Surface Acoustic Wave DMMP Gas Sensor with a Porous Graphene/PVDF Molecularly Imprinted Sensing Membrane

In this paper, surface acoustic wave (SAW) sensors containing porous graphene/PVDF (polyvinylidene fluoride) molecularly imprinted sensitive membrane for DMMP gas detection were investigated. A 433 MHz ST-cut quartz SAW resonator was used to convert gas concentration changes into frequency shifts by the sensors. The porous graphene/PVDF film was fabricated on the sensor’s surface by using the tape-casting method. DMMP molecules were adsorbed on the porous structure sensing film prepared by the 2-step method to achieve the specific recognition effect. The sensitivity of the sensor could reach −1.407 kHz·ppm−1. The response time and recovery time of the SAW sensor with porous graphene/PVDF sensing membrane were about 4.5 s and 5.8 s at the concentration of 10 ppm, respectively. The sensor has good anti-interference ability to most gases in the air.


Introduction
Chemical warfare agents (CWAs) [1] are fast-acting lethal compounds even at low concentration levels. It is necessary to develop CWAs detection technology with high sensitivity, good selectivity, strong anti-interference ability, fast response and compatibility with the current Internet of things technology. Nerve gases [2] are the most dangerous agents of chemical warfare and mass destruction, can cause irreversible damage to the nervous system within seconds and are fatal if exposure occurs for even a few minutes. Therefore, it is very dangerous for researchers to directly study such gases' sensing characteristics in the laboratory. Almost all studies examine simulants, which have no toxicity or less toxicity, but similar functional groups, structures and properties as nerve gases. DMMP [3,4] is a well-known nerve agent simulator, specifically of Sarin.
As a new and efficient recognition technology, molecularly imprinted technology [5,6] integrates materials science, polymer science, chemical engineering, biochemistry and other disciplines and can recognize specific analytes. Due to the characteristics of predetermination, recognition and practicability, it has been widely used in many fields, such as chromatographic separation [7,8], solid-phase extraction [9,10], biomimetic sensing [11], enzyme catalysis [12], clinical drug analysis [13] and so on. In addition, sensing films prepared by molecularly imprinted technology can also be integrated with electrochemical sensors and microelectronic sensors for detecting trace analytes.
Recently, surface acoustic wave (SAW) sensors [14][15][16] have been extensively used in liquid [17,18] and gas [19][20][21] detection because of their low cost, high precision, miniatur-ization, and compatibility with semiconductor technology. The SAW sensor consists of a sensing film and a conversion element (SAW resonator or delay line). A sensing film is the core of the SAW sensor. After it adsorbs the analyte to be measured, the SAW converter converts the adsorptions into frequency or phase shifts of the sensor to obtain the device's response. Due to the high sensitivity and selectivity, the polymer [22][23][24], semiconductor [25,26] and graphene [27,28] sensing films have mostly been used in SAW sensors.
In this paper, we investigated a surface acoustic wave (SAW) sensor with porous PVDF layer sensing film based on molecularly imprinted recognition technology for simulant DMMP detection. A 433 MHz SAW one-port resonator based on ST-quartz substrate and aluminum interdigital transducers (Al-IDTs) was selected to convert the change of gas concentration into frequency shift. Figure 1 is a schematic diagram of the fabrication process for the one-port SAW resonator as a based device. The substrate was ST-quartz with a SAW velocity of 3158 m·s −1 . The aluminum IDTs were fabricated using a standard UV-photolithography process. Each resonator consisted of an IDT with 160 pairs of electrodes and two reflectors on both sides of IDTs with 270 electrodes. The pitch and acoustic apertures were 1.8 µm and 172 µm, respectively. Moreover, the resonant frequency of the SAW devices was approximately 434.9 MHz. The sensing films were coated on both reflectors' IDTs on the SAW resonator to reduce the propagation loss. Recently, surface acoustic wave (SAW) sensors [14][15][16] have been extensively use in liquid [17,18] and gas [19][20][21] detection because of their low cost, high precision, mi iaturization, and compatibility with semiconductor technology. The SAW sensor consis of a sensing film and a conversion element (SAW resonator or delay line). A sensing fil is the core of the SAW sensor. After it adsorbs the analyte to be measured, the SAW co verter converts the adsorptions into frequency or phase shifts of the sensor to obtain t device's response. Due to the high sensitivity and selectivity, the polymer [22][23][24], sem conductor [25,26] and graphene [27,28] sensing films have mostly been used in SAW se sors.

Preparation of the SAW Sensor with a Porous Graphene/PVDF Sensing Layer
In this paper, we investigated a surface acoustic wave (SAW) sensor with poro PVDF layer sensing film based on molecularly imprinted recognition technology for sim ulant DMMP detection. A 433 MHz SAW one-port resonator based on ST-quartz substra and aluminum interdigital transducers (Al-IDTs) was selected to convert the change gas concentration into frequency shift. Figure 1 is a schematic diagram of the fabrication process for the one-port SAW re onator as a based device. The substrate was ST-quartz with a SAW velocity of 3158 m·s The aluminum IDTs were fabricated using a standard UV-photolithography process. Ea resonator consisted of an IDT with 160 pairs of electrodes and two reflectors on both sid of IDTs with 270 electrodes. The pitch and acoustic apertures were 1.8 μm and 172 μm respectively. Moreover, the resonant frequency of the SAW devices was approximate 434.9 MHz. The sensing films were coated on both reflectors' IDTs on the SAW resonat to reduce the propagation loss. The porous graphene/PVDF sensing films were fabricated as the sensing membran First, 0.2 mmol of citric acid monohydrate (CAM), 0.02 mL of dimethyl methyl phosph nate (DMMP), 8 mg of graphene and 0.6 g of polyvinylidene fluoride (PVDF) powder w dispersed in 10 mL of N, N-dimethylformamide (DMF) with the assistance of ultrason dispersion to form CAM/DMMP/PVDF/graphene/DMF polymeric precursor solutio Then, 1 mL polymeric precursor solution was coated on the SAW resonator to obtain t porous film using a tape-casting method [29], followed by a drying treatment at 85 °C f 30 min in an oven.

Preparation of the SAW Sensor with a Porous Graphene/PVDF Sensing Layer
The preparation process of the sensing film on the SAW resonator is shown in Figu 2a. The graphene and recrystallized CAM micro/nanoparticles were uniformly disperse in the cured graphene/PVDF/DMMP membrane (Figure 2(a1)). Next, the SAW device w immersed in the concentrated NaHCO3 (0.01 mol/L) aqueous solution for 1.5 h at 25 ° Due to polymer swelling, the CAM nanoparticles near the surface of the sensing film w react with NaHCO3 at the interface between PVDF and graphene. Then, micro holes we formed with the emission of CO2. At the same time, DMMP on the inner wall of the micr hole and the surface of the film was discharged with CO2 ( Figure 2(a2)). Thus, the poro The porous graphene/PVDF sensing films were fabricated as the sensing membrane. First, 0.2 mmol of citric acid monohydrate (CAM), 0.02 mL of dimethyl methyl phosphonate (DMMP), 8 mg of graphene and 0.6 g of polyvinylidene fluoride (PVDF) powder was dispersed in 10 mL of N, N-dimethylformamide (DMF) with the assistance of ultrasonic dispersion to form CAM/DMMP/PVDF/graphene/DMF polymeric precursor solution. Then, 1 mL polymeric precursor solution was coated on the SAW resonator to obtain the porous film using a tape-casting method [29], followed by a drying treatment at 85 • C for 30 min in an oven.
The preparation process of the sensing film on the SAW resonator is shown in Figure 2a. The graphene and recrystallized CAM micro/nanoparticles were uniformly dispersed in the cured graphene/PVDF/DMMP membrane (Figure 2(a1)). Next, the SAW device was immersed in the concentrated NaHCO 3 (0.01 mol/L) aqueous solution for 1.5 h at 25 • C. Due to polymer swelling, the CAM nanoparticles near the surface of the sensing film will react with NaHCO 3 at the interface between PVDF and graphene. Then, micro holes were formed with the emission of CO 2 . At the same time, DMMP on the inner wall of the micro-hole and the surface of the film was discharged with CO 2 (Figure 2(a2)). Thus, the porous graphene/PVDF sensing films with DMMP molecular imprinting was formed. Finally, the devices were rinsed with ethanol and water to give the SAW sensors ( Figure 2(a3)). graphene/PVDF sensing films with DMMP molecular imprinting was formed. Finally, the devices were rinsed with ethanol and water to give the SAW sensors ( Figure 2(a3)).

Characterization and Measurement
The morphologies of devices were characterized using scanning electron microscopy (FE-SEM; Merlin Compact, Zeiss, Oberkochen, Germany) and atomic force microscopy (AFM; Agilent 5500, Agilent Technologies, Santa Clara, CA, USA). The crystalline phases of materials were analyzed by X-ray diffraction (XRD; D/max-2500/PC, Rigaku, Tokyo, Japan). The responses of the SAW resonators and sensors were measured by a network analyzer (ENA-E5071C, Keysight, Santa Rosa, CA, USA).
For sensing vapors, a DMMP gas delivery and SAW measurement system was developed as Figure 3, which was included (1) standard N2 source, (2) massflow controller (MFC) for the mixture of N2/ DMMP, (3) MFC for N2, (4) heating device, (5) DMMP source, (6) prepared SAW sensor in a chamber, and (7) network analyzer. The nitrogen from the source was divided into two paths, one carrying the DMMP and the other only carrier to the sensor surface. Both were precisely controlled by MFC. The gas flow rate of the N2/DMMP path was maintained at 250 mL·min −1 , and the flow rate of the pure nitrogen path can be adjusted. The flask containing DMMP is heated to 30 °C or 50 °C. Using the molar volume at different temperatures and flow differences between two gas channels, the concentration of the DMMP scan be achieved [30,31]. The SAW sensor was placed in a sealed chamber with a volume of about 500 mL. A pump was connected to the chamber's outlet to realize the weak negative pressure in the test system, which could ensure the dynamic circulation of gas and the tightness of the system. Before testing, the SAW sensor, fixed on the test fixture, was placed in an N2 environment and preheated to 80 °C for 1 h to remove any gas absorbed on the sensing film.

Characterization and Measurement
The morphologies of devices were characterized using scanning electron microscopy (FE-SEM; Merlin Compact, Zeiss, Oberkochen, Germany) and atomic force microscopy (AFM; Agilent 5500, Agilent Technologies, Santa Clara, CA, USA). The crystalline phases of materials were analyzed by X-ray diffraction (XRD; D/max-2500/PC, Rigaku, Tokyo, Japan). The responses of the SAW resonators and sensors were measured by a network analyzer (ENA-E5071C, Keysight, Santa Rosa, CA, USA).
For sensing vapors, a DMMP gas delivery and SAW measurement system was developed as Figure 3, which was included (1) standard N2 source, (2) massflow controller (MFC) for the mixture of N2/ DMMP, (3) MFC for N2, (4) heating device, (5) DMMP source, (6) prepared SAW sensor in a chamber, and (7) network analyzer. The nitrogen from the source was divided into two paths, one carrying the DMMP and the other only carrier to the sensor surface. Both were precisely controlled by MFC. The gas flow rate of the N2/DMMP path was maintained at 250 mL·min −1 , and the flow rate of the pure nitrogen path can be adjusted. The flask containing DMMP is heated to 30 • C or 50 • C. Using the molar volume at different temperatures and flow differences between two gas channels, the concentration of the DMMP scan be achieved [30,31]. The SAW sensor was placed in a sealed chamber with a volume of about 500 mL. A pump was connected to the chamber's outlet to realize the weak negative pressure in the test system, which could ensure the dynamic circulation of gas and the tightness of the system. Before testing, the SAW sensor, fixed on the test fixture, was placed in an N 2 environment and preheated to 80 • C for 1 h to remove any gas absorbed on the sensing film.   Figure 4a is the image of a one-port SAW resonator fabricated by a standard integrated circuit manufacturing method. Figure 4b,c shows the OM and SEM images of IDTs, which confirm that the IDTs finger width and gap are equal to 1.6 um, respectively. Figure  4d is the cross-sectional SEM images of IDTs, showing that the thickness of aluminum electrodes is about 108.5 nm.  Figure 5a is the SEM images of graphene/PVDF/CAM/DMMP membrane, showing that the graphene and recrystallized CAM micro/nanoparticles were uniformly dispersed in the film. After immersion in the NaHCO3 solution, the recrystallized CAM precipitated in the form of carbon dioxide and formed a porous structure on the surface of the sensing film ( Figure 5b). The thickness of the sensing film was about 2.47 μm (Figure 5c). Figure 6 shows AFM images of the Al IDTs, PVDF/DMMP film and porous graphene/PVDF sensing film. The roughness (RMS) of the IDT electrodes on one-port SAW resonator, PVDF/DMMP /SAW, and porous graphene/PVDF /SAW was 1.14 nm, 10.9 nm and 119 nm, respectively. With the formation of the porous structure sensing film, the roughness of the sensor surface increased significantly, which led to the S11 of the device decreasing from −8.21 dB to −2.68 dB (Figure 7a). Moreover, after the sensing film was fabricated on  Figure 4a is the image of a one-port SAW resonator fabricated by a standard integrated circuit manufacturing method. Figure 4b,c shows the OM and SEM images of IDTs, which confirm that the IDTs finger width and gap are equal to 1.6 um, respectively. Figure 4d is the cross-sectional SEM images of IDTs, showing that the thickness of aluminum electrodes is about 108.5 nm.  Figure 4a is the image of a one-port SAW resonator fabricated by a standard inte grated circuit manufacturing method. Figure 4b,c shows the OM and SEM images of IDT which confirm that the IDTs finger width and gap are equal to 1.6 um, respectively. Figur 4d is the cross-sectional SEM images of IDTs, showing that the thickness of aluminum electrodes is about 108.5 nm.  Figure 5a is the SEM images of graphene/PVDF/CAM/DMMP membrane, showin that the graphene and recrystallized CAM micro/nanoparticles were uniformly disperse in the film. After immersion in the NaHCO3 solution, the recrystallized CAM precipitate in the form of carbon dioxide and formed a porous structure on the surface of the sensin film (Figure 5b). The thickness of the sensing film was about 2.47 μm (Figure 5c). Figure shows AFM images of the Al IDTs, PVDF/DMMP film and porous graphene/PVDF sen ing film. The roughness (RMS) of the IDT electrodes on one-port SAW resonato PVDF/DMMP /SAW, and porous graphene/PVDF /SAW was 1.14 nm, 10.9 nm and 11 nm, respectively. With the formation of the porous structure sensing film, the roughnes of the sensor surface increased significantly, which led to the S11 of the device decreasin from −8.21 dB to −2.68 dB (Figure 7a). Moreover, after the sensing film was fabricated o  Figure 5a is the SEM images of graphene/PVDF/CAM/DMMP membrane, showing that the graphene and recrystallized CAM micro/nanoparticles were uniformly dispersed in the film. After immersion in the NaHCO 3 solution, the recrystallized CAM precipitated in the form of carbon dioxide and formed a porous structure on the surface of the sensing film (Figure 5b). The thickness of the sensing film was about 2.47 µm (Figure 5c). Figure 6 shows AFM images of the Al IDTs, PVDF/DMMP film and porous graphene/PVDF sensing film. The roughness (RMS) of the IDT electrodes on one-port SAW resonator, PVDF/DMMP /SAW, and porous graphene/PVDF /SAW was 1.14 nm, 10.9 nm and 119 nm, respectively. With the formation of the porous structure sensing film, the roughness of the sensor surface increased significantly, which led to the S 11 of the device decreasing from −8.21 dB to −2.68 dB (Figure 7a). Moreover, after the sensing film was fabricated on the reflective grating, the resonant frequency of the sensor showed a negative frequency shift of about 294 kHz due to the mass-loading effect [32].

Characterization and Measurement Results
Micromachines 2021, 12, x FOR PEER REVIEW 5 of 10 the reflective grating, the resonant frequency of the sensor showed a negative frequency shift of about 294 kHz due to the mass-loading effect [32].   The temperature coefficient of frequency (TCF), which is an important parameter to characterize the temperature stability of the SAW sensor, was observed in the range of 25-100 °C. As shown in Figure 7b, the preparation of the sensing film did not lead to significant changes in the temperature stability of the device. The TCF of the bare SAW resonator and SAW sensor with porous graphene/PVDF sensing membrane was −1.4305 ppm/°C and −1.5624 ppm/°C. Figure 8 shows the static responses of the SAW gas sensor with different DMMP concentrations in the range from 0 to 10 ppm. The sensor was tested more than 10 times at each concentration value. The raw data, statistical analysis and residual plots are shown in Table S1 and Figure S2. The response of the SAW sensor decreased rapidly with increasing DMMP concentration in the range from 0 to 10 ppm. The sensitivity of the sensor, which can be obtained from the slope of the linear fitting (solid black line in Figure 8), was    The temperature coefficient of frequency (TCF), which is an important parameter to characterize the temperature stability of the SAW sensor, was observed in the range of 25-100 °C. As shown in Figure 7b, the preparation of the sensing film did not lead to significant changes in the temperature stability of the device. The TCF of the bare SAW resonator and SAW sensor with porous graphene/PVDF sensing membrane was −1.4305 ppm/°C and −1.5624 ppm/°C. Figure 8 shows the static responses of the SAW gas sensor with different DMMP concentrations in the range from 0 to 10 ppm. The sensor was tested more than 10 times at each concentration value. The raw data, statistical analysis and residual plots are shown in Table S1 and Figure S2. The response of the SAW sensor decreased rapidly with increasing DMMP concentration in the range from 0 to 10 ppm. The sensitivity of the sensor, which can be obtained from the slope of the linear fitting (solid black line in Figure 8), was    The temperature coefficient of frequency (TCF), which is an important parameter to characterize the temperature stability of the SAW sensor, was observed in the range of 25-100 °C. As shown in Figure 7b, the preparation of the sensing film did not lead to significant changes in the temperature stability of the device. The TCF of the bare SAW resonator and SAW sensor with porous graphene/PVDF sensing membrane was −1.4305 ppm/°C and −1.5624 ppm/°C. Figure 8 shows the static responses of the SAW gas sensor with different DMMP concentrations in the range from 0 to 10 ppm. The sensor was tested more than 10 times at each concentration value. The raw data, statistical analysis and residual plots are shown in Table S1 and Figure S2. The response of the SAW sensor decreased rapidly with increasing DMMP concentration in the range from 0 to 10 ppm. The sensitivity of the sensor, which can be obtained from the slope of the linear fitting (solid black line in Figure 8), was The temperature coefficient of frequency (TCF), which is an important parameter to characterize the temperature stability of the SAW sensor, was observed in the range of 25-100 • C. As shown in Figure 7b, the preparation of the sensing film did not lead to significant changes in the temperature stability of the device. The TCF of the bare SAW resonator and SAW sensor with porous graphene/PVDF sensing membrane was −1.4305 ppm/ • C and −1.5624 ppm/ • C. Figure 8 shows the static responses of the SAW gas sensor with different DMMP concentrations in the range from 0 to 10 ppm. The sensor was tested more than 10 times at each concentration value. The raw data, statistical analysis and residual plots are shown in Table S1 and Figure S2. The response of the SAW sensor decreased rapidly with increasing DMMP concentration in the range from 0 to 10 ppm. The sensitivity of the sensor, which can be obtained from the slope of the linear fitting (solid black line in Figure 8), was about −1.407 kHz/ppm (R = 0.99346). The linear-regression equation, parameters, statistics and R-value are shown in Figure S3.

Static Responses of DMMP Sensors
Micromachines 2021, 12, x FOR PEER REVIEW 6 about −1.407 kHz/ppm (R = 0.99346). The linear-regression equation, parameters, stat and R-value are shown in Figure S3.

Dynamic Responses and Sensing Mechanism
The frequency shift of the SAW sensors based on the electric loading effect ca correlated with the sheet conductivity and the surface capacitance per length of the ing film using the following relationship [33,34]: where 0 v is the effective SAW velocity of the device, 2 K is the electromechanical pling coefficient of the piezoelectric substrate, S σ is the sheet conductivity, and the surface capacitance per length. In the process of adsorbing DMMP gas, S C , whi determined by the structure of the device, remains constant. The porous graphene PVDF film could both absorb DMMP gas and resulted in a decrease of resonant frequ due to the shift of the sheet conductivity. The sheet conductivity of porous phene/PVDF film was about 6.1 × 10 −3 S/cm under pure N2 and 2.7 × 10 −3 S/cm und ppm DMMP gas. As shown in Figure 9, the dynamic responses of the prepared SAW sensors with rous PVDF sensing film to fast DMMP concentration (0, 3, 5, 10 ppm) changes were in tigated. The blue bubble was the DMMP concentration-time curve. The gas concentr was determined by the ratio of nitrogen and DMMP, while 0 ppm is pure nitrogen equal gas flow. The response time and recovery time were defined as the time requ for the frequency shift to achieve 10% and 90% of the change for the adsorption an sorption processed, respectively [35]. It can be seen from Figure 9 that the response of the sensor was clearly faster than the recovery time. For the concentration of 3 pp ppm and 10 ppm, the response time was about 2.9 s, 3.6 s, and 4.5 s, and the recovery was about 3.7 s, 4.9 s, 5.8 s, respectively. When DMMP gas was introduced into the s tive membrane, the molecularly imprinted sensitive membrane could quickly captur analyte. Moreover, the hydrogen bond structure between DMMP and PVDF needs a tain gas flux to desorb when the gas concentration decreased. Figure 2b shows th adsorption and hydrogen bond formation diagram. In the multilayer sensing film, P

Dynamic Responses and Sensing Mechanism
The frequency shift of the SAW sensors based on the electric loading effect can be correlated with the sheet conductivity and the surface capacitance per length of the sensing film using the following relationship [33,34]: where v 0 is the effective SAW velocity of the device, K 2 is the electromechanical coupling coefficient of the piezoelectric substrate, σ S is the sheet conductivity, and C S is the surface capacitance per length. In the process of adsorbing DMMP gas, C S , which is determined by the structure of the device, remains constant. The porous graphene and PVDF film could both absorb DMMP gas and resulted in a decrease of resonant frequency due to the shift of the sheet conductivity. The sheet conductivity of porous graphene/PVDF film was about 6.1 × 10 −3 S/cm under pure N2 and 2.7 × 10 −3 S/cm under 10 ppm DMMP gas. As shown in Figure 9, the dynamic responses of the prepared SAW sensors with porous PVDF sensing film to fast DMMP concentration (0, 3, 5, 10 ppm) changes were investigated. The blue bubble was the DMMP concentration-time curve. The gas concentration was determined by the ratio of nitrogen and DMMP, while 0 ppm is pure nitrogen with equal gas flow. The response time and recovery time were defined as the time required for the frequency shift to achieve 10% and 90% of the change for the adsorption and desorption processed, respectively [35]. It can be seen from Figure 9 that the response time of the sensor was clearly faster than the recovery time. For the concentration of 3 ppm, 5 ppm and 10 ppm, the response time was about 2.9 s, 3.6 s, and 4.5 s, and the recovery time was about 3.7 s, 4.9 s, 5.8 s, respectively. When DMMP gas was introduced into the sensitive membrane, the molecularly imprinted sensitive membrane could quickly capture the analyte. Moreover, the hydrogen bond structure between DMMP and PVDF needs a certain gas flux to desorb when the gas concentration decreased. Figure 2b shows the gas adsorption and hydrogen bond formation diagram. In the multilayer sensing film, PVDF was used to protect the metal interdigital electrode, and porous graphene could effectively increase the specific surface area of the sensing film and increased the adsorption of the gas to be measured. When the sensitive membrane adsorbs DMMP gas, DMMP molecules pass through the porous structure, the O atom in the P = O group of DMMP and the H atom of PVDF form an H-bond [36] structure, which achieves specific recognition. The synergetic effect between PVDF and porous graphene was proven by the enhancement of sensitivity.
Micromachines 2021, 12, x FOR PEER REVIEW synergetic effect between PVDF and porous graphene was proven by the enhance sensitivity.  Figure 10 shows the selectivity of the SAW DMMP sensor. The SAW sensor have a marked response to pure N2, O2 and Ar. Although the responses could be in by pure CO2 and ethanol gases, it was also far less than DMMP of 3 ppm. These suggest that due to the preparation of molecularly imprinted sensitive membr sensor had strong selectivity for DMMP. Furthermore, as shown in Figure 11, the r frequencies for different DMMP concentrations were recorded for 15 days to ve stability of the sensors. When the time of storage increased, the frequency shift porous graphene/PVDF SAW sensor were nearly constant.   Figure 10 shows the selectivity of the SAW DMMP sensor. The SAW sensor did not have a marked response to pure N 2 , O 2 and Ar. Although the responses could be interfaced by pure CO 2 and ethanol gases, it was also far less than DMMP of 3 ppm. These results suggest that due to the preparation of molecularly imprinted sensitive membrane, the sensor had strong selectivity for DMMP. Furthermore, as shown in Figure 11, the response frequencies for different DMMP concentrations were recorded for 15 days to verify the stability of the sensors. When the time of storage increased, the frequency shifts of the porous graphene/PVDF SAW sensor were nearly constant.

Selectivity and Stability
Micromachines 2021, 12, x FOR PEER REVIEW synergetic effect between PVDF and porous graphene was proven by the enhance sensitivity.  Figure 10 shows the selectivity of the SAW DMMP sensor. The SAW sensor have a marked response to pure N2, O2 and Ar. Although the responses could be in by pure CO2 and ethanol gases, it was also far less than DMMP of 3 ppm. These suggest that due to the preparation of molecularly imprinted sensitive membr sensor had strong selectivity for DMMP. Furthermore, as shown in Figure 11, the r frequencies for different DMMP concentrations were recorded for 15 days to ve stability of the sensors. When the time of storage increased, the frequency shift porous graphene/PVDF SAW sensor were nearly constant.

Conclusions
In this paper, a DMMP gas sensor based on the SAW device with a porous phene/PVDF molecularly imprinted sensing membrane was developed. The sen membrane with a porous structure could improve the adsorption capacity of DMMP Due to the molecularly imprinted recognition technology used in the sensing film SAW sensor, which had an excellent anti-interference ability, could detect DMMP sp ically. The sensitivity of the sensor could reach −1.407 kHz·ppm −1 . The response tim the SAW sensor with porous graphene/PVDF molecularly imprinted sensing memb was faster than 4.5 s, and the recovery time was faster than 5.8 s at the concentration ppm. Furthermore, the developed SAW sensor could also be used for real-time and on monitoring of nerve gas.

Conclusions
In this paper, a DMMP gas sensor based on the SAW device with a porous graphene/PVDF molecularly imprinted sensing membrane was developed. The sensing membrane with a porous structure could improve the adsorption capacity of DMMP gas. Due to the molecularly imprinted recognition technology used in the sensing film, the SAW sensor, which had an excellent anti-interference ability, could detect DMMP specifically. The sensitivity of the sensor could reach −1.407 kHz·ppm −1 . The response time of the SAW sensor with porous graphene/PVDF molecularly imprinted sensing membrane was faster than 4.5 s, and the recovery time was faster than 5.8 s at the concentration of 10 ppm. Furthermore, the developed SAW sensor could also be used for real-time and online monitoring of nerve gas.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/mi12050552/s1, Figure S1: Responses of the bare SAW and sensor with sensitivity film for TCF, Figure S2: Residual plots for the data of Figure 8, Table S1: Statistical Analysis of the Experimental Data.