrogen bond of cotton short fiber and makes the cellulose transformation from type I to type II completely. It was reported that the C-O-C stretching vibration absorption peak of β-(1,4) glycosidic bond appears at 897 cm −1 in cellulose type I, which is also the "amorphous band" of cellulose (Fan et al. 2019). However, it can be seen from Fig. 3a that the stretching vibration peak of C-O-C appears at 894 cm −1 , and the intensity of the absorption peak increases significantly relative to the intensity of CH 2 , further confirms the transformation of cellulose from type I to type II. The absorption peaks at 1015 cm −1 and 1648 cm −1 correspond to the stretching vibration of C-O-C and the O-H bending vibration of adsorbed water (Yuan et al. 2021). As shown in Fig. 3b, the absorption peak at 2892 cm −1 is due to the stretching vibration of C-H bond. The absorption peak at 3320 cm −1 can be attributed to the stretching vibration of O-H . Compared with RCB, RCPB and RC, this peak intensity of RCPB/PAN/RCPB is the highest, indicating that the surface O-H content of RCPB/PAN/RCPB is the highest. Since the surface of RCPB contains PDA, the O-H content is also slightly higher than that of RCB. The characteristic peak of nitrile group (-C≡N) in PAN nanofiber membrane appears at 2242 cm −1 , and the peak at 1731 cm −1 is attributed to the C=O bond vibration, resulting from the hydrolysis of PAN nanofiber membrane or the residue of DMF Wang et al. 2013). The stretching vibration absorption peak of C-O-C at 1057 cm −1 shifts obviously (Yuan et al. 2021), which also indicates that RCPB was successfully combined with PAN nanofiber membrane.
The XPS analysis of RC, RCB and RCPB is shown in Fig. 4a. It can be seen that the curve of RCPB includes peaks of C 1 s, O 1 s, Bi 4f, W 4f and N 1 s. The high resolution scanning spectra of C1s are shown in Fig. 4b. The RC membrane gives rise to three peaks at 284.80, 286.51 and 287.89 eV, corresponding to C-C, C-O and C=O (You et al. 2014), respectively. In RCB and RCPB membranes, the binding energy of C-O shifts to a lower direction, which are 286.46 eV and 286.47 eV, respectively. Similar phenomena are also observed in C=O, and the binding energies are 287.78 eV and 287.73 eV (Fu et al. 2015), respectively. The high-resolution XPS spectra of O 1 s are shown in Fig. 4c. The peak at 532.79 eV belongs to the C-O-C bond of RC membrane. With the introduction of nanoparticles, the C-O-C of RCB and RCPB membranes moves to a higher binding energy. In addition, a new O peak is found in RCB and RCPB membranes, corresponding to the surface lattice oxygen (532.33 eV and 532.55 eV) (Liu et al. 2021).

3
Vol.: (0123456789) Fig. 4 a Wide-scan XPS spectra of RC, RCB and RCPB; high-resolution XPS C 1 s b and O 1 s c spectra of RC, RCB and RCPB; high-resolution XPS Bi 4f d and W 4f e spectra of RCB and RCPB; high-resolution XPS N 1 s f spectra of RCPB Figure 4d is the high resolution spectrum of Bi 4f in RCB and RCPB membranes. The peaks at 159.32 and 159.05 eV correspond to Bi 4f 7/2 , and the peaks at 164.63 and 164.26 eV correspond to Bi 4f 5/2 . Compared with RCB, the binding energy of Bi 4f (4f 7/2 and 4f 5/2 ) in RCPB moves to a lower binding energy (about 0.3 eV). The spin orbital splitting of Bi 4f in RCB and RCPB membranes (between Bi 4f 7/2 and 4f 5/2 ) is about 5.3 eV, which is consistent with BWO nanoparticles, indicating that the oxidation state of Bi in RCB and RCPB membranes is + III (Liu et al. 2021). The high resolution XPS spectra of W 4f are shown in Fig. 4e. The peaks at 35.52 and 35.26 eV correspond to W 4f 7/2 , 37.83 and 37.38 eV correspond to W 4f 5/2 , which confirms that the oxidation state of tungsten in RCB and RCPB is + VI (Wan et al. 2019). Compared with BWO and PDA/ BWO (Fig. 4c), it also moves to lower binding energy. The N 1 s spectrum in Fig. 4f is fitted to two peaks at 399.77 eV and 401.50 eV, which belongs to − NH 2 and − NH − groups in PDA/BWO (Zhao et al. 2020). The change of binding energy in XPS spectra is attributed to the strong interaction between nanomaterials (Zhang et al. 2010). Therefore, the shifts of C 1 and O 1 s proved that there is a strong interaction between RC and PDA/BWO nanoparticles in RCPB.
Thermogravimetry was used to analyze the thermal stability of materials. Figure 5a, b show the TG and DTG curves of RC, RCB, RCPB and RCPB/PAN/RCPB membranes under N 2 atmosphere, respectively. The first stage of thermal weight loss occurs at about 100 °C, and the weight loss ratio is about 5%, which is caused by the high temperature volatilization of water contained in the sample (Fu et al. 2015;Jia et al. 2012). Maintaining constant weight at 120-245 °C is the induction period of cellulose depolymerization (Cai et al. 2020). The cellulose in RC, RCB, RCPB and RCPB/PAN/RCPB membranes decomposes with the increase of temperature (Fu et al. 2015), and the initial temperatures are 285.3, 308.4, 304.6 and 253.4 °C, respectively. The peak temperatures of weight loss ratio are 311.4, 324.4, 318.0 and 320.4 °C, respectively. With the increase of temperature, the final char residue mass is 12.31, 33.19, 31.84 and 57.4%, respectively. It can be seen from the initial temperature and residual mass of thermal decomposition that the thermal stability of RCB and RCPB membranes is significantly improved compared with that of RC due to the addition of nanoparticles. Compared with RCB, the thermal stability of RCPB decreases slightly, which may be due to the presence of PDA on the surface. The residual mass of RCPB/PAN/RCPB is 1.8 times higher than that of RCPB, indicating that the thermal stability of RCPB/PAN/RCPB nanofiber membrane is significantly improved again after the composite of RCPB and PAN nanofiber membrane.

Performance of RCPB/PAN/RCPB composite membrane
The 3D profile of RC, RCB, RCPB and RCPB/ PAN/RCPB membranes is shown in Fig. 6. The roughness of RC is 0.227 μm. With the addition of BWO and PDA/BWO nanoparticles, the roughness of membrane increases gradually (Chen et al. 2021), and the roughness of RCPB is 2.4 times higher than that of RC. After the introduction of PAN nanofiber membrane, the roughness of RCPB/ PAN/RCPB composite increases to 0.658 μm, which is slightly higher than that of RCPB, and the increase of roughness is also conducive to improve the hydrophilicity of membrane. It can be also seen from the 3D profile that the dispersion of BWO and PDA/BWO nanoparticles in the RC matrix is relatively uniform, which is consistent with the SEM analysis.
The surface wettability of membrane was investigated by the water contact angle analysis. The excellent wettability can effectively accelerate the interaction of the catalyst and the reactant, so as to exhibit better performance (Qiu et al. 2022). Figure 7 shows the water contact angle images of RC, RCB, RCPB and RCPB/PAN/RCPB membranes. Figure 7a shows that RC has good hydrophilicity with the water contact angle of 31.0°. This is because the surface contains a large number of hydroxyl groups, which makes RC have high water adsorption and transport capacity, and is conducive to improve the photocatalytic activity (Qiu et al. 2022). With the addition of BWO and PDA/BWO nanoparticles, RCB and RCPB also has good hydrophilicity, and the initial water contact angle gradually decreases. The initial water contact angle of RCPB is only 22.0° (Fig. 7c), which is due to the large amount of hydrophilic hydroxyl and amino groups on the PDA surface. Figure 7d shows the water contact angle images of the RCPB/ PAN/RCPB composite, and the initial water contact angle is 69.0°, retaining good hydrophilicity. Wenzel equation believes that the roughness of the membrane affects the wettability of the membrane surface to a certain extent. The increase of roughness can make the membrane more hydrophilic.
The mechanical properties of RC, RCB, RCPB and RCPB/PAN/RCPB membranes are studied by tensile test (Fig. 8). RC membrane has good tensile strength up to 37.0 MPa. The tensile strength of RCB and RCPB is 23.4 and 29.2 MPa, respectively, which is slightly lower than that of RC. According to the literature, the introduction of nanoparticles may not always improve the mechanical properties of the membrane, which may be due to the stress concentration caused by the agglomeration of nanoparticles . However, when the PAN nanofiber membrane was introduced to combine with RCPB, the tensile strength and elongation at break of RCPB/PAN/ RCPB are significantly improved compared with that of RCPB. The tensile strength and elongation at break are 32.1 MPa and 5.34%, which are increased by 1.1 and 2.5 times, respectively, indicating that RCPB/ PAN/RCPB has good mechanical properties.
Catalyst surface charge is one of the important factors affecting the synergistic effect of adsorption and catalysis . As shown in Table S1, the Zeta potential of RCPB/PAN/RCPB in deionized water is the largest, which is −27.4 mV, while the Zeta potentials of RCB and RCPB are −10.8 and −23.7 mV, respectively. It is well known that the enhancement of negative potential will strengthen the electrostatic interaction between the catalyst and Cr(VI), thus accelerating the adsorption process and promoting the catalytic reaction. Accordingly, the introduction of polydopamine and PAN nanofiber membrane enhances the Zeta potential of RCPB/ PAN/RCPB, improving its adsorption capacity for Cr(VI).

Photocatalytic reduction of Cr(VI)
In order to study the photocatalytic activity of the synthesized RCPB, the performance of Cr(VI) reduction under visible light was investigated. In the absence of catalyst, the photoreduction of Cr(VI) itself can be ignored. During the experiment, the pH of the solution was fixed at 3.0, the catalyst concentration was 1.0 g L −1 , and the initial concentration of Cr(VI) was 10 mg L −1 . The effects of different membrane materials and membrane thickness on Cr(VI) reduction were discussed. It can be seen from Fig. 9 that the photoreduction percent of Cr(VI) by RC under visible light is 62.3%, and the reaction rate constant is 0.0084 min −1 . However, the RC has no reduction effect on Cr(VI) after 150 min under dark state. After 120 min of illumination, the reduction percent of Cr(VI) by RCB reaches 84.2%.
For RCPB system, the effect of membrane thickness on photoreduction of Cr(VI) was discussed. Compared with RCPB-0.3, the reduction percent and reaction rate of RCPB-0.2 are significantly better, indicating that the membrane thickness significantly affects the photocatalytic activity. Therefore, RCPB-0.2 (abbreviated as RCPB) was chosen as the photocatalyst for subsequent experiments. After reaction for 120 min under visible light, the reduction percent of Cr(VI) by RCPB was as high as 98.2% (with the reaction rate constant of 0.0329 min −1 ), while it was only 23.2% in dark, indicating that light-harvesting plays a major role in the reduction of Cr(VI) relative to adsorption.
Cr 2 O 7 2− is the main existing form of Cr(VI) in acid medium. In general, when the pH is lower than the point of zero charge of RCPB, Cr(VI) can be strongly adsorbed on the surface of positively charged RCPB by electrostatic interaction, which is beneficial to improve the photocatalytic activity. However, when the catalyst surface is negatively charged with increasing pH, an increase in the Coulomb repulsion between Cr 2 O 7 2− and RCPB inhibited the reduction of Cr(VI) (Zhao et al. 2020). Therefore, excessive H + at low pH The curve of -ln(C t /C 0 ) and illumination time t in the photocatalytic reduction of Cr (VI) by different membranes; c The reaction rate constant k of different membranes will promote the reduction of Cr 2 O 7 2− to Cr(III) by photogenerated electrons (Xiong et al. 2020).
Accordingly, the catalyst concentration was 1.0 g L −1 and the initial concentration of Cr(VI) was 10 mg L −1 during the experiment, and the effect of initial pH (2.0-4.0) on the photocatalytic reduction of Cr(VI) by RCPB was investigated. As shown in Fig. 10, the initial pH has a significant effect on the adsorption and photoreduction of Cr(VI). The reduction percent of Cr(VI) by RCPB decreases with increasing pHs. After 120 min illumination, the k value of Cr(VI) reduction is only 0.0143 min −1 at pH 4.0, while the k values are 0.0933 and 0.0329 min −1 at pH 2.0 and pH 3.0, which are increased by 5.5 and 1.3 times, respectively. However, due to the strong acidity of pH 2.0 system, subsequent experiments were conducted at pH 3.0.
In addition, the optimum dosage of RCPB was assessed. During the experiment, pH 3.0 and the initial Cr(VI) concentration of 10 mg L −1 were fixed. It can be seen from Fig. 11 that the reduction percent of Cr(VI) increases significantly with the increase of RCPB dosage. When the dosage of photocatalyst was 0.5 g L −1 , the reduction percent of Cr(VI) is only 73.8% after 120 min of illumination. When the dosages of photocatalyst were 1.0 and 1.5 g L −1 , the reaction rate constants of Cr(VI) reduction are 0.0329 and 0.0826 min −1 , respectively, and the photoreduction percent of Cr(VI) reaches 98.2%, 1.3 times higher than that of 0.5 g L −1 system. Considering the Fig. 10 a The effect of initial solution pH on RCPB reduction of Cr(VI); b The curve of -ln(C t /C 0 ) and illumination time t in the photocatalytic reduction of Cr (VI) by RCPB; c The reaction rate constant k of RCPB system at different pHs resource saving, 1.0 g L −1 catalyst was used in subsequent experiments.
To study the optimal dosage of RCPB/PAN/RCPB during the experiment, pH 3.0 and initial Cr(VI) concentration of 10 mg L −1 were fixed. It can be seen from Fig. 12 that the reduction percent of Cr(VI) increases significantly with the increase of catalyst dosage. When the photocatalyst dosage was 1.0 g L −1 , the reduction percent of Cr(VI) is 82.6% after 120 min irradiation, and the reaction rate constant is only 0.0151 min −1 . When the dosage of photocatalyst was 1.5 and 2.0 g L −1 , the k values after 120 min of irradiation are 0.0586 and 0.0869 min −1 , respectively, and the photoreduction percent of Cr(VI) could both reach 99.7%, increasing by 17.1%. Considering the resource saving, the optimum photocatalyst dosage of RCPB/PAN/RCPB was 1.5 g L −1 . After the composite of RCPB and PAN nanofiber membrane, not only the mechanical properties of RCPB/PAN/RCPB are improved, but also the reduction percent of Cr(VI) by RCPB/PAN/RCPB is 99.7% within 120 min under the same experimental conditions (the reaction rate of RCPB/PAN/RCPB is slightly inferior to that of RCPB) (Fig. S2).
Cyclic stability is an important indicator to evaluate the photocatalytic performance. In order to explore the reuse performance of RCPB, four cycles of Cr(VI) reduction were conducted under visible light. After each experiment, the membrane was washed and dried for the next use. During the experiment, the dosage of RCPB was fixed at 2.5 g L −1 , pH was 3.0 and the initial concentration of Cr(VI) was 10 mg L −1 . As shown in Fig. S3a, the respective reduction percent of Cr(VI) is 99.9%, 97.0%, 86.2%, and 80.4% after four times of reuse, indicating that the RCPB is relatively stable. On the other hand, after four cycles, the photocatalytic reduction efficiency of Cr(VI) by RCPB/PAN/RCPB only decreases from 99.7 to 84.6% (Fig. 13a). The decrease of the reduction percent may be due to the loss of the quality of the powder catalyst during the cleaning process and the blockage of the membrane pores during the circulation. Due to the existence of PAN nanofiber membrane, the RCPB loaded on the upper and lower surfaces shows more active sites. Compared with the four-cycle experiments of RCPB (Fig. S3), the Cr(VI) reduction percent by RCPB/PAN/RCPB increases to a certain The active substances for photocatalytic reduction of Cr(VI) by RCPB/PAN/RCPB were investigated through capture experiments. During the experiment, 10 mM KBrO 3 (Zhang et al. 2010) and 5 mM EDTA  were used as photogenerated e − and h + capture agents, respectively. As shown in Fig. 14, after the addition of KBrO 3 into the reaction suspension, it is obvious that the whole reduction process is completely inhibited, and the Cr(VI) concentration in the solution remains at 10 mg L −1 after 120 min of illumination. On the other hand, the addition of EDTA improves the photocatalytic reduction performance of RCPB/PAN/RCPB for Cr(VI), and the reduction percent of Cr(VI) reaches 99.3% within 60 min, while the reduction percent of Cr(VI) in the control group is 98.2% after 120 min of illumination. This is mainly because the h + generated in the photocatalytic reaction was captured, which inhibited the recombination of e − and h + pairs, and made more e − participate in the reduction of Cr(VI). Therefore, photogenerated e − is the main active substance for the photocatalytic reduction of Cr(VI) by RCPB/PAN/ RCPB.

Analysis of photoelectric characteristics
In the photo-driven catalysis, the effective separation of photogenerated e − −h + pairs is crucial for Cr(VI) reduction. The charge separation and transfer properties of photocatalyst were analyzed by EIS, transient photocurrent response and PL tests. Figure 15a shows the EIS diagrams of RC, RCB, RCPB and RCPB/ PAN/RCPB membranes. It is obvious that the order of charge transfer resistance is RC > RCB > RCPB/ PAN/RCPB > RCPB, indicating that the resistance of charge transfer at RCPB interface is the smallest, which is conducive to charge transfer and separation (Kim et al. 2017). In addition, as shown in Fig.  S4 and Fig. 15b, RC, RCB, RCPB and RCPB/PAN/ RCPB membranes have obvious transient photocurrent responses under visible light. The photocurrent density of RCPB is the largest, which is 1.9 and 1.5 times higher than that of RC and RCB, respectively. This indicates that RCPB has the highest separation efficiency of e − −h + pairs, which further verifies that RCPB has the optimal photocatalytic reduction performance. The results also show that the introduction of PAN nanofiber membrane increases the recombination of e − −h + pairs to a certain extent, resulting in a slight decrease in the photocatalytic activity of RCPB/PAN/RCPB. In general, the higher PL peak intensity represents the lower separation efficiency of photogenerated carriers. As shown in Fig. 15c, the PL intensity of RCPB is significantly lower than that of RCB, indicating that RCPB has a low e − −h + pair recombination rate (Liu et al. 2021).
Because the light response of photocatalyst is closely related to its catalytic performance, the composite membrane was analyzed by UV-vis diffuse reflectance spectroscopy and the corresponding band gap energy was calculated. As shown in Fig. 16a, the RC has a slight response between 335 and 400 nm. The absorption edge of RCB is 420 nm, while the absorption edge of RCPB shows a red shift. The light absorption in the 400-800 nm region is also significantly increased, and the absorption edge can reach 650 nm, indicating that the light harvesting ability of RCPB is significantly improved. RCPB/ PAN/RCPB has good light response between 200 and 800 nm. The band structure is of great significance for exploring the photocatalytic mechanism (Liu et al. 2021). Through the calculation of Eq. (3), the band gap energies (E g ) of RCB, RCPB and RCPB/ PAN/RCPB membranes are 2.45, 2.25 and 2.43 eV, respectively (the insert in Fig. 16a, b). Subsequently, the Mott-Schottky (MS) curves of RC, RCB, RCPB and RCPB/PAN/RCPB were assessed at 1000 Hz. The MS curve shows a positive slope (Fig. 16c), indicating that both RCB and RCPB are n-type semiconductors. The flat band positions (E fb ) of RCB and RCPB are −0.09 V (vs. Ag/AgCl, pH 7.0) and − 0.19 V (vs. Ag/AgCl, pH 7.0), respectively. In general, the conduction band potential of n-type semiconductor is close to the flat band potential. According to Eq. (4), the conduction band potentials (E CB ) of RCB, RCPB and RCPB/PAN/RCPB are about 0.13 V (vs. NHE, pH 7.0), 0.03 V (vs. NHE, pH 7.0) and 0.12 V (vs. NHE, pH 7.0), respectively. Combined with E g , E CB and Eq. (5), the valence band positions (E VB ) of RCB, RCPB and RCPB/PAN/ RCPB are 2.58, 2.28 and 2.55 V, respectively.
(3) h = A(h − E g ) n/2 (4) E(NHE) = E(Ag∕AgCl) − E θ + 0.059pH Mechanism of photocatalytic reduction of Cr(VI) by RCPB/PAN/RCPB Based on the above analysis, Fig. 17 shows the mechanism of photocatalytic reduction of Cr(VI) in water environment by RCPB/PAN/RCPB composite under visible light. PAN nanofiber membrane is only used as carrier to improve the mechanical properties of RCPB. Capture experiments shows that the main active substances involved in the photocatalytic reduction of Cr(VI) in RCPB system is e − . The reduction potential of RCPB/PAN/RCPB is higher than −O 2 /O 2 − (−0.33 V), so it is difficult to generate O 2 − . RCPB/PAN/RCPB produces e − −h + pairs [(Eq. (6)] under visible light, and the e − on the surface of RCPB/PAN/RCPB can reduce Cr(VI) to Cr(III) [(Eq. (7)].

Conclusions
In this study, PDA/BWO was loaded on the RC by blending, and the RCPB with photocatalytic activity under visible light was synthesized. PAN nanofiber membrane was introduced to prepare RCPB/PAN/ RCPB sandwich structure membrane by scraping method, which can not only enhance charge separation and interfacial charge transfer, but also increase the specific surface area by 1.9 times relative to RCB, thus providing more active sites for Cr(VI) reduction. EIS, transient photocurrent response, DRS and PL further confirmed that RCPB has a low e − −h + pair recombination rate. Compared with RCPB, the tensile strength and elongation at break increases by 1.1 and 2.5 times, respectively, indicating that RCPB/PAN/ RCPB possesses excellent mechanical properties. After four cycles, the photocatalytic reduction percent of Cr(VI) by RCPB/PAN/RCPB remains at 84.6%, higher than that by RCPB, indicating that RCPB/ PAN/RCPB possesses satisfactory cycle performance.