A. Structure characterization
Fourier-transform infrared spectroscopy (FTIR) and X-ray diffraction (XRD) were used to characterize three Nano-structure samples.
FTIR (JASCO, model: 460 Plus) studies of the TiO2 grown in all three structures showed the characteristics of the formation of high purity product with the least amount of impurity. Fig. 4 shows a sequence of FTIR spectra for TiO2 nanowire, nanoparticle and thin-film in the range of 500–4000 cm−1. As reported in the literature  the peaks at 3440 cm−1 in the spectra are due to the stretching and bending vibration of the -OH group. The peaks at 1620 cm-1 show stretching vibrations of Ti-O-Ti. Further, the peak observed at 690 cm−1 is due to the vibration of the Ti-O-O bond. Therefore, it is obvious that The FTIR spectra clearly exhibit the presence of Ti-O bonds in all the three samples.
Moreover, XRD (Inel, model: EQUINOX3000) was conducted to determine the phase analysis of each sample. In our XRD equipment monochromatic optic Kα1 or Kα1/2 were used. Fig. 5 presents the XRD pattern of the synthesized samples where the peaks of anatase and rutile nanoparticles are well recognized. As shown in Fig. 5 both nanoparticle and nanowire samples show only the anatase phases, but the rutile phase is observed only in a thin-film sample. XRD pattern specifies the phases and sizes of crystalline materials, providing a determination of the ratio of the anatase phase to the rutile phase in the TiO2 Nano-structures. Three peaks were considered for the study, and using the Debye- Scherer formula, the grain size was calculated for all three peaks. From Fig. 5, considering the peak Full Width at Half Maximum (FWHM), the average particle size is calculated. This size was calculated for nanowire and nanoparticles samples less than 100 nm while for thin-film sample more than 100 nm was obtained. Peaks in XRD pattern at 2θ=25° (101) and 2θ=48° (101) indicated the presence of the anatase phase while peaks at 2θ=27.5° (110) demonstrated the occurrence of rutile phase. All three peaks suggest that the prepared samples are in a hexagonal structure. The pattern matches well with the standard JCPDS files # 21-12729 . The intensity of XRD peaks of the nanowire and Nano-particle TiO2 samples shows that the formed nanoparticles are crystalline and broad diffraction peaks indicate small size crystallite. Since in the present study, the sensors were tested at room temperature, there is no concern about the phase change.
B. Electrical measurement
Gas sensing measurements have been conducted under constant temperature conditions at room temperature with different ammonia concentrations.
For measurements and tests of TiO2 samples the circuit shown in Fig. 6 is used. The silver paint was used to connect the external contact to Au electrodes. The electrical characteristics of the TiO2 sensors were obtained from the I-V curve of the electrodes junction measured by the KEITHLEY 228A voltage/current source and Sanwa PC7000 Handheld multimeter. To dynamic gas sensing measurements in this investigation, studied devices were placed in a chamber with a continuously gas flow. The humidity control was performed using silica gel. Besides, temperature and humidity sensors are placed in the sensing chamber for online monitoring.
In the first step, the response of the samples toward the different amounts of NH3 gas has been measured. As shown in Fig. 7 the highest response was observed in the nanowire TiO2 sample toward different concentrations. The response to ammonia gas in the nanoparticle TiO2 was very close to the response of the nanowire sample, but the response of the thin-film TiO2 sample was approximately half that of the nanowire TiO2 sample. As shown in the SEM pictures (Fig. 1 and Fig. 2) in the nanowire and nanoparticle samples with high porosity, a larger surface-to-volume ratio is seen because the surface arrangement of atoms is slightly different from that of the body  and so surface oxygen molecules are easily removed from the surface and create oxygen vacancies . Further, as it is shown in Fig. 7 when the inlet NH3 gas concentration increased, the response also increased in all three samples. At a high concentration of 200 ppm, the percentage of response differences between the thin-film sample and the other two samples was less compare to lower NH3 concentration. This phenomenon implies that at higher concentrations of gas, other defects such as interstitial ions Ti3+ and Ti4+ in the titanium dioxide lattice play a role in the gas sensing mechanism . The amount of oxygen vacancies in the TiO2 at the room temperature is not sufficient for gas sensing. But, in Nano-structure TiO2, due to the presence of a very large volume to surface ratio, there is a lot of surface oxygen with dangling bonds, which can be absorbed by ammonia molecules and extracted from TiO2. This process can lead to the release of carriers in titanium dioxide and thus, a change in the Fermi level. When surface of Nano-structures TiO2 is exposed to gas molecules, the oxyanions react with the ammonia molecules and deliver a large number of carries to the grains . In a non-porous structure, gas molecules react by physically and then chemically adsorbed on the oxide semiconductor surface, decomposed by semiconductor oxygen atoms. The result of this surface interaction is the reduction of oxygen, resulting in an increase in the number of free electrons in the oxide body, which increases conductivity . In the polycrystalline structure, the determining factor for the conduction of the sensitive layer is the potential barrier height between the grains. At the grain level, oxygen can be absorbed to the grain surface in forms of O2-, O- and O2- creating a potential barrier at the grain boundaries. So typically, the semiconductor oxide sensor properties increase with decreasing grain size . Therefore, nanowire and nanoparticle samples that have smaller grain size boundaries will show higher response.
The change in conductivity in metal oxide gas sensors is due to two factors: change the carrier density or bending of energy bands in the presence of gas . The speed of carrier response or bending change was examined after applying the gas to the surface. Fig. 8 shows response and recovery times after inserting the NH3 gas and disconnecting the flow of the gas. Response time for the nanowire and nanoparticle samples have been recorded around 40 seconds while for thin-film sample showed a response time of around 70 seconds. Recovery times in three samples were measured approximately 15 seconds. As discussed earlier the nanowire and nanoparticle films exhibited more pores than thin films. So due to the greater porosity of the nanowire and the nanoparticle samples, the number of molecules reaching the surface increase and create lower Response Times. However, porosity does not have much effect on recovery time.
In the nanocrystals, the interaction of the gas molecules with other molecules like water molecules adsorbed on the surface of the grains has a greater effect on the electrical conductivity of the sensitive layer . The change in the number of atoms absorbed at the surface controls the mobility of the carriers by changing the number of distributed carriers . The presence of water molecules can interfere with the gas molecules absorbed on the surface. Two different tests were performed to accurately examine the effect of humidity.
First, the fabricated sensors considered as the humidity sensor. Humidity sensing was performed at room temperature and 760 mmHg pressure. As shown in Fig. 9 (a) the response of the samples to humidity is very low compared to ammonia gas. All three sensors showed a very small response to change in relative humidity. Most sensitivities observed for thin-film, nanoparticle and nanowire were 0.019, 0.035 and 0.034, respectively at 80% relative humidity. In this experiment dry environment is considered below 10% relative humidity and concentration [gas] in the response equation (1) for 80% relative humidity was equal to 50 ppm. Since the response of samples is very low in the presence of different humidity, these samples cannot be a humidity sensor.
In the second experiment, changes in humidity during ammonia gas sensing were investigated. Fig. 9(b) shows the changes in the response of the different samples versus humidity. The response values have been obtained from the results shown in Fig. 7. As shown in Fig. 9(b) the response decrease slightly with increasing humidity up to 60%. Though the response decreases drastically above 60% of humidity’s.
As shown in Fig. 9(b) humidity effect toward 50 ppm NH3 gas response is less than 3% up to 60% relative humidity for samples. Also the sensor response to 10-90 relative humidity has experimented for all three samples.
To characterize the response of the three sensors at the higher operating temperature we measured the senility of the samples at various temperatures from 25 °C to 200 °C. As reported in the literature  the sensing mechanism TiO2 change with temperature. These can be due to the different oxygen densities in TiO2 with various temperatures. It is believed that higher temperatures create lattice defects in TiO2 . Oxygen vacancies at higher temperatures diffused rapidly with high mobility into the TiO2 body as so these define the conductivity of TiO2 sensors . As Fig. 10 illustrates, increasing the temperature to around 200 °C resulted in a significant increase in response. This increase was observed more than 3.5 times in the nanoparticle sample compared to the sample operated at room temperature. Further, the effect of temperature on the response of the nanoparticle sample was greater than that of the nanowire sensor. The difference is because the nanoparticle sample has a higher porosity than the other films increasing the properties of the body O2 in the sensing mechanism.
To consider the thermal and electrical stability of the fabricated samples, two tests were performed without inserting gas into the test chamber: First, we applied a voltage of 40V DC for 24 hours  followed by testing the samples in an oven at 400 °C for 8 hours . After the tests, samples left over for three months under normal room conditions. Gas sensing tests were performed on the samples again. As shown in Fig. 11 the results showed that the nanowire sample was more stable than the other samples. Significant differences were observed in the response of the nanoparticle and thin-film in the response time and the recovery time for the fabricated samples and those tested after three months. Condition 1 shows the results of the response just after the fabrication of TiO2 and using Condition 2 exhibits the results obtained after 3 months.