2.1. Structure, Composition and Morphology
Powder X-ray diffraction (XRD) was used to characterize the phase composition and purity of the as-synthesized products. The yellow powder (precursors) obtained from the oil bath reaction agreed well with reported orthorhombic tungsten oxide hydrate (WO
3·H
2O, JCPDS No. 84-0886), as shown in
Figure 1a (Pattern A) [
29]. The WO
3·H
2O precursors were converted into WN and WN/WO
3 composites by nitridation in NH
3 atmosphere and oxidation in air, under heat treatment at 700 °C and 200 °C for 3 h, respectively.
Figure 1a (Patterns B and C) shows the XRD of WN and WN/WO
3 composites. As shown in pattern b, all the diffraction peaks matched well with those of hexagonal WN (JCPDS No. 65-2898). For the composite (Pattern C), the crystal phase was a mixture of hexagonal WN (JCPDS No. 65-2898), a big convex appearing in the position of 2θ 20°–30°, and a small convex in the position of 2θ 50°–60°, which can be attributed to amorphous WO
3. Thus, the final product is a composite material composed of hexagonal WN and amorphous WO
3 (WN/WO
3). The composites underwent a mass change during the transformation from WN to WO
3 by calcination in air. By using this mass change, we can calculate the ratio of WN to WO
3. Thermogravimetric analyses of the products at temperature range of 0–1000 °C are shown in
Figure 1b. The DSC-TGA results of the WN/WO
3 composite nanosheets are displayed in
Figure 1b. It was observed from the TGA curve that the specimen shows a weight loss of about 1%, which relates to adsorbed water and other trace substances. When the temperature reached about 211.5 °C, the weight of the specimen began to increase. When the temperature reaches about 457 °C, the weight nearly reaches stability. Moreover, the DSC curve exhibited an obvious exothermal peak at 426 °C, which corresponded to the obvious weight gain of approximately 8.86% between 211 and 457 °C in the TGA curves. The significant weight gain was attributed to the reaction of the WN with oxygen to generate WO
3. According to the weight gain of approximately 8.86%, the WN content of the composites is calculated to be 55.9%.
The morphology of WN/WO
3 composites was characterized through SEM and TEM analysis, and the results are shown in
Figure 2. According to
Figure 2a–c, the WN/WO
3 composites exhibited square shaped nanosheets with uniform lateral dimensions mainly ranging from 80 to 150 nm (
Figure 2a,b) and thickness of 10 to 30 nm (
Figure 2c). In addition, relatively homogeneous pores with average diameter of 10 nm were dispersed uniformly at the surfaces of WN/WO
3 composites. The porous feature may have contributed to the volume shrinkage caused by transformation of orthorhombic WO
3·H
2O to hexagonal WN.
Figure 2d shows the HRTEM image of the WN/WO
3 composites, where the lattice fringes d-spacings of 0.239 and 0.204 nm are identified and correspond to the (111) and (200) planes of hexagonal WN (JCPDS No. 65-2898), respectively. The inset of
Figure 2d presents the selected area electron diffraction (SEAD) of WN/WO
3 composites. The distinctive diffraction rings and spots on the SEAD pattern match the crystal structure identification obtained from the HRTEM and XRD. The diffraction rings are indexed to cubic WN (JCPDS No. 65-2898) (111), (200), (220), (311), (400), and (331) planes, respectively.
2.2. Gas Sensing Characteristics
Dynamic sensing performance of WN/WO
3 composites-based sensors to various concentrations of NO
2 at room temperature is shown in
Figure 3a. NO
2 is an oxidizing gas, so when an NO
2 molecule adsorbs on the surface of WN/WO
3 composites, it captures electrons from the composites, resulting in the increase of resistance of the composites-based sensors upon exposure to NO
2.
Figure 3b plots the sensor response as a function of NO
2 concentration, and shows that with an increase in NO
2 concentration, the response of the sensor also exhibits an increasing trend. Interestingly, the sensitivities to NO
2 concentrations from 5 to 100 ppb and 100 to 1000 ppb exhibit linear dependence, fitting well into two linear curves which have different slopes. It is well understood that electrical conductance shows strong dependence on carrier and mobility (σ =
neμ
n, where
n is the concentration of electrons,
e is electronic charge, and μ
n is electron mobility), and that electrical resistivity is the reciprocal of electrical conductance. When the WN/WO
3 composites sensor is exposed to low concentration of NO
2, the amount of adsorbed NO
2 is so sparse that the degradation of carrier could be almost negligible, and the response of the sensor exhibits a “high increase model”. However, when exposed to high concentration of NO
2, the effect of NO
2 adsorption will be enhanced, and the degradation of carrier mobility becomes relatively remarkable, resulting in a decrease in the rate of sensitivity increase under abundant NO
2 adsorption [
30]. The detection limit of the WN/WO
3 composites sensor for NO
2 is calculated to be approximately 1.28 ppb based on signal-to-noise ratio of 3 (see the Section “Calculation of Theoretical Limit of Detection Using Signal/Noise Ratio” in
Supplementary Materials), which is far below the immediately dangerous to life or health (IDLH) values (NO
2: 20 ppm) defined by the U.S. National Institute for Occupational Safety and Health (NIOSH). The recommended NO
2 exposure limit defined by the NIOSH is 1 ppm [
31].
Selective detection of target gas from various interference gases is still an unsolved drawback for chemiresistive gas sensors [
32]. The responses of the sensor based on WN/WO
3 composites to common vehicle exhaust, including SO
2, CO
2, moisture, CO, NH
3, ethanol and NO
2, were measured to quantify the selectivity, which is shown in
Figure 4. As clearly illustrated in
Figure 4, the sensor shows a high response to NO
2, which is more than 2 times higher than that for other interference gases, indicating the excellent selectivity to NO
2 as opposed to other selected interference gases. The selective detection to NO
2 gas might be due to the high reactivity and large electron affinity (2.28 eV) of NO
2, in comparison with pre-adsorbed oxygen (0.43 eV) and other test gases [
33]. In addition, the responses of the sensor based on WN/WO
3 composites to 200 ppb NO
2 at RT under different humidities were tested and the result is shown in
Figure S1. It can be observed that, with increasing relative humidity, the response of the sensor decreases because of the competition between water molecules and NO
2 molecules for the reacting sites [
34]. However, the sensor shows a high response of 17.8 even at high humidity of 90 RH%.
Figure 5a shows the response and recovery times of the WN/WO
3 composites-based sensor. With increasing NO
2 concentration, the response time and recovery time both experience a decrease. The reason why the response and recovery times are substantially longer at low concentration is unclear and needs further study. The following can be considered as a plausible explanation. At low concentration of ambient NO
2, the NO
2 molecules occupy a lower percentage of air relative to oxygen molecules, so the probability of NO
2 molecules arriving and being captured by the surface of the sensing materials (WN/WO
3) is relatively smaller, resulting in a long response. When the concentration of NO
2 increases, the probability of NO
2 molecules arriving and being captured by the surface of sensing materials becomes higher, decreasing the response time. As for the recovery time: at low concentration of NO
2, the amount of adsorbed NO
2 is too small, making it difficult to desorb completely, resulting in a relatively longer recovery time. When the concentration is higher, the concentration difference between the surface of WN/WO
3 and ambient atmosphere is large enough to desorb NO
2 molecules quickly, reducing the recovery time. Similar observations have also been reported by other investigators [
35,
36]. In addition, the sensor shows excellent reversibility properties with a response time of less than 80 s and a recovery time of less than 180 s, regardless of the NO
2 concentration. Stability is another key quality indicator in the development of gas sensors for real markets. Thus, we measured the response of the WN/WO
3 composites-based sensor for 7 weeks, once a day (every 24 h) at room temperature for a week and then once a week for 6 weeks, shown in
Figure 5b. We can observe that the sensor experiences a loss in response of less than 12.27% after 7 weeks of aging. The response of the sensor comes into saturation after 2 weeks. The results indicate that the WN/WO
3 composites sensor exhibits good long-term stability.
Dynamic sensing transients of WN/WO
3 composites and WO
3 nanosheets to 100 ppb NO
2, at room temperature and 300 °C (the optimum working temperature of WO
3, data shown in
Figure S2) respectively, are shown in
Figure 6. It can be observed that the
Ra of WN/WO
3 composites is lower than that of WO
3 nanosheets, which confirms that the introduction of WN into WO
3 can decrease the resistance of the materials. In addition, the responses of WN/WO
3 composites and WO
3 nanosheets sensors to 100 ppb NO
2 were 23.7 and 6.2, respectively, which suggests that gas sensing responses towards NO
2 can be enhanced by partial nitrogenizing of WO
3. Furthermore, the introduction of WN into WO
3 apparently decreases the working temperature from 300 °C to room temperature.
2.3. Gas Sensing Mechanism
Typically, the gas sensing mechanism of WO
3 belongs to surface-controlled type, where the adsorption and desorption of NO
2 molecules on the surface of WO
3 play a vital role [
24]. In ambient air, the oxygen molecules will adsorb on the surface of WO
3 and trap electrons from the conduction band of WO
3 to form ionic oxygen species (O
2−, O
− and O
2−), decreasing the electron concentration of the surface of WO
3, resulting in a high-resistance depletion layer at the surface [
37]. When the NO
2 gases come in, the NO
2 molecules, which are highly electrophilic, will capture electrons from the conduction band of WO
3, and will also react with the adsorbed ionic oxygen species. This will lead to a further decrease in electron concentration, increasing the resistance of WO
3. The reactions taking place on the surface of WO
3 are as follows.
The as-fabricated WN/WO
3 composites-based sensor, when compared with WO
3 sensor, exhibited enhanced sensing performances. The reasons why WN/WO
3 composites-based sensor shows enhanced gas sensing characteristics are still unclear and worthy of further investigation. The following can be considered as a plausible explanation based on our experimental results. Firstly, WN exhibits metallic properties with a high electron concentration of about (5–6) × 10
20 cm
−3 and a resistivity of (1–2) × 10
−1 Ω at room temperature [
38]. However, WO
3 is an n-type semiconductor, whose resistivity is far larger than that of WN. Thus, the electrons will transfer from the high concentration side to the low concentration side because of diffusion effect, resulting in a lower resistivity of the composites when compared with bare WO
3. The lower baseline of resistivity of WN/WO
3 composites will eventually increase the response (
S = (
Rg/
Ra − 1) × 100%) of the sensor. In addition, WN and WO
3 themselves did not show any sensing properties for NO
2 at room temperature (
Figure S3). However, the WN/WO
3 composites exhibit excellent sensing characteristics to NO
2 at room temperature. This may largely prove that there might be some synergetic effect in our WN/WO
3 composites, leading to room temperature sensing for NO
2.