Plasma Physics
Plasma Physics
1 Introduction
11
12
14
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.1.1
14
Vlasov model . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
18
2.2.1
22
2.2.2
Integration for
dependent variables . . . . . . . . . . . . . . .
23
2.2.3
25
26
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
26
29
33
3.3.1
Numerical results . . . . . . . . . . . . . . . . . . . . . . . . . .
33
35
3.4.1
35
3.4.2
37
3.4.3
38
3.4.4
Numerical analysis . . . . . . . . . . . . . . . . . . . . . . . . .
39
3.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
42
44
4.2.1
46
47
4.3.1
Numerical results . . . . . . . . . . . . . . . . . . . . . . . . . .
48
49
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
54
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
55
58
59
61
66
68
5.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
73
74
77
Bibliography
78
List of Figures
3.1 The real frequency (blue) and growth rate (red) for Bernstein waves are
plotted versus normalized k? , keeping ( !p )2 = 20, and varying u0 (the
thermal spread parameter): (a) is for u0 = 50; (b) is for u0 = 10; (c) is
for u0 = 5; at which point the instability is substantially reduced; and
(d) is for u0 = 2, at which no growth rate is visible, i.e., stable Bernstein
wave is recovered. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
34
1)2 ];
3.2 Plot of the thermal ring distribution in normalized variables, y = exp[ u20 (x
where x =
v?
.
v0
v0
40
(red) for the O mode are plotted versus normalized wave number b0 =
k?
, for
! 2pe
2
= 20, R =
T?
Tk
v0
= 0; (b)
49
4.2 The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the zeroth-harmonic purely-growing mode,
dimensional space
T?
k ; Tk
, for
! pe 2
(n=0)
max
, in two
50
4.3 The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the rst harmonic cyclotron mode,
sional space
T?
k ; Tk
, for
! pe 2
(n=1)
max
, in two dimen-
51
4.4 The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the second harmonic cyclotron mode,
mensional space
T?
k ; Tk
, for
! pe 2
(n=2)
max
, in two- di-
52
5.1 The real frequency (blue) and the growth rate (red) for the electromagnetic ordinary-mode versus
(! pe =
2
e)
= 2,
66
5.2 The same as Fig. 1, except that the drift speed is nite, as indicated by
Pe = 1. In this case, the O-mode instability begins to operate. . . . . .
67
5.3 The same as Fig. 2, except that the drift speed is increased to Pe = 1:5.
In this case, the O-mode instability operates over a wider range of
e.
68
5.4 The lled contour plot (with colourmap) of the maximum growth rate,
max =
e,
2
e)
= 20,
= 10, and
70
5.5 The same as Fig. 4, except Pe = 0:5. In this case, the maximum growth
rate contours shown deviation from the bi-Maxwellian case. Note that
the analytical marginal stability curve, shown in red, does not show good
agreement with the lowest maximum growth rate contour. . . . . . . .
71
5.6 The same is Fig. 5.5, except now the drift speed is increased to Pe = 1.
Again, the anaylatical marginal stability curve is not is good agreement
with the lowest maximum growth rate contour. . . . . . . . . . . . . .
72
By
Fazli Hadi
Department of Physics
University of Peshawar
Peshawar Pakistan
2015
Fazli Hadi
as a partial fulllment for the requirement of the
Degree
of
DOCTOR OF PHILOSOPHY
in
PHYSICS
Abstract
The investigation of perpendicularly propagating modes, excitations and associated instabilities at the order of electron cyclotron frequencies in non-relativistic plasmas is important for understanding many astrophysical and laboratory phenomenon
such as noise emission and absorption, solar wind plasma, large number of experiments
lunched with satellites to provide in situ data on the properties and nature of plasmas
in the earth and other planetary magnetospheres, the achievement of thermonuclear
fusion and so forth. In this context, the Vlasov model is employed for electron-ion
plasma in which the ion dynamics are ignored. Using kinetic theory approach for
homogenous collisionless magnetized plasma, we derive the general expression for the
conductivity tensor in cylindrical polar coordinates. Modes of non-relativistic electrons
are investigated for perpendicular propagation in non-Maxwellian plasma. For this
purpose thermal ring and drifting Maxwellian distribution functions are used to derive
dierent modes.
In this thesis we particularly focus on analytical and numerical solution of the dispersion relation for electrostatic Bernstein wave and electromagnetic O mode which
propagate perpendicular to the ambient magnetic eld. The classic Bernstein waves
may be intimately related to banded emissions detected in laboratory plasmas, terrestrial and other planetary magnetospheres. However, the customary discussion is based
upon isotropic thermal velocity distribution function. In order to understand how such
waves are excited one needs an emission mechanism, i.e., instability. In non-relativistic
collision-less plasmas, the only known Bernstein wave instability is that associated with
a cold perpendicular velocity ring distribution function. However, cold ring distribution is highly idealized. The present thesis generalizes the cold ring distribution to
Acknowledgments
It is great pleasure for me to express my thanks to all those who contributed
in many ways to the completion of my PhD studies and made it an unforgettable
experience for me. First and foremost, I am extremely indebted to my PhD Supervisor
Dr Anisa Qamar, because this work would not have been possible without her kind
guidance, support and encouragement.I successfully overcome many di culties and
learned a lot under her guidance .
I say heartful thank to Dr Ghulam Murtaza (Salam Professor) for developing skills
in theoritical plasma physics and building condence in me to start my work on kinetic
theory of plasma physics at "Salam Chair in Physics" Government College University
(GCU) Lahore.
I would like to express my heartfelt gratitude and sincere thanks towards my foreign
supervisor Dr Peter H. Yoon for providing me an opportunity to work at the University
of Maryland (UMD) College Park, United State of America (USA) for six months in
2014 under International Research Support Initiative Program (IRSIP) of Higher
Education Commission (HEC). I am extremely indebted to him for his precious time
he has spent in numerous discussions with me during my stay at UMD. He patiently
supervised and guided me in the right direction. I also deeply appreciate Sadaqat Jan
for his kind help and assistance during my stay at College Park, UMD, USA.
I take this opportunity to sincerely acknowledge HEC, for providing nancial assistance in the form of Indigenous 5000 PhD Scholarship Schemeand IRSIP.
I would also like to thank all my colleagues and friends at Salam Chair GCU
Lahore for meaningful discussions, cooperation and help whenever I needed in research
work. My gratitude also goes to Dr Fraz Bashir (COMSATS) Lahore for helping in the
Certicate
It is certied that the work contained in this dissertation entitled Study of
Perpendicularly Propagating Modes in Magnetized Nonrelativistic Plasma with
Loss-cone Distributionis based on theoritical investigation. This work has been
carried out by Mr. Fazli Hadi under the supervision of Dr. Anisa Qamar and is
accepted in its present form by the Department of Physics, University of Peshawar as
satisfying the thesis requirements for the degree of Doctor of Philosophy in Physics.
Supervisor
Submitted through
Associate Professor
Chairman
Department of Physics
Department of Physics
University of Peshawar
University of Peshawar
Approval
This thesis, prepared by Fazli Hadi in partial fulllment of the requirements for
the degree of Doctor of Philosophy in Physics, has been approved and accepted by the
following:
_____________
Dr. Anisa Qamar (Supervisor)
Associate Professor
Department of Physics
University of Peshawar
_______________
Dr. Yaseen Iqbal (Chairman)
Professor
________________
(External Examiner)
............................
Department of Physics
.............................
University of Peshawar
..............................
Chapter 1
Introduction
A plasma may be dened as a statistical system containing mobile charge particles
[1]. According to general principles of statistical mechanics, the Maxwellian distribution arises when plasma is in thermal equilibrium. However when a plasma is heated,
substantial temperature dierences develop between the plasma species i.e. electrons
and ions, because the ions and electrons are in thermal equilibrium among themselves,
but they may not be necessarily in thermal equilibrium with each other [2]. It happens
frequently in plasma since the energy exchange between electron-electron or ion-ion by
collisions are both faster than collisional electron-ion energy transfer due to the huge
mass dierences, therefore non equilibrium eects are prominent in plasmas [2, 3]. A
realistic system may be far away from thermodynamic equilibrium state, because the
plasma may be subject to a variety of dierent eects. For example, spatial variation
of physical quantities such as density, temperature, and intensity of the magnetic eld
which is present almost in all laboratory and astrophysical plasmas. More importantly,
background turbulence that may be regarded as a quasisteady state, can take away the
distribution far from Maxwellian. Plasmas subjected to the inuence of external force
elds would create ows of particles in the natural space environments and in laboratories, consequently leads the distribution functions to non-Maxwellian shape [5, 4].
Such non-Maxwellian distributions can be regarded as sources of free energy in plasma.
By free energy we mean, particles kinetic energy that can be transferred to uctuating elds (the inverse of Landau damping). Two forms of free energy associated
with particle distributions are: i) a non-Maxwellian energy distribution with an excess of high energy particles for example kappa distribution, which have super thermal
particles and ii) velocity distribution that is anisotropic, like bi-Maxwellian with temperature parallel to magnetic eld lines Tk not equal to the temperature perpendicular
to the eld lines T? (the parallel and perpendicular are taken with respect to external
magnetic eld B0 = B0 z). The occurrence of such distributions is a well established
feature of the solar wind plasma from both the observational [6, 7, 8] and theoritical
point of view [9] . The free energy sources associated with particles distributions may
also exist in connected form, for example the loss-cone distribution is a combination
of both non-Maxwellian energy distribution and an anisotropic velocity distribution.
During an instability the free energy associated with a non-Maxwellian distribution
is converted into radiations and as a result changes in the distribution of particles
occur which leads to loss of particles from the plasma system [5]. However not all
the deviations from the Maxwellian distributions leads to instabilities. In an innite,
homogeneous plasma having anisotropy in velocity distribution immersed in a uniform
magnetic eld B0 , if the distribution function f (v) decreases monotonically with velocity v, that is
@f (v)
@v
< 0 for all v; then the plasma is stable [10, 11] and correspond
@f (v)
@v?
necessary condition for instability in a plasma having perpendicular population inversion [12] and [13]. There are other conditions for instabilities such as the temperature
anisotropy, which does not involve positive gradient, so we are specically paying attention in this thesis to those free energy sources involving perpendicular population
inversion.
Naturally occurring plasma in space, such as solar wind, magnetospheric plasma
and some laboratory plasmas which have curved and converging magnetic eld lines
are anisotropic. Loss-cone and ring type distributions (limiting case of Loss-cone distribution) usually occur when plasma is conned in a magnetic trap or in a mirror
magnetic eld [10, 14]. In such situations only those particles located in a certain
domain of the phase space are conned while others in the Loss-cone are lost. The
6
distribution arises from such a special limit in the phase space will naturally have a
shape dierent from Maxwellian. In such circumstances the system keeps those particles which have more kinetic energy in the perpendicular direction than in the parallel
direction with respect to the external magnetic eld. The resulting inequity in the energy contents will then create a plasma wave instability [1]. Since for realistic Loss-cone
distribution function the mathematical formulation is very di cult and even analytically intractable. Therefore various simpler models of the Loss-cone distributions have
been adopted in literature to study electrostatic and electromagnetic waves, for example partially lled Loss-cone models [15] and [16], partial shell distribution [17, 18] and
Dorry-Guest-Harris (DGH) [19, 20] type of Loss-cone distribution.
An extreme case of Loss-cone distribution is delta-function perpendicular ring distribution. One of the aims of the present thesis is to investigate plasma instabilities
driven by the delta-function cold ring distribution accompanied by thermal spread.
For this purpose, one needs a Loss-cone model that naturally reduces to the coldring model. But none of the three models discussed above shrinks to a true deltal
function ring distribution except the DGH model f0a
v ? ; vk
l
= na f0a vk g0a
(v? )
which approaches the delta ring model for extremely large values of Loss-cone index l (where l = 1; 2; 3; :::::::;), but this is impractical. The function of interest is
l
(v? ) = (
g0a
1
2
?a l!)
(v? =
2l
?a )
(v? =
?a )
ular velocities. Pioneering work with such type of distributions were provided by Dory,
Guest and Harris. It is worthwhile to note here, that f0a vk is an arbitrary even
function of the parallel velocities, normalized to 1. The DGH type of distributions
represents a class of distribution function, which reduces to the Maxwellian for l = 0
1
and to the ring (cold delta-function) when l ! 1; i.e. g0a
(v? ) = (2 v? )
(v?
v0 ) ;
here v? and v0 are microscopic perpendicular electron velocity and the average perpendicular ringvelocity respectively. For 0 < l < 1, they are peaked at the nonzero
value v0 =
?a l 2
[14].
The pioneering work with distribution having perpendicular velocity gradient were
performed by Tataronis to explain the odd half harmonics [21]. But shortly after the
Ogo 5 magnetospheric observations Fredrick et al [22] seized upon such distribution
7
with a thermal spread in the perpendicular distribution to show the propagation of unstable odd half harmonics for realistic and large values of the ratio of plasma frequency
to gyrofrequency (! p = )2 [16] for electrons. Since the cold delta-function ring distribution function can be viewed as a highly idealized Loss-cone model which is unlikely
to be realized in natural situations, it is necessary to work with a more general model,
that is the thermal ring distribution [23] which is focused in this thesis.
Another type of non-Maxwellian distribution function is drifting bi-Maxwellian
which is found in countermoving plasma streams. Countermoving plasma streams are
omnipresent to astrophysical sources when supersonic plasma ows experience obstruction, such as the interaction of solar wind with planetary magnetosphere or comets.
The counterstreaming plasma is a special class of velocity anisotropic particle distribution function and are subjected to a variety of kinetic plasma instabilities. These
instabilities either in the form of counterstreaming beam or bi-Maxwellian with large
enough temperature anisotropies, which quickly dissipate their energy in the form of
electromagnetic uctuations are extremely signicant to the formation of shock structures [2, 24].
The existence of cyclotron harmonic waves in equilibrium plasma is the consequences of nite size of Larmour radii of given plasma specie i.e., electron or ion due
to the thermal motion [25]. The cyclotron resonance phenomenon were discovered by
articial satellite in the ionosphere and were then studied under laboratory conditions
using cyclotron waves propagating in perpendicular direction to external magnetic
eld. The experimental observations of the cyclotron harmonics eects, for instance
noise radiations absorption and propagation in both space and laboratory plasmas
have stimulated the study of cyclotron harmonic waves [22, 26, 27]. The above observations can be explained on the basis of perpendicularly propagating modes having
frequencies near to the harmonics of the electron or ions cyclotron frequencies [28].
The existence of cyclotron waves were rst conrmed by see reference [29] which focus
on the energy loss in the form of electron cyclotron radiations from a thermonuclear
reactor. Their observations indicate that, substantial radiations emission take place
at higher harmonics were rst considered only due to fundamental electron cyclotron
8
frequency. But the experimental study shows the emission of radiations (noise) from
plasma at the second and third harmonic frequency while absorption at the electron
cyclotron frequency. Similarly electromagnetic ion cyclotron (EMIC) waves often observed in various regions of magnetosphere ranging from a few to tens of RE (the radius
of earth). Considerable literature has been devoted to the properties of EMIC waves
such as, polarization, excitation mechanism and wave particle interactions by a number of authors. EMIC waves play an important role in heating and ion accelerating
processes. For example resonant absorption of EMIC waves is one of the main mechanism for accelerating and heating of high speed solar wind streams. Heating of heavy
ions in earth magnetosphere has been attributed to EMIC waves.
A great many physical phenomenon associated with cyclotron harmonic waves, that
are observed in many laboratory and natural space plasma environments are noise emission and absorption both at electron and ion cyclotron frequency harmonics [26, 27].
Bernstein waves [28] with frequencies near the harmonics of electron and/or ion cyclotron frequencies are natural candidates to explain these observations. There are
several other applications of Bernstein wave concept to multiple-harmonic cyclotron
emissions both in laboratory and space, but just to name a few representative examples spanning many decade, the resonance eects at cyclotron harmonic frequencies
observed by the early Aloutte topside sounder rocket are attributed to Bernstein waves
in the ionosphere [30]. More recently, observations provided by space probes Voyager 1 and 2 have revealed the existence of similar electrostatic emission in Jovian
magnetosphere [31], which naturally lend themselves to an interpretation in the light
of Bernstein wave paradigm. Bernstein waves are invoked to explain the naturally
occurring banded emission observed in the terrestrial magnetosphere [32] and in the
magnetosphere of other planets, e.g., Jupiter [33], Saturn [12, 34, 35], Uranus [36], and
even Neptune [37]. Other applications of Bernstein wave in laboratory plasmas may
include the following examples: Bernstein waves having no density cuto and damps on
the electron cyclotron resonance or its harmonics have been proposed as a viable source
for heating fusion plasma and for driving plasma currents via mode conversion [38].
The transportation of external momentum and energy to electrons has been demon9
strated in Stellerator [39] and in high- beta tokamak plasma with spherical tori such as
the National Spherical Torus Experiment (NSTX) [40] and the Mega Amp Spherical
Tokamak (MAST) [41].
The O mode instability in high- plasmas is of great concern in astrophysical and
laboratory applications. The experimental observation of the electric eld uctuations
made by one of the probe spacecraft IMP-6 in the outer magnetosphere reported by
Gurnett and Shaw [42]. They observed very strong electromagnetic radiation bands
that are harmonically related in the frequency spectrum with electric eld uctuation
parallel to the external magnetic eld. They suggested that such type of phenomenon
may be attributed to the O mode instability. Hamasaki [43, 44], Davidson and Wu [45]
examined a stationary plasma with temperature anisotropy and for the rst time it was
shown that in a temperature anisotropic plasma having T? =Tk < 1 and su ciently large
k
(where
growing modes exist. A few years latter the prediction of propagating instability with
the real frequency (! r being non-zero) has been made by Cheng in steady state plasma
to determine the instability boundaries. The justication of the instability boundaries
predicted by the instability criteria which require the instability to be absolute is also
made through the numerical solutions of the dispersion relation. In this case the growth
rates are found to be typically of the order of the cyclotron frequency [10].
It is well known that the solar wind plasma is characterized by the temperature
anisotropy. However, the predicted anisotropy does not follow the macroscopic theory.
In the absence of collisions or signicant heat ux, this is commonly interpreted as
the result of plasma instabilities. In the literature various models of the marginal
stability criteria have been built for high beta regime [46]. However, the existence
of similar marginal stability conditions for low-beta regime is still an open question.
The solar wind stable plasma congurations are found in a rhombic shape around
k
are only partially understood in terms of marginal or more general instability studies:
The high anisotropy, high parallel plasma beta i.e., Ap > 1;
pk
dened by the mirror instability, the low anisotropy, high parallel plasma beta Ap < 1;
10
pk
> 1; boundary is determined by the rehose instability and the high anisotropy,
pk
of parallel propagating LH-ploarized Alfven waves, the low anisotropy, low parallel
plasma beta Ap < 1;
1.1
pk
On the basis of above discussion, we have chosen to investigate the stability analysis for perpendicularly propagating modes in a uniform and static magnetic eld specially the electrostatic Bernstein wave and electromagnetic O mode are studied for
plasmas with thermal ring (Loss-cone type) distributions in perpendicular velocities.
The stability analysis of the electromagnetic O mode have been discussed in counterstreaming plasma as well, having the following motivations to proceed with:
Loss-cone plasma have been the focus of various studies in the recent years [14,
48, 49, 50, 51, 52, 53] due to their occurrence, such as in planetary magnetospheres
and solar atmosphere. The visible aurora is a particular manifestation of the loss of
energetic particles from magnetosphere. Variety of processes in plasma e.g. particles
losses, particular modes of wave propagations, radiations characteristics and harmonic
emissions related to the loss-cone have been studied through electrostatic and electromagnetic instabilities [23]. Signicantly Loss-cone plasma have also been found in
laboratory, such as devices for studying the thermonuclear fusion, electron cyclotron
resonances and so forth. The research regarding the Loss-cone instabilities have been
motivated and facilitated by various space craft observations for instance, Alouette-1
[26, 27], IMP-6 [54], GEOS [55, 56, 57]. The study of the stability analysis of O mode
in Loss-cone plasma is of interest because of its possible application to the explanation
of solar wind plasmas. As is well known from several observations [6, 7, 8, 58] the solar
wind stable plasma is in the shape of rhomb around the
and A = T? =Tk ' 1. For large values
hose instabilities [59], but for small values of parallel plasma beta suitable instabilities
have not been identied so far. As the solar wind is the only plasma where detailed
11
in situ satellite observations are possible, these measurements can also be relevant for
other plasma with comparable parameters e.g., the interstellar medium [60]. Therefore
it might be helpful to examine the instability conditions of dierent modes to get a
better understanding why specic plasma systems can grow unstable.
Cyclotron harmonic waves have been observed in many regimes of plasma environments for instance in thermonuclear fusion studies the high noise radiations near
the electron and ion cyclotron harmonic frequencies (n
ce ,
ci )
larly in many laboratory experiments the transmission resonances near electron cyclotron and its harmonics (n
ce )
[14, 26, 27, 61]. The electron cyclotron emissions are of great practical importance
in the explanation of many processes in fusion devices, in laboratory discharges, and
in the ionosphere. Plasma ringing phenomena have been excited by Alouette-1 and
other ionosphere probes at the electron plasma frequency, upper hybrid frequency, and
cyclotron frequency harmonics. It has been suggested that observation of this eect is
closely linked to the warm plasma electrostatic waves which could normally carry the
energy away and their frequencies are nearly equal to those mentioned above [62].
1.2
The present thesis consists of six chapters mainly focus on the stability analysis
of the perpendicular propagation of the order of electron cyclotron harmonics frequencies in magnetized nonrelativistic Loss-cone plasma. For this reason the electrons are
treated nonrelativistically while the ions provide a stationary background for neutralizing purpose. Chapter 1 started with the introduction to non-Maxwellian plasma
described by non-Maxwellian distributions and mainly concentrates on Loss-cone distribution function. The stability of perpendicular propagating modes i.e., Bernstein
wave and ordinary mode are also discussed. In chapter 2 the basic elements of kinetic
model for arbitrary distribution function has been overviewed. By using linearized
Vlasov model various components of the generalized dielectric tensor has been calculated which involve detailed mathematical formulation and steps. In chapter 3 the
12
electrostatic electron Bernstein wave instability is considered for thermal ring distribution, also dierent Loss-cone models are employed to check that either the Bernstein
wave is stable or not against each model. It is found and also demonstrated that the
Bernstein wave is excited by a special type of Loss-cone distribution i.e., cold ring
distribution and not excited by every Loss-cone distribution function which has free
energy. In chapter 4 we extend the stability analysis to electromagnetic O mode by
considering the purely growing mode along with higher harmonics for thermal ring
distribution function. The marginal stability curves are plotted for fundamental frequency and also for higher harmonics which shows the range of instabilities for O mode
in (
k ; T? =Tk )
space for each harmonic. The purely growing ordinary mode instability
is studied by using drifting bi-Maxwellian distribution function in chapter 5. It is justied that the O mode instability is aected by inclusion of higher harmonics in the
calculations. The marginal stability curve and the analytical growth rate curves are
compared for increasing values of streaming velocity. Finally in chapter 6 we presents
the summary of the key results of the thesis along with the potential future directions.
13
Chapter 2
Basic Concepts of Kinetic Theory
2.1
Introduction
14
most powerful computers and solving the classic electromagnetic many body problem
for a plasma is a hopeless task.
In uid description of plasma some information about the plasma are suppressed.
For instance the information about the distribution of velocities of particles with in the
uid element, because the uid variables are functions of position and time but not of
velocities. The physical properties of plasma, that depends on the microscopic detail
can be discovered in a six-dimensional phase space. Thus instead of starting with the
density of the particles n (x; t) at position x and time t we begin with the so called
distribution function f (x; v; t) which is the density of particles in phase space [64, 65].
Therefore it is customary to use the distribution function to describe a plasma. The
distribution function is dened such that dn = f (x; v; t)d3 xd3 v represent the number
of particles at a given time t having positions in the range between x and x + dx and
velocities in the range between v and v + dv. As time progresses the particle motion
and acceleration causes the number of particles in these x and v ranges to change and
so the distribution will also change. This temporal evolution of the distribution gives
more detailed information about the system than a uid description. Similarly the
evolution of f (x; v; t) to characterize a class of particles having the same (x; v), is far
advantageous as compared to keep track of the trajectories of individual particles [66].
All informations about a system under study can be extracted from the distribution
function. If the distribution function f (x; v; t) is known then all macroscopic i.e., uid
quantities can be computed. The particles density in conguration space na (x; t) of
particles of species a is the integral of the particle distribution function over velocity
R
space na (x; t) = fa (x; v; t)d3 xd3 v. Other macroscopic quantities for instance the uid
velocity and pressure can be calculated by using dierent integrals of the distribution
function.
2.1.1
Vlasov model
We consider a uniform plasma in which electrons are mobile while ions provides
a xed neutralizing background with an externally applied magnetic eld B0 under
equilibrium condition. A small perturbation from equilibrium position of electrons
15
@fa
@t
(2.1)
coll
The left hand side of the Eq. (2.1) is the total time derivative in six-dimensional phase
space of an arbitrary distribution function. The term
Fokker-Planck collisional term. Here rv =
@fa
@t coll
@
x+ @v@x y+ @v@x z
@vx
velocity space. The collision eects strongly depends on the temperature of the plasma
and can be neglected in su ciently hot plasma, because thermal velocities of the plasma
particles are too large. Therefore in a collisionless plasma such as space plasma where
thermal energies of plasma particle are very large one cane ignore collision eects. A
rough justication for this approximation is that we are interested in frequencies much
higher than collision frequency !
coll .
16
(2.2)
where aa is the acceleration of the ath specie of charge q having rest mass m; while the
internal electric and magnetic eld E (r; t) and B (r; t) arises due to the presence and
motion of all charged species inside the plasma [67]
aa =
qa
dv
=
dt
ma
E (r; t) +
B (r; t)
c
(2.3)
E+
B
c
rv fa = 0.
(2.4)
B=
4
1 @E
J+
;
c
c @t
E=
1 @B
,
c @t
r B =0,
17
(2.5)
(2.6)
(2.7)
r E=
"0
(2.8)
Induced charge and current density ( , J) emerges in plasma due to perturb electric
and magnetic elds which are macroscopic quantities and can be expressed as moments
of the particle distribution function
na (r; t) qa
J=
na (r; t)qa
fa (r; v; t)d3 v,
(2.9)
(2.10)
In the above equations d3 v = dvx dvy dvz is the volume of the phase space. The summation is taken over all particles species i.e. electrons, protons and ions, fa (r; v; t) is
the phase space density and qa is the charge while na (r; t) is the density of the particles. Equation (2.2) to (2.10) is a complete set of self consistent equations to be solved
simultaneously. For example, in an iterative procedure assuming starting approximate
values for E (r; t) and B (r; t), Eq. (2.2) can be solved to yield fa (r; v; t) for the various
species. Using the calculated fa in Eq. (2.9) and (2.10) leads to values for the charge
and current densities ( , J) in the plasma, which can further be substituted in Maxwell
equations and solved for E (r; t) and B (r; t). These values are then plugged back into
the Vlasov equation, and so on, in order to obtain a self-consistent solution for the
single particle distribution function [67].
2.2
18
f1 (r; v; t). The electric and magnetic elds can also be written as E (r; t) = E1 (r; t)
and B (r) = B0 (r) +B1 (r; t) in Eq. (2.2), in which the equilibrium electric eld is
taken as zero. We shall drop the subscripts denoting the species in order to show that
we are dealing with the electrons dynamics only
@f1
q
@f1
+ v:
+ (v
@t
@x
m
@f1
q
+ (E1 + v
@v
m
B0 ) :
B1 ) :
@f0
= 0:
@v
(2.11)
Equation (2.11) is the linearized Vlasove equation for non-relativistic collisionless electron plasma placed in a uniform external magnetic eld B0 = B0z . Non-relativistic velocity in cylindrical coordinates is v = (v? cos ', v? sin ', vk ) while the non-relativistic
electron cyclotron frequency is given by
ce
eB0
.
me c
charge particle motion consist of a constant parallel translation with speed vk . To understand the dynamics of the plasma completely we need linearized Maxwells equation
which are given below
r
B1 =
r
1 @E1 4
+
J1 ;
c @t
c
(2.12)
@B1
,
@t
E1 =
(2.13)
(2.14)
r B1 =0,
r E1 =4 ne
(2.15)
f1 dv.
We start from Eq. (2.11) to nd the solution of small perturbations f1 (r; v; t) from
some plasma equilibrium state f0 (r; v). Taking linearized Eqs. (2.11)-(2.12) and Eq.
(2.13) and using Fourier Laplace transform [65] i.e.,
(E; B; f ) =
dt e
st
1
1
dx
e
(2 )3
ik x
(E1 ; B1 ; f1 ) ,
@ (E1 ; B1 ; f1 )
@t
= s$ (E1 ; B1 ; f1 ) ,
@ (E1 ; B1 ; f1 )
@t
= ikF (E1 ; B1 ; f1 ) ,
and
19
here s =
i!. The Vlasove Eq. (2.11) and the wave equation from Maxwells curl Eqs.
(s + ik v) f1 + e E1
(k
@f0
@p
E1 )
e
(v
c
B0 )
@f1
,
@p
(2.16)
and
s2 + c 2 k 2
c 2 ki kj + 4 s
ij
Eij = 0,
(2.17)
cki ckj
Eij = 0,
s s
(2.18)
ij
or
ij
4
s
c2 k 2
s2
ij
or
ij
N2
Ni Nj
N2
ij
(2.19)
Ej = 0.
The refractive index is dened as Ni;j = cki;j =!, represents the ith and j th component
of the refractive index. The dielectric permittivity tensor
ij
ij
ij
ij
can be dened as
(2.20)
in the direction of static applied magnetic eld B0 . The last term on right side in Eq.
(2.11) can be expressed as
v0
B0
c
@f1
=
@p
@f1
,
@'
(2.21)
here ' is the azimuthal angle. After rearranging Eq. (2.11) we obtain
@f1
@'
@
@
+v
@t
@x
f1 =
E1 +
B1
c
@f0
.
@v
(2.22)
Substituting for right hand side in above equation from Eq. (2.21) which takes the
form
( 0) =
E1 +
20
B1
c
@f0
.
@v
(2.23)
(s + ik v)
( 0)
f1 =
(2.24)
which is an inhomogeneous dierential equation [68] of order rst. To solve this dierential equation we use the integrating factor technique. For this purpose it is convenient
to nd the solution of homogeneous part rst which is
1
'
'0
or
1
s + ikk vk ('
sin '0 )
(2.25)
Now using G ('; '0 ) as integrating factor [69] the general solution of Eq. (2.24) can be
written as
f1 =
G ( 0)
( 0) d 0.
(2.26)
The integral in Eq. (2.26) is an indenite integral with only the upper limit taken
into account. The lower limit should not be specied, and its contribution should be
ignored. Using the above equation the current density from Eq. (2.10) takes the form
J = n0 e
'
(2.27)
('0 ) and G ('0 ) are inserted in Eq. (2.27) from Eq. (2.23) and Eq.
J = n0 e
'
0
d' exp
"
'0
(s + ik v) d'
00
e E1 +
B1
c
@f0 3
d p.
@p
(2.28)
The current density J can be written in terms of conductivity tensor by using Ohms
law as
Ji =
ij
21
Ej ,
(2.29)
here
ij
is the conductivity tensor. Combining Eq. (2.28) and Eq. (2.29) which after
simplication becomes as
ij
= n0 e2
ij
= n0 e
'
d'0 exp
1
"
'0
(s + ik v) d'00
'
@f0
1
@f0
+ (v E) k
@p
!
@p
exp
1
1
dpk
dp? p?
d'
@f0
@p
(k v) d3 p ,
(2.30)
'
d'0
s + ikk vk ('
(cos '0
1
!
1 , sin '
1, 2
sin '0 )
+ cos '0
3)
(2.31)
@f0
i
@f0
+ kk v k
@p? s
@p?
v?
@f0
@pk
@f0
,
@pk
and
i
@f0
k? v k
s
@p?
v?
@f0
@pk
with these denitions we now perform the ' and -integration in Eq. (2.29) in the
following sections.
2.2.1
22
='
'0 as under
sin ('
00
)
B v? cos ' cos('
B
B v sin ' cos('
)
B ?
@
vk cos('
) 1
0
=2
1
X
B
B
B
B
@
in
))]
vk sin('
n2
v J 2 (b) 1
b2 ? n
v? cos ' (
v? nb Jn2 (b)
v? Jn02 (b) 1
n
v J 2 (b) 1
b k n
3)
n
b
C
C
v? sin ' ( 2 + cos('
) 3) C
C
A
vk ( 2 + cos('
) 3)
in
v J (b)Jn0 (b) 1
b ? n
in
v J (b)Jn0 (b) 1
b ? n
+ cos('
n
b
n
b
C
C
C,
(2.32)
3 C
A
in above Eq. (2.32) we have used generating function for the Bessel function of integer
order
exp [ib sin '] =
1
X
n= 1
and
exp [ ib sin ('
)] =
1
X
)] Jm (b) ,
n= 1
2.2.2
Integration for
dependent variables
d exp
s + ikk vk
in
23
s + ikk vk
in
using the results of and -integration in Eq. (2.31), we may write the dyadic equation
in (2.19) as
ij (k; !)
Z 1
Z
2 X
2
me ! pe
dp? p?
s a
0
i;j
n2
v J 2 (b) 1
b2 ? n
B
B
B
B
@
1
1
dpk
n=1
X
n=
1
s + ikk vk + in
1
v? nz Jn2 (b)
in
v J (b)Jn0 (b) 1
b ? n
in
v J (b)Jn0 (b) 1
b ? n
ij
v? Jn02 (b) 1
n
v J 2 (b) 1
b k n
vk Jn2 (b)
n
b
3
n
b
n
b
C
C
C.
(2.33)
3 C
A
v? 2 + nb 3
1 vk
=
(s + ikk vk + in )
(s + ikk vk + in )
1
s
vk
@f0
@p?
v?
@f0
@pk
+1
X
n [Jn (b)]2 = 0,
n= 1
+1
X
Jn (b)Jn0 (b) = 0,
n= 1
[Jn (b)]2 = 1.
n= 1
When the above steps are executed the Eq. (2.33) can be written as
"ij =
ij
2
me ! 2pe
s
with
2
Lij = 2 m! 2pe
s
p? dp? dpk
1
X
1
(s + ikk vk + in )
1
X
vk
p? dp? dpk
v
1 ?
vk
@f0
@p?
v?
Mij + Lij ,
(2.34)
@f0
@pk
(2.35)
(2.36)
and
0
B
B
Mij = B
B
@
n2 2
J (b)v?
b2 n
in
J (b)Jn0 (b)v?
b n
n 2
J (b)vk
b n
in
J (b)Jn0 (b)v?
b n
Jn02 (b)v?
iJn (b)Jn0 (b)vk
n 2
J (b)vk
b n
C
C
iJn (b)Jn0 (b)vk C
C.
A
v
Jn2 (b) v?k
The Eq. (2.34) represents the expression for generalized dielectric tensor in a homogenous plasma which can be used for an arbitrary equilibrium distribution function f0 .
Now the components of dielectric tensor with the denition of $ =
24
! kk v z @
v?
@v?
+ kk @v@ z
and S =
2.2.3
P R
n
dv !
1
n
kk vz
The generalized dielectric tensor components takes the form given below which
demonstrate a remarkable property called Onsagar symmetry, that is the hot plasma
dielectric tensor always satises the symmetry relations
zy
xy
yx ,
xz
zx ,
yz
function f0 .
xx (k; !)
=1+
yy (k; !) = 1 +
xy (k; !) =
2 2
! 2pe
2 n Jn
S
v
?
!2
b2
$f0 ,
(2.37)
! 2pe h 2 0 2 i
S v? (Jn ) $f0 ,
!2
(2.38)
yx (k; !) = i
0
! 2pe
2 nJn Jn
S
v
?
!2
b
! 2pe
nJn2
S v? vz
xz (k; !) = zx (k; !) =
!2
b
yz (k; !)
zz (k; !)
=1+
zy (k; !)
! 2pe
dvvz 2
!
@
@vz
$f0 ,
$f0 ,
! 2pe
S (v? vz Jn Jn0 ) $f0 ,
!2
vz @
v? @v?
! 2pe
f0 + 2 S (vz Jn )2 $f0 .
!
(2.39)
(2.40)
(2.41)
(2.42)
25
Chapter 3
Stability Analysis of Bernstein
Wave
3.1
Introduction
The classic Bernstein wave [28] pertain to stable oscillation in Maxwellian plasmas. However, in order to discuss naturally occurring banded emissions in planetary
magnetospheres, for instance, one needs an emission mechanism, i.e., the instability of
Bernstein waves driven by a free energy source. In the literature, the only discussion
of Bernstein wave instability is a series of papers based on the cold, delta-function
perpendicular ring distribution function [21, 26, 27, 71, 72, 73]
f v ? ; vk =
(v? v0 )
G vk ;
2 v?
(3.1)
26
f=
1
2A
A = exp
exp
(v?
v02
v0 )2
2
v0
G vk ;
1 + erf
v0
(3.2)
The above model represents the thermal ring distribution with normalization constant A and thermal speed of the electrons in perpendicular direction to external magnetic eld. This model reduces to Eq. (3.1) in the limit of v0 = ! 1, and to thermal
Maxwellian model for v0 = ! 0, so that one may not only discuss the stabilization of
the ring instability of the Bernstein wave by thermal spread, but also demonstrate the
transition from unstable situation to stable oscillation.
In spite of the caveat previous eorts to discuss the excitation of Bernstein waves relied on the assumption of an isotropic background population plus a tenuous Loss-cone
electron population, in which various models of the Loss-cone distribution are adopted.
For instance, [22] adopted a Gaussian ring distribution similar to Eq. (3.2) in order
to discuss the excitation of multiple-harmonic electron cyclotron waves. Similarly,
[19, 20] adopted the Dorry-Guest-Harris (DGH) type of Loss-cone distribution, Fe
2l
2
v?
exp ( lv?
=a2 ) G vk for similar analyses, [15] and [16] respectively employed par-
distribution, Fe
+ (1
2
2
) (v?
=a2 )] exp ( v?
=a2 ) G vk ; Fe
2
exp ( v?
=a2 )] G vk , and [17] and [18] made use of the partial shell
exp
(v
v0 )2 =a2
1 + tanh vk =v
2
; where v 2 = v?
+ vk2
. Because of the freedom associated with the real frequency being determined by the
27
core Maxwellian electrons, various models of the energetic Loss-cone electrons can be
utilized in the growth rate formula in order to discuss the multiple-harmonic Bernstein
mode excitation. However, as noted, when the entire electron population possesses
a non-Maxwellian feature, the real frequency and the growth rate must be computed
concomitantly.
When a population of energetic electrons is directed into such a region having
increasing magnetic eld, it is well known that the Loss-cone distribution with perpendicular population inversion, @F=@v? > 0 develops. The cold delta-function ring
distribution function Eq. (3.1) can be viewed as a highly idealized Loss-cone model.
[21, 26, 27, 71, 72, 73], were the rst to systematically investigate the Bernstein wave
instabilities associated with such a distribution function. However, as noted, the cold
ring model is unrealistic. We thus consider a more general model Eq. (3.2) in our
analysis.
This chapter has two basic ideas. First is the investigation of the Bernstein wave
excitation by taking thermal ring distribution function in non-relativistic plasma while
the second is based on the study of some sample Loss-cone distribution functions in
context of Bernstein wave which have free energy but unable to excite Bernstein wave.
The excitation of Bernstein waves requires a su ciently high perpendicular velocity
gradient associated with the electron distribution function. In contrast to situations
where the core electrons provide the real frequency dispersion relation, while the energetic components are used for the growth rate computation, in such an approach, the
total electron velocity distribution function is modeled by f = f0 + (ne =n0 ) fe , were f0
is the Maxwellian distribution, ne =n0 is the ratio of energetic-to-background electron
number density, and fe is the model velocity distribution for the energetic electrons.
In the Earths magnetosphere and other planetary situation, where multiple-harmonic
electrostatic electron cyclotron waves are observed, it is customary to assume that the
magnetic eld lines in the magnetosphere contain isotropic population of core electrons
and a tenuous component of the loss-cone electrons, which facilitates the growth rate
analysis. However, such an assumption is not necessarily always valid. In the case
of Earths auroral eld lines, which is amenable to direct satellite measurements, it
28
is well known that during the so-called inverted-V events, the core isotropic electrons
are removed along from the eld lines by large-scale parallel electrostatic potential
drop, leaving only the Loss-cone type of energetic electrons to persist. Such a situation is precisely the physical model we have in mind, where the bulk electrons possess
non-Maxwellian velocity distribution function.
3.2
Mathematical formalism
B0 = B0 ez is chosen along the z-axis, while the direction of wave propagation vector k
is conned to the xz-plane making an angle
k cos z and N
ck
!
index. The signs ? and k refers to perpendicular and parallel components with respect
to B0 . The general plasma wave dispersion relation is given by [24, 47, 74]
Nk2
xx
xy
yx
zx
xz
+ N? Nk
N2
yy
+ N? Nk
zy
zz
(3.3)
= 0,
yz
N?2
zy
xz
zx
= 0. When the above quantities are plugged into Eq. (3.3) it implies
xx
yx
xy
N?2
yy
zz
29
= 0,
0
N?2
(3.4)
xx
2
xy
N?2 +
yy
zz
N?2 = 0,
(3.5)
for non-diagonal elements to be very very small, we can get two decoupled modes i.e.
the pure transverse X mode
N?2 =
yy ,
(3.6)
(3.7)
= 0.
Using the dielectric tensor component from Eq. (4.16) (c.f. chapter 2) we get the
general dispersion relation for electron Bernstein wave,
0=1+
X ! 2pe
a
where $ =
! @
v? @v?
and S =
P R
n
!2
dv !
2
S v?
1
n
n2 Jn2 (b)
b2
$fe ,
(3.8)
0=1+
1
X
n=
where ! 2pe =
4 n0 e2
m
1
2
! 2pe
! (! n )
1
dv
v? n2 Jn2
b2
@f (v? )
,
@v?
eB0
mc
(3.9)
is the (non-relativistic)
electron gyro-frequency and Jn (b) is the Bessel function of the rst kind of order n
with argument b =
k? v ?
30
eld B0 while e, m and c stand for the unit electric charge, rest mass of the electron,
and the speed of light in vacuum respectively. When summation over positive values
of n is taken only Eq. (3.9) can also be written as
0=1+
1
X
n=1
2n2 ! 2pe
! 2 n2
(3.10)
The above Eq. (3.10) therefore represents the generalized dispersion relation of electron
Bernstein wave for any arbitrary distribution function f (v? ). In our analysis of the
Bernstein wave, we use thermal ring distribution function which is given by
f=
1
2A
(v?
exp
v0 )
2
G vk .
(3.11)
The above model represents the thermal ring distribution with normalization constant
A and thermal speed
netic eld. Microscopic perpendicular electrons velocity and the average perpendicular
ringvelocity are given by v? and v0 respectively. We assume that the distribution is
R1
separable into v? and vk , f = f (v? ) G vk where 1 dvk G vk = 1. Under such an
assumption, the parallel velocity integration can be performed trivially.
1=
2
2A
dv? v? exp
v? v 0
v0 )2
(v?
x+
v0
v0
x2
t2
(3.12)
dt,
and
2
erfc(z) = p
t2
dt = 1
erf (z) .
With the above denitions of the error functions leads to the following form
1=
1
exp
A
v02
2
v0 p
(1 + erf (z)) ,
A = exp
v02
2
v0
1 + erf
v0
(3.13)
The thermal ring distribution model in Eq. (3.11) reduces to cold ring distribution
Eq. (3.14) in the limit of
v0
a
v0
a
! 0, so
that one may not only discuss the stabilization of the ring instability of the Bernstein
waves by thermal spread, but also demonstrate the transition from unstable situation
to stable oscillation.
It is worthwhile to note that the classic electrostatic Bernstein waves [28] pertain
to stable oscillation in Maxwellian plasmas. However, in order to discuss naturallyoccurring banded emissions in planetary magnetospheres, for instance, one needs an
emission mechanism, i.e., the instability of Bernstein waves driven by a free energy
source. In the literature, the only discussion of Bernstein wave instability is a series
of papers based on the cold, delta-function perpendicular ring distribution function
[21, 26, 27, 71, 72, 73]
f v ? ; vk =
(v? v0 )
G vk ,
2 v?
(3.14)
32
3.3
Specically we shall carry out the Bernstein wave linear stability analysis on the
basis of thermal ring distribution given in Eq. (3.11). The general dispersion relation
for Bernstein wave in Eq. (3.10) for thermal ring distribution takes the following form
2
0=1
2A
When we put
0 = 1
1
X
n=1
dv =
2 v0
A
Jn+1
2n2 ! 2p
! 2 n2
R1
0
v? Jn2 (b)
dv
(v?
b2
v0 ) exp
1
X
!2
! 2p
n2
x + Jn
k? v 0
v?
v0
v?
v0
v?
v0
(3.15)
(3.16)
v0
1 exp
v0
,
a
1
! 2p
2u40 X
A n=1 ! 2 n2
0=1
v0 )2
(v?
n=1
k? v 0
dxx2 (x
1)
v?
v0
1 (3.17)
.
[Jn+1 (b0 x) + Jn
k? v 0
, and x =
(b0 x)]2 e
v?
v0
u20 (x 1)2
(3.18)
The above dispersion relation is solved numerically which demonstrate the existence of
multiple-harmonic electron cyclotron instability, particularly for small values of thermal
spread u0 .
3.3.1
Numerical results
In Figure 3.1 we have depicted the stabilization of the Bernstein wave. We have
carried out the complex root solving procedure. Our rst aim is to show the stabilization of the Bernstein-ring instability [21, 26, 27, 71, 72, 73], by thermal spread in
the Gaussian ring distribution. For this purpose, we propose the natural dimensionless
wave number b0 =
k? v 0
Figure 3.1: The real frequency (blue) and growth rate (red) for Bernstein waves are
plotted versus normalized k? , keeping ( !p )2 = 20, and varying u0 (the thermal spread
parameter): (a) is for u0 = 50; (b) is for u0 = 10; (c) is for u0 = 5; at which point the
instability is substantially reduced; and (d) is for u0 = 2, at which no growth rate is
visible, i.e., stable Bernstein wave is recovered.
in Eq. (3.18) for u0 = 2; 5; 10; 50. Here the increasing values of u0 corresponds to decreasing values of thermal spread : The input parameters are also shown in the gure
captions and titles. The blue lines corresponds to the real frequency Re(z) =Re
while the red curve represents the growth rate Im(z) = Im
. In the case of u0 = 50
Fig. 3.1(a) shows the solution that is virtually identical to the classic ring instability
[21, 26, 27, 71, 72, 73]. As the thermal spread is increased to u0 = 10, Fig. 3.1(b)
begins to show the suppression of the high-harmonic mode growth rates. As u0 is reduced to 5, and to 2, it can clearly be seen from the numerical plots that the Bernstein
wave approaches to stable solution. Figure 3.1 also depict the transition from unstable
solution to classic Bernstein mode solution as the ring speed is reduced.
34
3.4
@f
@v?
regions of v? and only such type of distributions are unstable to ute mode kk = 0
propagations [21, 72, 73], but this is not always true for a non-Mawellian distributions
that have free energy and cause instability. In the following there are some sample
model loss-cone distribution functions which have gradient in their perpendicular velocities but are unable to produce Bernstein wave excitation. It is interesting to note
that the model distributions adopted by various authors [15, 16, 19, 20] in order to
investigate the Bernstein mode excitation by a tenuous energetic electrons loss-cone
which can be actually used within the context of Eq. (3.10). Specically, the analytic
loss-cone models available in the literature correspond to the DGH model and two
loss-cone models mentioned earlier
f (1) =
f (2) =
f (3) =
2l
(l + 1)l+1 v?
e
2 l!
2l
G vk
2 (1 +
G vk
2
A
A
(l+1)
1+
2
v?
2
2
v?
2
G vk ,
2
v?
2
Be
2
v?
2
2
Av?
2
(3.19)
(3.20)
(3.21)
> 0. For the
third model, A > B and 0 < B < 1. Now we analyze the Bernstein wave dispersion
relation for above distributions Eqs. (3.19)-(3.21) which are respectively calculated in
the following section.
3.4.1
In this part we study the dispersion relation given in Eq. (3.10) for model distribution given by Dory et al. [75] Loss-cone type distribution dened in (3.19), where
l = 0; 1; 2; 3; ::::::::::, G vk is an even function of the parallel velocities, normalized to
35
1=2
dv? v?
=2
2l
v?
2l
(l+1)
dxx2l+1 e
2
v?
2
(l+1)x2
=2
N (l + 1)
l+1
dxx2l+1 e
x2
N (l + 1)l+1 l! .
Inserting Eq. (3.19) into Eq. (3.10), we can manipulate the dispersion relation for
Bernstein wave as
0 = 1+
2
1
! 2pe l + 1 X
2n2
2
! l! n= 1 k?
4 @ l 1
@ 2
6
6
6
6
4
2l
4 @ l
@ 2
l
2l+2
or
1
! 2pe 1 X
1 n2
0=1+
! l! n= 1 l ! n
"
R1
0
udu
Jn2 e
u2
2
0=1+
! 2pe
! 2 n2
n=1
"
2
l
l 1
@
@ l
l
l
2
l
u2
2
udu Jn2 e
4 @
@ 2
2l
1
X
1
!
R1
udu Jn2 e
@
@ l
pk? u
l+1
u2
2
7
7
7,
7
5
pk? u
l+1
l 1
4 @
@ 2
2l+2
k? u
l+1
2n2
2
l! l+1
l
#
(3.22)
#
(3.23)
In ( l ) e
(3.24)
this is the dispersion equation for the Dory Guest-Harris Loss-cone type distribuR1
tion function. The above modied form is obtained by solving the integral 2 0 udu
Jn2 e
u2
2
pk? u
l+1
argument
2
= k?
In ( l ) e
=2 (l + 1)
36
3.4.2
measures the
positive slope and it will diminish the free energy that cause unstable growth and
therefore the growth rate. The normalization for distribution in Eq. (3.20) can be
performed according to
1=2 N
=2
2
v?
dv? v? 1 +
1
x2 e
dxx 1 +
2
v?
2
x2
0
2
N (1 +
).
(3.25)
Inserting Eq. (3.20) in Eq. (3.10) we can nd the Bernstein wave dispersion relation
for the given distribution
0=1
! 2pe
!
0=1
0=1
1
1+4
1
X
2n2
k2
n= 1 ?
! 2pe
!
1
1+4
! 2pe
!
1
1+4
0=1
1
X
n=1
2
2
1
X
n2
n= 1
1
!
n2
! n
1
! 2pe
! 2 n2
dxxJn2
k?
1
X
n=
(1
(1
2n2 e
2 1+4
4) In ( ) e
4 )
In ( )
n2 In ( ) e
4 [In ( )
+4
4 + 4x2 e
In ( ) e
+ 4n2 In0 ( ) e
In0 ( )] .
x2
(3.26)
The above equation gives the dispersion relation for partially lled loss-cone. Where we
R1
2
have made use of the following denitions of the modied Bessels functions, 2 0 dxxe x Jn2 (bx) =
R1
2
0
2 2
In ( ) e and 2 0 dxx3 e x Jn2 (bx) = In ( ) e
; also = k?
=2 2 is the argument of the modied Bessel function.
37
3.4.3
Here we calculate again the normalization constant for the model distribution
mentioned in Eq. (3.21)
Z
1=2 N
=2
2
v?
2
dv? v? e
x2
dxxe
Be
Z
NB
2
v?
2
,
Ax2
dxxe
N
N
N
B
B
,
A
.
Inserting Eq. (3.21) in Eq. (3.10) we can nd the Bernstein wave dispersion relation
for the given distribution
0=1
1
X
! 2pe A
2n2
2
! A B n= 1 k?
k?
dxxJn2
B2
0=1
n=1
where
1
!
n
k
p? x
A
dxxJn2
0
1
X
! 2pe
! 2 n2
BIn ( 0 ) e
2n2
AIn ( ) e
2A
B
,
0
(3.27)
=A. With these models, the velocity integrals in the Bernstein mode
0=1+
1
X
n=1
a(1)
n
"
2n2
=
l
l! ll+1
a(2)
n =
a(3)
n
2n2 e
1+
2n2
=
A B
(k)
an ! 2p
! 2 n2
2 @
l
@ l
In ( )
AIn ( ) e
l 1
1
l
[In ( )
(3.28)
#
(3.29)
In0 ( )] ,
(3.30)
BIn ( 0 ) e
0
38
2 @
l
@ l
(3.31)
where
l
2 2
k?
2 (l + 1)
,
2
2
k?
2
2
2
2
k?
2A
2
2
In the above equations In ( ) is the modied Bessel function of the rst kind of
order n. We have attempted to obtain unstable Bernstein mode solutions on the basis
of the above dispersion relations, but the numerical solutions all corresponded to stable
oscillations possessing no nite growth rate. As we shall discuss subsequently in more
detail, the reason seems to be that even though the models in Eqs. (3.19)(3.21) do
represent electrons possessing free energy source, the perpendicular velocity gradient
i.e.,
@f
,
@v?
3.4.4
Numerical analysis
In order to prove the above speculation, it is useful to plot in Fig. 3.2 the thermal
ring distribution Eq. (3.11) in normalized form, namely, y = exp
normalized perpendicular velocity x =
v?
,
v0
u20 (x
1)2 , versus
v0
,
a
plots y =
y=
A
A B
(1+
e
x2 )e
1+
x2
x2
Be
Ax2
3.2(a) shows that as the ratio u0 increases, the model approaches the delta-function
ring distribution. In order to reduce the model to the Maxwellian distribution, all one
needs to do is to take u0 ! 0 while holding
none of the three models shown in Eqs. (3.19)(3.21) reduces to the true delta-function
ring distribution. Only the DGH model approach to delta ring model for extremely
large values of "l", but this is impractical.
One can make use of the ndings from Fig. 3.1 to achieve our second aim, i.e.,
to explain why none of the model distributions in Eqs. (3.19)(3.21) yield unstable
solutions. According to Fig. 3.1, when we increase the thermal ring spread (or decrease
the parameter u0 to 5), most of the Bernstein mode cyclotron instabilities disappeared,
and when we further increase the thermal spread (or equivalently, reduce u0 to 2),
39
the situation corresponded to stable Bernstein waves. Upon visual inspections, the
three models shown in Fig. 3.2 at best correspond to the thermal ring distribution
with u0 on the order of 2 or so. It is little wonder then, that none of the model
distributions in Eqs. (3.19)(3.21) yielded unstable solutions. Based upon such a
nding, we may thus conclude that unless the perpendicular velocity gradient
@f
@v?
is
su ciently high, Bernstein mode instability cannot be excited despite the fact that the
underlying perpendicular velocity distribution of the electrons may possess free energy.
40
3.5
Summary
In this chapter we have carried out the nonrelativistic Bernstein wave stabilization
by thermal spread associated with ring distribution function using the complex root
solving procedure. Our rst aim is to show the stabilization of the Bernstein-ring instability by thermal spread in the Gaussian ring distribution. It is shown that increasing
thermal spread leads to stabilization of the Bernstein wave. Similar procedure is used
for some sample loss-cone model distribution function, but the result is stable solution.
It is concluded that the loss-cone distribution having free energy for the excitation of
the wave cant always causes excitation of the Bernstein wave, because their must be
euough free energy required for Bernstein wave excitation. This fact is demonstrated
by plotting the sample distribution function for various increasing values of thermal
spread, which leads to Maxwellian distribution function.
41
Chapter 4
Ordinary mode instability
associated with thermal ring
distribution
4.1
Introduction
Davidson and Wu [45] rst discussed the purely growing electromagnetic ordinary O mode instability for high beta plasmas. Their model distribution was the biMaxwellian in its mathematical form. They showed that the O mode propagating in
perpendicular direction with respect to the ambient magnetic eld can become unstable for su ciently high beta and for excessive parallel electron temperature anisotropy,
Tk > T? , where Tk and T? represent the parallel and perpendicular Maxwellian temperatures, respectively. The electron beta is dened by
= 8 n0 kB Tk =B02 , kB denoting
the Boltzmann constant, n0 being the ambient plasma density, B0 being the ambient
magnetic eld intensity.
Recently, however, this instability has received renewed attention because of its
possible application to the solar wind [46, 47, 76, 77]. It is well known that the solar
wind plasma is characterized by the temperature anisotropy [6, 7, 8, 58, 59, 78, 79, 80],
but it also well known that the observed temperature anisotropy does not follow the
macroscopic theory. Instead, the anisotropy upper bound closely follows various mar-
42
ginal instability criterion for various temperature anisotropy-driven plasma instabilities. In literature various empirical models of the marginal stability curve have been
constructed for high-beta regime [59], but whether similar marginal stability conditions
exist for low-beta regime is still an open question.
However, it was suggested that the O mode instability for counter-streaming plasma
distributions may provide the low-beta marginal stability criterion [46, 47, 76, 77].
The rationale for adopting the counter-streaming model is based on the fact that both
the solar wind electrons and protons are sometimes characterized by such features.
For the electrons, it is well known that sometimes, there exists a highly eld-aligned
streaming component known as the strahl, which is present in the background of the
core electrons. Similarly, for the ions, in some cases, there exists a small eld-aligned
component streaming with respect to the background component. It is in this context
that the purely-growing O mode instability has recently received renewed attention.
In the present chapter we revisit the high-beta O mode instability, but unlike the
original work by Davidson and Wu [45], we consider a situation where the electron
velocity distribution function may contain a perpendicular population inversion. The
reason for considering such a model is because when the solar wind plasma is trapped in
a magnetic mirror-like geometry, which may exist in a magnetic cloud structure, or in
the immediate downstream of the quasi-perpendicular region of the Earths collisionless
bow shock, the perpendicular distribution may possess a loss-cone, or equivalently,
perpendicular population inversion. If the Loss-cone feature is su ciently pronounced,
then it is possible that cyclotron harmonic modes may also get excited in addition to
the purely-growing branch.
Indeed, observations near the Earth bow shock show that loss-cone feature associated with the electron distribution function and harmonic electron cyclotron waves are
detected [81, 82, 83, 84]. Although such waves are customarily interpreted in terms of
electrostatic Bernstein wave framework, sometimes they also contain signicant transverse polarization indicating that such waves may either by obliquely propagating Bernstein modes or possibly the cyclotron-harmonic ordinary modes, which is the focus of
the present investigation.
43
4.2
Mathematical formalism
E (r; t) +
B=
r
r
B (r; t)
c
rv f (r; v; t) = 0, (4.1)
1 @E 4
+
J,
c @t
c
(4.2)
1 @B
.
c @t
E=
(4.3)
s2 + c 2 k 2
ij
c 2 ki kj + 4 s
Ej = 0.
ij
(4.4)
From above, the dispersion relation for perpendicular propagation can be written as
0
B
B
det B
B
@
c2 kk2
xx
xy
yy
c2 k 2
yy
xx
xy
!2
!2
0
N?2 +
zz
2
xy
zz
2
c2 k?
!2
C
C
C = 0,
C
A
N?2 = 0,
(4.5)
(4.6)
the above equation (4.6) gives two decouple modes, that are the X mode
2
c 2 k?
=
!2
yy
1+
2
xx
xx yy
44
(4.7)
zz
we write
zz
zz
(4.8)
! 2pa
=1+ 2
!
vk @
v? @v?
@
@vk
dvvk
fa +
P R
n
! 2pa
S
!2
vk Jn (b)
dv !
kk vk
and $ =
fa ,
(4.9)
! kk vk @
v?
@v?
+ kk @v@ k .
zz
! 2pa
= 1+ 2
!
dvvk
@
@vk
vk @
v? @v?
Z
2
! 2pa X
vk Jn (b)
dv
+ 2
! n
! n
kk v k
fa
kk v k @
@
+ kk
v?
@v?
@vk
fa .
(4.10)
For perpendicular propagation kk = 0;therefore the above equation leads to the following form
zz
Z
! 2pa
vk @
@
fa
= 1+ 2
dvvk
!
@vk v? @v?
XZ
! 2pa
@fa
+
dvvk2 Jn2 (b)
.
! (! n a ) n
v? @v?
(4.11)
By using equation (4.11) we take the Davidson and Wu [1970] expression for O mode
dispersion relation
2 2
! =c k +
! 2pa
1
X
! 2pa
n= 1
n2
(! 2
2
a
n2
2)
a
vk2
2
dvJn (b)
v?
@fa
.
@v?
(4.12)
The above general dispersion relation for O mode can also be written as,
2
c 2 k?
=1
!2
! 2pa
!2
1+
45
1
X
a (n)n
n2 2a
n=1
2
a
!2
(4.13)
a (n)
where
d3 v = 2
R1
dvk
1
R1
0
dv
k? v ?
Jn2
vk2 @fa
,
v? @v?
(4.14)
trons and a = p for protons), k? is the wave number perpendicular to the ambient
magnetic eld B0 , ! pa = (4 e2a n0 =ma )1=2 is the plasma frequency for any species a,
a
= ea B0 =ma c is the gyro frequency, Jn (b) is the Bessel function of rst kind of
order n, and fa is the particle velocity distribution function normalized to unity i.e.,
R 3
d vfa = 1. Here, ea and ma , stand for the charge and mass of species a respectively
and c correspond to the speed of light in vacuum.
4.2.1
In order to model the Loss-cone, we adopt a relatively simple model. That is, we
assume that the velocity distribution function has a unique property that can be written
as product of the reduced parallel distribution and perpendicular distribution, fa (v) =
fa (v? )Ga (vk ). When the distribution is separable into functions that only depend on
parallel or perpendicular components of velocity then the dispersion equation and hence
the stability of the computation of the dispersion relation for perpendicular mode is
insensitive to the shape of the parallel velocity part of the distribution i.e., Ga (vk ).
In what follows, we consider only the electron response. Upon dening the eective
parallel temperature
Tk = m
1
1
(4.15)
where the species subscripts a = e have been omitted. It follows that the detailed
mathematical form of G(vk ) becomes unimportant in the subsequent discussion. However, for the sake of concrete demonstration, we implicitly assume that G(vk ) is given
by the Maxwellian form, G(vk ) = m=2 kB Tk
46
1=2
exp
f (v? ) =
A0 =
dx x exp
1
2A
v0
exp
0
"
1
exp
2
(v?
v0 )
2
v02
p v0
(4.16)
erfc
v0
(4.17)
where erf c(x) is the complementary error function. Note that the eective perpendicular temperature can be dened by
T? = m
A2 =
dx x3 exp
3
dv? v?
f (v? ) =
v0
m 2 A2
,
2A0
(4.18)
A2 =
1
2
v02
2
+ 1 exp
v02
p v0
v02
2
3
2
erfc
v0
(4.19)
We thus consider more realistic thermal ring model in our following linear analysis
of O mode dispersion relation.
4.3
In the present section we discuss the linear dispersion relation for the purelytransverse ordinary (O) mode when the underlying plasma is characterized by a Losscone, or a thermal ring perpendicular velocity distribution function. In the extreme case
of cold ring perpendicular distribution, Cheng [10] discussed the cyclotron harmonic
instability associated with the ordinary electromagnetic mode. However, the cold ring
is an ideal limiting case of the loss-cone distribution, which is unlikely to be realized
in natural situations.
When the model distribution Eq. (4.16) is used in Eq. (4.14) and then in Eq.
47
0 = R z
(b0 )2 A2
A2
+ b0 2
A0
k A0
(x u0 )2
Z 1
1
1
A2 X
dx Jn (b0 x) [Jn 1 (b0 x)
bz 2
A0 n=1 z 2 n2 0
0 2
(x u0 )2
(4.20)
,
z=
, b0 =
k?
, R=
T?
,
Tk
! 2pe
2
, u0 =
v0
8 n0 Tk
.
B02
(4.21)
In the above R is the temperature ratio, u0 is the normalized perpendicular ring speed,
and
O mode instability situation rst considered by Davidson and Wu [45]. On the other
hand, in the limit of u0 ! 1, we recover the limiting case considered by Cheng [10]. In
the numerical example to be discussed subsequently, we consider the input parameters
corresponding to
= 20, R = 0:1,
4.3.1
Numerical results
Shown in Fig. 4.1 are the O-mode dispersion relation for ring speed u0 = 0; 1; 2; 3; 4; 5:
The blue curves correspond to the real frequency, Re(z) = Re(!= ), while the red
curves depict the growth rate Im(z) = Im(!= ). The horizontal axis is the normalized
perpendicular wave number b0 = k? = . The rst case (a) of u0 = v0 =
= 0 corre-
spond to the classic O mode instability [45, 76]. In this case, only the purely-growing
instability corresponding to the harmonic mode number n = 0 is excited. However, as
the ring speed is increased from u0 = 1 to u0 = 2, the rst harmonic cyclotron instability near Re(!)
u0 = 4 (panel e) the second harmonic instability growth rate further increases until
u0 = 5, when the third-harmonic mode [Re(!)
48
Figure 4.1: The real frequency Re(z) =Re ! (blu) and growth rate Im(z) =Im !
(red) for the O mode are plotted versus normalized wave number b0 = k? , for =
! 2pe
v0
so forth. The limit of u0 ! 1 is the cold ring case considered by Cheng [10], but even
for moderate u0 = 5, we already have a dispersion relation that is quite similar to the
cold ring case.
4.4
In this section we plot the maximum growth rate for the purely-growing branch
(corresponding to the harmonic mode characterized by n = 0 in the Bessel function
summation) and for harmonic modes characterized by n = 1 and n = 2 over the phase
space (
k ; T? =Tk )
49
= (! pe =
2
e)
Figure 4.2: The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the zeroth-harmonic purely-growing mode,
(n=0)
max
, in two dimensional
T?
= !pe = 20, and for (a) u0 = 0, (b) u0 = 1, (c) u0 = 2, (d)
space
k ; Tk , for
u0 = 3, (e) u0 = 4, (f) u0 = 5.
k ; T? =Tk ),
for
(n=0)
max =
, in two-dimensional
(d) u0 = 3, (e) u0 = 4, and (f) u0 = 5. The contour levels indicate the numerical values
of
(n=0)
max =
. Note that the case of u0 = 0 is the same as that shown by Hadi et al. [76].
For higher values of the ring speed, u0 , it can be seen that the range of instability in
(
k ; T? =Tk )
k ; T? =Tk )
for which n = 0
mode is stable, especially for increasing n0 . Figure 4.3 plots the maximum growth rate
for the rst-harmonic cyclotron branch (corresponding to the rst harmonic mode char-
50
Figure 4.3: The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the rst harmonic cyclotron mode,
! pe 2
T?
k ; Tk
, for =
(e) u0 = 4, (f) u0 = 5.
(n=1)
max
! pe
k ; T? =Tk )
to 5. Figure 4.3 shows that the rst-harmonic cyclotron instability does not exist for
u0 = 0, but as u0 is increased to u0 = 1, the n = 1 mode instability begins to get
excited, especially for high
k.
instability becomes even more prominently excited and the range of instability extends
to lower
(n=2)
max =
second-harmonic mode does not exist for u0 = 0 and u0 = 1, at least over the range of
parameters considered in the gure, but as u0 , increases to 2 and beyond, the second-
51
Figure 4.4: The lled counter plot (with coulor mape) of the maximum growth rate
correspond to the second harmonic cyclotron mode,
T?
k ; Tk
! pe 2
, for =
(e) u0 = 4, (f) u0 = 5.
(n=2)
max
52
harmonic mode becomes unstable, and the maximum growth rate generally increases
monotonically as a function of u0 . The fact that higher harmonics are excited for nite
u0 i.e., for distribution with perpendicular population inversion, means that the system
contains more free energy. That is, in addition to the parallel temperature anisotropy,
there also exists perpendicular ring feature. This additional free energy leads to the
excitation of cyclotron harmonic instability.
4.5
Conclusion
In this chapter, the instability associated with electromagnetic O mode is revisited and extended the analysis to another direction. For such an end, we considered a
thermal ring perpendicular distribution. By doing so, we made a connection between
the two earlier works by Davidson and Wu [45] and Cheng [10]. We have demonstrated that the O mode instability for thermal ring distribution may be excited for
cyclotron harmonics as well as the purely-growing branch, depending on the value of
the normalized ring speed u0 .
The present chapter also investigated the general O mode instability characteristics
as a function of input parameters
maximum growth rate curve. It is shown that for increasing values of ring speed u0 the
range of instability in (
k ; T? =Tk )
we move to the high harmonics mode (rst harmonic mode n = 1) cyclotron harmonic
mode instability does not exist for lower values of ring speed. As the thermal ring
speed u0 is increased, the maximum growth rate curve increases. Finally the maximum
growth rate is plotted for second harmonic mode (n = 2) cyclotron instability in the
(
k ; T? =Tk )
space. It is shown that even for higher values of ring speed as compared
53
Chapter 5
Ordinary Mode Instability For Low
Beta Plasmas
5.1
Introduction
Davidson and Wu [45] who simply displayed in a gure that displays the demarcation between the stable and unstable regime without explanation. While the analysis
in reference [24] was still restricted to the bi-Maxwellian electron distribution. The
analysis of the purely growing O mode were extended by Ibscher and his collaborators
54
[46, 47] to the counter-streaming beams possessing the temperature anisotropy, and
found a signicant modication to the marginal stability condition, especially in the
low-beta regime. Finally, Ibscher and Schlickeiser [77] boldly applied the ndings in
[46, 47] to the solar wind proton anisotropy boundary by assuming that the temperature anisotropy and beta for the electrons and protons are the same, so that the results
obtained for the electrons can be automatically translated for the protons as well.
While these developments are signicant and interesting, it should be noted that
the work presented in references [24, 45, 46, 47, 77] approach the O mode instability
problem by solely referring to the marginal stability analysis, but did not actually solve
the dispersion relation for the O mode. The numerical solutions are not only interesting
in and of themselves, but they may be essential, for instance, in the dynamical theory of
the O mode instability, such as in the quasilinear formalism. It is for this reason that we
revisit the O mode instability from the standpoint of direct numerical solutions of the
O mode dispersion relation. An important outcome of the present analysis is that we
may construct the maximum growth rate contours in (
k ; T? =Tk )
space so as to verify
and validate the marginal stability threshold curve constructed in [24, 45, 46, 47, 77].
In this chapter, rst we want to show that in perpendicular propagation only the O
mode retains the temperature anisotropy while other mode i.e., X mode does not. However the later exhibit the thermal eects only in the perpendicular direction. We derive
the O mode dispersion relation for counter-streaming plasma distributions possessing
the temperature anisotropy and numerically solve the dispersion relation.
5.2
Mathematical formalism
The basic plasma model we assume here, consist of collisionless plasma immersed
in a uniform external magnetic eld B0 having equal number of plasma charged species
i.e., electrons and ions, and free from any electric eld. We also assume that the ions
provide a uniform positive xed background because their dynamics are ignored. For
exploring such a plasma we utilize Vlasove model consisting of collisionless Vlasove
55
equation
@fa (r; v; t)
e
+ v rr fa (r; v; t) + (E (x; t)) rv fa (r; v; t) = 0,
@t
m
(5.1)
r
r
1 @E 4
+
J;
c @t
c
(5.2)
1 @B
.
c @t
E=
(5.3)
c 2 ki kj + 4 s
ij
ij
Ej = 0.
(5.4)
From above, the dispersion relation for perpendicular propagation (kk = 0) can be
written as
B
B
det B
B
@
2
c2 k ?
!2
xx
xy
xy
2
c2 k?
!2
yy
zz
xx
yy
N?2 +
2
xy
zz
2
c2 k?
!2
C
C
C = 0.
C
A
N?2 = 0,
(5.5)
(5.6)
2
c2 k ?
.
2
!
zz
2
c 2 k?
,
!2
56
(5.7)
zz (k; !) = 1 +
XZ
dvvk
! 2pa
!2
vk @
v? @v?
@
@vk
fa +
X ! 2pa
!2
S v k Jn
$fa ,
(5.8)
S=
XZ
dv
$=
1
!
kk vk
kk v k @
@
+ kk
.
v?
@v?
@vk
The genral dispersion relation for the O mode takes the form
2
c 2 k?
=
!2
zz
=1+
X ! 2pa Z
!2
X n=1
X
a
! 2pa
! (! n
1
n=
a)
vk @
v? @v?
@
@vk
d3 vvk
fa
k? v ?
d3 vvk2 Jn2
@fa
.
v? @v?
(5.9)
zz
X ! 2pa
=1
!2
(5.10)
N?2 =
yy
2
xy
1+
(5.11)
xx yy
xy
xx ; yy ;
the
dispersion relation in Eq. (5.11) for X mode results two decoupled modes i.e., the
pure transverse X mode N?2 =
yy ,
xx
= 0. The Bernstein
mode is discussed in detail in chapter 4 for thermal ring distribution function while for
bi-Maxwellian distribution it is given below
xx
=1+
XX
a
xx
=1
X
a
! 2pa
! (! n
! 2pa
! (! n
57
a)
a)
v? n2 Jn2 @fa
dv
,
b2 @v?
3
X n2 In ( a )e
n
(5.12)
(5.13)
yy
=1+
XX
a
yy
=1
XX
a
xy
= i
2
c2 k ?
)
!2
! 2pa
! (! n
! 2pa
! (! n
XX
a
xy
yy
may be written as
a)
d3 v v? (Jn0 )
n2 In e
a)
! 2pa
! (! n
@fa
,
@v?
In e
(5.14)
(5.15)
a)
X X ! 2pa In e
i
! (! n
a
n
d3 v
(5.16)
a)
(5.17)
Close observations shows that, of the two modes it is clear that X mode does not
support any anisotropy-driven instabilities since the dispersion relation is identical to
the case of isotropic thermal distributions. Consequently, only the O mode support
the possibilities of instabilities. Now we study the O mode instability in detail for
counterstreaming plasma in the follwoing sections.
5.3
To obtain the general dispersion relation for the electromagnetic O mode propagating in perpendicular direction to the uniform ambient magnetic eld B0 in homogeneous
and collisionless plasma we follow the component of the dielectric tensor
2
X ! 2pa
c 2 k?
= 1+
!2
!2
X ! 2pa X Z
!2
d3 vvk
@
@vk
vk @
v? @v?
2
dv
v k Jn
n
kk v k
58
fa +
kk v k @
@
+ kk
v?
@v?
@vk
fa , (5.18)
X ! 2pa X Z
!
P1
vk @
v? @v?
fa +
vk Jn 1 @fa
.
dv
! n v? @v?
3
(5.19)
2
X ! 2pa
c 2 k?
=1+
!2
!2
2
X ! 2pa
c 2 k?
=
1
+
!2
!2
@
@vk
d3 vvk
#
1
2
X
v
@f
n
@f
k
a
a
a
,
d3 v vk
+
Jn2 (z)
@vk
v? @v? (! n a )
1
"
@fa X ! 2pa
d vvk
+
@vk
!2
3
dv
1
X
vk2
2
Jn (z)
v?
n= 1
(5.20)
@fa
n a
. (5.21)
@v? (! n a )
a (n) =
where
d3 v = 2
R1 R1
1
X ! 2pa
a
!2
d3 v Jn2
1+
k? v ?
a
1
X
a (n)n
n2 2a
n= 1
2
a
!2
(5.22)
vk2 @fa
,
v? @v?
(5.23)
and a = p for protons), k? is the wave number perpendicular to the ambient magnetic
eld, ! pa = (4 e2a n0 =ma )1=2 is the plasma frequency for species a,
= ea B0 =ma c is
the gyro frequency, Jn (x) is the Bessel function of rst kind, and fa is the particle
velocity distribution function. Here, ea , ma , and c stand for the charge and mass for
species a and the speed of light in vacuo, respectively. It is noticeable that
a (n)
is
independent of !.
5.4
Our model distribution function in this study (The ordinary mode instability for
low beta plasma) is drifting bi-Maxwellian particle velocity distribution. In astrophysical sources, when supersonic plasma ows encounter obstacle e.g., such as when solar
59
wind interact with planetary magnetosphere comets gives rise to the countermoving
plasma streams.The counterstreaming plasma are subject to a variety of kinetic plasma
instabilities. As in [46, 47, 77], we adopt the following form for the distribution function
fa (v? ; vk ) =
as M
1
exp
3=2 v 2 v
?a kas
(5.24)
2
v?
2
v?a
exp
vk + Vas
2
vkas
. (5.25)
The above model represents the counter-streaming Gaussian beam distribution posP
sessing temperature anisotropy. The parameter as = nas = s nas is the ratio of the
density of the stream s to the density of entire plasma, Vas , v(?;k)as = (2T(?;k)as =ma )1=2
are the streaming velocity and the thermal velocities of stream s. The distribution
P
P
function satises the relations s as = 1 and s as Vas = 0, which represent the to-
tal charge and current neutrality conditions, and eective parallel and perpendicular
temperatures as well as the ratio of the two temperatures are given by
e
kB Tka
= ma
dv
ma
=
2
kB T?a
e
Tka
T?a
vk2 fa
ma X
=
2 s
Z
as
2
as vkas
2V 2
1 + 2 as
vkas
(5.26)
2
ma v?a
,
2
!
2Vas2
1+ 2
.
vkas
2
d3 v v?
fa =
2
vkas
2
v?a
(5.27)
(5.28)
as
X
s
1
2
3=2 v 2 v
?a kas
2
as
1=2
2
v?
2
v?a
dv? v? exp
dxx exp
x2
1
1
60
dy exp
Z
"
1
1
vk + Vas
2
vkas
dvk exp
Vas
y+
vkas
X
s
as
.(5.29)
s as
= 1. The
T?a
2
ma v?a
=
.
2
(5.30)
as
dvvk fa =
vkas
1=2
as
1
1=2 v
kas
Vas
vkas
dy y
1
1
y2
vk + Vas
2
vkas
dvk vk exp
X
as Vas
(5.31)
as Vas
(5.32)
= 0.
5.5
Making use of the model distribution (5.24) in the dispersion relation (5.22) and
R
a
solving the integral dvvk @f
for streaming distribution function in Eq. (5.23) becomes
@vk
zz
XX
! 2pa
as 2 2
!
vk
vk
Vas
= 1 2
v? dv?
dvk vk
+ 2
2
2
vkas v?a vkas
0
1
s
!
2
2
vk + Vas
v?
1
exp
exp
2
2
3=2 v 2 v
v?a
vkas
?a kas
Z 1
Z
1
XX
X
! 2pa
k? v?
2
2
v? dv? dvk vk2 Jn2
as
! (! n )
0
s
n= 1
!
2
2
vk + Vas
1
v?
exp
exp
.
2
2
3=2 v 4 v
v?a
vkas
?a kas
61
(5.33)
XX
s
2
1
1
2
XX
Z
as
! 2pa
as 2 2
!
1
X
! (!
!
XX
dxxe
y2
as
1
1
2
1
1
Z 1
2
vkas
2
2
) v?a
0
"
! 2pa 1
as 2
!
2
XX
x2
Vas2
2
vkas
! 2pa
n= 1
dy y 2 +
2
c 2 k?
= 1
!2
dxxe
2
vkas
1
x2
2
v?a
y2
k? v?a
Jn2
vk +Vas
vkas
=y
#
Vas2
e
2
v?a
y2
= x and
(5.34)
2
vkas
2
v?a
1
X
n=
dy
"
v?
v?a
2
vkas
! 2pa
2
! (! n ) v?a
1
Vas2
2
v?a
2
1 vkas
+ 2
2 v?a
In ( a ) e
, (5.35)
when we write the innite summation as the summation over positive harmonics only,
the above equation reduced to
2
c 2 k?
= !2
XX
2
as ! pa
XX
s
2
as ! pa
2
vkas
2
v?a
V2
1 + 2as
vkas
1
X
2n2
n=1
In ( a ) e
2
(!
n2 2 )
Here In ( a ) is the modied Bessel function of the rst kind with argument
2 2
k?
v?a =(2
2
a ).
(5.36)
e
Note that if we make use of the eective temperature Tka
dened in
Eq. (5.26), then Eq. (5.36) becomes identical to the dispersion relation rst obtained
in [45] for bi-Maxwellian distributions, if we simply replace the Maxwellian parallel
temperature by the eective temperature. However, [46, 47, 77] dened the parallel
temperature on the basis of the Maxwellian thermal spread associated with each drifting
beam component
kB Tkas =
2
ma vkas
(5.37)
rather than on the basis of taking the second parallel velocity moment of the total
distribution function. If one permits such an approach, then the dispersion relation
Eq. (5.22) can be used to discuss the extension of the O mode instability to low-beta
62
regime. In the present thesis, we shall adopt the approach taken in [46, 47, 77].
This equation is structurally similar to the O mode dispersion relation for purely
bi-Maxwellian plasma. Consequently, the conclusion that only purely growing mode
exists is equally applicable, rewriting
n2
!2
n2
1+
!2
n2
!2
(5.38)
2 2
c k =! +
XX
2
as ! pa
XX
a
2
vtk
2
as ! pa 2
vt?
2
vtk
2V 2
1 + 2as
vtk
2Vas2
1+ 2
vtk
2
as ! pa 2
vt?
1
X
1
X
n=
!2
n2
!2
1
In ( a )e
I ( )e
2 n a
(5.39)
n= 1
or
2 2
c k =! +
2
as ! pa
XX
a
+2
XX
a
2
as ! pa
2
vtk
2
vt?
2
as ! pa
2
vtk
2
vt?
2V 2
1 + 2as
vtk
2
as ! pa 2
vt?
XX
2
vtk
2V 2
1 + 2as
vtk
1+
2Vas2
2
vtk
1
X
1
X
n=1
!2
n2
!2
In ( a )e
I ( )e
2 n a
n=1
I0 ( a )e
P1
n=1 In ( a )e
+I0 ( a )e
(5.40)
a
!2
1+2
XX
a
XX
a
2
as ! pa
+c2 k 2 +
2
vtk
2
as ! pa 2
vt?
XX
a
XX
a
s
2
vtk
2
as ! pa 2
vt?
2V 2
1 + 2as
vtk
2
vtk
2
as ! pa 2
vt?
2V 2
1 + 2as
vtk
63
XX
a
1+
!
2Vas2
2
vtk
I0 ( a )e
1
X
In ( a )e
2 Tka
as ! pa
T?a n=1 ! 2 n2
!
a
P1
n= 1 In ( a )e
(5.41)
or
!2
1+
"
XX
a
2
vtk
2
as ! pa
2
vt?
XX
s
1+
2Vas2
2
vtk
1+
2Vas2
2
vtk
2
as ! pa 2
vt?
XX
a
2
vtk
2
vtk
2Vas2
1+ 2
vtk
2
as ! pa 2
vt?
2In ( a )e
! 2 n2
!
!
a
c2 k 2
I0 ( a )e
(5.42)
! 1
2Vas2 X 2In ( a ) Exp (
1+ 2
1+
vkas n=1
n2 2 + 2
s
" 2
!
#
XX
vkas
2Vas2
2
2
=
1+ 2
1
c2 k?
as ! pa
2
v?a
vkas
s
!
2
2
XX
v
2V
2 kas
1 + 2 as I0 ( a ) Exp ( a ) .
as ! pa 2
v?a
vkas
s
2
"
XX
v2
2 kas
as ! pa 2
v?a
a)
(5.43)
R =
XX
a
2
c 2 k?
L=1+
2
as ! pa
2
vkas
2
v?a
XX
a
XX
a
"
2V 2
1 + 2 as
vkas
2
as ! pa
2
as ! pa
2
vkas
2
v?a
(5.44)
= R=L,
2
vkas
2
v?a
2V 2
1 + 2 as
vkas
2V 2
1 + 2 as
vkas
a) ,
I0 ( a ) exp (
1
X
2In ( a ) e
n2 2a +
n=1
(5.45)
(5.46)
Note that Eq. (5.43) is only a formal expression for the growth rate that cannot be
directly used to solve for
, since
the above expression is useful for considering the threshold condition, or the marginal
condition for the instability. Since the denominator L is positive denite, the general
64
stability condition can be discussed on the basis of the requirement that R > 0.
Following Refs. [46, 47, 77] , particularly, Ref. [77], we consider two components
(s = 2) counter-streaming beam system characterized by Vas = Va1 for s = 1 and
Va2 for s = 2, respectively, which leads to the following relations that result
P
from the charge and current neutrality conditions, i.e. the two constraints s as = 1
P
and s as Vas = 0, leads to
Vas =
a2
a1 ,
=1
and Va2 =
a1 Va1
(5.47)
a2
Also [77] assumed that the thermal spread associated with the two counter-streaming
2
2
= vka2
Gaussian beams is the same, vka1
2
vka
. We further introduce the following
ke
kp
=
=
2
4 n0 me vke
8 n0 kB Tke
,
B02
2
4 n0 mp vkp
8 n0 kB Tkp
(5.48)
B02
ke
B02
B02
(5.49)
2
es Ves
.
2
s is Vis
(5.50)
For two streams when moving in opposit directions the above equation can be written
2
e1 Ve1
2
i1 Vi1
2
e2 Ve2
:
2
i2 Vi2
+
+
(5.51)
2
e1 )] Ve1
2
i1 )] Vi1
1
=
i1 1
e1
Ve12
:
2
e1 Vi1
i1
(5.52)
Pe =
X 4 n 0 me
B02
s
2
es Ves
4 n 0 me (
2
e1 Ve1
B02
65
2
e2 Ve2 )
4 n0 me Ve12
B02
1
e1
.
e1
(5.53)
Figure 5.1: The real frequency (blue) and the growth rate (red) for the electromagnetic
ordinary-mode versus e (logrithmic horizontal scale) for = (! pe = e )2 = 20, T? =Tk =
0:2, k = 2, = 10, and Pe = 0. The choice of Pe = 0 means that this case corresponds
tothe bi-Maxwellian distribution. In this case, there is no purely-growing instability.
kp
ke
assumption as in [77], who further assumed that T?p = T?e and Tkp = Tke . [77] chooses
= 10. We do the same in the present thesis, and we do not vary this parameter. The
parameter
is related to the ratio of ion versus electron drift speeds. The physical
5.6
Since the electron and proton temperatures as well as the betas are assumed to
be equal [77], we may designate the temperature ratio and the parallel beta without
66
Figure 5.2: The same as Fig. 1, except that the drift speed is nite, as indicated by
Pe = 1. In this case, the O-mode instability begins to operate.
k.
stability condition on the basis of the condition R > 0 see Eq. (5.46). They report that
the necessary condition for the instability can be represented by
T?
< 1
Tk
f(
k)
(5.54)
f(
k)
= 2L(
k)
+ 2:2I0 [L(
L(
k)
= ln 0:45(
k )]
Pe
1+
(5.55)
+ Pe ) .
(5.56)
f(
k )= k
on the max-
imum growth rate contours in order to test the validity of the above approximate
formula. Before we do that, however, let us display the actual numerical dispersion
relation for the O mode, including the stable solutions, for some sample input parameters.
67
Figure 5.3: The same as Fig. 2, except that the drift speed is increased to Pe = 1:5.
In this case, the O-mode instability operates over a wider range of e .
5.7
The numerical plots of real and imaginary parts of the dispersion relation in Eq.
(5.22) can be obtained by numerical solution of the dispersion relation, which can be
expressed in dimensionless form
0 = z
"
z2
+
R
2
1+
2Pe
k
N
X
2In ( e )e
n2 z 2
n=1
1 + I0 ( e )e
2Pe
I0 ( e )e
z2 =
!2
;
2
! 2pe
;
2
e
R=
(5.57)
T?
:
Tk
Figures (5.1)-(5.3) plot the solutions. The input parameters are indicated in the
gure titles. In the case of Pe = 0 Fig. (5.1), since the system is the simple biMaxwellian distribution, the purely-growing O mode instability is absent. This is
68
because the present choice of input parameters do not happen to satisfy the instability
threshold criterion for bi-Maxwellian system. However, as the drift speed is increased,
as indicated by Pe = 1 in Fig. (5.2), the O mode instability becomes operative over
a range of
e.
Note that the growth rate associated with the O mode instability is
shown in red curve. The real frequency associated with the unstable O mode is zero,
as a careful visual inspection reveals, so that the assumption of purely growing mode
is validated. Those roots having zero growth rates have nite frequencies (the real
part of the complex frequency is shown in blue), and so are irrelevant to the present
discussion concerning the purely growing O mode instability. However, these roots
are also displayed for the sake of completeness. In Fig. (5.3), we show the case of
even higher drift speed, corresponding to Pe = 1:5. In this case, the purely growing
O mode instability (with zero real frequency) operates over a wider range of
than
the previous case of Pe = 1. Note also that the general O mode dispersion relation
contains not only the stable fast electromagnetic wave, but also numerous cyclotron
harmonic modes intersecting the fast mode. as can be seen from Figs. 5.1-5.3, all
high harmonic cyclotron modes and fast ordinary mode are stable (i.e., zero growth
or damping rates). As noted, these stable waves bear no direct relevance to the O
mode instability of present interest. We note, however, that it is important to retain
su ciently high number of harmonic terms in the summation over the Bessel function
series in order to obtain accurate numerical value for the O mode instability. This is
because the unstable mode with zero real frequency (i.e., the zeroth-harmonic purelygrowing instability) can attain maximum growth rates that can be many times the
magnitude of the electron cyclotron frequency. To put it another way, even though
the purely growing mode is associated with the zeroth cyclotron harmonic (n = 0),
in solving for the purely-growing solution, it is necessary to retain su cient number
of harmonic terms in Eq. (5.57). In the numerical scheme we typically kept up to
10th-order cyclotron harmonic summation i.e., N = 10 in Eq. (5.57).
We now carry out a series of numerical root solution, as in Figs. 5.1-5.3, over a wide
range of input parameters. We choose the same set of input parameters as in Figs.
5.1-5.3, namely,
= (! pe =
2
e)
= 20 and
Figure 5.4: The lled contour plot (with colourmap) of the maximum growth rate,
max = e , in two dimensional T? =Tk , k space for the same set of input parameters
as in Figs. 5.15.3, namely, = (! pe = e )2 = 20, = 10, and Pe = 0. In this case,
since the system is the bi-Maxwellian the lowest maximum growth rate contour and the
approximate marginal stability curve (shown in red) are identical to that of Davidson
and Wu.
70
Figure 5.5: The same as Fig. 4, except Pe = 0:5. In this case, the maximum growth
rate contours shown deviation from the bi-Maxwellian case. Note that the analytical
marginal stability curve, shown in red, does not show good agreement with the lowest
maximum growth rate contour.
71
Figure 5.6: The same is Fig. 5.5, except now the drift speed is increased to Pe = 1.
Again, the anaylatical marginal stability curve is not is good agreement with the lowest
maximum growth rate contour.
and
(
k ; T? =Tk ).
case, the lowest maximum growth rate contour and the approximate marginal stability
curve Fig. 5.3, shown in red, are in good agreement, and both are consistent with the
marginal stability threshold investigated by Davidson and Wu [45]. As we increase the
drift speed to Pe = 0:5 Fig. 5.5 the maximum growth rate contours show deviation from
the bi-Maxwellian case [46, 47]. Now, the low-beta regime also becomes unstable. Note
that the analytical marginal stability curve (5.54)-(5.56) does not show good agreement
with the lowest maximum growth rate contour. As we further increase the drift speed
to Pe = 1, a large swath of low-beta regime becomes unstable. However, as with the
previous case of Pe = 0:5, the analytical marginal stability curve (5.54)-(5.56) is not in
good agreement with the lowest maximum growth rate contour. This shows that while
the essential conclusion of [46, 47, 77] is certainly correct, their construction of the
marginal stability curve on the basis of approximate analysis needs improvement. In
72
closing, we note that the stability analysis carried out thus far pertains to the electrons
only, and the ions play a relatively minor role (via the parameter ). However, as
noted previously, Ibscher and Schlickeiser [77] applied the ndings in [46, 47], which
is conrmed in the present chapter of this thesis, to the solar wind protons. This
was possible because they assumed that the temperature anisotropy and beta for the
electrons and protons are the same, so that the results obtained for the electrons can
be automatically translated to the protons as well.
5.8
Summary
higher harmonics cyclotron mode and the fast mode are stable although. Similarly we
compare the analytical marginal stability curve with the maximum growth rate counter
for bi-Maxwellian case when pe = 0, both are in good agreement and are consistent
with the Davidson and Wu [45] results. For increasing drift speed the maximum growth
rate counter clearly deviates from bi-Maxwellian case and the analytical growth rate
curve shows a large disagreement with the lowest maximum growth rate contour.
73
Chapter 6
Summary and Conclusion
In chapter 1 we have given a brief description of Maxwellian and non-Maxwellian
plasma and highlighted some non-Maxwellian distribution functions which acts as a
source of free energy. It is pointed out particularly that not all the distributions having
free energy can cause instability but only a special class of distribution functions have
this ability. We have then presented an overview of various modes for perpendicular
propagations like electrostatic Bernstein wave and electromagnetic O mode with its
cyclotron harmonic eects which is the main focus of our present work.
The basic kinetic models governing the magnetized non-relativistic collisionless plasmas were presented in chapter 2. In this context, a brief introduction of kinetic theory
along with the Vlasov model are presented. Using Vlasov model a relativistic expression for conductivity tensor is derived in cylindrical polar coordinates by solving
linearized collisionless Boltzmann equation and Maxwells equations. The conductivity tensor components are kept general that can be used for any kind of specie a i.e.,
electron and ion of plasma. However the derived expressions are of great importance
and are used for electron plasma in which the ion dynamics are ignored in this thesis.
Further the elements of the conductivity tensor are presented in terms of rst kind
of Bessel function of order n and their derivatives. These components can be used to
obtain the dispersion relation for various electrostatic and electromagnetic modes that
exist in plasma such as Bernstein wave, R & L waves, O mode, X mode, Alfven waves
etc.,. Particularly in this thesis we have focused on perpendicularly propagating modes
74
was rst discussed by Davidson and Wu [45]. While the O mode instability
for cold ring distribution was discussed by Cheng [10]. These instabilities remained
largely forgotten until quite recently, when it was suggested that the bi-Maxwellian O
mode instability might be relevant to the solar wind anisotropy problem [46, 47, 76, 77].
The recent studies of [46, 47, 76, 77] extended the bi-Maxwellian O mode instability
by considering a system of counter-streaming plasma distributions, which might be
relevant to the solar wind distributions under certain circumstances. In this chapter,
the analysis of the O mode instability are extended to another direction. For solar wind
plasma trapped in magnetic mirror-like geometry such as magnetic clouds or in the
vicinity of the Earths collisionless bow shock environment, the velocity distribution
function may possess a perpendicular inversion feature. In order to investigate the
inuence of such a feature, we considered a thermal ring perpendicular distribution.
By doing so, we made connection with the earlier work by Cheng [10]. We have
demonstrated that the O mode instability for thermal ring distribution may be excited
for cyclotron harmonics as well as the purely-growing branch. It is also investigated
that the general O mode instability characteristics as a function of input parameters
k
and the temperature ratio T? =Tk , by plotting the maximum growth rate contour
k ; T? =Tk ).
ndings may be useful for future interpretation of the solar wind data, especially when
the measurements in the solar wind contains su cient information on the velocity
distribution function.
By making use of the Vlasov-Maxwell system of equations, in chapter 5 the dispersion relation of perpendicularly-propagating O mode including the higher harmonic
eects is obtained with a counter-streaming bi-Maxwellian distributions for the electrons and protons in order to further investigate and verify the ndings reported in
[46, 47, 77]. By directly plotting the numerically-obtained maximum growth rates in
the
ke ; T?e =Tke
carried in [46, 47, 77], we have conrmed the validity of their ndings. We also found,
76
however, the approximate analytical marginal stability curve obtained in these references show some deviations from the maximum growth rate contours, indicating that
an improved formula is called for.
In spite of this minor discrepancy, the conclusion reached by [77] that, under a
judicious choice of parameters, the O mode instability indeed is capable of providing
the basic explanation for the lower left corner of the Bale-diagram [7] for the solar
wind protons is conrmed.
6.1
Future work
From the contents of this thesis, it is clear that the plasma characterized by thermal ring distribution function modies the properties of perpendicularly propagating
electrostatic and electromagnetic modes such as Bernstein wave and O mode. It is
pointed out in this work that there exist various theoretical and numerical methodologies to explore the dynamics of linear waves which can be used for possible explanation
of solar wind. A straight forward extension of the present work is to investigate the
behavior of such waves in relativistic plasma characterized by thermal ring distribution
of electrons. We are interested in extending this work in the near future by employing
quasi linear theory as it is also a common feature of space plasma that can helps us in
modifying our results.
77
Bibliography
[1] S. Ichimaru, Basic Principles of Plasma Physics, W. A. Benjamin, Inc. (1973).
[2] R. J. Goldston and P. H. Rutherford, Introduction to Plasma Physics, c IOP
Publishing Ltd. (1995).
[3] D. A. Gurnett, and Bhattacharjee, Introduction to Plasma Physics with Space
and Laboratory Application, Cambridge University Press, (2005).
[4] S. Zaheer, G. Murtaza, and H. A. Shah, Phys. Plasmas 11, 5 (2004).
[5] B. Lehnert Plasma Phys 9 (1967)
[6] J. C. Kasper, a. J. Lazarus, and S. P. Gary, Geophys. Res. Lett. 29, 1839 (2002).
[7] S. D. Bale, J. C. Kasper, G. G. Howes, E. Quataert, C. Salem, and D. Sundkvist,
Phys. Rev. Lett. 103, 211101 (2009).
[8] B. A. Maruca, J. C. Kasper, and S. D. Bale, Phys. Rev. Lett. 107, 201101 (2011).
[9] E. Marsch, Living Rev. Solar Phys. 3, 1 (2006).
[10] C. Z. Cheng J. Plasma Phys. 13, 335 (1975).
[11] G. Beke, J. L. Hirsheld and S. C. Brown Physical Review 122, 4 (1961)
[12] T. J. Birmingham, J. K. Alexender, M. D. Desch, R. F. Hubbard, and B. M.
Pedersen, J. Geophys. Res. 86, 8497 (1981).
[13] G. Ahirwar, Res. J. Engineering Sci. 1, 6 (2012).. Lett. 122, 4 (1961).
[14] K. F. Lee, Phys. Rev. Lett. 6, 356 (1972).
78
79
80
83