Functionalization of Molybdenum Disulfide via Plasma Treatment and 3-Mercaptopropionic Acid for Gas Sensors

Monolayer and multilayer molybdenum disulfide (MoS2) materials are semiconductors with direct/indirect bandgaps of 1.2–1.8 eV and are attractive due to their changes in response to electrical, physicochemical, biological, and mechanical factors. Since the desired electrical properties of MoS2 are known, research on its electrical properties has increased, with focus on the deposition and growth of large-area MoS2 and its functionalization. While research on the large-scale production of MoS2 is actively underway, there is a lack of studies on functionalization approaches, which are essential since functional groups can help to dissolve particles or provide adequate reactivity. Strategies for producing films of functionalized MoS2 are rare, and what methods do exist are either complex or inefficient. This work introduces an efficient way to functionalize MoS2. Functional groups are formed on the surface by exposing MoS2 with surface sulfur vacancies generated by plasma treatment to 3-mercaptopropionic acid. This technique can create 1.8 times as many carboxyl groups on the MoS2 surface compared with previously reported strategies. The MoS2-based gas sensor fabricated using the proposed method shows a 2.6 times higher sensitivity and much lower detection limit than the untreated device.


Introduction
As a 2D material, graphene has been demonstrated to exhibit differing electrical and optical properties depending on factors such as electrical and magnetic fields [1], strain [2], stacking geometry [3], and edge chirality [4,5] However, many have realized the applicability limitations of graphene since it does not have a bandgap, which has raised interest in transition-metal dichalcogenide (TMD) materials. TMD materials consist of hexagonal metal atoms (M) sandwiched between two chalcogen atomic layers (X), which form 3D crystals through the lamination of adjacent sheets through van der Waals interactions. TMD materials are potentially useful in electronics owing to their superconductivity [6] and charge density wave effects [7].
Monolayer and multilayer molybdenum disulfide (MoS 2 ), one of the most studied TMDs, are semiconductors with direct/indirect bandgaps of 1.2-1.8 eV [8,9] and are highly attractive due to the considerable environment-dependent changes in their electrical [10][11][12][13], physicochemical [14], biological [15], optical [16], and mechanical properties [17]. Particularly, a field-effect transistor with monolayer MoS 2 exhibited highly favorable electrical properties such as a mobility of 200 cm 2 /(V s), on/off ratio of 108, and sub-threshold swing of 70 mV/dec [11]. Since these electrical properties of MoS 2 were revealed, related studies have increased, and research toward extending its applicability has focused on large-area deposition and growth [18][19][20][21] and functionalization [22][23][24][25]. While significant progress has been made toward large-scale MoS 2 , studies on functionalization techniques have yet to be actively conducted. Functional surface ligands consist of an anchor group that can attach to gas molecules, nanocrystals, biomaterials, etc. Particularly, functional groups can help dissolve the particles and provide adequate reactivity. Therefore, functionalization strategies for these materials are necessary. Detailed strategies must be established since some functionalization techniques for carbon nanotubes are not suitable for TMD materials.
Attempts have been intermittently made to introduce new MoS 2 functionalization strategies. Recently, a report described an approach to directly and simply functionalize MoS 2 using nitrilotriacetic acid [22]. However, the method requires a transition-metal cation with a high sulfur affinity and octahedral coordination since the direct anchoring of organic ligands to the sulfur surface is impossible. Although their new method was simple, the application of transition-metal cations complicates the process. Chen et al. and Chu et al. introduced methods for functionalizing MoS 2 using 4-nitrobenzenediazonium tetrafluoroborate [23] and cysteine [24], respectively, after generating sulfur vacancies, but neither described how to generate the sulfur vacancies or gave detailed experimental methods. The direct synthesis of various organic functional groups on MoS 2 thin films has also been reported [25], but the method is less efficient than the proposed method and lacks optimization.

Materials and Methods
Tri-layer MoS 2 films grown through CVD were obtained from 6carbon (Shenzhen, China). Gold paste (C2041206P2) as a gas sensor pad was purchased from Gwent (Pontypool, UK). 3-Mercaptopropionic acid (MPA) was purchased from Sigma-Aldrich (St. Louis, MO, USA), and a 40 mM MPA solution was prepared using a mixture of 99% ethanol and deionized (DI) water.
The fabricated MoS 2 -based gas sensor was characterized by scanning electron microscopy (SEM) (Hitachi, Tokyo, Japan), and the change in resistance in response to the constituent gas concentration was measured using an LCR meter (Hioki, Nahano, Japan). The chemical composition for each experimental step was investigated using X-ray photoelectron spectroscopy (XPS) (Thermo, MA, U.S.A). The sulfur vacancies and MoS 2 thickness after plasma treatment were measured by Raman spectroscopy (LabRAM, Horiba KOREA, Bucheon, Korea) and atomic force microscopy (AFM) (Shimadzu, Kyoto, Japan), respectively.
The MoS 2 film grown by CVD on a 3 mm × 3 mm SiO 2 /Si substrate was plasma-treated at 10 W for 2 s in an Ar + atmosphere to create sulfur vacancies. The plasma-treated MoS 2 film was then exposed to the MPA solution overnight to form coordinate bonds between the HS groups in MPA and sulfur vacancies. To fabricate the gas sensor pad, the gold paste was screen-printed onto both ends of the functionalized MoS 2 film. Figure 1 shows a schematic diagram and SEM image of the functionalized MoS 2 -based gas sensor. The MoS2 film grown by CVD on a 3 mm × 3 mm SiO2/Si substrate was plasma-treated at 10 W for 2 s in an Ar + atmosphere to create sulfur vacancies. The plasma-treated MoS2 film was then exposed to the MPA solution overnight to form coordinate bonds between the HS groups in MPA and sulfur vacancies. To fabricate the gas sensor pad, the gold paste was screen-printed onto both ends of the functionalized MoS2 film. Figure 1 shows a schematic diagram and SEM image of the functionalized MoS2-based gas sensor.

Results and Discussion
The presence of functional groups on a sensing material is a crucial factor in determining its performance since they can increase the sensitivity. Figure 2 shows the functionalization process of the MoS 2 films. The first step involves the generation of sulfur vacancies by plasma treatment. Since MoS 2 is a 2D material, it is highly susceptible to physical damage, and thus a weak and chemically inactive plasma treatment should be performed. Since the minimum power of our home-built plasma equipment is 10 W, the plasma power was fixed at 10 W, and Ar ions were used as plasmons for their chemical inactivity. As shown in Figure 3a, AFM analysis was conducted to examine the physical changes in the MoS 2 film over the treatment period. The thickness of a three-layer MoS 2 film is theoretically 2.3 nm [8,9,26], which was the same as the thickness measured by AFM analysis. As the treatment time increased, the MoS 2 film was physically etched at a rate of 0.6 Å/s. The etch rate of the Ar plasma was notably slower than those reported for previous studies using oxygen plasma [11], indicating that Ar plasma is more suitable for the generation of sulfur vacancies in MoS 2 films.

Results and Discussion
The presence of functional groups on a sensing material is a crucial factor in determining its performance since they can increase the sensitivity. Figure 2 shows the functionalization process of the MoS2 films. The first step involves the generation of sulfur vacancies by plasma treatment. Since MoS2 is a 2D material, it is highly susceptible to physical damage, and thus a weak and chemically inactive plasma treatment should be performed. Since the minimum power of our home-built plasma equipment is 10 W, the plasma power was fixed at 10 W, and Ar ions were used as plasmons for their chemical inactivity. As shown in Figure 3a, AFM analysis was conducted to examine the physical changes in the MoS2 film over the treatment period. The thickness of a three-layer MoS2 film is theoretically 2.3 nm [8,9,26], which was the same as the thickness measured by AFM analysis. As the treatment time increased, the MoS2 film was physically etched at a rate of 0.6 Å/s. The etch rate of the Ar plasma was notably slower than those reported for previous studies using oxygen plasma [11], indicating that Ar plasma is more suitable for the generation of sulfur vacancies in MoS2 films.   Figure 3b shows the Raman spectra of the MoS2 films that were plasma treated for different durations. Numerous theoretical studies have indicated that the frequency difference between the E 1 2g and the A1g Raman modes is dependent on the thickness of the MoS2 film [18]. The layers of MoS2 are held together via van der Waals forces. As the number of layers decreases, less incident energy is lost through the vertical vibrations of the molecules, resulting in a decrease in the frequency of the A1g peak [18][19][20][21]. Conversely, the E 1 2g peak is affected more by the interatomic coulomb potential, which decreases with fewer layers due to the reduction in interactions between atoms. As a result, the E 1 2g frequency increases. This increase in E 1 2g frequency corresponding to horizontal vibrations is less sensitive to the number of layers than the decrease in A1g frequency corresponding to vertical vibrations. The A1g peak occurring above the bi-layer rather than a single layer is less sensitive to the MoS2 thickness. However, the E 1 2g peak is sensitive to the MoS2 thickness. The reduced sensitivity to thickness of the E 1 2g mode has been explained by the less-efficient inter-layer coupling of the in-plane phonons [27]. As the MoS2 becomes the generation of sulfur vacancies was confirmed to be 2 s, since the physical thickness of sulfur is 0.176 nm thinner, the E 1 2g frequency increases and the A1g frequency decreases, thereby reducing the frequency difference. This frequency difference for the MoS2 film treated for 8 s was reduced from 23.21 for the untreated film to 21.89 cm −1 . A previous paper reported this frequency difference for a bilayer MoS2 film as 21.61 cm −1 [28]. Therefore, the Raman analysis indicates that the treatment time required to etch one layer was 8 s in the proposed method. Thus, to form sulfur vacancies in a 3-layer MoS2 film without removing a layer, a treatment time of less than 8 s is required. As demonstrated by the AFM analysis (Figure 3a), the treatment time for the generation of sulfur vacancies was confirmed to be 2 s, since the physical thickness of sulfur is 0.176 nm.
The XPS S 2p and Mo 3d deconvolutions further supported the generation of sulfur vacancies by the plasma treatment ( Figure 4). The Mo 3d and S 2p peaks of the MoS2 film appeared at 229.3 and 162.2 eV, respectively, which have previously indicated an MoS2 film with three layers [25]. The S/Mo ratio can be calculated as the area ratio of the S 2p orbital representing the number of S atoms and the Mo 3d orbital representing the number of Mo atoms. Using the integral, the area of each orbital according to the XPS result was calculated, and the S/Mo ratio was calculated to compare before and after plasma treatment. The untreated MoS2 film showed the ideal S/Mo ratio of 1.91, while that of the plasma-treated MoS2 film was 1.51, demonstrating the formation of sulfur vacancies [29].  Figure 3b shows the Raman spectra of the MoS 2 films that were plasma treated for different durations. Numerous theoretical studies have indicated that the frequency difference between the E 1 2g and the A 1g Raman modes is dependent on the thickness of the MoS 2 film [18]. The layers of MoS 2 are held together via van der Waals forces. As the number of layers decreases, less incident energy is lost through the vertical vibrations of the molecules, resulting in a decrease in the frequency of the A 1g peak [18][19][20][21]. Conversely, the E 1 2g peak is affected more by the interatomic coulomb potential, which decreases with fewer layers due to the reduction in interactions between atoms. As a result, the E 1 2g frequency increases. This increase in E 1 2g frequency corresponding to horizontal vibrations is less sensitive to the number of layers than the decrease in A 1g frequency corresponding to vertical vibrations. The A 1g peak occurring above the bi-layer rather than a single layer is less sensitive to the MoS 2 thickness. However, the E 1 2g peak is sensitive to the MoS 2 thickness. The reduced sensitivity to thickness of the E 1 2g mode has been explained by the less-efficient inter-layer coupling of the in-plane phonons [27]. As the MoS 2 becomes the generation of sulfur vacancies was confirmed to be 2 s, since the physical thickness of sulfur is 0.176 nm thinner, the E 1 2g frequency increases and the A 1g frequency decreases, thereby reducing the frequency difference. This frequency difference for the MoS 2 film treated for 8 s was reduced from 23.21 for the untreated film to 21.89 cm −1 . A previous paper reported this frequency difference for a bilayer MoS 2 film as 21.61 cm −1 [28]. Therefore, the Raman analysis indicates that the treatment time required to etch one layer was 8 s in the proposed method. Thus, to form sulfur vacancies in a 3-layer MoS 2 film without removing a layer, a treatment time of less than 8 s is required. As demonstrated by the AFM analysis (Figure 3a), the treatment time for the generation of sulfur vacancies was confirmed to be 2 s, since the physical thickness of sulfur is 0.176 nm.
The XPS S 2p and Mo 3d deconvolutions further supported the generation of sulfur vacancies by the plasma treatment ( Figure 4). The Mo 3d 5 2 and S 2p 3 2 peaks of the MoS 2 film appeared at 229.3 and 162.2 eV, respectively, which have previously indicated an MoS 2 film with three layers [25]. The S/Mo ratio can be calculated as the area ratio of the S 2p orbital representing the number of S atoms and the Mo 3d orbital representing the number of Mo atoms. Using the integral, the area of each orbital according to the XPS result was calculated, and the S/Mo ratio was calculated to compare before and after plasma treatment. The untreated MoS 2 film showed the ideal S/Mo ratio of 1.91, while that of the plasma-treated MoS 2 film was 1.51, demonstrating the formation of sulfur vacancies [29].  For the functionalization of the MoS2 surface, the MoS2 film with sulfur vacancies was exposed to an MPA solution. MoS2 has been used as a hydrodesulphurization catalyst because surface sulfur vacancies tend to form covalent bonds with sulfur-containing groups. As a result, molecules with thiol functional groups chemically bond to the sulfur vacancies formed on the MoS2 surface [30,31]. The chemical components of the MoS2 films with sulfur vacancies treated in solutions of MPA at different concentrations were measured by XPS to determine the optimized MPA concentration (Figure 5a). The measured absorbance spectra presented four peaks; the first at 285 eV was assigned to sp 2 -hybridized carbon atoms, and the other three at 286.06, 288.51, and 288.97 eV were assigned to oxygen-containing hydroxyl (C−O), carbonyl (C=O), and carboxyl groups (−COO), respectively. The presence of these oxygen-containing group features implies that the functional groups were stably bonded to the MoS2 film [32]. From the XPS results for different MPA concentrations, the atomic ratio corresponding to carboxyl groups increased for concentrations up to 55 mM and then decreased, potentially because high concentrations of MPA may form inter-MPA disulfide bridges (Figure 5b). For the functionalization of the MoS 2 surface, the MoS 2 film with sulfur vacancies was exposed to an MPA solution. MoS 2 has been used as a hydrodesulphurization catalyst because surface sulfur vacancies tend to form covalent bonds with sulfur-containing groups. As a result, molecules with thiol functional groups chemically bond to the sulfur vacancies formed on the MoS 2 surface [30,31]. The chemical components of the MoS 2 films with sulfur vacancies treated in solutions of MPA at different concentrations were measured by XPS to determine the optimized MPA concentration (Figure 5a). The measured absorbance spectra presented four peaks; the first at 285 eV was assigned to sp 2 -hybridized carbon atoms, and the other three at 286.06, 288.51, and 288.97 eV were assigned to oxygen-containing hydroxyl (C−O), carbonyl (C=O), and carboxyl groups (−COO), respectively. The presence of these oxygen-containing group features implies that the functional groups were stably bonded to the MoS 2 film [32]. From the XPS results for different MPA concentrations, the atomic ratio corresponding to carboxyl groups increased for concentrations up to 55 mM and then decreased, potentially because high concentrations of MPA may form inter-MPA disulfide bridges (Figure 5b).  Figure 5 also compares the effectiveness of the proposed and previous method for the formation of functional groups. The exfoliation process by n-butyllithium deforms the crystal structure of MoS2 and introduces defects mainly at the edges of the exfoliated 2D material [25]. Conversely, the defects introduced by the plasma treatment were evenly generated on the MoS2 surface, and thus more defects were created compared with the exfoliation process. As previously mentioned, since thiol functional groups covalently bond to the defect sites, the number of defect sites is a crucial factor in the functionalization of the nanosheets. As illustrated by the C 1s deconvolution in Figure 5, approximately 1.8 times more carboxyl groups were produced using the proposed strategy compared with the amount generated by the previously published methods.  Figure 5 also compares the effectiveness of the proposed and previous method for the formation of functional groups. The exfoliation process by n-butyllithium deforms the crystal structure of MoS 2 and introduces defects mainly at the edges of the exfoliated 2D material [25]. Conversely, the defects introduced by the plasma treatment were evenly generated on the MoS 2 surface, and thus more Nanomaterials 2020, 10, 1860 6 of 9 defects were created compared with the exfoliation process. As previously mentioned, since thiol functional groups covalently bond to the defect sites, the number of defect sites is a crucial factor in the functionalization of the nanosheets. As illustrated by the C 1s deconvolution in Figure 5, approximately 1.8 times more carboxyl groups were produced using the proposed strategy compared with the amount generated by the previously published methods.
As shown in Figure 6, the analytical performance characteristics of the functionalized MoS 2 films were confirmed using an NH 3 gas sensor. The electrical properties of MoS 2 are highly sensitive to charge transfer between adsorbed molecules and the 2D films. According to theoretical studies, NH 3 interacts weakly with perfect films and induces little charge transfer [33,34]. From the theoretical studies discussed above, it can be expected that the functionalized MoS 2 -based gas sensor has a higher sensitivity than the pristine MoS 2 -based sensor. Therefore, the adsorption of NH 3 onto defect-free films is difficult, whereas MoS 2 films with functional groups can easily adsorb the gas. Additionally, the adsorption barrier of functionalized MoS 2 films is further lowered by pre-dissociated oxygen atoms, which results in an increase in the charge transfer rate [35]. Both the untreated and functionalized devices exhibited an increase in resistance upon exposure to NH 3 gas [36]. Figure 6c presents the calibrated response curves for different gas concentrations of the untreated and functionalized MoS 2 -based sensors. The device with untreated MoS 2 had a sensitivity of 0.0075 at a gas concentration range of 0-22 ppm. On the other hand, the functionalized MoS 2 -based sensor showed a sensitivity of 0.02 for the same range, which is 2.6 times higher. This enhanced sensitivity is attributable to a reduced electron-withdrawing ability from the formation of hydrogen bonds between the functional groups and polar NH 3 molecules [37]. Moreover, the device with the functionalized MoS 2 film showed a detection limit two times lower than that of the untreated device.
Nanomaterials 2020, 10, x FOR PEER REVIEW 7 of 9 Figure 6. Responses of (a) untreated and (b) functionalized MoS2-based sensors for different NH3 concentrations at room temperature. (c) Variation in R/R0 of the MoS2 films exposed to different NH3 concentrations, where R0 is the resistance of the device in the outgassed state and R is the steady-state resistance when exposed to the NH3 gas.

Conclusions
In this study, a technique for the stable functionalization of MoS2 was investigated. Although it is difficult to produce functional groups on 2D materials, it is possible to stably form functional groups on MoS2 films through treatment with weak plasma and successive exposure to MPA. Sulfur Figure 6. Responses of (a) untreated and (b) functionalized MoS 2 -based sensors for different NH 3 concentrations at room temperature. (c) Variation in R/R 0 of the MoS 2 films exposed to different NH 3 concentrations, where R 0 is the resistance of the device in the outgassed state and R is the steady-state resistance when exposed to the NH 3 gas.

Conclusions
In this study, a technique for the stable functionalization of MoS 2 was investigated. Although it is difficult to produce functional groups on 2D materials, it is possible to stably form functional groups on MoS 2 films through treatment with weak plasma and successive exposure to MPA. Sulfur vacancies were formed on the MoS 2 film surface by a minimal plasma treatment (10 W and 2 s), and then carboxyl groups were formed through coordinate bonds at the sulfur vacancies using an MPA solution. The stable functionalization of MoS 2 using this strategy was confirmed by the performance of an NH 3 sensor. The functionalized MoS 2 -based device at low gas concentrations exhibited a 2.6 times greater sensitivity than the device with the untreated MoS 2 film since the adsorption of NH 3 onto perfect films is not only difficult but also causes weak charge transfer. Furthermore, the MoS 2 films with functional groups provided lower detection limits since they allow for easier gas adsorption. The proposed strategy is suitable for the fabrication of sensors based on other 2D materials in addition to MoS 2 -based devices.