The interface of in-situ grown single-layer epitaxial MoS2 on SrTiO3(001) and (111)

SrTiO3 (STO) is a versatile substrate with a high dielectric constant, which may be used in heterostructures with 2D materials, such as MoS2, to induce interesting changes to the electronic structure. STO single crystal substrates have previously been shown to support the growth of well-defined epitaxial single-layer (SL) MoS2 crystals. The STO substrate is already known to renormalize the electronic bandgap of SL MoS2, but the electronic nature of the interface and its dependence on epitaxy are still unclear. Herein, we have investigated an in-situ physical vapor deposition (PVD) method, which could eliminate the need for ambient transfer between substrate preparation, subsequent MoS2 growth and surface characterization. Based on this, we then investigate the structure and epitaxial alignment of pristine SL MoS2 in various surface coverages grown on two STO substrates with a different initial surface lattice, the STO(001)(4 × 2) and STO(111)-(9/5 × 9/5) reconstructed surfaces, respectively. Scanning tunneling microscopy shows that epitaxial alignment of the SL MoS2 is present for both systems, reflected by orientation of MoS2 edges and a distinct moiré pattern visible on the MoS2(0001) basal place. Upon increasing the SL MoS2 coverage, the presence of four distinct rotational domains on the STO(001) substrate, whilst only two on STO(111), is seen to control the possibilities for the formation of coherent MoS2 domains with the same orientation. The presented methodology relies on standard PVD in ultra-high vacuum and it may be extended to other systems to help explore pristine two-dimensional transition metal dichalcogenide/STO systems in general.

(Some figures may appear in colour only in the online journal) * Author to whom any correspondence should be addressed.
Original content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence. Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.

Introduction
The cubic perovskite, SrTiO 3 (STO), is a versatile transition metal oxide substrate used for the growth of thin film interfaces and as support for two-dimensional (2D) materials systems with interesting electronic or optoelectronic properties. Application examples include high-T C superconducting FeSe [1,2] or YBa 2 Cu 3 O 7−δ [3] and the technologically relevant CdTe films for photovoltaic technology [4,5]. The temperature-dependent dielectric constant of STO of ≈300 at 300 K, which increases to ∼2 × 10 4 at 4.2 K [6], is furthermore very interesting for single-layer (SL) transition metal dichalcogenides (TMDCs) as calculations have shown a strong influence of the surrounding dielectric environment on the TMDC electronic structure [7]. SL MoS 2 is the prototypical example of the family of TMDCs considered in future optoelectronics based on 2D materials [8][9][10][11][12][13]. Specifically for SL MoS 2 sandwiched between high dielectric materials, a bandgap reduction of 0.9 eV of is expected. The environmental screening effect may also be important for low-dimensional electronic devices such as field effect transistors. So far, however, the systematic analysis of structural and electronic effects at the interface of the MoS 2 /STO system has been prevented by lack of uniform samples that for example allows analysis with angle-resolved photoemission spectroscopy (ARPES).
The main production methods for SL TMDC's on substrates are exfoliation and deposition (e.g. in [14,15]) or chemical vapor deposition (CVD) [16][17][18][19][20]. The interest in producing large wafer scale TMDCs has recently directed attention towards growth of large domains controlled by the epitaxy with the substrate [21]. Chen et al has investigated the epitaxial growth of isolated SL MoS 2 islands grown by CVD on three low-index single-crystal STO substrates, and found that the epitaxial interface is strongly controlled by commensuration between the MoS 2 lattice and the specific STO surface [16]. Dumcenco et al [22], and more recently Lai et al [23], came to similar conclusions for distinct triangular shaped MoS 2 flakes produced by CVD on α-Al 2 O 3 sapphire, showing that substrate preparation is essential for controlled epitaxial growth of MoS 2 on dielectrics. Li et al managed to grow wafer scale SL MoS 2 by controlling the alignment of the MoS 2 edges with particular step edge directions on a specially prepared miscut sapphire Al 2 O 3 (0001) substrate [24]. Moreover, Huang et al investigated a substrate-dependent luminescence for MoS 2 grown by CVD on the STO(111) and STO(100) [20]. However, the previous studies of MoS 2 epitaxy have relied on samples transferred through air between substrate preparation, CVD station and microscopy analysis, and have thus left unanswered questions related to the surface state of STO, the epitaxy of the MoS 2 and role of domain boundaries when the coverage is increased towards the full monolayer.
To facilitate the analysis of the epitaxial structure of MoS 2 /STO with scanning tunneling microscopy (STM), we have adopted a direct in-situ growth method by physical vapor deposition (PVD) of Mo under sulfiding conditions, previously applied for MoS 2 and WS 2 SLs on Au(111) [41][42][43][44][45], TiO 2 [46] and graphite/graphene surfaces [47,48]. The in-situ growth approach has the general advantage of producing samples with a controllable coverage towards the full MoS 2 monolayer with a lower defect density and high degree of crystallinity. Such samples are necessary for example in lowtemperature STM/spectroscopy (STM/STS) studies and for ARPES. We find that an ex-situ wet etching preparation procedure of the STO surface followed by an in-situ PVD deposition and high temperature sulfidation scheme is an efficient scheme to synthesize pristine SL MoS 2 islands with a welldefined interface on both STO facets. We structurally characterize the produced epitaxial MoS 2 /STO samples by room temperature STM and find that the SL MoS 2 is structurally rather unperturbed by the substrate. For low MoS 2 coverage, we observe that individual MoS 2 islands preferably align with the zig-zag edges parallel to the ⟨110⟩ directions of STO, on either of the facets. For coverages closer to the monolayer, we find that coherent SL MoS 2 domains are easier grown on the STO(111) compared with STO(001), which is explained by the a better epitaxial alignment on the hexagonal (111) substrate and the existence two different rotational epitaxies for the latter system. The synthesis method reported here should allow further exploration of the interesting MoS 2 /STO interface, and in particular enable a better experimental determination of electronic structure and band gap modifications SL MoS 2 induced by this dielectric substrate.

Methods
We use 0.5 wt.% Nb-doped single crystal SrTiO 3 (001) and 0.7 wt.% Nb-doped single crystals SrTiO 3 (111) EPI polished on one side (MTI Corporation) and cut into 2 mm × 10 mm samples. The samples were ultrasonically cleaned with acetone and isopropanol before prepared ex-situ by annealing at 950 • C for 2 h and consecutively annealed at 1000 • C for 2 h. The samples were sonicated for 10 min in 60 • C deionized water and afterwards annealed at 1000 • C for 2 h. The cooleddown samples are sonicated for 10 min in 60 • C deionized water again and lastly annealed 1000 • C for 5 h. All ex-situ annealing was performed with the samples in a cleaned alumina boat in a tube furnace exposed to air.
The ex-situ prepared STO samples were then loaded into an ultra-high vacuum (UHV) system with a base pressure of ∼3 × 10 −10 mbar equipped with a home-built variable temperature STM. The sample temperature was measured by a K-type thermocouple mounted directly in contact with the mounted crystals. All STM experiments were performed at room temperature. Inspection of the sample at this stage (figure S1) revealed atomically flat substrates, however, with a low concentration of remnant impurities visible as protrusion on the terraces (see figure S1(D)). The water-leached samples were therefore gently Ar sputtered and annealed at 900 • C by e-beam heating in vacuum, leading to the surfaces illustrated in figure 1. X-ray photoelectron spectroscopy (XPS) survey scans after the ex-situ treatment (figure S2) reveal trace amounts of Na and adventitious carbon, whereas some Si was also observed in some cases after high-temperature annealing. The observed impurity levels are expected to have little impact for the observed growth morphology of SL MoS 2 investigated here, but we note that they may ultimately have a significant influence on the electronic properties of the MoS 2 /STO interface. The Nb-doped STO samples were used here to ensure sufficient conductivity for STM. In this regard, we note that we never observed Nb in any of the XPS spectra, which indicates that dopants in our samples do not migrate to the surface during the high-temperature annealing and that they are present in the surface region in negligible concentrations.
Mo was supplied by PVD from a mini e-beam evaporator (Oxford Applied Instruments, EGCO-4). The Mo deposition rate was calibrated by comparison with SL MoS 2 growth on Au(111). H 2 S was supplied from a gas bottle (H 2 S, AGA, nominal purity 2.8) and dosed during Mo deposition by backfilling the UHV chamber, and through a tube doser directed onto the substrate in the high-temperature crystallization step (see below). The STM was equipped with a Pt/Ir tip with the bias voltage applied to the sample.

Results and discussion
The STO preparation produces atomically flat well-defined terraces on both substrates, reflecting extended rectangular terraces with a width of 100-300 nm on STO(001) and terraces with a hexagonal outline and a width of >50 nm on STO(111) (see figure 1). The sputtering induced a slightly higher concentration of steps and in some cases pits in the surfaces compared with the un-sputtered samples (figure S1(B)), but it was necessary to removed impurities as noted in the methods section. From the high resolution STM images, the surface terminations resulting from the high-temperature annealing were determined to be uniform c(4 × 2) and (9/5 × 9/5) reconstructions for the (001) and (111) facet, respectively (figures 1(A) and (B)). Similar in-situ sputtering and annealing of STO(001) surfaces have previously been shown to produce the c(4 × 2) reconstruction, which is suitable for growth of TMDCs due to the hexagonal symmetry of the polyhedral Ti-center [49]. The c(4 × 2) reconstruction is here imaged in the inset of figure 1(A) as a rectangular pattern with two spots per unit cell due to a tip-mode, which images the surface O-lattice instead of the polyhedral Ti-center that has been reported previously [25]. Likewise, for STO(111) the incommensurate (9/5 × 9/5) reconstruction appears in STM images as a hexagonal lattice of protrusions with a lattice periodicity of ∼9.8 Å (figure 1(B), inset) and with a slight undulation in height of the protrusions due to the mismatch with the underlying lattice, which is consistent with previous reports for this Ti-terminated reconstructed STO(111) surface [37,50]. In both cases, the reconstructed lattices appear pristine, except for a few dark depressions which may be associated with a low level of residual impurities (see figure S2).
The subsequent synthesis of SL MoS 2 on the STO surfaces is carried out by a PVD and sulfidation sequence in a H 2 S gas implemented schematically as shown in figure 1(C). We first backfill the chamber with H 2 S gas to a pressure of ∼2 × 10 −6 mbar. In this atmosphere, Mo is deposited onto the sample from an e-beam evaporator at a rate of ∼0.07 ml min −1 for 5 min. After Mo deposition, the local pressure of H 2 S at the sample surface is increased during the ∼30 min annealing step (1000 K), by bringing a gas doser tube in close proximity (∼2 mm) to the sample surface. The use of the doser is estimated to increase the local pressure by a factor of ∼100, ensuring that H 2 S is in high excess relative to Mo during crystallization. The H 2 S atmosphere was maintained throughout the annealing until the sample temperature was 450 K. This sequence in figure 1(C) constitutes one growth cycle with . The annealing temperature of 1000 K was chosen as a compromise between promoting the diffusivity and reactivity of Mo-species at higher temperatures versus any unwanted reaction with the STO, such as partial oxidation of Mo-species and sulfurization of STO. Lower annealing temperatures and lower H 2 S gas pressures generally resulted in the formation of other, less well-defined surface structures (see figure S3), which probably represent partially sulfided Mo, in line with observations on other metal oxide surfaces [46,51].  figure S5) and characteristic truncated triangular, or almost hexagonal, island shapes reflecting the hexagonal symmetry of the MoS 2 crystal lattice. We successfully obtained atom-resolved STM images of the MoS 2 lattice on the basal plane of the islands, as seen in the inset in figure 2(B). The lattice displays a hexagonal pattern of bright protrusions with a lattice parameter of 3.2 ± 0.1 Å. This value agrees well with the expected value for a MoS 2 (0001) lattice oriented in parallel with the substrate, and is similar to literature values for the basal plane S-S distance in PVD grown MoS 2 on Au(111) [44] and CVD grown MoS 2 on STO [16]. The hexagonal MoS 2 shape seen in figure 2(A) is also typical for MoS 2 grown by PVD in a sulfiding atmosphere at high temperature (here 1000 K), unlike CVD growth of MoS 2 which primarily results in triangular shaped particles [16,22,52]. The equilibrium shape of a SL MoS 2 particle is fundamentally controlled by the relative edge free energies of the two different zig-zag edge terminations of a SL MoS 2 sheet (often referred to as ( 1010 ) Mo edge or (1 010 ) S edges, respectively, see figure S6) [53,54]. The specific edge free energies are sensitive to the chemical potential through the variation in S edge coverage, and thus, the predominant hexagonal MoS 2 shape in figure 2(A) implies a near equal stability of the ( 1010 ) Mo edge or (1 010 ) S edges under the synthesis conditions applied for MoS 2 growth on STO(001) here. Oppositely, a triangle reflects that exclusively one of the zig-zag edge types is present under such synthesis conditions. A similar shape variation between triangles and hexagonal shapes controlled by synthesis conditions is seen for SL MoS 2 particles grown on Au(111) in various sulfiding/sulforeductive (in H 2 /H 2 S gas) conditions [55] or by CVD [56].
The second growth cycle produces merged MoS 2 islands (figure 2(B)) with the formation of some defect lines (black arrows) arising from the boundary between two opposite rotational domains. Similar line defects have been reported on CVD grown MoS 2 on SiO 2 [57], h-BN [58] and PVD grown MoS 2 on Au(111) [44,59]. When two islands of same orientation relative to the substrate are fused together, a seamless junction is formed. However, when two islands of opposite rotational domains form a structural boundary between them, it will consist of opposing edges of the same type (e.g. S-edge vs S-edge), whereby a line defect is formed [60]. Interestingly, we also occasionally observe the formation of similarly shaped particles with a lower height relative to the surrounding substrate (black circle in figure 2(B)). This we associate with partially embedded MoS 2 islands on a one step lower STO terrace.
To investigate the epitaxy between the MoS 2 and the underlying STO, we first measured the angles between the edges of the truncated MoS 2 islands (figure 2(A)) with respect to the STO [100] direction (see figure 1(A)) determined in the STM images of the clean STO surface. Six peaks are grouped in two main domain rotations at +15 • (red) and −15 • (blue) with respect to the [100] direction ( figure 2(C)). These angles, also found by Chen et al for a CVD grown sample [16], gives a crystallographic relationship, where one of the ⟨1100⟩ MoS2 directions (i.e. along a (1 010 ) zigzag S edge, see figure S6) is aligned parallel to one of the ⟨110⟩ STO directions. To illustrate this, we consider the substrate structure of the STO(001)c(4 × 2) reconstruction [30,49] where we have depicted the surface O-atoms (blue balls) of the five-fold Ti-centered polyhedra (green area) in figure 2(D) (for clarity, we have neglected any other atoms in the surface). We have overlaid a structural model of a triangular MoS 2 island fulfilling the crystallographic rule of one of the edges (the ⟨1100⟩ MoS2 direction, see figure S6) being parallel to one of the ⟨110⟩ STO directions. This places the two equivalent edge directions parallel to the irrational ⟨ 0⟩ STO directions as shown with red arrows in figure 2(D), producing the +15 • domain. As noted in [16], the epitaxial alignment of MoS 2 in this configuration can be rationalized by the 7:4 coincidence lattice observed along one of the edges in figure 2(D). Mo edges, also exposed in the hexagonal MoS 2 particle, are aligned along the ⟨110⟩ STO . This means that each of the two domain rotations may exhibit an additional 60 • rotational symmetry due to hexagonal shape of the MoS 2 islands, and that four possible rotational domains are thus possible for the angle distribution in figure 2(C). Since we could not resolve the edge type directly in the STM images here, it is not possible to confirm whether both Mo edges and S edges, or only one them, align along the ⟨110⟩ STO . The observation of straight anti-domain boundaries in figure 2(B) (black arrows), however, show that hexagonal MoS 2 particles with oppositely aligned S or Mo edges do form on the substrate (i.e. for two particles rotated by 60 • relative to each other and merged along the same edge type).
Annealing of the c(4 × 2) reconstructed STO(001) surface in 2 × 10 −6 mbar H 2 S at 1000 K in a separate experiment did not alter the periodicity of the STO reconstruction as determined by STM. Therefore, we do not expect the surface to change STO reconstruction during deposition of Mo and subsequent post-annealing in H 2 S. In this way, our model is consistent with a nearly perfect 7:4 lattice epitaxy between the lattice parameter of ⟨1100⟩ MoS2 (S-S distance) and ⟨110⟩ STO (Sr-Sr distance). The agreement with previous work indicates that the same epitaxial bonding is resulting from CVD growth on air-transferred STO and the in-situ PVD synthesis, but the triangular morphology of the CVD grown MoS 2 particle in that study [16] means that only two symmetry equivalent orientations were present. It is likely that the epitaxial alignment of the MoS 2 is facilitated by the coordination of edge atoms rather than the basal plane atoms. Assuming that the Mo and S edge terminations both facilitate the bonding to the STO substrate in an equal way, there are four different ways to place a hexagonal MoS 2 island on the substrate to account for figure 2, as opposed to two for a triangular MoS 2 island. Evidently the possibility of four different rotational domains for MoS 2 /STO(001) further limits the possibilities for the formation of a fully coherent, large domain SL MoS 2 film on STO(001), which is also well-reflected in the complex morphology obtained in figure 2(B) at high coverages using the cyclic PVD deposition method.

SL MoS 2 /SrTiO 3 (111)
Next, we used the same synthesis method for growth of SL MoS 2 islands on the STO(111) surface prepared with the hexagonal (9/5 × 9/5) incommensurate reconstruction. Intuitively one should expect a better lattice match between two hexagonal structures such as MoS 2 (0001) and STO(111), compared with the STO(001). Figure 3(A) shows a large-scale STM image of the surface after one growth cycle. The formation of individual MoS 2 islands is again identified by the clear hexagonal shapes and elevated appearance above the STO substrate. Surprisingly, a second growth cycle mainly produces an increased number of separated MoS 2 islands, with only a slightly increased size of the individual islands ( figure 3(B)). Similar to STO(001), we observe defect lines (black arrow) arising from merging of two islands of opposite rotational domains resulting in structural boundary defects. The inset in figure 3(B) displays an atom-resolved STM image of the basal plane displaying bright protrusions with a lattice parameter of 3.2 ± 0.1 Å, again corresponding to the MoS 2 (0001) plane.
A measure of the angle of the MoS 2 edges to the STO [110] direction for the first two growth cycles displays three peaks separated by 60 • (figure 3(C)), thus revealing a simpler epitaxial alignment than on STO(001). This means that for MoS 2 /STO(111) only two rotational domains exist, which can tentatively be represented by a crystallographic relationship with each of the zig-zag edges, the ( 1010 ) Mo edge or (1 010 ) S edge, aligned parallel to the ⟨110⟩ STO directions. We illustrate this by the schematic drawing in figure 3(D) of the (9/5 × 9/5) reconstructed surface represented in a simplified form, with one grey ball per STO unit cell overlaid with a structural model of a triangular MoS 2 island. Note that the (9/5 × 9/5) STO(111) surface has not been solved in atomistic detail [37], but it likely reflects an incommensurate Ti-rich reconstruction which could possess sites with varying interaction with MoS 2 . We, however, do not observe any preference for substrate bonding of the basal plane sites of MoS 2 , and therefore mainly assign the orientational alignment to be caused mainly by interaction with the MoS 2 edges.  4(B)). Apart from the 3.2 Å periodicity of the basal plane S-S distances in MoS 2 , the periodicity of the moiré superstructure is measured to have 63 ± 3 Å periodicity and a rotation of the superstructure of 15 ± 1 • with respect to the unit cell vector the MoS 2 lattice. Keeping the scanning parameters constant and by applying small bias pulses, it was possible to change the apex of the tip to obtain different tip modes. In one such tip mode, the MoS 2 basal plane appeared with a regular hexagonal pattern with a distinctly different periodicity of 9.8 ± 0.7 Å ( figure 4(C)). This periodicity is not associated with the MoS 2 film, but it is in good agreement with the periodicity of the (9/5 × 9/5) reconstruction itself, with a lattice periodicity of 9.96 Å. Apparently, for some tip modes and biases within the nominal band gap of MoS 2 , the tunneling current is due to electronic states inside the band gap, which carry a contribution from the underlying (and still reconstructed) STO. Similar effects have been observed for STM imaging of SL MoS 2 grown on the rutile TiO 2 (110) surface [46]. This suggests that the STO substrate structure, or at least the periodicity underneath, is preserved when the MoS 2 layer is grown.
The exact size and relative rotation of the hexagonal moiré unit cell vectors (white in figure 4(B)) is highly sensitive to the exact rotation of the MoS 2 (0001) lattice relative with respect the underlying STO surface. The observed 15 • rotation between the moiré unit cell vector and a corresponding unit cell vector of the MoS 2 (0001) lattice in figure 4(B) is, in fact, not entirely reproduced from the simple alignment in figure 3(D), and thus requires a slight refinement to the crystallographic relation presented in figure 4(B) for MoS 2 /STO(111). We did a more detailed structural modelling (see also figure S7) to determine the angular difference of MoS 2 and STO that reproduces the overall 15 • moiré rotation. A good match is found for a slight rotation difference of ∼0.8 • of the MoS 2 edges, i.e. the MoS 2 lattice appears very nearly, but not exactly, aligned with the high symmetry direction in the underlying STO. This is further represented in the ball model based on the schematic model in figure 3(D). We performed STM line scans across the moiré pattern in figure 4(B) (shown in figure S8) and report a height corrugation of 56 ± 2 pm over the moiré unit cell, which is half the corrugation observed for MoS 2 on Au(111) [42]. STM heights are not straightforwardly associated with the geometrical height, as the STM images constitute a convolution of geometry and local electronic structure, which in extreme cases leads to an inverted contrast for a moiré pattern on MoS 2 [61]. However, a simple interpretation is that the depressions in the moiré pattern are regions with no S-atoms on top of the protrusions constituting the (9/5 × 9/5) reconstruction. The other domain is with S-atoms partially on top. Alternatively, the depressions are correlated with lower local density of electronic states due to hybridization between STO and MoS 2 , which gives rise to the distinct protrusion in the middle of the moiré depression (indicated by a white circle in figure 4(C)). The height of the central protrusion is measured to 14 ± 2 pm (figure S8) for both figures 4(B) and (C). It is interesting that we do not see this protrusion as clearly in other moiré pattern depressions, suggesting the state could be due to a defect, e.g. an oxygen substituted S-atom [62,63] on the STO-side of the MoS 2 or a cation impurity.
It would be of great interest to obtain high resolution STM images of the MoS 2 edges on a high-k dielectric such as STO, in order to investigate presence of onedimensional edge states, which are present in MoS 2 [64]. Such one-dimensional edge states have been shown to exhibit second-order nonlinear optics [52] and calculations show how the metallic edge states of MoS 2 may exhibit magnetic behavior [65]. However, despite extensive attempts in this study, atomic resolution STM imaging of the edge region was unsuccessful. The edges themselves appear structurally well-defined from the STM images, and we therefore attribute such difficulties to low conductivity in the edge region. This interestingly indicates that the edge states in MoS 2 are significantly altered by the presence of the STO substrate.

Conclusions
The structural analysis provided here by high-resolution STM shows that MoS 2 can be synthesized as well-defined single layers on SrTiO 3 (001)-c(4 × 2) and SrTiO 3 (111)-(9/5 × 9/5) reconstructed surfaces. Both MoS 2 /SrTiO 3 systems adopt an epitaxy that orients the zig-zag edge directions along the substrate high symmetry directions. For STO(001), a complicated epitaxy develops reflecting that the SL MoS 2 grows in four different possible rotational domains. In the perspective of growing electronically uniform heteroepitaxial layers needed for example in ARPES studies, the (001) substrate is thus not optimum since rotational domains with a rather limited size are formed. For STO(111) the number of domain rotations is reduced to two and, correspondingly, larger and more uniform SL MoS 2 layers can be grown, although still with an inappropriate level of domain formation. Further optimization of the growth conditions reported here, e.g. inspired by the synthesis reported recently for Au(111) substrates which also features two epitaxial orientations during epitaxial growth of TMDCs [43,66], is therefore needed to provide single-orientation epitaxial SL MoS 2 on STO.
In extension of this work, we also suggest that the growth recipe should be extended to other electronically interesting SL TMDCs such as WS 2 , WSe 2 MoSe 2 which have been synthesized by PVD as epitaxial layers on single-crystal metal substrates. An extension of the STO substrate preparation and growth procedure reported here may be useful for future studies of such systems that have demonstrated enhanced superconductivity by ionic gating [67,68] or the formation of lateral heterostructures made from these TMDCs. The in-situ synthesis for epitaxial monolayers on STO is moreover highly attractive for exploration of TMDC heterostructures on STO, in line with the recent work on epitaxial 2D materials such as h-BN and graphene supported on noble metals, such as MoS 2 and WS 2 /Au(111) and MoS 2 /Gr [69][70][71][72][73]. The method may also open up for controlled growth of new epitaxial 2D systems, of e.g. in-situ grown stacked 2D heterostructures or laterally connected 2D systems on STO.

Data availability statement
The data cannot be made publicly available upon publication because they are not available in a format that is sufficiently accessible or reusable by other researchers. The data that support the findings of this study are available upon reasonable request from the authors.