The discovery of graphene has opened the door to study other two-dimensional (2D) materials,^[1–5] such as 2D metal chalcogenides (2DMCs), black phosphorus, and so on.^[6] These materials have attracted great interest owing to their completely different electronic structures from those of bulk materials. Furthermore, novel electronic properties and photoresponse can be achieved, due to the integration of two or more 2D materials.^[7,8] On account of its atomic-level thickness, the layered heterostructure also provides excellent optical transparency and good mechanical flexibility.^[9]

Graphene is expected to be used in many electronic devices due to its high conductivity and high carrier mobility. However, its low photoresponse limits its potential.^[10–12] Single layer MoS₂ has high optical absorption and distinct photoluminescence (PL). Nonetheless, it has low carrier mobility, which affects its responsivity. Therefore, the two materials can complement each other in the MoS₂/graphene heterostructures. The single layer MoS₂ can enhance the optical absorption, while transfer photo-generated carriers effectively to graphene.^[13] New photoelectric devices have been developed based on these heterostructures, such as high gain photodetectors and optical responding memories.^[14] MoS₂ is used as the sensitizer of light, and graphene with high mobility is used as highway to transport the photo-excited carrier. In this case, the combination of high optical absorption and fast electron transportation makes the MoS₂/graphene phototransistors more responsive (up to 10⁸ A/W) than those with only graphene or MoS₂.^[15,16]

Nonetheless, the existence of grain boundaries (GBs) leads to the deterioration of physical properties of graphene, such as the decrease of chemical reactivity,^[17] thermal conductivity,^[18] and carrier mobility.^[19] This, means that growing single-crystal graphene (SCG), identifying and modifying its GBs are very important for practical applications. The GBs of graphene are disordered topological defects, which strongly affect the physical and chemical properties of graphene,^[20] therefore, its application in devices. Recent research shows that the distribution of defects at grain boundaries is linear periodic.^[21] In general, nonhexagonal rings (such as five-, seven-, and eight-membered rings) exist at the grain boundaries to mediate the lattice mismatch between two graphene domains.^[22] The inherent topological defects can accommodate lattice defects, but retain three folds of sp² bonding of carbon atoms.^[21] Topological defects are also the basis of polycrystalline graphene. Such defects may adversely affect the thermal and electronic properties of graphene, such as reducing the electronic mobility. The growth of MoS₂ on graphene depends on van der Waals interaction, which can be used to mark the graphene GBs.^[23] Therefore, it is crucial to investigate the influence of graphene GBs on the photoelectric response of MoS₂/graphene heterostructures.

In this paper, single crystal MoS₂ is deposited on hexagonal single-crystal graphene (HSCG) by molecular beam epitaxy (MBE). There is only one orientation of MoS₂ on SCG. The spectral characteristics and photoresponse of the samples are measured. The G and 2D peaks of graphene demonstrate a blueshift after growing a monolayer of MoS₂. The quenching of PL of MoS₂ is remarkable, while the PL peaks at the MoS₂/graphene GBs shift toward short wavelength. Besides, the photocurrent signal in the MoS₂/HSCG heterostructures is successfully measured without bias.

The HSCGs were grown by inserting transition metal into Cu pocket.^[24,25] SLG was grown on the Cu pocket via the low-pressure-chemical-vapor-deposition process.^[23] After that, the polymethyl methacrylate (PMMA)-assisted method was used to transfer the graphene onto SiO₂/Si substrates^[26] for further synthesis. MoS₂ crystals were grown by an MBE system using MoO₃ and S powder as precursors.^[23] The aqueous solution (2 M NaOH) at 90 °C was used to transfer the MoS₂/graphene onto Cu grids. Finally, electron beam lithography was used to prepare a pair of Ti/Au electrodes.^[27]

The surface morphology of materials was characterized by a scanning electron microscope (SEM) (Zeiss Sigma). A Raman spectrometer (Alpha 300, WITec) with a 488 nm laser excitation source was used to measure Raman and PL spectra. The photocurrent was measured using the Raman spectrometer with a 532 nm laser. TEM (JEM-2100, operated at 200 keV with a point-to-point resolution of 0.19 nm) was used to probe the heterostructures jointly with selected area electron diffraction (SAED) patterns.

A few number of HSCG domains on Cu pocket are observed by SEM, as shown in Fig. 1(a). Besides, the sizes of the HSCG domains are about 40–60 μm. An SLG membrane up to 5 cm (Fig. 1(b)) was also prepared. It is a common method to transfer the grown graphene onto target substrates using poly(methyl methacrylate) PMMA.^[26] Then the MoS₂ single crystals were synthesized on the graphene/SiO₂/Si substrates using the MBE method. After that, a hot alkali etching method was used to transfer the samples onto Cu grids.^[23,28] There is only one orientation for MoS₂ grown on HSCG (Fig. 1(c)), while various orientations of MoS₂ are observed when deposited on polycrystalline graphene. Therefore, the location and the size of graphene grain boundaries can be identified in this way easily, as shown in Fig. 1(d).

Fig. 1.

	Figure Option ViewDownloadNew Window
	Fig. 1. SEM image of (a) hexagonal single-crystal graphene and (b) single layer graphene membrane grown on cooper foil. (c) Enlarged SEM image of single crystal MoS2 grown on HSCG domain. (d) Enlarged SEM image of MoS2/SLG GBs heterostructures. (e) TEM image of MoS2. (f) SAED pattern of (e).

To confirm the orientation of the MoS₂/graphene heterostructures, high resolution TEM (HR-TEM) images of the heterostructures and corresponding diffraction patterns were obtained by using TEM and SAED. Figure 1(e) shows a triangular MoS₂ single crystal on the graphene film, whose edges can be exhaustively scrutinized. The SAED clearly shows two sets of hexagonal patterns from MoS₂ and graphene without any deviation from the lattice orientation, which suggests that the MoS₂ single crystals are grown epitaxially on the graphene substrates through van der Waals (Fig. 1(f)).

Figure 2(a) displays an SEM image of the MoS₂/HSCG heterostructures. The Raman spectra of the graphene show several differences before and after the growth of MoS₂, as shown in Fig. 2(b). Before MoS₂ deposition, the characteristic G and 2D peaks of graphene are observed, while the D band which represents defects in graphene is absent. The intensity ratio of 2D/G band is almost 2 : 1, confirming the growth of high quality SLG.^[29,30] After MoS₂ deposition, a broad background of PL from MoS₂ is observed, confirming that both the graphene and MoS₂ are present. Furthermore, the intensity of the 2D peak is reduced, but a D peak with weak intensity emerges. Note, the G and 2D peaks also show a blueshift. This indicates that the growth of MoS₂ slightly influences the quality of graphene.

Fig. 2.

	Figure Option ViewDownloadNew Window
	Fig. 2. (a), (e) SEM images of MoS2/graphene heterostructures. (b) Raman spectrum of the SLG and heterostructures. (c) Raman spectrum of the MoS2 and heterostructures. (d) PL spectrum of MoS2 on SLG and SiO2. (f) Enlarged SEM image corresponding to the red rectangle in (e).

It is known that elevated temperature, charge transfer, and strain affect the position of the 2D peak of graphene. Since the spectrum is measured at room temperature, the blueshifts of the G and 2D peaks are not due to increasing temperature. On the other hand, the van der Waals contact between graphene and MoS₂ tends to change the doping level,^[31] while the increase of the doping concentration generally leads to a reduction of the I_2D/I_G ratio.^[5] Therefore, the blueshift in this case may due to doping.

The Raman spectra with characteristic MoS₂ features are measured and shown in Fig. 2(c). The A_1g and peaks are located at 404 cm⁻¹ and 384 cm⁻¹ respectively. It confirms the growth of the single layer of MoS₂.^[32–34] Unsurprisingly, no MoS₂ related peaks are observed in the graphene samples. The Raman spectra for MoS₂ domains grown on SiO₂ are also investigated for comparison. The A_1g and peaks for MoS₂ on SiO₂ show a higher intensity compared with those on graphene. It is known that single layer MoS₂ has a strong PL.^[11] The PL spectra of MoS₂ grown on HSCG, graphene GBs, and SiO₂ are shown in Fig. 2(d). The intensity of PL is strongly suppressed for the heterostructures. It is clear that the PL of the MoS₂ grown on graphene is strongly quenched, which proves the existence of electronic interaction between graphene and MoS₂. Interestingly, the characteristic PL features of MoS₂ grown on graphene GBs show a blueshift, which may result from the lattice mismatch. It can be seen from Fig. 2(f) that MoS₂ grows more densely on the graphene GBs, without the typical triangle shape. This may be the reason for the blueshift of the PL. MoS₂ domains directly grown on graphene show stronger interaction with graphene, which is crucial for effective interlayer coupling.

Figures 3(a) and 3(b) show the Raman mapping of graphene grown by CVD method. These two figures prove that the graphene grown by inserting Mo sheet into copper pocket has high quality. Figures 3(c) and 3(d) show the G and 2D peaks of graphene after MoS₂ growth by MBE, respectively. There are defects on the surface of graphene after MoS₂ growth since the 2D band intensity decreases. Based on the mapping of MoS₂ A_1g and bands (Figs. 3(e) and 3(f)), MoS₂ tends to grow at graphene GBs and folding. Figure 3(g) is the MoS₂/HSCG PL mapping which shows that the PL distribution on the graphene surface is relatively uniform. This makes the subsequent measurement of photocurrent feasible.

Fig. 3.

	Figure Option ViewDownloadNew Window
	Fig. 3. Raman mapping images of the G peak (a) and 2D peak (b) of graphene, G peak (c) and 2D peak (d) of the graphene after MoS2 growth. (e), (f) Raman mapping images of A1g and bands of the MoS2 grown on HSCG. (g) PL mapping of the MoS2 grown on HSCG.

The photocurrent signal is successfully captured by irradiation of a 532 nm laser without bias, which proves the presence of photoinduced electron transfer from MoS₂ to graphene. Figure 4(a) is a schematic diagram of the device and figure 4(b) is the optical micrograph of the MoS₂/HSCG heterostructure. Raman intensity mapping of the G band of graphene shown in Fig. 4(e) confirms the growth of high quality graphene. Therefore, the photocurrent signal is not likely caused by breakage of graphene. The boundaries between graphene and SiO₂ can be seen clearly in the mapping of photocurrent of the device, as shown in Fig. 4(c).

Fig. 4.

	Figure Option ViewDownloadNew Window
	Fig. 4. (a) Schematic of MoS2/HSCG heterostructure device. (b) Optical micrograph of the MoS2/HSCG heterostructure and corresponding (c) Raman photocurrent mapping and (d) the A1g band of MoS2. (e) Raman intensity mapping of the G band of graphene.

The photocurrent near the MoS₂/graphene interface in Fig. 4(c) originates from the interaction of photogating, hot carriers, and other factor. The different position and strength of the metal at the upper and lower ends of the picture prove that the device has formed a path in the circuit. Meanwhile, the photocurrent was also measured at the place where metal–graphene contacted. This is because the potential barriers and built-in electric field might emerge at the electrode/graphene and MoS₂/graphene interfaces due to change of Fermi level of adjacent graphene.^[13]

The photothermoelectric effect (PTE) will form a photo-generated voltage V_PTE and generate a photocurrent in graphene,^[35,36] due to the interaction between electrons. The photoresponse can be calculated by V_PTE = (S₂ – S₁)ΔT, where S₁ and S₂ are the Seebeck coefficients, and ΔT is the temperature difference. Therefore, the photothermoelectric effect is one of the causes of the photocurrent in this case.

As shown in Fig. 4(d), the distribution of MoS₂ is not uniform. Note the intensity is positively correlated with the density of MoS₂. By comparing Figs. 4(c) and 4(d), the photocurrent signal captured in the white circle corresponds to the intensity of the A_1g band of MoS₂, which proves that the photocurrent is due to the presence of MoS₂. However, it is worth mentioning that figures 4(c) and 4(d) do not correspond to each other exactly, which may be caused by the following reasons. The size of MoS₂ is at nm level, which is smaller than the spatial resolution of Raman mapping, which also affects the interpretation of the experimental results. If the MoS₂ is too small or the density is low, the photocurrent is hard to detect. Therefore, the spatial resolutions of Raman intensity and photocurrent are not perfectly identical. In contrast, no photocurrent is observed on MoS₂/SLG membrane heterostructures, which may be due to the topological defects of the polycrystalline grain boundaries. The topological defects affect the transmission of electrons and reduce the electrical properties of graphene, which eventually affects the collection of photocurrent.

In summary, the MoS₂/graphene heterostructures are prepared and the spectra and photoresponse are systematically investigated. It is observed that MoS₂ can mark the position of graphene grain boundaries, and the PL of MoS₂ is quenched after growing on graphene. The PL peak position of MoS₂ grown on graphene grain boundaries shows a blue-shift due to the lattice mismatch. Besides, the photocurrent signal is observed in MoS₂/hexagonal single-crystal graphene heterostructures without a bias, but not in MoS₂/polycrystalline graphene heterostructures. The electron scattering at the graphene grain boundaries affects the optical response of MoS₂/graphene heterostructures. The effect of graphene GBs on MoS₂/graphene heterostructures may have wide applications in optoelectronics devices.