Effect of Noble Metal Addition on Co3O4-Based Gas Sensors for Selective NO Detection

P-type semiconductor-based gas sensors have been fabricated for the selective detection of 50–200 ppb nitric oxide (NO). To fabricate the sensors, cobalt oxide (Co3O4) powders containing a noble metal (Pd, Ag, Pt, or Au) at 10 wt% were prepared by a colloidal mixing method and deposited on interdigitated Pt electrodes. The sensing responses to 50 and 200 parts per billion (ppb) NO and 25 and 100 parts per million (ppm) H2 in air were investigated. The sensors with 10 wt% Pd, Ag, Pt, or Au loaded on the Co3O4 powder demonstrated a greater response to NO than Co3O4 with no noble metal content. The addition of Ag was found to suppress the H2 response, a phenomenon that is attributed to the surface modification effects of Ag on the Co3O4 surface. Furthermore, the sensor with 5 wt% Pd and 5 wt% Ag loaded on the Co3O4 showed a high response in the presence of 50 ppb NO, as well as good selectivity against H2.


Introduction
Human breath includes small amounts of biomarker gases, such as hydrogen (H 2 ), nitric oxide (NO), carbon monoxide (CO), methane (CH 4 ), and various volatile organic components (VOCs), at low concentrations in the range from several parts per billion (ppb) to parts per million (ppm). (1)(2)(3) NO concentration in human breath is associated with airway inflammatory disorders such as asthma and bronchiectasis. (4)(5)(6) Although monitoring NO concentration in human breath is one of the best noninvasive screening tests for the early diagnosis of these inflammatory disorders, this screening process requires an analytic device with sufficient accuracy at the ppb level.
Semiconductor gas sensors based on metal oxides have been considered as promising candidates for NO measurement because of their low cost, compact size, and direct electronic interface. (7)(8)(9)(10)(11) Among the metal oxides, n-type tungsten oxide (WO 3 ) appears to be a promising material for use in a NO gas sensor. (8,9) In a previous publication, we reported on the reliable response of a p-type cobalt oxide (Co 3 O 4 ) gas sensor to NO gas at 0.5-5 ppm in air. (10) Furthermore, because that Co 3 O 4 gas sensor showed a low resistance of below 1 kΩ, the peripheral circuit used to measure the sensor resistance does not require a large resistance, reducing the device cost. However, to the best of our knowledge, the Co 3 O 4 -based gas sensor's responses to NO gas at several hundred ppb have not been studied. In the case of breath analysis, sensors for the detection of NO concentrations in the range from several ten to several hundred ppb require selectivity against other gases such as H 2 , CO, CH 4 , and VOCs. The H 2 concentration level in human breath has been measured at the several tens of ppm level, and it exists at higher concentrations than the other gases. (12) Moreover, the Co 3 O 4 -based gas sensor in our previous publication showed a low response even at 0.5 ppm NO, and responses to NO gas at several hundred ppb have not been studied. (10) Therefore, the addition of a noble metal to these metal oxide semiconductors is critical for improving the sensitivity and selectivity of these gas sensors. (13,14) We reported that the NO gas sensing properties of the Pd-loaded Co 3 O 4 sensor with 0.1-30 wt% Pd content were estimated and the sensor with 10 wt% Pd content showed a high response and signal-to-noise ratio. (15) In this study, Co 3 O 4 powder with 10 wt% noble metal (Pd, Ag, Pt, and Au) content was prepared by the colloidal mixing method, and the responses of these sensors to 50 and 200 ppb NO in air were investigated, in reference to the response of a sensor based on pure Co 3 O 4 . (14,15) To discuss the selectivity against other gases as an initial evaluation, their responses to H 2 in air were also investigated.

Materials and Methods
Five distinct colloidal suspensions were prepared by mixing commercial Co 3 O 4 powder (Sigma-Aldrich) with 10 wt% Pd, 10 wt% Ag, 10 wt% Pt, 10 wt% Au, and 5 wt% Pd + 5 wt% Ag particles. "10 wt% Pd" means that the total Pd content is 10 wt% in the powder. The particle size of the Co 3 O 4 powder was 20-30 nm, and the average particle sizes of the Pd, Ag, Pt, and Au in the colloidal suspension (Tanaka Kikinzoku Kogyo K.K.) were approximately 4, 15, 2, and 3 nm, respectively.
The Co 3 O 4 powder was stirred in an ethanol solution, followed by the addition of the Pd, Ag, Pt, or Au colloidal suspension. Subsequently, the solution was stirred at 70 °C until the ethanol evaporated. The resulting solid residue was baked in air at 400 °C for 2 h to make a sensing material powder with 10 wt% metal content. A ceramic paste of the sensing material powder was prepared by mixing an organic dispersant consisting of 10 wt% ethyl cellulose and 90 wt% terpineol. The weight ratio of the sensing material powder to the organic dispersant was 1:16. The paste was dispensed on a surface-oxidized Si substrate consisting of 2.5 × 4 mm 2 Pt interdigital electrodes with line and space definitions of 15 μm each. These units were then baked in air at 400 °C for 2 h to produce sensors with sensing membranes of 2-6 μm thickness on the Si substrate. A pure Co 3 O 4 sensor was also prepared in a similar manner. Figure 1 shows a schematic illustration of the sensor. The sample powders were characterized by X-ray diffraction (XRD) analysis and transmission electron microscopy (TEM). The XRD analysis was carried out using a Smartlab diffractometer (Rigaku Corporation) equipped with a copper source (CuKα) and a one-dimensional high-speed detector (D/teX Ultra 250). The X-ray generator was operated at 40 kV and 30 mA. The TEM analysis was performed using a JEM-2010 instrument (JEOL Ltd.).
The NO gas responses of the sensors were investigated using a flow test chamber heated to 100-300 °C in an electrical tube furnace. After placing a sensor in the chamber, air and NO in air were allowed to flow into the chamber alternately, at a flow rate of 200 mL/min. Air and NO in air were prepared by mixing with 99.99% N 2 gas, N 2 -balanced 10 ppm NO gas, and 99.5% O 2 gas. The O 2 content in air and NO in air was 20 vol%. The NO concentration was controlled to the values of 0, 50, and 200 ppb in air. The sensor resistance in various gaseous atmospheres was measured by a two-probe method at 10 s intervals using a K2700 digital multimeter (Keithley). We defined the sensor response value (S) using the following equation: S = R g /R a , where R g denotes the resistance after 15 min of NO gas exposure, and R a denotes the resistance in air prior to NO gas exposure.
The response of the sensors to H 2 gas was investigated using the same apparatus as was used to measure the NO gas response. The H 2 gas concentration was controlled to the values of 0, 25, and 100 ppm in air.  Fig. 2 (b), a small broad peak at 33.8° was observed in the pattern of the Co 3 O 4 sensor loaded with 10 wt% Pd. This pattern was assigned to PdO (JCPDS card no. 75-0584), and is attributed to the oxidation of Pd to PdO during the baking at 400 °C for 2 h. In Fig.  1(e), a peak with a broad width was observed at 38.2° in the pattern of the sensor loaded with 10 wt% Au. This peak was assigned to Au (JCPDS card no. 04-0784). No other peaks were observed in the patterns of the sensors loaded with 10 wt% Ag or 10 wt% Pt. The crystallite sizes (D) were calculated using the Scherrer equation D = kλ/βcosθ, where k (= 0.94) is the shape factor, λ is the X-ray wavelength, β is the full width of the diffraction peak at half its maximum, and θ is the Bragg diffraction angle.  Fig. 3(c), and dark phases of approximately 5 nm diameter can be observed in Figs. 3(d) and 3(e). These dark phases are considered to be the metal particles. No dark phases are apparent in the image of the sensor loaded with 10 wt% Pd in Fig. 3 (b); this is attributed to the phenomenon reported by Matsushima et al., (16) in which PdO particles on a SnO 2 surface were reported to be less visible in TEM images. Because the Pd in the powder existed as PdO, it showed less contrast to Co 3 O 4 , and therefore, it was indistinguishable from the Co 3 O 4 in the TEM image. We used Pd, Ag, Pt, and Au because of their dispersion properties in colloidal solutions. It is observed from Fig. 3 that the metal nanoparticles are bonded with the Co 3 O 4 particles without obvious aggregation, even though the metal weight ratio is 10 wt%. + 10 wt% Au powders, respectively, during exposure to 0, 50, and 200 ppb NO in air at 100 °C. When all sensors were exposed to the NO gas, the sensor resistances began to increase. Because the majority carrier of the Co 3 O 4 is the hole, the holes combine with electrons, and the hole concentration at the Co 3 O 4 surface decreases, resulting in the increasing resistance of the Co 3 O 4 . (17) When Co 3 O 4 is exposed to air, the O 2 absorbed on the Co 3 O 4 surface, resulting in the formation of absorbed oxygen ions (O 2 − , O 2− , or O − ). When a reductive NO gas flows onto the Co 3 O 4 , the NO gas combines with adsorbed oxygen ions at the Co 3 O 4 surface, releasing electrons back to the conduction band. Therefore, the resistance of the sensors increased with increasing exposure to NO gas. All the sensors showed a distinct response to the NO gas, even at 50 ppb. The sensor resistance was higher in the presence of 200 ppb NO than in the presence of 50 ppb NO. In Fig. 4(a), the resistance of the pure Co 3 O 4 membrane exposed to 50 ppb NO did not return to R a , even after 15 min of air exposure. This result shows that adsorbed NO on the pure   In addition, the resistance of NO − exposed sensors decreased during exposure to 0 ppb NO air. The time required to return to 90% of the steady-state value of R a or R g decreased with increasing operating temperatures. This suggests that increasing operating temperatures promoted the combustion of NO molecules or the adsorption of O 2 on the Co 3 O 4 surface. Figures 7(a)  Unlike the other metal-loaded Co 3 O 4 sensors, the sensing response of the Ag-loaded Co 3 O 4 sensor decreased with increasing operating temperature. This is attributed to the difference in the metal particle size, because the Ag particle is larger than the Pd, Pt, and Au particles. This relationship between the sensing response and the size of the metal particle will be investigated in the future. Figure 8 shows the sensor resistances of the pure Co 3 O 4 , Co 3 O 4 + 10 wt% Pd, Co 3 O 4 + 10 wt% Ag, Co 3 O 4 + 10 wt% Pt, and Co 3 O 4 + 10 wt% Au powders in air as a function of temperature. To determine each sensor's resistivity or conductivity, the thickness and porosity of each sensor's membrane should be estimated as precisely as possible. This is difficult to perform with sufficient accuracy, and we therefore plotted only the resistance in air (R a ) on the vertical axis. However, the temperature dependences of the resistance and resistivity are the same. Figure 8 shows decreasing resistance with increasing operating temperature. The resistance of the Co 3 O 4 membrane increased with the addition of Pd and increased even more with the addition of Pt. This increase in the sensor resistance is probably caused by a decrease in the hole concentration at the Co 3 O 4 surface. On the other hand, the addition of Ag or Au was not expected to cause a decrease in the hole concentration. The resistance of the Ag-or Au-loaded Co 3 O 4 membrane was slightly smaller than that of the pure Co 3 O 4 membrane. It is assumed that Ag or Au in the sensor membrane has not only some surface modification effects that enhance the sensor response, but also a conductive effect connecting the Co 3 O 4 particles and decreasing the sensor resistance.

Results and Discussion
When the sensors were exposed to H 2 gas, the resistances of all the sensors increased. The H 2 sensor responses (S H2 ) were estimated using S H2 = R g /R a . Figures 9(a)  , reducing gases (such as CO and CH 4 ), H 2 O, and VOCs. When the sensor is exposed to a oxidizing gas, the resistance of the sensor decreased. (17) On the other hand, when the sensor is exposed to H 2 O and VOCs, the sensor response is reported to be inaccurate. (18) Therefore, the Co 3 O 4 -based sensor must include a gas separation filter to remove the interference from gases such as NO 2 , CO 2 , CH 4 , CO, H 2 O, and VOCs. Among the noble-metal-loaded Co 3 O 4 -based sensors, the Ag-loaded sensor showed lower responses toward 25 and 100 ppm H 2 than the others. From this result, the Ag-loaded Co 3 O 4 sensor is considered to demonstrate good selectivity against H 2 gas. Yamazoe et al. reported that the gas behaviors of Pd-SnO 2 and Pt-SnO 2 were controlled by a chemical interaction in which noble metals assist the redox processes of the semiconductor, and the gas behavior of Ag-SnO 2 was controlled by an electronic interaction in which noble metals interact electronically with the semiconductor as an electron donor or acceptor. (13) The selectivity against H 2 gas may be affected by the different surface modification effects of the various noble metals on the Co 3 O 4 -based sensor membrane.
To improve the sensor response to NO, further investigation is required to detect sub-ppb levels of NO with sufficient accuracy. Sakai et al. have reported that the multiple-noble-metal (Pd, Pt, and Au)-loaded SnO 2 showed higher sensor response toward VOCs than a single-noble-metal-loaded SnO 2 . (19) To estimate the effects of multiple noble metal loadings, we investigated the responses of the Pd/Ag-loaded Co 3 O 4 sensor because the Pd-loaded Co 3 O 4 sensor showed the highest response toward 50 ppb NO, and the Ag-loaded Co 3 O 4 sensor showed the lowest response toward 100 ppm H 2 . Figures 10(a) and 10(b) show the responses of the Co 3 O 4 sensor loaded with 5 wt% Pd and 5 wt% Ag powders toward NO and H 2 as a function of operating temperature. The high responses were S NO = 1.20 and 1.47 for 50 and 200 ppb NO, respectively, at 200 °C. In addition, the responses for 25 and 100 ppm H 2 were S H2 = 1.26 and 1.47, respectively, at 200 °C. The Co 3 O 4 sensor loaded with 5 wt% Pd and 5 wt% Ag showed a higher NO response and a lower H 2 response than the Co 3 O 4 sensors loaded with a single noble metal. To enhance these sensor properties, the optimization of the doping ratio of Pd to Ag and Co 3 O 4 sensors loaded with multiple noble metals will be further investigated in the future.

Conclusions
In this study, NO gas sensors based on Co 3 O 4 as a sensitive functional material were manufactured, and their sensing properties were investigated. The sensors with 10 wt% noble metal (Pd, Ag, Pt, or Au) require less time to return to 90% of the steady-state values of R a or R g and have higher responses toward NO gas than Co 3 O 4 -based sensors with no noble metal content. Among the noble-metal-loaded Co 3 O 4 sensors, the sensor loaded with 10 wt% Ag content demonstrated good selectivity against H 2 gas. The Co 3 O 4 sensor loaded with 5 wt% Pd and 5 wt% Ag showed better NO response and lower H 2 response than the Co 3 O 4 sensors loaded with a single noble metal.