Hexavalent Chromium Removal via Photoreduction by Sunlight on Titanium–Dioxide Nanotubes Formed by Anodization with a Fluorinated Glycerol–Water Electrolyte

In this paper, titanium–dioxide (TiO2) nanotubes (TNTs) are formed by anodic oxidation with a fluorinated glycerol–water (85% and 15%, respectively) electrolyte to examine the effect of fluoride ion concentration, time, and applied voltage on TNT morphologies and dimensions. For fluoride ion concentration, the surface etching increases when the amount of ammonium fluoride added to the electrolyte solution increases, forming nanotube arrays with a clear pore structure. At a constant voltage of 20 V, TNTs with an average length of ~2 µm are obtained after anodization for 180 min. A prolonged anodization time only results in a marginal length increment. The TNT diameter is voltage dependent and increases from approximately 30 nm at 10 V to 310 nm at 60 V. At 80 V, the structure is destroyed. TNTs formed at 20 V for 180 min are annealed to induce the TiO2 anatase phase in either air or nitrogen. When ethylenediaminetetraacetic acid is added as a hole scavenger, 100% hexavalent chromium removal is obtained after 120 min of sunlight exposure for nitrogen-annealed TNTs.


Introduction
In recent years, the application of TiO 2 as a photocatalyst to reduce heavy metals such as hexavalent chromium [Cr(VI)] has gained academic attention. Cr(VI) is used in several industrial processes, including chrome plating, stainless steel manufacturing, chrome pigment formation, and leather tanning. Among these, chrome plating is responsible for producing large amounts of wastewater containing Cr(VI) ions. Fortunately, Cr(VI) compounds can be eliminated from industrial wastewater by chemical precipitation, flotation, adsorption, ion exchange, and electrochemical deposition [1].
The reduction of Cr(VI) to chromium [Cr(III)] is also conducted to treat Cr(VI)-laden wastewater by photocatalysis [2]. This process is beneficial because, whereas Cr(VI) is mobile, corrosive, and toxic to humans, Cr(III) is non-toxic and an essential dietary element. Industrial activities such as metal finishing and chrome plating produce wastewater with large amounts of Cr(VI), which can contaminate water bodies and soil. The uptake of Cr (VI) in plants from Cr(VI)-contaminated soil was reviewed by Shahid et al. in 2017 [3], and the existence of Cr(VI) in rice was reported by Rudzi et al. in 2018 [4]. Preventing Cr(VI) from contaminating the environment can be achieved by properly removing the ions from their discharge point. As mentioned, Cr(VI) can be reduced to Cr(III) by photocatalysis process. This process requires a photocatalyst, such as titanium dioxide (TiO 2 ), which is very effective due to the negative position of the conduction-band electrons that can reduce Cr(VI) to Cr(III).
Zwilling and co-workers [13] fabricated highly ordered porous anodic TiO 2 films in chromic acid containing hydrofluoric acid. This so-called first-generation electrolyte resulted in short TNTs (~500 nm) due to excessive etching. This was followed by the development of a second-generation electrolyte, i.e., a buffered solution containing sodium fluoride or ammonium fluoride (NH 4 F), which produces TNTs longer than 5 µm [24]. Even longer TNTs (>10 µm) have been formed using third-generation organic fluoride electrolytes, such as ethylene glycol, glycerol, and formamide. However, organic electrolytes require oxygen species for stable oxide growth [25][26][27][28][29][30][31]; water is commonly added as the oxygen provider. Recent research has reported the use of alkaline species for electrolytes to obtain even longer TNT formation, such as potassium hydroxide, lithium hydroxide, and sodium hydroxide [32][33][34][35][36][37][38]. The use of hydrogen peroxide has also been explored, and TNTs with a grassy surface structure have been reported [39].
In this work, TNTs are produced by Ti anodization in glycerol-water electrolyte. Anodic film grew in glycerol-water electrolyte was examined, including the success of self-aligned TNTs formation, by varying NH 4 F content, time, and anodization voltage. To the best of our knowledge, research on TNTs formation with the electrolyte in question is lacking, especially with respect to using photocatalysts for Cr(VI) reduction in synthetic wastewater. In fact, there is a complete lack of work reporting on the use of highly serrated TNTs for Cr(VI) reduction under sunlight. The formation TNTs with serrated walls, could increase the surface area for more catalytic sites that can enhance reactions. Long TNTs with a clear surface opening are also preferred for the catalytic process to take advantage of both the interior and the exterior of the nanotubes.
TiO 2 is a wide-bandgap semiconductor, and, therefore, electron-hole pair generation is only possible under ultraviolet (UV) radiation. However, sunlight activation can be achieved by the mid-gap states induced by the defects created when TNTs are heat-treated in a reduced atmosphere or in a nitrogen-containing environment. In this paper, the formed TNTs are annealed in nitrogen, and their ability to reduce Cr(VI) via photoreduction to Cr(III) is reported. Moreover, to ensure an effective reduction process, ethylenediaminetetraacetic acid (EDTA) was used as a hole scavenger. This is necessary, as a hole scavenger can suppress the recombination of electron-hole pairs, hence the increased in the number of free electrons available for Cr(VI) reduction.  Figure 1 shows the cross-section FESEM morphologies of anodized Ti in the waterglycerol electrolyte with various amounts of NH 4 F; insets are the surface images of the corresponding samples. Anodization was conducted at 20 V for 90 min. From the micrographs, it is evident that the amount of NH 4 F added to the electrolyte influences the length and diameter of the TNTs. Figure 2 summarizes the findings, from which it is evident that, apart from the 0.1 wt.% NH 4 F sample, a nanotubular structure was successfully formed with the length and diameter affected by the amount of fluoride ions. The TNT length increased from 1.6 to 2.0 µm when anodized with the electrolyte containing 0.3 and 0.5 wt.% NH 4 F, respectively. However, higher concentrations of NH 4 F (>0.7 wt.%) reduced the length, which can be attributed to severe chemical dissolution at the TNTs surface due to the increased number of fluoride ions. Moreover, the TNT diameter increased, albeit marginally, with an increase in NH 4 F content.  Figure 1 shows the cross-section FESEM morphologies of anodized Ti in the waterglycerol electrolyte with various amounts of NH4F; insets are the surface images of the corresponding samples. Anodization was conducted at 20 V for 90 min. From the micrographs, it is evident that the amount of NH4F added to the electrolyte influences the length and diameter of the TNTs. Figure 2 summarizes the findings, from which it is evident that, apart from the 0.1 wt.% NH4F sample, a nanotubular structure was successfully formed with the length and diameter affected by the amount of fluoride ions. The TNT length increased from 1.6 to 2.0 µm when anodized with the electrolyte containing 0.3 and 0.5 wt.% NH4F, respectively. However, higher concentrations of NH4F (> 0.7 wt.%) reduced the length, which can be attributed to severe chemical dissolution at the TNTs surface due to the increased number of fluoride ions. Moreover, the TNT diameter increased, albeit marginally, with an increase in NH4F content. The formation of TiO2 by anodic oxidation is described by Equation (1). At the initial stage of the anodization process, a barrier layer comprising defective TiO2 was formed. Pores then developed because of the chemical dissolution of the barrier layer (Equation (2)) and for discrete nanotubes formation, the pores must be separated.  The formation of TiO 2 by anodic oxidation is described by Equation (1). At the initial stage of the anodization process, a barrier layer comprising defective TiO 2 was formed. Pores then developed because of the chemical dissolution of the barrier layer (Equation (2)) and for discrete nanotubes formation, the pores must be separated.

Results and Discussion
During the anodization process, ion migration across the anodic film is essential to increase the oxide thickness. The migration of fluoride ions is believed to be faster than any other inward-migrating ions, and, thus, a fluoride-rich layer (FRL) will form at the oxide-Ti interface [40]. The bottom part of the nanotubes, as seen in the cross-sectional images, is scallop-shaped, indicating a flow mechanism in the oxide pore formation. According to this mechanism, FRL will eventually be displaced as inter pore materials. Pores are separated when the FRL is dissolved by water, resulting in discrete nanotubes. At larger amounts of NH 4 F, a thicker FRL is thought to form around the pores, and, as the electrolyte is excess with water, the dissolution of this FRL will result in TNTs with very thin walls. As seen from the FESEM images, the sample prepared in the electrolyte with 1.0 wt.% NH 4 F had walls thinner than 10 nm, and the severe dissolution resulted in an inhomogeneous surface structure. From here, 0.5 wt.% NH 4 F was chosen as the optimum amount of fluoride ion in the electrolyte, as it is adequate for surface etching, and the nanotubes formed also had considerable length, as shown in Figure 2.

Effect of Anodization Voltage
Apart from NH 4 F content, anodization voltage is also known to play an essential role in tuning TNT dimensions [20]: a higher anodization voltage results in a higher oxidation rate, resulting in the formation of more H + ions in the electrolyte (Equation (1)). If the number of H + ions in the electrolyte increases, chemical dissolution increases, resulting in larger and longer TNTs. Nevertheless, there is a limit to the voltage applied, as a high voltage may induce an overly rapid electric field dissolution, destroying the nanotubular structure. As reported by Lockman et al., depending on the electrolyte used, there is a threshold voltage at which the TNT structure is formed; TNTs are formed above the threshold voltage, while a lower voltage would result in insufficient electric field dissolution, generating compact anodic film or film with small pores [20]. Figure 3 shows the FESEM morphologies, whereas Figure 4 summarizes the anodic film features as a function of applied voltage. Ti anodized at 1 V has a rather compact surface oxide without any noticeable microstructure, as shown by Figure 3a. Small pits with 5-8 nm diameters can be observed in Figure 3b for the sample anodized at 3 V. Increasing the anodization voltage to 5 V results in larger pits (diameter~10 nm) with an increased distribution, as shown in Figure 3c.
The corresponding cross-sectional FESEM image shows that the thickness of the anodic layer is approximately 80 nm. TNTs formed on the foil with an average diameter of 30 nm (Figure 3d As the anodization voltage was increased to 30 V, the diameter increased to 150 nm ( Figure 3g); at 40 V, it increased to 250 nm ( Figure 3h); however, the nanotubes length for both samples did not differ by much (~2.5 µm). Anodization at 60 V resulted in TNTs with a diameter and length of 310 nm and 2.6 µm, respectively (Figure 3i); for this electrolyte, it appears that voltage affects TNT diameter but not length. At 80 V, the nanotubular structure was destroyed ( Figure 3j) due to the overaccelerated electric field dissolution causing severe polarization and weakening of the Ti-O bond, which led to the formation of an irregular porous oxide structure. Figure 3k shows the TEM image of the 20 V sample taken from the bottom part of the TNTs, demonstrating TNTs with scallop-shaped barrier layer of ∼20 nm thickness. If compared with anodic TNTs formed in ethylene glycol [32][33][34][35][36][37][38], the thickness of the barrier layer for the abovementioned sample is rather large, perhaps because of the excess water in the glycerol, which increases oxidation. In Figure 3l, a single nanotube is shown with serrated walls. structure. As reported by Lockman et al., depending on the electrolyte used, there is a threshold voltage at which the TNT structure is formed; TNTs are formed above the threshold voltage, while a lower voltage would result in insufficient electric field dissolution, generating compact anodic film or film with small pores [20]. Figure 3 shows the FESEM morphologies, whereas Figure 4 summarizes the anodic film features as a function of applied voltage. Ti anodized at 1 V has a rather compact surface oxide without any noticeable microstructure, as shown by Figure 3a. Small pits with 5-8 nm diameters can be observed in Figure 3b for the sample anodized at 3 V. Increasing the anodization voltage to 5 V results in larger pits (diameter ~10 nm) with an increased distribution, as shown in Figure 3c. of an irregular porous oxide structure. Figure 3k shows the TEM image of the 20 V sample taken from the bottom part of the TNTs, demonstrating TNTs with scallop-shaped barrier layer of ∼20 nm thickness. If compared with anodic TNTs formed in ethylene glycol [32][33][34][35][36][37][38], the thickness of the barrier layer for the abovementioned sample is rather large, perhaps because of the excess water in the glycerol, which increases oxidation. In Figure 3l, a single nanotube is shown with serrated walls.

Effect of Variation Reaction Time
Foils were then anodized at 20 V but at varying times from 1 to 720 min; 20 V was selected, as it produced TNTs with a diameter smaller than 100 nm. A time variation experiment was performed to investigate the influence of anodization duration over TNT dimensions. The FESEM images in Figure 5 show the morphologies of the anodic film, and Figure 6 summarizes the anodic film features formed at different anodization times. As seen in Figure 5a, anodization for 1 min resulted in anodic films with a thickness of ~10 nm and randomly distributed pits on the surface. The pit diameter increased to ~50

Effect of Variation Reaction Time
Foils were then anodized at 20 V but at varying times from 1 to 720 min; 20 V was selected, as it produced TNTs with a diameter smaller than 100 nm. A time variation experiment was performed to investigate the influence of anodization duration over TNT dimensions. The FESEM images in Figure 5 show the morphologies of the anodic film, and Figure 6 summarizes the anodic film features formed at different anodization times. As seen in Figure 5a, anodization for 1 min resulted in anodic films with a thickness of 10 nm and randomly distributed pits on the surface. The pit diameter increased to~50 nm after anodization for 15 min (Figure 5b). From the cross-sectional morphology, it is evident that a self-aligned nanotubular structure formed beneath this pitted region, indicating inadequate surface etching. A more obvious nanotube structure was obtained for the sample anodized for 30 min (Figure 5c). The TNT opening was also very clear, suggesting that surface etching is time dependent. Moreover, TNTs length was longer for this sample than for the 15 min sample.
The length increased further with an increase in anodization time; at 60 min, the TNT length increased to 700 nm; at 90 min, it increased to 1.5 µm; and, at 120 min, it increased to 1.8 µm, as shown in Figure 5d-f, respectively. However, the length remained constant at 2 µm for samples anodized for 180-720 min (Figure 5g-i). The TNT diameter was~80 nm for all samples, indicating that the diameter is independent of anodization time. Similar findings have been reported [18].

Effect of Annealing Environment.
Samples prepared by anodization at 20 V were annealed either in air or nitrogen at 450 • C for 3 h. The XRD pattern in Figure 7a shows that the as-anodized TNTs are amorphous, and, after annealing, anatase TiO 2 (ICSD: 98-010-7874) can be identified from peaks at 25.4 • , 48.1 • , and 54.7 • , corresponding to the (011), (020), and (121) planes, respectively. Minimal differences can be observed for samples annealed in air and nitrogen, apart from a slight shift in the (011) peak, as shown in Figure 7b. This shift is attributed to lattice distortion [41], perhaps induce by the substitution of nitrogen within the crystal lattice, which results in strain [42][43][44]. Since the ionic radius of nitrogen (r 2+ N = 1.46 Å) is larger than that of oxygen (r 2+ O = 1.38 Å), the substitution of nitrogen in the O-site of the crystal structure may result in the TiO 2 crystal lattice expanding because of tensile strain [42].
sample than for the 15 min sample.
The length increased further with an increase in anodization time; at 60 min, the TNT length increased to 700 nm; at 90 min, it increased to 1.5 µm; and, at 120 min, it increased to 1.8 µm, as shown in Figure 5d-f, respectively. However, the length remained constant at ~2 µm for samples anodized for 180-720 min (Figure 5g-i). The TNT diameter was ~80 nm for all samples, indicating that the diameter is independent of anodization time. Similar findings have been reported [18].  Samples prepared by anodization at 20 V were annealed either in air or nitrogen at 450 °C for 3 h. The XRD pattern in Figure 7a shows that the as-anodized TNTs are amor- at 25.4°, 48.1°, and 54.7°, corresponding to the (011), (020), and (121) planes, respectively. Minimal differences can be observed for samples annealed in air and nitrogen, apart from a slight shift in the (011) peak, as shown in Figure 7b. This shift is attributed to lattice distortion [41], perhaps induce by the substitution of nitrogen within the crystal lattice, which results in strain [42][43][44]. Since the ionic radius of nitrogen ( = 1.46 Å) is larger than that of oxygen ( = 1.38 Å), the substitution of nitrogen in the O-site of the crystal structure may result in the TiO2 crystal lattice expanding because of tensile strain [42]. Anatase crystallite size for air-and nitrogen-annealed TNTs was calculated using the Debye-Scherrer equation at the (011) anatase peak according to the full width at the half maximum of the diffraction pattern. The crystallite sizes were 26.29 and 36.79 nm for nitrogen-and air-annealed TNTs, respectively. The smaller crystallite size for the nitrogenannealed TNTs can be attributed to growth suppression during annealing in the nitrogen environment [45].
To further study the changes in phase structure after annealing in air and nitrogen environment, Raman spectroscopy on the TNTs was carried out, and the results are shown in Figure 8. When the TNTs were annealed at 450 • C, both samples exhibited three typical modes corresponding to A 1g (515 cm −1 ), B 1g (398 and 515 cm −1 ), and E g (144, 197, and 640 cm −1 ), respectively, in the Raman spectrum [39]. The peaks in these modes demonstrate the presence of anatase phase as the predominant crystal structure. The anatase peaks of TNT are labelled with A ( Figure 8a). The intensity of the dominant anatase peak at 144 cm −1 was higher for the TNTs annealed in nitrogen compared to air. Moreover, from the enlarged image of the dominant peak in Figure 8b, it can be observed that the bandwidth increased for the nitrogen-annealed sample. These results indicate that the crystallinity of the anatase phase was enhanced and the crystallite size decreased after annealing in nitrogen, which is consistent with the results of the XRD measurement. Anatase crystallite size for air-and nitrogen-annealed TNTs was calculated using the Debye-Scherrer equation at the (011) anatase peak according to the full width at the half maximum of the diffraction pattern. The crystallite sizes were 26.29 and 36.79 nm for nitrogen-and air-annealed TNTs, respectively. The smaller crystallite size for the nitrogenannealed TNTs can be attributed to growth suppression during annealing in the nitrogen environment [45].
To further study the changes in phase structure after annealing in air and nitrogen environment, Raman spectroscopy on the TNTs was carried out, and the results are shown in Figure 8. When the TNTs were annealed at 450 °C, both samples exhibited three typical modes corresponding to A1g (515 cm −1 ), B1g (398 and 515 cm −1 ), and Eg (144, 197, and 640 cm −1 ), respectively, in the Raman spectrum [39]. The peaks in these modes demonstrate the presence of anatase phase as the predominant crystal structure. The anatase peaks of TNT are labelled with A ( Figure 8a). The intensity of the dominant anatase peak at 144 cm −1 was higher for the TNTs annealed in nitrogen compared to air. Moreover, from the enlarged image of the dominant peak in Figure 8b, it can be observed that the bandwidth increased for the nitrogen-annealed sample. These results indicate that the crystallinity of the anatase phase was enhanced and the crystallite size decreased after annealing in nitrogen, which is consistent with the results of the XRD measurement. The optical properties of the annealed TNT films were determined using diffuse reflectance UV-Vis spectroscopy, the results of which are shown in Figure 9. The energy bandgap of the samples was calculated using the Kubelka-Munk equation [46]: (1 − )  The optical properties of the annealed TNT films were determined using diffuse reflectance UV-Vis spectroscopy, the results of which are shown in Figure 9. The energy bandgap E g of the samples was calculated using the Kubelka-Munk equation [46]: where R is the diffuse reflectance of the sample. Using the Tauc relation, the following equation was obtained: where A is the proportionality constant, F(R) is the K-M function, and hυ is photon energy [47,48]. The F(R) function can be multiplied by hυ using a corresponding coefficient (n).
The n values of 0.5 and 2 were used to estimate the direct and indirect bandgap oxide, respectively. Then, the extrapolation of (F(R) × hυ) 2 or (F(R) × hυ) 0.5 = 0 was conducted to obtain the direct and indirect E g values, respectively.

Photoreduction of Hexavalent Chromium Ions on Titanium-Dioxide Nanotubes
Cr(VI) photoreduction experiments were conducted by exposing 10 ppm Cr(VI) solution to natural sunlight on a sunny day. Cr(VI) photoreduction was very slow without EDTA addition, as shown in Figure 10; less than 10% Cr(VI) removal was recorded. The percentage of removal increased when EDTA was added to the Cr(VI) solution: 50% and 63% for air-and nitrogen-annealed TNTs, respectively, after 60 min of sunlight exposure. The total reduction of Cr(VI) was achieved for nitrogen-annealed TNTs after 120 min of sunlight exposure, possibly because of the narrow energy bandgap for this sample [46]. Air-annealed TNTs only achieved a 70% reduction after 120 min of sunlight irradiation.
Nitrogen insertion is possible within the TiO2 lattice, as observed from the anatase peak shift in the XRD. Substituting nitrogen in the lattice resulted in the formation of N-2p states above the valence band (VB) of TiO2 [51], which reduced the TiO2 bandgap and thus enabled the adsorption of visible light for electron excitation (Equation (5)) from the CB, as depicted in Figure 11.

Photoreduction of Hexavalent Chromium Ions on Titanium-Dioxide Nanotubes
Cr(VI) photoreduction experiments were conducted by exposing 10 ppm Cr(VI) solution to natural sunlight on a sunny day. Cr(VI) photoreduction was very slow without EDTA addition, as shown in Figure 10; less than 10% Cr(VI) removal was recorded. The percentage of removal increased when EDTA was added to the Cr(VI) solution: 50% and 63% for air-and nitrogen-annealed TNTs, respectively, after 60 min of sunlight exposure. The total reduction of Cr(VI) was achieved for nitrogen-annealed TNTs after 120 min of sunlight exposure, possibly because of the narrow energy bandgap for this sample [46]. Air-annealed TNTs only achieved a 70% reduction after 120 min of sunlight irradiation.  The reduction potential of HCrO4 − /Cr 3+ under the standard condition is E 0 = +1.35 V vs. normal hydrogen electrode (NHE) at pH = 0, whereas the CB potential for TiO2 is +0.05 eV vs. NHE at pH = 0 [52]. Since the reduction potential of HCrO4 − /Cr 3+ is more positive than the CB, electron reduction is possible: The reduction of Cr(VI) under acidic conditions (pH ≤ 2) can be described in Equation (6) [53]: The band edge positions of the CB and VB for the TNTs was calculated by applying the concept of electronegativity using Equations (7) and (8): Nitrogen insertion is possible within the TiO 2 lattice, as observed from the anatase peak shift in the XRD. Substituting nitrogen in the lattice resulted in the formation of N-2p states above the valence band (VB) of TiO 2 [51], which reduced the TiO 2 bandgap and thus enabled the adsorption of visible light for electron excitation (Equation (5)) from the CB, as depicted in Figure 11.  The reduction potential of HCrO4 − /Cr 3+ under the standard condition is E 0 = +1.35 V vs. normal hydrogen electrode (NHE) at pH = 0, whereas the CB potential for TiO2 is +0.05 eV vs. NHE at pH = 0 [52]. Since the reduction potential of HCrO4 − /Cr 3+ is more positive than the CB, electron reduction is possible: The reduction of Cr(VI) under acidic conditions (pH ≤ 2) can be described in Equation (6) [53]: The band edge positions of the CB and VB for the TNTs was calculated by applying the concept of electronegativity using Equations (7) and (8): = − + 0.5 (7) Figure 11. Proposed mechanism of Cr(VI) photoreduction of TNTs under sunlight irradiation.
The reduction potential of HCrO 4 − /Cr 3+ under the standard condition is E 0 = +1.35 V vs. normal hydrogen electrode (NHE) at pH = 0, whereas the CB potential for TiO 2 is +0.05 eV vs. NHE at pH = 0 [52]. Since the reduction potential of HCrO 4 − /Cr 3+ is more positive than the CB, electron reduction is possible: The reduction of Cr(VI) under acidic conditions (pH ≤ 2) can be described in Equation (6) [53]: The band edge positions of the CB and VB for the TNTs was calculated by applying the concept of electronegativity using Equations (7) and (8): where E VB is the VB edge potential at the point of zero charge, E CB is the CB edge potential, and X is the absolute electronegativity of the semiconductor (the X value for TiO 2 is 5.81 eV) [54], which is defined as the geometric average of the absolute electronegativity of the constituent atoms, and E e is the energy of free electrons at the hydrogen scale (approximately 4.5 eV). Table 1 shows the calculated E CB , E VB , and E g values for TNTs formed in glycerol-water electrolyte and after annealing. As shown in Figure 10, Cr(VI) reduction was sluggish without EDTA added to the solution. This could be due to several reasons, such as the fast recombination process [56] or the reoxidation of Cr(III) to Cr(VI) [57] by • OH. • OH radicals are well known as a strong oxidizing agent and are produced when holes react with H 2 O, which can be expressed by Equation (9) [58]: • OH concentration can be suppressed by ensuring they react with EDTA, as this results in their degradation to CO 2 and H 2 O, which can be expressed by Equation (10): Holes are also known to be oxidizing, and EDTA can be used to capture holes; capturing holes can also reduce the recombination of electron-hole pairs. This can be expressed by Equation (11):

The Effect of Initial Hexavalent Chromium Concentration
A reduction experiment using Cr(VI) of varying concentration was performed using a Cr(VI) solution of 5, 10, 15, and 20 ppm. The experiment ( Figure 12) showed a very fast reduction of 5 ppm Cr(VI) solution with 100% reduction after 90 min of sunlight exposure. As shown previously, 100% removal of Cr(VI) was observed after 120 min for 10 ppm solution. At higher ppm level of Cr(VI) reduction times, 80% removal was seen for 15 ppm solution and 75% for 20 ppm solution.

Reaction Kinetics Modeling
The pseudo-first-order Langmuir-Hinshelwood model [59] was used to understand the reaction kinetics: The resulting integrals of Equation (12) can be expressed as a pseudo-first-order equation when the adsorption is relatively weak/or the reactant concentration is low: where R denotes the photocatalytic reaction rate; k is the apparent photocatalytic pseudofirst-order reaction rate constant; C o and C denote the initial Cr(VI) concentration and the Cr(VI) concentration at a given time t, respectively. The model validity is determined by the correlation coefficient of determination (R 2 ): a value close to 1 suggests that the model is ideal for explaining Cr(VI) kinetics. Figure

Reaction Kinetics Modeling
The pseudo-first-order Langmuir-Hinshelwood model [59] was used to understand the reaction kinetics: The resulting integrals of Equation (12) can be expressed as a pseudo-first-order equation when the adsorption is relatively weak/or the reactant concentration is low: where R denotes the photocatalytic reaction rate; k is the apparent photocatalytic pseudofirst-order reaction rate constant; and denote the initial Cr(VI) concentration and the Cr(VI) concentration at a given time t, respectively. The model validity is determined by the correlation coefficient of determination (R 2 ): a value close to 1 suggests that the model is ideal for explaining Cr(VI) kinetics. Figure 13 plots −ln vs. time; the k value is the slope of the line fitted on the plot of −ln vs. t.
The inset of Figure 13a provides the k and R 2 values for Cr(VI) metal ions. The pseudo-first-order Cr(VI) removal rate increases when EDTA is introduced. Based on the k value, the photocatalytic activity of nitrogen-annealed TNTs with EDTA as a hole scavenger is roughly seven times larger (k = 0.0249 min −1 ) than that for nitrogen-annealed TNTs without EDTA (k = 0.0039 min −1 ). Similar results were obtained for the k value of air-annealed TNTs with EDTA (k = 0.0101 min −1 ): photocatalytic activity was roughly five times greater than it was in the absence of EDTA as a hole scavenger (k = 0.0022 min −1 ).
For Cr(VI) solutions with different concentrations, Cr(VI) reduction appears to follow the Langmuir-Hinshelwood model and can be described by pseudo-first-order kinetics, as confirmed by the obtained straight line and R 2 value of ≅1 (Figure 13b). A similar observation was reported for Cr(VI) photoreduction by Shaban et al. [60]. The reaction rate   Natural sun-Current The inset of Figure 13a provides the k and R 2 values for Cr(VI) metal ions. The pseudofirst-order Cr(VI) removal rate increases when EDTA is introduced. Based on the k value, the photocatalytic activity of nitrogen-annealed TNTs with EDTA as a hole scavenger is roughly seven times larger (k = 0.0249 min −1 ) than that for nitrogen-annealed TNTs without EDTA (k = 0.0039 min −1 ). Similar results were obtained for the k value of air-annealed TNTs with EDTA (k = 0.0101 min −1 ): photocatalytic activity was roughly five times greater than it was in the absence of EDTA as a hole scavenger (k = 0.0022 min −1 ).
For Cr(VI) solutions with different concentrations, Cr(VI) reduction appears to follow the Langmuir-Hinshelwood model and can be described by pseudo-first-order kinetics, as confirmed by the obtained straight line and R 2 value of ∼ =1 (Figure 13b). A similar observation was reported for Cr(VI) photoreduction by Shaban et al. [60]. The reaction rate constant for Cr(VI) reduction was 0.0303, 0.0243, 0.0145, and 0.0093 min −1 for the 5, 10, 15, and 20 ppm solutions, indicating a fast reduction in diluted solutions after sunlight irradiation for 120 min. Table 2 compares the Cr(VI) reduction efficiency obtained in the present paper with the existing literature, from which it is evident that the photocatalytic reduction of nitrogenannealed TNTs is comparable with fluorine-and carbon-doped TNTs under UV-Vis light irradiation [45,[61][62][63][64][65]. Furthermore, by adding a small amount of EDTA (1 mM), a higher efficiency of Cr(VI) photoreduction was obtained compared to the addition of other scavengers, such as phenol (10 mM) and tartaric acid (6 mM), at higher amounts under UV-Vis light irradiation. [60,66,67].

Materials and Methods
Ti foils (99.96% pure; thickness: 0.13 mm; Stream Chemical, Newburyport, USA) were cut into square sections (10 × 10 mm) for the experiment. Before anodization, the foils were degreased in an ultrasonic bath of acetone (J.T. Baker-9254, Center Valley, PA, United States) and then ethanol (95.7% pure; Samchen, Shah Alam, Selangor, Malaysia) for 15 min each. The foils were then rinsed with deionized water and dried. The cleaned foils were then placed in an electrochemical cell with a restricted area of 5 × 10 mm exposed to the electrolyte. The anodization experiment was conducted using a two-electrode electrochemical cell at room temperature with a platinum cathode (diameter of 2 mm; 75 mm in length; Metrohm, Herisau, Switzerland) and Ti foil as the anode using a DC power source (Agilent E3647A, Santa Clara, CA, USA). The distance between the anode and cathode was 30 mm. The electrolyte was a glycerol-water (85% and 15%, respectively) solution (Merck, Darmstadt, Germany). Ammonium fluoride (NH 4 F, Merck, Darmstadt, Germany)) was added to the electrolyte in different amounts: 0.1, 0.3, 0.5, 0.7, 0.9, and 1.0 wt.%. To examine the anodization time effect, a 0.5 wt.% NH 4 F electrolyte bath was used. The anodization time varied from 1 to 720 min at a constant voltage of 20 V. The effect of the anodization voltage was examined by varying the voltage from 1 to 80 V for 90 min in the same electrolyte bath (fixed sweep rate of 0.1 V/s).
The anodized Ti foil was removed from the electrolyte and rinsed with deionized water and dried in air naturally. The anodized Ti was then annealed using a horizontal tube furnace (Lenton 1200, Derbyshire, United Kingdom) at 450 • C for 3 h in air or nitrogen. The morphologies of the anodized foils were observed by field emission scanning electron microscopy (FESEM; Variable Pressure Zeiss Supra 35, Oberkochen, Germany). Transmission electron microscopy (TEM; JEOL, JEM-2100F, Tokyo, Japan), at an acceleration voltage of 200 kV, was used to provide detailed observations of nanotube structure. X-ray diffraction (XRD; Bruker D8, Bruker GmBH, Karlsruhe, Germany) and Raman spectrometer (REN-ISHAW inVia 9P1567, Charfield, United Kingdom) were conducted to determine the TNT phase and crystal analysis. UV-visible (UV-Vis) spectrophotometers were used to attain the diffuse reflectance spectra (Cary 5000, UV-Vis-NIR, Agilent, Santa Clara, United States) and conduct the Cr(VI) photoreduction (Varian Cary 50, budd lake, NJ, United States).
For the Cr(VI) removal evaluation, 100 ppm Cr(VI) was prepared by dissolving 0.0283 g of potassium dichromate salt (K 2 Cr 2 O 7 , A.R. grade, Merck, Germany) in 100 mL of deionized water. Then, the Cr(VI) stock solution was diluted to obtain the 5-20 ppm Cr(VI) solution. The pH of the solution was lowered using hydrogen chloride (Merck, Germany). For the hole scavenger, 1 mM of Ethylenediaminetetraacetic acid (EDTA; AJAX Chemicals, Sydney, Australia) was added to the Cr(VI) solution. During the test, four anodized foils with a surface area of 1 × 0.5 cm 2 were immersed in the Cr(VI) solution. Before light irradiation, the solution was left in the dark for 30 min to achieve adsorptiondesorption equilibrium. The solution was then exposed to sunlight with an average light intensity of~1000 Wm −2 , which was measured with a solar power meter (TM-207, Tenmars, Taiwan, China). During light irradiation, 2 mL of the aliquot sample was withdrawn every 15 min. The reduction of Cr(VI) was determined using 1,5-diphenylcarbazide (Merck, Germany), and, based on the UV-Vis results, the Cr(VI) concentration in the solution was calculated using Equation (14) at λ max = 540 nm: where C denotes the concentration of Cr(VI) at a given time (at λ max = 540 nm), and C o is the initial concentration (before sunlight irradiation).

Conclusions
In this study, photoreduction of Cr(VI) using anodized TNTs under natural sunlight irradiation was demonstrated. The TNTs were formed by the anodic oxidation of Ti foil in a fluorinated glycerol-water electrolyte. The effects of NH 4 F concentration in the electrolyte, anodization voltage, anodization duration, and annealing atmosphere (air or nitrogen) on the nanotubes were systematically described. The minimum voltage for TNTs formation was identifiezd as 5 V. To obtain TNT arrays with a clear, open top structure exhibiting a considerable length, anodization must be conducted using an electrolyte with an NH 4 F concentration of >0.5 wt.% for more than 30 min. A higher NH 4 F content led to rigorous etching forming irregular TNTs. The diameter of the TNT formed increased with voltage in the range of 10 to 60 V, while anodization at 80 V resulted in the collapse of the tubular structure. TNTs with an average length of~2 µm were obtained after anodization for