Changyong
Lan
*a,
Xinyu
Jia
a,
Yiyang
Wei
a,
Rui
Zhang
a,
Shaofeng
Wen
a,
Chun
Li
a,
Yi
Yin
a and
Johnny C.
Ho
bc
aState Key Laboratory of Electronic Thin Films and Integrated Devices, School of Optoelectronic Science and Engineering, University of Electronic Science and Technology of China, Chengdu 611731, China. E-mail: cylan@uestc.edu.cn
bDepartment of Materials Science and Engineering, City University of Hong Kong, 999077 Hong Kong, P. R. China
cInstitute for Materials Chemistry and Engineering, Kyushu University, 816-8580 Fukuoka, Japan
First published on 9th November 2023
Two-dimensional (2D) materials have become a hot topic in materials science, electronics, optoelectronics, and other fields. However, the practical applications of 2D materials rely heavily on the reliable synthesis of large-area, high-quality materials, which still poses a significant challenge. In this study, we present a detailed investigation into the Na2WO4-assisted synthesis of WS2. Our findings reveal that the substrate temperature and the sequence and duration of introducing S vapor are critical factors in manipulating the morphology of the WS2 products. Monolayer, thick film, and one-dimensional nanostructures can be obtained by varying the substrate temperature and the introduction sequence of S vapor. Furthermore, the introduction sequence and duration of S vapor can significantly impact the monolayer films' optical and electrical properties. Films synthesized with the introduction of S vapor before the evaporation of the W source exhibited strong photoluminescence (PL) emission, with a greater contribution from excitons. In contrast, films synthesized with the introduction of S vapor after the evaporation of the W source showed reduced PL emission, with a greater contribution from trions. Additionally, field effect transistors based on films synthesized with the introduction of S vapor before the evaporation of the W source displayed a larger threshold voltage and higher electron mobility. These findings suggest that the Na2WO4-assisted synthesis method for WS2 is highly controllable and pave the way for utilizing these monolayer WS2 materials for technological applications.
The utilization of molten salt has been demonstrated to effectively lower the melting point of metal oxide, leading to a significant increase in the vapor pressure of metal oxide, which facilitates the growth of various types of 2D materials.33–35 Recently, we have reported quasi-epitaxial growth of large single-domain monolayer WS2 films on sapphire substrates by employing Na2WO4.36 Nevertheless, a comprehensive investigation of the growth parameters, particularly the substrate temperature and duration of introducing the S vapor, has yet to be conducted. In this study, we carried out a detailed investigation to study the impacts of substrate temperature, sequence, and duration of introducing S vapor on the growth behavior of WS2. We found that the substrate temperature, the S vapor introduction sequence, and duration are critical factors in manipulating the morphology of WS2 products. Wire-like structures are obtained at low substrate temperatures via the vapor–liquid–solid mechanism. The high substrate temperature is beneficial for the formation of a continuous monolayer film. The duration of introducing S vapor can significantly impact the optical and electrical properties of the monolayer films. These findings suggest that the Na2WO4-assisted synthesis method for WS2 is highly controllable and pave the way for utilizing these monolayer WS2 materials for technological applications.
Fig. 1 Schematic of the setup and temperature profiles. (a) Schematic of the setup. (b) Temperature profiles for zone I, zone II, and the heating belt. (c) Temperature distribution in zone II. |
To begin with, we investigate the effects of introducing S vapor before the evaporation of the W precursor. At a lower substrate temperature of 800 °C, most of the resulting WS2 products are deposited in subzone A, as observed on the corresponding substrates (Fig. 2a). The quantity of deposition materials decreases as the distance from zone I increases towards subzones B, C, and D. At a higher substrate temperature of 900 °C, a continuous film is formed in all four subzones. Nonetheless, the continuous film exhibits numerous thick domains that increase in quantity with distance from zone I. At a medium temperature, the deposition quantity decreases in subzone A and increases in other subzones (Fig. 2b).
Fig. 2 Optical microscope images of the products obtained in zone II when S was heated before the evaporation of the W source. The temperatures of zone II are (a) 800 °C, (b) 850 °C, and (c) 900 °C. |
In zone I, WO3 reacts with Na2WO4 to form NaxWOy, which is in a liquid state above approximately 700 °C.37 At a temperature of 930 °C, NaxWOy vaporizes, forming a vapor. Prior to the evaporation of NaxWOy, S is introduced, which reacts with the evaporated NaxWOy to form WS2 species. As a result, although some NaxWOy species may still be transported to zone II, most of the vapor species transported to this zone are WS2 species. According to the classical nucleation theory, the driving force for nucleation is supersaturation. The chemical potential difference (Δμ) between the deposited crystal and the vapor phase can be estimated by Δμ = kTln(s), where k is the Boltzmann constant, and T is the temperature in Kelvin. s is defined as P/Peq, where P is the actual partial vapor pressure of WS2, and Peq is the saturation vapor pressure above the crystal.38 The growth rate of the crystals is expected to vary linearly with s. The saturation vapor pressure Peq is strongly related to temperature. Usually, Peq depends exponentially on temperature, such that a small decrease in temperature leads to a significant reduction in Peq and an increase in s. At a low deposition temperature of 800 °C, a large temperature gradient is observed in subzone A (Fig. 1c), resulting in the fast growth of WS2 crystals. Additionally, the nucleus density is proportional to s, which explains why the thick deposition with small crystals is observed in subzone A. Because most of the source species are exhausted in subzone A, the source species transported downstream to zone II become scarce, reducing P and, subsequently, s. Consequently, the growth rate decreases, leading to sparser WS2 flakes in the downstream side of zone II. This observation indicates that the growth of WS2 at a low deposition temperature is limited by mass transport. As the deposition temperature increases, the temperature gradient decreases in subzone A, leading to a slower crystal growth rate. Accordingly, the growth of WS2 in subzone A is less mass-transport limited. However, for a setting temperature of 900 °C, the temperature gradient in subzones C and D becomes large, resulting in thick deposition in these areas.
Based on the above discussion, attaining a homogeneous temperature distribution is crucial for obtaining a uniform film. A setting temperature of 900 °C for zone II can achieve this homogeneity in subzones A and B. The homogeneous temperature distribution enables the production of a uniform monolayer film. However, thick domains observed in Fig. 2c may indicate overgrowth, which prompted an investigation into the time evolution of WS2 growth at a temperature of 900 °C in zone II (Fig. 3). With a short growth time of 15 min, dispersed monolayer WS2 triangles are observed in subzones A, B, and C, while subzone D shows a continuous film with thick domains and particles. The latter observation can be attributed to the large temperature gradient in the subzone. When the growth time is increased to 25 min, a continuous monolayer film is formed in subzones A, B, and C. In contrast, a thick film was formed in subzone D. These results suggest that an appropriate growth time is critical for attaining the continuous monolayer WS2 film in subzones A, B, and C. In any case, an excessively long growth time of 40 min leads to overgrowth, resulting in a film with numerous thick domains (Fig. 2c). This indicates that the growth of WS2 is not fully self-limited, and defects in the monolayer WS2 may act as nucleation sites for the overgrowth of WS2.
Fig. 3 Optical microscope images of zone II taken as a function of growth time. The temperature in zone II is set to be 900 °C. |
Next, we investigate the impact of introducing S vapor after the evaporation of the W precursor. At a deposition temperature of 800 °C, most of the WS2 crystals are deposited in subzone A, as shown in Fig. 4a. Increasing the temperature to 850 °C would result in the observation of monolayer WS2 flakes collected in subzone A, accompanied by a decrease in the quantity of the monolayer WS2 in subzone B (Fig. 4b). In subzones C and D, thick flower-like WS2 materials are witnessed. When the temperature of zone II is raised to 900 °C, the continuous monolayer WS2 film can be obtained in subzones A, B, and C, but thick dispersed WS2 domains are still present in subzone D (Fig. 4c).
Fig. 4 Optical microscope images of the products obtained in zone II when S was heated after the evaporation of the W source. The temperatures of zone II are (a) 800 °C, (b) 850 °C, and (c) 900 °C. |
At a temperature of 930 °C in zone I, NaxWOy vaporizes and is transported to zone II by the carrier gas. Without introducing the S vapor, liquid droplets of NaxWOy may form on the substrates in zone II. The deposition rate of these liquid droplets is linearly carried with s, where s = P/Peq. Here, P is the actual partial pressure of NaxWOy, and Peq is the saturation pressure above liquid NaxWOy. At a lower temperature of 800 °C, a large temperature gradient is formed in subzone A, resulting in the formation of numerous NaxWOy droplets on the substrate. When S vapor is introduced, the liquid droplets would react with S to form WS2 crystals. These crystals aggregate later to form a thick WS2 film. Since most of the NaxWOy species are deposited in subzone A, the amount of NaxWOy vapor transported downstream to subzones B, C, and D is significantly reduced, leading to a decrease in the product in those regions. When the temperature in zone II is increased to 850 °C, the density of NaxWOy droplets is reduced due to a decrease in the temperature gradient in subzone A. The reaction of the NaxWOy droplets with S vapor results in the formation of thick WS2 crystals on the substrate, which act as nuclei for the monolayer WS2 growth. The temperature gradient in subzone A is greater than in subzone B, resulting in a faster WS2 growth rate in subzone A and larger monolayer WS2 triangles. As the temperature gradient gradually increases, the density of droplets in subzone C and subzone D increases, forming a flower-like product. When the temperature in zone II reaches 900 °C, a uniform monolayer WS2 film forms in subzones A, B, and C, indicating the uniform deposition of WS2 on the substrates. However, the large temperature gradient in subzone D would lead to thick WS2 deposition. The above results and discussion indicate that achieving a uniform temperature distribution in the deposition zone is crucial for forming a uniform monolayer WS2 film. Additionally, it is important to precisely control the growth time to prevent overgrowth.
Notably, introducing S at an appropriate time can lead to the formation of nanowires/nanobelts on substrates when the temperature of zone II is 850 °C, as shown in Fig. S1 (ESI†). Subzone A exhibits a thick WS2 film due to a significant temperature gradient, while nanowires and nanobelts are observed in subzones B, C, and D. The formation of one-dimensional (1D) nanostructures via the vapor–liquid–solid (VLS) mechanism can be readily explained due to the formation of NaxWOy droplets on the substrate. In zone II, NaxWOy species are deposited onto the substrate as tiny droplets, reacting with the S vapor to form WS2 that dissolves in the droplets. Upon reaching supersaturation, WS2 is deposited onto the substrate through continuous deposition from the droplets, ultimately creating 1D WS2 nanostructures. Previous studies have reported the growth of MoS2 nanobelts on the NaCl substrate, with the formation of one-dimensional (1D) MoS2 nanostructures attributed to the VLS mechanism.39 Here, the construction of NaxWOy liquid droplets is critical to the growth of 1D WS2, and thus, S should be introduced after the formation of NaxWOy liquid droplets and not too late. Otherwise, tiny droplets will aggregate into larger ones that are too heavy for the deposited WS2 to push away, forming only large crystals. Small NaxWOy liquid droplets can be formed accidentally in other synthesis conditions, resulting in the frequent observation of 1D-like WS2 nanostructures in zone II (Fig. S2, ESI†).
Based on the obtained results, we propose mechanisms for the Na2WO4-assisted growth of WS2, as illustrated in Fig. 5. When S is introduced before the evaporation of NaxWOy (Fig. 5a), most of the vapor species transported downstream are WS2. At high deposition temperatures, the deposition of WS2 is slow, resulting in the growth of monolayer WS2 triangles and, thus, a monolayer film. However, if the growth time is too long, overgrowth may occur on the defects or grain boundaries of the as-formed monolayer, leading to films with thick domains. At low deposition temperatures, the deposition of WS2 is fast, leading to the growth of thick WS2 triangles and, thus, a non-uniform thick WS2 film. When S is introduced after the evaporation of NaxWOy (Fig. 5b), the vapor species transported downstream in the early stage is NaxWOy. The NaxWOy liquid droplets can form on the substrate, and when S is introduced, NaxWOy would react with S to form WS2 species that can dissolve in the droplets. After supersaturation, WS2 will deposit from the droplets. If the droplets are large enough, WS2 will gradually deposit from them, forming large crystals. If the droplets are of adequate size, a VLS mechanism can occur, forming 1D WS2. If the droplets are too small, they will transform into WS2 nanocrystals immediately upon reacting with S, which will act as the nuclei of monolayer WS2, forming a monolayer WS2 film. In fact, most monolayer WS2 triangle flakes exhibit a thick dot at their center (Fig. S3, ESI†). It is worth noting that the thick dots were consistently observed at the center of monolayer WS2 triangles, regardless of the sequence and duration of the S introduction. This observation suggests the crucial role of the thick dot in the growth of monolayer WS2. In addition, the monolayer WS2 film can be scaled up if a horizontal tube furnace with a larger diameter and longer heating zone is used, according to the growth mechanism.
Fig. 5 Growth mechanism. (a) S is introduced before the evaporation of NaxWOy. (b) S is introduced after the evaporation of NaxWOy. |
To assess the quality of the monolayer film deposited at 900 °C, Raman and PL spectroscopy were employed. The film synthesized with S introduction before the evaporation of the W source is labeled as sample #1, and the film synthesized with S introduction after the evaporation of the W source is labeled as sample #2. Both films exhibited similar Raman spectra (Fig. 6a), consistent with the monolayer WS2.40,41 However, their PL spectra are different, as shown in Fig. 6b. The PL spectrum of sample #1 is stronger than that of sample #2, and the peak shapes are different. Deconvolution of the peaks (Fig. 6c and d) reveals three peaks corresponding to exciton (X0: ∼2.01 eV), trion (XT: ∼1.98 eV), and defect-related emission (XD: ∼1.93 eV).42 The intensity ratios between X0 and XD are 1.38 and 0.42 for samples #1 and #2, respectively, indicating that the contribution from XT is larger for sample #2. The formation of XT is strongly related to the free electron concentration in WS2, with higher concentrations contributing more to the PL emission.43 Therefore, sample #2 has a higher free electron concentration. Though higher electron concentration is beneficial for the formation of trions, the overall possibility of the formation of both excitons and trions reduces, leading to the weak PL intensity for sample #2.44 Intrinsic n-doping for WS2 is commonly attributed to S vacancies,45 inferring that the higher free electron concentration in sample #2 suggests a higher defect concentration overall. Due to the introduction of S vapor before the evaporation of NaxWOy for sample #1, a reduced number of S vacancies is anticipated compared to sample #2, which is consistent with the observed PL results. These results suggest the high crystal quality of sample #1 in comparison with sample #2.
Back-gated field effect transistors (FETs) were fabricated to evaluate the electrical properties of the films. The structure of the devices is shown in the inset of Fig. 6e, SiO2 with a thickness of 285 nm was used as the dielectric and n+-doped Si was used as the backgate electrode. Fig. 6e shows the typical transfer curves of FETs based on samples #1 and #2, indicating that both exhibit n-type conductivity as the drain–source current increases with the gate voltage. Both curves exhibit hysteresis due to the charge traps in the WS2 channel, and/or in the interface between WS2 and SiO2.46 Log plots of the transfer curves are presented in Fig. 6f to provide a better understanding of the FETs. Both FETs' current on/off ratio exceeds 106, indicating a good device performance. Based on sample #1, the FETs display a larger threshold voltage than those based on sample #2, suggesting the higher n-doping in sample #2, consistent with the PL results. To further confirm this hypothesis, threshold voltages were extracted from tens of FETs, and the statistical diagrams are presented in Fig. 6g, which demonstrate that FETs based on sample #2 have a smaller threshold voltage. This result again indicates that sample #1 has a lower electron concentration than sample #2. The electron mobilities were extracted from the transfer curves, and the statistical diagrams are presented in Fig. 6h. FETs based on sample film #1 exhibit a higher average mobility, indicating better quality than sample #2, consistent with the observed PL results. Notably, both the threshold voltages and mobilities were extracted from the forward sweeping for simplicity. Due to the small hysteresis, the difference for forward and backward sweeping is small.
Footnote |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ma00867c |
This journal is © The Royal Society of Chemistry 2023 |