6.1
Beam contamination on twisted bilayer MoS
2(a) DF-TEM overview of the
twisted bilayer MoS2 sample,
courtesy of the Philip Kim group.
(b) BF-LEEM overview of the same
flake, dominated by circular rings of carbon surface contamination.
Although domains in the twisted bilayer MoS2 sample were success- fully imaged in TEM (as shown in Figure 6.1a), the exposure to high en-
42 Few layer molybdenum disulfide
ergy electrons from the TEM beam had resulted in carbon surface contami- nation, hindering successful imaging of the flake in LEEM. Still large scale features and general structure remains recognisable from the TEM image.
Furthermore the LEED diffraction spots were too broad for successful dark field measurements on the twisted bilayer area of interest. In Figure 6.1b the LEEM image is shown and compared to the TEM image in Figure 6.1a.
6.2
Characterization of native bilayer MoS
2To characterize the second MoS2sample of a flake of varying thickness on hBN, both a bright field overview as well as IV-curves measurements were taken. The overview was taken at a landing energy of 11.4 eV, to have rea- sonable contrast between different layers and the hBN. The stitched result is shown in Figure 6.2.
Figure 6.2: Stitched overview of the MoS2flake on hBN. Indicated in red are the
µLEED areas used for the layer comparison in Figure 6.4 and in yellow the area
with different layer counts shown in Figures 6.3 and 6.5. Note that the contrast be-
tween the single layer and underlying hBN is low here.Inset:Optical microscope
image of the flake before transfer.
42
6.2 Characterization of native bilayer MoS2 43
Figure 6.3: Dark Field image of the area indicated in yellow in Figure 6.2 (ro-
tated90◦compared to there) The image
is taken in the tilted alignment at E0 =
20 eV. No new structures appear com- pared to BF images.
In Dark Field measurements of the flake no new, smaller scale structure was visible for any en- ergy up to 100 eV, an image with typical features is shown in Figure 6.3. The lack of additional struc- ture indicates there are no different stacking domains present, in turn indicating no stress or strain in this sample sufficient to form domains, as expected for a exfoliated native multilayer.
The lack of small scale struc- ture enabled the use of illumination apertures to perform µLEED mea- surements on different areas in or- der to further ‘fingerprint’ the ma- terial.
In Figure 6.4,µLEED IV-curves are shown of areas of different thicknesses as indicated in Figure 6.2.
0 20 40 60 energy (eV) 10−8 10−7 10−6 10−5 10−4 10−3 in te n si ty (a.u .) 0 20 40 60 energy (eV) 0 20 40 60 energy (eV) SPECULAR FIRSTORDERA FIRSTORDERB
Figure 6.4: Comparison of the intensity of the reflection µLEED spot and both equivalence classes of primary refracted spots for monolayer, bilayer and triple
layer areas of the MoS2 flake on hBN. Clear differences can be spotted for the
different layer counts. Data was measured in HDR mode.
We check the consistency of this data by calculating the lattice constant of the MoS2from it, using the formula derived at page 18. For the measured data, the diffraction spots appear atE=19.8 eV. Substituting this we have:
a= 2πh¯ 1 2 p (6me)19.8 eV =0.318 nm
44 Few layer molybdenum disulfide
Compared to the literature value of 0.316 nm at 380 K [32], this is in nice agreement, taking into account this data was taken at 680 K.
Of particular interest was an area where different layers of the MoS2 flake form areas of different thicknesses, indicated in yellow in the overview in Figure 6.2. Keeping in mind that this flake was created by exfoliation, the assumption is that different areas here originate from different layers of the bulk crystal, with its 2 layer unit cell. Indeed in Figure 6.5 different areas show different behaviour. The signal-to-noise ratio here is worse than in 6.4, due to the smaller illumination aperture used to confine the signal to a single area. 10−11 10−10 10−9 10−8 10−7 10−6 1L in te n si ty (a.u .) SPECULAR FIRSTORDERA FIRSTORDERB 0 20 40 60 energy (eV) A 10−11 10−10 10−9 10−8 10−7 10−6 2L in te n si ty (a.u .) 0 20 40 60 energy (eV) B
Figure 6.5: Comparison of the intensity of the reflection µLEED spot and both equivalence classes of primary refracted spots for different monolayer (top) and
bilayer (bottom) areas of the MoS2 flake on hBN. Areas and used diffraction spots
are indicated on the right. Data was measured in HDR mode. Clearly visible is the
inversion of the diffracted curves fromAtoBregions.
6.3
Discussion
When comparingµLEED curves for MoS2on hBN (See Figure 6.4), various features should be noted. Firstly for the primary spot on single layer, only two minima are visible up to 11 eV, whereas for a higher layer count an additional minimum at 4 eV appears.
Secondly small peaks in the minima at 12 eV and 19 eV seem to move through the minima to lower energies for increasing layer number.
Finally the symmetry between the six diffraction spots is broken, as has been observed before by Yeh et al. [33], where they argued this was due to 44
6.3 Discussion 45
penetration of the electrons to deeper layers of the crystal, while measuring on a monolayer (!). We argue instead that this threefold symmetry is due to the threefold symmetry of the bonding of both the molybdenum and sulfur atoms. This argument is supported by the curves from the squares region in Figure 6.5: monolayer areas stemming from different layers in the bulk show inverted equivalence classes for the different first order spots, corresponding to the different bonding directions in the two layers of the 2H unit cell.
As VLEED in this way is mostly susceptible to the surface layer, we can expect the IV-LEED curves for a bilayer (or even higher layercount) to most closely resemble the IV-LEED from a monolayer oriented in the same way as thetop layer. Indeed in Figure 6.5, the curves for the 2LA area closely resemble those of the 1LA area and the equivalence classes are roughly inverted compared to 1LB. From our assumption it would then follow that the monolayer in 1LA lays on top of the layer exposed at 1LB. Applying this way of reasoning to theµLEED data in Figures 6.4 and 6.5, we can now build up a picture of which layers from the bulk make up what areas on the flake, as shown in Figure 6.6.
1L(B) 2L(B) 3L 2LA 1LA
hBN
Figure 6.6:Schematic sideview of the layer structure of the different areas derived
from theµLEED data, where the areas from Figure 6.4 are indicated as 1L, 2L and
3L.
Worth noting is that the diffraction spots were measured at exactly equal positions, confirming exact inversion of layer orientation in this case. For superficially twisted bilayers this technique would however also enable de- termination of the twist angle.
Finally we determine whether the placement of MoS2 on a hBN flake was successful in increasing the flatness of the sample. As depicted in Fig- ure 6.7, the width of the specular spot was determined for µLEED mea- surements on a set of samples and different layer counts. This was done by fitting a Gaussian profile to a linecut of the spot perpendicular to the direction of the prism dispersion.
Firstly for a separate (exfoliated) MoS2on Si sample we measure a FWHM similar to what Yeh et al. find. As mentioned before, the peak width for the first twisted bilayer MoS2 on SiN sample was way worse with a FWHM
46 Few layer molybdenum disulfide −1.0 −0.5 0.0 0.5 1.0 k ( ˚A−1) 0.0 0.2 0.4 0.6 0.8 1.0 N o rm al ize d in te n si ty 2L on Si: FWHM=0.82 ˚A−1
2L on SiN (post TEM): FWHM=1.61 ˚A−1
1L on hBN: FWHM=0.15 ˚A−1
2L on hBN: FWHM=0.14 ˚A−1
3L on hBN: FWHM=0.13 ˚A−1
Figure 6.7: Profile of the specular spot (0,0) in diffraction for different substrates
and layer counts of MoS2atE0=32 eV
a factor of 2 larger. On hBN however the FWHM for the bilayer is al- most a factor 6 better compared to the Si substrate. In addition we observe nowhere near as much broadening of the diffraction peak for the mono- layer on hBN compared to what Yeh et al. observe on Si: we observe a factor of 1.16 broadening from the trilayer to the monolayer, where they had observed a broadening of a factor of 4. As such we can conclude that our strategy to use hBN as an atomically flat substrate for the MoS2 was successful.
46