A Dual-Channel MoS2-Based Selective Gas Sensor for Volatile Organic Compounds

Significant progress has been made in two-dimensional material-based sensing devices over the past decade. Organic vapor sensors, particularly those using graphene and transition metal dichalcogenides as key components, have demonstrated excellent sensitivity. These sensors are highly active because all the atoms in the ultra-thin layers are exposed to volatile compounds. However, their selectivity needs improvement. We propose a novel gas-sensing device that addresses this challenge. It consists of two side-by-side sensors fabricated from the same active material, few-layer molybdenum disulfide (MoS₂), for detecting volatile organic compounds like alcohol, acetone, and toluene. To create a dual-channel sensor, we introduce a simple step into the conventional 2D material sensor fabrication process. This step involves treating one-half of the few-layer MoS₂ using ultraviolet–ozone (UV-O3) treatment. The responses of pristine few-layer MoS₂ sensors to 3000 ppm of ethanol, acetone, and toluene gases are 18%, 3.5%, and 49%, respectively. The UV-O3-treated few-layer MoS₂-based sensors show responses of 13.4%, 3.1%, and 6.7%, respectively. This dual-channel sensing device demonstrates a 7-fold improvement in selectivity for toluene gas against ethanol and acetone. Our work sheds light on understanding surface processes and interaction mechanisms at the interface between transition metal dichalcogenides and volatile organic compounds, leading to enhanced sensitivity and selectivity.


Introduction
Environmental pollution, caused in part by the Industrial Revolution, continues to have a detrimental impact on our planet.One of the significant pollution sources is volatile organic compounds (VOCs), which are produced by various human activities from paint manufacturing to pharmaceutical industries and from hydrocarbon fuel burning to refrigerants [1].VOCs also have direct negative health effects on humans, such as eyes, nose, and throat irritation, respiratory problems, nausea, and damage to the central nervous system and other organs [2,3].It is also reported that these gases can cause cancer [4].VOCs can cause ground-level ozone formation through atmospheric photochemical reactions with NO x , harming other living organisms and the environment [5].Due to these significant concerns, it is critical to measure and monitor VOCs both indoors and outdoors.
While achieving high sensitivity is important in gas sensor development, the primary challenge is to improve the selectivity of sensors to specific gases in a mixture of gases [27].To improve selectivity, there are several proposals for complex active materials or sensor structures.For example, Yan et al. investigated the selectivity of ethanol gas on a MoS 2 nanosheet decorated with SnO 2 nanoparticles.They reported that the response of the sensor to 500 ppm ethanol gas at 280 • C is 160%, which is higher than its response to ammonia, formaldehyde, and acetone gases [28].Chen et al. investigated the selectivity of a MoS 2 gas sensor decorated with Au nanoparticles to enhance the sensor response from 12% to 32% compared to the pristine MoS 2 for acetone gas [26].Sing and Sharma demonstrated the high responsivity of a MoS 2 /WO 3 composite structure for hydrogen sulfide and ethanol gases and reported a gas response of 17% for 5 ppm ethanol gas at 260 • C and 15% for 0.5 ppm hydrogen sulfide gas at 320 • C [29].Verma and Yadav showed good selectivity with a WO 3 /WS 2 heterostructure gas sensor for acetone gas at 132.5 and 17 at 1 and 1000 ppm concentrations compared to other gases such as ethanol, ammonia, formaldehyde, methanol, toluene, water, xylene, and ethylene glycol [30].Hussain et al. investigated the selectivity of WSe 2 nanosheets and WSe 2 /reduced graphene oxide hybrid sensors and reported that the hybrid sensor has a response of 5.5% for 100 ppm ethanol at 180 • C while the WSe 2 sensor has a 2.75 times lower response.Additionally, they investigated the sensors' selectivity for ethanol compared to other gases such as methanol, toluene, isopropyl alcohol, ammonia, and acetone [31].Nanorod-decorated nanosheets, including vanadium oxide (V 2 O 5 ) and samarium (III) oxide (Sm 2 O 3 ), have also been employed in toluene sensors as well as acetone sensors and showed good selectivity and sensitivity [32,33].
In this study, instead of using complex or hybrid active material systems, we propose a dual-channel MoS 2 gas sensor for VOCs consisting of two sensors in the same active material by modifying half of the MoS 2 nanosheet via a simple step of ultraviolet-ozone (UV-O 3 ) exposure.Therefore, the device is composed of a pristine and a UV-O 3 -treated MoS 2 sensing element on the same chip.In our previous study [34], we have shown that an appropriate UV-O 3 treatment causes a p-type doping effect in the MoS 2 as a result of the substitution and adsorption of oxygen radicals (O, O 2 , and O 3 ).The UV-O 3 treatment in this study will provide a simple mechanism to regulate the free electron density in the active sensing material, leading to an improved selectivity for polar gas molecules.Sensor fabrication and characterization techniques are presented in the experimental section.The sensing performance and different mechanisms between pristine and UV-O 3 -treated sensors are discussed in the results and discussion section.

Surface Characterization
Bulk single-crystal MoS 2 was purchased from HQ Graphene.The Scotch tape mechanical exfoliation method was used to obtain few-layer MoS 2 flakes.We first pressed the Scotch tape repeatedly onto the bulk material, and then peeled it off slowly.The process was repeated with fresh tape until flakes of the desired thickness were obtained.We then placed the flakes on clean glass substrates to perform surface characterizations before and after UV-O 3 treatment.The water contact angle (WCA) is a crucial parameter used to characterize the wettability of a surface by water.We employed the sessile drop method to measure the WCA of the clean Borofloat ® glass (PI-KEM, UK) surface and the MoS 2 flakes surface at a macroscopic level.As the size of a single flake is limited, we prepared a sample with a high density of few-layer MoS 2 flakes covering an area much larger than the water droplet (2 microliters).A high-resolution camera was used to capture the images of the droplet profile.We measured the WCAs before and after 4 min UV-O 3 treatment.
Electrostatic force microscopy (EFM imaging was performed using an atomic force microscope in tapping mode (XE100, Park System, Suwon, Republic of Korea) with a conductive tip (NSC14/Cr-Au).The tip bias was set to 1 V and the electrostatic-forceinduced phase-shift signal was acquired at a height of 10 nm from the surface.

Device Fabrication
The Scotch tape and polydimethylsiloxane (PDMS)-assisted mechanical exfoliation method was used to obtain few-layer MoS 2 flakes.Using a microscope, we identified flakes with the desired number of layers on PDMS (Gel-pak, Hayward, CA, USA).The flake was then transferred from PDMS to a SiO 2 (300 nm)-on-Si substrate with pre-patterned electrodes.The post-transfer sample was then coated with a resist layer of polymethyl methacrylate (PMMA, Allresist GmbH, Strasberg, Germany) to allow a window to be opened using the electron beam lithography process (EBL, FIE Versa3D Dual Beam, Hillsboro, OR, USA).Half of the 5-layer flake was exposed to air.This area of the MoS 2 nanosheet was exposed to UV-O 3 (Ossila, Sheffield, UK) for 4 min.The remaining PMMA resist was then removed in acetone.The electrode fingers were patterned on the sample by EBL with a bilayer resist, MMA/PMMA (Allresist GmbH, Strasberg, Germany), and Au/Cr (60 nm/10 nm) were deposited as the contact electrodes.A lift-off process was carried out in warm acetone.The fabricated sensor was mounted onto a ceramic chip holder with wires bonded for electrical characterization.

Photoluminescence and Raman Spectroscopy Measurements
Micro-photoluminescence (micro-PL) and micro-Raman measurements were carried out using a custom-built micro spectroscopy set-up, equipped with a thermoelectric cooled CCD (Newton BEX2-DD, Andor, Oxford Instruments, UK) and two gratings (300 and 1800 grooves/mm) in a spectrometer (Shamrock 500i, Andor).To optically excite the samples, a 532 nm CW laser (Gem532, Novanta Photonics, Taunton, UK) was used, and the excitation laser beam was focused to a spot that was ∼0.8 µm in diameter on the samples placed on an XYZ translation stage via a 100x NIR objective (NA = 0.7).Spectra from the samples were collected via the same objective.

Gas-Sensing Setup
The VOC vapor was generated from cooled bubblers that were immersed in a thermally controlled bath, with synthetic air as the carrier gas.The gas stream saturated with the VOC analyte was then diluted with pure synthetic air to achieve the desired gas concentration.The total gas flow rate was set to 300 sccm by the computer-controlled mass flow controllers.The current-time characteristics of the gas sensors were measured under a constant bias of 1 V using a KEITHLEY 6517B Electrometer.A LakeShore 340 (Lake Shore Cryotronics, Westerville, OH, USA) temperature controller was used to control the ambient temperature.All the gas-sensing experiments were carried out by introducing the VOC target gas mixed with dry air for 10 min followed by a flow of pure dry air for 10 min in each cycle.Gas-sensing experiments were carried out by monitoring the current change of the sensors with a constant voltage supply (1 V).The response of the sensor was calculated based on the resistivity changes of the active material so that it could be compared independently of the device geometry (with different channel widths and lengths).The response is , where ρ i and ρ s represent the resistivities of the gas before and after contact with the sensor, respectively.The resistivity of the sensor is defined as ρ = (R * w)/l , where R, w, and l represent resistance, contact width, and channel length, respectively [35].

Surface Characterization
We carried out WCA measurements and EFM imaging to characterize the surface of few-layer MoS 2 before and after the UV-O 3 treatment.Before the UV-O 3 treatment, we observed that the WCAs are very different between a blank substrate (Borofloat ® glass, 29 • ± 2 • ) and a glass substrate covered with MoS 2 flakes (78 • ± 15 • ), as shown in Figure 1a,c, indicating that the blank glass substrate is much more hydrophilic than the MoS 2 surface, which is consistent with the literature [36].The large uncertainty in the WCA values originated from the variation in flake density over a large area of the sample.After a 4 min UV-O 3 treatment, the WCA increased by 15% for the MoS 2 sample (becoming slightly more hydrophobic) but decreased by 14% for the blank reference sample (slightly more hydrophilic); see Figure 1b,d.UV-O 3 treatment is known to make glass surfaces more hydrophilic by introducing hydroxyl (OH) groups that attract water molecules.The mild increase in hydrophobicity on the MoS 2 surface can be attributed to the oxidation of the surface [37,38].
compared independently of the device geometry (with different channel widths and lengths).The response is defined as R = * 100%, where  and  represent the resistivities of the gas before and after contact with the sensor, respectively.The resistivity of the sensor is defined as  = ( * )/ , where R, w, and l represent resistance, contact width, and channel length, respectively [35].

Surface Characterization
We carried out WCA measurements and EFM imaging to characterize the surface of few-layer MoS2 before and after the UV-O3 treatment.Before the UV-O3 treatment, we observed that the WCAs are very different between a blank substrate (Borofloat ® glass, 29° ± 2°) and a glass substrate covered with MoS2 flakes (78° ± 15°), as shown in Figure 1a,c, indicating that the blank glass substrate is much more hydrophilic than the MoS2 surface, which is consistent with the literature [36].The large uncertainty in the WCA values originated from the variation in flake density over a large area of the sample.After a 4 min UV-O3 treatment, the WCA increased by 15% for the MoS2 sample (becoming slightly more hydrophobic) but decreased by 14% for the blank reference sample (slightly more hydrophilic); see Figure 1b,d.UV-O3 treatment is known to make glass surfaces more hydrophilic by introducing hydroxyl (OH) groups that attract water molecules.The mild increase in hydrophobicity on the MoS2 surface can be attributed to the oxidation of the surface [37,38].EFM is an electrical mode in AFM to map variations in the sample's electric field and reveal information about the surface potential and charge distribution.The surface potential is obtained relative to the tip potential.In our EFM measurements, we kept the tip EFM is an electrical mode in AFM to map variations in the sample's electric field and reveal information about the surface potential and charge distribution.The surface potential is obtained relative to the tip potential.In our EFM measurements, we kept the tip potential and the distance from the tip to the surface constant and obtained surface potential mapping and the average surface potential from the active region of the sensing devices.The average surface potential changes from 0.094 mV to −112.5 mV after a 4 min UV-O 3 treatment; see Figure 1e,f.This reduction in the surface potential is due to the reduction in the Fermi level of the MoS 2 flake, resulting in p-type characteristics [39].

Photoluminescence and Raman Spectroscopy of MoS 2
The layer number of the flake was further confirmed based on the peak wavelength of the PL spectrum.A PL peak wavelength at 890 nm indicates that the flake is a five-layer (5 L) MoS 2 , see the highlighted area in Figure 2a (I).In our previous study, we observed that 4 min of UV-O 3 exposure time was optimal to produce p-type-doped MoS 2 while maintaining above 80% of the initial PL intensity [34].Therefore, half of the active material was treated for 4 min while the other side was kept pristine under the resist, see Figure 2a (IV), in order to create two channels in the same material for sensing.Figure 2b,c show the PL and Raman spectra of the pristine and the UV-O 3 -treated MoS 2 , respectively.The PL spectra of two sides of the material were used as a control indicator of whether the UV-O 3 treatment degraded the material or not.With maintaining 70% of the initial PL intensity, it can be concluded there is no structural degradation in the 4 min treated material.On the other hand, the comparison of the Raman spectra between the pristine and treated materials was used as an indicator if the treated material indeed turned into being p-type-doped.The Raman spectrum of MoS 2 is characterized by two main vibration modes, E 2g and A 1g , representing the in-plane and out-of-plane vibrations, respectively [40].It is known that the E 2g mode is sensitive to the strain while the A 1g vibration mode is sensitive to the doping effect [40][41][42].Kang et al. and Shi et al., in different studies, showed that changes in the line width and blueshifts (toward the higher frequency) of the A 1g vibration mode are indications of the p-type doping effect in MoS 2 [41,43].The Raman spectrum in Figure 2c shows 0.8 cm −1 of blueshift in the A 1g vibration mode, which can be attributed to the p-type doping effect in MoS 2 .After the PL and Raman examinations, the electrodes were fabricated to achieve two sensing channels in the same device.
Nanomaterials 2024, 14, x FOR PEER REVIEW 5 of 15 potential and the distance from the tip to the surface constant and obtained surface potential mapping and the average surface potential from the active region of the sensing devices.The average surface potential changes from 0.094 mV to −112.5 mV after a 4 min UV-O3 treatment; see Figure 1e,f.This reduction in the surface potential is due to the reduction in the Fermi level of the MoS2 flake, resulting in p-type characteristics [39].

Photoluminescence and Raman Spectroscopy of MoS2
The layer number of the flake was further confirmed based on the peak wavelength of the PL spectrum.A PL peak wavelength at 890 nm indicates that the flake is a five-layer (5 L) MoS2, see the highlighted area in Figure 2a (I).In our previous study, we observed that 4 min of UV-O3 exposure time was optimal to produce p-type-doped MoS2 while maintaining above 80% of the initial PL intensity [34].Therefore, half of the active material was treated for 4 min while the other side was kept pristine under the resist, see Figure 2a (IV), in order to create two channels in the same material for sensing.Figure 2b,c show the PL and Raman spectra of the pristine and the UV-O3-treated MoS2, respectively.The PL spectra of two sides of the material were used as a control indicator of whether the UV-O3 treatment degraded the material or not.With maintaining 70% of the initial PL intensity, it can be concluded there is no structural degradation in the 4 min treated material.On the other hand, the comparison of the Raman spectra between the pristine and treated materials was used as an indicator if the treated material indeed turned into being p-typedoped.The Raman spectrum of MoS2 is characterized by two main vibration modes, E2g and A1g, representing the in-plane and out-of-plane vibrations, respectively [40].It is known that the E2g mode is sensitive to the strain while the A1g vibration mode is sensitive to the doping effect [40][41][42].Kang et al. and Shi et al., in different studies, showed that changes in the line width and blueshifts (toward the higher frequency) of the A1g vibration mode are indications of the p-type doping effect in MoS2 [41,43].The Raman spectrum in Figure 2c shows 0.8 cm −1 of blueshift in the A1g vibration mode, which can be attributed to the p-type doping effect in MoS2.After the PL and Raman examinations, the electrodes were fabricated to achieve two sensing channels in the same device.

Sensing Performance
The current-time (I-t) characteristics were investigated to examine the saturation time, i.e., time before the current saturates with the applied voltage.Figure 3a,b show the room-temperature I-t characteristics of pristine and treated sensors, respectively.The p-type doping effect due to the UV-O 3 treatment results in reduced contact resistance, and hence the measured current increases 50 times after the treatment.Although there is a difference in the magnitude of the current for these two sensing channels, the current saturates after 5 min of applied voltage in both channels.This saturation time was taken into consideration in the sensing experiments.
of the 5-layer MoS2 for the UV-O3 treatment, V.After patterning the electrode fingers to connect th MoS2 flake and the prefabricated contact pads in the MMA/PMMA resist; VI.Final device after met allization and lift-off process; (b) PL and (c) Raman spectra of the pristine (blue) and the UV-O3 treated (red) MoS2.

Sensing Performance
The current-time (I-t) characteristics were investigated to examine the saturation time, i.e., time before the current saturates with the applied voltage.Figure 3a,b show th room-temperature I-t characteristics of pristine and treated sensors, respectively.The p type doping effect due to the UV-O3 treatment results in reduced contact resistance, and hence the measured current increases 50 times after the treatment.Although there is difference in the magnitude of the current for these two sensing channels, the current sat urates after 5 min of applied voltage in both channels.This saturation time was taken into consideration in the sensing experiments.Gas measurements were carried out with the following flow sequence, with the ga flow system illustrated in Figure 3c: 10 min flow of dry air, followed by 10 min flow o target gas mixed with dry air, and then 10 min flow of dry air followed by a flow of a Gas measurements were carried out with the following flow sequence, with the gas flow system illustrated in Figure 3c: 10 min flow of dry air, followed by 10 min flow of target gas mixed with dry air, and then 10 min flow of dry air followed by a flow of a different concentration of gas mixture.In this study, we aimed to show how postgrowth functionalization of 2D materials can affect their gas-sensing performance.To understand the mechanism behind the change in performance, we eliminated all other environmental/external effects, such as 2D material layer thicknesses and humidity [44].We examined the sensing performance at room temperature (RT) and 100 • C. The response of the sensors for each gas was measured with seven different concentrations in the range of 300 ppm and 15,000 ppm.The individual range was chosen based on their occupational environment gas exposure limit.Figure 4 shows the real-time monitoring of the current change of the pristine MoS 2 gas sensor with varying gas concentrations at RT and 100 • C. While there is no response in the sensors for VOC gases at RT, reasonable responses were observed in the pristine sensor to ethanol, acetone, and toluene at 100 • C. Figure 6 shows the real-time monitoring of the current change in a few-layer UV-O3treated MoS2 sensor with varying gas concentrations at RT and 100 °C.Reasonable responses are observed for ethanol and acetone at 100 °C compared to no notable response at RT.The limits of detection (LODs) are estimated to be <300 ppm, <1500 ppm, and <500 ppm for ethanol, acetone, and toluene, respectively.Characterizing the dual-channel sensors at a wider range of temperatures may offer further improvement in sensitivity and The pristine MoS 2 used in this study is intrinsically n-doped.The main gas-sensing mechanism of the pristine MoS 2 sensors is the same as any n-type bulk semiconductorbased sensor.In an n-type semiconductor where electrons are the majority charge carriers, the oxygen molecules in the air are physically adsorbed, by trapping free electrons, onto the surface, resulting in space charge layers that are called electron depletion layers (EDLs).Because the physically bonded oxygen reacts with sensing molecules and forms chemical bonds, the resistance of the sensor reduces.The types of chemisorbed ionized-oxygen species (O − , O 2 − , or O 2− ) are largely affected by the working temperature as seen in Equations ( 1)-( 4) [45].
O 2(gas) →O 2(ads) When a pristine MoS 2 sensor is operated at 100 • C, the O 2 − ion plays a more important role in detecting VOCs.The sensing mechanism in the pristine MoS 2 sensor is illustrated in Figure 5. Ethanol, acetone, and toluene gases react with surface oxygen ions.The interaction mechanism of the chemical reactions at 100 °C is given by Equations ( 5), ( 6) and ( 7) for ethanol, acetone, and toluene, respectively.C2H5OH(g) + 3O2 − (ads) →3H2O(g) + 2CO2(g) + 3e − (5) As a result of these reactions, different amounts of CO2, H2O, and numbers of electrons are generated with the presence of ethanol, acetone, and toluene, respectively.The removal of O2 − ions from the surfaces due to reactions with the target gas, for example, ethanol, causes a thinner EDL region and a decrement in the potential barrier, which results in an increment in the sensor current as shown in Figure 6.The trend of different responses to different gases follows the amount of required O2 − for each gas molecule; see Equations ( 5), ( 6) and (7).Therefore, we can conclude that the response of the sensors was Ethanol, acetone, and toluene gases react with surface oxygen ions.The interaction mechanism of the chemical reactions at 100 • C is given by Equations ( 5)- (7) As a result of these reactions, different amounts of CO 2 , H 2 O, and numbers of electrons are generated with the presence of ethanol, acetone, and toluene, respectively.The removal of O 2 − ions from the surfaces due to reactions with the target gas, for example, ethanol, causes a thinner EDL region and a decrement in the potential barrier, which results in an increment in the sensor current as shown in Figure 6.The trend of different responses to different gases follows the amount of required O 2 − for each gas molecule; see Equations ( 5)- (7).Therefore, we can conclude that the response of the sensors was dependent on the thickness of EDLs [11,46].Figure 6 shows the real-time monitoring of the current change in a few-layer UV-O 3 -treated MoS 2 sensor with varying gas concentrations at RT and 100 • C. Reasonable responses are observed for ethanol and acetone at 100 • C compared to no notable response at RT.The limits of detection (LODs) are estimated to be <300 ppm, <1500 ppm, and <500 ppm for ethanol, acetone, and toluene, respectively.Characterizing the dual-channel sensors at a wider range of temperatures may offer further improvement in sensitivity and selectivity, which can be investigated in future work.The responses of the pristine few-layer MoS 2 sensors to 3000 ppm ethanol, acetone, and toluene gases at 100 • C are 18%, 3.5%, and 49%, respectively.The UV-O 3 -treated MoS 2 sensors show responses of 13.4%, 3.1%, and 6.7%, respectively.Therefore, such a dual-channel sensing device demonstrates a 7-fold improvement in selectivity for toluene against ethanol and acetone.Due to the toluene molecules requiring two to three times more O 2 − (ads) ions to react with compared to the ethanol and acetone molecules (see Equations ( 5)-( 7)), the response to toluene gas is expected to be affected much more significantly by the different EDL layers formed on top of the active surface before and after the UV-O 3 treatment compared to the other two gases, resulting in an enhanced selectivity for toluene.
It has been revealed that UV-O 3 treatment leads to the formation of S-O bonds in the MoS 2 .The O − ions can also exist at interstitial sites and S-vacancies [47].The physical and oxygen-mediated chemical bonds in the UV-O 3 -treated sensors result in a p-type characteristic in the MoS 2 .Therefore, UV-O 3 -treated sensors showed a reduced response to toluene compared to acetone and ethanol due to its large molecular structure and more ionized oxygen required to bind (Equation ( 7)). Figure 7 shows the comparison of the responses for the pristine and UV-O 3 -treated MoS 2 sensors.The response of the UV-O 3treated sensor reduces for all VOCs except acetone at higher concentrations.This reduction is attributed to switching the type of the free carriers from electron-dominant to holedominant, resulting in a reduction in the dynamic EBL thickness of the sensor surface during the operation.To achieve a better resolution and higher sensitivity, a larger number of sensing channels can be pursued, inspired by the structure proposed by Rabchinskii et.al. [16], with large area flakes or wafer-scale MoS 2 grown by chemical vapour deposition.Furthermore, the layer-thickness-dependent sensitivity can be explored in future studies.3.5%, and 49%, respectively.The UV-O3-treated MoS₂ sensors show responses of 13.4%, 3.1%, and 6.7%, respectively.Therefore, such a dual-channel sensing device demonstrates a 7-fold improvement in selectivity for toluene against ethanol and acetone.Due to the toluene molecules requiring two to three times more O2 − (ads) ions to react with compared to the ethanol and acetone molecules (see Equations ( 5), ( 6) and ( 7)), the response to toluene gas is expected to be affected much more significantly by the different EDL layers formed on top of the active surface before and after the UV-O3 treatment compared to the other two gases, resulting in an enhanced selectivity for toluene.
It has been revealed that UV-O3 treatment leads to the formation of S-O bonds in the MoS2.The O − ions can also exist at interstitial sites and S-vacancies [47].The physical and oxygen-mediated chemical bonds in the UV-O3-treated sensors result in a p-type characteristic in the MoS2.Therefore, UV-O3-treated sensors showed a reduced response to toluene compared to acetone and ethanol due to its large molecular structure and more ionized oxygen required to bind (Equation ( 7)). Figure 7 shows the comparison of the responses for the pristine and UV-O3-treated MoS2 sensors.The response of the UV-O3treated sensor reduces for all VOCs except acetone at higher concentrations.This reduction is attributed to switching the type of the free carriers from electron-dominant to holedominant, resulting in a reduction in the dynamic EBL thickness of the sensor surface during the operation.To achieve a better resolution and higher sensitivity, a larger number of sensing channels can be pursued, inspired by the structure proposed by Rabchinskii et.al. [16], with large area flakes or wafer-scale MoS2 grown by chemical vapour deposition.Furthermore, the layer-thickness-dependent sensitivity can be explored in future studies.

Conclusions
We have designed and fabricated a few-layer MoS2-based dual-channel gas sensor platform for volatile organic compounds, including ethanol, acetone, and toluene.The multi-pixelated sensor platform consists of two sensors side by side made from the same material.While one-half of the material is kept as pristine, i.e., an electron-dominant sensor, the other half of the material is treated by UV-O3 to obtain a hole-dominant sensor.UV-O3 treatment is used to control the free electron density in the active material and change the thickness of the electron depletion layers.Other techniques such as oxygen plasma or annealing in air or an O2 environment can also be employed for the same purpose, but we believe that UV-O3 treatment is more reliable and reproducible.The responses of pristine MoS2 gas sensors to 3000 ppm of ethanol, acetone, and toluene gases were found to be 18%, 3.5%, and 49%, and those of UV-O3-treated sensors were 13.4%, 3.1%, and 6.7%, respectively, with 0% humidity at 100 °C.The proposed double-channel

Conclusions
We have designed and fabricated a few-layer MoS 2 -based dual-channel gas sensor platform for volatile organic compounds, including ethanol, acetone, and toluene.The multi-pixelated sensor platform consists of two sensors side by side made from the same material.While one-half of the material is kept as pristine, i.e., an electron-dominant sensor, the other half of the material is treated by UV-O 3 to obtain a hole-dominant sensor.UV-O 3 treatment is used to control the free electron density in the active material and change the thickness of the electron depletion layers.Other techniques such as oxygen plasma or annealing in air or an O 2 environment can also be employed for the same purpose, but we believe that UV-O 3 treatment is more reliable and reproducible.The responses of pristine MoS 2 gas sensors to 3000 ppm of ethanol, acetone, and toluene gases were found

Figure 1 .
Figure 1.Surface characterization before and after UV-O3 treatment.(a-d) The water contact angles (WCA) were measured on a blank Borofloat ® glass and a Borofloat ® glass with a high density of fewlayer MoS2 flakes before (a,c) and after (b,d) a 4 min UV-O3 treatment, respectively; (e,f) are the electrostatic force microscopy (EFM) images of the 5 L MoS2 flake surface before and after UV-O3 treatment, respectively.

Figure 1 .
Figure 1.Surface characterization before and after UV-O 3 treatment.(a-d) The water contact angles (WCA) were measured on a blank Borofloat ® glass and a Borofloat ® glass with a high density of few-layer MoS 2 flakes before (a,c) and after (b,d) a 4 min UV-O 3 treatment, respectively; (e,f) are the electrostatic force microscopy (EFM) images of the 5 L MoS 2 flake surface before and after UV-O 3 treatment, respectively.

Figure 2 .Figure 2 .
Figure 2. (a) Optical microscope images (with a scale bar of 50 μm) of the sensor in different fabrication steps: I. As-exfoliated few-layer MoS2 flake on polydimethylsiloxane (PDMS) with the 5-layer area highlighted with red dotted lines; II.Contact electrodes fabricated on SiO2-on-Si substrate with a 200 μm × 200 μm clear window for transferring 2D materials; III.After transferring the few-layer MoS2 flake onto SiO2-on-Si substrate; IV.After opening a window in the PMMA resist layer on half

Figure 3 .
Figure 3. Current-time (with a constant bias of 1 V applied) plots of (a) pristine and (b) UV-O3 treated MoS2, respectively, and (c) a schematic illustration of the gas measurement system.

Figure 3 .
Figure 3. Current-time (with a constant bias of 1 V applied) plots of (a) pristine and (b) UV-O 3 -treated MoS 2 , respectively, and (c) a schematic illustration of the gas measurement system.

Nanomaterials 2024 , 15 Figure 6 .
Figure 6.The responses of UV-O3-treated few-layer MoS2 gas sensors exposed to (a) ethanol, (b) acetone, and (c) toluene gases at 100 °C (top) and at RT (bottom).The background drifts in the current values are due to the carriers stabilizing in the sensor.

Figure 4 .
Figure 4. Real-time monitoring of current change in a few-layer pristine MoS 2 -based gas sensor with varying gas concentrations of (a) ethanol, (b) acetone, and (c) toluene gases at 100 • C (top) and room temperature (RT, bottom).

15 Figure 5 .
Figure 5. (a) Illustration of the sensing mechanism, (b) response of the pristine MoS2-based gas sensor to dry air and ethanol gas at 100 °C, and (c) illustration of the sensing mechanism of UV-O3treated sensor.

Figure 5 .
Figure 5. (a) Illustration of the sensing mechanism, (b) response of the pristine MoS 2 -based gas sensor to dry air and ethanol gas at 100 • C, and (c) illustration of the sensing mechanism of UV-O 3 -treated sensor.

Figure 4 .
Figure 4. Real-time monitoring of current change in a few-layer pristine MoS2-based gas sensor with varying gas concentrations of (a) ethanol, (b) acetone, and (c) toluene gases at 100 °C (top) and room temperature (RT, bottom).

Figure 6 .
Figure 6.The responses of UV-O 3 -treated few-layer MoS 2 gas sensors exposed to (a) ethanol, (b) acetone, and (c) toluene gases at 100 • C (top) and at RT (bottom).The background drifts in the current values are due to the carriers stabilizing in the sensor.