Robust Room-Temperature NO2 Sensors from Exfoliated 2D Few-Layered CVD-Grown Bulk Tungsten Di-selenide (2H-WSe2)

We report a facile and robust room-temperature NO2 sensor fabricated using bi- and multi-layered 2H variant of tungsten di-selenide (2H-WSe2) nanosheets, exhibiting high sensing characteristics. A simple liquid-assisted exfoliation of 2H-WSe2, prepared using ambient pressure chemical vapor deposition, allows smooth integration of these nanosheets on transducers. Three sensor batches are fabricated by modulating the total number of layers (L) obtained from the total number of droplets from a homogeneous 2H-WSe2 dispersion, such as ∼2L, ∼5–6L, and ∼13–17L, respectively. The gas-sensing attributes of 2H-WSe2 nanosheets are investigated thoroughly. Room temperature (RT) experiments show that these devices are specifically tailored for NO2 detection. 2L WSe2 nanosheets deliver the best rapid response compared to ∼5–6L or ∼13–17L. The response of 2L WSe2 at RT is 250, 328, and 361% to 2, 4, and 6 ppm NO2, respectively. The sensor showed nearly the same response toward low NO2 concentration even after 9 months of testing, confirming its remarkable long-term stability. A selectivity study, performed at three working temperatures (RT, 100, and 150 °C), shows high selectivity at 150 and 100 °C. Full selectivity toward NO2 at RT confirms that 2H-WSe2 nanosheet-based sensors are ideal candidates for NO2 gas detection.


■ INTRODUCTION
Fossil fuel combustion and automotive emissions always result in highly toxic emissions. Some of the most commonly known pollutants are nitrogen dioxide (NO 2 ), hydrogen disulfide (H 2 S), ammonia (NH 3 ), and acetone, to name a few. These gases inflict severe detrimental effects on human health and are the primary reason behind lung diseases, such as respiratory irritation syndrome, emphysema, and chronic bronchitis. 1−3 So, developing highly sensitive and selective gas sensors for air monitoring and human health protection becomes an important issue. As a result, the health hazards caused especially by NO 2 emission are widely acknowledged by the World Health Organization (WHO) in air pollution guidelines wherein NO 2 is considered as one of the dangerous pollutants in higher concentrations. Among the various gas sensors, resistance-type gas sensors are the most attractive and practical for use in sensing toxic analytes and explosive gases due to their facile fabrication, ease of operation, low cost, and miniaturization. 4−6 In general, these resistance-type gas sensors are based on metal oxide semiconductors (MOS). MOS has achieved great success due to its excellent performance in sensing different harmful gases, such as NO 2 , NH 3 , CO, and H 2 S. However, their long response/recovery times are still significant barriers to the large-scale implementation of these technologies. 7−10 Most MOS gas sensors require high operating temperatures to achieve fast responses with short response/recovery times. 11 In this regard, 2D materials (2DMs) and especially transition-metal chalcogenides (TMCs) or transition-metal di-chalcogenides (TMDCs) specify a massive range of unique properties that prove their usefulness in applications toward gas sensing. 12,13 Semiconducting two-dimensional (2D) TMDCs of the type MX 2 , where M = Mo, W and X = S, Se, are encouraging materials to be explored in areas of gas-sensing because they are exceptionally sensitive to the ambient conditions. These classes of materials also have significant response as a result of their high surface-to-volume ratio. Recent reports indicate that nanosheet-like structure of WSe 2 and MoS 2 /SnO 2 heterostructure-based gas sensor have high response toward nitrogen dioxide (NO 2 ), an outstanding selectivity compared with other gases, and an excellent longterm stability (up 8 weeks). 4,14 Other examples include large area functionalization of TMDC-like WS 2 nanosheet gas sensors which show a reliable response toward both acetone and NO 2 . Recent literature also sheds light on the areas of porous TMCs which have commendable responses and recovery times toward formaldehyde sensing and rich electrochemistry in energy storage systems such as supercapacitors, batteries, and other applications. All such promising applications of TMCs, and most importantly TMDCs, enable us to verify the rich versatility in these classes of materials as promising entities toward emerging applications. 15−17 The extremely high surface sensitivity, enhanced surfacearea-to-volume ratios coupled with the superior electrical features, flexibility, and ease of integration into different devices enable them to compete toward a class of highperformance chemical sensors. 18−20 Considering the device specifications and the urge toward tailoring these materials for interacting with various chemical species at the non-covalent level, it helps to focus and understand their selectivity and sensitivity toward the detection of different analytes, including gases, ions, and small biomolecules. The interactions between individual sheets of 2DMs and the target molecules/ions occur either by physisorption (i.e., through non-covalent interactions of molecular units onto the basal planes of 2DM layers) or through chemisorption (i.e., chemical reactions between reactive species and 2DM), leading to covalent bonds onto their basal planes. 21 However, the non-covalent interactions become the primary connections when quick response and the fast recovery are prerequisites during real-time monitoring.
The family of TMDCs is among the most studied 2DMs in recent years due to their vast applicability. 22 TMDCs are semiconductors of the type MX 2 , as mentioned before. 23 TMDCs are promising materials due to their unique chemical and physical properties, including semiconducting properties, high surface-area-to-volume ratios, absorption coefficients, and adjustable and direct band gaps. 24 The bulk TMDCs can be easily exfoliated, yielding mono-layered or few-layered 2DMs where one atomic layer of the metal is sandwiched in between two layers of X atoms. 22 The chemical versatility of TMDs and their reactivity, together with their natural abundance, result in the ever-growing interest in those materials. Till date, 2Dlayered TMDCs, such as molybdenum disulfide (MoS 2 ), tungsten disulfide (WS 2 ), and tungsten selenide (WSe 2 ), have been explored as sensitive 2D layered materials that can detect NO 2 owing to their large surface-to-volume ratio with high surface activities and excellent compatibility with the current Si-based fabrication technologies for electronic devices. 25−27 Among the widely researched TMDCs, the 2H variant of tungsten di-selenide (2H-WSe 2 ) is considered as a potential candidate among TMDCs in many applications such as chemical sensors, solar cells, energy storage, and photocatalysis. 28−31 Some of the recent works on the WSe 2 -layered structure show that this material is extensively used in applications, such as in building new-generation sodium-ion batteries (SIB). The WSe 2 /carbon composite architecture has excellent performance as a new and efficient anode material in SIB owing to its multilayer structure and large interlayer spacing that helps facilitate and further increase the intercalation of sodium ions. 32,33 WSe 2 has shown its capability as a suitable candidate for solar water splitting. Hybrid structure such as carbon nanotube/WSe 2 exhibits superior performance in photodegradation of the organic dye methyl orange. 34,35 Highperformance flexible solar cells are also widely investigated using multiple layers of WSe 2 due to which high-power conversion efficiency is achieved. 36 Among all the widely researched TMDCs, 2H-WSe 2 is a commonly known semiconductor which shows its ability toward detecting toxic gases such as NO 2 . 27 Previous reports by several groups demonstrate a p-type field-effect transistor based on mechanically exfoliated 2H-WSe 2 monolayers for NO 2 detection where palladium (Pd) was used as the source and drain electrodes to lower the contact resistance for hole injection. 24 Upon exposure to 0.05% of NO 2 , the source−drain current increased 5 orders of magnitude due to the decrease of the Schottky barrier and the increase of p-doping. 24 Furthermore, in devices with a top-contact gate, the on−off ratio was also significantly improved after NO 2 exposure. 24 TMDCs are also characterized by an abundance or absence of central chalcogen atoms, functioning as suitable coordination sites for specific heavy metal ions. 20 For many of these reasons, as cited above, TMDCs are now being considered promising sorbents toward detecting various harmful gases that cause serious health issues.
Here, we report the simple incorporation of liquid-phase exfoliated (LPE) 2H-WSe 2 [synthesized using standardized reaction parameters in an ambient pressure chemical vapor deposition (AP-CVD)] flakes/nanosheets from bulk 2H-WSe 2 grown on W foil directly using either ethanol or isopropanol as the exfoliating solvents. The flakes or sheets of 2H-WSe 2 have been successfully integrated into selective and sensitive sensors toward NO 2 gas. In the case of TMDC fabrication, the choice of the synthetic method also dictates the properties that they exhibit. Solvothermal, soft lithography, electrospinning, surface etching, brush-coating, microwave-assisted hydrothermal, spray deposition, CVD, and so forth remain as some of the most widely researched areas for morphological tuning. 37 However, CVD holds a few advantages for TMDC fabrication. CVD in itself is responsible for a facile approach toward both bottomup and top-down syntheses of TMDCs. Considering the previous reports of WSe 2 being studied toward NO 2 sensing, the CVD-based approaches in WSe 2 preparation involved the reaction between WO 3 powder and elemental Se on the sapphire substrate in the presence of Ar/H 2 , 38 or deposition of the WSe 2 film and diethyl selenide on the Si/SiO 2 substrate, 39 or mechanical exfoliation and transfer of WSe 2 flakes on top of a prepatterned Si/SiO 2 surface. 40 So, there are no reports available on the gas-sensing properties and mechanisms in LPE 2D 2H-WSe 2 (prepared by an AP-CVD method). According to our previous work on the growth of 2H-WSe 2 directly on W foil and elemental Se using a two furnace CVD system, we have shown the formation of a phase-pure material which can be exfoliated into bilayer sheets. Our previously adopted bulk 2H-WSe 2 synthesis is also highly scalable, which holds advantage with respect to the yields obtained. Further optimization in this work is the study toward modulating the total number of layers of 2H-WSe 2 by preparing a homogeneous dispersion in ethanol that can be easily dropcasted onto the TiW/Pt pads (simply by controlling the total number of droplets of the dispersion). This ultimately leads to a variation in response to the NO 2 gas sensing. This particular study, according to the best of our knowledge, is the first report of APCVD-grown LPE 2H-WSe 2 from bulk 2H-WSe 2 grown on tungsten (W) foil, as per our recent report on 2H-WSe 2 growth 41 acting as a chemical sensor for NO 2 detection (primarily based on sensing concerning droplet modulation). The pristine 2H-WSe 2 is prepared according to our previous work, where a surface reaction between W foil and elemental Se (in the absence of any reductant) drives the formation of 2H-WSe 2 on the exposed part of the W foil. 41 The bulk 2H-WSe 2 is then exfoliated following the already-optimized parameters (at two frequencies of 37 and 80 kHz). The stability of the as-prepared dispersions was previously checked in different solvents. The homogeneity of the distributions was monitored for a specific period (starting from 1 day till 60 days) for the two sets of frequencies. In our gas-sensing application, 59 kHz is chosen as the optimum working frequency for exfoliating these nanosheets from the bulk 2H-WSe 2 . By modulating the total number of droplets on the surface of the gas-sensing device, we report the influence and interaction of 2H-WSe 2 droplets toward the adsorbed gases on the surface, including selectivity and sensitivity toward NO 2 (in addition to NH 3 and H 2 S). We thoroughly investigate the effect of the number of droplets (leading to several layers of 2H-WSe 2 with different thicknesses) on the sensing characteristics, including gas response, long-term stability, and selectivity. We believe that this work will be of great interest in understanding and fabricating a low-power consumption NO 2 sensor for its future miniaturization.

■ RESULTS AND DISCUSSION
A comparison of the X-ray diffractogram analysis of exfoliated tungsten di-selenide in an Argon atmosphere synthesized at 1 h reaction time in the air (at 300°C) is shown in Figure 1a. The results indicate that the pristine material synthesized according to our previous synthetic protocol is a phase pure material with a hexagonal crystal system of space group , respectively (which is in complete correlation with our previous work on 2H-WSe 2 synthesis). Different oxidation parameters varied to understand the thermal stability of the pristine 2H-WSe 2 . The XRD spectra were recorded to understand the extent of oxidation and the air-stability of the bulk as-prepared 2H-WSe 2 . Previous reports on monolayer WSe 2 (synthesized by a bottom-up approach 42,43 ) have indicated that the single-layer 2H-WSe 2 is highly prone to oxidation because of air-induced protrusions at its edges. However, the oxidation with air exposure occurs mostly at step edges after the exposure of 2H-WSe 2 in air for 9 weeks, while it is not detected on the internal terraces; therefore, the oxidation of the step edges in WSe 2 appears to be a self-terminating process.
Our studies, on the other hand, are involved as a combination of bottom-up bulk synthesis of 2H-WSe 2 with a top-down exfoliation method in ethanol and isopropanol as solvents. First, we investigated the effects of the ambient conditions toward our bulk material (i.e., the 2H-WSe 2 synthesized on the W foil). Although no in situ reduction was used for our synthesis, the bulk material when exposed to air at room temperature (RT) showed no visible or morphological changes. The bulk 2H-WSe 2 after 4 months of being kept in the ambient conditions still remains the same without any changes in its phase, indicating that our process is very suitable and prevents ambient oxidation of the material. In order to understand the effects of oxidation, we varied two reaction parameters, namely the temperature and the total oxidation time. For a variation in the reaction temperature, the oxidation of the pristine 2H-WSe 2 is started from RT till 300°C . Each of these oxidations at different temperatures was carried out for 12 h (maximum time limit for each oxidation temperature). As evident from Figure 2, the bulk material is intact and the incorporation of oxygen is not recorded in the qualitative energy-dispersive X-ray spectrometry (EDXS) spectrums for the material oxidized at 50, 100, 150, and 200°C . However, at 300°C, we see an almost complete oxidation of the 2H-WSe 2 into WO 3 (after 1 h oxidation of the bulk). For a variation in the total reaction time, we kept the oxidation temperature fixed at 200°C. This is because the gas-sensing measurements were carried out starting from RT till 150°C. Figure 1a depicts the change in the XRD peak intensities starting from 1 h. At 200°C, 1 h, the material is still in the pristine 2H-WSe 2 form and the peaks can be easily correlated to the ICDD no.: 00-038-1388. The anisotropic line broadening of 2H-WSe 2 is also evident till 5 h of the total reaction time. But around 5 h of the oxidation, we see the formation of a tiny hump around 58.59°, which is the clear indicator of the formation of the (332) plane of monoclinic WO 3 . At 8 and 12 h oxidation times, the (332) plane of WO 3 is very intense, suggesting a preference in growth toward that crystallographic plane. The other crystallographic planes can then be easily identified (for 5, 8, and 12 h) which are (022) at The material heat treated beyond 12 h of oxidation at 200°C in its qualitative EDXS measurements in the high-resolution scanning electron microscopy (HRSEM) shows the presence of O, W, and Se, indicating that some WSe 2 might still be unconverted. The morphology of the monoclinic WO 3 heated at 300°C for 12 h shows that the flake-like morphology is still maintained but it consists of tiny particulates growing as aggregates on the surface of the bulk. At temperatures of 300°C, there is partial incorporation of oxygen into the system when being annealed in air at reaction times greater than 12 h. Still, the stability of the pristine material is well maintained both at RT and till 150°C . Figure 1(a_1−a_3) shows the changes associated with the gradual evolution in the morphology, where the flake-like or sheet-like structure is maintained for 2H-WSe 2 . Transmission electron microscopy (TEM) measurements helped to identify the morphology and to confirm the structure of the assynthesized pristine 2H-WSe 2 . The bright-field TEM images help us estimating the average sizes of the 2H-WSe 2 sheets (ranging between 50 and 80 nm), which is consistent with our previous reports. The lattice fringes, as shown in Figure 1h, is calculated to be around 0.284 nm [for the (100) plane] and 0.325 nm [for the (004) plane] corresponding to the ICDD database no.: 00-038-1388 for 2H-WSe 2 . High-resolution TEM (HRTEM) measurements of the WSe 2 nanosheets helped us to identify honeycomb-like lattices, which are very typical of the 2H-variant of WSe 2 . Corresponding fast Fourier transforms (FFTs) of the three selected areas highlighted in red (marked as 1, 2, and 3) in the flakes also confirm the crystal structure of our pristine 2H-WSe 2 . The EDXS measurements obtained from TEM/HRTEM, as well as from HRSEM imaging, show a repeated deviation in the stoichiometry of 1:2 (W/Se), indicating that the material probably contains Se vacancies. Also, a long-running study on the air-annealing of 2H-WSe 2 at temperatures starting from 50°C till 200°C (at 1 h) shows little to almost no incorporation of oxygen into the system, as shown in Figure 2. However, as indicated above, the heat treatment of 2H-WSe 2 at 300°C evidently leads to the formation of monoclinic WO 3 which is also ascertained by the qualitative EDXS spectra, as shown in Figure 2w.
The exfoliation parameters of the pristine 2H-WSe 2 also varied according to our previous report. 41 Since both ethanol and isopropanol proved to be ideal low-boiling solvents for our exfoliation experiments, further analyses in ethanol dispersions were carried out. The main idea was to find out the optimal stability of the dispersions obtained after exfoliations because it would help in subsequent simple integration of the 2H-WSe 2 flakes into the sensor-based system. The ethanol-based dispersions were prepared by the same method as described before (1 h sonication at 80 and 37 kHz, respectively). The stability and homogeneity of these ethanol-based dispersions were checked for extended periods (starting from day 1 till day 60), as shown in Figure 3a  WSe 2 was observed. We can see that the W foil used during the exfoliation optimization procedure (which came into focus once the 2H-WSe 2 flakes started to settle down in the dispersion). Atomic force microscopy (AFM) image in Figure  3g, along with the height profile, shows the clear presence of layers or sheets of WSe 2 after drop-casting on the Si/SiO 2 substrate. 41 These observations indicate that a lower sonication frequency (at 100% power output) coupled with a higher sonication time is another reliable method to obtain fewlayered flakes of 2H-WSe 2 without any severe changes in its final morphology. The stability of the dispersions of 2H-WSe 2 prepared at 37 kHz and 80 kHz frequencies was also compared after letting them sit for 60 days to understand the extent of dispersibility in either case. After 60 days, we observe that the 80 kHz dispersion is still homogeneous, while the 37 kHz dispersion has started to settle down, as shown in Supporting Information ( Figure S1a,b). The optimization of frequency and output power is, therefore, crucial in our experiments because it is a significant indicator of the resulting stability and the final morphology of the flakes. The X-ray photoelectron spectroscopy (XPS) measurements of the exfoliated 2H-WSe 2 at 80 and 37 kHz (1 h sonication) are also compared. We observe the presence of only W and Se in the Supporting Information ( Figure S2a− d), where W can be resolved into W 4f 7/2 and W 4f 5/2 at 32.5 and 34.63 eV and the doublet peaks at 54.5 and 55.3 eV corresponding to selenium in the freshly exfoliated material.
The effect of modulating the number of droplets on the number of layers is also investigated. Figure 4 shows us an evident variation in the total number of layers concerning the number of droplets. We drop-casted the ethanol-based solution of 2H-WSe 2 on the Si/SiO 2 substrate to observe whether we obtain a lesser number of layers for lesser droplets and more number of layers for a greater number of droplets. Starting from two droplets, we obtained two layers of 2H-WSe 2 (∼1.7 ± 0.3 nm), which is very consistent with our previous report. 41 Increasing the number of droplets to 8 and 12 gave us consecutive height profiles of ∼5.5 ± 0.3 nm (indicating ∼6L) and ∼11 ± 0.3 nm (indicating ∼13L), respectively. We also performed a similar measurement for exfoliation and dropcasting of 2H-WSe 2 at 37 kHz, which provided us a similar variation. For 1 h sonication of 2H-WSe 2 in ethanol at 37 kHz showed 14.0 nm (indicating ∼16L) for only two droplets of the drop-casted 2H-WSe 2 . This supported the fact that at lower frequencies and lower exfoliation times, the layers of 2H-WSe 2 are still quite agglomerated. Higher sonication times of ∼8 h provided the corresponding values [∼1.003 ± 0.3 nm for 2 droplets (indicating ∼2L), ∼4.0 ± 0.3 nm for 8 droplets (indicating ∼5L), and ∼15 ± 0.3 nm for 12 droplets (indicating ∼17L) for 2H-WSe 2 (as shown in Supporting Information, Figure S3)]. So, we can assume that at the workable frequency of 59 kHz and ∼8 h exfoliation led us to approximately 2L for 2 droplets, ∼5−6L for 8 droplets, and ∼13−17L for 12 droplets in our gas-sensing experiments. This helps us assume that a variation in the number of droplets will lead to an almost controllable variation in the total number of WSe 2 layers drop-casted onto the substrate. Such a variation in the resulting gas-sensing measurements is discussed further in the following sections.
Gas-Sensing Performance. The sensing mechanism of the 2H-WSe 2 -based conductometric sensors can be explained by the surface reaction-induced charge transfer through gas molecule adsorption. This induced charge transfer leads to a change in the sensor's conductance. The charge transfer characteristics is mainly controlled by factors such as donor/ acceptor characteristics of gas molecules regarding their oxidizing or reducing property and the inherent conductivity type of the semiconducting material (n-or p-type) used for such studies. The sensing mechanism of p-type WSe 2  semiconductors under NO 2 exposure is described in Figure 5. This mechanism explicitly consists of two consecutive surface reactions, starting with oxygen adsorption coming from ambient air and followed by NO 2 molecules adsorption. Adsorption of oxygen molecules usually takes place on the exposed surface. Oxygen takes electrons from the valence band of WSe 2 , creating a hole accumulation region (HARpositive charge carriers), as represented by eq 1 and described by Figure 5b. 44,45 We should note that the sensors are tested at working temperatures (WT) below 200°C, ranging from RT to 150°C. Therefore, the O 2 that was absorbed was similar to a molecular species (O 2 − ). 44 The injection of the NO 2 modifies the HAR due to combined effects such as electron affinity of NO 2 , physisorption/chemisorption rate, and 2H-WSe 2 defect sites, mostly Se vacancies in particular. As an oxidizing gas with high electronic affinity, NO 2 is an electronacceptor that could capture more electrons from the 2H-WSe 2 surface (eq 2 and Figure 5a) as well as interact with preadsorbed oxygen to form NO 3 − (eq 3 and Figure 5c). 45 In both cases, as the NO 2 adsorption process continues, the holes' density increases and HAR becomes more concentrated in holes, thereby increasing electrical conductance. In TMDCs, especially in transition-metal di-selenides, defects such as Se vacancies (usually denoted as V Se ) have a critical fundamental rule in the sensing mechanism toward gas molecule species contributing to carrier charge transfer. It is confirmed from previous reports that the adsorption of NO 2 on the 2H-WSe 2 surface is considered to be unfavorable without the existence of chalcogen vacancies (V Se ). 46 In our case, the presence of Se vacancies is supported by electron paramagnetic resonance (EPR), as shown in Supporting Information ( Figure S4). Exfoliated 2H-WSe 2 exhibits an S-shape signal at ∼337.5 mT (g = 2.00), indicating the presence of Se vacancies. The stronger S-shaped signal at 7 K is simply correlated with the higher concentration of Se vacancies. The existence of such vacancies is typical for semiconducting materials, and they are usually paramagnetic in nature. 47,48 Therefore, V se behaves as active sites and provides a high chemical activity to the adsorbed molecules such as NO 2 , which can tune further HAR, impacting the electrical conductance and the final response. 49 (2) As already explained before, tungsten di-selenide is not thermally stable at temperatures beyond 200°C and longer oxidation times (>12 h) due to the ambient oxidation of 2H-WSe 2 . However, all sensors are tested at lower temperatures, which is advantageous for developing low-power consumption devices. Figure 6a,b exhibits the dynamic response toward NO 2 and the electrical conductance of WSe 2 (2D), WSe 2 (8D), and WSe 2 (12D) at working temperature (WT) ranging from RT to 150°C. The three as-fabricated sensors' conductance increases with the WT, which confirms the semiconducting nature of the synthesized 2H-WSe 2 nanosheets. After NO 2 exposure, the sensors' electrical conductance increased up to a maximum value. After switching off NO 2 , the sensors' electrical conductance return to the initial value as airflow is restored regardless of the thickness (or the number of layers) formed by drop-casting. This behavior is consistent with the sensing mechanism of p-type semiconductors toward oxidizing gases. The electron-acceptor property of NO 2 upon interacting with p-type semiconductor causes electrons extraction from the valance band, enhancing the holes carrier concentration and leading to high electrical conductance, as explained in the previous section. There is incomplete recovery of WSe 2 sensors at RT. Indeed, after increasing the temperature from 50 to 150°C , the recovery rate of the sensors becomes faster. The conductance recovers the baseline in response to the thermal energy that increases the reaction's kinetics (desorption/ adsorption rate), which allows full desorption of the NO 2 molecules, and in turn, drastically improves the recovery of these sensors. 50 The role of the total number of droplets (number and the total thickness of the layers) used while conducting experiments on sensing properties are investigated. Figure 7a shows the dynamic responses of the different 2H-WSe 2 layers toward 10 ppm of NO 2 at RT, while Figure 7b shows their responses. The responses increased as the WT decreased from 150°C to RT. The response of the bilayer 2H-WSe 2 (2D) sensor upon 10 ppm of NO 2 was 361, 114, 113, and 97.9% when it was tested at RT, 50, 100, and 150°C, respectively. The high response obtained at RT is attributed to possible selenium vacancies (as reported previously for many TMDCs and also 2H-WSe 2 ). This happens because, on a broader note, the defect sites created in metal di-chalcogenide, such as in the case of di-sulfides or di-selenides (S or Se), participate primarily in gas molecule adsorption reactions. Previous density functional theory studies confirmed that the adsorption of N 2 molecules on MoS 2 is negligible in the absence of disulfide vacancies (VS 2 ). 51 Thus, these defects are essential at low temperatures as N 2 molecules deplete the material carrier charge through the VS 2 and change the electronic conduction. The three sensors [WSe 2 (2D), WSe 2 (8D), and WSe 2 (12D)] show the highest responses at RT. Especially, a bilayered 2H-WSe 2 -based sensor [WSe 2 (2D)], which has a thickness of around 1.7 nm (thereby indicating two flakes or two sheets), shows the highest response compared to the sensors containing a greater number of layers (5.5 nm for 8 droplets and 11 nm for 12 droplets). Therefore, it can be inferred that the electronic properties in 2H-WSe 2 are completely thickness dependent. Decreasing the overall thickness leads to the higher surface to volume ratio, which  increases the gas molecule adsorption leading to a higher response. Indeed, the high surface to volume ratio of the WSe 2 (2D) sample is due to the bilayer formed by drop-casting two droplets.
In contrast, 8 and 12 droplets probably lead to a bulk-like formation (more than eight layers) with a low surface to volume ratio. 52−54 Such observations were reported previously by He et al., who investigated the effect of the thickness of MoS 2 on the NO 2 response in a flexible transistor, showing a sensitivity improvement toward NO 2 when the thickness decreased from 18 to 2 nm due to the increase in surface area at low thickness. 55 Late et al. reported synthesizing different MoS 2 layers (L) ranging from 3L to 5L, corresponding to the thickness ranging from 1.4 to 3.3 nm. The high response toward both reducing and oxidizing gases such as NO 2 and NH 3 was achieved for 5L MoS 2 ; 53 therefore, a direct link connecting the thickness (number of the layers) of the twodimensional material and sensing mechanism is still under consideration and needs further investigation and development. The dissimilarity of thickness/response dependency is attributed to two factors that can affect sensor functionality: (1) without taking into account the thickness of the multilayers constructed, increasing the layers is prominent for a high response since the creation of more interspaces provides higher gas molecule intercalation between the layers leading to high sensitivity as the gas is likely to be intercalated in the layer− layer interface 56 and (2) single or bilayer promotes high surface to volume ratio, which increases the adsorption rate resulting in a higher response. However, if we rely on the second factor in the case of bilayer 2H-WSe 2 , even the interlayer spacing is considered as a subfactor for sensor functionality. 57 In addition to the above investigations, the sensing performance of 2H-WSe 2 is also studied toward other gases such as NH 3 and H 2 S. These gases are also of particular interest in our case because of the many devastating effects they pose on human health and the environment. Figure S5a− c (Supporting Information) shows the dynamic response of the 2H-WSe 2 (2D) sensor toward NO 2 (2−10 ppm), H 2 S (5−25 ppm), and NH 3 (5−25 ppm), respectively. The dynamic responses of 2H-WSe 2 (8D) and 2H-WSe 2 (12D) toward NH 3 and H 2 S are given in the Supporting Information (Figures S6  and S7). The best WT for high detection of NH 3 is 150°C, whereas 100°C is optimal for H 2 S, and, as already discussed, RT for NO 2 . It is worth noting that even the WT of 150°C satisfies the low-power consumption of 2H-WSe 2 nanosheetbased sensors. As shown in the dynamic response, the electrical conductance decreases on NH 3 and H 2 S exposure. This behavior is primarily due to the electron-donor nature of reducing gases such as NH 3 and H 2 S. During the interaction of these gases with the p-type material, the reducing compounds provide or reinject the electrons into materials instead of extracting the latter as with NO 2 . The injected electrons recombine with holes, reducing the density of hole carriers and decreasing the electrical conductance of the p-type material. 58,59 The electron transfer induced by reducing gases oppositely affects the hole carrier concentration compared to oxidizing gases. Figure 8a shows the response against NO 2 concentrations. The response upon NO 2 exposure increased stepwise from 250 to 361% when the concentration was increased from 2 to 6 ppm, reaching the maximum response at the critical concentration (8 ppm) and creating an enhanced sensor surface interaction. Followed by increasing the NO 2 concentration at 10 ppm, the surface was fully covered with NO 2 molecules, which excludes new NO 2 molecules' possibility of reacting with the surface due to the saturation level. 60 Therefore, the response remained stable (unchangeable) upon exposure to 10 ppm. The response saturation is also investigated in many reports. This usually happens when the sensor is operating at low temperature (as per our RT investigations), and it is attributed to the molecular adsorption, which sometimes leads to a poisoning effect in the sensor if no kinetic energy (U.V. or heat) is applied to overcome the issue of the gas molecules detaching from the surface. 61,62 The high response toward 2 ppm and the response stability at/from 6 ppm of NO 2 show the ability to detect NO 2 at lower concentrations (even at the ppb level). As shown in Supporting Information (Figure S8), the bilayer 2H-WSe 2 can sense even 200 ppb (0.2 ppm) of NO 2 gas. Ideally, 200 ppb is considered as the experimental detection limit as well as the threshold exposure limit of NO 2 recommended by American Conference of Governmental Industrial Hygienists. 4,63 In fact, an ideal NO 2 sensor should have a practical limit of detection (LOD) equal to or lower than this value. Our results hold a high significance in many medical applications, such as utilizing chemical gas sensors in breath analyzers of the human exhaled breath where NO 2 is considered as a biomarker for early diagnosis of asthma prediction and its associated risks. In the case of asthma detection from human exhaled breath, monitoring the exhaled breath requires detecting an extremely low concentration of NO 2 in the presence of a high relative humidity (RH) level (higher than 70%), which makes our  The response upon H 2 S was increased from 10 to 33% when the concentration was increased from 5 to 25 ppm and increased from 17 to 49% upon NH 3 exposure maintaining the same NH3 concentration range as H 2 S (Figure 8b,c). Moreover, the LOD is also calculated. To estimate the LOD of the sensor, the experimental data are fitted using the power law, as indicated in Supporting Information (Figure S9a−c). LODs toward NO 2 , H 2 S, and NH 3 are about 100, 4829, and 1561 ppb, respectively.
A comparison of the obtained sensing performance toward some studies reported in the literature on the 2H-WSe 2 -based NO 2 sensor is summarized in Table 1.
To investigate the selectivity of the 2H-WSe 2 sensor, we tested the bilayer 2H-WSe 2 (2D) toward fixed concentrations of different gases such as H 2 , NH 3 , acetone, and H 2 S. The selectivity was studied at three temperatures: 150, 100°C, and RT, as shown in Figure 9. The 2H-WSe 2 (2D) showed a full selectivity toward NO 2 at RT. Indeed, 2H-WSe 2 is entirely insensitive to the other gases. By heating the sensor at 100°C and even at 150°C, the response was slightly increased toward other gases, especially NH 3 and H 2 S. Despite this, the 2H-WSe 2 sensor remains selective to NO 2 . Moreover, the longterm stability of 2H-WSe 2 (2D) is also investigated, as shown in Supporting Information ( Figure S10). The sensor offers good stability toward low concentration of NO 2 even after 9 months. It is worth noting that many measurements toward several gases at different temperatures are performed as a longterm study, all of which show the high stability of 2H-WSe 2 in detecting low NO 2 concentration (2 ppm). Besides, the estimated response and recovery times of bilayer 2H-WSe 2 (2D) toward NO 2 concentrations (2, 4, 6, 8, and 10 ppm) are calculated and illustrated in Supporting Information ( Figure  S11). The sensor also shows a fast response and fast recovery time (220 and 1700 s, respectively) at RT, based on the fact that the response and recovery times of the sensor are limited only to the filling time of the test chamber (approximately 4−5 min to fill the 1L test chamber). Finally, the effect of humidity on the NO 2 (2 ppm) response is also studied by changing the RH from 20 to 90% at RT ( Figure 10). As can be seen, the sensor is sensitive even at a high humidity level (90%), which also proves the sensors' suitability for diagnostics in exhaled breath analysis, where real-time applications often require high humidity near the normal exhaled breath (80% to 90%).

■ CONCLUSIONS
The scalable synthesis and profoundly simple approach of integrating 2H-WSe 2 nanosheets (prepared as per an AP CVD

■ EXPERIMENTAL SECTION
Synthesis of 2H-WSe 2 . The synthetic procedure is the same as per our previous reports of growing bulk 2H-WSe 2 on W foil using elemental Se (Alfa Aesar, 99.999% purity) and W foil (Alfa Aesar, 99.999% purity) in an AP-CVD-assisted process. 41 We prepared the pristine W foil after etching it in dilute hydrochloric acid (0.01 M) and sonicating it in isopropanol (for removing the native oxide layer on the exposed W foil's surface). The W foil was introduced into the furnace (F2) at 900°C. Then, the elemental Se was introduced into the furnace F1 at 650°C in order to initiate the WSe 2 formation in the W foil surface. All experiments were performed in an Ar atmosphere in the absence of hydrogen (for eliminating the possibility of H 2 Se formation). The bulk 2H-WSe 2 formed on the W foil surface was characterized and then was exfoliated into nanosheets following exfoliation in ethanol as a solvent. This 2H-WSe 2 dispersion was utilized for the gas-sensing measurements (by modulating the total number of drops of this dispersion). The air-annealing of 2H-WSe 2 is performed from 50 to 200°C to observe microstructural changes and changes in the original phase (which determines the stability of the material in ambient and elevated temperatures in the absence of an inert atmosphere). The exfoliation parameters for obtaining the 2H-WSe 2 powder from the bulk 2H-WSe 2 grown on W foil were optimized as per the previous report by our group and checked for further stability in the resulting dispersions. The exfoliation times are checked thoroughly to obtain the best quality flakes [Supporting Information (Figures S12 and S13)], and the stability and homogeneity of resulting dispersions were also checked. The difference in the layers of 2H-WSe 2 obtained from their corresponding liquid dispersions was investigated by following up the exfoliation time, and the number of bulk 2H-WSe 2 synthesized on W foil (till 8 h). While preparing the dispersions during prolonged sonication of 8 h at 37 kHz, 100 W, and 80 kHz, 100 W, we made sure to replace the water in the sonication bath [using an ultrasonic bath (Elmasonic P)] every 1 h to prevent the rise in temperature during this extended sonication time.
Characterization. X-ray diffraction measurements of bulk tungsten di-selenide (WSe 2 ) on tungsten (W) foil is performed using a Bruker D8 advanced diffractometer. We analyzed a 2 cm × 2 cm square sample of bulk tungsten di-selenide (WSe 2 ). The data were collected in the 2θ range from 10 to 80°, with a step size of 0.002°a nd scanning rate of 0.4°/min. The X-ray generator is operated at 40 kV and 30 mA with Cu Kα radiation (λ = 1.542 Å). The bulk tungsten di-selenide (WSe 2 ) on tungsten (W) foil is further analyzed by X'pert HighScore Plus software to ensure that the product obtained is from a highly repeatable procedure as well as to confirm the effects of air annealing on the material starting from 50°C to 200°C . Similar HRSEM of the bulk material and air-annealed material is performed using the instrument Magellan 400 FEI. Additional analyses with AFM measurements were performed. This measurement was carried out in a Bio Fast Scan scanning probe microscope (Bruker A.X.S.). For image processing, Gwyddion software is used. The "flatting" and "planefit" functions are also directly applied to each image. Further HRTEM analysis of the exfoliated WSe 2 sample was carried out in a JEM 2100, JEOL (operating at 200 keV). The TEM images are analyzed using Gatan digital Micrograph. XPS measurements are performed using the Thermo Scientific Nexsa spectrometer. The samples were irradiated with a soft X-ray source (∼1.5 KeV) under ultrahigh vacuum conditions (∼10 −10 to 10 −9 torr), and their   The conductometric sensor design is described in Figure 11. 69,70 The design comprises an alumina substrate acting as a support for the tungsten di-selenide sensing material, Pt electrodes to measure the electrical conductance, and a Pt heater to induce temperature for gas− surface reaction pads to provide perfect adhesion between the different sensor elements and the substrate. First, polycrystalline alumina (Al 2 O 3 ) substrates (2 mm × 2 mm, 99.9% purity, Kyocera, Japan) having a thickness of 250 μm are ultrasonically cleaned for 20 min acetone, rinsed by distilled water, and then dried in air. TiW/Pt Pad deposition was performed using DC-magnetron sputtering (DC-MS) (3 min, 300°C, pressure 5 × 10 −3 mbar, and 75 W Ar plasma). The interdigitated Pt electrodes (IPE) and Pt heater were deposited on top and backside of Al 2 O 3 , respectively, utilizing DC-MS at the same conditions. The deposition time was 20 min. The final transducer was mounted on a transistor outline package using electro-soldered gold wires.
After synthesizing bulk 2H-WSe 2 using AP-CVD on the tungsten (W) foil, the 2H-WSe 2 suspension was prepared by scratching and dispersing a fixed amount of material into 20 mL of ethanol. A properly dispersed 2H-WSe 2 solution was obtained by ultrasonication of the solution for 8 h (with 1 h break in between) at maximum power and a high frequency of 59 kHz at RT. The dispersion containing 2H-WSe 2 nanosheets was drop-cast on the top of IPE to fabricate the conductometric device, as shown in Figure 11. Three 2H-WSe 2 sensor batches with different droplet numbers of 2H-WSe 2 (corresponding to a different number of layers) were fabricated herein and labelled as WSe 2 (2D), WSe 2 (8D), and WSe 2 (12D) corresponding to two, eight, and twelve droplets, respectively. All sensors are stabilized by heating them at 150°C using power supply for 3 days (72 h) after device production and before the gas testing. It is important to note that after each drop, devices were kept for 20 min in the hood for the complete evaporation of the solvent. The volume of a drop of the liquid dispersion is equal (approximately) 0.045 cm 3 (as explained in Supporting Information Section, Page S2).
Gas-Sensing Measurements. For functional characterization in the presence of target chemical molecules, the sensors were placed inside a stainless-steel chamber. This is because our system can host up to eight sensors simultaneously for performing the electrical measurements. The required analyte or gas concentration is obtained by mixing synthetic air with the correct amount of gas from a certified bottle (S.O.L., Italy). The total flow inside the chamber is fixed at 200 sccm. The electrical conductance of the sensors is measured continuously by applying 1 V and recording the generated current using a picoammeter (Keithley 6485). The electrical behavior of sensors is investigated toward different chemical compounds, including reducing (NH 3 , H 2 S, acetone, and H 2 ) and oxidizing (NO 2 ) ones. Each gas is injected for 30 min, and then a synthetic airflow is restored for 1 h. The RH is fixed at 20% at 20°C, and different WTs were also additionally investigated, ranging from RT to 150°C. 2H-WSe 2 nanosheets have p-type semiconducting nature; therefore, the gas response was calculated using the formulae where G air and G gas are the electrical conductances of 2H-WSe 2 in the air (conductance baseline) and gas, respectively.