Atomically Flat Zigzag Edges in Monolayer MoS2 by Thermal Annealing

Transcription

1 Atomically Flat Zigzag Edges in Monolayer MoS2 by Thermal Annealing Qu Chen 1, Huashan Li 2, Wenshuo Xu 1, Shanshan Wang 1, Hidetaka Sawada 1,3,4, Christopher S. Allen 1,4, Angus I. Kirkland 1,4, Jeffrey C. Grossman 2, Jamie H. Warner 1* 1 Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, United Kingdom 2 Department of Materials Science and Engineering, Massachusetts Institute of Technology, 77 Massachusetts Avenue, Cambridge, MA, 02139, USA. 3 JEOL Ltd., Musashino, Akishima, Tokyo , Japan 4 Electron Physical Sciences Imaging Center, Diamond Light Source Ltd, Didcot, Oxfordshire, *Jamie.warner@materials.ox.ac.uk; OX11 0DE, United Kingdom Supporting Information Experimental Methods Monolayer MoS2 Synthesis and Transfer. The MoS2 domains were grown on a SiO2 (300 nm)/si substrate (University Wafer), which was first sonicated in acetone and 2-propanol sequentially, followed by an oxygen plasma treatment before growth. The growth was carried out in a CVD system using 20 mg of MoO3 powder (99.5%, Sigma-Aldrich) and 500 mg of S (99.5%, Sigma-

2 Aldrich) as the precursor with Ar as carrier gas under atmospheric pressure. The S powder was placed at the central area of the first furnace in the outer 1 inch quartz tube, while the MoO3 was loaded separately upstream in the second furnace in an inner quartz tube having a smaller diameter of 1 cm. The growth system was first flushed with 500 sccm of Ar gas for 30 min, followed by a pre-introduction of S vapor by heating the S powder to 180 C for 10 min under an Ar flow rate of 150 sccm. The second furnace was maintained at 200 C at the same time to avoid any deposition of solid S on the substrate surface. This ensured that the reaction occurred under a S-rich atmosphere effectively controlling the initial MoS2 nucleation density. The second furnace was heated at a rate of 40 C min 1 to 800 C, while the MoO3 powder reached an approximate maximum temperature of 300 C. The reaction was conducted at 800 C for 20 min with 10 sccm Ar. After completion of the reaction a fast cooling process was applied to quickly stop the growth. Figure S1 shows an optical image of the as synthesized MoS2 which is confirmed as monolayer using Raman and photoluminescence spectroscopy. A thin film of PMMA was then spin-coated on the MoS2 surface after growth, followed by floatation on 1mol/L KOH to etch the SiO2. The PMMA/MoS2 film was then rinsed by transfer onto deionized water. The rinsed film was subsequently transferred onto a prefabricated in situ heating chip and dried in air for 3hours followed by baking at 180 C for 30 min to increase the adhesion between the MoS2 and the heating chip. The PMMA was then removed by submerging the heating chip in acetone for 12 hours.

3 Figure S1. CVD synthesized monolayer MoS 2. a) Optical image of monolayer MoS 2 domains on a Si chip showing uniform contrast. b) Raman spectrum of a domain area with peak at 381 cm -1 and A 1g peak at 401 cm -1. c) Photoluminescence (PL) at the same position with a significant peak at 670 nm. b) and c) are indicative of monolayer MoS 2. Transmission Electron Microscopy with an in Situ Heating Holder. AC-TEM was conducted using Oxford s JEOL JEM-2200MCO TEM operated at 80 kv accelerating voltage with a CEOS imaging aberration corrector. TEM data was recorded using a Gatan Ultrascan 4K 4K CCD camera with 2s acquisition times. ADF-STEM was conducted using an aberration corrected JEOL ARM300CF STEM equipped with a JEOL ETA corrector operated at an accelerating voltage of 60 kv located in the electron Physical Sciences Imaging Centre (epsic) at Diamond Light Source. Dwell times of 5 20 µs and a pixel size of nm px 1 were used for imaging with a convergence semi-angle of 31.5 mrad, a beam current of 44 pa, and inner-outer acquisition angles of mrad. High-temperature imaging up to 800 C was performed using a commercially available in situ heating holder from DENS Solutions (SH30-4M-FS). Heating the sample was achieved by passing a current through a platinum resistive coil imbedded in the TEM chip (DENS Solutions DENS-C-30). The resistance of the platinum coil was monitored in a four-point configuration,

4 and the temperature was calculated using the Callendar-Van Dusen equation (with calibration constants provided by the manufacturer). Slits were fabricated in the Si3N4 membranes using focused ion beam milling before transferring the MoS2. Image Processing and Simulation. ImageJ was used to process the AC-TEM and ADF images. For TEM images, a bandpass filter (between 100 and 1 pixels) and a Gaussian blur were carefully applied to minimize long-range uneven illumination and reduce noise. For ADF images only the brightness and contrast of the images was adjusted for visualisation. Multislice image simulations for ADF images were performed using the multislice method implemented in the JEMS software with supercells generated from DFT calculations. Parameters for image simulations were based on the experimental condition of the JEOL ARM300CF. The chromatic aberration at 60 kv is 0.89 mm with an energy spread of 0.42 ev. The probe size is 65 pm and the convergence semi angle is 31.5 mrad. The angle range for dark field imaging is from 49.5 mrad to 198 mrad. Spherical aberration is 5 µm. Density Functional Theory. First-principles calculations within density-functional theory (DFT) were employed to investigate the electronic structure of our prototype systems using the VASP (v5.4) package. 1 Plane-wave and projector-augmented-wave (PAW) type pseudopotentials 2 were implemented with the GGA-PBE exchange-correlation functional. 3 The kinetic-energy cutoff was set to 400 ev during geometric optimization, which was increased to 600 ev to ensure the convergence of electronic structures. The van der Waals interaction was accounted for by the DFT-D2 method of Grimme, 4 with a 50 Å cutoff radius for pair interactions. Dipole correction was applied to correct the leading errors introduced by the large dipole moment in finite periodic simulation box in the direction across the nanoribbon. The structures were relaxed until all forces were smaller than 0.01 ev/ Å. The Brillouin zone was sampled by a dense Monkhorst-Pack 5 k-point grid of A large spacing of

5 15 Å was constructed to avoid fake interactions in non-periodic directions. The energy penalty of the sub-processes describing the etching on edge is defined as = + Where and are the energies of initial and final states respectively, is the energy of a single Mo or S atom calculated using the same simulation box as those for nanoribbons. Since the ejection of a S atom is a one-step process without complicated intermediate states as confirmed by our testing simulations, may serve as an estimation of the average activation energy for ejecting the entire line of S atoms to vacuum as shown in Figure 3h-j. Alternatively, the barrier and minimum energy path for Mo atom diffusion (Figure 3i) were determined by the climbing image nudged elastic band (CI-NEB) method 6 with 5 images. The force tolerance for convergence was set to 0.03 ev/ Å, with an 8 5 rectangular supercell containing 121 atoms and sampling on a single Γ point.

6 Figure S2 Estimation of the states of hydrocarbon contaminates at different conditions. AC-TEM images taking at a) room temperature, b) 500 C, with electron beam irradiation (electron dose ~ 10 / / ) during the increase of the temperature. c) 800 C, with electron beam irradiation during the increase of the

7 temperature. d) 800 C, with the electron beam blocked until the temperature reached 800 C. Insets in a- d show the power spectrum from the FFT. e-h) Images corresponding to a-d), but using a mask of the FFT power spectrum to remove the contributions from the MoS 2. Masks are shown in the insets in e-h). i) Temperature profile and imaging points for a-c). j, k) Magnified images after smoothing and applying a false colour for the two boxes highlighted in f) showing the local crystalline graphene-like areas. l, m) Magnified images after smoothing and applying a false colour for the two boxes highlighted in g) showing the larger local crystalline graphene-like areas. Scale bars in a-d), 1 nm; j-k), 0.2 nm; l-m), 0.4 nm.

8 CVD grown MoS2 has to undergo a polymer-based wet-transfer process during the preparation of a TEM sample which will leave a certain amount of hydrocarbon residue on the sample surface. These hydrocarbons are amorphous and are difficult to observe directly using AC-TEM. The unavoidable amorphous hydrocarbon thin films correspond to the uniform halo shown in the power spectrum (inset in Figure S2a). After Fourier filtering to mask the reflections due to MoS2 the amorphous carbons on the surface can be directly observed (Figure S2b). Increasing the temperature under electron beam radiation induces crystallization of the amorphous carbons, as shown in the insets in Figure S1b and S1c, where the uniform halo of carbon becomes a sharp ring with a diameter of 0.21/nm, corresponding to the {10-10} crystal plane of graphene. The presence of the ring in the power spectrum calculated by the FFT indicates a polycrystalline structure within the carbon film (Figures S2j to S2m) and with increasing temperature the ordered structures expand. The polycrystalline carbon film remains intact even when holes form in the MoS2 lattice, as shown in Figure S3. During the propagation of these holes, the edge contours are irregular, with no atomically smooth edges. In order to eliminate the interference of the carbon film the electron beam was blanked until the temperature was stable at 800 C, as in the temperature profile shown in Figure 1f in the main text. Under this condition the surface of MoS2 did not show any form of carbon contamination (Figures S2d and h).

9 Figure S3 a-f) Propagation of a hole in MoS2 at 800 C with a carbon film covering the surface. The contour of the hole is highlighted by yellow dashed line. Scale bar in a-f) = 1 nm.

10 Figure S4 Series of TEM images showing (a) the creation and (b-i) the propagation of hole in MoS2 j) Schematic of the overlapping contours of the holes representing the evolution of the pore. Figure S3 shows the formation and propagation of a hole in the MoS2 lattice in the presence of a crystallized carbon film at elevated temperature (the situation in Figure S2c). The irregular shape of the edges is indicative of kinetic control of the dominant chemical etching process. Mo atoms tend to agglomerate into small clusters attached to edge sites. The formation and propagation of a hole in MoS2 with a clean surface (the situation in Figure S2d) is radically different, as shown in Figure S4. A small pore (Figure S4b) is first opened at the tip of a line defect (highlighted in pale blue in Figure S4a). The pore then grows larger through the removal of atoms from the edges. All edges generated during the hole

11 propagation lie along zigzag directions and are atomically sharp indicative of thermodynamic control. It is observed that the shapes of the pore during growth are either hexagons or truncated triangles with longer S-edges and shorter Mo-edges with the included angle between two adjacent edges always 120, i.e., two Mo-edges (S-edges) do not connect directly. By overlapping frames from the image series we are able to understand the propagation rate along each edge. Figure S5. Temperature = 800 C. AC-TEM images from MoS 2 with two S vacancies shown in a) (the left one is the same vacancy shown in Figure 1b). b) and c) show other areas with no visible S vacancies. The electron dose rate is ~ 10 / /. Scale bar in a-c), 2 nm. With a carbon-free surface, S vacancies generated by the electron beam are rapidly filled as the S atoms are highly mobile with the migration energy of 2.33 ev and not trapped by contaminants. Therefore, S vacancies are rarely observed due to the slow time resolution (ca. 10 s/frame) of the acquisition (Figure S5). Higher-time-resolution (ca. 2 s/frame) ADF image sequences (Movie 1) reveals the fast recovery process of the S vacancies.

12 Figure S6 ADF images showing a) a connection point of the two types of edges, b) a smooth Mo-edge and c) a smooth Mo-replaced S-edge. Scale bars in a-c) = 0.5 nm.

13 Figure S7 ADF images of an S edge a) before and b) after reconstruction. c-g) illustration of a possible reconstruction route calculated from DFT. Red circles represent S atoms in the subsequent frame. Values are in ev/s atom. Scale bars in a) and b) = 0.5 nm. The intermediate state before the formation of reconstructed S edge is a Mo terminated Klein-like edge. S vacancies are highly mobile on the clean MoS2 lattice and rapidly migrate to the edge site to replace the outermost line of S atoms, resulting in reconstruction of the edge. This transition is extremely fast and the intermediate Klein edge state is rarely observed (Figure S7a is the only example we have captured in our experiments). Figures S7c-f) illustrate a plausible route for the reconstruction. The intermediate state (Figure S7a) observed is the model shown in Figure S7e and

14 Figure S7g corresponds to the reconstructed S edge (Figures 2b and Figure S7b). The values shown in the

15 figure indicates the energy required for ejection of the marked S atoms. A more energetically favourable path for the elimination of the outermost line of S atoms is from S vacancy migration, which only requires 2.33 ev per single S vacancy. Figure S8 MoS2 edges during the cooling process. a-c) TEM images confirming the stabilities of atomically flat edges during cooling. d) Temperature profile with beam conditions during cooling. e) ADF image of the same edge at e) 800 C and f) room temperature after cooling.

16 Atomically flat MoS 2 edges remain when the temperature drops as shown in Figure S8. However, carbon contaminants rapidly absorb back onto the MoS 2 surface during cooling and hence efforts were taken to minimize exposure to electron beam during cooling. Our observations suggest no obvious deterioration in the profile of the edges on cooling, demonstrating the stability of the smooth edges at low temperatures. Layer-by-layer peeling also occurs takes along Mo-replaced S edges, as demonstrated in Figure S9a to d. The nanocrystal at the left end of the edge grows simultaneously with the etching of the edge, as a result of the accumulation of Mo atoms. The edge surface shows migrating Mo atoms along the edge. Excess Mo atoms at the edge also lead to the formation of a MoS nanowire, as shown in Figure S9e to g.

17 Figure S9 In-situ observation and DFT simulations of the behaviour of a Mo-replaced S edge under electron beam irradiation. a-d) AC-TEM image sequence of a Mo-replaced S edge showing a transformation from atomically smooth structure to a step in 2 min. Yellow dashed lines in d) indicates Mo-replaced S edges and blue lines Mo-edges. e-g) Formation of a MoS nanowire from relocation of atoms from adjacent edge areas. Time duration: ca. 10 s. h) 3D atomic model for a MoS nanowire, with cyan and gold balls representing Mo atom and S atoms respectively. i-l) Etching process along the Moreplaced S edge simulated using DFT, with purple and yellow balls representing Mo and S atoms respectively. Red and blue circles highlight the S atoms missing in the sequential frames. All values shown are in unit of ev/s atom. Scale bars in a-g) = 1 nm. DFT calculations show that the etching process starts from the line-by-line elimination of S atoms closest to the edge, which results from either ejection of S atoms (energy for each step shown in Figures S9i to l) or more probably, migration of S vacancies from the MoS2 lattice. When three rows of Mo atoms connect (Figure S9k), the structure is unstable and leads to two

18 different structures; firstly the formation of 3D MoS nanowire (Figures S9e-g and l), secondly the migration of the outermost line of Mo atoms to the end of the edge, as shown in Figures S9ad, leaving a reconstructed S edge. This process repeats after the first layer of atoms are fully etched. Figure S10 DFT calculated MoS2 nanoribbon model and corresponding PDOS diagrams with different factor decompositions. a, b) Top and side view of a MoS 2 nanoribbon with one Mo edge (lower edge) and one f Mo-replaced S edge (upper edge). c-e) PDOS diagrams of the nanoribbon, for y>0 and y<0 corresponding to the spin-up and down states respectively. The electronic structure is sensitive to the edge element, coordination and configuration. A normal MoS2 ribbon (Figure S11a) provides higher magnetization (3.16 µb) than that for the ribbon structure experimentally observed (2.47 µb) due to the contribution from p orbitals at the S edge, while the mono-s edge (Figure S11b) has a moment of 1.99 µb.

19 Figure S11 Electronic and magnetic properties of MoS2 nanoribbons with two different edge structures. a) Top and side view of a ribbon with a Mo edge (lower) and a di-s edge (upper). b) Top and side view of a ribbon with a Mo edge (lower) and a mono-s edge (upper). c, e) Corresponding PDOS diagrams for the structure in a). d, f) Corresponding PDOS diagrams for the structure in a). Movie S1 S vacancy migration with S vacancies highlighted in yellow circles. Movie S2 Dynamics of etching along Mo edge. Movie S3 Layer-by-layer removal of Mo-replaced S edge. Movie S4 Narrowing and reconstruction of a MoS 2 nanoribbon.

20 References (1) Kresse, G.; Furthmüller, J. Phys. Rev. B 1996, 54 (16), (2) Blöchl, P. E. Phys. Rev. B 1994, 50 (24), (3) Perdew, J. P.; Burke, K.; Ernzerhof, M. Phys. Rev. Lett. 1996, 77 (18), (4) Grimme, S. J. Comput. Chem. 2006, 27 (16), (5) Pack, J. D.; Monkhorst, H. J. Phys. Rev. B 1977, 16 (4), (6) Henkelman, G.; Uberuaga, B. P.; Jónsson, H.; Henkelman, G. J. Chem. Phys. 2000, 113 (22),