Vertical and in-plane heterostructures from WS2/MoS2 monolayers.

TLDR: A one-step growth strategy for the creation of high-quality vertically stacked as well as in-plane interconnected heterostructures of WS2/MoS2 via control of the growth temperature is reported.

Abstract: Layer-by-layer stacking or lateral interfacing of atomic monolayers has opened up unprecedented opportunities to engineer two-dimensional heteromaterials. Fabrication of such artificial heterostructures with atomically clean and sharp interfaces, however, is challenging. Here, we report a one-step growth strategy for the creation of high-quality vertically stacked as well as in-plane interconnected heterostructures of WS2/MoS2 via control of the growth temperature. Vertically stacked bilayers with WS2 epitaxially grown on top of the MoS2 monolayer are formed with preferred stacking order at high temperature. A strong interlayer excitonic transition is observed due to the type II band alignment and to the clean interface of these bilayers. Vapour growth at low temperature, on the other hand, leads to lateral epitaxy of WS2 on MoS2 edges, creating seamless and atomically sharp in-plane heterostructures that generate strong localized photoluminescence enhancement and intrinsic p-n junctions. The fabrication of heterostructures from monolayers, using simple and scalable growth, paves the way for the creation of unprecedented two-dimensional materials with exciting properties.

Figures

Figure 1. Schematic of the synthesis and the overall morphologies of the vertical stacked and in-plane WS2/MoS2 heterostructures. (A-D) Schematic, optical and SEM images of the vertical stacked WS2/MoS2 heterostructures synthesized at 850 ˚C, showing the bilayer feature and the high yield of the triangular heterostructures. (E-H) Schematic, optical and SEM images of the WS2/MoS2 in-plane heterojunctions grown at 650 ˚C. (G) is an optical image of the interface between WS2 and MoS2 with enhanced color contrast, showing the abrupt change of contrast at the interface. SEM images are presented in reverse contrast. The green, purple and yellow spheres in (A) and (E) represent W, Mo and S atoms, respectively. (I) Schematic of the synthesis process for both heterostructures.

Figure 4. Atomic structure of the lateral heterojunctions between WS2 and MoS2 monolayers. (A, B) Atomic resolution Z-contrast STEM images of the in-plane interface between WS2 and MoS2 domains. Small roughness resulting from interfacial steps can be seen in A. The red dashed lines highlight the atomically sharp interface along the zigzag-edge direction. The orange and pink dashed lines depict the atomic planes along the arm-chair and zigzag directions, respectively, indicating the WS2 and MoS2 domains share the same crystal orientation. (C, D) Atomic resolution Z-contrast images of the atomically sharp lateral interfaces along the zigzag (C) and armchair (D) directions. The atomic models on the right correspond to the structure in the highlighted regions. Scale bars: (A) 1 nm; (B-D) 0.5 nm.

Figure 3. Raman and PL characterization of the WS2/MoS2 vertical heterostructure. (A) Optical image of a WS2/MoS2 heterostructure used for Raman characterization. (B) Raman spectra taken from the four points marked in (A), showing that the monolayer region is pure MoS2, while the double layer area is the superposition of MoS2 and WS2 monolayers. (C, D) Raman intensity mapping at 384 cm-1 and 357 cm -1, respectively. The lower Raman intensity at the center of the triangle in (C) is due to the coverage of WS2. (E) Optical image of a WS2/MoS2 heterostructure used for PL characterization. (F) PL Spectra taken from the four points marked in (E), showing the characteristic MoS2 PL peak at the monolayer region and three peaks at the bilayer region. (G) PL intensity mapping at 680 nm shows localized PL enhancement around the step edge of the bilayer region. (H) PL spectra of CVD-grown WS2/MoS2 bilayer, WS2/MoS2 bilayer made by mechanical transfer, and CVD-grown MoS2 and WS2 bilayers, respectively. All spectra were taken at the same laser intensity and plotted

Figure 5. Raman and PL characterizations of in-plane WS2/MoS2 heterojunction. (A) Optical microscopy image of a triangular in-plane WS2/MoS2 heterojunction for Raman and PL characterization. (B) Raman spectra taken from the points marked by 1-3 in its inset, showing characteristic WS2 (point 1) and MoS2 (point 3) peaks in the outer and inner triangles, respectively, and their superposition at the interface region (point 2). (C) Combined Raman intensity mapping at 351 cm-1 (yellow) and 381 cm-1 (purple), showing the core-shell structure with WS2 as the shell and MoS2 as the core. (D) PL spectra of the points marked by 1-5 in its inset. The peak positions for spectra 1 and 5 are 630 nm and 680 nm, respectively, indicating pure WS2 and pure MoS2. The PL peak shifts as approaching the interface (points 2 and 4). At the interface (point 3), a stronger broad peak at 650 nm shows up. (E) PL spectra at the interface (point 3), at the intersection of interface (point 6) and the superposition of spectra from pure MoS2 (point 5) and pure WS2 (point 1). (F) Combined PL intensity mapping at 630 nm (orange) and 680 nm (green). (G) PL intensity mapping at 650 nm, showing localized response around the interface. The optical image with interface highlighted

Figure 2. STEM-Z-contrast imaging and elemental mapping of the stacked WS2/MoS2 heterostructures. (A) Low-magnified false-colored Z-contrast image of the sample, where monolayer MoS2 is shown in blue, monolayer WS2 in green, and WS2/MoS2 bilayer in orange. (B) Zoom in view of the region highlighted in (A). (C) Z-contrast image intensity profile along the highlighted dashed line in (B), showing the distinct contrast variation among the different monolayers and bilayer region. (D) Elemental mapping of Mo, W, and S from the whole area shown in (B). (E) Z-contrast image of the bilayer region with 2H stacking orientation. The brighter columns are overlapping columns of W and S2, while the less bright columns are overlapping of S2 and Mo. The green arrow points to the atomic positions where W atom is replaced by Mo in the WS2 layer, which has similar intensity to its neighboring site. Below: Image intensity profile acquired along the yellow rectangle in (E). (F) Z-contrast image of the step edge of the WS2/MoS2 bilayer. The green dash line highlights the step edge,

Full text

1
Vertical and In-Plane Heterostructures from WS
2
/MoS
2
Monolayers
Yongji Gong
1, 2 †
, Junhao Lin
3, 4 †
, Xingli Wang
5 †
Gang Shi
2
, Sidong Lei
2
, Zhong Lin
6
, Xiaolong Zou
2
,
Gonglan Ye
2
, Robert Vajtai
2
, Boris I. Yakobson
2
, Humberto Terrones
7
, Mauricio Terrones
6, 8
, Beng
Kang Tay
5
, Jun Lou
2
, Sokrates T. Pantelides
3, 4
, Zheng Liu
5
, Wu Zhou
3 *
, Pulickel M. Ajayan
1,2 *
1 Department of Chemistry, Rice University, Houston, TX 77005, USA
2 Department of Materials Science & NanoEngineering, Rice University, Houston, TX 77005, USA
3 Materials Science & Technology Division, Oak Ridge National Lab, Oak Ridge, TN 37831, USA
4 Department of Physics and Astronomy, Vanderbilt University, Nashville, TN 37235, USA
5 School of Materials Science and Engineering, School of Electrical and Electronic Engineering
Nanyang Technological University, 639798, Singapore
6 Department of Physics and Center for 2-Dimensional and Layered Materials, The Pennsylvania State
University, University Park, Pennsylvania 16802, USA
7 Department of Physics, Applied Physics and Astronomy, Rensselaer Polytechnic Institute,
Johnson-Rowland Science Center, 110 Eighth Street, Troy, NY 12180, USA.
8 Department of Chemistry, Department of Materials Science and Engineering & Materials Research
Institute, The Pennsylvania State University, University Park, PA 16802, USA
† Y.G., J.L. and X.W. contributed equally to this work
* Corresponding authors: wu.zhou.stem@gmail.com; ajayan@rice.edu
Layer-by-layer stacking or lateral interfacing of atomic monolayers has opened up new
unprecedented opportunities to engineer two-dimensional (2D) heteromaterials.
Fabrication of such artificial heterostructures with atomically clean and sharp interfaces,
however, is challenging. Here, we report a one-step growth strategy for the creation of
high-quality vertically stacked as well as in-plane interconnected heterostructures of
WS
2
/MoS
2
via control of the growth temperature. Vertically stacked bilayers with WS
2
epitaxially grown on top of MoS
2
monolayer are formed with preferred stacking order at
high temperature. Strong interlayer excitonic transition is observed due to the type II
band alignment and to the clean interface of these bilayers. Vapor growth at low
temperature, on the other hand, leads to lateral epitaxy of WS
2
on MoS
2
edges, creating
seamless and atomically sharp in-plane heterostructures that generate strong localized
photoluminescence enhancement and intrinsic p-n junctions. The fabrication of
heterostructures from monolayers, using simple and scalable growth, paves the way for
the creation of unprecedented two-dimensional materials with exciting properties.

2
Heterostructures have been the essential elements in modern semiconductor industry, and play
a crucial role in high-speed electronics and opto-electronic devices
1,2
. Beyond conventional
semiconductors, two-dimensional (2D) materials provide a wide range of basic building
blocks with distinct optical and electrical properties, including graphene
3
, hexagonal boron
nitride
4,5
, and transition-metal dichalcogenides (TMDs)
6-9
. These atomic monolayers could
also be combined to create van der Waals heterostructures, where monolayers of multiple 2D
materials are stacked vertically layer-by-layer, or stitched together seamlessly in plane to form
lateral heterojunctions. Many novel physical properties have been explored on such van der
Waals heterostructures, and devices with improved performance have been demonstrated
10-14
.
The lateral heterojunctions could also lead to exciting new physics and applications. For
example, the semiconducting monolayer TMDs can serve as building blocks for p-n junctions
and other opto-electronic devices
15-17
. However, the fabrication of 2D heterostructures with
clean and sharp interfaces, essential for preserving opto-electronic properties driven by the
interlayer or intralayer coupling, remains challenging. Van der Waals heterostructures could
be created by stacking different 2D materials using mechanical transfer techniques
12
.
However, the stacking orientation cannot be precisely controlled, and the interface between
layers can be easily contaminated
18,19
, not to mention the challenge for massive production of
the samples. Lateral heterostructures, in contrast, can only be created via growth. Both
vertical and in-plane heterostructures of semimetallic graphene and insulating h-BN have
been recently demonstrated via chemical vapor deposition (CVD)
20-24
; however, direct growth
of heterostructures consisting of different semiconducting monolayers has not been achieved.
Here, we report a scalable single-step vapor phase growth process for the creation of highly
crystalline vertical stacked bilayers and in-plane interconnected WS
2
/MoS
2
heterostructures,
respectively, under different growth temperature. Atomic resolution scanning transmission
electron microscopy (STEM) imaging reveals that high temperature growth yields
predominantly vertically stacked bilayers with WS
2
epitaxially grown on top of the MoS
2
monolayer, following the preferred 2H stacking. In contrast, the low temperature growth
creates mostly lateral heterostructures of WS
2
and MoS
2
within single hexagonal monolayer
lattice, with atomically sharp heterojunctions along both the zigzag and armchair directions.

3
The vertical and lateral heterostructures are further verified by Raman and photoluminescence
(PL) spectroscopy characterization. Strong interlayer or intralayer excitonic interaction
between MoS
2
and WS
2
are observed by PL spectroscopy for the first time on these two types
of heterostructures, owing to their clean and sharp interfaces. Specifically, a new bandgap of
1.42 eV is observed in the bilayer heterostructure, arising from the interlayer excitonic
transition between MoS
2
and WS
2
25,26
; whereas a strong localized PL enhancement is
observed at the lateral interface between MoS
2
and WS
2
, presumably due to the increased
excitonic recombination of the as-generated electron-hole pairs at the atomically sharp
interface
27
. These two types of heterostructures are further demonstrated to be building blocks
for high mobility field effect transistors (FET) and planar monolayer p-n junctions, indicating
their potential for constructing unique devices.
Synthesis and Morphology
Figure 1I shows the scheme for the growth of WS
2
/MoS
2
heterostructures. Molybdenum
trioxide (MoO
3
) powder is placed in front of the bare SiO
2
/Si wafer for the growth of MoS
2
,
while mixed power of tungsten and tellurium is scattered on the wafer for the growth of WS
2
.
The addition of tellurium helps to accelerate the melting of tungsten powder during the
growth (Fig. S4). Sulfur powder is put upstream within the low temperature zone. Argon is
used to protect the system from oxygen and carry sulfur vapor from the upstream of the tube
during the reaction. The difference in their nucleation and growth rate gives rise to sequential
growth of MoS
2
and WS
2
, instead of Mo
x
W
1-x
S
2
alloy, and the precise reaction temperature
determines the structure of the final product: vertical stacked bilayers are preferred at ~850 ˚C,
while in-plane lateral heterojunctions dominate when the synthesis was carried out at ~650˚C
(see Methods for more details). A brief discussion of the possible mechanism of the
temperature-selective growth is provided in the Supplementary Information. This simple,
scalable growth process creates clean interfaces between the two monolayer components,
which is advantageous over mechanical transfer of layers.
The morphology of the WS
2
/MoS
2
vertical and in-plane heterostructures was examined by
optical microscopy, scanning electron microscopy (SEM), and atomic force microscopy

4
(AFM). Figures 1A - 1D are the schematic and typical optical and SEM images of the vertical
stacked heterostructures, showing individual WS
2
/MoS
2
bilayer triangles and high yield of
heterostructures. The bilayers can be easily distinguished from monolayers via optical
contrast (Fig. 1B), with MoS
2
monolayers showing light purple color and the bilayer regions
in much darker purple. The domain size of the bottom MoS
2
layer is typically larger than 10
μm. Both totally covered and partially covered WS
2
/MoS
2
bilayer (Fig. S5) can be found,
providing different geometries for device fabrication. The schematic and morphology of
WS
2
/MoS
2
in-plane heterostructures is shown in Figs. 1E - 1H, where the lateral interface
between monolayer MoS
2
and WS
2
can be easily distinguished by the contrast difference.
SEM and optical images shown in Fig. S6 demonstrate the high-yield of such in-plane
heterostructure obtained from this growth method. The difference in bilayer or monolayer
morphology of these two types of heterostructures is further verified by AFM images
presented in Fig. S7.
Vertically stacked heterostructure - bilayer
The atomic structure of the vertical stacked WS
2
/MoS
2
bilayers was studied by Z-contrast
imaging and elemental mapping on an aberration-corrected STEM (see Methods). Figure 2A
shows the morphology of the as-transferred stacked WS
2
/MoS
2
heterostructure in a
low-magnified Z-contrast image, where the image intensity is directly related to the averaged
atomic number and the thickness of the sample
28-30
. A WS
2
monolayer would, therefore,
display higher image contrast than a MoS
2
monolayer, while the image intensity from the
bilayer heterostructure is roughly the sum of that from its two monolayer components. In
order to highlight the different regions in the sample, the image in Fig. 2A is shown in a false
color scale. Most of the sample is covered by continuous bilayer heterostructure (orange
region), while at some intentionally induced broken edges (see Methods) both of the
individual monolayers can be identified (with MoS
2
shown in blue and WS
2
shown in green).
Figure 2B shows a magnified image of the region highlighted in Fig. 2A. The obvious
contrast step across the two individual layers, as shown by the image intensity line profile in
Fig. 2C, demonstrates the presence of separated MoS
2
and WS
2
monolayers instead of a
homogenous Mo
x
W
1-x
S
2
alloy. Elemental mapping of Mo, W and S (Figs. 2D and S8) from

5
the same region unambiguously confirms that MoS
2
and WS
2
are well separated into two
atomic layers, forming vertical bilayer heterostructures.
Figures 2E and 2F show atomic resolution Z-contrast images from the bilayer region and a
step edge of the WS
2
/MoS
2
heterostructure, respectively. The alternative bright and dark
atomic column arranging in the hexagonal lattice suggests the as-grown stacked WS
2
/MoS
2
heterostructure preserves the 2H stacking, where the bright and dark columns are W and Mo
atom aligned with a S
2
column, respectively, as illustrated in Fig. 2G. The WS
2
/MoS
2
heterostructure grown by our one-step growth method is found to have predominantly the 2H
stacking, which exemplifies the advantage of this direct growth method over mechanical
transfer method where the stacking orientation of the heterostructure cannot be well
controlled. As a side note, Mo substitution in the WS
2
layer can be occasionally observed, as
indicated by the reduced contrast at the W atomic sites (green arrows in Fig. 2E and the
associated intensity line profile). Similarly, some trace amount of W atoms is also found to
substitute into the MoS
2
layer (Fig. 2F). However, the substitution is at a fairly low
concentration (~ 3%, see Fig. S9 for details), which would only have minimum effect on the
properties of the MoS
2
and WS
2
monolayers.
Raman and PL spectroscopy were used to further characterize the vertical bilayer
heterostructure. As shown in Figs. 3A and 3B, Raman spectra collected from the light purple
area (points 1 and 2) show only the E
’
(at 383.9 cm
-1
) and A
1
’
(at 405.3 cm
-1
) peaks of MoS
2
monolayer
8,9
, confirming the bottom layer is MoS
2
31
. In the bilayer region (point 3 and 4 in
the dark purple area), however, two additional peaks located at 356.8 cm
-1
and 418.5 cm
-1
are
observed, which can be assigned to the overlapping 2LA(M) & E
’
and A
1
’
modes, respectively,
of the top WS
2
monolayer
6,32
(details in Fig. S10). Raman intensity mapping using the MoS
2
E
’
mode at 384cm
-1
and the WS
2
E
’
mode at 357cm
-1
further demonstrate the formation of
WS
2
/MoS
2
bilayer stacks, as shown in Figs. 3C and 3D.
The PL spectra (Fig. 3F) acquired from the monolayer region (points 1 and 2 in Fig. 3E) show
only a strong peak at the wavelength of 680 nm, corresponding to the 1.82 eV direct excitonic

Frequently asked questions

Q1. What contributions have the authors mentioned in the paper "Vertical and in-plane heterostructures from ws2/mos2 monolayers" ?

8 Department of Chemistry, Department of Materials Science and Engineering & Materials Research Institute, The Pennsylvania State University, University Park, PA 16802, USA † Y. G., J. L. and X. W. contributed equally to this work * Corresponding authors: wu. zhou. stem @ gmail. 

Q2. Why is the lateral interface a bit diffuse?

Note that due to the large laser spot size (~1 μm) used in their experiment, the lateral interface in the Raman and PL mappings appears a bit diffuse, and the Raman spectrum from the interface area correspond to signals from both sides of the atomically abrupt heterojunction. 

Q3. How is the bandgap observed in the bilayer heterostructure?

a new bandgap of 1.42 eV is observed in the bilayer heterostructure, arising from the interlayer excitonic transition between MoS2 and WS2 25,26; whereas a strong localized PL enhancement is observed at the lateral interface between MoS2 and WS2, presumably due to the increased excitonic recombination of the as-generated electron-hole pairs at the atomically sharp interface27. 

Q4. What can be achieved in randomly assembled heterostructures?

The specific orientation relationships and ordering between the individual monolayer domains can lead to specific interface electronic properties which cannot be obtained in randomly assembled van der Waals hetero-materials. 

Q5. What could be the interesting application of the WS2 heterostructures?

Such scalable methods to grow engineered 2D heterostructures could lead to interesting applications such as vertically stacked FET devices and planar monolayer devices. 

Q6. How many PL peaks are observed in the stacked WS2/MoS2?

In addition, the 1.82 eV (680 nm) and 1.97 eV (630 nm) PL peaks observed in the stacked WS2/MoS2 bilayer are almost vanished in CVD-grown MoS2 bilayer and WS2 bilayer. 

Q7. What is the effect of the CVD method on the WS2/MoS2 hetero?

This observation suggests that the MoS2 and WS2 layers in the bilayer heterostructure, on one hand, behave as individual monolayers, and, on the other hand, generate new functionalities (a new direct band gap) of WS2/MoS2 heterostructure via interlayer coupling owing to the clean interface. 

Q8. What is the advantage of using their method for the direct growth of crystalline heterostructures?

These results highlight the advantage of using their CVD method for the direct growth of crystalline heterostructures, in which layer transfers are not needed and clean interface could be readily obtained. 

Q9. What is the lateral roughness of the WS2/MoS2 interface?

The interfacial steps most likely originate from small fluctuations of the MoS2 growth rate at the nm-scale, and their presence contributes to the overall roughness of the lateral WS2/MoS2 interface. 

Q10. What is the morphology of the WS2/MoS2 heterostructure?

Elemental mapping of Mo, W and S (Figs. 2D and S8) fromthe same region unambiguously confirms that MoS2 and WS2 are well separated into two atomic layers, forming vertical bilayer heterostructures. 

Q11. How do the authors demonstrate the high quality of the CVD-grown heterostructures?

In order to illustrate the high quality of the CVD-grown heterostructures, the authors demonstrate high-mobility back-gating vertically stacked WS2/MoS2 field-effect transistors (FETs) (Fig. 3I, Fig. S14). 

Q12. How much is the roughness of the lateral interface estimated?

The overall roughness of the lateral interface is estimated to be ~4 unit cells over a width of 15 nm (Figs. S17 and S18), and the authors expect that this could be further reduced by optimizing the CVD growth conditions.