Since the discovery of graphene in 2004, there has been a surge in research focused on the properties of two-dimensional (2D) materials [1]. Graphene, which has exceptional electronic and structural properties, has been used for applications in transistors [2–5], photovoltaics [6–10], and photodetectors [11]. Its inherent 2D nature has sparked interest in its use in flexible electronic components [12–14]. However, the drawbacks of graphene are its stringent growth requirements, which are primarily constrained to specific substrates, and the challenging nature of the associated manufacturing processes [15–21]. Furthermore, the absence of a sizeable band gap in graphene limits its use in electronic devices.

To address these shortcomings, research focus has shifted considerably toward transition metal dichalcogenides as promising 2D materials. Among these, 2D MoS₂ exhibits a large direct bandgap (approximately 1.9 eV) and strong photoluminescence properties, particularly at the monolayer level [22,23]. These attributes make MoS₂ a versatile candidate for integration into field-effect transistors (FETs) [24–27] and sensor [28–32] applications. Furthermore, its potential spans a broad spectrum of optoelectronic devices, including photodetectors [33–36] and photovoltaics [37,38]. Additionally, single-crystalline MoS₂ can spontaneously nucleate even on amorphous films, such as SiO2, allowing more seamless heterogeneous integration with other semiconductors. However, despite its potential application in diverse fields, achieving a uniform thickness of MoS₂ over a large area remains challenging.

Conventionally, the most widely recognized method for 2D MoS₂ synthesis involves the chemical vapor deposition of MoO₃ and S precursors. However, this method faces challenges in terms of precise thickness control and large-area growth. Recent research has succeeded in achieving monolayer MoS₂ growth on a scale of up to 4 in [39]. Nevertheless, uniformity becomes a limiting factor as the growth area increases. Therefore, an innovative solution-based thermolysis synthesis method was developed using a Mo precursor [40]. This approach facilitates precise control of MoS₂ thickness through the manipulation of precursor [(NH₄)₂MoS₄, (NH₄)₆Mo₇O₂₄·4H₂O, Na₂MoO₄] concentration. This method allows precise control over the MoS₂ thickness by adjusting the precursor concentrations, achieving large-scale growth of up to 6 in [41–46]. Furthermore, its solutionbased process characteristics allow its growth on flexible substrates, and 20-inch large-scale growth using roll-to-roll processes has been demonstrated [47]. However, there is limited systematic research on the thickness uniformity and crystallinity of MoS₂ during solutionbased thermolysis.

In this study, we investigated the process-dependent (i.e., precursor concentration, temperature, spin-coating speed, and choice of solvent) thickness uniformity and crystallinity of 2D MoS₂ synthesized by solution-based thermolysis.

2.1. 2D MoS₂ synthesis

A 2 × 2 cm² c-plane sapphire substrate was ultrasonicated for 5 min in acetone, isopropyl alcohol, and deionized water. Subsequently, ultraviolet (UV)-ozone treatment (20 mW/cm²) was performed for 30 min to improve the hydrophilicity of the sapphire surface before spin coating. The precursor solutions were prepared using (NH₄)₂MoS₄ (Sigma Aldrich, 99.97 %) with N, N-Dimethylformamide (DMF), a DMF + n-butylamine mixture, and a DMF + n-butylamine + ethanolamine mixture at concentrations of 0.024, 0.012, and 0.0048 M.

The solution was sonicated for 30 min before spin coating, which was performed at speeds of 2,000 and 4,000 rpm, followed by baking at 60 °C for 15 min. Subsequently, a MoS₂ thin film was synthesized via a two-step annealing process in a tube furnace. During the first annealing process, the film was held in an N₂/H₂ gas atmosphere at a pressure below 1 Torr, and the temperature was maintained at 300 °C for 30 min to remove solvent residue. The second annealing process occurred in an N₂ atmosphere at various temperatures (700, 800, 900, and 1,000 °C) for 30 min. In the second annealing step, additional S powder was heated to provide subplementary sulfur, preventing the decomposition of MoS₂ during the process and preventing MoS₂ oxidation. A schematic of the process flow is shown in Fig. 1.

Figure 1. Process flow of solution-based thermolysis for 2D MoS₂ synthesis

2.2. Characterization

Raman spectra of the synthesized 2D MoS₂ samples were obtained by dm500i (Dongwoo Optron) Raman spectrometer using a 532-nm laser line and a spot size of 1 μm. The surface morphology of the MoS₂ samples was analyzed by field emission scanning electron microscopy (FE-SEM) using a JSM-6701F (JEOL) and by atomic force microscopy (AFM) using an NX20 (Park Systems) instrument.

The crucial factors to consider in the solution-based synthesis of 2D MoS₂ are the uniformity and crystallinity of the layer. Choosing a solvent with high polarity that effectively dissolves the precursor (NH₄)₂MoS₄ is essential. Among the candidate solvents, such as DMF, dimethyl sulfoxide, and ethylene glycol, selecting a solvent with a low viscosity is crucial for uniform coating. Therefore, DMF with a viscosity of 0.92 mPa·s was chosen. In terms of the process, a one-step annealing method for growing MoS₂ directly at elevated temperatures is not suitable for obtaining uniform stoichiometric MoS₂ because of the possibility of decomposition of MoS₂ in a strongly reducing environment [43]. Hence, we adopted a two-step annealing process [48]. In the first step, the spin-coated sample was annealed at 300 °C under the N₂/H₂ ambient. The gas was introduced for 30 min, while maintaining a pressure of less than 1 Torr. This step led to the decomposition of (NH₄)₂MoS₄ into MoS₃. The first annealing step ensured that any residue from this step did not interfere with the subsequent MoS₂ growth in the second annealing step. The process temperature was set at 300 °C to minimize decomposition of MoS₂ in the presence of H₂ [49]. In the second annealing step, the process temperature was increased to 700–1,000 °C, and the atmosphere inside the reactor was adjusted to N₂ gas. During the second annealing, vaporized sulfur was introduced using N₂ as the carrier gas, facilitating a 30 min synthesis of MoS₂. A sulfur-rich environment was created to promote the formation of stoichiometric MoS₂ [43].

3.1. Temperature-dependent thickness and crystallinity of MoS₂

The thickness and crystallinity of the MoS₂ were analyzed using Raman spectroscopy, specifically by examining the spacing and full width at half-maximum (FWHM) of the E¹_2g and A_1g peaks. Figure 2 shows the peak spacing in the range of 23.12–25.31 cm⁻¹ (four to five layers) [43], regardless of the concentration of the solution and speed at which the solution is coated. Surprisingly, the Raman data indicate that an increase in the annealing temperature led to an increase in the MoS₂ thickness. The observed decrease in FWHM from 700 (7.44 cm⁻¹) to 1,000 °C (5.47 cm⁻¹) suggested an improvement in MoS₂ crystallinity. The diffusion rate of sulfur (S) through the layers and parallel diffusion along the van der Waals bonding plane in MoS₂ varies with annealing temperature. Therefore, the increase in the thickness with increasing annealing temperature can be attributed to the diffusion of S through the layers, provided that the precursor concentration is sufficiently high. However, as discussed below, the overall thickness uniformity could only be enhanced with a proper solvent, even with 900 and 1,000 °C annealing [50,51].

Figure 2. Raman spectra of MoS₂ with the precursor concentration and spin-coating speed of (a) 0.024 M and 2,000 rpm, (b) 0.024 M and 4,000 rpm, (c) 0.012 M and 2,000 rpm, and (d) 0.012 M and 4,000 rpm with various annealing temperatures (700–1,000 °C) during the second annealing step.

3.2. Thickness control

Using DMF, additional analysis was conducted on the synthesis and thickness control of MoS₂ by changing the first annealing atmosphere from N₂/H₂ to H₂. The first annealing was conducted at 300 °C in an H₂ gas atmosphere, while the second annealing remained consistent with the previous process, using only N₂ gas at 900 °C. For this analysis, Raman spectra were collected from various regions of the sample, as shown in Fig. 3(a). Possibly due to a stronger reducing environment created by the H₂-only atmosphere, thickness of MoS₂ could be reduced to two to three layers, as suggested by the Raman peak spacing of 19.67 cm⁻¹ in Fig. 3(a). The Raman spectra obtained from five different regions of the sample showed almost identical profiles with similar peak spacings. However, the AFM analysis [Fig. 3(b)] revealed that a triangular MoS₂ domain was formed, and the MoS₂ film was not continuous. The formation of the MoS₂ domains was attributed to the use of DMF, as discussed below.

Figure 3. (a) Raman spectra from five different regions of a MoS₂ sample (solvent: DMF only, first annealing with H₂ at 300 °C) and (b) AFM image.

3.3. Surface coverage and thickness uniformity

Surface coverage and thickness uniformity strongly depend on the viscosity of the solvent. DMF has the lowest viscosity among the solvents tested in this study; however, the SEM images in Fig. 4 suggest that the solution with only DMF shows incomplete surface coverage due to the aggregation of the MoS₂ precursor [indicated by the white circles in Fig. 4(a)], regardless of the precursor concentration. This is likely the reason why the DMF-based process showed no concentrationdependent thickness control (Fig. 2) and poor surface coverage [Fig. 3(b)]. Additionally, pinholes were observed in the synthesized MoS₂, indicating nonuniform growth. This was further confirmed by Raman spectroscopy, which was conducted at five different points on the sample, as shown in Fig. 5. For the samples processed at a low rpm, peaks were observed only near the center of the sample, with no MoS₂ peaks detected in the side regions. Although this trend was less pronounced in the samples processed at high rpm, there were still points on the sample where MoS₂ peaks were not detected. This was attributed to the formation of ionic thiomolybdate ($Mo{S}_4^{2-}$) clusters when DMF was coated onto the substrate surface. As shown in the SEM image, this cluster formation led to the aggregation of the powder, causing the observed clustered morphology [52].

Figure 4. Top-view SEM images of MoS₂ with the precursor concentration and spin-coating speed of (a) 0.024 M and 2,000 rpm (solvent: DMF only) with the white circles indicating the features formed due to agglomeration of the precursor and (b) 0.012 M and 2,000 rpm (solvent: DMF only).

Figure 5. Raman spectra from five different regions of MoS₂ samples with (a) 0.024 M and 2,000 rpm (solvent: DMF only) and (b) 0.0012 M and 2,000 rpm (solvent: DMF only).

Another solvent containing an amine group was added to DMF to enhance the coating uniformity. Initially, a solution was prepared using DMF and n-butylamine with a viscosity of 0.5 Pa·s in a 1:1 ratio. Despite improving the wettability of the sapphire substrate surface through UV-ozone treatment before spin coating, the solution did not coat the substrate surface because of its lower viscosity and adhesion compared to the original solution, regardless of the concentration. To address this issue, ethanolamine, a higher viscosity amine group solvent with a viscosity of 24.1 mPa·s, was added in the subsequent step [53]. The two-step annealing process was performed using with a 0.024-M and 0.0048-M solutions. With a 0.024-M solution, the peak spacing was approximately 25.31 cm⁻¹, indicating a 5-layer thick MoS₂ film. Raman spectra were obtained at five different points across the sample, as shown in Fig. 6(a). When the precursor concentration of the solution was lowered to 0.0048 M, the peak spacing became approximately 19.67 cm⁻¹, indicating MoS₂ with two to three layers. With the thinnest MoS₂ sample synthesized at 900 °C shown in Fig. 6(b), the Raman spectra were taken from 12 different points across the sample with an average spacing of 19.6 cm⁻¹ and a standard deviation of 0.417. This result confirms the successful synthesis of uniformly thick MoS₂. Figures 3 and 4 suggest that the control parameters, such as concentration and annealing temperature, had minimal impact on the domain size without a uniform distribution of precursors. Compared with the MoS₂ films shown in Figs. 3(b) and 4, these films are continuous. A top-view SEM image in Fig. 6(c) taken from the 900 °C sample shows no distinguishable domains. The AFM image in Fig. 6(d) also suggests the absence of microscale domains. The root-meansquare roughness of the surface is 0.34 nm.

Figure 6. Raman spectra from different regions of MoS₂ samples with (a) 0.024 M (solvent: DMF + n-butylamine + ethanolamine mixture), (b) 0.0048 M (solvent: DMF + n-butylamine + ethanolamine mixture), (c) top-view SEM image, and (d) surface morphology of MoS₂ synthesized with the mixed solvent of 900 °C synthesized MoS₂.

In this study, solution-based thermolysis of 2D MoS₂ was investigated by exploring the impact of process variables (i.e., coating speed, solvent viscosity, solution concentration, annealing ambient temperature, and annealing temperature) on the crystallinity, thickness uniformity, thickness controllability, and surface coverage of 2D MoS₂. First, the coating speed had no impact on the thickness control. Raman analysis showed that the crystallinity of MoS₂ improved as the process temperature increased. Although there were variations in the growth thickness of the 2D MoS₂ with varying temperatures in the second annealing step, the impact was considered marginal. Although DMF has a low viscosity, aggregation of the precursors leads to poor surface coverage and a lack of thickness control by the precursor concentration. With addition of an amine group solvent to prevent the formation of ionic thiomolybdate ($Mo{S}_4^{2-}$) clusters, the films exhibited excellent surface coverage and uniform thickness in the 2-by-2 μm² sample. The number of layers of MoS₂ could also be adjusted by changing the concentration of (NH₄)₂MoS₄ in the solution.

This research was subported by the BK21 Four project funded by the Ministry of Education, Korea (2120231314753) and the Basic Research Laboratory program funded by the Ministry of Science and ICT (RS-2023-00222070).

The authors declare no conflicts of interest.