Structural and Optical Properties of Tungsten Disulfide Nanoscale Films Grown by Sulfurization from W and WO3

Tungsten disulfide (WS2) was prepared from W metal and WO3 by ion beam sputtering and sulfurization in a different number of layers, including monolayer, bilayer, six-layer, and nine-layer. To obtain better crystallinity, the nine-layer of WS2 was also prepared from W metal and sulfurized in a furnace at different temperatures (800, 850, 900, and 950 °C). X-ray diffraction revealed that WS2 has a 2-H crystal structure and the crystallinity improved with increasing sulfurization temperature, while the crystallinity of WS2 sulfurized from WO3 (WS2-WO3) is better than that sulfurized from W-metal (WS2-W). Raman spectra show that the full-width at half maximum (FWHM) of WS2-WO3 is narrower than that of WS2-W. We demonstrate that high-quality monocrystalline WS2 thin films can be prepared at wafer scale by sulfurization of WO3. The photoluminescence of the WS2 monolayer is strongly enhanced and centered at 1.98 eV. The transmittance of the WS2 monolayer exceeds 80%, and the measured band gap is 1.9 eV, as shown by ultraviolet-visible-infrared spectroscopy.


Introduction
Tungsten disulfide (WS 2 ) is a two-dimensional (2-D) material consisting of a covalently bonded sheet of W atoms filled between two trigonal sheets of S atoms. WS 2 is mainly composed of three structures, hexagonal (2-H), trigonal (1T), or rhombohedral (3R) phase. Among them, the 2H phase structure is relatively stable and exhibits better optical properties [1]. The interlayers of WS 2 are bound only by weak van der Waals forces, and the interlayer spacing is~0.6 nm [2,3]. WS 2 has been extensively studied due to its unique layer-dependent properties, such as the ability to absorb 5-10% of incident sunlight [4], and its unique band structure. Monolayer WS 2 exhibits a direct band gap [5] of 1.98 eV [6], while multilayer WS 2 shows an indirect band gap [7] of about 1.3 eV [8,9]. This phenomenon can be attributed to the lack of Coulomb repulsion between the p z orbitals of the chalcogenide elements in adjacent layers, leading to stabilization of the Γ-state valence band [10]. WS 2 has been investigated for applications such as solar cells [11], hydrogen evolution reactions (HER) [12], electrocatalysis [13], batteries [14], and transistors [15]. However, the efficiency of most of the devices remains low. Therefore, a better understanding of the formation and physical properties of WS 2 is essential. Many efforts have been made to improve the application value of WS 2 , such as controlling the formation mechanism, including studying the growth mode of the films to obtain a uniform, large-area [16], and well-oriented crystals. There are three common methods for depositing WS 2 thin films: the stripping method [17], the chemical vapor Nanomaterials 2023, 13, 1276 2 of 9 deposition (CVD) [18,19], and the direct vulcanization synthesis method [20]. In this study, we seek to obtain high-quality WS 2 films by comparing tungsten (W) metal and oxide (WO 3 ) sulfurization processes. The structural and optical properties of the WS 2 thin films were investigated in detail for future applications in next-generation optoelectronic devices.

Experimental
The W and WO 3 films were prepared on c-axis Al 2 O 3 (1 cm 2 , provided by Bangjie Material Technology). First, the Al 2 O 3 substrate was ultrasonicated in acetone for 5 min to remove dirt and then rinsed with methanol. The cleaned c-axis sapphire was then inserted into the ion beam sputtering (IBS, Commonwealth Scientific) chamber for the growth of W or WO 3 films. Subsequently, the grown W or WO 3 films were sulfurized to obtain WS 2 films, as schematically shown in Figure S1a,b of the Supplementary Document. The sputtering of W films was performed at a base pressure of about 5 × 10 −6 Torr. Ar (99.999% purity) was introduced into the chamber at a flow rate of 5 sccm to initiate sputtering process. On the other hand, the WO 3 films were grown by flowing Ar and oxygen at flow rates of 5 and 3 sccm, respectively. The sulfurization of the W and WO 3 films was carried out in a horizontal quartz-tube furnace. The W or WO 3 films were placed in the center of the quartz-tube furnace while about 3 g of sulfur (99.999% purity) was placed next to the tube lid. The tube was then evacuated to 5 × 10 −2 Torr. During the sulfurization process, nitrogen gas (99.999%) flowed into the chamber while maintaining a pressure of 0.7 Torr. In this work, the sulfurization process is maintained at 900 • C for thickness-dependent studies.
To study the crystallinity behavior, the nine-layer sample of WS 2 -W was sulfurized at different temperatures of 800, 850, 900, and 950 • C. All sulfurization processes were carried out for 20-30 min and cooled naturally to room temperature. The elemental compositions of the WS 2 films were examined using X-ray photoelectron spectroscopy (XPS, Thermo Fisher Scientific with an Al Kα light source). The binding energies were referenced to the NIST-XPS database. The WS 2 phase was identified by micro-Raman (Horiba Jobin Yvon Lab RAM HR) equipped with an 1800-cycle grating, an objective lens with 100× magnification, and a laser wavelength of 532 nm. X-ray absorption near-edge spectroscopy (XANES) was performed to identify the local electron structure of WS 2 . The XANES was performed at the 07A1 beamline of the National Synchrotron Radiation Research Center (Hsinchu, Taiwan). The crystal structure of WS 2 films was examined by X-ray diffraction (XRD, D8 by Bruker AXS Gmbh) equipped with a Cu 2 keV light source. A high-resolution transmission electron microscope (HR-TEM, JEOL JEM-2100F CS STEM) equipped with a dual-beam focus ion beam was conducted to study the microstructure and thickness at the atomic scale. The electrons were accelerated at 12 kV and a magnification of 500 K. The TEM was also equipped with an energy-dispersive X-ray spectrometer (EDS). Further analysis was performed using micro-photoluminescence spectroscopy (PL) to study the band gap of WS 2 and ultraviolet-visible-infrared spectroscopy to understand the light absorption properties of the samples. Figure 1a,b shows the XRD of WS 2 -W at different sputtering thicknesses and sulfurization temperatures. All the XRD peaks marked with an asterisk match the substrate Al 2 O 3 peaks. The presence of lattice planes (002), (004), (006), and (008) suggests that the structure of WS 2 is a 2H (hexagonal) phase [5]. Moreover, the interlayer spacing of the WS 2 was estimated to be~0.62 nm using the Bragg equation, which agrees with the theory and previous results [21,22]. TEM investigation has been carried out to observe the atomic stacking of WS2 (ninelayer sample). In order to improve the conductivity of the film, about 5 nm of platinum was deposited on the surface of WS2. Platinum was also used to protect the sample from oxidation. Figure 1c shows the TEM image of the nine-layer of WS2 stacked along the caxis with the platinum capping layer. The total thickness of the WS2 layer was estimated to be ~5.37 nm with an interplanar spacing of 0.62 nm (Figure 1d), which agrees with the XRD result.

Elemental Composition Analysis
The elemental composition of WS2 was qualitatively examined by X-ray photoelectron spectroscopy (XPS). Figure 2 shows high-resolution XPS spectra in W 4+ and S 2-energy regions. Figure 2a exhibits the tungsten signals at 32.6, 34.8, and 38.2 eV, which can be ascribed to W 4f7/2, W 4f5/2, and W 5f3/2, respectively, with a spin-orbit splitting of (4f7/2-4f5/2) = 2.2 eV. The existence of high-intensity signals at 32.6 and 34.8 eV also suggests the formation of 2H-WS2, which concurs with previous reports [24,25]. Figure 2b shows the spectrum of the doublet S 2p at 163.5 eV and 162.3 eV, representing S 2p1/2 and S 2p3/2, respectively [26,27]. These peaks are attributed to the divalent sulfide ions (S 2-), which are also associated with the formation of 2H-WS2. The elemental composition of WS2 was estimated to be 56.0, 28.7, and 15.3% (in atomic percentage) for S 2p, W 4f, and O 1s, in order. The oxygen signal is from the sapphire substrate. Therefore, the W and S ratio is very close With increasing film thickness, the intensity of XRD signals (002), (004), (006), and (008) increases, while the FWHM of each peak is reduced, as shown in Figure 1a, indicating that the crystallinity of the films improves with the number of WS 2 layers. The absence of the (002) peak for the monolayer of WS 2 is due to the diffraction limit. The tungsten films were sulfurized at different temperatures of 800, 850, 900, and 950 • C to obtain the optimum temperature for formation WS 2 (nine-layer sample). The corresponding XRD results are shown in Figure 1b, indicating that the crystallinity of WS 2 improves with increasing sulfurization temperature. Additionally, to study the local structure of WS 2 , the X-ray absorption near-edge structure (XANES) was also performed at W L 3 -edge. Figure 1c exhibits the spectrum of WS 2 with tungsten and tungsten trioxide as references. The XANES of WS 2 agrees with the previous result [23]. In addition, it also confirms that the WS 2 (black path) has a 2H (hexagonal) structure.
TEM investigation has been carried out to observe the atomic stacking of WS 2 (ninelayer sample). In order to improve the conductivity of the film, about 5 nm of platinum was deposited on the surface of WS 2 . Platinum was also used to protect the sample from oxidation. Figure 1c shows the TEM image of the nine-layer of WS 2 stacked along the c-axis with the platinum capping layer. The total thickness of the WS 2 layer was estimated to be~5.37 nm with an interplanar spacing of 0.62 nm (Figure 1d), which agrees with the XRD result.

Elemental Composition Analysis
The elemental composition of WS 2 was qualitatively examined by X-ray photoelectron spectroscopy (XPS). Figure 2 shows high-resolution XPS spectra in W 4+ and S 2− energy regions. Figure 2a exhibits the tungsten signals at 32.6, 34.8, and 38.2 eV, which can be ascribed to W 4f 7/2 , W 4f 5/2 , and W 5f 3/2 , respectively, with a spin-orbit splitting of ∆E P (4f 7/2 − 4f 5/2 ) = 2.2 eV. The existence of high-intensity signals at 32.6 and 34.8 eV also suggests the formation of 2H-WS 2 , which concurs with previous reports [24,25]. Figure 2b shows the spectrum of the doublet S 2p at 163.5 eV and 162.3 eV, representing S 2p 1/2 and S 2p 3/2 , respectively [26,27]. These peaks are attributed to the divalent sulfide ions (S 2− ), which are also associated with the formation of 2H-WS 2 . The elemental composition of WS 2 was estimated to be 56.0, 28.7, and 15.3% (in atomic percentage) for S 2p, W 4f, and O 1s, in order. The oxygen signal is from the sapphire substrate. Therefore, the W and S ratio is very close to 2:1, confirming that the samples are in good stoichiometry. Additionally, the elemental composition of WS 2 was also qualitatively examined using EDS ( Figure S2). Nanomaterials 2023, 13, x FOR PEER REVIEW 4 of 9 to 2:1, confirming that the samples are in good stoichiometry. Additionally, the elemental composition of WS2 was also qualitatively examined using EDS ( Figure S2).

Micro-PL Spectroscopy
Micro-PL has been performed to study the optical and electronic properties of WS2. It is found that the PL signal comes from the energy gap of the direct transition when the thickness is reduced below two atomic layers. The PL of WS2 is strongly suppressed as thickness increases above two atomic layers, in which the indirect band gap is completely formed. Figure 3 shows the PL peaks of the monolayer and bilayer of WS2. The peaks are centered at 1.98 and 1.95 eV, respectively, in good agreement with previous results (2.1-1.9 eV) [28] and with the DFT-LDA calculations for the monolayer and the bilayer of WS2 [29,30]. The PL intensity shows a slight redshift (~3 eV) and a threefold decrease in intensity as the layer increases from the monolayer to the bilayer of WS2 [3]. The peak shift is likely due to the proportional relationship between layer number and carrier concentration, which also contributes to Coulomb scattering [31]. This result is in line with the theoretical prediction that the WS2 monolayer has a direct band gap at the Γ point [32].

Ultraviolet-Visible-Infrared Spectroscopy
The optical properties of WS2 were further investigated using an ultraviolet-visibleinfrared (UV/visible/NIR) spectrophotometer, as shown in Figure 4. It was observed that the transmittance of WS2 increased from 40% (nine-layer) to over 80% (monolayer) in the UV-visible region (Figure 4a). The band gap of the WS2 monolayer is about 1.9 eV, as estimated from the Tauc plot of the absorption spectrum (Figure 4b).

Micro-PL Spectroscopy
Micro-PL has been performed to study the optical and electronic properties of WS 2 . It is found that the PL signal comes from the energy gap of the direct transition when the thickness is reduced below two atomic layers. The PL of WS 2 is strongly suppressed as thickness increases above two atomic layers, in which the indirect band gap is completely formed. Figure 3 shows the PL peaks of the monolayer and bilayer of WS 2 . The peaks are centered at 1.98 and 1.95 eV, respectively, in good agreement with previous results (2.1-1.9 eV) [28] and with the DFT-LDA calculations for the monolayer and the bilayer of WS 2 [29,30]. The PL intensity shows a slight redshift (~3 eV) and a threefold decrease in intensity as the layer increases from the monolayer to the bilayer of WS 2 [3]. The peak shift is likely due to the proportional relationship between layer number and carrier concentration, which also contributes to Coulomb scattering [31]. This result is in line with the theoretical prediction that the WS 2 monolayer has a direct band gap at the Γ point [32]. to 2:1, confirming that the samples are in good stoichiometry. Additionally, the elemental composition of WS2 was also qualitatively examined using EDS ( Figure S2).

Micro-PL Spectroscopy
Micro-PL has been performed to study the optical and electronic properties of WS2. It is found that the PL signal comes from the energy gap of the direct transition when the thickness is reduced below two atomic layers. The PL of WS2 is strongly suppressed as thickness increases above two atomic layers, in which the indirect band gap is completely formed. Figure 3 shows the PL peaks of the monolayer and bilayer of WS2. The peaks are centered at 1.98 and 1.95 eV, respectively, in good agreement with previous results (2.1-1.9 eV) [28] and with the DFT-LDA calculations for the monolayer and the bilayer of WS2 [29,30]. The PL intensity shows a slight redshift (~3 eV) and a threefold decrease in intensity as the layer increases from the monolayer to the bilayer of WS2 [3]. The peak shift is likely due to the proportional relationship between layer number and carrier concentration, which also contributes to Coulomb scattering [31]. This result is in line with the theoretical prediction that the WS2 monolayer has a direct band gap at the Γ point [32].

Ultraviolet-Visible-Infrared Spectroscopy
The optical properties of WS2 were further investigated using an ultraviolet-visibleinfrared (UV/visible/NIR) spectrophotometer, as shown in Figure 4. It was observed that the transmittance of WS2 increased from 40% (nine-layer) to over 80% (monolayer) in the UV-visible region (Figure 4a). The band gap of the WS2 monolayer is about 1.9 eV, as estimated from the Tauc plot of the absorption spectrum (Figure 4b).

Ultraviolet-Visible-Infrared Spectroscopy
The optical properties of WS 2 were further investigated using an ultraviolet-visibleinfrared (UV/visible/NIR) spectrophotometer, as shown in Figure 4. It was observed that the transmittance of WS 2 increased from 40% (nine-layer) to over 80% (monolayer) in the UV-visible region (Figure 4a). The band gap of the WS 2 monolayer is about 1.9 eV, as estimated from the Tauc plot of the absorption spectrum (Figure 4b).

Raman Spectra of WS2-W and WS2-WO3
Raman spectroscopy is widely used for elemental analysis, layer stacking order, and doping effect of transition metal dichalcogenides [33]. Figure 6 shows the Raman spectra of WS2-W and WS2-WO3 with the different number of stacked layers. Figure 6a,b shows two main Raman peaks centered at ~354 and ~420 cm −1 , respectively. These Raman peaks agree with previous work [21,34,35]. It was observed that the A1g peaks overlap with the Raman peaks of sapphire (380 and 417cm −1 ) [36] when the Raman peaks were deconvoluted using a multi-peak Lorentzian fitting method [37].

Comparison of WS 2 -W and WS 2 -WO 3 3.4.1. XRD of WS 2 -W and WS 2 -WO 3
In order to obtain high-quality WS 2 films, we compared the sulfurization of W and WO 3 films, termed as WS 2 -W and WS 2 -WO 3 , respectively. The samples were sulfurized side by side at 900 • C. The XRD results are shown in Figure 5. It was revealed that the diffraction peaks of WS 2 -WO 3 are sharper and narrower than those of WS 2 -W with similar thickness, suggesting that the crystallinity of WS 2 -WO 3 is better than that of WS 2 -W.

Raman Spectra of WS2-W and WS2-WO3
Raman spectroscopy is widely used for elemental analysis, layer stacking order, and doping effect of transition metal dichalcogenides [33]. Figure 6 shows the Raman spectra of WS2-W and WS2-WO3 with the different number of stacked layers. Figure 6a,b shows two main Raman peaks centered at ~354 and ~420 cm −1 , respectively. These Raman peaks agree with previous work [21,34,35]. It was observed that the A1g peaks overlap with the Raman peaks of sapphire (380 and 417cm −1 ) [36] when the Raman peaks were deconvoluted using a multi-peak Lorentzian fitting method [37].

Raman Spectra of WS 2 -W and WS 2 -WO 3
Raman spectroscopy is widely used for elemental analysis, layer stacking order, and doping effect of transition metal dichalcogenides [33]. Figure 6 shows the Raman spectra of WS 2 -W and WS 2 -WO 3 with the different number of stacked layers. ous results [38,39]. Additionally, for both WS2-W and WS2-WO3 samples, as the film thickness decreases, the A1g signals show a slight red shift of about ~2.3 cm −1 (Figure 6d). The redshift of A1g is likely associated with reducing the atomic restoring forces. When the number of WS2 layers is decreased, the long-range Coulomb interaction between effective charges and dielectric screening is enhanced, leading to increased restoring force between the S-S atoms [3]. Figure 6. (a,b). Deconvolution of the Raman spectra of WS2-W and WS2-WO3, respectively, using Lorentz fitting method; (c) the intensity ratio of 2 / 1 as a function of the number layers of WS2 (d) the peak positions of A1g, 2 1 , and 2LA with respect to the number layer of WS2. Additionally the deconvolution of the Raman spectra for bilayer, six-layer, nine-layer can be seen in the supple mentary documents.
Due to the dielectric screening effect, the Raman signal is sensitive to the interactions between atoms in the interlayer and long-range Coulomb interactions [40]. The Raman wavevector and the intensity of WS2 are strongly correlated with the layer thickness at the nanoscale level. In this work, we also used Raman spectra to calibrate the thickness of WS as a monolayer, bilayer, six-layer, and nine-layer [31,39]. The Raman spectra confirmed that WS2 was well-formed as WS2-W and WS2-WO3. It is clear that the peak intensity and Figure 6. (a,b). Deconvolution of the Raman spectra of WS 2 -W and WS 2 -WO 3 , respectively, using Lorentz fitting method; (c) the intensity ratio of I 2LA /I A 1g as a function of the number layers of WS 2 ; (d) the peak positions of A 1g , E 1 2g , and 2LA with respect to the number layer of WS 2 . Additionally, the deconvolution of the Raman spectra for bilayer, six-layer, nine-layer can be seen in the Supplementary Documents. Figure 6a,b shows two main Raman peaks centered at~354 and~420 cm −1 , respectively. These Raman peaks agree with previous work [21,34,35]. It was observed that the A 1g peaks overlap with the Raman peaks of sapphire (380 and 417cm −1 ) [36] when the Raman peaks were deconvoluted using a multi-peak Lorentzian fitting method [37]. Deconvolution of Raman peaks also revealed three strong peaks, which are E 1 2g (in-plane displacement of W and S), A 1g (out-plane displacement of S-S atoms), and 2LA (M) (second order Raman peak), which are centered at~356,~418, and~351 cm −1 , respectively. Furthermore, the approximate distance between A 1g and E 1 2g of WS 2 -W is~61.8, 63.8, 61.6, and 64.1 cm −1 for monolayer, bilayer, six-layer, and nine-layer samples, respectively, which is not much different from the distance for WS 2 -WO 3 of about 61.8, 63.7, 63.2, 62.7 in the same order. Figure 6c shows the I 2LA /I A 1g intensity ratio of WS 2 -W reduces from the monolayer (1.8) to the nine-layer (1.3). However, WS 2 -WO 3 shows a slightly different trend, where the intensity ratio increases from the monolayer (1.3) to the bilayer (1.8), and then decreases to the nine-layer (1.3). The results suggest that the second-order Raman peak intensity (I 2LA ) is almost twice that of the first-order I A 1g . This is consistent with the previous results [38,39]. Additionally, for both WS 2 -W and WS 2 -WO 3 samples, as the film thickness decreases, the A 1g signals show a slight red shift of about~2.3 cm −1 (Figure 6d). The redshift of A 1g is likely associated with reducing the atomic restoring forces. When the number of WS 2 layers is decreased, the long-range Coulomb interaction between effective charges and dielectric screening is enhanced, leading to increased restoring force between the S-S atoms [3].
Due to the dielectric screening effect, the Raman signal is sensitive to the interactions between atoms in the interlayer and long-range Coulomb interactions [40]. The Raman wavevector and the intensity of WS 2 are strongly correlated with the layer thickness at the nanoscale level. In this work, we also used Raman spectra to calibrate the thickness of WS 2 as a monolayer, bilayer, six-layer, and nine-layer [31,39]. The Raman spectra confirmed that WS 2 was well-formed as WS 2 -W and WS 2 -WO 3 . It is clear that the peak intensity and FWHM of E 1 2g for WS 2 -WO 3 are narrower than that of WS 2 -W (Table S1 in the Supplementary Materials). The average estimation of FWHM of WS 2 -WO 3 is~7.2; however, the WS 2 -W is~10.1, suggesting WS 2 -WO 3 has better crystallinity than WS 2 -W. This is consistent with the XRD results ( Figure 5). This phenomenon can be attributed to the fact that sulfur can replace oxygen more effectively at a relatively lower temperature [41], while WS 2 formed from tungsten metal requires a higher temperature to facilitate metal-sulfur bonding [42]. As a result, the defect density in WS 2 -WO 3 is likely to be lower than in WS 2 -W.

Conclusions
We report a study of nanoscale WS 2 films with various number of layers, prepared by sulfurization of W and WO 3 at different temperatures (800 to 950 • C). The WS 2 films were inferred to be the 2H phase and c-axis oriented. The XRD shows that the crystal quality of the WS 2 films improved with increasing sulfurization temperature. The photoluminescence of the monolayer of WS 2 is strongly enhanced and centered at 1.98 eV. The transmittance of the monolayer WS 2 exceeds 80% and the band gap is 1.9 eV, revealed by ultravioletvisible-infrared spectroscopy. The Raman analysis shows that the FWHM of WS 2 -WO 3 is narrower than that of WS 2 -W, indicating the structure of WS 2 -WO 3 is superior to that of WS 2 -W, which is in good agreement with the X-ray diffraction result. We conclude that a large-area, high-quality WS 2 film can be prepared by sulfurizing WO 3 . The results are promising for applications in next-generation optoelectronic devices.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nano13071276/s1, Figure S1: Schematic diagram: (a). W-metal and WO 3 prepared on sapphire substrate by ion beam sputtering technique; (b). W-metal or WO 3 sulfurization process using a thermocouple-equipped furnace, the process was carried out inside a horizontal quartz tube with a diameter of 50 mm and length of 100 cm; Figure S2: shows the EDS depth analysis of the WS 2 film that was carried out with Line scan mode. The small picture in the upper left corner is the relationship between the scanning position and the ratio of each element; Figure S3: the deconvolution of Raman spectra of bilayer, six-layer, and nine-layer samples using a multi-peak Lorentz fitting to separate the A1g peak from the substrate and 2LA from E_2 g 1 . The (a-c) shows the WS2 sulfurized from tungsten metal and (d-f) shows the WS2 sulfurized from tungsten trioxide; Table S1: Summary of the intensity ratio of I 2LA /I A 1g and FWHM as a function of layer numbers.  Data Availability Statement: All data included in this study are available upon request by contact with the corresponding author.