The sensor developed in this study operates on the principle that changes in the coverage of the adsorbed or chemisorbed gas species on the sensing film cause a detectable change in the electrical conductance of the film. Gas sensing devices are typically designed to operate at elevated temperatures in order to activate the reactions which produce the sensor response and to reduce humidity effects [22]. However, a drawback of such devices is that the observed response may actually be the result of the presence of more than one gas in the atmosphere. Therefore, to enhance the “selectivity” of a device, its operating temperature must be optimized in order to reduce the effects of competing reactions. In the present study, a Pt micro-heater is fabricated on the sensing layer [23]. When an electrical voltage is applied to this heater, its temperature increases, generating a simultaneous heating effect in the sensing layer immediately above it. The optimal operating temperature for the sensor is determined by varying the temperature of the heater, i.e. by varying the applied voltage, and comparing the detection sensitivities at different temperatures.

As shown in Figure 1, the benzene sensor developed in this study comprises a quartz substrate, Pt micro-heater/ IDEs and a WO₃ sensing layer. The micro-heater provides an instantaneous and precise temperature control capability, while the IDEs facilitate the direct electrical measurement of the change in conductivity of the sensing layer induced by the presence of benzene in the atmosphere. The grain size of the sputtered WO₃ sensing layer is very small, and therefore both the sensitivity and the response time of the device are enhanced due to the increased contact area between the catalyst grains and the sensed gas.

The WO₃ films were prepared using an RF magnetron sputtering system with a W target of 99.99% purity [24–26]. The oxide layer was sputtered on a quartz substrate located at a distance of 11.4 cm from the W target. Sputtering was performed under a gas pressure of 0.01 torr with the target maintained at a constant RF power of 200 W. The reactive sputtering gas was a mixture of argon (50%) and pure oxygen (50%). Prior to deposition, the chamber was pumped to a background pressure of 10⁻⁶ torr for 1 h and a pre-sputtering process was performed for 10 min to remove any traces of contamination from the target surface. Subsequently, RF sputtering was performed for 10 h, resulting in a WO₃ film with a thickness of approximately 4.4 μm.

Figure 2 presents a schematic illustration of the fabrication process used to realize the current benzene sensor. Following the deposition of a thin layer of Cr (0.05 μm) which served as the adhesion layer for a Pt layer (0.2 μm) deposited using an electron-beam evaporation process on the quartz substrate, the WO₃ sensing layer was sputtering deposited. A standard lift-off technique was employed to pattern the Pt micro-heater and IDEs. The resistance of the heater was designed to be 30 Ω. Following the WO₃ sputtering deposition process, the sensor was annealed in dry air. In the annealing process, the sensor was heated from 25 °C to 500 °C over a period of 60 min, then maintained at a temperature of 500 °C for 90 min, and finally cooled back down to 25 °C over 2 h.

Figures 3 show SEM photographs of the as-deposited WO₃ thin films. From inspection, voids were found on the substrate after a deposition time of 100 min but crystals of WO₃ did not appear (Figure 3(a)). In Figure 3(b), WO₃ crystals appeared with many voids after a deposition time of 150 min. After a deposition time of 240 min, WO₃ crystals grew and connected with cracks which could enhance the sensitivity of the gas sensor due to the porosity on the sensing layer (Figure 3(c)). In Figure 3(c), the grain size is found to be approximately 0.2 μm, which is helpful to increase the contact area of the sensing layer with the sensed gas. Figures 4 present SEM images of sputtered WO3 thin film with deposition of 240 min after annealing. It can be seen smaller grains appeared after annealing. Hence, it is apparent that annealing enhances the sensing performance of the sensor by increasing the area/volume ratio in the sensing layer through reducing the grain size. Finally, the AFM image in Figure 5 reveals surface morphology of the sputtered WO₃ thin film. It can be found the materials of the sputtering deposited thin film are uniform and the average roughness is 11.42 nm in the current study.

Figures 6 show the diffraction patterns of WO₃ thin-film samples deposited at a sputtering temperature of 25 °C. In general, when a thin film is to be used as a sensing layer, it is desirable that the film has a non-random orientation. In Figure 6, the WO₃ film was not formed with good crystalinity until that the specified sputtering annealing conditions resulted in a WO₃ film with good crystallinity and a pronounced (002) preferred orientation with a deposition time of four hours.

Figure 7 presents the variation of the micro-heater temperature with the applied power. It can be seen that the temperature increases linearly with increasing power at a rate of approximately 54°C/W.

As shown in Figures 8, linear dependency is observed between the resistance and the benzene concentration from 0 to 5 ppm in Figure 8(a) and from 0 to 1.0 ppm in Figure 8(b), respectively, at different sensor working temperatures (i.e., micro-heater temperatures). The slopes of the plotted lines represent the sensitivity of the device and are found to be 1.00 KΩ ppm⁻¹ at 300 °C, 0.75 KΩ ppm⁻¹ at 250 °C and 0.60 KΩ ppm⁻¹ at 200 °C, respectively (Figure 9). Thus, it is apparent that the optimal working temperature of the benzene gas sensor is 300 °C. The minimum detection limit of the proposed sensor is also determined to be 0.2 ppm. Note that the detection limit is significantly lower than those reported for similar MEMS-based devices presented in previous studies.

In conventional gas measurement devices, the time required for benzene concentration measurement can vary from tens of seconds to minutes. However, a requirement exists for sensors with a real-time gas detection and measurement capability. Figures 10 present the time response of the benzene gas sensor at various working temperatures as the benzene concentration surrounding the sensor is increased from 0 to 6 ppm by abruptly removing a shutter near the sensor in the test chamber. From inspection, the average time constants of the gas sensor are found to be 70 s, 50 s and 35 s at working temperatures of 200 °C, 250 °C and 300 °C, respectively. Please note the resistance increases after the detection of the lowest resistance because the gas sensor reaches its saturation level.

Figures 11 present the repeatability results obtained by repeatedly increasing the concentration from 0 to 20 ppm, maintaining this concentration for a period, and then reducing to 0.

The results indicate that the benzene gas sensor has a recovery time of 120 s, 60 s and 30 s at working temperatures of 200 °C, 250 °C and 300 °C, respectively. The recovery time reduces as the sensing layer temperature increases to be 300 °C due to the offered thermal and kinetic energy for returning the chemical reactions within the sensing layer at the temperature for recovery characteristics.