NO and NO2 Sensing Properties of WO3 and Co3O4 Based Gas Sensors

Semiconductor-based gas sensors that use n-type WO3 or p-type Co3O4 powder were fabricated and their gas sensing properties toward NO2 or NO (0.5–5 ppm in air) were investigated at 100 °C or 200 °C. The resistance of the WO3-based sensor increased on exposure to NO2 and NO. On the other hand, the resistance of the Co3O4-based sensor varied depending on the operating temperature and the gas species. The chemical states of the surface of WO3 or those of the Co3O4 powder on exposure to 1 ppm NO2 and NO were investigated by diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy. No clear differences between the chemical states of the metal oxide surface exposed to NO2 or NO could be detected from the DRIFT spectra.


Introduction
Since environmentally hazardous gases include toxic and greenhouse effect gases, the threshold limit value, which is defined as the maximum concentration of a chemical allowable for repeated exposure without producing adverse health effects, is regulated by the American Conference of Governmental Industrial Hygienists [1]. Effective and inexpensive systems for the detection and quantification of environmentally hazardous gases are required. Standard air pollution measurements

OPEN ACCESS
are still based on time-consuming and expensive analytical techniques such as optical spectroscopy and gas chromatography [2,3]. Gas sensors have been considered as promising candidates for measurement of environmental pollution levels because of their low cost, high sensitivity, fast response, and direct electronic interface.
Environmentally hazardous gases can be classified into oxidizing gases (such as NO 2 , CO 2 , and Cl 2 ) and reducing gases (such as NO, H 2 S, CO, and C 2 H 5 OH). When an oxidizing gas is steamed on an n-type semiconductor surface, the concentration of electrons on the surface decreases and the resistance of the n-type semiconductor increases. In the case of a p-type semiconductor, the concentration of electrons on the p-type semiconductor surface decreases and the resistance of the p-type semiconductor decreases because the extracted electrons result in the generation holes in the valence band. When the reducing gas is streamed on a metal oxide semiconductor, the gas reacts with the oxygen ions on the semiconductor surface, releasing electrons back to the conduction band. Therefore, when the concentration of electrons on the semiconductor surface increases, the resistance of the n-type semiconductor decreases and that of the p-type semiconductor increases because the generated electrons recombine with holes [4].
Many kinds of NO x (NO and NO 2 ) gas sensors including metal oxide semiconductors [5][6][7] and solid electrolytes [8,9] have been investigated. Among metal oxide semiconductors, n-type semiconductors, specifically those based on WO 3 , which are highly sensitive, are promising candidates that can be used for the detection of NO x gas [10][11][12]. Despite the large number of reports on the use of metal oxide semiconductors for the detection of NO x , only a few make a clear distinction between the response toward NO and NO 2 . Because NO is easily oxidized to NO 2 in air, the detection of NO 2 gas is carried out via the oxidation of NO by an oxidizing agent such as alumina supported potassium permanganate or by oxygen in air over a catalyst such as Pt [13,14]. To develop a high-performance NO gas sensor, it is essential to understand the means of optimizing the semiconductor that constitutes the sensor.
In the present work, the gas sensing properties of the sensor element that uses both n-type WO 3 and p-type Co 3 O 4 toward the NO 2 and NO were examined. The oxidation of NO was a thermally activated reaction in air atmosphere so that the in situ observation of the chemical state of NO 2 and NO on the surface of the sensing material at the temperature of operation may provide important information on the gas detection mechanism. A diffuse reflectance infrared Fourier transform (DRIFT) spectroscopy is an excellent analysis to obtain the chemical state on the surface of the sensing material and has been performed to elucidate the gas detection mechanism of the gas sensors [15,16]. The chemical states of NO 2 and NO on the surface of the semiconductor oxide were investigated by the DRIFT spectroscopy at a specific working temperature could be changed to indicate the two temperature used in this project.

Experimental
WO 3 powder (99.5%, Wako Pure Chemical, Osaka, Japan) and Co 3 O 4 powder (average particle size: 20~30 nm, 99.8%, Sigma-Aldrich, St. Louis, MO, USA) were mixed with an organic dispersant, consisting of a mixture of ethyl cellulose and terpineol, to obtain a paste. The weight ratio of the mixture of ethyl cellulose and terpineol was 1:9. The weight ratio of the powder and the organic dispersant was 1:16. The paste was dispensed on a 5 × 9.5 mm 2 surface-oxidized Si substrate,

Sensors 201
which consi each. The su n-type WO 3 The cros field-emissio cross-section form the sa FE-SEM im   DRIFT spectrum was obtained. Then, the powder was purged to remove NO or NO 2 by air flow and the DRIFT spectrum was once again obtained. The spectra were recorded at a spectral resolution of 1 cm −1 with 256 scans. Figure 2a,b show the response of the WO 3 sensor element to NO 2 at 100 °C and 200 °C, respectively. When NO 2 gas was introduced, the resistance of the WO 3 sensor element increased with increase in the concentration of NO 2 . This is the typical response of an n-type oxide toward an oxidizing gas, leading to R g > R a . At 200 °C, R a of WO 3 decreased with temperature, and the sensor responses were higher than those at 100 °C. At 200 °C, the response of the WO 3 sensor element increased to S = 19.2 at 1 ppm of NO 2 and the response was adequately linear. The response and recovery times of the resistance of the WO 3 sensor element reduced with increasing the operating temperature, but the resistance did not reach the saturation even after 15 min for NO 2 exposure. The response time of the WO 3 sensor element was not so fast in comparison with the other sensors [5][6][7]. In order to reduce the response and recovery times, the optimum operating temperature is required.  When NO gas was introduced at 100 °C, the resistance of the sensor element immediately increased and then subsequently decreased within 10 min. No clear relationship between the gas concentration and sensor response could be observed. Therefore, the resistance of the sensor to 5 ppm of NO was smaller than that to 0.5 and 1 ppm NO. The unexpected changes of the sensor resistance of WO 3 are supposed to be under the influence of the low operating temperature of 100 °C, which is not high enough for desorption of the reaction product on the WO 3 . However, when the WO 3 sensor was exposed to NO gas at 200 °C, the resistance of the element increased with increase in the concentration of NO, similar to the response shown to NO 2 , as seen in Figure 2b. At 200 °C, the response of the WO 3 sensor element to 1 ppm NO was S = 2.2, and the response was adequately linear. The resistance of the WO 3 sensor element increased by exposure to NO 2 and NO at 200 °C. Although the resistance of the sensor based on the n-type WO 3 is considered to be decreased by exposure to reductive NO, the resistance increased, as shown in Figure 3a,b. Since NO with an unpaired electron is unstable state, NO easily reacts with oxygen in air to become a stable NO 2 . The reaction of NO to NO 2 proceeds with temperature [13]. It has previously been reported that NO could be partially oxidized to NO 2 and leading to adsorption of NO 2 on WO 3 (or Pt-doped WO 3 ) surface, which has been confirmed by temperature programmed desorption (TPD) analysis [17]. Therefore, we assumed that the resistance of the sensor on exposure to NO increased and the responses of the sensor toward NO exposure were smaller than those toward NO 2 exposure at 100 °C and 200 °C because of the partial oxidation of NO. Figure 4a,b show the response of the Co 3 O 4 sensor element toward NO 2 exposure at 100 °C and 200 °C, respectively. When NO 2 gas was introduced at 100 °C, the resistance of the sensor element immediately decreased. This may be attributed to the adsorption of NO 2 onto the surface of the p-type semiconductor Co 3 O 4 and to the role of NO 2 as an oxidizing gas at 100 °C. A linear relationship between the gas concentration and the sensor response was observed. The response of the Co 3 O 4 sensor element to 1 ppm of NO 2 at 100 °C was S = 2.2. The sensor resistance of NO 2 -exposed Co 3 O 4 did not reached to R a even after 15 min for air exposure; the recovery time of Co 3 O 4 sensor element was not so fast. In our preliminary experiment, the sensor resistance of NO 2 -exposed Co 3 O 4 reached the saturation within 60 min for air exposure. Therefore, the sensor resistance of NO 2 -exposed Co 3 O 4 in this work is also expected to reach the saturation within 60 min for air exposure. When the sensor was exposed to NO 2 at 200 °C, the resistance of the sensor element immediately increased and then gradually decreased on further exposure to NO 2 , which seems to indicate the role of NO 2 as a reducing gas at 200 °C. The decrease in the resistance of the Co 3 O 4 sensor increased with the concentration of the NO 2 gas. As a result, the resistance of the Co 3 O 4 element after 15 min of exposure to NO 2 decreased with the concentration of NO 2 gas. In our preliminary experiment, the sensor resistance of the Co 3 O 4 reached the saturation after 30 min for NO 2 exposure and did not reached to R a even after 60 min for NO 2 exposure. Therefore, the sensor resistance of the Co 3 O 4 in this work is also expected to reach the saturation within 30 min for NO 2 exposure. No clear linear relationship between the gas concentration and sensor response could be observed.   Figure 4a shows the R g decrease exhibited by the p-type Co 3 O 4 on exposure to NO 2 at 100 °C, however, the response of resistance R g of Co 3 O 4 reversed and R g increased on exposure to NO 2 at 200 °C. This contradictory result indicated a role of NO 2 as an oxidizing gas at 100 °C and as a reducing gas at 200 °C. However, it rermains unclear how an oxidizing gas such as NO 2 could become a reducing gas at a different temperature. As shown in Figure 5a,b, NO is expected to be adsorbed on the surface of Co 3 O 4 and act as a reducing gas, resulting in an increase in the sensor resistance. The sensor resistance of Co 3 O 4 on exposure to 5 ppm of NO at 100 °C increased within 3 min of exposure and then became slightly smaller than R a . The unexpected changes of the sensor resistance of Co 3 O 4 are supposed to be due to insufficient desorption of reaction product on the oxide surface, similar to the case of the WO 3 sensor. The reaction of Co 3 O 4 on exposure to 5 ppm of NO at 100 °C is assumed as follows: NO reacts with adsorbed oxygen on Co 3 O 4 surface and the sensor resistance increases; the reaction of NO and adsorbed oxygen generates NO 2 . Although the generated NO 2 is normally desorbed      At 200 °C, the peaks around 2,062, 1,861 and 1,421 cm −1 were also observed in WO 3 on NO 2 exposure. The peaks at about 2,062 and 1,861 cm −1 formed upon interaction with NO 2 at 200 °C were similar to those at 100 °C and the peaks could be assigned to the various overtones and combination modes of the bond between oxygen and tungsten in the lattice of the oxide. The peak at around 1,421 cm −1 could be assigned to nitrate species. This peak at 200 °C seemed much stronger than that at 100 °C and hence, the amount of the nitrate species on WO 3 surface at 200 °C was larger than that at 100 °C. When air was introduced to the WO 3 powder sensor after exposure to NO 2 , the intensity of the peaks at 100 °C and 200 °C decreased with the time of air flow, which indicated the desorption of NO 2 from the surface of WO 3 . Figures 9 and 10 show the DRIFT spectra of WO 3 powder on exposure to 1 ppm of NO at 100 °C and 200 °C, respectively. In Figure 9, negative peaks at around 2,062 and 1,861 cm −1 were seen upon the introduction of NO gas. WO 3 powder is supposed to be unreactive for NO at 100 °C because there was no difference between the spectra obtained on exposure to NO (Figure 9a-c) and on exposure to air (Figure 9d-f). However, as shown in Figure 10, the intensity of the peaks at around 2,062 and 1,861 cm −1 increased with the time of exposure to NO and decreased on introduction of air. The negative peaks of NO observed in Figure 9, Figure 10e,f may be attributed to possible glitches with the background subtraction. When NO reacts with adsorbed oxygen on the WO 3 surface, NO 2 formed and adsorbed on the surface of WO 3 , resulting in the increase in the intensity of the peaks shown in Figure 7 or Figure 8. Subsequently, NO 2 desorbed from the surface of WO 3 by air flow, and the intensity of the peaks reduced again. In the DRIFT spectra of WO 3 on exposure to NO 2 and NO, peaks corresponding to NO vibration and NO 2 asymmetric vibration bands could not be observed. Previous reports have shown the presence of peaks corresponding to CO vibration and CO 2 asymmetric vibration bands in the DRIFT spectra of SnO 2 exposed to CO, while peaks of W-O alone were observed in the DRIFT spectra of WO 3 exposed to CO [16,22]. We assumed that WO 3 may have insignificant interaction with NO 2 or NO gas in comparison with other metal oxide semiconductors. The positions of the peaks observed in the DRIFT spectra of WO 3 exposed to NO 2 were similar to those observed in the DRIFT spectra of WO 3 exposed to NO. Further, as shown in Figure 7, the peaks in the DRIFT spectra were small and there were hardly any differences between the DRIFT spectra. Therefore, the response of the sensor to NO exposure was more unstable than that to NO 2 exposure at 100 °C.  Figures 11 and 12 show the DRIFT spectra of Co 3 O 4 powder exposed to 1 ppm NO 2 at 100 °C and 200 °C, respectively. In Figure 11, four peaks around 1,609, 1,535, 1,430 and 1,268 cm −1 were observed which are reported to correspond to the NO vibration band of the bridging bidentate nitrate, the NO vibration band of the chelating bidentate nitrate, the NO 2 asymmetric vibration band of the monodentate nitrate, and the NO 2 asymmetric vibration of the bridging bidentate nitrate or chelating bidentate nitrate, respectively [20,21]. The intensity of the peaks increased with the time of exposure to NO 2 gas and the intensities did not decrease even after the replacing NO 2 flow with air flow.

DRIFT Spectra of Co 3 O 4 Powder
In Figure 12, two strong peaks at around 1,535 cm −1 and a weak peak around 1,268 cm −1 were observed, which could be assigned to the chelating bidentate nitrate. From the spectra shown in Figures 11 and 12, we could suggest that exposure of NO 2 gas to the Co 3 O 4 surface formed the bridging bidentate, chelating bidentate, and monodentate nitrates at 100 °C and the chelating bidentate nitrate at 200 °C.   Figures 13 and 14 show the DRIFT spectra of Co 3 O 4 powder exposed to 1 ppm NO at 100 °C and 200 °C, respectively. The peak patterns shown in Figures 13 and 14, are similar to the patterns shown by the DRIFT spectra of Co 3 O 4 powder exposed to NO 2 . Hence, the chemical states of the Co 3 O 4 surface achieved on exposure to NO 2 and NO seemed identical. Further, most peaks in the DRIFT spectra of Co 3 O 4 exposed to air and on exposure to NO 2 or NO gases were similar. Despite the changes observed in the resistance of Co 3 O 4 on exposure to various atmospheres, there was no clear difference in the DRIFT spectra of Co 3 O 4 powder exposed to various atmospheres. With the present data, it is difficult to discuss the origin of these contradictory results and we intend to investigate these in future. Figure 13. DRIFT spectra of Co 3 O 4 powder at 100 °C. The powders were exposed to 1 ppm of NO in air for (a) 2 min, (b) 20 min, and (c) 50 min. After NO exposure, the powders were purged by air for (d) 1 min, (e) 30 min, and (f) 60 min.
As in the case of the DRIFT spectra of Co 3 O 4 powder, no difference between NO 2 and NO exposures could be observed in the case of the DRIFT spectra shown by WO 3 powder. However, as shown in Figures 4 and 5, the sensor resistance of Co 3 O 4 varied depended on the operating temperature (100 °C or 200 °C) and the gas species (NO 2 or NO in air). It is to be noted that although the chemical state of the Co 3 O 4 surface exposed to NO 2 exposure was similar to that of the Co 3 O 4 surface exposed to NO at 100 °C, the sensor resistance decreased on exposure to NO 2 and increased on exposure to NO. With the change in the electron concentration on Co 3 O 4 surface, the sensor resistance changed. The resistance of the sensor on exposure to NO 2 decreased at 100 °C and increased at 200 °C. This can be viewed on the basis of the previous studies which report on the transitions from p-type to n-type behavior of several metal oxide sensors (or vice versa) [12,[23][24][25]. In particular, Zhang et al. showed that the transition of WO 3 was observed under 93 ppb NO 2 exposure at working temperature above 130 °C [12]. A similar transition seemed to have occurred in the case of the Co 3 O 4 sensor on exposure NO 2 at 200 °C in this study. On the other hand, no transition of WO 3 seemed to have occurred because NO 2 concentration was higher than 0.5 ppm in this study. Although the transition would be due to the oxygen adsorption and formation of inversion layer at the metal oxide surface, further works are necessary to clear the transition mechanism. In this work, we could not examine the DRIFT spectra of WO 3 or Co 3 O 4 powder in the sample cell but the DRIFT spectra of WO 3 or Co 3 O 4 film on the Si substrate with a platinum comb-type electrode; the difference between the reactions of the powder and the film might be present. Due to the difference between the reactions of the powder and the film, some contradictory results in this work are supposed to be appeared. The chemical states of NO 2 and NO on the sensor elements will be investigated in the future. Figure 14. DRIFT spectra of Co 3 O 4 powder at 200 °C. The powders were exposed to 1 ppm of NO in air for (a) 2 min, (b) 20 min, and (c) 50 min. After NO exposure, the powders were purged by air for (d) 1 min, (e) 30 min, and (f) 60 min.
In this study, we have dealt with the sensor responses of WO 3 and Co 3 O 4 elements on exposure to NO 2 or NO at different operating temperatures. In the case of the real-life application of the sensors, if the target gas consisting of an unknown mixture of NO 2 and NO is to be analyzed by WO 3 and Co 3 O 4 -based gas sensors, the results should be viewed carefully because the sensor response could be changed by the component ratio of NO 2 and NO in NO x .

Conclusions
We have investigated the gas sensing properties of the sensor element that uses n-type WO 3 or p-type Co 3 O 4 toward NO 2 and NO. Further, we also analyzed the chemical states of NO 2 and NO on the semiconductor oxide surfaces. The resistance of the WO 3 sensor at 100 °C and 200 °C increased with the concentration of NO 2 . The resistance of the WO 3 sensor to NO exposure first increased and then immediately decreased within 10 min at 100 °C, while at 200 °C, the resistance of the sensor increased. Since the sensor resistance of Co 3 O 4 exposed to NO increased at 100 °C and 200 °C, NO acted as a reducing reagent and oxidized to NO 2 . The sensor properties for NO exposure were consistent with the DRIFT spectra at 100 °C and 200 °C. On the other hand, since the sensor resistance of Co 3 O 4 exposed to NO 2 decreased at 100 °C, NO 2 acted as an oxidizing agent and the sensor properties for NO 2 exposure were consistent with the DRIFT spectra. At 200 °C, the sensor resistance of Co 3 O 4 exposed to NO 2 increased and the sensor properties for NO 2 exposure were inconsistent with the DRIFT spectra. Interestingly, no clear differences between the chemical states of the metal oxide surface exposed to NO 2 or NO could be detected from the DRIFT spectra of sensors based on either of the semiconductors. We think that NO was oxidized into NO 2 and was adsorbed on the surface of WO 3 or Co 3 O 4 as NO 2 .