Organic/Inorganic-Based Flexible Membrane for a Room-Temperature Electronic Gas Sensor

A room temperature (RT) H2S gas sensor based on organic–inorganic nanocomposites has been developed by incorporating zinc oxide (ZnO) nanoparticles (NPs) into a conductivity-controlled organic polymer matrix. A homogeneous solution containing poly (vinyl alcohol) (PVA) and ionic liquid (IL) and further doped with ZnO NPs was used for the fabrication of a flexible membrane (approx. 200 μm in thickness). The sensor was assessed for its performance against hazardous gases at RT (23 °C). The obtained sensor exhibited good sensitivity, with a detection limit of 15 ppm, and a fast time response (24 ± 3 s) toward H2S gas. The sensor also showed excellent repeatability, long-term stability and selectivity toward H2S gas among other test gases. Furthermore, the sensor depicted a high flexibility, low cost, easy fabrication and low power consumption, thus holding great promise for flexible electronic gas sensors.


Introduction
Nature has been so much stained and strained by the actions of humans, resulting in a rapid increase in environmental pollution. Therefore, its monitoring has become the need of the hour and a priority for the health of mankind. One of the threats that make our environment hostile is hazardous gases. Some gases are more toxic than others, depending on their effects that range from mild respiratory symptoms to eventual death. Enormous efforts have been invested in tackling this situation by developing sensors, which are highly sensitive, selective and stable toward the target gases, among others [1][2][3][4].
Occupational exposure to toxic gases has been a major concern in the petroleum industry, with hydrogen sulfide (H 2 S) being one of the contenders. It is also known as a "knock down gas", as exposure to it may lead to immediate loss of consciousness and death [5][6][7]. Overlong exposure to 10-500 parts per million (ppm) causes respiratory symptoms, and exposure to 500-1000 ppm can be fatal [5]. In addition to the petroleum industry, H 2 S gas is evolved in agricultural industry, garbage disposal sites, sewers, tanneries, wastewater treatment and mining industries, to name a few [8][9][10][11].
Recently, there have been numerous developments in the area of gas detection systems. The sensors exploit various principles, such as chemiresistive sensing [12], capacitive-based sensing [13], quartz crystal microbalance sensors [14], optical transduction [15] and organic field-effect transistor sensors. Among these, chemiresistive sensors that track the change in electrical conductivity or resistance of the sensing material have attracted much attention due to the high sensitivity, low cost, ease of fabrication and possibility of miniaturization.
Zinc oxide (ZnO) possesses a range of physicochemical properties that are apt for hazardous gas sensing applications. It is classified as an n-type II-VI semiconductor with a wide bandgap (3.37 eV), a large excitation binding energy (60 meV) and high electron mobility (400 cm 2 V −1 s −1 ) [16,17]. Additionally, ZnO is chemically stable, environmentally friendly and can be synthesized at a low cost [18], making it a promising material for the task. The crystalline nature of ZnO allows it to be grown in different structures, such as nanoparticles (0D), one-dimensional (1D), two-dimensional (2D) and three-dimensional (3D) structures [19]. The morphologies, such as nanorods, wires, needles, spheres, ellipsoids and flowers [20], enable controlling the surface area-to-volume ratio, thereby enhancing the utility of the material in a plethora of sensing applications. In addition to that, the bandgap of the material can be altered by doping ZnO with various materials, which improve the sensitivity of the sensor with varying operating temperatures [21][22][23][24][25][26][27].
The working principle of the ZnO-based sensors is usually evaluated at elevated temperatures of about 300-500 • C [28][29][30][31][32]. To attain these temperatures, external energy has to be provided to the sensor, which would mean an increase in the operational cost. Furthermore, at these temperatures, the flammable and explosive gases are more prone to explosions due to their low ignition point. Additionally, at higher temperatures, the stability of the material reduces, which would generate inaccurate results [33,34], and this also leads to shortening the life time of the sensor. The sensors that are developed to operate at room temperature do not require any additional heating elements; hence, their operational cost is reduced. In addition, they can be made portable, thus their risk of explosion is avoided.
Organic polymers, such as PVA, are characterized by their wide abundance and their known promising properties, such as flexibility, environmental friendliness, thermal stability and the ability to form electrolytes by virtue of their hydrophilic nature [35]. Moreover, the conductivity of PVA can be controlled by doping it with an ionic liquid (IL), such as glycerol [36][37][38]. These combined properties can be further exploited by doping the polymer matrix with materials that have an affinity toward the detection of H 2 S gas, so that changes in their conductivity or resistivity can be recorded.
The aim of this investigation is to develop an organic-inorganic hybrid gas sensor, where a matrix of PVA/IL polymer is doped with ZnO NPs, and to explore its gas sensing capabilities toward H 2 S gas, among other hazardous gases. The low fabrication and operational costs, flexibility and operational ability at room temperature enable the device to be deployed for monitoring these threats in real-time application scenarios.

Fabrication of the ZnO/PVA/IL Membrane
An amount of 50 mg of ZnO NPs was first dispersed in a 40 mL of distilled water (DW) using a vortex shaker. Then, 1000 mg of PVA granules was added to the dispersed NPs and stirred vigorously at 90 • C until a homogenous solution was obtained. After that, the mixture was doped with 1 mL of glycerol and was kept under continuous stirring for 20 min. The mixture formed was then cast onto a Petri dish and was dried in air at 70 • C for 18 h, resulting in a flexible membrane, as shown in Figure 1. The thickness of the membrane was determined to be 200 µm.

Characterization
ZnO NPs were subjected to analysis for their structural and morphological characteristics. A powder X-ray diffraction (PXRD) of the as-received ZnO NPs was performed using a Rigaku, MiniFlex 600-C instrument, Austin, TX, USA, with a Cu Ka X-ray with a scan range of 2-80 • and at a scan rate of 1 • /min. The morphology of the ZnO powder, as well as the membrane, was investigated using a Thermoscientific Quattro S SEM instrument (Waltham, MA, USA) at an operating voltage of 15 kV. The membrane was further analyzed for its composition by Fourier transform infrared (FT-IR) spectroscopy using a Thermo Nicolet, NEXUS, 470 FT-IR instrument. A KBr disk method over a scan range of 400-4000 cm −1 was used. The TGA plots were recorded using TGA-Q500, TA Instruments, New Castle, DE, USA, with the ramp rate of 20 • C/min and within a temperature range of 30-800 • C.

Characterization
ZnO NPs were subjected to analysis for their structural and morphological chara teristics. A powder X-ray diffraction (PXRD) of the as-received ZnO NPs was performe using a Rigaku, MiniFlex 600-C instrument, Austin, TX, USA, with a Cu Ka X-ray with scan range of 2-80° and at a scan rate of 1°/min. The morphology of the ZnO powder, well as the membrane, was investigated using a Thermoscientific Quattro S SEM instr ment (Waltham, MA, USA) at an operating voltage of 15 kV. The membrane was furth analyzed for its composition by Fourier transform infrared (FT-IR) spectroscopy using Thermo Nicolet, NEXUS, 470 FT-IR instrument. A KBr disk method over a scan range 400-4000 cm −1 was used. The TGA plots were recorded using TGA-Q500, TA Instrumen New Castle, DE, USA, with the ramp rate of 20 °C/min and within a temperature range 30-800 °C.

Sensor Fabrication and H2S Gas Sensing Tests
The sensor was fabricated as delineated in our previous work [36]. A mixture of th test gas, diluted with synthetic air, was introduced into the chamber using Bronkhor mass flow controllers (MFC). The test chamber was sealed and placed inside a fume hoo to maintain a non-humid atmosphere throughout the measurement. The device was ke at RT throughout the testing sequences. Keithley Instruments source measurement un (KI236) was used to record the response of the sensor, applying a constant bias voltage 4 V between the electrodes. Labview software (version 21.0.1-32-bit) was used to interfa and record the response of these units. Figure 2 shows a comparison between the PXRD patterns of the as-received Zn NPs, the PVA/IL membrane and the ZnO-doped PVA/IL membrane. Compared with th JCPDS PXRD pattern of pure ZnO (Card # 36-1451), the as-received ZnO shows a phas pure composition where all peaks of the standard ZnO were observed. The weakly cry talline PVA within the PVA/IL and composite matrices showed two broad peaks at 19 and 20.8° [39][40][41][42]. Additionally, the composite membrane confirmed the presence of th ZnO NPs, where its characteristic peaks representing the 010, 002 and 011 planes we observed.

Sensor Fabrication and H 2 S Gas Sensing Tests
The sensor was fabricated as delineated in our previous work [36]. A mixture of the test gas, diluted with synthetic air, was introduced into the chamber using Bronkhorst mass flow controllers (MFC). The test chamber was sealed and placed inside a fume hood to maintain a non-humid atmosphere throughout the measurement. The device was kept at RT throughout the testing sequences. Keithley Instruments source measurement unit (KI236) was used to record the response of the sensor, applying a constant bias voltage of 4 V between the electrodes. Labview software (version 21.0.1-32-bit) was used to interface and record the response of these units. The membrane was subjected to thermogravimetric analysis (TGA), as seen in Figure 3A. From the decomposition curves, we are able to observe that there is a gradual loss in weight due to the evaporation of water molecules and adsorbed moisture content. Additionally, the loss in weight between 300 and 550 • C is attributed to the removal of organic groups [41]. No appreciable weight loss beyond 700 • C was observed, inferring the formation of stable inorganic phases [43]. A slight variation in the thermal profile of the composite membrane was observed, which is due to the fact that ZnO NPs are thermally stable throughout the heating cycle. The recorded profile matches well with the reported literature [41], confirming the incorporation of ZnO NPs into the polymer matrix.

Structural and Morphological Characterization of ZnO NPs and ZnO/PVA/IL Membrane
The FTIR analysis from Figure 3B shows that the membrane has the functional peaks associated with PVA [42,44] and ZnO NPs [42,45], along with a slight shift in their positions, inferring the incorporation of the NPs in the polymer matrix. The broad absorption bands at 3272 cm −1 and 2938 cm −1 denoting the O-H and C-H stretching modes, respectively, are observed. The band at 555 cm −1 is attributed to the Zn-O stretching vibration mode [42], while the bands at 1142 cm −1 and 1329 cm −1 are ascribed to the primary and secondary alcohol in-plane bending modes [45]. The membrane was subjected to thermogravimetric analysis (TGA), as seen in Figure  3A. From the decomposition curves, we are able to observe that there is a gradual loss in weight due to the evaporation of water molecules and adsorbed moisture content. Additionally, the loss in weight between 300 and 550 °C is attributed to the removal of organic groups [41]. No appreciable weight loss beyond 700 °C was observed, inferring the formation of stable inorganic phases [43]. A slight variation in the thermal profile of the composite membrane was observed, which is due to the fact that ZnO NPs are thermally stable throughout the heating cycle. The recorded profile matches well with the reported literature [41], confirming the incorporation of ZnO NPs into the polymer matrix. The FTIR analysis from Figure 3B shows that the membrane has the functional peaks associated with PVA [42,44] and ZnO NPs [42,45], along with a slight shift in their positions, inferring the incorporation of the NPs in the polymer matrix. The broad absorption bands at 3272 cm −1 and 2938 cm −1 denoting the O-H and C-H stretching modes, respectively, are observed. The band at 555 cm −1 is attributed to the Zn-O stretching vibration mode [42], while the bands at 1142 cm −1 and 1329 cm −1 are ascribed to the primary and secondary alcohol in-plane bending modes [45]. Figure 4 shows the SEM and EDX analysis, respectively, of the ZnO/PVA/IL membrane. The image shows a uniform dispersion of the NPs incorporated into the membrane. However, ZnO NPs were observed in the form of agglomerates with a uniform size distribution, which is a common criterion of NPs dispersed in highly viscous PVA solutions. The elemental analysis confirms the presence of Zn and O in the composite membrane.  The membrane was subjected to thermogravimetric analysis (TGA), as seen in Figure  3A. From the decomposition curves, we are able to observe that there is a gradual loss in weight due to the evaporation of water molecules and adsorbed moisture content. Additionally, the loss in weight between 300 and 550 °C is attributed to the removal of organic groups [41]. No appreciable weight loss beyond 700 °C was observed, inferring the formation of stable inorganic phases [43]. A slight variation in the thermal profile of the composite membrane was observed, which is due to the fact that ZnO NPs are thermally stable throughout the heating cycle. The recorded profile matches well with the reported literature [41], confirming the incorporation of ZnO NPs into the polymer matrix. The FTIR analysis from Figure 3B shows that the membrane has the functional peaks associated with PVA [42,44] and ZnO NPs [42,45], along with a slight shift in their positions, inferring the incorporation of the NPs in the polymer matrix. The broad absorption bands at 3272 cm −1 and 2938 cm −1 denoting the O-H and C-H stretching modes, respectively, are observed. The band at 555 cm −1 is attributed to the Zn-O stretching vibration mode [42], while the bands at 1142 cm −1 and 1329 cm −1 are ascribed to the primary and secondary alcohol in-plane bending modes [45]. Figure 4 shows the SEM and EDX analysis, respectively, of the ZnO/PVA/IL membrane. The image shows a uniform dispersion of the NPs incorporated into the membrane. However, ZnO NPs were observed in the form of agglomerates with a uniform size distribution, which is a common criterion of NPs dispersed in highly viscous PVA solutions. The elemental analysis confirms the presence of Zn and O in the composite membrane.

Gas Sensing Performance
The composite membrane sensor was evaluated for its response against H2S gas exposure at different concentrations with respect to time at RT. Figure 5A shows the sensor response to different concentrations of H2S gas over a period of time. Figure 5B shows the current response with respect to the test gas concentration. The sensor response (S%) was calculated using Equation (1).
where Ia is the reference current when the sensor was exposed to air, and Ig is the current measured when the sensor was exposed to H2S. An increase in the response was observed as a result of increasing the concentration of the test gas. After each exposure to the test gas, the chamber was flushed with synthetic air to remove any residual gas molecules. It is also noted that while the chamber was flushed, the current values reduced to their starting values in the absence of the test gas, demonstrating the reversibility of the sensor. The sensor was further tested for its repeatability and long-term stability. Figure 6A demonstrates an excellent repeatability of sensing in the membrane with a near-identical current response for five cycles of exposure to H2S gas at a concentration of 100 ppm. Figure 6B shows the response of the membrane to 100 ppm of H2S gas exposed for 21 days, which demonstrates its long-term stability. It can be seen that the response is in the 94-99% region, with minimal error bars.

Gas Sensing Performance
The composite membrane sensor was evaluated for its response against H 2 S gas exposure at different concentrations with respect to time at RT. Figure 5A shows the sensor response to different concentrations of H 2 S gas over a period of time. Figure 5B shows the current response with respect to the test gas concentration. The sensor response (S%) was calculated using Equation (1).
where I a is the reference current when the sensor was exposed to air, and I g is the current measured when the sensor was exposed to H 2 S.

Gas Sensing Performance
The composite membrane sensor was evaluated for its response against H2S gas exposure at different concentrations with respect to time at RT. Figure 5A shows the sensor response to different concentrations of H2S gas over a period of time. Figure 5B shows the current response with respect to the test gas concentration. The sensor response (S%) was calculated using Equation (1).
where Ia is the reference current when the sensor was exposed to air, and Ig is the current measured when the sensor was exposed to H2S. An increase in the response was observed as a result of increasing the concentration of the test gas. After each exposure to the test gas, the chamber was flushed with synthetic air to remove any residual gas molecules. It is also noted that while the chamber was flushed, the current values reduced to their starting values in the absence of the test gas, demonstrating the reversibility of the sensor. The sensor was further tested for its repeatability and long-term stability. Figure 6A demonstrates an excellent repeatability of sensing in the membrane with a near-identical current response for five cycles of exposure to H2S gas at a concentration of 100 ppm. Figure 6B shows the response of the membrane to 100 ppm of H2S gas exposed for 21 days, which demonstrates its long-term stability. It can be seen that the response is in the 94-99% region, with minimal error bars. An increase in the response was observed as a result of increasing the concentration of the test gas. After each exposure to the test gas, the chamber was flushed with synthetic air to remove any residual gas molecules. It is also noted that while the chamber was flushed, the current values reduced to their starting values in the absence of the test gas, demonstrating the reversibility of the sensor.
The sensor was further tested for its repeatability and long-term stability. Figure 6A demonstrates an excellent repeatability of sensing in the membrane with a near-identical current response for five cycles of exposure to H 2 S gas at a concentration of 100 ppm. Figure 6B shows the response of the membrane to 100 ppm of H 2 S gas exposed for 21 days, which demonstrates its long-term stability. It can be seen that the response is in the 94-99% region, with minimal error bars. Furthermore, the sensor's response time, which is the time needed for the sensor to reach 90% of its maximum response, was also calculated. The average time response for the proposed sensor was 24 ± 3 s for 100 ppm of H2S gas at RT. Another parameter that would evaluate the sensor's performance would be the recovery time. The recovery time of the sensor is defined as the time required by the sensor to recover to 90% of its baseline signal after the target gas has been stopped. The average time for the sensor was calculated to be 112 ± 5 s for 100 ppm of H2S gas at RT. Table 1 outlines a comparison of the performance of the proposed sensor with the reported literature.
Finally, the sensor's selectivity was measured by exposing it to 100 ppm of H2, C2H4 and CO gases at RT. Figure 7 illustrates the response of the sensor to those gases, showing a significant difference between the response to H2S compared to the other gases, which indicates an excellent selectivity of the proposed sensor toward those gases.  Furthermore, the sensor's response time, which is the time needed for the sensor to reach 90% of its maximum response, was also calculated. The average time response for the proposed sensor was 24 ± 3 s for 100 ppm of H 2 S gas at RT. Another parameter that would evaluate the sensor's performance would be the recovery time. The recovery time of the sensor is defined as the time required by the sensor to recover to 90% of its baseline signal after the target gas has been stopped. The average time for the sensor was calculated to be 112 ± 5 s for 100 ppm of H 2 S gas at RT. Table 1 outlines a comparison of the performance of the proposed sensor with the reported literature. Finally, the sensor's selectivity was measured by exposing it to 100 ppm of H 2 , C 2 H 4 and CO gases at RT. Figure 7 illustrates the response of the sensor to those gases, showing a significant difference between the response to H 2 S compared to the other gases, which indicates an excellent selectivity of the proposed sensor toward those gases.

Gas Sensing Mechanism
The sensing mechanism of the ZnO/PVA/IL can be elaborated as a surface sensing mechanism [16,[55][56][57], which is based on a process of adsorption-oxidation-desorption. ZnO has a non-centrosymmetric wurtzite crystal structure with polar surfaces, as shown in Figure 8. When ZnO NPs, which are homogeneously embedded in a polymeric matrix, are flushed with synthetic air, the hydrophilic nature of the matrix allows the oxygen molecules to become adsorbed on the surface, which captures the electrons from the ZnO conduction band, leading to the formation of negatively charged ions (O 2 − , O − , O 2− ) [16] and thereby providing surface acceptors sites, as shown in Figure 8A.
Upon reaching equilibrium, the conductivity of the ZnO NPs reaches a steady state. Upon the exposure to H 2 S, as a reducing gas, a charge transfer process is initiated via the oxidization of the surface-adsorbed oxygen anions, providing additional free electrons that migrate into the conduction band of ZnO, as shown in Figure 8B. This process results in a reduction in the thickness of the depletion layer, which gradually reaches equilibrium; hence, the conductivity of the membrane is enhanced. When the chamber is flushed with synthetic air, the target gas molecules are gradually expelled from the membrane, which reduces the conductivity of the material to the base values recorded. The continuity of these steps explains the sensing mechanism of ZnO NPs, as depicted by our findings. Furthermore, the sensor's response time, which is the time needed for the sensor to reach 90% of its maximum response, was also calculated. The average time response for the proposed sensor was 24 ± 3 s for 100 ppm of H2S gas at RT. Another parameter that would evaluate the sensor's performance would be the recovery time. The recovery time of the sensor is defined as the time required by the sensor to recover to 90% of its baseline signal after the target gas has been stopped. The average time for the sensor was calculated to be 112 ± 5 s for 100 ppm of H2S gas at RT. Table 1 outlines a comparison of the performance of the proposed sensor with the reported literature.
Finally, the sensor's selectivity was measured by exposing it to 100 ppm of H2, C2H4 and CO gases at RT. Figure 7 illustrates the response of the sensor to those gases, showing a significant difference between the response to H2S compared to the other gases, which indicates an excellent selectivity of the proposed sensor toward those gases.

Gas Sensing Mechanism
The sensing mechanism of the ZnO/PVA/IL can be elaborated as a surface sensing mechanism [16,[55][56][57], which is based on a process of adsorption-oxidation-desorption. ZnO has a non-centrosymmetric wurtzite crystal structure with polar surfaces, as shown in Figure 8. When ZnO NPs, which are homogeneously embedded in a polymeric matrix, are flushed with synthetic air, the hydrophilic nature of the matrix allows the oxygen molecules to become adsorbed on the surface, which captures the electrons from the ZnO conduction band, leading to the formation of negatively charged ions (O2 − , O − , O 2− ) [16] and thereby providing surface acceptors sites, as shown in Figure 8A. Upon reaching equilibrium, the conductivity of the ZnO NPs reaches a steady state. Upon the exposure to H2S, as a reducing gas, a charge transfer process is initiated via the oxidization of the surface-adsorbed oxygen anions, providing additional free electrons that migrate into the conduction band of ZnO, as shown in Figure 8B. This process results The homogeneous distribution of the ZnO NPs within a polymeric matrix that contains a conductive IL is believed to provide an enhanced venue for the above-mentioned mechanism to take place at RT. This can be explained by the extensive H-bonding network that takes place between the -OH groups of the PVA polymeric matrix, as well as the IL. This network facilitates charge transfer across the composite membrane, following the adsorption of the oxygen species from air and their reduction by H 2 S in a steam of the latter gas [58]. At temperatures between 150 and 300 • C, a hybrid mechanism of surface reactivity and gas diffusion contributes to the sensitivity of the membrane, whereas above 300 • C, the sensitivity is limited to the gas diffusion phenomenon. Detailed equations representing the surface adsorption reactions were outlined by Kang et al. [16]. The sensitivity of the membrane reported in the current investigation would be supported by the former. Moreover, the even distribution of the ZnO NPs in the polymeric membrane would also support the enhancement of the sensitivity toward H 2 S gas in comparison to the membranes deprived of the NPs.

Conclusions
This investigation demonstrates the potential of fabricating high-performance H 2 S gas sensors based on organic-inorganic nanocomposites. The flexible mixed-matrix membrane was prepared by doping inorganic ZnO NPs into an organic PVA polymer together with IL. The fabricated membrane was investigated for its gas-sensing performance. The ZnOfree PVA/IL membrane did not show any sensing response toward H 2 S gas, whereas the ZnO-doped membrane showed good sensitivity, with a detection limit of 15 ppm and fast time response of 24 ± 3 s at RT, along with its high selectivity and long-term stability. Considering this low operating temperature, the requirement for external heating elements is not necessary, hence, the fabrication and operational costs are reduced. Moreover, the composites of the membrane are known to cause no harm to the environment, which makes them eco-friendly. Therefore, the proposed sensor could be implanted into electronic devices for potentially monitoring harmful gases in real time with a high efficiency.