Enhanced ethanol sensing properties of WO3 modified TiO2 nanorods

Pristine and WO3 decorated TiO2 nanorods (NRs) were synthesised to investigate n-n-type heterojunction gas sensing properties. TiO2 NRs were fabricated via hydrothermal method on fluorine-doped tin oxide coated glass (FTO) substrates. Then, tungsten was sputtered on the TiO2 NRs and thermally oxidised to obtain WO3 nanoparticles. The heterostructure was characterised by X-ray diffraction (XRD), scanning electron microscopy (SEM), and energy-dispersive X-ray (EDX) spectroscopy. Fabricated sensor devices were exposed to VOCs such as toluene, xylene, acetone and ethanol, and humidity at different operation temperatures. Experimental results demonstrated that the heterostructure has better sensor response toward ethanol at 200 °C. Enhanced sensing properties are attributed to the heterojunction formation by decorating TiO2 NRs with WO3.

a better NH 3 sensing performance [42]. Even though, MOX heterostructures have potential applications in chemical gas sensors [43][44][45], there are limited study on the sensing properties of WO 3 /TiO 2 NRs heterostructures. Especially, studies with various WO 3 loading are considerably poor.
In this study, WO 3 /TiO 2 NRs with various WO 3 loading were obtained for chemical gas sensors. The enhancing sensor performance of SMO gas sensors with decorating of highly ordered 1D n-type TiO 2 NRs by n-type WO 3 is the motivation of this study. Highly ordered TiO 2 NRs were synthesised by hydrothermal method on TiO 2 seed-layer coated FTO substrates. Then, WO 3 layers were coated on TiO 2 NRs by magnetron sputtering technique with different thicknesses. Structural and morphological characterisation of WO 3 /TiO 2 heterostructures were investigated. Gas sensor performances of heterostructures were studied against ethanol, toluene, xylene, acetone and humidity at various temperatures.

Fabrication of WO 3 modified TiO 2 NRs heterostructures
The WO 3 /TiO 2 heterostructures were obtained on the basis of different studies in the literature. A compact TiO 2 seed-layer with a thickness of approximately 50 nm was deposited on the FTO substrate to prevent shorting before the growth of TiO 2 NRs as reported previously in literature [46]. First, FTO substrates were purified by acetone, isopropanol and DI water in ultrasonic bath for 10 min, respectively. Then, Ti seed-layer was deposited by RF magnetron sputtering on the FTO substrates. Sputtering of Ti thin film (TF) process was performed in 5m Torr Ar atmosphere, with applied power of 100 W for 25 min. Finally, samples were annealed in the air atmosphere at 500 ˚C for 3 h. During the thermal oxidation process, Ti layer reacts with oxygen molecules in the air and transforms to TiO 2 layer [47].
TiO 2 NRs were synthesised by hydrothermal method on TiO 2 seed-layer coated FTO substrates. Firstly, 40 mL DI water and 40 ml HCl was mixed. Then, 0.9 ml TTIP was added drop by drop on previously mixed solution. Finally, the resulting mixture was stirred for 1 h at room temperature to obtain a homogeneous solution. This precursor solution was poured into a 250 mL autoclave, and TiO 2 seed-layer coated FTO substrates were placed vertically into the autoclave. Then the autoclave was sealed, placed into the temperature-controlled oven and thermally treated at 170 °C for 15 h [22,48]. After thermal treatment, the samples were removed from autoclave, rinsed in DI water and dried under dry air flow. WO 3 layer was deposited on TiO 2 NRs by the thermal oxidation method. Firstly, W layer was deposited by RF magnetron sputtering. Deposition of W layer was performed at 5 m Torr with RF power of 120 W for 1, 2, and 3 min under Ar atmosphere. Then, deposited W layer was thermally oxidised at 450 °C for 1 h to obtain the WO 3 layer [49]. Sensor fabrication process was illustrated in Figure 1. Pristine TiO 2 NRs sample was named as TiO 2, and WO 3 -modified TiO 2 NRs samples were named as WT-1, WT-2 and WT-3 in accordance with W deposition time of 1, 2, and 3 min, respectively. The morphological and structural characterisations of fabricated samples were performed by scanning electron microscopy (SEM) equipped with energy dispersive X-ray spectroscopy (EDX) (Philips XL 30S) and X-ray diffraction (XRD) (Rigaku D-max, RINT-2200 series, X-ray diffractometer with Cu-Kα radiation, λ = 0.15418 nm), respectively.

Gas sensing measurements
WO 3 -decorated TiO 2 NRs were examined for VOCs and humidity sensing performance. To perform electrical measurements, aluminium contact electrodes (thickness 200 nm) were evaporated on the samples with Leybold Univex 450 (city, country?) thermal evaporation system. Schematic illustration of sensor fabrication was given in Figures 2a and  2b. The sensors were placed into a test chamber with 1 L in volume. A high purity dry airline was connected to the test chamber. The dry air flow and the concentration of gases were controlled by flow meters and a multi gas controller -MKS 647C. Working temperature of devices was controlled by a Lakeshore 340 (city, country?). Keithley 6517A electrometer (city, country?) was used for current vs. time characteristics during gas sensing measurements. The atmosphere in the test chamber was cleaned by dry air flow. When the electric current reached a steady value, VOCs was sent to the test chamber. Humidity sensing performances of the sensors were also characterised. VOCs were generated by bubbling method [25]. Antoine's equation was used to calculate VOCs' concentration. All data were reported as a sensor response defined as follows [50]; S R = ∆I⁄I 0 , where ∆I is the change in the current value when sensors were exposed to target gas molecules, I 0 is the baseline current value measured under dry air flow condition. Response (t90 res ) and recovery (t90 rec ) times are defined as the time required the sensor to achieve 90% ∆I of its current form [51,52].

Material characterisation
The surface morphology and elemental analysis were performed by SEM and EDX, respectively. The SEM images indicate that TiO 2 NRs are vertically aligned and homogenously covered on the substrate surface as seen in Figure Figure 4 shows the presence of the titanium, tungsten and oxygen in the rods. EDX spectrum of the samples in Figure  4a shows that tungsten is not present in TiO 2 sample and atomic distribution of W are 0.5%, 0.94% and 1.11% in WT-1, WT-2 and WT-3 samples, respectively. The amount of W particles is correlated with sputtering time of W. The EDX mapping of WT-3 is given in Figure 4b. W particles homogeneously covered the surface of TiO 2 NRs as seen in Figure 4b.
XRD patterns of the samples are given in Figure 5. According to the XRD patterns, the diffraction peaks 36.1˚ and 62.8˚ were attributed to (101) and (002) crystal planes of rutile TiO 2 , respectively (PDF card number 00-021-1276). The intensity of the diffraction peaks 36.1˚ and 62.8˚, which refer to rutile TiO 2 , decrease in samples WT-1, WT-2 and WT-3. This can be explained by decorating of TiO 2 NRs with WO 3 [53]. In addition, there are no observed diffraction peaks related with WO 3 in the XRD pattern, which might be due to poor signal formation from small amount of material loading. In order to identify the WO 3 XRD pattern, WO 3 TF with a thickness of 50 nm (WO 3 -50) was deposited by RF magnetron sputter system, and subsequently thermal oxidised on FTO substrate. In Figure 6 comparative XRD results are shown for WO 3 -50, WT-3 and pristine FTO substrate. Diffraction peaks at 2θ = 23.2⁰, 24.54⁰, 33.1⁰ and 34⁰ can be assigned to monoclinic WO 3 (002), (200), (022) and (220) reflections, respectively (PDF card number 00-043-1035). The peaks which marked as "S" belong to FTO substrate (PDF card number 00-046-1088). where R is ethyl, i-propyl, n-butyl, etc. [54]. Firstly, hydrolysis occurs by reaction of TiO 2 precursor (TTIP) with water and TTIP transforms to titanium alkoxide. Then, titanium alkoxide forms a complex with water. The acidic platform controls the rate of complex formation. Finally, the high temperature and pressure condition accelerates hydrolysis process and appeared complex starts to deposit onto the substrate as TiO 2 NRs in rutile phase [55]. The fundamental reason of growing highly ordered NRs in the deposition process of titanium complexes onto the substrate is surface energy. In the TiO 2 phase, the lowest surface energy has (110) face. It means that [001] direction, parallel to (110) plane, is the theoretically preferable growth direction. The powerful (002) pick in XRD pattern is the proof of the growth of the highly aligned TiO 2 NRs along [001] direction [56].

Gas sensing properties
Gas sensor measurements of fabricated sensors were performed under toluene, xylene, acetone, ethanol, and relative humidity ambient in an operation temperature range between 100 °C and 250 °C. There was no observed sensor response signal from all the samples against any gases at 100 °C. Also, pristine TiO 2 and WO 3 /TiO 2 heterostructures could not sense acetone molecules for all operation temperatures. Operation temperature dependent sensor response results of all sensors are given in Figure 7 with bar diagrams. After all sensor measurements at different operation temperatures, 200 °C is identified as the optimal operation temperature for all sensors due to the highest sensor response values against each gas. At the optimal operation temperature, WT-1 sensor showed an excellent sensing performance against 1850 ppm ethanol as seen in Figure 7b. Sensor performance toward ethanol is drastically increased with the effect of WO 3 on the surface. Sensor response of WT-1 is 18-fold higher compared to pristine TiO 2 NRs at 200 °C. WO 3 plays a key role on the surface as catalyst and increases the sensor response. Also, sensor response values of WT-1 are highest against all tested gases due to catalytic effect of WO 3 . After the identifying optimal operation temperature, concentration dependence of sensor response was investigated at 200 °C. Concentration dependence of sensor response is given in Figure 8.    The investigation of concentration dependence of sensor response demonstrate that all sensors are most sensitive against ethanol molecules at optimal operation temperature. Sensor response of all samples against 1850 ppm ethanol and different ethanol concentrations for WT-1 sample at 200 °C are given in Figure 9.
It's clear that the signal returns to baseline after turning off the ethanol and purging with dry air. WO 3 modified TiO 2 NRs heterostructures showed enhanced sensor properties compared to the pristine TiO 2 NRs sensor. Sensing mechanisms of MOX sensors that are composed of only one type of materials have been studied and well explained in the literature [57,58]. Enhanced sensor properties can be attributed to the catalytic effect of WO 3 . In this case, WO 3 plays a role as a catalyst material in the reaction between analyte gas and TiO 2 . If the surface coverage of the WO 3 increase, the catalytic role of the WO 3 turns into a sensing layer, so a lower sensor performance generally is observed [59]. In Figure 3, SEM data also clarifies more coverage on the surface for WT-2 and WT-3. Previous works have also reported the enhanced sensor properties due to the catalyst role in heterostructures [17,[60][61][62]. WT-1 sample exhibited enhanced ethanol response than others and its concentration dependence sensor response performance was illustrated separately in Figure 9b. During the exposure, the response increased rapidly, then the rate of increment stopped and slightly declined to reach the saturation. While purging the sensor, the response decreases rapidly and reaches baseline. Response times (t90 res ) of WT-1 sensor are 6 min for each ethanol concentrations. The sensor showed a very stable sensing characteristic against ethanol. On the other hand, recovery times (t90 rec ) of WT-1 sensor are 15, 14, 8, and 6 min for 1850, 900, 450, and 225 ppm ethanol, respectively. These time values are better than the ones in our previous ethanol sensor studies [22,50].
SMO materials such as TiO 2 and WO 3 have oxygen deficiencies on crystalline surface due to their specific stoichiometry. As a result of this condition, free electrons appear in the conduction band of SMO material. Therefore, this type of semiconductors is named as n-type semiconductor. Gas sensing mechanism of n-type SMO materials generally can be explained with oxygen adsorption on the surface as given in Figure 10. When the n-type SMO was exposed to ambient, oxygen molecules in the air would be adsorbed on the surface of SMO with capturing by charge carriers (free electrons) (Figure 10a). Decrease in number of charge carriers leads to appearance of depletion layer between the grain boundaries that limits the electron transfer. Therefore, the depletion layer width and contact barrier height between two adjacent grains will be increased by the remaining number of oxygen molecules. The higher contact barrier leads to lower conductance of n-type SMO. When n-type SMO material is exposed to reducing gas molecules, these molecules react with the preadsorbed oxygen molecules (Figure 10b). Then, the depletion layer width and contact barrier height between two adjacent grains decreases again. As a result, this reaction leads to increasing of n-type SMO materials conductance [63].
When there is contact formed between TiO 2 and WO 3 , these two different n-type SMO materials behave as a new sensing material named as n-WO 3 /n-TiO 2 heterostructure, as shown in Figure 11. Because Fermi levels of TiO 2 is higher  . The schematic illustration of n-type metal oxide gas sensors grain boundary when exposed to a) air ambient and b) reducing gas ambient. than that of WO 3 , the electrons are transferred from conduction band of TiO 2 to the conduction band of WO 3 (Figure 11a). This process would continue until equalising of the Fermi level between WO 3 and TiO 2 occurs. The formation of n-n-type heterostructure leads to the creation of an electron depletion layer in TiO 2 and an electron accumulation layer in WO 3 (Figure 11b). The accumulation layer of WO 3 would be enhanced oxygen adsorption in air ambient [40,[48][49][50]64].

Conclusion
WO 3 /TiO 2 heterostructures were fabricated to investigate VOCs sensing performance. According to morphological characterisation, WO 3 uniformly covered the entire highly ordered TiO 2 NRs surface. XRD investigation shows that TiO 2 NRs was grown on rutile phase and was highly aligned along the [001] direction. XRD peaks of WO 3 on samples did not exist due to their small amounts. However, further investigation illustrated that WO 3 will grow in monoclinic phase with the thermal oxidation method. TiO 2 and WO 3 /TiO 2 heterostructures were tested against VOCs such as toluene, xylene, acetone and ethanol, and relative humidity. It was observed that n-n-type WO 3 /TiO 2 heterostructure advanced the sensor performance of TiO 2 NRs against almost all tested gases, except acetone, which is not detected with any sensors. WT-1 sensor showed the best sensor performance compared to TiO 2 , WT-2 and WT-3 sensors. Ethanol sensing response of WT-1 sensor was 18-fold higher than pristine TiO 2 NRs at 200 °C. The enhanced gas sensor performance of WO 3 / TiO 2 heterostructure is attributed to n-n type heterostructure formation that leads to the formation of depletion and accumulation layers. According to our findings, n-WO 3 /n-TiO 2 heterostructures have a high potential for ethanol sensor applications.