Magic clusters on group IV surfaces 7

7.1.2 Si magic clusters on Si111 From our results, Si magic clusters can be grown via 2 methods; spontaneous nucleation from high temperature heating/quenching of Si111 surface or select

Chapter 7: Conclusion

7.1 Summary

In this section, we will provide a summary of our key findings which will address the research objectives in the following the work;

1) Si magic clusters on 6H-SiC(0001)

2) Si magic clusters on Si(111)

3) Co-Si magic clusters on Si(111)

Following which, we will also briefly address the future work related to this work

7.1.1 Si magic clusters on 6H-SiC(0001)

In this work, we have established the occurrence of Si magic clusters on 6H-SiC(0001) This structure exists as one of the phases as the 6H-SiC(0001) surface undergoes phase transformation with progressive annealing from (3x3) → (6x6) clusters

→ (6x6) rings This process is facilitated by the formation of Si magic clusters which first occurs with Si popping out from the (3x3) surface as tetra-clusters (size~5.0±0.5Å), driven by the breaking of highly strained co-planar Si-Si bonds These tetra-clusters agglomerate to form larger clusters of uniform shape and size (~14.3±0.5Å), which self assembly to form (6x6) unit cells at higher temperatures This process is motivated by the minimization of dangling bonds and a consequent lowering of surface energy We propose a hexagonal cluster structure model consisting of 8 tetra-clusters or 32 Si adatoms to account for the observed cluster shape, size and (6x6) periodicity

We also show the formation of (6x6) rings from the (6x6) clusters at higher temperatures and consequently establish the absence of the (6√3x6√3) R30º reconstruction from line profile and auto co-relation analysis Although our XPS data shows that Si is lost from the surface during annealing, however the ring surface is still Si-rich and not graphitized STM data shows that smaller type “A” clusters (tetra-clusters, size~5.0±0.5Å) form from type “B” clusters (size~14.3ű0.5) and leads to the formation of (6x6) ring structure (size~15.5 Å ±0.5) consisting of type “C” Si adatoms (size~3.0ű0.5) when the surface is heated beyond 1000°C This shows that type “A” tetra-clusters participate in the phase transformations by mediating the surface structural

transformation We propose a model to account for this surface evolution through the selective removal of M x 4 Si atoms (Where M=number of clusters) at various annealing temperatures By removing M=19 per unit cell, we obtain a structural model of the ring structure consisting of locally organized type “C” Si adatoms arranged into hexagonal D1 and D2 formations, which accounts for the experimental observations

7.1.2 Si magic clusters on Si(111)

From our results, Si magic clusters can be grown via 2 methods; spontaneous nucleation from high temperature heating/quenching of Si(111) surface or selective growth from Si adatoms deposited on Si(111) by Si atom source

(I) Heating /quenching of Si(111)

In the first method, in addition to being able to form Si magic clusters from heating/quenching, we have also established the origin, formation and structure of Si magic clusters formed spontaneously on Si(111)-(7x7) in UHV STM data shows that uniformly shaped Si magic clusters of size ~ 14.0 ± 0.5Å are formed during the “1x1”→ (7x7) phase transformation Real time data shows that these clusters form from excess Si adatoms which are expelled from the surface This is attributed to the “1x1” phase having

a greater atomic density of 45 more Si atoms per (7x7) unit cell than the (7x7) phase Thus, Si magic clusters originate from the excess Si adatoms which have to be removed from the “1x1” structure in order to facilitate the structural change to (7x7) Si magic clusters are also shown to be mobile as evidenced by the cluster trails coexisting with the

“1x1” and (7x7) domains Hence diffusion of Si on Si(111) occurs via Si magic clusters which facilitates the transport of excess Si from the “1x1” regions to the step edges and consequently propagating the (7x7) domains Formation of Si magic clusters is therefore critical to the “1x1” to (7x7) transition as they accommodate the excess atoms during structural change and facilitate mass transport during (7x7) domain nucleation From the

dual biasing data, Si magic clusters of average size ~ 14.0 ± 0.5Å are found to comprise

of 3 sub-unit features of size ~4.5±0.5Å each, arranged in a isosceles triangular configuration with an average separation of ~7.5 ±0.5Å, ~5.7 ±0.5Å and ~5.7 ±0.5Å

(II) Growth of Si magic clusters from atom source

In the second part of this work, we have addressed how to fabricate stable mono-disperse magic clusters from Si adatoms deposited on (7x7)-Si(111) by;

(i) Demonstrating the self assembly of Si magic cluster via the use of an atom source instead of a cluster source In doing so, we avoid growth issues related to inconsistent cluster size and shape distribution typically attributed to the use of cluster source techniques

(ii) Demonstrating the stability and self assembly phenomena of Si magic clusters to form spatially well–ordered cluster arrays at low temperatures (<500oC) thereby establishing Si magic clusters as a new phase within the framework of Si nanostructures which has potential for device applications

(iii) Elucidating the growth mechanism of the Si magic cluster, which is shown to occur via a step-wise addition of tetra-clusters as fundamental building blocks, where X2

(comprising of n = 4 Si atoms) is formed from X1, X3 (comprising of n = 8 Si atoms)

is formed from X2 + X2 and X4 (comprising of n = 12 Si atoms) is formed from X2 +

X3 We also determine the respective relative formation energies from quantitative data to be E(X2) = E(X1)–0.09eV, E(X3) = 2E(X2)–0.05eV and E(X4) = E(X2)+E(X3

)-0.22eV It is unexpected that Si tetra-clusters are fundamental building blocks in this growth process and not Si adatoms as anticipated

(iv) Determining the structure of magic clusters in terms of its size (~ 13.5±0.5Å), shape (3 tetra-clusters of size ~ 4.5±0.5Å occupying TS positions arranged in a closed packed isosceles formation with separations of 5.7Å, 5.7Å and 7.5Å) and number of

Si atoms (n = 12)

We also compare Si magic clusters on Si(111)-(7x7) with Si magic clusters on 6H-SiC(0001) and noted the following;

• Both types of clusters are formed from excess Si adatoms expelled onto the surface when the substrate is heated to high temperatures

• This formation occurs during phase transformations on Si(111) and 6H-SiC(0001) Hence Si magic clusters play a significant role in mediating phase transformations

on different Si surfaces

• Both types of Si magic clusters (heat/quench on Si(111) vs Si magic clusters on SiC), nucleate via the assembly of Si tetra-clusters Therefore Si tetra-cluster structure function as building block units in the assembling Si magic cluster structures

• Si magic clusters on SiC are not mobile and are observed to exist with (6x6) periodicity Although Si magic clusters grown on Si(111)-(7x7) exhibit localized ordering on Faulted Half Unit Cell sites, however these clusters participate in (7x7) island formation and (7x7) domain propagation and are hence mobile in nature

• While clusters on both surfaces are round in shape, however the clusters on SiC hare slightly larger than clusters on S(111)

Thereby we have established Si magic clusters on Si(111) and 6H-SiC(0001) as a new phase within the framework of Si nanostructures, where Si tetra-clusters are fundamental building blocks in this growth process The observation of tetra-clusters grown from Si atoms deposited on Si(111) also suggests a natural tendency for Si adatoms to come together to form a Si4 cluster The agglomeration of the tetra-clusters to form larger Si magic clusters on both Si(111) and 6H-SiC(0001) surfaces also establishes the Si4 structure as a critical building block for the growth of Si magic clusters on Si(111)

7.1.3 Co-Si magic clusters on Si(111)

In this work, we have addressed how to grow a well-ordered array of mono-dispersed

Co-Si magic clusters as follows;

1) We find that θ=0.5ML is a critical Co coverage where there is no island formation

at any annealing temperature Higher coverages (θ=1.0ML or 2.0ML) lead to

island formation at various annealing temperatures

2) Formation of Co-Si magic clusters occurs after annealing RT deposited Co on Si(111) to 460oC These clusters are 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

3) We can selectively grow a full layer of Co-Si magic cluster array with (√7x√7)

periodicity on top of Si(111)-(7x7) by annealing θ=1.0ML of Co coverage to

610oC

4) At low coverages (θ=0.5ML), Co-Si magic clusters 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 In particular, the i=6 cluster configuration consisting of 6 clusters

can be seen in U-shaped formation, ring formation or closed packed hexagonal

structure The i=7 configuration are either closed packed hexagonal structures or

U-shaped formations

5) The number density of cluster configuration as a function of annealing temperature (460oC, 490oC and 530oC) shows that i=1 dominate the surface with some i=2, 3 and 4 at 460oC while i=6, i=7 and i>7 increase in occurrence at 490oC

At 530oC, there are even less single or paired clusters while more clusters now

exist as i=6, i=7 or i>7 Therefore STM data shows a preferential occurrence of i

= 6 and 7 over i = 1 to 5 at higher temperatures

6) Real time study of the dynamical behavior of these magic clusters shows evidence

of individual cluster diffusion and formation of i = 4, 5, 6 and 7 configurations

with different lifetimes leading to the self organization and ordering on a Si(111)-(7x7) template The self assembly process does not occur spontaneously and a configuration dependent critical nuclei (i*=6) exists The smallest stable configuration is one consisting of seven Co-Si magic clusters arranged in a

hexagonal closed packed formation (i = 7) This hexagonal closed packed

formation not only maximizes cluster co-ordination but also minimizes surface dangling bonds on an underlying Si(111)-(7x7) template Growth of cluster structures is translated via diffusion, attachment and self alignment of a magic clusters instead of adatoms Hence (√7x√7) structure is not a surface reconstruction but is due to self assembly of Co-Si magic clusters

7) Based on the dimensions of the various cluster features obtained from the line profile measurements under different tunneling biases, we propose the structure of the Co-Si magic cluster to consist of 3 Si adatoms with a uniform separation of 4.5Å sitting on top of 6 Co atoms arranged in a triangular configuration This cluster structure occupies a circular area of diameter ~9.0Å and sits directly above the rest-atom layer, on the hollow sites of the (7x7) DAS structure

8) We propose a diffusion mechanism which accounts for the motion of a single

Co-Si magic cluster gliding over the Co-Si(111) surface without disrupting the (7x7)

structure This mechanism involves (i) exchange of Si atoms between the top layer of the cluster and the underlying adatom layer of the (7x7)-DAS structure and (ii) the sequential breaking and formation of bonds between the base Co atoms of the cluster with the (7x7)-DAS Si rest atoms, to account for the diffusion

of Co-Si magic clusters observed on Si(111)-(7x7)

7.2 Future Work

From this work, we propose the following areas of interest for future work

1) In the elucidation of the respective structures of Si magic clusters on 6H-SiC(0001) and on Si(111) as well as Co-Si magic clusters on Si(111), the ball and stick models proposed are a result of the consideration of dangling bonds and comparison of sizes of clusters observed from STM measurements However the experimental observations are confined within the constraints of the STM measurements, as the existence of artifacts or distortion due to electronic effects could affect the accuracy of our size measurement Nevertheless, we used a slow scan speed through a large bias range to eliminate drift and electronic effects in ensuring accurate topographical representations of the surface features However, a first principle calculation of the structures would be useful as part of future work in confirming the cluster structure Considering the large of number of atoms and different possible types of bonding configurations, a theoretical approach similar to

that taken by Que et al [1-2] may be necessary in order to solve this structure efficiently

2) In the formation of (6x6) rings from the (6x6) clusters at higher temperatures on 6H-SiC(0001), XPS shows that although Si is lost from the surface during annealing, the ensuing ring surface is still Si-rich and not graphitized As we have demonstrated that large areas of the Si-rich (6x6) ring structures may be fabricated, it would be of interest to probe if this structure would eventually lead to the formation of graphene via further heating of the surface and inducing graphitization of SiC

Thus far graphene has been generated typically from the micro-mechanical cleaving of graphite [3] However, this approach has been plagued by inconsistent film quality, thus affecting the fabrication of graphene for practical electronic applications [4-5] Our proposed method of controlled graphitization of SiC to form large areas of graphene would be significant in overcoming the growth issues associated with typical mechanical approaches to graphene formation and thus allow us to tap its much feted electronic properties for further applications

As it is unclear as to how each layer of the SiC substrate develops from a Si-rich

to a graphitic surface according to high temperature treatment, our future work will address the nature and formation of this phase using STM and XPS In particular, we will study if the progressive loss of Si from the heating of the SiC surface results in surface structural change which follows the empirical rule of removing M x 4 Si atoms (where M=number of clusters) As most of our work is done on 6H-SiC(0001) substrates, it would also be interesting to extend this to 4H substrates, where a different atomic stacking sequence exists This would allow us to see if this mechanism still operates under the same conditions

At the same time we will probe if the introduction of Hydrogen gas would further enhance the formation of wide graphene terraces from our approach of SiC graphitization

In order to develop this area of work further, we would also probe the growth of binary magic clusters on the graphene surface As graphene, being non-reactive in nature, is not expected to alter the properties of the clusters, it would thus be interesting for us to probe