Enhanced Gas Detection by Altering Gate Voltage Polarity of Polypyrrole/Graphene Field-Effect Transistor Sensor

Abstract

This work introduces a new measurement methodology for enhancing gas detection by tuning the magnitude and polarity of back-gate voltage of a field-effect transistor (FET)-based sensor. The aim is to simultaneously strengthen the sensor response and accelerate the sensor recovery. In addition, this methodology can consume less energy compared with conventional measurements by direct current bias. To illustrate the benefits of the proposed methodology, we fabricated and characterized a polypyrrole/graphene (PPy/G) FET sensor for ammonia (NH₃) detection. Our experiment, simulation and calculation results demonstrated that the redox reaction between the NH₃ molecules and the PPy/G sensitive layer could be controlled by altering the polarity and the magnitude of the back-gate voltage. This proof-of-principle measurement methodology, which solves the inherent contradiction between high response and slow recovery of the chemiresistive sensor, could be extended to detect other gases, so as to improve global gas measurement systems. It opens up a new route for FET-based gas sensors in practical applications.

1. Introduction

The chemiresistive sensor (or chemiresistor) with two electrodes is the most common gas sensor. Its working principle is to detect gas concentration by measuring the current or resistance change caused by charge transfer between the target gas molecules and the sensitive layers. Such a measurement is implemented by continuously monitoring the current between two electrodes under a given direct current (DC) voltage. Another configuration for gas detection is based on the field-effect transistor (FET), which is a three-terminal device. The sensitive layer, which acts as the FET channel, lies between the source and the drain terminals and is insulated from the back gate by a dielectric layer. The drain current, flowing along the sensitive layer, can be modulated by the back-gate voltage. The drain current can also be changed by the electrons donated/withdrawn from the target gas molecules. During the sensing measurement of an FET-based sensor, the source is usually grounded, the drain is continuously biased with a given DC voltage, and the back gate is swept by DC voltages. The detection of the gas concentration depends on the drain current change caused by the charge transfer. Compared with the chemiresistor, the FET sensor is regarded as highly sensitive with good selectivity [1,2], due to its ability to deliver multi-parameter responses, such as the following: drain current, charge mobility, threshold voltage, flat-band voltage, work function, and on/off current ratio [3,4].

The working principles of both types of sensors are based on the current change measured via a DC voltage [5]. The biggest issue of the DC measurement is the slow response/recovery [6,7], and the current baseline drift [8], designated as “poisoning effect”. The charge transfer is a slow chemical procedure, generally of the order of tens to thousands of seconds. This chemical procedure is more efficient for adsorbed molecules with large binding energy [9]. More particularly, the charge transfer between adsorbed gas molecules and interface traps/defect sites are extremely slow [10,11]. These slow dynamic processes limit the sensor response and recovery speeds. Moreover, the stronger the interaction between adsorbed molecules and the sensitive layer, the higher the sensitivity to the gas, but the slower the recovery [12]. To overcome these problems, various technologies and methods have been proposed, such as heating [13,14], micro-hotplate [15], ultraviolet radiation [16], current stimulation [17], voltage bias [18], degassing [19], or chemo-selective coating [20]. However, all these solutions result in complex fabrication processes, increased energy consumption, high production costs, and bulky device sizes.

Kulkarni’s group [21] developed a heterodyne sensor, in which an alternating current (AC) was applied to the source of a graphene FET sensor to modulate the channel charge carriers and, simultaneously, to excite the perturbation of target molecular dipoles. They used the heterodyne mixing current as the sensing signal to detect the gas concentration. The signal parameters can be phase, amplitude, or frequency of the mixing current. These parameters contain rich information on the target molecules. The heterodyne sensor obtained a detection limit of 1 ppb and a response time of 0.1 s. However, heterodyne sensing can only be used for the detection of polar gases. Recently, Liu’s group [22] gated a graphene FET sensor by an AC voltage so as to limit the slow interaction between the adsorbed molecules and surface traps/defect states. Their technique depends on the phase lag between the channel current and the AC gate voltage to determine gas concentrations. For detecting water, methanol and ethanol vapors, they obtained a minimal baseline drift and a recovery speed that was ten times faster than that in the conventional DC measurement. Although slow interaction was excluded by this technique, it might be difficult for the sensor sensitivity to reach an ultra-high level. Indeed, the phase signal measurement only responds to weak adsorption and desorption of gas molecules above a certain distance from the graphene surface.

The objective of the present work is to introduce a new measurement methodology for gas detection, using FET-based sensors. By alternating the polarity and the magnitude of the back-gate voltage, the present method can not only enhance sensor response, but can also speed up sensor recovery.

2. Materials and Methods

2.1. Measurement Principle

The measurement principle of the present methodology is attributed to the redox reaction between a target gas and a sensitive layer, controlled by the electric-field effect. In order to clearly explain the measurement principle of the novel methodology, we focus our attention on a polypyrrole/graphene (PPy/G) FET sensor for ammonia (NH₃) detection. Hereafter, we limit our explanation and analysis to an accumulation-mode p-channel transistor. The current flow in this transistor is due to hole transport in the p-channel. The target gas could be any electron-donating gas (i.e., a reducing gas), such as NH₃, H₂, CH₄, H₂S, CO, etc. It could also be any electron-accepting gas (i.e., an oxidizing gas), like NO₂, H₂O, O₂, CO₂, SO₂, etc. Note that the description of the accumulation-mode n-channel transistor can readily be derived from the description of the accumulation-mode p-channel transistor by appropriately changing sign of the electric field. In general, both graphene and polypyrrole have p-type doping. As explained below, NH₃ molecules mainly adsorb on the PPy surface, while the role of graphene is to improve the charge collection towards the source and drain of the sensor. Graphene also plays a critical role in the deposition of the PPy layer (i.e., graphene acts as a support substrate and electrode during the electropolymerization process). It is important to note that the floating graphene has no screening effect on the electric field created by the back-gate voltage.

Figure 1 shows the schematic cross-sections of the PPy/G FET sensor, and Figure 2 schematically plots the drain current (I_DS) response of the PPy/G FET sensor for NH₃ detection under different polarities and values of the back-gate voltages. When the back gate of the PPy/G FET sensor is grounded (V_BG = 0 V), the behavior of the sensor is similar to a chemiresistor (Figure 1a). When the electron-donating gas molecules (such as NH₃) adsorb onto the PPy channel surface, electrons are transferred from NH₃ to the π backbone of the PPy. They neutralize holes in the p-type PPy, thereby, decreasing the current in the PPy channel (Figure 1b). For desorption in the NH₃-off state, the electrons go back to the NH₃ from the PPy channel and, then, the neutralized PPy reverts back to p-type, restoring the PPy channel current to its original value. However, this is a slow chemical procedure, which may not even result in complete restoration at room temperature (Figure 1c). It can be seen from Figure 2a that the recovery time of the sensor was long and the response curve of the sensor current looks like a shark-fin with a baseline drift.

In Figure 1d, the back gate of the PPy/G FET sensor is biased by a positive voltage (V_BG > 0 V). Holes are, therefore, pushed to the PPy channel surface, which facilitates PPy accepting more electrons from NH₃. As a result, the current in the PPy channel significantly decrease. When the sensor is exposed to NH₃, the redox reaction between NH₃ and PPy is described by the following:

NH₃ (g) + PPy ↔ PPy-NH₃

(1)

PPy-NH₃ + h⁺ ↔ PPy-NH₃⁺

(2)

In reaction (1), the electron-donating NH₃ molecules first react with PPy to form PPy-NH₃ complexes. With holes (h⁺), the PPy-NH₃ complexes then become positively charged ions (PPy-NH₃⁺) in reaction (2). The surface hole concentration can be modulated by the value of the back-gate voltage. Therefore, the number of the PPy-NH₃⁺ ions are increased by positive back-gate voltage. As a result, the increased PPy-NH₃⁺ ions are prone to adsorb many more electrons from the NH₃ molecules onto the PPy surface, thereby strengthening PPy chemical reduction and strongly decreasing sensor current (Figure 1e). In other words, when the PPy channel is exposed to electron-donating NH₃ molecules under positive back-gate voltage, many more NH₃ molecules and NH₃-donating electrons (reactions (1) and (2) in the forward direction) can be attracted, strongly increasing the sensor response. This is equivalent to an amplification effect, being created by the back-gate voltage. However, the desorption of NH₃ molecules is quite slow under positive gate voltage (Figure 1f). As shown in Figure 2b, the current response of the sensor changes into a larger shark-fin curve, with a larger baseline drift, and the sensor recovery time slows down. Although the sensor response is higher at V_BG > 0 V than at V_BG = 0 V, sensor recovery becomes problematic at V_BG > 0 V. The larger the back-gate voltage, the longer the full recovery time. The reason is that some NH₃ molecules and NH₃-donating electrons cannot easily leave the PPy channel surface since they have been attracted by (or recombined with) a number of holes under V_BG > 0 V.

To solve this problem, a negative voltage is applied to the back gate (V_BG < 0 V) and this withdraws holes from the PPy surface, as shown in Figure 1g. The PPy-NH₃⁺ ions decompose back into NH₃ and PPy (reactions (1) and (2) in the reverse direction), accelerates sensor current recovery. This means that the negative back-gate voltage helps electrons to move back to the NH₃ molecules. It is important to note that when applying a negative back-gate voltage, the following different situations may occur: (i) The channel current may turn back to its initial value (I_DS0). This happens when the number of holes withdrawn by the negative back-gate voltage equals the number of NH₃-donating electrons. (ii) The channel current may be larger/smaller than I_DS0. This happens when the withdrawn holes are less/more in number as compared to the number of donated electrons. Full recovery of the channel current to I_DS0 is achieved within the first (ideal) case. This case happens by tuning the negative back-gate voltage to a value at which the drain current turns back to its initial value (I_DS0) in the NH₃-off state.

According to the phenomena described above, we developed a new measurement methodology, in which different polarity voltages (such as a square wave voltage) were provided to the back gate of the FET sensor during the sensing measurement (Figure 2c). Hereafter, the NH₃-on state meant that NH₃ was injected into the testing chamber, while the NH₃-off state meant that NH₃ was switched off in the testing chamber. In the NH₃-on state, the positive half cycle of the square wave voltage (+V_BG) pushed holes to the PPy channel surface. This promoted PPy chemical reduction, and significantly increased the sensor current response. In the NH₃-off state, the negative half cycle of the square wave voltage (−V_BG) withdrew holes from the PPy channel surface. This boosted PPy chemical oxidation and, thus, accelerated the sensor recovery. The square wave voltage changed the response profile of the sensor current from a shark-fin curve (Figure 2a,b) to a nearly rectangular curve (Figure 2c). In other words, the redox reaction between NH₃ and PPy could be controlled by the square wave voltage. In brief, it is possible to increase the current response of the sensor and reduce the recovery time of the sensor by using the positive and negative half cycles of the square wave voltage, respectively.

DC measurements often require a tradeoff between having a thicker sensitive layer, allowing for higher sensitivity, or a thinner sensitive layer, allowing for quicker recovery. The need for a tradeoff can be easily solved in an FET sensor gated by different polarity voltages. This is the great advantage of our methodology over the common methods used in DC measurements conducted by FET sensors.

2.2. Measurement Setup

Figure 3 schematically displays the experimental setup proposed in this work for the sensing measurements. This setup is composed of a PPy/G FET gas sensor, a square wave generator, a current detector and a signal processor. The square wave generator is electrically connected to the back gate of the FET gas sensor to maximize the sensor response, and to accelerate the sensor recovery, for target gases. The current detector, connected between the source and drain of the FET sensor, monitors the modulated currents containing the gas concentration information during the sensing procedure. The signal processor is configured to communicate the information between the square wave generator and the current detector. The signal processor can control the starting time, the amplitude, the polarity and the period of the square wave voltage, according to the feedback signal from the current detector. Specifically, the current detector informs the signal processor when the drain current experiences a change. For example, when a p-type FET sensor is exposed to electron-donating gas and its drain current decreases, the generator applies positive voltage to the sensor back gate, which boosts the polymer reduction reaction and significantly enhances the sensor response (amplification effect). When the electron-donating gas is switched off and the drain current increases, the generator offers negative voltage to the sensor back gate, which promotes the polymer oxidation reaction and accelerates the sensor recovery. In addition, the square wave voltage can be used to evaluate the repeatability of the sensor.

2.3. Sensor Fabrication and Measurement

The fabrication details of the PPy/G FET gas sensor can be found in our previous works [23,24]. Briefly, the fabrication process includes the following four simple steps: (i) The Au interdigitated electrodes (IDEs), or a pair of electrode pads, were prepared on a SiO₂/Si substrate by the lift-off process. The SiO₂ thickness was 90 or 300 nm to allow easy observation of graphene with an optical microscope; (ii) The Al back-gate electrode (300 nm in thickness) was formed on the back of the SiO₂/Si substrate by electron-beam physical vapor deposition; (iii) Small-sized graphene oxide was deposited on top of the IDEs by the ultrasonic spray pyrolysis technique, while large-sized CVD graphene was transferred to the top of the electrode pads using poly(methyl methacrylate) as the supporting film; (iv) A thin layer of PPy was synthesized on the graphene surface by electropolymerization. This step was carried out at 20 °C in an electropolymerization bath, which included a solution of pyrrole (0.35 mL) and NaClO₄ (0.6 g) in acetonitrile (50 mL). The graphene on the substrate was used as the working electrode. The counter and reference electrodes were made of Pt and Ag/AgCl, respectively [25]. When a voltage pulse was applied to the working electrode, the PPy layer was synthesized on the graphene surface. The PPy synthesis ratio was 10 nm/cycle under a voltage pulse of 0.9 V. The samples were then washed with ethanol. Here, the PPy/G sensitive layer acted as the channel of the FET sensor and was exposed to NH₃, while the IDEs, or electrode pads, acted as the source/drain of the FET sensor. Our sensor structure is simpler, compared with the real FET device, since it omits the source/drain doping step.

Sensor measurements were carried out in a testing chamber with an internal volume of 42 cm³ (3 cm × 3.5 cm × 4 cm) at atmospheric pressure, with a temperature of 20 °C. Wet air (relative humidity of 50% RH) and NH₃ were used as carrier gas and target gas, respectively. The gas flow in the testing chamber was the piston mode. The desired NH₃ concentration was obtained by dilution of 100 ppm NH₃ with wet air. For 1 ppm NH₃ concentration, we mixed 10 mL/min of 100 ppm NH₃ in air with 490 mL/min of dry air and 500 mL/min of saturated wet air (100% RH). The total gas flow rate in the testing chamber was always kept at 1000 mL/min.

3. Results and Discussion

3.1. Experiment Results

The fabricated PPy/G FET sensors were used to experimentally analyze the benefits of the measurement methodology proposed in this work. In order to clearly observe the influence of the electric-field effect on the response and recovery of the sensors, we examined the I_DS variation of the sensors for detecting 1 ppm NH₃ under different values and polarities of V_BG. The I_DS measurements were performed at V_DS = −0.8 V and the source grounded. We used rectangular pulses for the ammonia injection. The examinations were carried out at room temperature and relative humidity of 50%.

Figure 4a shows the drain current/back-gate voltage characteristics for a fabricated PPy/G FET sensor in the absence of NH₃. It confirms that this sensor was a typical accumulation-mode p-channel transistor. In the NH₃-off state, the drain current increased with positive back-gate voltage, since holes were pushed to the PPy surface, which enhanced the carrier density of the p-channel. In the NH₃-on state, the drain current decreased with positive back-gate voltage since the pushed holes on the PPy surface were neutralized by the NH₃-donated electrons, reducing the carrier density of the p-channel. Figure 4b experimentally records the I_DS response of the PPy/G FET sensor under different back-gate voltage (V_BG). In Figure 4b, the vertical blue rectangles indicate the period of 1 ppm NH₃ injection in the testing chamber (the NH₃-on state) and the red line indicates the values and polarities of the back-gate voltages. In region AD, the back gate of the sensor was grounded (V_BG = 0 V). In region BC, where the sensor was exposed to NH₃, no significant current response was observed. In region DE, the back gate was biased at V_BG = +10 V and I_DS decreased with time until it reached a plateau. The sensor showed a current response of 16%. We defined the current response as (I_DS0 − I_DS)/I_DS0, where I_DS0 is the initial current at V_BG = 0 V in the NH₃-off state and I_DS is the minimum current at V_BG = 10 V in the NH₃-on state. This indicated that positive back-gate voltage really strengthened the sensor response. However, I_DS did not recover to its original value (I_DS0) in region EF for the NH₃-off state. This might have been due to the fact that the NH₃-donating electrons were trapped in the defect states of the sensitive layer, and, thus, the withdrawing of these electrons was very slow (difficult). Then, a larger V_BG = +30 V was applied to the back gate in region FJ. For the NH₃-on state in region FH, a larger current response of 21% was obtained. This confirmed that the larger the positive back-gate voltage, the larger the response of the sensor current. A larger positive voltage produced a stronger electric field to force more holes to the PPy surface. This provided more opportunity for the reaction between NH₃ and PPy to be completed. For the NH₃-off state in region HI, I_DS could not completely return to I_DS0, since the current curve tended to a plateau in this region. Moreover, the sensor recovery was slower at V_BG = +30 V (recovery started at point H) than at V_BG = +10 V (recovery started at point E). At point I, a negative voltage (V_BG = −30 V) was provided to the back gate and the sensor recovery became much faster. Namely, the recovery time was less than 10 s from point I to J. This proved the fact that the negative back-gate voltage accelerated the sensor recovery. The reason for this in the physics context can be found in Section 2.1 (measurement principle). Briefly, the negative back-gate voltage withdrew holes from the PPy surface, which caused the donated electrons to quickly return back to the NH₃ molecules. However, if the absolute value of the negative voltage was too large, the quantity of holes was higher than the quantity of donated electrons and, then, I_DS had a value higher than I_DS0, as shown at point K. We performed this examination with several PPy/G FET sensors and obtained similar current response trends. The advantages of our measurement methodology were well reflected in the results of the experiments. They are expressed by the simulation results and the calculations in analytical modeling below.

The present measurements were performed manually. In the future, we intend to automatize the measurements by using a signal processor, as described in Section 2.2. In the real measurements, the NH₃ injection might not be a perfect pulse and, thus, the drain current (I_DS) change might fluctuate. However, I_DS always tends to decrease in the NH₃-on state, while I_DS always tends to increase in the NH₃-off state. Therefore, we may set an a priori threshold current (I_th) for an automatic measurement. When I_DS, measured by the current detector, reaches I_th (indicating the NH₃-on state), the signal processor automatically switches on the square wave generator and applies a positive half cycle of the square wave voltage (V_BG > 0 V) to the back gate at point D. The signal processor tracks, in real-time, changes in the I_DS during exposure to NH₃. When I_DS is larger than I_th (indicating the NH₃–off state), the signal processor automatically switches the positive half cycle to the negative half cycle of the square wave (V_BG < 0 V) at point E, causing I_DS to quickly increase to I_DS0. In this case, the positive half cycle of the square wave pushes holes to the PPy surface and increases the sensor response, while the negative half cycle of the square wave quickly withdraws holes from the PPy surface and accelerates the sensor recovery. Under a square wave voltage, the sensor shows a high response and a rapid recovery for NH₃ detection at room temperature.

As discussed above, the redox reaction of PPy with NH₃ plays an important role. However, graphene not only serves as a support material for electropolymerization, but also accelerates the electron transport between the source and drain, shortening the response/recovery times. Under the electric-field induction, the electrons donated by NH₃ may pass through the PPy to react with the sp2-bonded carbon atoms of graphene. This synergistic effect may intensify the sensor sensitivity. Moreover, the low contact resistance of graphene with Au electrodes would result in a high signal/noise ratio and low energy consumption.

3.2. Simulation Results

Using the ATLAS software from SILVACO, we simulated the I_DS response of the PPy/G FET sensor in detecting electron-donating gases under different V_BG. The 2D simulation is based on an accumulation-mode p-channel transistor, the channel of which is made of thin polymer film. In this kind of transistor, holes are responsible for the current flow, which is similar to the behavior of our p-type PPy/graphene sensor. The important physical parameters used in the present simulation are listed in

Table 1. Important physical parameters used in the present simulation.

Parameter Name	Property	Value Range
Front-gate voltage (VFG)	Variable	0 to 0.1 V
Back-gate voltage (VBG)	Variable	−3 to 3 V
Front-gate oxide (SiO2)	Constant	20 nm
Back-gate oxide (SiO2)	Constant	300 nm
Channel thickness	Constant	100 nm
Channel length	Constant	2 µm
Front-gate length	Constant	2 µm
Back-gate length	Constant	3 µm
Polymer doping concentration	Constant	4 × 1016 cm−3
Drain voltage (VDS)	Constant	−0.8 V

. A 100-nm thick p-type polymer was used to mimic the PPy/G sensitive layer. The channel length of the transistor was 2 µm, which was the same as the inter-finger distance in the PPy/G gas sensor. The source/drain without doping was directly in contact with Au electrodes. The length of each Au electrode was 0.5 µm. The transistor had a back-gate electrode and the thickness of the back-gate oxide was 300 nm. The length of the back gate was 3 µm, and it overlapped the source, channel and drain.

As already mentioned, the electron-donating gas was equivalent to a positive voltage applied on the front-gate of the accumulation-mode p-channel transistor. Therefore, we used a front-gate electrode to induce the donated electrons by electrostatic doping [26]. The I_DS change caused by electron transfer from NH₃ to the sensitive layer could be computed and demonstrated with the ATLAS software. The simulation was performed as follows. We first extracted I_DS0 in the NH₃-off state when V_BG and V_FG equaled zero. Then, I_DS for V_FG = +0.1 V and V_FG = 0 V were set as the current corresponding to the NH₃-on state and the NH₃-off state, respectively. Finally, with V_FG fixed at +0.1 V, we simulated the I_DS variations by changing V_BG. The simulation was carried out at V_DS = −0.8 V.

Figure 5 demonstrates the simulation results, which show the I_DS response of the accumulation-mode polymer p-type channel transistor versus time at different V_BG for the electron-donating gas on/off states. The simulation results are presented by following the experimental route as shown in Figure 4. We first kept V_FG = 0 V (in the gas-off state) and V_BG = 0 V to simulate the initial value of the drain current (I_DS0) in region AB. Region BC at V_BG = 0 V shows the I_DS variations by switching V_FG from 0 to +0.1 V and +0.1 to 0 V to emulate the electron-donating gas-on state and gas-off state, respectively. The transistor exhibited an I_DS decrease of 43% at V_BG = 0 V when the gas was on. In region CD, the transistor was reset to I_DS0 by switching V_FG = 0 V and V_BG = 0 V. Then, we applied V_BG = +1 V to the transistor in the gas-on state (V_FG = +0.1 V) to simulate the I_DS variations. In this case, a change of 92% in I_DS was obtained. Further, we increased V_BG from +1 to +3 V in the gas-on state (in region FH). A larger change (>100%) in I_DS appeared. The I_DS response of the transistor was twice as high at V_BG = 3 V as it was at V_BG = 1 V. It was clear that the I_DS response of the transistor increased with a positive V_BG. However, the I_DS recovery of the transistor was slow and even incomplete under positive V_BG (in region HI). This issue was solved by applying a negative voltage to the back gate. When V_BG = −3 V, I_DS increased sharply and even exceeded I_DS0 (in region JK). The I_DS profile in Figure 5 is qualitatively consistent with the experimental result shown in Figure 4. This confirmed the sensing mechanism proposed in the previous section.

3.3. Calculation Results Based on a Sensor Modeling

Figure 6 shows the schematic cross-section of an accumulation-mode p-channel thin film transistor, which was used to model the behavior of the PPy/G FET sensor. In our modeling, the electron transfer behavior between the target gas and the sensitive layer was equivalent to the application of a front-gate voltage (V_FG). In other words, the donated electrons in the p-channel were induced by “electrostatic doping” [27]. When a negative drain voltage (V_DS) was applied to the FET sensor, the drain current flowing through the PPy thin film (I_DS) was the sum of two contributions: a surface accumulation channel (I_acc) and a body current (I_body):

I_DS = I_body + I_acc

(3)

In the thin film configuration, the drain current could be modified by V_FG (the front-gate voltage) and V_BG (the back-gate voltage). NH₃ is an electron-donating gas. Therefore, it was equivalent to applying a positive V_FG to the front gate of the thin film transistor. In this case, the drain current could be modulated not only by the gas concentrations (altering V_FG), but also by V_BG.

We adapted Colinge’s modeling of the thin-film accumulation-mode silicon-on-insulator p-channel MOSFET [28] to a polymer channel, and then calculated I_DS as a function of V_FG for different values of V_BG. The parameters used in our calculation were as follows: polymer channel length, width and thickness were 2, 1, and 0.1 µm, respectively; thicknesses of the front-gate oxide and back-gate oxide were 1 and 90 nm, respectively; charge densities of the top and bottom interfaces were 5 × 10¹⁰ and 10¹¹ cm⁻², respectively; doping concentration of the PPy channel was set to 4.5 × 10¹⁸ atoms/cm³; hole mobility in the PPy body was 0.1 cm²/Vs; PPy band gap was 2.32 eV [29]; PPy permittivity was 500 [30]; drain voltage V_DS was fixed at −0.8 V.

Figure 7 presents the calculation results based on the analytical modeling in ref. [28]. It is clear that for a given V_FG (i.e., for a given NH₃ concentration), I_DS increased and decreased by applying negative and positive V_BG, respectively. Moreover, the larger the absolute value of V_BG, the larger the I_DS change. These results theoretically supported the sensing mechanism mentioned above. This model was also suitable for electron-accepting gases, simply by changing the sign of the front-gate voltage. Our calculation results were also consistent with the steady-state linear behavior of accumulation-mode long-channel p-MOSFET [31].

4. Conclusions

We developed a new measurement methodology for enhancing gas detection by tuning the magnitude and polarity of back-gate voltage of a field-effect transistor (FET)-based sensor. A polypyrrole (PPy)/graphene FET sensor was used to examine the proposed measurement methodology for detecting an electron-donating gas (NH₃). Our experimental results, supported by analytic modeling and ATLAS 2D simulation, demonstrated that the redox reaction of the PPy sensitive layer with NH₃ molecules was controlled by the back-gate voltage polarity and value of the FET sensor. Positive back-gate voltage pushed holes from the p-type PPy to the surface, which boosted the PPy chemical reduction, significantly increasing the sensor response for NH₃ detection. On the other hand, negative back-gate voltage withdrew holes from the p-type PPy surface and promoted its chemical oxidation, accelerating the sensor recovery in the NH₃-off state. This proof-of-principle measurement methodology has the ability to measure low gas concentrations with a large response and fast recovery at room temperature. It can be of great interest for FET sensor networks in the field of environmental monitoring and disease diagnosis. It can also be of great interest for the Internet of Things, and the automotive and food/agriculture industrial sectors.