Fabrication of Stacked MoS2 Bilayer with Weak Interlayer Coupling by Reduced Graphene Oxide Spacer

We fabricated the stacked bilayer molybdenum disulfide (MoS2) by using reduced graphene oxide (rGO) as a spacer for increasing the optoelectronic properties of MoS2. The rGO can decrease the interlayer coupling between the stacked bilayer MoS2 and retain the direct band gap property of MoS2. We observed a twofold enhancement of the photoluminescence intensity of the stacked MoS2 bilayer. In the Raman scattering, we observed that the E12g and A1g modes of the stacked bilayer MoS2 with rGO were further shifted compared to monolayer MoS2, which is due to the van der Waals (vdW) interaction and the strain effect between the MoS2 and rGO layers. The findings of this study will expand the applicability of monolayer MoS2 for high-performance optoelectronic devices by enhancing the optical properties using a vdW spacer.


Results and Discussion
illustrates the preparation process of stacked MoS 2 . The chemical vapor deposition (CVD)-grown monolayer MoS 2 films were transferred onto the 300-nm-thick of SiO 2 /Si substrate using the conventional wet transfer method. (See Methods section for details) To prepare stacked bilayer MoS 2 , another CVD-grown monolayer MoS 2 film was transferred onto the monolayer-MoS 2 /SiO 2 /Si template. First, we confirmed the basic optical properties of the monolayer MoS 2 by using the PL and Raman analysis (Supporting Information (SI), Fig. S1). Figure 1b shows the PL intensity map of monolayer MoS 2 flakes and stacked MoS 2 flakes. Normally, the PL intensity of bilayer MoS 2 is much lower than that of monolayer MoS 2 because it has an indirect band gap 4,5 .
Interestingly, we observed that the integrated PL intensity of stacked MoS 2 flake (II) is slightly higher than that of the monolayer MoS 2 flake (I). It is implied that stacked bilayer MoS 2 flake has different optical features that are irrelevant to the layer-dependent optical properties of MoS 2 .
To confirm the PL peak position of the sample, we extracted the PL spectrum from the stacked MoS 2 flakes marked I and II in Fig. 1b. The A and B exciton peak positions at ~1.83 and ~1.98 eV of I and II were almost the same (Fig. 1c). Figure 1d shows the Raman spectra of I and II. We observed the two Raman modes at approximately 383.5 and 401.8 cm −1 of monolayer region I in the stacked MoS 2 , corresponding to the E 1 2g and A 1g modes 20 . The distance between the E 1 2g and A 1g modes was 18.3 cm −1 . The A 1g and E 1 2g modes of the region II were shifted compared to those of region I, but the distance between the two Raman modes is around 18.2 cm −1 , which is related to the monolayer properties of MoS 2 20 . We confirmed the same trend from other stacked MoS 2 flakes, regardless of stacking angle (see SI, Fig. S2). Figure 1e shows the PL intensity map of the same stacked MoS 2 sample after thermal annealing at ~300 °C for 1 h under nitrogen gas. After thermal annealing, the integrated PL intensity was reduced for all of the stacked MoS 2 flakes. The PL intensity of region II is decreased compared to that of region I. In region II of stacked MoS 2 , the intensity of exciton peak A decreases significantly and is slightly shifted (Fig. 1f). These PL results show a similar characteristic to the natural bilayer 21,22 . In addition, the distance between the E 1 2g and A 1g modes increased from ~18.2 to ~20.6 cm −1 , which is somewhat similar to the bilayer properties (Fig. 1g) 20 . From the PL and Raman results, we found that annealed stacked MoS 2 regions have bilayer properties.
To investigate the phenomena of the different optical properties of the stacked MoS 2 before and after thermal annealing, we carried out photothermal induced infrared resonance (PTIR) spectroscopy, which simultaneously provides topographical information and infrared absorbance (see SI, Fig. S3) 23 . Fig. 2a,b show the AFM topography images and height profiles of stacked MoS 2 before and after thermal annealing, respectively. The www.nature.com/scientificreports www.nature.com/scientificreports/ topography images provide distinguishable contrast between the monolayer (I) and stacked (II) MoS 2 region. We measured the thickness of the stacked MoS 2 from region I to II. The thickness of the dashed line at the stacked MoS 2 before thermal annealing was ~3.0 nm. This value is much larger than the expected thickness of the stacked bilayer MoS 2 22 . However, after thermal annealing, the thickness of the stacked MoS 2 decreased to ~1.6 nm, which is almost identical to the thickness of the bilayer MoS 2 22 . According to the previous results, it is assumed that defects or organic molecules exist in between the stacked flakes 21 . Fig. 2c presents the PTIR absorbance spectra of regions I and II of the stacked MoS 2 sample before and after thermal annealing treatment. We observed a peak at approximately 1728 cm −1 , corresponding to the C=O mode of the poly(methyl methacrylate) (PMMA) residues on the overall area of the stacked MoS 2 sample before thermal annealing 24 . Normally, PMMA is used as a supporting material during the wet-transfer process. However, polymer residues usually remain on the surface of the TMD thin film after the transfer process, and it is difficult to completely remove them from the TMD thin film 21,25 . Therefore, the stacked bilayer MoS 2 are thicker than the original bilayer MoS 2 because polymer residues are present on MoS 2 . The polymer residue suppresses the interlayer interaction between stacked MoS 2 monolayers. Thus, the PL and Raman results of stacked MoS 2 before annealing showed monolayer characteristics ( Fig. 1b-d). In contrast, after thermal annealing, the intensity of the C=O mode decreased. This indicates that the concentration of the PMMA is reduced (Fig. 2c). The thermal annealing removed the polymer residues and caused interlayer interaction in stacked MoS 2 . Thus, the thickness of stacked MoS 2 decreased after annealing, and it exhibited bilayer characteristics. From these results, the polymer residues between stacked MoS 2 flakes can act as a spacer, which can retain the intrinsic properties of monolayer MoS 2 prior to thermal annealing. However, during the device-fabrication process, the thermal annealing process inevitably decreases the resistivity of the interface for better device performance 26 . Therefore, it is necessary to find the optimum spacer in the stacked MoS 2 that blocks the intercoupling between flakes without damage in the thermal annealing process. Figure 3a shows an illustration of the sample-preparation process of the stacked MoS 2 with spacer layer. The CVD-grown monolayer MoS 2 films were transferred onto the 300-nm SiO 2 /Si substrate using the wet-transfer method. The prepared GO solution was spin-coated at 500 rpm for 5 s, followed by 1500 rpm for 60 s on top of the monolayer MoS 2 on the SiO 2 /Si substrate. The average thickness of the coated rGO on the MoS 2 sheet was approximately 5 ± 2 nm (see SI, Fig. S4). To prepare hybrid stacked MoS 2 , another CVD-grown monolayer MoS 2 film was transferred onto the GO-coated monolayer MoS 2 /SiO 2 /Si template. By thermal annealing treatment at 350 °C for 3 h, we fabricated the stacked MoS 2 with an rGO spacer. To investigate the reduction of GO, we carried out Raman spectroscopy measurement of the GO before and after thermal annealing (see SI, Fig. S5). Figure 3b is the AFM topography image of the stacked MoS 2 with rGO spacer. According to the AFM topography image, the monolayer and stacked MoS 2 region can be distinguished. Additionally, we observed wrinkles and some bubbles on the stacked MoS 2 with rGO sample. These seems to be formed during the transfer process (see SI, Fig. S6). The thickness of the monolayer MoS 2 with rGO (green dot) is approximately 1.1 nm and that of the stacked bilayer MoS 2 with rGO region (orange dot) is approximately 2.7 nm (see SI, Fig. S7). For the spatially resolved optical characterization, confocal PL and Raman measurements were performed for stacked MoS 2 with rGO spacer. Figure 3c shows the PL spectra obtained from each position of the sample. We observed the A and B excitons (~1.86 and 2.0 eV) of MoS 2 and the Raman G and D bands (~2.13 and 2.16 eV) of the rGO at PL spectra 12 . As shown in Fig. 3c, the PL intensity of stacked bilayer MoS 2 with rGO (M/rGO/M) region is twice that of rGO on monolayer MoS 2 (rGO/M) or monolayer MoS 2 on rGO (M/rGO) regions. In previous reports, PL quenching was observed in MoS 2 over mechanically exfoliated graphene (MEG) because MEG has semi-metallic properties that result in a metal-semiconductor interface 27,28 . On the other hand, rGO show semiconductor behavior because of residual oxygen functional groups after thermal annealing 29,30 . Therefore, PL quenching is not observed in the interface between rGO and MoS 2 stacked structure. Figure 3d   www.nature.com/scientificreports www.nature.com/scientificreports/ each other. As observed in Fig. 3d, some position of rGO/M region has a higher PL intensity rather than the M/ rGO/M region. To compare the intensity, we extracted the PL spectra of bright spot of rGO/M region and M/ rGO/M region (see SI, Fig. S8). According to previous report, GO induce the p-type doping of monolayer MoS 2 because of the functional groups of GO 12 . We believe that the cause of high PL intensity in the rGO/M region is p-type doping by residue of functional group after thermal annealing. A slight red shift in the A exciton peak is observed in the M/rGO/M region. In addition, the PL peak position of the stacked MoS 2 with rGO sample is different (Fig. 3e). We confirmed that the PL results at stacked MoS 2 with rGO spacer showed different characteristics from those of the stacked MoS 2 without rGO spacer.
To clarify the cause for the considerably changed optical and structural properties of stacked MoS 2 with rGO spacer, we have performed confocal Raman mapping. Figure 4a shows the Raman intensity map of the stacked MoS 2 with rGO. The Raman intensity map provides distinguishable contrast between the stacked and monolayer MoS 2 regions. Figure 4b shows the local Raman spectra of each position. The peak positions of the two Raman modes of stacked MoS 2 with rGO spacer are different from those of the stacked MoS 2 without spacer. The A 1g and E 1 2g modes of stacked MoS 2 with rGO were further shifted compared to those of stacked MoS 2 without spacer. In order to understand these results, we focus on the E 1 2g and A 1g modes of MoS 2 , depending on the region. Figure 4c presents the Raman intensity of the E 1 2g and A 1g modes of the MoS 2 according to each position. The higher Raman intensity of the M/rGO/M region originates from the increased scattering cross section by stacked MoS 2 with rGO. Figure 4d shows the peak position of the two Raman modes according to the position of the stacked MoS 2 with rGO spacer. The E 1 2g and A 1g modes of the stacked MoS 2 with rGO spacer were observed to shift to the opposite direction compared to that of pristine monolayer MoS 2 (M). Interestingly, we observed that the red shift of E 1 2g on M/rGO/M is large compared to M/rGO and rGO/M, which is shown in the Raman peak position map in Fig. 4e. Furthermore, the full width at half maximum (FWHM) of the E 1 2g peak of M/rGO/M broader than that of other regions (Fig. 4f). This indicates that interfacial strain is generated more on the M/rGO/M than on the M/ rGO and rGO/M 31,32 . In contrast, the blue shift of A 1g on M/rGO and rGO/M is larger than the shift observed in M/rGO/M, which is confirmed in the Raman map shown in Fig. 4g. The FWHM of the A 1g peak of M/rGO/M slightly broader than that of other regions (Fig. 4h). This means that the van der Waals (vdW) interlayer interaction between the MoS 2 and rGO on M/rGO and rGO/M is stronger than that of M/rGO/M 31 . According to the previous results, the blue shift of A 1g of MoS 2 can originate from the p-doping effect by the functional group of GO 12,33 . The A 1g peak appeared to be more strongly shifted as the amount of GO functional groups increased 12 . In this study, rGO was prepared with GO by thermal annealing. Thus, most functional groups of GO were removed and only a small amount of the functional group remained on the surface 25 . In addition, we confirmed that the A 1g mode of stacked MoS 2 with rGO was further shifted compared to stacked MoS 2 with GO (see SI, Fig. S9). This means that p-doping of MoS 2 by rGO has a negligible effect on the shift of the A 1g peak observed in our stacked MoS 2 with rGO. Therefore, it is concluded that the stacked MoS 2 with rGO spacer is in close contact with vdW interaction between MoS 2 and rGO. We suggest that the rGO can serve as a spacer to decouple the top and bottom MoS 2 and retain the direct band gap of MoS 2 .

Conclusions
In conclusion, we successfully fabricated stacked bilayer MoS 2 with rGO spacer. We observed that the decreased PL intensity in the stacked MoS 2 layers without spacer because of bilayer properties. Meanwhile, when the rGO is used as a spacer in between MoS 2 sheets, PL intensity shows sum up of PL from two individual MoS 2 sheets. It clearly indicates that rGO is a suitable material to decouple the top and bottom MoS 2 and retain the direct band gap of MoS 2 . In the Raman results, we observed the vdW interaction and the strain effect between the MoS 2 and www.nature.com/scientificreports www.nature.com/scientificreports/ rGO layers. We expect that these results will provide a fundamental understanding of the interlayer interaction of stacked 2D materials and enable further development of stacked MoS 2 -based devices.

Methods
Conventional wet transfer method. Poly(methyl methacrylate) (PMMA) (Micro Chem, 4 wt% in chlorobenzene) was coated onto monolayer MoS 2 grown on a SiO 2 /Si substrate to serve as a supporting layer for the transfer process. After being coated with PMMA, the sample was floated on a 1 M potassium hydroxide (KOH) solution at 80 °C to remove the SiO 2 layer. Subsequently, only monolayer MoS 2 with PMMA remained on the KOH solution. The remaining PMMA/MoS 2 was washed with DI water to remove any residual KOH etchant. The washed PMMA/MoS 2 was transferred to the SiO 2 /Si substrate. After the drying process was complete, the PMMA was removed using acetone.
preparation of the Go. GO was synthesized from natural graphite (Alfa Aesar, 99.999% purity, 200 mesh) by modified Hummers' method. First, 5 g of graphite powder and 350 mL of 10 M Sulfuric acid (H 2 SO 4 ) were blended. KMnO 4 (15 g) was slowly added over approximately an 1 h. Stirring was continued for 2 h in a cooled water bath. The mixture was strongly stirred for 3 days at room temperature. Deionized water was added and stirring for 10 min. The mixture was stirred for 2 h at room temperature after the addition of an aqueous solution of H 2 O 2 (30 wt%). Aqueous solution of HCl (35 wt%) was then added and stirred for 30 min at room temperature. After the supernatant solution was decanted, deionized water was slowly added and stirred for 30 min. The GO solution 1 g/l in water was sonicated for 1 h to exfoliate the GO sheets. To obtain dispersed GO, centrifugation at 10,000 rpm was performed for 1 h, and the supernatant solution was decanted.
Characterization Methods. PL spectra were obtained using a confocal PL spectrometer equipped with an objective lens with high numerical aperture of 0.7 and a diode-pumped solid-state laser (532 nm). Confocal