Motion of Light Adatoms and Molecules on the Surface of Few-Layer Graphene

Transcription

1 Supporting information Motion of Light Adatoms and Molecules on the Surface of Few-Layer Graphene Franziska Schäffel 1,*, Mark Wilson 2, Jamie H. Warner 1 1 Department of Materials, University of Oxford, Parks Road, Oxford OX1 3PH, United Kingdom 2 Department of Chemistry, Physical and Theoretical Chemistry Laboratory, University of Oxford, South Parks Road, Oxford OX1 3QZ, United Kingdom * franziska.schaeffel@materials.ox.ac.uk S1

2 Details of molecular dynamics computer simulations Molecular dynamics computer simulations were performed using a Tersoff II (T2) potential model 1. A T2 potential is chosen as it retains a relatively simple functional form (and hence is computationally tractable) and reproduces the basic bulk structural and energetic properties of the diamond and graphite structures. In addition, the model accounts well for carbon nanotube stability 2. Whilst the T2 potential does not reproduce key defect energies (as determined, for example, by density-functional-based electronic structure calculations 3 ), it does provide a useful base from which to assess the general validity of a modelling approach and its relationship to the on-going experiments whilst retaining contact with earlier simulation studies. Graphene sheet configurations are obtained from simulations on a single sheet containing 960 atoms and corresponding to a density of 0.36 atoms Å 2 in a tetragonal cell (20 12 unit cells). The short range nature of the T2 potential means that, despite the periodic replication of the simulation cell in three dimensions, the graphene sheets do not interact with each other. Molecular dynamics simulations are performed (with the graphene sheet initially confined to the xy plane) at T = 300 K (maintained by Nosé-Hoover thermostats 4 ). Single configurations are extracted at time intervals of 100 ps and are used to initiate the adsorption simulations outlined below. For single atom adsorption events an atom is propelled towards the graphene sheet varying the initial atom coordinate (i.e. relative to the underlying graphene sheet), the atom velocity (kinetic energy) and the direction of motion (the velocity vector). The starting coordinate above the graphene sheet is chosen so as to be beyond the range of the T2 potential and so the projected atom is initially in a ballistic regime. The velocity vectors are chosen by generating unit vectors on a grid {v x, v y, v z } where {v x,v y } = {0, 1, 2, 9} and v z = {1, 2, 3 9}. The magnitude of the velocity is then determined to correspond to the required kinetic energy. For the adsorption events the thermostats are turned off to prevent heat flow from the (hot) projected atom to the thermostat heat sink. The 1 J. Tersoff, Phys. Rev. B, 37, 6991 (1988) 2 C. L. Bishop & M. Wilson, Mol. Phys., 106, (2008). 3 L. Tsetseris & S. T. Pantelides, Carbon, 47, 901 (2009); L. Li, S. Reich & J. Robertson, Phys. Rev. B, 72, (2005); O. V. Yazyev, I. Tavernelli, U. Rothlisberger & L. Heln, Phys. Rev. B, 75, (2009) 4 S. Nosé & M. L. Klein, Mol. Phys. 50, 1055 (1983); W. G. Hoover, Phys. Rev. A 31, 1695 (1985) S2

3 large number of atoms in the graphene sheet (960) means that the relative heat associated with the projectile atom has no significant effect on the temperature of the system as a whole which remains at around the equilibrated graphene sheet value of T = 300 K. Results are obtained by averaging over 10 projectile atom starting configurations for each graphene sheet and over a further 10 starting graphene sheet configurations. To investigate the possible molecular chain motion observed experimentally simulations are performed in which a four carbon atom chain is adsorbed to a single graphene sheet and then energized (reflecting the potential use of the electron beam). Energy is transferred to the chain by identifying the surface tether atom (the atom on the graphene sheet which is linked to the chain) and the chain atom furthest away from this point (defining a vector, r chain ). Angular velocities are then imparted on the chain atoms (about the tether atom) varying the thermal energies and angles of projection relative to the graphene surface. After several picoseconds of molecular dynamics the excess thermal energy of the chain (i.e. above that of the room temperature equilibrated graphene sheet) dissipates. At this point the potential tether points are identified and the chain is re-energized. Determination of the number of layers During long electron irradiation at a high beam current density of approximately 0.1 panm 2 the few layer graphene (FLG) sheet disintegrates. One layer at a time is selectively removed from the FLG flake opening up regions with holes (Figure S1) containing discrete step edges between the layers. In Figure S1 each step is labelled representing the respective layer number. The FLG flake discussed in this contribution consist of a total of six graphene layers. S3

4 Figure S1. FLG sheet consisting of six layers: After long electron irradiation (158 min) a hole has formed in the middle. Thus the number of layers can be determined. Determination of the number of atoms lost during electron exposure Figures S2a and S2b shows high resolution transmission electron microscope (HR-TEM) images of an area on the FLG flake which contains two clusters. From the modified images (Figures S2a and S2b ), where the contribution of the crystalline graphene lattice has been subtracted using a Fourier mask, it can easily be observed that the cluster size decreases with exposure time. Thorough analysis of the area A covered by these clusters confirms the removal of atoms from them. In Figure S3c the number of lost atoms N has been determined under the assumptions of a density ρ = 2 g/cm 3 and the height h of approx. 3 nm (9 times the van der Waals distance) using, with N A being Avogadro s number and M C the relative atomic mass of carbon. From these data the average number of atoms lost per seconds is determined to be ~ 8 atom/s which is a reasonable value for typical clusters covering a surface area of nm 2. S4

5 Figure S2. Determination of the number of atoms lost during electron irradiation: a, b) HR-TEM images showing two clusters mainly consisting of amorphous carbon, including labels in a). Image a) was acquired 149 min after the starting electron exposure of the FLG flake, image b) was taken only 3 min later, after a total irradiation time of 152 min. a, b ) Modified HR-TEM images of a) and b) where the contribution of the graphene lattice has been subtracted from the image. These images highlight the removal of carbon from the clusters with exposure time. c) Number of atoms lost from the clusters with respect to the exposure time. An average number of 8 atoms/s are lost. S5

6 Image processing procedures Figure S3 shows the different procedures of image processing applied during analysis of the HRTEM micrographs. Digitized HRTEM micrographs (A) are analyzed by fast Fourier transforms (FFT). The 2D FFT of a lattice fringe image forms a pattern similar to a diffraction pattern, i.e. the position of a spot in the FFT pattern indicates a lattice spacing. An inclusive mask is then applied to all the spot positions in the FFT and an inverse FFT (B) is calculated from this artificial reciprocal image. By simply adding and subtracting the unfiltered image (A) and the inverse FFT (B) a Fourier enhanced micrograph (A+B) or a micrograph without the contribution of the periodic graphene lattice (A-B) can be obtained. The software used for image processing is DigitalMicrograph. Figure S3. Procedures of image processing. S6

7 Time sequence: formation of an elongated molecule The time sequence shown in Figure S4b taken from the marked area in Figure S4a presents the formation of an elongated molecule, possibly a hydrocarbon chain. While the atoms marked white in Figure S4b exhibit a very faint contrast at the beginning of the sequence, their contrast gets more pronounced after 45 s. After 125 s the feature rotates into the plane of the FLG flakes and reveals the elongated morphology of the molecule. Figure S4. Formation of an elongated molecule: a) HRTEM image including FFT of the FLG flake; b) Time sequence of the area marked red in S4a. A strong contrast becomes apparent after 45 s. After 125 s an elongated feature is visible. S7

8 TEM simulations of NMP on graphene We have carried out TEM simulations of graphene with varying number of layers and placed an additional NMP molecule onto its surface. The simulations are summarized in Figure S5. The number of graphene layers increases from top to bottom from 1 to 6. The defocus is varied between -8 nm and 8 nm in 4 nm steps from left to right. A negative spherical aberration coefficient of C S = mm was used in these simulations. Figure S5. TEM image simulations of NMP on graphene. From top to bottom the number of graphene layers increases from 1 to 6. From left to right the defocus value is varied from -8 nm to 8 nm in 4 nm steps. The spherical aberration coefficient C S was mm. The scale bar is 1 nm. From imaging single layer graphene it is known that the use of a negative spherical aberration coefficient and a positive defocus gives rise to white atom contrast of the graphene layer on a black S8

9 background which is reproduced in our simulations (top row, right columns). It is noteworthy that under the same conditions the NMP molecule adds dark contrast features to the simulated image. At negative defocus the NMP molecule adds white contrast to a single layer graphene in black atom contrast (left columns). Interestingly, when increasing the number of graphene layers the contrast of the underlying graphene switches from white to black (or vice versa). However, the contrast of the NMP molecule remains the same. For an even number of graphene layers the graphene now shows black atom contrast at positive defocus (rows 2, 4 and 6). This contrast switching has also previously been reported and experimentally proven by Warner. 5 It is also worth noting that for an odd number of layers (rows 3 and 5) a triangular contrast pattern of the graphene layers is observed as previously reported. 5 While the contrast of the graphene undergoes severe changes with increasing layer number the appearance of the NMP molecule does not change; it remains visible as dark contrast at positive defocus (apart from a relative contrast weakening as the layer number increases). Additional analysis of time sequence presented in Figure 5 of the manuscript Figure S6 shows magnified and cropped views of the images shown in Figures 5c-5f together with a dot pattern matching the spot patterns in the TEM micrographs in the lower panel. Figure S6. Enlarged views of the TEM micrographs presented in the time sequence in Figure 5, together with dot patterns matching the spot pattern in the micrographs (lower panel). The three main contrast features are again highlighted as filled red circles and the clockwise rotation of the molecule is indicated in the figures. In Figure S6a a fourth feature is marked with an empty red 5 J. H. Warner, Nanotechnology 21, (2010) S9

10 circle. In the following two pictures in the time sequence this feature is no longer apparent (Figures S6b and S6c), however it reappears slightly elongated in Figure S6d. Some features to the bottom left of the molecule are marked with a black arrow in Figures S6a-S6c. These are most likely the remains of the nearby graphene edge that has been sputtered by the electron beam. In Figure S6d where the molecule has shifted noticeably on the FLG xy-plane (cp. Figure 5 in the manuscript) these features are not visible anymore. Now that the molecule is clearly not attached to the graphene edge anymore, this contrast pattern is used for comparison with TEM simulations. Altogether six contrast spots are apparent now and marked with filled and empty red circles. The clockwise rotation between the position of the spot pattern in Figures S6a and S6b is approximately 60. In Figure S6d the molecule has rotated clockwise for another 55. Figure S7 shows a summary of simulated TEM images of NMP as well as molecules based around an NMP molecule where extra groups have been added at various positions along the five-membered NMP ring structure. S10

11 Figure S7. Summary of TEM simulations of (row I) an NMP molecule and (rows II-XI) molecules based around NMP. Columns (A) and (B) show a side and a top view of the molecule, respectively. The color code is: grey carbon, white hydrogen, purple nitrogen and red oxygen. The atomic positions to which extra groups have been attached within the NMP molecule are marked in the NMP side view (row I, column B). (C) TEM simulation of a free molecule, i.e. without underlying graphene. (D, E) TEM simulations of the molecule on five-layer graphene with a C S of mm and mm, respectively. (F, G) TEM simulations of the molecule on five-layer graphene with (C S = mm) with graphene contrast subtracted. S11

12 Figure S7, row I shows atomic models and simulated TEM images of a 1-Methyl-2-pyrolidone (NMP) molecule. The contrast pattern used to match the dark contrast spots in the experimentally observed micrographs (cf. Figure S6a) has been overlaid onto the simulated image (column F). While the three distinct features, marked with full red circles, are in good agreement, the fourth feature in the simulated image appears a lot stronger and elongated as compared to the experimentally observed image. More TEM simulations have been carried in order to find a better match. The observed contrast of simulated NMP molecules with an extra methyl group at positions 1 (row II), 3 (row XI) and 5 (row X) is much like that of the pure NMP molecule. Attaching a methyl group in position 4 (row IV), however, seems to create an additional spot which also correlates well with the experimentally observed spot pattern rotated clockwise by approximately 160. The reason this extra spot does not appear in the aforementioned simulations (rows II, X, XI) is due to the arrangement of the additional group with respect to the underlying graphene. If the position of the methyl group coincides with a carbon atom position in the upper graphene layer it gives rise to stronger contrast. Figure S8 shows simulated TEM images that clarify this and demonstrates the existing limitations of determining the structure of light molecules on graphene. Figure S8: Simulated TEM micrographs of 1,4-Dimethyl-2-pyrolidone at different xy positions and rotations with respect to the underlying five-layer graphene. Row (I): atomic model with the top graphene layer highlighted in yellow. Row (II): simulated images obtained with C S = mm and graphene contrast subtracted. S12

13 Here the 1,4-Dimethyl-2-pyrolidone molecule that provided the extra contrast spot (Figure S7, row IV) has been translated and rotated with respect to the underlying graphene layers. Clearly, the observed simulated contrast strongly differs comparing the Figures between (A), (B), (C) and (D). In (A) the extra group coincides with the position of a carbon atom in the upper graphene layer giving rise to a strong contrast spot. In (B) and (D) the group lies above the position of a carbon atom in the second layer. The feature can still be made out in the simulated micrograph, however with decreased contrast. In (C) the molecule has been positioned so that the extra methyl group does not coincide with the graphene lattice. In this case the contrast spot vanishes from the simulated image. This also explains why attaching the methyl group to positions 1, 3 and 5 of the NMP molecule (cf. Figure S7, rows II, X, XI) did not result in the observation of extra spots in the simulated images. This however does not compromise the general idea of processing experimentally observed as well as simulated micrographs in this way, since groups that overlap with positions of carbon atoms remain visible. Some molecular information will however be lost in both cases which highlights the necessity for using strictly the same image processing protocols on both HRTEM and simulated images. Furthermore, the simulation study shows that the contrast of groups may also be weakened when tilting the molecule with respect to the underlying graphene. Comparing rows IV to VI of Figure S7, which show simulated images of the same molecule with different tilt angles, reveals a weakening of the contrast of the extra spot (green arrow) as well as the convergence of the contrast spots. Adding an ethyl group to position 4 results in broadening of the extra spot (Figure S7, rows VII and VIII). Making the added group even longer then provides a sixth spot that correlates well with the sixth spot in the experimentally observed contrast pattern. This 1-Methyl-4-propyl-2-pyrolidone molecule provides the best match of the experimentally observed contrast in this simulation (keeping of course in mind that contrast spots may disappear when not coinciding with the carbon positions in the graphene lattice as mentioned above). S13

14 Determination of the length and relative orientation of the molecules Figure S9 shows the area(s) of the graphene flake from which the time sequence in Figures 6 and 7 has been derived. The inset in Figure S9a shows the corresponding FFT with the armchair (ac) and zigzag (zz) direction marked in red and yellow, respectively. In Figure S9b the graphene contrast has been subtracted as previously discussed. The inset shows the first frame of the time sequence (Figure 6) in which both molecules appear to be oriented along an armchair directions (red lines). Figure S9. a) Unfiltered TEM micrograph and respective FFT with armchair (ac) and zigzag (zz) directions highlighted; b) Same TEM micrograph with the graphene contrast subtracted. The areas marked yellow and green highlight the respective areas in Figures 6 and 7 in the manuscript. The atom marked white acts as positional reference. The inset shows one frame of the time sequence with two molecules highlighted that align well with the armchair direction; c, d) Two micrographs from the time sequence in Figure 6. Some molecules are marked with numbers; e) Simulated TEM micrographs of cumulene chains consisting of four (top), five (middle) and six carbon atoms (bottom); f) Lengths of the molecules contrast in the experimentally obtained (1-6) and simulated (7-9) TEM micrographs. S14

15 Figures S9c and S9d show 2 frames from the time sequence. The length of the contrast of the molecules (numbered 1-6) has been determined and is provided in Figure S9f. For comparison, Figure S9e presents simulated TEM images of cumulene chains consisting of four (top), five (middle) and six (bottom) carbon atoms. The respective lengths (7-9) are also given in Figure S9f. From this analysis is can be derived that the chains consist of 4 to 5 carbon atoms. Movie S1. Frames of the time sequence in Figures 6 and 7 assembled into a movie to visualize the dynamic behavior of the molecules. S15