Achievement of High-Response Organic Field-Effect Transistor NO2 Sensor by Using the Synergistic Effect of ZnO/PMMA Hybrid Dielectric and CuPc/Pentacene Heterojunction

High-response organic field-effect transistor (OFET)-based NO2 sensors were fabricated using the synergistic effect the synergistic effect of zinc oxide/poly(methyl methacrylate) (ZnO/PMMA) hybrid dielectric and CuPc/Pentacene heterojunction. Compared with the OFET sensors without synergistic effect, the fabricated OFET sensors showed a remarkable shift of saturation current, field-effect mobility and threshold voltage when exposed to various concentrations of NO2 analyte. Moreover, after being stored in atmosphere for 30 days, the variation of saturation current increased more than 10 folds at 0.5 ppm NO2. By analyzing the electrical characteristics, and the morphologies of organic semiconductor films of the OFET-based sensors, the performance enhancement was ascribed to the synergistic effect of the dielectric and organic semiconductor. The ZnO nanoparticles on PMMA dielectric surface decreased the grain size of pentacene formed on hybrid dielectric, facilitating the diffusion of CuPc molecules into the grain boundary of pentacene and the approach towards the conducting channel of OFET. Hence, NO2 molecules could interact with CuPc and ZnO nanoparticles at the interface of dielectric and organic semiconductor. Our results provided a promising strategy for the design of high performance OFET-based NO2 sensors in future electronic nose and environment monitoring.


Introduction
Since air pollution has become an urgent global problem with the development of industry and technology, detecting gases, especially toxic gases, as the basis for controlling air pollution, has become increasingly significant [1]. One of the most common and detrimental air pollutant oxidizing gases is nitrogen oxides, including nitrogen dioxide (NO 2 ), which is produced and released into atmosphere from combustion and automotive emission. In addition to contributing to the formation of fine particle pollution, NO 2 is linked with a number of adverse effects on the respiratory system such as chronic bronchitis, emphysema, and respiratory irritation at low concentrations [2][3][4]. The potential detrimental impact of NO 2 emission on public health and the environment has led to extensive scientific and

ZnO/PMMA Hybrid Preparation
ZnO nanoparticles were prepared according to the method reported in the previous literature [26], and the average size of as-synthesized quasispherical ZnO NPs was 4.9 nm. ZnO nanoparticles were divided from methanol by centrifugation and dispersed in chloroform/methanol (50 mL, v/v = 90: 10) to obtain a stock solution. PMMA (average molecules weight~120,000) was dissolved in anisole with a concentration of 200 mg/ml. The obtained solution was mixed with the prepared ZnO nanoparticles dispersion (v/v = 1:1). The ZnO/PMMA hybrid dielectric was fabricated using spin coating process. Figure 1 shows the molecular structures of pentacene, CuPc and PMMA and schematic structure of the top-contacted OFET-based sensors with only CuPc/pentacene heterojunction (device A) and both ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction (device B). The OFETs were processed according to the following procedure. Indium tin oxide (ITO) coated glass was used as substrate and gate electrodes. Prior to the spin-coating of the dielectric layers, the substrates were successively ultrasonically cleaned in acetone, deionized water and isopropyl alcohol. ZnO/PMMA hybrid and PMMA, functioned as the gate dielectric, were spin-coated on ITO substrate at room temperature (25 • C) and baked in an oven at 90 • C for 2 h. Subsequently, 30 nm pentacene and 5 nm thick CuPc were thermally evaporated in a vacuum of~2 × 10 −4 Pa at a rate of 0.2 Å/s successively. Finally, the source and drain electrodes of 50 nm gold (Au) were thermally deposited using a shadow mask at a rate of 10 Å/s. The length and width of the channel were 100 µm and 1 cm, respectively.

ZnO/PMMA Hybrid Preparation
ZnO nanoparticles were prepared according to the method reported in the previous literature [26], and the average size of as-synthesized quasispherical ZnO NPs was 4.9 nm. ZnO nanoparticles were divided from methanol by centrifugation and dispersed in chloroform/methanol (50 mL, v/v = 90:10) to obtain a stock solution. PMMA (average molecules weight ~ 120,000) was dissolved in anisole with a concentration of 200 mg/ml. The obtained solution was mixed with the prepared ZnO nanoparticles dispersion (v/v = 1:1). The ZnO/PMMA hybrid dielectric was fabricated using spin coating process. Figure 1 shows the molecular structures of pentacene, CuPc and PMMA and schematic structure of the top-contacted OFET-based sensors with only CuPc/pentacene heterojunction (device A) and both ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction (device B). The OFETs were processed according to the following procedure. Indium tin oxide (ITO) coated glass was used as substrate and gate electrodes. Prior to the spin-coating of the dielectric layers, the substrates were successively ultrasonically cleaned in acetone, deionized water and isopropyl alcohol. ZnO/PMMA hybrid and PMMA, functioned as the gate dielectric, were spin-coated on ITO substrate at room temperature (25 °C) and baked in an oven at 90 °C for 2 h. Subsequently, 30 nm pentacene and 5 nm thick CuPc were thermally evaporated in a vacuum of ~2 × 10 −4 Pa at a rate of 0.2 Å/s successively. Finally, the source and drain electrodes of 50 nm gold (Au) were thermally deposited using a shadow mask at a rate of 10 Å/s. The length and width of the channel were 100 µm and 1 cm, respectively.

Device Test and Data Analyses
The electrical characteristics of all the devices were measured with a Keithley 4200 sourcemeter (Tektronix, Shanghai, China) under ambient conditions at room temperature.
The morphologies of the organic semiconductor were characterized with atomic force microscopy (AFM) (Agilent, AFM 5500) in a tapping mode. The OFET sensor was stored in an airtight test chamber (approximately 0.02 L). Dry air and 100 ppm standard NO2 gases (anhydrous) were purchased from Sichuan Tianyi Science and Technology Co., Chengdu, China, and a mixture with the appropriate concentration was introduced into the test chamber by a mass flow controller (S48 300/HMT, Beijing Boriba Metron Instruments Co., Beijing, China). NO2 gas response characteristics of the OFET sensors were measured with a variation of drain-source current, which

Device Test and Data Analyses
The electrical characteristics of all the devices were measured with a Keithley 4200 sourcemeter (Tektronix, Shanghai, China) under ambient conditions at room temperature.
The morphologies of the organic semiconductor were characterized with atomic force microscopy (AFM) (Agilent, AFM 5500) in a tapping mode. The OFET sensor was stored in an airtight test chamber (approximately 0.02 L). Dry air and 100 ppm standard NO 2 gases (anhydrous) were purchased from Sichuan Tianyi Science and Technology Co., Chengdu, China, and a mixture with the appropriate concentration was introduced into the test chamber by a mass flow controller (S48 300/HMT, Beijing Boriba Metron Instruments Co., Beijing, China). NO 2 gas response characteristics of the OFET sensors were measured with a variation of drain-source current, which acted as a function of time. Also, the transfer curves in various concentrations of NO 2 were systematically characterized. The field-effect mobility of device was extracted in the saturation regime from the highest slope of |I DS | 1/2 vs. V G plots by using Equation (1): where L and W are the channel length and width, respectively. C i is the capacitance (per unit area) of the dielectric, V G is the gate voltage, and I DS is the drain-source current. Figure 2 depicts the representative transfer plots of devices A and B. Both devices A and B have the typical behavior of a p-type transistor. Device A exhibits a field-effect mobility (µ), a current on/off ratio (I on /I off ), a threshold-voltage (V T ), and a sub-threshold slope(SS) of 0.13 cm 2 ·V −1 ·s −1 , 2.0 × 10 3 , −12 V, and 3.0 V/dec, respectively. In contrast, the values of µ, I on /I off , V T , and SS for device B are 0.006 cm 2 ·V −1 ·s −1 , 8.9 × 10 1 , −15 V and 15 V/dec, respectively. It is obvious that the device performance of CuPc/pentacene heterojunction OFETs based on pure PMMA is much higher than that of based on ZnO/PMMA hybrid dielectric. Compared with our previous results [22], ZnO nanoparticles can lead to more serious performance decrease on CuPc/pentacene heterojunction-based device than the pentacene-based device. This phenomenon indicates that CuPc may also play an essential role as well as ZnO nanoparticles, i.e., they have a synergistic effect. acted as a function of time. Also, the transfer curves in various concentrations of NO2 were systematically characterized. The field-effect mobility of device was extracted in the saturation regime from the highest slope of |IDS| 1/2 vs. VG plots by using Equation (1):

Results and Discussion
where L and W are the channel length and width, respectively. Ci is the capacitance (per unit area) of the dielectric, VG is the gate voltage, and IDS is the drain-source current. Figure 2 depicts the representative transfer plots of devices A and B. Both devices A and B have the typical behavior of a p-type transistor. Device A exhibits a field-effect mobility (µ), a current on/off ratio (Ion/Ioff), a threshold-voltage (VT), and a sub-threshold slope(SS) of 0.13 cm 2 ·V −1 ·s −1 , 2.0 × 10 3 , −12 V, and 3.0 V/dec, respectively. In contrast, the values of µ, Ion/Ioff, VT, and SS for device B are 0.006 cm 2 ·V −1 ·s −1 , 8.9 × 10 1 , −15 V and 15 V/dec, respectively. It is obvious that the device performance of CuPc/pentacene heterojunction OFETs based on pure PMMA is much higher than that of based on ZnO/PMMA hybrid dielectric. Compared with our previous results [22], ZnO nanoparticles can lead to more serious performance decrease on CuPc/pentacene heterojunction-based device than the pentacene-based device. This phenomenon indicates that CuPc may also play an essential role as well as ZnO nanoparticles, i.e., they have a synergistic effect. Then, the OFETs were exposed to NO2 atmosphere with various concentrations that ranged from 0 to 15 ppm. All the devices were exposed to a specific concentration of NO2 for 5 min before measuring. The gate voltage VG was from 20 to −40 V and the drain voltage VD was −40 V. As shown in Figure 3, the curves of device A exhibit a slight shift, but that of device B shifts more significantly.

Results and Discussion
To intuitively illuminate sensing property, several parameters were calculated from the transfer curve to evaluate the performance of OFET sensors, such as Ion, µ, VT, and SS. The variation of multiple parameters defined as ΔR = (RNO2 − RAIR)/RAIR × 100% is presented in Figure 4. As shown in Figure 4a,b, the variations of Ion and µ of device A and device B show an opposite trend along with the increasing concentration of NO2, the Ion and µ of device B increase by 193% and 69% at 15 ppm Then, the OFETs were exposed to NO 2 atmosphere with various concentrations that ranged from 0 to 15 ppm. All the devices were exposed to a specific concentration of NO 2 for 5 min before measuring. The gate voltage V G was from 20 to −40 V and the drain voltage V D was −40 V. As shown in Figure 3, the curves of device A exhibit a slight shift, but that of device B shifts more significantly.
To intuitively illuminate sensing property, several parameters were calculated from the transfer curve to evaluate the performance of OFET sensors, such as I on , µ, V T , and SS. The variation of multiple parameters defined as ∆R = (R NO2 − R AIR )/R AIR × 100% is presented in Figure 4. As shown in Figure 4a,b, the variations of I on and µ of device A and device B show an opposite trend along with the increasing concentration of NO 2 , the I on and µ of device B increase by 193% and 69% at 15 ppm NO 2 , while that of device A decreases by 30% and 28%. Since the I on and µ of pentacene-based OFETs will increase significantly when exposed to NO 2 atmosphere [17], CuPc might have a different impact on devices A and B. The V T of device B presents a remarkable decrease about 80%, while that of device A shows nearly no change. Because V T is usually referred to charge trapping at the dielectric/semiconductor interface, the more hole charges trap at the interface, the stronger negative gate voltage is needed to turn the transistor on, and vice versa. Thus, the dielectric/semiconductor interface of device A has less trap sites than device B after being exposed to NO 2 , and the NO 2 interact places do not locate at this interface. NO2, while that of device A decreases by 30% and 28%. Since the Ion and µ of pentacene-based OFETs will increase significantly when exposed to NO2 atmosphere [17], CuPc might have a different impact on devices A and B. The VT of device B presents a remarkable decrease about 80%, while that of device A shows nearly no change. Because VT is usually referred to charge trapping at the dielectric/semiconductor interface, the more hole charges trap at the interface, the stronger negative gate voltage is needed to turn the transistor on, and vice versa. Thus, the dielectric/semiconductor interface of device A has less trap sites than device B after being exposed to NO2, and the NO2 interact places do not locate at this interface.   NO2, while that of device A decreases by 30% and 28%. Since the Ion and µ of pentacene-based OFETs will increase significantly when exposed to NO2 atmosphere [17], CuPc might have a different impact on devices A and B. The VT of device B presents a remarkable decrease about 80%, while that of device A shows nearly no change. Because VT is usually referred to charge trapping at the dielectric/semiconductor interface, the more hole charges trap at the interface, the stronger negative gate voltage is needed to turn the transistor on, and vice versa. Thus, the dielectric/semiconductor interface of device A has less trap sites than device B after being exposed to NO2, and the NO2 interact places do not locate at this interface.   Furthermore, SS is proportional to the trap density at the interface of dielectric and organic semiconductor, and the trap density (N) can be extracted by Equation (2): where q is the electronic charge, k is Boltzmann's constant, T is absolute temperature, and C is the areal capacitance of the dielectric structure. So the SS is proportional to the N. As shown in Figure 4d, the SS increases constantly to 60% in device B at 15 ppm NO 2 concentration. Nevertheless, the SS of device A is almost unchanged under the concentration of 0.5-15 ppm NO 2 . So the interaction in device A between NO 2 and dielectric/organic semiconductor interface is different from that of device B.
To study the reason for the deviation in sensing performance, AFM was utilized to observe the morphologies of active films (AFM images of dielectrics are shown in Figure S1). As shown in Figure 5a,b, the grain size of pentacene in device B is much smaller than that in device A, indicating that ZnO nanoparticles embedded in PMMA dielectric act as impurities and influence the morphology of pentacene film by three aspects: decreasing the grain size, deepening the grain boundary, and disordering molecular arrangement [27]. From Figure 5c, it can be clearly observed that the ultrathin film of CuPc in device B forms a relatively homogeneous film. Furthermore, SS is proportional to the trap density at the interface of dielectric and organic semiconductor, and the trap density (N) can be extracted by Equation (2): 2 10 where q is the electronic charge, k is Boltzmann's constant, T is absolute temperature, and C is the areal capacitance of the dielectric structure. So the SS is proportional to the N. As shown in Figure  4d, the SS increases constantly to 60% in device B at 15 ppm NO2 concentration. Nevertheless, the SS of device A is almost unchanged under the concentration of 0.5-15 ppm NO2. So the interaction in device A between NO2 and dielectric/organic semiconductor interface is different from that of device B.
To study the reason for the deviation in sensing performance, AFM was utilized to observe the morphologies of active films (AFM images of dielectrics are shown in Figure S1). As shown in Figure  5a,b, the grain size of pentacene in device B is much smaller than that in device A, indicating that ZnO nanoparticles embedded in PMMA dielectric act as impurities and influence the morphology of pentacene film by three aspects: decreasing the grain size, deepening the grain boundary, and disordering molecular arrangement [27]. From Figure 5c, it can be clearly observed that the ultrathin film of CuPc in device B forms a relatively homogeneous film. From the above discussion, it can be deduced that the enhancement of the sensing properties is attributed to the synergistic effect of ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction: (1) ZnO nanoparticles embedded in PMMA will dramatically decrease the grain size and enlarge the depth of grain boundary, which not only increases the potential barrier but also makes the subsequent CuPc molecules partially diffuse into the interface of pentacene and hybrid dielectric. Moreover, the nitrogen atoms around the Cu atom have a relatively high electron density, which can act as hole-charge traps. Thus, the Ion and µ of device B are much lower than those of device A; (2) When device B is exposed to NO2, as shown in Figure 6, NO2 can diffuse directly into the interface of organic semiconductor and hybrid dielectric, then interact with ZnO nanoparticles and CuPc. As is well known, NO2 is a strong oxidizing gas with very high electron affinity, so it will interact with the surface of ZnO nanoparticles through surface-adsorbed oxygen ions [23]. Similarly, CuPc molecules can interact with NO2 and form charge complex due to the delocalized π-electrons which are readily ionized [13]. NO2 can weaken the impact of ZnO nanoparticles and CuPc at or near the dielectric/organic semiconductor interface on charge transport effectively, so Ion, µ and VT are significantly improved. As a result, the high relative responses can be achieved. The main difference between devices A and B is that CuPc affects the charge conducting channel directly as a functional receptor under the action of ZnO nanoparticles. Device A just forms a simple vertical heterojunction; the sensing mechanism is dependent on the energy level disordering after being exposed to NO2. Because the NO2-induced domain fracture originates at the CuPc/Au interface, it is proposed that the NO2-induced domain fraction also degrades the CuPc/Au electrical contacts, the increased From the above discussion, it can be deduced that the enhancement of the sensing properties is attributed to the synergistic effect of ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction: (1) ZnO nanoparticles embedded in PMMA will dramatically decrease the grain size and enlarge the depth of grain boundary, which not only increases the potential barrier but also makes the subsequent CuPc molecules partially diffuse into the interface of pentacene and hybrid dielectric. Moreover, the nitrogen atoms around the Cu atom have a relatively high electron density, which can act as hole-charge traps. Thus, the I on and µ of device B are much lower than those of device A; (2) When device B is exposed to NO 2 , as shown in Figure 6, NO 2 can diffuse directly into the interface of organic semiconductor and hybrid dielectric, then interact with ZnO nanoparticles and CuPc. As is well known, NO 2 is a strong oxidizing gas with very high electron affinity, so it will interact with the surface of ZnO nanoparticles through surface-adsorbed oxygen ions [23]. Similarly, CuPc molecules can interact with NO 2 and form charge complex due to the delocalized π-electrons which are readily ionized [13]. NO 2 can weaken the impact of ZnO nanoparticles and CuPc at or near the dielectric/organic semiconductor interface on charge transport effectively, so I on , µ and V T are significantly improved. As a result, the high relative responses can be achieved. The main difference between devices A and B is that CuPc affects the charge conducting channel directly as a functional receptor under the action of ZnO nanoparticles. Device A just forms a simple vertical heterojunction; the sensing mechanism is dependent on the energy level disordering after being exposed to NO 2 . Because the NO 2 -induced domain fracture originates at the CuPc/Au interface, it is proposed that the NO 2 -induced domain fraction also degrades the CuPc/Au electrical contacts, the increased density of domain boundaries would therefore act to trap carriers near the contacts and induce positive uncompensated charge, which is consistent with the decrease of I on [13].
Sensors 2016, 16,1763 7 of 10 density of domain boundaries would therefore act to trap carriers near the contacts and induce positive uncompensated charge, which is consistent with the decrease of Ion [13]. The real-time response curve of device B for NO2 detection was also studied (Figure 7). The pulse of each NO2 concentration was 10 min. During the recovery process, NO2 gas was removed, and the sensor was exposed to dry air for 10 min. It is obvious that, when exposed to different NO2 concentrations, device B has a fast response. In addition to the response of fresh OFET sensor, the environmental stability under ambient atmosphere is of critical importance to the practical application of OFET-based sensors and the overall lifetime of the device [28,29]. Thus, the sensor that was tested was stored in ambient air with a relative humidity of ~50% for 30 days, the testing process was similar to the aforementioned fresh device characterization. The variation of its output current at VG = −40 V when exposed to different NO2 concentrations (ranging from 0.5 to 15 ppm) is shown in Figure 8a. As shown in Figure 8b, the relative change of Ion at 0.5 ppm NO2 is more than 10 times higher than that of the fresh device (more than twice at 15 ppm). Moreover, after being exposed to 15 ppm NO2, the device still can detect 0.5 ppm NO2 effectively, and exhibit a high stability to constantly detection. As lots of H2O and O2 is absorbed on the surface of ZnO nanoparticles, a large enhancement of sensitivity may be attributed to the concerted efforts of H2O, O2 and NO2 [30][31][32]. The real-time response curve of device B for NO 2 detection was also studied (Figure 7). The pulse of each NO 2 concentration was 10 min. During the recovery process, NO 2 gas was removed, and the sensor was exposed to dry air for 10 min. It is obvious that, when exposed to different NO 2 concentrations, device B has a fast response.  The real-time response curve of device B for NO2 detection was also studied (Figure 7). The pulse of each NO2 concentration was 10 min. During the recovery process, NO2 gas was removed, and the sensor was exposed to dry air for 10 min. It is obvious that, when exposed to different NO2 concentrations, device B has a fast response. In addition to the response of fresh OFET sensor, the environmental stability under ambient atmosphere is of critical importance to the practical application of OFET-based sensors and the overall lifetime of the device [28,29]. Thus, the sensor that was tested was stored in ambient air with a relative humidity of ~50% for 30 days, the testing process was similar to the aforementioned fresh device characterization. The variation of its output current at VG = −40 V when exposed to different NO2 concentrations (ranging from 0.5 to 15 ppm) is shown in Figure 8a. As shown in Figure 8b, the relative change of Ion at 0.5 ppm NO2 is more than 10 times higher than that of the fresh device (more than twice at 15 ppm). Moreover, after being exposed to 15 ppm NO2, the device still can detect 0.5 ppm NO2 effectively, and exhibit a high stability to constantly detection. As lots of H2O and O2 is absorbed on the surface of ZnO nanoparticles, a large enhancement of sensitivity may be attributed to the concerted efforts of H2O, O2 and NO2 [30][31][32]. In addition to the response of fresh OFET sensor, the environmental stability under ambient atmosphere is of critical importance to the practical application of OFET-based sensors and the overall lifetime of the device [28,29]. Thus, the sensor that was tested was stored in ambient air with a relative humidity of~50% for 30 days, the testing process was similar to the aforementioned fresh device characterization. The variation of its output current at V G = −40 V when exposed to different NO 2 concentrations (ranging from 0.5 to 15 ppm) is shown in Figure 8a. As shown in Figure 8b, the relative change of I on at 0.5 ppm NO 2 is more than 10 times higher than that of the fresh device (more than twice at 15 ppm). Moreover, after being exposed to 15 ppm NO 2 , the device still can detect 0.5 ppm NO 2 effectively, and exhibit a high stability to constantly detection. As lots of H 2 O and O 2 is absorbed on the surface of ZnO nanoparticles, a large enhancement of sensitivity may be attributed to the concerted efforts of H 2 O, O 2 and NO 2 [30][31][32]. Selectivity is a crucial parameter and an open issue for practical sensing applications, which usually relies on the specific interaction or energy modulation between the organic semiconductors and the analytes [33,34]. Another common kind of air pollutant oxidizing gas of sulfur oxides, sulfur dioxide (SO2) was also investigated by using device B. As shown in Figure 9, the VT increases and the Ion decreases along with the increase of SO2 concentration. Surprisingly, this result is opposite to the conventional phenomenon of oxidizing gases [35]. This result may be due to the interaction between ZnO and SO2 at room temperature, yielding SO4 2− , which can act as the hole trap sites at the interface of dielectric and organic semiconductor [36,37].

Conclusions
In conclusion, the ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction were used to achieve a high response OFET-based NO2 gas sensor, exhibiting a dramatic variation of Ion, μ, and VT after being exposed to NO2. The Ion and VT increased by 193% and 77% at 15 ppm NO2, and by 5% and 8% at 0.5 ppm NO2. Moreover, after storing at atmosphere for 30 days, the relative change of Ion at 0.5 ppm NO2 was more than 10 times higher (about 50%) than that of a fresh device. The high performance of this OFET-based sensor was attributed to the synergistic effect of ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction. In addition, the sensor with synergistic effect could clearly distinguish the oxidizing gas of NO2 from SO2 with opposite Ion variation. Using the synergistic effect of dielectric and organic semiconductor is demonstrated to be an effective method for device engineering in OFET-based gas sensors. Selectivity is a crucial parameter and an open issue for practical sensing applications, which usually relies on the specific interaction or energy modulation between the organic semiconductors and the analytes [33,34]. Another common kind of air pollutant oxidizing gas of sulfur oxides, sulfur dioxide (SO 2 ) was also investigated by using device B. As shown in Figure 9, the V T increases and the I on decreases along with the increase of SO 2 concentration. Surprisingly, this result is opposite to the conventional phenomenon of oxidizing gases [35]. This result may be due to the interaction between ZnO and SO 2 at room temperature, yielding SO 4 2− , which can act as the hole trap sites at the interface of dielectric and organic semiconductor [36,37]. Selectivity is a crucial parameter and an open issue for practical sensing applications, which usually relies on the specific interaction or energy modulation between the organic semiconductors and the analytes [33,34]. Another common kind of air pollutant oxidizing gas of sulfur oxides, sulfur dioxide (SO2) was also investigated by using device B. As shown in Figure 9, the VT increases and the Ion decreases along with the increase of SO2 concentration. Surprisingly, this result is opposite to the conventional phenomenon of oxidizing gases [35]. This result may be due to the interaction between ZnO and SO2 at room temperature, yielding SO4 2− , which can act as the hole trap sites at the interface of dielectric and organic semiconductor [36,37].

Conclusions
In conclusion, the ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction were used to achieve a high response OFET-based NO2 gas sensor, exhibiting a dramatic variation of Ion, μ, and VT after being exposed to NO2. The Ion and VT increased by 193% and 77% at 15 ppm NO2, and by 5% and 8% at 0.5 ppm NO2. Moreover, after storing at atmosphere for 30 days, the relative change of Ion at 0.5 ppm NO2 was more than 10 times higher (about 50%) than that of a fresh device. The high performance of this OFET-based sensor was attributed to the synergistic effect of ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction. In addition, the sensor with synergistic effect could clearly distinguish the oxidizing gas of NO2 from SO2 with opposite Ion variation. Using the synergistic effect of dielectric and organic semiconductor is demonstrated to be an effective method for device engineering in OFET-based gas sensors.

Conclusions
In conclusion, the ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction were used to achieve a high response OFET-based NO 2 gas sensor, exhibiting a dramatic variation of I on , µ, and V T after being exposed to NO 2 . The I on and V T increased by 193% and 77% at 15 ppm NO 2 , and by 5% and 8% at 0.5 ppm NO 2 . Moreover, after storing at atmosphere for 30 days, the relative change of I on at 0.5 ppm NO 2 was more than 10 times higher (about 50%) than that of a fresh device. The high performance of this OFET-based sensor was attributed to the synergistic effect of ZnO/PMMA hybrid dielectric and CuPc/pentacene heterojunction. In addition, the sensor with synergistic effect could clearly distinguish the oxidizing gas of NO 2 from SO 2 with opposite I on variation. Using the synergistic effect of dielectric and organic semiconductor is demonstrated to be an effective method for device engineering in OFET-based gas sensors.