1. Introduction
As one kind of transition metal dichalcogenide (TMD) [
1], molybdenum disulfide (MoS
2) is the best known material for two-dimensional (2D) crystal research after graphene [
2]. With its reduced number of layers, MoS
2 exhibits many excellent properties [
3,
4], such as good optical transparency [
5], high electron mobility (up to 200 cm
2/V·s), and direct band-gap structure (Eg = 1.8 eV) [
6]. It can be employed to fabricate field effect transistors (FETs) with a high current on/off ratio [
5,
7], sensitive photodetectors [
8,
9], light emitting diodes (LEDs) [
10,
11], and heterojunction solar cells [
12,
13]. Thus, due to its unique semiconductor properties and wide applications, 2D MoS
2 attracts great attention. It is considered a potential candidate in atomic-scale semiconductor science [
14,
15].
Recently, the main preparation methods of MoS
2 have included hydrothermal synthesis [
16], tape auxiliary mechanical exfoliation [
17], liquid-phase exfoliation [
14], physical vapor deposition (PVD) [
18], and chemical vapor deposition (CVD) [
19]. Compared with other methods, CVD is an efficient method to massively synthesize an MoS
2 coating. The CVD method can also alter the shape of MoS
2 domains from triangular nanosheets to continuous films by controlling synthesis parameters, such as the quantity of the reactants, the temperature of precursors, and the carrier gas flow rate. Previous studies have proven that the nucleation density of MoS
2 played a key role in the deposition process, leading to the quality and shape control of MoS
2 domains [
20,
21,
22,
23,
24,
25]. It is reported that, before the growth of MoS
2, adding a “seed layer” on the substrates can adjust the nucleation density of MoS
2 and control the shape of MoS
2 domains [
20,
21,
22]. With its hexagonal lattice structure, graphene can be chosen as a seed layer to form MoS
2 nuclei. However, the pre-treatment process of adding a seed layer requires the addition of steps to the CVD method, and part of the seed material is toxic [
21]. The rest of the seed compound will introduce an unfavorable factor in changing the electronic or optical characteristics of MoS
2 samples. Hence, it is necessary to find a simplified process to control the nucleation density of MoS
2.
In this paper, we propose a simple CVD method at atmospheric pressure without hydrogen which can increase formula flexibility by controlling the heating temperature of MoO3 powder and sulfur powder. The results show that the size of MoS2 grown domains is 10 μm with fast growth. The shape of MoS2 domains vary widely, from discrete darts and triangles to continuous film, on substrate. We deduce that the carrier flow rate distribution in the quartz tube leads to a variation in the nucleation density of MoS2, resulting in the eventual shape distribution of MoS2 domains. Therefore, controlling the carrier gas flow rate can be an effective approach to controlling the shape and coverage of MoS2 domains in the CVD method. These results are of significance for understanding the growth regulation of 2D MoS2.
3. Results and Discussion
The crystal features of MoS
2 grown on the SiO
2/Si substrates were analyzed. As shown in
Figure 2a, MoS
2 nanosheets are successfully deposited on the SiO
2/Si substrate. According to the optical contrast of MoS
2 nanosheets [
23], the film thickness in the inner position of the sample is relatively thinner than that in the edge position. In order to further investigate the surface morphology of the sample, we used SEM to examine the MoS
2 nanosheets. In
Figure 2b, we can find a distinct layered effect where the color depth in the edge position of the sample is deeper than that in the internal position. This is similar to the thickness distribution of MoS
2 nanosheets observed using OM.
To further confirm the number of layers, we chose two spots in the sample to be characterized by Raman and PL spectroscopy. One spot (blue spot) is in the internal position (region 1) and the other spot (red point) is in the edge position (region 2). It was found that there were two obvious Raman peaks in
Figure 2c. E
12g represents the in-plane vibrational mode between the molybdenum atom and the sulfur atom. A
1g stands for the out-of-plane vibrational mode between sulfur atoms [
24]. Δk, the Raman frequency difference between E
12g and A
1g, can determine the number of layers [
25]. The two Raman peaks of the blue spot are located at 384.14 and 404.96 cm
−1, so the Δk is 20.82 cm
−1. This Δk corresponds to monolayer MoS
2 [
26]. Similarly, in the red spot, the two Raman peaks are located at 384.30 and 409.01 cm
−1, and the Δk is 24.71 cm
−1, which corresponds to few-layer MoS
2 [
15,
26]. With the decreased number of layers, the band-gap of MoS
2 gradually shifts from the indirect band-gap to the direct band-gap. In terms of
Figure 2d, the PL spectra of monolayer MoS
2 in the blue spot, we can see two resonant points at 678.5 nm (1.82 eV) and 622 nm (1.99 eV). The two resonant points correspond to A
1 (the maximal splitting valence band) and B
1 (the minimum conduction band), the direct exciton transition of monolayer MoS
2. The PL spectra was fitted with Gaussian curves. The full width at half maximum (FWHM) of peak at 678.5 nm is 30.2 nm and that of 622 nm is 23.8 nm. On the other hand, the PL spectra of few-layer MoS
2 in the red spot show weak PL intensity. The A
1 peak is at 672 nm (1.84 eV), and the B
1 peak is at 622 nm (1.99 eV). Furthermore, we used AFM to measure the thickness of the sample. According to the measurement results shown in
Figure 2e, the height (marked with a white line) between the internal position and the edge position of the sample is h = 2.4 nm. The height (labeled with a white line) between the edge position and the substrate is H = 3.3 nm. Thus, the height between the internal position and the substrate is about 0.9 nm, which is consistent with the thickness of monolayer MoS
2 [
27].
The size and coverage of the MoS
2 domains are highly dependent on the distribution of the samples in the spatial location of the substrate [
27]. To better observe this phenomenon, we created an XY-coordinate system where the bottom left corner of a substrate (
Figure 3a) is taken as the origin O. As shown in
Figure 3b, the y-axis is along the airflow and the x-axis is vertical the airflow. In the rectangular coordinate system, at y = 1.0 mm, 10 points were selected on the x-axis (in the direction of vertical airflow) for observation. According to the distribution characteristics of MoS
2 domains, nine representative images were selected for display, as shown in
Figure 3c–k. At x = 0.1 mm (
Figure 3c), owing to the low evaporation concentration of MoO
3, only small black nuclei appeared on the SiO
2/Si substrate. At x = 1 mm (
Figure 3d), the generated MoS
2 domain appeared as small triangles and darts discretely. The triangular side of the largest domain reached up to 7 μm. At x = 1.5 mm (
Figure 3e), with a larger domain size, regular triangles were formed. The side length of the largest triangular domain is about 15 μm. At x = 2.5 mm (
Figure 3f), it is observed that part of MoS
2 triangles are connected together to form some irregular film. The largest side of the triangular domain in this area is above 20 μm. Furthermore, large-scale MoS
2 film has continuous coverage in the range of x = 3.5 to 13.5 mm (
Figure 3g). As depicted in
Figure 3h–k, contrary to the distribution in
Figure 3c–f, the size of MoS
2 samples decreases with the further increase of x. From y = 0 to 6.2 mm, a similar distribution of the MoS
2 samples can be found along the x direction.
The differences of MoS
2 domains in the direction vertical to the airflow (x-axis) are shown distinctly in the above OM images. Along the x-axis, the shapes of MoS
2 were changed in the following order: small nucleated particles, small triangles, larger triangles, then large-sized film. Then, the film size gradually shrunk and became sparser. From
Figure 3c–f, it can be inferred that that the vapor concentration of MoS
2 increased continuously along the positive direction of the x-axis before MoS
2 film formation. In
Figure 3d, MoS
2 domains would start growing from hexagonal nuclei with three Mo-zz and three S-zz sides. In this area, the Mo:S ratio condition was <1:2 so that small triangles and darts formed. From
Figure 3f–h, it can be seen that a sufficient supply of MoS
2 vapor results in large triangles and continuous film. In this area, the Mo:S ratio condition was ≥1:2 [
27]. After the vapor concentration of MoS
2 reached the maximum (i.e., filming phenomenon occurring), the MoS
2 layer began to become discontinuous with the growth of the x-axis (
Figure 3h–k), presenting a relatively sparse, discrete distribution of triangular MoS
2 films. Meanwhile, there was a shrinkage in size and quantity with respect to these triangles. Based on the above analysis, we can deduce that there was an obviously a gradient distribution of MoS
2 domain size in a cross-section vertical to the direction of airflow due to the difference of MoS
2 vapor concentration.
To explore the size distribution rule of the MoS
2 domain on the substrate, we chose five sections, as shown in
Figure 3e–i. Each section had the same area (20,164 μm
2) and labeled as Sections 1–5. The number of effective nucleation points (i.e., the nucleation points with MoS
2 geometric area greater or equal to 0.5 μm
2) and the nucleation density (i.e., the number of effective nucleation points per unit area) within the chosen section were statistically measured. According to statistical numbers in
Table 1, the highest nucleation density is in Section 3, similar to the optical micrograph in
Figure 3g. This area has the largest size of MoS
2 film. Thus, the distribution rule of MoS
2 domains on the substrate can be summarized as follows. For the same substrate along the direction vertical to the airflow, the nucleation density is related to the distance of the midcourt-line position of the substrate. The size of MoS
2 thin film is larger as the distance is closer. Inversely, the larger the distance is, the smaller the size of the MoS
2 thin film is and the fewer nucleation density of MoS
2 is.
To explain the relationship between the nucleation density of MoS
2 and the size distribution of MoS
2 domains, we employed the laminar flow theory to analyze the airflow distribution. The largest speed of carrier airflow occurred in the center of the quartz tube, and the speed near the inwall of the quartz tube is close to zero [
28,
29]. Thus, along the direction vertical to the airflow, the speed of the carrier gas (Ar) is larger at the center surface of substrates than that their side. Since faster carrier gasses can transport more reactants in the same time, the area with a faster carrier gas can form more effective nucleation points, resulting in an increase of the crystal growth size [
30]. As depicted in
Figure 3g, the nucleation density also increased in the center region of substrates and MoS
2 film formed. On the contrary, the area with a slower carrier gas can form less effective nucleation points, resulting in a decrease of nucleation density. As shown in
Figure 3g, it is noted that a high nucleation density will increase growth points in the center area. An abundant supply of MoS
2 vapor will make isolated MoS
2 domains connect together. As shown in
Figure 3e,i, when close to the side of the substrates, there is lower nucleation density, and enough MoS
2 vapor enables the growth of larger sized triangles in this area. Furthermore, when at the side of the substrates, the lack of nuclei and vapor of MoS
2 led to small triangles and discrete darts. Thus, domain morphology is highly dependent on nucleation density [
31]. Thus, controlling the speed of carrier gas will be an effective approach for regulating nucleation density. Using this approach can also adjust the formation of MoS
2 domains [
19,
22].
To further explore the influence of carrier gas flow rates on the nucleation density, we prepared MoS
2 samples under the same conditions with different carrier gas flow rates, from 10, 40, 80, 120, 160, 200, to 240 sccm. Optical micrographs of MoS
2 samples in the center point (x = 7.5 mm, y = 7.0 mm) of substrates with different carrier gas flow rates are shown in
Figure 4a–h. The relationship between different carrier gas flow rates and their corresponding nucleation density is shown in
Figure 4i. While the gas flow rate was 10 sccm, the nucleation density of MoS
2 was 0.0061 N/μm
2. Only MoS
2 nuclei were found on substrates (
Figure 4a). Although low gas flow rates lead to high concentrations of S vapor to fully sulfurize MoO
3, it suffers from low transfer efficiency of the MoS
2 vapor. As a result, few MoS
2 nuclei can be deposited on substrates. By increasing gas flow rates in the range of 0–160 sccm, the concentration of S vapor decreased, but the concentration of MoS
2 vapor was still enough. Therefore, the nucleation density of MoS
2 increased. Thus, the growth of MoS
2 was promoted in low flow rates. While the gas flow rate was 160 sccm, nucleation density reached the top value of 0.2912 N/μm
2 and large-scale films of MoS
2 were formed (
Figure 4e). Even gas flow rates further increased from 160 to 280 sccm. The concentration of S vapor was not enough to maintain reactions of MoS
2 synthesizing [
32]. Therefore, the nucleation density of MoS
2 decreased and the shape of MoS
2 domains changed from disconnected film, to large triangles, then to small triangles (
Figure 4f–h). Thus, the growth of MoS
2 was suppressed in high flow rates. Finally, according to these experiment results, controlling the carrier gas flow rate can also control the shape and coverage of MoS
2 domains.