Chemical Vapor Transport Deposition of Molybdenum Disulfide Layers Using H₂O Vapor as the Transport Agent

Abstract

Molybdenum disulfide (MoS₂) layers show excellent optical and electrical properties and have many potential applications. However, the growth of high-quality MoS₂ layers is a major bottleneck in the development of MoS₂-based devices. In this paper, we report a chemical vapor transport deposition method to investigate the growth behavior of monolayer/multi-layer MoS₂ using water (H₂O) as the transport agent. It was shown that the introduction of H₂O vapor promoted the growth of MoS₂ by increasing the nucleation density and continuous monolayer growth. Moreover, the growth mechanism is discussed.

1. Introduction

Molybdenum disulfide (MoS₂) layers, having unique optical and electrical properties, have attracted extensive interest in the fields of energy generation, electronics, and sensors [1,2,3,4,5,6,7]. The growth of large-scale, high-quality MoS₂ layers targeted for silicon integrated device fabrication is still challenging. Vapor deposition has been the predominant method for the growth of large-scale, continuous MoS₂ monolayer or few layers films in recent years [8,9,10]. Molybdenum oxides and sulfur are generally used as precursors of MoS₂. For example, Lee et al. heated MoO₃ powder in sulfur vapor and obtained MoS₂ monolayer and multi-layer films [11]. In this method, MoO₃ was first reduced by sulfur vapor to form MoO_3−x, which was then further reacted with sulfur vapor to form MoS₂ [12]. MoO₃ acted as a nucleation center promoting crystal growth as well as the introduction of crystal defects. The introduction of defects plays two important roles; one is to promote nucleation for multi-layer growth, and the other is to tailor the electrical properties [13,14,15,16]. The MoS₂ domains grown with this method showed different morphologies, e.g. triangle, hexagon, three-point star, as a function of the different atomic ratio of sulfur to molybdenum [17,18].

MoS₂ powder is another commonly used starting material. Wu et al. [19] heated MoS₂ powder at 900 °C in the center of a tube furnace and obtained a MoS₂ monolayer on an insulating substrate downstream of the precursor in a lower temperature zone (~650 °C). The usage of single-precursor MoS₂ powder as the source of Mo and S avoided the introduction of impurities and heterogeneous nucleation during the growth of MoS₂ flakes. Therefore, the MoS₂ monolayer showed a regular triangular shape and high optical quality. This vapor-solid growth method is suitable for the deposition of high-quality monolayer single crystal flake. However, in our recent work, we found that the nucleation is difficult to initiate and the growth temperature window is very narrow, ca. ~50 °C [20]. These issues could be attributed to the low vapor pressure of MoS₂ powder. The chemical vapor transport method was generally used for the growth of crystals with a solid precursor that has low vapor pressure. For example, Pisoni et al. reported the growth of MoS₂ single crystals using I₂, Br₂, and TeCl₄ as transport agents [21]. The transport agent converts MoS₂ into high vapor pressure intermediates, which undergo the reverse reaction to deposit MoS₂ onto the substrate. However, the vapor transport agents used in this study are highly toxic and reactive, which could limit their widespread use.

To overcome this limitation, we have investigated ways to improve the nucleation density of MoS₂ using various additives. In this paper, we report the chemical vapor transport growth behavior of MoS₂ monolayer or multi-layer films by using MoS₂ powder as the precursor and water (H₂O) vapor as the transport agent. In the nucleation stage, H₂O vapor was introduced into the deposition system and acted as a chemical transport agent. Our mechanistic study suggests that water reacted with MoS₂ to form MoO₂, which promoted the nucleation of MoS₂. In the previously mentioned growth methods, the sulfur comes from the sublimation of sulfur or MoS₂ powder and the sulfur flow rate is out of control. Here, the sulfur was formed through the reaction of MoS₂ and water, which provides us a possible way to adjust the sulfur flow rate by controlling the water vapor flow rate. In the second stage, H₂O vapor was cut off and MoS₂ continuously grew through a simple physical vapor transport process. This novel approach combined the heterogeneous nucleation and homogeneous growth to control the crystal size and thickness of the MoS₂ layer. The thickness of the MoS₂ film obtained ranged from a monolayer to multiple layers. The lateral size of the single-crystal domain is up to 300 μm.

2. Materials and Methods

MoS₂ Layers Synthesis

MoS₂ was prepared by modifying a previously reported vapor deposition method using a silicon wafer with a 300-nm layer of oxide (SiO₂/Si) as the substrate [19]. The schematic of the vapor deposition setup is shown in Figure 1. MoS₂ powder (99.5% purity, Aladdin, Shanghai, China) was used as the precursor. Before use, the precursor (0.5 g) was loaded into a small quartz glass boat (70 mm in length) and put in the center of the tube furnace (1 inch in diameter, Hefei Kejing Materials technology Co. Ltd., Hefei, China). Before growth, the precursor was flushed under Ar/H₂ (70 sccm, H₂ 5%, total pressure of 75 Torr. sccm: standard cubic centimeter per minute) for 10 min at room temperature to remove the air and water absorbed on the precursor. The substrate was put downstream close to the furnace wall.

For the MoS₂ growth, the precursor was heated to 1000 °C from room temperature in 30 min under Ar/H₂(75 Torr, Ar/H₂ 70 sccm) and kept at 1000 °C for 1 h. The furnace was then turned off and cooled from 1000 °C to room temperature. During the above heating process, the temperature of the substrate ranged from 710 to 850 °C. The water (H₂O) vapor was introduced into the furnace by turning on/off the water valve, which connects the water tube and the Ar/H₂ inlet. For typical growth, the water valve was kept open during the whole heating stage. For studies on the influence of H₂O on the growth of MoS₂, we kept the valve open during the heating stage but limited the water exposure during the synthesis.

Optical microscope imaging of the sample was conducted with a Jiangnan MV3000 digital microscope (Nanjing Jiangnan Novel Optics Co. Ltd., Nanjing, China). Tapping mode atomic force microscopy (AFM) was performed on an Agilent 5500 (Agilent Technologies, Palo Alto, CA, USA) in air. Raman spectrum and photoluminescence (PL) were acquired on a Renishawin Via micro-Raman spectroscope (Renishaw, London, UK) with a 532 nm solid-state laser at room temperature. X-ray diffraction (XRD) was carried on a Thermo ARLXTRA (Thermo Electron, Waltham, USA) and ultraviolet visible diffuse reflection spectroscopy (UV-Vis DRS, not including specular reflection) was performed on Shimadzu MPC-3100 (Shimadzu, Tokyo, Japan) with an integrating sphere.

3. Results and Discussion

3.1. MoS₂ Flakes Grown in the Presence of H₂O Vapor

MoS₂ flakes were prepared on the substrate using H₂O and MoS₂ powder as illustrated in Figure 2a. Figure 2b shows the separated triangular MoS₂ flakes grown on the SiO₂/Si substrate with the H₂O vapor valve kept open during the heating of the furnace and growth of the MoS₂ flakes. The thickness of the flakes ranged from monolayer to multiple layers. The triangles in dim and uniform color indicate the uniform monolayer MoS₂. The bright color triangles are attributed to multi-layer MoS₂ with a pyramid-shape structure. The flake lateral size ranged from ca. 20 to 40 μm.

The success of the growth and thickness of the MoS₂ flakes were confirmed by Raman spectroscopy. Figure 2c displays the typical Raman spectra of monolayer and multi-layer MoS₂ flakes corresponded to the images in Figure 2b. The E_2g and A_1g modes of MoS₂ were observed. The frequency difference between the E_2g and A_1g mode is thickness-dependent. With the increase of the layer number, the frequency difference valve will increase. The E_2g and A_1g peaks positions are at 385.0 cm⁻¹ and 404.1 cm⁻¹ (383.3 cm⁻¹ and 409.1 cm⁻¹) with a frequency difference of 19.9 cm⁻¹ (25.8 cm⁻¹), indicating that the thickness of flakes is monolayer (multi-layer) [22].

Besides Raman spectra, PL is generally used for the identification of the thickness of the MoS₂. Mak et al. studied the relationship between the PL quantum yield and layer number. They found that the PL quantum yield drops quickly with the increase of the layer number. Bulk MoS₂ is an indirect-gap semiconductor showing negligible PL. Few-layer MoS₂ shows weak PL due to the confinement effects. Monolayer MoS₂ is a direct-gap semiconductor giving out bright PL [23]. Figure 2d shows the typical photoluminescence spectra (PL) both of the monolayer and multi-layer MoS₂ flakes corresponded to the images in Figure 2b. The excitation wavelength was 532 nm. The PL peaks of monolayer MoS₂ are located at 674.5 nm and 622 nm, which are attributed to the A1 and B1 direct excitonic transition emission of the MoS₂ monolayer, respectively [9,17,24]. We observed that the PL intensity of the monolayer is much stronger than that of the multi-layer sample.

3.2. Effect of H₂O Vapor on the MoS₂ Layers Growth

To investigate the effect of H₂O vapor on the MoS₂ growth, we limited the time the synthesis was exposed to water vapor. After the precursors were heated to 1000 °C, the water valve was closed after a fixed amount of time during the growth stage: Figure 3a–d 0 min (least water exposure), Figure 3e–h 10 min, Figure 3i–l 20 min, and Figure 3m–p 60 min (most water exposure).

Shown in Figure 3a–d, the shape of the MoS₂ prepared without the introduction of H₂O is a separated island. Meanwhile in Figure 3e,i–k,m–o, continuous, large-area MoS₂ films were observed. This may be due to the presence of H₂O vapor, which enhanced the diffusibility of molybdenum and sulfur atoms at domain boundaries, resulting in the continuous growth of monolayer MoS₂. The large optical contrast in Figure 3e–p indicates the formation of multiple layers and/or clusters, which may be due to the formation of high heterogeneous nucleation density and the Stranski-Krastanov growth mode. The formation of heterogeneous nucleation will be discussed below. Besides the continuous film obtained as described above, the domain size of MoS₂ prepared with H₂O (shown in Figure 3f–h,p) was larger than those prepared without H₂O (shown in Figure 3b,c). Figure 4 shows the magnified optical image of the same sample that is shown in Figure 3e. The lateral size of the triangle-shaped MoS₂ flakes ranges from 24 μm to 372 μm. The average lateral size of the MoS₂ flakes prepared without H₂O introduction was 13 ± 6 μm, while the average size increased to 159 ± 80 μm based on the statistical calculation of the size of the isolated flakes shown in Figure 3a–h, respectively.

The water introduction also has an effect on the thickness of MoS₂ flakes. Based on the frequency difference (24.7 cm⁻¹, Figure S1) between the E_2g and A_1g mode of MoS₂ and uniform color contract, we can conclude that the MoS₂ flakes prepared without water exposure in Figure 3a–d is multi-layer. In contrast, in those samples prepared in the presence of water (Figure 3e–p), monolayer MoS₂ was observed (as discussed at the end of Section 3.2). Therefore, the introduction of water can reduce the thickness of the MoS₂ flakes.

It is reported that water molecules and carbon atoms can intercalate between the two-dimensional material and the substrate [25,26,27]. Although we do not have enough evidence to show the presence of the water intercalation in our sample at high growth temperature (710 °C to 850 °C), we suspect that the molecular structure of water vapor possibly intercalates into the interlayer of the MoS₂ flakes or the interface between the MoS₂ and SiO₂/Si substrate, which affects the absorption, desorption, and diffusion of the precursor atoms and even the final monolayer growth.

From Figure 3m–o, we can observe some bright features. The white spots are multi-layer MoS₂. The area with green and yellow color we suspected to be amorphous MoS₂, MoO₂, or even organic contamination. To reduce the organic contamination, the silicon wafer substrate was cleaned with hot piranha solution (7:3 concentrated H₂SO₄:35% H₂O₂) for 10 min at room temperature, and the vapor deposition system was flushed under Ar/H₂ to remove air-borne contamination before MoS₂ growth.

Figure 5 shows the typical UV-Vis DRS of the MoS₂ film corresponding to the images shown in Figure 3e–h. The UV-Vis DRS peaks at 665 and 610 nm match the two PL emission peaks (Figure 2d), and are due to the characteristic A1 and B1 direct excitonic transitions of MoS₂, respectively [28].

AFM is a commonly used technique for two-dimensional material thickness measurement. Here, we conducted multiple scans of the thickness of the monolayer MoS₂ at an edge of the MoS₂ flake by AFM. Figure 6 shows the typical AFM image of the edge of the monolayer MoS₂ triangle shown circled in black in Figure 3h. A straight trench with a width of ca. 150 nm was observed on the substrate surface, which divided the substrate into two sections. The bottom of the trench is the SiO₂/Si substrate. The left side of the trench is MoS₂ particles, and the right side of the trench is monolayer MoS₂. The thickness of the MoS₂ flake is 0.9 ± 0.1nm (Figure 6, Figure S2, and Table S1), indicating that the flake is monolayer. This thickness value, although it significantly deviates from the expected thickness of monolayer MoS₂ (0.615 nm), is consistent with other AFM measurements of single-layer MoS₂ deposited on a SiO₂ substrate [29,30]. In fact, the discrepancy that the measured value by AFM is larger than the theoretical value is common phenomena in the measurement of the thickness of two-dimensional monolayer materials, such as graphene [31]. The discrepancy was attributed to the instrument offset due to tip-substrate interaction as well as adsorbed molecules between the monolayer and the SiO₂ substrate [26,31]. From the AFM image in Figure 6, we can see that monolayer MoS₂ is smooth and continuous. We measured the root-mean-square (RMS) surface roughness over a 1 μm × 1 μm area. The RMS was 0.22 nm. The trench is probably formed through the rapid diffusion of MoS₂ nucleation along the direction perpendicular to the domain edge. Detailed study of the trench will be reported in future study. In addition to the trench, there are also many white particles on the surface of the MoS₂ flake. These particles should be MoS₂ formed during the growth of the MoS₂ flake, or even contaminations formed during the transport of the sample.

3.3. Mechanism of MoS₂ Growth in the Presence of H₂O Vapor

The results obtained in Figure 3 suggest that H₂O vapor promoted the growth of MoS₂ film. We hypothesized that the H₂O vapor reacted with MoS₂ powder to give molybdenum oxide. Then the molybdenum oxide evaporated and deposited on the substrate, acting as heterogeneous nucleation center, from which the molybdenum oxide reacted with sulfur at a lower temperature and transformed into the MoS₂ layer [12]. The following reactions should have occurred during the growth of MoS₂ [32]:

$${\mathrm{MoS}}_2+4{\mathrm{H}}_2\mathrm{O}\stackrel{1000°\mathrm{C}}{\to }{\mathrm{MoO}}_2+{\mathrm{H}}_2\mathrm{S}+{\mathrm{SO}}_2+3{\mathrm{H}}_2$$

(1)

$$2{\mathrm{H}}_2\mathrm{S}+{\mathrm{SO}}_2\stackrel{∆}{\to }3\mathrm{S}+2{\mathrm{H}}_2\mathrm{O}$$

(2)

$${\mathrm{MoO}}_2+2\mathrm{S}+2{\mathrm{H}}_2\stackrel{∆}{\to }{\mathrm{MoS}}_2+2{\mathrm{H}}_2\mathrm{O}$$

(3)

To verify this hypothesis experimentally, we used XRD to test the composition of the precursor annealed at 1000 °C for 20 h in H₂O vapor and H₂/Ar atmosphere. We indeed found that all of the XRD peaks in Figure 7 were indexed according to the monoclinic molybdenum dioxide (MoO₂) (JCPDS NO. 00-032-0671). This result agrees with our hypothesis that molybdenum oxide was formed. The growth process essentially is a chemical vapor transport process. The H₂O vapor acts as transport agent.

4. Conclusions

In summary, we have successfully prepared monolayer/multi-layer MoS₂ through a H₂O vapor-modified vapor deposition method on a SiO₂/Si substrate. The growth of MoS₂ is highly sensitive to the presence of H₂O. The results reveal that H₂O increases the nucleation density of MoS₂ flakes. The Raman, PL, and AFM revealed that both monolayer and multi-layer MoS₂ were formed. Under extended water exposure, a continuous MoS₂ film was formed. Using XRD, we showed that MoO₂ was formed by the reaction between MoS₂ and water, which resulted in the observed enhancement in the nucleation and growth.