Metal Induced Growth of Transition Metal Dichalcogenides at Controlled Locations

Abstract

Metal induced nucleation is adopted to achieve the growth of transition metal dichalcogenides at controlled locations. Ordered arrays of MoS₂ and WS₂ have successfully been fabricated on SiO₂ substrates by using the patterned Pt/Ti dots as the nucleation sites. Uniform MoS₂ monolayers with the adjustable size up to 50 μm are grown surrounding these metal patterns and the mobility of such layer is about 0.86 cm²/V·s. The crystalline flakes of WS₂ are also fabricated extending from the metal patterns and the electron mobility of these flakes is up to 11.36 cm²/V·s.

Introduction

Two-dimensional (2D) materials have attracted considerable interest due to their unique electrical, optical, thermal and mechanical properties that do not exist in their bulk form^{1,2,3,4,5,6,7,8,9,10}. Thus 2D materials are regarded as potential candidates for future logic devices¹¹, integrated circuits¹², and optoelectronics devices¹³. Transition metal dichalcogenides (TMDs) are important 2D materials, which have been researched extensively. Although single and few-layer TMD have been synthesized by many methods^{14,15,16,17,18,19}, the grown TMD materials are small flakes instead of a whole continuous film throughout the substrate and the flakes with various shapes distributed randomly on the substrate^20,21,22,23, which limits the large scale devices fabrication and hence their applications.

An alternative is to grow TMD at predesigned locations where the electronic devices with certain function will be defined, which provides a promising way to overcome small crystal size and random distribution for TMD used in large scale devices fabrication. To our knowledge, there are very limited reports on the growth of TMD at defined locations. Han et al.²⁴ declared that the crystalline MoS₂ monolayer can be grown at predefined locations by using lithographically patterned islands of MoO₃ or ammonium heptamolybdate as seed materials. Su et al.²⁵ reported that the layered semiconductor SnS₂ arrays was grown on the patterned substrate by using thin-film pads of Pd/Cr, Cr, SiO₂/Cr, and Ni as nucleation sites. Godin et al.²⁶ also reported the growth of polycrystalline WS₂ monolayers on patterned substrates by controlling surface energy via oxygen-plasma treatment. Lou et al.²⁷ found that the MoS₂ triangular crystals are commonly nucleated and formed on the step edges of SiO₂. However, it is still a challenge to grow various category transition metal dichalcogenides at controlled locations by a generic growth protocol.

In this study, metal induced nucleation is proposed to be a simple way to grow transition metal dichalcogenides at controlled locations. Ordered arrays of MoS₂ and WS₂ were successfully grown on SiO₂ substrates by using the patterned Pt/Ti dots as the nucleation sites. Uniform MoS₂ monolayers are grown surrounding the metal patterns and the size of the MoS₂ monolayer can be easily controlled by adjusting the size of the used metal pattern. Moreover, the mobility of such MoS₂ layer is measured to be about 0.86 cm²/V·s. Under the same protocol, the crystalline WS₂ flakes are also grown extending from the metal patterns and the electron mobility of the flakes is up to 11.36 cm²/V·s.

Results and Discussion

Growth protocol of the 2D TMDs

As shown in Fig. 1, the TMD are grown in a two-zone CVD furnace. The pre-patterned metal (Pt/Ti) arrays are fabricated on SiO₂ covered Si wafer by e-beam lithography. Solid reactive precursors, sulfur powder and MoO₃ powder (or the ball-milled WO₃ and NaCl powder) were placed in the heating zone-I and zone-II, respectively. When the furnace was heated up, the solid precursors were sublimated into the quartz tube and transferred to the substrate by the carrier gas, Ar. The Pt/Ti patterns are expected to act as the nucleation sites for the growth of TMD.

Figure 1

(a,b) Schematic of the growth process of the monolayer MoS₂ arrays using the patterned Pt/Ti as cores deposited on SiO₂/Si substrate, and (c) Experimental setup of the CVD system.

Characterization of MoS₂

Figure 2a shows a typical SEM image for MoS₂ grown on the substrate by this metal induced nucleation method. It is obvious that the circular films exactly follow the patterned metal dots to form a well ordered array. The inset of Fig. 2a, the magnified image for a single unit, clearly demonstrates that the bright metal dot is in the middle of the dark circular film, indicating that this dark film actually grows from the metal dot. Figure 2c shows a series of Raman spectra taken at various points on a circular MoS₂ film (shown in Fig. 2b). There are two prominent peaks at ~387 cm⁻¹ and 407 cm⁻¹ at all the Raman spectra. It is well known that these two peaks are ascribed to the in-plane mode E¹_2g and out-of-plane mode A_1g of the MoS₂, respectively. The presence of these two Raman peaks unambiguously gives the evidence that these dark films are indeed MoS₂ grown on the substrate. On the other hand, these two Raman modes of E¹_2g and A_1g, are very sensitive to the layer thickness^28,29, which provide a convenient and reliable means to determine the thickness of MoS₂ film. The frequency difference value (Δ) between these two peaks are about 19.5 cm⁻¹ ~20.0 cm⁻¹ at the points of 1, 2, 6 and 7 in Fig. 2b, respectively, confirming that the thickness of the MoS₂ film surrounding the Pt/Ti dots is one monolayer. The vibration modes of MoS₂ are also observed in the spectrum taken at the metal dot (point 4 in Fig. 2b), indicating that the MoS₂ is also grown on the top of the metal. Moreover, relative larger frequency differences (about 22∼27 cm⁻¹) are obtained on the top and near the Pt/Ti dots, suggesting that the MoS₂ films at these places are multilayers (Figure S2). The Raman mapping with the peak difference as the indicator (407 cm⁻¹) given in Fig. 2d further reveal that the MoS₂ film grown out of the Pt/Ti dot is very uniform monolayer. In addition, the PL spectrum in Figure S3 exhibits the strongest emission at ~1.83 eV for the MoS₂ films grown around the metal dots, which is in agreement with the previous reports on MoS₂ monolayer^30,31. The thickness of the MoS₂ film derived from AFM measurement in Fig. 2(e) is ~0.85 nm, which is also consistent with the values for MoS₂ monolayer³², and the results also revealed that there are some cracks in the MoS₂ units in Fig. 2(e), it is possible that the cracks are ascribe to the domain boundaries of the polycrystalline MoS₂, where some domain boundaries of the polycrystalline MoS₂ are slightly oxidized in air. It is worth noting that the dimension of MoS₂ monolayer grown by this method can be easily controlled by adjusting the size of the metal dots. Figure 2f–h gives the SEM images of MoS₂ monolayer grown around the metal dots with the various sizes under the same growth condition. A SiO₂/Si substrate containing an array of the metal dots with various sizes was used to grow the MoS₂ monolayer around various metal dots at the same time, as shown in Fig. 2f–h. It is obvious that the dimension of the MoS₂ monolayer increases from 18.3 μm to 53.5 μm as the size of the metal dot varies from 3 μm to 10 μm. Careful examinations reveal that the area of the grown MoS₂ film linearly depends on the circumference of the metal dot, indicating that the radial growth rate of MoS₂ around the metal dot might be kept the same during the growth process. (See Figure S4).

Figure 2

(a) SEM images of the patterned growth of the MoS₂ layers. Inset: high magnification image of a typical MoS₂ flake. (b) Optical image of a typical MoS₂ flake (c) Raman spectra correspond to postions 1–7 in (b). (d) The E¹_2g peak intensity mapping for the MoS₂ flake in (b). Due to the sample drift during the Raman mapping measurement, the shape of the flake seems distorted a bit as compare to the optical image in (b). (e) AFM images of the edge of the MoS₂ layer on the substrate, and (f–h) SEM images of the different dimensions MoS₂ units.

Transmission electron microscopy (TEM) was used to characterize the crystal structures of the obtained MoS₂ film. Figure 3c shows the low magnification image of MoS₂ film as well as the metal dot (the center black area) that transferred onto Cu grid. The boundary of the transferred the film was outlined by the red dashed line in Fig. 3a. Figure 3(b) shows a typical high-resolution TEM (HRTEM) image. The periodic atom arrangement is clearly observed and the specific inter-planar distances for the (100) plane is measured to be about 0.285 nm, which gives the direct evidence for the crystalline nature of such MoS₂ monolayer. The selected area electron diffraction (SAED) was taken on different location as marked by the numbers in Fig. 3a. In Figure 3c and d, there is only one set of the hexagon diffraction pattern at location 8 and 12 but with different orientation, indicating that the MoS₂ film in these areas are crystalline monolayer but in different crystalline domains. The cracks on the AFM image in Fig. 2e represent the domain boundaries between different domains, indicating the polycrystalline nature of the MoS₂ film. The presence of two set of the hexagon spots in the SAED patterns for the areas of the point 10 and point11 (Fig. 3(e)) shows that the few-layers MoS₂ is grown on the top and near the Pt/Ti dots. The element distribution in the MoS₂ film is measured by energy dispersive spectroscopy (EDS). Although the elements Pt and Ti can be found in the multilayer area on the top of the metal dot (Fig. 3(f)), they are not observed in the regions far away from the metal dot (Fig. 3(g)), and only the S and Mo elements are measured. These results suggested that the metal dot only acts as a nucleation site and the metals, Pt and Ti, do not diffuse into the MoS₂ film during the growth process, at least in our measurement accuracy.

Figure 3

TEM images of the MoS₂ units, TEM images of (a) the MoS₂ unit and (b) the typical HRTEM image for the MoS₂ unit, the SAED pattern for (c) and (d) the position of point 8 and point 9, respectively, (e) the position of point 10 and point 11, and EDS analysis for the position of (f) point 10, (g) point 11.

Characterization of WS₂

WS₂ films are also successfully grown on the substrate by using the same growth protocol. Figure 4(a) shows an optical image of the WS₂ films grown on the prepatterned substrates in which the films appear much brighter than the metal dots. Although the WS₂ films are not uniformly circular shape like the MoS₂ films, these WS₂ films certainly follow the periodicity of the metal pattern, indicating that the metal dots do act as the nucleation sites during the film growth process. The SEM image of the WS₂ unit in Fig. 4(b) clearly shows that the irregular WS₂ flakes extend from the center metal dot, in contrast to a symmetric circular shape of the MoS₂ monolayer in Fig. 2a. The chemical reaction process different to those of MoS₂ may account for the irregular shape of the WS₂ grown around the metal dot. The typical Raman spectrum for the films is shown in Fig. 4(c). Two peaks located at 360.1 cm⁻¹ and 426.1 cm⁻¹ are observed, which are the well known E¹_2g and A_1g modes for WS₂³³. Figure 4(d) shows the Raman mapping image for the WS₂ unit in Fig. 4(b) by using the peak position (426.1 cm⁻¹) as an indicator. The perfect match between the Raman mapping image and the SEM image directly reveals that the WS₂ films are not grown on the bare substrate but around the metal dots, further confirming that the metal dots are indeed the nucleation sites for the growth of WS₂ films. The sharp bright spots with a hexagonal periodicity in the SAED measurement for the WS₂ flakes and the periodic atom arrangement with the specific interplanar distances of ~0.270 nm assigned to the (100) plane in the HRTEM image confirm the crystallinity nature of the WS₂ flakes with high quality, as shown in Fig. 4(e) and (f).

Figure 4

(a) Optical image of the WS₂ layers induced grown on the substrate, (b) SEM image of the WS₂ unit, (c) the typical Raman spectra of the WS₂ unit, (d) the peak position maps for the WS₂ unit, (e) TEM images of the WS₂ layers and the inset for the typical images of the SAED pattern, (f) the typical HRTEM image of the WS₂ layers.

Electrical measurements

In order to characterize the electronic properties of these grown TMD films, field effect transistors (FETs) were fabricated by e-beam lithography directly on the growth substrate without any transfer processes. Typical measurement results were shown in Fig. 5, where both the MoS₂ and WS₂ devices show good n-type field effects. The on/off ratio of the MoS₂ FET is about 10⁵, while it is only about 10⁴ for the WS₂ FET. The extracted electron mobility are 0.86 cm²/V·s for MoS₂ and 11.36 cm²/V·s for WS₂, respectively. The mobility of MoS₂ is relatively lower as compared with the previous work²⁴, which is originated from the polycrystallinite nature of the grown MoS₂ monolayer. During the fabrication process of these devices, the FETs were purposively built far from the pre-defined Pt/Ti dots to avoid the influence of the metal dots. As a consequence, the channel of the MoS₂ FET is actually a MoS₂ monolayer and that of the WS₂ FET is a multilayered WS₂. Therefore, the above measured results are understandable because that the multilayered channel is more difficult to be switched off than a monolayered channel but the higher mobility can be achieved in multilayered channel. The electrical properties of the FETs based on the grown MoS₂ and WS₂ films unambiguously confirm the high quality of the TMD films grown by this metal indunced nucleation method. Such observations also support our argument that large scale devices fabrication can be easily achieved via the defined location growth of TMD.

Figure 5

The electrical property of the measured MoS₂ FETs (a,b) and WS₂ FETs (c,d) The channel length and width are 2.82 μm and 24.5 μm for the MoS₂ FETs, and 4.4 μm width and 4 μm length for the WS₂ FETs.

In summary, we reported a metal induced growth method for the transition metal dichalcogenides grown at controlled locations. Where the high quality monolayer MoS₂ arrays are grown orderly around the core of Pt/Ti patterned on the substrates, and the WS₂ multilayers are also prepared at a controlled location by a simple CVD technology. The mobility of the MoS₂ films and the crystalline WS₂ flakes are about 0.86 cm²/V·s and up to 11.36 cm²/V·s, respectively. The results are hopeful for facilitating device fabrication for the integrated devices based on the transition metal dichalcogenides.

Methods

Growth process of the MoS₂

Pure MoO₃ powder was placed in a quartz boat at the centre of furnace, cleaned substrates with patterned Pt/Ti cores were placed on the downstream, and a separate quartz boat with sulfur powder was placed on the upstream, which was heated up to 190 °C using a separate heating system. The furnace was heated from room temperature to 850 °C at a ramp rate of 15 °C/min under an argon (99.999%) flow of 100 sccm. The temperature was held constant for 30 min during the MoS₂ growth, and the furnace chamber was then rapidly cooled to room temperature by opening the furnace door.

Growth process of the WS₂

Pure WO₃ powder and NaCl powder with the molar ratio of 1.4:1 were mixed and ball-milled in a grinding container for 2 h, using alcohol as a solvent, then dried at 95 °C³⁴. According to the ref. 34, WO₃ is the tungsten precursor but NaCl acts as a growth promoter. Afterwards the ball-milled powders were placed in a quartz boat at the centre of furnace, a separate quartz boat with sulfur powder was placed on the upstream, which was heated up to 190 °C using a separate heating system, and cleaned substrates with patterned Pt/Ti cores were placed on the downstream. The furnace was heated from room temperature to 900 °C at a ramp rate of 15 °C/min. The temperature was held constant for 30 min during the WS₂ growth, and the furnace chamber was then rapidly cooled to room temperature by opening the furnace door.

Device fabrication and testing

The MoS₂ and WS₂ field effect transistors were fabricated by e-beam lithography using Raith 150. After exposure, the source and drain electrodes (Ti/Au film of 10/30 nm thick) was deposited by using e-beam evaporator (K. J. Lesker with base vacuum of 7×10⁻⁸ torr) followed by lift-off process. The bottom Si was used as gate electrode. The filed effect properties of the fabricated devices were measured by using probe station and Keithley 4200 Semiconductor parameter analyzer at room temperature in air.

Characterizations

The morphologies and microstructures of transition metal dichalcogenides were characterized by optical microscopy, scanning electron microscopy (FEI-Quanta 200 F), atomic force microscope (Veeco Dimension 3100) and transmission electron microscopy (JEOL JEM-2100 at 80 keV). Raman and PL spectra were taken by Horiba Jobin Yvon LabRAM H8000 system with laser excitation wavelength of 488 nm.