Preparation of Fluorine-Doped TiO2 Photocatalysts with Controlled Crystalline Structure

1 Institute of Materials Science, National Center for Scientific Research (NCSR) “Demokritos”, 15310 Attica, Greece 2 Institute of Physical Chemistry, National Center for Scientific Research (NCSR) “Demokritos”, 15310 Attica, Greece 3 Department of Chemistry, University of Ioannina, 45110 Ioannina, Greece 4 State Key Laboratory of Advanced Technology for Materials Synthesis and Processing, Wuhan University of Technology, Wuhan 430070, China


INTRODUCTION
In the last decades, the TiO 2 is regarded as a promising material for organic pollutants remediation due to its properties [1].Its wide application though is still retarded by the low efficiency of the photo-oxidative process.The development of titania-based photocatalysts activated by visible light (λ ≥ 385 nm) remains a great challenge.Anion doping of TiO 2 films and powders with elements like nitrogen [2][3][4][5], sulfur [6,7], carbon [8,9], and fluorine [10][11][12][13][14][15][16][17][18] has been investigated in order to red-shift the absorbance to be achieved.Narrowing of the band gap of TiO 2 upon doping was often considered responsible for the enhanced visible light activity of these materials.Theoretical DOS calculations revealed intermediate energy levels available in the band gap of TiO 2 for most of the elements except fluorine [1].Park and Choi [10] reported that surface fluorination (F-TiO 2 ) enhances the photocatalytic degradation of certain substrates due to the electron-trapping sites formation.The enhanced photocatalytic activity of F-doped powders has been attributed by  to several beneficial effects produced by Fdoping and mainly to the creation of surface oxygen vacancies.Yu et al. [14] reported fluorine ions incorporation in the TiO 2 lattice using an NH 4 F source.They attributed the stronger absorption in the UV-visible range to the presence of the F − ions that favor anatase phase formation and improve the crystallinity of TiO 2 .However, Yamaki et al. [15] claimed that F-doping causes a modification of the conduction band edge and introduction of impurity states near its bottom.Our recent study on fluorine-modified titania films [18] revealed improved anatase crystallinity upon fluorine doping without the effective band gap of TiO 2 to be affected.
Anatase is usually produced when TiO 2 is synthesized by sol-gel method [19].Calcination of anatase at high temperatures is applied to obtain rutile phase.Mixtures of anatase and rutile (e.g., Degussa P25 consists of ∼75% anatase and ∼25% rutile) are reported to exhibit enhanced photocatalytic activity in degradation of organic pollutants as the electronhole recombination rate is decreased in the composite system [20].Cheng et al. [21] reported that rutile phase is developed International Journal of Photoenergy preferably from a solution with low pH.Klein et al. [22] produced rutile (100%) titania powders at low temperature using nitric and hydrochloric acids as catalysts.
In this work, we study the influence of CF 3 COOH fluorine precursor on the structure and the properties of TiO 2 powders.

EXPERIMENTAL
Fluorine-(F-) doped TiO 2 powders have been prepared by sol-gel route.Tetraethyl orthotitanate (TEOTi) Merck was used as titanium source.Different amounts of fluoroacetic acid (Aldrich Chemical Co., Milwaukee, Wis, USA) (atomic ratios F/Ti = 0, 10/90, 20/80) were dissolved in deionized water (molar ratio H 2 O/TEOTi = 18) under stirring at room temperature and the TEOTi was added to the solution dropwise.The sols were stirred in closed beakers at room temperature for 24 hours for completion of the hydrolysis process.Consequently, they were dried at 90 • C in air for 12 hours for water and alcohol evaporation.The xerogels obtained were denoted as TF0, TF10, and TF20 for the undoped TiO 2 and the samples with F/Ti = 10/90 and 20/80 atomic ratios, respectively.The as-prepared xerogels were calcinated at 400, 500, and 600 • C for 1 hour in a Carbolite muffle furnace with heating rate of 5 • C/min.Differential thermal analysis (DTA) and thermogravimetry (TG) measurements were performed using NETZSCH equipment model STA 449.The samples with a mass ∼14 mg were placed in alumina crucible and heated in air flow (30 mL/min) from room temperature up to 1000 • C. The heating rate applied was 10 • C/min.
The crystalline structure of the powders was investigated using X-ray diffraction (XRD) technique.SIEMENS D500 diffractometer with secondary graphite monochromator and CuKα radiation was used to obtain the X-ray diffraction patterns of the samples.The measurements were performed using the following combination of slits: 1.0 • /1.0 • /1.0 • as aperture diaphragms, 0.15 • as detector diaphragm, and 0.25 • as diffracted beam monochromator diaphragm.The measured 2θ range between 20 • and 80 • was scanned in steps of 0.03 • /2s.The size of the crystallites responsible for the Bragg reflection was determined using the Scherrer equation where d is the crystallite diameter in Å, k the shape constant (0.9), λ the wavelength in Å, θ the Bragg angle in degrees, and b the observed peak width at half-maximum peak height in rad.The phase content (%) of the samples was calculated using the formula [14] W where W R is the content of the rutile phase (W A = 100−W R ), A A and A R are the integrated intensities of anatase (101) and rutile (110) peaks, respectively, and K A is a coefficient equal to 0.886.The surface areas of the powders were determined by the Branauer-Emmet-Teller (BET) method.The equipment COULTER 5200 USA was used for the nitrogen adsorption/desorption measurements at 77 K.All the samples were degassed at 573 K for 24 hours.
The presence of the fluorine in the TiO 2 , the valence state of the dopant, as well as the bonding element was analyzed by X-ray photoelectron spectroscopy (XPS).Nonmonochromatized MgKα X-rays from a dual-anode VSW TA-10 source were used.The ultra-high vacuum (UHV) chamber was baked at 150 • C for 12 hours to reach the 10 −10 mbar pressure range.Measurements were performed for the samples without Ar ion sputtering (as received), as well as after a 45-minute light Ar + ion sputtering in order to remove most of the superficial carbon contamination.
UV-vis diffuse reflectance spectra of the samples in the wavelength range 200-900 nm were obtained using a UV-2100 Shimadzu instrument.The widths of the effective band gap of the samples were determined using Kubelka-Munk phenomenological theory [23].
The photocatalytic activity of the powders was evaluated by photocatalytic oxidation of acetone (CH 3 COCH 3 ) using a 7L reactor (INNOVA).An amount of 0.3 g of each powder was used for the preparation of aqueous suspensions which were placed in petri dishes and dried at 100 • C for 2 hours.The samples were placed into the reactor where acetone was injected.The initial humidity in the reactor was controlled by dryer containing CaCl 2 .The acetone vapour was allowed to reach adsorption equilibrium with the photocatalyst prior measurements.Its initial concentration was about 400 ppm.A 15 W UV lamp (Cole Parmer Instr.Co., Chicago, Ill, USA) was used for illumination.The illumination time applied was 60 minutes.The amounts of acetone, carbon dioxide, and water vapour in the reactor were monitored.During the photocatalytic reaction, the acetone concentration decreased steadily.The photocatalytic activity of the powders was quantitatively evaluated by comparing the apparent reaction rate constants k a of the decomposition reactions given that the degradation of acetone is known to be a reaction of pseudo first order [19].

RESULTS AND DISCUSSION
The TG, DTG, and DTA curves of undoped and F-doped TiO 2 powders are presented in Figure 1.For the undoped TiO 2 (Figure 1(a)), it can be observed that the total loss of weight (∼25%) took place in three not well-defined steps.The first is the main loss of weight (∼22.3%) that took place between ∼30 and ∼180 • C and it was related to the evaporation of water and alcohol included in the xerogel.The second (∼2.5%) and the third (∼0.4%) weak losses took place between ∼180-430 • C and ∼430-675 • C, respectively, during which the remained water and organics (∼2.9%) were gradually removed from the powders.The DTA curve reveals a weak exothermic phenomenon from 400 • C to 672 • C and it was connected to the process of crystallization of anatase.The endothermic effect after 670 • C and the shoulder at 732 • C were attributed to the process of anatase to rutile transformation.The DTA, TG, and DTG curves of the TF10 fluorinated xerogel are presented in Figure 1 results in complication of TG and DTA curves in comparison with the undoped sample.The mass decrease of the samples TCF10 took place in three steps.The first step of about 7.3% took place in the range of 30 to 150 • C and the second step of about 17% took place from 150 to 350 • C. The latter was connected to the decomposition and removal of organic radicals (CF 3 COO − ).In addition, in the range from about 390 to 700 • C, mass loss of about 3% took place that can be ascribed to the release of less volatile components and reaction products.Beyond ∼700 • C the DTA curve exhibits endothermic tendency which might be attributed to the anatase to rutile transformation.The thermal behavior of TF20 xerogel (not given) was similar to the TF10 sample.The X-ray diffraction patterns of the calcinated at 400, 500, and 600 • C samples are presented in Figure 2. In our study, pure anatase was recorded for the undoped titania (Figure 2        the crystallites (d, nm), as well as the anatase/rutile phase composition (W, %) of the powders are presented in Table 1.From the results displayed in Figure 2 and Table 1, it is evident that the crystallization process of TiO 2 depends on the presence and the quantity of the fluorine precursor, as well as the calcination temperature.The increase in the size of the crystallites with the increase of the calcination temperatures can be observed for all the samples.It is also evident that the polymorph phase composition is dependent on the quantity of the CF 3 COOH in the starting solution.It can be deduced that fluorine-doped titania powders with controlled anatase/rutile phase composition can be synthesized by altering the molar ratio of CF 3 COOH/TEOTi in the starting solution and the calcination temperature.
The influence of the F − ions on the crystallization process of TiO 2 has been studied and improved anatase crystallinity upon F-doping has been reported [14,18].Up to our knowledge, there is no report on direct synthesis and fluorination of TiO 2 powder with controlled anatase/rutile content.Moreover, the increase of the F/Ti molar ratio in the starting solution (sample TF20) leads to formation of pure rutile even at 400 • C. The crystallization of rutile alone or along with anatase phase at low temperatures is attributed to the acidic character of the fluorine precursor which provides low pH conditions during peptization.At these acidic conditions, the edge-shared bonding among the [TiO 6 ] octahedrons which is responsible for anatase formation during calcination is suppressed, while the corner-shared bonding that is responsible for the rutile formation is favored [21].
Nitrogen adsorption-desorption isotherm measurements were performed to determine the specific surface area, the porosity, and the pore size distribution in the samples.The N 2 adsorption-desorption isotherms for the undoped TiO 2 and the fluorinated samples calcinated at 400 • C clearly show a hysteresis in the desorption branch (Figure 3).These samples exhibited the highest BET surface area.Their pore size distribution curves obtained from the desorption branch are depicted in Figure 4. From the figures and the results presented in Table 2, it can be observed that the BET surface areas decrease significantly while the pore size distribution becomes wider with the increase of the calcination temperature for all samples.A decrease in the International Journal of Photoenergy porosity with the increase of the calcination temperature is also evident.The BET values for the fluorinated powders are much lower than those for the undoped TiO 2 powders when calcinated at the same temperatures.These findings are connected to the presence of fluorine containing organic precursor in the xerogels and the exothermic phenomena of its burning occurring during calcination.The heat release leads to the agglomeration of TiO 2 nanoparticles and to the decrease of the BET surface area.The process contributes to the improvement of the TiO 2 crystallinity as it was revealed by XRD results.
The chemical composition of the sample TF20 calcinated at 400 • C was studied by XPS analysis.After the removal of the superficial carbon contaminant (3-4 monolayers), the elements detected were titanium, oxygen, and fluorine.The spectra obtained for Ti 2p, O 1s, and F 1s regions (Figure 5) revealed that the powder consists of TiO 2 , where titanium and oxygen are present as Ti 4+ and O 2− .The valence states of the titanium and the oxygen were determined from the Ti 2p3/2-Ti 2p1/2 splitting (∼5.7 eV) and the difference of ∼71.5 eV between the binding energies of the Ti 2p3/2 peak (458.5 ± 0.1 eV) and the O 1s peak (530.0 ± 0.1 eV).The F 1s spectrum (Figure 5(c)) is nonsymmetric and the contribution near 684.3 eV indicates that the fluorine is presented in the sample mainly as TiF 4 [24] and/or to physically adsorbed F − on TiO 2 [19].The contribution near 685.4 eV could be related to TiOF 2 structures [12] and the tail near 688 eV to nonstoichiometric solid solution of F in TiO 2 of the TiO 2−x F x type [12,19].It can be also associated with C-F bonding in CF 3 residual groups [25].The fluorine concentration was estimated as ∼14 at % before and ∼16 at % after the Ar + ions sputtering.This level of doping can be connected to the high temperature of decomposition of the fluorine precursor which is evident from the thermal analysis.
The diffuse reflectance spectra (R) of the samples were used to assess the absorption of the photocatalysts in the UVvis region.The undoped TF0 sample and the TF20 samples revealed typical anatase and rutile spectra, respectively, and for this reason they are not shown.The absorption curves of the fluorinated TF10 samples with various anatase/rutile ratios are presented in Figure 6(a).The P25 absorption curve is also presented for comparison because of its mixed phase composition (anatase ∼75% and rutile ∼25%).To determine the band gap of the powders, the Kubelka-Munk method based on the diffuse reflectance spectra was employed.The (F(R)hv) 1/2 versus hv plots of the P25 and the TF10 samples are presented in Figure 6(b), where F(R) = (1−R)/2R.It can be observed that the band gap values obtained for the Degussa P-25 and the TF10 sample calcinated at 400 • C are close (3.08 eV and 3.07 eV, resp.).This fact can be connected to the similar anatase/rutile phase content of the samples.This is also consistent with the literature [11,12,19] and support the conception that the fluorine doping does not affect the band gap of TiO 2 .With the increase of the calcination temperature, the widths of the band gap for the fluorinated samples decrease from 3.07 to 3.03 eV.This "narrowing" of the effective band gap is attributed to the increase of the rutile content in the powders which is in accordance with our XRD results.The increase in the CO 2 concentration due to the decomposition of acetone and the calculated apparent rate constant of the photocatalytic reaction are presented in Figures 7 and 8, respectively.From our results, it can be concluded that the fluorinated TiO 2 powders exhibit enhanced activity in comparison to the reference photocatalyst Degussa P25.The best results are recorded for the powders calcinated at 400 • C which can be connected to the mixed anatase/rutile phase composition and the higher BET surface area in comparison to the samples calcinated at 500 and 600 • C. In addition, the activity of TF20 samples treated at 400 and 500 • C exceeds significantly the activity of the corresponding TF10 powders.This fact was attributed to the rutile rich composition of the powders and the higher level of fluorine doping.The drastic decrease in the activity of the treated at 600 • C samples is attributed to the decrease of the BET surface area, as well as to the absence of anatase phase in the powders.

CONCLUSIONS
Simultaneous fluorination and controlled low-temperature anatase/rutile phase crystallization of nanosized TiO 2 powders were achieved using sol-gel route.It was found that the anatase/rutile crystalline ratio depends on the amount of fluorine precursor used for fluorination as well as on the temperature of thermal treatment.The crystallization of pure rutile or anatase/rutile mixture at low temperatures is attributed to the acidic character of the fluorine precursor.Improved crystallinity and enhanced absorbance of the powders in the UV-vis region are recorded upon F-doping.The fluorine is presented in the powders mainly as metal (Ti) fluoride.
The width of the TiO 2 band gap was not affected by the presence of fluorine.The red shift of the absorption edge is attributed to the increased rutile content in the fluorine doped TiO 2 powders.The prepared fluorine doped TiO 2 powders calcinated at 400 • C exhibited higher photocatalytic activity than that of Degussa P-25 for decomposition of acetone under UV-vis illumination.
(a)) after thermal treatment at every temperature mentioned above.The XRD patterns of F-doped TiO 2 powders (Figures2(b) and 2(c)) revealed that a second crystalline phase, that of rutile, was developed along with the anatase phase at temperatures as low as 400 • C. The average sizes of

Figure 4 :
Figure 4: Pore size distribution in the undoped TiO 2 calcinated at different temperatures (a) and in the TF10 and TF20 powders calcinated at 400 • C (b).

Figure 7 :
Figure 7: Increase of the CO 2 concentration due to decomposition of acetone for fluorinated TF10 (a) and TF20 (b) titania powders.

Table 1 :
Phase content W (%) and average diameter d (nm) of the crystallites in the TiO 2 powders calcinated at different temperatures (A: Anatase, R: Rutile).

Table 2 :
BET surface area and porosity of TiO 2 powders calcinated at different temperatures.