TMCS
III, Leeds 18
th
Jan
2012
Slide
1
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
W
urtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
2
Semiconductors
Group IV: Si, Ge, C
Group III

V: GaAs, InAs, AlAs, GaP, InP, AlP, GaN, InN, AlN, GaSb, InSb, AlSb
Group II

VI: CdSe, ZnSe, ZnS, CdS, MgSe, ZnTe
InN
TMCS
III, Leeds 18
th
Jan
2012
Slide
3
Epitaxy
Epitaxy
: (Greek; epi "above" and taxis "in ordered manner")
describes an ordered crystalline growth on a monocrystalline
substrate.
Homo

epitaxy
(same layer and substrate material)
Hetero

epitaxy
(different layer and substrate material).
In the Hetero

epitaxy case growth can be:
Lattice Matched
: same, or very close, lattice constant of layer and
substrate
e.g. GaSb/InAs or AlAs/GaAs
Lattice Mismatched
: different lattice constant of layer and substrate
material e.g. InP/GaAs or InN/GaN.
TMCS
III, Leeds 18
th
Jan
2012
Slide
4
Lattice Matched and Mismatched Epitaxy
Lattice Matched
Lattice Mismatched
TMCS
III, Leeds 18
th
Jan
2012
Slide
5
Lattice Mismatched Epitaxy
Lattice Mismatched
In lattice mismatched heteroepitaxy the
layer material can be made to “adapt”
(can become smaller or larger) its in
plane lattice constant to match that of
the substrate (pseudomorphic growth).
Consequently volume conservation
(though volume is not perfectly
conserved) dictates that the lattice
constant in the growth direction needs
to become larger/smaller.
In this way the lattice periodicity is
maintained in the growth plane, but lost
in the growth direction.
TMCS
III, Leeds 18
th
Jan
2012
Slide
6
Quantum Mechanics in action
1D Quantum Well
In
0.52
Al
0.48
As
In
0.52
Al
0.48
As
In
0.84
Ga
0.16
As
AlAs
AlAs
Growth direction
Nanostructures: 2D Growth, 2D Growth + etching, 3D Growth
Green: Free Carrier
,
Red: Confinement
2D Multi Quantum Wires
Taurino et al Mat Sci and Eng B
,
67 (1999) 39
Scanning
Tunneling
Microscopy
TMCS
III, Leeds 18
th
Jan
2012
Slide
7
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
8
Elastic Strain
Semiconductors are produced by depositing liquid or gasses that
when coalesce and solidify follow the crystal structure of the “seed”,
usually a substrate of high crystalline quality.
During this deposition, often done in very small amounts (low growth
rate), as small as depositing one atomic layer at the time, if the layer
material has a bulk lattice constant larger than the substrate, then
the crystal will appear slightly deformed from its equilibrium state.
We refer to this material as “strained”.
Unstrained
Strained
We chose the axes vectors
x,y,z arbitrarily, but need to
be linearly independent.
Note that while the axes
vectors are chosen to be
unitary (in units of the lattice
constant) in the unstrained
case, the strained axes are
not necessarily unitary.
TMCS
III, Leeds 18
th
Jan
2012
Slide
9
Elastic Strain
Unstrained and strained axis can be easily related:
The numerical coefficient
ε
ij
define the deformation of all the
atoms in the Crystal.
The diagonal terms
ε
ii
control
the length of the axis, while the
off diagonal terms
ε
ij
control the
angles between the axis.
This picture is general
and valid for all types
of crystals, not just
simple cubic.
Unstrained
Strained
TMCS
III, Leeds 18
th
Jan
2012
Slide
10
Elastic Strain
Unstrained
Strained
This set of equations are in the
form of a mathematical entity
called Tensor.
The equations define the strained
position of any atom within the
crystal that upon strain moves
from
R
to
R’
.
Unstrained and strained
positions are written in terms of
the old and new axis:
Important: notice how the
coefficients
α
,
β
,
γ
are the same in the
unstrained and strained system
TMCS
III, Leeds 18
th
Jan
2012
Slide
11
Elastic Strain
We now substitute the new axis with the expressions for the distortion:
After a little manipulation and taking into account the expression for
R:
Provided the original position and the distortion tensor are known, this
expression gives a practical way of calculating the position of any atom
inside a strained unit cell.
TMCS
III, Leeds 18
th
Jan
2012
Slide
12
Strain Components
Often there is confusion between the terms strain and distortion. In
this lectures we follow the notation used in Jasprit Singh’s book, for
which the strain components e
ij
are different from the distortion
components but related to them by:
The final expressions for the off

diagonal terms e
ij
are an
approximation in the limit of small strain.
Dilation
: expresses how much the volume of the unit cell changes, and
in the limit of small strain is given by:
Biaxial Strain
: expresses how much the unit cell is strained in the z
direction compared to the x and y:
Uniaxial Strain
: strain in one direction only, e.g. if e
ij
= constant and
e
ii
=0 then the strain is uniaxial in the [111]
TMCS
III, Leeds 18
th
Jan
2012
Slide
13
Stress Components
Stress components
: the force components (per unit area) that
causes the distortion of the unit cell.
There are 9 components:
X
x
, X
y
, X
z
, Y
x
, Y
y
, Y
z
, Z
x
, Z
y
, Z
z
Capital letters: direction of the force
Subscript: direction normal to the plane on which the stress is
applied (x is normal to yz, y is normal to xz, z is normal to xy, )
X
x
x
y
The number of independent
components reduces when we consider
that in cubic systems (like diamond or
zincblende) there is no torque on the
system (stress does not produce
angular acceleration).
Therefore X
y
= Y
x
, Y
z
= Z
y
, Z
x
= X
z
And we are only left with 6:
X
x
, Y
y
, Z
z
; Y
z
, Z
x
, X
y
TMCS
III, Leeds 18
th
Jan
2012
Slide
14
Elastic constants
The stress components are connected to the strain components via
the small strain elastic constants:
In practice we never have to deal with all 36 elastic constants.
First of all it is always the case that c
ij
=c
ji
which reduced the total to 21.
Second in real crystals, particularly cubic, the lattice symmetry reduces
the number even more.
Therefore in ZB we only have 3 independent constants: c
11
,c
12
,c
44
In WZ there are 5: c
11
,c
12
,c
13
, c
33
, c
44
TMCS
III, Leeds 18
th
Jan
2012
Slide
15
Some more definitions
Elastic strain energy density for ZB
:
Bulk Modulus for ZB
:
Shear Constant for ZB
:
TMCS
III, Leeds 18
th
Jan
2012
Slide
16
Properties of Semiconductors
ZB
a
B
C’
c11
c12
c44
(Ǻ)
(Mbar)
(Mbar) (Mbar) (Mbar)
(Mbar)
Si
5.431
0.980
0.502 1.660
0.640 0.796
Ge
5.658 0.713 0.410
1.260
0.440 0.677
C
3.567 0.442 0.478
10.79
1.24 5.78
Ga

As
5.653
0.757
0.364 1.242
0.514
0.634
In

As
6.058
0.617
0.229
0.922
0.465 0.444
Al

As
5.662
0.747
0.288
1.131
0.555 0.547
Ga

P
5.451
0.921
0.440
1.507
0.628 0.763
In

P
5.869
0.736
0.269
1.095
0.556 0.526
Al

P
5.463
0.886
0.329
1.325
0.667 0.627
Ga

N
4.500
2.060
0.825
3.159 1.510 1.976
In

N
4.980
1.476
0.424
2.040
1.190 1.141
Al

N
4.380
2.030
0.698
2.961
1.565 2.004
Ga

Sb
6.096
0.567
0.270
0.927
0.378 0.462
In

Sb
6.479
0.476
0.183
0.720
0.354 0.341
Al

Sb
6.135
0.855
0.414
1.407
0.579 0.399
TMCS
III, Leeds 18
th
Jan
2012
Slide
17
Strain in Lattice Mismatched Epitaxy
Poisson ratio
: is a measure of the tendency of materials to
stretch in one direction when compressed in another. This ratio
depends on the substrate orientation and the type of crystal. For
cubic crystals including ZB:
Strain
: in pseudomorphic growth one can consider, independent of
the substrate orientation, strain to have only two components, one
parallel to the growth plane and one perpendicular.
Important: in [001] growth:
e
= e
xx
= e
yy
and
e
= e
zz
TMCS
III, Leeds 18
th
Jan
2012
Slide
18
Strain in [111] pseudomorphically
grown layers
Important: in [111] growth the combination of
e
and
e
results in a
strain tensor with e
xx
= e
yy
= e
zz
and e
xy
= e
xz
= e
yz
The distortions in this case are:
[111]
(1,1,1)
x
z
y
Important: the distortions are expressed
in the basis system where x, y and z are
aligned with the [100], [010] and [001]
directions.
Instead
e
and
e
are defined so that they
relate to strain in the (111) plane and the
[111] direction, respectively.
TMCS
III, Leeds 18
th
Jan
2012
Slide
19
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
20
Tetrahedral Bonding
In the Zincblende crystal, just like in the diamond one, atoms
bond together to form tetrahedrons.
Hence the individual atomic orbitals merge to form sp
3
hybrid orbitals
TMCS
III, Leeds 18
th
Jan
2012
Slide
21
Wurtzite
While Zinblende is the preferred crystal structure of III

As, III

P and
III

Sb, III

N tend to crystallize preferentially in hexagonal form. The
hexagonal crystal with a two atom basis consisting of cations and
anions is called Wurtzite.
View from the top
Perspective View
Wurtzite
Zincblende
Two adjacent tetrahedrons overlap in the z direction in WZ but not in ZB.
Hence second nearest neighbours in WZ are actually closer than in ZB at
equilibrium. The modified inter

atomic forces result in a slight reduction of
the interatomic distance between the first nearest neighbours.
TMCS
III, Leeds 18
th
Jan
2012
Slide
22
The 7
th
elastic parameter
Is a description based on 6 strain components enough to describe
all deformations in a ZB or WZ crystals?
The distance that the atom is
displaced by is characterized by
the
Kleinman parameter
With a the lattice constant and
γ
the shear strain.
This results in a crystal where the
atomic bonds are not all of the
same length.
Strain in the [111]
Only 3
identical
sp
3
orbitals
+
+
+
+

TMCS
III, Leeds 18
th
Jan
2012
Slide
23
Strain from atomic positions
Given the 5 coordinates of the atoms in a tetrahedron how do we
reverse engineer the strain?
This become a simple system of linear equations easily solvable.
The solution gives the 6 components of the strain tensor.
However the deformation on the position of the yellow atom, dependant
on the Keinman parameter, is still undetermined and requires a separate
calculation.
TMCS
III, Leeds 18
th
Jan
2012
Slide
24
The issue of local/global composition
Microscopists
refer to strain as difference in the bond lengths
compared to the host.
Theorists
think of strain as deformation of a material from its bulk
state.
Everyone else
does not usually know what they are talking about!!
Furthermore strain is a relative property (variation of e.g. bond
length compared to an initial state).
If dealing with an alloy and if wanting to take the theorist approach,
one needs to know what the lattice constant of the alloy is.
But what does composition mean?
It makes sense for a large uniform block, not for non uniform.
We take the approach of counting atoms up to second nearest
neighbour form the centre of the tetrahedron
TMCS
III, Leeds 18
th
Jan
2012
Slide
25
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
26
Modelling Strain in Real Structures
Because of its impact on the electronic properties strain in
semiconductor nanostructures always needs to be evaluated with
the highest possible accuracy.
Measurements (usually involving electron microscopy analysis) are
not usually sufficiently accurate, so modelling is the only viable
alternative.
Simple elasticity formulas are acceptable when dealing with
standard cases where strains are uniform or approximating strains
as uniform is acceptable, e.g. a simple quantum well.
They become useless however in complex quantum well
structures, wires and dots where strains are non uniform.
In time several methods have been developed ranging from
continuum, finite element, analytic and atomistic.
Atomistic methods are now widely used for quantum dots while
continuum methods are the preferred methods for quantum wells.
TMCS
III, Leeds 18
th
Jan
2012
Slide
27
Molecular Dynamics
•
Molecular Dynamics is a computer simulation in which a starting set
of atoms or molecules is made to interact for a period of time
following the laws of Physics (e.g Newton’s Laws).
•
In Semiconductor science one can build an atomistic model of a
strained crystal but if the strain is not known a priori then atoms are
not going to be in their equilibrium positions.
•
Then their motion paths are dictated by the “
force field
” generated
by the potential of the solid.
TMCS
III, Leeds 18
th
Jan
2012
Slide
28
Often the simulation does not require very large atomic motion.
For instance for calculating strain one might only want to allow small
atomic displacements from the crystal structure, without atom switching.
When Energy minimisation is the fundamental criterion and forces are
used to direct the geometry optimisation rather than predicting the final
positions, we are using a “Molecular Statics” simulation.
Molecular Dynamics
Initial Position of the atoms r
0
i
Evaluate the positions after
Δ
t
Potential V (r
0
i
)
Forces F=

grad
V (r
0
i
)
Velocities and acceleration
Repeat till Forces are low
TMCS
III, Leeds 18
th
Jan
2012
Slide
29
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
30
Valence Force Field
The “force field” that is generated by the potential of the atoms in
the solid can be represented as a 3 body potential.
In the Keating's Valence Force Field:
The Potential is the sum of the
potential energy between the
pairs of atoms i and j (two
body), plus a term that depends
on the angle between i,j and a
third atom k (three body).
d
ij
0
is the unstrained bond length of atoms
i
and
j
and
0
is the
unstrained bond angle (e.g. for zinc

blend cos
0
=

1/3), and
ijk
is the
angle between atoms
i
,
j
and
k
.
The local chemistry is contained in the parameters
and
, which are
fitted to the elastic constants
i
j
k
k
k
jki
R
j
R
i
R
k
P.N. Keating, Phys. Rev. 145, 637 (1966)
TMCS
III, Leeds 18
th
Jan
2012
Slide
31
Valence Force Field
The VFF is widely used for all types of nanostructures.
VFF is basically a parabolic approximation to the potential of solids
The main limitation is that there
are only 2 parameters (
and
) but
3 elastic constants even for
Zincblende!!!
R
0
R
V(R)
Binding Energy
Uniform: same
distortion in
x,y and z
Non Uniform: z
stretch, x,y compress
(by the same amount)
and viceversa
Ω
is the volume
occupied by one
atom
TMCS
III, Leeds 18
th
Jan
2012
Slide
32
Progress in Valence Force Field
Anharmonicity
correction
:
•
Ability to reproduce
anharmonic
effects is linked to the quality of prediction
of the phonon spectrum.
•
Some progress has been presented (e.g.
Lazarenkova
et al,
Superlattices
and Microstructures 34, 553 (2003)).
•
Not clear why phonon frequencies, elastic constants and mode
Grüneisen
parameters are not correlated (Porter et al J. Appl. Phys. 81, 96 (1997
)).
•
For
Ionicity
in
Zincblende
to solve this problem check recent
P. Han and G.
Bester, Phys. Rev. B
83
174304 (2011)
Ionicity
and
Wurtzite
:
•
Empirical potentials were historically developed for Si and
Ge
(pure covalent
bonds)
•
III

V are mainly covalent, partially ionic. II

VI are both covalent and ionic
•
Only for infinite crystals or systems were the charge is uniformly distributed
this it’s not a big deal.
•
Important in III

N WZ (Grosse and
Neugebauer
, PRB 63, 085207 (2001)),
and can be incorporated following
Ewald
summation scheme (codes available).
•
Also check Camacho
et al (
Physica
E,
Vol. 42, p. 1361 (2010)
)
“application
of Keating’s valence force field to non
–
ideal
wurtzite
materials
”
Distance between 2
ions, one of which is in
the central cell
TMCS
III, Leeds 18
th
Jan
2012
Slide
33
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
34
Stillinger

Weber
F. Stillinger and T. A. Weber, Phys. Rev. B 31, 5262 (1985)
The “force field” that is generated by the potential of the atoms in
the solid can be represented as a 3 body potential.
In the Stillinger

Weber potential:
i
j
k
k
k
jki
R
j
R
i
R
k
The Potential is the sum of the
potential energy between the
pairs of atoms i and j (two
body), plus a term that depends
on the angle between i,j and a
third atom k (three body).
This in an adaptation of the well known Lennard

Jones
potential used for liquefied noble gasses.
This potential works very well for Si in
diamond structure where
the
bond
angle cos
0
=

1/3.
The local chemistry is contained in the parameters A, B , p, q , a,
λ
and
γ
which are fitted to various material properties.
TMCS
III, Leeds 18
th
Jan
2012
Slide
35
Stillinger

Weber
The SW is not as widely used as VFF, but it has his niche
(thermodynamics of Si mainly).
In a way it should perform much better than VFF as it is not a
parabolic approximation to the potential of solids.
Parameterisations take into account the crystal phase diagram and
check that diamond is the lowest energy structure
Works
reasonably well for diamond

Si but not for other crystal structures.
TMCS
III, Leeds 18
th
Jan
2012
Slide
36
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
37
Tersoff Potential
The “force field” that is generated by the potential of the atoms in
the solid can be represented as a 3 body potential.
In the Tersoff potential:
i
j
k
k
k
jki
R
j
R
i
R
k
The Potential is the sum of the
potential energy between the
pairs of atoms i and j (two body),
multiplied times a term (b
ij
) that
depends on the angle between i,j
and a third atom k (three body).
The expression for b
ij
(known as bond order) is written
as to emulate the atomic coordination number Z.
Hence
ζ
is sometimes called the pseudo

coordination.
g(
θ
) and
ω
describe the angular and radial forces dependence.
TMCS
III, Leeds 18
th
Jan
2012
Slide
38
i
j
k
k
k
jki
g
θ
θ
eq
ω
Tersoff Potential
angular forces: resistance to bend
radial forces: resistance to stretch
•
When fitting to Bulk Modulus g(
θ
) is always g(
θ
eq
)
and
ω
ijk
==1
•
When fitting to Shear Constant g(
θ
)≠ (
θ
eq)
but
ω
ijk
==1
•
When fitting c
44
then both g(
θ
) ≠ (
θ
eq)
and
ω
ijk
≠
1
•
Hence the Kleinman parameter links angular and radial forces!!!
TMCS
III, Leeds 18
th
Jan
2012
Slide
39
Tersoff Potential
This potential describes covalent bonding and works very
well for different crystal structures for group IV and despite
the partial ionicity of the bond, group III

V.
The local chemistry is contained in the parameters A, B , r
e
,
α
,
β
,
γ
, c, d, h, n and
λ
,
which are fitted to various material
properties.
J. Tersoff, Phys Rev Lett
56
, 632 (1986) & Phys Rev B
39
, 5566 (1989)
Sayed et al,
Nuclear Instruments and Methods in Physics Research
102
, 232 (1995)
TMCS
III, Leeds 18
th
Jan
2012
Slide
40
Tersoff Potential
The TP is not as widely used as VFF, but its use is rapidly increasing
as parameterizations are improved.
Again it should perform much better than VFF as it is not a parabolic
approximation to the potential of solids.
As there are many parameters, parameterisations can take into
account many things, including the crystal phase diagram, all the
cohesive and elastic properties and many more.
Works
rather well for zincblende and diamond group IV and III

V but it is
not yet optimized for thermodynamic and vibrational properties.
D. Powell, M.A. Migliorato and A.G. Cullis, Phys. Rev. B 75, 115202 (2007)
TMCS
III, Leeds 18
th
Jan
2012
Slide
41
Progress in Tersoff
The Kleinman parameter
•
The many parameters need putting to good use.
•
Kleinman deformation is critical because expresses the balance between
radial and angular forces (Powell et al PRB 75, 115202 (2007))
DFT
DFT
Range of physical shear strains
Tersoff
Tersoff
Ionicity and Phonons
•
Ionicity, like VFF, is missing.
•
Crystal growth only possible if ionic contribution is included (Nakamura et
al J. Cryst. Growth
209,
232 (2000)
•
Phonons are still independent of elastic constants (Powell et al, Physica E
32, 270 (2006)
TMCS
III, Leeds 18
th
Jan
2012
Slide
42
Beyond Tersoff: bond order potentials
Π
versus
σ
–
bonding
•
Tersoff neglects Π
–
bonding. Is it of consequence?
•
Tersoff can to some extent reproduce surface
reconstruction energies (Hammerschmidt, PhD thesis)
Beyond
σ

bonding
•
It is generally possible to rewrite the b
ij
with expressions directly obtained
from tight binding. (D.G. Pettifor, “Many atom Interactions in Solids”,
Springer Proceedings in Physics 48, 1990, pag 64))
•
In this way the “bond order” can be explicitly obtained analytically to any
order (Murdick et al, PRB 73, 045206 (2006)).
•
The second moment approximation is essentially equivalent to Tersoff
(Conrad and Scheerschmidt, PRB 58, 4538 (1998))
TMCS
III, Leeds 18
th
Jan
2012
Slide
43
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
44
General Tips for MD
Building Models:
•
If possible try and use existing software
•
Try and guess final positions: it saves a lot of computational time
Empirical Potentials:
•
Codes that use VFF, SW and Tersoff are usually freely available!
•
IMD (Stuttgart), CPMD (IBM

Zurich) are parallel (for running on
clusters) and open source
•
Nemo
3
(Purdue)
uses VFF
•
Always check what version of the potentials are being used!!
Molecular Statics:
•
Make sure that the parameters that control the length of time the
simulation is running for are set to reasonable values
•
Build your simulation up in size to see what you can get away
with in terms of system sizes and check that results do not
depend on the size chosen
Strain:
•
Good strain algorithms exist and are freely available
•
If you write your own you need a nearest neighbour list. Usually
MD produces one
Gridding:
•
Strain is first obtained onto the atomic grid. Then to use it often
it needs converting to an ordered grid. One can use various
methods like Gaussian smoothing or weighted average.
TMCS
III, Leeds 18
th
Jan
2012
Slide
45
MD of QDs using Tersoff Potential
ε
xx
ε
z
z
ε
yy
Before MD
After MD
Fixed
PBC
Floating
TMCS
III, Leeds 18
th
Jan
2012
Slide
46
Outline
1.
Epitaxy
in Semiconductor Crystal Growth
2.
Elastic Description of Strain in Cubic Semiconductor Crystals
3.
Atomistic Description of Strain
4.
Molecular Statics and Force Fields
5.
Keating's Valence Force Field
6.
Stillinger

Weber Potential
7.
Tersoff
Potential
8.
Simulation of
Nanostructures
9.
Piezoelectricity in
Zincblende
and
Wurtzite
crystals
TMCS
III, Leeds 18
th
Jan
2012
Slide
47
Kleinman Parameter
The distance that the atom is displaced by is
characterized by the Kleinman parameter
With a the lattice constant and
γ
the shear strain.
This results in a crystal where the atomic bonds
are not all of the same length.
TMCS
III, Leeds 18
th
Jan
2012
Slide
48
+
+
+
+

Strain in the [111]
4
identical
sp
3
orbitals
Only 3
identical
sp
3
orbitals
+
+
+
+

Piezoelectricity
In the case of a uniaxial distortion the displacement is in the [111]
direction, and can still be characterized by the Kleinman parameter
The displacement of cations relative to anions in III

V semiconductors
results in the creation of electric dipoles in the polar direction which in
ZB is the direction that lacks inversion symmetry.
TMCS
III, Leeds 18
th
Jan
2012
Slide
49
Piezoelectricity in
Zincblende
The effect can be quantified by writing a general expression for the
polarization as a function of the so called “piezoelectric coefficients”
and the distortion components.
Convention is:
xx=1, yy=2, zz=3, yz=4, zx=5, xy=6
In ZB, for symmetry, the only non zero
coefficients are e
14
= e
25
= e
36
In
actual fact this picture is incomplete as only includes coefficients
linked to linear terms in the strain.
In the past 6 years the importance of including also
coeffiecients
linked
to quadratic terms in the strain (e.g.
𝛆
xx
2
or
𝛆
xy
𝛆
xz
)
has been
highlighted (so called non linear or second order
Piezo
effect).
•
M.A
.
Migliorato
et al
,
Phys. Rev. B 74, 245332 (2006)
•
L. C. Lew Yan
Voon
and M.
Willatzen
, J. Appl. Phys. 109, 031101 (2011) REVIEW
•
A.
Beya

Wakata
et al,
Phys. Rev. B 84, 195207 (2011)
TMCS
III, Leeds 18
th
Jan
2012
Piezoelectricity in
Wurtzite
Wurtzite
Zincblende
Spontaneous
polarization
Strain induced polarization
Quadratic terms in the strain (e.g.
𝛆
xx
2
or
𝛆
xy
𝛆
xz
)
are also important.
There is still some controversy between the early accepted values of
mainly for the spontaneous polarization coefficients
•
L. C. Lew Yan
Voon
and M.
Willatzen
, J. Appl. Phys. 109, 031101 (2011) REVIEW
•
J. Pal
et al,
Phys. Rev. B
84,
085211 (2011)
TMCS
III, Leeds 18
th
Jan
2012
Slide
51
Thank you!!!
Acknowledgments
:
Joydeep
Pal , Umberto
Monteverde
, Geoffrey
Tse
,
Vesel
Haxha
, Raman
Garg
(University of Manchester)
Dave
Powell (University of
Sheffield)
GP
Srivastava
(University of Exeter)
Comments 0
Log in to post a comment