Growth of a Large, Single-Crystalline WS₂ Monolayer for High-Performance Photodetectors by Chemical Vapor Deposition

Abstract

2D WS₂ is a promising candidate for the next generation nanoelectronics, spintronics, valleytronics, and optoelectronics. However, the uncontrollably large-area growth of WS₂ nanosheets and their unsatisfactory performance of the photodetectors based on WS₂ hindered its applications. Here, we proposed a CVD method using tungstic acid as the precursors to grow WS₂ flakes. After being characterized by AFM, Raman, PL, and TEM, we found the as-grown WS₂ flakes were high-quality structures. Then the photodetectors based on the as-grown WS₂ were fabricated, which exhibited high responsivity (7.3 A W⁻¹), a fast response rate (a response time of 5 ms and a recovery time of 7 ms), prefect external quantum efficiency (EQE) (1814%), and remarkable detectivity (D*) (3.4 × 10¹² Jones). Our works provided a new CVD method to grow some high-quality WS₂ nanosheets.

1. Introduction

Atomically thin tungsten disulfide (WS₂), a 2D crystal with some interesting and important properties, is a promising candidate for the next generation of nanoelectronics, spintronics, valleytronics, and optoelectronics [1,2,3,4,5,6,7]. For example, WS₂ has a direct bandgap in the visible range and high absorption relative to its thickness [5,8]. WS₂ exhibits ambipolar field-modulation behavior [9]. The theoretical calculations predict that it has a reduced effective mass, allowing higher carrier mobility [10]. Monolayer WS₂ has strong PL emission efficiency, stronger than other TMDCs [11,12]. WS₂ exhibits strong spin-orbit coupling and band splitting due to spin enabling spintronics/valleytronics [13,14,15,16]. WS₂ also has high nonlinear susceptibility, suggesting its use for nonlinear optical devices [17,18]. Nevertheless, most researchers on WS₂ are largely limited because of the relatively small lateral size of exfoliated flakes, not to mention the randomness of their shape, thickness, and crystal quality [19,20]. Liquid exfoliation has an advantage for the mass production of WS₂, but it is difficult to control defects, hindering its application to electronic devices [21,22,23]. CVD is a hopeful approach to grow large-area WS₂ flakes for extensive device applications. There are two common methods to synthesize WS₂ by supplying tungsten sources on inert substrates before sulfurization in CVD. One is thin tungsten films or thin tungsten oxide films deposited by various methods including e-beam evaporation, magnetron sputtering, and atomic layer deposition [11,24,25,26,27,28,29]. The other is that of tungsten oxides solid precursors vaporized with solid sulfur sources simultaneously during the CVD process [30,31,32,33,34]. As it is difficult to control the tungsten sources location, distribution, and uniformity precisely, and the poor adhesion of the sources to inert substrates, the controlled CVD growth of large-area uniform monolayer WS₂ remains a challenge. Some groups used metal substrates (Au) instead of the inert substrates to synthesize a large-area monolayer WS₂ film; they also tested the electrical properties of WS₂-based field-effect transistors after bubbling transfer from Au substrates to SiO₂/Si substrates [35,36,37]. Here we report a new method, spin-coating tungsten acid as a tungsten source on SiO₂/Si substrates directly before sulfurization, to grow single crystalline WS₂ monolayers with a size of up to hundreds of micrometers. We also found that photodetectors based on single crystalline WS₂ flakes have a high responsivity of 7.3 A W⁻¹ with a fast response rate of 5 ms, an external quantum efficiency (EQE) of 1814%, and a detectivity (D*) of 3.4 × 10¹² Jones.

2. Materials and Methods

The WS₂ flakes were synthesized by using tungstic acid on SiO₂/Si substrates as the precursors and sulfurizing via a chemical vapor deposition (CVD) method. First, 2.8 g of tungstic acid (99.9%, Alfa) was dispersed in the oxalic acid solution (0.15 M/L) by ultrasound. Then the tungstic acid colloidal solution was dispersed on a clean SiO₂/Si substrate by spin-coating. After that, the whole substrate was annealed in air at 100 °C for 1h. Finally, the SiO₂/Si substrates with tungstic acid were sulfurized by CVD. For CVD growth, the Si/SiO₂ substrates with the tungstic acid colloidal solution were placed at the center of the quartz tube, and the 0.1 g S powders (99.5%, Alfa) were placed on the upstream side of the Si/SiO₂ substrates. The quartz tube was flushed with Ar (5N) gas several times and purged to 0.1 Pa with a mechanical pump. The furnace temperature was raised to 900 °C in 30 min and kept for 30 min with a flow rate of 50 sccm Ar gas. After growth, the furnace was naturally cooled to room temperature. The synthesized WS₂ samples were characterized by an atomic force microscope (AFM, SPM9700, Shimadzu), Raman spectroscopy, and a transmission electron microscope (TEM, Tecnai G² F30 S-TWIN, FEI). The Raman spectra were recorded in the backscattering geometry at a 532 nm line with an argon ion laser Raman spectrometer (LabRAM HR800, Horiba JobinYvon). The Raman mappings were collected by alpha300 R, WITec GmbH, Ulm, Germany, the laser wavelength was 532 nm, and the scanning step interval was 300 nm.

We fabricated the photodetectors based on the WS₂ flakes by a standard photolithography procedure (MDA-400M, Midas). The 10/50 nm Ti/Au electrode patterns were deposited by an E-beam deposition system (Nexdep, Angstrom Engineering). The photodetector measurement devices contained a broadband laser-driver light source (LDLS, EQ-1500, Energetiq) calibrated by a UV-enhanced silicon photodiode in an ambient atmosphere, a semiconductor characterization system (4200-SCS, Keithley), and an oscilloscope (DSO-X 3052A, Keysight). The oscilloscope light pulse chopped was 500 nm, and the frequency was 3 Hz.

3. Results and Discussion

Figure 1a shows the optical image of the as-grown triangle WS₂ flakes on SiO₂/Si substrate; the majority of the flakes were more than 220 μm. The thickness of these triangle crystals was measured by atomic force microscopy (AFM) (Figure 1b). The AFM step height of the WS₂ flake was typically measured at 0.7 nm, which corresponds to a one-layer structure.

We have further investigated the layer number of the resulting atomically thin WS₂ triangles by Raman spectroscopy. As shown in Figure 2a, the strongest peak of the WS₂ triangle flake at about 350 cm⁻¹ includes three subpeaks, which are resolved by a multipeak Lorentzian fitting. According to the calculated phonon dispersion [38] and experimental studies [13,39,40,41] of 1L-WS₂, the in-plane vibrational E¹_2g(M) mode peak is at 343 cm⁻¹, the second-order mode of longitudinal acoustic phonon 2LA (M) peak is at 350 cm⁻¹, and the in-plane vibrational E¹_2g(Γ) mode peak is at 355 cm⁻¹, respectively. The other peaks at 418, 295, and 322 cm⁻¹ are attributed to the out-of-plane A_1g(Γ) mode, the combination modes of 2LA(M)-2E²_2g(Γ), and the combination modes of 2LA(M)-E²_2g(Γ), respectively. The spectral fingerprint of a monolayer WS₂ is the frequency separation of 62 cm⁻¹ between the E¹_2g(Γ) mode and the A_1g(Γ) mode [11]. Raman images (Figure 2b,c) plotted by extracting the intensity were acquired to demonstrate the uniformity of our WS₂ samples. The 2LA(M) phonon mode was much more intense than the A_1g(Γ) mode. The as-grown WS₂ triangle flake was uniform.

The optical properties of the monolayer WS₂ triangle flakes were further investigated by microphotoluminescence (PL). The monolayer WS₂ flake has a PL peak at about 642 nm, and its full width half maximum (FWHM) value is 22 nm (Figure 3a). The corresponding PL peak-integrated intensity, position, and width mappings of the monolayer WS₂ triangle flake are shown in Figure 3b–d, respectively. The results of the PL indicated the as-grown WS₂ was defect free and uniform.

The structures of the monolayer WS₂ triangle flakes were characterized by TEM. Figure 4a shows the low magnitude bright-field TEM image of a WS₂ triangular crystal. The high-resolution TEM (HRTEM) image (Figure 4b) shows the hexagonal lattice fringes, which indicates a perfect atomic structure with a lattice spacing of 0.27 nm, corresponding to the (100) planes. The corresponding selected area electron diffraction (SAED) pattern (Figure 4c) revealed the defect free nature of the WS₂ flake [36]. Then the elemental compositions of the WS₂ triangle flakes were acquired by EDX mapping and EDX spectrum. EDX mapping in Figure 4d,e indicates the W and S elements distribute evenly. EDX spectrum in Figure 4f demonstrates the atomic ratio of W and S is around 1:2, which is consistent with the original stoichiometry of WS₂. The TEM results further prove that as-grown WS₂ flakes are defect free.

To research the optoelectronic properties of the WS₂ flakes, photodetectors based on them were fabricated. The spectral response curve reaches a minimum at the wavelength of ≈645 nm in Figure 5a. Hence, the bandgap is about 1.92 eV by calculation, which is consistent with the PL results. Fitting the plot of photocurrent I_ph on light intensity P for the WS₂ flake as I_ph ≈ P^θ obtains the value of θ ≈ 0.96 (Figure 5b), hinting the as-grown WS₂ has very few defects or traps to photo-induced electron/hole pairs in the test power density range [42]. We further tested the cyclability of the photodetector under 500 nm incident light with the light on/off time interval of 30 s under a bias of 1 V (Figure 5c). The performances of the photodetector are stable. The response and recovery rates were 5 ms and 7 ms tested by an oscilloscope, respectively (Figure 5d). The photoresponsivity was 7.3 A W⁻¹, according to the R_λ = I_ph/PS, where I_ph is the photoexcited current, P is the light power intensity, and S is the effective area of the photodetector. The external quantum efficiency (EQE) was 1814%, according to the EQE = hcR_λ/eλ, where h is the Plank’s constant, c is the light velocity, R_λ is the photoresponsivity, e is the elementary electronic charge, and λ is the exciting wavelength. The specific detectivity (D*) was 3.4 × 10¹² Jones, according to the D* = R_λS ^1/2/(2eI _dark)^1/2, where I _dark is the dark current. The performance of our photodetector is more inspiring than the most reported WS₂ based photodetectors shown in

Table 1. Comparison of the key parameters of our photodetector.

Photodetectors	Fabrication Method	Rλ (A/W)	EQE (%)	Response Time (ms)	Reference
Multilayer WS2	CVD	92 × 10−6	-	5.3	[4]
Multilayer WS2	Exfoliated	5.7	1118	<20	[43]
Monolayer WS2	CVD	18.8 × 10−3	-	<4.5	[44]
Multilayer WS2	PLD	0.51	137	4.1 × 10−3	[45]
Monolayer WS2	CVD	3.07	763	370	[46]
Multilayer WS2	Drop casting	145 × 10−3	-	153.78	[47]
Monolayer WS2	CVD	7.3	1814	5	This work

.

4. Conclusions

In summary, we grew the monolayer WS₂ triangle flakes via a CVD method with the tungstic acid, and researched the performance of the photodetectors based on them. The utilization of tungstic acid colloidal solution could improve the uniformity of the tungsten sources on the substrates. The as-grown monolayer WS₂ flakes have a size of about 220 μm, a bandgap of about 1.92 eV, and no defects. The photodetectors based on them showed excellent performance, such as high responsivity of 7.3 A W⁻¹, large EQE of 1814%, and a fast response rate of 5 ms.