LETTER pubs.acs.org/NanoLett
Ripples and Layers in Ultrathin MoS2 Membranes Jacopo Brivio,† Duncan T. L. Alexander,‡ and Andras Kis*,† †
Electrical Engineering Institute and ‡Interdisciplinary Center for Electron Microscopy (CIME), Ecole Polytechnique Federale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland
bS Supporting Information ABSTRACT: Single-layer molybdenum disulfide (MoS2) is a newly emerging two-dimensional semiconductor with a potentially wide range of applications in the fields of nanoelectronics and energy harvesting. The fact that it can be exfoliated down to single-layer thickness makes MoS2 interesting both for practical applications and for fundamental research, where the structure and crystalline order of ultrathin MoS2 will have a strong influence on electronic, mechanical, and other properties. Here, we report on the transmission electron microscopy study of suspended single- and few-layer MoS2 membranes with thicknesses previously determined using both optical identification and atomic force microscopy. Electron microscopy shows that monolayer MoS2 displays long-range crystalline order, although surface roughening has been observed with ripples which can reach 1 nm in height, just as in the case of graphene, implying that similar mechanisms are responsible for the stability of both two-dimensional materials. The observed ripples could explain the degradation of mobility in MoS2 due to exfoliation. We also find that symmetry breaking due to the reduction of the number of layers results in distinctive features in electron-beam diffraction patterns of singleand multilayer MoS2, which could be used as a method for identifying single layers using only electron microscopy. The isolation of suspended single-layer MoS2 membranes will improve our understanding of two-dimensional systems, their stability, and the interplay between their structures, morphologies, and electrical and mechanical properties. KEYWORDS: Two-dimensional materials, dichalcogenides, MoS2, transmission electron microscopy
T
he fact that a pure two-dimensional (2D) system is unlikely to exist was supported by theoretical investigations started by Peierls,1 Landau, and Lifshitz.2 They predicted that a pure 2D lattice could not exist at nonzero temperatures. The Mermin Wagner theorem3 on the other hand anticipated that 2D materials would not exhibit long-range order due to fluctuations. The isolation of graphene4,5 showed that due to 3D ripples6,7 or interaction with a substrate, a quasi-2D system could exist. After graphene, boron nitride (BN)8,9 was the next example of isolated, freestanding 2D films. Even though graphene was the groundbreaking 2D material, the synthesis, exfoliation, and study of other 2D materials, such as BN or molybdenum disulfide (MoS2), will allow a deeper understanding of the physics of 2D systems and a wider range of practical applications. Here, we report on the microscopic structure of monolayer (1 L) and few-layer MoS2. MoS2 consists of 2D S Mo S slabs stacked together with an interlayer distance of 6.15 Å. Because of the relatively weak van der Waals force acting between the layers, it can be chemically intercalated10 12 and exfoliated using micromechanical cleavage13 or liquid phase exfoliation12,14 down to 1 L thickness. Single-layer MoS2 is a direct gap semiconductor,15 18 with a band gap of 1.8 eV (ref 17) due to quantum confinement.18 The presence of a direct band gap in single-layer MoS2 makes it very interesting for applications in nanoelectronics,19 optoelectronics, and energy harvesting. Although MoS2 has already been characterized by electron microscopy with single-atom sensitivity,12,14,20 no evidence of r 2011 American Chemical Society
the ripples that are understood to stabilize the structure of graphene has been reported, so it is not clear what are the mechanisms behind the stability of single-layer MoS2. At this point, it is also not clear if there is a straightforward way for distinguishing between 1 L MoS2 and thicker samples without the use of aberration-corrected transmission electron microscopy (TEM). We make use of a simple method21,22 to transfer single- and few-layer MoS2 of known thickness between different substrates, allowing us to perform high-resolution transmission electron microscopy (HRTEM) and optical and atomic force microscopy (AFM) imaging on the same mesoscopic crystal of MoS2 with a precisely determined thickness. We start by performing mechanical exfoliation23 of MoS2 on silicon substrates covered with 270 nm thick SiO2,24 resulting in ultrathin MoS2 crystals, such as the one shown in Figure 1a. This allows us to easily find and distinguish among 1 L, 2 L, 3 L, and thicker flakes simply by quantifying the optical contrast with an optical microscope.24 AFM operated in tapping mode (amplitude modulated, repulsive regime) is used to verify flake thickness (Figure 1b,c). Once an interesting MoS2 crystal has been identified, we coat the sample with a poly(methyl methacrylate) film, release the film by etching the oxide layer, and transfer it together with the mesoscopic MoS2 Received: July 1, 2011 Revised: September 23, 2011 Published: October 19, 2011 5148
dx.doi.org/10.1021/nl2022288 | Nano Lett. 2011, 11, 5148–5153
Nano Letters
LETTER
Figure 1. Monolayer MoS2 transfer, suspended membranes, and microscopic characterization. (a) Optical image of a 1 L MoS2 flake exfoliated onto SiO2. Scale bar is 5μm. (b) AFM topographical image of single-layer MoS2 acquired on SiO2 before its transfer onto a Si3N4 membrane. Scale bar is 500 nm. (c) Height profile along the black line in (b) shows a thickness of 5.3 Å, indicating that the region highlighted in part (a) is indeed 1 L MoS2. (d) Low-magnification TEM image of the same flake as in (a) and (b). Scale bar, 2 μm. (e) HRTEM image of 1 L MoS2. The corresponding diffractogram is shown in the inset. Scale bar is 2 nm. (f) Filtered HRTEM image based on image shown in (e).
crystal onto a patterned silicon nitride (Si3N4) membrane25 27 suitable for HRTEM imaging, Figure 1d f. We chose 20 nm thick Si3N4 as the support because it offers the best compromise between mechanical robustness and electron beam transparency. The polymer film is dissolved by soaking in Nmethyl-2-pyrrolidone (NMP) prior to imaging. HR imaging performed using a standard field-emission TEM (JEOL 2200FS) operated at 200 kV shows that the flakes exhibit crystalline order and are overall clean with only small amorphous regions. We occasionally see folded structures, shown in Figure S3 in the Supporting Information, formed when a portion of a 1 L folds over itself, resulting in Moire patterns in HRTEM images. Such structures were excluded from further studies. We do not observe significant damage or amorphization on the time scale of our observations, with the MoS2 being stable enough for HRTEM characterization up to an acceleration voltage of 200 kV. Single and double layers (1 L and 2 L, respectively) and thicker flakes were first characterized using selected area (SAED) and nanoarea electron diffraction (NED) patterns. A typical diffraction pattern is shown in Figure 2b. All the diffraction patterns we obtained show the hexagonal symmetry characteristic of 2H MoS2, thus verifying that the stable configuration of mechanically exfoliated MoS2 is 2H,28 30 just as in the case of bulk crystals. No difference in the lattice parameters between 1 L, 2 L, and thicker flakes was found within the sensitivity of the technique (please refer to Supporting Information for more detail). However, we observe an important distinction between diffraction patterns acquired from single- and multilayer MoS2,
which is due to symmetry breaking and can be used to identify single layers using only TEM. It is noted that diffraction patterns with sufficient quality for this purpose can be acquired both over holes and over the thin Si3N4 membrane. Successive diffraction spots belonging to the same {1100} family show different intensities in the case of single-layer flakes, whereas in the case of two- and multilayer MoS2 there is no observable difference. Variations in the intensities can be highlighted by calculating the intensity ratio of neighboring diffraction spots belonging to the {1100} family as shown in Figure 2b and c. This qualitative difference between diffraction patterns from single and multilayer MoS2 is due to the different symmetry of the 1 L with respect to the bulk unit cell. In detail, going from the bulk to the 1 L, the cell loses its six-fold symmetry; this results in small variations in the intensities within the same family of lattice plane. Starting from the two different unit cells, we simulate diffraction patterns and indeed verify that the difference in the spot intensities is expected for 1 L MoS2. The intensity differences measured experimentally are in fact larger than those simulated (see Figure 2c). This could be due to possible differences in electron distributions between different atoms caused by covalent bonding. As this effect is based on symmetry reduction only, we believe that it can also be observed in other layered dichalcogenides, for example, NbSe2 or MoSe2. Tilting the sample with respect to the electron beam provides another facile method for distinguishing between 1 L and thicker flakes. By changing the beam incidence angle, a stronger variation in the relative spot intensities within the same lattice plane family {1100} is observed for 2 L and thicker flakes compared to 1 L 5149
dx.doi.org/10.1021/nl2022288 |Nano Lett. 2011, 11, 5148–5153
Nano Letters
LETTER
Figure 2. TEM diffraction patterns from 1 L and 2 L MoS2 (0° tilt) can be used to distinguish between them. (a) Low-magnification image of the thicker part of the MoS2 flake. Scale bar is 1 μm. (b) SAED pattern of the region highlighted in (a) with no tilt applied. The hexagonal symmetry is evident, and the two sets of planes {1100} and {2110} are highlighted. Scale bar is 2 nm 1. (c) and (d) Histograms which summarize the ratios within the {1100} family spots for diffractograms acquired on 1 L and 2 L MoS2 crystals, respectively. Simulated ratios are shown for comparison.
flakes (Figure 3). As the sample is tilted up to 15°, the diffraction pattern for a 1 L is practically unchanged (Figure 3c and d). The spot intensity of 2 L and thicker samples however shows a much stronger dependence on the tilting angle (Figure 3e and f). Eventually, there is an almost complete dimming of one diffraction spot with respect to the others for a tilting angle Q = 15°. A qualitative explanation for this phenomenon can be found if we consider the sample geometry. The confinement of a standard, thin TEM sample in one direction leads to a relaxation of the Bragg condition, which is represented by reciprocal lattice rods (relrods) associated with each reciprocal lattice point. The relrods are more extended for thinner samples than for thicker ones, making it more difficult to leave the Bragg condition when the sample is tilted. Just as in the case of graphene,6 we observe a blurring in the NED spots caused by variations in the tilt angle. The widening of diffraction peaks with increasing tilt angle is much stronger for 1 L MoS2 with respect to 2 L and thicker flakes and indicates that there is a morphology of sub-μm ripples6 in the flake. Since our flakes are only 1 2 layers thick, we exclude the possibility that these features correspond to dislocation networks typical for 10 100 nm thick MoS2 flakes,31 34 as such dislocation networks occur because of sliding between two subsequent layers. The fact that diffraction peaks can always be fitted with a Gaussian profile is compatible with the presence of isotropic corrugations instead of an overall flake bending, in which case the broadening would
show a preferential direction.6 The presence of an isotropic corrugation of lateral length L and height h modifies the reciprocal representation of the flake. Instead of the ideal relrods perpendicular to the plane of the reciprocal hexagonal lattice of a perfectly flat surface, the reciprocal representation of a corrugated 1 L MoS2 crystal consists of 3D cones with a certain angle centered on the ideal rods. The aperture of the cones can be estimated by plotting the broadening of the diffraction peaks with changing incidence angle, as shown in Figure 4a. For 1 L membranes we find cone angles between 7° and 13°, corresponding to an average deviation of the surface normal from its mean direction of (5°, similarly to the case of graphene.6 In the case of 2 L MoS2 the surface normal deviation is on the order of 0.5° (Figure 4b), whereas for 2 L graphene a deviation of 2° has previously been reported.6 For 1 L MoS2, the typical ratio between the lateral ripple size L and their height h is ≈10. The knowledge of L is thus necessary to quantify the ripple height. Thanks to the small NED probe size, the upper limit for L is fixed to 50 nm, while the lower limit is given by the electron coherence length. We utilized HRTEM to estimate L from contrast variation in images, such as the one shown in Figure 1e and f. We find that L ≈ 6 10 nm, which corresponds to a ripple height h ≈ 6 10 Å. This suggests that freestanding MoS2 has a similar roughness to freestanding graphene, whereas bilayer MoS2 is 5150
dx.doi.org/10.1021/nl2022288 |Nano Lett. 2011, 11, 5148–5153
Nano Letters
LETTER
Figure 3. TEM diffraction patterns of 1 L and 2 L MoS2 with applied tilt. (a) and (b) Schematic drawing of the relative orientation between the incident electron beam and the sample. Monolayer MoS2 flake is shown as seen from the side and along the [0001] top direction, tilted by 0° and 15°. Crystal planes originating some of the {1100} diffraction spots are highlighted in red. (c) SAED pattern and the intensity plot along the line connecting two diffraction spots indicated with arrows for 1 L MoS2 with no tilt. (d) Diffraction pattern for a 1 L with tilt angle Q = 15°. (e) and (f) SAED diffraction patterns for 2 L MoS2 flakes with and without the tilt. In the case of 1 L MoS2, increasing the tilt angle does not result in strong variations of the diffraction spot intensities within the {1100} family. The situation changes dramatically in the case of 2 L MoS2. The respective intensity profiles taken along one direction in the DPs clearly show the almost total dimming of one peak for 2 L flake compared to the 1 L. This behavior can be used to distinguish between 1 L and thicker MoS2. Scale bars are 2 nm 1.
significantly flatter than bilayer graphene. This could indicate the van der Waals interaction between the layers in MoS2 is stronger than in graphene. In conclusion, we have presented the first electron microscopy study of unambiguously identified 1 L MoS2 and thicker flakes of known thickness. We have studied the topography of mesoscopic MoS2 crystals by means of HRTEM, SAED, and NED. We observe ripples in 1 L MoS2, making it the second 2D material after graphene in which ripples have been observed. The height of the ripples which characterize the surface of 1 L MoS2 is 6 10 Å. Such ripples could have a detrimental effect on the electron mobility in 1 L MoS2 and removing them could result in reaching ultimate limits of mobility in this material. We have also demonstrated two methods to distinguish among 1 L and thicker MoS2 flakes based only on TEM analysis and diffraction. We note that these two methods are also likely to be useful for other layer dichalcogenide materials. As in the case of graphene,6 the possibility of being able to distinguish unambiguously a single-layer MoS2 flake inside a TEM instrument opens the
way for future experiments based on electron microscopy. The fabrication of suspended MoS2 membranes will also enable further optical and mechanical studies of this material. Together with graphene and few-layer BN, the isolation of single-layer MoS2 could improve our knowledge of 2D systems and expand their field of applications.
’ MATERIALS AND METHODS Single layers of MoS2 are exfoliated from commercially available crystals of molybdenite (SPI Supplies Brand moly disulfide) using the scotch tape micromechanical cleavage technique method pioneered for the production of graphene. AFM imaging is performed using the Asylum Research Cypher AFM. FEI CM300 and JEOL 2200FS instruments operated at 200 keV acceleration voltage were used for TEM imaging. A NiOx reference was used to verify diffraction pattern calibrations; samples were left inside the microscopes for at least 30 min prior to TEM analysis for thermal and mechanical stabilization. Diffraction pattern simulations were carried out using the JEMS software package.35 5151
dx.doi.org/10.1021/nl2022288 |Nano Lett. 2011, 11, 5148–5153
Nano Letters
LETTER
carried out in part in the EPFL Center for Micro/Nanotechnology (CMI). We thank K. Lister (CMI) for technical support with e-beam lithography. This project was financially supported by Swiss National Science Foundation grant no. 200021_132102 and the Swiss Nanoscience Institute (NCCR Nanoscience).
’ REFERENCES
Figure 4. Ripples in 1 L and 2 L MoS2. (a) Evolution of NED diffraction spots with tilt angle for 1 L MoS2. Values of the peak center and full width at half maxima (fwhm) in the reciprocal space are reported for various angles. Each angle corresponds to a dotted line. From the peak broadening with increasing tilt angle, we can calculate the cone angle formed by the relrods in the reciprocal space, which is linked to the corrugation through the ratio h/L. For 1 L MoS2, the cone angles are between 7° and 13°, corresponding to an average deviation of the surface normal with respect to its mean direction of (5°, similar to those observed in graphene. (b) Fwhm dependence for the {2110} family by the tilt angle, in the case of 1 L and 2 L flakes. Our data suggest that 2 L MoS2 is much flatter than 1 L MoS2, with cone angles being between 1° and 2° for bilayer MoS2. From HRTEM images it is possible to directly infer the lateral dimension L of the corrugations, which we find to be between 6 and 10 nm for 1 L MoS2 (see Supporting Information).
’ ASSOCIATED CONTENT
bS
Supporting Information. This material is available free of charge via the Internet at http://pubs.acs.org.
’ AUTHOR INFORMATION Corresponding Author
*E-mail: andras.kis@epfl.ch.
’ ACKNOWLEDGMENT TEM experiments were performed at the EPFL Interdisciplinary Center for Electron Microscopy (CIME). We thank P. Stadelmann (CIME) for useful discussions and JEMS software support as well as M. Cantoni (CIME) and C. Hebert (CIME) for numerous discussions and support. Substrate fabrication was
(1) Peierls, R. E. Ann. Inst. Henri Poincare 1935, 5, 177. (2) Lifshitz, E. M.; Pitaevskii, L. P. Statistical Physics Part I; Pergamon: Oxford, U.K., 1980; Chpts 137 and 138. (3) Mermin, N. D. Phys. Rev. 1968, 176 (1), 250. (4) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Zhang, Y.; Dubonos, S. V.; Grigorieva, I. V.; Firsov, A. A. Science 2004, 306 (5696), 666–669. (5) Novoselov, K. S.; Geim, A. K.; Morozov, S. V.; Jiang, D.; Katsnelson, M. I.; Grigorieva, I. V.; Dubonos, S. V.; Firsov, A. A. Nature 2005, 438 (7065), 197–200. (6) Meyer, J. C.; Geim, A. K.; Katsnelson, M. I.; Novoselov, K. S.; Booth, T. J.; Roth, S. Nature 2007, 446 (7131), 60–63. (7) Fasolino, A.; Los, J. H.; Katsnelson, M. I. Nat. Mater. 2007, 6 (11), 858–861. (8) Pacile, D.; Meyer, J. C.; Girit, C. O.; Zettl, A. Appl. Phys. Lett. 2008, 92 (13), 133107–3. (9) Meyer, J. C.; Chuvilin, A.; Algara-Siller, G.; Biskupek, J.; Kaiser, U. Nano Lett. 2009, 9 (7), 2683–2689. (10) Joensen, P.; Frindt, R. F.; Morrison, S. R. Mater. Res. Bull. 1986, 21 (4), 457–461. (11) Divigalpitiya, W. M. R.; Morrison, S. R.; Frindt, R. F. Thin Solid Films 1990, 186 (1), 177–192. (12) Ramakrishna Matte, H. S. S.; Gomathi, A.; Manna, A. K.; Late, D. J.; Datta, R.; Pati, S. K.; Rao, C. N. R. Angew. Chem. 2010, 122 (24), 4153–4156. (13) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (30), 10451–10453. (14) Coleman, J. N.; Lotya, M.; O’Neill, A.; Bergin, S. D.; King, P. J.; Khan, U.; Young, K.; Gaucher, A.; De, S.; Smith, R. J.; Shvets, I. V.; Arora, S. K.; Stanton, G.; Kim, H.-Y.; Lee, K.; Kim, G. T.; Duesberg, G. S.; Hallam, T.; Boland, J. J.; Wang, J. J.; Donegan, J. F.; Grunlan, J. C.; Moriarty, G.; Shmeliov, A.; Nicholls, R. J.; Perkins, J. M.; Grieveson, E. M.; Theuwissen, K.; McComb, D. W.; Nellist, P. D.; Nicolosi, V. Science 2011, 331 (6017), 568–571. (15) Lebegue, S.; Eriksson, O. Phys. Rev. B 2009, 79 (11), 115409. (16) Splendiani, A.; Sun, L.; Zhang, Y.; Li, T.; Kim, J.; Chim, C.-Y.; Galli, G.; Wang, F. Nano Lett. 2010, 10, 1271–1275. (17) Mak, K. F.; Lee, C.; Hone, J.; Shan, J.; Heinz, T. F. Phys. Rev. Lett. 2010, 105 (13), 136805. (18) Kuc, A.; Zibouche, N.; Heine, T. Condens. Matter 2011, 1104, 3670. (19) Radisavljevic, B.; Radenovic, A.; Brivio, J.; Giacometti, V.; Kis, A. Nat. Nanotechnol. 2011, 6 (3), 147–150. (20) Kisielowski, C.; Ramasse, Q.; Hansen, L.; Brorson, M.; Carlsson, A.; Molenbroek, A.; Topsøe, H.; Helveg, S. Angew. Chem., Int. Ed. 2010, 49 (15), 2708–2710. (21) Jiao, L.; Xian, X.; Wu, Z.; Zhang, J.; Liu, Z. Nano Lett. 2008, 9 (1), 205–209. (22) Dean, C. R.; Young, A. F.; Meric, I.; Lee, C.; Wang, L.; Sorgenfrei, S.; Watanabe, K.; T.; Kim, P.; Shepard, K. L.; Hone, J. Nat. Nanotechnol. 2010, 5 (10), 722–726. (23) Novoselov, K. S.; Jiang, D.; Schedin, F.; Booth, T. J.; Khotkevich, V. V.; Morozov, S. V.; Geim, A. K. Proc. Natl. Acad. Sci. U.S.A. 2005, 102 (30), 10451–10453. (24) Benameur, M. M.; et al. Nanotechnology 2011, 22 (12), 125706. (25) Kis, A.; Csanyi, G.; Salvetat, J.-P.; Lee, T.-N.; Couteau, E.; Kulik, A. J.; Benoit, W.; Brugger, J.; Forro, L. Nat. Mater. 2004, 3, 153–7. 5152
dx.doi.org/10.1021/nl2022288 |Nano Lett. 2011, 11, 5148–5153
Nano Letters
LETTER
(26) Yuzvinsky, T. D.; Mickelson, W.; Aloni, S.; Konsek, S. L.; Fennimore, A. M.; Begtrup, G. E.; Kis, A.; Regan, B. C.; Zettl, A. Appl. Phys. Lett. 2005, 87 (8), 083103. (27) Yuzvinsky, T. D.; Mickelson, W.; Aloni, S.; Begtrup, G. E.; Kis, A.; Zettl, A. Nano Lett. 2006, 6, 2718–2722. (28) Wypych, F.; Solenthaler, C.; Prins, R.; Weber, T. J. Solid State Chem. 1999, 144, 430–436. (29) Gordon, R. A.; Yang, D.; Crozier, E. D.; Jiang, D. T.; Frindt, R. F. Phys. Rev. B 2002, 65 (12), 125407. (30) Bissessur, R.; Kanatzidis, M. G.; Schindler, J. L.; Kannewurf, C. R. J. Chem. Soc., Chem. Commun. 1993, 20, 1582–1585. (31) Pashley, D. W.; Presland, A. E. B. Proc. Eur. Reg. Conf. Electron Microscopy, Delft 1960, 1, 417–422. (32) Kamiya, Y.; Ando, K.; Nonoyama, M.; Uyeda, R. J. Phys. Soc. Jpn. 1960, 15, 2025. (33) H€orl, E. M. J. Appl. Phys. 1965, 36 (1), 253–261. (34) Isshiki, T.; Nishio, K.; Saijo, H.; Shiojiri, M.; Yabuuchi, Y.; Takahashi, N. Microsc. Res. Tech. 1993, 25 (4), 325–334. (35) Stadelmann, P. A. Ultramicroscopy 1987, 21 (2), 131–145.
5153
dx.doi.org/10.1021/nl2022288 |Nano Lett. 2011, 11, 5148–5153
