Photocatalytic degradation of industrial acrylonitrile wastewater by F–S–Bi–TiO2 catalyst of ultrafine nanoparticles dispersed with SiO2 under natural sunlight

Highly active photocatalyst, having certain anti-ionic interfering function, of F, S and Bi doped TiO2/SiO2 was used for the first time to degrade the organic pollutants in acrylonitrile industrial wastewater under natural sunlight. The photocatalyst were prepared and characterized by UV–Vis, XRD, TEM, EDS, Nitrogen physical adsorption and XPS technique. UV–Vis analysis revealed addition of F, S and Bi into the lattice of TiO2 led to the expansion of TiO2 response in the visible region and hence the efficient separation of charge carrier. The photocatalytic potential of as prepared catalyst to degrade acrylonitrile wastewater under simulated and natural sunlight irradiation was investigated. The extent of degradation of acrylonitrile wastewater was evaluated by chemical oxygen demand (CODCr). CODCr in wastewater decreased from 88.36 to 7.20 mgL−1 via 14 h irradiation of simulated sunlight and achieved regulation discharge by 6 h under natural sunlight, illuminating our photocatalyst effectiveness for refractory industrial wastewater treatment. From TEM results, we found that SiO2 could disperse the photocatalyst with different component distributions between the surface and the bulk phase that should also be responsible for the light absorption and excellent photocatalytic performance. The XPS analysis confirmed the presence of surface hydroxyl group, oxygen vacancies.

Scientific RepoRtS | (2020) 10:12379 | https://doi.org/10.1038/s41598-020-69012-z www.nature.com/scientificreports/ treatment of industrial wastewater under sunlight was a great challenge for the researchers. Watanabe 12 have reported that photocatalysis would cause a new environment revolution twenty years ago. It is well known that, semiconductor-based photocatalysts have been investigated as an auspicious material for the solar energy conversion in regard to the breakdown of hazardous organic pollutants 13 . Across various photocatalysts, titanium dioxide (TiO 2 ) is known as the most determined material because of its chemical stability, high oxidation potential, nontoxicity and physical stability 14 . However, the use of TiO 2 in photocatalysis are limited because of its certain drawbacks like: their large band gap 15 , which means that in solar energy processes, only UV light can be utilized and their low photocatalytic efficiency because of the fast recombination rate of electron-hole pairs. Therefore, many efforts have been promoted to reduce the bandgap of TiO 2 by doping or by band gap engineering [12][13][14][15][16][17][18][19] . In our previous study, the high photocatalytic activity of F-doped TiO 2 was attributed to the increase in the number and strength of surface acid sites 20 . It was explained that F-doping led to the creation of surface oxygen vacancies 17 , or the increase of Ti 3+ state 21 . On the other hand, higher photocatalytic activity of S doped sample was attributed to the increase in the surface Bronsted and Lewis acid sites 22 . Samantaray indicated that sulphate radical impregnation decreases the crystallite size and stabilized the anatase phase of TiO 2 23 . In addition, Bi doped TiO 2 exhibited a red shift in the optical adsorption and Bi 3+δ+ species played a vital role in minimizing the electron hole recombination 16 . According to Li et al. 24 , Bi doping into TiO 2 generates a new intermediate energy level below the conduction band edge of TiO 2 , extending the absorption in the visible region and enhanced their photocatalytic efficiency. On the other hand, SiO 2 was used usually as a supporter, and its dispersing effect on nanoparticle size as well as that with oxidativity has not been reported.
To improve the photocatalytic activity of TiO 2 for the decomposition of organic pollutants in acrylonitrile wastewater under solar light irradiation, we modified TiO 2 with the combination of silica and F, S, Bi doping (F-S-Bi-TiO 2 /SiO 2 ). The dispersion of SiO 2 produced ultrafine nanoparticles. We have found that silica dispersion changed the aggregation state, constituent distribution in amount and morphology of the nanocatalyst, which was responsible for light absorption and increased photocatalytic activity 20 . We first degraded the acrylonitrile simulated wastewater and then degraded the acrylonitrile wastewater. This photocatalyst exhibited excellent performances in both the photocatalytic decompositions of organic pollutants under simulated and natural sunlight. So, our approach is an important attempt for the photocatalytic treatment of industrial wastewater.

Characterization
The valence states on the surface of catalysts were analyzed by a Thermo ESCALAB 250XI X-ray photoelectron spectrometer (America) using Al Kα (hn = 1,486.6 eV) as a radiation source. The irradiation of simulated sunlight and natural sunlight intensity was measured with a FZ-A RADIOMETER irradiance meter (China). From 6 am-8 pm, it was measured at a certain interval as shown in Table S1. Tecnai G 2 F30 TEM was used to analyze the physical structural characteristics of the photocatalysts. The samples were ultrasonically dispersed in ethanol. The suspension was deposited on a Lacey-carbon film, which was supported on a copper grid. The particle size distributions of catalysts with or without SiO 2 dispersant were calculated with NIH software using TEM image treatment. The crystalline phases of the photocatalysts were determined by X-ray diffractometer (RIGAKU, D/Max 2500PC, Japan) at a scanning rate of 6° min −1 in the 2θ angle range of 10°-80° using Cu Kα radiation combined with nickel filter. The accelerating voltage and the applied current were 40 kV and 200 mA, respectively. Crystallite sizes were calculated according to Scherrer equation: where L, K, λ and θ are the average crystal size, the shape factor for spherical crystallites, the X-ray wavelength and Bragg diffraction angle, respectively. B, B mea and b ins are the breadths of intrinsic diffraction profile, the test sample diffraction integral profile and instrumental diffraction profile, respectively. UV-visible spectra were measured on UV-2450 UV spectrophotometer (Shimadzu Corporation, Japan). The range of the scanning wavelength was 200-800 nm. The BET specific surface area was measured by BELSOROP-MINI II (Japan) adsorption instrument and pore size distribution was analyzed by Barrett-Joyner-Halenda (BJH) method.
GC-MS (Agilent 7890A-5795C, America) was used for the component's analysis. The instrument was equipped with a DB-5 capillary column (length of 30 m, 0.25 mm i.d., 0.25 mm d.f.). The injector and MS transmission line temperatures were 250 and 310 °C, respectively. The oven temperature initiated at 40 °C (hold for 5 min), and then increased at 5 °C min −1 to 290 °C (hold for 2 min). The electron energy was set at 70 eV and the ion source temperature was 230 °C. The standard spectra in GC-MS database was used to identify the chemical constituents in wastewater. In the present study, refractory acrylonitrile wastewater was obtained from an acrylonitrile manufacturing plant. The acrylonitrile wastewater was pretreated through adsorption with microporous zeolite, HZSM-5 before photocatalytic degradation, COD Cr decreased from 582.4 to 88.36 mgL −1 , and after that there were no change in the values. The organic pollutants in the wastewater after adsorption were detected and the results were listed in Table 1.
Photocatalytic activities for acrylonitrile simulated wastewater. The photocatalytic activities of the photocatalyst used for acrylonitrile simulated wastewater were evaluated in a photocatalytic reaction system 20 . The quartz glass reactor was sealed after 180 mL of acrylonitrile wastewater (10 mgL −1 ) and 300 mg of the catalyst was placed in it. The mixture was magnetically stirred in the dark until the adsorption equilibrium attained. Then, the spherical Xenon short arc lamp (AHD350, 350 W) was turned on for 12 min. In the process of illumination, the reaction solution of 1 mL was taken out at the interval of 2 min, and after that photocatalyst was filtered out through a filter film of 0.45 μm, acrylonitrile concentration was measured by HP-LC with LC-2030 UV detector and Sunfire TM C18 column. The wavelength of detector was 210 nm and the mobile phase volume ratio of methanol to water was 3:7. Triplicate samples from each batch were taken for the tests.
Scientific RepoRtS | (2020) 10:12379 | https://doi.org/10.1038/s41598-020-69012-z www.nature.com/scientificreports/ The ion effects on photocatalytic activities of acrylonitrile simulated wastewater. The ion effects were examined by following procedure: appropriate amount of sodium sulfate and sodium chloride were added to 180 mL of solution to obtain 61.8 mgL −1 of sulfate radical and 22 mgL −1 of chloride ion consistent with Table 2. The ion solution containing acrylonitrile was used for the comparative trial.
Activity evaluation by COD cr and TOC measurements for acrylonitrile wastewater after adsorbed by HZSM-5. The catalyst contents and reaction device were used as the same as that in simulated wastewater. After adsorption equilibrium, the spherical Xenon lamp was turned on for 14 h. At given intervals of illumination, 4 mL of reaction solution was taken out and was filtered out through a filter film of 0.45 μm (all detection procedures in this study were performed according to or as per China national standard except for special mention). COD Cr values were detected with a Fast COD Detection Instrument (LH-5B-3B(V8)). TOC concentrations of the samples were measured via a Shimadzu TOC analyzer (TOC-L CPN, Japan). Concentration of inorganic ions, BOD and COD Cr in acrylonitrile raw wastewater, the wastewater after adsorption measured according to national standard methods were listed in Table 2.  Table 2. Concentration of inorganic ions, BOD and COD Cr in acrylonitrile raw wastewater and the wastewater after adsorption by HZSM-5.

Results
Effect of different ions on the activity and light adsorption performance. Figure 1a shows the photocatalytic performances of the prepared samples for acrylonitrile simulated wastewater under the effect of spherical Xenon short arc lamp. The degradation ratio of F-S-Bi-TiO 2 /SiO 2 catalyst after 4 min was 63%, which was much higher than TiO 2 -P25 and F-S-Bi-TiO 2 .
In order to demonstrate the interference of inorganic ions, a certain amounts of sodium sulfate (61.8 mgL −1 sulfate radical) and sodium chloride (22 mgL −1 chloride ion) were added into acrylonitrile simulated wastewater and the degradation was carried out by the catalyst under the same condition mentioned above. These results revealed that free sulfate radical, chloride ions, or sodium ions inhibited the degradation progress. This was similar with our previous studies which illustrated that, certain inhibition effect by free sulfate radical on the degradation process 22 . Figure 1b shows the UV-Visible spectra of TiO 2 P-25 (Degussa), F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 catalysts. The absorption of F-S-Bi-TiO 2 /SiO 2 catalyst was increased in the UV-visible region at 300-600 nm as compared to TiO 2 P-25. However, the UV-visible spectrum of F-S-Bi-TiO 2 became flat below 335 nm, showing weakest absorbance. Although, F-S-Bi-TiO 2 /SiO 2 exhibited lower absorption ability towards visible region (400-600 nm) than F-S-Bi-TiO 2 , showing the higher degradation ratio. The higher degradation ratio was attributed to the better dispersion of SiO 2 , because in F-S-Bi-TiO 2 /SiO 2 photocatalyst TiO 2 was well dispersed, reducing the agglomeration and enhancing the absorption in the UV region 22 . However, both the photocatalysts have the same weight but F-S-Bi-TiO 2 contains more S elements and absorbed more visible light. The Photocatalytic reaction mainly depends upon the UV light. XRD analysis. In order to know the crystal structure of the prepared catalyst, XRD patterns of F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 calcinated at 450 °C were recorded as shown in Fig. S2. The crystallite size of the catalysts was calculated with the most predominant peak of the anatase face (101) with the help of Scherrer equation (Table 3). It was clearly revealed from Table 3 that the addition of SiO 2 as a dispersant agent decreases the average crystal size, and increases the surface area of the photocatalyst.
TEM and EDS analysis. The TEM images of as prepared photocatalyst without SiO 2 were shown in Fig. 2a-c. From Fig. 2a, we observed the crystal with grains size of 11-60 nm, and they might be formed in different aggregated stages. Aggregation of ca. 4 nm of particles were produced on the rough end faces, where borderline disappeared in the interior of the large piece crystal (Fig. 2b). The other type gave out indistinct borderline in Fig. 2c. The crystallite faces of TiO 2 (d = 0.356 nm, (101)) and Bi 4 Ti 3 O 12 (d = 0.364 nm, (009)) were exhibited in Fig. 2b,c. Package morphology has been identified clearly by TEM images (Fig. 2d). The atoms are arranged   (Fig. 2f,g). As comparison to Fig. 2b, TiO 2 crystal with SiO 2 was also constituted by smaller particles of 2-4 nm with identical crystal face (Fig. 2f). Bi 4 Ti 3 O 12 (d = 0.211 nm, (2010)) crystal with borderline mark aggregated into a large crystal and also possessed identical crystal face (Fig. 2g). The explanation was that crystal growth might influence each other in same aggregate through the interface, connected among gel particles to form relatively larger identical crystal face, like as biomimetic crystallization. Based on the same principle of interaction of end face atoms, different crystal faces were developed at starting on an end face on the basis of total lowest-energy rule of the system (Fig. 2i). Bi 2 Ti 2 O 7 in Fig. 2j was regarded as the intermediate phase of Bi 4 Ti 3 O 12 formation 25 . A large number of bumps were also formed on the rough surfaces (Fig. 2k). The packages were formed by silk ribbon-like film twining (Fig. 2h). This film could be consisting of long identical crystal face and generated from stirring drawing. Relatively large mass of SiO 2 promoted drawing in late stage of gelation. The EDS images in Fig. 2 show that Si, O, Ti, F, S and Bi elements were evenly distributed on the surface of S-Bi-F-TiO 2 /SiO 2 catalyst 26 , which confirmed the conjecture that the elements were not detected in XRD. The particle size distribution of the catalyst was calculated with NIH software for all particles in random region and was shown in Fig. 3. The catalyst without SiO 2 showed a wide particle size distribution. The average diameter was 30.9 nm, which was smaller than crystal grain size (40.4 nm) as calculated by XRD (Table 3). Since the borderline was not identified by the program, we measured the diameters of particles artificially and calculated particle size distribution of the catalyst with SiO 2 on the basis of the same rule. The result was shown in Fig. 3b. It has been found that the average diameter of the particles is 12.3 nm, which was in agreement with the XRD result (13.9 nm). Compared to the catalyst without SiO 2 , nanoparticles sizes are apparently different. Nanoparticles sizes of the major parts were above ca, 20 nm for the catalyst without SiO 2 , contrarily, they were below 16 nm for the catalyst with SiO 2 and the major parts belonged to ultrafine nanoparticles 27 . Figure 4 represents the nitrogen adsorption and desorption isotherms of F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 calcined at 450 °C. Both F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 display type IV isotherm and H 2 hysteresis, which indicate the presence of mesoporous materials.

Nitrogen physical adsorption.
Moreover, the F-S-Bi-TiO 2 /SiO 2 sample has the similar pore structure with F-S-Bi-TiO 2 . The inset in Fig. 4 shows the plot for the pore size distribution determined by Barrett-Joyner-Halenda (BJH) method from the adsorption branch of the isotherm. The average pore diameters and total pore volume of F-S-Bi-TiO 2 were 26.0 nm, 0.30 cm 3 /g and for F-S-Bi-TiO 2 /SiO 2 are 14.1 nm, 0.67 cm 3 /g. Both of them exhibit mesopore rich structure. However, at all pressure region, the adsorption amount of N 2 on F-S-Bi-TiO 2 /SiO 2 was higher than that of F-S-Bi-TiO 2 , which indicates that a lot of relatively small mesoporous on the surface of F-S-Bi-TiO 2 /SiO 2 were formed under the action of SiO 2 . Therefore, the specific surface area of F-S-Bi-TiO 2 /SiO 2 (194.3 m 2 /g) was higher than that of F-S-Bi-TiO 2 (45.9 m 2 /g). And, greater adsorption capacity will lead to greater degradation rate, which was consistent with the experimental results. XPS analysis. XPS spectra of F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 were shown in Fig. 5. Since fluorine doping converted Ti 4+ state to Ti 3+ , and these Ti 3+ state was related to oxygen vacancies 28 , the content of F changes in both types of the catalysts were investigated. In Fig. 5a the F 1s spectrum of F-S-Bi-TiO 2 shows only one peak centered at binding energy 684.4 eV (represented by red color), indicated only surface fluoride species were present in F-S-Bi-TiO 2 . However, in case of F-S-Bi-TiO 2 /SiO 2 , the F 1s XPS spectra shows two peaks centered at binding energy 688.5 and 686.2 eV, respectively (represented by black color). The peak at binding energy 688.5 eV was attributed to the doped F into the substituted sites of TiO 2 lattice and produced mixed oxide structure of O-Ti-F 18,29,30 . The lower binding energy, centered at 686.2 eV was attributed to the surface fluoride species adsorbed on the surface of TiO 2 30 . This result shows that silica plays a very important role in stabiliz- www.nature.com/scientificreports/ ing the dispersion of F ions. Next, the O 1s spectrum of F-S-Bi-TiO 2 /SiO 2 catalyst was shown in Fig. 5b. The peak at binding energy 533.8 eV was attributed to oxygen present in surface hydroxyl species 29 , while the peak at binding energy 530.4 eV ascribed to lattice oxygen of TiO 2 16 . Next, the S 2p spectra of the catalyst was shown in Fig. 5c. A well symmetrical S 2p peak at binding energy 169.5 eV was observed, corresponding to S 6+ state of SO 4 2− species 31,32 . The binding energy at 164.6 eV was attributed to elemental sulfur 33 , which was overlapped with Bi 4f XPS spectra. Hence, the SO 4 2− species were mainly adsorbed on the catalyst surface, which improved surface acid strength and favored to the adsorption and degradation of the pollutants 22,34 . The Bi 4f XPS spectrum was shown in Fig. 5d. The peak at binding energy 159.5 eV was attributed to Bi 4f 7/2 , while the peak at binding energy 164.7 eV belonged to Bi 4f 5/2 . These values of binding energies were higher than the binding energy of Bi 3+ , indicating that bismuth existed in Bi 3+δ state and formed Bi-O-Ti bond 16 . The intensities of Bi 4f  www.nature.com/scientificreports/ peaks of F-S-Bi-TiO 2 /SiO 2 were decreased as compared to F-S-Bi-TiO 2 , which indicate that dispersion function of silica was selective, that was only favorable for the substitution of F for the lattice oxygen but not for the formation Bi oxides. The weight percentages of the elements in the F-S-Bi-TiO 2 /SiO 2 and F-S-Bi-TiO 2 were shown in Table S2.
Photocatalytic purification of industrial acrylonitrile wastewater. On the basis of above analysis (Tables 1 and 2) we found that, there were a lot of inorganic and organic substances in the industrial acrylonitrile wastewater and these substances inhibited the catalyst activity (Fig. 1a). Specially, low ratio of BOD to COD Cr ( Table 2) identified that biochemical treatment could not be used. On the other hand, since there was limitation in amount of adsorption (Fig S1), the use of adsorption to treat was also hardly to reach regulation discharge. Hence, photocatalytic degradation of industrial acrylonitrile wastewater was one of most promising technique. A stable light source was required to keep experimental repeatability as mentioned in Table S1. Figure 6 shows the change in the value of COD Cr as a function of illumination time under simulated sunlight. It was found that COD Cr value decreased from 88.36 to 7.20 mgL −1 and TOC concentration decreased from 39.45 to 2.57 mgL −1 for 14 h irradiation. F-S-Bi-TiO 2 /SiO 2 catalyst was recycled through filtration after reaction and the recovery ratio was more than 95%. The photocatalytic reaction performance of recycled F-S-Bi-TiO 2 /SiO 2 was demonstrated through repeated rounds under the same reaction conditions. The value of COD Cr was 30.67 mgL −1 after 14 h of illumination at the second round. After photocatalysis, the value of COD Cr was 45.86 mgL −1 at the fourth round. Figure 6 shows the photocatalytic activity of fresh F-S-Bi-TiO 2 /SiO 2 catalyst for the degradation of pollutants in industrial acrylonitrile wastewater under natural sunlight irradiation. The value of COD Cr decreased to 39 mgL −1 after 6 h irradiation, which has satisfied China national discharge standards.

Discussion
The actual textile dyeing wastewater was effectively oxidized by TiO 2 under UV radiation. The degradation percentage of the dye and COD Cr were 98.50% and 91.50%, respectively 11 . Dai et al. 25 investigated the adsorption purification of acrylonitrile production wastewater by a microporous zeolite, CS-Z1 and a visible lightdriven Ti-β-Bi 2 O 3 photocatalyst. Nanoporous Ti-β-Bi 2 O 3 was prepared via a solvothermal synthesis method in laboratory. Dai et al. was mainly interested in the theory, but not in application. They have not considered the inorganic ions effects. In this paper, we used industrial acrylonitrile wastewater as testing samples. The industrial acrylonitrile wastewater was pretreated by microporous zeolite, HZSM-5. After the treatment, the wastewater contained some inorganic and organic matter, as shown in Tables 1 and 2. Some of them were same or different to the pollutants reported by Dai et al. Table 2 only gives out a part of interfering inorganic ions And the existence of these ions could slow down the reaction rate (Fig. 1a). Our results showed that the prepared catalyst possesses certain anti-interfering ability. From the application point of view, we have used simple SiO 2 dispersing sol-gel for the synthesis of highly active ultrafine nanoparticle catalyst with certain anti-ion interfering function and for the first time we successfully degraded industrial acrylonitrile wastewater in only 6 h under natural sunlight. This is the most significant findings for the photocatalysis application in environment.
For most of the pollutants, close to zero discharge is a final goal that people pursue in environment protection. Sixto et al. 10 have demonstrated that photocatalytic purification of phenol containing wastewater by TiO 2 -P25 using sacrificial agent under ultraviolet part of sunlight. In this study, there was not any sacrificial agent, however, we still realized that the values of COD Cr and TOC near zero discharge. These results show the potential of as prepared photocatalyst in near zero discharge and high TOC removal efficiency of wastewater treatment. www.nature.com/scientificreports/ In our previous study, we found that the reaction rate of the catalyst with SiO 2 was several times faster than without SiO 2 . The later exhibited some general properties of a large bulk aggregate. On the other hand, there were a majority of ultrafine nanoparticles with acidic sites on the catalyst with SiO 2 20 . A large number of bumps increased the quantum size effect and UV absorption at 270-380 nm, which should raise activity. Several nanometers package of irregular crystal films have promotion effect on light absorption. Photon were generated by light which can go through several nanometers of irregular Bi 4 Ti 3 O 12 out layer film and get into TiO 2 (Fig. 2e), causes scattering at the rough interface of the both phases. Apparently, it should be favorable to light absorption and to promote the catalytic activity (Fig. 6). The bumps were different with films of packages in morphology, but we think that were consistent in functions, because their sizes were in approximate ranges of several nanometers (Fig. 2e, g). Apparently, there were much more Fion in the lattice, which could produce holes 19 and also contribute the high activity of the catalyst with silica dispersion.

conclusion
We built up a most simple and low cost approach for the preparation of ultrafine nanocatalyst of F, S and Bi doped TiO 2 with SiO 2 dispersing sol-gel particles and successively degraded the organic pollutants in acrylonitrile industrial wastewater to reach national discharge standard under 6 h of natural sunlight irradiation. In the prepared photocatalyst the doping of F, S and Bi causes the enhanced absorbance in the visible region. The results of photocatalytic activity evaluation demonstrated that COD Cr value reached to discharge standard still after four recycle uses under the simulated sunlight irradiation, and to near zero discharge for the fresh photocatalyst. These results exhibited effectiveness and potential of our photocatalyst for the treatment of complicated and refractory industrial wastewater. The XPS and EDS analysis implied that S and Bi were doped successfully. The photocatalyst size distribution has been identified by visible nano-aggregates, constructed with 2-4 nm of finer gel particles. It indicated the function of SiO 2 dispersing to form ultrafine nanoparticles. The nano-aggregates might form an identical lattice faces or different faces when gel particles crystallized. TEM results revealed that, the bump's numbers may also be responsible for the increase in the light adsorption and photocatalytic activity. The identical face might originate from silk ribbon film of package. The crystals of several nanometers of out layer films and rough core surfaces of packages also increased the light absorption and enhance their photocatalytic activity.