Magic clusters on group IV surfaces 6

image also shows that the clusters at 700oC remain similar in size and shape compared to cluster observed after annealing at 460oC.There is no significant change in the surface morpholog

Chapter 6: Co-Si magic clusters on Si(111) (7x7)

So far most of the studies have been focused on the formation and behavior of single material magic clusters such as Si clusters on SiC and Si(111) Hence it would be

interesting to see if binary magic clusters can be formed and progressively ordered from

Co deposited on Si(111) This chapter discusses data from STM and XPS which is used

to probe the evolution of Co deposited at θ=2ML, θ=1ML and θ=0.5ML on clean Si(111)-(7x7) (where θ = Co coverage) at room temperature and progressively annealed

to higher temperatures We will first study the global morphology by characterizing the various surface structures such as 3D islands or clusters that could occur at different annealing temperatures as a function of Co flux We will also determine if there is critical

Co coverage where clusters exist exclusively or annealing temperature which promotes ordering and formation of (√7x√7) surface structures

While it is generally assumed that the ordering of clusters occurs spontaneously,

we will also investigate the self assembly phenomena of clusters to resolve if the (√7x√7) structure is an intrinsic surface reconstruction or due to self assembly of clusters It would also be interesting to determine whether a stable critical nuclei exists to promote this self assembly process, analogous to that in thin film growth We will do this through the use

of fast scanning STM, to probe directly the process leading to the formation of (√7x√7) Finally, dual biasing STM will be used to analyze the shape and size of the various clusters as a function of tunneling voltage This information will allow us to elucidate the

structure of Co-Si clusters as well as better understand the diffusion mechanisms involved in cluster motion on the Si(111)-(7x7) surface

6.1 Global surface evolution of Co-Si on Si(111)

6.1.1 Global Morphology: 2.0ML of Co deposited on Si(111)-(7x7)

Clean Si(111)-(7x7) was first obtained by flashing the Si(111) sample to 1200oC

in-situ, before it is being cooled to room temperature and scanned by STM Fig 6.1(a) shows the 100nmx100nm STM image of this surface revealing well ordered (7x7) reconstruction 2.0ML of Co was then deposited via a solid Co source electron beam evaporator onto this surface at room temperature (RT) The surface is subsequently annealed to 460oC, 610oC, 700oC, 800oC and 900oC for 30min each time STM and XPS scans of the surface are taken after each annealing cycle and are shown in Fig 6.1(b) to Fig 6.1(f) (STM) as well as Fig 6.2 (XPS) respectively

The STM scan of the surface annealed at 460oC in Fig 6.1(b) shows the formation of bright hexagonal features which we identify as islands with flat top surfaces possessing an average size and height of ~ 10±5nm and ~ 30±5Ǻrespectively In order to characterize these 3D islands, we zoom-in on a high resolution 17.5nmx17.5nm STM scan of a typical hexagonal island as shown in Fig 6.1b(i) The scan reveals that the surface of the island consists of a well ordered array of round protrusions These features appear smaller than the cluster-like features observed to be surrounding the island Line profile measurements X and Y along the [ ]1 1 and [ ]211 azimuths show the average separation between adjacent protrusions features is estimated to be ~ 7.6±0.5Å This is twice the lattice parameter of Si(111)-(1x1) (3.8Ǻ), which indicates that the protrusions

possess a (2x2) periodicity.We also estimate the height of the island to be ~ 15.0±0.5Å (Ht) and also find smaller cluster-like features on the surface which appear to be disordered as indicated in the STM image Fig 6.1(b)(ii) shows a zoom-in 15.0nmx15.0nm STM image of these cluster-like features The line profile measurements

X and Y, show that each cluster has an average size of ~ 9.0±0.5Å and height of 1.6±0.1Å (Z) The scan also shows that these features are uniform in shape and size Hence we identify them as Co-Si magic clusters

When the surface is annealed to 610oC for 30min, the STM scan of the surface as shown in Fig 6.1(c), reveals that the 3D islands retain the same hexagonal shape, but are now larger in size and height (30±5nm and 30±5Ǻ) More interestingly, we now observe the appearance of bright and dark triangular domains coexisting alongside the 3D islands Zoom-in STM images of the surface as shown in Fig 6.1(c)(i) show that the bright domains are covered with the same clusters while the darker regions are due to the (7x7) reconstruction This suggests that the clusters appear to be highly mobile and could have diffused to form triangular domains, thus exposing the underlying (7x7) reconstruction

Further annealing of the surface to 700oC, as shown by STM in Fig 6.1(d) shows that the hexagonal 3D islands are larger in size and height than before (50±10nm and 50±5Ǻ), whilst its number density appears to be decreasing The scan also shows that the size of the darker (7x7) triangular domains are now also larger and its number density is also smaller as observed in the zoom-in scan shown in Fig 6.1(d)(i) The high resolution

image also shows that the clusters at 700oC remain similar in size and shape compared to cluster observed after annealing at 460oC.

There is no significant change in the surface morphology when the sample is heated to 800oC (Fig 6.1(e)), however it is interesting to note that the exposed (7x7) domains tend to occur in areas adjacent to the 3D islands We also observe that the clusters still retain their shape and size in spite of the higher annealing temperatures When the surface is annealed to 900oC, the cluster-like features are no longer observed

In fact, the 3D islands are now observed to be larger (70±10nm) and appear to be more triangular in shape rather than its original hexagonal shape as previously seen, as observed in Fig 6.1(f)

If we analyze the crystallographic direction of each of the 6 facets of the original hexagonal shape of the island are[ ]11 , 2 [ ]1 1 , [ ]211 , [ ]11 , 2 [ ]1 1 and[ ]211 , we find that the

[ ]11 , 2 [ ]1 1 and [ ]211 planes appear to be longer than the [ ]11 , 2 [ ]1 1 and[ ]211 planes This suggests that the growth speed along [ ]11 , 2 [ ]1 1 and[ ]211 , is faster than [ ]11 , 2 [ ]1 1and [ ]211 which accounts for the island shape transition from hexagonal to triangular as shown in Fig 6.3 It is anticipated that during island growth, the island formation will tend to adopt a shape with the lowest energy configuration, thereby exposing the planes with the lowest surface energy Hence it is not surprising that the stereographic projection

of Si (111) in the [111] azimuth show that the lower surface energy facets exist along the

[ ]11 , 2 [ ]1 1 and [ ]211 planes respectively as illustrated in Fig 6.3

The XPS spectra as shown in Fig 6.2a and Fig 6.2b corresponds to the Co 2p3/2and Si 2p core-level spectra taken from the XPS scans of the surface after each annealing cycle The XPS scan of the surface when Co is deposited at RT shows that the binding energy of Co 2p3/2 (778.5±0.1eV) is close to that of a pure metallic Co (778.3±0.1 eV) rich environment [1] The peak shape also appears asymmetric and is akin to a metallic-like Co environment The corresponding binding energy of Si 2p is shown to be 99.2±0.1

eV, which is similar to clean Si substrate (99.2±0.1eV) [1].When the surface is annealed

to 460oC, there is no noticeable energy shift in the Si 2p peak However, we detect a shift

in the binding energy of Co 2p3/2 to 778.9±0.1eV, which is close to the binding energy of

Co in a bulk CoSi2 crystal structure (778.8±0.1eV) This indicates the formation of Co-Si when the sample is annealed to 460oC, suggesting that the 3d islands and cluster features observed are Co-Si and not metallic Co We do not observed significant changes in the binding energies of the Si 2p peak (99.2±0.1eV) or the Co 2p3/2 peak (778.9±0.1eV) respectively when the surface is further annealed to 610oC or 700oC

From the global morphological and high resolution STM scans as well as XPS scans, we see that when 2.0ML of Co is deposited onto Si(111) at RT and annealed to

460oC, hexagonal-shaped Co-Si islands (size~ 10±5nm) form immediately The island surfaces are flat and possess a (2x2) reconstruction Co-Si magic clusters of average size and height of ~ 9.0±0.5Å and ~ 1.6±0.1Å are also observed to occur in areas adjacent to the islands At higher temperatures (610oC, 700oC and 800oC), the islands appear to grow larger while the clusters now form into triangular domains coexisting alongside the islands, thus exposing the underlying (7x7) reconstruction Since the clusters possess

uniform size and shape and they do not grow larger in size at higher temperatures, hence

we identify them as Co-Si magic clusters When the surface is annealed to 900oC, the 3D islands appear to be more triangular in shape and the clusters are no longer observed

From the experimental evidence, we have demonstrated that binary material

Co-Si magic clusters can be formed However these clusters are found to co-exist with 3D islands Hence it would be of interest to see if we would be able to preferentially form a surface dominated exclusively by magic clusters, devoid of 3D islands In the next few sections, we attempt to achieve this by lowering the Co coverage used from 2.0ML to 1.0ML and eventually 0.5ML

Figure 6.1: 100nmx100nm STM images of (a) clean Si(111)-(7x7)

at RT and after 2.0ML of Co deposited at RT and annealed for 30min at

(b) 460oC (c) 610oC (d) 700oC (e) 800o (f) 900oC

Figure 6.1(b)(i) shows 17.5nmx17.5nm image of the surface at

460oC and line profiles X and Y across the island surface in the

[ ]1 1 and [ ]211 azimuths show that the average separation between

protrusions is ~ 7.6±0.5Å Island height (Ht) is estimated to be ~ 15.0±0.5Å

Figure 6.1(b)(ii) shows a zoom-in 15.0nmx15.0nm scan of the

clusters at 460oC Line profile X, Y and Z show that the average size and height of each cluster is ~ 9.0±0.5Å and ~ 1.6±0.1Å

Figure 6.1(c)(i) shows a 30nmx30nm zoom-in of the surface

at 610oC Size and height of cluster is ~ 9.0±0.5Å and ~ 1.6±0.1Å

Fig 6.1(d)(i) shows a 30nmx30nm zoom in of surface annealed at

0 1 2 3 4 5 0

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 0

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.5 1 1.5 2

0.5 1 1.5

X[nm]

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0

0.2 0.4 0.6 0.8 1 1.2

X[nm]

0 0.5 1 1.5 2 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

1 2 3 4 5 6 7 8

0.5 1 1.5 2 2.5 3

X[nm]

0 1 2 3 4 5 6 0

0.5 1 1.5 2

X[nm]

3.5nm (2x2) Clusters

Y X

0.2 0.4 0.6 0.8 1

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.2 0.4 0.6 0.8 1

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 0

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.5 1 1.5 2 2.5

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.5 1 1.5 2

0.5 1 1.5

X[nm]

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 0

0.2 0.4 0.6 0.8 1 1.2

X[nm]

0 0.5 1 1.5 2 0

0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

1 2 3 4 5 6 7 8

0.5 1 1.5 2 2.5 3

X[nm]

0 1 2 3 4 5 6 0

0.5 1 1.5 2

X[nm]

3.5nm (2x2) Clusters

Y X

0.2 0.4 0.6 0.8 1

X[nm]

0 0.5 1 1.5 2 2.5 3 3.5 0

0.2 0.4 0.6 0.8 1

Figure 6.2 shows the XPS signal of (a) Si 2p signal at 99.3eV and (b) Co 2p signal at

778.9eV when Co (θ=2.0ML) is deposited at RT and annealed for 30min to 460oC, 610oC and 800oC

050000100000

Binding energy (eV) (a)

(b)

Figure 6.3 shows schematic diagram of (a) hexagonal island with 6 planes and (b) triangular island with 3 planes which are more exposed (c) illustrates the island growth evolving from hexagonal to triangular shaped islands

200nm

[121]− −[211]−

6.1.2 1.0ML of Co deposited on Si(111)-(7x7)

We deposit 1.0ML of Co on a clean Si(111)-(7x7) surface at room temperature (RT) and anneal the surface for 30min to 460oC, 610oC, 700oC , 800oC and 900oC, before scanning the surface with STM and XPS each time In this section, we discuss the morphological evolution using large scale 100nmx100nm (Fig 6.4) and zoom in 15nmx15nm (Fig 6.5) STM images together with the XPS data (Fig 6.6)

Fig 6.4a and Fig 6.5a both show the initial starting Si(111) surface consisting of well ordered (7x7) reconstruction Fig 6.5b shows the STM image of Co deposited at RT and annealed to 460oC We again see observe cluster-like features of size 9.0±0.5Ǻ, as shown by line profiles X and Y, which is similar to the ones previously observed We also find that the surface is dominated by these clusters with very little areas of (7x7) reconstruction exposed These clusters also appeared to be disordered with no sign of island formation The zoom in image seen in Fig 6.5b is also consistent with the large scale scan However, we also note that some of the Co-Si magic clusters tend to form into localized 2D hexagonal closed packed cluster configurations, as indicated by the dark circle shown in Fig 6.5b

Upon annealing to 610oC, as seen in Fig 6.4c, the global morphology appears largely unchanged, where Co-Si magic clusters still dominate the surface and there is no formation of 3D islands However, the high resolution STM scan in Fig 6.5c now shows that the Co-Si magic clusters are now arranged into a well ordered 2D array of clusters in

contrast to the disordered state previously observed This suggests that the clusters are likely to be mobile on the Si(111) surface at these temperatures The line profile measurement of the cluster-cluster separation, as illustrated in Fig 6.5g, shows that the average distance between adjacent clusters is ~ 10.5±0.5Å This suggests that the clusters possess a periodicity of √7 times that of (1x1)-Si(111) It is interesting to note that while the (√7x√7) structure has been previously reported [2], it has been largely assumed that this structure is a surface reconstruction due to an intrinsic rearrangement of the top few Co-Si layers on the surface and there has been in fact no observation/mention of Co-Si clusters However our observations of the highly mobile nature of Co-Si clusters on the Si(111)-(7x7) surface and their tendency to form hexagonal closed pack structures suggests that a self assembly of Co-Si clusters to form 2D cluster configurations could lead to the formation of well ordered (√7x√7) periodic structures

When the surface is annealed to 700oC, as shown in Fig 6.4d, we observe the formation of islands for the first time These islands are about ~ 20±5nm and are hexagonal in shape and its surrounding surface is still largely covered by Co-Si clusters The zoom-in scans in Fig 6.5d show that the clusters are no longer as well ordered, and are in fact exposing more of the underlying (7x7) reconstruction

When we anneal the surface to 800oC (Fig 6.4e), we observe that the 3D islands still retain its hexagonal shape, however the Co-Si clusters have now gathered into localized triangular domains, as seen by the formation of dark and bright triangular areas The corresponding high resolution scan of this surface in Fig 6.5e shows that the dark

and bright regions is due to exposed (7x7) reconstruction and Co-Si magic clusters occupied areas respectively

Upon the final anneal of the surface to 900oC, as shown in Fig 6.4f, we still see the occurrence of 3D islands, which now appear to be more triangular-like as opposed to its original hexagonal shape However we do not observed the dark and bright triangular regions as previously seen The zoom-in images in Fig 6.5f shows that the surface surrounding the 3D islands consists exclusively of the (7x7) reconstruction and the Co-Si magic clusters are not observed

The XPS spectra as shown in Fig 6.6a and Fig 6.6b corresponds to the Co 2p3/2and Si 2p core-level spectra taken from the XPS scans of the surface after each annealing cycle When Co is initially deposited at RT, the XPS scan shows that the binding energy

of Co 2p3/2 is 778.9±0.1eV, which is close to the binding energy of Co in a bulk CoSi2crystal structure (778.8±0.1eV) This observation is in contrast to the earlier RT XPS

scan of the surface when θ=2.0ML of Co was deposited The binding energy of Co 2p3/2

was observed to be 778.5±0.1eV which indicated a metallic Co rich environment and only shifted to 778.9±0.1eV when annealed to 460oC This could be attributed to the

detection of larger amounts of un-reacted metallic Co when θ=2.0ML is used as opposed

to θ=1.0ML, which suggests that the formation of Co-Si takes place when Co is

deposited first at RT on Si(111) The corresponding binding energy of Si 2p is shown to

be 99.2±0.1 eV, which is similar to clean Si substrate (99.2±0.1eV) [1] We do not observed significant changes in the binding energies of the Si 2p peak (99.2±0.1eV) or

the Co 2p3/2 peak (778.9±0.1eV) respectively when the surface is further annealed to 460

oC, 610oC or 700oC Hence we are able to conclude that the clusters observed under STM are Co-Si and not metallic Co

Hence from the STM data shown, we are able to delay the onset of 3D island

formation by depositing a lower Co coverage of θ=1.0ML compared to θ=2.0ML as

shown earlier The islands only appear when the surface is annealed at 700oC and this is preceded by the occurrence of a single layer of Co-Si magic clusters at 460oC We avoided island formation and were able to promote ordering amongst the Co-Si magic clusters via annealing at 610oC This led to the formation of (√7x√7) ordering of the clusters In fact we also observe from high resolution scans of magic clusters forming localized 2D hexagonal closed packed configurations from previously disordered states This suggests that the clusters are mobile at this temperature and there is a tendency for these individual clusters to self assemble into localized 2D surface arrangements before eventually leading to longer range ordering of clusters Hence we deposit an even lower

amount of Co (θ=0.5ML) to investigate the self assembly process in the next section.

Figure 6.4 shows 100nmx100nm STM scans of the surface (a) consisting of initial (7x7)

reconstruction (b) with θ=1.0ML of Co deposited at RT and annealed to 460oC (c) annealed to 610oC (d) annealed to 700oC (e) annealed to 800oC (f) annealed to 900oC

Figure 6.5 shows 15nmx15nm STM scans of the surface (a) consisting of initial (7x7)

reconstruction (b) with θ=1.0ML of Co deposited at RT and annealed to 460oC (c) annealed to 610oC to form cluster array with (√7x√7) periodicity (d) annealed to 700oC (e) annealed to 800oC (f) annealed to 900oC (g) line profile X and Y shows that clusters have average size of 9.0±0.5Å (h) line profile Z across 11 clusters shows average cluster separation to be ~10.5±0.5Å, which indicates a (√7x√7) periodicity

0.2 0.4 0.6 0.8 1 1.2

Figure 6.6 shows the XPS signal of (a) Si 2p signal at 99.2eV and (b) Co 2p signal at

778.9eV when Co (θ=1.0ML) is deposited at RT and annealed for 30min to 460oC, 610oC and 800oC

0 50000

Binding energy (eV)

0 50000

(b)

6.1.3 0.5ML of Co deposited on Si(111)-(7x7)

As the previous data suggest that the Co-Si clusters could be mobile and appear to form localized 2D configurations, consequently we proceeded to deposit a lower coverage of Co onto the Si(111) surface in order study the self assembly phenomenon of the Co-Si magic clusters, and investigate if the (√7x√7) structure is indeed a surface

reconstruction or due to self assembly of clusters We deposit θ=0.5ML of Co on a clean

Si(111)-(7x7) surface at room temperature (RT) and anneal the surface for 30min to

460oC, 610oC, 700oC , 800oC and 900oC, before scanning the surface with STM and XPS each time We discuss the morphological evolution using large scale 100nmx100nm STM images (Fig 6.7) together with the XPS data (Fig 6.8)

Fig 6.7a shows the initial starting Si(111) surface consisting of well ordered (7x7) reconstruction When 0.5ML of Co is deposited at RT onto this surface and annealed to

460oC, as seen in Fig 6.7b, we observed that the surface is dominated by Co-Si magic clusters, albeit in much lower number densities compared to when higher Co coverage was used The clusters are seen to be randomly distributed over the Si(111)-(7x7) reconstructed surface with no observation of 3D islands The STM scan not only shows individual and paired clusters but also 2D gathering of clusters to form various

geometries consisting of different number of clusters, i At 460oC, individual clusters

(i=1), appear to dominate the surfacein spite of the existence of other configurations of

clusters such as i=2, i=3 or i=4

When the surface is annealed to 610oC, as shown in Fig 6.7c, global morphological scans do not show any island formation However, the clusters are

observed to have formed new 2D configurations such as ring formations (i=6), hexagonal structures (i=7) and domains of clusters where i>7 on the Si(111)-(7x7) surface In fact the cluster configurations of i=1, i=2, i=3 and i=4 observed earlier appear to occur in

lower number densities

When we further anneal the surface to 700oC, as seen in Fig 6.7d, we find that the clusters have gathered into localized 2D triangular domains where the number of clusters

per domain is typically i>7 The number density of the various cluster configurations such as the ring formations (i=6) and hexagonal structures (i=7) appear to have decreased, while the cluster configurations of i=1, 2, 3, 4 and 5 appear to be even more scarce

When the surface is annealed to 800oC, as shown in Fig 6.7e, the 2D triangular

domains appear to have become larger in contrast to the occurrence of i≤7 cluster

configurations which continue to decrease even further This again suggests that the clusters prefer to come together to form the triangular domains observed This observation is consistent with the STM data shown in Fig 6.7f, where the surface upon annealing to 900oC, shows even larger areas of triangular domains consisting of clusters However unlike the surface evolution shown in Fig 6.1 and Fig 6.4, it is interesting to note that no formation of 3D islands is observed thus far

The XPS spectra as shown in Fig 6.8a and Fig 6.8b corresponds to the Co 2p3/2and Si 2p core-level spectra taken from the XPS scans of the surface after each annealing cycle When Co is deposited at RT, the XPS shows that the binding energy of Co 2p3/2 is 778.9±0.1eV, which is similar to the binding energy of Co in a bulk CoSi2 crystal structure where binding energy of Co 2p3/2 is 778.8±0.1eV This observation is consistent

with the previous RT XPS scan of the surface when θ=1.0ML of Co was deposited and differs from the metallic Co rich environment when θ=2.0ML of Co was deposited

When the surface was annealed to 460oC, 610oC and 700oC, XPS data does not reveal any significant energy shift in the Co 2p3/2 peak Hence the clusters observed under STM are Co-Si in nature and retain their chemical character with progressive annealing at higher temperatures The corresponding binding energy of Si 2p is shown to be 99.2±0.1

eV, which is similar to clean Si substrate (99.2±0.1eV) [1] We also do not observed significant changes in the binding energies of the Si 2p peak (99.2±0.1eV) when the surface is further annealed to higher temperatures

From the experimental data tracing the surface evolution of 0.5ML of Co deposited and annealed on Si(111), we observe the occurrence of the Co-Si magic clusters, but we do not see the formation of 3D islands at any annealing temperature This

is in contrast to the previous 2 Co coverages used (θ=2.0ML and θ=1.0ML), where 3D

island formation was detected after annealing at 460oC and 700oC respectively Hence we

find that θ=0.5ML is a critical coverage where formation of 3D Co-Si islands can be

avoided and Co deposited on Si(111) exists exclusively as Co-Si magic clusters upon annealing This has not been seen in previous work which typically shows the formation

of 3D islands or layers with bulk Co2Si or CoSi2 structures when Co is deposited and annealed on Si(111) Hence we are able to grow a new phase of Co-Si in the form of magic clusters within a 2D single layer on top of the Si(111)-(7x7) via careful control of deposition rate and substrate temperature

Figure 6.7 shows 100nmx100nm STM scans of the surface (a) consisting of initial (7x7)

reconstruction (b) with θ=0.5ML of Co deposited at RT and annealed to 460oC (c) annealed to 610oC (d) annealed to 700oC (e) annealed to 800oC (f) annealed to 900oC

20 nm

Figure 6.8 shows the XPS signal of (a) Si 2p signal at 99.2eV and (b) Co 2p signal at

778.9eV when Co (θ=0.5ML) is deposited at RT and annealed for 30min to 460oC, 610oC and 800oC

If we summarize the STM and XPS data of Co deposited at different coverages on Si(111)-(7x7), we find that;

(a) when θ=2.0ML of Co is deposited;

We observe the formation of clusters and islands with (2x2) reconstructed surfaces when the Si(111) surface is annealed to 460oC The clusters are found to be uniformly round in shape, possess an average size and height of 9.0±0.5Å and 1.6±0.1Å, and do not increase

in size when annealed at higher temperatures The binding energy of Co 2p3/2 at 460oC is shown to be 778.9±0.1eV, hence we identify the clusters as Co-Si magic clusters

(b) when θ=1.0ML of Co is deposited;

The same clusters dominate the (7x7) surface in the absence of islands at 460oC These clusters self organize into localized hexagonal closed pack configurations and eventually form into arrays with (√7x√7) periodicity at 610oC The cluster form into triangular domains while 3D islands are observed at 700oC as compared to annealing at 460oC when

to grow a 2D phase of Co-Si in the form of magic clusters on top of Si(111)-(7x7) This

is surprising as it is expected to form 3D islands or layers with bulk Co2Si or CoSi2structures when Co is deposited and annealed on Si(111) [2]

From the STM data, we also observe the formation of Co-Si magic clusters which

group into various 2D cluster configurations of i = 1, 2, 3, 4, 5, 6 and 7, where i is the number of clusters within each formation The number density of i = 1 cluster configuration appears to be the largest, followed by i = 2, 3, 4 and 5 initially at 460oC However progressive annealing of the surface resulted in the gradual decrease of these

cluster species, whilst the number densities of i = 6 and 7 are shown to increase The

clusters are subsequently observed to group into localized 2D triangular domains where

i>7 as the surface is further annealed It is important to note that the clusters do not coalesce into 3D features as no island formation is detected at any time, instead they are observed to gather into even larger 2D domains at higher temperatures

From the STM observations of cluster evolution as a function of temperature, the

clusters exhibit a tendency to come together to form small cluster configurations of i<7

before forming larger domains at this coverage (θ=0.5ML), while the previous coverage

(θ=1.0ML) shows that the same clusters form localized closed packed configurations leading to eventual (√7x√7) ordering This suggests that the clusters can diffuse and self assemble from single entities to form intermediate cluster configurations leading to ordered cluster arrays It would therefore be of interest to study how this process of cluster self assembly propagates cluster ordering more closely Hence we maintain a low

Co coverage (θ=0.5ML) and use high resolution STM to further probe the self assembly

phenomena of Co-Si magic clusters in the following section

6.2 Co-Si magic clusters

6.2.1 Types of cluster configuration and number density as a function of

temperature

In order to study the different combinations of cluster configurations more closely,

we deposit θ=0.5ML of Co onto Si(111)-(7x7) at RT and anneal to 460oC, before using high resolution 15nmx15nm STM snap shots to identify them, as shown by the dark circles indicated in the panel of images shown in Fig 6.9

Fig 6.9a shows a single Co-Si magic cluster, which we denote as i=1, which is

observed to sit preferentially on the brighter Faulted Half of the (7x7) reconstruction Fig

6.9b shows a cluster pair (i=2), which typically consists of 2 adjacent clusters with a separation of ~ 9.0±0.5Å We show that i=3 cluster trimers which consists of 3 clusters,

are observed to be arranged as 3 clusters in a row in Fig 6.9c(i) and (ii) or in a triangle

arrangement with 3-fold symmetry in Fig 6.9c(iii) We also identify the i=4 cluster

configuration in Fig 6.9d, comprising of 4 clusters arranged in (i) triangular geometry, (ii)

in a straight row and (iii) U-shaped configuration Fig 6.9e shows the i=5 cluster

configuration with 5 clusters arranged in (i) a row with a maximum of 2 nearest neighboring clusters (ii) closed packed hexagonal structure and (iii) triangular-like

formation The i=6 cluster configuration is shown in Fig 6.9f where 6 clusters can be

found in (i) U-shaped formation (ii) ring formation or (ii) closed packed hexagonal

structure The final cluster configuration of i=7 is shown in Fig 6.9g, where the clusters

are arranged in closed packed hexagonal structures in (i) and (ii) or U-shaped formations where clusters are strung across in a row

In order to investigate the number density trend of the various cluster configurations more closely, we counted the number density of each configuration when the surface (θ=0.5ML) is annealed to 460oC, 490oC and 530oC as seen in the 100nmx100nm STM scans in Fig 6.10(a)-(c) The evolution of cluster configuration as a function of annealing temperature is represented in the corresponding histograms next to

the STM scans The statistical data show that individual (i=1) clusters dominate the surface with some i=2, 3 and 4 at 460oC (Fig 6.10(a)) At 490oC (Fig 6.10(b)), ring

formations (i=6), hexagonal structures (i=7) and domains of clusters where i>7 are

observed to increase in occurrence When the surface is annealed to 530oC (Fig 6.10(c)),

we observe occurrences of even less single or paired clusters while more clusters now

exist as ring (i=6), hexagonal (i=7) or in cluster domain (i>7) configurations The STM data therefore shows that there is a preferential occurrence of i = 6 and 7 over i = 1 to 5 at

higher temperatures We note that cluster density is conserved throughout the various annealing temperatures, which suggests that the clusters are mobile and exhibit an energetic preference to gather into localized 2D ordered arrangements

From the data, we observe that these clusters self organize to form various 2D cluster configurations of different geometries and cluster numbers The experimental data

shows that there is a preferential formation of hexagonal closed packed i =7 cluster configurations from the initial i =1, 2, 3, 4, 5 and 6 at higher temperatures which lead eventually to growth of i >7 cluster domains with (√7x√7) periodicity In fact reported

works so far in general have assumed that the ordering of clusters occurs spontaneously