Dynamics of epitaxial graphene growth and adsorptions of cobalt 4

4 STM studies of Graphite, C-rich 63 and Graphene on 6H-SiC(0001) Three carbon surfaces i.e. highly oriented pyrolytic graphite (HOPG), C-rich (63x63)R30o/6H-SiC(0001) (hereafter 63) and graphene/6H-SiC(0001) are studied in this chapter. In-situ scanning tunneling microscopy (STM) is used as the main characterisation technique to elucidate the structure. The first section focuses on the study of HOPG under different sample bias which would be used as reference for comparison with epitaxial graphene. The transformation from 63 to graphene were investigated using adsorption of Cobalt (Co) acting as tracer to map the evolution of 63 and graphene terraces. The 63 surface, the precursor phase prior to graphene formation is found to be made up of regions which are more Si rich and also regions that are carbon rich. The formation of graphene is observed to begin from the step edges as Si desorption occurs and the growth process continues akin to a step-flow growth mode. Analysis of the atomic step heights at various stages of graphitisation shows that as the initial 63 surface converts to form graphene, three Si-C bilayers beneath collapses to regenerate an interface with also a 63 periodicity. Based on these observations, a structural mechanism for formation of single and multilayer graphene is proposed. In addition, kinetic analysis of the growth process reveals that the transformation occurs with an activation energy of 3.0 ± 0.4 eV, a value close to breaking a Si-C bond. STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) 4.1 (a) STM of highly oriented pyrolytic graphite (HOPG) (b) Fig. 4.1 (a) Wide scan XPS spectra and (b) high-resolution spectra of C 1s core-level recorded from both ex-situ and an in-situ prepared graphite. In (a), trace amount of oxygen (O 1s) are only found on ex-situ prepared surface. In (b), intensities between 285 and 286 eV (C-H related states) recorded from ex-situ sample are more pronounced than in-situ sample. This implies besides oxygen, ex-situ sample were also contaminated with traces of hydrocarbons. This section revisits the STM imaging (constant current mode) of graphite and serve as reference for similar STM studies of epitaxial graphene monolayer on 6HSiC(0001) in later section. We begin with using a more global scale technique i.e. XPS to compare the two surface preparation methods that were previously described in Section 3.2.1. Figures 4.1a and 4.1b show the wide scan (or survey scan) and the higher resolution scan of C 1s core-level for two graphite samples. One of them was prepared insitu (degassed in UHV at 600ºC) and the other was prepared ex-situ (top few layers removed using adhesive tape). The C 1s core-level for both surfaces are located at 284.4 eV which is the typical binding energy for sp2-type C-C bond of graphite. The asymmetric of C 1s from 285 eV onwards are due to final-state effect originated from the 94 Chapter semi-metallic properties of graphite [1,2]. Although the C 1s for both surfaces are having the same binding energy, the graphite sample prepared ex-situ is found to be contaminated with traces of physisorbed oxygen and hydrocarbons that are most likely introduced during sample loading. The presence of oxygen on ex-situ prepared graphite can be found in Fig. 4.1a where about atomic percentage (with respect to C 1s) of O 1s signal is detected. Besides oxygen, C 1s spectra in Fig. 4.1b also show that the ex-situ prepared graphite is also contaminated with traces of hydrocarbons where signals located between 285 and 286 eV are more intense than in-situ prepared sample. These additional intensities are associated with C-H (hydrocarbon)-related electronic states. Based on XPS results, it is clear that in-situ preparation method is a better way of preparing clean graphite. Hence only in-situ prepared graphite substrates are used for the following STM studies. (a) (b) 3.3 0.1Å atomic layer nm 50 nm 20.5 ± 0.1 Å atomic layers 200 nm 200 nm Fig. 4.2 STM images of (a) a clean graphite surface (2m x 2m). High resolution image (20nm x 20nm) in inset, which were recorded at sample bias, Vs= +0.02V and constant current of 2nA, shows the graphite surface is atomically flat; and (b) a clean surface with a step of 20.5 ± 0.1 Å or atomic layers height (3.35 Å x 6). Inset shows another surface with a mono-atomic step height. 95 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) Large-scale STM image in Fig. 4.2a shows the morphology of a clean graphite surface. The higher resolution image is provided as inset. As expected for this highly twodimensional material, the graphite surface is atomically flat with atomic corrugation of 0.14 Å. For our samples, terraces as wide as 1m x 1m without any steps are often observed. When steps are found, they are in multiples of the graphite interlayer spacing i.e. (3.35 x n) Å, where n is number of layer. Figure 4.2b shows two examples of these step heights i.e. n = and n = (inset). As seen from Fig. 4.2, the in-situ preparation technique provides an atomically clean surface where the graphite surface (terraces and steps) is free from adsorptions of contaminants. For atomically clean surface, it is fair to expect that atomic resolution STM imaging will reflect the honeycomb structure of graphite. Instead graphite is very often observed as hexagonal close-packed (hcp) structure with periodicity of 2.4 Å as shown in Fig. 4.3a [3,4,5]. Assigning this image to the ball-and-stick model of graphite in Fig. 4.3b seems to suggest a missing C atom at every alternate position. Therefore the image obtained is not true topographically. The highly asymmetric tunneling current between neighbouring atoms comes from dissimilarity of C atoms within a layer where the natural Bernal or AB stacking of graphite generates two non-equivalent sublattices namely β and α site. The β site has no neighbouring C atom directly above or below it and is responsible for the bright features seen in Fig. 4.3a while the α site has such neighbouring C atom from the adjacent layers [5,6] (see Appendix for structural details). These images show that albeit weak interaction from the second layer graphite (van der Waals type), the electronic effects manifest itself strongly in the STM imaging. It is also found that the imaging of graphite is very sensitive to tunneling conditions and tip modification where adjusting the tunneling conditions or change in tip states may sometimes reveal the 96 Chapter honeycomb structure of graphite as shown in Fig. 4.3c (the corresponding ball-and-stick model is shown in Fig. 4.3d). The ability to resolve honeycomb structure thus may not be sufficient as a guide to distinguish monolayer graphene from multilayer graphene. (b) (a)  2.4 Å  2.46 Å 2Å (c) (d)   2Å Fig. 4.3 Atomic resolution (2nm x 2nm) STM imaging of graphite obtained at sample bias, Vs (a) +0.09V and (c) +0.08V. The corresponding ball-and-stick model is provided in (b) and (d) respectively. The unit cell for graphite with lattice parameter 2.4 Å is drawn in every figure. Unlike carbon (C) atom at  site, C at  site has no neighbouring C directly underneath or above it. 97 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) 4.2 STM of C-rich (63x63)R30o/6H-SiC(0001) 4.2.1 (63x63)R30o surface STM image in Fig. 4.4 shows the morphologies of a clean (63x63)-R30o/6HSiC(0001) surface (hereafter 63). This surface was prepared at around 1130oC where extensive Si desorption has occurred. In Fig. 4.4a, the terraces are noticeably much smaller than graphite and they are probably inherited from the as-received wafer. The sizes of the terrace range from 100 nm to as wide as 500 nm. Higher magnification of this surface (area in the box) is shown in Fig. 4.4b. As seen this surface is clean with no contamination by foreign bodies. The periodic array of bright triangular-like features is arises from surface reconstruction of this 63 surface. (a) [1210 ] (b) [1 120 ] 200 nm 10 nm Fig. 4.4 (a) (1000nm x 1000nm) Filled-state STM image showing morphologies of a clean 63 surface with terraces ranging from 100nm to as wide as 500nm. The area in the box are recorded at higher magnification and presented in (b) where the atomic resolution reveals the surface is decorated with periodic array of triangular features. All STM images were acquired at sample bias, Vs= -1.5V. The boxed area in (b) are enlarged and presented in Fig. 4.5. 98 Chapter (a) 6x6 maxima 6x6 63 3 3 cluster [1210 ] nm [1 120 ] (b) 6. Å (c) 3 6x6 63 nm nm Fig. 4.5 (a) (30nm x 30nm) Filled states STM image of 63 surface obtained at -1.5V and 0.2 nA. The 63, 6x6 and 3 unit cells are drawn using various diamond unit cell as labelled. The regular arrays of triangular-like protrusions with 6x6 periodicity are named as 6x6 maxima. Each of this unit consists of a group of three clusters. Inset shows each cluster has a diameter of 6.1 Å. The four horizontal line drawn across the center of each 6x6 maxima show they are not orderly aligned. Besides 6x6 maxima, there are other individual clusters as observed in (a). Higher magnification in (b) shows that these clusters have a 3 periodicity with direction rotated 30o from 1x1. They are named as 3 clusters. The 6x6 array is more dominant than 63 as shown by self-correlation image in (c). 99 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) The boxed area in Fig. 4.4b is enlarged and presented in Fig. 4.5a for further discussion. As seen, the surface consists of periodic array of triangular-like protrusions with 6x6 periodicity (solid line diamond, 18.5 Å) aligned along the  1210  and  1120  directions. However as seen by the lines drawn across the center of these maxima (top right corner), they are not perfectly aligned. For ease of discussion, this feature will be named as 6x6 maxima. Each of the 6x6 maxima can be further resolved into three smaller atomic clusters that give rise to its triangular shape. The configuration of these three atomic clusters at times can be seen to be inverted as shown in Fig. 4.5a. The width of each of these atomic clusters is 6.1 ± 0.2 Å (inset). Interestingly this diameter is twice the length of the 1x1 unit cell of 6H-SiC(0001) i.e. 3.08 Å * 2. Besides 6x6 maxima, individual atomic clusters are seen scattering around the 6x6 maxima. By comparison, the 6x6 maxima appear brighter than these clusters. Higher resolution of these clusters is shown in Fig. 4.5b. As seen these clusters are aligned 30o from the 1x1 with local periodicity of 3x3. Besides the arrangement, the mean diameter of each 3 cluster i.e. 5.5 ± 0.2 Å is smaller than the 6x6 maxima and akin to length of 3 unit cell of SiC-1x1 i.e. 5.33 Å. As such, these atomic clusters will be named as 3 clusters. Selfcorrelation of this surface is shown in Fig. 4.5c where the 6x6 periodicity is more dominant than 63 (dotted line diamond). As presented in Chapter 1, flower-like [7] and honeycomb-like [8,9] surface has been separately reported for STM study of 63 phase. Initially it was thought that these different features are due to early and late stage of 63 formation. However, we found that these features are bias-dependant. Besides the flower-like pattern that we observed in Fig. 4.5, Fig. 4.6 shows the bias-dependant filled-states images of a 63 surface. The self- 100 Chapter correlation of each bias is shown as inset. As the sample bias, Vs increased to -1.8V, this tunneling condition seems to incline towards the electronic states of 6x6 maxima than the 3 clusters (Fig. 4.6a). As a result, the 6x6 maxima are now linked with each other, generating a honeycomb network while the 3 clusters become unresolved and they collectively appear as regular array of “holes” for the honeycomb. At higher bias (Fig. 4.6b), the 6x6 maxima now emerge less dominant. While the triangular shape of the 6x6 maxima is still noticeable, their brightness is almost similar as the 3 clusters. As a consequence, the regular array of 6x6 is almost concealed by the distribution of 3 clusters and the surface now appears amorphous-like. Nevertheless, the 6x6 periodicity is still prominent as shown by the self-correlation (inset). The implication is that this 63 surface is modulated by (i) 6x6 array of 6x6 maxima with each maxima consists of three atomic clusters and (ii) individual atomic clusters with local periodicity of 3. (a) -1.8V (b) -3.0V nm nm Fig. 4.6 Bias-dependant filled-states images of a 63 surface: (a) Vs = -1.8V and (b) Vs = -3.0V. In (a) the surface shows honeycomb structure and in (b) the surface looks amorphous-like. Tunneling current of 0.15 nA was used. Despite being a 63 under LEED [16,10], selfcorrelation of this surface for each bias (inset, 10nm x10nm) shows dominant 6x6 periodicity (small solid line diamond cell). The 63 periodicity (bigger dotted-line diamond cell) is also drawn for comparison. 101 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) (a) -1.15V, 0.15nA (b) -0.023V, 1.5nA 6.2 Å 12.6 Å 3.0 Å 14.4 Å nm (c) 6x6 nm (d) nm nm Fig. 4.7 Filled-state images (a) 1.15V and (b) 0.023V of a 63 surface. Insets (4nm x 4nm) highlight the 6x6 maxima under different tunneling bias. The 6x6 unit cell is represented by the diamond cell drawn in both images. Numbers in each image denotes the width of the features highlighted. As observed, the 6x6 maxima switch from three atomic clusters in (a) to groups of smaller protrusions in (b) as highlighted using arrows. The surfaces in (a) and (b) are enlarge in (c) and (d) respectively for clearer depiction. Insets in (c) and (d) show one unit of 6x6 maxima. As indicated by the crosses (x), the original states arise from the three atomic clusters in inset of (c) now becomes featureless as shown in inset of (d). Instead, ten bright protrusions (marked by dots), which are also arranged in triangular manner, are now emerged at different lateral position from the crosses. As measured in (b), each of these protrusions (~3 Å) is half the size of the initial atomic clusters (~6.1 Å) that appear together as trimer. 102 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) hexagon is reproduced in Fig. 4.23b using bulk SiC normal to [0001] as the structure. Due to the C3 symmetry, the six edges of the SiC are divided into two groups according to the number of dangling bonds per C atom, with the “fast” step (two dangling bonds) alternates with the “slow” step (single dangling bond). During graphene growth, the fast steps are anticipated to recede faster than the slow steps as reflected by the arrows. As a result of this preferential etching, the original hexagon-shaped terrace is reduced to an intermediate shape which akin to the remaining 63 terrace in Fig. 4.23a. The similarity between the two clearly shows that there are two types of step with different reactivity present on the 63 surface. The influence the step reactivity on graphitisation is seen earlier where the irregular 63 terraces becomes well-aligned when graphene growth starts (Fig. 4.18). 63 (a) (b) 63 7.0 5.2 2.2Å slow G C Si 63 2.5 fast 2.8 63 5.3 G G 100nm 2.5 7.5 63 2.6 63 6√3 G Fig. 4.23 (a) STM image (500nm x 500nm) of a partially graphitised surface. Arrows represent the direction of graphitisation; (b) First picture shows top view of fast and slow edges of bulk (1x1) due to hexagonal symmetry and the etching of these step edges at different rate generates intermediate structure similar to (a). 134 Chapter The abrupt disappearance of BL steps for 63|63 terraces on a partially graphitised and the re-emergence of BL steps heights when the surface is fully graphitised indicates that the 63 terraces, consisting of BL, collapses during formation of a monolayer graphene. More significantly, graphene that is formed is accompanied by the regeneration of a 63-like structure at the interface between the graphene and the bulk (Fig. 4.15). Via LEEM, Hannon et al. [31] observe BL of their 6H-SiC(0001) surface collapsed as 63 phase is formed. The implication therefore is that as the 63 surface converts to graphene, three BL directly underneath the graphene also collapse to support formation of a new 63-like layer at the interface. Quantitatively, carbon atoms from BL of SiC will give one monolayer of graphene. This indirectly implies that the 63 layer may contain sufficient carbon atoms to form graphene. Since the 63 surface, which was prepared between 1100oC to 1170oC, does not convert directly to graphene until the surface is annealed to 1200oC, we believe that there are significant amount of Si-C bonds still present on this surface i.e. from the 6x6 maxima on the 63 surface (see Section 4.2.3). 4.5.3 Mechanisms for graphitisation of 6H-SiC(0001) The main mechanism for graphitisation of 6H-SiC(0001) has been the collective collapse of BL due to the abundance of such half-unit cell steps on this surface (Fig. 4.21a). Mechanism leading to mono- and multilayer graphene will be proposed with the attempt to fit the morphologies and step heights observed to real atomic environment of SiC based on the following ideas: 135 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) (i) the 63 layer has sufficient carbon atoms to form graphene as observed in Section 4.5.3 and Hannon et al. [31]; (ii) desorption of Si is not kinetically limited at these temperatures, a Si-deficient or Crich 63 regenerated at the interface; and (iii) BL of SiC are needed and collapse whenever a 63 interface is formed. The cross section of a 63-terminated 6H-SiC is shown in Fig. 4.24a. Since the 63 structure is not fully resolved, the 63 layers are represented by a block for convenience. Underneath this 63 layer, perfect bulk structure of 6H-SiC is assumed with stacking sequence of CBABCA along the [0001]. Along the [0110] , one end may be terminated with single dangling bond or double dangling bonds depending on the stacking sequence. As observed, graphene growth starts from the step edges terminated with double dangling bond (or the “fast” step) as depicted in Fig. 4.24a. Two events take place i.e. (i) the Si-C bonds within 63 layers starts to break and allow this layer converts to graphene and (ii) the BL underneath the 63 layers collapses and form a new 63 layer at interface. The second-step process is accompanied by desorption of Si. The intrinsic structure of 6H-SiC, where the step terminated with double dangling bonds is extended over to BL, conveniently allow systematic formation of 63 and graphene. Both processes involve irreversible Si desorption and significant diffusion and rearrangement of C atoms at the step edges. 136 Chapter (a) 6√3 B 7.5Å (i) [0001] Si [0110] (ii) [2110] C 6√3 6√3 A C B A C (b) 6√3 Vacated Si position filled up by graphene 6√3 5.2Å 2.3Å 6√3 5.2Å 2.3Å 6√3 (c) Graphene 6√3 7.5Å 7.5Å Graphene 6√3 Graphene 6√3 Fig. 4.24 Mechanism of monolayer graphene growth that gives the 7.5 Å step height. See text for description of (i) and (ii). Orange and yellow balls represent Si atoms while dark grey balls represent C atoms. 137 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) The step heights of G|63 in Figs. 4.15, 4.21 to 4.23 suggest the graphene layer rests approximately 2.3 Å above the new 63 interface. By positioning the new graphene monolayer above the 63 interface according to this distance (see Fig. 4.12b), we found that this position coincides with the original Si position (from 1.89 + 0.45 = 2.34 Å, where 0.45 Å is vertical distance for a C-C bilayer) which are now vacant due to desorption. The 2.3 Å separation is 20% longer than the covalent Si-C distance (1.89 Å) which seems to suggest the graphene layer does not interact strongly with the underneath 63 interface. This in agreement with the recent photoemission [39] and STM [29] experiments that suggested metallic states of the first graphene are almost unperturbed by the underlying interface. Due to collapse of the underneath layers to form 63 interface, the graphene lies 5.2 Å (from 7.5 Å – 2.3 Å) lower than its initial 63 surface, which is observed for the step distribution of 63|G in Figs. 4.21b and Figs. 4.22 and 4.23. This height difference between the remaining 63 and its graphene make the partially graphitised surface significantly rougher than the initial 63 surface, as observed earlier in Fig. 4.18. The slow steps can also form graphene as shown on the left of Fig. 4.20b. It occurs in a similar manner as the fast steps except slower. Graphene growth continues until the 63 layer is completely exhausted as seen in Fig. 4.24c. Once graphene growth is completed, the BL step height i.e. 7.5 Å is recovered back. This step height is one of the dominant steps observed on fully graphitised surface in Fig. 4.21c. The roughness transition shown here is also consistent with STM observation in Fig. 4.18a-c. 138 Chapter [1210] 3.3 [1120] 7.5 7.5 3.3 w sl o sl o w 7.5 4.2 fast 3.3 4.2 fa s t 4.2 slow 100nm Fig. 4.25 Fully graphitised surface with “fast” and “slow” steps assigned according to step heights (numeric labels). See text for full description. Besides the 7.5 Å step, two unknown steps i.e. 3.3 and 4.2 Å are also observed on a fully graphitised surface as discussed earlier. STM images show that these two steps often come as a pair as seen in Fig. 4.25 but very often adjacent or opposite to a 7.5 Å step. The topmost surface in Fig. 4.25 with a hexagonal shape shows that the 3.3|4.2 Å pair is positioned 60o to a step with height of 7.5 Å and 120o to another 3.3|4.2 Å pair. These observations suggest that the formation of these steps height are related to C3 symmetry of SiC(0001). We attribute the 3.3 Å comes from second layer graphene formation. We believe most of the second layer graphene nucleate at the “fast” steps while the “slow” steps still converting to graphene. The mechanism is illustrated in Fig. 4.26 where continue from Fig. 4.24c, a 63 interface starts to convert to graphene while another BL underneath it collapse to form a new 63 interface. Due to formation of new layer graphene underneath the first graphene, a bilayer graphene structure is formed at this region as shown in Fig. 4.26b. This bilayer graphene assumes the graphite interlayer 139 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) spacing i.e. 3.3 Å. The 4.2 Å step height comes from 7.5 Å - 3.3 Å and explains why they often come as a pair. (a) Graphene Si 6√3 7.5Å 7.5Å Graphene Graphene 6√3 6√3 (b) 4.2Å Bilayer 6√3 7.5Å 3.3Å 6√3 6√3 Newly formed 63 interface Fig. 4.26 Mechanism for formation of bilayer graphene and step heights of 3.3 and 4.2 Å. The mechanism suggested above for monolayer and bilayer graphene can be extended to the third layer graphene in similar fashion. Since the third layer formation is not clear in our work, we adopt the result reported recently by Emtsev et al. [37]. In their Fig. 1i, depressions of Å and Å is observed at the step edges due to formation of second and third layer graphene. Adopting the same mechanism we suggested above, Fig. 140 Chapter 4.27 shows the origin of 4.2 and 8.4 Å step height associated with bi- and trilayer graphene formation. Due to significant collapse of the bulk below the surface to form a 63-like interface each time an additional graphene monolayer is formed, new step are created and the topmost graphene surface layer are seen “drapping” across the steps, forming a continuous film. This carpet-like structure has been observed by several other groups although no mechanism is put forward by them [40,41]. As seen in Fig. 4.26b, there is certain strain exerted on the graphene monolayer where the bonding of C atoms seemed stretched as it “draps” across the step. This may be one of the sources that contribute to the reduced carrier mobility of epitaxial graphene in comparison to exfoliated graphene [37]. The above proposed mechanism therefore explained the evolution of step heights observed as the surface evolved from 63 to partially graphitised and finally to graphene surface. 63 SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC Graphene Bilayer Trilayer 4.2Å SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC SiC 3.3Å SiC SiC SiC SiC SiC SiC 4.2Å SiC SiC SiC 3.3Å 4.2Å 3.3Å SiC SiC SiC SiC SiC SiC 3.3Å Fig. 4.27 Mechanism in Fig. 4.24 and 4.26 are extended for third layer graphene formation. 141 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) 4.5.4 Secondary mechanism of graphene growth (a) (b) G G 63 1.8 Å 2.5 Å 63 63 2.5 Å G 20nm 10 nm 20 nm (d) (c) 63 1200oC Si-C bond Si atoms breaking desorb – 1300oC G 63 Si 10 nm I II z (Å) C atoms diffuse & rearrange 0.7Å Graphene 2.3 Å 10 20 x (nm) 30 0.7 Å 63 40 IV III Fig. 4.28 (a)–(c) STM images showing some graphitisation characteristics: (a) (100nm x 100nm) Graphene (G) growth starts from step edges; (b) (60nm x 60nm) A incompletely graphitised 63 terrace as indicated by high density of Co clusters. Inset (120nm x 100nm) shows larger scale of this terrace; and (c) (50nm x 30nm) A incompletely graphitised 63 terrace without Co and the line profile (blue line) across this surface; (d) Proposed mechanism as 63 transforms to graphene. Yellow and grey represents C and Si layers respectively. They not represent the actual density of C and Si. Readers are referred to text for details. 142 Chapter The previous section established the main mechanism adopted by steps with BL thicknesses. Besides steps with BL height, other step heights i.e. BL and BL are also present on 63 surface although their amount are much less than BL step height (Fig. 4.21a). These step heights on occasion are also found to convert into graphene. Fig. 4.28a shows graphene growth commences from step edges of a 63 terrace with thickness of just BL (2.5 Å). This observation seem to support the idea put forward by Hannon et al. [31] where the 63 phase has sufficient carbon to support graphene growth provided that the surface is annealed at correct temperature regime to remove the Si clusters (6x6 maxima). The BL step height is also seen to shrink by 0.7 Å after graphene formed at the edges. Fig. 4.28b-c shows another characteristic of graphene growth where parts of the 63 terrace are not completely graphitised. One of the possible reasons for this occurring is possibly due to insufficient of C at this region towards the late stage of graphitisation. Due to the propagation direction of the graphene growth from step edges, the unconverted 63 often appear as long and narrow streaks (measuring 10-20 nm width) that are perpendicular to the step edges. Again the vertical distance between unconverted 63 and graphene is 0.7 Å as shown in Fig. 4.24c. The same vertical height difference can be measured between the Si-rich 6x6 maxima and the C-rich 3 clusters (see line profile in Fig. 4.8d). This measurement implies that once the Si-rich 6x6 maxima are removed at above 1200oC, the C-rich 3 clusters converts to graphene. Based on the observations above, a simplified schematic for graphene growth involving BL 63 terrace is proposed in Fig. 4.28d where Si-C bonds starts to break 143 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) from the edge of the 63 surface (step I and II). At the same time Si underneath the 63 also starts to desorb (hence more Si-C bond breaks) as the C layer on top of it is being released to grow graphene. This in turn generate a new 63 phase beneath the graphene due to extensive loss of Si (step III). As the C-rich 3 clusters are located 0.7 Å below the 6x6 maxima, the graphene generated from these clusters is predicted to have similar vertical position and hence 0.7 Å lower than the 6x6 maxima. Finally in step III-IV, graphene continues to grow from the edges and a complete graphene monolayer lies 2.3 Å above the newly generated 63 interface as suggested in Section 4.3. Besides growing from step edges, on occasion graphene islands are seen nucleated on terrace. These graphene islands can be found inside the boxed area in Fig. 4.22. These graphene often form next to where the deep pits were created. Hannon et al. proposed the pinning of 63 causes the formation of these pits [31]. We believe the formation of these deep pits perturbed the primary mechanism from taking place. 4.5.5 Kinetics of graphene formation The proposed mechanism for graphitisation requires the breaking of the Si-C bond at the step edges of the initial 63 layer. While Si-Si bonds in the Si-rich SiC phase are known to break at temperatures as low as 900oC [9], the stronger Si-C bond in the 63 layer is anticipated to occur above 1200oC, the temperature at which graphene starts to form. The kinetics describing the breaking of the Si-C bond at the 63 surface to generate free C atoms (C*) leading to graphene formation can be expressed as follows (Eq. 4.1): 144 Chapter k kb a   C(6 3)   CG  C*  k a (4.1) k d Si(6 3)  Si(v) where ka is the rate constant for breaking a Si-C bond, k-a the rate constant for a C* atom to re-attach to the 63 site (i.e. forming Si-C bond), kb the rate constant for C* atom attach to a graphene site (forming C-C bond) and kd the rate constant for desorption of Si from 63 into vapour (v) phase and this last process is irreversible and shall not be the rate limiting step at these temperatures. By treating the processes in eq. (4.1) as first order, the following rate equations can be written: d C(6 3)     k C  a  (6 3)   k a  C * dt d  C * dt d  CG  dt (4.2)  ka C(6 3)   k a  C *  kb  C *   (4.3)  kb  C * (4.4). C* is a highly reactive intermediate and is assumed to be consumed immediately after being produced. They are therefore present in minute amounts compare to C63 and CG at any stage of the graphitisation, and hence d  C * dt is very small compared to d C6  dt and d  CG  dt i.e. d  C * dt  . From eq. (4.3) we have   ka C(6 3)   C *   k a  kb (4.5). Equation (4.4) thus becomes, d  CG  dt  ka kb C(6 3)    k a  kb (4.6). Rebinding of Si and C to form Si-C bond in eq. (4.1) is unlikely since they are not stable 145 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) at these temperatures and desorption of Si is irreversible. Hence this gives ka and kb >> k-a. Equation (4.6) can be simplified to: d  CG  dt  ka C(6 3)    (4.7) Treating [C63] as fraction of area terminated with 63 and [C63]= - [CG], eq. (4.7) has an integrated form of: ln 1- CG   ka t (4.8) where [CG] is the fraction of area converted to graphene and ka is given by:  E  ka  A exp   a   k BT  (4.9) where A is an Arrhenius pre-exponential factor, Ea is the energy barrier for breaking Si-C bond and kB is Boltzmann constant. [1-CG] can be extracted from Fig. 4.20 and using Eqs. (4.8) and (4.9), ka at different temperatures (Fig. 4.29a) and the activation energy Ea (Fig. 4.29b) can be determined experimentally. As shown in Fig. 4.29, a good fit is obtained in both instances and Ea is found to be 3.0 ± 0.4 eV. (a) (b) Ea=3.0 ± 0.4 eV Fig. 4.29 (a) Integral plot according to Eq. (4.8) and (b) Arrhenius plot for rate constant, ka according to Eq. (4.9). 146 Chapter The extracted energy barrier is found closer to the Si-C bond energy (3.3eV) than Si-Si (2.3eV) or C-C bond energy (3.6eV) [42]. The implication is that the 63 surface which contains Si-C bonds is the rate-limiting step in the formation of graphene at lower temperatures. This is consistent with the structural model proposed for 63 surface where it consists of Si-rich 6x6 maxima. The 0.3 eV energy different from the Si-C bond energy may be due to the fact that graphene growth commence from the step edges that may reduce the overall kinetic barrier. 147 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) 4.6 Conclusions The graphite surface shows various atomic resolutions that change under different tunnelling bias. There is dominance of electronic effect and hence may not provide useful topographic information. Using the honeycomb structure to determine single layer graphene over multilayer graphene may not be entirely reliable although this method has been employed by others to determine graphene thickness produced by them [29]. The 6√3 surface is found to be modulated by regular array of 6x6 maxima that are Si-rich and 3 clusters that are carbon rich. The combination of 6x6 maxima and 3 clusters may give rise to 6√3 diffraction pattern observed by LEED and bridge the discrepancy between LEED and STM. A simple model is proposed for 6x6 maxima which consist of three Si tetramers and six to ten SiC clusters. When graphene forms, a new 6√3-like layer is generated beneath the graphene monolayer as the interface. The electronic contributions from this interface are found still dominant under STM imaging. Employing the different adsorption behaviour of Co on 6√3 and graphene terraces, this mapping technique is used to monitor the key processes leading to graphene formation on 6H-SiC(0001). The Co decoration shows that graphitisation starts from step edges, akin to step-flow growth. The mapping also allows us to follow the evolution of step heights from 6√3  (6√3 + graphene)  graphene in great details where this information revealed that concurrent with the transformation of 6√3 to graphene, BL underneath the surface collapse to form a new 6√3 at the interface. This 63 interface acts as a buffer layer that protects graphene from direct interaction 148 Chapter with SiC bulk. The collapse of BL indicates that the 63 layer may have sufficient carbon atoms to form graphene. However, due to the presence of the Si-rich clusters (6x6 maxima), the 63 layer are prevented from directly converting to graphene. Kinetic study of this surface shows that breaking the Si-C covalent bond is the overall rate limiting step to the transformation, with an energy barrier of 3.0  0.4eV, in agreement with Si being present in the 63 phase. Based on the STM observations, a universal mechanism that is responsible for the step height evolution leading to formation of mono-, bi- and tri-layer graphene growth from 63 are established. The mechanism proposed shows that the graphene form via this mechanism is able to retain the original terrace size of 63 surface. For this reason the graphene size is pre-dictated by the size or quality of the precursor i.e. 63 surface. Hence, improving the quality of SiC wafer or 63 surface should be able to improve the quality of graphene. The result also implies that for any given SiC polytypes, fast steps are more desirable than slow steps. However, the 6H polytype seems to have advantage over other polytype as its intrinsic stacking (3 BL) support the formation of 63 naturally. 149 [...]... at 0.95V of graphitised 6H-SiC(0001) Both graphene and 6x6 maxima from 63 structure are seen in this image; (b) top panel: (60nm x 40 nm) filled states image of a partially graphitised surface The 63 and graphene (G) areas are identified by the density of Co clusters Bottom panel: line profile of graphene and 63 surface where graphene lies 2.3 Å above 63; and (c) top view and side view of a schematic... 0.1V (inset, 2nm x 2nm) shows graphene lattice; (c) and (d) distributions of cluster volume on 63 and graphene terrace respectively; and (e) line profiles of unadorned area of 63 (blue) and graphene (dotted line) Fig 4. 17a shows two types of terrace exist on a partially graphitised surface of 6H-SiC(0001) that previously annealed at 1230ºC As observed there are two types of terraces as mapped by two... distance of 3 clusters and 5x5 maxima with respect to 6x6 maxima As shown by the line profiles in Figs 4. 8b and 4. 8c where both of the STM images were captured at the same tunneling bias, the 3 clusters 105 STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) and 5x5 maxima are sitting ~0.7 Å lower than the 6x6 maxima Finally, extending the 6x6 maxima of 63 on the left of Fig 4. 8c to... this section and they are based on the following experimental observations i.e.:(i) two different atomic clusters with size ~ 6.1 Å and ~ 3.0 Å found on 6x6 maxima (Figs 4. 5a, 4. 7a and 4. 7b) and their position relative to each other (Figs 4. 7a and 4. 7b), and (ii) the atomic clusters in 6x6 maxima is Si-rich and consist of elemental Si-Si bonds which was detected by the photoemission study of this surface... cells for (6x6), 63 and 3 are also drawn 118 Chapter 4 Figure 4. 14 compares the 63 surface prior and after graphitisation occurs Both surfaces look quite similar where both of the 6x6 maxima and 3 clusters can be seen Each of these 6x6 maxima and 3 clusters still retain their orientation on graphitised surface i.e (6x6) and 3 respectively as shown by the inset of Fig 4. 14b However, these two... clusters of 5x5 in (a) and 3 clusters in (b) are highlighted using arrows and circles labeled A, B and C; (c) a 63 region next to a 5x5 region on the right The 6x6 maxima of 63 are extended to the 5x5 using open circles and these circles sit on the voids of 5x5; and (d) the line profile from (b) and (c) are shown as top and bottom panel respectively Both line shows similar vertical distance of 5x5 and. .. STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) (a) Cross-section Intrinsic (bulk) [0001] T4 adatom A A’ 1 2 3 [1120] [1100] 4 5 (c) (b) [0001] [1120] A 3 /4 5 1 [1100] 2 H3 T4 6.16 Å A’ 3.08 Å (d) 6x6 Fig 4. 12 Proposed structure for 6x6 maxima (a) Cross-section of 6x6 maxima before and after termination with one Si adatom filling the T4 site of first Si layer; (b) top view of the proposed... reveals graphene lattice (Fig 4. 17b inset), but not on terraces with high density of Co clusters Above observations show the high density of Co clusters nucleates on terraces terminated by 63 and the low density of Co clusters nucleates on the graphene On both surfaces, Co appears as 3-dimensional (3D) dome-shaped clusters Figures 4. 17c and 4. 17d show the volume distributions of Co clusters on 63 and graphene. .. The black and red diamonds represents the (1x1) unit cell of graphene and SiC(0001) respectively For simplicity, each 6x6 maxima is represented using a group of three spheres Further STM investigation of this graphitised surface reveals that this 63-like layer is actually at the interface beneath the graphene In the lower half of Fig 4. 15a, the 119 STM studies of graphite, C-rich 63 and graphene on... where the line profiles show that the 63 surface is twice as corrugated as graphene 1 24 Chapter 4 It is evident from the above observations that adsorption of Co can be used as a tracer to map the evolution of 63 and graphene terraces at different stages of graphitisation In the next section, the evolution of the surface using this method will be followed and the key processes leading to graphene formation . 2 .4 Å 2 .46 Å STM studies of graphite, C-rich 63 and graphene on 6H-SiC(0001) 98 4. 2 STM of C-rich (63x63)R30 o /6H-SiC(0001) 4. 2.1 (63x63)R30 o surface STM image in Fig. 4. 4. very often observed as hexagonal close-packed (hcp) structure with periodicity of 2 .4 Å as shown in Fig. 4. 3a [3 ,4, 5]. Assigning this image to the ball -and- stick model of graphite in Fig. 4. 3b. of graphite, C-rich 63 and graphene on 6H-SiC(0001) 94 4. 1 STM of highly oriented pyrolytic graphite (HOPG) This section revisits the STM imaging (constant current mode) of graphite and