Dynamics of epitaxial graphene growth and adsorptions of cobalt 6

6 Adsorption of Co on Graphene surfaces at elevated temperatures Growth of Co at elevated temperatures induced ordering of Co on graphene/6HSiC(0001) surfaces. Flat top islands with thickness of just a few atomic layers but spreading as wide as 50 nm are observed. Besides these flat islands, two other cluster-like structures are also formed on the surface. They are labelled as dim clusters for features with thickness less than Å and bright clusters for features with thickness between to Å but smaller in width than the dim cluster. Using the 63 interface as a reference, the dim and bright clusters are found to preferentially nucleate on top of the C-rich 3 clusters and Si-rich 6x6 maxima respectively. The dim clusters are often found cluttering around the edges of the flat island while distribution of bright clusters is random. Post-growth annealing experiments further reveal that the flat islands formation is primarily mediated by the dim clusters. Scaling analysis performed on the flat islands reveals that a single dim cluster can transforms into a flat island i.e. i*= 0. By annealing to higher growth temperatures, bright clusters are also observed to contribute towards formation of flat islands by releasing Co adatoms via dissolution. More interestingly, both STM and XPS reveal that the flat islands are sandwiched between the 63 interface and graphene surface. These flat islands, as a consequence, are also stable against ambient oxidation. Co adsorption at elevated temperatures 6.1 Adsorption of Co on graphene at elevated temperatures (a) 450oC Co (b) RT 4.9 h= 4.1Å 10.1 8.0 10.0 I 9.4 II 11.6 22.3 [1210 ] 50 nm 4.8 [1 120 ] 50 nm (c) 2.5 z (nm) 2.0 II 1.5 h = 22.3Å 1.0 I h = 10.0Å 0.5 0.0 20 40 60 80 100 x (nm) 120 140 160 180 Fig. 6.1 STM images (400nm x 400nm) showing morphologies of Co after deposited at (a) 450oC and (b) room temperature (RT) on graphene surfaces. Line profiles in (c) are cross-section of islands in (a) and (b). The line profiles shows Co islands grown at 450oC are 2D multilayers with a flat top while Co grown at RT are 3D dome-shaped cluster. The height, h for other the islands in (a) are given by the numeric labels placed next to them. Figure 6.1a shows the morphology of Co on graphene/6H-SiC(0001) at 450oC. It differs dramatically from those grown at room temperature as shown in Fig. 6.1b. The diameter of Co features grown at 450oC ranges from to 50 nm while the Co features grown at room temperature ranges from 5-13 nm. Although bigger in size, the Co islands grown at 450oC shows that these Co islands have flat top and with a height, h of just a few atomic layers* (Fig. 6.1c, line profile I). The thicknesses of other flat islands in Fig. * Assuming a close-packed structure for Co i.e. hcp (0001) or fcc (111), one atomic layer has a height of Å. 207 Chapter 6.1a are given by the numeric labels placed next to them and all these islands have flat top. In comparison, the Co grown at room temperature are highly 3-dimensional (3D) and dome-shaped with heights that are twice the former (Fig. 6.1c, line profile II). The geometry of Co islands grown at 450oC i.e. multilayer 2D islands terminated with flat top evidently shows that growth at elevated temperature induces crystal ordering in Co structures. Flat top islands are typical characteristic for multilayer islands with crystal ordering. This has been seen in multilayer 2D Pb islands grown on Si(111) in which close-packed structure forms [1,2,3]. Graphene, clean Co 15 min, 450oC Co 30 min, 450oC Co 100 nm Co 45 min, 450oC Co 60 min, 450oC Co 90 min, 450oC Fig. 6.2 STM images (400nm x 400nm) of sequential growth of Co on graphene/6H-SiC(0001) at 450oC. Co flux used is (1.6  0.2) x 10-2 ML/s. Figure 6.2 follows the sequential growth of Co on a graphene surface that was held at 450oC and exposed to Co flux of (1.6  0.2) x 10-2 ML/s. After 15 minutes of Co deposition, formation of Co islands is observed to begin at the edges of graphene terraces 208 Co adsorption at elevated temperatures (highlighted by arrows). Nucleation on terraces is seen only to take place later and in this series of growth, it occurs after 30 minutes of deposition. The density and size of the Co islands increase with the amount of Co deposited, obeying to the typical diffusionmediated nucleation and growth from vapour phase. At 90 minutes, the shape of the Co islands differs from preceding coverage. This difference is caused by coalescence of islands where two or more islands join up as their growth zone overlap with each other. (b) 25 (a) > ML Frequency (%) 20 15 10 5.5 Co 4.0 nm 9.5 13.5 17.5 21.5 Co island height (A) 4.00 200 nm Fig. 6.3 (a) STM image (1000nm x 1000nm) of 100 minutes of Co grown on graphene/6HSiC(0001) at 585oC and (b) height distribution of Co flat islands extracted from STM image shown in (a). Figure 6.3a is an example of the Co morphology at coverage higher than those shown in Fig. 6.2. In this case more than 80% of this surface is covered with Co. Based on the morphology in Fig. 6.3a, the Co islands appear to wet the surface and they maintained a flat top island geometry rather than forming 3D dome-shaped islands. Approximately 70% of the Co islands have thickness exceeding ML (Fig. 6.3b). The Co 209 Chapter islands appear irregular or fractal-like after coalesced with neighbouring islands. Interestingly, for any given large island, the height of this island is quite uniform. Considerable rearrangements of materials after coalescence must have occurred, otherwise two or more heights arising from coalescence of islands with different heights should be observed. It is noted in Fig. 6.1a that prior to coalescence, Co islands of various heights are formed on the graphene surface. In order to study the growth of Co islands in more details, STM study of the surface morphology after Co deposition at various temperatures and time were performed. The results are shown in Fig. 6.4. It is evident that at any of the growth temperatures, the Co flat islands formation begins by preferentially nucleating along the edges of terrace before nucleation on terraces occurs. For growth at a constant temperature, the density and size of these flat islands increase with coverage and coalescence of these flat islands are seen at higher coverages. As the growth temperature increased, the island density decreases while the island size increases. This change can be understood since the Co diffusion length increases with temperature and with longer diffusion length, the system inclines towards island growth than nucleation [4,5,6]. Hence, the densities of island are lower but islands are bigger in size at higher temperature. 210 Co adsorption at elevated temperatures (a) 10 min, 585oC 30 min, 585oC 50 min, 585oC 100 min, 585oC 10 min, 635oC 30 min, 635oC 50 min, 635oC 80 min, 635oC 30 min, 720oC 50 min, 720oC 100 min, 720oC 200nm (b) 200nm 100nm (c) 10 min, 720oC 200nm Fig. 6.4 A series of STM images showing growth of Co islands on graphene held at (a) 585oC, (b) 635oC and (c) 720oC. Co flux used is (1.6  0.2) x 10-2 ML/s. All images are 800nm x 800nm except b(i) (500nm x 500nm). The colour scales are adjusted for each image to enhance the details of the morphologies. (a) 18 (b) 18 450oC -2 450 C 14 16 585oC o Island density (x10 / cm ) 635oC 720oC 12 Island density (x 10 / cm ) 16 10 o 585 C 635oC 720oC Nsat 14 12 J 10 0 20 40 60 80 100 Deposition time (min) 120 20 40 60 80 100 Deposition time (min) 120 Fig. 6.5 (a) Density of Co flat islands with deposition time and growth temperatures. Solid line for each curve is guide for the eye; and (b) fitting of pre-coalescence regimes in (a) according to N(t)= Nsat[1-exp(-J(t-ti)/Nsat)]. 211 Chapter The densities of these flat islands as a function of deposition time for each growth temperature are measured and they are shown in Fig. 6.5a. Both Nsat and the initial nucleation rate, J are seen decreases with temperature. In addition, an incubation period prior to formation of flat islands can be observed for all growth temperatures where the density of flat islands is zero up to a period of at least 10 minutes. This behaviour is in sharp contrast to the 3D clusters growth at room temperature. To extract quantitative values of Nsat and J, the growth regimes prior to saturation are fitted according to Eq. (5.1) i.e. N (t )  N sat 1  exp   J (t  ti N sat   where ti is the incubation period. The fitting for each growth temperature (solid lines) is provided in Fig. 6.5b and the values of J and Nsat used to fit the curves are tabulated in Table 6.1. As discussed earlier, both Nsat and J decrease with temperatures and they are resulted from longer diffusion length of Co adatoms at higher temperatures. Table 6.1 J and Nsat values for various growth temperatures of Co on graphene/6H-SiC(0001). Substrate temperature (oC) J (cm-2s-1) Nsat (cm-2) 450 1.7 x 107 1.5 x 1010 585 9.6 x 106 9.6 x 109 635 7.2 x 106 8.1 x 109 720 5.0 x 106 6.1 x 109 The present of the incubation period would suggest the need to achieve a critical adatom concentration on the surface before nucleation of islands occurs. It is anticipated that apart from nucleation and growth, any other kinetic processes that compete for adatoms and subsequently deplete their concentration would inevitably result in a longer period of time before nucleation of islands occurs. Therefore, if process such as desorption of Co adatoms becomes significant at high growth temperature, a longer time is needed to achieve the critical adatom density for nucleation and hence resulted in a 212 Co adsorption at elevated temperatures noticeable incubation period as seen in Fig. 6.5a. This period would thus be longer as temperature increases since desorption rate also increases. However, when we examine the STM images at higher magnification, the imaging suggests that the desorption process is not the only reason for the existence of an incubation period. “hole” Bright cluster Co flat island Bright Dim nm 63 interface Co flat island Dim cluster Co flat island 10 nm Fig. 6.6 STM image (150nm x 100nm, Vs= -2.7V) showing besides Co flat islands, two type of clusters also exist on the surface i.e. the smaller but brighter (marked by solid arrow, bright cluster) and the bigger but darker cluster (marked by broken arrow, dim cluster). Inset (20nm x 20nm) highlights the adsorption site of bright and dim cluster relative to the structure of 63. The bright clusters reside on top of the corner of 6x6 maxima while the dim clusters reside on top of the “hole” (3 clusters) of 63. Due to close proximity of maxima and the “hole”, very often the bright and dim clusters are next to each other. The graphene surface was held at 450oC and exposed to Co flux of 1.6 x 10-2 ML/s for 45 minutes. Figure 6.6 shows the morphology of the surface surrounding three flat islands. At this bias and length scale, graphene is not resolvable under STM but the unadorned area of the surface is dominated by the 63-like structure (appears as honeycomb) at the interface. As depicted, besides the three Co flat islands (named as island hereafter), 213 Chapter smaller structures i.e. small clusters are also seen nucleating with density and nucleation rate significantly higher than the islands. Closer examination of these clusters reveals that there are two types of clusters depending on their brightness and size. One type appears brighter but smaller (marked by solid arrow, called bright cluster hereafter) while the other type appears wider but darker (marked by broken arrow, called dim cluster hereafter). At Vs= -2.7 V, the bright clusters measure approximately 20 Å (width) x 1.5 Å (height) while the dim clusters measure approximately 35 Å (width) x 0.7 Å (height). Besides the difference in size (or brightness), these clusters are found to sit on different adsorption site as shown in the inset. In relative to the structure of the 63 interface, the bright clusters are found to preferentially reside on top of Si-rich 6x6 maxima of 63 while the dim clusters reside on top of the “hole” of the nanomesh which is also known to be the site where the C-rich 3 clusters are located. Due to the close proximity of 6x6 maxima and 3 clusters of 63, a bright cluster can also be found sitting next to a dim cluster. These clusters are also found for all the other elevated temperatures. Figure 6.7 compares the coverage-dependant cluster and island densities at growth temperature of 720oC. During the characterisation, a number of STM imaging have poor resolution of the 63 interface and resulted in difficulties to identify the dim and bright clusters. Hence in the density plot of clusters in Fig. 6.7, the bright and dim clusters are counted collectively. Nevertheless, the data is still meaningful for the following discussions. The formation of clusters is seen to begin almost immediately after deposition started and their nucleation rate is much faster than the flat islands. Their density reached saturation rapidly in comparison to the density of flat islands. In fact it is noticeable that nucleation of flat islands begins when the cluster density is approaching saturation. Thereafter the density 214 Co adsorption at elevated temperatures of flat islands increases gradually and eventually reaches a plateau. On average, the collective densities of clusters (bright and dim) are more than two orders higher than the densities of flat islands. The rapid nucleation of clusters also implies that they have lower kinetic barrier on the graphene surface. At 720oC, based on their density at the early stage of nucleation i.e. at t = 20 minutes, the population ratio of island to cluster is 1:1600, giving an energy difference of 0.63 eV†. Island / cluster density (x1012/ cm2) 1.2 o T = 720 C 1.0 cluster 0.8 0.008 flat island 0.006 0.004 0.002 0.000 20 40 60 80 100 Deposition time (min) 120 Fig. 6.7 Comparison of cluster and island densities that both found nucleating on graphene at 720oC. Solid line for each curve is guide for the eye. Figure 6.8 shows high resolution STM images of flat islands, bright and dim clusters captured at both filled- and empty-states. It is clear from this bias-dependant studies that regardless of the states they are imaged under, the presence of two types of clusters i.e. bright and dim clusters is physical and not due to electronic effect. As shown, the change in the apparent heights of Co flat islands (between 4.0 and 4.1 Å) is almost † The energy is approximated from n1/n2 = exp (ΔE/kBT), where n1=population of flat island and n2=population of dim cluster. 215 Co adsorption at elevated temperatures -3.5 ln (N flat / N dim) -4.0 -4.5 -5.0 Eeff,flat= 0.7  0.1 eV -5.5 0.9 1.0 1.1 1.2 1.3 1000/T Fig. 6.17 Arrhenius plot for Nflat/Ndim according to Eq. (6.5). The slope yields an energy barriers of 0.7  0.1 eV. Experimentally the determination of Nflat/ Ndim at various temperatures would require one to measure the density of dim clusters and also the density of flat islands of a given coverage i.e. at a fixed deposition time. The time frame chosen however should also be in the pre-coalescence regime. The dim cluster density is obtained from Fig. 6.14e while the corresponding flat island density is obtained from the same experiments extracted from STM images of several microns. The Arrhenius plot according to Eq. (6.5) is found to yield a straight line as shown in Fig. 6.17. The activation energy barrier i.e. Eo is determined to be 0.7  0.1 eV. We will discuss the implication of these energy barriers in Section 6.4 where the mechanism for the growth of flat islands is discussed in more detail. 231 Chapter 6.3 Surface intercalation of Co flat islands STM studies of Co growth (Section 6.2.1) and annealing experiments (Section 6.2.2) suggest that the Co adatoms penetrate the graphene monolayer where nucleation of dim clusters which mediates the formation of flat islands occurs at the 63 interface and beneath the graphene surface. The implication is that the Co features, in particular the large flat island are covered by graphene. Direct evidence for this is seen in Fig. 6.18a where filled-state image shows an area of an unadorned graphene with 63 at the interface (upper left) next to a flat island. The lattice of graphene above the 63 interface is clearly visible and more significantly this graphene monolayer is seen to be a continuous layer which covers the flat island as indicated in Fig. 6.18b. This graphene monolayer as shown in Fig. 6.18c is also seen as a continuous layer over the flat island even when the height of this flat island changed. Self correlation of the real-space atomic resolution imaged on this flat island is provided as an inset. The periodicity of the structure is akin to lattice parameter of graphene i.e. 2.46 Å but it appears to be under some level of stress as the STM images shows graphene lattice is distorted as it is dilated near the edges of the flat islands. 232 Co adsorption at elevated temperatures (a) Graphene (b) Co flat island (c) nm (b) 2.65Å (c) 2.3Å nm nm Fig. 6.18 Filled-state STM image (Vs= -0.2V) of (a) (30nm x 30nm) a Co flat island next to an exposed graphene surface. The highlighted area in (a) are presented as (b) (10nm x 10nm) and (c) (5nm x 5nm) respectively. Inset (2nm x 2nm) in (c) is self-correlation extracted from imaging on Co flat islands in (a). 233 Chapter (a) Co poly Co 4s+3d Intensity (a.u.) Intensity (a.u.) (b) VB Co/G Co/G Co poly 20 18 16 14 12 10 Binding energy (eV) 27.5 42.5 57.5 72.5 -2 (d) Si 2p Si-C Si-Si Intensity (a.u.) (e) 106 104 27.5 42.5 57.5 72.5 102 100 98 Binding energy (eV) 96 C-C sp2 94 C 1s 291 289 287 285 283 Binding energy (eV) 281 (f) 782 780 778 776 Binding energy (eV) 27.5 42.5 57.5 72.5 158 Co-Co 774 Si 2s Si-C Si-Si 156 154 152 150 Binding energy (eV) 27.5 148 146 O 1s 42.5 57.5 C-Si 293 160 Intensity (a.u.) 108 784 Intensity (a.u.) Intensity (a.u.) (c) Co 2p3/2 27.5 42.5 57.5 72.5 Co poly 279 72.5 538 536 534 532 530 Binding energy (eV) 528 526 Fig. 6.19 Angle-resolved XPS (Al K 1486.6eV) of Co/graphene after exposed to air for more than months: (a) valence band, (b) Co 2p3/2, (c) Si 2p, (d) Si 2s, (e) C 1s and (f) O 1s. Co was deposited at 1.6 x10-2 ML/s for 100 minutes on a graphene surface held at 610oC where 80% of the graphene is covered by Co flat islands as shown by the STM image (1000nm x 1000nm) in (a). The XPS acquired from a polycrystalline Co is also provided in (a) and (b) for comparison. Legend provided in each core-level indicates the take-off angles from 27.5o to 72.5o. 234 Co adsorption at elevated temperatures Apart from STM, XPS studies of Co/graphene samples exposed to air were also carried out. Figure 6.19 shows the angle-resolved XPS of Co/graphene (Ts= 610oC) that had been exposed to air under ambient conditions for more than months. Bulk Co signals (valence band and Co 2p3/2) acquired from sputtered-clean polycrystalline Co foil is also provided for comparison. Survey scan from XPS shows presence of oxygen on this exposed Co/graphene surface. Despite that, both of the valence band (Fig. 6.19a) and Co 2p3/2 (Fig. 6.19b) spectra show Co remained in metallic states and stable against ambient oxidation. The Fermi edge arises from Co 4d+3s remained sharp and identical to the Fermi edge from bulk Co. The Co 2p3/2 core-level spectra are also identical with the bulk Co signals and both of them are located at 778.4  0.2 eV and asymmetric. Typical Co oxides peaks i.e. the Co3+ and Co+ states which normally located between 781 and 780 eV respectively are obviously not present in this Co 2p3/2. This observations also directly imply that there is no chemical interaction between Co flat islands and graphene or 63 interface. Cross-checked with oxygen (O 1s in Fig. 6.19f) also shows absence of any metallic oxides that normally located at around 530 eV. Instead typical air-related oxides that located at around 532 eV are recorded. Other core-levels spectra from the substrate i.e. Si 2p (Fig. 6.19c), Si 2s (Fig. 6.19d) and C 1s (Fig. 6.19e) also shows non-oxidised peaks, otherwise peaks at 104 eV (Si4+ 2p), 155 eV (Si4+ 2s) and 286-290 eV (C-O related C 1s) will be observed. All of the core-levels show sub-peaks that are related the graphene surface as discussed earlier in Section 5.2.1. In particular, the elemental Si peaks are still observable in Si 2p (99.2 eV) and Si 2s (150.9 eV) respectively and they are not oxidised. As for C 1s core-level (Fig. 6.19e), the C-Si signal is visible initially and subsequently not detectable as the take-off angles are more surface sensitive. Only C 1s signal related to graphene (sp2) located at 284.5 eV can be detected. The humps located between 285.5 and 286 eV (peak C5 and C6 in Section 5.2.1) which associated to 3 235 Chapter clusters are also not detectable after deposition of Co at elevated temperature. The pristine state of all the elements strongly suggest that the surface is terminated with graphene and hence protecting the Co and 63 interface underneath against oxidation. Further evidence is also provided when the intensities of the core-levels as a function of take-off angle are analysed as shown in Fig. 6.20. (b) 100 80 Atomic percentage (%) Atomic percentage (%) (a) 100 C 1s 60 40 Si 2s 20 80 C 1s (284.5 eV) 60 40 20 Co 2p3/2 Co 2p3/2 20 40 60 o Take-off angle ( ) 80 20 40 60 80 o Take-off angle ( ) Fig. 6.20 Angle-dependant atomic percentage of (a) C 1s, Si 2s and Co 2p3/2 and (b) C 1s (284.5 eV, graphene component) vs. Co 2p3/2. Figure 6.20a shows the signal of C 1s increases with take-off angles while the signals of Si 2s and Co 2p3/2 decreases, implying that the system is terminated with carbon while both Si and Co are located at the interface. As seen earlier in Fig. 6.19e, these C species at the surface are graphene. As the take-off angles increases, graphitic sp2 C 1s (284.5 eV) increases and becomes the only component detectable at take-off angle of 72.5o. The peak areas of this component are extracted via peak fitting and plotted against the Co signals as presented in Fig. 6.20b. The graphene C 1s signals is seen to increase while the Co signals decreases with the take-off angles. This observation further substantiates the STM results and the proposed structure where the Co flat islands are 236 Co adsorption at elevated temperatures burrowed beneath the graphene monolayer. (a) 63 nm (b)(ii) (b)(i) 2.42 Å nm nm (c) (d) nm nm Fig. 6.21 (a) STM image of Co flat island grown on (63x63)R30o/6H-SiC(0001) surface at 610oC. Similar to Co/graphene surface grown on elevated temperature, the Co island on 63 surface is also covered by a layer of graphene as depicted in (b)(i) where the honeycomb structure of graphene is clearly noticeable. The self-correlation of this graphene lattice is shown in (b)(ii) where a hexagonal close- packed structure with a lattice parameter similar to graphite i.e. 2.42 Å is revealed. This graphene monolayer is also seen extending to different layer of this flat island as seen in (c) and (d). This graphene monolayer is believed to be originated from the 3 carbon clusters on the 63 surface where this formation process is catalysed by Co. As a comparison, we also examined the growth of Co on a 63 surface. Figure 6.21a shows the STM image of a flat island formed on 63 surface at 610oC. Recording this island at atomic resolution reveals that the surface of this flat island is surprisingly terminated with graphene layer where the honeycomb lattice is clearly observed in Fig. 6.21b(i)). Self-correlation of this image shows lattice parameter of graphite i.e. 2.42 Å 237 Chapter (Fig. 6.21b(ii)). In addition this graphene monolayer is seen extending continuously to a different layer of the flat island (Figs. 6.21c and 6.21d). The STM observation in Fig. 6.21 clearly shows intermixing of Co with the 63 surface where the additional carbon atoms to form graphene would come from the C-rich 3 clusters. Hence penetration of Co into 63 can in fact catalyse the conversion of 63 into graphene at temperatures well below the temperature window necessary for graphitisation of 6H-SiC(0001). In addition, intermixing of Co with 63 would further support the observation seen in Fig. 6.13a-b where depressions of 0.7 Å lower than the 63 interface are observed. 238 Co adsorption at elevated temperatures 6.4 Mechanism of Co flat island formation on graphene Having established that the flat islands, dim and bright clusters are beneath the graphene, the following mechanism for the formation of flat islands from the Co vapour phase is proposed to account for the growth behaviour and dynamics of bright, dim clusters and flat islands seen in Sections 6.1 to 6.3. Co atoms from the vapour phase first adsorbed on the graphene surface. These adsorbed Co atoms (adatoms) are mobile and they diffuse readily on the surface. Apart from the supply of Co atoms from the vapour phase, the concentration of Co adatoms on the surface also depends on two other events, namely desorption as well as penetration of Co adatoms through the graphene layers and diffuse into the region between the graphene and 63 interface. These kinetic processes are illustrated in Fig. 6.22a(i). Although not illustrated, the main channel for penetration will be via step edges and defects of graphene. Beneath the graphene layer, these Co adatoms diffuse and aggregate with each other to form bright or dim clusters. As the dim clusters are a lot shallower than the bright clusters, Co adatoms also penetrated into the 63 interface in forming the dim clusters, from which nucleation and growth of multilayer flat islands occurs (Fig. 6.22a(ii)). The reaction pathways and potential energy diagram for a Co atom from vapour phase to formation of cluster or island are illustrated in Fig. 6.22b. The left side of the diagram shows the pathways to formation of bright cluster while right side of the diagram shows the pathway leading to the formation of dim cluster and subsequently the flat island. For the left side of the diagram, the adsorbed Co adatom can either desorbed from the surface (Edes) or penetrate through the graphene layer. Penetration requires the Co adatoms to surmount an effective barrier, E1. As such E1 is likely to consist of the surface 239 Chapter surf,G diffusion barrier on graphene, ( Edif ) and also a penetration barrier,( EpG ). At the adatom/63 interface, the Co adatoms will need to overcome an energy barrier E2 leading to formation of bright clusters. This energy barrier is likely to have contribution of surf,6 ) and also a nucleation barrier for i*, ( Eibright ). surface diffusion barrier on 63, ( Edif * Both E1 and E2 can be written as follow: surf,G E1 Edif , EpG   (6.6a) surf,6 E Edif , Eibright *  (6.6b) And the total energy barrier leading surmounted by Co adatoms from vapour phase to formation of bright clusters, Ebright has the contribution from both E1 and E2: Ebright  E1, E  (6.6c)  surf,G surf,6 Ebright Edes , Edif , EpG , Edif , Eibright *  (6.6d) This effective barrier, Ebright has been deduced experimentally i.e. 0.5  0.1 eV (see Fig. surf,G surf,6 6.14f). This effective barrier would have contribution from Edes , Edif , EpG , Edif and Eibright . * Similarly, the Co adatoms on the right side of the diagram are also subjected desorption, diffusion of graphene surface and penetration through the graphene layer. At the interface, apart from diffusion barrier on 63 surface, the Co adatoms are subjected to additional barriers i.e. nucleation barrier for dim cluster, ( Eidim * ) and penetration into 63 interface, Ep6 before formation of dim clusters. These barriers are depicted as E3 in Fig. 6.22b and can be written as follow: 240 Co adsorption at elevated temperatures  surf,6 E Edif , Ep6 , Eidim *  (6.7a) And the total energy barrier the right side leading to formation of dim cluster is given as follow: Edim  Edes , E1, E 3  surf,G surf,6 Edim Edes , Edif , EpG , Edif , Ep6 , Eidim * (6.7b)  (6.7c). Experimentally, the effective barrier surmounted by Co adatoms from vapour phase leading to nucleation of dim clusters, Edim is found to be 1.1  0.1 eV (see Fig. 6.14e). surf,6 surf,G , EpG , Edif , Ep6 and This effective barrier would have contribution from Edes , Edif Eidim * . The main kinetic process for transformation of dim cluster into flat island involves rearrangement of Co adatom which includes adatoms hoping up and down and edge diffusion as depicted in Fig. 6.22a(ii). The total barrier of these processes is represented as E4 in Fig. 6.22b and it has been previously deduced experimentally as Eo i.e. 0.7  0.1 eV (see Fig. 6.17). The bright clusters on the other hand also contribute to the growth of flat islands although it requires dissolution to release adatoms. The possible kinetic pathway for dissolution is given as dotted line in Fig. 6.22b. This would require them to surmount an energy barrier, E5 which includes binding energy of Co-Co to release the adatoms. The released adatoms are then subjected to diffusion barrier (E2) before captured by a dim cluster or a growing flat island. However this pathway is most likely the secondary channel that limited during growth at a constant temperature and requires the surface 241 Chapter operating at higher temperatures for e.g. annealing. The origin for these bright clusters is most likely similar to the Co clusters form on clean 63 surface at room temperature i.e. nucleation via diffusion and aggregation of Co adatoms. They have similar geometry (aspect ratio) and both preferentially nucleate next the 6x6 maxima. 242 Co adsorption at elevated temperatures (a) (i) desorption Vapour phase diffusion penetration Graphene (G) Co adatom penetration 63 Co adatom nucleation (bright cluster) Nucleation (dim cluster) bulk (ii) hop-up hop-down Edge diffusion Flat island bulk (b) Vapour phase desorption Edes LEFT nucleation penetration RIGHT penetration E1 E5 E2 E2 Adatoms/ 63 (interface) nucleation E1 transformation E3 Adatoms/ G (surface) Adatoms/ 63 (interface) E4= 0.7 eV Annealing Bright cluster Dim cluster Annealing Flat island, i*=0 Fig. 6.22 (a) Mechanism and (b) energy diagram for formation of clusters (bright & dim) and flat islands. When a bright cluster undergoes dissolution especially driven by annealing, it will follow the dotted-line pathway before captured by a flat island. 243 Chapter It is believed that embedding the Co underneath the graphene monolayer is primarily motivated by the system trying to minimise the surface energy. Surface burrowing only occurs if the surface energy of adsorbate is much higher than the surface energy of the host where such huge energy difference creates a capillary force that induce embedding of surface adatoms by the host surface [10]. In this respect, the surface energy of Co (2709 mJ/m2) [11,12] is almost fifty times higher than graphene (46.7 mJ/m2) or graphite flakes (54.8 mJ/m2) [ 13 ]. Burrowing the Co will keep the surface energy minimum which in this case preserving the surface energy of graphene. Similar reason has been put forward for gold (Au) burrowed underneath bismuth (Bi) surface where Au has lower surface energy than Bi [10]. Surface intercalation of metal atoms beneath graphene monolayer has been reported only recently and literature on such work is limited. Shikin et al. reported intercalation of Au underneath a graphene monolayer that pre-grown on Ni(111) surface [14]. Unlike our system, their surface intercalation occurs via annealing of Au overlayer deposited at room temperature. Based on their spectroscopy studies, they predicted this Au form uniform 2D-islands at the interface between graphene and Ni(111). However, the presence of clusters was not reported. A recent work using STM and STS by Bremlal et al. reported similar intercalation behaviour by Au between graphene and SiC(0001) [15]. Similar to Shikin et al. [14], the Au were first deposited on graphene/SiC(0001) at room temperature and annealed to 727oC. They reported that the Au does not interact with graphene; similar to our XPS results where metallic Co is observed. Unlike Refs. 14 and 15, we found that annealing the 3D dome-shaped Co clusters that formed at room temperature does not result in surface intercalation. Hence, the kinetic pathways leading to surface intercalation of Co for graphene/Co/SiC(0001) differ from those found for graphene/Au/SiC(0001). 244 Co adsorption at elevated temperatures 6.5 Conclusions Growth at elevated temperatures induced ordering of Co on graphene/6HSiC(0001) surface. In contrast to 3D dome-shaped Co clusters at room temperature, large multilayer 2D islands are found formed on graphene surface for growth temperatures chosen in present works i.e. between 450oC and 750oC. Besides multilayer islands, two other smaller features are also found formed on the surface i.e. clusters with apparent height less than Å (hereafter dim cluster) and thicker clusters with height between to Å but with width smaller than dim cluster (hereafter bright cluster). STM images captured from various as-deposited surfaces show that the dim clusters have the tendency to clutter around the flat islands and depleted zones of dim clusters surrounding the flat islands are also observed. Post-growth annealing shows increased of flat islands density while the cluster density decrease rapidly. This relation clearly suggests that the dim clusters are mediating the nucleation and growth of flat islands. Scaling analysis performed onto the size distribution of flat islands shows a i*= 0, suggesting the dim clusters are smallest stable nuclei on the surface and upon incorporation of nearby clusters or capturing mobile adatoms, they will grow into flat islands. Both STM and XPS also revealed that the flat islands are burrowed underneath graphene monolayer indicating formation of flat islands required penetration of Co adatoms through the graphene lattice. The driving force for this unique kinetic process is believed to be instigated by the desire of system to minimise the surface energy. The bright clusters on the other hand also contribute to the formation of flat islands although their contribution is indirect and requires dissolution at higher temperature. Similar to dim cluster, they are also formed beneath the graphene monolayer. 245 Chapter Their formation are resulted from aggregation of Co adatoms that are trapped between the graphene monolayer and 63 interface. Due to this unique graphene/Co/SiC structure, graphene provides natural protecting layer of only one atomic layer thick for Co islands against ambient oxidation. XPS revealed that the Co remained in metallic states with the 4s+3d electron states similar to those bulk polycrystalline Co. Coupled with high mechanical strength and transparency of graphene, this structure may be useful for storage and recording applications. 246 [...]... significantly (Fig 6. 11f) Both of the density and size of Co flat islands continue to increase with annealing temperatures (Figs 6. 11c and 6. 11d) The evolution of flat island and cluster density as a function of annealing temperatures are provided in Figs 6. 11i and 6. 11j respectively It is evident that the initial rapid increase of island density from 500oC to 60 0oC is accompanied with rapid decrease of cluster... respectively according to Eq (6. 1) 6. 2.4 Energetics of Co flat islands Scaling of size distributions similar to Section 5.4 was also performed on the flat islands to deduce i* The volume distributions of the Co flat islands for two growth temperatures i.e 450oC and 63 5oC is shown in Fig 6. 15a where Ns is the number of island with size v (volume) and NT is the total number of islands sampled for the distribution... (a) 6 3 5 nm (b)(ii) (b)(i) 2.42 Å 1 nm 1 nm (c) (d) 2 nm 2 nm Fig 6. 21 (a) STM image of Co flat island grown on (6 3x63)R30o/6H-SiC(0001) surface at 61 0oC Similar to Co /graphene surface grown on elevated temperature, the Co island on 6 3 surface is also covered by a layer of graphene as depicted in (b)(i) where the honeycomb structure of graphene is clearly noticeable The self-correlation of this graphene. .. poly 279 72.5 538 5 36 534 532 530 Binding energy (eV) 528 5 26 Fig 6. 19 Angle-resolved XPS (Al K 14 86. 6eV) of Co /graphene after exposed to air for more than 6 months: (a) valence band, (b) Co 2p3/2, (c) Si 2p, (d) Si 2s, (e) C 1s and (f) O 1s Co was deposited at 1 .6 x10-2 ML/s for 100 minutes on a graphene surface held at 61 0oC where 80% of the graphene is covered by Co flat islands as shown by the... height of the bright and dim clusters as the bias changes 2 16 Co adsorption at elevated temperatures 6. 2 Formation of flat islands: Cluster-mediated epitaxy This section will focus on probing the role of bright and dim clusters on formation of Co flat islands STM images of Co adsorbed at different temperatures and post -growth annealing will be examined The Arrhenius behaviour of the cluster and island... minutes for growth between 550oC to 700oC The STM images for this series of experiment using the same Co flux used in Figs 6. 5a are shown in Figs 6. 14a-d The density of bright and dim clusters are counted and plotted according to Eq (6. 1) They are shown in Figs 6. 14e and 6. 14f respectively In Figs 6. 14a-d, both size and density of the clusters reduces with temperature implying the effective coverage of Co... function of take-off angle are analysed as shown in Fig 6. 20 (b) 100 80 Atomic percentage (%) Atomic percentage (%) (a) 100 C 1s 60 40 Si 2s 20 80 C 1s (284.5 eV) 60 40 20 Co 2p3/2 Co 2p3/2 0 0 20 40 60 o Take-off angle ( ) 80 20 40 60 80 o Take-off angle ( ) Fig 6. 20 Angle-dependant atomic percentage of (a) C 1s, Si 2s and Co 2p3/2 and (b) C 1s (284.5 eV, graphene component) vs Co 2p3/2 Figure 6. 20a... form dim or bright clusters and eventually the formation of flat islands 229 Chapter 6 (a) 18 (b) -12.0 -12.5 ln N sat ln J 17 16 15 -13.0 -13.5 (2/3)Eeff = 0.19  0.1 eV Eeff = 0.27  0.1 eV -14.0 14 -14.5 0.8 1.0 1.2 1.4 1000/T(K) 1 .6 0.8 1.0 1.2 1.4 1 .6 1000/T(K) Fig 6. 16 Arrhenius plot of (a) J and (b) Nsat J and Nsat are values from Table 6. 1 As the formation of flat island appears primarily to be... islands STM studies of Co growth (Section 6. 2.1) and annealing experiments (Section 6. 2.2) suggest that the Co adatoms penetrate the graphene monolayer where nucleation of dim clusters which mediates the formation of flat islands occurs at the 6 3 interface and beneath the graphene surface The implication is that the Co features, in particular the large flat island are covered by graphene Direct evidence... is seen in Fig 6. 18a where filled-state image shows an area of an unadorned graphene with 6 3 at the interface (upper left) next to a flat island The lattice of graphene above the 6 3 interface is clearly visible and more significantly this graphene monolayer is seen to be a continuous layer which covers the flat island as indicated in Fig 6. 18b This graphene monolayer as shown in Fig 6. 18c is also . Both of the density and size of Co flat islands continue to increase with annealing temperatures (Figs. 6. 11c and 6. 11d). The evolution of flat island and cluster density as a function of annealing. site of bright and dim cluster relative to the structure of 6 3. The bright clusters reside on top of the corner of 6x6 maxima while the dim clusters reside on top of the “hole” (3 clusters) of. (min) Island density (x10 9 / cm -2 ) (b) N sat J Fig. 6. 4 A series of STM images showing growth of Co islands on graphene held at (a) 585 o C, (b) 63 5 o C and (c) 720 o C. Co flux used is (1.6