Scalable synthesis of WS2 on graphene and h-BN: an all-2D platform for light-matter transduction

By exhibiting a measurable bandgap and exotic valley physics, atomically thick tungsten disulfide (WS2) offers exciting prospects for optoelectronic applications. The synthesis of continuous WS2 films on other two-dimensional (2D) materials would greatly facilitate the implementation of novel all-2D photoactive devices. In this work we demonstrate the scalable growth of WS2 on graphene and hexagonal boron nitride (h-BN) via a chemical vapor deposition approach. Spectroscopic and microscopic analysis reveal that the film is bilayer-thick, with local monolayer inclusions. Photoluminescence measurements show a remarkable conservation of polarization at room temperature peaking 74% for the entire WS2 film. Furthermore, we present a scalable bottom-up approach for the design of photoconductive and photoemitting patterns.


Introduction
Two-dimensional (2D) transition metal dichalcogenides (TMDs) have recently raised interest in the scientific community and beyond, due to their remarkable optoelectronic properties [1][2][3] and prospected integration into flexible and transparent platforms. In particular, tungsten disulfide (WS 2 ) has been appreciated as the TMD of choice for photoactive devices. Indeed, atomically thick WS 2 does not only feature a direct optical band gap of about 2 eV at the K and K′ symmetry points of the Brillouin zone [4][5][6], but also exhibits long exciton life and coherence times, causing remarkable photoluminescence (PL) [4]. Most notably, as a direct consequence of giant spin−orbit coupling and spin-valley coupling, bilayer WS 2 presents an astonishing preservation of polarization at room temperature [7]. Up to now, polarization conservation has been achieved in conventional solidstate systems by adopting technologically unpractical cryogenic temperatures. Hence, WS 2 represents an exciting playground for light-matter interaction studies and a promising platform for valleytronic applications. To date, most of the remarkable properties of WS 2 have been observed by using mechanically exfoliated flakes. In particular, valley polarization has been measured for micrometer-sized WS 2 grains on SiO 2 [7,8]. The scalable synthesis of continuous WS 2 films displaying high polarization coherence is instrumental in the implementation of a WS 2 -based valleytronic technology. Moreover, incorporation of such WS 2 films within van der Waals heterostacks would pave the way for the realization of novel all-2D optoelectronic devices. Chemical vapor deposition (CVD) is the most suitable technique for the scalable synthesis of highly crystalline 2D heterostacks [9,10]. However, such an approach is not trivial as weak interlayer interactions favor three-dimensional (3D) island growth [11]. The formation of multi-layer islands is typically avoided by adopting short reaction times which lead, however, to the synthesis of isolated crystals [11][12][13][14]. To date, the only works on direct growth of WS 2 on other 2D-materials have Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI. demonstrated isolated sub-micrometer grains of fewlayer thick WS 2 on graphene [13] and isolated grains on hexagonal boron nitride (h-BN) substrates [14].
In this work we show that continuous atomicthick WS 2 films can be synthesized on appealing 2D substrates, i.e. graphene and h-BN, by using CVD. Spectroscopic, microscopic and PL measurements indicate that on both substrates the synthesized WS 2 is bilayer-thick, with local monolayer inclusions. PL measurements reveal that the polarization of the incident light is highly retained. The polarization anisotropy retrieved from PL polarization maps over areas of thousands of square-microns peaks at 74% at room temperature. These results indicate the suitability of scalable CVD-grown WS 2 films to absorb and conserve quantum information in the form of polarized light and to open exciting prospects for the adoption of van der Waals heterostacks in valleytronics. Furthermore, we present a scalable bottom-up approach for the design of photoconductive and photoemitting patterns.

WS 2 growth
To synthesize WS 2 , we used a vapor-phase reaction from solid sulfur (S) and tungsten trioxide (WO 3 ) powders in a horizontal quartz tube. The selected precursors have previously led to high quality 2D crystals of WS 2 with large domain sizes on classical bulk insulators [4,15,16]. Sulfurization of WO 3 powder (Sigma Aldrich, 99.995%) was carried out within a horizontal hot-wall furnace (Lenton PTF). The furnace comprises an inner hot zone, in which 10 mg of WO 3 were placed, and a cooler outer zone in which 1 g of sulfur (S) (Sigma Aldrich, 99.998%) was placed (about ∼10 cm away from the hot-zone edge). The growth temperature within the hot-zone was set to 900°C. This final temperature was reached at a rate of 1°C s −1 and was kept for 1 h and 30 min. The substrates were placed face-up next to WO 3 powder within the same crucible. Before the temperature ramp up, the chamber was pumped down to a pressure of ∼5×10 −2 mbar. Argon was flown during the temperature ramp with a flux of 500 sccm, leading to a pressure of 4.5 mbar, which kept the sulfur solid. After reaching 900°C, the Ar flux was suddenly reduced to 80 sccm, which reduced the furnace pressure to 1.3 mbar thus initiating sulfur evaporation and WO 3 sulfurization. The combination of a sudden evaporation of S at the growth temperature of 900°C and high reactant flow rates, were found to balance the competing processes of sublimation, reaction, transfer, diffusion and precipitation of the reactants in favor of fast and dense WS 2 growth. WS 2 synthesis was performed on h-BN flakes exfoliated on quartz, CVD graphene transferred on quartz, and epitaxial graphene on silicon carbide (SiC) [17,18]. The choice of transparent substrates anticipates the use of the investigated heterostacks for optoelectronic applications.

PL microscopy
PL microscopy experiments were carried out with a Leica SP5 confocal laser scanning microscope using a 63x (NA 1.2) water-immersion objective. As light source, pulsed solid state laser diode at 640 nm was used (Picoquant). The luminescence arising from the sample was collected through the objective and filtered with a 640 notch filter and with a 610 long pass filter (Chroma). To measure luminescence polarization, a broadband (420−680) polarizer beam splitter was introduced in the light path. The two arising beams are collected with two identical fiber couple SPAPDs (Picoquant). The alignment of the polarizer beam splitter principal axis and the excitation laser polarization plane was proved by an additional internal polarizer filter to be better than 5°.
See supporting information for experimental details about substrates preparation, Raman analysis, transmission electron microscopy (TEM), scanning electron microscopy (SEM) and atomic force microscopy.

Results and discussion
3.1. Strong light polarization conservation in WS 2 synthesized on h−BN We first analyze the properties of WS 2 synthesized on h-BN, an ideal substrate to preserve near-pristine properties of graphene [19] and other 2D materials. Figure 1(a) shows a SEM micrograph of a representative h-BN flake exfoliated on quartz and fully covered with WS 2 . No isolated triangular crystal [11,15,20] is observed and the uniform distribution of WS 2 on the flake is confirmed by Raman spectroscopy, which is instrumental in identifying the nature and the thickness of TMDs [21,22]. The intensity ratio of the inplane vibrational mode E 2g 1 and of the out-of-plane vibrational mode A 1g is indicative of the thickness of WS 2 when using specific excitation wavelengths such as that adopted in this work, i.e., 532 nm [21,23,24]. As visible in figure 1(b), the A E 1g 2g 1 ratio is approaching 1, indicating bilayer coverage [23]. Interestingly, only along h-BN crystal anomalies (clearly visible in the optical image reported in the inset in panel (b)), a smaller ratio is observed, suggesting monolayer-thick regions. We further corroborate the monolayer/ bilayer nature of the synthesized film by analyzing the overtone peak located at 310 cm −1 . This peak, which is attributed to rigid interlayer shear forces, diminishes with the number of layers, and ultimately disappears when approaching the monolayer limit [25,26]. Indeed, as shown in supporting information, the 310 cm −1 peak is barely detectable.
Time-and polarization-resolved PL spectroscopy is used to investigate the optical properties of the synthesized WS 2 film. This technique allows us to extract both the lifetime of the excitons created by the incoming photons and the conservation of their polarization. PL measurements were performed by using linearly polarized laser light with a near-resonant excitation energy (1.94 eV). To quantify the degree of polarization anisotropy of the emission, P, the following relation is used: is the intensity of PL with parallel (perpendicular) polarization with respect to the polarization of the excitation. As expected, PL emission is observed across the entire h-BN flake (see figure 1(c)).
A significantly stronger intensity-compatible with a transition to direct band-gap-is observed along h-BN crystal boundaries (compare map to inset in panel (c)), where monolayer WS 2 patches were indeed detected via Raman. In figure 1(e), we plot the PL intensity histogram (vertical graph), the PL polarization histogram (horizontal graph) and a 2D correlation histogram (intensity plot) obtained by analyzing the representative flake in panel (c). Remarkably, the peak polarization anisotropy is about 74% over an area of thousands of square micrometers. The elongated 2D histogram indicates a strong correlation between intensity and polarization, which we elaborate further by plotting separate histograms for areas attributed to monolayer WS 2 and bilayer WS 2 (red and blue in panel (e)), based on the PL intensity. The two histograms are shown in figure 1(f) and define the polarization conservation to be 43% for monolayer WS 2 and 74% for bilayer WS 2 , respectively. These values are in close agreement with those recently measured for isolated mono-and bilayer WS 2 crystals on SiO 2 (i.e., 30% and 80%) with circularly polarized light [8]. The slightly higher (lower) values measured for monolayer (bilayer) found in this work can be explained by the fact that both thicknesses occur below the resolution threshold of the PL setup (∼250 nm), as also indicated by the wide tails of the histograms. Furthermore, the mean excitonic lifetime measured for the film via time-resolved PL is ∼350 ps, which is in agreement with that reported in literature for bilayer WS 2 [27]. It is to note that a continuous atomically thick polycrystalline WS 2 film such as that presented in this work does not pose an obstacle for coherent photoabsorption/emission, since the exciton radius in WS 2 is as small as ∼1-2 nm [28].

Epitaxial growth of WS 2 on CVD graphene
We now turn to the analysis of WS 2 synthesized on graphene, an ideal substrate to transduce the collected quantum information into a photocurrent. Figure 2(a) is a SEM micrograph of CVD graphene single-crystals after growth of a continuous WS 2 film. The crystallinity of the WS 2 layer is evaluated by analyzing a partial growth via TEM imaging and selected area electron microdiffraction (details in supporting information). As shown in figure 2(b), the partial process leads to the formation of WS 2 single-crystals with an epitaxial relation to the graphene substrate. The single-crystal domains appear to be several hundreds of nanometers in size. The Raman map reported in panel (c) indicates that most of the graphene grains are covered by bilayer WS 2 (i.e., ( ) I A E 1g 2g 1 ∼ 1) [23] (see also supporting information). Local monolayer inclusions are observed especially in proximity and within transfer-generated tears of graphene (which are visible in the optical image in panel (c)).
3.3. Self-organized photo-emitting/photoconductive patterns of WS 2 on epitaxial graphene As PL is quenched in WS 2 /graphene stacks because of enhanced electron transfer to gapless graphene [29], one might wonder whether the synthesized WS 2 films still presents a robust polarization conservation. To gain some insight into this, we also consider epitaxial graphene on SiC(0001) [30] as a perfect platform to investigate the optical properties of the synthesized WS 2 crystals. On a typical epitaxial monolayer graphene sample, monolayer areas alternate with zerolayer areas following the atomic terraces of the SiC surface [31]. Zerolayer, or buffer layer, graphene is the first carbon-rich layer forming on top of SiC and, although the hexagonal lattice is retained, 30% of the constituting atoms are covalently bonded to the substrate [32]. Local breaking of the sp 2 hybridization  disrupts the semi-metallic nature of the layer and hence its quenching properties. In the SEM micrograph reported in figure 3(a), one can appreciate triangular WS 2 grains obtained with a partial growth covering monolayer graphene (light gray contrast) and zerolayer graphene (mid gray contrast). Indeed, the PL intensity from a sample covered with WS 2 bilayers (see thickness assessment in Supporting Information) is altered along lines parallel to the terrace directions (see panel (c)). A measurable photoemission is detected only in correspondence of WS 2 /zerolayer stacks. Analogously to the h-BN substrate, an elongated 2D histogram of PL polarization and intensity is observed and the polarization conservation is again peaked at 72%. The use of epitaxial graphene on SiC as a substrate for WS 2 synthesis presents a scalable bottom-up approach for patterning ribbons, which can either emit or transfer the charge to the graphene underneath. The outstandingly high mobility and free path length in graphene potentially allows to transport the quantum information from the absorbed light via spin-polarized electrons.

Conclusion
In conclusion, we have demonstrated the CVD synthesis of continuous atomically thick WS 2 films on 2D substrates. The synthesized films display remarkable polarization conservation at room temperature, as high as 74%. We show that by adopting epitaxial graphene on SiC as growth substrate, one can define in a bottomup fashion photoemitting and photoconducting ribbons. The scalable synthesis and design on 2D substrates of WS 2 films with outstanding optical properties is instrumental in the development of novel all-2D quantum optoelectronic and valleytronic devices.