Vapour phase growth and grain boundary structure of molybdenum disulphide atomic layers

ARTICLESPUBLISHED ONLINE: 9 JUNE 2013 | DOI: 10.1038/NMAT3673

Vapour phase growth and grain boundarystructure of molybdenum disulphide atomic layersSina Najmaei1†, Zheng Liu1†, Wu Zhou2,3†, Xiaolong Zou1, Gang Shi1, Sidong Lei1, Boris I. Yakobson1,Juan-Carlos Idrobo3, Pulickel M. Ajayan1 and Jun Lou1*Single-layered molybdenum disulphide with a direct bandgap is a promising two-dimensional material that goes beyondgraphene for the next generation of nanoelectronics. Here, we report the controlled vapour phase synthesis of molybdenumdisulphide atomic layers and elucidate a fundamental mechanism for the nucleation, growth, and grain boundary formationin its crystalline monolayers. Furthermore, a nucleation-controlled strategy is established to systematically promote theformation of large-area, single- and few-layered films. Using high-resolution electron microscopy imaging, the atomic structureand morphology of the grains and their boundaries in the polycrystalline molybdenum disulphide atomic layers are examined,and the primary mechanisms for grain boundary formation are evaluated. Grain boundaries consisting of 5- and 7- member ringsare directly observed with atomic resolution, and their energy landscape is investigated via first-principles calculations. Theuniformity in thickness, large grain sizes, and excellent electrical performance signify the high quality and scalable synthesisof the molybdenum disulphide atomic layers.

Graphene has shown many fascinating properties as asupplement to silicon-based semiconductor technologies1–4.Consequently, great effort has been devoted to the devel-

opment and understanding of its synthetic processes5–8. However,graphene’s high leakage current, due to its zero bandgap energy,is not suitable for many applications in electronics and optics9,10.Recent developments in two different classes of materials—transition metal oxides and sulphides—have shown great promisein filling the existing gaps10–12. The successful demonstration ofmolybdenum disulphide (MoS2)-based field-effect transistors11(FET) has prompted an intense exploration of the physicalproperties of few-layeredMoS2 films13–17.

MoS2 is a layered semiconductor with a bandgap in the range1.2–1.8 eV, whose physical properties are significantly thickness-dependent13,14. For instance, a considerable enhancement in theMoS2 photoluminescence has been observed as the materialthickness is decreased14. The lack of inversion symmetry insingle-layer MoS2 and developments in controlling the availablevalley quantum states in this material have brought valleytronicapplications closer to reality15. However, the large-scale synthesis ofhigh-qualityMoS2 atomic layers is still a challenge.

Recent top-down approaches to obtain large-area MoS2 thinfilms have received considerable attention18–20. Nevertheless, thelack of uniformity in thickness and size undermines the viabilityof such approaches. Other scalable MoS2 synthesis techniques havefocused on the direct sulphurization of molybdenum-containingfilms21–23. Recent work on the solid phase sulphurization ofmolybdenum thin films revealed a straightforward method forlarge-area MoS2 synthesis22. However, the suboptimal qualityof samples prepared using this synthetic route—exemplified bytheir low carrier mobility (0.004–0.04 cm−2 V−1 s−1), as com-pared with mechanically-exfoliated samples (0.1–10 cm2 V−1 s−1),and small grain sizes (∼20 nm)—remains a problem22.Alternatively, Liu et al. have exploited the sulphurization of

1Department of Mechanical Engineering and Materials Science, Rice University, Houston, Texas 77005, USA, 2Department of Physics and Astronomy,Vanderbilt University, Nashville, Tennessee 37235, USA, 3Materials Science and Technology Division, Oak Ridge National Laboratory, Oak Ridge,Tennessee 37831, USA. †These authors contributed equally to this work. *e-mail: [email protected]

ammonium tetrathiomolybdate films23. However, this methodbears a considerable level of complexity in precursor preparationand in achieving sufficient quality, thus limiting its feasibility23.

Traditionally, the sulphurization of molybdenum trioxide(MoO3) has been the primary approach in MoS2 nanomaterialsynthesis and it has been widely studied24–31. Making use of itslow melting and evaporation temperatures, Lee et al. demonstratedthat MoO3 is a suitable precursor for chemical vapour deposition(CVD) of MoS2 thin films24. This work studied the nucleationand growth process in CVD-grown MoS2 atomic layers facilitatedby seeding the substrate with graphene-like species, but lackeda comprehensive characterization of grains and grain boundarystructures24. The evolution in graphene synthesis suggests that anin-depth understanding of the CVD process for single-crystallinedomain nucleation and growth is essential to the large areapreparation of high-quality materials5–8. Here, we develop a CVD-based procedure for the large-area synthesis of highly crystallineMoS2 atomic layers by vapour-phase MoO3 sulphurization.Furthermore, we evaluate the growth process, grain morphology,and grain boundary structure of the polycrystalline MoS2 atomiclayers and characterize their corresponding electrical performance.

To understand the growth process of MoS2, the sourcesupply was controlled by dispersing MoO3 nanoribbons withvaried concentrations on a substrate, away from the designatedgrowth region, while all other parameters were kept constant(Supplementary Fig. S1). The experiments show that the synthesisof MoS2 was limited by the diffusion of vapour-phase MoO3−x:decreasing the density of dispersed nanoribbons systematicallyslows or stops the growth at certain points, providing a means toexplore the growth process. Several stages were observed during theMoS2 atomic layer growth. Initially, small triangular domains werenucleated at random locations on the bare substrate (Fig. 1a). Then,the nucleation sites continued to grow and formed boundarieswhen two or more domains met (Fig. 1b,c), resulting in a partially

754 NATURE MATERIALS | VOL 12 | AUGUST 2013 | www.nature.com/naturematerials

© 2013 Macmillan Publishers Limited. All rights reserved

http://www.nature.com/doifinder/10.1038/nmat3673

mailto:[email protected]

http://www.nature.com/naturematerials

NATURE MATERIALS DOI: 10.1038/NMAT3673 ARTICLES

350

Inte

nsity

(a.

u.)

400 450

Δ~25.6 cm¬1

Δ~21.1 cm¬1

Δ ~20.6 cm¬1

Shift (cm¬1)

500 550

MoS2

SiO280 μm 15 μm

2 μm 20 μm 10 μm 500 μm

a b c d

e f g h

Figure 1 | The fundamental growth process of MoS2 films and controlled nucleation. a–d, SEM images showing the growth process of MoS2 from smalltriangles to continuous films. e, Optical image of a large-area continuous film synthesized on substrates with rectangular shaped patterns. The patterns aredistinguishable by their bright blue colour and the areas between them are completely covered by continuous MoS2 layers. f, The close-up optical image ofthe as-synthesized MoS2 film on the patterned substrate, demonstrating that the method described typically results in single- and bilayered films betweenthe pillars and thicker samples on the pillars; these patterns act as nucleation promoters. g, Raman spectra acquired from different regions highlighted in f,where the numbers presented refers to the difference (1) between the out-of-plane (A1g) and in-plane (E1

2g) Raman peaks appearing in the∼380–410 cm−1 range, showing the thickness and its small variability in the sample. h, Large-area, approximately 1× 1 cm2 film transferred to a newsubstrate from a patterned substrate using conventional polymer-base transfer techniques.

continuous film. This process can eventually extend into large-areasingle-layered MoS2 continuous films if sufficient precursor supplyand denser nucleation sites are provided (Fig. 1d). Furthermore,we determined sulphur concentration and chamber pressure astwo key parameters that control the prospects of the materialgrowth (Supplementary Figs S2 and S3). These parameters andtheir synergistic effects on the morphology, nucleation, and growthof MoS2 crystals were used to control the MoS2 synthesis inour experiments, as detailed in the Supplementary Information.These observations show that, in contrast to graphene, whichnucleates abundantly on copper substrates without any treatment8,the limiting factor in the growth of large-area, highly crystallineMoS2 films is the rare and complicated nucleation process of thismaterial on bare SiO2 substrates. These observations are also inagreement with previous studies on CVD-grownMoS2 (ref. 24) andelucidate the importance of efficient control of nucleation, essentialto the large-area growth of theMoS2 atomic layers.

In the quest for feasible strategies to control the nucleationprocess, we take advantage of some of our common experimentalobservations. Our experiments show that MoS2 triangular domainsand films are commonly nucleated and formed in the vicinity ofthe substrates’ edges, scratches, dust particles, or rough surfaces(Supplementary Fig. S4). We used this phenomenon to control thenucleation by strategically creating step edges on substrates usingconventional lithography processes (Fig. 1e). Patterned substrateswith a uniform distribution of rectangular SiO2 pillars (40×40 µm2

in size, 40 µm apart, and ∼40 nm thick) were directly used in theCVD process for MoS2 growth (Fig. 1e). The pillars facilitate a highdensity of domain nucleation and the continued growth allows theformation of large-area continuous films (Supplementary Fig. S5).Raman spectroscopy measurements indicate that the MoS2 atomiclayers grow on the surface of the pillars as well as on the valley spacebetween them, suggesting that both the surface roughness and edgesmay contribute to the nucleation process (Fig. 1f,g). The as-grownfilms are predominantly single-layered, with small areas consisting

of two or more layers at the preferred nucleation sites (Fig. 1g). Ourobservations show that the pattern-based growth process followspressure and sulphur concentration dependencies similar to thegrowth ofMoS2 onpristine substrates. The strategies to enhance thisapproach, such as variations in the size and geometry of the pillars,and their effects on the growth process are discussed in Supplemen-tary Figs S6 and S7. The inherent dependence of this approach onthe edge-based nucleation resembles observations and theoreticalpredictions in the growth of some other layeredmaterials32,33.

Theoretical studies have revealed a significant reduction in thenucleation energy barrier of graphene at step edges, as comparedwith flat surfaces of transition metal substrates33. We propose thatsimilar edge-based catalytic processes are also involved in the initia-tion ofMoS2 growth. The simple lithographymethod applied in thepreparation of silicon oxide patterns and step edges on insulatingsubstrates provides a unique and robust strategy for the growth oflarge-area, high-quality MoS2 films that can be readily transferredto any substrate or applied to fabricate devices (Fig. 1h). Thisstrategy can potentially be adopted to grow large-area MoS2 filmson other insulating substrates, such as hexagonal boron nitride,to take advantage of its atomically smooth surface and possibleelectrical performance enhancement. Further experiments revealthat MoS2 growth on both patterned and bare substrates followsthe same process: the initial nucleation of triangular domainsand their subsequent growth and coalescence. As we establish theknowledge of the MoS2 atomic layer growth, it becomes evidentthat the nucleation and growth of the triangular domains, as wellas the formation of boundaries among these domains, play animportant role in the development of large-areaMoS2 atomic layers.Therefore, the characterization of the triangular domains becomesessential to understanding the advantages and limitations of thissynthetic route and the proposed growthmechanism.

A representative atomic force microscopy (AFM) image of aMoS2 domain is shown in Fig. 2a. The thickness of these triangles isapproximately 0.7 nm, corresponding to one MoS2 atomic layer,

NATURE MATERIALS | VOL 12 | AUGUST 2013 | www.nature.com/naturematerials 755




ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT3673

4 µm200 nm 5 Å

1.0

0.8

0.6

0.4

Z (

nm)

X (µm)

0.2

0.0

0.0 0.4 0.8 1.2 1.6

¬0.2

a b c

Figure 2 | Characterization of the MoS2 triangular domains. a, AFM image from a MoS2 triangular flake. Nucleation sites at the centre of the MoS2

triangle are also visible. Inset: line scan demonstrating the thickness of a single layered MoS2. b, Bright-field TEM image of the MoS2 triangular flakes.Inset: the SAED pattern acquired from this triangle demonstrates it is a single crystal with hexagonal structure. c, STEM-ADF image representing thedefect-free hexagonal structure of the triangular MoS2 flake. The brighter atoms are Mo and lighter ones are S2 columns. Inset: the FFT pattern of theimage confirming the hexagonal MoS2 structure.

with high level of uniformity in thickness, as also confirmedby Raman spectroscopy (Supplementary Fig. S8). The chemicalcomposition of the sample was confirmed by X-ray photoelectronspectroscopy (XPS) and electron energy loss spectroscopy analysis(Supplementary Fig. S9). The equilateral geometry of the triangleswith perfect 60◦ angles suggests the single-crystalline nature ofthese nanostructures, with edges parallel to a specific lattice ori-entation (Fig. 2a). Transmission electron microscopy (TEM) andaberration-corrected scanning transmission electron microscopy(STEM) were then applied to examine the crystal structure of MoS2triangles and atomic layers. The selected area electron diffraction(SAED) pattern from a typical MoS2 triangle (Fig. 2b) presentsonly one set of six-fold symmetry diffraction spots, confirming theirsingle-crystallinity with hexagonal structure. The atomic structureof the single-layer triangle is shown by the STEM annular darkfield (ADF) image in Fig. 2c, where the alternating brighter anddimmer atomic positions in the hexagonal rings correspond to Moand S2, respectively. Quantitative analysis of the STEM-ADF imageintensity confirms that the sample is single-layer MoS2 with two Satoms at each S2 site34. Moreover, a uniform and almost defect-freestructure was observed across the entire single-crystal domain.Nevertheless, through the formation of grain boundaries whenthese single-crystalline domains merge, polycrystalline films ofMoS2 form. It is known that the grain boundaries in polycrystallinetwo-dimensional materials degrade their physical properties and,to overcome these problems, large grain sizes are desirable35.Therefore, characterization of crystallinity in the large-area MoS2films is necessary to assess the quality of thematerial.

We evaluate the crystallinity of the large-area continuous MoS2films by examining the microstructure of these atomic layers usingdark-field TEM imaging36–38. A total of 507 dark-field TEM imagesacquired from the same diffraction spot were used to construct thedark-field TEM image (Supplementary Fig. S10). The∼15×15 µm2

region shown in Fig. 3a contains two areas of single-layer MoS2and demonstrates a contrast between the different areas of uniformcrystal orientations. The white regions represent single-layeredMoS2 with the same lattice orientation. The dark-field TEM images,along with additional results presented in the SupplementaryInformation, demonstrate the large grain size (several tens ofmicrometres) of the MoS2 films. This is consistent with the averagesize of the triangular domains observed in our experiments beforethey coalesce, and reaffirms our hypothesis for the MoS2 growthprocess. Furthermore, the thickness uniformity over regions greaterthan tens of micrometres further exemplifies the high quality ofthese samples. Nevertheless, grain boundaries that mediate the

crystal orientation transitions between the grains are one of themajor sources of defects inmaterials andwarrant closer scrutiny.

An example of a grain boundary in MoS2 atomic layers isshown in Fig. 3b, where the SAED pattern collected from thehighlighted area reveals three sets of six-fold symmetry spots(Fig. 3c), corresponding to relative rotations of ∼5◦ and 30◦ incrystal orientation among the three grains. The location of thegrains and their boundaries can be identified from the false-coloured dark-field TEM image presented in Fig. 3d.

Two possible mechanisms of boundary formation in MoS2were identified. In some instances, traditional grain boundariesconsisting of chemical bonds between the two single-layered grainsare formed and the in-plane growth stops (Figs 3d and 4a). Attimes, the growth could proceed with the nucleation of a secondlayer at the boundaries (Supplementary Figs S11a, S12 and Fig. 3d),as seen in the pale blue stripe along the grain boundary betweenthe green and blue grains shown in Fig. 3d, indicated by the whitearrow. This is a common feature, visible even in optical images,that can be used to distinguish the grain boundaries. The othermode of boundary formation occurs when one grain continuesto grow on top of the other without forming chemical bondsbetween the two grains, resulting in an overlapped bilayered region(Supplementary Fig. S11b and Fig. 4g–i). The mechanisms of grainboundary formation can be explored by taking a closer look at thestructure of the grain junctions inMoS2 atomic layers.

A combination of aberration-corrected STEM-ADF imaging andfirst-principles calculations based on density functional theory wasapplied to provide an in-depth analysis of the atomic configurationat grain boundaries in single-layer MoS2. It has been well-documented that grain boundaries in graphene can be describedas arrays of dislocations formed by 5|7-fold carbon rings36–38. How-ever, little is known about the detailed grain boundary structures inbinary systems such as h-BNandMoS2 (refs 39,40). Identification ofthe detailed atomic structure of dislocations and grain boundariesin h-BN and MoS2 has been primarily a theoretical investigation41.Figure 4a shows an atomically-resolved STEM-ADF image from agrain boundary with ∼21◦ tilt angle in false colours. The darkeratomic columns in Fig. 4a correspond to two S atoms on top ofeach other, while the brighter spots are single Mo atoms. A detailedanalysis of the STEM-ADF image indicates that all dislocationsat this observed grain boundary have the shortest Burgers vector(1, 0), following the same notation as Burgers vectors in graphene35(Supplementary Fig. S13), which strongly suggests that removingor inserting a semi-infinite stripe of atoms along the armchairdirection can form these dislocations. A closer look at the regions





NATURE MATERIALS DOI: 10.1038/NMAT3673 ARTICLES

2 µm 0.5 µm 200 nm

a b c d

Figure 3 | Grains in single-layered MoS2 films. a, Dark-field TEM image acquired from the continuous MoS2 film using the same diffraction spot (collagedfrom∼159 individual dark-field images). Most of the 15× 15 µm2 imaging area is covered by grains with the same crystal orientation and demonstrates thelarge single-crystal grain size of the MoS2 film. b, Bright-field TEM image of the MoS2 atomic layer containing a junction of three different grains. c, SAEDpattern acquired from the highlighted area in b, showing three sets of six-fold symmetry diffraction patterns with relative rotations of∼5◦ and 30◦.d, False-colour dark-field TEM image using the three diffraction spots marked in c, showing the presence of three grains.

1 nm

200 nm 2 nm 2 nm

1

2 3

Mo

S

4 5 15°c

ba

g h i

b

c

d e

f

Figure 4 | Line defects in the single-layered MoS2 films. a, STEM-ADF image of a MoS2 grain boundary with its common dislocations highlighted.b,c, Close-ups of the regions highlighted in a. The purple circles are Mo and yellow ones are S2 columns. d, Schematic of dislocations observed in MoS2

grain boundaries. e,f, Schematic of the Mo-oriented 5|7-dislocation core and its S2 substituted counterpart corresponding to b,c, respectively. The yellow,orange, and purple spheres represent top and bottom S and Mo, respectively. g, False-colour dark-field TEM image showing the presence of two adjacentgrains. The white solid line marks the grain boundaries, while the yellow dashed line marks the overlapped grain junction. h, STEM-ADF image of anoverlapped junction between two grains. Inset: FFT of the image showing two sets of MoS2 diffraction spots. i, False-coloured FFT-filtered image of hshowing the distribution of the two grains in cyan and yellow.

highlighted by the light blue and red rectangles in Fig. 4a reveals theatomic configurations of the dislocation cores that are commonlyobserved at the MoS2 grain boundaries. The Mo-oriented dislo-cations consist of 5|7-fold rings (Fig. 4b) and their variations, withsulphur substitutions at one of the sharingMo sites (Fig. 4c).

Basic dislocation theory42 combined with density functionaltheory calculations were employed to understand the atomic struc-ture and formation energy of the pristine and sulphur-substituted5|7-fold rings (Supplementary Information). Owing to the uniquetriple-layered structure ofMoS2, the dislocation cores extend three-dimensionally, with two layers of 5|7-fold rings joining at Mo sites

in the middle layer (Fig. 4d–f). The rings comprise a Mo-orienteddislocation core structure formed by fifteen atoms, with two Moatoms constituting aMo–Mo bond shared by the 5- and 7-fold rings(Fig. 4e). In eachMo-oriented 5|7-dislocation, there are five distinctMo sites that could be substituted by S atoms (Fig. 4e). At Mo sites1, 4 and 5, six S–S homo-elemental bonds are introduced after S2substitution (that is, two S atoms replacing one Mo), whereas thesharedMo–Mobond in themiddle remains unchanged. In contrast,for S2 substitution at the middleMo sites 2 and 3, theMo–Mo bondis replaced by two Mo-S bonds, and only four S-S bonds are gen-erated, making this type of S2 substitution energetically favourable





ARTICLES NATURE MATERIALS DOI: 10.1038/NMAT3673

4,000

107

Single-layered

Few-layered

106

On/

off r

atio

105

0.1 1 10

2,000I ds (

nA)

I ds (

nA)

0

¬150 ¬100 ¬50 0Vbg (V) Mobility (cm2 V¬1 s¬1)

50 100

4.27 cm2 V¬1 s¬1

150

100100 V40 V0 V¬60 V

50

0

¬50

¬100

¬0.6¬0.4 ¬0.2 0.0

Vds (V)

0.2 0.4 0.6

a b

Figure 5 | Electrical properties of MoS2 films. a, Room-temperature characteristics of the FET devices with 5 V applied bias voltage. Inset: draincurrent–voltage curves acquired for back gate voltages Vbg of−60, 0, 40 and 100 V representing a good ohmic contact. b, A summary of the mobility andon/off ratio measurements on MoS2 FET devices, demonstrating a mobility range of 0.2–18 cm2 V−1 s−1 and a maximum on/off current ratio of 6× 106.

(Fig. 4f). In sulphur-rich conditions, the formation energies for theS2 substitution at Mo sites 1–5 are 5.2, −0.8,−0.9, 1.0, and 1.5 eV,respectively. With the lowest negative formation energy, the S2 sub-stitution at Mo site 3 is the most stable among the various possiblestructures, consistentwith our experimental observations (Fig. 4c).

As well as the aforementioned traditional grain boundaries,overlapped junctions may also form when two MoS2 grains merge.Figure 4g shows a false-colour dark-field TEM image from twograins with 30◦ rotation. As a result of the formation of in-planecovalent bonds, the red and green grains form grain boundaries inthe upper part, whereas in the lower part of the junction, the twograins continue to grow on top of each other, forming a bilayeredoverlapped region at the junction. A closer look at such an over-lapped junction, Fig. 4h, exposes the distinct moiré pattern, which,combined with the two sets of six-fold symmetry diffraction spots(shown in the inset), corroborates the claim that only two grainsare present in this region. Fourier filtering each set of diffractionspots allows the spatial extension of the grains to be mapped out(Fig. 4i and Supplementary Fig. S14). The Fourier filteringmappingsuggests that the two layers continue to grow on top of each otherwithout forming any in-plane chemical bonds. Consequently, theinteraction between these grains at the overlapped junctions couldbe mediated by weak van der Waals forces. Moreover, Fig. 4gdemonstrates that the formation of conventional grain boundariesand the overlapped junctions are competing processes, but a cleardependence on the degree of lattice orientationmismatch in the twograins was not observed in our experiments.

To evaluate the electrical performance of the materials, wemeasure their field effect carrier mobilities. FET devices werefabricated by patterning the films using lithography and reactiveion etching (Supplementary Fig. S16). These devices have a channellength and width of 100 µm and 10 µm, respectively. All devicesdemonstrated FET characteristics of n-type semiconductors, similarto mechanically-exfoliated samples (Fig. 5a; ref. 11). Mobilitymeasurements obtained under good ohmic contact (Fig. 5a andinset) show an average of 4.3± 0.8 cm2 V−1 s−1 for the multipledevices, with film thicknesses ranging from single to multiple layers(Fig. 5b), which is within the range of reported experimental valuesformechanically-exfoliated samples. Themeasured on/off ratios forthe devices reach a maximum value of 6×106 for gate voltages inthe range−150 to 150V and a source-drain bias of 5 V, comparableto previous measurements11.

In conclusion, our results demonstrate the vapour phase growthof MoS2 atomic layers by nucleation, growth, and grain boundaryformation of single-crystalline domains. A straightforward method

for the controlled nucleation of molybdenum disulphide andlarge-area synthesis of films was demonstrated. The coalescence ofthe grains leads to the formation of grain boundaries composed ofarrays of dislocations of pristine and sulphur-substituted 5|7-foldrings. Alternatively, some grains join by simply growing ontop of each other, without forming any chemical bonds. Thehigh quality of the MoS2 atomic layers, exemplified by theirelectrical performance, presents a significant step towards scalablepreparation of this material.

MethodsTo synthesize MoS2 atomic layers in a vapour phase deposition process,MoO3 and pure sulphur were applied as precursor and reactant materials,respectively. Insulating SiO2/Si substrates were used, and the experiments wereperformed at a reaction temperature of 850 ◦C. More details are described in theSupplementary Information.

A scanning electron microscope (FEI Quanta 400) was used to study themorphology and distribution of MoS2 triangles and films on silica substrates.Raman spectroscopy performed on a Renishaw inVia microscope characterizedthe structure and thickness of the films at 514.5 nm laser excitation. Filmthicknesses and topographical variations in the samples were measured using AFM(Agilent PicoScan 5500).

TEM was employed to study the morphology, crystal structure, and defectsof the MoS2 samples. The TEM experiments were performed on a JEOL 2100Fmicroscope operating at 200 kV, and a FEI Titan 80–300 TEM operating at60 kV. The diffraction patterns of individual MoS2 triangles were obtained fromthe whole triangle using a selected-area aperture. For the large-size dark-fieldTEM image of the MoS2 film (Supplementary Fig. S10), 507 dark-field TEMimages were taken at the same diffraction spot under the same microscopesettings, with a spacing of 3 µm for each frame, under the dark-field imagingmode. The images were then aligned and stitched by Microsoft Image CompositeEditor. To recheck every composite, BF TEM images at low magnification wereacquired (Supplementary Fig. S10a). False-colour dark-field TEM images wereobtained by red, green and blue image construction of individual dark-field imagesacquired at the dark-field mode by selecting the spots at each pattern set with asmall objective aperture.

STEM imaging and spectroscopy analysis were performed on anaberration-corrected Nion UltraSTEM-100 operating at 60 kV (ref. 43). Theconvergence semi-angle for the incident probe was 31 mrad. ADF images weregathered for a half-angle range of ∼86–200mrad. Electron energy loss spectrawere collected using a Gatan Enfina spectrometer, with an electron energy lossspectroscopy collection semi-angle of 48 mrad. The ADF images presented in thismanuscript are low-pass filtered to reduce the random noise in the images. TheADF image shown in Fig. 4a was prepared by partially filtering out the direct spot inthe FFT of the image. This filtering process helps to remove the contrast generatedby the surface contamination (Supplementary Fig. S15).

TEM samples were prepared using a standard PMMA-based transfer, wherethe samples were spin-coated and submerged in basic solutions such as KOH.Subsequently, acetone and DI water were used to clean the TEM grid, allowing forthe suspension of the films in the solution and the transfer to TEM grids. We notethat this method tends to damage the MoS2 film by creating small holes, as seen inFig. 3b. Samples for STEM analysis were prepared by placing the as-grown MoS2





NATURE MATERIALS DOI: 10.1038/NMAT3673 ARTICLESfilms onto lacey-carbon TEM grids, and immersing them into 2% HF solution for20 s.Most of theMoS2 films and triangles were then transferred onto the grid.

XPS (PHI Quantera) was performed using monochromatic aluminiumKα X-rays, and the obtained data was analysed with the MultiPak software.FET devices were made by a photolithography process using photoresist S1813,mask aligner (SUSS Mask Aligner MJB4), and O2 reactive ion etching (RIE,Phantom III). The photoresist was removed by acetone and PG-REMOVER. Asimilar lithography process was used to prepare the photoelectric devices, with theaddition of the photoresist LOR5B as an adhesive layer to achieve an undercut.The electrodes (Au/Ti, 50 nm/3 nm) were deposited using the e-beam evaporator.Electrical measurements were performed in a Lakeshore probe station undervacuum conditions (10−5 torr), and FET and I–V measurements were carried outusing two Keithley 2400 source meters. We estimated the charge carrier mobilityin these devices using the equation µ= [dIds/dVbg]× [(L/(WCiVds)], wheredIds/dVbg is the slope, L andW are the channel length and width, respectively. Thecapacitance between the channel and the back gate per unit area is estimated to be∼1.2×10−4 Fm−2 (Ci = ε0εr/d , where ε0 is the permittivity of free space, εr = 3.9and d= 285 nm). The theoretical estimate for the capacitance of SiO2 will provideus with a lower limit estimate for the calculated mobilities. The photoelectricmeasurements were performed in a home-built electrical and environmentalnoise-reducing chamber, using a Stanford Research SR830 lock-in amplifiercombinedwith an optical chopper for signalmodulation and noise filtration.

Received 7October 2012; accepted 30 April 2012; published online9 June 2013

References1. Novoselov, K. S. et al. Electric field effect in atomically thin carbon films.

Science 306, 666–669 (2004).2. Novoselov, K. S. et al. Two-dimensional gas of massless Dirac fermions in

graphene. Nature 438, 197–200 (2005).3. Zhang, Y., Tan, Y. W., Stormer, H. L. & Kim, P. Experimental observation

of the quantum Hall effect and Berry’s phase in graphene. Nature 438,201–204 (2005).

4. Rycerz, A., Tworzydlo, J. & Beenakker, C. W. J. Valley filter and valley valve ingraphene. Nature Phys. 3, 172–175 (2007).

5. Li, X. et al. Large-area synthesis of high-quality and uniform graphene films oncopper foils. Science 324, 1312–1314 (2009).

6. Wang,H. et al. Controllable synthesis of submillimeter single-crystalmonolayergraphene domains on copper foils by suppressing nucleation. J. Am. Chem. Soc.134, 3627–3630 (2012).

7. Li, X. et al. Large-area graphene single crystals grown by low-pressurechemical vapor deposition of methane on copper. J. Am. Chem. Soc. 133,2816–2819 (2011).

8. Yu, Q. et al. Control and characterization of individual grains and grainboundaries in graphene grown by chemical vapour deposition. Nature Mater.10, 443–449 (2011).

9. Palacios, T. Graphene electronics: Thinking outside the silicon box. NatureNanotech. 6, 464–465 (2011).

10. Osada, M. & Sasaki, T. 2D inorganic nanosheets: Two-dimensional dielectricnanosheets: Novel nanoelectronics from nanocrystal building blocks. Adv.Mater. 24, 209–209 (2012).

11. Radisavljevic, B., Radenovic, A., Brivio, J., Giacometti, V. & Kis, A. Single-layerMoS2 transistors. Nature Nanotech. 6, 147–150 (2011).

12. Schwierz, F. Nanoelectronics: Flat transistors get off the ground. NatureNanotech. 6, 135–136 (2011).

13. Mak, K. F., Lee, C., Hone, J., Shan, J. & Heinz, T. F. Atomically thin MoS2: Anew direct-gap semiconductor. Phys. Rev. Lett. 105, 136805 (2010).

14. Splendiani, A. et al. Emerging photoluminescence in monolayer MoS2.Nano Lett. 10, 1271–1275 (2010).

15. Mak, K. F., He, K., Shan, J. & Heinz, T. F. Control of valley polarization inmonolayer MoS2 by optical helicity. Nature Nanotech. 7, 494–498 (2012).

16. Yin, Z. et al. Single-layer MoS2 phototransistors. ACS Nano 6, 74–80 (2011).17. Ghatak, S., Pal, A. N. & Ghosh, A. Nature of electronic states in atomically thin

MoS2 field-effect transistors. ACS Nano 5, 7707–7712 (2011).18. Zeng, Z. et al. Single-layer semiconducting nanosheets: High-yield preparation

and device fabrication. Angew Chem. Int. Ed. 50, 11093–11097 (2011).19. Coleman, J. N. et al. Two-dimensional nanosheets produced by liquid

exfoliation of layered materials. Science 331, 568–571 (2011).20. Balendhran, S. et al. Atomically thin layers of MoS2 via a two step thermal

evaporation-exfoliation method. Nanoscale 4, 461–466 (2012).21. Lin, Y-C. et al. Wafer-scale MoS2 thin layers prepared by MoO3 sulfurization.

Nanoscale 4, 6637–6641 (2012).22. Zhan, Y., Liu, Z., Najmaei, S., Ajayan, P. M. & Lou, J. Large-area vapor-phase

growth and characterization of MoS2 atomic layers on a SiO2 substrate. Small8, 966–971 (2012).

23. Liu, K. K. et al. Growth of large-area and highly crystalline MoS2 thin layers oninsulating substrates. Nano Lett. 12, 1538–1544 (2012).

24. Lee, Y. H. et al. Synthesis of large-area MoS2 atomic layers with chemical vapordeposition. Adv. Mater. 24, 2320–2325 (2012).

25. Margulis, L., Salitra, G., Tenne, R. & Talianker, M. Nested fullerene-likestructures. Nature 365, 113–114 (1993).

26. Feldman, Y., Wasserman, E., Srolovitz, D. J. & Tenne, R. High-rate, gas-phasegrowth of MoS2 nested inorganic fullerenes and nanotubes. Science 267,222–225 (1995).

27. Weber, T., Muijsers, J. C., van Wolput, J. H. M. C., Verhagen, C. P. J. &Niemantsverdriet, J. W. Basic reaction steps in the sulfidation of crystallineMoO3 to MoS2, as studied by X-ray photoelectron and infrared emissionspectroscopy. J. Phys. Chem. 100, 14144–14150 (1996).

28. Li, X. L. & Li, Y. D. Formation of MoS2 inorganic fullerenes (IFs) by thereaction of MoO3 nanobelts and S. Chem. Eur. J. 9, 2726–2731 (2003).

29. Li, Q., Newberg, J. T., Walter, E. C., Hemminger, J. C. & Penner, R. M.Polycrystalline molybdenum disulfide (2H-MoS2) nano- and microribbons byelectrochemical/chemical synthesis. Nano Lett. 4, 277–281 (2004).

30. Song, X. C., Zhao, Y., Zheng, Y. F. & Yang, E. Large-scale synthesis of MoS2bucky onions. Adv. Eng. Mater. 9, 96–98 (2007).

31. Van der Zande, A. M. et al. Grains and grain boundaries in highly crystallinemonolayer molybdenum disulfide. Nature Mater. 12, 554–561 (2013).

32. Han, G. H. et al. Influence of copper morphology in forming nucleation seedsfor graphene growth. Nano Lett. 11, 4144–4148 (2011).

33. Gao, J., Yip, J., Zhao, J., Yakobson, B. I. & Ding, F. Graphene nucleation ontransition metal surface: Structure transformation and role of the metal stepedge. J. Am. Chem. Soc. 133, 5009–5015 (2011).

34. Krivanek, O. L. et al. Atom-by-atom structural and chemical analysis byannular dark-field electron microscopy. Nature 464, 571–574 (2010).

35. Yazyev, O. V. & Louie, S. G. Electronic transport in polycrystalline graphene.Nature Mater. 9, 806–809 (2010).

36. Huang, P. Y. et al. Grains and grain boundaries in single-layer graphene atomicpatchwork quilts. Nature 469, 389–392 (2011).

37. Kim, K. et al. Grain boundary mapping in polycrystalline graphene. ACS Nano5, 2142–2146 (2011).

38. Yazyev, O. V. & Louie, S. G. Topological defects in graphene: Dislocations andgrain boundaries. Phys. Rev. B 81, 195420 (2010).

39. Liu, Y., Zou, X. & Yakobson, B. I. Dislocations and grain boundaries intwo-dimensional boron nitride. ACS Nano 6, 7053–7058 (2012).

40. Shi, Y. et al. van der Waals epitaxy of MoS2 layers using graphene as growthtemplates. Nano Lett. 12, 2784–2791 (2012).

41. Zou, X. L., Liu, Y. Y. & Yakobson, B. I. Predicting dislocations and grainboundaries in two-dimensional metal-disulfides from the first principles.Nano Lett. 13, 253–258 (2012).

42. Hirth, J. P. & Lothe, J. Theory of Dislocations 2nd edn (Wiley, 1982).43. Krivanek, O. L. et al. An electron microscope for the aberration-corrected era.

Ultramicroscopy 108, 179–195 (2008).

AcknowledgementsThis work was supported by the Welch Foundation grant C-1716, the NSF grantDMR-0928297, the US Army Research Office MURI grant W911NF-11-1-0362, the USOffice of Naval Research MURI grant N000014-09-1-1066, and the NanoelectronicsResearch Corporation contract S201006. This research was also supported in part bythe National Science Foundation through grant No. DMR-0938330 and a WignerFellowship through the Laboratory Directed Research and Development Programof Oak Ridge National Laboratory, managed by UT-Battelle, LLC, for the USDepartment of Energy (WZ); Oak Ridge National Laboratory’s Shared ResearchEquipment (ShaRE) User Program (JCI), which is sponsored by the Office of BasicEnergy Sciences, US Department of Energy. The computations were performed atthe Cyberinfrastructure for Computational Research funded by NSF under GrantCNS-0821727 and the Data Analysis and Visualization Cyberinfrastructure funded byNSF under Grant OCI-0959097.

Author contributionsJ.L., P.M.A., J-C.I. and B.I.Y. proposed and supervised the project. S.N. developed andperformed the synthesis experiments. S.N., Z.L., G.S. and S.L. performed the materialand device performance characterizations. Z.L. performed the dark-field-TEM imagingexperiments. W.Z. performed the STEM characterization and part of the dark-field-TEMimaging experiments. X.Z. performed the grain boundary calculations. All authors helpedin analysing and interpreting the data. S.N., Z.L., W.Z., X.Z. and J.L. wrote the paper andall authors discussed and revised the final manuscript.

Additional informationSupplementary information is available in the online version of the paper. Reprints andpermissions information is available online at www.nature.com/reprints. Correspondenceand requests for materials should be addressed to J.L.

Competing financial interestsThe authors declare no competing financial interests.





http://www.nature.com/reprints


Documents

Vapour phase growth and grain boundary structure of molybdenum disulphide atomic layers