SUBMITTED TO THE
UNIVERSITY OF LUCKNOW
Doctor of Philosophy
IN
PHYSICS
BY
DEPARTMENT OF PHYSICS
UNIVERSITY OF LUCKNOW
LUCKNOW226 007
INDIA
2014
Dedicated
to
My Parents
CERTIFICATE
This is certify that all the regulations necessary for the submission
of Ph.D. thesis of Mr. Satya Prakash Singh have been fully observed.
I (Satya Prakash Singh) hereby, declare that the thesis entitled “Study
of szigetti effective charge on the surface of condensed materials at
nano scale parameter” being submitted to Lucknow university in
fulfillment of the requirements for the degree of Philosophy in Physics is
the original work of mine and has not previously formed the basis for the
award of any degree, diploma or any other similar title or recognition.
Last but not least, I would like to mention the encouragement and
cooperation given to me by my family members specially my father,
mother& wife throughout the research work.
Preface iii
Chapters Page No
1. Introduction 118
2. Theory of phonon,plasmons and polaritons 1936
2.1 Surface phonon excitations
2.2 Modifed plasmon –matter interaction
2.3 Locally excited plasmons on the material surface
2.4 Theoritical & experimental study of plasmons
2.5 Excitation of surface polaritons on the surface of
Materials
3 Dielectric functions due to lattice vibrations 3755
3.1 Dielectric constant and relative permittivity
3.2 Dielectric functions for materials
3.3 Polarization of dielectric materials
3.4 Dielectric function for polar semiconductor
3.5 Dielectric function for materials
4. Szigetti’s dielectric theory 5675
4.1 Szigetti’s first relation
4.2 Anharmonic correction
4.3 Spectroscopic implication of Szigeti relation
4.4 Aspects of the szigetti charge
5 Filtering properties of materials 76149
5.1 Modified Bloch’s hydrodynamical model
5.2 Special dispersion relations for two mode coupling
5.3Study of three mode coupling under special conditions
5.4 Local theory approximation for k>>1
5.5 Dielectric function on the surface of materials
5.6 Expression of szigetti effective charge for k>>1
5.7 Band attenuation properties of materials without
magnetic Field & with magnetic field
5.8 Comparitive study of filtering properties
5.9 Variation of the width of allowed band of polarized
materials with szigetti effective charge for k>>1
5.10 Advatages of szigetti effective charge study and Its
applications
6 Summary and conclusions 150171
7. References 172183
8. Publications 184
PREFACE
The study of surface polariton waves at the interface of isotropic, homogenous dielectric
media has recently recived considerable attention surface polaritons are defined as the coupled modes
of photons and elementary surface excitations of solids, so surface polariton waves propagates as
electromagnetic waves along the interface of two media but decay in nonoscillatory exponential
manner in a direction perpendicular to the interface. The surface properties of these modes depend
upon the type of materials that form interface .therefore these modes provide a sensitive probe for the
study of solid surface. the fact that the surface polaritons may be excited at some place on the surface
of a solid and detected at some other place on the surface so it makes it surface polaritons an extremely
sensitive tool for determining surface properties [v.m agranovich(1978)]
The frequency of the surface polariton waves depends upon the type of elementary surface
excitations to which the EM wave (photons) couples .The important elementary excitations in dielectric
media are plasnons, phonons, excitons etc. a study of these surface polariton mode give an insight into
the type of elementary excitations sustained by the medium .the presence of polariton waves at the
surface affect the optical properties of the medium , the study of dispersion of these surface waves
considering reflection and refraction at surface and interfaces ,which may be refraction at surfaces and
interfaces , which may be referred to as crystal optics of surfaces , plays a fundamental role in the
phyics of surfaces.
In view of the above discussion, in the present thesis, the Bloch’s Hydrodynamical model is
utilized to study the coupled surface polariton waves sustained by spatially –dispersive polar semi
conductor medium for k≠0 bounded by a nondispersive dielectric medium. Since the properties of
surface polariton waves are strongly dependent on the geometry of the interface and for spherical
geometry of the interface are derived and the effect of spatial dispersion and interface are divided and
the effect of spatial dispersion and interface geometry on surface polariton is studied. The dispersion
relation for both these cases is obtained by applying certain suitable boundary condition at the
interface. Also since the dielectric function plays important role in determining the dispersion constant
(wave vector) dependence of the dielectric function of the dispersive polar semiconductor medium is
also studied . Since the optical properties of the medium are affected by the presence of surface waves,
the refractive index of the polar semiconductor medium is also derived and its frequency &wave vector
dependence are studied . The inferences obtained from the study of the dielectric function and the
refractive index , when analyzed along with the dispersion curves obtained for surface polariton waves
leads to important conclusion regarding the effect of surface polariton waves on the optical properties
the of the polar semiconductor medium. Also the dispersion curves provide information regarding the
wave vector region for which the surface waves mixed photonsurface excitation character.
It is observed that the inclusion of spatial dispersion plays an important role in the study of
surface polariton waves at polar semiconductor dielectric interfaces. The surface polariton waves at the
i
spherical interfaces become radius independent when spatial dispersion is neglected and as a result
becomes independent of the curvature of the interface. The exclusion of spatial dispersion , therefore
,will not give correct result for small spheres . For the plane interfaces too, the inclusion of spatial
dispersion effects is important, and leads to the infinite lifetime of surface polaritons.
The effect of D.C Magnetic field on two mode coupling (SPSOP) cylindrical polar
semiconductor for k=0 and k≠0 have been discussed by Dr. K.S. Srivastava and D.K Singh (1997). The
effect of D.C Magnetic field on spatial dispersion relation for three mode coupling in spherical polar
semiconductor for k≠0 gives an important result in the study of surface polariton waves at
semiconductor dielectric interface.
Szigeti published four seminal papers on the dielectric behaviour of crystals during the
period 19491961. Szigeti’s theory is applicable to isotropic and anisotropic, ionic and covalent crystals
with different structures. Szigeti’s theory connects dielectric, spectroscopic and elastic properties. An
important outcome of Szigeti’s theory is the concept of the effective ionic charge (s). It is pointed out
that s correlates with a number of physical properties and is a measure of ionicity of the interatomic
bond. Since Szigeti’s work, several theoretical models have been proposed to account for the fact that s
< 1. These models provide an insight into the complex polarization mechanisms in solids. This review
summarizes Szigeti’s work and the work that followed; the implications and applications of Szigeti’s
theory are discussed.
ii
CHAPTER
CHAPTER1
INTRODUCTION
1
related to study of solids. Maximum information regarding the surface
properties of a medium may obviously be obtained by studying the
propagation, not the bulk, but of surface electromagnetic energy just on
the surface.
When the dipole moment density and also the displacement field
depends not only on the value of electric field at a particular point but
also on the value assumed by the electric field in the near vicinity of that
point, then material is said to be exhibit “spatial dispersion”. The
inclusion of spatial dispersion in the study of surface polaritons leads to
important conclusion like the finite life time of the surface modes even
collision less plasma state..
2
induces and the resulting wave that propogates in the medium consists of
EM field (photons) coupled with elementary excitations of the medium.
This coupling is stronger when the frequency of the incident waves is
comparable with the frequency of the elementary modes of the medium.
The EM waves that arise as a result of this coupling are termed as
polariton waves [14, 15].
3
semiconductors are well known and have been recently employed by
several workers [24, 3739] to study these waves. The most widely used
effective methods that have been employed for obtaining the dispersion
relation, are the Maxwell’s equation method [3941] and the quantum
mechanical random phase approximation [RPA] [42].
4
polar semiconductors at plane surface [143] and cylindrical surface are
studied [144145].
The matter is classified in three states Solid, liquid and gas. The
distinction lies in difference in bond strength or the strength of
intermolecular forces which decreases considerably as substances change
from solid to liquid and the bonds break or intermolecular forces
practically disappear as liquid changes to gaseous state. Bonds can be
broken by increasing the kinetic energy of atoms and molecules, i.e. by
increasing the temperature of the substance. Sir William Crooks, in 1879,
studied electrical discharge of gases and observed that when the electrical
discharge was passed through the gas, the gas was ionized and the
5
temperature was increased in the process and called this collection of
positive and negative ions as the fourth state of matter. Langmuir [53]
called this as “plasma” and defined it in a more comprehensive way as a
system containing high and approximately equal concentration of
positively and negatively charged particle which are relatively mobile and
interact with each other collectively via long range coulomb forces.
6
dimensional quantum wires [123] and in the deviation of additional
boundary conditions at the interface between two conductors [124].
A survey of literature shows that all these studies [ 81 83] have
been made for metal, in which surface phonon waves are absent and
independent plasmons exist. Srivastava and Tondon [128] for the first
time modified the Bloch’s Hydrodynamical equations for metals [125]
to study the interaction between these two modes in semiconductors. The
modified Bloch’s equation for the semiconductor [128] may be written
as
_
Dv _ 1 _ _ _ _ _
d ρ (n' )
m = −e E + ∇× B − mv ∇− ∇ ∫ 0n ( r ,t ) (1.1)
Dt c n'
_
_ _
4π _ 1 ∂ D
∇× B = J+ (1.2)
c c ∂t
∂n _ _
= −∇ .(n v) (1.3)
∂t
_ _
4π e _ _
∇.E = N ( r ) − n ( r , t) (1.4)
E
+
Y[Y2(az+b)Y+az] = 0 (1.5)
For lower value of k nearly linear variation with wsp/wt for radius
1A. & 2A. .
When radius increases then variation of wsp/wt with respect to K
shown lesser linear variation.
Thus for higher value K it shows non linear variation. Hence for
higher value of K, it shows non linear agreement.
But at radius 8A. It shows again perfect linear variation for high
value of K.
Again linear variation between wsp/wt versus K disturb for radius
16A. for high value of K.
Variation of wsp/wt versus K for different radius observe that
variation is either perfectly linear or not perfectly linear.
When radius increases then it’s linear variation changes with
wsp/wt versus K, thus for high radius linear variation changes.
8
Author observed that:
Thus author obverse that all medium shows same type of variation
only transoil shows different type of variation.
9
Now Author observed that that
Thus for low value of K all medium shows linear variation while
for high value of K only lice shows different variation when we plot 3D
graph between wsp/wt versus K for different medium.
So author observe that all medium shows linear variation while lice
medium not shows linear variation.
10
Again author observe for different radius that
11
Author observe at constant radius variation of InP:
Thus for InP at lone value of K all medium including lice shows
linear variation but for higher value of K, only lice shows different
variation also for high value of K.
12
In 3D graph we take wsp/wt along one axis. Wave vector K along
other axis while different radius consider along third dimension.
The wsp/wt varies linearly with respect to K (1 A.  10 A.) for
cylindrical surface of InP at different radius.
At different radius InP medium shows same linear variation.
wsp/wt increases vertically when we move from top to bottom,
while this ratio is decreases when we move Horizontally from left
to right.
13
In this graph we are comparing variation of wsp/wt with wave rector K
for different substances KF, InP & Mgo.
wsp/wt in all substances increases from top to bottom when K
increases.
wsp/wt is lower for KF but in other two substances InP & Mgo wsp/wt
increases rapidly [slightly greater value].
When we plot comparative graph for cylindrical surface of different
substance like KF, InP & Mgo with respect to wave vector K at
same radius 1 A., then all substances have same linear variation.
This dispersion curves shows that the frequency of the lower mode
changes slowly and frequency of the upper mode varies rapidly
with the wave vector K.
14
For KF variation of wsp/wt with K varies with lesser effect where as
for other two substances InP Mgo show more effective linear
variation. At this radius (2 A.) both InP & Mgo having nearly same
value of wsp/wt in comparison to radius (1 A.).
15
wsp = wsop i.e. where two uncoupled modes intersect. The stronger
coupling is observed at wsp/wt = 1.
Thus the surface acts as band pass filter [BPF]. Agian their is no
propagation when ε(W) is negative.
for , the surface become transparent again act as High pass filter
(HPF). from graph we observe band width (∆) of band pass filter is given
by
W
∆ 1
W
W
∆ 1
W
Gives values at which the surface will act as high pass filter. from
study of Szigeti effective charge e∗ versus ∆ with the help of data in
table. from this the best fit of data we find the following relationship
between e∗ and ∆ as
∗
∆=0.074+0.099 (1.7)
from table our calculated values agree well with the experimental values
of Hass and Henvis.
16
We can conclude that polar semiconductor surface acts as a band
pass filter for ω <ω<ωt and as high Pass Filter for ωω+. The es2 is
measure of iconicity of polar semiconductors. thus width of band (∆)
increases with increase in ionicity. If there is more ionic character, band
of allowed frequency is wider and surface becomes high pass filter at
higher value of frequency ω.
There are some new features which are not seen when we observe
general effective charge in place of Szigetti effective charge so this
charge provides to be very useful tool in theoretical and experimental
study of Surfaces in the present work. The next chapter i.e. chapter2
deals with elementary ideas of theory of phonon, Plasmon & polariton. In
this we study the existence of surface Plasmon and surface polariton in
polar semiconductor. Their interaction discussed with the help of its
dispersion relation.
18
CHAPTER
CHAPTER2
THEORY OF PHONON, PLASMONS & POLARITONS
In the present work I shall study only the solid state plasma and
electron moving in a lattice of ions will be charge carriers. Due to the
motion of free electrons oscillations are set up in the system which are
known as plasma oscillation.
e
∞
(2.1)
∇ E = − ∫ δω = ∫ ∫ exp(− iωt )E ( x, ω )dω .dx
trajectory
2π −∞
dp (ω )
∞
∇E = ∫ hω dω (2.2)
0
dω
1 (2.3)
∇ 2V (r , ω ) = − ρ (r , ω )
ε oε
∇.[ε oε (ω )∇V (r , ω )] = − ρ (r , ω )
(2.4)
20
Basically equations (2.1) and (2.2) allow the total phonon excitation
probability to be determined using the solution of equations (2.3) and
(2.4). However, it is convenient to treat the surface and phonon
excitations separately for this purpose it can be noted that the general
solution of equation (2.3) and (2.4) is the sum of the homogeneous and
the particular solution of the problem. The two terms represent the
induced potential (Surface Phonon) and the direct potential (Volume
Phonon) respectively.
V (r , ω ) = V ind (r , ω ) + V p (r , ω ) (2.5)
Since the induced potential responsible for the surface losses can
be calculated separately. It is useful to introduce the notation of surface
Phonon excitation Probability dpsur (w)/dw. Starting from equations (2.1)
& (2.2) it can be shown that the surface excitation probability for the
geometry shown in figure 2.1 is given by the following expression.
dP sur e
∞ ∞
iω (Y '−Y ) ind ro = (xo , y ' ,0 )
2 ∫
= dY ' ∫ dY lm exp V (r , r ) (2.6)
r = ( xo , y,0)
o
dω πhV − ∞ −∞ V
e
∇ 2V (r , ro ) = δ (r − r o ) (2.7)
ε oε (ω )
e
∇.[ε (ω )∇V (r , ro )] = δ (r − ro ) (2.8)
εo
21
If it is an isotropic medium. The potential distribution therefore
is quasielectrostatic and frequency dependent for each point along the
trajectory of the incident electron.
d 2 P volume (ω ) e2
θc 2π − qo
2
4π 3V 2ε o h ∫0 ∫0 q p 2ε ⊥ (ω ) + qc 2ε II (ω )
= θd θ d φ (2.9)
dωdY
Fig.2.1
The integration over the angles θ and φ takes into account all
electrons scattered with in an angle smaller then the cutoff angle θc [119].
23
ω 2p
ε 1 (ω ) = ε L − 2 (2.13)
ω
In absence of any external field, the electric field arises only due to
polarization charges and by symmetry it has the same magnitude but
opposite direction at the two sides of the interface, i.e.
E z1 = − E z 2 at z=0 (2.16)
ε1 (ω ) = −ε 2
ω 2p
εL − = −ε 2
ω s2
ω p2
= εL + ε2
ω s2
ω p2
ω s2 =
εL +ε2
24
Fig 2.2
ωp
or ωs = (2.17)
[ε L + ε 2 ]12
ωp
ω s=
2
ω + ω p 2
wsp = p1 (2.18)
2
25
R.H Rithcie [78] was the first to predict, theoretically, the
existance of surface plasmon in thin metallic foils and at the single plane
surface bounded by vaccum.
Later it was confirmed by stern and Ferrell. Both pointed out that a
part of plasmons excited in a thin metallic film by energetic electrons,
should be able to decay into photons of frequency equal to bulk plasma
frequency ω p .
( ) t
( ) ( )
D r , t = ∫ dt ' ∫ a 3r '∈RA r − r ' , t − t ' E ' r ' , t
0
(2.19)
where ∈RA (r − r ' , t − t ') is the real space, real time response function.
The displacement at a particular position r and at particular time t
26
depends upon electric field over a region of space about r (i.e. the
response of medium is non local or spatial dispersion has been included)
and at earlier time t ' . Thus
( ) ( )
ε R r − r ' , t − t ' = ε R r − r ' , t − t ' → for t ' < t (2.20)
( ) ( )
D K , ω = ∫ d 3r ∫ dt D r , t e −ωt × ei K r (2.21)
E (K , ω ) = ∫ d r ∫ dt E (r , t )e × e
3 − ωt iK r
(2.22)
∈ (K , ω ) = ∫ d r ∫ dt ∈ (r , t )e × e
3
R
− ωt iK r (2.23)
multiplying eqn (2.22) and (2.23) we find with the help of (2.19)
( ) (
D K , ω =∈ K , ω E K , ω )( ) (2.24)
Hence for a non local system the response function (which is the
dielectric function) is a function of wave vector and frequency, In general
a tensor. However a local response can be imposed by requiring
( ) ( )
∈R r − r ' , t − t ' = δ r − r ' ∈R (t − t ') (2.25)
(
δ r − r' = 1 ) for r = r ' (2.26)
=0 for r ≠ r '
27
wavelength λ is much larger (or K is very small) then depends on the
electric field. Since in this limit
( )
lim K → 0 ∈ K , ω =∈ (ω ) (2.27)
( ) (
D K , ω =∈ (ω )E K , ω ) (2.28)
28
Only very few experimental attempts to investigate different types
of plasmonic waveguides with subwavelengh locolization have been
under taken so far .
30
Strong subwave length localization of wp modes has been
demonstrated and reasonable propagation distance have been predicted,
which makes wp modes a good candidate for the design of subwave
length plasmanic waveguides.
31
complex entity. The energy stored in the wave is shared among the
incident field and the excitation induced in medium. Thus a polarization
wave is actually a composite wave or a coupled mode of the incident
electromagnetic field with the elementary excitations of the medium that
may coupled to electromagnetic field by virtue of their electric and
magnetic characteristics [86, 87]. Any elementary excitation of a medium
which gives rise to an electric and magnetic fields of the incident
Electromagnetic wave. The incident electromagnetic radiation can couple
simultaneously more then one predominant elementary excitation modes
of the medium if they have comparable frequencies.
Thus the term polariton is broad and general term for coupled
electromagnetic (EM) modes in solids [88]. However when a specific
designation is required for a particular case, For example, in the
frequency region where EM waves in a medium couple with optical
phonon or plasmon excitations of a dielectric medium, the coupled
modes are designated respectively as phonon polariton or plasmon
polaritons as the case may be. Similarly in the case of magnetic materials,
the polarisation modes are termed as magnon polaritons. In the present
work only the polariton mode is nonmagnetic, isotropic dielectric media
have been studied.
32
The coupled EM modes or polariton modes in a medium can be
both bulk modes [86,87] or surface modes [81] layer or are bounded
along the interface of two media such that they propagate without decay
in a wave like manner along the interface between two media having
different electrical or magnetic properties but decrease exponentially with
distance in the direction perpendicular to the interface.
34
wave vector as well frequency. To understand this nonlocal limit [85], a
dispersive dielectric medium is considered in which there occurs a
surface vibrational excitation of frequency ‘ ω ’ due to an incident
wavelength ‘ λ ’ then for λ >>d.
Where ‘d’ is the inter atomic spacing. This type of response of the
medium to the incident radiation is called the local response of the
medium, under such circumstances the propagating polariton waves is to
be within the local theory limits. This condition is easily satisfied by the
long wavelength surface optical Phonon waves so that the surface phonon
polariton modes can be studied accordingly within the local theory limits.
In general the relation between the displacement field and the electric
field for homogenous dielectric medium is given by [85]
D (ω , k ) = ε (ω , K )E (ω , K ) (2.29)
( )
lim K =0 ∈ K , ω = ε (ω ) (2.30)
35
obtain the dispersion relation for surface Plasmon polariton waves
taking spatial dispersion into account the nonlocal form of the
dielectric function is required.
36
CHAPTER
CHAPTER3
DIELECTRIC FUN
FUNCTIONS DUE TO LATTICE
VIBRATIONS
38
C
k=
C0
Vo
hence k=
Vd
ε
εr= k=
ε0
39
polariton spectrum. Here the electron gas in metals has been taken the
dielectric constant ∈ of electrostatic is defined in terms of the electric field
E and the polarization P, the dipole moment density
D = ε o E+P (3.1)
ρ = ρ ext + ρ i nd (3.2)
ρ (ρ ext + ρ ind )
divE = = (3.4)
εo εo
d 2x
m = −eE
dt 2
eE
x=
mω 2
e2E
The dipole moment of one electron is − ex = −
mω 2
40
The polarization defined as the dipole moment per unit volume, it
is
ne 2
P = − nex = − E (3.5)
mω 2
D(ω ) P(ω )
ε (ω ) = =1+ (3.6)
ε o E (ω ) ε o E (ω )
ne 2
∈ (ω ) = 1 − (3.7)
ε o mω 2
ne 2
ωp = (3.8)
ε om
ω p2
∈ (ω ) = 1 − (3.9)
ω2
41
3.3 Polarization of Dielectric materials
Figure 3.1
42
separation between the atoms depending upon the location of ion pair.
This polarization is also independent of Temperature.
Figure 3.2
43
Figure 3.3
Figure 3.4
p = p e + pi + po + p s
p = p e + pi + po
1 1 1 M +M −
= + so M = (3.10)
M M+ M− M+ + M−
44
When external electric field is applied to a substance. It gets
polarized giving rise to a polarization field which always opposes to the
applied field. The macroscopic field in the insulator is
E = Ee − E p (3.11)
4π
Eeff = E + P (3.12)
3
For a crystal with cubic symmetry. This field is always greater than
the macroscopic field. ‘P’ is known as Polarization vector.
45
distributed about the nucleus). The dipole moment is proportional to the
effective field and is given by α + Eeff and α − Eeff on positive and negative
Z e u + + α + E eff
− Z e u − + α − E eff
( )
P = Z e u + − u − + (α + + α − )E eff
[
P = N Z e x + (α + + α − )E eff ] (3.13)
(
M + u + = − µ u + − u − + Z e E eff ) (3.14)
(
M − u − = µ u + − u − − Z e E eff )
( ) (
M u + − u − = − µ u + − u − + Z e Eeff ) (3.15)
M X = − µ X + Z e Eeff (3.16)
46
The frequency dependence of the dielectric function is considered
by taking A.C field as
X = X o e − iωt (3.18)
2
− M ω X = − µ X + Z e Eeff − M ω 2 X = − M ω 2 X + Z e Eeff (3.19)
µ
where ω = is optical mode frequency which the lattice would
m
1
X = (3.20)
(
M ω −ω2
2
)Z E
e eff
(Z )2
P=N e
+ (α + − α − ) Eeff (3.21)
(
M ω2 − ω2 )
(Z e )2
α= + (α − α ) (3.22)
2
(
M ω − ω
2
+
) −
Here the first term in the atomic or ionic polarizability and the
second term is electronic polarizability.
47
The dielectric function due to lattice ε l (ω ) or the background
dielectric function is related to polarizability α by Clausius Mossotti
relation
∈L (ω ) − 1 4π
= Nα (3.23)
∈L (ω ) + 2 3
∈L (ω ) − 1 4π (Ze )
2
= N + (α + − α − ) (3.24)
∈L (ω ) + 2 (
2
3 M ω − ω 2 )
eqn
∈L (ω ) − 1 4π ( Ζe) 2
= Ν α + + α − + (3.25)
∈L (ω ) − 1 3 Μ (ω 2 − ω 2 )
∈0 −1 4π
ω >> ω , = Ν[α + + α − ] (3.26)
∈∞ +2 3
48
High frequency means the frequency high enough to render lattice
polarizability zero. This range comes out to be in optical region and
therefore is related to the index of refraction ‘n’. Thus ∈0 is greater than
∈∞ .
∈0 −1 ∈∞ −1 4π Ze 2
= + Ν (3.27)
∈0 +2 ∈∞ +2 3 Mϖ 2
∈L (ω ) − 1 4π 4π Ze 2 N
= N [α + + α − ] + (3.29)
∈L (ω ) + 2 3 3 ω 2
Mϖ 1 − 2
2
ϖ
∈ L (ω ) − 1 ∈∞ −1 ∈0 −1 ∈∞ −1 1
= + − 2
∈ L (ω ) + 2 ∈∞ +2 ∈0 +2 ∈∞ +2 ω
(1 − 2 )
ϖ
2 ∈∞ +2 2
∈∞ ω − ∈0 ∈ +2 ϖ
∈L (ω ) = 0
(3.30)
∈ +
ω 2 − ∞ 2ϖ 2
∈0 +2
49
∈∞ +2 2
ωt 2 = ϖ (3.31)
∈0 +2
D =∈ (ω ) E = E + 4π P (3.32)
∇•D = 0
E= − ∇ψ (3.33)
div P = 0 ∈ (ω ) = ∞
curl P = 0 ∈ (ω ) = 0 (3.34)
div P = 0 curl P = 0
50
again in cubic crystal D is parallel to E (i.e D are parallel to
p ). So considering the spatial dependence
D D0
i kr
E = Re E 0 e (3.35)
P P
0
D = 0 or D, Eand P ⊥ K (3.36)
for above conditions (a) & (b) in eqn (3.37(1) are combining, we
get
(a) E = 0 D, Eand P ⊥ K , ε (ω ) = ∞
(3.37ii)
(b) ) D = 0 D, Eand PIIK ε (ω ) = 0
51
4π 4π
E eff = 0 + P ⇒ E eff = P (3.38)
3 3
3 ε∞ −1
P = ZeNx + ε + 2 E eff (3.39)
4π ∞
3 ε ∞ − 1 4π
P = ZeNx + ε + 2 3 P (3.40)
4π ∞
ε∞ + 2
then P = Zenx (3.41)
3
The equation of motion (Eq 3.20) may be written as with the help
of earlier equations
2 4π
M (ω 2 − ωt ) x = Ze P (3.42)
3
2 4π ε ∞ + 2
M (ω 2 − ωt ) x = Ze ZeN x
3 3
2 ε∞ + 2
M (ω 2 − ωt ) = 4πN ( Ze) 2
9
ε∞ + 2 2
or ωt 2 = ω (3.43)
ε° + 2
by
E = −4π P
4π
E eff = −4π P + P
3
8π
E eff = − P (3.44)
3
ε + 2
P = ZeN x ∞ (3.45)
3
( 2 2
)
M ωt − ω x = − Ze E eff (3.46)
8π 8π ε∞ + 2
= − Ze − P = 8πZe P , ( 2
M ωl − ω 2 = ) N ( Ze) 2 (3.47)
3 3 3 3ε ∞
53
∈o ∈∞ +2 2
2
ωl = ω (3.48)
∈∞ ∈o +2
Equation (3.44 and 3.48) give the relation between the frequencies
of two modes as
∈o 2
ωl 2 = ωt (3.49)
∈∞
∈o − ∈ ∞
∈L (ω ) =∈ ∞ + (3.50)
ωt 2 − ω 2
For ionic compounds the static dielectric contant is greater than the
high frequency dielectric constant which also confirms with the help of
(3.49).
55
CHAPTER 4
SZIGETI’S DIELECTRIC THEORY
P = αE, (1)
ε − 1 = 4πα. (2)
This relation does not take into account of the internal field. When
the internal field is included, the Clausius–Mosotti relation
56
(ε − 1)/(ε + 2) = (4π/3)α. (3)
εs = ε∞ + (z2e2N)/π µt ν2 (5)
µ 1=m11+m21 (6)
57
single transverse optic infrared active mode. Agreement of eq. (5) with
experiment is not satisfactory.
M = Mi +Mu, (7)
(8)
58
only firstorder terms in the summation. Szigeti [2] obtained the
following relation:
The value of s for some alkali halides calculated by Szigeti [2] are
given in table 1. The input data (εs, ε∞, νt) used by Szigeti was from
earlier sources [8,9]. Values calculated by Lowndes and Martin [10] from
their own more recent and accurate data are also included in table 1. The
values in both sets are mutually consistent.
59
A common feature in the s values is that they are significantly less
than unity. Szigeti [2] himself pointed out that the deviation of s from
unity could be due to a partial homopolar (partial ionic) character or due
to interpenetration of the ions. In deriving eq. (9), a harmonic crystal was
assumed as implied by eq. (8). Szigeti [3] qualitatively surmised that this
may not affect the s values. However, he examined this aspect rigorously
in 1959 [4]. These and other aspects will be considered in later sections.
60
(11)
where the Q’s are the normal coordinates and α0, βij and γijk
are constants. Q0 is the transverse optic infrared active mode. Here the
first term contains the effect of displacement of ions and also the
electronic distortions while the second and third terms are due entirely to
electronic deformation. Similarly, he expressed the potential energy W as
W = Wh +W’ (12)
and
(14)
where M and W are given by eqs (11) and (12) and Eext is the
external field. Writing the Hamiltonian and applying perturbation theory,
Szigeti [4] showed that
εs − ε∞ = η + G, (16)
61
where η is the harmonic and G the anharmonic contributions. η is
the second term in eq. (9) whereas G is a function of the two dielectric
constants and the constants in eqs (11) and (14). where α is the linear
coefficient of thermal expansion, ψ the compressibility and T the
temperature. s, the Szigeti charge corrected for anharmonicity is given by
63
of s independently of Szigeti’s relation was certainly desirable. Several
models were developed to account for the s values and, more particularly,
to account for the deviation of s from unity. These models are discussed
in this section. Yamashita [16] and Yamashita and Kurosawa [17] treated
the problem by a quantum mechanical variation method. They calculated
the ionic charge for LiF. Narasimhan [18] extended Yamashita’s method
to MgO. Yamashita’s method has not found further application as it
involves more parameters than can be evaluated from experimental data.
Szigeti [3] himself suggested that the mutual distortions of the ions due to
their overlap could be a cause of the s values being <1.
Born and Huang [6] pointed out that such distortions would result
in a ‘distortion dipole moment’ m(r) and that this dipole moment would
cause s to be <1. However, the treatment of Born and Huang was
qualitative. Lundquist [19] used the Heitler–London approach to evaluate
the overlap distortion moment for NaCl. Dick and Overhauser [20]
reexamined Szigeti’s theory. They pointed out that apart from the
electronic polarization and polarization due to displacement of ions
included in Szigeti’s theory, it is necessary to include ‘some further
polarization’. Thus, they introduced the concepts of shortrange
interaction polarization and exchange charge polarization into the theory.
These arise from charge redistributions which occur when ions move with
resulting changes in electron overlap. Formulating their theory in terms of
the shell model, Dick and Overhauser calculated the integrals Sij
necessary to estimate the exchange charge. Here e_, A, µ and λ are
parameters occurring in the theory; these can be theoretically estimated
for each crystal. The values of s calculated from eq. (17) are given in
table 2. These values follow the same general trends as the original
Szigeti (or Lowndes and Martin) values but are seen to be systematically
64
larger and hence closer to unity. Dick and Overhauser concluded that the
shortrange interaction polarization and the exchange charge polarization
are responsible for the deviation of s from unity. Hanlon and Lawson [21]
observed that the values of the Szigeti charge s of alkali halides lie on a
smooth curve when plotted against the difference in the polarizabilities of
the negative and positive ions. In order to explain this empirical
observation, Hanlon and Lawson modified Szigeti’s theory by including
the contribution of a mechanical polarization. When an electric field is
applied, each ion will be polarized in the direction of the field. But the
ions will also move relatively closer. Their electron shells will
mechanically repel one another, resulting in a relative shift of each
electron cloud with respect to its nucleus. This shift corresponds to an
additional polarization. In the positive ion, this mechanical polarization
enhances the electrical polarization, but in the negative ion it detracts.
Hanlon and Lawson showed that this mechanical polarization contributes
to the value of s.
65
The concept of the deformation dipole moment originally
introduced to explain the Szigeti charge values was applied by Hardy to
the problem of lattice dynamics of alkali halide crystals. Hardy [24]
included the interactions due to the deformation dipole moment in the
dynamical matrix for KCl to obtain the dispersion curves. This method
was applied to study the lattice dynamics of several other alkali halide
crystals with NaCl and CsCl structures [25,26].
66
has revived the deformation dipole model originally introduced by Hardy
in the 1960s. Mahan [29] made ab initio calculations of the
polarizabilities of alkali ions and used them to calculate the refractive
index and the effective ionic charge by employing improved
‘calculational technology’. Mahan derived the perturbation equations for
the deformation dipole tensor by the combined use of selfconsistent
treatment of local density approximation and the spherical solid model.
The effective ionic charge s is calculated from
s = (1 − τ) (20)
and
67
denoted by q*. Though Szigeti [2] had mainly the alkali halides in view,
he evaluated the value of s for
Several other cubic diatomic crystals like TlCl, MgO and ZnS,
triatomic crystals like CaF2 and anisotropic crystals like TiO2. In the same
spirit, the Szigeti charge has been evaluated by several workers for
crystals belonging to different structures. Mitra and Marshall [31] quoted
values for the fully covalent diamondtype crystals and to several
partially covalent crystals like ZnS. s values for several crystals are given
in table 3.
68
factors could be that ‘the bond has a partial homopolar character’ but he
did not pursue this possibility in any of his later papers. However, from
time to time, the view has been expressed that the Szigeti charge is a
measure of the ionicity of the bond in the crystal [15,36,41,43]. There are
several measures of the ionicity. A commonly employed measure is the
Phillips ionicity fi [44]. Phillips plotted fi against an effective charge S
and obtained a linear plot. However, the effective charge S used by
Phillips has a different definition and, further, he considered only crystals
with the zinc blende structure. Gervais [39] examined the correlation of fi
and the Szigeti charge s by plotting s against exp(fi) for over 60 crystals
belonging to several structures. The plot (figure 3) was linear with data
points evenly scattered around the straight line. Thus, the Szigeti charge s
can be considered as a measure of ionicity of the bond.
s = krm, (22)
69
Where m is 0.5 for all plots and the constant k has a different value
for each ion. From the plots it is seen that k increases with the size of the
anion. Lowndes and Martin [10] found that a plot of s and the radius ratio
(r+/r−) for the alkali halides is a smooth curve (figure 5) with a separate
curve for each halogen (or alkali ion). Koh [43] obtained linear s vs.
(r−/r) plots (figure 6). Assuming equations of the type
s = a(r−/r) + b, (23)
70
αs2 = constant. (24)
This relation holds well in the case of the alkali halides with NaCl
structure. Earlier, Megaw [47] proposed the relation
71
This relation explains the difference in the thermal expansion of
crystals with different valencies. Sirdeshmukh [46] pointed out that, in
view of Szigeti’s theory, eq. (26) reduces to eq. (24).
72
the plot of A and (1−s) (figure 11) was linear. A linear plot was also
obtained for (1 − s) and A of alkali halides doped with Cr+ ions.
The first term is due only to change in temperature and the second
represents change due to thermal expansion. The net value of (1/s)(ds/dT)
is due to the result of the two competing contributions at different
temperatures.
73
The results obtained by them were provisional as they used
theoretical values for the mode Gruneisen parameter γt. Shanker et al
[58], on the other hand, calculated γt from the experimental data of
Lowndes and Rastogi [59] on temperature variation of the transverse
optical phonon frequency. Using these experimental γt values, they
calculated (V/s)(ds/dV ) from eq. (43). These values of (V/s)(ds/dV ) are
given in table 4. Shanker et al [58] also calculated (V/s)(ds/dV ) from
various dielectric models. These values are also given in table 4. The data
given in table 4 reveal the following features:
74
frequency data of Ferraro et al [65] is used while it is nonlinear with the
data of Fertel and Perry [66]. Thus, the frequency data plays an important
role.
75
CHAPTER 5
FILTERING PROPERTIES OF MATERIALS
_
Dv _ 1 _ _ _ _ _
d ρ (n ') (5.1)
m = − e E + ∇ × B − m v ∇ − ∇ ∫ 0n ( r , t )
Dt c n'
_
_ _
4π _ 1 ∂ D
∇× B = J+ (5.2)
c c ∂t
∂n _ _
= − ∇ .(n v) (5.3)
∂t
_ _
4π e _ _
∇.E = N ( r ) − n ( r , t) (5.4)
E
+
D
Equation (5.1) is the Euler’s equation of motion. The operator is co
Dt
moving time derivative given by
76
D ∂ ∂ ∂ ∂
= + Vx + Vy + Vz (5.5)
Dt ∂t ∂x ∂y ∂z
_
Where Vx , Vy and Vz are the components of velocity V ( r , t ) in x, y
Dv
and z directions respectively. gives the acceleration of an electron in
Dt
the fluid and may be written with the help of equation (4.5) as
Dv ∂v _
= + (v. ∇)v (5.6)
Dt ∂t
The first term on the right hand side of equations (5.1) is force
exerted on a free electron by the electromagnetic field which is produced
due to the density of fluctuations of the electron gas. The electric and
magnetic fields may be written in terms of scalar and vector potential as
_
_ _
1∂A
E = −∇ Φ − (5.7)
c ∂t
_ _ _
B = ∇× A (5.8)
The second term of equation (5.1) is the restoring force due to the
short range interactions with the background i.e. the interaction of the
electrons with lattice vibrations. υ represents the collision frequency and
is assumed to be constant. The last term represents the force due to
pressure ‘P’ and an electron in the mass of the fluid. The pressure is taken
to be of Fermi type assuming that the electrons are totally free i.e.
5
h 2 (3π 2 )2 3 n 3 5
P ( n) = = ξn 3
(5.9)
5m
h 2 (3π 2 ) 2 / 3
And ξ = (5.10)
5m
77
With these substitutions eqn. (5.1) reduces to
_
∂ v
m
∂t
(5.11)
1 ∂ A e
n ( r ,t )
d ρ ( n ')
= e
∂
+ ∇φ − v × ( ∇ × A ) − m ( v .V ) v − m v V − ∇ ∫
c t c 0
n '
Where ‘m’ and ‘e’ represent the mass and charge of a free electron
and n( r , t ) is the instantaneous electronic concentration at a position r .
2 ε ∂2 ε ∂φ 4π e
∇ − 2 2 A = ∇ + nv (5.12)
c ∂t c ∂t c
the effective charge of the ith ion. Since this equation describes the
motion of free electrons only or independent plasma oscillations, the
dielectric equation used is approximated to the frequency independent
value ε with the help of equation (5.7) and choosing the gauge ∇. A = 0
equation (5.4) reduces to –
78
4π e
∇ 2φ =
n( r , t ) − N + ( r ) (5.13)
ε
e
−∇ψ = V − A (5.14)
mc
∂ψ
∇
∂t
2 n (5.15)
1 e e e 1 dP(n ')
= ∇ ∇ψ − A − ∇φ + v A − ∇ψ + ∇ ∫
2 mc m mc m 0 n'
2
V × ∇ × Α = ∇ ∇ψ −
e 1 e
( ) V .∇ V +
mc 2 mc
(5.16)
2 n
∂ψ 1 e e 1 dp(n ')
= ∇ψ − A − φ + ∫ (5.17)
∂t 2 mc m m 0 n'
2 ε ∂2 ε ∂φ 4π e e
∇ − 2 2 A = ∇ + n −∇ψ + A (5.18)
c ∂t c ∂t c mc
∂n e
= ∇. n ∇ψ − ( A) (5.19)
∂t mc
79
4π e
and ∇ 2ψ =
n( r , t ) − N + ( r ) (5.20)
ε
Here author shall give only non relativistic treatment that is the
speed of electromagnetic propagation c has been taken to be much larger
than phase velocity v of surface wave. Therefore, for c>> v equations
(5.14) and (5.17) to (5.20) reduce to
v = −∇ψ (5.21)
n
∂ψ 1 eφ 1 dP(n ')
= ( ∇ψ ) − + ∫
2
(5.22)
∂t 2 M m 0 n'
∂n
= ∇[ n∇ψ ] (5.23)
∂t
4π e
∇ 2φ = ( n − N+ ) (5.24)
ε
n(r ,t ) = n0 (r ) + λ n1 (r , t ) + λ 2 n2 (r , t ) + +
ψ (r ,t ) = ψ 0 (r ) + λψ 1 (r , t ) + λ 2ψ 2 (r , t ) + +
80
Where n0 >> n1 >> n2 so that only first perturbation is sufficient to
describe the system, λ is a parameter and is a measure of perturbation.
e 1 n dP ( n ')5 / 3
φ0 = ξ ∫ (5.26)
m m 0 n'
e 5
φ0 = ξ n2 / 3 (5.27)
m 2m
4π e
and ∇ 2φ = [ n0 − N + ] (5.28)
ε
In first order
n
∂ψ e 1 dP (n ')5/ 3 ∂ψ 1 e 5
= φ1 + ξ ∫ or = φ1 + ξ n1/0 3 n (5.29)
∂t m m 0 n' ∂t m 3m
∂n1
= ∇(n0 ∇ψ T ) (5.30)
∂t
4π e
and ∇ 2φ1 = n1 (5.31)
ε
n0 (r ) = n0 (5.32)
81
Equations (5.29) to (5.31) then modify for semiconductor to give
∂ψ 1 e 5
= − φ1 + ξ n1/ 3 n1 (5.33)
∂t m 3m
∂n1
= n0 (∇ 2ψ 1 ) (5.34)
∂t
4πe
∇ 2φ1 = n1 (5.35)
ε
∂2 vF 2
2
2 + ω p − ∇ n1 = 0 (5.36)
∂t 3
1/ 2
5ξ
vF = n02/ 3 (5.37)
m
2 vF2 k 2 2 vF2 2 2
−ω + ω 2
p + n1 ( r , t ) = 0 or ω =ω + k p (5.39)
3 3
82
This is the dispersion relation for bulk plasma. It shows that the
plasma energies are shifted by a term k 2 . If one takes into account the
finite speed of propagation of hydrodynamic disturbance in the electron
gas. For a metal as ε = 1 , one has only to modify ω p from the value
1/ 2 1/ 2
4π n0 e 2 4π n0 e 2
to
εm m
The relation (5.39) differs from the bulk plasma dispersion relation
of Pines [129], obtained by using quantum mechanical treatment of
random phase approximations viz.
vF2 2 h 2 k 2
ω 2 = ω p2 + k + +−−−−−−− (5.40)
3 4m 2
Now the difficulty arises when the author want to apply the
hydrodynamic equations to solid state electron gas for relatively high
plasma frequencies. Futter [130] has considered this point in his study of
83
layered plasmas and suggested that for a Fermi gas of D dimensions, the
factor 3 in the value of β = vF / 3 should be replaced by ‘D’ and to use β
acoustic (sound) wave, but when ω >> vF ,then β becomes the speed of
optical wave. Therefore in this region the term due to restoring force
equation (5.15) may be neglected safely.
ω 2 = ω 2p + β 2 k 2 (5.41)
∂2
And 2 + ω p2 − β 2∇ 2 n1 (r , t ) = 0 (5.42)
∂t
84
Where
β = (3 / 5)vF Or β = 3π 2 n0
3 h 1/ 2
5 m
(β ∇
2 2
+ ω 2 − ω p2 ) ∇.E = 0 ( 5.43)
And (c ∇ 2 2
)
+ ε Lω 2 − εω p2 ∇ × E = 0 (5.44)
Therefore
(∇ 2
− γ 2 ) ∇.E = 0 (5.45)
85
(∇ 2
− γ 2 )∇× E = 0 (5.46)
ω p2 − ω 2
Where γ= (5.47)
β2
ε (kω )ω p2 − ε L (kω )
α= (5.48)
c2
the solution of eqn (5.45) and eqn (5.46) can be written in terms of
scalar potential function ψ γ and ψ α [96]
E γ = ∇ψ γ (5.49)
α
ˆ ψα )
E 1 = ∇ × ( rr (5.50)
α 1 (α )
E2 = (∇ × E ) (5.51)
α
r
Where r is radius of cylinder and r̂ the unit vector along r . The
scalar potential functions ψ γ and ψ α can be expanded in terms of
cylindrical harmonics, as
ψ γ (r ) = ∑ψ γ l (r )γ lm (θ ).ei k .r (5.52)
l ,m
And ψ α = ∑ψ α (r )Y
l ,m
l
m
(θ ).eik .r (5.53)
86
∂2 1 ∂2 ∂2 1 ∂
∇2 = + + + (5.54)
∂r 2 r 2 ∂θ 2 ∂z 2 r ∂r
∂2 1 ∂ 1 ∂2 ∂2
= 2 + + 2 2
+ 2 ψ γ (5.57)
∂ r r ∂r r ∂θ ∂Z
∂ ∂
ψ γ = ∑ψ γ l ( r ) Yl m (θ ) eikr
∂r ∂r lm
= ∑ψ γ l ( r ) Yl m (θ )( ik ) eik .r + ∑ ψ rl ( r ) Yl m (θ ) eikr
lm l ,m
∂2 ∂
ψ γ = ∑ψ γ l ( r ) Yl m (θ )( ik ) eik .r + ∑ψ rl' ( r ) Yl m (θ ) eikr
∂r 2
∂r lm l ,m
87
( lm
m '2
= ∑ψγ l ( r ) Yl (θ )( ik ) e + ∑ψ rl ( r ) Yl (θ )( ik ) e
ik .r m
l ,m
ikr
)
2
+ ∑ψ γ' l ( r ) Yl m (θ )( ik ) eik .r + ∑ ψ rl'' ( r ) Yl m (θ )( ik ) eikr
lm l ,m
But
∇.E = ∇ 2ψ γ
∂2 1 ∂ 1 ∂2 ∂2
= 2 + + 2 2
+ 2 ∑ψ rl ( r ) Yl m (θ ) eikr
∂r r ∂r r ∂θ ∂Z lm
∂2
(
= 2 ∑ψ γ l ( r ) Yl m (θ ) eikr +
∂r lm
1 ∂
)
∑ψ γ l ( r ) Yl m (θ ) eik .r
r ∂r lm ( )
1 ∂2 ∂2
+ 2
r ∂θ 2
∑
lm
(
ψ γl ( r ) Yl
m
(θ ) e ikr
+ 2
∂z lm
)
∑ψ γ l ( r ) Yl m (θ ) eik .r (5.58)
∇.E = ∑ψ γ l (r ) ylm (θ ) + ∑ψ γ' l (r ) ylm (θ ) (ik )2 eikr
lm lm
+ ∑ψ γ' l (r ) ylm (θ ) + ∑ψ γ'' l (r ) ylm (θ ) (ik )eikr
lm lm
1
+ ∑ψ γ l (r ) ylm (θ )(ik )eikr + ∑ψ γ' l (r ) ylm (θ )eikr
r lm lm
''
1
+
r2
∑ψ γ l (r ) ylm (θ )(ik )eikr
lm
(5.59)
1
∇.E = ∑ψ γ l (r )Yl m (θ )(−k 2 )eikr + ∑ ψ γ l (r )Yl m (θ )eikr
lm r lm
88
= ∑ψ γ l ( r )Yl m (θ )(ik − k 2 )eikr (5.60)
lm
(∇ 2
− γ 2 ) ∇.E = ( ∇ 2 − γ 2 ) ∑ a*ψ γ l ( r )Yl m (θ )e ik .r = 0
lm
(5.62)
∂2 1 ∂ 1 ∂2
2 ∂2
∇ = 2 + + +
∂r r ∂r r 2 ∂θ 2 ∂Z 2
lm
∂ 1 ∂ 1 ∂2 ∂2 2
2
= 2 + + 2 2 + 2 −γ .∑a*ψγl (r)Ylm(θ)eik.r = 0
∂r r ∂r r ∂θ ∂Z lm
∂2 1 ∂
(∇ 2
−γ 2
)∑a ψ *
γl ( r )Yl (θ )em ik .r
= 2 + .∑ a ψ γ l ( r )Yl (θ )e
* m ik .r
lm ∂r r ∂r lm
1 ∂2
+
r 2 ∂θ 2
∑ a ψ γ (r )Y
lm
*
l l
m
(θ )eik .r − (γ 2 )∑ a *ψ γ l ( r )Yl m (θ )eik .r = 0
lm
∂2 1 ∂ ∂2 1 ∂
∑ a ψ γ l ( r )Yl (θ )e
* m ik . r
2+ 2 +
∂r r ∂r ∂r r ∂r lm
−(γ 2 )∑ a*ψ γ l ( r )Yl m (θ )eik .r
lm
89
1 ∂2
2 ∑
= 2 a*ψ γ l (r )Yl m (θ )eik .r (5.63)
r ∂θ lm
r2
×
∑ a*ψ γ l (r )Yl m (θ )eik .r
lm
∂ 2 1 ∂ 2
∑ a ψ γ l (r )Yl (θ )e − (γ )∑ a ψ γ l (r )Yl (θ )e = l
* m ik .r 2 * m ik .r
2 +
∂r r ∂r lm lm
Then
∂2 1 ∂
∑ a ψ γ l (r )Yl (θ )e
* m ik . r
2+
∂r r ∂r lm
l2
− γ 2 + 2 ∑ a*ψ γ l ( r )Yl m (θ )eik .r
r lm
l2
= 2 ∑ a*ψ γ l (r )Yl m (θ )eik .r
r lm
∂2 1 ∂
∑ a ψ γ l (r )Yl (θ )e
* m ik .r
2+
∂r r ∂r lm
2 l2 (5.64)
− γ + 2 ∑ a*ψ γ l (r )Yl m (θ )eik .r = 0
r lm
Now
∂
∑
∂r lm
a*ψ γ l ( r )Yl m (θ )eik .r
(a)
= ∑ a*ψ γ l (r )Yl m (θ )(ik )eik .r + ∑ a*ψ γ' l (r )Yl m (θ )(ik )eik .r
lm lm
And
90
∂2
2 ∑
a*ψγ l (r)Yl m (θ )eik.r = ∑a*ψγ l (r)Yl m (θ )(ik)2 eik.r
∂r lm lm
+∑a*ψγ' l (r)Yl m (θ )(ik)eik.r + ∑a*ψγ' l (r)Yl m (θ )eik.r
lm lm
+∑a ψγ l (r)Yl (θ )e
* '' m ik .r
lm
∂2 ∂
= ∑ a 2 ψ γ l + 2(ik ) ψ γ l − k 2ψ γ l Yl m (θ )eik .r
*
(b)
lm ∂r ∂r
∂ ∂2
Substituting the values of and 2 from (a) and (b) in eqn (5.64)
∂r ∂r
∂2 1 ∂
∑ a ψ γ l (r )Yl (θ )e
* m ik .r
2+
∂r r ∂r lm
2 l2
− γ + 2 ∑ a*ψ γ l (r )Yl m (θ )eik .r = 0
r lm
∂ 2 ∂ 2 m ik .r
2 ψ γ l (r ) + 2(ik ) ψ γ l (r ) − k ψ γ l (r ) Yl (θ )e
∂r ∂r
1 ∂ m
∑ a*
γl
+
r
ψ ( r )(ik ) +
∂r
ψ γl ( r ) Y
l
(θ )eik .r
=0 (5.65)
lm
l
2
− γ 2 + 2 ∑ a*ψ γ l (r )Yl m (θ )eik .r
r lm
91
∂2 ∂
2 ψ γ l (r ) + 2(ik ) ψ γ l (r ) − k 2ψ γ l (r )
∂r ∂r
1 ∂
m ik .r −ω t
γl
+ ψ ( r )(ik ) + ψ γl ( r ) Y (θ )e =0
r ∂r (5.66)
l
2
− γ + 2 ∑ a ψ γ l (r )
2 *
r lm
∂2 1 ∂ 2 2 l2
ψ γ l (r ) + ψ γ l (r ) − γ + k + 2 ψ γ (r ) = 0 (5.67)
∂r 2 r ∂r r
∂ 1
2(ik ) ψ γ (r ) + ψ γ (r ) = 0 (5.68)
∂r r
2 2 2
Let us assume that γ + k = q
∂2 1 ∂ 2 l2
ψ γ (r ) + ψ γ (r ) − q + 2 ψ γ (r ) = 0 (5.69)
∂r 2 r ∂r r
ψ γ ( r ) = ΑΙ l ( qr )
2 2 2
where q = γ + k (5.70)
92
ψ γ l (r , t ) = ∑ ΑΙ l ( qr )Yl m (θ )ei ( kr −ωt ) r<R
l ,m
=0 r>R (5.71)
4πe
∇ 2φl (r , t ) = ψ rl (r , t ) (5.72)
ε
Now
4πe i ( kr −ωt )
∇ 2 ∑ φl (r )Yl m (θ )e i ( kr −ωt ) = ∑ l ΑΙ ( qr )Yl
m
(θ )e
l ,m ε l ,m
∂2 1 ∂ 1 ∂2 ∂2
2 + + 2 2
+ 2 ∑φl (r )Yl m (θ )e i ( kr −ωt )
∂r r ∂r r ∂θ ∂z l ,m
4π e i ( kr −ωt )
∑ ΑΙl (qr )Yl (θ )e
m
=
ε l ,m
∂2 1 ∂ 1 ∂2
2 + + ∑ φ l ( r )Yl m (θ )e i ( kr −ωt )
∂r r ∂r r 2 ∂θ 2 lm
93
= ∑ Α1Ι l (qr )Yl m (θ )ei ( kr −ωt ) (5.75)
l ,m
Taking z as constant
∂2 1 ∂
2 + ∑ φl (r )Yl m (θ )e i ( kr −ωt ) − ∑ Α1Ι l (qr )Yl m (θ )e i ( kr −ωt )
∂r r ∂r lm l ,m
1 ∂2
= 2 2 ∑ φl (r )Yl (θ )e
m i ( kr −ωt )
(5.76)
r ∂r lm
r2 ∂ 2 1 ∂ i ( kr −ωt )
2 + φ
l ( r ) − Α Ι ( qr ) e
∑ φl (r )ei (kr −ωt )
lm
∂r r ∂r
1 l
1 ∂2 m
= m 2 Yl (θ ) (5.77)
Yl (θ ) ∂r
∂ 2 1 ∂
2 + φl (r ) i ( kr −ωt ) 2
2
r
∂r r ∂r e =l (5.78)
∑ φl (r )ei ( kr −ωt ) −Α Ι ( qr )
lm 1l
l 2 ∑ φ l (r )e i ( kr −ωt )
∂ 2
1 ∂
2 + φ l (r ) − Α1Ι l (qr ) = lm
e i ( kr −ωt ) (5.79)
∂r r ∂r r 2
or
∂2 1 ∂ l2
2 + − 2 φl (r ) = Α1Ι l ( qr )e i ( kr −ωt ) (5.80)
∂r r ∂r r
94
On Simplifying, we get
∂2 2 ∂ 2 l2
φl ( r ) + φl ( r ) − k + 2 φl (r )
∂r 2 r ∂r r
(5.81)
∂ 1
+2(ik ) φl (r ) + φl (r ) = ΑΙ l (qr )ei ( kr −ωt )
∂r r
∂2 2 ∂ 2 l2
φl (r ) + φl (r ) − k + 2 φl (r ) = Α1 I l (qr ) (5.82)
∂r 2 r ∂r r
∂ 1
φl ( r ) + φl ( r ) = 0 (5.83)
∂r r
C
φl (r ) = (5.84)
r
φ l ( r ) = C .F + P.I (5.85)
Where I l (qr ) and K l (qr ) are modified cylindrical Bessel function of first
and second kind respectively. The value of these modified cylindrical
Bessel function may be calculated in terms of the following relation
95
π
I l ( x) = I 1 (5.87)
2x l+ 2
π
K l ( x) = K 1 (5.88)
2x l+ 2
Α1 I l ( qr )
P.I = (5.89)
q2 − k 2
Substituting the eqn (5.89) and (5.86) in eqn (5.85), we get the
general solution as
Α1 I l (qr)
φl (r ) = BI l (qr) + CKl (qr) + (5.90)
q2 − k 2
4πeΑ
Where B and C are constant and Α1 =
2 2 2
and q = γ + k
ε
Α1 I l ( qr )
ψ γl ( r ) = + BI l (qr ) + CK l ( qr ) (5.91)
q2 − k 2
φli r =R
= φle r=R (5.92)
96
And
∂ ∂
ε1 φli = ε 2 φle (5.93)
∂r r = R ∂r r = R
B1 = M 1 Α (5.94)
C1 = N1 Α (5.95)
Where
1 1
N1 = M 1 I l (kR) + 2 I l (qr ) (5.97)
K l (kR) r
This with the help of eqn (5.94) and eqn (5.95), equation (5.91) can be
written as
ψ γl ( r ) = ΑI l ( qr ) + BI l ( kr ) + CK l ( kr ) (5.98)
= Α[I l ( qr ) + M 1 I l ( kr ) + N 1 K l ( kr ) ]
= ΑX l (γkr ) (5.99)
Where X l (γkr ) = I l ( qr ) + M 1 I l ( kr ) + N 1 K l ( kr )
97
ψ γ = ∑ψ γl Yl m (θ )e ikr = ∑ ΑX l (γkr )Yl m (θ )e ikr (5.100)
lm lm
∂
Erlγ = ∇ψ γ = ΑX l (γ kr )Yl m (θ )
∂r
∂
or Erlγ = ΑX l (γ kr )Yl m (θ ) (5.101)
∂r
∂ ∂ m
Eθγl = Α. X l (γ kr ) ∂θ Yl (θ ) (5.102)
∂r
1
B= (∇ × E ) (5.103)
iω
From eqn (5.50) and eqn (5.101) and (5.102) it is clear that
Brlγ = 0 (5.104)
γ
And Bθ l = 0 (5.105)
Similarly eqn (5.52) along with eqn (5.51) yields the electric and
magnetic field components as
r Α 2Yl (α kr ) ∂ m
Eθ(αl2 ) = Yl (θ ) (5.107)
iα r ∂θ
98
and
Brl(α2 ) = 0 (5.108)
αΑ 2Yl (α kr )
Bθ(αl2 ) = (5.109)
ω sin θ
The function
ψ δ l (r ) = Α3 [ hl (δ r ) + N 3 K l (kr ) + M 3 I l ( kr ) ] (5.111)
where
1
N2 = [ M1I l (kR) + Il ( p1R)] (5.113)
kl (kR)
1 p2 ε h p (δ R ) K l ( kR ) − ε B hl (δ R ) K l ( kR )
' '
M2 = (5.114)
k ε B I l ( kR ) K l' ( kR ) − ε I l' ( kR ) K l' ( kR )
Where p = kR (5.115)
2 ω2
δ = εB (5.116)
c2
99
Where p1 = α 2 + k 2 and p2 = δ 2 + k 2 hl (δ R ) is the
cylindrical Hanke function and given in equation (5.111) and ε B is the
dielectric of the bounding medium.
δ l 2ψ δ l (r )Yl m (θ )
Erl =
iδ (5.117)
2
rψ (r ) δ Yl m (θ )
E = δl
δ
θl (5.118)
iδ r ∂θ
Brlδ = 0 (5.119)
δ ∂ψ δ l (r )∂Yl m (θ )
B =
θl (5.120)
ω sin(θ )∂ 2θ
Eθ P
= Eθ B
at r=R (5.121)
Dr P
= DB at r=R (5.122)
Bθ P
= Bθ B
at r=R (5.123)
100
Where the subscript ‘P’ and ‘B’ denotes the polar semiconductor
and bounding medium respectively. ‘R’ represents the boundary of the
polar semiconductor cylinder and the bounding medium such that
Now let us assume that from equation (5.110) and (5.111), we get
ψ rl (r ) = Α2 yl (α kr ) (5.125)
And ψ δ l ( r ) = Α3 zl (α kr ) (5.126)
Where zl (δ kR ) = hl (δ R ) + M 3 K l ( kR ) + N 3 I l ( kR ) (5.127)
101
ε1 (kω )l 2Yl (α kR ) Rε1 ( kω ) X l' (γ kR ) ε B (kω )l 2 zl (δ kR )
' '
[ RYl (α kR)] X l (γ kR ) [ Rzl (δ kR)] =0 (5.130)
iα 2Yl (α kR ) 0 −i 2δ 2 z (δ kR )
2 1 − Ω2
γ = (5.133)
ξ2
And δ 2 = ε B ( kω ) − Ω 2 (5.134)
Now the two mode coupling eqn of SPSOP electro state surface
modes by taking the limit ∝R→O, R → O.
102
ε
ε̅ l Ω
ω
ω
So that eqn becomes
* Ω + 0 (5.135)
ω
ω
[Il1(x)  Il+1(x) ]

I1(x) = (5.136)
and I' (x) = [Il1(x) + (l+1) Il+1(x)] (5.137)
ω 1
.'ε l 1 ϵ∞ l('1 ! (0
.'ε l 1 ϵ∞ I
!"#$ /%
"&$ /% ω
ω
ε̅ l 0 '1 ! (0 .ε̅ l'1 ! (
!"#$ /% !"#$ /%
/% "&$ /%
ω
ω "&$ ω
ω
.'ε l 1'1
(0 +0
!"#$ /%
!"&$ /% ω
(5.138)
ε3 ε4
ε∞ ε4
ω2SOP = ωt 2 (5.139)
103
The uncoupled surface plasma frequency may also be obtained
from eqn (5.101) by neglecting the contribution due to surface optical
phonon modes i.e. setting ωt=0 and ε∞ 6666
, ε so eqn (5.138) becomes
1
7 ! 18 + 1
ε̅ε
!"#$ /% ω9 4 ε ε̅ !"#$ /%
"&$ /% ω 4 !"&$ /%
(5.140)
104
AB$ ABC D B6
ABC D E BC
(l+1)B = [ ] 
=0 (5.141 )
+ .FG  F ] Q =0 (5.144)
OL&$ KN OL#$ KN KL&$ KN KL#$ KN
AOL&$ KNA OL#$ KN AKL&$ KNA KL#$ KN
interface for KJ 0.
phonon, polar ton and Plasmon for polar semiconductor of cylindrical
eqn(5.144), we get F as
105
DE
F = FS  F̅
D
(5.145)
FS =
B∞ TBU
T
where
Y = & Z =
D DE
DV DV
(5.146)
Y[Y2(az+b)Y+az] = 0 (5.147)
GB6[ BB∞[
B B∞[ \ B∞[
Where a = , b= (5.148)
only Y=0 , does not give any meaningful solution, then we get
y2  (az+b)Y+az =0 (5.149)
The above equation is quadratic in Y and for each mode l gives two
roots for the given value of the constant a,b for Z.
106
OU KN O KN O$ KN O] KN
OU KN O KN O$ ^N O] KN
C= C=
where P = √` a
QR + i = j Vf =j m 3o pq /k
1efc g k k
B6 h l l
and
rq s = aq s =
tuf vw tuf vw
w w
r s = a s =
xyt vw tuf vw xyt vw tuf vw
w w w w
 
r s = ry s  I1(x) a s = ay s  K1(x)
k k
w w
rk s = r s  I1(x) ak s = a s  K1(x)
l l
w w
ωt = 1.39x1012sec1
107
from eqn(5.149) We get
Y2=(az+b)Yaz (5.151)
108
But at radius 8A. It shows again perfect linear variation for high
value of K.
Again linear variation between wsp/wt versus K disturb for radius
16A. for high value of K.
Variation of wsp/wt versus K for different radius observe that
variation is either perfectly linear or not perfectly linear.
When radius increases then it’s linear variation changes with
wsp/wt versus K, thus for high radius linear variation changes.
109
For same radius neoprene use a medium variation of wsp/wt initially
increases but as K=5, it changes abnormally again at K=7 it fallow
simillar path as in epoxy resine medium follow.
When we increase K more than 9 (K > 9) than it increases rapidly.
Except trans oil medium other medium like mica, quartz, Bee wax,
transformer oil and vaccum shows linear increment of wsp/wt with
K.
Thus wsp/wt increases with increment in value of K
For KF substance at .5A.radius. It is observed that trans oil
medium show different variation otherwise other seven medium
shows same kind of linear variation.
110
Ratio of wsp/wt increases when we move left to right for different
medium with constant value of K.
For same medium wsp/wt increases when K increases.
All medium like Epoxy Resine, Neoprene, Mica, Quartz, Lice, Bee
Wax, Transformer Oil & Vaccum share linear variation between
wsp/wt versus K. Thus for different medium variation are same.
For lice medium Ratio of wsp/wt increases with increase in K but
when K>8 then decreases rapidly.
Ratio of wsp/wt becomes constant at K = 9 or greater value for lice
medium.
Even at wave vector K = 0 all medium have same value of wsp/wt.
But when value of wave vector is higher (i.e. K>8) then for lice
medium its linear variation vanishes.
Thus for low value of K all medium shows linear variation while
for high value of K only lice shows different variation when we plot 3D
graph between wsp/wt versus K for different medium.
111
Ratio of wsp/wt is lowest at 5A. radius and wsp/wt is maximum are
in cylindrical shape increases then more space to pass waves. Thus
wsp/wt increases with increases in radius.
For Mgo substance in epoxy resine medium there is constent
variation, It is free from radius.
It has only single effect when radius is more then large amount of
wave pass trough it easily otherwise less amount of wave pass
trough it.
When K = 0 then there is no propagation of waves value of K
increases then wave propagation also increases.
wsp/wt has max value at 16A. means at this radius waves passes
easily through the cylindrical geometry. It cappers for highest value
of K = 10.
112
When we plot graph wsp/wt versus K. Then InP substance for
cylindrical surface of 0.5 A. radius of different medium, then all
medium shows linear variation but lice shows different variation.
In lice medium wsp/wt increases but its value remains constant at
5.040028. So for lice, wsp/wt increases with increment in K but for
K = 7 or more then this, wsp/wt deceases rapidly.
Thus for InP at lone value of K all medium including lice shows
linear variation but for higher value of K, only lice shows different
variation also for high value of K.
113
Calculation
0.576112 0.458197
0.945957 0.814658 0.758192 0.753698 0.753417
1.344748 1.268568 1.257262 1.256526 1.25648
1.768659 1.733666 1.730612 1.73042 1.730408
2.210058 2.195285 2.194229 2.194164 2.19416
2.660146 2.65354 2.653105 2.653078 2.653076
3.112725 3.109517 3.109312 3.109299 3.109299
3.565661 3.563975 3.563869 3.563863 3.563862
4.018341 4.017393 4.017334 4.01733 4.01733
4.470037 4.470037 4.470038 4.470036 4.470036
5
4
Wsp/Wt
1A
3 2A
2 4A
1 8A
0 16A
1 3 5 7 9
Fig. 5.1
114
Calculation
epoxy resine
Neoprene
6 Mica
Wsp/Wt
4 Quartz
2 Lice
0 Bee Wax
1 2
3 4 Transformer
5 6
7 Vacuum
8
9
10
11
K
Fig. 5.2
115
Calculation
Epoxy resen
10 Neoprene
Wsp/Wt
Mica
5 Quartz
Lice
0
1 2 Wax
3 4
5 6 Trasformer oil
7 8
9 Vacuume
K 10
Fig. 5.3
116
Calculation
10
Wsp/Wt
.5A
5
1A
0 2A
1 2 4A
3 4
5 6 7 8A
8
9
10 16A
K
Fig. 5.4
117
Calculation
epoxy
neopren
mica
quartz
1 2
3 lice
4
5
6 beewax
7
8 trance oil
9
vacuume
10
K
Fig. 5.5
118
Calculation
10
Wsp/Wt
1A
5
2A
0
4A
1 2 3 4
5
6
7
8
9
10
K
Fig. 5.6
119
5.3 Special dispersion relation three for mode coupling
form of FS Qz given by the equation (5.48) for polar semiconductor. this
from (5.131) by substituting the frequency and wave vector dependent
substitution leads to
}
B∞ KDΩ BU KD V
R{' (γkR)  ~
}
E
}
Ω V
}
E
Q
F∞ zQΩ
Fq zQ
QR
F̅zQ
Q Ω
Ω
QR
120
Yrl ( + WA rA (pr)
=A A (
Because Ml =0
Nl = 0
and M2 = M3 = N2 = N3 = 0 (5.159)
RI’()(F∞ Ω
Fy
F̅ Ω
F∞ Ω
Fy V Ω m ′
D V D D
D E D R
D E
(5.160)
n (5.160) is the dispersion relation for spherical surface Plasmon and
[150] for K → ∞.
polariton coupled models for polar semiconductors in the case of K = 0
121
The dispersion relation for surface plasma, polaritons and phonons
including spatial dispersion effect as given.
} }
B∞ MDΩ Bc MD V B∞ MDΩ B∞ MD V
~ .'F̅zQ
 ~
} }
E E
} }
Ω V Ω V
RX1(γkr)
} }
E E
DE
Ω _F∞ [Fy 'F̅
1 F∞ a ( 0 Ω1
'1 Fy
DE
D V D V
F̅ * z F̅0 Ω
F̅z + 0
D
E
D V
(5.162)
13.6z + 0 (5.163)
D
> = \
a
DV V
w= and
122
Then F∞ W6[F̅( )  Fy +(1+F∞ K12] W4 + [(1+Fy K12+
E
V
F̅ 0 W2 F̅
E E
V V
K12 = 0 (5.164)
Now equation (5.165) gives three values of Q, Q, &Qk, for
different value of a .
We observe from the graph that the frequencies w1, w2 and w3 vary
with respect to propagation constant K. At high value of K, w1 & w2 are
always constant but the third coupling frequency w3 always increases with
respect to K.
123
Calculation
100
w
50
w3
0 w2
10 20 w1
30
40
50
10 20 30 40 50
w1 0.6501 0.6486 0.6483 0.6482 0.8481
w2 1.1607 1.1575 1.03818 1.03798 1.03789
w3 10.6681 23.3565 33.034 45.932 57.6918
Fig. 5.7
124
Calculation
100
w
0
10 20 w3
30 40 w2
50 w1
10 20 30 40 50
w1 0.90419 2.858 0.903706 0.9036 0.903667
w2 1.06 1.05215 1.05092 1.0505 1.0503
w3 10.1384 20.7643 31.5085 38.2408 50.8616
Fig. 5.8
125
5.4 Local Theory approximation
F£ +
¤D ¦ω §¨
Bc ¥ Bc ¥ ©ω
=1+ P=  E
FQ + 1
§¨
Bc ©ª
So that
§¨
Bc h
ωp 2 =
Thus FQ=1
ω
ω
(5.166)
FQ=FS Q 
ω
ω
(5.167)
126
FS Q is dielectric constant of Lattice vibration. The coupling of
the electric field E of the photon with the dielectric polarisation P of the
transverse optical phonon is described by Electromagnetic wave eqn
C2k2E=ω2 (E+4πP)
E + (ω2ωt2) P = 0
«¬
(5.169)
ω
c k 4πω
®® ®®
= 0 (5.170)
ω
ω
«¬
©
1π«¬
At k=0 there are two roots ω=0 for photon and ω2 = ωt 2 +
for polarization the dielectric function obtained from eqn (5.168)
FQ = F∞ +
1π«¬ /
ω ω
(5.171)
127
Fy = F∞ +
1π«¬
ω
FS Q + F∞ + .FS
F∞ 0
ω
ω ω
B∞ ω
BU ω;
By LST Relation =
So F∞ ω@ = Fy ω
FS (ω) = F∞ + .ω@
ω 0
B∞
ω ω
Then
= F∞ 71 8
ω; ω
ω ω
FS (ω)=
B∞ D± B∞ D
ω ω
FS (ω)=
B∞ ω Bc ω
ω ω
or (5.172)
128
5.5 Dielectric function on the surface materials
FQ = F (∞)  = F (∞)71
8
1π«¨ ω
©ω ω
(5.173)
QR =
1π«¨
Where
ε∞©
So ε =0 at ω=ωp
= c2µ2E
³ ´
³
εω, k=
¸ ¬
ω
129
¹ Complex : for ω real, K is Complex and the waves are
damped in space.
¹ + ∞ ∶
an applied force, thus the poles of εω, K
System has infinite response in the absence of
the medium.
εω, O=1 and εO, K = 1+
ωÉ Ê9
ω Å
So
same limit as Q → O. Thus great care must be taken with the dielectric
Where VS is screening potential but k→ O does not approach the
The Total dielectric function, lattice Plus electrons, but without the
electronic polarizability of the ion cores, is
¹Ë, Ì=1 +
ËÍ Î ÏÐ Î
ËÎ ÌÎ
(5.174)
130
Above eqn is valid when we consider both frequency and
propagation constant k.
ε∞ ω εU ω
ε@ (ω) =
ω ω
ε(W)= ε∞ 71 8
Ω ωÉ
ω ω ω
(5.175)
1π«¨9 ∗ ε∞
Ω = 7 8
ε∞ k
Where (5.176)
ε∞ εÔ
ω
ω
ε@ (ω) =
(5.177)
ω
ω
ε∞ εÔ
ε∞ 71 8=
Ω
ω
ω ω
ω
ω
ω
131
ε∞ ε∞ εÔ
ε∞
1 =
1Ω«¨9 ∗
ω
ω ω ε∞ k
ω
ω
ω
on simplifying
1
ω ε∞ + =7ε
ε∞ 8
1Ω«ε∞ ¨9 ∗
kω
ω ω
ω
(5.178)
1Ω«ε∞ ¨9 ∗
= ε
kω
So ε∞ +
1Ω«ε∞ ¨9 ∗
ε  ε∞ =
kω
kω
eÓ ∗ = 7 87 87 8
εÔ ε∞
1Ω « ε∞
Õ
kω
eÓ = 7 8 .MVØ 0 7 8
∗ εÔ ε∞ Õ
1Ω ε∞
Or (5.179)
«
Va = , Va is volume of one unit cell. By eqn(5.179) Szigeti
effective charge calculated.
ωÉ
ω
ε(ω) = 1
132
None dielectric for polar semiconductor is as
ωÉ
ε(ω) = ε@ (ω) 
ω
ω
ε∞ ε3
ω
ε@ (ω) = ω
ω
ω ω
ε∞ εÔ
So eqn (5.167) become FQ =
ω ω
ω ω
ω ω
graph we observed that there exit two SPSOP models. These modes do
not propagate when ε(ω) is negative.
ω ω
ω ω
i.e <
Ã
ω& ω
ω ω
i.e <1.
133
F = ε@ δij  ε̅ .ω2δij  ωciωuj + iδijkωck]
ωÉ
ω ω ωÝ
(5.180)
1π§3 ¨
where ω¦ =
ε̅ ©∗
>
¨
©∗
and ωc =
= 0, iJj
=o otherwise
134
Calculation
20
w/wt
10
0
0.5 1
1.5 2
2.5
3
Fig. 5.9
135
Calculation
80
60
40
evX2
20
evX2
0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.2 1.4 1.6 1.8 2 2.2
20
40
60
X
Fig. 5.10
136
The variation of w/wt versus propagation constant K in presence of
DC magnetic filed for NaF tube shown in fig (5.9). It shows the
dispersion curves for the coupled SP and SoP waves in NaF cylindrical
tube radius 100A.. Thus dispersion curves shows that the frequency of the
upper mode (Ώ3) varies slowly with the wave rector K. when K is more
than 2 it increases rapidly.
Lower mode (Ώ1) varies very slowly and remains almost constant
when value of wave vector is equal to or greater than unit. But Ώ2
mediator mode lies in between upper and lower mode is exactly constant
for all value of wave vector (K). So there exist a band gap between the
two modes showing that the coupled modes with frequencies lying in this
region can not be excited.
The author observes that there are four range of EM wave for
which cylindrical surface behaves as filtering property of InP cylinder.
These ranges are :
137
third order filtering property of InP for EM waves. For 0.95 to 1.00
EM wave decreases.
iv. For range X = 1.00 to 1.79 the surfaces of InP does not allow to
pass EM waves because at this range X, εvX2 becomes negative and
at X = 1.1, εvX2 =  43.5334.
138
vector K at same radius 1 A., then all substances have same linear
variation.
This dispersion curves shows that the frequency of the lower mode
changes slowly and frequency of the upper mode varies rapidly
with the wave vector K.
Thus wsp/wt verses K is almost linear increment of substances KF, InP &
Mgo at same radius cylindrical surface (1 A.) these behaves like same
nature.
139
At this radius (16 A.) both InP & Mgo have same value of wsp/wt so
both have same type (kind) and equal linear variation.
Calculation
0.576112 0.731606 0.798947
0.945957 1.192797 1.254347
1.344748 1.690038 1.769384
1.768659 2.214776 2.304098
2.210058 2.761879 2.847247
2.660146 3.318774 3.394872
3.112725 3.878741 3.94537
3.565661 4.439395 4.497908
4.018341 4.999995 5.051902
4.470037 5.560347 5.606916
KF Inp MgO
140
Comparative study of 1A radius cylidrical substances
Wsp/Wt
10
5
0
1 2 3 4 5 KF
6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
KF 0.5761 0.9459 1.3447 1.7686 2.2100 2.6601 3.1127 3.5656 4.0183 4.4700
InP 0.7316 1.1928 1.6900 2.2147 2.7618 3.3187 3.8787 4.4393 5 5.5603
MgO 0.7989 1.2543 1.7693 2.3041 2.8472 3.3948 3.9453 4.4979 5.0519 5.6069
Fig. 5.11
10
5
0
1 2 3 4 5 Series1
6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
Series1 0.5761 0.9459 1.3447 1.7686 2.2100 2.6601 3.1127 3.5656 4.0183 4.4700
Series2 0.7989 1.2543 1.7693 2.3041 2.8472 3.3948 3.9453 4.4979 5.0519 5.6069
Series3 0.7316 1.1928 1.6900 2.2147 2.7618 3.3187 3.8787 4.4393 5 5.5603
Fig. 5.12
141
Calculation
0.242957 0.242957
0.753417 0.996888 0.996888
1.25648 1.597475 1.597475
1.730408 2.175628 2.175628
2.19416 2.745552 2.745552
2.653076 3.311498 3.311498
3.109299 3.87521 3.87521
3.563862 4.437539 4.437539
4.01733 4.998952 4.998952
4.470036 5.559727 5.559727
KF Inp MgO
6
4
2
0
1 2 3 4 KF
5 6 7 8 9 10
1 2 3 4 5 6 7 8 9 10
KF 0 0.7534 1.2564 1.7304 2.1941 2.6530 3.1093 3.5638 4.0173 4.4700
InP 0.2429 0.9968 1.5974 2.1756 2.7455 3.3115 3.8752 4.4375 4.9989 5.5597
MgO 0.2429 0.9968 1.5974 2.1756 2.7455 3.3115 3.8752 4.4375 4.9989 5.5597
Fig. 5.13
142
5.9 Variation of the width of allowed band of polarized
materials with Szigeti effective charge
ωÉ à
ω
εÔ
ε(ω) = ε∞ Þ
ß
ω
ω ωÕ
ω
(5.181)
ω ω
ωä /
ω =
ε ε ω ä
ε3 ε3 ωÉ ωÉ
1 ε∞
ω
ω ω
(5.182)
∞ ∞ ε∞ ε∞
Thus the surface acts as band pass filter [BPF]. Agian their is no
propagation when ε(ω) is negative.
¢
ω ω#
ω ω
for , the surface become transparent again act as High pass
filter (HPF). from graph we observe band width (∆) of band pass filter is
given by
∆¦æ + 1
ω&
ω
(5.183)
Similarly ∆ç¦æ + 1
ω#
ω
(5.184)
143
study of Szigeti effective charge eÓ∗ versus ∆ with the help of data in
gives values at which the surface will act as high pass filter. from
∆ =0.074+0.099
¨9∗
¨
(5.185)
from table our calculated values agree well with the experimental values
of Hass and Henvis.
pass filter for ω <ω<ωt and as high Pass Filter for Qw¢ω+. The es2 is
We can conclude that polar semiconductor surface acts as a band
¨9∗
¨
When we plot graph between band gap (∆) and , we obtain we
From the graph we observe that the band width (∆) of the band
depends on єy , єê , QR and Q . thus, the width
è&
è
pass filter is given by 1
144
è#
è
, which gives the value at which the surface will act as high on the
145
Calculation
0.1
0.08
0.06
0.04
0.02
0
0 0.2 0.4 0.6 0.8 1
e*/e
Fig. 5.14
146
5.10 Advantage of szigetti effective charge study and its application
(1/M) = 1/m1 + 1 / 2 m2
mass of ion with lesser valency. The value of t∗ obtained by szigetti [2]
Where m1 is the mass of ion with higher valency and m2 is the
for CaF2 type crystals. These value of es are of the same order as but
systematically lower than those for the highly ionic alkali halides.
The long wave lattice dynamics of the fluorite lattice was studies
by Axe [73]. For he obtained the same relation [(eqn) as obtained by
szigeti [3] except that the definition of the reduced mass used by Axe
[73]. The value of es obtained by Axe and his coworkers for several
crystals with caF2 structure. The es values for tow oxides with CaF2
structure are distinctly low and indicate ‘Increased covalency’ [75].
147
provided an insight in the complex polarization mechanisms in solids.
Some new results have been discussed. There are : 
148
it shows behavior of refractive index with excitation intensity Io. In
this figure it shows variation of excitation intensity in absence and
presence of szigetti effective charge.
149
CHAPTER 6
SUMMARY AND CONCLUSIONS
150
PhononPolariton for k≠o. The effect of Spatial dispersion on the
Properties of Surface Polariton wave have also been studied by
comparing the result obtained when Spatial dispersion is taken into
account, with those obtained in the local limit i.e. when spatial dispersion
is neglected.
When the dipole moment density and also the displacement field
depends not only on the value of electric field at a particular point but
also on the value assumed by the electric field in the near vicinity of that
point, then material is said to be exhibit “spatial dispersion”. The
inclusion of spatial dispersion in the study of surface polaritons leads to
important conclusion like the finite life time of the surface modes even
collision less plasma system. The properties of surface polaritons can be
studied by the help of their dispersion relation. The non local dielectric
function is both frequency and wave vector dependent although in local
limit it becomes wave vector independent and depends only on
frequency. The non local dielectric function can be calculated either by
using the quantum mechanical or classical approach. The dispersion
relation for surface plasmon, surface optical phonon & surface polariton,
all modes has been obtained and an attempt to study the effect of spatial
dispersion on surface plasmon surface optical phonon modes of polar
semiconductor.
151
These modes indicate the type of elementary excitations that give
rise to the polariton waves. The incident electromagnetic radiation can
couple to more than one type of elementary excitations simultaneously if
their frequencies are comparable. These waves arise as a result of
coupling of the EM radiation and those elementary surface excitations of
the medium that may couple on linear manner to the incident EM field by
virtue of their electric of magnetic character. The predominant surface
excitations in the case of dielectric media are surface plasmons, surface
phonon, surface excitons etc. The fact that surface EM waves at a metal
surface involves the coupling of EM radiation .with surface plasmons was
first explain by Stern . Also surface of optical lattice vibrations that exist
at the surface of ionic and polar materials have a great tendency to couple
with electromagnetic radiation. These vibrations lead to surface phonon,
polariton modes . In case of polar semiconductors, surface Plasmon
modes and surface Phonon modes have comparable frequencies so that
there is a possibility of interaction between the two leading modes i.e.
surface plasmon & surface polariton modes.
152
The above mentioned techniques have been successfully employed
by several workers to study surface polariton modes in metal,
semiconductors and polar semiconductors for different interface
geometries. The theoretical study of these modes has also recently
received great attention. The properties of the surface modes can be
studied by deriving their effect on the reflection and refraction of incident
EM waves plays a fundamental role in the Physics of surfaces.
153
between these two modes in se miconductor. The dispersion for surface
Plasmon and polariton is given as below
ε 1 ( k ω ) l 2Yl (α kR ) X l (γ kR ) ( − i 2δ 2 z (δ kR ) − 0 )
[ RYl (α kR ) ] ( − i 2δ 2 z (δ kR ) )
− ε 1 ( k ω ) X l' (γ kR )
− [ Rz l (δ kR ) ] iα Yl (α kR )
' 2
−ε B (kω)l 2 zl (δ kR) 0 − Xl (γ kR) × iα 2Yl (α kR) = 0 (6.1)
ωt = 1.39x1012sec1
154
Z=( = (4.46)2 = 19.89, = 3.7 x 102/A
Y2 = (az+b)Yaz (6.2)
155
Thus for higher value K it shows non linear variation. Hence for
higher value of K, it shows non linear agreement.
Again linear variation between wsp/wt versus K disturb for radius
16A. for high value of K.
Variation of wsp/wt versus K for different radius observe that
variation is either perfectly linear or not perfectly linear.
When radius increases then it’s linear variation changes with
wsp/wt versus K, thus for high radius linear variation changes.
156
For same radius neoprene use a medium variation of wsp/wt initially
increases but as K=5, it changes abnormally again at K=7 it fallow
similar path as in epoxy resine medium follow.
When we increase K more than 9 (K > 9) than it increases rapidly.
Except trans oil medium other medium like mica, quartz, Bee wax,
transformer oil and vacuum shows linear increment of wsp/wt with
K.
Thus wsp/wt increases with increment in value of K
For KF substance at .5A.radius. It is observed that trans oil
medium show different variation otherwise other seven medium
shows same kind of linear variation.
157
Ratio of wsp/wt increases when we move left to right for different
medium with constant value of K.
For same medium wsp/wt increases when K increases.
All medium like Epoxy Resine, Neoprene, Mica, Quartz, Lice, Bee
Wax, Transformer Oil & Vaccum share linear variation between
wsp/wt versus K. Thus for different medium variation are same.
For lice medium Ratio of wsp/wt increases with increase in K but
when K>8 then decreases rapidly.
Ratio of wsp/wt becomes constant at K = 9 or greater value for lice
medium.
Even at wave vector K = 0 all medium have same value of wsp/wt.
But when value of wave vector is higher (i.e. K>8) then for lice
medium its linear variation vanishes.
Thus for low value of K all medium shows linear variation while
for high value of K only lice shows different variation when we plot 3D
graph between wsp/wt versus K for different medium.
158
When value of K increases then wsp/wt increases Radius of cylinder
increases then wsp/wt also increases.
Ratio of wsp/wt is lowest at 5A. radius and wsp/wt is maximum are
in cylindrical shape increases then more space to pass waves. Thus
wsp/wt increases with increases in radius.
For Mgo substance in epoxy resine medium there is constent
variation, It is free from radius.
It has only single effect when radius is more then large amount of
wave pass trough it easily otherwise less amount of wave pass
trough it.
When K = 0 then there is no propagation of waves value of K
increases then wave propagation also increases.
wsp/wt has max value at 16A. means at this radius waves passes
easily through the cylindrical geometry. It cappers for highest value
of K = 10.
159
wsp/wt versus K is almost linear increment in all medium except
lice medium.
Value of wsp/wt is high in vaccum medium and low in epoxy resine
but the range is nearly equal for same value of K.
When we plot graph wsp/wt versus K. Then InP substance for
cylindrical surface of 0.5 A. radius of different medium, then all
medium shows linear variation but lice shows different variation.
In lice medium wsp/wt increases but its value remains constant at
5.040028. So for lice, wsp/wt increases with increment in K but for
K = 7 or more then this, wsp/wt deceases rapidly.
Thus for InP at lone value of K all medium including lice shows
linear variation but for higher value of K, only lice shows different
variation also for high value of K.
160
Author observe that for InP in epoxy resine medium variation of
wsp/wt is linear with K at different radius such that 1 A. & 2 A., Both
shows linear variation in the epoxy resine medium as in fig. no. 6.
The author observe from the graph that the frequencies w1, w2 and
w3 vary with respect to propagation constant K. At high value of K, w1 &
w2 are always constant but the third coupling frequency w3 always
increases with respect to K.
161
If author take InP substance to study the surface behavior. It is
observed that as propagation constant K increases there is no variation in
w1 and w2 but w3 varies very slowly till K = 20 but sharply above K
=20[as shown in fig 5.8].
= c2 2
E
=є(ω,k)E
and D , then we have the dispersion relation for
electromagnetic εw, k =
162
+ ∞ ∶ % System has infinite response in the absence of
an applied force, thus the poles of εW, K
define the frequency of the free oscillations of
the medium.
01
23
So εW, O =1 and εO, K = 1+
*
same limit as W O. Thus great care must be taken with the dielectric
function near the origin of the WK plane.
The Total dielectric function, lattice Plus electrons, but without the
electronic polarizability of the ion cores, is
01
23
εW, K =1
+
*
(6.4)
7 7=
ε∞ 6 7 9 :ε; < 78 >
So eqn become 5 = 8
%
(6.5)
7 7
6 9 :( 6 9
78 78
163
Where ε? , ε∞ are low frequency dielectric constants and high
01
frequency dielectric constant respectively. the values of chosen
08
Form graph we observed that there exit two SPSOP models. These
modes do not propagate when (W) is negative.
0 0@
i.e <
08 08
0@ 0
i.e
08
A 0 <1
8
01
= εB ij  ε DW2 ij  WciWuj + i ijkWck] (6.6)
0
0
:0C
FπGH I
where WE =
εJ∗
IL
and Wc =
J∗
164
it's value ij = 1, i=j
= 0, i j
=o otherwise
Lower mode (Ώ1) varies very slowly and remains almost constant
when value of wave vector is equal to or greater than unit. But Ώ2
mediator mode lies in between upper and lower mode is exactly constant
for all value of wave vector (K). So there exist a band gap between the
165
two modes showing that the coupled modes with frequencies lying in this
region can not be excited.
The author observes that there are four range of EM wave for
which cylindrical surface behaves as filtering property of InP cylinder.
These ranges are :
166
Now comparing the filtering properties of different substances as
167
wsp/wt increases with increment in value of K.
For KF variation of wsp/wt with K varies with lesser effect where
as for other two substances InP Mgo show more effective linear
variation. At this radius (2 A.) both InP & Mgo having nearly same
value of wsp/wt in comparison to radius (1 A.).
Thus the surface acts as Band Pass Filter [BPF]. Agian their is no
0 0M
propagation when (W) is negative for " , the surface become
08 08
transparent again act as High pass filter (HPF). from graph we observe
band width ( ) of band pass filter is given by
0@
∆LEO + 1 % (6.7)
08
0M
Similarly ∆QEO + 1 % (6.8)
08
Gives values at which the surface will act as high pass filter. from
study of Szigeti effective charge eS∗ versus with the help of data in
table. from this the best fit of data we find the following relationship
between eS∗ and as
I3∗
=0.074+0.099< > (6.9)
I
169
from table our calculated values agree well with the experimental values
of Hass and Henvis.
The long wave lattice dynamics of the fluorite lattice was studies
by Axe. For es he obtained the same relation [(eqn) as obtained by szigeti
except that the definition of the reduced mass used by Axe . The value of
es obtained by Axe and his coworkers for several crystals with caF2
structure. The es values for tow oxides with CaF2 structure are distinctly
low and indicate ‘Increased covalency’. Szegeti’s theory is by far the
most comprehensive theory of dielectric. It is applicable to diatomic as
well as polyatomic crystals to ionic as well as covalent crystal. Though
170
the original theory assumed a harmonic cystal szigetti modified it to make
it applicable to anharmonic crystal. The most important contribution of
szigeti’s theory is the concept of the eefiective ionic charge, it has found
applications, the most important being its utility as an ionicity parameters.
Two equations for the effective ionic charge by Havinga have been
systematically applied to a number of alkali halides.
171
REFERENCES
7. Suen LE, Xie SS, Liu W, Zhou WY, Liu ZQ, Tang DS, Wang G
and Quin LX 2000 Nature 403 384
9. Tan ZK, Sun HD, Wang J. Chen J. and LiG 1998 appl Phys. Lett.
73 2287
12. Blase X Bonidict LX, Shirly EL and Louine SA 1994 Phys. Rev.
Lett. 72 1878
172
13. Dresselhous MS, Dresselhous G, and Ek Lund PC 1996 Science
of fullerenes and carbon nanotube (San Diego CA Acadmic).
15. Bendict LX, Crespi VH, Louiev SA and Cohen ML 1995 Phy
Rev B 52 14935
17. Tang ZK, Tang LY, Wang N., Zhang XX, Wen GH, Li GD
Wang JN, Chna CT and Sheng P 2001 Science 2922 462
18. Ajayan PM, Iijima S. and Ichiahshi T., 1992 Phys. Rev. B 52
14935
19. Bach Told A., Henry M., Terrier C., Strunk C., Schonen Berger
C., Saletat , J.R. Banarad J.M. and Forro, L. 1998Appl Phys.
Lett. 73 274
21. Bezrydin A., Verschveren, A.R. M., Tans, S.J. and Dekker
C.1998 Phy. Rev. Lett. 80 4036
173
23. Bonard J.M., Stura T. Salvetat J. P. Maier , F. Stockl T. Duschl ,
K. Forrol, Deheer, W. and Chatelian A. 1997 B Adv, matter 9,
827
24. Bursil, L.A. Stadlemann, P.A., Peng, J.L. and Prawers, 1992
Phys. Rev. B. 49 2883
26. Chu Y.T. Warmack, R.J. Richhie, R.H. Little, J. W. Becker, R.S.
and Eerrell, T.L. 1984 Partical Acceralation 16. 13
34. Ferrall, T.L. and Echenique, P.M. 1985 Phy. Rev. Lett. 55 1561
174
35. Erank, S. Poncharal, P. Wang, Z.L. and Deheer,W. A. 1998
Science 280 1579
37. Hubbard J., 1955 A Proc Phy. Soc. A., 68 976 1955 b ibid 68 441
41. Krats Chmer W., Lamb, L.D. Fosti RO Populos, K., and
Huffman, D.R. 1990 nature 347 354
42. Kroto H. W. Healh J.R. Obrian, S.C. Curi, R.F. and Smalley R.F.
1985 Nature 381 162
43. Kuzuo R. Teruchi M. and Tanaka M., 1992 Jap. J. Phys. P. 2,31,
L 1484
44. Lucas, A.A. Henrad, L. and Lambin P. 1994 Phy. Rev. B. 49,
2888 1995 Nuclear Instrument Meth B 96 465
45. Mintmire, J.W. Dunlap, B.I. and White, C.T. 1992 Phy. Rev.
Lett. 68 631
175
48. Saito R., Eujita, G. Dreseellhous G. and Dreseellhous M.S. 1992
Apll. Phys. Lett. 60, 2204
176
62. IiJima, Sumio, Nature (Lodon, United kingdom), 3564, (6348),
1991
63. D.A. Bochvar and E.G. Galpern, Dokl. Akad. Mauk. USSR, 209,
(610, 1973)
64. I.V. Stankevich, M.V. Nikerov and D.A. Bochvar, Russ. Chem.
Rev., 53, (640, 1984
65. H.W. Kroto, J. R. Heath, S.C.O. Brien, R.F. Curl and R.E.
Smally, Nature, 318,(162, 1985)
69. J.W. Ding, X.H. Yan and J.X. Cao, “Analytical relation of band
gaps to both chirality and diameter of single wall carbon
nanotubes”, Physical Review B, vol. 66.p073401, 2002
71. M. Grayson, et. al., Phys. Rev. Lett., 80, 1062 (1998)
177
73. T.W. Odam, J. Huang, P. Kim and C. N. Lieber, Nature 391,
Imperial college Press, 1998
82. C Kittle, Quantum Theory of Solids (John Wiley and Sons, Inc
New York, p.46 (1963)
83. J.J. Hopefield, J. Phys. Soc. Japan (Suppl. 21, 77) (1966).
86. D.L.Mills and e.Burstein, Rep. Prog. Phys 37, 890 (1974)
178
87. Kuin huang, Proc. Royal Soc., London, A 208, 352365
(1951)
91. R.J.Bell, R.W. Alexander and A.C. Ward, Surf. Sci. 48, 253
(1975)
93. R.H. Ritchie, in Polaritons, E. Burstein and P.De Inarttni, et. al.
(Pirgaman Press, Inc. New York) p217 (1972)
179
100. Weing Que, George Kirczenow and Eleuterid, Castano Phys.,
Rev., B43, 14079 (1991)
102. A.E. Quiliz, Sc. Ving and J.J. Quim, Phys, Rev. B.11,2118
(1975)
105. V.V. Brystain and Yu A Firsov Sov. Phys. Solid State 14 981
(1972).
180
111. BEZRYADIN, A., VERSCHUEREN, A.R. M., TANS, S. J., and
DEKKER, C., 1998, Phys. Rev. Lett., 80, 4036.
112. FERRELL, T.L., and ECHENIQUE, P.M., 1985, Phys. Rev. Lett.,
55, 1562.
113. BAUSELLS, J., RIVACOBA, A., and ECHENIQUE, P.M. 1987, Surf.
Sci. 190, 1015.
114. ECHENIQUE, M., HOWIE, A., and WHEATLEY, D., 1987, Phil. Mag.
B, 56, 335.
116. Chu, Y.T., Warmack, R.J., Ritchie, R.H., Little, J.W., Becker,
R.S., and FERRELL, T.L., 1984, Particle Acceleration, 16, 13.
117. HUBBARD, J., 1955a, Proc. Phys. Soc. A, 68, 976; 1955b, ibid 68,
441.
181
124. P.M.Morse and H. Feshbade (Methods of theoretical Physics) p
658
182
138. V.V. Briskin, Yu. M. Gerbstein and D.N. Mirlin, Sov. Phy. 27,
1057 (1971)
139. I.I. Reshina, Yu. M. Gerbstein and D.N. Mirlin, Sov. Phys. Solid
State 15, 2012 (1974)
143. K.S., S.N. Fatima and S.K. Shukla Indian J.Physics 70 (B) PP
144. K.L. Kleiver and R. Fuchs, Adv. Chem. Phys. 27, 355 (1974)
145. D.L. Mills and E. Burstien, Rep. Prog. Phys. 37, 890 (1974)
151. E.E. Havinga and A.J. Bosman, Phys. Rev. 140 A291 (1965)
183
PUBLICATIONS
184