Introduction

The significance of 2D materials has been manifested by a plethora of fascinating functionality demonstrated in graphene, a monolayer material of carbon atoms1,2,3,4. In contrast to graphene, which is a semi-metal with no bandgap by nature, MoS2 monolayers offer an attractive semiconductor option due to a direct bandgap of 1.8 eV5,6. This non-trivial bandgap makes MoS2 monolayers a better candidate than graphene to revolutionize many electronic and photonic devices that are currently dominated by traditional group IV or III–V semiconductor materials7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23. Considerable efforts have been dedicated to achieve scalable synthesis of high quality MoS2 monolayers that is necessary for the development of practical devices. These include sulfurization of MoO3 or Mo layers pre-deposited on substrates, or thermolysis of Mo compound (NH4)2MoS424,25,26,27,28,29,30,31. However, none of the existing processes provides the capability to exclusively produce uniform MoS2 monolayers over a large area up to centimeters. Additionally, none of these processes is able to precisely control the layer number of MoS2 films. For instance, MoS2 monolayers resulted from these processes always co-exist with by-products of thicker layers. As the properties of MoS2 materials strongly depend on the layer number, this unsatisfactory uniformity and controllability poses a formidable challenge for the improvement of device performance. Here we present a new, self-limiting chemical vapor deposition (CVD) approach that can exclusively grow high-quality monolayer or few-layer MoS2 films over an area of centimeters on various substrates such as silicon oxide, sapphire and graphite. The layer number of the MoS2 films can be precisely controlled as 1, 2, 3 and beyond. The synthesized film shows unprecedented uniformity with no other layers found by thorough characterizations. It exhibits optical and electrical quality comparable to the MoS2 exfoliated from bulk materials and can be transferred to other arbitrary substrates.

Results

We grew MoS2 films at high temperatures (> 800°C) using MoCl5 and sulfur as precursor materials (we found the quality of resulting thin films worse at lower growth temperatures). The precursor materials can react at elevated temperatures to produce MoS2 species, which may subsequently precipitate onto receiving substrates to yield MoS2 films (Fig. S1). The layer number of the resulting MoS2 films can be very precisely controlled by controlling the amount of MoCl5 used in experiments or the total pressure in the synthetic setup (details see Studies of growth mechanism in the Supplementary Information). We have successfully grown monolayer or few-layer MoS2 films on various substrates including silicon oxide, sapphire and graphite. Fig. 1 a–b show optical images of the MoS2 monolayer and bilayer films grown on sapphire substrates. The optical images of MoS2 monolayer and bilayer films grown on silicon substrates with 280 nm thick silicon oxide (SiO2/Si) are also given as insets. Scratches were intentionally introduced to show the color contrast between the thin film and the substrate (no MoS2 film exists in the scratched area). We confirm the chemical composition of the thin films is MoS2 using x-ray photoelectron spectroscopy (XPS) characterizations (Fig. S2). From the optical images, we can find that the thin films are continuous and uniform over a large area. We can also find that the color contrasts of MoS2 monolayer and bilayer films bear little difference on sapphire but are substantially different on SiO2/Si, the monolayer showing pink color while the bilayer more bluish. This substantial difference in color contrast may provide a facile approach to determine the thickness of 2D MoS2 materials.

Figure 1
figure 1

Controlled synthesis of MoS2 monolayer and bilayer films.

(a) and (b), Optical images of the MoS2 monolayer and bilayer films grown on sapphire substrates, respectively. The insets are optical images of the MoS2 monolayer and bilayer films grown on SiO2/Si substrates. The scale bars in the insets are 80 μm. (c) and (d), AFM height profiles for typical MoS2 monolayer and bilayer films grown on sapphire, respectively. The insets show the AFM images from which the height profiles are extracted. A larger version of the AFM images is given in Fig.S3 of the Supplementary Information. (e), Raman spectra of the MoS2 monolayer (1 L) and bilayer (2 L) films grown on sapphire. Also given are the Raman spectra of exfoliated MoS2 monolayer and bulk MoS2 dispersed on SiO2/Si substrates. The two characteristic Raman modes are labeled. The peak at 418 cm−1 is from the sapphire substrate.

Full size image

We characterized the synthesized thin films using atomic force microscope (AFM) and Raman spectroscopy. The thickness of the synthesized MoS2 monolayers and bilayers are shown 0.68 nm and 1.40 nm (Fig. 1c–d), in consistence with exfoliated MoS2 (Fig. S3–S5). Fig. 1e shows the Raman spectra collected from the as-grown MoS2 thin films on sapphire with different layer numbers (the Raman spectra for the MoS2 films grown on SiO2/Si is given in Fig. S6). The Raman spectra of exfoliated MoS2 monolayers and bulk MoS2 materials are also given in Fig. 1e as references. Two characteristic Raman modes can be found in the spectra, the A1g mode associated with the out-of-plane vibration of sulfur atoms and the E12g mode related with the in-plane vibration of Mo and sulfur atoms32,33,34. The full width at half maximum (FWHM) of the E12g peak may be used as an indicator for crystalline quality. The E12g FWHM of the synthesized MoS2 monolayer is 4.2 cm−1, close to that of the exfoliated monolayer, 3.7 cm−1. This suggests a good crystalline quality in the synthesized film. Additionally, the frequency difference (Δk) between the E12g and A1g modes has been known closely related with the layer number and can be used to determine the thickness of MoS2 materials32,33,34. We can find that the frequency difference Δk of the synthesized thin film increases with the thickness (Fig. 1e), consistent with what were observed on exfoliated MoS232,33,34. A systematic correlation between the value of Δk and the layer number of the synthesized film is given in Table S1 of the Supplementary Information. The Δk of the synthesized film is generally larger than that of exfoliated MoS2. For example, the Δk of the synthesized monolayer is found around 20.4 cm−1, different from that of the exfoliated monolayer, 19.4 cm−1. This could be related with certain crystalline imperfection, for example, smaller crystalline grains in the synthesized thin film. Other groups also observed a larger Δk in synthesized MoS2 materials than exfoliated counterparts26.

The synthesized MoS2 thin films show unprecedented uniformity. We can find the as-grown MoS2 thin films continuous and uniform cross an area of centimeters under optical microscope or by eyes, as illustrated by Fig. 1a and 2a. To more quantitatively assess the uniformity over a large area, we divided the monolayer MoS2 film grown on a typical sapphire substrate in size of 1 cm × 3 cm into eight areas (Fig. 2b inset). We measured numerous Raman spectra at each of the areas and plotted the representative results in Fig. 2b. We can find that the peak positions of the E12g and A1g modes remain identical in all the spectra. The profile of the Raman peaks, such as the FWHM, shows negligible difference. This indicates that the synthesized thin film is indeed a homogeneous monolayer with similar crystalline quality across the entire substrate. We also performed Raman mapping over an area of 10 μm × 10 μm in the as-grown monolayer film and plotted the measured frequency difference Δk in a 2D image (Fig. 2c). All the Δk are found in a range of 20.3 – 20.7 cm−1, confirming the film a homogeneous monolayer. To further confirm the uniformity, we performed multiple AFM measurements at different areas on the film and found similar thickness and surface topology in all the AFM images (Fig. S8). A typical AFM measurement (5 μm × 5 μm) is given in Fig. 2d. It shows a continuous and smooth surface (roughness < 0.2 nm) with no step and void observed. Similar uniformity can also be found at the synthesized bilayer MoS2 film (Fig. S9–S10). We should note that the size of the synthesized thin film (1 cm × 3 cm) is limited by the size and temperature uniformity of our current synthetic setup. We believe that by optimizing the synthetic setup we can produce wafer-scale MoS2 monolayer or few-layer films.

Figure 2
figure 2

Large-area uniformity of the synthesized MoS2 monolayer and bilayer films.

(a), A picture for as-grown MoS2 monolayer (1 L) and bilayer (2 L) films, along with a NCSU logo. Underneath the films is a blank sapphire wafer, which is transparent and used as a refence. (b), Raman spectra collected from eight different areas of the MoS2 monolayer film. The inset schematically illustrates how the eight areas distribute across the substrate. (c), Map of the Raman frequency difference Δk collected from an area of 10 μm × 10 μm in the MoS2 monolayer film. All the measured Δk are in the range of 20.3 – 20.7 cm−1. Mapping step: 0.5 μm. (d), Perspective view of a typical AFM image collected from the MoS2 monolayer film. A topview of this AFM image is given as Fig. S7 in the Supplementary Information.

Full size image

The synthesized MoS2 films exhibit a very good crystalline quality as indicted by the Raman measurements (Fig. 1e and 2b). To further elucidate the crystalline structure, we characterized the thin film using aberration-corrected scanning transmission electron microscope (STEM). To do that, we transferred the synthesized film to TEM grids following a process reported in references25. The synthesized film is found as polycrystalline with crystalline grains varying from tens to hundreds of nanometers as indicated by selected area electron diffraction patterns (Fig. S11). We used high angle annular dark field (HAADF) imaging to investigate the atomic arrangement in the synthesized MoS2 monolayer and bilayer films. The monolayer and the bilayer can be identified from the folding edge of the transferred thin film (Fig. 3a inset). Fig. 3a shows typical STEM high angle annular dark field (HAADF) image of synthesized monolayers and corresponding fast Fourier transformation (FFT) pattern. The hexagonal atomic arrangement shown in the STEM image and the FFT pattern indicate that the basal plane of the synthesized thin film is (001), i.e. the c-axis of MoS2 materials perpendicular to the thin film. From the STEM image we can directly measure the lattice constant as a = b = 0.32 nm, consistent with bulk MoS2 materials. As STEM-HAADF imaging is dependent on the atomic number, we can identify the positions of Mo and S atoms in the monolayer by the difference in image contrast (Mo atoms show much higher contrast than S atoms), as shown in Fig. 3b. Additionally, STEM-HAADF imaging provides a credible way to identify the stacking sequence in MoS2 bilayers. We can find a substantial difference between the images of MoS2 monolayer (Fig. 3b) and bilayer (Fig. 3c). This difference indicates an AB stacking sequence in the MoS2 bilayer, where the Mo atoms of one layer overlap with the position of the sulfur atoms in the other layer (Fig. S13). The experimental observation matches the modeled image for the AB stacking very well (Fig. 3c).

Figure 3
figure 3

Characterization of the crystal structure of synthesized MoS2 monolayer and bilayer films.

(a), High angle annular dark field (HAADF) image of typical MoS2 monolayer film. Upper inset is the corresponding FFT pattern, which is not indexed for visual convenience (an indexed version is given in Fig. S12a). Lower inset shows the folded edges of MoS2 monolayer and bilayer films. The lattice constant and important crystalline directions are given as shown. (b) and (c), HAADF of MoS2 monolayer and bilayer films. The image of the bilayer film (c) is filtered and the original version is given in Fig. S12b. The insets show modeled crystal structures of MoS2 monolayer and bilayer films with blue and yellow dots corresponding to Mo and S atoms, respectively. The sulfur atom in the inset of (c) is not given for visual convenience.

Full size image

The synthesized thin films show excellent optical quality that is comparable to exfoliated MoS2. Fig. 4a shows the PL spectra collected from exfoliated and synthesized MoS2 monolayers. Raman peaks of the MoS2 materials are also included and the (photoluminescence) PL spectra are normalized to the intensity of corresponding Raman peaks. The normalization is to better correlate the measured PL intensity with the intrinsic luminescence efficiency of the materials6. Two peaks can be found in the PL spectra, ~ 620 nm and ~ 670 nm and can be correlated to the B1 and A1 excitons of MoS2, respectively6. We can find similar PL peaks from both synthesized and exfoliated MoS2 monolayer. This suggests a comparable optical quality in the synthesized film with the exfoliated MoS2. We should note that the synthesized monolayer for the PL measurement was grown on sapphire substrates. For the convenience of comparison with the exfoliated MoS2, which were dispersed onto SiO2/Si substrates, the synthesized monolayer was transferred to SiO2/Si substrates for the PL measurement (see the effect of substrates on PL in Fig. S14). We find that the PL from the MoS2 monolayer film grown on SiO2/Si substrates is much weaker than the films grown on sapphire (Fig. S15), probably due to a poorer crystalline quality in the films grown on SiO2/Si substrates than those grown on sapphire as reported previously25. Additionally, the PL efficiency of exfoliated MoS2 has been well known substantially decreasing with the increase in the layer number due to an evolution of the bandgap with the layer number. We can find that the PL efficiency of the synthesized MoS2 bilayer is indeed substantially smaller than that of synthesized monolayers (Fig. S16). This suggests that the bandgap of the synthesized MoS2 film bears a similar dependence on the layer number as exfoliated MoS2. To illustrate the uniformity of the optical quality, we performed PL mapping over an area of 20 μm × 20 μm in the synthesized monolayer. The measurement demonstrates a uniform distribution in the PL intensity (Fig. 4a inset), confirming the uniformity as suggested by the Raman measurements shown in the preceding text.

Figure 4
figure 4

Optical and electrical characterizations of the synthesized MoS2 monolayer films.

(a), Photoluminescence (PL) of the synthesized and exfoliated MoS2 monolayers. Both were dispersed on SiO2/Si substrates. The synthesized monolayer was grown on sapphire substrates and then transferred to SiO2/Si substrates for the PL measurement. Using the sample substrate for the PL measurement is for the convenience of comparison. Our experimental results show the PL of MoS2 films may be affected by the substrate (Fig. S14) (b), Ids-Vds curve for a field effect transistor (FET) made with the synthesized MoS2 monolayer with gating voltages Vgs 0, 10, 20, 40 V. Inset, transfer characteristic for the FET with a source drain bias Vds of 2 V.

Full size image

We also evaluated the electrical quality of the synthesized MoS2 monolayer film. We transferred the thin film grown on sapphire to degenerately doped silicon substrates covered with 280-nm-thick SiO2 and fabricated bottom-gate transistors by evaporating 5 nm Ti/200 nm Au electrodes on top of the MoS2 thin film. We performed electrical measurements at room temperature and in ambient environment using the degenerately doped silicon substrate as a back gate and deposited Au electrodes as the source and drain. Fig. 4b shows typical output characteristic and transfer characteristic (inset) curves of the device. These results indicate that the device is an n-type channel field effect transistor, consistent with other works on exfoliated MoS27,35,36. The linear dependence of the current Ids on the source-drain bias Vds suggests that the contacts between the MoS2 thin film and the electrodes are ohmic. This ensures that the observed field effect behavior is from the monolayer MoS2 channel rather than Schottky barriers at the contact. From the measurements, we can derive the field-effect mobility of charge carriers in the device as 0.003–0.03 cm2/V.s by using the expression μ = [dIds/dVg] × [L/(WCiVds)], where L and W is the length and width of the channel, respectively, Ci is the capacitance of the gating oxide and Vg is the bias applied to the back gate. This is reasonably comparable to previous reports on exfoliated monolayers measured in similar conditions (ambient environment, back gating with silicon oxide as the gate oxide)7,35,36. We can thus conclude that the synthesized thin film has a reasonably comparable electrical quality with exfoliated MoS2.

Discussion

The observed remarkable uniformity and controllability in the synthesized MoS2 film strongly suggests that this synthetic process is self-limiting, i.e. the growth may be automatically stopped by some intrinsic factors of the synthetic process. Unlike the self-limiting mechanism for the growth of graphene37, which is related with the limited solubility of carbon source in the catalyst of copper, our experimental results suggest that the partial pressure of gaseous MoS2 species and the interaction of MoS2 thin films with substrates play key roles in this self-limiting mechanism.

We found in experiments that the layer number of MoS2 thin films increases with the amount of precursor materials (Table S1, Fig. S17–S18) and the total pressure in the synthetic setup (Fig. S19–S20). We can correlate this to the effect of the two parameters on the partial pressure of gaseous MoS2 species PMo, which we find increasing with the amount of precursor materials and the total pressure (details see Supplementary Information). We believe that the partial pressure PMo plays an important role in control of the layer number.

To elucidate the role of PMo, we examine the dynamics of the synthetic process and find that the precipitation reaction MoS2(g) → MoS2(s) is the rate-determining step, where g and s refer to gaseous and solid phases, respectively (details see Supplementary Information). The precipitation reaction is governed by two pressures, the partial pressure of gaseous MoS2 species (PMo) and the equilibrium vapor pressure (referred as vapor pressure) of MoS2 in solid phase (PoMo). The role of PMo in control of the layer number is essentially rooted in its role in the thermodynamics and kinetics of the precipitation reaction. The difference between the partial pressure PMo and the vapor pressure PoMo provides the thermodynamic driving force for the precipitation reaction. The partial pressure must be larger than the vapor pressure, i.e., PMo > PoMo, to drive the precipitation of gaseous MoS2 species. Additionally, based on the law of mass reaction, we know that the partial pressure PMo also dictates the rate of the precipitation, a larger PMo driving a faster precipitation.

The observed dependence of the layer number of MoS2 films on experimental parameters (the amount of precursor materials and the total pressure) suggests that a larger partial pressure PMo tends to give rise to a thicker film (Fig. S21–S22). While a larger PMo can cause a larger precipitation rate, we do not think that this kinetic effect would be the major reason for the observed control of layer number. Should the precipitation rate be the major mechanism for the control, it would request a perfect control in timing to stop the precipitation right at the end of the formation of each individual layer, in order to produce the observed uniform MoS2 films with precisely controlled layer number. This is most unlikely in our synthetic setup. The growth of large-area, highly uniform MoS2 thin films in our experiments suggests that this precipitation is a self-limiting process, i.e. the precipitation automatically stops at the end of the formation of each individual layer.

We believe that the self-limiting mechanism may lie in a thermodynamic balance between the partial pressure of gaseous MoS2 species (PMo) and the vapor pressure of MoS2 thin films (PoMo) on the substrate. The vapor pressure of MoS2 thin films could depend on, more specifically, increase with the layer number. As PMo is requested larger than PoMo to provide the thermodynamic driving force, the precipitation of gaseous MoS2 species may be automatically stopped at a specific layer number by control of PMo. For instance, an exclusive growth of MoS2 monolayer could be achieved by controlling the partial pressure PMo between the vapor pressures of MoS2 monolayer (PoMo,1) and bilayer (PoMo,2) films as PoMo,1 < PMo < PoMo,2. In this case, the larger vapor pressure of MoS2 bilayers may automatically prevent the continuous growth once a MoS2 monolayer film is formed, regardless a continuous supply of MoS2 species. This indeed matches our observations that the lay number of MoS2 films shows a step-function dependence on the amount of precursor materials (Table S1). Our hypothesis on the self-liming mechanism can also be supported by the observed dependence of the layer number on growth temperatures. We find that the layer number of MoS2 film increases with the temperature of receiving substrates decreasing (Fig. S23). This can be understood that the vapor pressure of MoS2 thin films PoMo decreases with the temperature and the same partial pressure PMo can drive a deposition of thicker films at lower temperatures.

The dependence of the vapor pressure of MoS2 thin films on the layer number might be related with the interaction with substrates. The vapor pressure indicates a capability of atoms escaping from solid phase into gas phase. The interaction of MoS2 overlayers with substrates could substantially suppress the escaping of MoS2 atoms and subsequently cause a decrease in the vapor pressure. This interaction with substrates is expected to relax with the layer number (Fig. S24). As a result, the vapor pressure of MoS2 thin film may increase with the layer number. Similar suppression effects of substrates on the vapor pressure have been well demonstrated in other materials adsorbed on substrates, for instance, polymers38. This dependence of vapor pressures on the layer number of MoS2 films may provide the possibility of self-limiting growth. We also find in experiments that a precise control of the layer number generally tends to be more difficult for thicker films, for instance, larger than 4, often resulting a mixture of layer numbers. This suggests that the difference between the vapor pressures of MoS2 films with neighboring layer number (PoMo,n+1 - PoMo,n, n refers to the layer number) tends to shrink with an increase in the layer number, which makes the selective growth more difficult.

To further examine the role of the interaction with substrates, we studied the growth on different receiving substrates, the traditional three-dimensional (3D) bonded materials such as sapphire (referred as 3D substrates) and highly ordered pyrolytic graphite (HOPG, from Ted Pella) that has a similar layered structure as MoS2 (referred as 2D substrates). We grew MoS2 films on these different substrates under identical experimental conditions (the different substrates were placed side by side in the tube furnace). We can find a substantial difference between the films grown on sapphire and HOPG (Fig. S25). With the same experimental conditions, while the film grown on sapphire substrates is monolayer (Δk 20.4 cm−1), the one grown on HOPG is bilayer (Δk 22.4 cm−1). We can exclude out the possibility that the observed difference in the Raman shifts is caused by the difference in substrates (see Supplementary Information). This result confirms that the interaction of MoS2 films with substrates plays an important role in the self-limiting growth. The strength of the interaction of 2D material overlayer with conventional 3D substrates e.g. sapphire, is well known different from that with 2D substrates (such as HOPG). This difference has been extensively manifested by the van del Waals epitaxial growth of layered materials on 2D substrates39. In our experiment, the different interactions of MoS2 films with sapphire and HOPG might cause different vapor pressures. Therefore, the same partial pressure of MoS2 can give rise to a growth of thicker MoS2 film on HOPG.

In conclusion, we demonstrate a new chemical vapor deposition (CVD) approach that can provide capabilities to produce large-area, high quality MoS2 monolayer and few-layer films with unprecedented uniformity and controllability. While more studies would be necessary to better understand the growth mechanism, our experimental results strongly suggest that this synthetic process is self-limiting dictated by the balance between the equilibrium pressure of MoS2 thin films and the partial pressure of MoS2 species in the vapor phase. As a final note, this synthetic approach can be readily scaled up to produce wafer-scale 2D MoS2 for the development of practical electronic and photonic devices. It is poised to enable 2D MoS2 as a useful functional material to revolutionize a wide range of fields including information technology, optoelectronics, chemical/biological sensing and spintronics.

Methods

MoS2 thin films were synthesized in a tube furnace as illustrated in Fig. S1. In a typical growth, 1–50 mg of molybdenum chloride (MoCl5) powder (99.99%, Sigma-Aldrich) was placed at the center of the furnace and 1 g of sulfur powder (Sigma-Aldrich) were placed at the upstream entry of the furnace. Receiving substrates were placed downstream in a distance of 1–7 centimeters away from the center of the furnace. In typical experiments, the furnace was heated up at a rate of 28°C/min under a flow of Ar gas, held at a setting temperature for 10 min and then naturally cooled down to room temperature. Typical conditions for high-quality MoS2 thin film growth include a temperature of 850°C (the temperature for sulfur source is estimated ~ 300°C), a flow rate of 50 sccm and a pressure around 2 Torr. The layer number of the synthetic MoS2 thin film can be controlled by control of the amount of MoCl5 precursor or the total pressure in the synthetic setup (see Supplementary Information).

The crystal structure of the resulting products were analyzed with transmission electron microscope (TEM, JEOL-2010) and probe aberration-corrected scanning transmission electron microscope (STEM, FEI Titan 80–300 STEM) operated as 200 KV under STEM mode. The as-grown thin film was transferred to TEM grid for the structure characterization. The thickness and surface topology were measured using atomic force microscope (AFM, Veeco Dimension-3000). Raman mapping and photoluminescence (PL) measurements were carried out using Horiba Labram HR800 Raman Microscopy with an excitation wavelength of 532 nm. X-ray photoelectron spectroscopy were performed on SPECS XPS using an Mg Kα x ray source. The field-effect transistor device was fabricated by evaporating Ti/Au (5/200 nm) electrodes directly onto top of MoS2 films transferred to degenerately doped Si substrates with 280 nm thick silicon oxide. A copper grid (100mesh, 30 micron spacing, Ted Pella) was placed on top of the thin film as mask for the electrode fabrication. This gives a relative large channel with a length L = 30 μm and a width W = 230 μm. The electrical measurements were performed in ambient conditions using a probe station (Karl Suss PSM6).