The growth scale and kinetics of WS2 monolayers under varying H2 concentration

The optical and electronic properties of tungsten disulfide monolayers (WS2) have been extensively studied in the last few years, yet growth techniques for WS2 remain behind other transition metal dichalcogenides (TMDCs) such as MoS2. Here we demonstrate chemical vapor deposition (CVD) growth of continuous monolayer WS2 films on mm2 scales and elucidate effects related to hydrogen (H2) gas concentration during growth. WS2 crystals were grown by reduction and sulfurization of WO3 using H2 gas and sulfur evaporated from solid sulfur powder. Several different growth formations (in-plane shapes) were observed depending on the concentration of H2. Characterization using atomic force microscopy (AFM) and scanning electron microscopy (SEM) revealed etching of the SiO2 substrate at low concentrations of H2 and in the presence of an Ar carrier gas. We attribute this to insufficient reduction of WO3 during growth. High H2 concentrations resulted in etching of the grown WS2 crystals after growth. The two dimensional X-ray diffraction (2D XRD) pattern demonstrates that the monolayer WS2 was grown with the (004) plane normal to the substrate, showing that the WS2 conforms to the growth substrate.

and WSe 2 9 and tuning of the band gap through uniaxial strain 10 . Stacks of different TMDCs have been shown to create almost a novel material through the creation of new shared exciton states 11 and through femtosecond scale charge separation 12 . TMDCs have also been explored for photodetectors, optical modulators, bio-imaging devices, mode-locked lasers, ultrafast saturation, and solar cells by using TMDCs alone or in graphene/TMDC heterostructures [13][14][15][16][17][18] . WS 2 has been less investigated than MoS 2 simply because of the greater difficulty in producing samples through exfoliation. WS 2 , like MoS 2 , is of interest for optoelectronics because of its direct band gap in the visible range and high absorption relative to its thickness 19,20 . WS 2 monolayers have strong PL emission, stronger than other TMDCs such as MoS 2 21 . WS 2 also exhibits strong spin-orbit coupling and band splitting due to spin enabling spintronics/valleytronics, which was first demonstrated in MoS 2 [22][23][24] . WS 2 also has high nonlinear susceptibility, suggesting its use for nonlinear optical devices 25,26 . However, most device research to date has been largely based on mechanically exfoliated layers which does not allow high throughput manufacturing. Large scale deposition (polycrystalline chip scale or wafer scale growth, similar to what has already been demonstrated in graphene 27 and MoS 2 28 ) and large grain size WS 2 monolayers are essential for further application research and eventual commercialization. To date, Scientific RepoRts | 5:13205 | DOi: 10.1038/srep13205 CVD crystal growth of WS 2 has been shown to produce single crystal flakes hundreds of micrometers in size 29 .
Here we demonstrate polycrystalline WS 2 monolayer growth up to mm 2 coverage and show the effect of H 2 concentration on the crystal size, nucleation density, total areal density, and growth formation. Growth is performed by low pressure (LP) CVD from solid WO 3 and S sources. We utilize Raman spectroscopy, optical microscopy, scanning electron microscopy (SEM), atomic force microscopy (AFM) and two dimensional X-ray diffraction (2D XRD) to characterize growth formation, crystallinity, substrate orientation, existence of monolayer growth, and height of the deposited WS 2 layers. We furthermore demonstrate etching of WS 2 or etching of the SiO 2 substrate and explain these two effects in terms of the reaction chemistry. Finally, we report on a newly observed growth mode for single-crystal WS 2 monolayers and propose a sequential growth model to explain our observations. Figure 1a is an illustration of our experimental setup. The experiment was contained within a 3′ ′ diameter quartz tube with a mechanical pump at one end and gas (H 2 and Ar) introduced at the other with a controlled flow rate. The operating pressure during growth was approximately 9 Torr with the H 2 / Ar mixture gas and 5 Torr with only H 2 gas delivered. The center of the tube rests in a furnace which was raised to 900 °C and placed in the tube were two stacked substrates facing each other, one for the tungsten (W) source (WO 3 evaporated by electron beam onto silicon) and for growth (an oxidized silicon substrate). Sulfur was provided by evaporation from solid powder which was placed in a ceramic crucible outside of the central furnace region, at a lower temperature. For large area growth, the sulfur position was such that it starts to evaporate when the substrate is between 800 and 850 °C. Hydrogen was introduced at 650 °C. Growth proceeded for thirty minutes, after which the furnace was cooled and hydrogen delivery ceased. Figure 1b-e are SEM images of several WS 2 monolayers grown with 30, 40, 50 and 60 sccm of H 2 gas with 100 sccm Ar each. We note that there was no WS 2 deposition when H 2 was not introduced, but small, single crystal WS 2 triangles were observed when the H 2 /Ar gas mixture was provided. The images show that when the flow rate of H 2 was increased, both the size and density of crystals increased. Fig. 1f,g show the increasing size and total surface coverage measured by image analysis using ImageJ. The average size of a WS 2 crystal increased from 4.7 μ m 2 to 10.8 μ m 2 over the experimental range, and total area coverage increased from 2.2% to 20%. The increase in total area coverage is not only due to increase in crystal size, but also due to increase in nucleation density, which increased approximately four times, from one nucleation site per 213 μ m 2 to one per 54 μ m 2 in the experimental range.

Results and Discussion
Supplying H 2 gas without Ar during the reduction and sulfurization process of WO 3 resulted in increased single crystal size and nucleation density of WS 2 monolayer crystals, compared to the same experimental conditions but with Ar. In addition to the increased size of single crystal WS 2 monolayers, under certain conditions continuous mm 2 coverage of polycrystalline WS 2 monolayer growth was observed. Figures 2a-c show SEM images of WS 2 monolayers grown with 45, 50 and 60 sccm of H 2 gas and no Ar. Figure 2d shows the same trend as when Ar was present-when the H 2 flow rate was increased the average size of the WS 2 single-domain crystals increased. The crystals were larger than when Ar was present, from 79.8 μ m 2 under 45 sccm of H 2 and up to 432.7 μ m 2 under 60 sccm. Figure 2e shows polycrystalline WS 2 growth with more than 85% surface coverage, grown with 60 sccm H 2 . Figure 2f is a larger area SEM image of the polycrystalline growth, showing up to mm 2 surface coverage. Each cm 2 chip exhibited these large coverage regions along with regions of isolated single crystal growth such as the one used for the data in Fig. 2d. Figure 2g is a large-area optical image stitched to show the extent of large-area growth. The color contrast shows predominantly monolayer growth along with bilayer and thick growth in some regions. Within the red box, a 1.1 mm square, the monolayer coverage is 89.5%, calculated through thresholding and pixel counting performed with Gimp and ImageJ. Figure 3a,b show the Raman spectroscopy and photoluminescence (PL) peaks of mono, bi, and few layer WS 2 samples grown in our furnace. All spectra were taken with 532 nm excitation. Figure 3a gives the E 1 2g and A 1g phonon modes of mono, bi and multi-layer of WS 2 , located at 350.4 and 418.2 cm −1 for mono and bi-layer, confirming WS 2 growth 30 . In the multilayer WS 2 samples, the E 1 2g and A 1g modes were slightly blue-shifted to 352.5 and 421.3 cm −1 , consistent with values reported in literature 30 . As anticipated, the intensities of the two Raman peaks decreased with the number of layers, and the difference was small while not proportional to the number of layers. The frequency difference between the two modes for multi-layer WS 2 was smaller than that of mono or bi-layer WS 2 . All of these effects are consistent with previous reports 31 and support the conclusion that our growth products are WS 2 . Figure 3b presents the PL spectra of mono and bi-layer WS 2 samples. As observed by other groups, the PL intensity for monolayer samples was much higher than other two samples (bi or multi-layer) 32 . The inset of Fig. 3b shows the PL peaks of a bi-layer sample magnified to be visible, still measurable though appearing as flat when plotted alongside the monolayer sample. It is difficult to observe the PL peak of multi-layer WS 2 because the band gap changes from direct to indirect when the number of WS 2 layers changes from mono to multi-layer 33 . The center of the PL peak for mono and bi-layer samples was at 641.4 nm, corresponding to 1.93 eV, similar to values reported in literatures 29,34,35 . Furthermore, the sharp and intense PL peak indicates the high quality of the WS 2 monolayer. While the measured FWHM value of the PL peak from exfoliated WS 2 was 75 meV, the measured FWHM of the PL peak from our CVD-grown WS 2 was 40 meV, which is comparable to high quality CVD-grown WS 2 reported in other literature if not better 36 . Figure 3c shows the 2D x-ray diffraction data. The sharp and bright point indicates that the film was well crystallized. Comparing the data to powder diffraction data calculated from a WS 2 model using Mercury, the q value corresponds to the (004) plane.
The AFM and SEM images in Fig. 4 give further information on the effect of H 2 on crystal growth. There are two etching modes: substrate etching at low H 2 concentrations and WS 2 etching at high H 2 concentrations. In our experiment, we used a WO 3 -deposited SiO 2 /Si substrate as a tungsten source which was placed on top of a clean SiO 2 /Si substrate as a growth substrate. Also considering the presence or lack of Ar, we obtained four different types of WS 2 deposition samples: either the top (i.e., WO 3 deposited substrate) or bottom (i.e., oxidized Si substrate), and either with or without Ar flow during growth. Figure 4a depicts an AFM image of a WS 2 monolayer grown on the bottom substrate with no Ar. The step height measurement of 1 nm gives further evidence of monolayer growth. Figure 4b,c show AFM and SEM images of a WS 2 monolayer on the bottom substrate grown under a H 2 /Ar mixture. These images show the WS 2 grown in a 6 nm etched pit, indicating that WS 2 was deposited only after SiO 2 substrate etching. We observed the etched SiO 2 only on the bottom (growth) sample grown in the combination H 2 and Ar environment. We did not observe SiO 2 etching in any of the other three sample types, including the top (source) substrate during the same growth as the indented sample. This phenomena can be explained by considering the reduction and sulfurization process of WO 3 . When the H 2 / Ar mixture gas is used for WS 2 deposition, the concentration of H 2 gas is not high enough, resulting in the WO 3 not being fully reduced before sulfur gas is supplied. Then the clean bottom SiO 2 /Si substrate can be etched by a chemical which includes hydrogen, oxygen and sulfur, of which there are a few in the literature known to etch SiO 2 37 , while the top substrate is protected by the remaining WO 3 . Only when the WO 3 is completely reduced can WS 2 growth begin on the top source substrate. Thus, Ar dilutes the hydrogen, in turn increasing available oxygen during the time of sulfur delivery. When only H 2 gas is used, the high concentration of hydrogen fully reduces the WO 3 before sulfur is provided. However, as a second effect, when the flow rate of H 2 was high, above 60 sccm, the monolayer WS 2 deposited during the growth period was etched to the substrate surface nonuniformly. Figure 4d shows an etched WS 2 monolayer, located near the edge of the substrate. Comparing images of WS 2 growth under different hydrogen concentrations supports the explanation that the residual H 2 causes etching of the already grown WS 2 crystals after all the sulfur powder has evaporated. This is shown by the WS 2 being etched more at the edge of the chip, where the H 2 concentration is the highest due to our sandwich configuration. Also, the boundary of the original single crystal WS 2 monolayer can be observed due to incomplete etching, showing that the crystal was grown and later etched.
In our experiment, several growth formations for monolayer samples were observed, and the formations were also found to depend on the concentration of H 2 . Figure 5 shows the different growth processes and the resulting morphology differences. Figure 5a-c show different growth formations grown in an H 2 /Ar mixture environment. Figure 5a shows WS 2 monolayer growth under low H 2 /Ar flow rate conditions, where all the WS 2 monolayers are a clear triangle shape and less than 5 μ m in side length. Figure 5b shows monolayer growth under high H 2 /Ar flow rate, where we can observe a barb shaped triangle growth process, observed and explained by Cong et al. 38 . Figure 5c is a well-known 2D material growth mode, observable when a hexagonal crystal structure has different growth rates on alternate faces 39 . When we supplied only H 2 , we observed two different growth processes, shown in Fig. 5d-f. The growth shown in Fig. 5e,f is a new growth mode not reported in literature to our knowledge. Our interpretation on the temporal evolution of a single crystal of WS 2 is illustrated in Fig. 5g-j, showing the in-plane sequential growth, creating the multi-apex WS 2 triangle shape. The growth directions at the corners are shown by the solid arrows, while the dashed lines show the outline of the overall expansion. In this growth, W atoms are abundant on the SiO 2 surface due to the reduction of WO 3 by H 2 , as the sample is deposited under a high concentration of H 2 . We attribute this unique formation of the multi-apex WS 2 triangle shape to the ratio between W and S during the growth; the WS 2 monolayer becomes a triangle shape when the ratio of W to S is higher than 1:2 40 . In that paper, Wang et al. demonstrated that only for reactant delivery in the ratio 1:2 will hexagonal crystals grow, otherwise growth of three faces will be faster than the other three and triangles will form. In the terminated WS 2 monolayer (Fig. 5a), each side of the terminated triangle of WS 2 is an active site for in-plane expansion of the monolayer crystal. Figure 5h shows the schematic after the termination of the first expansion, forming an additional apex in each side of the triangle, of which growth state is shown in Fig. 5e. Figure 5i shows a second expansion step and Fig. 5j depicts the terminated second expansion process. Figure 5f is an SEM image after the third expansion process. This process repeats to enlarge the size of WS 2 monolayers. Figure 6 shows an AFM scan and Raman and PL for a typical step-growth sample. The Raman and PL signatures were taken on the area grown during the first enlargement step after the termination of the first WS 2 monolayer growth. The AFM scan shows that the growth is in-plane with matched thickness for the subsequent step, and the strong PL signal confirms that it is monolayer growth.

Conclusion
We have demonstrated large scale WS 2 monolayer deposition, with up to 433 μ m 2 single crystals and mm 2 size continuous polycrystalline films. We have furthermore elucidated the effect of the concentration of H 2 during the reduction and sulfurization process; the relatively simple relationship between increasing domain size with H 2 flow rate as well as substrate and sample etching and its effect on growth morphology. We have shown that controlling H 2 concentration is crucial for large area WS 2 deposition. In the presence of an Ar carrier gas, increasing the local pressure, the crystal size varied relatively little (a few micrometers) between low and high flow rates. In a H 2 only environment, the flow rate had a dramatic effect on growth. In addition, in the conditions of a high concentration of H 2 and a low concentration of sulfur gas, the grown WS 2 was etched. Raman spectroscopy and AFM images confirm monolayer growth in accordance with other groups' findings. XRD measurement confirms the grown WS 2 is well crystallized with the (004) plane normal to the substrate. Finally, we have offered an explanation for a new growth mode for WS 2 single crystal monolayers which works in stages.

Methods
5 nm thick WO 3 was evaporated from pellets onto a source substrate which was sandwiched with a clean second substrate for growth, with no space. The sandwiched sample was loaded into the middle of 3′ ′ quartz tube. For sulfur, we adopted the commonly published method for MoS 2 and WS 2 growth of placing solid sulfur powder in the furnace tube upstream of the growth area. The ambient gas was purged out by mechanical pump to the base pressure of 850 mTorr. As the furnace was ramped in temperature at 15 °C/min, the reaction proceeded by reduction of WO 3 by hydrogen and subsequent sulfurization of the WO 3 . The growth temperature was 900 °C. Ar gas was introduced from 150 °C to reduce moisture and ambient gas and H 2 gas was supplied from 650 °C (increasing temperature) to 700 °C (decreasing temperature). The deposition pressure depends on the gas type and amount of flow rate. The best result was obtained at 4.5 Torr deposition pressure under 60 sccm H 2 flow rate. The reduction and sulfurization reactions require a higher temperature than the sulfur evaporation. By placing the sulfur at different places outside of the main heating area of the furnace, it evaporated at different times relative to the substrate temperature. At the optimized location for our furnace setup, the sulfur powder started to evaporate at 830 °C furnace temperature and all sulfur powder was used up after about 30 minutes.