Sensitivity Enhancement of Acetone Gas Sensor using Polyethylene Glycol/Multi-Walled Carbon Nanotubes Composite Sensing Film with Thermal Treatment

There is a need to develop a chemiresistive gas sensor equipped with a thermostat over a wide area for the sensor, which can protect the sensor from the influence of ambient temperature due to the uniform temperature of the thermostat. In this paper, we demonstrated an acetone gas sensor based on a polyethylene glycol (PEG)/Multi-walled Carbon Nanotubes (MWCNTs) composite film, which was equipped with a thermostat. The sensor was operated at modest working temperatures for sensor sensitivity enhancement. The optimum design of the polyimide-based thermostat with widely uniform thermal distribution was investigated in detail. It was found that the temperature uniformity of the thermostat was achieved using double spiral geometry. The experimental results of the sensor response showed that the PEG/MWCNTs composite film with a moderate working temperature revealed a higher sensitivity than that without thermal treatment. Moreover, the sensing mechanisms of the PEG/MWCNTs composite gas sensor to acetone vapor were studied as well.


Introduction
Acetone [(CH 3 ) 2 CO] as a disease-specific biomarker [1][2][3] and a flammable industrial solvent can be hazardous and life threatening when human beings are exposed to its high concentration. The permissible exposure limit-time weighted average (PEL-TWA) of acetone is 750 ppm (1800 mg/m 3 ) and the permissible exposure limit-short time exposure limit (PEL-STEL) is 1000 ppm (2400 mg/m 3 ) [4]. Therefore, the detection of acetone in the environment or in exhaled breath has attracted attention in order to develop relative wearable and portable gas sensors for fire accident prevention and diagnosis of diabetes.

Preparation of Composite Sensing Film
Four different concentrations of PEG (0.5 g, 1.0 g, 1.5 g, and 2.0 g) were dissolved in DI water to prepare a 100 g PEG solution, and then the mixtures were sonicated for 3 h to achieve uniform PEG solutions. The sensing film was prepared drop-casting to form the bilayer sensor structure of the PEG/MWCNTs composite gas sensor. The top layer was a PEG film, and the bottom layer was MWCNTs film.

Optimum Design of Thermostat for Gas Sensor
For a sensor equipped with a thermostat for gas sensing, thermal distribution affects the performance of the arranged polymer sensing element. The optimized design for thermal distribution is essential to the thermostat, which ensures each sensor element stays at the same operating temperature. Therefore, the simulation objective is to design and optimize a thermostat for thermal distribution that can minimize the uneven heating problem of a gas sensor. For this reason, different thermostat geometries and heating coil materials were studied to search for the optimized thermal distribution and minimized power consumption [23][24][25][26]. The thermal-electric simulations of the thermostat were developed using a finite element tool ANSYS Multiphysics 17.2. Features of the thermostat and meshed model are shown in Figure 1.
Polymers 2017, 9, x FOR PEER REVIEW 3 of 13 is essential to the thermostat, which ensures each sensor element stays at the same operating temperature. Therefore, the simulation objective is to design and optimize a thermostat for thermal distribution that can minimize the uneven heating problem of a gas sensor. For this reason, different thermostat geometries and heating coil materials were studied to search for the optimized thermal distribution and minimized power consumption [23][24][25][26]. The thermal-electric simulations of the thermostat were developed using a finite element tool ANSYS Multiphysics 17.2. Features of the thermostat and meshed model are shown in Figure 1. Furthermore, the optimum geometry of thermostat was implemented to decide the material of heating coil. Four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. Figure 2 illustrates the simulation model and meshed model.

Gas Sensor Fabrication
When the optimum geometry and material of the thermostat were determined, the PEG/MWCNTs gas sensor was fabricated by a flexible printed circuit technology. Figure 3 illustrates the proposed fabrication process flow: (A-1.1) The top layer of the sensor was the interdigitated electrodes (IDEs). Copper with 35 μm thickness was placed on the 50 μm thick polyimide.  Furthermore, the optimum geometry of thermostat was implemented to decide the material of heating coil. Four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 • C and 80 • C, respectively. Figure 2 illustrates the simulation model and meshed model. is essential to the thermostat, which ensures each sensor element stays at the same operating temperature. Therefore, the simulation objective is to design and optimize a thermostat for thermal distribution that can minimize the uneven heating problem of a gas sensor. For this reason, different thermostat geometries and heating coil materials were studied to search for the optimized thermal distribution and minimized power consumption [23][24][25][26]. The thermal-electric simulations of the thermostat were developed using a finite element tool ANSYS Multiphysics 17.2. Features of the thermostat and meshed model are shown in Figure 1. Furthermore, the optimum geometry of thermostat was implemented to decide the material of heating coil. Four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. Figure 2 illustrates the simulation model and meshed model.

Gas Sensor Fabrication
When the optimum geometry and material of the thermostat were determined, the PEG/MWCNTs gas sensor was fabricated by a flexible printed circuit technology.

Gas Sensor Fabrication
When the optimum geometry and material of the thermostat were determined, the PEG/MWCNTs gas sensor was fabricated by a flexible printed circuit technology. Figure 3 illustrates the proposed fabrication process flow:  3) The geometry of the optimize thermostat was designated as 20 mm × 20 mm. The width and spacing of the thermostat line were 220 and 280 µm, respectively. The double-spiral shape wire was patterned on a substrate using UV lithography and wet etching (A-3) The fabricated IDEs were aligned and laminated onto the thermostat via acrylic adhesives. 3) The geometry of the optimize thermostat was designated as 20 mm × 20 mm. The width and spacing of the thermostat line were 220 and 280 μm, respectively. The double-spiral shape wire was patterned on a substrate using UV lithography and wet etching (A-3) The fabricated IDEs were aligned and laminated onto the thermostat via acrylic adhesives. A schematic diagram of the bilayer sensor fabrication process is illustrated in Figure 4. It included two steps regarding the composite sensing film.
(1) In the first step, the MWCNTs was sonicated for 10 min in an ultrasonic bath at room temperature.
A 4 μL solution of the MWCNTs dispersion was then deposited on the IDEs by a microjet and then placed in an oven to furnish the MWCNTs film at 80 °C for 6 hours to form the conductive layer. (2) In the second step, A 8 μL solution of PEG was then deposited on the MWCNTs layer to form the film. The fabricated device was placed in the oven to completely evaporate the solvent at 80 °C for 12 h.  A schematic diagram of the bilayer sensor fabrication process is illustrated in Figure 4. It included two steps regarding the composite sensing film.
(1) In the first step, the MWCNTs was sonicated for 10 min in an ultrasonic bath at room temperature.
A 4 µL solution of the MWCNTs dispersion was then deposited on the IDEs by a microjet and then placed in an oven to furnish the MWCNTs film at 80 • C for 6 hours to form the conductive layer. (2) In the second step, A 8 µL solution of PEG was then deposited on the MWCNTs layer to form the film. The fabricated device was placed in the oven to completely evaporate the solvent at 80 • C for 12 h.
Polymers 2017, 9, x FOR PEER REVIEW 4 of 13 3) The geometry of the optimize thermostat was designated as 20 mm × 20 mm. The width and spacing of the thermostat line were 220 and 280 μm, respectively. The double-spiral shape wire was patterned on a substrate using UV lithography and wet etching (A-3) The fabricated IDEs were aligned and laminated onto the thermostat via acrylic adhesives. A schematic diagram of the bilayer sensor fabrication process is illustrated in Figure 4. It included two steps regarding the composite sensing film.
(1) In the first step, the MWCNTs was sonicated for 10 min in an ultrasonic bath at room temperature.
A 4 μL solution of the MWCNTs dispersion was then deposited on the IDEs by a microjet and then placed in an oven to furnish the MWCNTs film at 80 °C for 6 hours to form the conductive layer. (2) In the second step, A 8 μL solution of PEG was then deposited on the MWCNTs layer to form the film. The fabricated device was placed in the oven to completely evaporate the solvent at 80 °C for 12 h.   Figure 5 describes the testing apparatus for the gas sensor. The gas generator was a dilution flow system that utilized a standard gas generator (KIN-TEK Analytical, Inc., 670C, La Marque, TX, USA) to evaporate an acetone solvent through a mass flow controller in order to obtain the stable concentration and temperature of acetone vapor. The measurement procedure consisted of several steps in each testing cycle. First, the thermostat was heated. Afterwards, high purity nitrogen (N 2 , ≥99.99%, background gas) was added into the glass test chamber for 30 min to obtain a stable reference baseline. Nitrogen was used as the carrier and gas was purged instead of air to exclude the influence of humidity and oxygen contained in air. Afterward, the evaporated acetone gas (target gas) was passed into the glass test chamber for 500 s, and gas molecules were adsorbed into the sensing film. Then, nitrogen was passed for 500 s for desorption from the sensing film. The flow rates of both the target gas and the background gas were controlled to be under 200 mL/min using a mass flow controller.

Experiment
Polymers 2017, 9, x FOR PEER REVIEW 5 of 13 Figure 5 describes the testing apparatus for the gas sensor. The gas generator was a dilution flow system that utilized a standard gas generator (KIN-TEK Analytical, Inc., 670C, La Marque, TX, USA) to evaporate an acetone solvent through a mass flow controller in order to obtain the stable concentration and temperature of acetone vapor. The measurement procedure consisted of several steps in each testing cycle. First, the thermostat was heated. Afterwards, high purity nitrogen (N2, ≥99.99%, background gas) was added into the glass test chamber for 30 min to obtain a stable reference baseline. Nitrogen was used as the carrier and gas was purged instead of air to exclude the influence of humidity and oxygen contained in air. Afterward, the evaporated acetone gas (target gas) was passed into the glass test chamber for 500 s, and gas molecules were adsorbed into the sensing film. Then, nitrogen was passed for 500 s for desorption from the sensing film. The flow rates of both the target gas and the background gas were controlled to be under 200 mL/min using a mass flow controller.

NI Labview
…... The sensor response of the PEG/MWCNTs composite gas sensor to an acetone was investigated under a measurement system. The resistance change of an individual sensor was acquired using a voltage divider method of the fabricated sensor that was installed in a chamber when exposed to acetone. The output voltages of each sensor element were collected via an 8-channel DAQ device (National Instruments Co., USB-6003, Austin, TX, USA). Afterwards, the individual sensor resistance was calculated by Ohm's law. The collected signal of the resistance eliminated the non-demand trend via post-processing by MATLAB. Finally, the normalized resistance changes (ΔR/R%) was displayed in real time on a personal computer.

Thermostat Performance
The thermal distributions of different type heater coils, including meander, double meander, double spiral, and complex double spiral are listed in Table 1. As a result, the optimum design was the double spiral thermostat which provided the best uniform thermal distribution and minimum thermal variation on the surface of the heating zone. Table 1. Selected geometry of the thermostat and related thermal distribution of each thermostat. The sensor response of the PEG/MWCNTs composite gas sensor to an acetone was investigated under a measurement system. The resistance change of an individual sensor was acquired using a voltage divider method of the fabricated sensor that was installed in a chamber when exposed to acetone. The output voltages of each sensor element were collected via an 8-channel DAQ device (National Instruments Co., USB-6003, Austin, TX, USA). Afterwards, the individual sensor resistance was calculated by Ohm's law. The collected signal of the resistance eliminated the non-demand trend via post-processing by MATLAB. Finally, the normalized resistance changes (∆R/R%) was displayed in real time on a personal computer.

Thermostat Performance
The thermal distributions of different type heater coils, including meander, double meander, double spiral, and complex double spiral are listed in Table 1. As a result, the optimum design was the double spiral thermostat which provided the best uniform thermal distribution and minimum thermal variation on the surface of the heating zone. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C.

Double Meander
Polymers 2017, 9,  Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C.

Double Spiral
Polymers 2017, 9,  Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C. Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C.

Complex Double Spiral
Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 • C and 80 • C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 • C and 80 • C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 • C.
Furthermore, the four different heating coil materials, including aluminum, platinum, copper, and stainless steel, were selected to evaluate the power consumption of the thermostat at the operating temperatures of 40 °C and 80 °C, respectively. The results are listed in Table 2, indicating that the use of stainless steel as the coil material offered the minimum power consumption. The power consumptions at the operating temperatures of 40 °C and 80 °C were about 179.48 and 631.78 mW, respectively. Figure 6 shows the optimum thermal distribution result of the thermostat by using aluminum coil at 40 °C.    The fabricated gas sensor equipped with the thermostat is shown in Figure 7. The thermal distribution of the thermostat was examined using a thermal image camera (Thermoteknix Systems Ltd, MIRICLE 307K-25, Cambridge, UK). The infrared thermal image is shown in Figure 8. Temperature uniformity over the heated area size of the heater corresponds approximately to 18 mm × 16 mm. The fabricated gas sensor equipped with the thermostat is shown in Figure 7. The thermal distribution of the thermostat was examined using a thermal image camera (Thermoteknix Systems Ltd, MIRICLE 307K-25, Cambridge, UK). The infrared thermal image is shown in Figure 8. Temperature uniformity over the heated area size of the heater corresponds approximately to 18 mm × 16 mm.

Characterizations
The top-view images of PEG/MWCNTs composite film and cross-sectional morphology of gas sensor were observed by a NOVA NANO SEM 450 (FEI Co., Hillsboro, OR, USA) with 10 kV acceleration voltage. The scanning electron microscope (SEM) morphologies are shown in Figure 9. The thickness of the PEG/MWCNTs composite sensing film fabricated by drop-casting method was about 217.6 nm, and MWCNTs were well wrapped in PEG film.

Characterizations
The top-view images of PEG/MWCNTs composite film and cross-sectional morphology of gas sensor were observed by a NOVA NANO SEM 450 (FEI Co., Hillsboro, OR, USA) with 10 kV acceleration voltage. The scanning electron microscope (SEM) morphologies are shown in Figure 9. The thickness of the PEG/MWCNTs composite sensing film fabricated by drop-casting method was about 217.6 nm, and MWCNTs were well wrapped in PEG film.  The different concentrations of PEG were examined by thermogravimetric analysis (TGA) (Perkin Elmer, Inc., Pyris 1 TGA, Waltham, MA, USA) from room temperature to 500 • C. The TGA results of four different concentrations of PEG are shown in Figure 10, which revealed that the preparation of four different concentrations met the requirement.
(c) (d) Figure 9. SEM morphologies of (a) top-view of PEG/MWCNTs composite film, (b) cross-section micrograph of the gas sensor, (c) close view of the interface between electrode and PEG/MWCNTs composite film, (d) close view of PEG/MWCNTs composite film.
The different concentrations of PEG were examined by thermogravimetric analysis (TGA) (Perkin Elmer, Inc., Pyris 1 TGA, Waltham, MA, USA) from room temperature to 500 °C. The TGA results of four different concentrations of PEG are shown in Figure 10, which revealed that the preparation of four different concentrations met the requirement.

Response of Single Film and Composited Film to Acetone Gas
MWCNTs play the role of a p-type semiconductor. MWCNTs that interact with target gas change the conductivity due to the charge transfer between electron-donating or electronwithdrawing molecules via Van der Waals force or donor-acceptor interaction. Moreover, polymer

Response of Single Film and Composited Film to Acetone Gas
MWCNTs play the role of a p-type semiconductor. MWCNTs that interact with target gas change the conductivity due to the charge transfer between electron-donating or electron-withdrawing molecules via Van der Waals force or donor-acceptor interaction. Moreover, polymer films adsorb gas molecules, and the conductivity is changed by a redox reaction due to the addition or withdrawal of electrons [27][28][29].
The examined response of a single material sensing film can help to distinguish the intrinsic gas sensing property and interaction. Figure 11 shows that each 1.0% PEG sensor and 60 ppm MWCNTs sensor operated at two different temperatures (room temperature and 50 • C) can detect an acetone gas concentration of 6294.3 ppm. The response of the 1.0% PEG sensor was increased when exposed to acetone and cyclically recovered when exposed to nitrogen. In contrast, the MWCNTs sensor was responsive to both gases. Furthermore, when the operating temperature was 50 • C, the response was more intense than that at room temperature. The results implied that PEG could be the candidate sensing film to detect acetone. Moreover, MWCNTs was responsive to each gas, and the response was increased by increasing the operating temperature.
Polymers 2017, 9, x FOR PEER REVIEW 9 of 13 films adsorb gas molecules, and the conductivity is changed by a redox reaction due to the addition or withdrawal of electrons [27][28][29].
The examined response of a single material sensing film can help to distinguish the intrinsic gas sensing property and interaction. Figure 11 shows that each 1.0% PEG sensor and 60 ppm MWCNTs sensor operated at two different temperatures (room temperature and 50 °C) can detect an acetone gas concentration of 6294.3 ppm. The response of the 1.0% PEG sensor was increased when exposed to acetone and cyclically recovered when exposed to nitrogen. In contrast, the MWCNTs sensor was responsive to both gases. Furthermore, when the operating temperature was 50 °C, the response was more intense than that at room temperature. The results implied that PEG could be the candidate sensing film to detect acetone. Moreover, MWCNTs was responsive to each gas, and the response was increased by increasing the operating temperature.  Figure 12 shows the response of four types of PEG sensing films exposed to acetone vapor of 6294.3 ppm at different operating temperatures, including room temperature (R.T., 25 °C), 40 °C, 45 °C, and 50 °C. When the sensor was operated at a higher operating temperature, the response and signal-to-noise ratio (SNR) were better than those at room temperature. The results exhibited that the PEG chains adsorbed more acetone gas molecules at higher operating temperatures and then enhanced the change of conductivity to dominate the response by a redox reaction due to the addition or withdrawal of electrons [30,31].  Figure 12 shows the response of four types of PEG sensing films exposed to acetone vapor of 6294.3 ppm at different operating temperatures, including room temperature (R.T., 25 • C), 40 • C, 45 • C, and 50 • C. When the sensor was operated at a higher operating temperature, the response and signal-to-noise ratio (SNR) were better than those at room temperature. The results exhibited that the PEG chains adsorbed more acetone gas molecules at higher operating temperatures and then enhanced the change of conductivity to dominate the response by a redox reaction due to the addition or withdrawal of electrons [30,31].
(a) (b) Figure 11. Sensor response of 1.0% PEG and MWCNTs at (a) Room temperature and (b) 50 °C. Figure 12 shows the response of four types of PEG sensing films exposed to acetone vapor of 6294.3 ppm at different operating temperatures, including room temperature (R.T., 25 °C), 40 °C, 45 °C, and 50 °C. When the sensor was operated at a higher operating temperature, the response and signal-to-noise ratio (SNR) were better than those at room temperature. The results exhibited that the PEG chains adsorbed more acetone gas molecules at higher operating temperatures and then enhanced the change of conductivity to dominate the response by a redox reaction due to the addition or withdrawal of electrons [30,31].

Sensor Response of Acetone with Thermal Treatment
The operating temperature and the acetone vapor concentration deeply affected the response of the PEG/MWCNTs composite sensing material. Therefore, the acquirement of sensor resistance is a straightforward method to observe the phenomenon, which provides much information including the kinetics of atmospheric oxygen adsorption, electron-donating/withdrawing molecules via Van der Waals force, and donor-acceptor interactions [32,33]. To observe the optimum working temperature, the response of the gas sensor fabricated by PEG/MWCNTs was examined as a function of the operating temperature for exposing to 317.2, 465.9, and 604.6 ppm diluted acetone vapors, as shown in Figure 13. The acetone response was increased as a function of the concentration of PEG. This means that the good interaction of PEG surface with the target gas molecules induced the charge transfer in MWCNTs. The response was increased and then achieved its maximum value at 45 • C for each concentration of acetone, possibly because the moderate working temperature provided the thermal degradation of PEG in its side chains to enhance the Van der Waals interaction of PEG/MWCNTs gas sensor surface with vapor molecules [34]. The sensor response of acetone was found to be reduced at a low temperature, which may be due to the weakly dipolar with weak hydrogen bond properties of physically absorb on conducting polymers [35]. Therefore, a high-performance gas sensor is required to identify the proper working temperature for different target gases. thermal degradation of PEG in its side chains to enhance the Van der Waals interaction of PEG/MWCNTs gas sensor surface with vapor molecules [34]. The sensor response of acetone was found to be reduced at a low temperature, which may be due to the weakly dipolar with weak hydrogen bond properties of physically absorb on conducting polymers [35]. Therefore, a highperformance gas sensor is required to identify the proper working temperature for different target gases. The stability test results of 1.5% PEG/MWCNTs composite sensing film at 45 °C in the presence of 604.6 ppm acetone have been examined, as shown in Figure 15. The initial response of the gas sensor is about 0.76% but the response dropped to 0.41% after 100 cycles. The decreased response of PEG/MWCNTs composite sensing film might be the aging issue which due to the incomplete desorption of acetone molecules, water vapor influence and thermal effect. In a practical application, the PEG/MWCNTs composite gas sensor requires calibrating and monitoring the degradation to improve the reliability. The stability test results of 1.5% PEG/MWCNTs composite sensing film at 45 • C in the presence of 604.6 ppm acetone have been examined, as shown in Figure 15. The initial response of the gas sensor is about 0.76% but the response dropped to 0.41% after 100 cycles. The decreased response of PEG/MWCNTs composite sensing film might be the aging issue which due to the incomplete desorption of acetone molecules, water vapor influence and thermal effect. In a practical application, the PEG/MWCNTs composite gas sensor requires calibrating and monitoring the degradation to improve the reliability.
The stability test results of 1.5% PEG/MWCNTs composite sensing film at 45 °C in the presence of 604.6 ppm acetone have been examined, as shown in Figure 15. The initial response of the gas sensor is about 0.76% but the response dropped to 0.41% after 100 cycles. The decreased response of PEG/MWCNTs composite sensing film might be the aging issue which due to the incomplete desorption of acetone molecules, water vapor influence and thermal effect. In a practical application, the PEG/MWCNTs composite gas sensor requires calibrating and monitoring the degradation to improve the reliability.

Conclusions
In this study, an optimum thermostat was created with stainless steel (SUS304) as the double spiral heating element on a polyimide substrate for the application in a gas sensor. Through numerical modeling, the optimum thermostat was designed and fabricated, aimed at improving the uniformity of thermal distribution on the heating zone of the sensor. The PEG/MWCNTs composite film was fabricated on IDEs by drop-casting method.
A comparative study was conducted to investigate response behaviors to acetone of both the single sensing film and composite sensing film. The composite film operating at a modest temperature was found to have some merits over the single film, such as better selectivity and sensitivity. Moreover, four different concentrations of PEG were prepared and formed as the PEG/MWCNTs composite gas sensor where the 1.5% PEG/MWCNTs composite sensing film that operated at 45 °C was found to be more sensitive than other types. Additionally, the sensor response study was carried out for the operating temperature at four different conditions, 45 °C was found to show a better response than other thermal treatment conditions. Furthermore, this kind of gas sensor equipped with a thermostat to immune ambient temperature influence was promising for environmental and industrial applications.

Conclusions
In this study, an optimum thermostat was created with stainless steel (SUS304) as the double spiral heating element on a polyimide substrate for the application in a gas sensor. Through numerical modeling, the optimum thermostat was designed and fabricated, aimed at improving the uniformity of thermal distribution on the heating zone of the sensor. The PEG/MWCNTs composite film was fabricated on IDEs by drop-casting method.
A comparative study was conducted to investigate response behaviors to acetone of both the single sensing film and composite sensing film. The composite film operating at a modest temperature was found to have some merits over the single film, such as better selectivity and sensitivity. Moreover, four different concentrations of PEG were prepared and formed as the PEG/MWCNTs composite gas sensor where the 1.5% PEG/MWCNTs composite sensing film that operated at 45 • C was found to be more sensitive than other types. Additionally, the sensor response study was carried out for the operating temperature at four different conditions, 45 • C was found to show a better response than other thermal treatment conditions. Furthermore, this kind of gas sensor equipped with a thermostat to immune ambient temperature influence was promising for environmental and industrial applications.