Two-dimensional (2D) materials with their unique properties originating from the quantum confinement offer more development choices for modern electronic devices [1–6]. Up to now, the vdWHs have been evolved into several functional devices such as field-effect transistors [7], high-sensitivity chemical sensors [8], and photodetectors with high performance [9–11]. Notably, the van der Waals interlayer interaction provides the vdWHs with more novel characteristics than that of the monolayer 2D materials [12–16]. One of the new characteristics is the remarkable photoluminescence (PL) quenching phenomenon from neutral excitons occurring in type-II heterojunctions as a result of enhanced PL from the local excitons. So vdWHs provide a unique platform for studying the circular polarization properties of localized excitons [17–22]. The present work demonstrates a way of capture and study of the distinct circular polarization properties of the vdWHs [23].

To date, most of the reported vdWHs were made with 2D materials by micro-mechanical peeling, a process limiting their large-scale production and practical applications. Meanwhile, the observation and investigation of circular polarization of local excitons in WSe₂/WS₂ heterostructures have been challenging. The physical vapor deposition method avoids the above problems because of the closed preparation environment with the protection of inert gas. With the clean interface and high quality, the physical vapor deposition products are often utilized in the excitons research and non-linear optics [24–28]. In this article, we demonstrate a bidirectional-flow physical vapor deposition method for growing vertical WSe₂/WS₂ heterostructures in large quantities. During the preparation process, a reverse carrier flow was used for cooling to reduce undesirable nucleation during the growth of the WSe₂/WS₂ heterostructures. The sharp and clean interfaces in such heterostructures enable the photoluminescence of neutral excitons in the constituent monolayers to be quenched significantly, creating more local excitons. Additionally, the existence of the circular polarization of local excitons in the WSe₂/WS₂ heterostructures was successfully proved with the polarization degrees of 7.17% for σ^- detection and 4.78% for σ⁺ detection. In a recent study, a single-photon emission from the localized excitons was observed in monolayer TMDC materials [29]. Local excitons play a critical role in quantum emitters with enhanced spontaneous emission rates [30,31]. The enhancement of the spontaneous radiative emission is crucial for the design of LEDs [32]. The present study of localized excitons in vdWHs may provide insight for the development of new LEDs [33].

The high-quality WSe₂/WS₂ heterostructures were synthesized by a two-step physical vapor deposition method. In previous experiments, the mechanical transfer method was widely used. While the mechanical transfer method is more primitive, during the transfer process, the nanosheets could be lightly damaged, wrinkled, blistered, and contaminated with organic matter. Therefore, a large number of defects would be introduced. Because of the high density of defect states, multiple defect exciton peaks could be detected within a narrow band range in the same PL spectrum, leading to a rather complex spectrum for the identification and investigation of the local excitons. Also, the density of native defects in the CVD-grown product is much higher than the mechanical exfoliated product, which is not beneficial for the identification of excitons [34]. The physical vapor deposition method, however, has less control over site specific growth as the nucleation spots are rather random. It is also more costly and time consuming compared with the mechanical transfer method.

By contrast, the physical vapor deposition method is efficient in producing high-quality vdWHs with uniformity and few defects. High-quality products also make it possible to study a single defect. Moreover, compared to traditional preparation, the physical vapor deposition method can yield more heterostructures. As shown in Fig. 1(a), the vertical heterostructures were synthesized in a tube furnace (OTF-1200X-II KJ Group) using a physical vapor deposition system at atmospheric pressure. In the traditional CVD growth process, thermal degradation and undesired nucleation are the critical barriers for stable formation of heterostructures [35,36]. A tube furnace with an upgraded pneumatic system could solve these problems [37]. For the first step, the WS₂ powder was placed in an aluminum oxide boat, which was propelled into the heating zone of a 1-inch quart tube. Then a piece of polished SiO₂ (285nm)/Si was placed (facing up) in another boat at the downstream position as the growth substrate. The center zone of the furnace was heated to 1160 ℃ under atmospheric pressure with a reverse flow of 250 sccm argon gas. When the target temperature was reached, the WS₂ vapor source was carried downstream by a forward flow of 100 sccm argon gas for a growth period of 2 minutes. After the WS₂ growth, the furnace cooled down naturally with an inverse argon flow of 250 sccm. The second step was the epitaxial growth of WSe₂ on top of WS_2. The SiO₂/Si substrate with the as-grown WS₂ nanosheets was placed at the same position in another similar physical vapor deposition system for the WSe₂ growth. During the temperature-increasing stage, a reverse flow of 250 sccm cold argon gas cooled the existing WS₂ nanosheets to restrain thermal degradation. The center zone of the furnace was heated to 1100 ℃ under atmospheric pressure. After reaching the desired temperature for WSe₂, the inverse Ar gas was switched to a 110 sccm forward flow to carry the vaporized WSe₂ source to the substrate. Then growth temperature was maintained for 3 minutes for the epitaxial growth of WSe₂. Finally, the substrate naturally cooled down to room temperature with the inverse cold argon flow. The growth process of both films was schematically shown in Fig. 1(b), where the blue triangular nanoplate is WS_2, and the smaller triangular nanoplate on top of WS₂ is WSe₂. Due to the different growth temperatures of TMDCs, the order of heterostructure grown by physical vapor deposition method is fixed. The prepared WSe₂/WS₂ vertical heterostructures are shown in Fig. 1(c). As can be seen, each heterostructure displays two distinct regions of different contrasts. The central triangular region of darker contact is WSe₂, sitting on top of a larger triangular region of lighter contrast.

 

Fig. 1. Growth of the WSe₂/WS₂ heterostructures. (a) Schematic diagram of the tube furnace used for the growth of the WSe₂/WS₂ heterostructures in the experiment. The arrows in different colors show two opposite directions of flow, with the green being the forward flow. (b) Schematic diagram of the physical vapor deposition method for the growth of the heterostructures. The top of the diagram shows the deposition process of molecules for the WSe₂ layer, while the bottom of the diagram indicates the process of the epitaxial growth of WSe₂ on top of the WS₂ layer. (c) Optical images of the WSe₂/WS₂ heterostructures showing the smaller triangular-shaped WSe₂ on top of the relatively larger triangular-shaped WS₂.

Download Full Size | PDF

Raman and PL studies were conducted at a low-temperature using a WITec Alpha 300R Confocal Raman System to confirm the types of excitons and the layer number of the WSe₂ and WS₂. A 532 nm laser was applied to irradiate the heterostructure to obtain Raman and PL spectra and mapping images. As shown in Fig. 2(a), the WS₂ Raman spectrum matches well with the WS₂ spectrum of the WSe₂/WS₂ heterostructures. There were minor Raman shifts between that of WSe₂ and the heterostructure for both the characteristic peaks of out-of-plane A_1g and in-plane E_2g¹ modes as a result of the coupling between WS₂ and WSe₂. The frequency difference between the A_1g and the E_2g¹ modes of WSe₂ is consistent with the splitting of 11-12 cm^-1 in the monolayer WSe₂. These features in the Raman spectra provide evidence that the heterostructure is composed of monolayer WSe₂ and WS₂ [38,39]. The optical microscope images (insets of Figs. 2(c)-(e)) of synthesized WSe₂/WS₂ heterostructures on the SiO₂/Si substrate reveal the two concentric triangular regions of WS₂ and WSe₂ with slightly different optical contrasts. The red regions in the spatially-resolved Raman maps are WS₂ obtained with the A_1g peak of WS₂ at 250 cm^-1, while the green regions are WSe₂ generated with the E_2g peak of WSe₂ at 350 cm^-1. In the Raman mapping images, the heterostructures show distinct boundary and uniform colors in both the overlapping and non-overlapping regions, demonstrating the high quality of the physical vapor deposition product and the advantages of the bidirectional-flow physical vapor deposition method that enabled the subsequent study on the excitons from these heterostructures.

 

Fig. 2. Raman mapping and photoluminescence quenching in the WSe₂/WS₂ heterostructures. (a) Raman spectra of WSe₂ and WS₂ and their heterostructure. The red line is a spectrum from the WSe₂/WS₂, and the blue line is the WS₂ spectrum obtained from the edge of the heterostructure where there was no WSe₂ present. As a comparison, the Raman spectrum (black line) from a monolayer WSe₂ sample is also presented. (b) PL spectra of WSe₂, WS_2, and their heterostructure. (c)-(e) Raman mapping images of the WSe₂/WS₂ heterostructures. Mapped with their characteristic peaks, the green regions are WSe_2, and the red regions are WS_2. The insets are optical images of the WSe₂/WS₂ heterostructures. (f)-(h) PL mapping images of the WSe₂/WS₂ heterostructures showing photoluminescence quenching resulting from the band-structure of the type-II heterostructure. The insets in (c)-(h) are the optical images of the WSe₂/WS₂ heterostructures.

Download Full Size | PDF

As previously reported, the type of excitons in the TMDC heterostructures depends on the band alignment of the heterostructures. To probe the alignment of the band structure in the WSe₂/WS₂ heterostructure in more detail, the heterostructure was investigated by PL spectra (Fig. 2(b)). The PL peak position is to locate at 1.87 eV for WS₂, corresponding to the bandgap of 2-3 atomic layers of WS₂, while the PL peak is at 1.65 eV for WSe₂. Both these PL peaks were observed in the heterostructure, but they were much weaker than that measured in the separate monolayers of WS₂ and WSe₂. The PL intensities of the heterostructure were quenched by a factor of ∼500-1000. As shown in the PL mapping images of Figs. 2(f)-(h), while the PL intensity was strong and uniform outside the heterostructure region, the PL intensity was too weak to be revealed in the heterostructure region, indicating the type-II band alignment of the WSe₂/WS₂ heterostructures. For monolayer band structures, carriers are generated under optical excitation and consumed on account of the radiative recombination. Interestingly, in WSe₂/WS₂ heterostructures, the charge carriers are transferred between the WSe₂ and WS₂ layers induced by the particular type-II band alignment. The hole and electron carriers are separated into different layers, so the PL intensities of intralayer excitons significantly decrease. This dramatic quenching proves that the heterostructures synthesized by the physical vapor deposition method have a sharp interface that helps the transfer of carriers and highlights the defect excitons of the monolayer materials. As a contrast, the neutral excitons in the WSe₂/WS₂ heterostructures prepared through chemical vapor deposition and assembling by a PMMA stamping method were not completely quenched because of the contaminated interfaces. There is another possible way to induce the observed quench phenomenon. A direct-to-indirect bandgap transition also could block the recombination of carriers to quench PL of the WSe₂ and WS₂ nanosheets. However, in the previous report, researchers investigated the origin of PL quench in WSe₂/WS₂ heterostructures and conluded that such an assumption of bandgap transition is highly unlikely experimentally [27]. The observed local excitons in the current study were due to the significant suppression of the PL intensity of neutral excitons as a result of the band alignment in the WSe₂/WS₂ heterostructures.

For a deeper understanding of the localized excitons, we further studied the PL properties of the WSe₂/WS₂ heterostructures under a low-temperature environment (65K-300K). Figure 3 displays the temperature-dependent PL spectra in the different areas of WSe₂/WS₂ nanoplates. In the WS₂ area, there are two emission peaks corresponding to charged exciton peak (X^-) and neutral exciton peak (X⁰) in Fig. 3(a). As the temperature rose, the peak broadened, and the peak position shifted to the low-energy region. The following Varshni equation is commonly used to describe the relationship between the PL peak position and temperature in semiconductors [40]:

where α and β are fit parameters and ${E_g}(0)$ is the bandgap at absolute zero temperature. The Varshni equation was applied to fit the trend of PL peak positions with the changing of temperature, as shown in Fig. 3(c). For the charged excitons in the WS₂ region, the calculated fit parameters are $\alpha \textrm{ = 3}\textrm{.86} \times \textrm{1}{\textrm{0}^{\textrm{ - 4}}}eV/K$, $\beta ={-} 34K$ and ${E_g}(0 )= 2.045\textrm{ }eV$. In Fig. 3(c), the solid red curve shows the fitting result. The calculated parameters are similar to those of a previous study [41]. As the temperature decreased, the peak width of the PL spectrum gradually narrowed, making the identification of the localized excitons easier. Compared with regular excitons, the electrons of the localized excitons were trapped in the bandgap by a potential well (Fig. 3(d)), so the localized excitons (Fig. 3(b)) were induced by defects located below the X^- exciton peak of WS₂ in the heterostructure region. The interaction between atomic-scale defects and excitons occurred noticeably at atmospheric pressure. Therefore the defect-related peaks can be evolved in the PL spectrum. The defects were induced by the vacancies of sulfur atoms stemming from the oxidation of WS₂ molecules after preparation. The current observation and results are consistent with a previous report [42]. Remarkably, the defects of WS₂ in such heterojunctions had only one peak, meaning that only a single defect state was introduced, and this is also the advantage of fabricating samples by the physical vapor deposition technique. More crucially, the recognition of local excitons in such heterostructures enabled the study of circular polarization.

 

Fig. 3. Normalized photoluminescence spectra under variable temperature. (a) The PL spectra of the WS₂ region under different temperatures (65K-300K). The dashed line indicates the trends of peak shifting. (b) The PL spectra of the localized excitons in the WSe₂/WS₂ heterostructures under different temperatures (65K-300K). (c) X^- Peak energy as a function of temperature. The solid line indicates a fit using the Varshni equation. (d) Energy level diagram of WS₂/WSe₂ heterostructure.

Download Full Size | PDF

To better understand the circular polarization of localized excitons, the polarization-resolved photoluminescence spectra of the WSe₂/WS₂ heterostructures were investigated at T = 65K. In Figs. 4(a)-(d), the circular polarization of localized excitons was observed evidently. For instance, the position of the peak near 1.9 eV revealed that there were localized excitons of WS₂ [43]. The degree of the PL polarization can be quantified by the helicity:

The four polarization-resolved PL spectra in the heterostructure regions for two samples (sample 1 for Figs. 4(a) and (b); sample 2 for Figs. 4(c) and (d)) reveal that the different degrees of PL valley-polarized emissions at 1.9 eV of the WS₂ localized excitons are 7.17%, 4.78%, 6.13%, and -0.39%, respectively. The difference between σ⁺ and σ^- light intensities originated from the valley optical selection rule [44–46]. When the sample was excited by σ^- light, carriers were generated at − K valley and recombined to emit σ^- light, while no carriers were generated at + K valley. However, there was considerable emission of σ⁺ light at + K valley because of the intervalley scattering (top panel of Fig. 4(e)). On the contrary, only carriers at + K valley were excited when the incident light was σ⁺ light (bottom panel of Fig. 4(e)).

 

Fig. 4. Valley polarization of localized excitons. (a) and (c) Polarization-resolved PL spectra for the WSe₂/WS₂ heterostructures at 65K using σ- excitation. (b) and (d) Polarization-resolved PL spectra for the WSe₂/WS₂ heterostructures at 65K using σ+ excitation. In (a) and (b), the red line represents the intensity of left-handed light emitted by the heterostructures. The blue line represents the intensity of right-handed light emitted by the heterostructures. The olive-green dotted line indicates the corresponding degree of PL polarization. (e) Schematic diagram of optical selection rules. (f) Schematic diagram of the optical selection rules of the local excitons in WS₂. The purple lines indicate defect energy levels.

Download Full Size | PDF

For sample 1, it exhibited stronger emission of σ^- light than σ⁺ light when excited by σ^- light (Fig. 4(a)). When excited by σ⁺ light, the situation was opposite (Fig. 4(b)). As can be seen in Fig. 4(f), one defect energy level exists below the bottom of the conduction band, containing two quasi-degenerated spin-down and spin-up electronic states [47,48]. The transition between the valence band and the defect energy level holds the same optical selection rules as the transition between the valence band and conduction band at the ±K valley [49]. When the incident light was σ^- light, carriers generated at -K valley, and then the electrons were trapped at the defect energy level. The electrons localized at the defect level and holes localized in valence band combined radiatively, emitting σ^- light. Likewise, the emission of σ⁺ light was also considerable because of the intervalley scattering(top panel of Fig. 4(f)). The opposite situation can be detected when the sample was excited by σ⁺ light (bottom panel of Fig. 4(f)). The eventual degrees of polarization were both positive for sample 1. This phenomenon occurred in five of the eight samples. However, there was also another situation, as shown in sample 2 (Figs. 4(c) and 4(d)). When the incident light was σ⁺ light, the degree of polarization was negative. For sample 2, under the excitation of σ^- light, the σ^- light emission exhibited more vigorous intensity than that of the σ⁺ light. Thus, the polarization degree was positive. While the sample was excited by σ⁺ light, the σ^- light emission still exhibited stronger intensity than that of σ⁺ light and showed a negative polarization degree. The polarization degrees under the excitation of σ^- and σ⁺ lights showed opposite signs. The possible reason might relate to the contributions from multiple excitonic bands and the spin flipping during the charge trapping process [50]. The difference between the two situations originated from a distinct lattice mismatch and the different contributions from excitonic bands. However, the underlying physical principles behind the above phenomenon remain to be further explored.

In summary, we report the synthesis of 2D TMDC heterostructures by a bidirectional-flow physical vapor deposition method. The optical properties of the WSe₂/WS₂ heterostructures were characterized. The PL quenching phenomenon was distinctly observed and could be attributed to the type-II band alignment. The circular polarization of localized excitons in the heterostructure region was studied under low-temperature conditions, and the trends of PL spectra of the charged excitons in WS₂ are in a good agreement with the Varshni’s theory. The novel phenomenon of local excitons needs to be further explored in combination with theoretical calculations. The present work not only helps to understand the behaviors of localized excitons in nano-scale heterostructures but may provide insight towards quantum emitters in TMDC materials.