MOCVD of Hierarchical C‐MoS2 Nanobranches for ppt‐Level NO2 Detection

In the past decades, toxic gas emissions have increased significantly owing to the rapid growth of industry and road transportation. Therefore, monitoring major pollutants, such as NO2, is crucial to protecting human health. The 2D materials that contain numerous adsorption sites and exhibit ultrahigh chemical reactivity can be used as sensor materials to detect these toxic gases. Herein, highly uniform, large‐area carbon‐incorporating hierarchical MoS2 nanobranches are synthesized by metal–organic chemical vapor deposition (MOCVD). An in situ carbon‐incorporation method that uses the carbon impurity present in the precursor as the seed during the MOCVD process is employed to form a hierarchical structure containing abundant adsorption sites. A gas sensor based on the resulting C‐MoS2 nanobranches contains many edge sites exhibits high adsorption energy, and consequently, has high NO2 gas sensitivity. Hence, this hierarchical C‐MoS2 gas sensor shows excellent sensing properties, exhibiting a device response of 1.67 at an extremely low NO2 concentration (≈5 ppb). The limit of detection of the gas sensor for NO2 is calculated to be low (≈1.58 ppt), further confirming its exceptional performance. Thus, the hierarchical C‐MoS2 nanobranches deposited herein provide novel insights regarding the properties of 2D materials and are highly suited for fabricating high‐performance NO2 sensors.

In the past decades, toxic gas emissions have increased significantly owing to the rapid growth of industry and road transportation. Therefore, monitoring major pollutants, such as NO 2 , is crucial to protecting human health. The 2D materials that contain numerous adsorption sites and exhibit ultrahigh chemical reactivity can be used as sensor materials to detect these toxic gases. Herein, highly uniform, large-area carbon-incorporating hierarchical MoS 2 nanobranches are synthesized by metal-organic chemical vapor deposition (MOCVD). An in situ carbon-incorporation method that uses the carbon impurity present in the precursor as the seed during the MOCVD process is employed to form a hierarchical structure containing abundant adsorption sites. A gas sensor based on the resulting C-MoS 2 nanobranches contains many edge sites exhibits high adsorption energy, and consequently, has high NO 2 gas sensitivity. Hence, this hierarchical C-MoS 2 gas sensor shows excellent sensing properties, exhibiting a device response of 1.67 at an extremely low NO 2 concentration (%5 ppb). The limit of detection of the gas sensor for NO 2 is calculated to be low (%1.58 ppt), further confirming its exceptional performance. Thus, the hierarchical C-MoS 2 nanobranches deposited herein provide novel insights regarding the properties of 2D materials and are highly suited for fabricating high-performance NO 2 sensors.
field-effect transistor sensors are drawing significant attention for sensing gases owing to their simple manufacturing process and low cost. Semiconductor gas sensors detect molecules based on the direction and degree of electron movement by adsorbing the molecules to detect onto their surface when these molecules are present in the air. Many researchers have described the gassensing behaviors of micro-and nanostructured metal-oxide semiconductors, and most commercially available gas sensors are based on metal oxides. [10][11][12] Nevertheless, the operating temperatures of these gas sensors are high. Hence, their use in dangerous environments is limited. Moreover, their sensitivity and stability are degraded owing to the changes in the microstructure at high temperatures. [13,14] The 2D transition-metal dichalcogenides (TMDCs) are considered promising alternatives to conventional metal-oxide-based sensing materials. [15] Among TMDCs, MoS 2 is a prospective material for the development of high-sensitivity gas sensors because of its excellent electronic and chemical properties, inherently high surface-to-volume ratio, and strong affinity for various molecules. [16][17][18] The edge sites of MoS 2 , which are rich in unsaturated bonds, exhibit much higher chemical catalytic activity than the basal plane, thus making it suitable for electrochemicalsensing applications that require a high degree of molecular adsorption. [19][20][21] Cho et al. [22] fabricated vertically aligned MoS 2 nanosheets and could achieve a NO 2 gas sensitivity that was five times higher than that of horizontally aligned MoS 2 nanosheets. Islam et al. [23] integrated vertically aligned 2D MoS 2 layers onto a stretchable substrate with "serpentine" patterns for NO 2 gas sensing. The sensor could reliably detect NO 2 gas in concentrations of 5-30 ppm with a sensitivity of 160%-380%. These edge-rich MoS 2 -based gas sensors exhibit a higher sensitivity than sensors based on 2D MoS 2 films and do not require sensitivity enhancement treatment. However, they cannot detect NO 2 at the ppb level. Therefore, novel strategies must be devised to decrease their lower detection limit.
Herein, we propose a method for synthesizing hierarchical C-MoS 2 nanobranches via an in situ carbon-doping method that uses the precursor impurity as the seed during metal-organic chemical vapor deposition (MOCVD). The growth mechanism of the C-MoS 2 nanobranches was analyzed using X-ray photoelectron spectroscopy (XPS) and transmission electron microscopy (TEM). Based on their edge-rich structure, the C-MoS 2 nanobranches were used for gas sensing at room temperature. The resulting gas sensor exhibited excellent sensitivity with respect to NO 2 at the ppb level, and its properties met the requirements for NO 2 detection. Density-functional theory (DFT) calculations showed that the carbon in the precursor impurity not only acts as the seed for the vertical growth of the nanobranches but also increases the adsorption energy for the target gas molecules. The sensor could detect NO 2 in a concentration of 5 ppb with a response of 1.67 (R 0 /R gas ). The limit of detection (LOD) of the sensor, calculated as per the IUPAC procedure, was %1.58 parts per trillion (ppt), indicating that the hierarchical C-MoS 2 gas sensor is highly suited for ppt-level NO 2 detection.

Results and Discussion
Highly uniform C-MoS 2 films were grown using a cold-wall-type MOCVD reactor for different synthesis times. A substrate temperature of 250°C was maintained during the growth process to grow the hierarchical C-MoS 2 nanobranches. Figure 1a shows a schematic of the MOCVD growth process, which used Mo(CO) 6 as the precursor and H 2 S as the reactant gas. The flow rates of the  H 2 S and carrier (Ar) gases were controlled using a mass flow controller. A showerhead-type reactor connected to a gas line was used to ensure uniform flow. Figures S1 (Supporting Information) and 1b-d show the scanning electron microscopy (SEM) images of different C-MoS 2 films grown under the same conditions but for different growth times. After 2 min of growth, the bottom layer of the film sample was formed, and partially vertical MoS 2 could be observed, indicating that the growth of the film had begun ( Figure 1b). As the growth time increased to 40 min, the size and height of the petal-like vertical C-MoS 2 nanostructure increased to approximately 150 nm (Figure 1c). After 240 min of growth, the height of the vertical C-MoS 2 film increased to approximately 900 nm (Figure 1d), and nanobranches were formed on the petal-like vertical C-MoS 2 film. Figure 1e-h schematically illustrates the growth mechanism of the hierarchical C-MoS 2 nanobranches. During the initial stage of the synthesis process, the MoS 2 bottom layer grows in a layer-by-layer manner until a certain critical thickness is reached, and the carbon in the Mo(CO) 6 precursor is partially incorporated into the MoS 2 film under the optimized partial pressure conditions (Figure 1e). The decomposition rate of the Mo(CO) 6 precursor depends on the substrate material; [24] hence, in the area where the carbon is incorporated, the decomposition rate of Mo(CO) 6 is faster than that of the pristine MoS 2 film. These differences in the decomposition rate of the precursor and the growth rate of the film result in the growth of the petal-like vertical MoS 2 film (Figure 1f ). With an increase in the growth time (Figure 1f ), elemental carbon is incorporated into the vertical MoS 2 petals (Figure 1g). Consequently, branch-like hierarchical C-MoS 2 films are formed (Figure 1h). Figure S2 (Supporting Information) shows cross-sectional SEM images of the samples measuring 150 and 900 nm in height. Figure S2a, Supporting Information, shows the state in which only the vertical C-MoS 2 film grows; this is in keeping with the process shown in Figure 1g. Moreover, the structure with a large number of nanobranches shown in Figure S2b, Supporting Information, is consistent with the process shown in Figure 1h.
To further elucidate the growth mechanism of the fabricated C-MoS 2 nanobranches, we characterized the samples grown for different growth times using Raman spectroscopy, photoluminescence (PL) spectroscopy, XPS, and TEM.   corresponds to the starting point of the vertical growth process, exhibits peaks related to in-plane (E 1 2g , at 385.48 cm À1 ) and out-of-plane (A 1g , at 405.91 cm À1 ) vibration modes. The frequency spacing of these peaks is 20.43 cm À1 , which can be attributed to the monolayer and bilayer nature of the vertical MoS 2 film, as confirmed in the high-resolution TEM (HRTEM) image of the 10 nm sample shown in Figure S3 (Supporting Information). The C-MoS 2 samples with heights of 150 and 900 nm exhibited distinct peaks corresponding to in-plane (E 1 2g , at 383.20 cm À1 ) and out-of-plane (A 1g , at 408.17 cm À1 ) vibration modes. The intensity of the A 1g peak increased with an increase in the sample height because outof-plane vibrations occur preferentially in the case of vertical MoS 2 films. [25][26][27] Figure S4 (Supporting Information) shows the enlarged section of the spectra showing the low intensity. [28] The inset shows only a detail of the 2LA(M)/A 2u region and Si peak, but no distinctive peaks related to Mo 2 C could be observed at 661, 818, and 990 cm À1 . This may be because the carbon incorporated in our C-MoS 2 film is unlikely to exist as crystalline Mo 2 C. [29] Further optical analysis of the samples was performed using PL spectroscopy, and the results are shown in Figure 2b. In the case of the 10 nm sample, the PL peak was detected at 1.88 eV, while for the 150 nm sample, the PL peak was slightly blueshifted to 1.89 eV. The 900 nm sample showed a broad PL peak corresponding to A and B excitons. This peak, which was centered at 1.81 eV, was shifted by 80 meV compared with that of the 150 nm sample. As the surface structure and carbon content of the films changed, new emissions were detected, and the peaks broadened. Figure 2c shows the A 1g /E 1 2g peak intensity ratio and PL peak positions for seven samples with different heights. The A 1g /E 1 2g intensity ratio initially increased with an increase in the height of the C-MoS 2 sample and plateaued eventually. These results indicate that the PL peak position tends to blue-or redshift because of the formation of nanobranches. Figure 2d shows the atomic ratio of carbon as a function of the sample height. The atomic ratio of carbon increased until the formation of the nanobranches because of the carbon impurity, which promoted branch formation, and its incorporation into the vertical MoS 2 film. In the case of the 150 nm sample with nanobranches, the atomic ratio of carbon decreased with an increase in the growth time. This is because only the surface layer of the nanobranches, which had a low carbon content, could be analyzed, as the X-rays used for XPS have a penetration depth of less than 10 nm. From these results, we concluded that the carbon impurity acts as the seed for vertical growth and that the amount of carbon incorporated within MoS 2 in the nanobranches is lower than that in the vertical film. Figure 2e compares the XPS spectra of the samples with heights of 10 (top), 150 (middle), and 900 nm (bottom). The high-resolution core-level C 1s spectra can be deconvoluted into four peaks, that is, two large peaks at 284.2 and 284.6 eV, which are attributable to the C-Mo and C-C bonds, respectively, and two smaller peaks at 285.8 and 288.9 eV, which are attributable to the C-O-C and C = O bonds, respectively. The presence of the C-Mo peak indicates that some of the C atoms form chemical bonds with the Mo atoms. [30] The Mo 3d spectrum of the 10 nm sample could be deconvoluted into two peaks corresponding to the chemical states S-Mo-S (229.5/232.6 eV) and S-Mo-C (228.6/232.0 eV), which indicated that a large amount of C was substitutionally doped into the MoS 2 lattice. [31] The low-intensity peak at 233.4 eV, which corresponded to the Mo 5þ species, indicated that the C-MoS 2 may have been partially oxidized. [32,33] Furthermore, the S 2p peaks observed at 163.5 and 162.3 eV were attributable to the doublet of S 2p 1/2 and S 2p 3/2 , respectively. [34] The Mo 3d and S 2p peaks of the 150 and 900 nm samples were shifted to lower binding energies by 0.2 eV. The intensity of the C 1s peak was strong in the case of the 10 nm film, which corresponded to the initial growth stage. Moreover, the Mo 3d and S 2p peaks of the 10 nm sample were upshifted because the surface layer of this film had a higher dopant carbon content than that of other films. Table S1 (Supporting Information) shows the atomic compositions and S/Mo ratios of all the deposited film samples. Figure 2f,g shows cross-sectional HRTEM images of a C-MoS 2 nanobranch grown during the vertical growth of the 900 nm C-MoS 2 film. To verify that the carbon impurity promoted the nucleation and growth of the vertical MoS 2 film and its nanobranches, we obtained the intensity line profiles and measured the interlayer spacings across the dotted cyan line in Figure 2g ( Figure 2h). The measured interlayer distance in C-MoS 2 was 0.67 nm, which corresponds to the reported interlayer spacing of MoS 2 . [35,36] Because the carbide interlayer spacing was 0.28 nm, [37] the measured distance of 0.14 nm corresponds to the C-Mo bond, while the distance of 0.26 nm corresponds to the S-Mo bond. Figure S5b (Supporting Information) shows the intensity line profiles measured along the five lines marked in Figure S5a, Supporting Information. Figure 2i shows a schematic of the substitutional doping of C for S in the MoS 2 lattice for a layer formed on a nanobranch, on the basis of the aforementioned line profiles. The schematic is in accordance with the XPS results.
The C-MoS 2 films were grown in a layer-by-layer manner until a certain critical thickness was reached. Once this critical thickness had been reached, film growth continued vertically. To confirm this, we obtained cross-sectional TEM images of the samples with heights of 70 and 900 nm and compared the thicknesses of their bottom layers ( Figure S6, Supporting Information). Both the samples had approximately 10 nm thick bottom layers, despite the differences in the growth times. Figure 3a shows schematics of the C-MoS 2 gas sensor fabricated on the SiO 2 /Si substrate and charge transfer between a NO 2 molecule and the C-MoS 2 channel material. Most conventional MoS 2 films grown by CVD are n-type semiconductors. [38][39][40] However, the C-MoS 2 films synthesized in this study exhibited p-type characteristics. To discuss gassensing mechanism, reducing (or an electron donor) and oxidizing (or an electron acceptor) gas molecules should be considered. Examples of reducing gases are SO 2 , H 2 S, NH 3 , and CH 4 , and the examples of oxidizing gases are nitrogen oxides (NO X ), oxygen, and ozone. When MoS 2 is exposed to NO 2 gas, the NO 2 molecules, which are electron acceptors, readily remove electrons from MoS 2 and transform into NO 2 À (NO 2 þ e À ! NO 2 À ). [41,42] Thus, the NO 2 gas molecules quickly adsorb electrons, increasing the hole concentration in MoS 2 ; this results in the downshifting of its Fermi level to the valence band. [43,44] Consequently, the adsorption of NO 2 increases the degree of p-type doping of MoS 2 , which, in turn, reduces the resistance of the MoS 2 gas sensor. [45] To evaluate the sensing performances of the films with different heights, C-MoS 2 gas sensors based on these films were exposed to 200 ppb NO 2 gas. The sensor response was calculated using Equation (1) response where R gas is the resistance of the sensor in a target gas environment and R 0 is the initial resistance of the sensor. [46] Figure 3b shows that the response of the sensor based on the 150 nm sample was 1.36 and that the response increased to 3.86 (900 nm sample) once the hierarchical nanobranches were formed. This result suggests that the unique hierarchical C-MoS 2 nanobranch structures are highly sensitive to NO 2 gas. Figure S7 (Supporting Information) shows the initial resistivity of the sensors. We confirmed that the carbon ratio decreases with the growth time of C-MoS 2 , while the resistivity increases. The high response of the sensor to a low concentration of NO 2 gas indicates that the carbon impurity affects not only the growth mechanism of the hierarchical structure but also the adsorption energy of the gas molecules. To elucidate the effects of carbon incorporation on the gas-adsorption behavior of MoS 2 , we simulated the theoretical adsorption of NO 2 gas molecules on C-doped MoS 2 using DFT. Atomic schematics showing the configurations corresponding to the adsorption of NO 2 on the pristine MoS 2 and C-doped MoS 2 monolayers are shown in Figure 3c,d, respectively. The adsorption energy (E ads ) of the NO 2 gas molecules was estimated using Equation (2) where E MoS 2 þ NO 2 is the total energy of the NO 2 absorbed by the MoS 2 or C-doped MoS 2 monolayer, E MoS 2 is the energy of the MoS 2 monolayer, and E NO 2 is the energy of a NO 2 gas molecule. Initial distance between NO 2 gas molecule and the pristine MoS 2 monolayer, and C-doped MoS 2 monolayer was both approximately 1.8 Å ( Figure S8, Supporting Information). In the structure after complete geometric relaxation, the distance between the NO 2 gas molecule and the pristine MoS 2 monolayer was 3.34 Å, and the C-doped MoS 2 monolayer was 1.37 Å. The calculated adsorption energies of the NO 2 gas molecules on the pristine MoS 2 and C-doped MoS 2 monolayers were À0.14  and À1.77 eV, respectively. The value of E ads , which is negative for exothermic processes, was indicative of strong interactions. [47] Therefore, it can be concluded that NO 2 binds more strongly to the C-doped MoS 2 than to pristine MoS 2 , with the difference in their adsorption energies being 1.63 eV. The adsorption characteristics were also analyzed based on the differences in the charge densities of the two monolayers. Figure 3e,f shows schematics that highlight the differences in the charge densities of the pristine MoS 2 and C-doped MoS 2 monolayers with adsorbed NO 2 , respectively. The area in yellow is the electron accumulation area, while the cyan area is the electron consumption area. NO 2 acts as an electron acceptor, and a greater amount of charge is transferred in the case of C-doped MoS 2 , as evidenced by the charge depletion region above the C-top site.
For the adsorption of NO 2 onto the surfaces of the pristine MoS 2 and C-doped MoS 2 monolayers, the isosurface electron density values were set to 0.01 e Å À3 . The results of the charge analysis based on the Bader charges revealed that the adsorbed NO 2 accepts 0.03 and 0.54 e À from the pristine MoS 2 and C-doped MoS 2 monolayers, respectively. The charge transfer between the adsorbed gas molecules and the MoS 2 film results in a change in the sensor resistance when the sensor is exposed to the analyte gas. Thus, the strong interactions between the adsorbed molecules and MoS 2 led to significant conduction changes. Hence, charge transfer had a determining effect on the performance of the gas sensor. From these results, it can be concluded that the carbon impurity not only acts as the seed for the growth of the hierarchical nanobranch structure but also helps improve the performance of the gas sensor based on the C-MoS 2 .
To examine the thermodynamic stability of the carbon-doped MoS 2 system (C-MoS 2 ) and to deeper understanding of the C-doping process, the binding energy of C atom on the MoS 2 substrate was computed. A monolayer of sulfur vacancyincorporated MoS 2 (V S -MoS 2 ) was prepared as the base substrate, consisting of 25 Mo and 49 S atoms, where the vacancy was positioned at the S-top site. Then, the C atom was inserted at the S-vacancy resulting in C-MoS 2 structure, as depicted in Figure S9 (Supporting Information). As the stability indicator of C-doping, the binding energy (ΔE bind ) of C atom on the V S -MoS 2 surface was defined using Equation (3) where E CÀMoS 2 and E V S ÀMoS 2 are the total energies of the C-doped and S-vacancy contained MoS 2 system, respectively. The E C is the energy of C atom in vacuum. The resultant binding energy (ΔE bind ) is À7.23 eV, and its high negative value indicates that the C-substitution at the S-vacancy is thermodynamically stable. This computational data led us to conclude that the C-doping preferentially occurs at the S-vacancy domain, which is frequently observed during the deposition process, [48,49] and in turn, the C-MoS 2 structure is formed as seen in Figure 2. www.advancedsciencenews.com www.small-structures.com The 900 nm sample, which showed the highest sensitivity, was exposed to NO 2 in various concentrations (5,20,50,80, and 100 ppb) (Figure 4a). Figure 4b shows the functional relationship between the response of the C-MoS 2 gas sensor and NO 2 gas concentration. The C-MoS 2 gas sensor exhibited a linear response in the 5-100 ppb range, with the coefficient of determination (R 2 ) being 0.9776. Another important parameter of chemical sensors is their LOD, which is defined as LOD = k s B /S, where s B is the standard deviation from the device noise for zero analyte concentration ( Figure S10, Supporting Information), k is the expansion factor (a k value of 3 is recommended by IUPAC), and S is the sensitivity of the sensor at low concentrations. [50] Thus, the LOD of sensor was %1.58 ppt for s B = 0.00088 and S = 1.67 (at 5 ppb). This result indicates the potential capability of the sensor to detect NO 2 at the ppt level. The repeatability of the C-MoS 2 gas sensor was evaluated by continuously exposing it to NO 2 in different concentrations for four response-recovery cycles, as shown in Figure 4c. Based on the results of the continuous measurements of the sensor response, it can be concluded that the sensor showed good repeatability over the investigated concentration ranges, including at an extremely low concentration of 5 ppb. Finally, to evaluate the selectivity of the C-MoS 2 gas sensor, it was exposed to 100 ppb NO 2 , 100 ppb H 2 S, 1 ppm NH 3 , and 10 ppm CO 2 . The sensor responses with respect to the different analyte gases are plotted in Figure 4d. The results confirmed the remarkable selectivity of the C-MoS 2 gas sensor with respect to NO 2 gas. To further evaluate the sensing performance of the hierarchical C-MoS 2 sensor, it was compared with previously reported MoS 2 gas sensors, as shown in Table 1. The response time (t response ) is defined as the time required reaching the 90% of maximum response value in 5 ppm NO 2 gas. [51] In the same way, recovery time (t recovery ) is defined as a time taken by the sensor device to change its resistance from maximum value (in 5 ppm NO 2 ) to 10% above the base resistance ( Figure S11, Supporting Information). According to the table, the LOD of the hierarchically structured C-MoS 2 gas sensor for NO 2 gas is the lowest owing to its unique nanobranch structure, which contains the carbon impurity.

Conclusion
In summary, we fabricated gas sensors based on hierarchical C-MoS 2 nanobranches synthesized by MOCVD. The carbon impurity in the Mo(CO) 6 precursor was used as the seed for vertical growth during the CVD process. The morphology, atomic structure, and composition of the grown C-MoS 2 nanobranches were investigated using SEM, TEM, XPS, and PL and Raman spectroscopies to elucidate their growth mechanism. The gassensing properties of sensors based on C-MoS 2 films of different heights were evaluated; the sensor based on the 900 nm sample exhibited the best performance. This was owing to its high surface-to-volume ratio and the large number of edge sites of the hierarchical C-MoS 2 nanobranches. In particular, the gas sensor exhibited a response of 1.67 for an NO 2 concentration of 5 ppb, and its calculated LOD for NO 2 was approximately 1.58 ppt. DFT calculations were performed to elucidate the effects of the incorporated carbon in MoS 2 on the gas-sensing properties. The C-doped MoS 2 showed superior electron-transfer characteristics and higher adsorption energy compared with those of the pristine MoS 2 monolayer. Thus, it was confirmed that the carbon impurity acts as the seed for the growth of the hierarchical structure and improves the gas-adsorption properties. Furthermore, the proposed method of incorporating carbon for growth during the MOCVD process provides a facile route to synthesize large-area 3D nanostructures via a one-step process. The scope of this study can be extended to applications involving high chemical reactions other than gas sensing, and potentially enabling nanostructure formation of various materials.

Experimental Section
C-MoS 2 Growth: C-MoS 2 was grown in a shower-head-type reactor using Mo(CO) 6 (≥98%, Sigma-Aldrich, USA) as the Mo precursor and high-purity H 2 S (99.9%, noble gas, Republic of Korea) as the reactant gas. A p-type Si wafer with a 100 nm thick SiO 2 layer (1-10 Ω cm, Crystal Bank, Republic of Korea) was used as the substrate. The Si wafer was pretreated with a 0.1 wt% potassium hydroxide (KOH, 99.99%, Sigma-Aldrich, USA) solution for 10 min to make its surface hydrophilic and then rinsed with deionized water. The pretreated substrate was then loaded into a load-lock chamber connected to the MOCVD process chamber. The heating block in the CVD reactor was preheated to 350°C prior to the growth process. The synthesis was performed at a partial pressure ratio of 1:15 for the different periods. Device Preparation: The C-MoS 2 gas sensor was fabricated using the e-beam evaporation method. A stencil mask was used to prevent contamination by the chemical solutions used during the fabrication process. To use C-MoS 2 as the active material of the gas sensor, wafers coated with C-MoS 2 were cut to dimensions of 10 Â 20 mm, and the edge part of C-MoS 2 was removed using the scotch-tape method. The electrode was prepared by depositing 500 nm thick Ti and 100 nm thick Au layers using an e-beam evaporation method under a pressure of approximately 10 À7 Torr.
Materials and Device Characterization: Morphological analysis was performed using SEM (Hitachi S-4800) and TEM (FEI Titan 3 G2 60-300) systems with thermal field-emission guns operating at 300 kV. The samples for cross-sectional TEM imaging were prepared using a focused ion beam. The binding energies of the Mo, S, and C atoms were measured using XPS (Kratos Axis-Supra). The Raman and PL measurements were performed using a confocal Raman spectroscope (Renishaw inVia) with a 488 nm laser.
Gas-Sensing Experiments: The fabricated C-MoS 2 gas sensor was set in the sensing probe chamber shown in Figure S12 (Supporting Information) and placed in the enclosed chamber, and the gas-sensing experiments were conducted using an analyte gas diluted with N 2 gas. The concentration of the analyte gas was controlled by modulating the flow rates of the analyte gas and the N 2 dilution gas using two mass flow controllers. The measurements were performed using target NO 2 concentrations of 5, 20, 50, 80, 100, and 200 ppb. The change in the resistance of the C-MoS 2 sensor was measured using a Keysight B2985A high-resistance meter.
Computational Methods: All the DFT simulations were performed using the Quantum ESPRESSO packages. The electron exchange and correlation energies were treated by adopting the generalized gradient approximation of the Perdew-Burke-Ernzerhof functional. The projector-augmented wave potential was applied for the electron-ion interactions. The DFT-D3 method was used to deal with the van der Waals interactions. A hexagonal supercell (5 Â 5 Â 1) was constructed with 25 Mo atoms and 50 S atoms. In addition, a vacuum space of 15 Å was introduced along the vertical axis to avoid the effects of the other unit cells. One sulfur atom at the top site was replaced with a carbon atom to design C-doped MoS 2 . The Brillouin zone was sampled with a Γ-centered Monkhorst-Pack k-point with dimensions of 4 Â 4 Â 1. For structural optimization, the atomic force and energy were converged until the Hellmann-Feynman forces were less than 10 À4 Ry Bohr À1 , and the energy criterion was lower than 10 À8 Ry. A kinetic energy cutoff of 50 Ry was used for the electronic wave expansion during all the calculations. The differential charge density of the adsorption system was calculated using Equation (4) Δρ ads ¼ ρ MoS 2 þ NO 2 À ρ MoS 2 À ρ NO 2 (4) where ρ MoS 2 þ NO 2 is the charge density of the NO 2 -absorbed MoS 2 or C-doped MoS 2 monolayer, ρ MoS 2 is the charge density of the MoS 2 monolayer, and ρ NO 2 is the charge density of a NO 2 gas molecule.

Supporting Information
Supporting Information is available from the Wiley Online Library or from the author.