Synthesis, Characterization of g-C 3 N 4 /SrTiO 3 Heterojunctions and Photocatalytic Activity for Organic Pollutants Degradation

: Perovskite-structure SrTiO 3 (STO) and graphitic carbon nitride (g-C 3 N 4 , CN) have attracted considerable attention in photocatalytic technology due to their unique properties, but also suffer from some drawbacks. The development of composite photocatalysts that combine properties of the individual semiconductors with enhanced charge separation is the current major trend in the photocatalysis ﬁeld. In this study, SrTiO 3 /g-C 3 N 4 (CNSTO) composites with different ratios (10, 20, 30, 40 and 50% g-C 3 N 4 ) were prepared with a sonication mixing method. The samples were characterized by X-ray diffraction (XRD), scanning electron microscopy (SEM), N 2 porosimetry, Fourrier transform infra-red spectroscopy (FT-IR), UV-Vis diffuse reﬂectance (DRS) and dynamic light scattering (DLS). STO spherical particles were successfully loaded on the g-C 3 N 4 planes forming heterojunction composite materials. The photocatalytic activity was tested against the degradation of methylene blue (MB) dye under simulated solar light (SSL) irradiation following ﬁrst-order kinetics. The photocatalytic activity followed the trend: 20CNSTO > 30CNSTO > 40CNSTO > 50CNSTO ≈ 10CNSTO, in accordance with the amount of • OH radicals determined by ﬂuorescence spectroscopy. A Z-scheme mechanism was proposed for the enhanced photocatalytic degradation of MB as evidenced by trapping experiments with scavengers. Finally, signiﬁcant stability and reusability was exhibited, indicating that such composites are of potential interest for photocatalytic treatments under sunlight irradiation.


Introduction
Photocatalysis with semiconductors is an advanced oxidation process for organic pollution abatement that has received great interest due to several advantages, such as the use of ambient conditions of temperature and pressure, the use of solar light, the absence of fouling, lack of mass transfer limitation, and the mineralization of organic pollutants into carbon dioxide, water and inorganic ions [1][2][3][4].
Titanate perovskites, ATiO 3 (A = Ca, Sr, Ba, etc.), are semiconductors with a wide band gap and interesting electronic, optical, magnetic and photocatalytic properties. They are considered promising materials for photocatalytic processes because of their strong resistance to photo corrosion, suitable oxidation potential and their high physicochemical stability [5][6][7][8]. Among them, SrTiO 3 is the most promising material for photocatalytic applications [9][10][11][12]. It is a cubic perovskite (Pm3m, a = 3.9 Å) n-type semiconductor with an indirect band gap of 3.1-3.7 eV depending on the crystal structure and

XRD Analysis
The XRD patterns (Figure 1) of all the prepared g-C 3 N 4 /SrTiO 3 photocatalysts were assigned to SrTiO 3 perovskite phase with cubic symmetry (JCPDS no. 79-0176). The main peaks, at about 32.4 • , 39.9 • , 46.4 • , 57.8 • , 67.8 • and 77.2 • , represent the SrTiO 3 (1 1 0), (1 1 1), (2 0 0), (2 1 1), (2 2 0) and (3 1 0) surfaces, respectively. The XRD pattern of g-C 3 N 4 with hexagonal symmetry is also presented in Figure 1 (JCPDS no. 87-1526). The weak peak at 13.1 • (110) and the strong one at 27.4 • (200) represent the g-C 3 N 4 surfaces. The sharp peaks in all patterns indicate that the obtained powders are highly crystalline, and that they have no impurities. The XRD data was performed by Rietveld refinement, as reported in [53]. The crystal size of all the materials was calculated by appropriate software, using a Williamson and Hull [54][55][56]-type plotting method, and ranged from 17.1 nm for the 10CNSTO to 29.0 nm for the 30CNSTO, as shown in Table 1. The refinement parameters of % crystal phase, cell parameters (a, b and c), strain analysis and R 2 are also presented in Table 1. The lattice constants (a, b, c) of SrTiO 3 in the composites are slightly higher than those of pure SrTiO 3 . It can be observed that the crystal size values of the materials present a small variation, increasing along with the % content of CN until 30CNSTO, and then they decrease. In the composites 40CNSTO and 50CNSTO, the peak at 27.4 • that corresponds to g-C 3 N 4 is hardly observed. This is because g-C 3 N 4 is partly exfoliated in the mixing process by sonication. As a result, lamellar structure of g-C 3 N 4 was formed and so the crystallization degree is limited. The same observation has been reported in other studies, too [57,58]. In the rest of the materials, the diffraction peaks did not change after the introduction of g-C 3 N 4 . Hydrodynamic particle size (median diameter) measurements were also performed by the dynamic light scattering (DLS) and the values ranged from 0.302 µm for the CN to 0.345 µm for the STO, indicating the formation of aggregates in aqueous solutions. c) of SrTiO3 in the composites are slightly higher than those of pure SrTiO3. It can be observed that the crystal size values of the materials present a small variation, increasing along with the % content of CN until 30CNSTO, and then they decrease. In the composites 40CNSTO and 50CNSTO, the peak at 27.4° that corresponds to g-C3N4 is hardly observed. This is because g-C3N4 is partly exfoliated in the mixing process by sonication. As a result, lamellar structure of g-C3N4 was formed and so the crystallization degree is limited. The same observation has been reported in other studies, too [57,58]. In the rest of the materials, the diffraction peaks did not change after the introduction of g-C3N4. Hydrodynamic particle size (median diameter) measurements were also performed by the dynamic light scattering (DLS) and the values ranged from 0.302 μm for the CN to 0.345 μm for the STO, indicating the formation of aggregates in aqueous solutions.   Representative nitrogen adsorption-desorption isotherms of the 10CNSTO and the 20CNSTO catalysts are presented in Figure 2. The photocatalysts are non-porous materials and the isotherms belong to type II, according to IUPAC classification [59]. Their specific surface areas (SSA) are 29.6 and 32.3 m 2 /g, respectively. Representative SEM images of CN, STO and 20CNSTO, 30CNSTO are shown in Figure 3. CN (Figure 3a) presented some sheet layers and sheet stacks with a smooth surface and irregular shape. The SEM image of STO (Figure 3b) clearly revealed spherical particles. The images of 20CNSTO and 30CNSTO (Figure 3c,d) showed that, after mixing with g-C 3 N 4 , the spherical particles of SrTiO 3 were deposited in the CN sheet-stacks. catalysts are presented in Figure 2. The photocatalysts are non-porous materials and the isotherms belong to type II, according to IUPAC classification [59]. Their specific surface areas (SSA) are 29.6 and 32.3 m 2 /g, respectively. Representative SEM images of CN, STO and 20CNSTO, 30CNSTO are shown in Figure 3. CN (Figure 3a) presented some sheet layers and sheet stacks with a smooth surface and irregular shape. The SEM image of STO ( Figure 3b) clearly revealed spherical particles. The images of 20CNSTO and 30CNSTO (Figure 3c,d) showed that, after mixing with g-C3N4, the spherical particles of SrTiO3 were deposited in the CN sheet-stacks. The

FT-IR Spectroscopy
The FT-IR spectra of the g-C3N4/SrTiO3 photocatalysts, CN and STO are presented in Figure 4. In all samples except CN, the shoulder below 1000 cm −1 appears because of the SrTiO3 crystal lattice vibrations [60]. Bands at around 858 and 596 cm −1 are caused by the stretching vibration of the Sr-O and Ti-O bonds, respectively [61]. The band at 1637 cm −1 for STO is due to the bending vibration of -OH (caused by bending water) [60,62]. The absorption peak at around 3443-3447 cm −1 could be caused by the stretching vibrations of lattice hydroxyls from Ti-OH, perturbed by nearby Sr atoms

FT-IR Spectroscopy
The FT-IR spectra of the g-C 3 N 4 /SrTiO 3 photocatalysts, CN and STO are presented in Figure 4. In all samples except CN, the shoulder below 1000 cm −1 appears because of the SrTiO 3 crystal lattice vibrations [60]. Bands at around 858 and 596 cm −1 are caused by the stretching vibration of the Sr-O and Ti-O bonds, respectively [61]. The band at 1637 cm −1 for STO is due to the bending vibration of -OH (caused by bending water) [60,62]. The absorption peak at around 3443-3447 cm −1 could be caused by the stretching vibrations of lattice hydroxyls from Ti-OH, perturbed by nearby Sr atoms or by Sr-OH [60]. For CN, the peak at around 815 cm −1 can be attributed to the s-triazine ring vibrations. The observed peaks in the range of 1253-1636 cm −1 can be ascribed to the stretching vibrations of aromatic C-N and C≡N in the heterocycles [45]. The peak at around 2173 cm −1 is assigned to cyano group stretch, which can be attributed to loss of ammonia [62]. The broad peak at 3180-3340 cm −1 can be ascribed to stretching vibration N-H or N=H from uncondensed amine groups [45]. The above-mentioned characteristic peaks of CN and STO are present in the composite materials. The observed shifts of characteristic peaks of g-C 3 N 4 in the range of 1253-1636 cm −1 for the composite materials indicate the weaker bond strengths of C=N and C-N, and reveal the existence of interactions between g-C 3 N 4 and STO. Similarly, shifts in STO characteristic peaks at 584 and 3432 cm −1 were observed. Transmittance(%)

UV-Vis Spectra
The diffuse reflectance spectroscopy (DRS) results for all photocatalysts are presented in Figure  5a. The absorption edge of pure SrTiO3 (STO) was about 390-395 nm, as expected, so no response to visible irradiation was observed. On the contrary, the absorption edge of g-C3N4 was about 445 nm, indicating visible light response. The Eg values (Table 2) of the photocatalysts were calculated with the use of Kubelka-Munk plots, which are presented in Figure 5b. The Kubelka -Munk plot of 20CNSTO is presented separately in Figure 5c. In both DRS and Kubelka-Munk plots of each photocatalyst, it can be observed that there are bands of both CN and STO. Also, it can be observed that, with the increment of CN in the composites, the band of STO decreases, while the band of CN increases. All the composite samples displayed a significantly enhanced visible light absorption compared to pristine STO.

UV-Vis Spectra
The diffuse reflectance spectroscopy (DRS) results for all photocatalysts are presented in Figure 5a. The absorption edge of pure SrTiO 3 (STO) was about 390-395 nm, as expected, so no response to visible irradiation was observed. On the contrary, the absorption edge of g-C 3 N 4 was about 445 nm, indicating visible light response. The E g values (Table 2) of the photocatalysts were calculated with the use of Kubelka-Munk plots, which are presented in Figure 5b. The Kubelka -Munk plot of 20CNSTO is presented separately in Figure 5c. In both DRS and Kubelka-Munk plots of each photocatalyst, it can be observed that there are bands of both CN and STO. Also, it can be observed that, with the increment of CN in the composites, the band of STO decreases, while the band of CN increases. All the composite samples displayed a significantly enhanced visible light absorption compared to pristine STO.
20CNSTO is presented separately in Figure 5c. In both DRS and Kubelka-Munk plots of each photocatalyst, it can be observed that there are bands of both CN and STO. Also, it can be observed that, with the increment of CN in the composites, the band of STO decreases, while the band of CN increases. All the composite samples displayed a significantly enhanced visible light absorption compared to pristine STO. Absorbance(a.u.) (2) (3) (4) (5)     The evolution of fluorescence spectra intensity of 2-hydroxyterephthalic acid (OHTA) for the 20CNSTO photocatalyst at different intervals within an irradiation time framework of 120 min is displayed in Figure 6 as a representative example. It can be seen that the fluorescence intensity increases along with irradiation time. The kinetics of • OH radicals formation for all photocatalysts are shown in Figure 7. The ability of the photocatalysts to generate • OH radicals follows the trend: 20CNSTO > 50CNSTO > 30CNSTO > 10CNSTO > 40CNSTO. The 20CNSTO material showed greater • OH formation ability compared to all other prepared composites, which is consistent with the photocatalytic kinetics described in the next paragraph. Thus, it was considered the optimum ratio for such composites. are shown in Figure 7. The ability of the photocatalysts to generate • OH radicals follows the trend: 20CNSTO > 50CNSTO > 30CNSTO > 10CNSTO > 40CNSTO. The 20CNSTO material showed greater • OH formation ability compared to all other prepared composites, which is consistent with the photocatalytic kinetics described in the next paragraph. Thus, it was considered the optimum ratio for such composites.

Photocatalytic Activity
The photocatalytic activity of all catalysts towards the degradation of MB under UV-Vis and visible irradiation is presented in Figure 8 and Figure 9, respectively. The degradation of MB in both cases followed first-order kinetics. As expected, the degradation kinetics under visible light irradiation was slower than under UV-Vis (simulated solar) irradiation. The apparent rate constants (k), the corresponding correlation coefficients (R 2 ) and half-lives (t1/2) of all photocatalysts are shown in Table 3. According to the determined apparent rate constants, the photocatalytic activity under both UV-Vis and visible irradiation had the following trend: 20CNSTO > 30CNSTO > 40CNSTO > 50CNSTO ≈ 10CNSTO. The highest photocatalytic activity of the 20CNSTO catalyst is also verified by its ability to form • OH radicals. The corresponding apparent degradation rate constants trends

Photocatalytic Activity
The photocatalytic activity of all catalysts towards the degradation of MB under UV-Vis and visible irradiation is presented in Figures 8 and 9, respectively. The degradation of MB in both cases followed first-order kinetics. As expected, the degradation kinetics under visible light irradiation was slower than under UV-Vis (simulated solar) irradiation. The apparent rate constants (k), the corresponding correlation coefficients (R 2 ) and half-lives (t 1/2 ) of all photocatalysts are shown in Table 3. According to the determined apparent rate constants, the photocatalytic activity under both UV-Vis and visible irradiation had the following trend: 20CNSTO > 30CNSTO > 40CNSTO > 50CNSTO ≈ 10CNSTO. The highest photocatalytic activity of the 20CNSTO catalyst is also verified by its ability to form • OH radicals. The corresponding apparent degradation rate constants trends for UV-Vis and visible light irradiation are presented in Figures 8 and 9. The initial increase in the degradation efficiency can be attributed to the increment of CN content which benefits charge transfer in the materials interface and causes greater response into the visible light region. The greater increment of g-C 3 N 4 amount, though, decreases the effective heterointerfaces in the composites, which is unfavorable for the charge transfer [46,63] while the entrained decrease of STO, which has a higher oxidation potential valence band led to the decrease of • OH radicals formation.
Finally, the stability of the best catalyst (20CNSTO) was investigated for three consecutive photocatalytic cycles (Figure 10). The photocatalyst presented quite stable photocatalytic activity among the repeated cycles and about 95% of the initial degradation efficiency was maintained. This fact suggests that the 20CNSTO photocatalyst has good reusability. The slight decrease in the photocatalytic efficiencies could occur because of the accumulation of later stage products into the catalyst surface after the first catalytic cycle or because of small losses of catalyst during the recovery procedure due to the good dispersibility in the aqueous solution.
photocatalytic cycles (Figure 10). The photocatalyst presented quite stable photocatalytic activity among the repeated cycles and about 95% of the initial degradation efficiency was maintained. This fact suggests that the 20CNSTO photocatalyst has good reusability. The slight decrease in the photocatalytic efficiencies could occur because of the accumulation of later stage products into the catalyst surface after the first catalytic cycle or because of small losses of catalyst during the recovery procedure due to the good dispersibility in the aqueous solution.  among the repeated cycles and about 95% of the initial degradation efficiency was maintained. This fact suggests that the 20CNSTO photocatalyst has good reusability. The slight decrease in the photocatalytic efficiencies could occur because of the accumulation of later stage products into the catalyst surface after the first catalytic cycle or because of small losses of catalyst during the recovery procedure due to the good dispersibility in the aqueous solution.

Mechanism Analysis
Generally, superoxide radicals (O2 •− ), hydroxyl radicals (OH • ) and the photogenerated holes (h + ) have important role in the photocatalytic process. To propose the proper photocatalytic mechanism for the activity of the photocatalyst, trapping experiments took place with the use of scavengers. The used scavengers were isopropanol (IPA), formic acid (FA), N2, superoxide dismutase (SODred), acetonitrile /N2, and sodium azide (NaN3) as OH • , h + , O2 •− , OH • /O2 •− and (OH • + 1 O2) scavengers, respectively. The apparent rate constants (k) and the corresponding correlation coefficients (R 2 ) of 20CNSTO under the effect of each scavenger are presented in Table 4. The redox potentials of g-C3N4 (conduction band, CB = −1.4 eV vs. NHE (Normal Hydrogen Electrode), valence band VB = +1.3 eV vs. NHE) are more negative than those of SrTiO3 (CB = −0.2 eV vs. NHE, VB = +3.0 eV vs. NHE) [52,63]. Based on the experimental results that are presented in Table 4, the classical Type II photocatalytic mechanism, i.e., the migration of electrons from the CB of CN to the CB of STO with concurrently migration of holes from VB of STO to VB of CN, is excluded. More specifically, in such mechanisms, the accumulated holes in the VB of g-C3N4 couldn't produce • OH from the oxidation-adsorbed water molecules or OHions because the VB potential was less positive than the redox potential E 0 (OH•/H2O) (+2.68 eV vs. NHE) and E 0 (OH-/•OH) (+1.99 eV vs. NHE). However, according to the experimental results, • OH are indeed produced, as evidenced by OHTA fluorescence, as well as by the degradation rate decrease in the presence of IPA, NaN3 and   Table 4, the classical Type II photocatalytic mechanism, i.e., the migration of electrons from the CB of CN to the CB of STO with concurrently migration of holes from VB of STO to VB of CN , is excluded. More specifically, in such mechanisms, the accumulated holes in the VB of g-C 3 N 4 couldn't produce • OH from the oxidation-adsorbed water molecules or OHions because the VB potential was less positive than the redox potential E 0 (OH•/H2O) (+2.68 eV vs. NHE) and E 0 (OH − /•OH) (+1.99 eV vs. NHE). However, according to the experimental results, • OH are indeed produced, as evidenced by OHTA fluorescence, as well as by the degradation rate decrease in the presence of IPA, NaN 3 and acetonitrile. In addition, the molecular O 2 could also not be photoreducted to • OH in the CB of SrTiO 3 , as the CB potential was more positive than the redox potential E 0 in the presence of acetonitrile/N 2 as a scavenger showed the oxidation of MB by the holes in the VB of SrTiO 3 . As a result, a z-scheme mechanism is proposed for the photocatalytic activity of the composites. In the z-scheme mechanism, the photogenerated electrons in the CB of the STO will combine with the photogenerated holes in the VB of CN, while the accumulated holes the with high oxidation potential in the VB of STO and electrons with high reductive potential in the CB of CN could easily produce • OH and O 2 •− , which participate in the oxidative degradation of MB. In conclusion, the results of the trapping experiments showed that OH • and O 2 •− were the major oxidant species, followed by a minor contribution of 1 O 2 , and the generation of such species can only be rationalized by a z-scheme mechanism.

Materials and Chemicals
Urea (

Preparation of g-C 3 N 4 and g-C 3 N 4 /SrTiO 3 Heterojunctions
For g-C 3 N 4 synthesis, urea was preheated at 80 • C for 24 h in an alumina crucible and then calcined at 500 • C for 4 h with the heating rate of 10 • C /min [64,65]. The composite photocatalysts were prepared by a sonication mixing method. Appropriate stoichiometric amount of SrTiO 3 and g-C 3 N 4 were suspended in double distilled water separately under sonication for 1 h. Then both solutions were mixed, and the whole solution was treated again under sonication for 90 min (Hielscher UP100H Teltow, Germany ultrasonic processor, Amplitude 85%). The as-prepared samples contained different g-C 3 N 4 to SrTiO 3 amounts and had the following %wt content and code names (in brackets): 10% (10CNSTO), 20% (20CNSTO), 30% (30CNSTO), 40% (40CNSTO), 50% (50CNSTO) [57]. Also, for the comparison for the heterojunction oxides, g-C 3 N 4 which had been sonicated for 1 h (CN) was also used.

Texture Characterization of the Heterojunctions
Crystallinity and phase identification of the photocatalytic materials were defined by powder X -ray diffraction (XRD) using a Bruker Advance D8 XRD instrument (Billerica, MA, USA), which generates monochromated Cu K a (λ = 1.5418 Å) radiation with a continuous scanning rate in the range of 10 < 2θ < 90 in steps of 0.02 • and rate 0.01 • θ/sec. The patterns were determined with the use of the Joint Committee on Powder Diffraction Standards (JCPDS) database. The results were studied with Rietvield refinement by a suitable computer program.
The N 2 adsorption-desorption isotherms at 77K were obtained by porosimetry using a Quantachrome Autosorb -1 instrument (Bounton Beach, FL, USA). All samples (≈ 0.1 g) were degassed for 4h at 353 K for the elimination of any moisture and condensed volatiles. Brauner-Emmet-Teller (BET) method at relative pressure between 0.05-0.3 was used for the calculation of the specific surface area (SSA). The morphology of the photocatalysts was observed by scanning electron microscopy (SEM) by a JEOL JSM 5600 instrument (Tokyo, Japan).
Particle size measurements were carried out after 10 min of sonication with a Shimadzu SALD-2300 (Kyoto, Japan) laser diffraction particle size analyzer in dynamic light scattering (DLS) mode. The point-of-zero charge (PZC) of the materials was measured by the mass titration method, as reported elsewhere [66].

Fourier Transform. Infrared Spectroscopy (FT-IR)
The chemical structure of all heterojunctions was recorded by Fourier transform infrared spectroscopic (FT-IR) analysis. The analysis was carried out with an instrument by Thermo Scientific (Nicolet iS5) (Waltham, MA, USA). Spectral grade KBr (≥99%, Sigma-Aldrich St. Louis, MO, USA) was used as a reference. The materials were ground with KBr in 1:3 ratio and made into pellets using an appropriate hydraulic press. The pellet was scanned at 0.964 cm −1 in the range 4000-400 cm −1 .

UV-Vis.-Diffuse Reflectance Measurements
The absorbance spectra of the g-C 3 N 4 /SrTiO 3 heterojunctions were obtained by a Shimadzu 2600 (Kyoto, Japan) spectrophotometer which was equipped with an ISR-2600 (Kyoto, Japan) integrating sphere at room temperature with BaSO 4 as reference sample in the range of 200-800 nm.

Photocatalytic Experiments and Analytical Methods
The photocatalytic experiments were conducted with Suntest XLS+ apparatus (Atlas Linsengericht, Germany) under UV-Vis irradiation (simulated solar light, λ > 290 nm). A xenon lamp (2.2 kW), jacked with special 290 nm cut-off glass filter, was the light source. During the experiments, the irradiation intensity was maintained at 500 W m −2 . Experiments under visible light irradiation (λ > 400 nm) were performed using LED flood lamps (LG SMD, LED, 45 pcs, Seoul, Korea) 2 × 50 Wm −2 . The photocatalytic activity was tested against the degradation of methylene blue (MB).
For both UV-Vis and visible irradiation experiments, the photocatalysts were suspended in double distilled water by sonication for 10 min then transferred in an appropriate Pyrex glass reactor (250 mL) and stirred using a magnetic stirrer. An initial concentration of 5 mg L −1 of MB and 200 mg L −1 of photocatalyst was used in all experiments. Prior to irradiation, the suspension is magnetically stirred for 30 min in the dark to ensure substrate adsorption/desorption equilibrium established on the catalyst surface. During irradiation the temperature was kept at 23 ± 1 • C by water circulation in the jacket of the reactor and air-circulation.

Determination of • OH Radicals by Fluorescence Measurements
TA was used as a probe for the determination of hydroxyl radical formation rate. Aqueous solution contained of NaOH and TA (5 × 10 −4 M) was prepared and then 20 mg of photocatalyst powder was suspended in the photocatalytic reactor and stirred for 30 min under UV-Vis irradiation. The irradiation conditions are described in the next paragraph. Portions of 5 mL of the suspension were collected at different time intervals and filtered with 0.45 µm membrane filter. A Shimadzu RF-5300PC (Kyoto, Japan) fluorescence spectrophotometer was used in order to measure the intensity of the fluorescence peak at 425 nm with 310 nm excitation, which is attributed to 2-hydroxyterephthalic acid (OHTA), and it is known to be proportional to the amount of • OH radicals produced. The concentration of • OH was calculated by a calibration curve plotting the fluorescence intensity of standard OHTA (TCI, > 98% TCI, Tokyo Chemical Industry, Tokyo, Japan) solutions.

Conclusions
Visible-light active SrTiO 3 /g-C 3 N 4 photocatalysts have been synthesized by a sonication mixing method. The optimum fortification level of g-C 3 N 4 loading was 20%. All the prepared catalysts presented photocatalytic performance towards the decolorization of MB. Among them, the 20CNSTO material showed the best photocatalytic activity for the degradation of MB in both UV-Vis and visible