NO2 sensing properties of WO3-decorated In2O3 nanorods and In2O3-decorated WO3 nanorods

In2O3 nanoparticle (NP)-decorated WO3 nanorods (NRs) were prepared using sol–gel and hydrothermal methods. The In2O3 NRs and WO3 NPs were crystalline. WO3 NP-decorated In2O3 NRs were also prepared using thermal evaporation and hydrothermal methods. The NO2 sensing performance of the In2O3 NP-decorated WO3 NR sensor toward NO2 was compared to that of the WO3 NP-decorated In2O3 NR sensor. The former showed a high response to NO2 due to a significant reduction of the conduction channel width upon exposure to NO2. In contrast, the latter showed a far less pronounced response due to limited reduction of the conduction channel width upon exposure to NO2. When the sensors were exposed to a reducing gas instead of an oxidizing gas (NO2), the situation was reversed, i.e., the WO3 NP-decorated In2O3 NR exhibited a stronger response to the reducing gas than the In2O3 NP-decorated WO3 NR sensor. Thus, a semiconducting metal oxide (SMO) with a smaller work function must be used as the decorating material in decorated heterostructured SMO sensors for detection of oxidizing gases. The In2O3 NP-decorated WO3 NR sensor showed higher selectivity for NO2 compared to other gases, including reducing gases and other oxidizing gases, as well as showed high sensitivity to NO2.


Introduction
Despite the numerous merits of semiconducting metal oxides (SMOs) as sensor materials there are still certain limitations, such as their relatively low response to gases at room temperature and dissatisfactory selectivity [1]. To address the dissatisfactory sensing properties, various strategies have been attempted, including noble metal catalyst doping, heterojunction formation, and radiationassisted treatment with energetic particles including ion beams, electrons, and ultraviolet (UV) lights [2][3][4]. Of these techniques, heterostructure formation is plausibly most widely studied and is used for the fabrication of chemiresistive nanostructured gas sensors. There are several types of heterostructures including p-n, n-n and p-p heterostructures. Generally, p-p heterostructures are less commonly utilized because of their inferior sensing properties, whereas n-n heterostructures are as widely utilized as the p-n counterparts because of their superior sensing properties [5]. However, strangely, n-n heterostructures have not been studied as intensively as the p-n congeners. The enhanced sensing properties of n-n heterostructures are mainly due to the resistance modulation at the n-n heterojunctions in n-n heterostructures. Various heterostructure combinations are known, such as a simple mixture of two different types of n-SMOs [6], bi-layer type n-n nanostructures [7], n-n core-shell structures [8], a single type of n-SMO nanostructure decorated with another type of n-SMO nanoparticles (NPs) [9], etc.
This study focuses on, decorated n-n heterostructures. WO 3 O 3 NRs are compared and the differences in the sensing properties of these four nanostructures are analyzed and the origin of the differences is discussed in detail.

Preparation of In 2 O 3 nanoparticles-decorated WO 3 nanorods
High purity In 2 O 3 NPs were synthesized using a solgel method [10].

Preparation of WO 3 nanoparticles-decorated In 2 O 3 nanorods
In 2 O 3 NRs were synthesized using a thermal evaporation method [12]. A 3 mm thick gold film-coated p-type Si (100) substrate was placed on the top of an alumina boat containing a mixture of In 2 O 3 powders and positioned at the center of a horizontal quartz tube furnace. The furnace was heated to 900 °C and maintained at that temperature for 30 min under argon gas at a constant flow rate of 200 cm 3 /min. The WO 3 NPs were synthesized using a hydrothermal method [13]. WO 3 powders (2 mL) were dissolved in 48 mL of hydrochloric acid in sonicater. The pH of the solution was controlled at 7 using sodium hydroxide. After sonication of the solution for 6 h the precipitated powders were collected by removing the liquid, leaving the powders behind. The powders were placed into a hydrothermal synthesizer containing ethanol and the synthesizer was placed in an oven and heated at 180 °C for 12 h. WO 3 NPs were synthesized in the hydrothermal synthesizer. The substrate on which the In 2 O 3 NRs were synthesized by the thermal evaporation method was placed in a beaker containing ethanol and then ultrasonicated to separate the In 2 O 3 NRs from the substrate. Meanwhile, the WO 3 NPs synthesized by the hydrothermal method were dispersed in ethanol. The two solutions (In 2 O 3 NRs dispersed in ethanol and the WO 3 NPs dispersed in ethanol) were mixed and the mixed solution was exposed to UV (254 nm) irradiation for 12 h using a UV lamp. The mixed solution was then annealed under argon atmosphere at 400 °C for 1 h in an annealing furnace.

Fabrication of chemiresistive sensors
The

Characterization
The microstructures and phases of the synthesized NR samples were examined by scanning electron microscopy (SEM) and X-ray diffraction (XRD), respectively. The microstructures and phases of the samples were examined further by transmission electron microscopy (TEM).

Gas sensing tests
The NO 2 sensing performances of the fabricated sensors were examined using a custom-made gas sensing system. The concentration of NO 2 gas was controlled precisely in the concentration range of 5-200 ppm by mixing NO 2 with dry synthetic air using the mass flow controllers. Electrical measurements to examine the sensing properties of the sensors were conducted at room temperature under 50% relative humidity. The detailed sensing test procedure is described elsewhere [14]. The response of the sensors to NO 2 was evaluated by using the R g /R a ratio, where R g and R a are the resistances of the sensor measured in the presence of air and NO 2 , respectively. The response and recovery times were determined by measuring the times required to reach 90% of the total change in the resistance of the sensor after exposure of the sensor to the analyte gas and ambient air, respectively.  taller and sharper than WO 3 peaks, which might be due to the larger volume of In 2 O 3 NRs than those of the WO 3 NPs. Figure 3a, b present the high-resolution TEM image and corresponding selected area electron diffraction (SAED) pattern of the In 2 O 3 NP-decorated WO 3 NRs. The regularly aligned fringes in both the WO 3 and In 2 O 3 regions suggest that the WO 3 and In 2 O 3 nanostructures are both crystalline. The corresponding spotty electron diffraction (ED) pattern in Fig. 3b reveals that the WO 3 and In 2 O 3 nanostructures are single crystals.

Results and discussion
The temperature-dependent responses of all four different sensor materials to NO 2 are presented in Fig. 4. The responses of all the four sensor materials to NO 2 tended to increase with increasing temperature up to 300 °C, and then to decrease with further increases in the temperature. This result suggests that 300 °C is the optimal operating temperature of the sensors in detecting the NO 2 . All the sensing tests hereafter were conducted at 300 °C. At too low operating temperature (250 °C or lower), the NO 2 molecules may not have enough energy to overcome the energy barrier of adsorption, and fail to be adsorbed on the surface of the sensor materials, WO 3 and In 2 O 3 . However, at too high operating temperature (350 °C or higher), adsorption failure might also occur because the rate of desorption may outweigh that of adsorption [15]. Figure 5a-d present the dynamic response curves of the four different sensors toward NO 2 . All the sensors showed stable and reversible response and recovery behavior. The resistances of the sensors increased when an oxidizing gas (NO 2 ) was supplied, and recovered to the initial value when the NO 2 supply was stopped and the sensors were exposed to ambient air. This response toward the oxidizing gas is in accord with the sensing behavior of n-type semiconductors. As is well known, both WO 3 and In 2 O 3 are n-type semiconductors. The resistance changes increased as the NO 2 concentration was increased. The starting resistances of the pristine and WO 3 NP-decorated In 2 O 3 NRs was markedly lower than the pristine and In 2 O 3 NPs-decorated WO 3 NRs, respectively, which might be due to the much lower resistivity of In 2 O 3 than that of WO 3 . Figure 6 shows the responses of the four different sensors to NO 2 as a function of the NO 2 concentration. The response of the In 2 O 3 NP-decorated WO 3 NRs to NO 2 far exceeded those of the other three sensors over the entire NO 2 concentration range. The more pronounced response of the In 2 O 3 NP-decorated WO 3 NR sensor to NO 2 than that of the pristine WO 3 NRs and the greater response of the WO 3 NPs-decorated In 2 O 3 NRs sensor to NO 2 than that of the pristine In 2 O 3 NRs is plausibly due to the resistance modulation at the WO 3 -In 2 O 3 heterojunction formation [16]. Contrarily, the much stronger response of the In 2 O 3 NP-decorated WO 3 NR sensor to NO 2 than that of the WO 3 NP-decorated In 2 O 3 NR sensor is very interesting. The origin of this difference in the response of the heterostructured sensors with inverse configuration is discussed in detail in the next section. Figure 7a, b show the response and recovery times of the four different sensors toward NO 2 as a function of the NO 2 concentration. As expected, the response and recovery times of the In 2 O 3 NP-decorated WO 3 NR sensor were shorter than those of the pristine WO 3 NRs. In contrast, the response and recovery times of the WO 3 NPdecorated In 2 O 3 NR sensor were longer than those of the pristine In 2 O 3 NR sensor. Comparison of the response and recovery times of the In 2 O 3 NP-decorated WO 3 NR sensor with those of the WO 3 NP-decorated In 2 O 3 NR sensor, interestingly, shows shorter response and recovery times for the former in the higher NO 2 concentration range, whereas longer response and recovery times for the lower NO 2 concentration range than the latter. Shorter response and recovery times are commonly associated with a higher response for gas sensors.
The response of the In 2 O 3 NP-decorated WO 3 NR sensor to various gases is shown in Fig. 8. The sensor showed a much stronger response to NO 2 than to the other oxidizing gases such as O 3 and SO 2 or reducing gases such as CO, CH 4 and H 2 S, demonstrating the selectivity and sensitivity of the In 2 O 3 NP-decorated WO 3 NR sensor toward NO 2 . The selectivity of the sensor toward NO 2 against other gases might be related to the different optimal operating temperatures of the sensor for different target gases. The response of a sensor material to a certain gas might depend on many factors such as solid solubility of the gas in the material, the decomposition rate of the adsorbed molecule at the material surface, the charge carrier concentration in the material, the Debye length in the material, the catalytic activity of the material, the orbital energy of the gas molecule, etc. The dissociation (or reduction) rate of an oxidizing gas such as NO 2 is determined by these factors. Therefore, each gas has the characteristic optimal dissociation temperature at which its dissociation rate is maximized. The In 2 O 3 -decorated WO 3 nanorod sensor fabricated in this study showed higher response fortunately to NO 2 than other gases at 300 °C because of the higher dissociation rate of NO 2 at the surface of In 2 O 3 and WO 3 at the temperature, but it might show higher responses to other gases than NO 2 at different temperatures [17][18][19][20][21][22]. Figure 9a-d illustrate the sensing mechanism of the In 2 O 3 NP-decorated WO 3 NR sensor toward NO 2 . Earlier studies reported that the response of a base sensor material could be enhanced by decoration with another type of SMO NPs, mainly because of the greater modulation of the width of the depletion layer or the conduction channel, resulting in the greater modulation of the sensor resistance [23][24][25]. n-Type WO 3 has a larger work function (qΦ) than n-type In 2 O 3 (Fig. 9a). Accordingly, if WO 3 and In 2 O 3 are in contact, even under vacuum, electron transfer from In 2 O 3 (with a larger work function) to WO 3 (with a smaller work function) tends to occur until electronic equilibrium is attained between WO 3 and In 2 O 3 , as shown in Fig. 9a. Consequently, electron-accumulation and electron-depletion layers are formed in the WO 3 and In 2 O 3 regions, respectively. The schematic shows the In 2 O 3 NP-decorated WO 3 NRs with an accumulation layer with a width of W 11 , formed by electron transfer from the In 2 O 3 NPs to the WO 3 NRs (Fig. 9b). In ambient air, the surfaces of the WO 3 NR and In 2 O 3 NP adsorb oxygen molecules and the adsorbed oxygen molecules are ionized by the capture of the free electrons in the WO 3 and In 2 O 3 surface regions (Fig. 9c). Consequently, a depletion layer with a width of W 12 is formed in the surface region of WO 3 . The schematic shows a decorated WO 3 NR with a depletion layer formed via ionization of adsorbed oxygen molecules and an accumulation layer formed by electron transfer from the In 2 O 3 to the WO 3 (Fig. 9c). When NO 2 gas is supplied, NO 2 and O 2 molecules are both adsorbed by the In 2 O 3 and WO 3 surfaces. The adsorbed NO 2 molecules are converted into NO 2 − or NO [26,27] and the adsorbed oxygen molecules are converted into oxygen ions by capturing electrons from the WO 3 and In 2 O 3 surface regions. Consequently, a thicker depletion layer (with a width of W 22 ) (Fig. 9d) is formed than that formed in ambient air. The schematic shows a WO 3 NR with a depletion layer with a width of W 22 as well as the accumulation layer with a width of W 21 formed by the electron transfer from the In 2 O 3 to the WO 3 (Fig. 9d).
The sensing mechanism of the WO 3 NP-decorated In 2 O 3 NR sensor toward NO 2 is illustrated in Fig. 9e, f.  As discussed above, electron-accumulation and depletion layers are formed in the WO 3 and In 2 O 3 regions, respectively. Thus, a depletion layer with a width of W 31 forms on the In 2 O 3 side of the WO 3 -In 2 O 3 interface (Fig. 9e). In ambient air, oxygen molecules are adsorbed by the In 2 O 3 NR surface and ionized by accepting the electrons from the In 2 O 3 and WO 3 surface regions. Consequently, a depletion layer with a width of W 32 is formed in the In 2 O 3 surface region. A WO 3 NP-decorated In 2 O 3 NR with a depletion layer formed due to the ionization of adsorbed oxygen molecules and a depletion layer formed by electron transfer from the WO 3 NP is shown in Fig. 9e. Under NO 2 atmosphere, a thicker depletion layer (with a width of W 42 ) than that generated in ambient air is formed due to the adsorption and ionization of both NO 2 and O 2 molecules (Fig. 9f ). Note that no electron-accumulation layer is formed in the In 2 O 3 NR throughout the on-off cycling of the NO 2 gas supply.
Under ambient air and NO 2 , there was no big difference in the basic response of the In 2 O 3 NPs-decorated WO 3 NRs versus that of the WO 3 NPs-decorated In 2 O 3 NRs. A relatively thin depletion layer is formed in both samples upon exposure to air and a thick depletion layer is generated upon exposure to NO 2 . Consequently, the width of the conduction channel of the In 2 O 3 NP-decorated WO 3 NRs formed upon exposure to NO 2 is much smaller than that of the WO 3 NPdecorated In 2 O 3 NRs formed in ambient air. The conduction channel of the In 2 O 3 NP-decorated WO 3 NRs has a room for substantial reduction upon exposure to NO 2 because the conduction channel width has already been expanded due to the formation of an accumulation layer by the transfer of electrons from the In 2 O 3 NP to the WO 3 NR. In contrast, the conduction channel of the WO 3 NPs-decorated In 2 O 3 NRs was already shrunken due to the formation of the electron-depletion layer via electron transfer from the In 2 O 3 NR to the WO 3 NP. Accordingly, the conduction channel of the In 2 O 3 NR has little room for further reduction upon exposure to NO 2 [28].
The response, S is defined as R g /R a . for the oxidizing gas NO 2 and S is proportional to A a /A g because the resistance R = ρl/A, where ρ, l and A are the density, length and cross-sectional area of the conductor (channel, here) [29]. S can be expressed as the ratio of the conduction channel width for an analyte gas to that for air, S = W a 2 /W g 2 because A = πW 2 , where W is the conduction channel width. Therefore, the In 2 O 3 NP-decorated WO 3 NR sensor has a higher response, S to NO 2 because of the far smaller conduction channel width, W g in NO 2 atmosphere. In contrast, the WO 3 NP-decorated In 2 O 3 NR sensor has a lower response, S because

Conclusions
The sensing properties of the In 2 O 3 NP-decorated WO 3 NR sensor toward NO 2 were compared to those of the WO 3 NP-decorated In 2 O 3 NR sensor. The response of the former sensor to NO 2 was more pronounced than that of the latter due to the significant reduction of the conduction channel width of the former sensor upon exposure to NO 2 . The conduction channel of the In 2 O 3 NP-decorated WO 3 NR sensor had room for sufficient reduction as it was already expanded by electron transfer from the In 2 O 3 NPs to the WO 3 NRs. In contrast, the WO 3 NP-decorated In 2 O 3 NR sensor showed a lower response due to insufficient reduction of the conduction channel width upon exposure to NO 2 . The conduction channel of the WO 3 NP-decorated In 2 O 3 NR sensor had little room for further reduction due to prior shrinkage associated with electron transfer from the In 2 O 3 NRs to the WO 3 NPs. For the detection of a reducing gas instead of an oxidizing gas, the magnitude of the sensor response would be reversed. Therefore, choosing a proper decorating material in fabricating n-SMO NR sensors decorated with n-SMO NPs is important in obtaining high sensitivity. An SMO with a smaller work function must be chosen as a decorating material in a decorated heterostructured sensor for oxidizing gas detection. In contrast, an SMO with a larger work function must be chosen as the decorating material for heterostructured sensors geared toward the detection of a reducing gas.