# General synthesis of two-dimensional van der Waals heterostructure arrays

## Abstract

Two-dimensional van der Waals heterostructures (vdWHs) have attracted considerable interest1,2,3,4. However, most vdWHs reported so far  are created by an arduous micromechanical exfoliation and manual restacking process5, which—although versatile for proof-of-concept demonstrations6,7,8,9,10,11,12,13,14,15,16 and fundamental studies17,18,19,20,21,22,23,24,25,26,27,28,29,30—is clearly not scalable for practical technologies. Here we report a general synthetic strategy for two-dimensional vdWH arrays between metallic transition-metal dichalcogenides (m-TMDs) and semiconducting TMDs (s-TMDs). By selectively patterning nucleation sites on monolayer or bilayer s-TMDs, we precisely control the nucleation and growth of diverse m-TMDs with designable periodic arrangements and tunable lateral dimensions at the predesignated spatial locations, producing a series of vdWH arrays, including VSe2/WSe2, NiTe2/WSe2, CoTe2/WSe2, NbTe2/WSe2, VS2/WSe2, VSe2/MoS2 and VSe2/WS2. Systematic scanning transmission electron microscopy studies reveal nearly ideal vdW interfaces with widely tunable moiré superlattices. With the atomically clean vdW interface, we further show that the m-TMDs function as highly reliable synthetic vdW contacts for the underlying WSe2 with excellent device performance and yield, delivering a high ON-current density of up to 900 microamperes per micrometre in bilayer WSe2 transistors. This general synthesis of diverse two-dimensional vdWH arrays provides a versatile material platform for exploring exotic physics and promises a scalable pathway to high-performance devices.

## Main

Two-dimensional (2D) transition-metal dichalcogenides (TMDs), including 2D semiconducting TMDs (s-TMDs), such as MoX2 and WX2 (where X = S, Se, Te), and their metallic counterparts (m-TMDs), such as VX2, NbX2 and TaX2, have attracted intense interest as atomically thin building blocks for fundamental studies and novel devices at the limit of single-atom thickness. In particular, van der Waals (vdW) integration of 2D materials creates a new generation of vdW heterostructures (vdWHs) beyond the limits set by lattice matching and processing compatibility requirements in conventional bonded heterostructures1,2,3,4. It thus allows unprecedented flexibility to combine materials with radically different chemical compositions, crystal structures or lattice orientations, producing unique electronic and photonic characteristics or other exotic properties beyond the reach of the existing material systems and enabling totally new device functions6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30. However, most vdWHs to date are created by an arduous micromechanical exfoliation and manual restacking process, which—although versatile for fundamental studies and proof-of-concept demonstrations—is clearly not scalable for practical technologies. To explore the full potential of vdWHs requires a robust and scalable synthesis of vdWH arrays with precisely controlled chemical compositions, electronic properties and spatial locations31,32,33, which remains a standing challenge for the field.

Although considerable progress has been made in synthesizing 2D lateral heterostructures, multi-heterostructures and superlattices34,35,36,37,38,39,40,41,42,43,44, the controlled synthesis of high-quality 2D vdWHs is much less explored. Despite some recent reports of vdW epitaxial growth of 2D vdWHs45,46,47,48, many of these efforts rely on chance nucleation and growth with rather limited synthetic control. For example, the controlled synthesis of periodic 2D vdWH arrays, a necessary step towards scalable integration, has not been realized to date.

Here we report a general synthetic approach to regular arrays of 2D vdWHs between s-TMDs and m-TMDs regardless of the lattice differences. In brief, we first pattern periodic arrays of nucleation sites on monolayer or bilayer s-TMDs (for example, WSe2, WS2, MoS2), on which m-TMDs may selectively nucleate and grow to form periodic m-TMD/s-TMD vdWH arrays. This approach is general and not limited to a material with a specific chemical composition or lattice structure. We show that a wide range of 2D vdWHs, including VSe2/WSe2, NiTe2/WSe2, CoTe2/WSe2, NbTe2/WSe2, VS2/WSe2, VSe2/MoS2 and VSe2/WS2, can be produced with atomically sharp, nearly ideal vdW interfaces and widely variable moiré superlattices. These metal/semiconductor vdWHs exhibit an atomically clean interface and enable the construction of high-performance electronic devices with highly consistent device performance, delivering a high ON-current density of up to 900 μA μm−1 in bilayer WSe2 transistors.

## Synthesis of m-TMD/s-TMD vdWH arrays

To produce high-quality m-TMD/s-TMD vdWH arrays, it is essential to precisely control the nucleation and growth processes. In particular, the random undesired nucleation of m-TMDs must be prevented to ensure the controlled nucleation and growth of the vdWHs by design. To this end, we choose large-area 2D s-TMDs (for example, WSe2, MoS2, WS2) as the vdW epitaxial substrates. The atomically clean, dangling-bond-free surface of these 2D crystals ensures minimum surface defects and prevents uncontrolled chance nucleation. With such a substrate, we can thus intentionally pattern periodic arrays of defects as the exclusive nucleation sites for site-specific growth of m-TMDs to form vdWH arrays (Fig. 1).

For simplicity, we focus on WSe2 as an example substrate in our discussions unless otherwise mentioned. Large-area monolayer or bilayer WSe2 was first prepared via a well-developed reverse-flow chemical vapour deposition (CVD) system41. A focused laser irradiation combined with raster scan was used to create a periodic array of defects on the freshly grown WSe2 (Fig. 1b). The direct laser cauterization patterning approach allows the creation of local defects at specific locations without introducing any contamination on other areas of the underlying WSe2, thus minimizing undesired random nucleation in subsequent growth. Atomic force microscopy (AFM) studies revealed that the laser-induced defect arrays in monolayer WSe2 have a depth of about 0.3 nm (Extended Data Fig. 1), suggesting that the monolayer WSe2 is only partially damaged at the patterned sites.

The pre-patterned WSe2 was then placed into a separate furnace for synthesizing top-layer m-TMDs (for example, VSe2). Notably, the growth temperature of m-TMDs (Tsubstrate = 600 °C for VSe2) is considerably lower than that of s-TMDs (Tsubstrate = 850 °C for WSe2), which prevents the thermal degradation of the WSe2 substrates during the second growth step. It has been reported that multilayer 1T-VSe2 may prefer to nucleate and grow from the edge sites of s-TMDs or other defective sites47. Distinct from the uncontrolled nucleation and growth at edges or other random defects, the artificially created defect arrays in large-area WSe2 crystals function as the exclusive nucleation sites for highly selective growth of 2D VSe2 crystals to produce periodic VSe2/WSe2 vdWH arrays.

Precise control of the crystal nucleation enables the growth and tailoring of vdWH arrays with various periodic arrangements and periodicities. Figure 2a shows that the top-layer VSe2 nanoplates grow into a rectangular pattern on the WSe2 substrate, in which two distinct regions can be found with different optical contrast, corresponding to the VSe2 regions (yellow regions) and the WSe2 region (pink region). The VSe2 nanoplates typically exhibit a well-faceted hexagonal geometry of identical orientation, suggesting a well-defined vdW epitaxial relationship with a fixed orientational alignment with the underlying WSe2 substrate. Noticeably, there is typically only one single-crystal VSe2 nanoplate at each pre-defined nucleation site, indicating highly controllable nucleation and growth of VSe2 crystals at the predesignated locations. As a result, a highly uniform rectangular array of VSe2 nanoplates was obtained on WSe2 to form periodic vdWH arrays (Fig. 2a). With the flexibility of the laser cauterization patterning process to create the nucleation sites by design and the highly specific control of the nucleation and growth of VSe2 nanoplates on WSe2, vdWH arrays with variable periodicities or lattice arrangements may be readily produced (for example, a hexagonal array of the VSe2/WSe2 vdWHs shown in Fig. 2b).

With our growth process, the lateral size of the epitaxial VSe2 nanoplates on WSe2 may be controlled by the growth duration. For example, for growth durations of 6, 8 and 10 min, the width of the VSe2 nanoplates is controlled to be 6.0 ± 0.4, 7.6 ± 0.4 and 9.0 ± 0.2 μm (Fig. 2c–e). With the ability to define the nucleation sites with the designed periodicity and to control the size of the resulting VSe2 nanoplates, it is possible to tailor the edge-to-edge distance between the neighbouring VSe2 nanoplates. For example, the corresponding edge-to-edge distances achieved in VSe2/WSe2 vdWH arrays with 10-μm lateral periodicity are 4.0 ± 0.4, 2.3 ± 0.4 and 1.1 ± 0.1 μm (Fig. 2f–h), after VSe2 growth durations of 6, 8 and 10 min, respectively. The ability to control the edge-to-edge distance in (VSe2/WSe2)–WSe2–(VSe2/WSe2) heterostructure arrays offers a synthetic pathway to vdWH transistors with tunable channel length.

The AFM studies further demonstrated the highly uniform nature of the heterostructure arrays (Fig. 2i). Raman and photoluminescence studies were conducted to probe the spatial modulation of the structural and optical properties in the VSe2/WSe2 vdWH arrays. The Raman spectrum from the overlapping VSe2/WSe2 region exhibits two characteristic peaks at 206 and 257 cm−1, corresponding to the A1g modes of 1T-VSe2 and 2H-WSe2, respectively; whereas the Raman spectrum from the bare WSe2 region exhibits only one prominent A1g resonance mode of WSe2 at 257 cm−1 (Extended Data Fig. 2a, b). These features suggest pure phases and a vertically stacked structure. The Raman mapping images clearly show distinct spatial modulation and confirm the formation of the vertically stacked vdWH arrays (Fig. 2j–k). Similarly, the photoluminescence studies exhibit a strong photoluminescence peak at 776 nm for the bare part and a much weaker emission for the overlapping region (Extended Data Fig. 2c). The photoluminescence mapping studies (Fig. 2l) show similar features to the Raman mapping studies, confirming that the optical quality of the WSe2 substrate is largely retained during the growth of VSe2.

## Nucleation and growth mechanism

To probe the selective nucleation and growth mechanism, we conducted structural and composition analysis of the patterned nucleation sites on WSe2 before and after the nucleation and growth of VSe2 (Fig. 3). The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) image reveals a circular morphology of the laser-irradiated area with darker contrast compared with the unirradiated area (Fig. 3a). Elemental mapping by energy dispersive spectroscopy (EDS) showed an apparently lower Se signal in the laser-patterned area (Fig. 3b), while the W signal exhibited little variation compared with the unirradiated area (Fig. 3c). The Se-to-W atomic ratio in the patterned area is around 1.33, considerably less than that (about 2.00) in the unpatterned area (Fig. 3d), suggesting that the top-layer Se atoms are mostly vaporized and the W atoms are largely intact. This is not surprising considering that the Se atoms are much more volatile. Thus, the laser-patterning process leaves a largely W-terminated surface at the patterned sites.

After the growth of VSe2, the nucleation sites retained the circular shape (Fig. 3e). The EDS elemental analysis showed that the concentration of Se in the nucleation site is apparently lower (Fig. 3f) than that in the region outside the nucleation site, while the concentration of V is obviously higher in the nucleation site (Fig. 3h) and the distribution of W does not show an obvious contrast (Fig. 3g). The atomic ratio of V to Se is about 4:3 at the nucleation site, in contrast to the nearly ideal ratio of 1:2 outside the nucleation sites. The enrichment of V at the nucleation sites is intriguing and may be caused by the high concentration of V radicals adsorbed on defective points during the initial nucleation stage (see Fig. 3j; to be further discussed below).

## Structural characterization of VSe2/WSe2 vdWHs

Next we characterized the detailed atomic structures of the VSe2/WSe2 vertical heterostructures. Planar-view HAADF-STEM images of the different surface regions revealed well-defined 1T-VSe2 and 2H-WSe2 (Extended Data Fig. 3a, c). A careful analysis of atomic-resolution HAADF-STEM images gave lattice constants of 3.36 Å and 3.29 Å for VSe2 and WSe2 (Extended Data Fig. 3b, d), respectively, matching well with the expected database values (Joint Committee on Powder Diffraction Standards card numbers 89-1641 and 38-1388, respectively). The selected area electron diffraction (SAED) pattern taken at the bare WSe2 region shows a single set of hexagonally arranged diffraction spots that could be indexed to the WSe2 (Fig. 4a), while that from the overlapping VSe2/WSe2 region shows two sets of identically orientated diffraction spots (Fig. 4b), with the inner and outer set corresponding to the 1T-VSe2 and 2H-WSe2, respectively.

The high-resolution STEM image revealed a clear moiré superlattice structure formed between the identically oriented VSe2 and WSe2 lattices (Fig. 4c) with a smallest periodically repeating cell of 14.7 ± 0.1 nm, corresponding to 44 × 44 VSe2 unit cells stacked on 45 × 45 WSe2 unit cells, which is consistent with the theoretical atomic model (inset). A careful analysis of the atomic model revealed three types of local alignment in the moiré unit cell (designated as i, ii and iii in Fig. 4c) that show high resemblance but in fact have highly distinct atomic stacking orders, namely, the single V atom arrangement, V stacking over Se, and V stacking over W (Extended Data Fig. 4). These planar-view TEM and STEM studies and analyses confirm a fully oriented vdW epitaxial relationship.

We also conducted cross-sectional STEM studies to investigate the interface characteristics of the VSe2/WSe2 vertical heterostructures. A typical cross-sectional HAADF-STEM image clearly revealed the vdWH formation between multilayer VSe2 and bilayer WSe2 (ML-VSe2/2L-WSe2) supported on SiO2 substrate (Fig. 4d). The corresponding elemental mapping for W, Se, V and Si further revealed the spatially resolved elemental distribution with an abrupt composition change across the heterostructure interface (Fig. 4e–h). A high-resolution STEM image taken at the ML-VSe2/2L-WSe2 interface clearly resolved the Se and V atoms in the VSe2 layers and the W and Se atoms in the WSe2 layer (Fig. 4i). The image confirmed an atomically sharp boundary along the heterostructure interface, without any apparent atomic mixing or defects. A careful analysis of the line profile reveals a distance of 6.1 Å from the V to V atoms between neighbouring layers, and a distance of 2.9 Å from the Se to Se atoms between neighbouring VSe2 layers, both consistent with theoretical values. In addition, the V-to-W distance and the Se-to-Se distance across the heterojunction interface was also determined to be 6.1 Å and 2.9 Å, respectively, which are essentially the same as the intrinsic distance between the natural vdW layers, suggesting a nearly ideal vdW interface across the vertical heterojunction.

## General and scalable synthesis of vdWH arrays

We used the same strategy for the synthesis of a wide range of m-TMD arrays on WSe2 substrate and obtained a series of m-TMD/WSe2 vdWH arrays, such as NiTe2/WSe2, CoTe2/WSe2, NbTe2/WSe2 and VS2/WSe2 (Fig. 5a, c, Extended Data Fig. 5). A highly specific nucleation on the pre-patterned sites was observed in all cases. For instance, similar to the growth of the VSe2/WSe2 heterostructure arrays, periodic NiTe2/WSe2 and CoTe2/WSe2 arrays were uniformly grown on a patterned WSe2 substrate (Fig. 5a–d). The HAADF-STEM image revealed a lattice constant of 3.84 Å for NiTe2 (Fig. 5e) and a well-ordered moiré superlattice in NiTe2/WSe2 vdWH, with a smallest periodically repeating cell of 2.3 ± 0.1 nm (Fig. 5f, g), corresponding to 6 × 6 NiTe2 unit cells stacked on 7 × 7 WSe2 unit cells (Fig. 5h). Similarly, the CoTe2 showed a lattice constant of 3.80 Å (Fig. 5i), and the CoTe2/WSe2 heterostructure exhibited well-ordered moiré patterns with a smallest repeating cell of 2.6 ± 0.1 nm (Fig. 5j, k), corresponding to 7 × 7 CoTe2 unit cells stacked on 8 × 8 WSe2 unit cells (Fig. 5l).

The formation of well-defined moiré superlattice structures between diverse 2D materials opens a new pathway to tunable lateral potential modulation and holds substantial promise for exploring unique electronic and photonic characteristics that cannot be reached in conventional materials, as highlighted by recent studies revealing Mott-like insulator and superconducting behaviour in magic-angled graphene and moiré excitons in twisted TMD heterojunction bilayers25,26,27,28,29,30. It should be noted that, despite extraordinary excitements on moiré superlattices, studies so far have been largely limited to mechanically stacked heterostructures. Our demonstrated ability to synthetically produce diverse 2D vdWH arrays with widely tunable moiré superlattices offers an exciting material platform for exploring fundamental physics and/or creating totally new device functions.

In addition, we have further shown that 2D vdWH arrays can be readily grown on different s-TMD substrates, including monolayer MoS2 and WS2 (Extended Data Figs. 6, 7). In particular, by using the highly oriented monolayer MoS2 films grown on sapphire wafers49, we show that much larger-scale periodic arrays of VSe2/MoS2 vdWHs (>12,000 separated VSe2/MoS2 2D vdWHs) can be readily grown on continuous monolayer MoS2 films with a nucleation and growth yield up to about 99%. The less than 100% yield is largely due to particulates in a typical laboratory environment and can be further improved when the experiment is conducted in a more controlled clean room environment. These studies clearly confirm the generality, controllability and scalability of our approach for the growth of periodic m-TMD/s-TMD vdWH arrays.

## Synthetic vdW contacts for 2D transistors

With the successful growth of well-defined m-TMD/s-TMD heterostructure arrays with an atomically clean interface, m-TMDs may serve as a unique class of synthetic vdW contacts for s-TMDs without exposing the functional contacting interface to any lithography process and thus avoid lithography-induced contamination and deposition-induced damage to the atomically thin s-TMDs16. Furthermore, the bond-free vdW interface between the m-TMDs and s-TMDs may also greatly suppress the interface dipoles and metal-induced gap states. Thus, this general synthesis of 2D vdWH arrays holds considerable promise for a synthetic pathway to high-performance vdW contacts to s-TMDs.

To evaluate the merit of using such vdWH arrays as the synthetic vdW contacts for 2D transistors, we fabricated (VSe2/WSe2)–WSe2–(VSe2/WSe2) transistors, where the neighbouring VSe2 nanoplates function as the vdW source and drain electrodes (Fig. 6a). The output characteristics of a typical bilayer WSe2 transistor with VSe2 vdW contacts show a linear drain-source current–voltage (IdsVds) relationship (red curve in Fig. 6b, Extended Data Fig. 8a), indicating a satisfactory Ohmic contact formation. The transfer characteristics (IdsVgs, where Vgs is the gate-source voltage) reveal a typical p-type-transistor behaviour with an ON/OFF current ratio of nearly 107 (red curve in Fig. 6c, Extended Data Fig. 8b). In comparison, the bilayer WSe2 transistors with directly deposited Cr/Au metal contacts exhibited apparently nonlinear behaviour with a much smaller ON-state current and smaller ON/OFF ratio (about 104) (black curve in Fig. 6b,c). We also tested similar WSe2 devices with other deposited metal contacts (for example, Au or Pt), which all show similar performance, probably due to the universal deposition-induced damage and Fermi level pinning effect as recognized previously16. These studies clearly demonstrate that the synthetic VSe2 vdW contacts offer considerable advantages over the typical lithographically defined electrodes and compare well with the most optimized vdW metal contacts16.

An analysis of the transfer curve gives a two-terminal carrier mobility of up to 137 cm2 V−1 s−1, which is among the highest two-terminal mobility values reported for CVD-grown bilayer WSe2 (ref. 50), highlighting the high quality of the synthetic VSe2 vdW contacts. To further evaluate the reliability and reproducibility of the synthetic vdW contacts, we fabricated and measured a large number of devices with either type of contact. Among all the devices measured, the devices with synthetic vdW contacts essentially showed a unity device yield, with all devices exhibiting p-type-transistor behaviour with highly consistent performance. The cumulative mobility distribution shows a rather narrow spread around 100 cm2 V−1 s−1 (red curve in Fig. 6d). In contrast, the devices with deposited metal contacts showed a much poorer performance and broader distribution, with the majority of the devices showing mobility around 10 cm2 V−1 s−1 or lower, although higher-performance devices were also occasionally observed (black curve in Fig. 6d). Such a broad two-terminal mobility distribution could be attributed to lithography- or deposition-induced contamination or damage at the contacting interface with the bilayer WSe2, which has been well recognized in previous studies16. The robust synthesis of (VSe2/WSe2)–WSe2–(VSe2/WSe2) heterojunction arrays offers synthetic vdW contacts without exposing the atomically thin WSe2 channel to the aggressive lithography processes, thus helping to retain the high electronic performance of the atomically thin semiconductors and ensuring highly consistent device performance. In addition, we further improved our vdWH contacted devices by using thinner gate dielectrics (70 nm SiNx) and achieved a highest ON-state current density of up to 900 μA μm−1 while retaining an ON/OFF ratio of over 107 (Fig. 6e, f).

Altogether, we have demonstrated a general strategy for highly controllable growth of a series of periodic vdWH arrays between diverse m-TMDs and s-TMDs regardless of the lattice differences. Detailed structural analyses revealed an atomically sharp composition modulation, a nearly ideal heterojunction vdW interface, a well-defined vdW epitaxial relationship and clearly resolved moiré patterns with widely variable moiré periodicities. We further showed that such atomically clean heterojunctions can function as highly reliable synthetic vdW contacts for creating high-performance devices with excellent device performance, delivering a high ON-current density of up to 900 μA μm−1 in a bilayer WSe2 transistor. The ability to produce diverse 2D vdWH arrays with nearly intrinsic vdW interfaces and widely tunable moiré superlattices opens a synthetic pathway to a versatile material platform for fundamental studies and a scalable pathway to high-performance vdWH devices.

## Methods

### Growth of WSe2

The WSe2 monolayers or bilayers were grown in a CVD system using a thermally evaporated solid WSe2 source with the reverse flow to minimize nucleation sites and obtain large-area single crystals41,51. The WSe2 powder was placed in a quartz boat located at the centre heating zone of the furnace, and a piece of SiO2 (285 nm)/Si substrate was placed at the downstream end of the furnace as the growth substrate (Tsubstrate = 850 °C). The centre heating zone was purged with ultrahigh-purity argon gas (99.999%) for 2 min before heating to 1,180 °C under ambient pressure with 80 sccm reverse flow of argon. After reaching the desired growth temperature, the chemical vapour source was carried downstream under a forward argon gas flow of 80 standard cubic centimeters per minute (sccm) for a growth period of 4 min. The growth process was then terminated and the furnace was cooled naturally.

### Direct laser cauterization patterning WSe2

The periodic array of defects was created in WSe2 monolayer or bilayers by raster scanning of a focused laser beam (488 nm, 50 mW) with programmed patterns, with each point defect created by about 1–5-s exposure under a ×100 objective. These period defect points can be readily observed under an optical microscope.

### Growth of VSe2/WSe2 vertical heterostructure arrays

The vdW epitaxial growth of VSe2 on WSe2 was conducted with a two-zone furnace. Commercial selenide powder (1.0 g; 99.99%, Alfa) was placed in the middle of the first zone and vanadium chloride powder (0.1 g; 99%, Alfa) in a ceramic boat was placed in the middle between the two zones. A piece of SiO2/Si substrate with pre-patterned WSe2 was placed in the downstream zone as the growth template. Before the heating process, a 1,000-sccm argon flow was used to purge the reaction chamber. Then, the source zone and substrate zone were heated up to about 380 °C and 600 °C within 16 min under a mixed argon (80 sccm) and hydrogen (2 sccm) flow. After 10-min growth, the heating process was terminated and the furnace was naturally cooled to room temperature without changing the carrier gases.

### Growth of NiTe2/WSe2 vertical heterostructure arrays

The vdW epitaxial growth of NiTe2 on WSe2 was conducted with a two-zone furnace. Tellurium powder (0.1 g) (99.99%, Alfa) in a ceramic boat was placed in the upstream zone and nickel dichloride powder (0.1 g) (>98%, Energy Chemical) in a ceramic boat was placed in the middle of the second temperature zone. A piece of SiO2/Si with pre-patterned WSe2 as substrate was tilted above the nickel dichloride powder. Before the heating process, the system was purged with ultrahigh purity argon gas (Rizhen, about 99.999%) to completely remove oxygen and moisture from the quartz tube. Next, the first zone (tellurium) and the second zone (nickel dichloride) were heated up to 600 °C and 550 °C within 15 min under a mixed hydrogen (5 sccm) and argon (60 sccm) flow. After 15-min growth, the heating process was terminated and the furnace was naturally cooled to room temperature without changing the carrier gases.

### Growth of CoTe2/WSe2 vertical heterostructure arrays

The vdW epitaxial growth of CoTe2 on WSe2 was conducted with a two-zone furnace. Tellurium powder (0.1 g) (99.9%, Alfa) was placed in a ceramic boat at the upstream end of the quartz tube furnace. Colbalt chloride (CoCl2) powder (0.1 g) (99.7%, Aladdin) was placed in a quartz boat at the centre of the furnace. A piece of SiO2/Si with pre-patterned WSe2 was tilted above the CoCl2 powder as the growth substrate. The quartz tube was purged with ultrahigh-purity argon gas (99.999%) before ramping up to 580 °C in about 20 min (tellurium powder was kept at 420 °C) under a constant flow of argon (65 sccm) and hydrogen (6.5 sccm) under atmospheric pressure. The temperatures were held for 15 min before shutting off to allow the sample to naturally cool to ambient temperature.

### Characterization of heterostructure arrays

An optical microscope (DP27, Olympus), an atomic force microscope (Bioscope system, Brucker) and Raman spectroscopy (inVia Reflex, Renishaw with 532-nm laser as the excitation source) were used to characterize the morphology, the thickness, and the Raman and photoluminescence spectra of the resulting vertical heterostructure arrays. TEM characterizations and EDS analyses were performed using a JEM-2100F, JEOL, operating at 200 kV and STEM (Titan Cubed Themis G2 300; JEOL JEM-ARM300CF S/STEM, accelerating voltage, 300 kV).

Density functional theory calculations implanted in CASTEP packages52 were performed to study the nucleation of VSe2 on the patterned WSe2 substrate. We have chosen the generalized gradient approximation (GGA) with Perdew–Burke–Ernzerhof (PBE) to describe the exchange-correlation energy53,54,55. In particular, we treated (4s, 4p), (5d, 6s, 6p) and (3d, 4s, 4p) as the valence states for Se, W and V, respectively. Meanwhile, the cut-off energy of the plane-wave basis was set to be 330 eV with ultrasoft pseudopotentials. A 0.07-Å−1 reciprocal separation was applied in the k-points for the energy minimization based on the Broyden–Fletcher–Goldfarb–Shannon (BFGS) algorithm56. The convergence thresholds were set as 5 × 10−6 eV per atom for the total energy and 0.0005 Å per atom for the inter-ionic displacement. The binding calculation was performed on a 4 × 4 × 1 WSe2 layer. The binding energy was calculated according to:

$${E}_{{\rm{ads}}}={E}_{{\rm{total}}}-{E}_{{\rm{slab}}}-{E}_{{\rm{adsorbate}}}$$

where the Eads is the adsorption energy, Etotal is the total energy of the system with adsorbate, Eslab is the energy of the optimized surface and Eadsorbate is the energy of the isolated adsorbate.

### Device fabrication and characterization

For all devices, the device layout patterns were defined using electron-beam lithography and all metal films were deposited using electron-beam evaporation with a standard lift-off process. For the fabrication of (VSe2/WSe2)–WSe2–(VSe2/WSe2) transistors, the neighbouring VSe2 nanoplates functioned as the vdW source and drain electrodes, with the edge-to-edge distance between the neighbouring VSe2 nanoplates defining the semiconductor channel length. The Cr/Au electrodes were fabricated on top of the VSe2 (without directly contacting or damaging the WSe2 layer) as external electrical probes. For the control devices, the conventional metal contacts (Cr/Au, Au or Pt) were directly patterned and deposited on bilayer WSe2 to form the source and drain electrodes. The WSe2 outside the active channels was etched using reactive ion etching to further define the channel width. The silicon substrate was used as the gate electrode with 285-nm SiO2 or 70-nm SiNx back-gate dielectrics. All electronic transport measurements were performed using an Agilent B1500A semiconductor analyser at room temperature in vacuum.

## Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding authors upon reasonable request.

## References

1. 1.

Geim, A. K. & Grigorieva, I. V. Van der Waals heterostructures. Nature 499, 419–425 (2013).

2. 2.

Novoselov, K., Mishchenko, A., Carvalho, A. & Neto, A. C. 2D materials and van der Waals heterostructures. Science 353, aac9439 (2016).

3. 3.

Liu, Y. et al. Van der Waals heterostructures and devices. Nat. Rev. Mater. 1, 16042 (2016).

4. 4.

Liu, Y., Huang, Y. & Duan, X. F. Van der Waals integration before and beyond two-dimensional materials. Nature 567, 323–333 (2019).

5. 5.

Dean, C. R. et al. Boron nitride substrates for high-quality graphene electronics. Nat. Nanotechnol. 5, 722–726 (2010).

6. 6.

Britnell, L. et al. Field-effect tunneling transistor based on vertical graphene heterostructures. Science 335, 947–950 (2012).

7. 7.

Yu, W. J. et al. Vertically stacked multi-heterostructures of layered materials for logic transistors and complementary inverters. Nat. Mater. 12, 246–252 (2013).

8. 8.

Yu, W. J. et al. Highly efficient gate-tunable photocurrent generation in vertical heterostructures of layered materials. Nat. Nanotechnol. 8, 952–958 (2013).

9. 9.

Lee, C.-H. et al. Atomically thin p–n junctions with van der Waals heterointerfaces. Nat. Nanotechnol. 9, 676–681 (2014).

10. 10.

Fang, H. et al. Strong interlayer coupling in van der Waals heterostructures built from single-layer chalcogenides. Proc. Natl Acad. Sci. USA 111, 6198–6202 (2014).

11. 11.

Cheng, R. et al. Electroluminescence and photocurrent generation from atomically sharp WSe2/MoS2 heterojunction p–n diodes. Nano Lett. 14, 5590–5597 (2014).

12. 12.

Liu, Y. et al. Toward barrier free contact to molybdenum disulfide using graphene electrodes. Nano Lett. 15, 3030–3034 (2015).

13. 13.

Cui, X. et al. Multi-terminal transport measurements of MoS2 using a van der Waals heterostructure device platform. Nat. Nanotechnol. 10, 534–540 (2015).

14. 14.

Withers, F. et al. Light-emitting diodes by band-structure engineering in van der Waals heterostructures. Nat. Mater. 14, 301–306 (2015).

15. 15.

Wang, C. et al. Monolayer atomic crystal molecular superlattices. Nature 555, 231–236 (2018).

16. 16.

Liu, Y. et al. Approaching the Schottky–Mott limit in van der Waals metal–semiconductor junctions. Nature 557, 696–700 (2018).

17. 17.

Hong, X. et al. Ultrafast charge transfer in atomically thin MoS2/WS2 heterostructures. Nat. Nanotechnol. 9, 682–686 (2014).

18. 18.

Rivera, P. et al. Valley-polarized exciton dynamics in a 2D semiconductor heterostructure. Science 351, 688–691 (2016).

19. 19.

Xu, W. et al. Correlated fluorescence blinking in two-dimensional semiconductor heterostructures. Nature 541, 62–67 (2017).

20. 20.

Zhang, C. et al. Interlayer couplings, moiré patterns, and 2D electronic superlattices in MoS2/WSe2 hetero-bilayer. Sci. Adv. 3, e1601459 (2017).

21. 21.

Huang, B. et al. Layer-dependent ferromagnetism in a van der Waals crystal down to the monolayer limit. Nature 546, 270–273 (2017).

22. 22.

Gong, C. et al. Discovery of intrinsic ferromagnetism in two-dimensional van der Waals crystals. Nature 546, 265–269 (2017).

23. 23.

Song, T. et al. Giant tunneling magnetoresistance in spin-filter van der Waals heterostructures. Science 360, 1214–1218 (2018).

24. 24.

Unuchek, D. et al. Room-temperature electrical control of exciton flux in a van der Waals heterostructure. Nature 560, 340–344 (2018).

25. 25.

Cao, Y. et al. Unconventional superconductivity in magic-angle graphene superlattices. Nature 556, 43–50 (2018).

26. 26.

Cao, Y. et al. Correlated insulator behaviour at half-filling in magic-angle graphene superlattices. Nature 556, 80–84 (2018).

27. 27.

Seyler, K. L. et al. Signatures of moiré-trapped valley excitons in MoSe2/WSe2 heterobilayers. Nature 567, 66–70 (2019).

28. 28.

Tran, K. et al. Evidence for moiré excitons in van der Waals heterostructures. Nature 567, 71–75 (2019).

29. 29.

Jin, C. H. et al. Observation of moiré excitons in WSe2/WS2 heterostructure superlattices. Nature 567, 76–80 (2019); correction 569, E7 (2019).

30. 30.

Alexeev, E. M. et al. Resonantly hybridized excitons in moiré superlattices in van der Waals heterostructures. Nature 567, 81–86 (2019); correction 572, E8 (2019).

31. 31.

Han, G. H. et al. Seeded growth of highly crystalline molybdenum disulphide monolayers at controlled locations. Nat. Commun. 6, 6128 (2015).

32. 32.

Kang, K. et al. Layer-by-layer assembly of two-dimensional materials into wafer-scale heterostructures. Nature 550, 229–233 (2017).

33. 33.

Chen, P., Zhang, Z. W., Duan, X. D. & Duan, X. F. Chemical synthesis of two-dimensional atomic crystals, heterostructures and superlattices. Chem. Soc. Rev. 47, 3129–3151 (2018).

34. 34.

Huang, C. et al. Lateral heterojunctions within monolayer MoSe2–WSe2 semiconductors. Nat. Mater. 13, 1096–1101 (2014).

35. 35.

Gong, Y. et al. Vertical and in-plane heterostructures from WS2/MoS2 monolayers. Nat. Mater. 13, 1135–1142 (2014).

36. 36.

Duan, X. D. et al. Lateral epitaxial growth of two-dimensional layered semiconductor heterojunctions. Nat. Nanotechnol. 9, 1024–1030 (2014).

37. 37.

Li, M.-Y. et al. Epitaxial growth of a monolayer WSe2–MoS2 lateral p–n junction with an atomically sharp interface. Science 349, 524–528 (2015).

38. 38.

Chen, J. et al. Lateral epitaxy of atomically sharp WSe2/WS2 heterojunctions on silicon dioxide substrates. Chem. Mater. 28, 7194–7197 (2016).

39. 39.

Mahjouri-Samani, M. et al. Patterned arrays of lateral heterojunctions within monolayer two-dimensional semiconductors. Nat. Commun. 6, 7749 (2015).

40. 40.

Zhao, M. et al. Large-scale chemical assembly of atomically thin transistors and circuits. Nat. Nanotechnol. 11, 954–959 (2016).

41. 41.

Zhang, Z. et al. Robust epitaxial growth of two-dimensional heterostructures, multiheterostructures, and superlattices. Science 357, 788–792 (2017).

42. 42.

Sahoo, P. K., Memaran, S., Xin, Y., Balicas, L. & Gutiérrez, H. R. One-pot growth of two-dimensional lateral heterostructures via sequential edge-epitaxy. Nature 553, 63–67 (2018).

43. 43.

Xie, S. et al. Coherent, atomically thin transition-metal dichalcogenide superlattices with engineered strain. Science 359, 1131–1136 (2018).

44. 44.

Zhang, Y. et al. Edge-epitaxial growth of 2D NbS2–WS2 lateral metal–semiconductor heterostructures. Adv. Mater. 30, 1803665 (2018).

45. 45.

Ai, R. Q. et al. Growth of Single-crystalline cadmium iodide nanoplates, CdI2/MoS2 (WS2, WSe2) van der Waals heterostructures, and patterned arrays. ACS Nano 11, 3413–3419 (2017).

46. 46.

Yang, T. et al. Van der Waals epitaxial growth and optoelectronics of large-scale WSe2/SnS2 vertical bilayer p–n junctions. Nat. Commun. 8, 1906 (2017).

47. 47.

Zhang, Z. et al. Epitaxial growth of two-dimensional metal–semiconductor transition-metal dichalcogenide vertical stacks (VSe2/MX2) and their band alignments. ACS Nano 13, 885–893 (2019).

48. 48.

Wu, R. et al. Van der Waals epitaxial growth of atomically thin 2D metals on dangling-bond-free WSe2 and WS2. Adv. Funct. Mater. 29, 1806611 (2019).

49. 49.

Yu, H. et al. Wafer-scale growth and transfer of highly-oriented monolayer MoS2 continuous film. ACS Nano 11, 12001–12007 (2017).

50. 50.

Gao, Y. et al. Ultrafast growth of high-quality monolayer WSe2 on Au. Adv. Mater. 29, 1700990 (2017).

51. 51.

Zhang, Z. W. et al. Ultrafast growth of large single crystals of monolayer WS2 and WSe2. Natl Sci. Rev. https://academic.oup.com/nsr/advance-article/doi/10.1093/nsr/nwz223/5698323 (2020).

52. 52.

Clark, S. J. et al. First principles methods using CASTEP. Z. Kristallogr. 220, 567–570 (2005).

53. 53.

Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).

54. 54.

Hasnip, P. J. & Pickard, C. J. Electronic energy minimisation with ultrasoft pseudopotentials. Comput. Phys. Commun. 174, 24–29 (2006).

55. 55.

Perdew, J. P. et al. Atoms, molecules, solids, and surfaces: applications of the generalized gradient approximation for exchange and correlation. Phys. Rev. B 46, 6671–6687 (1992).

56. 56.

Head, J. D. & Zerner, M. C. A Broyden–Fletcher–Goldfarb–Shanno optimization procedure for molecular geometries. Chem. Phys. Lett. 122, 264–270 (1985).

## Acknowledgements

The authors at Hunan University acknowledge the support from National Natural Science Foundation of China (grant numbers 51991340, 51991343 and 51872086), and the Hunan Key Laboratory of Two-Dimensional Materials (grant number 2018TP1010). The planar TEM studies were conducted at the Center for Electron Microscopy at Tianjin University of Technology. The cross-sectional STEM experiments were conducted using the facilities in the Irvine Materials Research Institute (IMRI) at the University of California, Irvine. The work at University of California, Irvine was supported by the Department of Energy (DOE), Office of Basic Energy Sciences, Division of Materials Sciences and Engineering under grant DE-SC0014430.

## Author information

Authors

### Contributions

Xidong Duan conceived and designed the research. J. Li, X. Yang, Yang Liu and B. Huang contributed equally to this work. J. Li developed the defect-induced nucleation and synthesized a series of metal–semiconductor vertical heterostructure arrays. J. Li and X. Yang performed device fabrication and electrical measurements. Yang Liu and Xiangfeng Duan contributed to data analysis. X. Yang and Z.Z. participated in the investigation of growth mechanism and TEM characterizations. B. Huang conducted the theoretical calculations and wrote the related discussions with support from M.Z.S.; B.Z. and H.M. participated in the materials growth. R.W. and W.D. participated in the fabrication of field-effect transistors. Z.W. and G.Z. provided large-area materials on sapphire substrates. K.W. and J. Luo performed planar TEM studies. Z.Y.L., X. Yan and X.P. performed the cross-sectional HAADF-STEM characterizations. B. Li performed the Raman and AFM measurements. Yuan Liu and Y.H. contributed to discussions. J. Li, X. Yang, Yang Liu, Xidong Duan and Xiangfeng Duan co-wrote the manuscript with input from all of the authors. All authors discussed the results and commented on the manuscript.

### Corresponding authors

Correspondence to Xidong Duan or Xiangfeng Duan.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Extended data figures and tables

### Extended Data Fig. 1 Laser-patterned nucleation sites.

a, Optical microscopy image of WSe2 with periodically patterned defects. b, AFM image of WSe2 with patterned defects. c, The height profile of the white circle region in b, exhibiting a depth of about 0.3 nm.

### Extended Data Fig. 2 Raman and photoluminescence spectra of VSe2 and VSe2/WSe2.

Raman spectra of VSe2 on SiO2/Si (a) and VSe2/WSe2 vertical heterostructure (b). c, Photoluminescence spectra of the bare WSe2 and the overlapping VSe2/WSe2 vertical heterostructure.

### Extended Data Fig. 3 HAADF-STEM analysis of VSe2 and WSe2.

Atomic-resolution HAADF-STEM image of VSe2 (a) and the corresponding intensity profile of V (b). c, d, Atomic-resolution HAADF-STEM image of WSe2 (c) and the corresponding intensity profile of W (d).

### Extended Data Fig. 4 High-resolution view of simulated moiré structures.

Zoom-in view of four locations of moiré structures marked in Fig. 4c (i, ii, iii, i), showing three distinct atomic arrangements, corresponding to the single V atom arrangement (i), V stacking over Se (ii), and V stacking over W (iii). The two panels labelled ‘i’ are identical. The red, blue and yellow spheres correspond to V, W and Se, respectively.

### Extended Data Fig. 5 General synthesis of m-TMD/WSe2 vdWH arrays.

Optical microscopy images of the NbTe2/WSe2 vdWH arrays (a) and the VS2/WSe2 vdWH arrays (b).

### Extended Data Fig. 6 Large-scale VSe2/MoS2 vdWH arrays on continuous monolayer MoS2 thin film.

a, Typical photograph of highly oriented monolayer MoS2 continuous films grown on 2-inch sapphire wafer. b, c, Optical microscopy images of large-scale periodic VSe2/MoS2 vdWH arrays grown on continuous MoS2 thin films taken with ×10 magnification objective (b) and ×20 magnification objective (c). d–g, High-magnification optical microscopy images of periodic VSe2/MoS2 vdWH arrays collected in different regions of b, suggesting highly uniform growth of VSe2/MoS2 vdWH arrays.

### Extended Data Fig. 7 Characterizations of VSe2/WS2 vdWH arrays.

a, Typical optical microscopy image of a VSe2/WS2 vdWH array. b, Raman spectra of the bare WS2 and the overlapping VSe2/WS2 vertical heterostructure. c, d, Raman intensity mapping image of VSe2/WSe2 vdWH arrays at resonant peaks of 353 cm−1 (WS2; c) and 206 cm−1 (VSe2; d). e, Photoluminescence spectra of the bare WS2 and the overlapping VSe2/WS2 vertical heterostructure. f, Photoluminescence intensity mapping image at 658 nm (WS2 emission).

### Extended Data Fig. 8 Electrical characterization of a bilayer WSe2 transistors with synthetic vdW contacts.

a, b, Output (a) and transfer (b) curves of a typical device with synthetic VSe2 vdW contacts on 285-nm SiO2/Si. The channel length is about 2.0 μm.

## Rights and permissions

Reprints and Permissions

Li, J., Yang, X., Liu, Y. et al. General synthesis of two-dimensional van der Waals heterostructure arrays. Nature 579, 368–374 (2020). https://doi.org/10.1038/s41586-020-2098-y

• Accepted:

• Published:

• Issue Date:

• ### Electronic structure and magnetic properties of two-dimensional h-BN/Janus 2H-VSeX (X = S, Te) van der Waals heterostructures

• Yingdan Xu
• , Shengmei Qi
•  & Wenbo Mi

Applied Surface Science (2021)

• ### Tunable electronic, optical, and spintronic properties in InSe/MTe2 (M = Pd, Pt) van der Waals heterostructures

• Donghui Wang
• , Weiwei Ju
• , Dawei Kang
• , Tongwei Li
•  & Haisheng Li

Vacuum (2020)

• ### Moiré is More: Access to New Properties of Two-Dimensional Layered Materials

• Yao Xiao
• , Jinglu Liu
•  & Lei Fu

Matter (2020)

• ### Enhanced Optoelectronic Performance of CVD-Grown Metal–Semiconductor NiTe2/MoS2 Heterostructures

• Xiaokun Zhai
• , Xing Xu
• , Jiangbo Peng
• , Fangli Jing
• , Qinglin Zhang
• , Hongjun Liu
•  & Zhanggui Hu

ACS Applied Materials & Interfaces (2020)

• ### Programmable Single‐Crystalline PbI 2 Microplate Arrays and Their Organic/Inorganic Heterojunctions

• Jinjin Zhao
• , Haohao Li
• , Yuchen Qiu
• , Jiangang Feng
• , Junchuan Yang
• , Ke He
• , Penghua Wu
• , Naizheng Wang
• , Hanfei Gao
• , Lei Jiang
•  & Yuchen Wu