WO₃ nanorods were synthesized through hydrothermal technology. In a typical procedure, 0.01-mol Na₂WO₄·2H₂O and 0.02-mol NaCl were dissolved in 60-ml deionized water and stirred for 30 min by magnetic stirring apparatus. Na₂WO₄·2H₂O was served as a precursor and NaCl was used as structure directing agents (SDAs) which have been proven to facilitate the controllable growth of nanostructure WO₃. The pH value of the solution was fixed at 2.5 which was adjusted by using 3-M HCl solution. Then, the obtained uniform mixture was transferred into a 100-ml Teflon-lined stainless steel autoclave and heated at 180 °C for 24 h in an electric oven. After naturally cooling down to room temperature, the resultant precipitates were washed with deionized water and ethanol by a centrifuger several times and dried in a vacuum at 80 °C for 10 h. Finally, the slurry of as-prepared tungsten oxide nanorods was prepared by ultrasonically mixing tungsten oxide nanorods and terpineol, which acted as a vehicle binder and solvent. In this experiment, all reagents were purchased from the Guangfu Fine Chemical Research Institute (Tianjin, China).

Firstly, a pair of interdigital Pt electrodes was deposited on the cleaned alumina substrates by RF magnetron sputtering using a shadow mask. In detail, the DPS–III ultrahigh vacuum facing-target magnetron sputtering system was used to deposit Pt electrical contacts and operating parameters were listed as follows: the RF power was 90 W, the base vacuum pressure was 4.0 × 10⁻⁴ Pa and the working pressure was 2 Pa. Secondly, the coating slurry was prepared by mixing the nanorods (30 mg) in terpineol (0.2 ml) and then ground in an agate mortar for 30 min to obtain a homogeneous paste. The resulting paste was coated on those alumina substrates by the spin-coating method. Thirdly, the WO₃ nanorods sensors were annealed at 300, 400, and 500 °C for 2 h in the air with a heating rate of 6 °C/min, respectively. Then the gas-sensing characteristics of as-prepared sensors were tested in a computer-controlled static gas sensing characterization system which consists of a 30-L sealed organic glass test chamber, a professional digital multi-meter, a data acquisition system and a flat heating plate with an adjustment range from room temperature (RT) to 300 °C.^[26] The programmable multi-meter was connected to the data acquisition system, which was used for continuously recording the resistance changes of the sensor and the sampling interval was set to 1 s. The schematic diagram of the sensor and the measurement system are depicted in Fig. 1. When exposed to NO₂, the sensor response presented a p-type feature at room temperature. Hence, the response is defined as S=Ra/Rg, where R_a and R_g are the resistances of the sensor element in the presence of atmospheric air and target gas respectively. The time taken by the sensor to achieve 90% of the total resistance change is defined as the response and recovery time.

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. (color online) The schematic diagram of the sensor and the measurement system.

The morphology, element composition and crystal structure of the WO₃ nanorods were characterized using a field emission scanning electron microscope (FESEM, FEI Nanosem 430), energy dispersive spectroscopy (EDS) attached on the FESEM, transmission electron microscope (TEM, TECNAIG2F-20) and x-ray diffraction (XRD, RIGAKU D/MAX 2500V/PC, Cu Kα radiation).

Typical FESEM image of as-synthesized WO₃ nanorods is shown in Fig. 2(a), from which we can see that those nanorods exhibit a one dimension structure with lengths in the range of 3 μm–5 μm and cross sections in the range of 50 nm–150 nm. Nevertheless, there still exist nanorod bundles which are induced by larger surface areas and their intrinsic defects such as shear planes, oxygen vacancies, and stacking faults.^[27,28] Further characterizations of the morphology and crystallinity of the WO₃ nanorods were investigated by TEM and high resolution transmission electron microscope (HRTEM) as shown in Fig. 2(b) and Fig. 2(c). It can be seen that the lattice fringe separation is about 0.323 nm, corresponding to the lattice space of the hexagonal WO₃ (200) plane. In order to investigate the morphology evolution of WO₃ nanorods with increasing annealing temperature, the SEM images of annealed WO₃ nanorods coated on ceramic wafer by spin coating are illustrated in Fig. 3. The appearance of annealed nanorods is changed pronouncedly compared with that of raw WO₃ nanorods (Fig. 2(a)). After being processed at 300 °C for 2 h, the diameter of all the nanorods is still in the range of 50 nm–150 nm but the lengths reduce sharply through a fracture of nanorods (Fig. 3(a)). The morphology feature remains stable after annealing at 400 °C (Fig. 3(b)). When increasing the temperature to 500 °C, it is apparent that the morphological characteristic of the nanorods almost disappears completely, and nanoparticles which display pronounced facet structures with polygonal cross-sections are formed instead, indicating that a possible agglomeration occurs (Fig. 3(c)).

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (color online) SEM, TEM, and HRTEM images of WO3 nanorods.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. SEM images of WO3 nanorods after (a) 300-°C, (b) 400-°C, and (c) 500-°C thermal treatment.

The surface morphology of the square-marked area and the corresponding EDS spectrum of WO₃ nanorods with annealed samples are shown in Fig. 4. Every EDS spectrum contains the characteristic peaks of C, O, Na, and W, and all the annealed sensors still include the characteristic peaks of Al and Pt. The elements concentration of (a) the pristine WO₃ nanorods together with (b) 300-°C-annealed, (c) 400-°C-annealed, and (d) 500-°C-annealed sensor are illustrated in Table 1.

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (color online) The surface morphology of selected area and corresponding EDS spectrum of (a) WO3 nanorods, (b) 300-°C-, (c) 400-°C-, and (d) 500-°C-annealed samples.

Tab1e 1.

Tab1e 1.

Tab1e 1. .   (a) (b) (c) (d) Element Weight% Weight% Weight% Weight% C 0.7 0.67 0.56 0.78 O 26.85 24.69 22.82 23.99 Na 2.05 1.55 1.76 2.37 Al 0 1.88 3.68 0 W 70.39 54.14 54.70 41.49 Pt 0 17.08 16.47 31.38 Totals 100 100 100 100

Tab1e 1. .

The elements concentration of (a) the pristine WO₃ nanorods together with (b) 300-°C-annealed, (c) 400-°C-annealed, and (d) 500-°C-annealed sensor are illustrated in Table 1. The sum of the weight concentrations is equal to 100%.The atomic ratio of W and O should be 1:3 as a result of the chemical reaction, but their measuring value of atomic ratio calculated by EDS quantitative analysis software is (a) 1:4, (b) 1:5, (c) 1:5, and (d) 1:7. We surmised the additional O atomic of the pristine WO₃ nanorods and 500-°C-annealed samples may come from adsorbed oxygen and NaO which is the residual decomposition product of Na₂WO₄·2H₂O. Besides the above three O elements supplies, a portion of O elements come from Al₂O₃ substrate owing to that we have detected an Al element in the 300-°C- and 400-°C-annealed samples. The Pt elements in the EDS may come from the alumina substrates deposited with a pair of interdigital Pt electrodes. This can be certified by the result of XRD analysis. The appearance of the C element can be attributed to the SEM grid.

In order to estimate O elements weight% derived from adsorbed oxygen, we have calculated the O elements weight% derived from other oxide. For example, the relative molar mass of W and O in the WO₃ molecule is 184:48. The weight% of W elements is 70.39% in the pristine WO₃ nanorods, so the weight% of O elements in WO₃ is 18.36%.The results are shown in Table 2.

Tab1e 2.

Tab1e 2.

Tab1e 2. .   (a) (b) (c) (d) Total O elements weight% 26.85 24.69 22.82 23.99 O elements derived from WO3 18.36 14.12 14.30 10.82 O elements derived from Al2O3 – 1.67 3.27 – O elements derived from Na2O 0.71 0.54 0.61 0.82 O elements derived from adsorbed oxygen 7.78 8.36 4.64 12.35

Tab1e 2. .

On the basis of Table 2, we further estimated the weight% of adsorbed oxygen/μm² of (a) the pristine WO₃ nanorods together with (b) 300-°C-annealed, (c) 400-°C-annealed, and (d) 500-°C-annealed sensor. The results are shown in Table 3. The weight% of adsorbed oxygen/μm² increased by the way of heat treatment. Considering the sample room of EDS or SEM is pumped into a vacuum state before scanning, the adsorbing power between oxygen and WO₃ nanorods should be a chemical bond instead of a physical bond. The chemical bond contains an electrovalent bond and a covalent bond, which were produced by lots of oxygen vacancies. Thus, we consider the oxygen vacancies increased in the nanorods by the way of heat treatment, which can be further confirmed by the XRD.

Tab1e 3.

Tab1e 3.

Tab1e 3. .   (a) (b) (c) (d) O elements derived from adsorbed oxygen 7.78 8.36 4.64 12.35 Scanning area of the square-marked area (μm2) 295.50 55.25 14.40 1.44 Weight% of adsorbed oxygen/μm2 0.03 0.15 0.32 8.57

Tab1e 3. .

The crystalline phases of WO₃ nanorods annealed at different temperatures together with the pristine WO₃ nanorods were analyzed by x-ray diffraction. The XRD patterns of the original WO₃ nanorods match well with JCPDS NO. 33-1387 for hexagonal WO₃ as indicated in Fig. 5(a). In order to avoid the WO₃ nanorods being polluted, we directly take the prepared sensors as a test specimen for XRD analysis. Therefore, the main crystalline phases of annealed ones are Al₂O₃ (JCPDS NO. 46-1212) and Pt (JCPDS NO. 87-0646) except hexagonal WO₃ (JCPDS NO. 75-2187). However, it is noteworthy that a new crystalline phase of cubic W₃O (JCPDS NO. 02-1138) present in annealed samples as shown in Figs. 5(b), 5(c), 5(d). Its three characteristic peaks which are the (210) peak with I=100 (2θ=40.040), the (320) peak with I=60 (2θ=66.761), and the (421) peak with I=70 (2θ=88.895) match together the Pt (111) peak with I=100 (2θ=39.796), the Al₂O₃ (214) peak with I=23 (2θ=66.519), and the Al₂O₃ (0210) peak with I=9 (2θ=88.994). W₃O is a kind of metastable and non-stoichiometry tungsten oxide, which has a large number of oxygen vacancies. As the temperature is increased to 500 °C, the characteristic peak of Pt (2θ=67.859) also appears. This is a result of the high annealing temperature making the Pt electrodes be exposed to the outside.

Fig. 5.

	Figure Option ViewDownloadNew Window
	Fig. 5. (color online) XRD patterns of (a) freshly-obtained WO3 nanorods, (b) 300-°C-, (c) 400-°C-, and (d) 500-°C-annealing WO3 nanorods.

We have examined the 300-°C annealed sample by sensing 1-ppm NO₂ from room temperature to 250 °C and the sensor showed two completely different sensing behaviors. The WO₃ nanorods sensor showed a p-type response at the range of 25 °C–50 °C while an n-type response at the range of 50 °C–250 °C. This work has been published in Ref. [29]. The p-type semiconductor based on WO₃ nanorods operating at room temperature is rarely reported. Consequently, this paper is focused on the sensing property of WO₃ nanorods sensor at room temperature.

Figure 6 depicts the dynamic responses of the gas sensors based on the WO₃ nanorods annealed at (a) 300 °C, (b) 400 °C, and (c) 500 °C to varying NO₂ concentration at room temperature (RT). Figure 7 illustrates the relationship between (a) the response time, (b) the recovery time and NO₂ concentration for the sensors based on annealed WO₃ nanorods at RT. As shown in Figs. 6(a)–6(c) and Fig. 7(a), the WO₃ nanorods sensors are highly sensitive to NO₂ for the resistance of all the sensors decrease abruptly for the NO₂ injection period (shadow areas) and can approach the steady-state values within 75 s. In particular, the response time of the 400-°C-annealed sample reduce to 1 s when the sample is exposed to NO₂ gas with a concentration of either 3 ppm or 5 ppm. As shown in Figs. 6(a)–6(c) and Fig. 7(b), after the removal of NO₂, the resistance of all the annealed WO₃ nanorods sensors increases and is able to recover to the original value within 200 s. The recovery time reduces to 97 s for 3-ppm NO₂ and 152 s for 5-ppm NO₂, respectively. Figure 6(d) illustrates the response of the sensors based on annealed WO₃ nanorods to varying NO₂ concentration at RT. The variation trend of all annealed WO₃ nanorods sensors is similar in that the response increases apparently with increasing NO₂ concentration. In the case of the 400-°C-annealed sample, there is an obvious increase in the response, with the maximum responses of about 1.3 ppm to 0.5 ppm, 1.7 ppm to 1 ppm, 2.3 ppm to 3 ppm and 3.2 ppm to 5 ppm NO₂, respectively. When the heat treatment temperature is increased to 500 °C, the gas sensing property suddenly gets worse. The maximum response is 1.5, when even exposing to 5 ppm. In addition, there is only a slight enhancement in response with increasing NO₂ concentration. We assume that the deteriorated sensing properties probably are related with the reduced specific surface areas (Fig. 3(c)). Excepting the recovery time of 500-°C-annealed sensor is shorter than 400-°C-annealed sensor to 5 ppm NO₂, in every NO₂ concentration, the response time and the recovery time of the 400-°C-annealed sample (red column) is shorter than the 300-°C-annealed sample (yellow column) and the 500-°C-annealed sample (blue column) as shown in Fig. 8.

Fig. 6.

	Figure Option ViewDownloadNew Window
	Fig. 6. (color online) Dynamic responses of the gas sensors based on the WO3 nanorods annealed at (a) 300 °C, (b) 400 °C, and (c) 500 °C to varying NO2 concentration at room temperature. (d) Relationship between the responses and NO2 concentration for the sensors based on annealed tungsten WO3 nanorods at room temperature.

Fig. 7.

	Figure Option ViewDownloadNew Window
	Fig. 7. (color online) Relationship between (a) the response time (b) recovery time and NO2 concentration for the sensors based on annealed tungsten WO3 nanorods at room temperature.

Fig. 8.

	Figure Option ViewDownloadNew Window
	Fig. 8. (color online) Schematic energy-level diagram of (a) n-type semiconductor, (b) the formation of depletion layer, (c) the formation of inversion layer.

In a word, from Fig. 6 and Fig. 7 we can see that the 400-°C-annealed sample has more excellent sensing performance at room temperature compared with both 300-°C- and 500-°C-annealed WO₃ nanorods sensors. We suppose that the excellent gas sensing properties of the 400-°C-annealed sample could be attributed to its moderate crystallinity, larger specific surface areas and more oxygen vacancies. Therefore, the excellent gas sensing properties of hydrothermal-prepared WO₃ nanorods make it an ideal candidate for detecting NO₂ at room temperature.

When operated at high temperature, tungsten oxide sensors are known to behave as n-type semiconductors. One of the conceivable mechanisms for the p-type conductive behavior at RT is the formation of an inversion layer on the surface of WO₃ nanorods. Owing to the fact that bulk defects equilibrate quite slowly at RT, surface conduction is dominant. The surface of the semiconductor is composed of a surface layer and space charge layer. Because of the existence of large density of surface states which mainly was induced by dangling, structural defects and gas adsorption, too many charges accumulated on the surface layer, which led to an aggregation layer, depletion layer, or even inversion space-charge layer below the surface layer of the semiconductor as shown in Fig. 8. In the energy-level diagram of the n-type semiconductor, the Fermi level approaches the bottom of the conduction band and above the intrinsic level, as shown in Fig. 8(a). When the WO₃ nanorods sensors are exposed to air, oxygen molecules are absorbed on the surface of WO₃ nanorods and form chemisorbed oxygen species by capturing electrons from the conduction band of WO₃ nanorods (Eqs. (1) and (2)). Then the energy level of WO₃ nanorods is upswept and the depletion layer is formed, as shown in Fig. 8(b). When the WO₃ nanorods sensors are exposed to NO₂, as a powerful oxidizing gas with strong electron-withdrawing power, NO₂ is capable of competitively capturing electrons not only from the conduction band of WO₃ nanorods but also from chemisorbed oxygen species (Eqs. (3)–(5)). Besides the high surface areas that make the surface more actively adsorb O₂ or NO₂, there exists a non-stoichiometry monoclinic W₃O which has a large amount of oxygen vacancies. It easily adsorbs O₂ or NO₂ to stabilization.^[30] Then with the continuous accumulation of negative charges, too many negative charges aggregate in the surface layer and the inversion layer in which holes become the main charge carrier is gradually generated, as shown in Fig. 8(c). In the energy-level diagram, the Fermi level appears below intrinsic level.

For practical use, the repeatability of the gas sensors is also an important quality factor. Reversible and reproducible responses of the 400-°C-annealed gas sensor upon periodic exposure to NO₂ gas of 1 ppm are presented in Fig. 9 with five switching on/off cycles. Notably, the sensor shows a similar response value and response–recovery time, which suggest an excellent repeatability and stability. It is well known that the high crystallinity of the WO₃ nanorods is responsible for the high stability of the sensor.^[31]

Fig. 9.

	Figure Option ViewDownloadNew Window
	Fig. 9. (color online) Electrical resistance changes of the sensor based on 400-°C annealed WO3 nanorods as a function of time for five cycles at room temperature with 1 ppm of NO2.