Undoped and F doped SnO₂ films were deposited on microscopic glass substrates by sol-gel method using dip-coating technique. The salt precursor (SnCl₂・2H₂O) was dissolved into 5 ml of hydrochloric acid (HCl) and heated at 60˚C for 10 min. 25 ml of ethanol was added to obtain a solution of 0.1 M. In order to promote the gel formation, some drops of Triethanolamine (TEA) are added to the prepared solution. The added NH₄F into pre-mixed solution is the source of fluorine doping atoms. The resulting solutions were stirred for 2 hours at 60˚C, a clear and homogeneous solution was obtained, it was left to age for 24 h at room temperature resulting in a gel-like structure. The immersed glass substrates were withdrawn vertically from the solution at the rate of 8 cm/min. For film formation with suitable thickness, 15 dipping were repeated, between two successive dipping the deposited films were dried at 100˚C and finally annealed at 450˚C in air for 90 minutes.

Films structural properties were determined by XRD using Philips X’Pert system under CuKα (λ_Cu = 0.154056 nm) radiation whereas the 2θ scanning range was between 20˚ and 80˚. The films’ composition and morphology were determined by using various measurements: energy-dispersive X-ray spectroscopy (EDS) with the electron-beam energy set to 7 keV. The micro-Raman measurements were performed using a Renishaw Raman spectrometer. Fourier transformation infrared (FTIR) spectra were recorded using a Perkin-Elmer 2000 spectrometer. The optical transmission in the UV-visible range (200 - 900 nm) measurements was performed using Shimadzu UV-3101PC spectrophotometer. Films thicknesses and optical band gaps were deduced from optical transmission data fitting. The thickness calculation procedure is based on the comparison between the experimental transmittance data with the theoretical one calculated using Swanepoel theory [19] , where film thickness and refractive index are varied till convergence. The films electrical properties were studied by two probes method measurements in dark and at room temperature, in a coplanar structure formed by two golden strips electrodes deposited by DC sputtering onto films surface.

The X-ray diffraction spectra of FTO films prepared with different fluorine doping levels in the precursor solution are shown in Figure 1. As can be seen, all films are polycrystalline with (110) preferred growth orientation except the one deposited with 10 wt.% F which exhibits an amorphous state. The strong (110) peak is accompanied by other peaks such as (101), (200), (220) and (211) but with substantially lower intensities, confirms the tetragonal structure of the prepared SnO₂ films (JCPDS N˚. 41-1445) [20] . As seen, the introduction of fluorine in SnO₂ network does not affect its structure but it reduces the peak intensities probably due to the reduction of film thickness. Small peak located at 31.77˚ assigned to the (111) direction of SnO₂ cubic phase ((JCPDS 50-1429) is detected in films a and c (Figure 1). The same remark has been reported by Keskenler et al. [21] . in tungsten and fluorine co-doped SnO₂ prepared by sol-gel technique.

To evaluate the preferred growth direction, the texture coefficient TC(hkl) for each direction (hkl) was estimated using the following formula [22] .

T C ( h k l ) = I ( h k l ) I 0 ( h k l ) × [ 1 n ∑ 1 n I ( h k l ) I 0 ( h k l ) ] − 1 (1)

where I(hkl) and I₀(hkl) are respectively the measured and the standard intensity in the JCPDS card of the plane (hkl) and n is the reflection. The larger the deviation of texture coefficient from unity, the higher will be the preferred orientation.

The variation of texture coefficient estimated along (110), (101) and (211) directions with the fluorine content for SnO₂:F thin films are listed in Table 1. As can be seen, the direction (110) remains the preferential orientation in undoped or doped SnO₂, the same conclusion has been reported by several authors [23] [24] [25] .

The lattice constants “a” and “c” were determined from XRD results using the equation [26] :

1 d h k l 2 = h 2 + k 2 a 2 + l 2 c 2 (2)

where d_hkl is the inter-planar distance, (hkl) are the Miller indexes. As reported in Table 1, the calculated values are in agreement with the standard data (a₀ = 4.738 Å, c₀ = 3.187 Å). No notable change was observed with increasing F ion concentration, this may be due to the comparable ionic radius of F⁻ (1.17 Å) to O⁻² (1.22 Å).

The films crystallite size and internal lattice strain given in Table 1 have been estimated from Hall-Williamson equation [27] expressed as:

β cos θ λ = 1 D + ε sin θ λ (3)

where β is the FWHM (full wide that half-maximum) of diffraction peaks, θ is the Bragg angle, λ is the X-rays wavelength, D is the crystallite size and ε is the internal strain. D and ε are estimated from the fit of the plot βcos(θ)/λ versus sin(θ)/λ calculated for different peaks. The crystallite size is deduced from the plotted intercept with the y-axis yields to, while the strain is equal to the slope of the plot as reported in Figure 2.

The variations of the crystallite size and the internal strain are reported in Figure 3. Increasing fluorine content reduces the crystallite size from 25 to 12.5 nm and enhances the internal strain in the film network. The same trend has been reported by several authors [28] . In general, it is reported that introduction of doping atoms reduces the crystallite size regardless the host network such as ZnO, TiO₂, SnO₂ [29] [30] [31] . This may originate from the early stage of nucleation during film growth, actually, any foreign atom acts as a nucleation center, therefore the nucleation centers is increased in doped films causing the reduction of crystallite size.

The FTIR spectra of un-doped and F-doped SnO₂ films are illustrated in Figure 4. The spectrum is formed with several absorption peaks assigned to Vibrational modes of SnO₂ confirming its formation. The obtained spectra are comparable to the reported ones in the literature [32] [33] .

Peaks at low wave numbers (500 - 1000 cm⁻¹) can be attributed to SnO₂. For the un-doped film, the peaks located at 675 and 780 cm⁻¹ were assigned to O-Sn-O and Sn-O stretching vibrations modes, respectively [34] .

An additional peak located at 804 cm⁻¹, which becomes inconspicuous as fluorine concentration increases can be assigned to oxygen vacancies (V²⁺ ) presence in the O-Sn-O group. The oxygen vacancies modify the symmetrical O-Sn-O group to O-Sn- V o 2 + resulting in the weakening of O-Sn-O vibration due to the different bond lengths (O-Sn-O: 2.597 Å, Sn-O: 2.053 Å) [35] . Moreover, it is interesting to note that a weak peak near 408 - 410 cm⁻¹ is observed in highly fluorine doped films. This peak is assigned to Sn-F vibration frequency

[35] [36] . This indicates that few amounts of fluorine are introduced in SnO₂ film despite the large ratio of fluorine in the starting solution. The absence of this peak in un-doped one and in F (10 wt.%) doped SnO₂ films suggests that fluorine atom is not incorporated in SnO₂ network at low doping level which is confirmed by EDS analysis.

The vibration frequencies of O-Sn-O display a shift from 780 to 760 cm⁻¹ and a decrease in the intensity with fluorine concentration increase in the solution as shown in Figure 5. This shift can be ascribed to the increase in lattice deformation due to the interstitial fluorine ions F i − in the lattice. Doping could allow the fluoride ions to occupy beside the oxygen sites, the interstitial positions,

causing a disorder in the film network this is confirmed by the optical measurements.

The formation of undoped and F doped SnO₂ films might be achieved through the following reactions [37] [38] :

・In the solution:

・

・ For un-doped SnO₂ deposition:

S n ( O H ) 2 + 1 2 O 2 → S n O 2 + H 2 O ( g ) (5)

An analysis of the precipitate, formed in the bath during solution film formation, was performed by FTIR spectroscopy, in order to investigate the fluorine atoms presence in the solution and in SnO₂ lattice and to have an insight about the favorable fluorine bonding in the solution or in the film. Therefore, the precipitate resulted from solution contained 40 wt.% fluorine was analyzed by infrared spectroscopy.

The FTIR spectrum of this powder is reported in Figure 6. Strong peaks located at 1250, 1745 and 3540 cm⁻¹ attributed to H-F bond vibration modes [39] are present in the spectrum indicating that a large amount of fluorine remains in the solution and do not incorporate in the solid film. In addition, according to the formation enthalpy of H-F and Sn-F bonds, that are respectively Δ H f ° = −569.7 and −476.7 Kj/mol, H-F is easily formed than Sn-F. Hence the affinity of fluorine to bonding hydrogen in solution is larger than to bonding tin (Sn) in films. Consequently, little quantity of fluorine ions is incorporated in the films

during dip coating deposition. Moreover, the formation of FTO depends also on the mobility of fluorine ions in the solution and its availability on the reacting substrate surface. Thereafter, this explains, the absence of fluorine in films at low doping level and the inferiority of the electrical conductivity in FTO films prepared by sol-gel by comparison to the other techniques as mentioned below.

Several bands assigned to the compound NH₄Cl are present in the spectrum [40] this confirm the formation of this compound according to the reaction (4).

Raman scattering is generally used to investigate the material crystallinity, structural defects and side effects of crystallites. Raman spectrum of unhoped and F-doped SnO₂ thin films are shown in Figure 7 presents the room temperature. As can be seen, three fundamental Raman peaks located at 634, 475 cm⁻¹ correspond respectively to A_1g and B_2g models are present in the different spectrum, they are assigned to the symmetric and asymmetric stretching of Sn-O bonds [40] , these are reported in the literature [41] [42] . The appearance of these modes confirms the cassiterite tetragonal structure of SnO₂ as deduced from XRD analysis [43] . A shift in the A_1g mode towards the lower wave number, accompanied with broadening of the Raman bands is observed in SnO₂:F, could arise from the strain in the crystallite [44] and the disorder in the film network. This suggests that the fluorine incorporation is accompanied by the disorder formation in the film structure as confirmed as well by the shift of the O-Sn-O vibration frequency in FTIR spectrum.

Compositional analysis of the elements present in the films was performed on different samples; a typical recorded EDS spectrum is shown in Figure 8. In

Table 2 we have reported the elementary composition of different films. The most important feature is the absence of fluorine in supposed 10 wt.% F doped SnO₂. While at higher doping level 40 wt.%, few amounts of fluorine is incorporated in SnO₂ films network. This indicates that the fluorine was successfully incorporated into the tin oxide network but with a small amount. It is worth noting that the fluorine content in the films was very low compared with the initial percentage in the starting solution. This is due, as suggested from FTIR analysis, to the segregation of fluorine and its preference to remain in the solution rather than to be introduced in the film network. The amount of oxygen in the films is overestimated due to the oxygen originating from the glass substrate. However,

the alkali elements, such as Ca, K, Mg, Na and Si originate from the used glass substrate due to the low film thickness.

The optical transmission spectra of FTO films deposited with different fluorine concentrations are shown in Figure 9. The average transmittance in the visible range was reduced from 65% to 60% with increasing the doping level. A shift in the absorption edge towards the higher photon energy was observed with fluorine doping at (40 wt.%), this may be attributed to the Burstein-Moss shift due to the increase in free electrons concentration [45] [46] [47] .

Films optical band gaps are estimated, as shown in Figure 10, from the plot of (αhν)² versus photon energy (hν), according to Tauc formula for direct band gap semi-conductors [48] :

( α h υ ) 2 = B ( E g − h υ ) (6)

where α is the absorption coefficient, B is a constant, h is Planck’s constant, Eg is the energy band gap and ν is the photon frequency.

Due to the disorder in film network, localized states appear near the bands edges causing band tails formation. The states are responsible for the low energy photons absorption. The absorption coefficient in the band tail expressed as [49] :

α ( h υ ) = α 0 exp ( h υ E 00 ) (7)

where α₀ is the pre-exponential factor, hν the photon energy and E₀₀ ( known also as Urbach energy) is the bandtail energy width [50] . E₀₀ is estimated from the inverse slope of ln(α) versus (hν) linear plot.

The direct band-gap values obtained in this work are ranged from 3.3 to 3.7 eV (Table 3), this is approximately the same values reported in [51] - [56] . Figure 11 shows the variations of the band gap and band tail width E₀₀ with various fluorine doping levels. The Urbach energy rises with fluorine doping indicating that doping atoms induce disorder in films network [50] . This is consistent with band broadening observed in Raman spectra (Figure 7) and the strain increase with fluorine concentration as reported in Table 1. An increase in E₀₀ causes the optical band gap shrinking. This suggests that the optical gap is controlled by the disorder in the film network rather than by free electrons due to the doping effect.

The electrical properties of films were characterized by tow probes technique. The results are reported in Table 3 and Figure 12, they are comparable to the reported ones in films prepared by sol-gel technique [24] [25] . Clearly, the resistivity decrease, by two orders of decade, with increasing F-doping levels as shown in (Figure 12). This reduction confirms the incorporation of fluorine in tin oxide; each F⁻ anion substitutes an O²⁻ anion in the lattice. This yields to an

increase in free electrons concentration and to a decreases in films resistivity. However, as reported in Table 4, the resistivity of FTO films prepared by sol-gel is relatively high by comparison to their values in the ones prepared by other techniques. This discrepancy is mainly due to the low amount doping fluorine hosted in the SnO₂ network as concluded from Infrared spectroscopy and films composition analysis.

Once the electrical resistivity and the optical transmission T of films are known, the figure of merit can be calculated using Haacke’s equation [52] :

F = T 10 R s h (8)

where R_sh is sheet resistance defined as the film resistivity divided by the film thickness. The obtained figure of merit values at different wavelengths are reported in Table 5. These values are close to the reported ones in sol-gel deposited FTO [53] .