Electrospun Membranes Anchored with g-C3N4/MoS2 for Highly Efficient Photocatalytic Degradation of Aflatoxin B1 under Visible Light

The degradation of aflatoxin (AF) is a topic that always exists along with the food and feed industry. Photocatalytic degradation as an advanced oxidation technology has many benefits, including complete inorganic degradation, no secondary contamination, ease of activity under moderate conditions, and low cost compared with traditional physical, chemical, and biological strategies. However, photocatalysts are usually dispersed during photocatalytic reactions, resulting in energy and time consumption in the separation process. There is even a potential secondary pollution problem from the perspective of food safety. In this regard, three electrospun membranes anchored with g-C3N4/MoS2 composites were prepared for highly efficient photocatalytic degradation of aflatoxin B1 (AFB1) under visible light. These photocatalytic membranes were characterized by XRD, SEM, TEM, FTIR, and XPS. The factors influencing the degradation efficiency of AFB1, including pH values and initial concentrations, were also probed. The three kinds of photocatalytic membranes all exhibited excellent ability to degrade AFB1. Among them, the photocatalytic degradation efficiency of the photocatalytic membranes prepared by the coaxial methods reached 96.8%. The experiment is with an initial concentration of 0.5 μg/mL (500 PPb) after 60 min under visible light irradiation. The mechanism of degradation of AFB1 was also proposed based on active species trapping experiments. Moreover, the prepared photocatalytic membranes exhibited excellent photocatalytic activity even after five-fold use in the degradation of AFB1. These studies showed that electrospun membranes anchored with g-C3N4/MoS2 composites have a high photocatalytic ability which is easily removed from the reacted medium for reuse. Thereby, our study offers a highly effective, economical, and green solution for AFB1 degradation in the foodstuff for practical application.

When the photocatalysts mentioned above were used to degrade AFB 1 and DONs, the photocatalysts were generally suspended during the photocatalytic process [22][23][24][25][26][27][28][29]. As a result, the photocatalyst powders were easy to agglomerate and the separation process after the photocatalytic reaction required a lot of energy, which limited its large-scale application [31]. It is an attractive solution to prepare membranes by electrospinning as the carrier of photocatalysts. Electrospinning can produce fibers of tens to hundreds of nanometers in diameter with good mechanical properties, which can easily immobilize and recycle photocatalysts [32,33]. Thus, the energy consumption in the separation process and possible secondary pollution are reduced. Up to now, we have not found any reports on photocatalytic degradation of AFB 1 using photocatalysts immobilized on electrospun membranes. AFB 1 is often produced during the storage, transportation, and production of foods or food ingredients [2,3]; so, the safety and stability of photocatalysts must be considered. Among the numerous photocatalysts, graphitic carbon nitride (g-C 3 N 4 ) has gained the intensive attention of many researchers, as this metal-free polymeric n-type semiconductor is non-toxic, chemically stable, thermally stable, and easily modified [34]. However, the pristine g-C 3 N 4 is usually restricted by unsatisfactory photocatalytic efficiency due to insufficient solar light absorption and the fast recombination of photogenerated electron-hole pairs [35]. In order to improve the photocatalytic efficiency of g-C 3 N 4 , it is a reasonable strategy to construct heterostructures with other narrow-band gap semiconductors to provide more active sites and inhibit the recombination of photogenerated charges. Molybdenum disulfide (MoS 2 ) consists of three-dimensional stacked atomic layers with direct and indirect band gaps of 1.90 eV and 1.20 eV. It has become one of the most popular emerging co-catalysts due to its appropriate band structure, low cost, non-toxic, and exhibits excellent sunlight harvesting capability [36]. Therefore, it is a good idea to composite g-C 3 N 4 with MoS 2 to form effective heterostructures to enhance the visible light absorption and reduce the recombination of photogenerated electron-hole pairs owing to their matching band-edge positions for photocatalytic application [37]. To the best of our knowledge, the attempt to use electrospun membranes anchored with g-C 3 N 4 /MoS 2 to degrade AFB 1 under visible light irradiation has not been reported.

Results and Discussion
Based on the above considerations, we prepared g-C 3 N 4 /MoS 2 composites by calcination and hydrothermal methods and investigated their photocatalytic properties. Then, the prepared photocatalysts were dispersed in the polymer electrospinning solution synthesized by polyacrylonitrile (PAN), and flexible electrospun membranes with different structures anchored with g-C 3 N 4 /MoS 2 composites were prepared by uniaxial and coaxial methods, respectively. The as-prepared photocatalysts and flexible electrospun membranes (S 1 , S 2 , and S 3 ) were characterized by scanning electron microscopy (SEM), transmission electron microscopy (TEM), X-ray diffraction (XRD), Fourier-transform infrared spectroscopy (FTIR), X-ray photoelectron spectroscopy (XPS), and diffuse reflectance spectra (DRS). The photocatalytic efficiency of electrospun membranes for degradation of AFB 1 under visible light irradiation in an aqueous medium was investigated. Effects of factors such as pH value and the initial concentration of AFB 1 were also studied. Active species trapping experiments analyzed the mechanism of photocatalytic degradation of AFB1. In addition, the effect of recycling on photocatalytic efficiency was also evaluated.

Characterization of the PAN-g-C 3 N 4 /MoS 2 Electrospun Membranes
To study the morphologies of electrospun membranes anchored with g-C 3 N 4 /MoS 2 prepared by different processes, S 1 , S 2 , and S 3 were examined by SEM ( Figure 1). It could be seen spindle-like beads wrapped with g-C 3 N 4 /MoS 2 on S 1 (Figure 1a), which indicated the photocatalysts were successfully immobilized on electrospun membranes. Many other researchers have prepared a series of photocatalytic membranes by similar methods [38]. However, most of the photocatalysts in this kind of membrane were wrapped by polymers, which hindered light absorption and was not conducive to the migration of photogenerated charges to the active sites. Therefore, polyethylene oxide (PEO) was added into the electrospinning solution, which is very soluble in water, and the obtained electrospun membranes were treated with an ultrasonic water bath to expose more photocatalysts. From the red circles marked (Figure 1b), it could be confirmed that pores formed by removing PEO after post-treatment, so that more photocatalysts were exposed and the photocatalytic efficiency was enhanced accordingly. To further expose the photocatalysts, coaxial electrospinning and ultrasonic water washing treatment were adopted to prepare S 3 . Compared with S 1 and S 2 , the spindle-like beads were greatly reduced, and the photocatalysts that were completely exposed due to PEO could be obliterated. The way electrospun nanofibers bound the photocatalysts ( Figure 1c) and wave-like folds caused by the removal of PEO could be observed in the bright area around the red circle. With the increase in photocatalysts exposure, it can be speculated that the photocatalytic efficiency should be improved correspondingly.
The morphologies of the g-C 3 N 4 /MoS 2 composites were further studied by TEM and HRTEM ( Figure 2). It was observed that the well-crystallized MoS 2 lines were loaded on g-C 3 N 4 ( Figure 2a). Furthermore, many clear lattice fringes were shown in the HRTEM image (Figure 2b), indicating that good crystallinity has been obtained. Three sets of different lattices were found with the d-spacing of 0.62 nm, 0.32 nm, and 0.27 nm, respectively, corresponding to the (002) plane of MoS 2 , the (002) plane of g-C 3 N 4 , and the (110) plane of MoS 2 , respectively [39]. Meanwhile, the interface between g-C 3 N 4 and MoS 2 could also be perceived, indicating that the heterostructures were successfully formed between g-C 3 N 4 and MoS 2 .  Figure 1. SEM images of electrospun membranes anchored with g-C3N4/MoS2 prepared by different processes: (a) S1~mostly wrapped, (b) S2~partially exposed, and (c) S3~fully exposed.
The morphologies of the g-C3N4/MoS2 composites were further studied by TEM and HRTEM ( Figure 2). It was observed that the well-crystallized MoS2 lines were loaded on g-C3N4 ( Figure 2a). Furthermore, many clear lattice fringes were shown in the HRTEM image (Figure 2b), indicating that good crystallinity has been obtained. Three sets of different lattices were found with the d-spacing of 0.62 nm, 0.32 nm, and 0.27 nm, respectively, corresponding to the (002) plane of MoS2, the (002) plane of g-C3N4, and the (110) plane of MoS2, respectively [39]. Meanwhile, the interface between g-C3N4 and MoS2 could also be perceived, indicating that the heterostructures were successfully formed between g-C3N4 and MoS2. Figure 1. SEM images of electrospun membranes anchored with g-C 3 N 4 /MoS 2 prepared by different processes: (a) S 1~m ostly wrapped, (b) S 2~p artially exposed, and (c) S 3~f ully exposed.
The crystal structure and composition of g-C 3 N 4 /MoS 2 , S 1 , S 2 , and S 3 were confirmed with X-ray diffraction (XRD). In addition, the XRD pattern of g-C 3 N 4 and MoS 2 was displayed to be compared with g-C 3 N 4 /MoS 2 ( Figure S1), which provided more detailed data. As shown in Figure S1a, several diffraction peaks could be observed at 2θ = 14.5 • , 32.  [40]. Compared with the standard card, the diffraction peaks of g-C 3 N 4 /MoS 2 and MoS 2 shifted slightly to a bigger angle, which might be due to the residual stress in the material [41]. As shown in Figure S1b, the diffraction peak of g-C 3 N 4 , which appeared at 2θ = 13.14 • , was assigned to the (001) plane, attributed to the triazine unit, and the strong peak located at 28.02 • was the typical (002) diffraction plane ascribed to the inter-planar stacking of the aromatic system in g-C 3 N 4 (JCPDS: 87-1526) [29]. By contrast, the diffraction peak of g-C 3 N 4 /MoS 2 shifted to a smaller angle, implying the interaction between the g-C 3 N 4 and MoS 2 . Through the Scherrer formula (Supplementary Information), the crystallite size of g-C 3 N 4 /MoS 2 at the (002) plane could be estimated to be 98 Å, more significant than the crystallite size of g-C 3 N 4 The crystal structure and composition of g-C3N4/MoS2, S1, S2, and S3 were confirmed with X-ray diffraction (XRD). In addition, the XRD pattern of g-C3N4 and MoS2 was displayed to be compared with g-C3N4/MoS2 ( Figure S1), which provided more detailed data. As shown in Figure S1a [40]. Compared with the standard card, the diffraction peaks of g-C3N4/MoS2 and MoS2 shifted slightly to a bigger angle, which might be due to the residual stress in the material [41]. As shown in Figure S1b, the diffraction peak of g-C3N4, which appeared at 2θ = 13.14°, was assigned to the (001) plane, attributed to the triazine unit, and the strong peak located at 28.02° was the typical (002) diffraction plane ascribed to the inter-planar stacking of the aromatic system in g-C3N4 (JCPDS: 87-1526) [29]. By contrast, the diffraction peak of g-C3N4/MoS2 shifted to a smaller angle, implying the interaction between the g-C3N4 and MoS2. Through the Scherrer formula (Supplementary Information), the crystallite size of g-C3N4/MoS2 at the (002) plane could be estimated to be 98 Å, more significant than the crystallite size of g-C3N4 at the (002) plane (88 Å), which might be attributed to the improvement of crystallinity after annealing.
The XRD patterns of S1, S2, and S3 were generally very similar ( Figure 3a) since they were all composed of PAN and g-C3N4/MoS2. The only difference lay in the spatial structure of the photocatalysts and PAN nanofibers. Obvious diffraction peaks belonging to MoS2 and g-C3N4 could be observed at 2θ = 14.72° and 27.5° in the XRD patterns of S1, S2, and S3, respectively. Additionally, wide bumps could be observed in the range of 15-30°, similar to the work of Xie et al. [42], representing the amorphous PAN macromolecules. The results of XRD patterns could confirm the successful combination of g-C3N4/MoS2 composites and PAN electrospun membranes. Other diffraction peaks of g-C3N4/MoS2 were not found in the XRD patterns of S1, S2, and S3 due to the low content of photocatalysts and the amorphous nature of PAN.
The FTIR spectra of the different electrospun membranes were measured (Figure 3b). For pure PAN electrospun membrane, the peaks at 2934 cm −1 , 2242 cm −1 , 1728 cm −1 , 1450 cm −1 , and 1093 cm −1 were assigned to the stretching vibration of methylene -CH2-, stretching vibration of C≡N, stretching vibration of C=O, bending vibration of -CH2-, and stretching vibration of the C-N bonds [42][43][44]. Compared with pure PAN electrospun membrane, the C-N stretching vibration absorption peak of g-C3N4 located at 1235 cm −1 and 1640 cm −1 , and the characteristic peak of the 3-s-triazine structure located at 814 cm −1 , appeared in the FTIR spectra of S1, S2, and S3 [45,46]. Therefore, the FTIR results further demonstrated the successful loading of photocatalysts on electrospun membranes. However, due to the low content of MoS2, its characteristic peaks failed to be observed. It The XRD patterns of S 1 , S 2 , and S 3 were generally very similar ( Figure 3a) since they were all composed of PAN and g-C 3 N 4 /MoS 2 . The only difference lay in the spatial structure of the photocatalysts and PAN nanofibers. Obvious diffraction peaks belonging to MoS 2 and g-C 3 N 4 could be observed at 2θ = 14.72 • and 27.5 • in the XRD patterns of S 1 , S 2 , and S 3 , respectively. Additionally, wide bumps could be observed in the range of 15-30 • , similar to the work of Xie et al. [42], representing the amorphous PAN macromolecules. The results of XRD patterns could confirm the successful combination of g-C 3 N 4 /MoS 2 composites and PAN electrospun membranes. Other diffraction peaks of g-C 3 N 4 /MoS 2 were not found in the XRD patterns of S 1 , S 2 , and S 3 due to the low content of photocatalysts and the amorphous nature of PAN. REVIEW 7 should be noted that the intensity and area of the peaks assigned to g-C3N4 increase turn from S1 to S3, indicating more photocatalysts were exposed, which was benefici improve photocatalytic efficiency. The chemical status and bonding structures of the PAN-g-C3N4/MoS2 electros membranes were analyzed by X-ray photoelectron spectroscopy (XPS). The full-scale survey spectra revealed the existence of C, N, Mo, and S elements ( Figure 4). In addi the peak differentiation imitating the four elements was studied to further understand detailed composition ( Figure 5). The XPS spectra of C 1s could be deconvoluted into peaks (Figure 5a), wherein the peaks at 284.5 eV and 286.3 eV were attributed to the C-C bonds and C-NH2 species of the g-C3N4 [33]. The peak at 284.7 eV (sp 2 C-C) belon The FTIR spectra of the different electrospun membranes were measured ( Figure 3b). For pure PAN electrospun membrane, the peaks at 2934 cm −1 , 2242 cm −1 , 1728 cm −1 , 1450 cm −1 , and 1093 cm −1 were assigned to the stretching vibration of methylene -CH 2 -, stretching vibration of C≡N, stretching vibration of C=O, bending vibration of -CH 2 -, and stretching vibration of the C-N bonds [42][43][44]. Compared with pure PAN electrospun membrane, the C-N stretching vibration absorption peak of g-C 3 N 4 located at 1235 cm −1 and 1640 cm −1 , and the characteristic peak of the 3-s-triazine structure located at 814 cm −1 , appeared in the FTIR spectra of S 1 , S 2 , and S 3 [45,46]. Therefore, the FTIR results further demonstrated the successful loading of photocatalysts on electrospun membranes. However, due to the low content of MoS 2 , its characteristic peaks failed to be observed. It should be noted that the intensity and area of the peaks assigned to g-C 3 N 4 increased in turn from S 1 to S 3 , indicating more photocatalysts were exposed, which was beneficial to improve photocatalytic efficiency.

2023, 15, x FOR PEER
The chemical status and bonding structures of the PAN-g-C 3 N 4 /MoS 2 electrospun membranes were analyzed by X-ray photoelectron spectroscopy (XPS). The full-scale XPS survey spectra revealed the existence of C, N, Mo, and S elements ( Figure 4). In addition, the peak differentiation imitating the four elements was studied to further understand the detailed composition ( Figure 5). The XPS spectra of C 1s could be deconvoluted into four peaks (Figure 5a), wherein the peaks at 284.5 eV and 286.3 eV were attributed to the sp 2 C-C bonds and C-NH 2 species of the g-C 3 N 4 [33]. The peak at 284.7 eV (sp 2 C-C) belonged to C 1s of PAN, and the peak at 288.5 eV could be attributed to the carbon in N-C=N [47]. The XPS spectra of N 1 s had three peaks at 398.7 eV, 400.0 eV, and 401.1 eV, respectively (Figure 5b), which could be attributed to the sp 2 hybridized nitrogen in C-N=C, tertiary nitrogen N-(C) 3 groups, and free amino groups (C-N-H) [33]. Three peaks in the high-resolution XPS spectra of Mo 3d at 225.8 eV, 228.7 eV, and 231.9 eV were further revealed (Figure 5c), belonging to S 2s, Mo 3d 5/2 , and Mo 3d 3/2 , respectively [47]. It could be confirmed that the Mo element in g-C 3 N 4 /MoS 2 was mainly presented in the state of Mo 4+ . Regarding the XPS spectra of S 2p (Figure 5d), two major peaks at 162.4 eV and 163.5 eV could be attributed to S 2p 3/2 and S 2p 1/2 , respectively [47]. The XPS results verified that the g-C 3 N 4 /MoS 2 was successfully anchored with electrospun PAN membranes. The chemical status and bonding structures of the PAN-g-C3N4/MoS2 electros membranes were analyzed by X-ray photoelectron spectroscopy (XPS). The full-scale survey spectra revealed the existence of C, N, Mo, and S elements ( Figure 4). In addit the peak differentiation imitating the four elements was studied to further understand detailed composition ( Figure 5). The XPS spectra of C 1s could be deconvoluted into f peaks (Figure 5a), wherein the peaks at 284.5 eV and 286.3 eV were attributed to the C-C bonds and C-NH2 species of the g-C3N4 [33]. The peak at 284.7 eV (sp 2 C-C) belon to C 1s of PAN, and the peak at 288.5 eV could be attributed to the carbon in N-C=N [ The XPS spectra of N 1s had three peaks at 398.7 eV, 400.0 eV, and 401.1 eV, respectiv (Figure 5b), which could be attributed to the sp 2 hybridized nitrogen in C-N=C, tert nitrogen N-(C)3 groups, and free amino groups (C-N-H) [33]. Three peaks in the h resolution XPS spectra of Mo 3d at 225.8 eV, 228.7 eV, and 231.9 eV were further revea (Figure 5c), belonging to S 2s, Mo 3d5/2, and Mo 3d3/2, respectively [47]. It could be c firmed that the Mo element in g-C3N4/MoS2 was mainly presented in the state of M Regarding the XPS spectra of S 2p (Figure 5d), two major peaks at 162.4 eV and 163.5 could be attributed to S 2p3/2 and S 2p1/2, respectively [47]. The XPS results verified that g-C3N4/MoS2 was successfully anchored with electrospun PAN membranes.   Figure 6a illustrates the DRS spectra of g-C 3 N 4 and g-C 3 N 4 /MoS 2 powders. Compared with pure g-C 3 N 4 , the absorption of g-C 3 N 4 /MoS 2 has stronger intensity at the UV-visible light range and an obvious red-shift, which meant that the compounding of MoS 2 effectively broadens and strengthens the light absorption. The heterojunction constructed between g-C 3 N 4 and MoS 2 changes the optical properties of hybrid materials, promoting the light absorption, and could improve the photocatalytic activity under visible-light irradiation.  Figure 6a illustrates the DRS spectra of g-C3N4 and g-C3N4/MoS2 powders. Compa with pure g-C3N4, the absorption of g-C3N4/MoS2 has stronger intensity at the UV-vis light range and an obvious red-shift, which meant that the compounding of MoS2 ef tively broadens and strengthens the light absorption. The heterojunction constructed tween g-C3N4 and MoS2 changes the optical properties of hybrid materials, promoting light absorption, and could improve the photocatalytic activity under visible-light irr ation.
The results of UV-Vis DRS were used to calculate the band gap energy (Eg) of material through the Kubelka-Munk formula (1): where α, h, ν, and C are the absorption coefficient, Planck constant, optical frequency, constant, respectively. The value of n is determined by the material properties. Thro the Kubelka-Munk formula, the integral band gap of g-C3N4/MoS2 could be estimate be 2.75 eV, while that of g-C3N4 was approximated to be 2.9 eV (Figure 6b). Moreove C3N4/MoS2 with a narrower band gap should have better photocatalytic performance cording to a previous study [48]. Furthermore, the transient photocurrent (TPC) response of the as-prepared S1, S2, S3, and PAN electrospun membrane was displayed (Figure 7) under the condition of light on and off illuminating by a visible light source (Xe lamp, λ ≥ 420 nm). It is known that the higher the photocurrent intensity, the higher the separation rate of photogenerated carriers. Obviously, PAN electrospun membrane had no response to visible light radiation, whereas the photocurrent density of S1, S2, and S3 significantly increased in turn when the Xe lamp was turned on, indicating that more photogenerated charges were generated, which was mainly due to the increasingly exposed g-C3N4/MoS2 from S1 to S3. Therefore, the photocatalysts could be completely exposed by optimizing the preparation method to not only enhance the harvest of light but also promote the transfer of photogenerated charges from the inner to the surface, which might improve the photocatalytic efficiency effectively. The results of UV-Vis DRS were used to calculate the band gap energy (E g ) of the material through the Kubelka-Munk formula (1): where α, h, ν, and C are the absorption coefficient, Planck constant, optical frequency, and constant, respectively. The value of n is determined by the material properties. Through the Kubelka-Munk formula, the integral band gap of g-C 3 N 4 /MoS 2 could be estimated to be 2.75 eV, while that of g-C 3 N 4 was approximated to be 2.9 eV (Figure 6b). Moreover, g-C 3 N 4 /MoS 2 with a narrower band gap should have better photocatalytic performance, according to a previous study [48]. Furthermore, the transient photocurrent (TPC) response of the as-prepared S 1 , S 2 , S 3 , and PAN electrospun membrane was displayed (Figure 7) under the condition of light on and off illuminating by a visible light source (Xe lamp, λ ≥ 420 nm). It is known that the higher the photocurrent intensity, the higher the separation rate of photogenerated carriers. Obviously, PAN electrospun membrane had no response to visible light radiation, whereas the photocurrent density of S 1 , S 2 , and S 3 significantly increased in turn when the Xe lamp was turned on, indicating that more photogenerated charges were generated, which was mainly due to the increasingly exposed g-C 3 N 4 /MoS 2 from S 1 to S 3 . Therefore, the photocatalysts could be completely exposed by optimizing the preparation method to not only enhance the harvest of light but also promote the transfer of photogenerated charges from the inner to the surface, which might improve the photocatalytic efficiency effectively. Furthermore, the transient photocurrent (TPC) response of the as-prepared S1, S2, S3, and PAN electrospun membrane was displayed (Figure 7) under the condition of light on and off illuminating by a visible light source (Xe lamp, λ ≥ 420 nm). It is known that the higher the photocurrent intensity, the higher the separation rate of photogenerated carriers. Obviously, PAN electrospun membrane had no response to visible light radiation, whereas the photocurrent density of S1, S2, and S3 significantly increased in turn when the Xe lamp was turned on, indicating that more photogenerated charges were generated, which was mainly due to the increasingly exposed g-C3N4/MoS2 from S1 to S3. Therefore, the photocatalysts could be completely exposed by optimizing the preparation method to not only enhance the harvest of light but also promote the transfer of photogenerated charges from the inner to the surface, which might improve the photocatalytic efficiency effectively. Figure 7. Transient photocurrent response curves of S1, S2, S3, and PAN electrospun membrane. Figure 8 shows the photocatalytic degradation of RhB (10 mg/mL) over g-C3N4/MoS2 with different mass ratios of MoS2 under visible light irradiation. It can be seen that g- Figure 7. Transient photocurrent response curves of S 1 , S 2 , S 3 , and PAN electrospun membrane. Figure 8 shows the photocatalytic degradation of RhB (10 mg/mL) over g-C 3 N 4 /MoS 2 with different mass ratios of MoS 2 under visible light irradiation. It can be seen that g-C 3 N 4 /MoS 2 (1%) had the highest photocatalytic activity, the degradation rate of RhB over which was close to 85% after 90 min. On the other hand, the degradation rate of g-C 3 N 4 and MoS 2 to RhB was about 32% and 20%, respectively, obviously inefficient in comparison with that of the composite photocatalyst. These results confirmed that the strategy of small amount of compounding MoS 2 with g-C 3 N 4 was workable to promote photocatalytic activity, and the best mass ratio of MoS 2 in g-C 3 N 4 /MoS 2 is 1%.

Photocatalysis and Recycling Performance
The photocatalytic performances were comparatively evaluated by photocatalytic degradation of AFB 1 aqueous solution under visible light irradiation, and AFB 1 aqueous solution without photocatalytic membrane was used as the control group (Figure 9). Before photocatalytic degradation under visible light irradiation, the AFB 1 aqueous solution immersed with S 1 , S 2 , and S 3 was kept in darkness for 30 min to achieve adsorption/desorption equilibrium, and the duration of photocatalytic reaction was 60 min. C3N4/MoS2 (1%) had the highest photocatalytic activity, the degradation rate of RhB over which was close to 85% after 90 min. On the other hand, the degradation rate of g-C3N4 and MoS2 to RhB was about 32% and 20%, respectively, obviously inefficient in comparison with that of the composite photocatalyst. These results confirmed that the strategy of small amount of compounding MoS2 with g-C3N4 was workable to promote photocatalytic activity, and the best mass ratio of MoS2 in g-C3N4/MoS2 is 1%. The photocatalytic performances were comparatively evaluated by photocatalytic degradation of AFB1 aqueous solution under visible light irradiation, and AFB1 aqueous solution without photocatalytic membrane was used as the control group (Figure 9). Before photocatalytic degradation under visible light irradiation, the AFB1 aqueous solution immersed with S1, S2, and S3 was kept in darkness for 30 min to achieve adsorption/desorption equilibrium, and the duration of photocatalytic reaction was 60 min.
It could be observed that for the blank experiment without a photocatalytic membrane, the concentration of AFB1 was unchanged under visible light irradiation. The photocatalytic activity of S1, S2, and S3 was significantly improved, and the photodegradation efficiency was up to 65.5%, 79.2%, and 96.8% in 60 min, respectively (Figure 9a). These results showed that the degradation of AFB1 was mainly due to a photocatalytic reaction. As we speculated, the efficiency of photocatalytic degradation of AFB1 by S1, S2, and S3 increased in turn. S3 showed greatly higher photocatalytic efficiency with a degradation rate of 31.3% and 17.6% higher than S1 and S2, respectively. This implied that g-C3N4/MoS2 anchored on electrospun PAN membranes played an important role in the photocatalytic activity of AFB1 degradation. As the g-C3N4/MoS2 anchored on S3 were utterly exposed, the light-harvesting ability was enhanced compared with S1 and S2. Thus, many photogenerated charges were produced in g-C3N4/MoS2 and more easily transferred to the surface of the photocatalyst because they were not wrapped by the polymer. More importantly, this fully exposed g-C3N4/MoS2 provided more active sites and greatly enhanced the photocatalytic efficiency. The high-performance liquid chromatography (HPLC) chromatogram of AFB1 aqueous solution concentrations with the irradiation time was also demonstrated (Figure 9b). It could be observed that for the blank experiment without a photocatalytic membrane, the concentration of AFB 1 was unchanged under visible light irradiation. The photocatalytic activity of S 1 , S 2 , and S 3 was significantly improved, and the photodegradation efficiency was up to 65.5%, 79.2%, and 96.8% in 60 min, respectively (Figure 9a). These results showed that the degradation of AFB 1 was mainly due to a photocatalytic reaction. As we speculated, the efficiency of photocatalytic degradation of AFB 1 by S 1 , S 2 , and S 3 increased in turn. S 3 showed greatly higher photocatalytic efficiency with a degradation rate of 31.3% and 17.6% higher than S 1 and S 2 , respectively. This implied that g-C 3 N 4 /MoS 2 anchored on electrospun PAN membranes played an important role in the photocatalytic activity of AFB 1 degradation. As the g-C 3 N 4 /MoS 2 anchored on S 3 were utterly exposed, the light-harvesting ability was enhanced compared with S 1 and S 2 . Thus, many photogenerated charges were produced in g-C 3 N 4 /MoS 2 and more easily transferred to the surface of the photocatalyst because they were not wrapped by the polymer. More importantly, this fully exposed g-C 3 N 4 /MoS 2 provided more active sites and greatly enhanced the photocatalytic efficiency. The high-performance liquid chromatography (HPLC) chromatogram of AFB 1 aqueous solution concentrations with the irradiation time was also demonstrated (Figure 9b).
In a typical photocatalytic process, many factors affect photocatalytic performance. Besides the basic properties (crystal structure, particle size, specific surface area, and surface hydroxyl group) and carrier of the photocatalysts, external environmental factors such as light source, irradiation time, temperature, pH value, and initial concentration of reactants also make a certain sense [49]. In this study, the influence of pH values and initial concentrations of AFB 1 on photocatalytic efficiency was estimated, which were two variable factors in practical application. S 3 was used to study the photocatalytic efficiency at pH values of 3, 5, 7, and 9, whereas the concentrations of AFB 1 were kept constant (Figure 9c). It was observed that the degradation of AFB 1 was suppressed in an acidic aqueous solution. With the increase in pH value, the photocatalytic degradation rates of AFB 1 increased accordingly. In the neutral solution with a pH value of 7, nearly 17% of AFB 1 was adsorbed after 30 min. However, in the acidic solution with pH values of 3 and 5, only 8% and 13% of AFB 1 were adsorbed, indicating that the high photocatalytic degradation efficiency might come from high adsorption. The photocatalytic membranes and AFB1 (pH = 5) were positively charged in an acidic solution [26]. The absorption of AFB1 on the active site was low due to the repulsive force between the photocatalytic membranes and AFB1 [26,38]. Subsequently, the photocatalytic efficiency was weakened.
In a typical photocatalytic process, many factors affect photocatalytic performance. Besides the basic properties (crystal structure, particle size, specific surface area, and surface hydroxyl group) and carrier of the photocatalysts, external environmental factors such as light source, irradiation time, temperature, pH value, and initial concentration of reactants also make a certain sense [49]. In this study, the influence of pH values and initial concentrations of AFB1 on photocatalytic efficiency was estimated, which were two variable factors in practical application. S3 was used to study the photocatalytic efficiency at pH values of 3, 5, 7, and 9, For the same reason, in an alkaline solution with a pH value of 9, there was a similar repulsive force between the photocatalytic membranes and AFB 1 . However, the photocatalytic degradation efficiency was not decreased but instead slightly increased. The reason might be that AFB 1 was unstable in the alkaline environment [50]. To investigate the effect of the AFB 1 initial concentration on the photocatalytic degradation efficiency, S 3 was soaked in different initial concentrations of AFB 1 (0.5-4 µg/mL, i.e., 500-4000 PPb) with a pH value of 7 (Figure 9d). It was observed that the photocatalytic degradation efficiency was inversely related to AFB 1 initial concentration. The AFB 1 degradation efficiencies were 97.5% and 63.3% at initial concentrations of 500 and 4000 PPb, respectively. This could be assigned to a constant number of active sites on the photocatalytic membrane. With the increase of initial concentrations and the proceeding of the photocatalytic reaction, competitive adsorption of AFB 1 and its intermediates on the photocatalytic membranes would be aggravated, subsequently affecting the harvest of light and forming a barrier against photoexcitation in g-C 3 N 4 /MoS 2 [28,51].
For the practical application of the photocatalytic membranes, five consecutive photocatalytic experiments were carried out using S 3 under the same experimental conditions with proper washing and drying after each cycle (Figure 9e). The reproducibility results of AFB 1 degradation by S 3 showed that although the degradation pace decreased slightly after each photocatalytic degradation test, the degradation rate reached more than 85% overall. The slight decrease in degradation rate might be due to the contaminant of reused samples during the recovery step by the intermediate products produced in the photocatalytic degradation of AFB 1 . The recyclability of the photocatalytic membranes verified the possibility of practical application and a better economic benefit.
To better understand the mechanism of photocatalytic degradation of AFB 1 by the PAN-g-C 3 N 4 /MoS 2 electrospun membranes, the active species trapping experiments were carried out using S 3 under the same conditions described above (Figure 9f). Isopropanol (IPA), 1,4-benzoquinone (BQ), and ammonium oxalate (AO) were employed as the scavengers for hydroxyl radicals (•OH), super-oxide anion radicals (•O 2 − ), and photogenerated holes (h + ), respectively [52]. After 60 min of visible light irradiation, the degradation rate of AFB 1 without a sacrificial agent was 96.8%, and for others with scavengers IPA, BQ, and AO, the degradation rate was 90.2%, 88.1%, and 15.4%, respectively. Therefore, it could be confirmed that h + was the main active specie in the reaction process.

Mechanism for Enhanced Degradation Performance
Based on the previous results, the possible photocatalytic mechanism of AFB 1 degradation by the PAN-g-C 3 N 4 /MoS 2 electrospun membranes was proposed ( Figure 10). It could be regarded that g-C 3 N 4 /MoS 2 anchored on PAN electrospun membranes was simultaneously excited under visible light irradiation and produced photo-induced electrons and holes. According to previous studies and band gap values estimated by the Kubelka-Munk formula, the conduction band of g-C 3 N 4 (−1.22 eV) is higher than that of MoS 2 (−0.12 eV), and the valence band of MoS 2 (1.78 eV) is lower than that of g-C 3 N 4 (1.68 eV) [53]. The photo-induced electrons produced in g-C 3 N 4 can be easily transferred to the conduction band of MoS 2 through the interface, and the photo-induced holes produced in MoS 2 transfer to the valence band of g-C 3 N 4 in a similar manner. As a result, the photo-induced electrons are gathered in the conduction band of MoS 2 , and the photo-induced holes are gathered in the valence band of g-C 3 N 4 , which leads to photo-induced electrons and holes to separate effectively. Therefore, the probability of photo-induced electron-hole recombination is hindered, and the photocatalytic efficiency is improved accordingly. However, the conduction band potential of MoS 2 is more positive than the potential of E(O 2 /•O 2 − ) (−0.12 V > −0.33 V) [54]; the electrons on the conduction band of MoS 2 cannot react with O 2 to generate •O 2 − . For the same reason, the holes on the valence band of g-C 3 N 4 cannot generate •OH, as the valence band of g-C 3 N 4 is more negative than the potential of E(OH − /•OH) or E(H 2 O/•OH) (1.56 V < 1.99 or 2.4 V) [55]. Thereby, rich holes in the valence band of g-C 3 N 4 act as the main reactive species to oxidize AFB 1 directly, consistent with the results of active species trapping experiments. The reaction formulas are as follows: g-C 3 N 4 /MoS 2 + hν → e − (CB) + h + (VB) (2) separate effectively. Therefore, the probability of photo-induced electron-hole recombination is hindered, and the photocatalytic efficiency is improved accordingly. However, the conduction band potential of MoS2 is more positive than the potential of E(O2/•O2 − ) (−0.12 V > −0.33 V) [54]; the electrons on the conduction band of MoS2 cannot react with O2 to generate •O2 − . For the same reason, the holes on the valence band of g-C3N4 cannot generate •OH, as the valence band of g-C3N4 is more negative than the potential of E(OH − /•OH) or E(H2O/•OH) (1.56 V < 1.99 or 2.4 V) [55]. Thereby, rich holes in the valence band of g-C3N4 act as the main reactive species to oxidize AFB1 directly, consistent with the results of active species trapping experiments. The reaction formulas are as follows: AFB1 + h + → CO2 + H2O + intermediate products (3) Figure 10. The photocatalytic mechanism of PAN-g-C3N4/MoS2 electrospun membranes for degradation of AFB1 [53].

Conclusions
Three kinds of flexible electrospun membranes anchored with g-C3N4/MoS2 composites were synthesized via uniaxial or coaxial electrospinning techniques. Due to more g-C3N4/MoS2 photocatalysts being exposed and more active sites being produced, the photocatalytic efficiency of S1, S2, and S3 increased gradually. The degradation efficiency of AFB1 solution with a concentration of 500 PPb (50 mL) was up to 97% in 60 min under visible light irradiation with 0.025 g S3. The mechanism of photocatalytic membranes degradation of AFB1 in the photocatalytic process was proposed based on active species trapping experiments, and the reusability and stable activity were confirmed after five cycles of photocatalytic degradation experiments. Thus, the PAN-g-C3N4/MoS2 electrospun membranes were proved as high photocatalytic activity, easy separation, good reusability, and potential practical application in the foodstuff for the degradation of AFB1. Figure 10. The photocatalytic mechanism of PAN-g-C 3 N 4 /MoS 2 electrospun membranes for degradation of AFB 1 [53].

Conclusions
Three kinds of flexible electrospun membranes anchored with g-C 3 N 4 /MoS 2 composites were synthesized via uniaxial or coaxial electrospinning techniques. Due to more g-C 3 N 4 /MoS 2 photocatalysts being exposed and more active sites being produced, the photocatalytic efficiency of S 1 , S 2 , and S 3 increased gradually. The degradation efficiency of AFB 1 solution with a concentration of 500 PPb (50 mL) was up to 97% in 60 min under visible light irradiation with 0.025 g S 3 . The mechanism of photocatalytic membranes degradation of AFB 1 in the photocatalytic process was proposed based on active species trapping experiments, and the reusability and stable activity were confirmed after five cycles of photocatalytic degradation experiments. Thus, the PAN-g-C 3 N 4 /MoS 2 electrospun membranes were proved as high photocatalytic activity, easy separation, good reusability, and potential practical application in the foodstuff for the degradation of AFB 1 .

Preparation of g-C 3 N 4 /MoS 2
As shown in Scheme 1, the g-C 3 N 4 powders were prepared by calcining melamine at 550 • C for 3 h (5 • C/min). The MoS 2 powders were prepared by hydrothermal process. In a typical procedure, 20 mg sodium molybdate and 25 mg thioacetamide were dissolved in 30 mL deionized water under magnetic stirring for 20 min. Then, the above solution was poured into a stainless-steel autoclave, and the reaction temperature was controlled at 200 • C by the oven for 24 h. Following several times washing with deionized water and ethanol, the resultants dried at 60 • C for 10 h under vacuum were MoS 2 powders. at 200 °C by the oven for 24 h. Following several times washing with deionized w ethanol, the resultants dried at 60 °C for 10 h under vacuum were MoS2 powder The g-C3N4/MoS2 composites were fabricated by low-temperature calcina the mass ratio of MoS2 in g-C3N4/MoS2 was determined as 1% in this study. Firstl g-C3N4 and 2 mg MoS2 powders were dispersed in NMP and absolute ethano tively, and ultrasonicated for 60 min. The two solutions were then mixed and s 12 h, and the precipitates obtained after centrifugation were washed with deioniz and ethanol several times and dried at 80 °C for 10 h under vacuum. Secondly cipitates were ground to powders and followed by annealing at 400 °C for 2 ramping speed of 5 °C/min in a nitrogen atmosphere. Finally, the g-C3N4/MoS2 co were ball-milled for 3 h after cooling to room temperature for future use. Accord above scheme, the g-C3N4/MoS2 composites with different MoS2 mass contents 0. 2%, and 2.5% were prepared by changing the amount of MoS2 added.

Preparation of PAN-g-C3N4/MoS2 Electrospun Membranes
Three kinds of PAN-g-C3N4/MoS2 electrospun membranes were fabricated trospinning (Scheme 2). For the first one, a certain amount of g-C3N4/MoS2 compo added into DMF and ultrasonicated for 1 h to disperse the photocatalysts. Subs PAN was added and stirred for 2 h to obtain a yellow-grey solution. The concen PAN in DMF was 12 w/v%, and the contents of g-C3N4/MoS2 composites to DM Scheme 1. The schematic illustration of the fabrication of g-C 3 N 4 , MoS 2 , and g-C 3 N 4 /MoS 2 .
The g-C 3 N 4 /MoS 2 composites were fabricated by low-temperature calcination, and the mass ratio of MoS 2 in g-C 3 N 4 /MoS 2 was determined as 1% in this study. Firstly, 198 mg g-C 3 N 4 and 2 mg MoS 2 powders were dispersed in NMP and absolute ethanol, respectively, and ultrasonicated for 60 min. The two solutions were then mixed and stirred for 12 h, and the precipitates obtained after centrifugation were washed with deionized water and ethanol several times and dried at 80 • C for 10 h under vacuum. Secondly, the precipitates were ground to powders and followed by annealing at 400 • C for 2 h with a ramping speed of 5 • C/min in a nitrogen atmosphere. Finally, the g-C 3 N 4 /MoS 2 composites were ball-milled for 3 h after cooling to room temperature for future use. According to the above scheme, the g-C 3 N 4 /MoS 2 composites with different MoS 2 mass contents 0.5%, 1.5%, 2%, and 2.5% were prepared by changing the amount of MoS 2 added.

Preparation of PAN-g-C3N4/MoS2 Electrospun Membranes
Three kinds of PAN-g-C 3 N 4 /MoS 2 electrospun membranes were fabricated by electrospinning (Scheme 2). For the first one, a certain amount of g-C 3 N 4 /MoS 2 composites was added into DMF and ultrasonicated for 1 h to disperse the photocatalysts. Subsequently, PAN was added and stirred for 2 h to obtain a yellow-grey solution. The concentration of PAN in DMF was 12 w/v%, and the contents of g-C 3 N 4 /MoS 2 composites to DMF was 3 w/v%. The prepared solution was then injected into a plastic syringe with a metal needle driven by a syringe pump at a flow rate of 1.5 mL/h for electrospinning. The applied voltage was 10 kV, and the distance from the metallic needle to the aluminum foil surface was 15 cm. After electrospinning, the electrospun membranes were dried at 60 • C under vacuum for 12 h, recorded as S 1 . The second one was prepared according to S 1 with some modifications. Typically, the polymer added into DMF was changed to PAN/PEO (PAN: PEO = 2:1, wt%), while keeping the total concentration of the polymer constant with S 1 (12 w/v%). After drying at 60 • C under vacuum for 12 h, the electrospun membranes were immersed in deionized water, sonicated in a water bath for 1 h, and placed at 60 • C for 24 h to fully wash out PEO. The washed electrospun membranes were dried at 60 • C under vacuum for 12 h, recorded as S 2 . The third one was prepared by a simple coaxial electrospinning technique. The core solution with concentration PAN 12 w/v% was prepared similarly to S 1 without adding g-C 3 N 4 /MoS 2 composites. The sheath solution was prepared with PEO, g-C 3 N 4 /MoS 2 , and DMF similar to S 1 . The concentration of PEO in DMF was set to 7 w/v%, and the contents of g-C 3 N 4 /MoS 2 composites to DMF were 3 w/v%. The core and sheath solution was pumped out at rates of 1.5 mL/h using two syringe pumps, and the applied voltage and the distance from the metallic needle to the aluminum foil surface were set to be the same as both S 1 and S 2 . The resultant electrospun membranes were washed with deionized water and dried at 60 • C under vacuum for 12 h, recorded as S 3 .
w/v%. The prepared solution was then injected into a plastic syringe with a metal needle driven by a syringe pump at a flow rate of 1.5 mL/h for electrospinning. The applied voltage was 10 kV, and the distance from the metallic needle to the aluminum foil surface was 15 cm. After electrospinning, the electrospun membranes were dried at 60 °C under vacuum for 12 h, recorded as S1. The second one was prepared according to S1 with some modifications. Typically, the polymer added into DMF was changed to PAN/PEO (PAN: PEO = 2:1, wt%), while keeping the total concentration of the polymer constant with S1 (12 w/v%). After drying at 60 °C under vacuum for 12 h, the electrospun membranes were immersed in deionized water, sonicated in a water bath for 1 h, and placed at 60 °C for 24 h to fully wash out PEO. The washed electrospun membranes were dried at 60 °C under vacuum for 12 h, recorded as S2. The third one was prepared by a simple coaxial electrospinning technique. The core solution with concentration PAN 12 w/v% was prepared similarly to S1 without adding g-C3N4/MoS2 composites. The sheath solution was prepared with PEO, g-C3N4/MoS2, and DMF similar to S1. The concentration of PEO in DMF was set to 7 w/v%, and the contents of g-C3N4/MoS2 composites to DMF were 3 w/v%. The core and sheath solution was pumped out at rates of 1.5 mL/h using two syringe pumps, and the applied voltage and the distance from the metallic needle to the aluminum foil surface were set to be the same as both S1 and S2. The resultant electrospun membranes were washed with deionized water and dried at 60 °C under vacuum for 12 h, recorded as S3. Scheme 2. The schematic illustration of the fabrication of S1, S2, and S3.

Characterization of PAN-g-C3N4/MoS2 Electrospun Membranes
The morphologies of the PAN-g-C3N4/MoS2 electrospun membranes were observed by SEM (ZEISS Sigma, Aalen, Germany), and the microstructure of g-C3N4/MoS2 composites were observed by TEM (JEM-2100F). XRD patterns was obtained with an X-ray diffractometer (MiniFlex 600, Tokyo, Japan) at a scanning speed of 2°/min. FTIR spectra were analyzed on a Vector-22 spectrometer. High-resolution XPS spectra were analyzed by an X-ray photoelectron spectrometer. DRS was detected by a UV/VIS spectrophotometer (Shimadzu UV-3600 Plus, Tokyo, Japan). TPC curves were tested on a three-electrode electrochemical workstation (CHI600E, Beijing, China) with PAN-g-C3N4/MoS2 electrospun membranes/glassy as the working electrode, Ag/AgCl as the reference electrode, and platinum wire as the counter electrode, respectively. The electrolyte was 0.1 M Na2SO4 aqueous solution.

Photocatalytic Degradation Experiment
Scheme 2. The schematic illustration of the fabrication of S 1 , S 2 , and S 3 .

Characterization of PAN-g-C 3 N 4 /MoS 2 Electrospun Membranes
The morphologies of the PAN-g-C 3 N 4 /MoS 2 electrospun membranes were observed by SEM (ZEISS Sigma, Aalen, Germany), and the microstructure of g-C 3 N 4 /MoS 2 composites were observed by TEM (JEM-2100F). XRD patterns was obtained with an X-ray diffractometer (MiniFlex 600, Tokyo, Japan) at a scanning speed of 2 • /min. FTIR spectra were analyzed on a Vector-22 spectrometer. High-resolution XPS spectra were analyzed by an X-ray photoelectron spectrometer. DRS was detected by a UV/VIS spectrophotometer (Shimadzu UV-3600 Plus, Tokyo, Japan). TPC curves were tested on a three-electrode electrochemical workstation (CHI600E, Beijing, China) with PAN-g-C 3 N 4 /MoS 2 electrospun membranes/glassy as the working electrode, Ag/AgCl as the reference electrode, and platinum wire as the counter electrode, respectively. The electrolyte was 0.1 M Na 2 SO 4 aqueous solution.

Photocatalytic Degradation Experiment
The degradation of AFB 1 was evaluated in an aqueous medium under visible light irradiation by a 300 W xenon lamp with a 400 nm cut-off filter. Samples from electrospun membranes were cut into a circular shape (2 cm in diameter and approximately 0.025 g in weight) and fixed on a bracket, immersed in 50 mL of AFB 1 aqueous solution (500 PPb). Then, it was placed in the dark for 30 min to establish the adsorption/desorption equilibrium before light irradiation. The distance between the xenon lamp and the aqueous surface was 10 cm. In the progress of the photocatalytic degradation, 0.5 mL of the AFB 1 aqueous solution was collected every 10 min and then added 0.25 mL glacial acetic acid and 0.25 mL trifluoroacetic acid. The mixed solution was put in a water bath at 70 • C for 40 min to enhance the fluorescence emission intensity of AFB 1 when detected by HPLC. The concentration of the AFB 1 was analyzed by the HPLC on Waters-600 equipped with a UV/Vis detector (emission wavelength at 365 nm) and C-18 Phenomenex reverse phase column (250 × 4.6 mm i.d., 5 µm) at a flow rate of 1 mL/min with an isocratic system composed of water: methanol: acetonitrile (70:20:10). Different factors were also analyzed, such as pH values (4-10) and initial concentration of AFB 1 . The AFB 1 solution without electrospun membranes upon irradiation was also monitored in order to quantify the photocatalytic degradation of AFB 1 . The stability of the electrospun membranes was evaluated over 5 continuous cycle experiments under visible light irradiation. After each cycle, the electrospun membranes were rinsed with deionized water for continued use.
To explore the mechanism of degradation of AFB 1 by the electrospun membranes, active species trapping experiments were carried out by using the addition of IPA (1 mM