Kinetic Monte Carlo Simulations of One-

dimensional Nanowires Grown From Co-

deposition of In and Sn on Si(100) surface

Darwin B. Putungan
Physics Division, Institute of
Mathematical Sciences and
Physics, UPLB
Outline of the Presentation
• Brief Summary of the Work
• Introduction
▫ Computational materials modeling and simulation
▫ The Problem
▫ Significance of the Study / Point of Emphasis
• Brief Review of Literature
▫ One-dimensional atomic wires
▫ The Kinetic Monte Carlo Method
• Methodology
▫ Simulation Model
▫ Model Parameters
• Results and Discussions
▫ Chemical Selectivity
▫ Dimer Ordering
• Summary and Conclusion
• Remarks Regarding the Work
Brief Summary of the Work
• A growth model for co-deposition of Sn and In
on Si(100) at room-temperature (RT) was
simulated using Kinetic Monte Carlo (KMC).

• This was done in order to shed light on the

chemical selectivity and lack of dimer ordering
seen in [Jure et al., App. Surf. Sci. 162
638(2000)], a Scanning Tunneling Microscopy
(STM) study.
Introduction
• Computational materials modeling and
simulation
▫ Very important tool in the understanding of
materials’ properties, both in the atomic and
macroscale regimes.
▫ Bridge the gap between theory and
experimentation.
Introduction
• Computational materials modeling and
simulation
▫ Of utmost necessity in the following cases:
 When experimental analysis is too expensive or too
impractical .
 In the design of new materials – new
nanostructures, fuel storage, catalysis systems, etc.
Introduction
• The Problem – In-Sn co-deposition experiments
established the following:

i. Mixed In-Sn are rare

i. “Chemical selectivity”.
ii. Number of mixed In-Sn dimers - hardly affected
by an increase in In flux relative to Sn by 100:1.
• They thought increasing In flux rate could increase
the number of mixed dimers
• They were wrong!
Introduction
• The Problem –

▫ One goal of the Jure, et al. study was to fabricate:

 Nano-alloy
 Useful as contacts in atomic-scale devices.
 But as presented, mixed In-Sn dimers were found to
be negligible.

▫ That is one big problem.

Introduction
• The Problem – previous In-Sn co-deposition
experiments established the following:

ii. Analysis of the surface morphology – no

ordering of dimers (i.e., In-In and Sn-Sn
alternation).
Introduction
• The Problem –

▫ Another goal of the team of Jure, et. al:

 Fabricate ordered atomic wires,
 alternation of In-In and Sn-Sn dimers.
 Possible solution in controlling important electrical
properties in atomic-scale devices such as conductivity
and resistivity.
 But the team did not observe any evidence of ordering.

▫ Another big problem.

Introduction
• The Problem, in short, can be stated in two
simple questions:

1) Why does the number of mixed In-Sn dimers

very small? Why is it not affected by flux rate?
2) Why there is no ordering? How can one quantify
this seemingly qualitative observation?
Introduction
• Significance of the Study / Point of Emphasis

▫ KMC modeling:
 Useful in providing insights into the key atomistic processes
controlling growth.

▫ Knowledge from experiments + Knowledge from KMC =

 Better control of nanostructures.

▫ Could pave the way for the fabrication of new

nanostructures:
 Nano-alloys - useful in the much-anticipated atomic-scale
devices.
Introduction
• Brief Review of Literature

▫ Nanowires – too big for atomic-scale devices.

▫ Needs one-dimensional atomic chains / wires /

islands
 One atom wide and few atoms long.

▫ Observed in depositing Group III and IV atoms on

Si(100) (e.g., Al, In, Ga, Pb, Sn)
Introduction
• Brief Review of Literature

1D Pb atomic wires on Si(100)-2x1 (L. Jure et. al., Phys. Rev. B 61 2000)
Introduction
• Brief Review of Literature

1D Sn atomic wires on Si(100)-2x1 (A. A. Baski et. al, Phys. Rev. 44 1991 )
Introduction
• Brief Review of Literature

Co-deposited In and Sn 1D wires on Si(100)-2x1 (Jure et. al., Appl. Surf. Sci.
162-163 2000)
Introduction
• Brief Review of Literature – Kinetic Monte Carlo
(KMC)

▫ A class of Monte Carlo methods

 Used in the study of large systems of interacting
particles at the long time scale.
Introduction
• Brief Review of Literature – Kinetic Monte Carlo
(KMC)

▫ Bortz-Kalos-Liebowitz (BKL) Algorithm

 The most common implementation of KMC
simulation
Introduction
• Brief Review of Literature – Kinetic Monte Carlo
(KMC)
Pi  ni ri
Where Pi is the probability
ri that a certain process ri will
ri occur; ni is the number of
species capable of executing
ri the process ri .
ri

P 1 i
Methodology
• The atomistic lattice-gas model

▫ Model created in order to simulate the formation and

growth of 1D atomic wires.

▫ Pioneered by MA Albao [Phys Rev B 72 035426 (2005)]

▫ Subsequently used by other groups in their 1D structures

researches, some of them are:
 Javorsky et al., WDS 08 Part III, 95-100,2008,
 Ostadal et al., JPCS 100 (2008) 072006,
 RB Stinchcombe Phys. Rev B 77 (2008)
 Tokar,Physica B: Condensed Matter 398, 2, 2007.
Methodology
• The atomistic lattice-gas model

▫ In and Sn atoms on Si(100) surface tend to form

dimers.
 Minimum stable configuration on the surface.

▫ Parallel Dimer Model

Methodology
• The atomistic lattice-gas model

The parallel dimer model. Filled circles denote the adatom

species and unfilled circles denote silicon atoms. Small
unfilled circles denote second layer silicon atoms
Methodology
• The atomistic lattice-gas model – Features

▫ The following atomistic processes were included

in the simulation:
 Deposition
 Diffusion of adatoms
 Nucleation of adatoms (and growth of the atomic
wires)
 Decay (hopping of atoms from already grown atomic
wires).
Methodology
• The atomistic lattice-gas model – Features

▫ Si(100)-2x1 surface:
 square lattice, n x n, of adsorption sites.

▫ Atomic hopping/diffusion rates:

 Governed by Arrhenius law h   exp  Eact / k BT 
 v is the attempt frequency (v = 1013 /s), T is the temperature, Eact =
activation energy.

▫ Eact:
 For the mentioned atomistic processes (such as hopping, diffusion, etc.)
were calculated from the experiments done by P. Kocan et al., Surf. Sci.
(2007) and Density Functional Theory (DFT) calculations done by our
collaborator Dr. Feng Chuan-Chuang of the National Sun Yat Sen
University in Taiwan.
Methodology
• The atomistic lattice-gas model – Model
parameters
▫ Uniform flux rate for each metal:
 10-4 ML/s (1 ML = 6.78 x 1014 atoms/cm2 )
 Except for the case where In:Sn flux rate ratio is
100:1, where Fin = 10-2 ML/s.

▫ C-type defect density (serves as anchoring sites):

 Constant throughout the simulation
 Set to 0.006 / site
Methodology
• The atomistic lattice-gas model
Illustration of the atomistic
lattice-gas model adopted in
our simulations. Gray cubes
represent adatoms which
comprise an atomic row
while blue cubes are rapidly
diffusing free adatoms.. The
black squares representing
C-defects act as anchoring
site for adatoms to nucleate
into islands. Gray squares
are aggregation sites,
whereas red signs indicate
blocked or forbidden sites.
Results and Discussions
• Chemical Selectivity

▫ The interaction energy between In and Sn atoms

were varied from and the number of mixed In-Sn
dimers were monitored.

▫ This was done for both 1:1 and 100:1 In to Sn flux

ratio.
Results and Discussions
• Chemical Selectivity

Ratio of mixed dimers to homogeneous dimers.

Results and Discussions
• Chemical Selectivity

▫ Observations from simulations:

 Experimentally observed scarcity of mixed dimers is consistent
with relatively weak attraction or mild repulsion between the
atoms.
 Same observation for In:Sn flux rate ratio 100:1.

▫ We then postulated that:

 Number of mixed dimers does not depend on relative flux.
 It is controlled by the number of Sn-terminated islands.

▫ If this is true, the number of Sn-terminated islands should

be largely unaffected by relative flux rate.
 Results and Discussions
• Chemical Selectivity
• Indeed, that is so the case.

• The unaffected number of Sn-

terminated islands kept the number
if impingement between In and Sn.

• As a consequence, mixed In-Sn

dimers remain roughly constant.

• Simply flooding the surface with In

does not generally lead to an increase
in mixed dimers.

Number of Sn terminated atoms as a function of the

lateral interaction energy.
Results and Discussions
• Dimer Ordering

▫ We need to devise a way of quantifying dimer

ordering in order to study it.

▫ With that in mind, we created an ordering

correlation function we defined to be

g x   hx  hx  1 
Results and Discussions
• Dimer Ordering

g x   hx  hx  1 
▫ g(x) is the ordering correlation function
▫ h(x) gives the assigned numerical value to a dimer
type located at some arbitrary chosen site x
▫ h(x+1) is the numerical value of the dimer type
located at the nearest neighbor (nn) site
Results and Discussions
• Dimer Ordering

g x   hx  hx  1 
▫ How this works?
▫ If the chain is perfectly ordered (perfect
alternation of In and Sn), g = 1 for all x.
▫ If perfectly no ordering (homogeneous In or Sn), g
= 0 for all x.
▫ For random combination of In and Sn, 0<g<1.
Results and Discussions
• Dimer Ordering
1D nanowire with perfect
ordering; g = 1 for all dimer pairs

1D nanowire with perfect no

ordering; g = 0 for all dimer pairs

1D nanowire with arbitrary

ordering; g varies for every dimer
pair.
Results and Discussions
• Dimer Ordering – Important 1D wires labels:

Combined In(Sn) and Sn(In)

odd-sized atomic wire

Odd-sized In(Sn) atomic

wires
Results and Discussions
• Dimer Ordering

▫ The lack of ordering could be the effect of the

energy difference between the activation energies
of the mentioned atomic wire types.

▫ This was according to the researchers who did the

experiment.
Results and Discussions
• Dimer Ordering ΔE = 0.0 eV , Ecombined odd-sized ≤ Eodd-sized

1
0.9
0.8

Ordering Correla tiom

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9

Dimer pair
Results and Discussions
• Dimer Ordering ΔE = 0.40 eV , Ecombined odd-sized ≤ Eodd-sized

1
0.9
0.8

Ordering Correla tion

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9
Dimer pair
Results and Discussions
• Dimer Ordering ΔE = 0.00 eV , Ecombined odd-sized ≥ Eodd-sized
1
0.9
0.8

Ordering Correla tiom

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9

Dimer pair
Results and Discussions
• Dimer Ordering ΔE = 0.40 eV , Ecombined odd-sized ≥ Eodd-sized
1
0.9
0.8

Ordering Correla tion

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9
Dimer pair
Results and Discussions
• Dimer Ordering

▫ No ordering was observed for both cases.

▫ Thus, that the activation energy difference

between the two atomic wire types does not affect
ordering.
Results and Discussions
• Dimer Ordering

▫ We then hypothesized that the energy associated

with the odd-sized atomic wire types could induce
ordering.

▫ Let us see.
Results and Discussions
• Dimer Ordering E = 0.35 eV , Eodd-sized
1
0.9
0.8

Ordering Correlation
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9

Dimer pair
Results and Discussions
• Dimer Ordering E = 0.15 eV , Eodd-sized
1
0.9
0.8

Ordering Correla tion

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9
Dimer pair
Results and Discussions
• Dimer Ordering E = 0.00 eV , Eodd-sized
1
0.9
0.8

Ordering Correla tion

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
G1 G2 G3 G4 G5 G6 G7 G8 G9

Dimer pair
Results and Discussions
• Dimer Ordering – Insights

▫ Ordering depends on the activation energies of the

atomic wire types.

▫ These energies depend on the kind of the atom.

▫ Thus, the implication of this is that considerable

ordering cannot be induced for In and Sn
combination.

▫ Our suggestion – try other atoms.

Summary and Conclusion
• An atomistic lattice-gas model describing the growth
of 1D nanowires through In and Sn co-deposition
was developed.

• Two key experimental observations, chemical

selectivity and dimer ordering, were considered in
the study.

• We showed that in order to explain the apparent

chemical se-lectivity which manifested in the near-
absence of mixed dimers, the In-Sn binding must
have been slightly repulsive.
Summary and Conclusion
• This is turn, suggests instability of mix In-Sn
dimers.

• Dimer ordering is not governed by activation energy

difference between atomic wire types.

• It depends on the activation energy of the atomic

wires themselves.

• This implies that ordering depends on the atomic

species to be deposited.
Remarks Regarding the Work
• Research articles related to the work:

 Accepted in the 5th Asian Consortium

on Computational Materials Science
(ACCMS);
 Undergone first-round revision for
publication in the journal International
Journal of Modern Physics B.
“Innovation starts with basic research,
and to be used in applied research”

Thank you!