Magic clusters on group IV surfaces 2

Figure 2.4: Adatom or molecular site models on a SiC0001 surface: T1 represents the atop site; T4 represents the 3-fold site with an additional neighbor in the low-half top bilayer; H3 r

Chapter 2: Literature Review

2.1 Si Magic Clusters on 6H-SiC(0001)

2.1.1 Silicon Carbide

Silicon carbide as a substrate material has unique properties such as high temperature stability, thermal conductivity, fast carrier recombination and large band gap, which has led to the development of SiC based semiconductors electronic devices and circuits These applications are used in high-temperature, high power and high radiation conditions under which conventional semiconductors cannot adequately perform [1-3] such as the high temperature MOSFET [4] and the blue LED [5] In particular, the current push for continuing device miniaturization, have led to great interest in nano-structure formation on SiC surfaces

Silicon carbide is the only chemically stable form of silicon and carbon The basic unit of SiC consists of a covalently bonded tetrahedron of Si (or C) with a C (or Si) at the centre The bonding of silicon and carbon atoms is 88% covalent and 12% ionic with a distance of 1.89 Å between the Si and C atoms [6] Each Si(C) atom is surrounded by four C(Si) atoms respectively in strong tetrahedral sp3-bonds as shown in Figure 2.1 The crystalline structure consists of the close-packed stacking of bilayers of Si and C atoms parallel to the surface One set of atoms (Si or C) is shifted along the main axis of

symmetry by a quarter of the distance between the nearest similar layers The stacking sequence of the double layers follows one of three possible relative positions, arbitrarily labelled A, B and C as in closed packing of hard spheres One unique aspect of SiC is the stacking sequence of these double layers allows an endless number of different one-dimensional orderings (polytypes) without variation in stoichiometry This is the source

of SiC’s large number of crystallographic forms called polytypes Different crystal forms

of the same chemical composition are called polymorphs Polymorphism commonly refers to a three-dimensional change affected by either a complete alteration of the crystal structure or a slight shift in bond angles Polytypism is a special type of polymorphism that occurs in certain close-packed structures

Figure 2.1: Tetragonal bonding of carbon atom with the four nearest silicon neighbours The distances between 2 Si atoms and C-Si are about 3.08 Å and 1.89

Figure 2.2: Side view of different examples of SiC polytype crystal structure, along the )

0

11

( -direction (a) Hexagonal bilayer, (b) zincblende structure, 3C-SiC, (c) wurtzite structure, 2H-SiC, (d) 4H-SiC and (e) 6H-SiC The stacking direction is depicted by the enhanced Si-C bond train parallel to the (11 0)-direction (adapted from [7])

SiC exists not as a single crystal type but as a whole family of crystals known as polytypes which are named according to the periodicity of these layers For example, some of the other more common structures are, 4H and 3C (cubic lattice), as shown in Figure 2.2 [7] The exact physical, electrical and optical properties of SiC depend on the crystal structure adopted The stacking sequence of these polytypes can be described by the ‘ABC’ notation, where A, B and C represent the three sites available in separate

stacking layers in one sublattice For example, the cubic form of SiC called beta silicon carbide (β-SiC) or 3C-SiC, has an atom stacking sequence of ABCABCABC For

example, Schardt et al [7] studied the structure of the 3C and 6H polytypes and presented

a range of possible stacking sequences, which is shown in Figure 2.3 below He also investigated the arrangement of atoms on an ideally terminated surface and subsequently named the various site positions at the top, second and third layers T1, T4 and H3respectively This is presented in Figure 2.4

Figure 2.3: Cross-sectional view of the linear stacking in the cubic (3C) polytype (β-SiC) and three different staking sequences possible at the surface of the 6H polytype The different surface terminations are labeled according to the depth of orientation change (S1, CACBABC; S2, BCACBAB and S3, ABCACBA) (adapted from [7])

Figure 2.4: Adatom (or molecular) site models on a SiC(0001) surface: T1 represents the atop site; T4 represents the 3-fold site with an additional neighbor in the low-half top bilayer; H3 represents the 3-fold site without additional coordination (adapted from [7]).

While there are numerous polytypes (~250 different polytypes) that dictate different stacking sequences of the Si-C bilayer which results in varied and complex surface atomic arrangement as well as chemical identity, one of the more commonly used polytypes is the hexagonal 6H-SiC crystallographic bulk structure, which consists of stacking of Si-C bi-layers with the ABCACB sequence along the C-axis <0001> direction of the hexagonal structure and where each A, B or C corresponds to a SiC bi-layer, as shown in Figure 2.5 In an ideally bulk truncated configuration, the (0001) surface is terminated by Si atoms whereas the (0001)surface is terminated by C atoms

Si atom

C atom

Figure 2.5: Side view of 6H stacking sequence of 6H-SiC(0001)

As the surface structure and composition are strongly dependent on surface preparation conditions, the methodology of surface preparation is essential in determining the cleanliness, morphology and chemical make-up of the surface Hence the next section would be devoted towards reviewing the different approaches towards achieving clean and well ordered 6H-SiC(0001) surfaces in UHV Depending on annealing temperature, the 6H-SiC(0001) polytype exhibits a rich series of surface reconstructions ranging from Si-rich (3x3), (√3x√3)R30°, (5x5), (9x9), (2√3x2√13) to Si deficient (6√3x6√3) R30°

A

AB

C

B

CA

15Ǻ

These surface structures have been previously studied by a variety of techniques such as Low Energy Electron Diffraction (LEED), Reflection High Energy Electron Diffraction (RHEED), Auger Electron Spectroscopy (AES), Electron Energy Loss Spectroscopy (EELS) and STM [8-16] We would discuss the various literature in this area of work focusing on the significant surface structures which in turn would form the basis for our study of Si magic clusters on 6H-SiC(0001) substrate surfaces

2.1.2 SiC surface cleaning and preparation

SiC surfaces are strongly dependent on the methods in which the sample is prepared Conventional annealing in UHV yields a surface, which is considerably

carbonized or graphitised As early as 1975, Van Bommel et al [17] showed that

atomically clean ordered surfaces of 6H-SiC could be obtained by annealing in UHV at

800OC, but surface precipitation of graphite was observed even at this relatively low temperature, presumably associated with Si loss Annealing at higher temperatures required to volatilise surface oxides caused structural changes, which could not be reversed by additional annealing, since they are caused by the depletion of surface Si Subsequent work has confirmed that, while annealing in UHV can yield oxide-free surfaces, this preparation method suffers from three distinct disadvantages [18] Firstly, the results obtained are dependent on the type of sample Second, the surface phase transformations can proceed only in the direction of successively Si-poorer compositions Third, the most Si-poor phase was found to be accompanied by graphite precipitation and considerable disorder [18]

Since annealing in vacuum causes irreversible phase changes and can conceal low temperature phases, an alternative method of surface preparation based on the chemical

reduction of surface oxide at modest temperatures was applied Kaplan et al [19] reported

that virtually all oxide from 6H samples disappeared after several minutes exposure at

850OC to a Si flux With this method, Kaplan et al was able to obtain clean surfaces

without heating above 1000OC The arriving Si atoms convert surface oxide to SiO,

which is volatile in this temperature range, thus removing the oxide without otherwise disturbing the surface Contaminant C is similarly converted to SiC, thereby becoming part of the crystal structure By means of further Si exposure or annealing, the whole range of SiC surface structures from Si-terminated to C-terminated phases could be obtained and studied This method, although successful in obtaining clean SiC surfaces, does result in a high defect density We will consequently show that by adopting this method of surface cleaning we are able to prepare well ordered 6H-SiC(0001)-(3x3) reconstruction

2.1.3 6H-SiC(0001)-(3x3)

The 6H-SiC(0001) (3x3) reconstruction is a much-studied surface, with no less than 4 atomic structural models being proposed to describe it In this section, these 4 different models will be reviewed and later compared to the STM results obtained on (3x3) reconstruction

By using a Si flux to anneal Si-terminated 6H-SiC(0001) samples at 1120K,

Kaplan et al first observed the formation of a SiC(111)-(3x3) LEED pattern [19] Kaplan

identified the (3x3) phase with an adsorbed Si bilayer on the Si-terminated surface From

LEED and EELS results, Kaplan et al proposed a (3x3) structural model based on the

DAS (Dimer-Adatom-Stacking Fault) model of the Si(111)-(7x7) [20] This unit cell contains two adatoms per surface unit cell, six “rest” atoms in a first layer and eight atoms in a second layer on top of a silicon-terminated surface, as shown in Figure 2.6 The existence of the Si bilayer from the proposed EELS studies indicated a Si rich nature

of SiC(111)-(3x3) surface, and therefore the similarities between the Si(111)-(7x7) and the SiC(111)-(3x3) were assumed From previous inspection of the Si(111)-(7x7) model,

it was also shown that the same type of reconstructions can develop with a (2n+1)x(2n+1) periodicity (n=1,2,3,4….) [21] Periodicities (5x5), (9x9), (11x11) and (13x13) have been reported on Si surfaces [22-23] Therefore, a similar DAS reconstruction seemed a good candidate to account for the SiC(111)-(3x3) By modifying the stacking structure in the substrate, the DAS model also seemed to offer a good description for 4H- and 6H-SiC(0001)-(3x3) surface

However, similar studies by Kulakov et al involving the use of STM, showed that

the STM images has only one adatom in the unit cell [24], contrary to the DAS (3x3) structure which should give two well resolvable maxima per unit cell which would occupy (√3x√3)R30O positions As a consequence, a variant of the DAS model was proposed by Kulakov with one of the adatoms and its 3 nearest neighbours and the atom directly below removed This was done to suit the STM observation of the (3x3) surface

Figure 2.6: Schematic of initial atomic structure proposed by Kaplan for (3x3) reconstruction (adapted from [19])

An alternative is to view the (3x3) reconstruction model as from either C-rich or Si-rich tetrahedral clusters arranged in a (3x3) periodicity on top of the bulk SiC This

model was proposed by Li and Tsong et al in their STM studies where the imaging of the

empty and filled densities of states suggested the presence of C-rich and Si-rich tetrahedral distributed in a (3x3) geometry [25], presented in Figure 2.7 They observed a variety of surface features, bright spots, less bright spots and dark spots While the dark spots were attributed to the presence of vacancies, Tsong suggested that the different

Si

C

corrugation observed was due to isolated tetramers of various configuration sitting on the first layer of Si The bright protrusion were measured to have a diameter of 4Ǻ which is roughly the size of a tetrahedron, and a corrugation of 2.4Ǻ, which is close to the height

of a Si-C double layer Hence he concluded that the bright protrusions observed consisted

of a tetra-cluster with a C atom surrounded by 3 Si atoms or a Si atom surrounded 3 C atoms The possibility of a tetra-cluster made up entirely of Si or C atoms was also not ruled out

Figure 2.7: A structural model for the 6H-SiC(0001)-(3x3) reconstruction based on the STM images observed by Li and Tsong [25] It consists of (a) C-rich and (b) Si-rich tetrahedral clusters arranged in a 3x3 geometry as shown in (c) Both the (1x1) and the (3x3) unit cells are outlined in (c) (adapted from [25])

More recently U Starke et al [7, 26], determined the crystallographic structure of

the (3X3) reconstruction by joint application of quantitative LEED, STM, AES and holographic interpretation of the diffraction intensities By applying the density functional theory (DFT), he attempted to elucidate the structure by theoretical surface energy minimization A twisted Si adlayer reconstruction model was proposed, which is characterized by a full Si adlayer on top of the SiC substrate and a single Si tetramer per unit cell above is shown in Figure 2.8 Locally, the surface protruding atom resides in a

T4 site, resulting in bond angles (88.8°) to the trimer atoms below, deviating from the usual tetrahedral bond angles (109.5°) and bond lengths (2.47Ǻ) being expanded, compared to the bulk Si value (2.35Ǻ) He went on to show that the (3x3) phase is nearly planar, from its small maximum bucking amplitude of 0.27Ǻ Ideal tetrahedral bonding for the Si adlayer is absent, as the Si atoms have stretched or contracted bond lengths in different bond angles to achieve an optimised twisted surface with the lowest surface energy

Later theoretical work by Badziag [1] comparing the various suggested models for the (3x3) reconstruction supported only U Starkes’s model as the only energetically feasible model for the reconstruction In fact we will later show that the analysis of the STM observation of the (3x3) structure obtained in this study predominantly support U.Starke’s model and will use this structure as the starting platform to discuss the subsequent structural changes observed

Bottom layer:

Si-terminated 6H-SiC(0001)

in (1X1) position.

Middle layer: Si trimers

Top layer: Si adatom with one dangling bond

Top view of 6H-SiC(0001) (3X3) reconstruction

Twisted Si adlayer structural model proposed by

Top view of 6H-SiC(0001) (3X3) reconstruction Twisted Si adlayer

structural model proposed by U.Starke

3.08A

(3X3) Unit Cell

Figure 2.8: Ball and stick diagram describing atomic structure of the proposed U Starke

et al (3x3) model [7, 26]

IV atoms on Si(111) [27-28], he proposed that the (√3x√3) pattern might be due to 1/3 of

a monolayer of Si adatoms each bonding to 3 Si atoms of the SiC surface [19]

By annealing at 1170K, Owman and Mårtensson [29] observed a hexagonal array

of protrusions visible with some amounts of defects, using the STM The measured side length of the unit cell was 5.4Ǻ, which was in good agreement with the expected value of 5.3Ǻ for the (√3x√3) reconstructions They acquired STM images at bias voltages at both polarities, observing no significant difference in the appearance of the surface features The appearance of the STM images of the SiC(0001)- (√3x√3) surface strongly resemble those of the (√3x√3) adatom reconstruction observed for a 1/3 monolayer coverage of metals like Al, Ga, In or Pb on Si(111) [27-28], thus lending support to the model proposed by Kaplan [19] Hence, a possible configuration of the reconstruction is a similar type of adatom structure with either Si or C adatoms on top of a bulk like Si-C bilayer Owman then concluded that a Si adatom sitting on top of the T4 seemed most probable as this is observed for investigations of Si, Al or Ga adatoms behaving on Si(111) [30-31] and consequently this structural model was termed as the T4 model

They dismissed the notion that the T4 adatom might be C, as the larger covalent radii of

Si resulted in less bond angle stress, and hence would more probably be a Si adatom They also postulated that the existence of a C adatom on a H3 site could not be ruled out, and thereby resulting in a possible H3 model

More recently, Heinz et al [32]’s in-situ studies using STM, LEED and AES of

the (√3x√3) reconstruction is again best described by the T4 model Northrup and Negebauer [33] later identified the adatoms of the (√3x√3)R30O phase to be Si atoms in

T4 sites by first-principles total energy calculations This configuration was also found to

be the most thermodynamically favourable by Chang et al and Sabisch et al [34-35]

indicating that this surface is still Si rich at this substrate temperature A top and side view picture of the T4 model is presented in Figure 2.9 to allow us to better understand the atomic structure, where the top Si adatom is positioned at the T4 site above Si atoms

at the T1 site

Figure 2.9: T4 model: Top and side view of the 6H-SiC(0001)-(√3x√3)R30O surface The surface is terminated by a Si layer with an additional Si adatom in the T4 position [30-31]

T1 site

2.1.5 Si-rich 6H-SiC(0001)-(6x6)

Recently, Xie et al [36] observed a (6x6) reconstruction that consists of a Si-rich

phase on SiC(0001) that was obtained after annealing a Si-enriched (6x6) SiC(0001) surface at 900OC for a few minutes using RHEED By comparing AES data for (3x3), (6x6) phases with (√3x√3)R30 and (6√3x6√3)R30 phases, Xie et al found that both (3x3) and (6x6) phases are Si-rich Also, the Si auger peak for both the (3x3) and (6x6) patterns were similar, suggesting that both samples have a near-surface selvedge that is dominated by Si-Si bonding It was proposed that the (6x6) phase is an extension

6H-of the (3x3) tetrahedral reconstruction model with consecutive tetrahedral clusters missing, so that a long-range 6x6 periodicity is obtained As this is still an unreported surface structure, there were no proposed atomic model in the literature Recent STM

investigation of this Si-rich (6x6) by Ong et al [37] revealed the structure to consist of

clusters larger than the (3x3) tetra-clusters assembled in smaller domains as opposed to the long range ordering exhibited by the (3x3) reconstruction This work reported STM and XPS data that revealed the formation of an intermediate phase consisting of an ordered arrangement of Si clusters (diameter~14.3±0.5Å) which forms a unit cell with a (6x6) periodicity [37] These clusters possessed a uniform shape and size, which were attributed to Si magic clusters due to the unique dimensions and Si-rich nature of the surface This was surprising as the occurrence of magic clusters on SiC surfaces have not been reported before, nevertheless very little is understood with regards to the structure of these magic clusters as well as its formation and influence over surface morphological evolution

2.1.6 High temperature Si–deficient phases on 6H-SiC(0001)

Early observations in 1975 by van Bommel et al [17] of a (6√3x6√3)R30O LEED pattern was after annealing 6H-SiC samples in UHV at 1070K These authors proposed a model for the reconstruction which consisted of a single monocrystalline layer of graphite

on top of the bulk-terminated SiC crystal The resulting LEED pattern was explained by multiple scattering from the graphite and the underlying SiC surface

A (6x6) reconstruction pattern previously observed by STM on a localized range

by Tsukamoto et al [38] was initially attributed to the incommensurately adsorbed

graphite moiré pattern on SiC [17, 35] These structures were obtained after heating to temperatures well above 1200°C, and were reported to possess similar superstructures with rows running along the <11-20> direction with a unit cell of ~ 18Ǻ, superimposed

by a much denser grid with a unit cell of ~ 2.5Ǻ, as reported by Kulakov et al STM

observation [24, 39] The surface possessed long range order with much lower defect density than the (3x3) reconstruction and surface features could be described as having a honeycomb appearance This observed surface was popularly known as the graphitised (6√3x6√3)-R30o honeycomb structure

It was suggested to be possible to have a pseudoperiodic (6x6) structure where the surface unit cell has a cell edge that is exactly six times the Si-Si spacing, even though

the structure is non-periodic in two of the cell edges Chang et al [35] supposed that

LEED, which reflects long-range periodicity, would not see the “pseudoperiodic” 6x6 structure Instead, a larger periodicity in the form of 6√3x6√3R30 should be observed as

reported previously [17, 39 and 40] Through STM studies of the closely-related SiC(111) surface, a monocrystalline graphite layer was claimed to be directly observed

3C-on 6√3x6√3 rec3C-onstructed surfaces [35 and 41] Li and Ts3C-ong also reported an STM study of the surfaces created by Si evaporation onto the (6√3x6√3) surface where the results were interpreted as a confirmation of the graphite model for the (6√3x6√3) surface [25] Northrup and Neugebauer also further proposed that the (6√3x6√3) structure arises from a monocrystalline graphite overlayer on top of a (√3x√3) arrangement of Si adatoms

in view of the high stability calculated for the (√3x√3) structure with Si adatoms in T4positions [33]

There is much contention that the (6√3x6√3)-R30o reconstruction observed on diffraction techniques was not convincingly observed in real space imaging by STM An interesting study by Owman [40], using LEED and STM to probe SiC(0001) surface heated to 1000°C - 1200°C, revealed that although graphitisation did occur at temperatures beyond 1200°C, similar (6x6) structures could also be observed below that temperature with a significant trace of surface Si still present At this point, there was no indication of graphitic carbon within the C 1s spectra Therefore it would be more correct

to term this surface as Si-deficient rather than graphitised due to the observed loss of Si from the surface The observed the (6√3x6√3) LEED pattern was not detected on the STM, instead, (√3x√3), (5x5) and (6x6) structures were observed by the STM The LEED pattern was attributed to the scattering of a mixture of the STM observed (√3x√3), (5x5) and (6x6) structures Further heating of the surface above the temperature of 1250°C showed partial graphitisation and a modification of the (6x6) structure Owman

did not however propose a model to describe the formation or structure of the (5x5) or (6x6) reconstructions, which would be addressed later in the results and discussions section

2.1.7 Graphene

A single layer of graphite is called graphene due to the sp 2-bonding configuration

of the carbon atoms that are distributed in a hexagonal lattice This description is typically used for several layers of graphene up to about 10 layers, after which graphene with more than 10 layers result in bulk graphite [42] This form of material has attracted tremendous interest recently due to its unexpected physical and electronic properties For example, the electronic band structure of graphene shows a linear dispersion at the Fermi

energy at the K point of the surface Brillouin zone instead of a parabolic relation [43-44]

and the electron transport is governed by Dirac’s equation rather than Schrodinger’s equation [42, 45-46] In addition, graphene was also shown to have unconventional two-dimensional electron gas properties leading to quantum confinement [42, 45, 47], which

has led to prospective graphene based nano-electronics

Graphene had always been presumed to be thermodynamically unstable as a

free-standing layer [48] until Novoselov et al [45] used micro-mechanical cleaving of graphite

to obtain graphene Despite the success of this method, the most promising approach to obtain graphene for practical electronic applications has been identified to be the controlled graphitization of SiC for purposes of control over fabrication and quality of

film formation [49-50] While Chen et al have demonstrated that large areas of carbon

rich nano-mesh structures can be obtained from the heating of the 6H-SiC(0001) substrate [49], recent work have realized the absence of graphene-like electronic structure

from this phase and have in fact identified the less carbon-rich phase of (6√3x6√3) as possessing more significant graphene-like properties [51] This is because the onset of graphitization of the 6H-SiC surface is associated with the formation of the (6√3x6√3) phase However the nature of this surface reconstruction remains controversial as discussed in the previous section [51] In fact it is also unclear as to how each layer of the SiC substrate develops from a Si-rich to C-rich surface according to high temperature treatment Hence our work will address nature and formation of this phase using STM and XPS as outlined in Chapter 4

2.1.8 Summary

Despite the myriad of surface reconstructions observed and studied on the SiC surface, there has been very little coherent work discussing the evolution of these various structures and especially involving the occurrence of Si magic clusters In particular, the role of silicon adatoms or magic clusters in terms of their atomic re-arrangement, bond breaking and formation within the top few layers beginning with the SiC(0001)-(3x3) phase have been neglected Therefore we will address these issues related to Si magic clusters on 6H-SiC(0001) surface structure in Chapter 4 of this thesis

2.1.9 References

[1] P Badziag, Surf Sci., 822 (1998) 402-404

[2] P.J Lin-Ching, Bull Am Phys Soc 33 (1988) 298

[3] J.J Lander and J Morrison, Surf Sci 2 (1964) 553

[4] F Owman and P Mårtensson, Surf Sci 330 (1995) L639

[5] R Northrup, Phys Rev Lett 57 (1986) 254

[6] G.L Harris, Properties of Silicon Carbide, EMIS Data-reviews Series No 13,

INSPEC, the Institution of Electrical Engineers, London, United Kingdom, (1995) [7] J Schardt, J Bernhardt, U Starke and K Heinz, Phys Rev B 62, 15, (2000)

10335

[8] M Rohlfing and J Pollmann, Phys Rev Letters 84, 1, 2000

[9] V.M Bermudez, Appl Surf Sci 84 (1995) 45

[10] V van Elsbergen, T.U Kampen and W Monch, Surf Sci 365 (1996) 443

[11] T Tsukamoto, M Hirai, M Kusaka, M Iwami, T Ozawa, T Nagamura and T

Nakata, Surf Sci 371 (1997) 316

[12] I Forbeaux, J-M Themlin, V Langlais, L.M Yu, H Belkhir and J-M, Debever,

Surf Rev and Lett 5 (1998) 193

[13] X.N Xie, H.Q Wang, A.T.S Wee and K.P Loh, Surf Sci 478 (2001) 57

[14] M Naitoh, J Takami, S Nishigaki and N Toyama, Appl Phys Lett 75 (1999)

650

[15] X.N Xie and K.P Loh, Appl Phys Lett 77 (2000) 21

[16] W.J Ong, E.S Tok, H Xu and A.T.S Wee, Appl Phys Lett 80 (2002) 3406

[17] A.J Van Bommel, J.E Crombeen, A Van Tooren, Surf Sci 48, 2 (1975) 463

[18] P Kacknell, J Furthmuller and F Bechstedt, Diamond and Related Materials 6,

(1997), 1346-1348

[19] R Kaplan, Surf Sci 215 (1989) 111-134

[20] K Takayanagi, Y Tanishiro, S Takahashi and M Takahashi, Surf Sci 164

(1985) 367

[21] G.X Qian and D.J Chadi, J Vac Sci Tech B 4 (1986) 1079

[22] Y.N Yang and E.D Williams, Phys Rev Lett 72 (1994) 1862

[23] W Shimada, H Tochihara, T Sato and M Iwatsuki, Surf Sci 423 (1999) L291 [24] M.A Kulakov, G Henn, B Bullemer, Surf Sci 346(1996) 49-54

[25] L Li, I.S.T Tsong, Surf Sci 351 (1996) 141-148

[26] K Heinz, U Starke, J Bernhardt and J Schardt, Appl Surf Sci 162 (2000) 9-18 [27] P.J Lin-Ching, Bull Am Phys Soc 33 (1988) 298

[28] J.J Lander and J Morrison, Surf Sci 2 (1964) 553

[29] F Owman and P Mårtensson, Surf Sci 330 (1995) L639

[30] R Northrup, Phys Rev Lett 57, 254, 2986

[31] J.M Nicholls, B Reihl and J.E Northrup, Phys Rev B 35, 4137, 1987

[32] K Heinz, U Starke, J Bernhardt and J Schardt, Appl Surf Sci 162, 9-18, 2000

[33] J.E Northrup and J Neugebauer, Phys Rev B 52 (1995) R17001

[34] M Sabisch, P Krüger and J Neugebauer, Phys Rev B 58 (1998) 13712

[35] C.S Chang, I.S.T Tsong, Y.C Wang and R.F Davis, Surf Sci 256 (1991) 354 [36] X.N Xie and K.P Loh, Appl Phys Lett 77 (2000) 21

[37] E.S Tok, W.J Ong, H Xu and A.T.S Wee, Surf Sci 558 (2004) 145

[38] T Tsukamoto, M Hirai, M Kusaka, M Iwami, T Okawa, T Nagmura and T

Nagata, Surf Sci 256 (1991) 354

[39] M.A Kulakov, P Heuell, V.F Tsvetkov and B Bullemer, Surf Sci 315, 248-254,

1994

[40] F Owman and P Martesson, Surf Sci 369, 126 (1996)

[41] M.H Tsai, C.S Chang, J.D Dow and I.S.T Tsong, Phys Rev B 45 (1992) 1327 [42] A.K Geim and K.S Novoselov, Nat Mater 6 (2007) 183

[43] T Ohta, A Bostwick, T Seyller, K Horn and E Rotenburg, Science 313 (2006)

951

[44] K Emtsev, T Seyller, F Speck, L Ley, P Stojanov, J Riley and R Lecky, Mater

Sci Forum 556 (2006) 525

[45] K.S Novoselov, A.K Geim, S.V Morozov, V.I Fal’ko, M Katsnelson, I.V

Grigorieva, S.V Dubonos and A.A Firsov, Nature 438 (2005) 197

[46] M Katsnelson, K.S Novoselov and A.K Geim, Nat Phys 2 (2006) 620

[47] K.S Novoselov, A.K Geim, S.V Morozov, D Jiang, Y Zhang, S.V Dubonos,

I.V Grigorieva and A.A Firsov, Science 306 (2004) 666

[48] L.D Landau, Phys Z Sowjetunion 11, (1937) 26

[49] U Starke, Silicon Carbide – Major Advances, Springer, Berlin (2004) pp 281-316

[50] W Chen, H Xu, L liu, X Gao, D Qi, G Peng, S.C Tan, Y Feng, K.P Loh and

A.T.S Wee, Surf, Sci 596 (2005)176

[51] C Berger, Z Song, X Li, X Wu, N Brown, C Naud, D Mayou, T Li, J Hass,

A.N Marchenkov, E.H Conrad, P.N First and W.A Heer, Science 312, (2006)

1191

2.2 Si Magic Clusters on Si(111)-(7x7)

2.2.1 Si(111)

As a primary substrate material in the micro-electronics industry, the study of Si surfaces remains important, especially with the growing interest in the use of mono-dispersed particles in the fabrication of substrate supported nanostructures [1] Silicon has

a diamond structure, which is made up of silicon tetrahedral units, covalently bonded with an angle and bond length of 109.5° and 2.35Å, respectively The lattice parameter is 5.43Å [2] Figure 2.10 below illustrates the unit cell of bulk silicon

Figure 2.10: Diamond-structured silicon

There are 8 silicon atoms per unit cell, hence the volume density of silicon is calculated to be about 5.00 x 1022 atoms/cm3 The (001), (110) and (111) are the low-index planes The 3 low-index planes exhibit very different atomic arrangements in the ideally-terminated case In fact, their planar densities and density of dangling bonds differ

as well Among the 3 planes, (001) plane has the highest dangling bond density of 1.39 x

1015cm-2 In contrast, the (111) plane has only 8.00 x 1014cm-2, which is the lowest Hence, Si(111) plane is the most stable and easiest-cleaved plane of Si

In particular, the Si(111)-(7x7) surface has been widely studied by many investigators due to several factors; (1) It is the most thermodynamically stable crystal face (2) It is able to form the (7x7) reconstruction which is also a stable surface structure which most phases inevitably transform into (3) For STM investigations, the large unit cell and large corrugations are easier to image, while the stability against surface contamination makes the (111) surface easy to prepare (4) The Si(111)-(7x7) unit cell consists of faulted/unfaulted halves which possess available yet non-equivalent dangling bonds which in turn can function as preferential adsorption sites in promoting the self assembly of ordered nano particles

A cut through the covalent bonds of the bulk (111) plane would create two exposed surfaces with covalent bonds dangling in the direction normal to the surface, and into the vacuum In order to reduce the dangling bonds, the surface would adopt different structural reconstruction dependent on the type of treatments that the surface was subjected to The Si(111) surface for example is known to exhibit a variety of surface

reconstruction ranging from (2x1), (7x7), (9x9), (11x11), c(2x8), ( 3 x 3 ), (2x2) and c(4x2) under different sample preparation conditions However the most interesting and widely studied reconstruction is still the (7x7) unit cell which would be discussed in the following section

2.2.2 Si(111)-(7x7) reconstruction

In 1986, the Si(111)-(7x7) reconstruction was first observed by Binnig et al [3] as

a demonstration of the ability of the STM to image surfaces with atomic resolution This work led to the receipt of the Nobel Prize and generated much interest as the complex structural arrangement of the (7x7) confounded many in spite of the ease in preparing this surface structure In fact several structural models have been proposed to account for the

STM observation of (7x7) by Binnig et al, Chadi et al and Snyder et al [3-5], which have

unfortunately been less than convincing

It was Tersoff et al [6-7] and Tromp et al [8] who used theoretical calculations

based on the atomic charge superposition to compare the various proposed models with experimental STM images of the (7x7) reconstruction and found inconsistencies with

most of the models except for the one suggested by Takayanagi et al [9] This model

consisted mainly of a stacking fault in the outer double layer on half of the unit cell, which gave rise to the occurrence of corner holes, bounding twelve Si adatoms which were adsorbed on top of the Si(111) substrate The calculations based on this model agreed with the experimental results to within 0.1Å, which was a remarkable fit in view

of the approximations made during the calculations Thus consensus over this model as the best representation of the Si(111)-(7x7) surface reconstruction was reached The model was aptly named as the dimer-adatom-stacking fault (DAS) model, as it involved large displacements of atoms within a (111) double layer and introduced 12 adatoms per unit cell on top The DAS model was characterized by;

• Stacking fault in triangular half-unit cell

• Corner hole

• 9 dimers along boundary of faulted half unit cell

• 12 adatoms in T4 site with one angling bond each per unit cell

• 6 restatoms with one dangling bond each per unit cell

• bottom atom at the corner ring has one dangling bond

The side and plane view of this model generated from an ideally terminated Si(111) are

illustrated in the schematic as shown in Figure 2.11 and 2.12 respectively

(A) Side view of ideally terminated Si(111) surface

(B) Side view of Si(111)-(7x7)

Figure 2.11: Schematic diagram of the Si(111)-(7x7) surface (A) side view showing the

ideally terminated Si(111) surface (B) side view showing

<111>

2.349A2.15A

1.835A

1.435A

Dangling Bonds1st layer

2nd layer

3rd layer

1st layer adatoms

2nd layer restatoms

3rd layer dimers

4th layer bulk Si atoms

Figure 2.12: Schematic diagram showing plane view of the Si(111)-(7x7) reconstruction

The structural model of the (7x7) reconstruction consists of 4 layers of atoms The 4th layer is the bulk terminated (111) surface with one dangling bond per surface atom

As there are 64 of such atoms making up a (7x7) unit cell, this creates 64 dangling bonds within this layer These dangling bonds are saturated by the 3rd layer and forms dimer pairs as a result Hence this layer was termed as the dimer layer, where an atom in each corner of the (7x7) unit cell is also absent, giving rise to the observation of “corner holes”

as recorded by STM This layer is characterized by the formation of dimers, which connect the missing atoms, along the edge of the (7x7) unit cell and between the two triangular halves of the unit cell This occurrence of dimers at the edges of the (7x7) unit

Dimer pairs

Restatoms Adatoms

Bulk Si

<110>