Raman Spectral Analysis of Sputtered and Sulfurized Nanostructured WS₂ Films

Abstract

This study presents the Raman spectral characteristics and selected electrical parameter measurements of WS₂ films deposited by magnetron sputtering on sapphire and subsequently sulfurized. The analysis of the Raman spectra focuses on the positions and shifts of the E¹_2g and A_1g vibrational modes. The effects of different sputtering times on WS₂ films and the corresponding activation energy values were also investigated. From both physical and experimental perspectives, the Raman spectral features of WS₂ films were found to depend on the laser excitation wavelengths (532 nm and 632.8 nm) as well as on possible crystallographic defects and variations in the excitation point position. These defects have a significant influence on both the Raman spectra and the activation energies of the studied samples. The calculated activation energies (~ 0.15–0.19 eV) of the conduction charge carriers correlate with shallow defect-related energy levels indicated by the Raman characteristics.

1. Introduction

The main goal of this study is to analyze the Raman spectra and obtain information about the morphology of the WS₂ material versus position of Raman bands. Another goal was to obtain the activation energies of the charge carriers by electrical measurements and calculations, estimate their origin and correlate it with the Raman spectroscopy results. We investigate nanostructured tungsten disulfide (WS₂) films on sapphire substrates as a material for electronic applications, specifically as a gas sensor. For this, we need semiconductor material with a high surface-to-volume ratio. WS₂ has the largest direct band gap of about approximately 2.1 eV and indirect band gap of 1.3 eV. Its most distinguished characteristics are the transition from indirect band gap to direct band gap, when thinned down from a bulk to a monolayer.

The most prominent Raman bands of WS₂ are the E¹_2g and A_1g modes which are associated with in-plane and out-of-plane vibrations of W and S atoms. These two main vibrations are characteristic of layered transition metal dichalcogenides with a hexagonal structure. The E¹_2g band corresponds to the in-plane vibrations of sulfur and tungsten atoms in the layer, where the metal and chalcogen atoms vibrate in opposite directions. The frequency of the E¹_2g vibrations varies depending on the number of layers of the material. In the WS₂ monolayer, the E¹_2g mode is located at approximately 356 cm⁻¹. The A_1g mode represents oscillations of sulfur atoms that move perpendicular to the plane and out of plane, whereas the tungsten atoms remain relatively motionless at the equilibrium position (see Figure 1) [1,2,3,4,5]. Position of the A_1g band peak is also sensitive to the number of layers of WS₂; for example, in a monolayer, it is located at approximately 418 cm⁻¹, while in thicker layers, it shifts to higher wavenumber values. This shift is due to interactions between the layers of the material. The difference between both peaks serves as a sensitive indicator of the number of layers and interlayer interactions of WS₂. It has been found experimentally [6] that when WS₂ is a monolayer material, the difference between the dominant bands is less than ~60 cm⁻¹. In addition to the bands with E¹_2g and A_1g peaks, the Raman spectrum of WS₂ also contains higher-order vibrational modes, namely 2LA(M) and 2LA(T) [7]. Similar Raman features are also found in other layered transition metal dichalcogenides, such as WSe₂ and MoS₂. The main difference in the Raman spectra of WS₂, WSe₂ and MoS₂ is the number and position of their characteristic peaks. MoS₂ typically exhibits two prominent peaks, the in-plane E¹_2g mode around 383 cm⁻¹ and the out-of-plane A_1g mode around 407 cm⁻¹. In contrast, WSe₂ exhibits more peaks in its first-order spectrum, peaks located around 137 and 248 cm⁻¹ and mode A_1g around 251 cm⁻¹. The position of the single peak in few-layer samples of WSe₂ shifts with the number of layers, unlike MoS₂, which exhibits a clear evolution in the spacing between its two main modes [3,8,9,10].

Analysis of the characteristic Raman modes (E¹_2g and A_1g) provides insights into the crystalline quality, structural defects, and internal strain, all of which significantly influence the electrical and chemical behavior of the material. Shifts or intensity variations in the Raman bands upon gas exposure allow the monitoring of interactions between WS₂ and gas molecules, thereby contributing to the understanding of the charge transfer mechanisms involved in gas sensing. Furthermore, Raman spectroscopy enables the identification of different structural phases (e.g., 2H and 1T), which exhibit distinct electrical conductivities and consequently different sensing responses. Due to its capability for in situ measurements, Raman spectroscopy is also well suited for studying the dynamics of gas adsorption and desorption processes, making it an indispensable tool for optimizing the sensitivity, stability, and repeatability of WS₂-based gas sensors [11,12,13].

2. Materials and Methods

2.1. Preparation of WS₂ Films

The preparation of WS₂ films consists of the following steps. The substrate material is sapphire, and the plane on which the films were deposited has a crystallographic orientation C-plane [0001] (Cryscore optoelectronics Ltd., Jiaozuo, China). Here, we expect van der Waals epitaxy to play a major role as C-plane provide a hexagonal structure, which may align with WS₂ films under certain conditions. All sapphire substrates were cleaned ultrasonically using a sequence of acetone, isopropylalcohol, and deionized water.

WS₂ films were deposited onto polished C-plane sapphire substrates by magnetron sputtering from a WS₂ target that was declared by the manufacturer as stoichiometric (99.9% purity, fy MaTeck, Juelich, Germany) at room temperature. The sputtering vacuum chamber was evacuated to a pressure below 10–4 Pa before starting deposition using a turbo pump (Turbo HiPace 2300, fy Pfeiffer Vacuum, Prague, Czech Republic). Argon (99.999% purity, fy Linde Gas, Bratislava, Slovakia) inert gas flow 27 sccm was controlled by a mass flow controller (fy MKS Inc., Brno, Czech Republic). The total argon sputtering gas pressure was kept at 0.5 Pa and adjusted by a piezoceramic valve. A sputtering power of 45 W was used. The distance between the substrate and the WS₂ target was approximately 75 mm, and the substrate holder was in a constant position during sputtering without rotation. The thickness of WS₂ films was measured after deposition by Talystep through a photolithographically prepared step.

The second technological step was sulfurization. WS₂ films of different thickness were placed in the center of the heating zone together with another zone of sulfur powder (1 g of sulfur) and heated at a rate of 25 °C/min to a temperature of 800 °C. The annealing time at 800 °C was 30 min. Sulfurization was carried out in a nitrogen atmosphere at ambient pressure. After sulfurization, the temperature was decreased at a rate of 20 °C/min. Spontaneous cooling followed from a temperature of 200 °C.

The basic technology characteristics of the three samples presented in this work are in

Table 1. Basic technology characteristics of the three samples of WS₂.

Identification of WS2 Samples		Magnetron Sputtering Deposition		Sulfurization
		Deposition Time	Thickness of Deposited WS2 Film	Sulfurization Temperature	Sulfurization Time
1	WS92A	1 min	14 nm	800 °C	30 min
2	WS92B	2 min	30 nm	800 °C	30 min
3	WS91A	3 min	45 nm	800 °C	30 min

. The atomic ratio of W:S and impurities in the deposited WS₂ nanostructures were not investigated.

2.2. Characterization of WS₂ Films

Raman measurements of the WS₂ material sputtered on sapphire and subsequently sulfurized were conducted using two Raman spectrometers. The first Raman investigations were performed on confocal Raman microscope (alpha 300R, WITec, Ulm, Germany) using an excitation laser with a 532 nm wavelength. The laser power was kept below 1 mW. The spectral resolution of the entire Raman system was about 0.75 cm⁻¹. The second Raman investigations were carried out through micro-Raman measurements of the deposited WS₂ films performed using an ISA Dilor-Jobin YvonSpex Labram (HORIBA group division (FRANCE)) confocal system with 632.8 nm radiation from a He-Ne laser, using a back-scattering geometry and a laser power of 5 mW. Microscope objectives 10× and 80× were used to focus the laser beam onto a spot of approximately 5 µm in diameter, and to collect the scattered light which then passed through the spectrometer onto a CCD detector. Furthermore, a confocal hole with a diameter of 600 µm, a spectrograph entrance slit of 600 µm, and 600 grooves/mm diffraction grating were employed. Calibration measurements performed on single crystalline Si (100) showed a measurement uncertainty of ±2 cm⁻¹ in comparison with the catalogue value for silicon 520.7 cm⁻¹. All Raman spectra were obtained at room temperature. At least three measurements of Raman spectra were performed on individual samples at different locations. After Raman investigations, the morphologies of WS₂ films were observed by field emission scanning electron microscopy (FESEM, Quanta FEG 250, FEI Company, Hillsboro, OR, USA) in secondary electron (SE) mode. Electrical characterization was performed using the Ossila Four Point Probe (Ossila Ltd., Sheffield, UK), and activation energy calculations were carried out based on the Arrhenius plots of the samples.

3. Results and Discussion

3.1. FESEM Observations

The nanostructured WS₂ films were formed by sputtering WS₂ stoichiometric bodies of 14 nm, 30 nm and 45 nm thickness and by sulfurization under the same conditions for all samples listed in Table 1. Figure 2, Figure 3 and Figure 4 show the morphologies of nanostructured WS₂ films using FESEM. In Figure 2, a discontinuous seed layer of crystals is seen. We estimate that WS₂ of thickness 14 nm contains 20 layers [14]. In Figure 3, growing crystals are clearly visible, and in Figure 4, a flake nanostructure of WS₂ is developed. The WS₂ flakes are not uniformly distributed and are differently oriented. WS₂ crystals have strong in-plane chemical bonds but weak interplanar bonds, so they can form nanoflakes or nanolayers that are nanometers thick in one direction but can be micrometers wide and their active surface area can exceed 1000 square meters per gram [1,15].

3.2. Raman Spectroscopy

Figure 5, Figure 6 and Figure 7 show the Raman spectra of nanostructured WS₂ films. For an excitation wavelength of 532 nm, we observed prominent Raman peaks of WS₂ at 350 cm⁻¹ and 417–418 cm⁻¹. The first mode is asymmetric, and after deconvolution, it is clear that it is composed of the 2LA peak at ~350 cm⁻¹ and the E¹_2g at 355 cm⁻¹. The overlap of the E¹_2g and 2LA bands makes their location less reliable. Wavenumber separation of A_1g and E¹_2g Raman phonon modes excited by green laser is approximately 67 cm⁻¹ for sample WS92A, 67 cm⁻¹ for sample WS92B, and 69 cm⁻¹ for sample WS91A.

In the graphically recorded Raman characteristics of WS₂ samples obtained by excitation with a He-Ne laser (632.8 nm), two bands with centers at 345 cm⁻¹ to 349 cm⁻¹ are dominant, which approximately corresponds to the E¹_2g mode (in-plane vibrations of W and S atoms) and a band with a center at approximately 413–415 cm⁻¹, which corresponds to the A_1g mode (out-of-plane vibrations of S atoms). The Raman characteristics of WS₂ of the 632.8 nm laser excitation (Figure 5b, Figure 6b and Figure 7b) confirmed the fact that the distance between the dominant bands A_1g and E¹_2g increases with the increase in the thickness of the WS₂ film, i.e., with the increase in the number of layers. The difference between the recorded peaks of the dominant bands of the three samples under study is approximately 66 cm⁻¹, 67 cm⁻¹ and 68 cm⁻¹. As the thickness of WS₂ films increases from WS92A and WS92B to WS91A, the intensity of the Raman response also increases. The intensity ratio A_1g/E¹_2g of the Raman peaks varies depending on samples. The ratio is greater than 1 and increases with thickness samples. With the increase in the number of WS₂ layers, the band with the E¹_2g peak shifts to lower wavenumbers (see

Table 2. The value of separation and intensity ratio of the A_1g and E¹_2g.

Identification of WS2 Samples	532 nm Laser Excitation
	A1g (cm−1)	FWHM	E12g (cm−1)	FWHM	A1g–E12g (cm−1)	A1g/E12g Intensity Ratios
WS92A	416.75	7.25	349.79	11.54	66.96	0.63
WS92B	416.82	6.82	349.87	11.41	66.95	0.54
WS91A	418.20	6.30	349.28	11.19	68.92	0.76
Identification of WS2 Samples	632.8 nm laser excitation
	A1g (cm−1)	FWHM	E12g (cm−1)	FWHM	A1g–E12g (cm−1)	A1g/E12g Intensity Ratios
WS92A	415.28	14.13	349.19	16.18	66.09	1.01
WS92B	413.51	11.91	346.32	14.54	67.19	1.02
WS91A	413.74	10.22	345.22	14.39	68.52	1.04

).

3.3. Analysis of Raman Spectral Characteristics of WS₂ Films

In the Raman spectra of some locations of the WS92A and WS92B samples, one of the main E¹_2g bands is missing when excited by both lasers, so we do not see in-plane vibrations of tungsten and sulfur atoms. The decrease in the intensity of the E¹_2g mode may be related to defects, mainly sulfur vacancies, or amorphous regions in the WS₂ layers, where certain vibrations are not effectively scattered and the mode may be invisible and lost in the noise. This effect, of E¹_2g vibrations not being effectively scattered, was observed most in the thinnest WS92A film when excited by both lasers. In multilayer WS₂, the layers are bound to each other by van der Waals forces, and these interlayer interactions can change the dynamics of vibrations, which also leads to the attenuation of the E¹_2g mode. According to [3,4,14], E¹_2g vibrations can be suppressed even in monolayers. We assume that lattice defects have a significant effect on the Raman spectrum of the studied samples. The shift of the E¹_2g band to the side of smaller wave numbers and the moving away of the E¹_2g and A_1g bands from each other can be attributed to the effect of defects, and we mainly consider the effect of vacancies due to the deficit of sulfur atoms in the technological process. Vacancies change the local strain in the lattice. Defects disrupt the periodicity of the crystal lattice, which leads to a decrease in phonon coherence and therefore to a broadening (FWHM) of the full width at half maximum of both characteristic bands of E¹_2g and A_1g. The FWHM of both dominant modes of an ideal WS₂ crystal or monolayer is 3 to 6 cm⁻¹. The broadening of bands A_1g and for E¹_2g of samples WS92A, WS92B and WS91A when excited by a 532 nm laser and 632.8 nm laser shows a decreasing trend (see Table 2).

In addition to the effect of the shift of the dominant band E¹_2g relative to A_1g, we also observed other modes that are not typical in the Raman spectrum of WS₂ but may also be related to defects in the lattice, different orientations of layers or domains in the material, namely modes in the low-frequency region of the Raman spectrum at 260 cm⁻¹, 290 cm⁻¹ and 320 cm⁻¹ depending on the excitation.

The effect of higher photon energy of the excitation laser and shorter wavelength (532 nm) was manifested by the enhancement of the E¹_2g mode compared to the A_1g mode in all WS91A, WS92A and WS92B samples (see Figure 5a, Figure 6a and Figure 7a). The ratio of the peak intensity of the A_1g to E¹_2g bands ranges from 0.5 to 0.7 depending on the thickness of the WS₂ film in the Raman spectra excited by the 532 nm laser. At the same time, the intensity of the E¹_2g Raman modes excited by the green laser at most of the monitored locations of the WS92A and WS92B samples is higher in these thinner WS₂ layers compared to the Raman response of the thickest WS91A sample (see Figure 7a vs. Figure 5a and Figure 6a).

We explain the decrease in Raman intensity at a WS₂ film thickness of 45 nm (WS91A) for green excitation by the fact that the intensity of Raman modes may decrease with increasing WS₂ thickness if there are more homogeneities that reduce the effective Raman response.

The effect of the lower photon energy of the He-Ne laser and its longer wavelength (632.8 nm) was manifested by the enhancement of the A_1g mode with a stronger Raman signal compared to the excitation of the A_1g mode by a green laser, which is related to better penetration into the material of the WS₂ samples. For excitation with a 632.8 nm laser, the ratio of the intensity peaks of the A_1g to E¹_2g bands is ~1 and does not depend on the thickness of the WS₂ film of the monitored samples. The absence or weakening of the E¹_2g mode in the Raman spectrum of WS₂ can also be the result of the influence of experimental factors, such as the back-scattering configuration of the experiment, but also problems related to the observed sample, like impurities or inhomogeneities of the layer surface; the choice of the monitored location on the sample can also affect the intensity of the Raman signal.

We did not analyze the 2LA mode in the samples studied. The second-order longitudinal acoustic mode 2LA in the Raman spectrum of WS₂ excited by a He-Ne laser may be present or overlapped or absent or significantly attenuated near the E¹_2g band. To ensure the detection of the 2LA mode, it is important to optimize the experimental conditions or to make a theoretical estimate of its position by calculation.

Since we could not perform a compositional analysis of the studied WS₂ films, it is apparent from the study of available sources that there are interrelationships between chemical composition and structure. It was found that chemical composition of WS₂ films (determined, e.g., by EDS) has a direct influence on their Raman spectral characteristics. Deviations from the ideal W:S ratio of 1:2 typically indicate sulfur deficiency and the formation of sulfur vacancies, which strongly affect the vibrational modes of the material. Raman spectra of sulfur-deficient WS₂ commonly exhibit a slight blue shift of the in-plane E¹_2g mode and a red shift with reduced intensity of the out-of-plane A_1g mode. These spectral changes originate from local distortions of W–S bonds and modifications in interlayer coupling caused by missing sulfur atoms. Additionally, the presence of sulfur vacancies often leads to band broadening or the emergence of defect-induced Raman features, reflecting increased structural disorder. The combined analysis of EDS and Raman data thus provides complementary insight: while EDS confirms the elemental composition, Raman spectroscopy reveals the structural and bonding consequences of sulfur deficiency. Importantly, a moderate concentration of sulfur vacancies can enhance the gas sensing performance of WS₂ by providing additional active sites for gas adsorption and charge transfer, highlighting the crucial interplay between composition, structure, and functionality [13,16,17].

3.4. Electrical Characterization of WS₂ Films

The measurement setup consists of four non-destructive, round, soft-tipped probe contact used to measure sheet resistance. The system operates by applying a DC current between the two outer probes and measuring the resulting voltage drop between the two inner probes. All four probes are equally spaced, with a distance of 1.27 mm between each probe. For the most accurate results, measurements are typically taken at the center of the sample. From the Arrhenius equation, we know that

$$\sigma ={\sigma }_0 e^{-{\mathrm{E}}_{\mathrm{a}}/\mathit{KT}}, {\mathrm{E}}_{\mathrm{a}} = K \times 1000 \times M \left(\mathit{slope\; of\; graph}\right)$$

where T is temperature, K is the Boltzmann’s constant, E_a is activation energy, σ is conductance (S) dependent on temperature, and σ₀ is constant.

Sample WS92A, deposited at 1 min, was not characterizable by the four probe method, due to very high resistivity. Thinner films have comparatively higher resistivity, which is beyond the compliance of the instrument. The WS₂ films deposited in 2 min (WS92B) and 3 min (WS91A) and sulfurized with 1 g of sulfur have extremely high sheet resistance of 260 MΩ and 58 MΩ, respectively. Since sheet resistance is an intrinsic property of the material, it reflects the electrical characteristics of the sample. The measurements were taken at ambient room conditions. To avoid for potential variations in sheet resistance due to heat loss, each reading was recorded after achieving thermal equilibrium. The measurement temperature range is between 21 °C and 112 °C.

All resistance values given are the average of the measured values, as each measurement is made after reaching thermal equilibrium, so slight inaccuracies or errors may occur. Mathematically, the graph was linearly approximated with R² = 0.99 with negligible inaccuracies. Slope (92B—blue) = −0.96512 ± 0.00987, and slope (91A—orange) = −0.75897 ± 0.00753.

From the Arrhenius plots (see Figure 8 and

Table 3. Calculated activation energy values.

Identification of WS2 Samples	Deposition Time	Sheet Resistance (Four Probes Method)		Activation Energies (eV)
		Mean (Ohm/sq)	Standard Deviation
WS92A	1 min	---	---	---
WS92B	2 min	2.6 × 108	0.49 × 108	0.19 eV
WS91A	3 min	5.8 × 107	1.06 × 107	0.15 eV

and

Table 4. The separation of the A1g and E12g vs. activation energies.

Identification of WS2 Samples	532 nm Laser Excitation	632.8 nm Laser Excitation	Activation Energies (eV)
WS92A	66.96	66.09	--
WS92B	66.95	67.19	0.19 eV
WS91A	68.92	68.52	0.15 eV

), the activation energy for the conductivity of sulfurized WS₂ films was determined to be E_a ~ 0.19 eV for the layer sputtered for 2 min (WS92B) and E_a ~ 0.15 eV for the layer sputtered for 3 min (WS91A). These (low) activation energy values indicate that the electrical conductivity in the investigated temperature range is conditioned by charge carriers at shallow energy levels in the energy gap, i.e., defects (e.g., grain boundaries) and surface disorder states and traps (e.g., sulfur vacancies). The conductivity, as evident from the plot on Figure 8, therefore increases with temperature due to thermal excitation of charge carriers from these shallow states into the conduction band. Intrinsic conductivity, which is controlled by an intrinsic band mechanism, cannot be considered. In the case of sputtered WS₂ layers, the activation energy would have to be approximately E_g/2 ≈ 1.0 eV (E_g ≈ 2.3 eV) to excite electrons from near the Fermi level. From the perspective of research and development of WS₂ materials, this means that WS₂ in a state of low intrinsic conductivity shows increased sensitivity to surface changes, because energy defect states close to the conduction band (with a depth of ~0.15–0.19 eV) are easily ionized. For even small energy stimuli, e.g., temperature or change in surface potential or adsorption of molecules, if a gas molecule binds to the surface, the local carrier density changes and the conductivity of WS₂ changes noticeably even at low analytic concentrations. The low activation energy in WS₂ material means that the conductivity is controlled by shallow defect states, which is advantageous for sensor applications in gaseous environments. Both DFT (density functional theory) calculations and DLTS (deep level transient spectroscopy) measurements on layered (monolayer and multilayer) WS₂ show that sulfur vacancies form localized defect states just below the conduction band, i.e., they behave as shallow donor levels. This is also the reason why WS₂ films mostly exhibit n-type conductivity [18,19].

4. Conclusions

From a microscopic point of view, all tested samples have a relief character; therefore, it is not possible to record Raman maps of the samples. However, at individual points of all samples, it is possible to identify the crystalline morphological structure of WS₂ using the characteristic bands E¹_2g and A_1g when excited by the 532 nm laser and the 632.8 nm laser. According to the high photoluminescence background of the spectra, it can be concluded that the WS₂ nanoflakes layer contains crystallographic defects of various types. The intensity of the E¹_2g mode band is also probably slightly reduced in connection with layer defects. Defects also suppress or scatter out-of-plane vibrations, so the high intensity of the A_1g mode decreases relative to E¹_2g. The presence of defects in the lattice and different orientations of layers or domains are also indicated by sidebands next to the dominant bands, especially in the low-frequency part of the spectra. The asymmetric shape of the E¹_2g band also indicates lattice defects and defect zones in the film. The shift of the E¹_2g band to the side of smaller wavenumbers after excitation by a He-Ne laser and the moving away of the band peaks from each other with increasing material thickness can be interpreted as the influence of the increasing number of defects, for example grain boundaries, interstitial impurities, amorphous regions, but mainly sulfur vacancies, which change the local strain in the lattice; subsequently local defects also affect the energy structure of WS₂, and specifically sulfur vacancies lead to donor states near the conduction band.

From the Arrhenius plots, the activation energy for the conductivity of sulfurized WS₂ films was determined to be E_a ~ 0.19 eV for the film sputtered for 2 min (WS92B) and E_a ~ 0.15 eV for the film sputtered for 3 min (WS91A). The low activation energy indicates that the conductivity of the WS₂ material produced in our laboratory is controlled by the detected shallow defect states, which is advantageous for sensor applications in gaseous environments.