Chelated Nitrogen-Sulphur-Codoped TiO2: Synthesis, Characterization, Mechanistic, and UV/Visible Photocatalytic Studies

This study presents in detail the physicochemical, photoluminescent, and photocatalytic properties of carboxylic acid chelated nitrogen-sulphur-codoped TiO2. From the Fourier transform infrared spectroscopic study, it was revealed that the formate group formed bidentate bridging linkage while the acetate group coordinated in a bidentate chelating mode with a titanium precursor. In compliance with X-ray diffraction data, the anatase to rutile transformation temperature was extended due to carboxylic acid chelation and NS codoping. Raman analysis indicated four Raman peaks at 146, 392, 512, and 632 cm for the precalcined chelated TiO2; on incorporation with NS dopants, an increase in Raman intensity for these peaks was recorded, indicating the structure stability of the anatase phase. Furthermore, X-ray photoelectron spectroscopic study revealed the presence of anionic doping of nitrogen and cationic doping of sulphur in the lattice of TiO2. When evaluating the UV-visible photodegradation rate of 4-chlorophenol, the modified TiO2 (NS0.06-TFA) showed the highest photocatalytic activity. In connection with the activity tests, several scavenger agents were employed to elucidate the significance of the different reactive oxidizing species during the photocatalytic process. Moreover, the transfer pathways of photogenerated carriers and the photocatalytic reaction mechanism of modified TiO2 were also explained in detail.


Introduction
Titanium dioxide (TiO 2 ) is an effective photocatalyst, which is widely studied for the degradation of organic pollutants in water after the discovery of the TiO 2 -photoassisted electrochemical splitting of water reported by Fujishima and Honda in 1972 [1]. TiO 2 exists in nature in three polymorphic forms (anatase, rutile, and brookite), which differ only in the arrangement of their TiO 6 octahedra; anatase (tetragonal) consists of octahedral shearing vertices; rutile (tetragonal) is connected by edges; and in brookite (orthorhombic), both edges and vertices are connected. Thermodynamically, rutile is the most stable phase while both anatase and brookite are metastable, transferring to rutile under heat treatment, typically at temperature ranging between 600 and 700°C [2]. Among the TiO 2 crystalline forms, anatase and rutile are the most widely investigated polymorphs, with anatase generally found to be more photocatalytically active than brookite and rutile, but some reports illustrate the synergistic effect found in the composite mixture of anatase-brookite and anatase-rutile to be responsible for the enhanced photocatalytic activity [3,4]. However, because of the large band gap, anatase (3.20 eV) begins to absorb UV light around 388.5 nm, and brookite (3.14 eV) absorbs at around 395 nm, whereas the absorption onset of rutile (3.02 eV) occurs around 411 nm [5]. Due to the UV absorption of TiO 2 polymorphs, their usage under visible light irradiation is greatly restricted. Therefore, research studies begin to develop and synthesize TiO 2 that can absorb and make use of both UV (290-400 nm) and visible (400-700 nm) light to enhance photocatalytic process efficiencies.
Many researchers have focused on modifying TiO 2 with nonmetal nitrogen to be active in visible light irradiation; however, literature reports a different explanation for the decrease in N-TiO 2 band gap. For example, work by Asahi et al. [6] demonstrated that substitutional presence of nitrogen in the crystal lattice narrows the TiO 2 band gap through N2p orbital mixing with O2p orbital. However, Ihara et al. [7] suggested that addition of nitrogen only increases the stabilization of oxygen vacancies and as a consequence helps to improve the visible light activity of N-TiO 2 . While, work by Irie et al. [8] showed that nitrogen atoms can be substituted for the lattice oxygen sites to form an isolated impurity energy level (N2p) above the valence band (O2p) of TiO 2 , and under visible light, electrons are excited to the nitrogen impurity energy level. Many other research efforts have also been made with F, S, I, B, Br, C, Cl, and P to develop anionic-doped TiO 2 with improved visible light activity [2,9], and among these nonmetal dopants, sulphur doping in TiO 2 has received much attention due to its anionic (sulfide, S 2− ) and cationic (hexavalent (S 6+ ) and tetravalent (S 4+ )) characteristics. However, literature showed controversial reports regarding the photocatalytic activity of cationic and anionic doping of sulphur in TiO 2 . For example, Ohno et al. [10] reported that cationic S-doped TiO 2 showed better visible light absorption and photocatalytic activity compared to anionic S-doped TiO 2 ; in contrast, Szatmary et al. [11] reported low UV photocatalytic activity of cationic S-doped TiO 2 compared to Degussa P25. Chaudhuri and Paria [12] reported enhanced visible photocatalytic activity of anionic Sdoped TiO 2 , whereas the work of Rockafellow et al. [13] showed that sulphur atoms may act as deep hole trappers reducing the oxidizing power of the holes and hindering the generation of hydroxyl radicals. In spite of the extensive reports on cationic and anionic doping of TiO 2 , the visible photocatalytic activity has remained quite low, because monodoping can generate recombination centers for the photoinduced electrons and holes inside the TiO 2 [14]. To overcome the problem of electron-hole recombination, researchers tried codoping of TiO 2 with both metals and nonmetals, and the results showed enhanced electron-hole separation and increased absorption of photons in the visible range leading to enhanced photocatalytic activity of codoped TiO 2 against the degradation of aqueous organic compounds [15,16]. TiO 2 synthesis by the sol gel method involves very rapid hydrolysis of titanium alkoxide with water due to their susceptibility to nucleophilic attack. The rate of hydrolysis and condensation reactions which govern TiO 2 particle size, phase formation, and their transformation may be controlled by using chelating agents such as acetyl acetone, diols, acetic acid, urea, ammonium sulfate, and formic acid [17,18]. The alkoxide stability improves control over the reaction conditions; recently, it was concluded that in the presence of a silver dopant, the chelation of formic acid with titanium was weakened which favors earlier phase transformation [2]. In contrast, in the presence of vanadium ions, acetic acid favors bidentate chelating mode of linkage with titanium ions, and as a result, the transformation of anatase to rutile is suppressed [19]. In connection to such studies, it is hard to find a systematic study reporting the effect of nitrogensulphur codoping on carboxylic acid chelation, anatase to rutile transformation (ART), and UV-vis photocatalytic activity against the degradation of aqueous organic compounds; such work would be both scientifically and practically important. Therefore, the aim of the present work is to achieve the following objectives: (a) to develop a rapid, convenient, and simple preparation method for the synthesis of the photocatalyst; (b) to explain in a systematic way the binding mode behaviour of the two carboxylic acids such as formic and acetic acid with the alkoxide titanium precursor and the resulting effect on anatase to rutile transformation; and (c) to study the effect of an increase in molar concentration (mol%) of nitrogen-sulphur (NS) dopants on the physicochemical and UV-vis photocatalytic properties and on the hindrance of anatase to rutile transformation in carboxylic acid-modified TiO 2 . Finally, to elucidate a systematic and comprehensive mechanism to discuss the transfer pathways of UV-vis-photoinduced charge carriers in regard to explaining the decrease in electron-hole recombination rate and enhanced activity of the codoped powders against the degradation of 4-chlorophenol (4-CP) as a model pollutant.
In the typical experimental procedure for the synthesis of control TiO 2 , a molar ratio of 1 : 3 : 2 : 18 among titanium isopropoxide (TTIP), formic acid (CH 2 O 2 ), anhydrous ethyl alcohol (C 2 H 6 O), and water (H 2 O) was maintained. An appropriate volume of titanium isopropoxide (18 ml) was mixed with formic acid (7 ml) under strong magnetic stirring; for this solution, anhydrous ethyl alcohol (7 ml) was added dropwise under continuous stirring followed by the addition of water (18 ml). In the synthesis of nitrogensulphur-codoped TiO 2 , measured amount of thiourea as dopant was added to water to form a uniform solution before added to the flask containing the mixed solution of formic acid, titanium isopropoxide, and anhydrous ethyl alcohol. Instantaneous hydrolysis reaction occurs in the reacting mixture due to the addition of water (with or without dopants) which resulted into a milky sol of hydrated titanium hydroxide. The sol solution was kept on constant magnetic stirring for 2 h and then aged for gelation for 5 h at room temperature followed by drying in oven at 85°C for 24 h. Finally, calcination was accomplished by firing the dried powders at 700°C for 1 h in air to form crystalline TiO 2 nanoparticles.

2
International Journal of Photoenergy The dopant, thiourea, was added in the concentrations of 0, 0.02, 0.04, 0.06, 0.08, and 0.1 mol%, and the final sample powders were referred as T FA and (NS) x -T FA , where FA is formic acid, NS is nitrogen-sulphur dopants, and x is the dopant content (mol%). Similarly, control and nitrogen-sulphur-codoped catalyst powders were prepared following the same experimental procedure except using another acid catalyst; acetic acid and the final photocatalyst powders were labeled as T AA and (NS) x -T AA where AA is acetic acid. In addition, to compare the chelating effect of the carboxylic acids, control TiO 2 was also prepared by adding the required amount of water dropwise to the continuously magnetic stirred solution of titanium isopropoxide and ethyl alcohol. This photocatalyst powder was denoted by T control .

Characterization
Details. X-ray diffraction (XRD) patterns of control and doped TiO 2 powders were recorded with Phillips PW 1710 diffractometer using monochromatic high intensity Cu Kα radiations (λ = 0.15418 nm) at a scanning rate of 0.03°/sec and in the scanning range from 10 to 80°. The crystallite size (D) was estimated using the Scherrer equation (1), and the rutile phase content was calculated using (2) [20]. The peak intensity of [101] of anatase and [110] of rutile reflections located at 25.4°and 27.4°on 2θ (degrees) x-axis was used.
where k is a constant equal to 0.9, λ equal to 1.5418 Å is the X-ray wave length of Cu, β is full width at half maximum (FWHM) of the peak, and θ is the Bragg's angle in radians.

Rutile phase % =
where A A and A R represent the integrated intensities of anatase and rutile peaks. High-resolution transmission electron microscope (HRTEM) images of the prepared samples were observed with Phillips CM200 TEM operated at 200 kV. Infrared spectra (IR) was recorded on a Bruker Tensor 27 FTIR with OPUS data collection program (V 1.1); the energy range of 5000-400 cm −1 was used to evaluate the carboxylatebinding mode and the presence of different bonding vibrational frequencies; the precalcined and calcined catalyst powder was mixed with the reference potassium bromide (KBr). The thermogravimetric analysis was carried out with an apparatus (TGA-7) from Perkin Elmer (Norwalk, CT, USA) in the temperature range from 20°C to 800°C at a heating rate of 10°C·min −1 ; the experiments were carried out on 15-20 mg specimen under nitrogen atmosphere (40 ml/min). Raman spectra were acquired at room temperature using a SENTERRA confocal Raman microscope (Bruker, US) equipped with a 532 nm laser. Room temperature photoluminescence spectrum (PL) was recorded with FluroMax-2 spectrofluorometer at an excitation wavelength of 310 and 410 nm. Carry 5000 NIR UV-vis spectrophotometer (Varian Inc., US) with attached Praying Mantis™ diffuse reflection (DRIFTS) accessory was used to record absorption and diffuse reflectance spectra (DRS) in the range 190-800 nm; catalyst powder was mixed in KBr and pressed into a tablet; a KBr tablet made under the same conditions was used as a reference. The oxidation state and elemental compositions of the doping metals were analyzed using X-ray photoelectron spectroscopy (XPS, Thermo Scientific K-alpha, USA) using an X-ray source of AlKα (1486.6 eV). A survey spectrum was first recorded to identify all the elements present in the powder using a spot size of 300 μm and step size of 1.00 eV, followed by recording high-resolution spectra of the identified elements using a step size of 0.1 eV. Data spectrum processing was performed using the Avantage (Thermo Scientific) software.
The room temperature photocatalytic activities of the prepared catalyst powder were investigated for the degradation of 4-chlorophenol (4-CP) as a model pollutant. The cylindrical reactor contained 2 l aqueous solution of 0.007 g·l −1 of 4-CP and 0.1 g·l −1 of the prepared powder. A single UV light source (max. output at 254 nm, TUV 11 W 4P-SE, UV-Technik Speziallampen GmbH, Germany) protected in a quartz sleeve was placed vertically at the center point inside the reactor. The reactor solution under constant magnetic stirring was bubbled continuously with purified external air throughout the experiment. UV illumination was supplied for one-hour experimental duration. A sample of 5 ml was withdrawn at regular interval; before analysis, the sample suspension was filtered to remove the photocatalyst particles. The photocatalytic degradation of 4-CP was determined by performing the HPLC analysis procedure as stated elsewhere [21].
In the visible photocatalytic experiments, the reactor suspension was 1 l, and the concentration of 4-CP and the photocatalyst powder was kept the same as used in the UV experiments. The reactor solution was irradiated with visible light for 12 min experimental duration using a single 180 W Hg-medium pressure lamp (λ = 420-660 nm, intensity 100 mW/cm 2 ) supplied by Heraeus Nobel Light, Germany; protected in a quartz sleeve and mounted vertically at the center point inside the reactor, a UV blocking film (UVPS, USA) was used to eliminate any UV radiations in the emitted light. Moreover, the temperature of reactor suspension was maintained in the range of 25-30°C with cold circulating water. The pH of the reactor suspension was measured with Fisher AccumetTMAB15 pH meter equipped with a glass pH electrode, and a constant pH of 5.5-5.7 was measured throughout the photocatalytic activity experiments.

Results and Discussion
3.1. X-Ray Diffraction (XRD). Figure 1(a) shows the XRD patterns of control TiO 2 prepared in the absence and in the presence of carboxylic acids (acetic acid and formic acid) and calcined at 700°C. Rutile (99.0 wt.%) was the main phase in T control powder while T FA powder contained anatase (11.0 wt.%) and rutile (89.0 wt.%) phases, and T AA powder was composed solely of the anatase phase. Figure 1 Table 1 illustrates that the    [22] stating that addition of nitrogen increases the rutile percentage.
Further studying the effect of doping, crystal lattice distortion was observed as indicated from the change in anatase lattice parameter "c" (Table 1) meaning that variation in interatomic distances has occurred [23]. Thus, on doping, the Ti-O bonds will break and may reduce in numbers, implying that the number of octahedron molecules for rutile structure formation decreases, and as a result, rutile phase formation gets restrained. Moreover, anatase to rutile transformation (ART) is strongly particle size dependent [20]. Table 1 illustrates the crystallite sizes of the prepared powders; compared to control TiO 2 (T control , T FA , and T AA ), an increased inclusion of the dopants in TiO 2 lattice causes a decrease in the crystallite size of the samples ((NS) x -T FA ) employing formic acid. However, in acetic acid-modified samples ((NS) x -T AA ), no observable change in the crystallite size was noticed. This result indicates that NS codoping restrains the crystallite growth of codoped TiO 2 , so does the phase transformation. Defect concentration and grain boundary concentration can also govern phase transformation [24]; the influence of such effects increases with an increase in crystallite size via decrease in surface area. In addition to the restrained growth of the codoped TiO 2 crystallites, the similarity in the ionic radii of nitrogen (N 3− , 0.15 nm) to oxygen (O 2− , 0.14 nm) and that of sulphur (S 6+ , 0.03 nm and S 4+ , 0.04 nm) to titanium (Ti 4+ , 0.07 nm) further illustrates that nitrogen and sulphur dopants can substitute into the TiO 2 lattice, which minimizes the chance of NS migration to the anatase grain surface on calcination. As a consequence, surface defects in the anatase grain will decrease, which favors the decrease in anatase to rutile phase transformation.   [25] determined the mode of binding of the carboxylate group to the metal atom by calculating the difference (Δ) in the values of carboxylate stretches, Δ = υ(COO − ) asym − υ(COO − ) sym . From the spectra (Figure 3(a)), Δ = 216 cm −1 indicating that the formate group is bound with the titanium precursor by a bidentate bridging mode; the proposed scheme I is shown in Figure 4. The IR spectra of the samples (T AA , (NS) 0.06 -T AA , and (NS) 0.1 -T AA ) employing acetic acid in the synthesis procedure are shown in Figure 3(b); the T AA sample showed a peak at 1712 cm −1 assigned to the stretching vibration of υ(C-O) of the COOH group, and the peak at 1265 cm −1 is attributed to the symmetric bending vibration of the CH 2 group or to the bending vibration of the CH 3 group. Both of these peaks were weakened in the NS-codoped samples illustrating the inclusion of the dopant into the crystal lattice, while the weak shoulder at 1020 cm −1 may indicate the presence of carbonyl moieties coming from isopropoxide residues. The carboxylate peaks appear at 1536 and 1436 cm −1 which represent asymmetric carboxylate stretching (υ(COO − ) asym ) and symmetric carboxylate stretching (υ(COO − ) sym ), respectively [26,27]. The difference between these two carboxylate stretches gives the value of Δ = 100 cm −1 , illustrating that the acetate group is bound to titanium atoms through bidentate chelating modes, and the proposed scheme II is shown in Figure 4. Moreover, the weak shoulder at 1620 cm −1 in the samples ((NS) 0.06 -T AA and (NS) 0.1 -T AA ) is attributed to the O-H bending vibration mode showing that NS-codoped powders are more susceptible to OH groups.
The above analysis shows that under the same synthesis conditions and using the same molar (mol%) amount of carboxylic acid, formic acid, and acetic acid formed different types of binding linkages with the titanium precursor, and this is the reason why the samples behave in different ways to the phase transformation. It is obvious that titanium alkoxide modification with chelating agents (formic and acetic acid) could promote anatase phase at increased temperature. In compliance with the XRD data (Figure 1(a)), T AA powder is pure anatase, while T FA powder is composed of both anatase and rutile phases, demonstrating that the chelating structure of the acetates promotes the formation of polymeric chains with extensive cross-linking in comparison to the bridging structure of the formate group. These chains remain stable throughout much of the condensation process; as a consequence, anatase-rutile transformation temperature is extended. In contrast, in the absence of chelating acids, weak gel networks are formed in the powder T control ; only the addition of water causes increased hydrolysis which results in the formation of larger particles that thermodynamically favor phase transformation to rutile [2]. The observed results were further supported by the thermal analysis of the precalcined powders as explained also illustrate that no observable change in the intensity of the COO − stretches were recorded with the addition and an increase in the dopant content, which means that the NS dopants assist in maintaining a strong TiO 2 oligomer network. In support, a peak at 2060 cm −1 in the samples ((NS) 0.1 -T FA and (NS) 0.1 -T AA ) was developed, which is assumed to represent CN bonds [28]. Figure S2 (supplementary information) shows the IR spectra of samples (T control , T FA , (NS) 0.08 -T FA , T AA , and (NS) 0.08 -T AA ) calcined at 700°C. All the samples maintained the broad peak at 3178 cm −1 attributed to OH groups, and the OH peak intensity increases due to carboxylic acid chelation and NS codoping which is a positive sign for enhanced photocatalysis. It is obvious that these hydroxyl ions are main hole scavengers resulting in the formation of hydroxyl radicals required for the degradation of organic pollutants.    for the samples T FA and T AA where either formic acid or acetic acid was employed in the synthesis process as a chelating agent. As discussed earlier in the FTIR study, the use of NS dopants and an increase in their concentration do not induce a reduction in the intensity of carboxylate stretches. This indicates that titania formate or acetate complex is becoming stronger and results in the formation of a structurally strong oligomer that upon calcination easily formed a crystalline anatase TiO 2 . This result is supported by the Raman analysis; evidently, it can be seen that the intensities of the respective anatase of four Raman peaks are increased and their widths are broadened after NS codoping. This shows that the stability of anatase is promoted and, in addition, after calcination, will result into well crystalline anatase particles with decreased particle size. To further elaborate the effect of the addition of dopants, the insert in Figures 5(a) and 5(b) shows the Raman band shift and broadening in the E g mode (146 cm −1 ) of precalcined (NS) x -T FA and (NS) x -T AA (where x = 0 06 and 0.1 mol%) powders compared to T control , T FA , and T AA samples, respectively. The anomalous variation in the intensities of the Raman band, shifting towards lower wavenumbers and broadening, obviously showed the effect of NS dopants indicating that dopants effect the titanium carboxylate complex before crystallization has occurred. This also assured that nitrogen substitutes for oxygen atoms and sulphur substitutes for titanium atoms in the titania lattice. In comparison to (NS) x -T FA , the Raman spectra of (NS) x -T AA (where x = 0, 0.06, and 0.1 mol%) powders ( Figure 5(b)) showed a peak in the range 2800-3000 cm −1 ascribed to OH stretches [24], whose intensity increases with NS amount. This result is supported by the FTIR analysis; as shown in Figure 3(b), a broad band at 2800-3600 cm −1 is ascribed to the OH groups. This result further supports the formation of different modes of acetate and formate bonding with titanium as proposed previously ( Figure 4).

Figures 5(c) and 5(d)
show the Raman spectra of the powder calcined at 700°C. The insert (i) in Figure 5(c) is the Raman spectrum of T control which obviously demonstrates the anatase-rutile structure; the rutile peaks are at 141 cm −1 (B 1g ), 235 cm −1 (multiphonon process), 445 cm −1 (A 1g ), and 608 cm −1 (B 2g ) while a low intensity peak is observed at 512 cm −1 (A 1g ) which belongs to that of the anatase phase. The Raman analysis is ratified by the XRD results according to which the T control sample is composed of anatase-rutile (1 : 99) mixture. The insert (ii) in Figure 5(c) is the Raman spectrum of T FA ; the Raman modes stand for rutile at 141, 235, 445, and 608 cm −1 whereas anatase peaks were recorded at 396 (weak emerging peak), 512, and 636 cm −1 (a weak shoulder), respectively, which are in agreement with the XRD results which demonstrate that T FA powder is a composite mixture of anatase-rutile in the proportion of 11 : 89. This result further illustrates that due to the formic acid chelation characteristic, anatase formation is favored; the intensity of the rutile Raman peaks decreases while anatase Raman peaks are developing. The rest of the Raman spectra shown in Figure 5(c) designate that of (NS) 0.06 -T FA and (NS) 0.1 -T FA and those in Figure 5(d) are for (NS) x -T AA (where x = 0, 0.06, and 0.1 mol%). The peaks (142, 197 (E g ), 396, 516, and 635 cm −1 ) recorded in all the Raman spectra stand for anatase phase.
3.4. X-Ray Photoelectron Spectroscopy (XPS). X-ray photoelectron spectroscopy analysis was performed to determine the chemical composition and to identify the chemical states of the elements present in the codoped TiO 2 . In addition to the main peaks of titanium (Ti2p, 459.1 eV), oxygen (O1s, 530.3 eV), and carbon (C1s, 285.1 eV), the XPS survey spectrum of (NS) 0.06 -T FA sample clearly shows the peaks of the dopants nitrogen (N1s, 400.5 eV) and sulphur (S2p, 169.2 eV) compared to T FA powder as shown in Figure 6. The enlarged view for the N1s, C1s, and S2p peaks is also shown in Figure 6(a) insert. All the peaks were calibrated to the C1s peak which is attributed to the advantageous carbon from the XPS instrument itself and to the residual carbon from the precursor solution. The relative atomic percentage of the elements in the calcined powder ((NS) 0.06 -T FA ) estimated from the XPS data were determined to be about 39.0, 52.7, 4.5, 1.8, and 2.0 at.% for titanium, oxygen, carbon, nitrogen, and sulphur, respectively.
The deconvoluted peaks of Ti2p, O1s, N1s, and S2p are shown in Figures 6(b), 6(c), 6(d), and 6(e). The Ti2p spectrum in Figure 6 [29]. The high-resolution spectrum of N1s shown in Figure 6(d) is divided into four peaks: the first peak at lower binding energy of 399.1 eV is attributed to substitutional N in the O-Ti-N structure implying that lattice oxygen atoms are substituted by N atoms; the second and third peaks at 400.4 and 401.6 eV are attributed to Ti-N-O and Ti-O-N bonds illustrating an interstitial N state in the codoped TiO 2 powder; and the forth peak at the higher binding energy of 403.6 eV can be assigned to nitrogen species bound to various surface oxygen sites (such as NO or NO 2 ) [30,31]. Figure 6(e) shows the corresponding high-resolution spectrum of the S2p region; after deconvolution, the peak at 168.5 eV is attributed to S 4+ and the peak at 169.0 eV is assigned to S 6+ , suggesting that sulphur was incorporated into the TiO 2 lattice via direct substitution of titanium atoms [32]. Nitrogen atoms can occupy either substitutional or institutional sites within the TiO 2 crystal lattice; published literature states that both types of nitrogen insertion lead to localized states in the TiO 2 band gap which results in visible light absorption [22,33]. In addition, it is also reported in the work of Ihara et al. [7] that oxygen vacancies produced by   Figure S3 and Table S1, supplementary information Section S.2.); Li et al. [34] stated that F and F + centers are situated at 2.7 and 2.4 eV from the valence band, while Cronemeyer [35] reported that oxygen vacancy states existed at 0.75-1.18 eV below the conduction band minimum of TiO 2 . Electrons can be excited from the valence band to the oxygen vacancies (F, F + , and F ++ ) centers followed by transition to the conduction band of TiO 2 ; thus, band narrowing due to the F, F + , and F ++ centers caused by N-doping cannot be neglected.
Doping of sulphur into the TiO 2 lattice also helps in the band gap narrowing of TiO 2 , because sulphur atoms can alter TiO 2 conduction or valence band (or both of them) by producing impurity levels in the band gap of TiO 2 as a consequence improving the visible light absorption [29]. Ma et al. [36] reported that visible light absorption in S-doped TiO 2 can be attributed to the formation of O-Ti-S or Ti-O-S bonds, which could lead partial electron transfers from sulphur atoms to titanium and oxygen atoms. Figure 7 and Table 1 illustrate that narrowing of the band gap in T FA and (NS) x -T FA is more than that in T AA and (NS) x -T AA samples; two absorption onsets were recorded in (NS) x -T FA samples (Figure 7(a) insert, Tauc plot of (NS) 0.06 -T FA powder): a long absorption tail gives the value of 2.75 eV (denoted by full line) and the small absorption tail gives the value of 2.68 eV (denoted by dotted line). The cutoff energy values are measured by extrapolating the linear portion of the spectrum until it intersects the hυ axis (x-axis) at E g . The presence of the small absorption tail indicates the substitutional doping of nitrogen replacing the oxygen atoms in the TiO 2 lattice; as a result, an impurity band (mixing of N2p states with O2p states) is formed above the valence band of TiO 2 . While the long absorption tail in (NS) x -T FA and in (NS) x -T AA samples is attributed to both, institutional doping of nitrogen atoms into the TiO 2 lattice results in forming an isolated N impurity level above the valence band and mixing of S states with the conduction band (Ti3d). Therefore, the visible light response in NS-codoped TiO 2 powder may be ascribed to the combination of isolated N impurity levels and impurity levels due to mixing of the N states and S states with valence band O2p and Ti3d states, respectively. In addition, the formation of Ti 3+ centers and oxygen vacancies as depicted in XPS and PL studies may also help to improve the visible light absorption. Comparing the Tauc plot and absorption spectra of (NS) x -T FA powders to the (NS) x -T AA powders, the second small absorption tail is absent in the latter and this behaviour may be assigned to the difference in the chelating linkages of the two carboxylic acids. Furthermore, a blue shift in absorption spectrum of (NS) x -T FA and (NS) x -T AA powders was noticed on the increase in dopant content beyond 0.06 and 0.08 mol%, respectively. This is due to the decrease in the rutile fraction within the sample, and probably, 11 International Journal of Photoenergy the dopants exceed their solubility limit in TiO 2 , which diffuse to accumulate in the grain boundaries without contributing to the narrowing in the TiO 2 band gap [23]. Thus, an optimum amount of dopant addition in TiO 2 will bring the desired improvement in visible light characteristics of TiO 2 ; beyond such limit, further dopant addition will contribute towards the absorption blue shift.
3.6. Photocatalytic Activity. The UV and visible light photocatalytic activities of the control, commercially available Degussa P25 and NS-codoped TiO 2 powders, in addition to the effect of photolysis (absence of photocatalyst) were studied by monitoring the degradation of 4-chlorophenol (4-CP) as a model pollutant; the obtained results were shown in Figure 8. The UV activity trend presented in Figure 8(a) was as follows: (NS) 0.06 -T FA > T FA > Degussa P25 ≈ (NS) 0.08 -T AA > T AA > T control . Similarly, in visible light irradiation (Figure 8(b)), the (NS) 0.06 -T FA photocatalyst also showed higher activity than the (NS) 0.08 -T AA sample. It should be noted that the (NS) 0.08 -T AA powder has the highest activity among the (NS) x -T AA (where x is 0.02, 0.04, 0.06, and 0.1 mol% dopant concentration) and T AA samples. In the visible activity experiments, no results were presented for the control TiO 2 and Degussa P25 catalysts as they absorb light mainly in the UV region due to their high band gap value and they are not suitable for visible light photocatalysis. Moreover, experimentally (Figure 8(a)), it was observed that the degradation of the model pollutant 4-CP was not affected by applying direct UV and or visible light in the absence of photocatalyst (photolysis). This means that any change in the pollutant concentration can be ascribed only to the photocatalysis [21].
The highest photocatalytic activity of (NS) 0.06 -T FA powder can be ascribed to several reasons such as the composite nature of the sample (anatase/rutile), surface hydroxyl ions, narrowed band gaps, the enhanced formation of Ti 3+ centers, and oxygen vacancies due to NS codoping. As a result of synergistic effects between anatase-rutile and the charge trapping of dopant ions, the electron-hole recombination is decreased, and as a consequence, the charge carriers are diffused to the surface to be consumed in the process redox reactions. FTIR studies showed that as a result of chelation and NS codoping in TiO 2 , the surface hydroxyl groups (a broad absorption band (2800-3600 cm −1 ) centered at 3178 cm −1 ) are intact even at the high calcination temperatures of 700°C. These hydroxyl groups are very significant for the enhanced activity because they act as scavengers for the photogenerated holes to produce hydroxyl radicals which react with the organic compounds for their degradation. Doped powders have lower band gap values compared to control powders and thus are capable to be irradiated with longer wavelength light to produce exciton pairs that result in the increased production of super oxide anions (O 2

•−
) and hydroxyl radicals (OH • ); these radicals then take part in pollutant decomposition. Moreover, due to charge compensation, NS codoping promotes the formation of oxygen vacancies. During photocatalytic processes, these oxygen vacancies not only promote photocatalyst absorption for the lower energy photons but also acts as an active site for the formation of active species (O 2 •− and OH • radicals) required for the photocatalytic reactions [34,37]. Figure S4 (supplementary information) shows the pseudo-first-order kinetics of 4-chlorophenol degradation; under UV irradiations ( Figure S4a), the apparent rate constants, k app values (obtained from the slope of the plot −ln(C/C o ) versus irradiation time), for the powders T control , T FA , T AA , (NS) 0.06 -T FA , and (NS) 0.06 -T AA are 0.01, 0.05, 0.04, 0.07, and 0.04 min −1 , respectively. This shows that (NS) 0.06 -T FA powder has high activity among the prepared samples, and its rate constant value of 0.07 min −1 is nearly 1.4 times higher than T FA and 2.5 times higher than T control powders. Also, under visible light irradiations ( Figure S4b), the sample NS 0.6 -T FA showed a rate constant value of 0.22 min −1 which is nearly 1.6 times higher than (NS) 0.08 -T AA powder which showed the highest activity among the doped powders employing acetic acid in the photocatalyst synthesis process. ) are produced which have the potential to react directly by oxidative pathways, and it can also produce singlet oxygen or decompose to hydrogen peroxide (H 2 O 2 ), which is transferred to hydroxyl radicals during the photocatalytic process. To elucidate the formation of O 2 −• species, the scavenger superoxide dismutase (SOD) is added to the solution SOD red + 0 −• 2 + 2H + → SOD + H 2 O 2 [38], which resulted in a decrease of 40% in the degradation rate of 4-CP under our experimental conditions (Figure 9(a)). To test the formation of singlet oxygen during the photocatalytic process, the scavenger azide was added to the solution [39], which showed a decrease of 6% in the photodegradation rate; this result illustrates that singlet oxygen do not play a significant role in the degradation of 4-CP using (NS) 0.06 -T FA as a photocatalyst. Photoillumination of TiO 2 leads to the production of holes, which can react with adsorbed H 2 O to yield hydroxyl radicals (OH • ) on the catalyst surface. The scavengers potassium iodide (KI) and 2-propanol were used to measure the participation of holes and hydroxyl radicals in the photocatalytic process [40]; the result showed that the degradation rate decreased about 60 and 63%, respectively. In conclusion, these results showed that primarily holes and hydroxyl radicals followed by O 2 −• species were mainly responsible for the degradation of 4-CP. To further support the production of major reactive species, the hydroxyl radicals were monitored by the PL method using terephthalic acid (TA) as the 13 International Journal of Photoenergy fluorescence probe in the aqueous solution [21]. TA reacts easily with hydroxyl radicals to produce the highly fluorescent 2-hydroxy terephthalic acid TA + OH • → TAOH . Figure 9(b) shows the TAOH fluorescence intensity in the suspension solution with the highly active synthesized photocatalyst powder as a function of UV illumination time. It can be observed that the (NS) 0.06 -T FA powder produced a large amount of hydroxyl radicals compared to (NS) 0.08 -T AA , which is in agreement with the photocatalytic activity results shown in Figure 8. This result also indicates that a gain in the light absorption due to the decrease in the band gap of NS-doped powders helps the photoinduced carriers for producing hydroxyl radicals, which results in the enhanced degradation of 4-chlorophenol.
3.8. Mechanistic Study. In reference to the observed characterization and the UV-vis photocatalytic activity results, a probable mechanism explaining the transfer pathways of photogenerated charge carriers for the photodegradation of 4-chlorophenol is illustrated in Figure 10. The UV excitation of electrons from the valence band (VB) to the conduction band (CB) of TiO 2 is denoted by full bold lines while that of visible light is represented by the bold dashed lines.
The enhanced UV-vis photocatalytic activity for the degradation of 4-CP was achieved with (NS) 0.06 -T FA and (NS) 0.08 -T AA catalyst powders. The (NS) 0.06 -T FA sample consists of anatase-rutile; due to the synergistic effect, there is enhanced separation between the photoinduced electrons and holes, since the CB of rutile is slightly lower than that of anatase; thus, electrons in the anatase CB would immediately diffuse into the CB of rutile. Subsequently, the CB electrons react with the surface adsorbed oxygen molecules to produce superoxide anion radicals (O 2 −• ), which are consumed in the degradation of 4-CP.
DRS spectra showed that incorporation of NS dopants in the TiO 2 lattice through substitutional or interstitial doping can modify the TiO 2 electronic band gap. The Tauc plot of (NS) x -T FA powders showed two diffused reflection onsets in comparison to (NS) x -T AA powders. The small absorption tail is attributed to the substitutional doping of nitrogen in the TiO 2 lattice; as a result of substitutional N doping, occupied N2p-localized states (Figure 10(a)) appear above the O2p valence band, which narrows the band gap of TiO 2 and shifts the optical absorption edge to the visible light region [33]. Electrons can be excited from this occupied N2p impurity level to the surface defect (oxygen vacancies and Ti 3+ species) states and to TiO 2 CB band under visible light irradiation. The long absorption tail is ascribed to the interstitial doping of nitrogen and mixing of S states with conduction band (Ti3d) into TiO 2 . With interstitial N doping, localized states with π character are generated by the NO bond [33]. Two antibonding states (π-NO) lie above the O2p band and two deep energy bonding states (π * -NO) lie below the O2p states. It is the antibonding NO orbitals that facilitate visible light absorption as electrons can be excited from this occupied high energy state to the CB (Figure 9(b)). Moreover, some researchers believe that the interstitial N impurity which lies above the valence band may also act as a strong trapping site  14 International Journal of Photoenergy for the photogenerated holes [41] and as a consequence helps to decrease the charge carriers recombination rate. The S dopant may also induce an impurity level (S 6+ /S 4+ ) near the TiO 2 CB band which not only helps in the decrease of TiO 2 band gap but also helps to enhance production of electrons under visible light irradiations. Moreover, the Nimpurity level may also behave as a trapping site for the VB holes like the S impurity level (Figure 10(a)) acts for the CB electrons; as a result, they inhibit the recombination of photoinduced holes and electrons.
PL (supplementary information Section S.2.) and XPS analyses verify the presence of Ti 3+ centers and oxygen vacancies (F, F + ), which contributed to the low energy photon absorption in doped TiO 2 . Excess electrons in the CB or electrons excited from the valence band can cause the reduction of Ti 4+ species to produce Ti 3+ species in the bulk. These Ti 3+ centers lie below the bottom of the CB and may act as trapping sites for visible light photoexcited electrons; afterwards, the trapped electrons are diffused to the surface for the production of superoxide anions (O 2

−•
). Oxygen vacancy states introduced by doping into the TiO 2 lattice may act as twofold: firstly, by inducing the visible light absorption, as a result, electrons from the oxygen vacancy states (F, F + ), are excited to CB or to Ti 3+ centers, and secondly, the oxygen vacancies can trap the photoinduced VB electrons and act as reactive centers for the photocatalytic process [30].
Thus, it is believed that S and N impurity levels, π * -NO localized states, and Ti 3+ centers and oxygen vacancies act cooperatively to narrow the NS TiO 2 band gap, resulting in the visible light activity of NS-TiO 2 . As shown in Figure 10, the CB electrons or the electrons trapped at the sulphur impurity level react with surface-absorbed oxygen molecules to produce O 2 −• species, while the VB holes or the holes trapped at the nitrogen impurity level react with hydroxyl groups (OH − ) to produce hydroxyl radicals (OH • ). Afterwards, these radicals (O 2 −• , OH • ) take part in the degradation of organic pollutants under UV-vis light irradiations.

Conclusions
A systematic study was conducted to address the chelation of formic acid and acetic acid to the alkoxide precursor, and the effect of NS dopants on the physicochemical properties of carboxylic acid-modified TiO 2 and on the anatase to rutile transformation temperature was performed. XRD studies showed that the formic and acetic acid-modified TiO 2 samples were composed of a composite mixture of anataserutile phases and of a pure anatase phase, respectively, which was due to carboxylic acid chelation as evidenced by FTIR studies. In contrast, due to the absence of carboxylate linkages, the T control sample was consisted of mainly rutile (99.0 wt.%) phase. Greater visible absorbance for the (NS) x -T FA samples were measured compared to (NS) x -T AA due to the different carboxylate coordination with titanium atoms; this means that nitrogen and sulphur atoms are doped substitutionally in formic acid-modified TiO 2 with ease and in greater amounts compared to acetate-modified TiO 2 . When evaluating the photocatalytic performance of the prepared samples, the (NS) 0.06 -T FA powder showed the highest activity for the degradation of 4-chlorophenol under both UV and visible light irradiation. This enhanced photoactivity was attributed to the composite nature of the sample (anatase/ rutile), photogenerated holes and surface hydroxyl ions, the narrowed band gap, formation of Ti 3+ centers, and oxygen vacancies due to NS codoping. Finally, to verify the importance of the different reactive oxidizing species (ROS) involved in the photocatalytic process, some scavenger agents were employed, and the results showed that photogenerated holes, OH • radicals, and O 2 −• species were primarily responsible for the degradation of 4-CP. In conclusion, this study will indeed offer more opportunities to investigate other chelating compounds with titanium precursors and the effect of metals (alkali, transition, and rare earth metals) and nonmetals on the chelation and physicochemical and photocatalytic properties of the final TiO 2 material.

Conflicts of Interest
The authors declare that there is no conflict of interests regarding the publication of this paper.