Improved photodetector performance of SnO₂ nanowire by optimized air annealing

Figure 1 shows the XRD analysis of SnO₂ NWs measured at room temperature for different annealed samples at (350 °C, 550 °C, 650 °C, 750 °C, and 900 °C). The as-deposited GLAD SnO₂ NWs was found to be amorphous in nature (not shown in figure 1) which was already reported by the authors recently in [11]. Thereafter, on annealing, the sample at 350 °C the sample still retain its amorphous behaviour, while on increasing the annealing temperature above it, the diffraction peak intensities become sharper. This may be due to defect healing and grain growth induced by annealing. All the SnO₂ NW samples exhibited common diffraction peaks for SnO₂ at angles of 2Ө = 26.5°, 33.9°, 37°, 51.6°, 66° which are assigned to (110), (101), (200), (211) and (301) planes respectively obtained from JCPDS (88-0287) for tetragonal structure. Also, a peak at 23.5° was obtained for SnO of orientation (111) for 350 °C, 550 °C, 750 °C, and 900 °C annealed samples from the JCPDS (24-1342). It is observed that the diffraction peak intensity is increased on annealing and is maximum at 650 °C. The increased intensity suggests better crystallization. Similar reports were also observed by various other researchers for different annealing temperatures [14–16]. Therefore, the crystallinity of the 650 °C is found to be better than all the other deposited samples. However, on increasing the annealing temperature beyond 650 °C, the intensity of the peak reduces. This may be due to the degradation of crystallinity as a consequence of variation in the grain boundary energy which made unusual growth of grains at higher annealed temperatures. Moreover, with higher annealing temperature, (900 °C) it is reported that an amorphous layer is generally formed (silicate layer) at the interface [1]. The presence of such an interfacial layer may be the reason for a reduction in the peak intensity at 900 °C [17]. Furthermore, from figure 1, it is seen that a small peak for SnO is observed in the case of 350 °C sample and which is reduced to a very small peak in case of 550 °C. On increasing the annealing temperature beyond 550 °C, the sample undergoes an oxidation process where the direct transformation of SnO to SnO₂ phase occurs. However, in case of 650 °C, no peaks for SnO is observed which indicates a fully textured SnO₂ film growth. But, on increasing the annealing temperature beyond 650 °C, the SnO₂ phase destruction occurs and the SnO phase reappears. The SnO compound formation may also be due to the out-diffusion of oxygen into the silicon substrate to form an amorphous SiO layer at the bottom [18]. Also, the SnO₂ degradation ratio (I_SnO/I_SnO2) was calculated and found to be 1.16 and 1.22 in case of annealed 750 °C and 900 °C sample, respectively which indicates higher degradation of SnO₂ at higher annealing temperature.

Zoom In Zoom Out Reset image size

Using XRD, various other important parameters like microstrain, dislocation density, stacking fault, etc can be calculated in order to understand the structural modifications that might have occurred after annealing at different temperatures. The microstrain value has been calculated using the equation [19],

$$\begin{eqnarray*}&&\varepsilon =\displaystyle \frac{\beta \,\cos \,\theta }{4}.\end{eqnarray*}$$

The dislocation density (δ), represents the number of defects present in the crystal which is given by Williamson and Smallman's relation [19], δ = n/D²

where n is a factor equal to 1 providing minimum dislocation density and D is the grain size.

The effect of size and shape of the crystallite can be calculated using the Scherrer equation [19], $D=\tfrac{k\lambda }{\beta \,\cos \,\theta }$ where k is the shape factor of 0.94, λ is the wavelength of the x-ray, β denotes the FWHM and D denotes the particle size. Thus, from figure 2(a) it can be seen that with the increase in the intensity of the peaks, the FWHM decreases whereas, the grain size is increased with a maximum at 650 °C. From figure 2(b), the microstrain and dislocation density is found to be least at 650 °C, which indicates that the concentration of lattice imperfections is reduced at this temperature. A higher microstrain indicates a high level of defects in the grain boundaries. Moreover, on annealing, local rearrangement of atoms in the grain boundaries occur for which the grain boundary defects and dislocation density is reduced and the overall microstrain is also reduced [20]. The stacking fault helps to determine the deformation mechanism and dislocation behaviour in the samples. It associates with the interruption of a normal stacking sequence of a crystal plane, which significantly influences the defects, dislocations, and deformation affecting the performance of the material under different annealing conditions [19]. It is denoted by,

$$\begin{eqnarray*}&&{\rm{S}}{\rm{F}}=\left[\displaystyle \frac{2{\pi }^{2}}{45{(3\tan \theta )}^{\displaystyle \frac{1}{2}}}\right]\beta .\end{eqnarray*}$$

Zoom In Zoom Out Reset image size

On plotting as shown in figure 2 it is seen that the stacking fault is least for 650 °C annealed sample which also verifies the lower concentration of lattice imperfections in this case.

Texture coefficient (TC) of a plane represents texture of a particular plane concerning the preferred crystal orientation which is given as, ${\rm{T}}{\rm{C}}=\tfrac{I({h}_{i}{k}_{i}{l}_{i})}{{I}_{0}({h}_{i}{k}_{i}{l}_{i})}{\left(\tfrac{1}{N}\displaystyle {\sum }_{i=1}^{N}\tfrac{I({h}_{i}{k}_{i}{l}_{i})}{{I}_{0}({h}_{i}{k}_{i}{l}_{i})}\right)}^{-1}$ where I(h_ik_il_i) is the intensity of the diffraction peak of the sample, I₀(h_ik_il_i) is the intensity of the complete random sample and N is the number of diffraction peaks. TC > 1 indicates that the growth of the films has a higher degree of orientation to the 〈hkl〉 plane [16]. Thus, it is seen in figure 3 that the maximum TC value is obtained for 650 °C because the intensity of the peak is increased in this case which suggests the improvement of the crystal quality whereas it deteriorates for other annealed samples.

Zoom In Zoom Out Reset image size

Table 1 lists the various parameters of all the crystalline samples. From the table it can be concluded that the 650 °C sample gives the best crystallinity with minimum defect density as compared to all other samples.

The optical absorption measurement was performed on the GLAD synthesized SnO₂ NWs fabricated over quartz substrate for different annealed samples using UV–vis–NIR spectrophotometer. The absorption was carried out at room temperature under the wavelength range from 250 to 1000 nm as shown in figure 4. From the absorption graph, it is seen that all the samples absorb the maximum light in the UV region (250–400 nm) than in the visible region. Moreover, the absorption increases with annealing temperature and is highest in the case of the 650 °C sample. Thereafter, the absorption reduces with further increase in annealing temperature. The increased absorption in the case of the 650 °C annealed sample may be ascribed to the large generation of electron–hole pairs [21], improved crystallinity and higher grain sizes [22]. However, in the case of a 900 °C, 750 °C and 550 °C sample, absorption decreases due to the reduction in crystallinity and higher grain boundary defects (as observed from XRD) [21].

Zoom In Zoom Out Reset image size

Form the above analysis, it is observed that the 650 °C annealed sample showed the maximum improvement in the overall crystallinity among all the other annealed samples (350 °C, 550 °C, 750 °C and 900 °C) with reduced dislocation density and microstrain. Therefore, further investigation was carried out to understand its NW structural formation. Figure 5(a) shows the top-view image of the 650 °C annealed SnO₂ NWs deposited on the n-Si substrate. The image shows that the NWs were grown with an average diameter and inter NW distance (separation between the NWs) of ∼27 nm and ∼40 nm respectively. Figure 5(b) shows the cross-sectional FE-SEM images of the 650 °C annealed SnO₂ NW arrays. The image shows a uniform and vertically oriented SnO₂ NWs fabricated over SnO₂ TF/Si sample. The length of the NWs is calculated to be ∼185 nm and the thickness of SnO₂ TF is found to be ∼30 nm. The uneven growth of SnO₂ NWs (circle marked in both the figures) is due to the shadowing effect that is prevalent in the GLAD technique [11].

Zoom In Zoom Out Reset image size

The structural analysis of SnO₂ NWs was also done using TEM analysis along with SAED for the annealed sample at 650 °C. Figure 6(a) shows the group of SnO₂ NWs annealed at 650 °C. The magnified TEM image of a single NW is shown in figure 6(b). The length of the single NW is found to be ∼262 nm with a top diameter of ∼96.7 nm and a bottom diameter of ∼47.7 nm. Such an unsymmetrical pattern of GLAD NWs was also reported by the authors previously [2, 11]. The arrow in figure 6(b) indicates the growth direction. Moreover, the top inset shows the SAED patterns taken from the single NW signifies that the annealed 650 °C sample is polycrystalline in nature. The d-spacing measured from the magnified image (figure 2(b) (bottom inset), is found to be 0.34 nm which can also be verified from the XRD analysis. On calculation, using Bragg's law (from the XRD plot), the d-spacing is found to be ∼0.34 nm which is consistent with the above observation in TEM.

Zoom In Zoom Out Reset image size

Upon white light illumination, the photoresponse (I_Light/I_Dark) of the device increases in the reverse bias, depicting the Schottky diode characteristics. Figure 8 shows the variation of the photoresponse for voltage range −8 to 0 V (reverse bias) measured for the devices annealed at different temperatures 350 °C, 550 °C, 650 °C, 750 °C, and 900 °C. It is observed that the highest photoresponse is obtained for 650 °C annealed sample with a maximum value of ∼5 (at −8 V). This is due to the increased absorbance of the incident light which provides the generation of large number of electron–hole pairs in comparison to other annealed samples. The conversion to a near photoconductive mode makes the 650 °C device to contribute towards large photocarriers at the electrodes [11]. Another reason for such high value may be ascribed due to the improvement in the crystal quality and reduction in dislocation density which might increase the carrier concentration increasing the conductivity of the sample [24]. However, the photoresponse declines for the other annealed sample (350 °C, 550 °C, 750 °C and 900 °C) which may be due to the presence of bad crystallization and defects. However, in the case of the 650 °C device, the photoresponse increases linearly to the applied bias which signifies that the photocurrent can be tuned or controlled by varying the applied voltage at the electrode [27].

Zoom In Zoom Out Reset image size

Figure 9(a) shows the on–off switching response for the annealed samples under regular interval of 20 s and at a fixed bias of −2 V. The rise time and fall time of the photocurrent were calculated by using the second ordered exponential fitting equations [2].

$$\begin{eqnarray*}&&{I}_{{\rm{f}}{\rm{a}}{\rm{l}}{\rm{l}}}\left(t\right)={I}_{0}+{A}_{1}\exp \left(-\displaystyle \frac{t}{{\tau }_{{1}_{{\rm{f}}{\rm{a}}{\rm{l}}{\rm{l}}}}}\right)+{B}_{1}\exp \left(-\displaystyle \frac{t}{{\tau }_{{2}_{{\rm{f}}{\rm{a}}{\rm{l}}{\rm{l}}}}}\right)\end{eqnarray*}$$

$$\begin{eqnarray*}&&\begin{array}{l}{I}_{{\rm{r}}{\rm{i}}{\rm{s}}{\rm{e}}}\left(t\right)={I}_{0}+{A}_{2}\left(1-\exp \left(-\displaystyle \frac{t}{{\tau }_{{1}_{{\rm{r}}{\rm{i}}{\rm{s}}{\rm{e}}}}}\right)\right)\\ \,\,\,\,+\,{B}_{2}\left(1-\exp \left(-\displaystyle \frac{t}{{\tau }_{{2}_{{\rm{r}}{\rm{i}}{\rm{s}}{\rm{e}}}}}\right)\right),\end{array}\end{eqnarray*}$$

where I₀ is the dark current, A₁, A₂, B₁, B₂ are constants and τ_1rise, τ_2rise, τ_1fall, τ_2fall are first-and second-time constants for rising and falling respectively.

Zoom In Zoom Out Reset image size

Generally, two mechanisms are responsible for photoresponse in metal-oxide based photodetector. First, the generation and recombination of carriers related to the bulk given by τ_{1_rise} and τ_{1_fall} and secondly, the chemisorption process related to the surface given by τ_{2_rise} and τ_{2_fall}. Thus, on plotting the fitting graph as shown in figure 9(b), the rise time (τ_2rise) for 350 °C, 550 °C, 650 °C, and 750 °C are found to be 0.185 s, 0.184 s, 0.183 s, and 0.189 s respectively. Also, the fall time (τ_2fall) as shown in figure 9(c) are found to be 0.250 s, 0.251 s, 0.249 s, and 0.252 s for 350 °C, 550 °C, 650 °C, and 750 °C devices respectively. Moreover, equal rise (τ_1rise) and fall time (τ_1fall) of 2 ms is observed for 650 °C annealed sample which is much faster in comparison to the already reported as-deposited GLAD SnO₂ sample [2]. Moreover, the photocurrent of all the annealed devices are almost consistent and repeatable, indicating that the devices are stable. However, in case of a 900 °C annealed sample, the presence of the silicate layer increases the barrier which therefore increases the resistivity of the device. Besides, the defects in the 900 °C sample acted as a recombination site which hindered the separation of the electron–hole pairs. Therefore, no switching response is observed in the case of a 900 °C annealed sample.

Figure 10(a) shows the spectral response of all the fabricated annealed photodetectors (350 °C, 550 °C, 650 °C, 750 °C and 900 °C (maximized as inset)) measured from 280 to 500 nm wavelength at −2 V applied bias (fixed at turn on voltage) figure 10. From the figure, it can be seen that the 650 °C annealed device shows higher responsivity as compared to the other annealed samples. Also, the 650 °C annealed device show a maximum peak responsivity of 2.58 A W⁻¹ at ∼300 nm wavelength which may be ascribed to the band edge of SnO₂. In addition, the annealed 650 °C device showed 18 times more enhancement in responsivity in comparison to the as-deposited SnO₂ NW device, reported earlier by the authors [2]. As the 650 °C annealed sample showed improved crystallinity, reduced defects and dislocation density, and better responsivity, it can be concluded that the 650 °C annealed SnO₂ material is more stable and efficient for the photodetector performance.

Zoom In Zoom Out Reset image size

Further, the spectral response of the 650 °C annealed device is plotted as a function of various applied voltages in figure 10(b). On such, it is seen that the responsivity increases with the increase in the supply voltage since the photocurrent increases due to the efficient carrier collection. Thus, maximum responsivity of 2.58, 7.59, 15.97, 47.97 A W⁻¹ is found for 650 °C annealed sample at different voltage of −2 V, −3 V, −5 V and −7 V respectively. The 650 °C device was further analysed to measure various important parameters such as detectivity and noise equivalent power (NEP).

Detectivity is an important parameter for characterizing the photodetector performance and it signifies to detect the smallest weak output signal which can be expressed as

$$\begin{eqnarray*}&&{D}^{* }=\displaystyle \frac{{(AB)}^{1/2}}{{\rm{N}}{\rm{E}}{\rm{P}}}\,({\rm{c}}{\rm{m}}\,{\rm{H}}{{\rm{z}}}^{\frac{{\rm{1}}}{{\rm{2}}}}\,{{\rm{W}}}^{-1}),\end{eqnarray*}$$

where D* is the detectivity, A is the effective area (1.77 × 10⁻² cm²), NEP is the Noise Equivalent Power and B is the bandwidth of 1 KHz [11] for the shot noise which is dominant noise for frequency below 1 KHz. The incident light power density is 18.19 μW cm⁻² for 300 nm. By using the above equation, the detectivity is calculated and found to be 64.13 × 10¹⁰ Jones for 650 °C annealed sample at −2 V better than the as-deposited SnO₂ NW reported previously by the authors [11]. The increased detectivity may be attributed to the reduction of defects which resulted in the better response of the sample in comparison to the as-deposited sample.

The authors have found their result to be better than some of the recently reported works as tabulated below in table 2. Moreover, various other device parameters such as external quantum efficiency (EQE) and UV–Visible rejection ratio were also calculated. The device showed a high EQE of 1.07 × 10³%, indicating that the adsorbed photon was able to activate a greater number of electron–hole pairs. The UV–Vis rejection ratio (R₃₀₀/R₄₈₀) was found to 11.77 at −2 V indicating that the device is more sensitive to UV light and therefore can be potentially used as a visible-blind photodetector. However, more improvement is required which can be obtained by designing various nano-heterostructure using SnO₂ NWs.