UV photodetectors have attracted considerable attention in recent years due to their application in environmental, biological analysis, optical communication, flam detection, astronomy lithography and detection of missiles^[1,2]. Transparent conductive oxide (TCO) films such as tin oxide (SnO₂), zinc oxide (ZnO) and titanium oxide (TiO₂) are frequently utilized for the UV photodetectors application^[3–6]. In particular, most attention concerned tin oxide (SnO₂) semiconductor due to its special properties, which made it a required material for optical and optoelectronic applications^[7,8] such as wide band gap (3.6–4 eV)^[9,10], large exciton binding energy (about 131 eV)^[11], high optical transmittance^[12], chemical stability and n-type character due to native defects such as oxygen vacancies and tin interstitials, which can exist in two possible valence states: Sn²⁺ and Sn^4+[13]. A notable defect in SnO₂ and most metal oxides are oxygen vacancies^[14]. These defects may behave like traps for free electrons or holes, such as a recombination or generation center, which strongly influence the photoconductivity and photoresponse process^[15]. However, persistent photoconductivity and big recovery times are problematic characteristics due to deep traps^[16]. Information about these traps levels can be obtained from the decay curve of photoconductivity after cutoff of excitation^[17,18].

Previous studies^[19–22] demonstrated that the control of the concentration of oxygen vacancy by doping with different elements such as magnesium^[19], nickel^[20], tin^[21] and aluminum^[22–24] is among the ways to improve the performance of UV photodetectors in metal oxides^[25,26]. The doping process such Al, change the type of SnO₂ from n to p by substitution of Sn⁴⁺ by Al³⁺ ion and causing a disorder and some scattering centers, resulting in an increase in hole concentration with increasing the Al content^[27,28]. Also, the method and conditions of synthesis have a basic role in UV photodetectors properties such as solution concentration^[25], thin film thickness^[26], annealing temperature and time, the photoresponse being profoundly augmented with increasing annealing time^[29].

Elaboration conditions have a big influence on the properties and qualities of layers. SnO₂ thin films can be obtained by numerous methods of deposition^[30–33], among them, sol-gel dip coating technique presents several advantages such as simplicity, excellent homogeneity, the possibility of ease doping, low cost, large area substrate coatingand low reaction temperature. Moreover, Al-doped SnO₂ thin films can be obtained easily using this technique, which makes them very attractive in many applications such as gas sensors^[34] and solar cells^[35]. The thin films photoconductivity properties based on pure and doped SnO₂ have been studied by a number of research such as Sb-doped SnO₂^[36], As-doped SnO₂^[37], Al-doped SnO₂^[38], and the effect of twin boundaries on photocurrent decay of the pure SnO₂^[39]. However, the studies about the UV photoconductivity and photosensitivity in Al-doped SnO₂ thin films prepared by sol-gel dip coating are not enough.

In this work, we prepared Al-doped SnO₂, using sol-gel process. The influence of aluminum concentration on the performance of UV photodetectors based on SnO₂ thin films had been studied.

Using a sol-gel dip coating method, undoped and aluminum doped SnO₂ thin films were successfully deposited on glass substrates. Tin (II) chloride (SnCl₂·2H₂O) was dissolved in 30 mL of ethanol absolute (C₂H₆O), then a few drops of HCl were added to the solution for accelerating the hydrolysis reaction between the precursor and the solvent. Aluminum trichloridehexahydrate (AlCl₃·6H₂O), was added to the solution for Al doping (3 wt.%, and 5 wt.%). After vigorously stirring at 70 °C for 120 min, we obtained a homogenous solution. On the other hand, the glass substrate was cleaned with acetone, ethanol and deionized water for 10 min in the ultrasonic bath and then dried at room temperature. The substrates were immersed in the solution for 1 min with a withdrawal speed of 60 mm/min, and next dried at 200 °C for 10 min. After repeating this procedure 10 times, the films were annealed at 550 °C for 2 h.Fig. 1 shows the process of synthesizing pure and Al-doped SnO₂ thin films using the sol-gel process.

X-ray diffraction (XRD, Burker AXS-8D) with CuKα (λ = 1.541 Å) radiation was used to study the structural properties. The surface morphology of the SnO₂ films was performed by atomic force microscopy (A100-AFM). UV-visible spectrophotometer (JascoV-360) was used to study the optical properties in the range (200–1100 nm) to determine the band gap of samples.

Pure and Al-doped SnO₂ based UV photodetection measurements were performed using Ag electrodes in a planar interdigital configuration as shown inFig. 2. The Keithley (2401) source meter was used to measure the current–voltage (I–V) characteristics in dark and under UV illumination to determine the UV photoresponse. Illumination was performed by a UV lamp (VL-4LC, vilberlourmat) available in two wavelengths: 254 and 365 nm, and an intensity of 350µW/cm².

X-ray diffraction patterns for undoped and Al-doped SnO₂ thin films are shown inFig. 3. These films exhibit a polycrystalline nature in a tetragonal rutile structure (JCPDS 41-1445). No peaks related to Al or Al₂O₃were detected in the XRD pattern this may be caused to the low Al content and the Al³⁺ ions have been replaced by Sn⁴⁺ site without changing tetragonal structure^[40]. In previous studies, the SnO₂ thin films doped with Al and prepared by sol-gel don’t detect secondary phases related to Al doping^[41]. It can be seen that the (110) plane is the strongest diffraction peak. Other less-intense peaks are observed at (101), (200) and (210) planes. In addition, the intensity of the diffraction peaks decreases with increasing Al concentration, the augmentation of which reduces the film thickness. These results indicate to uniform distribution of Al ions across the SnO₂ lattice^[42] and agree with some reports^[41,43].

The crystallite size (D) can be calculated from the Scherrer's equation^[7]:

1D=Kλβcosθ,

whereK is a constant (shape factor, about 0.9),λ is the X-ray wavelength,β is the full width at half maximum of the XRD peak,θ is the Bragg diffraction angle.

The (110) peak was utilized to estimate the crystallite size. It can be noticed fromTable 1 that the crystallite size of the films decreases from 11.92 to 8.54 nm when Al concentration increases from 0 to 5 wt.% concentration. This proves that the incorporation of Al³⁺ into SnO₂ structure obstructs crystallization and prevents crystal growth^[44].

To calculate the lattice parametersa=bandc, the following equations have been used^[45]:

21dhkl2=h2+k2a2+l2c2,

3a=b=2d(hkl)=2d(110),

4c=11d(101)2−1a2.

FromTable 1, the lattice parameters and unit cell volume of deposited SnO₂ thin films correspond with the perfect valuesa =b = 4.738 Å,c = 3.186 Å andV = 71.5 Å. Furthermore, the lattice parameter increases from 4.70 to 4.742 Å when Al concentration increases from 0 to 3 wt.% and then decreases to 4.738 Å when Al concentration is equal to 5 wt.%. This change in a random manner in lattice parametersa andb may be due to a disturbance in the grains of the film after doping with Aluminum^[46].

Moreover, the strain(ϵ) of the films was estimated using the following relation^[40]:

5ϵ=βcosθ4.

The change of strain in the SnO₂ thin films observed inTable 1 can be explained by the variation in dopant concentration. The strain increases with increasing Al concentration because the incorporation, into the SnO₂ lattice of Al, which has an ionic radius (0.53 Å) smaller than Sn (0.69 Å), leads to defects and lattice distortions, resulting in a different type of stress. The mean crystallite size is estimated to be ~ 11.92 nm for the undoped sample and it shows a decreasing tendency as the Al content is increased. This behavior can be attributed to the change in lattice parameters^[47] and the increase of strain in the thin films that affect the normal growth of SnO₂, these results generally agree with the changes observed in the lattice parameters. Related results have been noted in the literature for Ni-doped SnO₂^[48] and Ti-doped SnO₂^[47]. The solubility of dopants is basically dependent on valence state and ionic radius value.

The calculation of the texture coefficientT_c(hkl) allows us to deduce the preferred orientation of growth via the following formula^[46]:

6Tc(hkl)=I(hkl)/I0(hkl)1n(∑I(hkl)/I0(hkl)),

whereI(hkl) is the measured intensity of the(hkl) peak,I0(hkl) is the standard intensity of(hkl) peak corresponding to the JCPDS data and n is the diffraction peaks observed.

Table 2 shows the texture coefficient values for (110), (101), (200) and (210) peaks. The (110) diffraction peak presented a high texture coefficient value for all deposited films, which means the (110) plan is the preferred orientation, this confirmed the XRD patterns results. The origin of preferential orientation along (110) can be interpreted by the periodic bond chain (PBC) theory^[46]. Via this theory, the SnO₂ crystals faces can be divided into flat (F), stepped (S) and kinked (K) according to 2, 1 and none number respectively of PBC parallel to the faces(hkl). Forms F and K are of great interest for cassiterite of SnO₂. Since they consist of (101) and (111) crystal planes, respectively^[49]. For the crystal grains formed by (101) F faces are two possible preferred orientations, namely (110) and (001). Since the (001) orientation was not observed by XRD, the (110) preferential orientation was formed by the (101) F faces^[50]. These results are similar to some previous reports^[51].

For analyzing the surface morphology of SnO₂ thin films doped with different Al doping concentrations, the atomic force microscope was used. The area of 3 × 3μm² was scanned. From 3D AFM images, which are obtained by high scan mode and illustrated inFig. 4. It can be clearly seen that the surface morphology of pure film is very different from Al-doped SnO₂ thin films.

The values of the root mean square roughness (RMS), the average roughness (R_a) and the grain size (D_a) for all films were evaluated using Gwyddion 2.60 program and reported inTable 1. It can be noted that the surface roughness is strongly dependent on the Al concentration. The values of RMS andR_a decreases from 5.06 to 2.01 nm and from 4.25 to 1.67 nm when Al concentration increases from 0 to 5 wt.%. The low values of RMS andR_a for prepared films are similar to the result which has been obtained from the In-doped SnO₂ thin films deposited with sol-gel spin coating technique^[52] and from Zr-doped SnO₂ thin films synthesized by spray pyrolysis^[53]. As for grain size, it also decreases with the increase in Al⁺³ ions concentration and it is in good agreement with the XRD results. As well, the size of the grains estimated from AFM is bigger than that determined through XRD. This is due to the fact that AFM reveals particle agglomerations, whereas XRD provides an average grain size^[53].

The optical transmission spectra of undoped and Al-doped tin oxide thin films elaborated by sol-gel dip coating technique on a glass substrate and annealed at 550 °C are plotted inFig. 5. All films display high transmittance greater than 80% in the 400–1100 nm interval of wavelength. According to Ref. [54], the low transmittance for Sn-doped films may be due to the increased optical scattering caused by the rough surface morphology.

From the observation of interference fringes, one can deduce that the film thickness decreases as Al concentration augments. The band gap energy values of Al-doped tin oxide thin films were calculated by using the Tauc equation^[55]:

7αhυ=A(hυ−Eg)1/2,

whereα is the absorption coefficient,A is constant,hυ is photon energy andEg is the band gap energy. The results are shown inFig. 6. The band gap value increases from 3.88 to 3.91 eV when Al concentration increases from 0 to 5 wt.%. Similar results have been reported by Ahmed^[41] and Baghri-Mohagheghi^[56]. The increase in gap energy after adding Al with 3 wt.% and 5 wt.% ratio may be due to a reduction in particle size and Burstein-Moos effect^[57].

Variation of photo and dark current with a voltage of undoped and Al-doped SnO₂ thin films in log-log scale as shown inFig. 7. It can be noticed that the curves are straight lines having a different slope with regard to different voltage according to the power law relationI∝V^r, wherer is the slope of different straight line segments^[18,29]. In the absence of illumination (Fig. 7(a)), the log-log plot ofI–V characteristics of pure and the sample doped with 3 wt.% aluminum can be divided into two separate regions. For pure SnO₂, the dark current varies super linearly (r = 1.24) at a voltage below 3.5 V. Above 3.5 V, it can show space charge limited current (SCLC) behavior (r = 3.04). It is significant that at the high voltage the dark current for undoped SnO₂ thin films engenders from space charge of excess carriers injected from one of the electrodes and the traps of materials also contribute to this behavior^[58,59]. The SCLC phenomena is reported by Aldemiret al. for photodiode based on Al-doped SnO₂ thin films prepared using spray pyrolysis^[38]. However, sub-linearly variation (r = 0.64) at the voltage below 3.5 V varies to super linearly variation (r = 1.77) at the voltage above 3.5 V is recorded with the sample doped with 3 wt.% of aluminum, the flow of trap limited and space charge limited current inside the material is responsible for this type of variation^[58,17]. The dark current for the sample doped with 5 wt.% Al varies sub-linearly (r = 0.83 < 1), the sublinear variation with applied voltage in this sample could be because of the emergence of blocking contacts that do not entirely refill the charge carriers after they are captured by electrodes^[58,60].

The dependence of photocurrent with applied voltage inFig. 7(b) shows super-linear behavior for all samples, the super-linear behavior of the dark current and photocurrent may be due to additional charge carriers being injected from one of the electrodes^[61].

Fig. 8 shows the photoconductivity rise and decay time spectra of undoped and Al-doped SnO₂ thin films. The measurements were obtained under 5 V bias voltages and UV light lamp with wavelength 365 nm at room temperature. One observation we can remark, the dark current (I_OFF) and photocurrent (I_ON) decrease with increased Al ions concentration. Furthermore, for pure and 5 wt.% Al-SnO₂ samples, when the light is turned on, the current increased quickly in the beginning, then it continues to slowly increase until the UV light is turned off again. As for 3 wt.% Al-SnO₂ sample, when the UV illumination is turned on, the current begins to increase until it reaches maximum value, then it decreases until the UV radiation turned off. The anomalous photocurrent behavior for 3 wt.% Al-SnO₂ may be due to the recombination of photo generation carriers^[62]. Similar retreat photoconductivity behavior was reported for SnO₂ nanoparticles that have been synthesized using the chemical co-precipitation method^[62] and for 4 wt.% In-doped ZnO sol-gel spin coated film^[63]. Another hand, when UV illumination was turned off, the current initially decrease very fast, later it continues to decrease gradually; the persistence effect photoconductivity phenomenon was reduced with the samples doped with aluminum.

In order to more explain, it is necessary to understand the mechanism of UV photoconductivity response in pure and Al-doped SnO₂ thin films.Fig. 9 schematizes the mechanism of UV photoconductivity for SnO₂ thin films. In the dark current, the molecules of oxygenO2(g) are stuck in the surface of thin films with an adsorption process by capturing free electrons; this produces a depletion region near the surface^[63], which leads to a decrease in the current at the surface of Al-SnO₂ thin films^[64]. The following reaction exemplifies this process.

8O2(g)+e−→O2−.

In the light condition, the electron–hole pairs produced by absorption of photon have at least equal energy as the band gap of Al-SnO₂ thin films. Moreover, the oxygen ionsO2− will recombine with the generated holes to be chemisorbed from the surface. This reduces the depletion layer near the surface of SnO₂ thin films and it allows to the photo generatedelectron and the free electron that generated by chemisorbed step migrate to conduction band and contributed in photosensitivity process when applying the bias voltage^[65].

9hν→e−+h+,

10O2−+h+→O2(g).

After returning to the dark again, the photocurrent decrease, due to recombination of the (e–h) pairs which was generated by absorption process and the repetition adsorption of oxygen^[66].

The sensitivity (S) orI_ON/I_OFF ratio is an important parameter for describing the performance of photodetectors. The film of 3 wt.% Al-SnO₂ shows the highest value of sensitivity and it was estimated at 273.85, while for the film 5 wt.% Al-SnO₂ decreased to 151.38.

Other parameters play a crucial role in the photodetector as the rise and decay time.Fig. 10 shows that using the exponential and Bi-exponential functions, the rise and decay time are approximated by the fitting curve.

11Iph=I0+A1e−t/tr,

12Iph=Iph(∞)+A2e−t/td1+A3e−t/td2,

whereI0 is the dark current,A1,A2,A3 are constants,tr is the rise time andtd1 andtd2 are the decay time constants respectively for a fast and a slow decrease of photocurrent. The fast decrease was attributed to recombination of photogenerated electron–hole pairs process, whereas the slow decrease was attributed to readsorption process^[67].

Table 3 shows the values of constant times for our films and other references. The 3 wt.% Al-SnO₂ shows the best rise and decay time constants values. There is improvement in the photoconductivity parameters as rise time, decay time andI_ON/I_OFF ratio for our films compare to previous values presented in the literature^[39,68,69].

Electron trap depth is calculated using the decay portion of the rise and decay curve when the current of decay is expressed by the bub model^[3,70]:

13I=I0e(−Pt),

whereI0 is the current when the light is turned off,I is the photocurrent versus time andP is the probability of escape of an electron from the trap persecond. The probability of an electron escaping from a trap has been described by Randal and Wilkins as^[71]:

14P=Se−E/KBT,

whereT is the absolute temperature,KB is the Boltzmann constant andS is the attempt to escape frequency (equal to 109 at room temperature).

The trap depth corresponding to different exponentials can be calculated by using the below equation^[18].

15E=KBT(lnS−lnlnI0/It).

The values of trap depths (E) for undoped and Al-doped SnO₂ are listed inTable 3. The changes in the trap depth values reveal the defect and disorder that occurs in thin films structure. For our deposited films, trap depth energy decreases with an increase in Al ratio from 0 to 5 wt.%, this indicates that the stability and idealism of crystalline structure decrease with increased Al concentration^[72], and leads to a reduction in the energy required to eliminate an electron from trap level^[73]. It is confirmed by XRD results, as the strain values increase with increasing Al concentration. The values of traps depth calculated for our deposited films is between 0.579 and 0.640 eV which is greater than the previous reported values for Europium doped SnO₂ nanoparticles prepared by the co-precipitation method which is 0.33 and 0.35 eV^[74], furthermore, our obtained values are very convergent to the taps depth values of ZnO thin films elaborated by sol-gel spin-coating technique^[61].

In the present work, we have studied the photoconductivity and defect levels in undoped and Al-doped SnO₂ thin films elaborated using low-cost sol-gel method. Films properties were investigated by X-ray diffraction (XRD), Atomic force microscopy (AFM) and UV-visible spectrophotometry. Presented results demonstrated that our samples show a polycrystalline tetragonal rutile structure and high homogenous surface; moreover, prepared films show high transparency in a visible wavelength region with an average transmittance between 80% and 90%. For photo-detection studies, 3 wt.% Al-doped SnO₂sample shows the highestI_ON/I_OFF ratio and the best rise and decay time parameters. The photoconductivity mechanism was interpreted by adsorption and desorption phenomena and the defect energy levels were calculated using the decay portion curve of photoconductivity spectra.

Through this study, we reached the possibility of fabrication of a UV photodetector with high sensitivity, fast response and good physical properties based on 3 wt.% Al-doped SnO₂ thin films using the sol-gel dip-coating technique.