Large-area, continuous and high electrical performances of bilayer to few layers MoS2 fabricated by RF sputtering via post-deposition annealing method

We report a simple and mass-scalable approach for thin MoS2 films via RF sputtering combined with the post-deposition annealing process. We have prepared as-sputtered film using a MoS2 target in the sputtering system. The as-sputtered film was subjected to post-deposition annealing to improve crystalline quality at 700 °C in a sulfur and argon environment. The analysis confirmed the growth of continuous bilayer to few-layer MoS2 film. The mobility value of ~29 cm2/Vs and current on/off ratio on the order of ~104 were obtained for bilayer MoS2. The mobility increased up to ~173–181 cm2/Vs, respectively, for few-layer MoS2. The mobility of our bilayer MoS2 FETs is larger than any previously reported values of single to bilayer MoS2 grown on SiO2/Si substrate with a SiO2 gate oxide. Moreover, our few-layer MoS2 FETs exhibited the highest mobility value ever reported for any MoS2 FETs with a SiO2 gate oxide. It is presumed that the high mobility behavior of our film could be attributed to low charged impurities of our film and dielectric screening effect by an interfacial MoOxSiy layer. The combined preparation route of RF sputtering and post-deposition annealing process opens up the novel possibility of mass and batch production of MoS2 film.

very low carrier mobility (0.02-7 cm 2 /Vs) 10,14,15,18,20 . Continuous CVD-MoS 2 films have been demonstrated using MoCl 5 without pre-treatment, but the reported carrier mobility is also very low (0.003-0.03 cm 2 /Vs) 18 . Sanne et al. 21 reported mobility value of 24 cm 2 /Vs and I on /I off current ratio exceeding 10 7 for top-gated MoS 2 FETs with high-k gate dielectric on Si 3 N 4 . Ma et al. 22 demonstrated the vapor-solid growth of few-layer MoS 2 films on (0001) oriented sapphire. They estimated room temperature mobility of 192 cm 2 /Vs from the space-charge limited transport regime of the film. Laskar et al. 23 attained large-area MoS 2 films on (0001) oriented sapphire using sulfurization of e-beam evaporated Mo. They reported field-effect mobility of ~12 cm 2 /Vs using Mott-Guirney law with the carrier density of 10 16 cm −3 . Still, the lack of pristine quality, and wafer-scale synthesis of continuous MoS 2 film on SiO 2 is a challenging issue to be addressed.
Recently, there are few attempts to revive the sputtering technique for the growth of thin MoS 2 film [24][25][26] . However, the reported films are either relatively thick or the reported electrical and optical properties are rare and poor [27][28][29] . Muratore et al. 27 and Qin et al. 28 reported the synthesis of continuous few-layer MoS 2 by sputtering method using a MoS 2 target. Tao et al. 30 reported MoS 2 film using Mo target sputtered in vaporized sulfur ambient, but the grown MoS 2 film also exhibited p-type behavior with hole mobility up to ~12.2 cm 2 /Vs and low on/ off current ratio of ~10 3 .
Herein, we report a simple and mass-scalable approach for thin MoS 2 films via MoS 2 -RF sputtering combined with the post-deposition annealing process for the first time. From Raman spectra and photoluminescence (PL), it has been shown that the crystalline quality of the as-sputtered MoS 2 films was highly enhanced through the post-deposition annealing process. Synthesized bilayer MoS 2 films exhibited high field-effect mobility of ~29 cm 2 / Vs and a current on/off ratio of ~10 4 . The mobility increased up to ~173-181 cm 2 /Vs, respectively, for few-layer MoS 2 films. To the best of our knowledge, the mobility value of our bilayer MoS 2 FETs is larger than any reported results of single to bilayer MoS 2 FETs grown on SiO 2 /Si with a SiO 2 gate oxide. Furthermore, the mobility value (~173-181 cm 2 /Vs) of our few-layer MoS 2 FETs is the highest ever for any MoS 2 FETs with a SiO 2 gate oxide. It is much higher than that of single crystal exfoliated MoS 2 flakes on SiO 2 /Si substrate 31 and comparable to the value of bulk MoS 2 , room temperature mobility limited by phonon-scattering 32 .

Results and Discussion
MoS 2 films of different thicknesses were deposited by adjusting RF magnetron sputtering time such as 1, 3, 5 and 15 min onto SiO 2 /Si, quartz and sapphire substrates. The substrate temperature was varied from RT to 500 °C. As-sputtered films were subjected to post-deposition annealing treatment at 700 °C in the sulfur and Ar environment to improve their crystallinity. The detailed scheme for preparation and annealing processes is illustrated in Fig. 1(a). Optical microcopy images of sulfurized MoS 2 films at 1, 3 and 5 min. sputtered on SiO 2 /Si substrate are shown in Fig. 1b Raman spectra of the as-sputtered MoS 2 films are shown in Fig. 2a-c. The as-sputtered MoS 2 films exhibit the E 1 2g and A 1g mode peaks with low intensity. It might be due to low crystalline quality and the presence of defects contributes to the broad and low intensity of the peaks. The strong substrate related peak is observed at 520 cm −1 . As the sputter time increases, the Raman scattering peak intensities are slightly enhanced. Additional peaks at ~820 and ~992 cm −1 are related to the oxygen bonds and characteristic peaks of MoO 3 (alpha(α )-MoO 3 ) 33 . The symmetric stretch of 820 cm −1 (A g , B 1 g ) is a terminal Mo = O bond and the 995 cm −1 (A g , B 1 g ) is an asymmetric stretch of the terminal Mo = O bond along the a-and b-axes 24,25,34 . MoS 2 films are highly sensitive to moisture and oxidize easily. It has been also proposed that conventional sputter-deposited MoS 2 film contains oxygen substituted for sulfur atoms in the MoS 2 crystal lattice during film growth 26 . Figure 2a-c shows that Raman spectra variation through post-deposition annealing. The Raman peak enhancement indicates that the high-temperature annealing in the presence of sulfur and Ar greatly improved the crystallinity of as-sputtered MoS 2 film. Moreover, MoO 3 -related peaks were significantly suppressed for the annealed MoS 2 films. Through the post-deposition annealing in sulfur and Ar, the MoO 3 is believed to be transformed into a crystalline MoS 2 structure 10,35 . For the 1 min-sample (MoS 2 sputtered for 1 min and annealed at 700 °C for 1 hour), the Raman peak difference between E 1 2g and A 1g mode is ~20.5 cm −1 , which is close to that of the exfoliated bilayer MoS 2 36 . Figure 2d,e shows the Raman spectra according to the different annealing times from 30 min to 3 hours. The peak intensities are increased slightly with increase of annealing time. In order to focus oxygen-related peaks more precisely, the Raman analysis was performed for thick MoS 2 films; as-sputtered films for 15 min at RT and 400 °C, and annealed MoS 2 films ( Figure S1). The thick film sputtered at RT exhibited strong MoO 3 peaks at ~822 and ~992 cm −1 (Figure S1c,d). The oxygen peak intensities were reduced at higher substrate temperature (400 °C), but decreased most through the post-deposition annealing at 700 °C (the as-synthesized film was originally sputtered at RT). Raman mapping was performed over an area of 30 μ m × 30 μ m for 1 min-sample as shown in Fig. 2f-h. The E 1 2g and A 1g mode peaks appear at ~384.82-384.92 (with a standard deviation 0.048 cm −1 ) and ~405.19-405.29 cm −1 (with a standard deviation 0.049 cm −1 ), respectively. The peak difference (∆ k) values are in the range of ~20.27-20.47 cm −1 (with a standard deviation 0.066 cm −1 ), corresponding to the MoS 2 bilayer 14,36 . For the 3 min-sample (MoS 2 sputtered for 3 min and annealed at 700 °C for 1 hour, Figure S2), E 1 2g and A 1g mode are located in the range of ~382.23-382.33 cm −1 (with a standard deviation 0.05 cm −1 ) and ~407.29-407.39 cm −1 (with a standard deviation 0.045 cm −1 ), respectively, with ∆ k values in the range of ~24.96-25.16 cm −1 (with a standard deviation 0.066 cm −1 ), corresponding to few-layer MoS 2 film 36 . For the 5 min-sample (MoS 2 sputtered for 5 min and annealed at 700 °C for 1 hour), the E 1 2g mode position downshifted to ~380.63-380.73 cm −1 (with a standard deviation 0.05 cm −1 ) and the A 1g mode upshifted to ~408.29-408.39 cm −1 (with a standard deviation 0.047cm −1 ). The ∆ k value is increased to ~27.56-27.76 cm −1 (with a standard deviation 0.070 cm −1 ), suggesting that film thickness increment. The Raman measurement was also performed for as-synthesized MoS 2 sputtered at various substrate temperatures from 200 to 500 °C ( Figure S3) and 500 °C, the Raman peak intensities are slightly varied and MoO 3 peaks at ~822 and ~993 cm −1 are reduced. XRD was performed to investigate the structural properties of MoS 2 film. XRD patterns of as-sputtered MoS 2 thin films and the corresponding annealed films are shown in Figure S4a-c. For as-sputtered films, only a silicon substrate-related peak at 2θ = 33° is observed, supporting the amorphous structure of RT-sputtered MoS 2 film. However, (002) lattice oriented diffraction line is observed at 2θ = 14.2° for annealed MoS 2 films. The strong (002) peak is present when the periodicity in c-axis is normal to the MoS 2 film plane which is in good agreement with the previous results 37,38 . As-sputtered MoS 2 films sputtered at higher substrate temperatures revealed a very weak (002) peak and intensity tends to increase with the increase of sputtering temperature from 200 °C to 500 °C( Figure S5). Thus, Raman and XRD analysis revealed that increase of sputtering temperature improves the film quality and reduces oxygen content but is not sufficient for obtaining high quality MoS 2 film; post-deposition annealing improves film quality the most. XPS analysis was used to measure binding energies of Mo and S atom. For the 1 min-sample, Mo 3d peaks at 229.1 and 232.2 eV are exhibited (Fig. 3a), which is attributed to the doublet of Mo 3d 5/2 and Mo 3d 3/2 , respectively 39 . Also sulfur atoms-related 2S pathetic peak is observed at 226.3 eV. S 2− peaks are also observed ( Fig. 3b) at 161.9 and 163.1 eV due to S 2p 1/2 and S 2p 3/2 , respectively. In addition, a peak at 235.9 eV corresponds to the Mo 6+ of MoO 3 40 . For the 3 min and 5 min-sample, the observed peaks are slightly shifted to lower binding energies, which may be due to the increment of the number of layers. All these results are in good agreement with the reported values for MoS 2 crystal 41 . The intensity of Mo 6+ peaks decreased with increasing growth time. The Mo 6+ peaks indicate that some oxygen is incorporated in the grown MoS 2 film. Oxygen can be incorporated as substitutional atoms at sulfur sites 42 , as atoms bound to Mo atoms at plane edges 26 , as an intercalant between basal planes as O 2 or moisture (H 2 O) 43 , or as an interfacial Mo-oxide layer due to Mo-oxygen bonding at the MoS 2 -SiO 2 interface 27,28 . XPS survey spectra of Figure S7 show that the total oxygen and silicon signal decreases with increasing sputtering time. This could be explained as the probability of electrons escaping from the SiO 2 substrate reduces exponentially with increasing MoS 2 thickness 31 . XPS depth profile analysis was performed to investigate the interfacial structure of the MoS 2 /SiO 2 film. A 1keV Ar ion beam was used for sputtering purpose. XPS survey spectra depict that increment of oxygen peak as well as decrement of Mo core level peak with the increase of etching time ( Figure S8c). The expanded view of Mo 3d core peak variations are displayed in the Fig. 4a as a function of etching time. Before the sputter etching, the peaks of Mo 4+ 3d states are the main part of the spectra, and a small amount of MoO 3 state exists on the surface. When the film is etched by ion beam, there is a chemical shift of its binding energy toward smaller values. The shift is attributed to the change in the chemical states of Mo 4+ from the film surface to inner 44 . The Mo 6+ peak of MoO 3 is highly suppressed after etching for 10 sec. So, the Mo 6+ peaks are mainly originated from the surface oxidation of MoS 2 . The peak shift proceeds until 60 sec. After 60 sec, the binding energy shifts back toward higher values. From the Fig. 4b, sulphur related S 2− peaks are decreased and broadened as etching proceeds due to the damage induced by Ar etching, and the peaks almost disappear after etching for 50~60 sec (Fig. 4b, Figure S8c, Supporting Information). On the contrary, Mo peaks still exist after 60 sec. Hence, it is highly likely that these Mo could be combined with oxygen atoms or Si atoms in SiO 2 and form as a molybdenum oxide (MoO x ), or molybdenum silicon oxide (MoO x Si y ) layer. The Si 2p peak in Fig. 4d is exhibited at ~102 eV before Ar etching, and it upshifts towards ~103.2 eV, which is the binding energy of SiO 2 . It is suspected that the Si 2p peak at ~102 eV is due to the MoO x -SiO x bonding 45 . The Si 2p binding energy at ~102 eV is very close to that of (MoO 3 )70(SiO)30 (102.5eV) 45 .
We later discuss that the interfacial layer can alter the electrical properties of MoS 2 film. The XPS depth profiling was also performed for a very thick MoS 2 film ( Figure S8) and observed results are also similar to few-layer MoS 2 . Figure 5a,b shows the cross-sectional high-angle annular dark-field (HAADF) image and the corresponding electron energy loss spectroscopy (EELS) spectra for 5 min-sample. For the position 1 and position 2, 'Si' and 'SiO 2 ' are detected at ~99 eV and ~105 eV, respectively, and 'O' is detected at ~525 eV. Therefore these two points are clearly SiO 2 . A sulfur is detected at ~160 eV from the region 3 and 4, and not from the position 1, 2 and 5, indicating that point 3 and 4 are MoS 2 . It is thought that position 2 looks bright due to higher scattering of Mo. The position 5 is an epoxy material exhibiting only C spectrum. The comparison of bright field and HAADF image (Fig. 5c,d) indicates that the region 2 is an interfacial layer of the MoS 2 /SiO 2 . It is suspected that during sputtering process, Mo adlayers are initially formed at the interface of MoS 2 /SiO 2 and the Mo layers diffused into SiO 2 during the annealing step, resulting in the formation of MoO x Si y layer. The diffused interfacial layer appears brighter due to higher scattering with heavier atoms in that region than that in pure SiO 2 film.
Luminescence properties were studied by PL analysis as shown in Figure S10. The PL peaks are very weak and broad for the as-sputtered films. As sputtering time increases, peak position is shifted to a higher wavelength since the film thickness increases 46,47 . The luminescence peak intensities are significantly increased for the annealed MoS 2 films ( Figure S10b). For the 1 min-sample, the major peak is located at ~662 nm (1.87 eV, A peak) and one minor peak at ~620 nm (2 eV, B peak), which corresponds to a direct excitonic transition at the K point of the Brillouin zone of MoS 2 . The energy difference (~0.13 eV) is due to the degeneracy breaking of the valence band, which is in a close agreement with the literature 48,49 . The measured FWHM value for direct transition of peak A is ~67 meV, which is similar to freely suspended samples of MoS 2 (50-60 meV) 50 and narrower than that of MoS 2 exfoliated onto SiO 2 (100~150 meV) 51 . The emission intensity gradually increases with red shift 52,53 as increase of annealing time as shown in Figure S10c. This strong luminescence behavior is due to bilayer MoS 2 with a highly crystalline structure and support our earlier observation by Raman and XRD analysis that crystalline quality improvement via annealing at 700 °C.
The thickness of the film was analyzed by AFM as shown in Fig. 6a-c. AFM scan was taken at a corner of the MoS 2 film patterned using photolithography and etching process. For the 1 min-sample, the estimated thickness is ~1.4 nm, which is approximately close to bilayer MoS 2 18,36 (Fig. 6a). The thickness is ~3.8 nm (~5-6 layers) and ~6 nm (~8-10 layers), for the 3 min and 5 min-samples, respectively. Film continuity and uniformity were explored by AFM topographical 2D images. The surface roughness (R a , average deviation) values over a scanned area of 5 μ m × 5 μ m are ~0.18 nm, 0.22 nm, ~0.19 nm for 1, 3, and 5 min as-sputtered MoS 2 films, respectively ( Figure S11). 2D topographical images of the annealed films are shown in Fig. 6(d-f). The surface roughness (Ra) values are ~0.25 nm, ~0.35 nm, and ~0.29 nm for 1, 3, and 5 min-sample, respectively. These low roughness values support the highly uniform and continuous MoS 2 films. We believe that a wafer-scale MoS 2 could be produced by optimizing the sputtering time and annealing process.
HRTEM analysis was performed to explore the crystalline structure of MoS 2 film (1 min-sample) as shown in Fig. 7. The lower magnification-HRTEM images are exhibited in Fig. 7a,b for a continuous MoS 2 film on the copper grid. Figure S12a shows the HRTEM image over an area of 39 nm × 30 nm for 1 min-sample. The film shows a continuous film with a hexagonal lattice structure. Several types of Moiré fringes are observed and  Figure S12 were analyzed using fast Fourier transformation (FFT) in Fig. 7d. The exhibited two inverse FFT images (Fig. 7e,f) are extracted from Figure 7c, showing that the two layers are rotated by ~26°. Figure 7g shows a different Moiré pattern (type A in Figure S12) consisting of two layers stacked in a low rotation angle, and the corresponding FFT image is shown in Fig. 7h. The continuous and uniform surface homogeneity was confirmed by FESEM images for 1, 3 and 5-min MoS 2 samples as shown in   7(i-k), respectively. A monolayer is also spotted in Figure S12 (type C). Figure S12b,c shows HRTEM images for the 3 min and 5 min-sample as a supporting information. Large area MoS 2 films with ~1 × 9 cm 2 area and its Raman spectra are shown in Figure S13.
We have fabricated MoS 2 FETs and performed I-V measurement to investigate electrical properties. The schematic diagram of MoS 2 FET structure is given in Figure S15a. The active areas of FETs were defined during the sputtering process using a metal-shadow mask. As-sputtered MoS 2 film exhibited very high resistance in the range between 16 GΩ and 0.2 GΩ . I d -V g and I d -V d plots of these devices are presented in Figure S14a-f. Our previous results showed that as-sputtered MoS 2 at RT are amorphous structure and are oxidized. As a result, as-sputtered film can exhibit in high channel resistance and low current and mobility 54,55 . Figure 8a shows that I d -V d curves of the 1 min-sample with respect to the back-gate voltages. Figure 8b shows the transfer characteristics of the annealed bilayer MoS 2 FET (1 min-sample). The field-effect mobility was extracted based on the slope of Δ I d /Δ V g fitted to the linear regime of the transfer curves using the following equation: where W is the width of the channel (200 μ m) L is the length of the channel (2300 μ m), C ox is the capacitance per unit area of the gate dielectric (1.15 × 10 −8 F/cm 2 ), V d is the applied drain voltage (V d = 1 V), and Δ I d /Δ V g is the slope of the linear part of the transfer plot (I d -V g ), or transconductance. The extracted transconductance, field-effect mobility, and on/off current ratio is ~2.9 × 10 −8 S, 29 cm 2 /Vs and ~10 4 , respectively, at V d = 1V. The linear drain current and the transconductance values at V d = 1V are displayed in the Figure S15b. The transfer characteristics and I d -V d curves for few-layer MoS 2 FETs (3 and 5 min-sample) are shown in Figure S15c,d. The extracted transconductance values are ~1.81 × 10 −7 S and ~1.73 × 10 −7 S for the 3 min and 5 min-sample, respectively, which are ~6 times greater than bilayer MoS 2 (1 min-sample). The current on/off ratio values are ~2 × 10 3 −4 × 10 4 for few-layer MoS 2 FETs. The extracted field-effect mobility is ~181and ~173 cm 2 /Vs for 3 min-sample and 5 min-sample, respectively. Table 1 compares field-effect mobility and I on /I off values of our results with previously reported MoS 2 FETs. A significant enhancement can be noted in our MoS 2 FETs. It is interesting to compare with the recent reported mobility of ~12 cm 2 /Vs for thin MoS 2 film, but the mobility decreased significantly to ~0.44 cm 2 /Vs for  Table 1 is due to the substrate effect such as sapphire or high-k gate oxide effect.
Besides, the mobility (173-181 cm 2 /Vs) of our few-layer MoS 2 is the highest value ever for any MoS 2 FETs with SiO 2 gate dielectrics. Ayari et al. 31 reported 10~50 cm 2 /Vs of mobility from single crystal exfoliated MoS 2 flakes with 8~40 nm thickness. Our sputtered-MoS 2 films have small grain sizes, which are smaller compared with an exfoliated MoS 2 . An important question then remains, what could be the possible mechanism for the high mobility behavior of our MoS 2 film? For current 2D crystal materials, electron mobility is mostly dominated by charged impurity scattering, and the mobility values achieved to date are far below the intrinsic potential in these materials 52 . We think that the high mobility behavior of our film could be attributed to low charged impurities of our film and dielectric screening effect by the interfacial MoO x Si y layer. In our process, MoS 2 films were directly sputtered on SiO 2 /Si substrate at high vacuum and transistors were fabricated without transfer step, while the conventional CVD-grown MoS 2 , except exfoliated MoS 2 , usually needs the wet-transfer process onto a desired dielectric substrate and it make high contamination. Since sputtering process is performed in a high vacuum chamber, the chemical residues and gaseous adsorbates could be minimized. In addition, the dielectric surface dangling bonds could be also minimized due to a strong interaction of Mo and O on SiO 2 of the interfacial layer. Thus, low charged impurities could reduce the Coulomb scattering, resulting in high mobility values in the sputtered-MoS 2 53 . It is also well known that a bulk α -MoO 3 possesses very high relative dielectric constants (> 500 for α -MoO 3 ) 57 . And the dielectric constants of an atomically thin α -MoO 3 is still high even though it is low compared with its bulk value 58 . Thus, the MoO x Si y could reduce Coulomb scattering effects due to its high-k value as well as low dielectric dangling bonds. We have also prepared MoO 3 film on SiO 2 /Si substrate via a reactive sputtering using Mo target. XPS data of the sulfurized MoS 2 from Mo target also have the MoO 3 peak similar to the previous results ( Figure S16). The as-sputtered MoO 3 exhibited very high resistance due to a wide bandgap of the material. On the other hand, the sulfurized few-layer MoS 2 FETs (from MoO 3 ) exhibited high mobility values (~44 cm 2 /Vs) ( Figure S17). This experiment also supports our hypothesis. The fact that few-layer MoS 2 has much higher mobility value than that of bilayer MoS 2 reflects a critical role of Coulomb interaction distance upon the mobility values since thicker film has longer interaction distance. We compared hysteresis in transfer curves of FETs made by exfoliated-MoS 2 , CVD-grown MoS 2 , and sputtered-MoS 2 (Figures S18 and S19). It is well known that the origin of hysteresis of conventional FETs is due to the trapping and detrapping of carriers 59   We compared the hysteresis under vacuum environment to prevent such extrinsic and environmental effects and focus on the trapping at the MoS 2 /SiO 2 interface 60 . The exfoliated-MoS 2 and CVD-grown MoS 2 exhibited large hysteresis in there I d -V g curves. On the contrary, the sputtered-MoS 2 film exhibited small hysteresis. Such improvement in the hysteresis can be attributed to the small trap at the MoS 2 /SiO 2 interface of the sputtered-MoS 2 film. It is thought that charge scattering due to charge trapping is reduced due to the interfacial layer and enhance the mobility behavior of our sputtered-MoS 2 film.

Conclusions
We have successfully demonstrated the growth of large-area and continuous bilayer to few-layer MoS 2 on SiO 2 /Si substrate via RF sputtering combined with the post-deposition annealing process. The crystalline quality of the as-sputtered films was substantially improved via annealing at 700 °C in the sulfur and argon environment. The bilayer MoS 2 FETs exhibited a high field-effect mobility of ~29 cm 2 /Vs and an on/off ratio of ~10 4 . The mobility value of our bilayer MoS 2 FETs is larger than any of latest results of single to bilayer MoS 2 grown on a SiO 2 /Si substrate with a SiO 2 gate oxide. The mobility for few-layer MoS 2 FETs increased to ~173-181 cm 2 /Vs. Our few-layer MoS 2 FETs exhibited the highest mobility value ever for any MoS 2 FETs with a SiO 2 gate oxide. It is presumed that the high mobility behavior of our film could be attributed to low charged impurities of our film and dielectric screening effect by the interfacial MoO x Si y layer. The combined synthesis route of MoS 2 -RF sputtering with the post-deposition annealing process could open up the possibility of mass and batch production of MoS 2 film. We believe our proposed strategy will pave the way for applications of MoS 2 in future electronics and optoelectronics.

Method
The various sizes of SiO 2 (300 nm)/Si substrates ranging from 1 × 1 cm 2 to 3 × 3 cm 2 were used for the film preparation process. All the substrates were cleaned in acetone, methanol, isopropyl alcohol (IPA) solution and deionized (DI) water and then dried and baked for 5 min. After loading the SiO 2 /Si substrates into a sputtering chamber, the chamber was vacuumed at 1 × 10 −6 Torr. Before the deposition process, the MoS 2 target (99.99% purity) was pre-sputtered in a pure argon (Ar) atmosphere for 5 min in order to remove the oxide layer on the surface of the target. The MoS 2 films were sputtered at various temperatures: RT, 200, 300, 400 and 500 °C. The chamber pressure was maintained at 10 mTorr during the deposition in an Ar atmosphere, and the RF power was kept constant at 25 W for 1 min. The temperature variation in the chamber was monitored through a thermocouple. The as-sputtered MoS 2 films were post-annealed at 700 °C under Ar and sulfur environment to improve the crystalline quality of the films. The as-deposited films were placed in an annealing chamber and heated up to 700 °C for 30 min, 1 hour, 2 hours, and 3 hours. The carrier gas flow rate was maintained at 100 sccm, and the pressure of chamber was kept at 2 × 10 −2 Torr.
Fabrication of the MoS 2 FET devices. The active area of MoS 2 FET was formed during sputtering using a shadow mask. This kind of shadow mask is to avoid any chemical contamination by traditional active area preparation route of photolithography or electron-beam lithography. The metal contacts of 6 nm-Ti/30 nm-Au were prepared by evaporation. After making the electrode contacts, the devices were annealed at 200 °C for 2 hour in a vacuum tube furnace with 100 sccm Ar flow. After the annealing, the resistance of devices decreased significantly. The electrical properties of the fabricated MoS 2 transistors were measured using the 2 probe method at room temperature in a vacuum chamber to avoid oxidation.
Characterization details of MoS 2 films. Synthesized MoS 2 films were analyzed by Raman spectroscopy (Renishaw invia RE04, 512 nm Ar laser) with a spot size of 1 μ m and a scan speed of 30 seconds. A Si substrate with a Raman peak of 520 cm −1 was used for calibration. X-ray photoelectron spectroscopy (XPS) (PHI 5000 Versa Probe, 25W Al Kα , 6.7 × 10 −8 Pa) and photoluminescense (PL) with a 512 nm wavelength was used. Laser radiation of PL was focused onto the MoS 2 film with a spot-size of around 1 μ m. FE-SEM (HITACHI S-4700) and atomic force microscopy (AFM) (Vecco Dimension 3100) were used to check the morphology and thickness of the films. TEM samples were prepared using lacey-carbon Cu grid. The atomic structure of MoS 2 thin films was characterized by a JEOL-2010F TEM with an accelerating voltage of 200 keV. Image acquisition and processing (FFT, IFFT, etc.) were performed using the Gatan Digital Micrograph software (Gatan Microscopy Suite 2.0). The crystallinity of the film was characterized by in-plan X-ray diffraction (XRD, Rigaku) with Cu-Kα radiation operated at 50 KV and 300 mA.