Keywords

1 Introduction

Because of their distinctive optical, electrical and mechanical properties, two-dimensional (2D) materials such as graphene and transition metal dichalcogenides (TMDs) are promising candidates for the high performance electronics and optoelectronics needed for next-generation sensing and energy applications [1,2,3]. For example, monolayer WS2 in the TMDs family has been reported as a promising strain sensing that are integrated into infrastructure such as building windows [5]. To realize their ultimate utilization in diverse applications, a feasible non-defective and quality-preserving transfer technique is essential to create functional devices or van der Waals heterostructures [6, 7]. The polymer-assisted wet transfer technique using poly(methyl methacrylate) (PMMA) is widely used because of its better applicability and material conservation [4, 8, 9]. However, this method usually involves soaking the sample in heated alkaline solution, resulting in chemical doping as well as a certain level of damage to the transferred materials [4]. To overcome this drawback, hydrophobic polystyrene (PS) has been adopted as the supporting polymer layer for the Chemical Vapor Deposition (CVD)-grown WS2 single crystal transfer process, which enables less exposure of the materials to the alkaline solution [10]. Nevertheless, there is still a lack of detailed characterization of the crystals before and after transfer to fully understand the doping effect of the transfer process.

Furthermore, because of their atomic thickness and reduced dielectric screening, the properties of 2D materials are greatly influenced by the underlying substrate [11]. For 2D TMDs, the reduced Coulomb screening gives rise to high exciton binding energies and quasiparticles of trions, which are charged excitons. Trions have been observed in doped monolayer TMDs [12]. The characteristics of excitons and trions, which significantly affect electronic transport and optical transitions in the monolayer TMDs, are found to be strongly coupled with the dielectric environment [13], as well as doping effects [14] induced by the underlying substrate. In this regard, non-uniform photoluminescence (PL) emissions of CVD-grown TMDs crystals on dielectric substrates are widely observed [4, 15,16,17,18]. However, the underlying physics have not been clearly demonstrated and further investigation is required to understand the mechanism of this non-uniform PL behavior.

In this study, the distinctive effects of the substrate and the influence of doping from the transfer process on the optical properties of CVD-grown monolayer WS2 triangular crystals were demonstrated. Optical and morphological characterizations indicated that the non-uniformity of the monolayer WS2 PL behavior is highly likely related to the intercalation of a layer of ambient water molecules between the WS2 crystal and the underlying hydrophilic substrate, where the intercalated water changes the dielectric environment and the doping state. Comparison was made between the commonly used PMMA transfer and the modified PS transfer methods in transferring single WS2 crystals from as-grown sapphire substrates to target SiO2/Si substrates. Detailed morphology mappings showed that PS transfer maintained better preservation of the crystals in transfer using PMMA, due to the avoidance of soaking in alkaline solution. Optical characterizations of crystals before and after PS transfer indicated that the residual alkali metal ions facilitated the formation of trions in monolayer WS2 under ambient conditions, leading to variations in PL behavior. In contrast to the transferred WS2 crystal on hydrophilic SiO2/Si substrates, where non-uniformity of PL along the edges remained, the transferred WS2 on hydrophobic PS presented uniform PL emission across the crystal, supporting the theory that water intercalation is the source of the inhomogeneous PL behavior on hydrophilic dielectric substrates. After removal of alkali dopants by annealing at 100 °C in an argon gas environment, the trion peak diminished, with only the exciton peak contributing to the PL emission, indicating that the trion peak formation closely correlated with the electron doping caused by alkali metal ions.

2 Methods

2.1 CVD Growth of Monolayer WS2 Crystals

The WS2 crystals were grown by CVD on substrates of sapphire [Al2O3 (0001)] using WO3 and S powders as precursors. The detailed CVD setup for monolayer single WS2 crystal growth can be found in Zhang et al. [4].

2.2 Polymer-Assisted Wet Transfer Using PMMA and PS

For PMMA wet transfer, the CVD-grown monolayer WS2 on the sapphire substrate was spin-coated at 3000 rpm for 60 s by PMMA (A4) and soft baked at 80 °C for 3 min. Next, the spin-coated sample was soaked in 2 mol/L KOH solution and heated to 100 °C on a hot plate for 1 h. Before soaking, the edges of the PMMA thin film on the sample were scratched with a scalpel to facilitate penetration of alkali between the polymer film and the substrate. After soaking, the PMMA film was separated from the as-grown sapphire substrate at a deionized (DI) water surface with the help of surface tension. The separated PMMA film floated on the surface of the DI water with the WS2 crystals attached. The film was then fished out by the target substrate from underneath (WS2 side). The PMMA film was removed by soaking in acetone and isopropanol (IPA) in turn.

For the PS transfer, PS (Mw ~ 192,000) in toluene solution (50 mg/mL) was used to spin-coat as-grown WS2 on a sapphire sample. The steps of the PS transfer process were the same as those introduced in Xu et al. [10]. Similarly, the PS film was fished out by SiO2/Si from underneath (WS2 side) and the polymer was washed off by acetone and IPA. A further cleaning step of soaking the sample in PG remover was used for thorough removal of the PS. To transfer WS2 onto a hydrophobic PS surface, a rigid substrate of SiO2/Si was used to attach the detached PS film from the top, leaving the WS2 sitting on the PS surface.

2.3 Optical and AFM Characterizations

Raman and PL characterizations were performed with a confocal microscope system (WITec alpha 300R) with ×50 objective lens and 532 nm laser excitation. A weak laser power of 50 µm and a short integration time of 1 s was adopted. The PL intensity mappings were obtained by totaling the PL intensity from 1.9 to 2.1 eV. The Raman and PL spectra were collected using a 600 line mm−1 grating. The atomic force microscopy (AFM) measurements were carried out on the Bruker Dimension Icon in tapping mode.

3 Results and Discussion

Characterizations of a representative as-grown monolayer WS2 crystal by CVD on an atomically flat sapphire substrate are presented in Fig. 1. The optical image in Fig. 1a shows a typical triangular single WS2 crystal ~5 µm in size with uniform contrast. The AFM morphological mappings of the crystal (Fig. 1b, c) revealed that the grown WS2 followed the atomic steps of the surface of the bare sapphire with step heights of 0.2 nm, indicating uniform height and atomic flatness. However, it was observed that the heights of the edges were slightly raised by ~0.37 nm, as shown in Fig. 1c, g. This was inferred to be raised by a single layer of water molecules from the ambient environment that intercalated between the sapphire and the monolayer WS2 crystal. This intercalated water layer can induce localized strong dielectric screening to the monolayer WS2 because water has a much higher dielectric constant value than sapphire [19].

Fig. 1
Six contour plots and 3 line graphs. A. Optical image. B and c. A F M images. D. Raman intensity. E. P L intensity. F. P L peak position map. G. Height nanometers versus distance micrometers. H. Intensity versus Raman shift. I. Intensity versus energy in electron volts of points 1, 2, and 3.

Characterizations of a CVD as-grown monolayer WS2 crystal. a Optical image; b AFM height scan; c zoomed-in AFM mapping of the region shown in the white square in (b); d–f Raman intensity, PL intensity and PL peak position mappings; g height profile along the yellow line in (c); h Raman spectrum collected in (d); i PL spectra taken from points 1–3 indicated in (e, f). Scale bars: 5 µm in (a, b, d–f); 1 µm in (c)

Full size image

This finding was further supported by the optical features reflected by Raman intensity, PL intensity and PL peak position mappings for this WS2 crystal on sapphire. A Raman spectrum collected from the crystal (Fig. 1h) shows a distinctive in-plane E′ Raman peak of monolayer WS2 at 350 cm−1. Although uniform Raman signals were detected across the crystal (Fig. 1d), distinctive non-uniform PL behavior was observed (Fig. 1e, f) where the edges of the crystal exhibited stronger PL emissions and redshifted peak positions. The regions of edges with varied PL behavior were consistent with the regions where the height was raised by water intercalation, indicating that the trapped water had a notable influence on the optical properties of the monolayer crystal.

To further understand the inconsistent PL behavior, three PL spectra were collected at points 1–3 from the edge to the center of the crystal as indicated in Fig. 1e, f and shown in Fig. 1i. Information regarding the three spectra is summarized in Table 1. Lorentzian fitting indicated that the PL spectra at points 2 and 3 featured a single peak each at 1.998 eV, coinciding well with the exciton peak of the CVD WS2 monolayers on sapphire [20]. For the PL spectrum collected at point 1 located at the edge of the crystal, however, the peak position was redshifted. Lorentzian fitting revealed two distinctive peaks of an exciton (\({X}^{0}\), magenta curve) at 1.999 eV and a trion (\({X}^{-}\), cyan curve) at 1.975 eV, with an energy difference of 24 meV. This value was in good agreement with the negative trion binding energy of the monolayer WS2 [21]. The exciton to trion intensity ratio was 1.4, agreeing well with the value for the monolayer WS2 with 10 V positive gate voltage [21], indicating n-type doping in the water-intercalated regions.

Table 1 Spectral information for WS2 crystal on sapphire (corresponding to Fig. 1i)
Full size table

Besides the change in peak composition and the shift in peak position, the integrated PL emission at the edge with water trapped underneath (point 1) also had fivefold higher intensity than the inner region of the crystal (points 2 and 3). This was likely due to the high dielectric constant (κ) of the water trapped underneath, which led to prolonged lifetimes of the quasiparticles and increased recombination efficiency of the excitons and trions [22, 23].

Next, the influence of the wet transfer method on the properties of the transfer monolayer WS2 was investigated. First, the morphological changes of the crystal transferred by processes using different polymers were studied. The transfer steps with the two different polymers (i.e., PMMA and PS) are schematically illustrated in Fig. 2a. As can be seen, compared with PMMA transfer, the PS film could be directly separated from the sapphire at the KOH solution surface with minimal exposure to the alkaline environment. The effect of the alkali exposure is reflected in the morphological characterization of the transferred crystals presented in Fig. 2b–e. Obvious damage can be observed along the edges of the PMMA-transferred crystal, with jagged shapes evident by the optical contrast (Fig. 2b) and AFM scan (Fig. 2d). This damage to the edges was presumed to be caused by KOH etching from the long soaking at elevated temperature [10]. Specifically, the raised edges of the crystals by ambient water intercalation prior to the transfer process caused them to be more easily subjected to etching by the alkali. In contrast, the morphology of the WS2 crystal transferred by PS indicated good preservation, including that at the edges (Fig. 2c, e).

Fig. 2
Five images. A. 2 flow diagrams. 1. P M M A, 1-hour soaking, and separation at the water interface. 2. P S, direct separation at K O H interface, and rinsing by water. B and d. Optical and A F M images of P M M A transfer. C and e. Optical and A F M images of P S transfer.

a Schematic of PMMA- and PS-based wet transfer methods. b, c Optical images of PMMA- and PS-transferred WS2 crystals on SiO2/Si; d AFM mapping of the PMMA-transferred crystal shown in the square in (b); e AFM mapping of the PS-transferred crystal in (c). Scale bars: 10 min (b, c); 2 min (d); 5 min (e)

Full size image
Fig. 3
Eight images and 2 graphs. Four pairs of images indicate the optical images, P L intensity mappings, P L peak position mappings, and P L width mappings of the crystal as transferred on silicon dioxide or silicon using P S and P G remover washed. I and j. Intensity versus energy for points 1, 2, and 3.

Optical characterization of the PS-transferred WS2 crystal onto SiO2/Si. a Optical image, b PL intensity mapping, c PL peak position mapping and d PL width mapping of the crystal as transferred onto SiO2/Si. e Optical image, f Raman intensity mapping, g PL intensity mapping and h PL peak position mapping of the crystal transferred onto SiO2/Si and soaked in PG remover. i PL spectra taken at points 1–3 in (b–d). j PL spectra taken at points 1–3 in (g, h). Scale bars: 5 µm in (a–h)

Full size image
Table 2 Trion and exciton peak information in spectra for WS2 crystal transferred onto SiO2/Si by PS transfer (corresponding to Fig. 3i)
Full size table

However, PS cannot easily be removed by acetone and IPA washing, so polymer residue on the crystals can be observed. Therefore, PG remover, an NMP-based solvent stripper, was further used to thoroughly remove the PS residue. As shown in Fig. 3, the polymer residue was cleanly removed with edges well preserved. These results suggested that PS transfer was superior to PMMA transfer, with better preservation of the morphology of the transferred 2D crystals.

The change in the optical properties of the WS2 crystal after the PS transfer process was investigated for the same crystal characterized in Fig. 1. To avoid confusion, the crystal before PG remover washing is here referred to as the “as-transferred crystal” and the crystal after PG remover washing is denoted as the “PG washed crystal”. Figure 3a–d shows the optical images and PL mappings of the as-transferred crystal. As can be seen clearly, non-uniform behavior of PL remains after the crystal was transferred onto SiO2/Si, with the edges exhibiting stronger PL intensity, redshifted peak position and narrowed peak width. Because SiO2 has a hydrophilic surface and a very small dielectric constant of 3.9 compared with water, it was deduced that this PL non-uniformity was induced by water intercalation. It has been reported that the trapped water can be removed by annealing at high source-drain bias or prolonged baking in air [24, 25]. Figure 3i and Table 2 shows three PL spectra collected at points 1–3 from the edge to the center. Unlike the spectra taken before the transfer, all three PL spectra for the as-transferred crystal feature trion emissions comparable to excitons. This strongly suggested that the WS2 crystal was subjected to doping after the transfer process. The doping was presumed to have been induced by the alkali metal ions of K+ from the transfer process acting as electron donors. The notable PL peak position blueshift compared with before transfer (~25–35 meV) can be explained by the release of strain induced by high temperature CVD growth after the PS transfer process [20].

After PG washing, the crystal was free of polymer residue, as indicated by the optical image in Fig. 3e. However, the optical characterizations suggested severe degradation of the optical properties of the crystal, reflected in the weak signals in Raman and PL intensity mappings in Fig. 3f–h. The PL spectra fittings in Fig. 3j show a weak single peak at the low energy of 1.950–1.975 eV for each spectrum of points 1–3, consistent with a defect-bound exciton emission derived from defects across the crystal [26]. Further study is required to address the negative impact of PG remover on the transferred WS2.

To further study the doping induced by alkaline solution, a CVD as-grown monolayer WS2 crystal was transferred onto a hydrophobic PS surface, using the PS wet transfer to eliminate the influence of intercalated water. As can be seen from the PL intensity, peak position and peak width mappings in Fig. 4a–c, the as-transferred WS2 crystal on PS exhibited uniform contrast over the entire crystal, with no variation along the edges. This observation strongly supported the theory that the non-uniform PL behavior along the edges stems from ambient water intercalation between the WS2 crystal and the underlying hydrophilic substrate. The PL spectra collected at points 1–3 on the as-transferred crystal on PS are shown in Fig. 4g, with details presented in Table 3. The features of the three spectra are nearly identical, implying consistent optical behavior over the crystal without trapped water along the edges. The negative trion emission across the crystal with a binding energy of 28 meV was strongly indicative of n-type doping of the alkali metal ions after the PS transfer process.

Table 3 Trion and exciton peak information in spectra of WS2 on PS by PS transfer (corresponding to Fig. 4g)
Full size table
Fig. 4
Six images and 2 graphs. Three pairs of images, a to f, indicate the P L intensity mappings, P L peak position mappings, and P L width mappings of the crystal as transferred on P S and as transferred on P S with Argon annealed at 100 degrees Celsius. G and h. Intensity versus energy for points 1, 2, and 3.

Optical characterization of a WS2 crystal transferred onto hydrophobic PS using the PS wet transfer method. a PL intensity, b PL peak position and c PL width mappings of the crystal as transferred onto PS. d PL intensity mapping, e PL peak position and f PL width mappings of the crystal transferred onto PS and annealed by Ar at 100 °C. g PL spectra taken at points 1–3 in (a–c). j PL spectra taken at points 1–3 in (d–f). Scale bars: 2 min (a–f)

Full size image

Ar annealing of the crystal at 100 °C for 1 h was then performed to confirm the effect of alkali doping. The optical mappings of the crystal after Ar annealing are presented in Fig. 4d–f, with the PL spectra at points 1–3 shown in Fig. 4h. At elevated temperatures, impurities, including the alkali dopants, could be detached from the crystal and exhausted by Ar gas flow. Thus, after Ar annealing, it can be seen in Fig. 4h that only a single peak was required to fit the PL spectrum for each point. The peak positions are at ~2.013 eV, corresponding to exciton emission energy. The suppression of trion peaks and the decrease of the PL intensity indicated that electron doping of the crystal was significantly diminished by removal of residue alkali during the annealing process. The exciton peak position was blueshifted compared with that of the crystal as transferred onto PS (2.003 eV) as a result of reduced electron doping, which also led to decreased intensity and narrowed width of the PL peaks after annealing.

4 Conclusions

We investigated the effects of polymer-assisted wet transfer process and type of substrate on the optical properties of CVD-grown monolayer WS2 crystals. We found that transfer using PS instead of the traditionally adopted PMMA led to better preservation of the morphology of the transferred WS2 crystals due to minimized exposure to alkali solution. The electron doping induced by the alkali metal ions from the transfer processes caused trion formation as well as intensified PL emission across the transferred crystal. Annealing by Ar at 100 °C can efficiently remove alkali residue on the transferred crystals. We also demonstrated that ambient water intercalation between the edges of the monolayer crystal and the underlying hydrophilic substrates induces variations in PL due to doping and dielectric screening effects. The non-uniform PL behavior of the WS2 crystals disappeared when the crystals were transferred onto a hydrophobic PS substrate. These results contribute to the understanding of trion formation in atomically thin TMDs via chemical doping and non-uniform PL behavior stemming from the dielectric screening effect of water intercalation when the material is deposited on hydrophilic substrates. This work has implications for the provision of consistent crystal behavior of 2D TMDs for their applications in sensing and energy harvesting and conservation devices.