Introductory Plasma and Fusion Physics

Majid Khan and Muhammad Kamran

ii
Introductory Plasma and Fusion Physics

BY
Majid Khan
Department of Physics
QUAID-I-AZAM UNIVERSITY ISLAMABAD
and
Muhammad Kamran
Department of Physics
COMSATS UNIVERSITY ISLAMABAD
ii
Contents

1 Introduction 1
1.1 Plasma: the fourth state of matter . . . . . . . . . . . . . . . . . . . 1
1.2 Temperature: A microscopic view . . . . . . . . . . . . . . . . . . . . 3
1.2.1 Velocity distribution function . . . . . . . . . . . . . . . . . . 3
1.3 Debye shielding . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.3.1 Debye potential in a uniform positive background . . . . . . . 8
1.3.2 A general case of non-uniform positive background . . . . . . 13
1.3.3 Number of particles in the Debye sphere . . . . . . . . . . . . 14
1.4 Concept of quasi-neutrality . . . . . . . . . . . . . . . . . . . . . . . . 15
1.5 Definition and criteria of a plasma . . . . . . . . . . . . . . . . . . . . 16
1.6 Some applications of plasmas . . . . . . . . . . . . . . . . . . . . . . 16
1.7 Potential around a moving charge (optional) . . . . . . . . . . . . . . 18
1.8 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2 Plasma orbit theory 21

2.1 Uniform force fields . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.1 Unmagnetized case with homogeneous E ~ . . . . . . . . . . . . 22
2.1.2 Homogeneous magnetic field with E ~ =0 . . . . . . . . . . . . 23
2.1.3 Combine electric and magnetic fields . . . . . . . . . . . . . . 27
2.2 Non-uniform magnetic fields . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.1 Grad-B (⊥ to B) ~ drift . . . . . . . . . . . . . . . . . . . . . . 32
2.2.2 Curvature drift . . . . . . . . . . . . . . . . . . . . . . . . . . 33
~ field variations: Magnetic mirror . . . . . . . . .
2.3 Axial B . . . . . . . 36
2.3.1 Invariance of magnetic moment . . . . . . . . . . . . . . . . . 38
2.3.2 The magnetic mirror effect . . . . . . . . . . . . . . . . . . . . 39
2.3.3 The loss cone . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
2.3.4 Calculation of lost fraction for an isotropic plasma . . . . . . . 41
2.3.5 Magnetic mirror and Van Allen belt . . . . . . . . . . . . . . . 43
~ finite Larmor radius effect . . . . .
2.4 Spatially varying E: . . . . . . . 44
~
2.5 Time varying E: Polarization drift . . . . . . . . . . . . . . . . . . . 46
2.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47

3 Fluid theory of plasmas 53

3.1 Classical magnetic materials v/s plasma . . . . . . . . . . . . . . . . 54
3.2 Classical treatment of dielectric and plasma . . . . . . . . . . . . . . 55
3.3 Dielectric constant for plasma in low frequency limit . . . . . . . . . . 56

iii
iv CONTENTS

3.4 The Fluid equations . . . . . . . . . . . . . . . . . . . . . . . . . . . 57

3.4.1 Convective derivative . . . . . . . . . . . . . . . . . . . . . . . 57
3.4.2 Continuity equation . . . . . . . . . . . . . . . . . . . . . . . . 59
3.4.3 Momentum (Force) balance equation . . . . . . . . . . . . . . 61
3.4.4 Pressure gradient force . . . . . . . . . . . . . . . . . . . . . . 61
3.4.5 Equation of state . . . . . . . . . . . . . . . . . . . . . . . . . 61
3.4.6 Collision (with neutrals) and momentum change . . . . . . . . 63
3.5 Fluid drift perpendicular to magnetic field . . . . . . . . . . . . . . . 63
3.6 Two main limits or approximations . . . . . . . . . . . . . . . . . . . 65
3.6.1 Fluid response . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.6.2 Adiabatic response . . . . . . . . . . . . . . . . . . . . . . . . 66
3.7 Plasma approximation . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.8 Stress tensor (optional) . . . . . . . . . . . . . . . . . . . . . . . . . . 67
3.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

4 Plasma waves: A fluid description 71

4.1 Review of waves and wave solutions . . . . . . . . . . . . . . . . . . . 71
4.1.1 Phase and group velocities . . . . . . . . . . . . . . . . . . . . 71
4.2 Electron plasma frequency . . . . . . . . . . . . . . . . . . . . . . . . 73
4.2.1 Linear theory . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
4.2.2 Collisional damping of plasma oscillations . . . . . . . . . . . 76
4.3 Langmuir waves: electron plasma waves . . . . . . . . . . . . . . . . . 77
4.4 Acoustic and ion-acoustic waves . . . . . . . . . . . . . . . . . . . . . 79
4.5 Plasma approximation . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.5.1 Validity of plasma approximation . . . . . . . . . . . . . . . . 82
4.6 Electrostatic electron Oscillations ⊥ to B ~ in a cold plasma . . . . . . 85
~
4.7 Electrostatic ion waves ⊥ to B . . . . . . . . . . . . . . . . . . . . . . 90
4.8 Electromagnetic waves in field free cold plasma . . . . . . . . . . . . 93
4.8.1 Cut off . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.8.2 Index of refraction . . . . . . . . . . . . . . . . . . . . . . . . 96
4.8.3 Skin depth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
4.8.4 Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
4.9 Electromagnetic waves ⊥ to B ~0 . . . . . . . . . . . . . . . . . . . . . 99
~
4.9.1 Ordinary waves: E1 k B ~0 . . . . . . . . . . . . . . . . . . . . . 99
4.9.2 Extra ordinary wave: E~1 ⊥ B~0 . . . . . . . . . . . . . . . . . 101
4.10 Electromagnetic Waves k to B ~0 . . . . . . . . . . . . . . . . . . . . . 105
4.11 MHD waves in a uniform plasma . . . . . . . . . . . . . . . . . . . . 109
4.11.1 Alfvén Waves . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
4.11.2 Magnetosonic waves . . . . . . . . . . . . . . . . . . . . . . . 115
4.12 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
Chapter 1

Introduction

In this chapter we shall introduce the basic definition of plasma state and the corre-
sponding parameters used to distinguish various plasmas. The concepts of velocity
distribution function and temperature have been reviewed briefly. At the end of
this chapter, we shall provide the plasma criteria and some applications of plasma
physics in laboratory and space environments. That will clearly indicate that by
plasma we do not refer to the one which runs in our veins.

1.1 Plasma: the fourth state of matter

We are well aware of the three commonly known states of matter, namely solids,
liquids and gases. These states can transform into each other by the so-called
phase transition mechanism, for example, by increasing (or decreasing) temperature.
Consider a solid ice-cube, when we heat it well enough it transforms to a liquid (i.e.
water). On further heating, it again changes the state and you can see a gaseous
phase in the form of vapors coming out of the surface of the liquid. Now the question
is what happens next if we keep heating it? If we provide sufficient energy to the gas,
it would eventually be ionized, i.e. the electrons leave the atom. Consequently, we
will have negative electrons along with positive nuclei (which are generally referred
to as ions). This state of matter has very different/interesting properties than
its counterparts and whose constituents interact mainly via electromagnetic forces.
An ionized gas when satisfy certain conditions (that will be discussed later in the
chapter) can be termed as a plasma – the fourth state of matter. The change of
state with temperature is schematically presented in Fig. 1.1.
Although the exact definition of a plasma is not very well known, most generally it
is a state of matter containing enough free charged particles such that its dynamical
behavior is dominated by corresponding electromagnetic interactions. The sun and
other starts are hot enough that the gases there are almost completely ionized, simi-
larly the interstellar (between the stars) media – due to action of stellar radiations –
is also in the plasma state. Thus, almost all the matter in universe can be described
as plasma, therefore, sometimes it is said that almost 99% of the visible universe is
in plasma state.

1
2 CHAPTER 1. INTRODUCTION

Figure 1.1: Change in the state of matter with increasing temperature as a result
of melting, vaporization and ionization.

However, our earth is an exception in this regard. To explain this, let us calculate
the degree of ionization in our atmosphere (mostly composed of Nitrogen gas having
an ionization energy of 14.5 eV), using the well-known Saha’s equation, which reads

ni T 3/2 e−U/kB T
≈C , (1.1)
nn ni
where C = 2.4 × 1021 is a dimensionless constant, ni and nn are, respectively, the
number densities of ionized and neutral atoms, kB = 1.38×10−23 J/K = 8.617×10−5
eV/K is the Boltzmann’s constant, T is the gas temperature and U is the ionization
energy of the considered gas. From Eq. (1.1).

n2i
≈ CT 3/2 e−U/kB T , (1.2)
nn
or, equivalently s
ni CT 3/2 −U/kB T
≈ e (1.3)
nn nn
Here we need to calculate the quantity nn , i.e. the number density of neutral particles
at standard conditions. This can be computed by using the equation of state p =
nn kB T . Substituting p = 1 atm = 1.01 × 105 Pa, and T = 300 K, results in the
famous Laschmidt number, namely nn = 2.44 × 1025 m−3 . Upon substituting all
these values in Eq.(1.1), we get
ni
≈ 1.14 × 10−122 → 0 (1.4)
nn
Hence, there is practically no ionization at the standard conditions around us. Even
without going into the derivation of Saha’s equation we can understand its physical
meanings. First, there is ni term on the right side, which means that as the number
of ionized atoms increase, the fraction can go down. This is because of the fact
that the probability of recombination of electrons and ions to form neutral atoms
increase with ni . The temperature dependence in Eq. (1.1) is more complicated
and is plotted in Fig. 1.2. Here we note that at very low temperatures the degree of
ionization remains low, however when kB T approaches the ionization energy Ui there
is – due to the exponential term – an abrupt enhancement in the ionization. This fast
increase in ionization makes the system fully ionized at high temperatures, such as in
astronomical objects where very large temperatures, in comparison with earth, have
been observed. Plasmas of various types are found in different environments, e.g.
in our sun (where typical temperatures are in millions of degree Kelvin), interstellar
1.2. TEMPERATURE: A MICROSCOPIC VIEW 3

medium (where the chances of recombination are relatively low), lightning (natural
electric discharge between clouds), fluorescent lamp (low-pressure mercury vapor
gas discharge) etc.

Figure 1.2: Fraction of ionized atoms is plotted against thermal energy for Nitrogen
gas at standard conditions of temperate and pressure.

Due to the charged nature of plasma species, i.e. electrons and ions, they interact
via long range forces with each other and with other external electromagnetic fields.
Generally, they respond collectively as they are kind of bounded.

1.2 Temperature: A microscopic view

1.2.1 Velocity distribution function
In a macroscopic sense the temperature refers to the degree of hotness or coldness of
a body and is measured by a thermometer. However, microscopically temperature is
the average kinetic energy of particles in a system. Any gas, in a thermal equilibrium,
has particles that are moving with different speeds in different directions, i.e. with
different velocities, and is described by the equilibrium velocity distribution function
(VDF). In general the distribution functions not only depend on velocity, but also
on spatial and time coordinates. The VDF is a central entity in plasma kinetic
theory that will be discussed in later chapters.
In simplest terms, we can define the VDF, f = f (r, v, t), as a quantity whose
integration over the velocity gives the number density of the particles, i.e.
ˆ∞
n= f dv (1.5)
−∞

Now, the question is how do we get the distribution function? Where does it come
from? What equation does it satisfy? At this stage, we do not know anything about
4 CHAPTER 1. INTRODUCTION

these questions and will be answered in later chapters when we discuss the kinetic
theory of plasma. Here we will assume that VDF is given.
In thermal equilibrium, it turns out that, the most probable VDF is the so-called
Maxwellian distribution (having a Gaussian or a bell-curve shape), which for parti-
cles of mass m is given by
2 2
mv
− 2k − v2
f = Ae B T
= Ae vt
(1.6)
where r
2kB T
vt = , (1.7)
m
p 2
is the thermal speed, v = vx + vy2 + vz2 is the speed of particle in Cartesian coor-
dinates and A is the normalization constant.

Calculations of A
First, let us consider 1D situation and using the corresponding Eq. (1.6) in Eq.
(1.5) to write ˆ ∞ v2
− 2
n=A e vt dv (1.8)
−∞
Now comparing the above equation with the following standard Gaussian integral
ˆ ∞ r
−αx2 π
e dx = , (1.9)
−∞ α
We have α = 1/vt2 and hence √
n = A πvt (1.10)
which provides the normalization constant for 1D case as
n
A= √ (1.11)
πvt
Thus, the normalized 1D Maxwellian distribution function reads
2
n − v2
f (v) = √ e vt (1.12)
πvt

Averaging over the distribution function

The average value of any quantity Q can be calculated by using the relation
´∞
Qf (v)dv
hQi = ´−∞∞ (1.13)
−∞
f (v)dv
For example, consider any quantity which does not depend on the velocity, i.e.
Q = c,where c is a constant. Therefore, we have
´∞ ´∞
cf (v)dv c f (v)dv
hci = ´−∞
∞ = ´∞ −∞
=c (1.14)
−∞
f (v)dv −∞
f (v)dv
which is what we expected for a constant.
1.2. TEMPERATURE: A MICROSCOPIC VIEW 5

Temperature as the averaging kinetic energy

The average value of kinetic energy for a Maxwellian velocity distribution can be
calculated by using Eq.(1.12) and Q = 1/2mv 2 in Eq.(1.13) to write

´∞ 1
mv 2 f (v)dv
 
1 2
mv = ´∞
−∞ 2
(1.15)
2 −∞
f (v)dv
´∞ 1 n − 2 v2

−∞ 2
mv 2 √πvt
e vt dv
= (1.16)
ˆ n
∞ 2
m − v2
= √ v2e vt
dv (1.17)
2 πvt −∞
(1.18)

Now using the integral identity

ˆ ∞
r
2 −αx2 1 π
xe dx = (1.19)
−∞ 2 α3
with α = 1/vt2 , we have

  ˆ ∞ v2
1 2 m 2 − vt2
mv = √ v e dv
2 2 πvt −∞
m 1√ 3
= √ πvt
2 πvt 2
1 mvt2
 
= (1.20)
2 2
 
1 m2kB T /m
=
2 2
1
= kB T
2
Hence, the average kinetic energy of the particles having a Maxwellian velocity
distribution function is related with temperature of the system and is equal to
1
k T.
2 B

Three dimensional case

For the 3D situation we have

(vx2 + vy2 + vz2 )

−
f (v) = A3d × e (1.21)
vt2

Calculations of normalization constant

Using Eq.(1.21) in Eq.(1.5) to write
6 CHAPTER 1. INTRODUCTION

˚ 2 +v 2 +v 2
vx y z
−
vt2
n = A3d e dvx dvy dvz
ˆ ∞ 2
vx
ˆ ∞ vy2 ˆ ∞ vz2
− − −
vt2 vt2 vt2
= A3d dvx e dvy
e e dvz (1.22)
−∞ −∞ −∞
√ √ √
= A3d × πvt × πvt × πvt
= A3d π 3/2 vt3

which provides the normalization constant as

n
A3d = (1.23)
π 3/2 vt3
Thus, the corresponding 3D Maxwellian distribution function reads

(vx2 + vy2 + vz2 )

n −
f (v) =) = 3/2 3 e (1.24)
π vt vt2

Averaging kinetic energy for 3D case

The average value of kinetic energy for a 3D Maxwellian VDF can be calculated by
using Eq.(1.24) and Q = 12 mv 2 = 21 m(vx2 + vy2 + vz2 ) in Eq.(1.13) to write
˝ (v 2 +v 2 +v 2 )

1 2
 1
2
m(vx2 + vy2 + vz2 )e− x vy2 z dvx dvy dvz
˝
t
mv = (vx2 +vy2 +vz2 )
(1.25)
2 e− dvx dvy dvz
v2 t

Here we have three identical terms, one for each component. Consider the term for
x-component and evaluate the corresponding integral as follows

˝ 2 − (vx2 +vy2 +vz2 )

m
A3d
 vx e vt2
dvx dvy dvz
Ix = ˝ (v 2 +v 2 +v 2 )
2  A3d
 e− x y2 z dvx dvy dvz vt
´ − vx
2 ˜ (vy2 +vz2 )
m vx2 e vt2
dvx e− vt2
dvy dvz
= ˝ (vx2 +vy2 +vz2 )
2 e− dvx dvy dvz
vt2
´ 2 − vx
2 ˜ − (vy2 +v2
z)
 
m vx e vt2
dv x e  2  dvy dvz
=  vt
2 ´ 2 ˜ 2
− (vy+v2
z)

(1.26)
e− vvx2 dvx e

v2 dvy dvz
´ 2 − vx2 
t t

m vx e vt2 dvx
= ´ 2
2 e− vvx2 dvx
√ t3
m πv
= × √ t
2 2 πvt
1 mvt2 1
= × = kB T
2 2 2
1.3. DEBYE SHIELDING 7

Likewise, the other two integrals, in y and z, will also provide 12 kB T each. Thus, we
have    
1 2 1
mv =3 kB T (1.27)
2 3d 2
In general the average energy of the particles corresponding to a Maxwellian distri-
bution equals to 21 kB T per degree of freedom (equipartition principle). Since these
two quantities are closely related that, it is customary to give temperature the units
of energy, i.e. eV. If kB T = 1eV , then
1eV
T [eV ] = −5
= 1.16 × 104 K
8.617 × 10 eV /K

Comments on temperature
The temperature is an average kinetic energy related with particle’s random motion
and it is possible for a plasma to have several temperatures at the same time.
For example, if there is an applied magnetic field then the motion along the field
line (not affected by a uniform magnetic field) has entirely different behavior than
perpendicular to it, where particle is associated with a circular motion. Thus, the
plasma can be associated with two temperatures T⊥ and Tk . Moreover, due to the
mass difference between electrons and ions in a plasma, they can also be associated
with different temperatures, Ti and Te .

Comments about f (v)

Consider the 1D version of Maxwellian VDF
2
n − v2
f (v) = √ e vt (1.28)
πvt
n
which has a Gaussian shape having the amplitude given by f0 = √πv t
and width
1
∆ = vt . As vt ∝ m , and if we have two species such that mb = 10ma , then
accordingly a will have greater themal speed than b, consequently in the distribution
function f0a < f0b and the width ∆0a > ∆ba as shown in Fig. 1.3.
In the context of plasmas, this implies that the electrons - due to much lighter mass -
has larger thermal speed in comparison with ions (assuming the same temperatures
for both species), i.e. r
vte mi
=  1. (1.29)
vti me
Thus, in comparison with ions, the electronic distribution function is much smaller
in magnitude and is extended over large speeds”.

1.3 Debye shielding

A single (stationary) charge q interacts with any other test charge qt via Coulomb
force with corresponding potential φc (r) ∝ q/r, where r is the separation between
test charge and q. Which means that the test charge, even at very large distances,
8 CHAPTER 1. INTRODUCTION

Figure 1.3: Comparison of the Maxwellian distribution function for two species
having different masses.

can feel the force from the source charge. Thus, for an isolated charge it has an
infinite interaction distance. Now, let us put the same test charge in a plasma -
composed of negative electrons and positive ions - and calculate the corresponding
potential around it. As shown in Fig. 1.4, this positive charge attracts plasma
electrons and repel the positive ions. For a cold plasma, i.e. there are almost no
thermal motions, there would be as many negative charges in the cloud to generate
the charge of magnitude qt . And if you draw a Gaussian sphere around the electronic
cloud the net charge enclosed in it is zero, and by Gauss’s law there is no electric
field outside this cloud (a perfect shielding). However, there are always thermal
motions in the system and thus the particles at the edge (where field is weaker)
of such spherical cloud can escape the potential well, i.e. there is never a perfect
shielding. Next, let us solve for the potential distribution φ(r) around the test
charge in a plasma. For that one can use the Poisson’s equation along with proper
boundary conditions. These conditions are (i) far away from the cloud we expect
no interaction, i.e. φ(r → ∞) = 0 and let (ii) φ(r → 0) = φ0 .

1.3.1 Debye potential in a uniform positive background

Due to the mass difference between electrons and ions, i.e. me (= m)  mi (= M ),

we can assume that the ions do not move but only provide a uniform background.
The Poisson’s equation for a system composed of electrons and singly ionized positive
ions reads
1.3. DEBYE SHIELDING 9

Figure 1.4: Schematic diagram of a positive test charge in a plasma environment.

It attracts the electrons and repels the positive ions. The radius of electronic cloud
around test charge is defined as the Debye length.

ρ
∇2 φ = −
0
1
= − (ρi + ρe ) (1.30)
0
1
= − (eni − ene )
0
Here ρ = qn denotes the charge density, i.e. ρi = eni with e denoting the funda-
mental electron charge. The ions are heavy and their density is almost the same as
the corresponding value in undisturbed region, i.e. ni = n0 . Whereas electrons are
assumed to follow the Gibbs (or Boltzmann) distribution with VDF given by

fe = Ae−H/(kB T )
2 +qφ)/(k
= Ae−(0.5mv BT )

 (1.31)
2

= A eeφ/kB T e−mv /2kB T

 2

= eeφ/kB T Ae−mv /2kB T = eeφ/kB T fM ax (v)

Note that as φ → 0, we get the undisturbed Maxwellian VDF, i.e. fe → fmax . The
corresponding number density can be evaluated as

ˆ ∞
ne = f dv
−∞
ˆ ∞
=e eφ(r)/kB T
× Afmax dv = eeφ(r)/kB T × n0 (1.32)
−∞
eφ(r)/(kB T )
= n0 e
10 CHAPTER 1. INTRODUCTION

Now consider the small perturbation limit, namely eφ(r)  kB T , which means that
the argument of exponential function is much smaller than unity and we can write
a Taylor series expansion in the form

ne = n0 eeφ(r)/(kB T )
"  2 #
eφ 1 eφ
= n0 1 + + + ....
kB T 2! kB T (1.33)
 
eφ
≈ n0 1 +
kB T

Here we have ignored the second and higher order terms, upon using Eq. (1.33) in
Eq. (1.30) we find

 
2 e eφ
∇ φ=− n0 − (n0 + n0 )
0 kB T
n0 e2
= φ (1.34)
0 kB T
1
= 2 φ,
λD

where

r
0 kB T
λD ≡ (1.35)
n0 e 2

is defined as the Debye length (after Peter Debye, 1884-1966), and can be simplified
as

r r
0 kB T
λD [m] = ×
e2 n0
s r
8.85 × 10−12 1.38 × 10−23 T
= × (1.36)
(1.6 × 10−19 )2 n0
s
T [K o ]
= 69
n0 [m−3 ]

Or, we can equivalently write by using the relation 1eV = 1.6 × 10−19 J as
1.3. DEBYE SHIELDING 11

s
kB T [J]
λD [m] = 1.859 × 1013
n0 [m−3 ]
s
1.6 × 10−19 T [eV ]
= 1.859 × 1013
n0 [m−3 ]
s (1.37)
T [eV ]
= 7436
n0 [m−3 ]
s
T [eV ]
≈ 7.43 × 103
n0 [m−3 ]
For example consider some plasma systems and corresponding Debye lengths:

(i) For typical fusion (a process in which small nuclei fuse to form heavy nuclei
along with lot of energy) reactor with minor radius a ∼ 1m: n = 1021 m−3 and
very high temperature around T = 10keV . The corresponding Debye length is
2.3 × 10−5 m, thus λD  a.

(ii) For ionosphere (a shell of plasma that surrounds the earth, stretching from
a height of about 50 to 1000 km) with n = 1011 m−3 and T = 0.05eV . The Debye
length for this system is 5.2 × 10−3 m. Note that λD is much smaller than the di-
mensions of the ionosphere.

(iii) For a glow discharge (plasma formed by the passage of electric current through
a gas): n = 1016 m−3 and temperatures T = 2eV , which implies λD = 10−4 m.

Note that typical separation between the atoms scales as ∆ ∝ 1/(n)1/3 and if
n ≈ 1016 m−3 , we find ∆ = 10−6 , i.e. ∆ < λD . Moreover, Bohr radius is about
10−10 m and hence no quantum mechanical treatment is required to deal with such
plasmas.

Solution of Poisson’s equation in spherical coordinates

We expect the cloud around the test charge to be spherical symmetric, i.e. inde-
pendent of the direction. Which, in spherical coordinates means that potential is
independent of the angular coordinates (θ, ϕ). Thus, the Poisson equation (1.34)
can be written as

1
∇2 φ = φ
λ2D
 
1 d 2 dφ 1
r = φ (1.38)
r2 dr dr λ2D
d2 φ 2 dφ 1
=⇒ + = φ
dr2 r dr λ2D
12 CHAPTER 1. INTRODUCTION

Assuming a general solution of the form

er/λD e−r/λD
φ(r) = A +B (r 6= 0) (1.39)
r r
However, from the boundary condition, φ(r → ∞) = 0, we must have A = 0,
moreover φ(r → 0) = φ0 and hence we obtain the for potential distribution as

e−r/λD
φ(r) = φ0 (1.40)
r
which is the required solution of Poisson’s equation and can be shown by substituting
above equation in Eq. (1.38) to write

e−r/λD
φ(r) = φ0
r 
dφ(r) φ0 r
= 2 −1 − e−r/λD
dr r λD
 
2 dφ(r) r
r = φ0 −1 − e−r/λD
dr λD
     (1.41)
d 2 dφ(r) 1 1 r
r = φ0 − − −1 − e−r/λD
dr dr λD λD λD
 
d 2 dφ(r) r
r = φ0 2 e−r/λD
dr dr λD
 
1 d dφ(r) φ0 1
2
r2 = 2 e−r/λD = 2 φ (QED)
r dr dr rλD λD

Notice that we have ignored the contribution of test charge in the charge density term
of Poisson’s equation. If we include that, there will be an additional ρt = qt δ(r) in
Eqs.(1.30) and (1.38), however as our calculations (in spherical coordinates) exclude
r = 0 contribution, thus this term does not contribute in the final potential relations
(see problem 1.11 for details).

Analysis of Debye potential

The potential around a test charge qt in plasma is given by Eq. (1.40) and is called
the Deby–Huckel (or simply Debye) potential, which is quite different than the usual
Coulomb’s potential of a point charge. The said relation is also used to define the
radius of electronic cloud around qt , i.e. the so-called Debye length
r
0 kB T
λ2D ≡ (1.42)
n0 e 2

• If there is no plasma, i.e. n0 → 0, or equivalently λD → ∞, the exponential

factor provides 1. Thus, the potential becomes Coulombic as we expect for an
isolated charged particle.
1.3. DEBYE SHIELDING 13

Figure 1.5: Comparison of Debye and Coulomb potential distributions. Here for the
purpose of demonstration, we have chosen λD = 2.

• The Debye potential depicts an exponential decay which scales with Debye
length, i.e. at a distance r = λD the potential drops by a factor of 1/e of its
corresponding Coulomb value. And after tens of λ0D s we expect φ → 0. In
Fig. (1.5) we compare the Coulombic and Debye potentials.

• If temperature of the plasma goes up, which implies that λD will also increase,
i.e. the spherical cloud will get bigger. Physically this make sense because of
the random motion particles can not be confined to a small spherical cloud.

1.3.2 A general case of non-uniform positive background

In the previous section we considered that ions provide a uniform background to
electrons, which follow the Gibbs distribution. However, in a more general steady
state case, both ions and electrons will follow the Boltzmann relation and thus Debye
length is expected to be modified. In this case the ion number density - in the weak
perturbation limit - is given by

ni = n0 e−eφ(r)/(kB Ti )
"  2 #
eφ 1 eφ
= n0 1 − + + ....
kB Ti 2! kB Ti (1.43)
 
eφ
≈ n0 1 −
kB Ti

Accordingly, the Poisson’s equation reads

14 CHAPTER 1. INTRODUCTION

  
2 e eφ eφ
∇ φ=− n0 − n0 − (n0 + n0 )
0 kB Ti kB Te
n0 e2
 
1 1
= + φ (1.44)
0 kB Ti kB Te
1
= 2 φ,
λD
where the modified Debye length is given by
−1/2
n0 e2
 
1 1
λD = + (1.45)
0 kB Ti kB Te
For the case if Ti  Te , then 1/Ti → 0 and ions (which in this case are hotter
species) do not contribute and thus we have only temperature of colder (electrons in
this case) in Debye length and vice versa. Thus, Debye length is determined mainly
by the temperature of colder species.

More generally there are also other charged particles in the system, e.g. in addition
to electrons and ions there may be some dust charged particles - such systems are
called dusty or complex plasmas. Then for a multi-component plasma consisting of
species of j kind the corresponding Debye length is given by

X  nj0 qj2  1 −1/2

λD = (1.46)
j
0 kB Tj
Or, equivalently we can write
1 X 1
= (1.47)
λ2D j
λ2Dj

1.3.3 Number of particles in the Debye sphere

While deriving an expression for Debye-Huckel potential we have considered a sta-
tistical concept, i.e. Gibbs distributions. All the statistics is valid only if you are
talking about large number of particles. Thus, the Debye sphere must have large
number of particles for the shielding to be a meaningful concept statistically. If the
plasma density is n then the number of particles in Debye sphere ND - also called
the plasma parameter- is given by

ND = n × Volume of Debye sphere

4
= n × πλ3D
3
 3/2
4πn 3 T [K] (1.48)
= × (69)
3 n
3/2
T [K]
= 1.38 × 106 1/2 −3
n [m ]
1.4. CONCEPT OF QUASI-NEUTRALITY 15

For a collective behavior of the system ND  1. All these particles inside the cloud
experiences the force of a test charge such that they shield out the region outside
λD where the potential decreases by almost 63% of φ0 .

1.4 Concept of quasi-neutrality

A plasma is always considered to be quasi-neutral system, which (for an electron-ion
plasma) means that number density of positive and negative charges must be the
same, i.e. ni ≈ ne . Now we need to show that why is it so? Let us start with the
Gauss’s law to write

~ = ρ
∇·E
0
e (1.49)
= (ni − ne )
0
For a slab (Cartesian) geometry let us approximate (estimate) divergence of electric
~ = dE/dx ∼ E/x, thus we find
field as ∇ · E
~ = e (ni − ne )
∇·E
0
ex (1.50)
=⇒ E = (ni − ne )
0
The corresponding electric force on a charge q in this field is given by
qex
F = qE = (ni − ne ) (1.51)
0
(i) If ni > ne , it would mean that E is positive, which will push positive charges
away and thus resulting in a decrease of such excess particles.
(ii) On the other hand if electrons are in excess, i.e. ne > ni , it would mean that E
is negative, which will push electrons away and thus resulting in a decrease of ne .

Therefore, this force will always tend to neutralize any charge imbalance between
positive and negative species, till ni ≈ ne and the system achieves an equilibrium
state (F = 0). Let us calculate the strength of this force density (N/m3 ) and
compare it with plasma kinetic pressure. For that consider a plasma with Te = 1eV ,
ne = 1019 m−3 and assuming ni 6= ne . That also means ∆n = ni − ne 6= 0, hence we
can write the force per unit volume as

F1 = ρE
ex∆n
= e∆n × (1.52)
0
x
= (e∆n)2
0
We can estimate F1 at x = 0.1m and by assuming only a slight deviation in densities
by taking ∆n/ne = 1% = 0.01, i.e. ∆n = 1017 . The resultant force density turns
16 CHAPTER 1. INTRODUCTION

out to be 3 × 106 N/m3 . Whereas the corresponding pressure force can be estimated
from Fp = p/x = [nkB (Ti + Te ]/x) ≈ (2ne kB T )/x = 16N/m3 . Hence the force
associated with an imbalance of charge species is much greater than the plasma
kinetic pressure, and enforces the charge neutrality in the plasma systems.

1.5 Definition and criteria of a plasma

Plasma is defined as the fourth state of matter consisting of a quasi-neutral gas -
of charged and neutral particles - which shows a characteristic collective behavior.
Hence quasi-neutrality and collective behavior are the two main characteristics which
distinguish the said state from an ionized gas. By former we mean that the ion
density is approximately equal to electron density, i.e. ni ≈ ne , however not much
neutral that the interesting electromagnetic effects vanish. Collective behavior refers
to the fact that long range electromagnetic interactions among charged particles are
dominant over the short range collisional interactions.
For any ionized gas to be considered a plasma the following three conditions must
be satisfied.
1. Let the shortest dimension of the system is L, then it must be much greater
than the Debye length, i.e.
L  λD (1.53)

2. The number of particles inside the Debye sphere must be large enough for the
Debye shielding to be statistically valid, thus we must have
ND  1 (1.54)

3. The frequency of plasma oscillations ω must be greater than the collision

frequency with neutrals ν, namely
ω>ν (1.55)
Here the plasma oscillations are associated with an electric field that is pro-
duced due to any charge imbalance in the plasma. Thus, this condition, in
other words, demands that the dynamics of the system must be governed by
electric (and/or magnetic) fields rather than by ordinary hydrodynamics.
The word plasma emerged from the Greek word, which means something molded.
The plasma state can be achieved when a neutral gas is heated and a major fraction
of atoms are dissociated into positive and negative charged carriers in the form of
an ionized gas. Every charged particle would interact simultaneously with all the
other ions through self-generated electric and magnetic fields that lead to a collective
behavior in plasmas.

1.6 Some applications of plasmas

(i) Gas discharges: Back in 1920’s Langmuir and Tonks were interested in running
a very high current through vacuum tubes. For that they need to apply more and
1.6. SOME APPLICATIONS OF PLASMAS 17

more potentials, which resulted in the ionization of inside gases and a plasma is
formed. These were considered to be weakly ionized plasmas with typical parame-
ters Te = 2eV and ne = 1014 − 1018 /m3 . The other examples in this category include
fluorescent light bulb, Neon lights, Mercury lamps, welding arcs etc. Such devices
are also used in various plasma processing, e.g. they are found next to some solids
for plasma deposition.

In the year 1927 Irving Langmuir (1881-1957) investigated a low-pressure Mercury

discharge and observed, in a long glass tube, a relatively cold luminous gaseous body
with homogeneous electrical and optical properties which was completely neutral to
the exterior. Nevertheless it was clear that this body contained free negative and
positive charge carriers, but equal numbers of them per unit volume. He knew well
that already in 1870 the British scientist Sir William Crooke (1832-1919) had rec-
ognized that a thin gas changes its character and attains a completely new state
when an electric current is passed through it. Langmuir was an inventive man and
searched a new name for this state of matter and he coined the name plasma for
a fully ionized, electrically neutral gas. He is therefore considered as the actual
founder of plasma physics and the most famous and outstanding pioneer of this
section of physics. As we will see later, Langmuir’s name is connected also to other
terms in plasma physics such as Langmuir waves and the Langmuir probe.

(ii) Space physics: Around 1930’s, when scientists were into radio waves, namely
the ground propagation of these waves. When they observed that if you propagate
waves up to the atmosphere they came back at some place else. Which give a hint
that they can communicate at long distances. This is due to the fact that there is an
ionized layer the so-called ionosphere, which is actually a plasma and can reflect back
the waves. In fact there are many more associated effects in this region. The typi-
cal parameters of such plasmas are already discussed in the context of Debye sphere.

(iii) Controlled fusion: Our sun (also known as the best fusion device) emits
charged particles, a group of which is also directed towards the earth, this stream
of ions is known as the solar wind having Te = 10eV , ne = 5 × 106 /m3 and typical
speed of 300km/s. The magnetosphere is the region of space surrounding earth
where the dominant magnetic field is the magnetic field of earth and is formed by
the interaction of the solar wind. The magnetic field of earth can deflect as well as
trapped such particles.

The plasma physics got a boost when it was proposed that fusion reactions (resulting
in a H-bomb) can be controlled to make a reactor. The said reactions involve fusion
of light nuclei (like nuclei of Deuterium and Tritium) producing a heavy counterpart
and lot of energy is released. These reactions take place in a plasma environment
and is the subject of major plasma community even today. A major portion of this
book is devoted to nuclear fusion, hence we postpone the discussion on this subject
to latter chapters.
18 CHAPTER 1. INTRODUCTION

1.7 Potential around a moving charge (optional)

While deriving the Debye-Huckel potential we have assumed that the test charge
is at rest. However, generally the charge particles enter the plasma with a cer-
tain speed and thus pass through it with a finite speed vtest . In this case we have
ρt = qt δ(r − tvtest ), which has to be included in the Poisson’s equation. The result-
ing potential distribution around this charge strongly depends on its speed. The
derivation of such potentials is beyond the scope of this book, however we briefly
discuss some of the scenarios that are found is such systems.

(i) For Maxwellian system with vtest = 0, the potential is referred as a shielded
potential, the test charge acquires a shielding spherical cloud, which is made up of
an excess of plasma specie of opposite charge and a deficiency of the like charged
specie, is known as Debye screening. The uniform plasma distribution deviates due
to presence of test charge which produces an extra potential field and thus a new
effective potential is constructed by adding these two.

(ii) For test particle moving at intermediate speed, i.e. vti < vtest < vte , where vti
and vte are the thermal speeds of ions and electrons, respectively. The shielding is
predominately done by the lighter species, namely electrons.

(iii) If test charge particles is moving very fast, whose speed is greater than the
electron thermal speed, i.e. vtest  vte . In this case it turns out that there is no
shielding at all, which is due to the fact that plasma particles do not get enough
time to response.

(iv) A slowly moving charge - having its speed less than ion thermal speed - is
shielded in such a way that electrons and ions equally participate in the shielding
process. More precisely it can be shielded by the cloud of opposite charges to form a
spherically symmetric short-range Debye-Huckel (DH) potential and/or long-range
far-field (FF) potential that falls as inverse third power of the distance ( 1/r3 ). On
the other hand, wake-field (WF) potential oscillates behind the test charge when its
speed resonating with the phase speed of a plasma mode. The oscillatory nature of
WF potential can be thought of as duck (test charge) moving through stationary
water (plasma) and the oscillatory water ripples (wake field) behind it. Obviously,
the greater the mass of test charge is, the greater effects can be seen behind it. That’s
why such WFs are quite often associated with heavy dust particles in plasmas, i.e.
in dusty plasmas.

1.8 Problems
P(1.1) Starting with a 2D Maxwellian distribution function verify the equipartition
theorem by calculate the average kinetic energy of the system.
1.8. PROBLEMS 19

P(1.2) Consider the following VDF

1 2 /v 2
f= e−v t (1.56)
π 3/2 vt3
(a) Show that it is normalized to unity.
(b) Using spherical coordinates show that average speed is given by
r
8kB T
hvi = (1.57)
πm
(c) Also derive the following

2  3kB T
v = (1.58)
m
P(1.3) Consider two plasma systems A and B. Calculate the ratio between their
Debye lengths, i.e. λA /λB if
(a) A is 10 times hotter than B, however B is 10 times denser.
(Ans: 10)
(b) A is 10 times hotter and 10 times denser than B.
(Ans: 1)

P(1.4) The electronic temperature inside a fluorescent light bulb is about 20 thou-
sands Kelvin. Why it does not feel that hot when you touch it?

P(1.5) Give a few examples of ionized gases which do not satisfy the plasma criteria.

P(1.6) When a particle (charged or neutral) passes through a neutral gas, it follows
Brownian motion, i.e. straight lines between random collisions. In plasma where
all the charged particles interact with this particles, what kind of trajectory do we
expect for a positive ions moving through it?

P(1.7) (i) Considering the following definition of Debye length, derive the relation
for plasma frequency ωp . r
kB Te −1
λD = ω (1.59)
m p
(Ans: ωp = ne2 /0 m)
(ii) Compute the plasma frequency and the Debye length for the following systems.

(a) Earth’s ionosphere with electron concentration of 106 cm−3 and electron tem-
perature of 0.2 eV.
(Ans: 5.6 × 107 rad/s, 3.3mm.)
(b) A cell of a typical plasma display with electron number density 1013 cm−3 and
Te = 0.2eV . Is the condition that the system dimension should be much greater
than the Debye length fulfilled?
20 CHAPTER 1. INTRODUCTION

(Ans: 1.78 × 1011 rad/s, 1.05µm)

(c) A welding arc with electron concentration ne = 1.6 × 1017 cm−3 and Te = 1.3eV .
(Ans: 2.3 × 1013 rad/s, 21nm)
(iii) Using the above definition and the plasma criterion λD  L to √ show that we
can write an alternate criterion in the form ωp  ω1 , where ω1 = vt /( 2L).

P(1.8) If we assume same temperatures for electrons and ions. Show that the Debye
length is independent of mass and is therefore comparable for different species.

P(1.9) Consider an ionized gas having following parameters n = 1016 m−3 and T =
2eV in a cubic box of length 0.01cm. If collision frequency is 106 rad/s. Can this
gas be categorized as a plasma?

P(1.10) A plasma with n = 1016 m−3 and T = 2eV is in a spherical box of radius
0.01m. Calculate the maximum number of Debye spheres that can exist in this
system.

P(1.11) The Debye potential at a distance r from a point charge q inserted in a

plasma is given by
q
φ(r) = e−r/λD (1.60)
4π0 r
show that the corresponding charge density is
 
1
ρ(r) = q δ(r) − e−r/λD (1.61)
4πrλ2D

Hints: First calculate the electric field using E ~ = −∇φ = −r̂ dφ/dr. Next, use the
Maxwell’s equation ∇ · E ~ = ρ/0 to find the charge density. You may need to use
the following vector identities
 
~r
∇· = 4πδ(r) (1.62)
r3
 
~r 1
∇· 2
= 2 (1.63)
r r
P(1.12) Write down the quasi neutrality conditions for the following plasmas:
(a) A dusty plasma composed of electrons, ions and negatively charge dust grains
having an atomic number Zd .
(b) An electron-positron ion plasma having an additional component, namely positron.

(c) Pair-ion-electron plasmas, whose constituents are positive and negative ions X ±
and electrons.
Chapter 2

Plasma orbit theory

As a first step to investigate the behavior of a collection of charged particle, it is

useful to develop some ideas about the motions of individual charged particles in
electromagnetic fields and the topic is referred to as orbit theory of plasma. In
this chapter we will review a single charged particle motion in various electric and
magnetic field configurations. Solving the equation of motion to determine the
trajectory of a single ion in various field configurations may not seem to be the
best way to deal with physics of plasmas, which is composed of many interacting
particles. Thus, it is worth asking a question why to bother with orbit theory of a
single charged particle? One attraction is that it is relatively easy to handle, beyond
that, the key concepts in orbit theory prove to be useful throughout plasma physics.
It is to be considered as a first step in learning how to deal with the complicated
multi-component plasma systems, where collective interactions play a central role.
In the single particle description it is assumed that the electric and magnetic fields
are not affected by the charged particle motion, i.e. microscopic field variations due
to an individual particle are not considered.

Guiding center approximation

Before we go deep in it, we need to be clear about the conditions under which
the particle orbit theory might be valid. We expect it to be useful in describing
the motion of highly energetic particles in low density plasmas, where particles
collisions are infrequent. While moving charges accordingly generate electromagnetic
fields which additionally act on other moving particles, for the case of low densities,
these induced fields – as well as the effect of collisions – may be neglected so that
the trajectories of charged particles may be considered to be governed entirely by
external field forces. In many situations of practical interest, the complex space and
time dependencies of E and B make it almost impossible to integrate the equation
of motion exactly. To ease the computational procedure, the motion of a charged
particle is averaged over an integer number of full revolutions around a magnetic
field line, thus, leading to the motion of the center of gyration also called the guiding
center (GC). The motion of the GC can be decomposed into traveling k and ⊥ to
B. The latter part is a result of different kinds of drifts across the field lines, arising
from curvature and non-uniformity of B, interaction with E, and in total constitutes
a slow drift of the GC across magnetic field. This method is valid for fields slowly

21
22 CHAPTER 2. PLASMA ORBIT THEORY

varying in space and time, i.e. when the gyro (Larmor) radius rL is much less than
the scale length of any field in-homogeneity and when the gyro (cyclotron) frequency
ωc is much greater than any characteristic field frequency. Alternatively we can state
that if the changes over a gyro-period are small, then it is convenient to average over
gyro motion and explain the motion of particle by the drift of its GC.
The basic equation describing motion of a non-relativistic charged particle j of mass
mj and charge qj in an electric field E and magnetic field B is
dvj
mj = qj (E + vj × B) + Fext
j (2.1)
dt
Here Fext
j represents any additional external force acting on the particle; for instance,
in astrophysical plasmas the gravitational forces may be significant and then are to
be accounted for in the equation of motion. In the following we consider, in the
order of increasing complexity, various possible configurations.

2.1 Uniform force fields

2.1.1 ~
Unmagnetized case with homogeneous E
In this case the equation of motion reads (here we drop the subscript j for symbolic
convenience)
d~v ~
m = qE
dt
d~v q ~
= E
ˆ v dt m ˆ
q ~ t
d~v = E dt (2.2)
v0 m 0

qE~
=⇒ ~v = v~0 + t (vf = vi + at)
m
d~x qE~
= v0 + t
dt m
Integrating again in the limit t = 0 → t and x = x0 → x, we find
~
1 qE
~x = x~0 + v~0 t + t2 (s = s0 + vi t + 0.5at2 ) (2.3)
2m
Thus, we get a uniform acceleration ~a = (q/m) E ~ of the charged particle along the
electric field. However, the particles are never accelerated to infinite speeds, but
rather limited by relativistic effects. If electric field is along z-axis then we will
get an accelerated motion along z-axis only, i.e. ax = ay = 0, az = qEz /m and
consequently vx = vx0 , vy = vy0 , vz = vz0 + qEz t/m. The important feature to note
is that the components of motion for this individual charged particle perpendicular
to the E-field do not change with time, however, the velocity component in the
direction of the E-field linearly varies with time. The particle is accelerated in the
direction of E for a positive charge and in the opposite direction for an electron, the
corresponding trajectory is depicted in Fig. 2.1.
2.1. UNIFORM FORCE FIELDS 23

Figure 2.1: Trajectory of a positively charged ion in a homogeneous electric field

[Harms 2000].

2.1.2 ~ =0
Homogeneous magnetic field with E
~ = B ẑ, in this
Let us assume the motion of a particle in a uniform magnetic field B
case the relevant force equation becomes

d~v ~
m = q~v × B
dt
d~v q ~
= ~v × B
dt m
d q
(vx x̂ + vy ŷ + vz ẑ) = (vy B x̂ − vx B ŷ) (2.4)
dt m
d qB
(vx x̂ + vy ŷ + vz ẑ) = (vy x̂ − vx ŷ)
dt m
d
(vx x̂ + vy ŷ + vz ẑ) = ωc (vy x̂ − vx ŷ) ,
dt
where
|q|B
ωc = (2.5)
m
is the cyclotron (or gyro) frequency. Thus, we can write various components of
velocity vector as following

v̇x = ωc vy =⇒ v̈x = ωc v̇y , (2.6)

24 CHAPTER 2. PLASMA ORBIT THEORY

v̇y = −ωc vx =⇒ v̈y = −ωc v̇x (2.7)

and
v˙z = 0 =⇒ vz = constant.
(2.8)
vz (t) = vz (0) ≡ vk
Hence the velocity along the field line remains un-affected. The corresponding z-
displacement is then given by
z(t) = z0 + vk t, (2.9)
whereas the perpendicular components in Eqs. (2.6) and (2.7) can be written in a
compact form as
v̈x,y = −ωc2 vx,y (2.10)
which is the equation of a harmonic oscillator, having solution of the form
vx,y = v⊥ e±iωc t , (2.11)
q
2 2
where ± sign stands for ions and electrons and v⊥ = vx0 + vy0 . Now let us consider
the x-component and integrating - for a mathematical convenience we choose time
integral limit −∞ → 0 - it to write
vx = ẋ = v⊥ eiωc t
ˆ
x ˆ t
dx = v⊥ eiωc t dt
x0
−∞ (2.12)
eiωc t

=⇒ x − x0 = v⊥ −0
iωc
x − x0 = −irL [Cos(ωc t) + iSin(ωc t)]
with the so-called Larmor (or gyro) radius is defined by
v⊥
rL = (2.13)
ωc
Comparing the real parts yields
x − x0 = rL Sin(ωc t) (2.14)
For the y-component, from Eq.(2.6), we have
1
vy = v̇x
ωc
1
ẏ = × iωc v⊥ eiωc t
ωc
ẏ = iv⊥ eiωc t
ˆ y ˆ t (2.15)
dy = iv⊥ eiωc t dt
y0 −∞
 iωc t 
e
=⇒ y − y0 = iv⊥
iωc
y − y0 = rL [Cos(ωc t) + iSin(ωc t)]
2.1. UNIFORM FORCE FIELDS 25

Figure 2.2: Trajectory of an isolated charged positive particle in a homogeneous

magnetic field [Harms 2000].

upon equating the real parts we find

y − y0 = rL Cos(ωc t) (2.16)

Squaring and adding Eqs. (2.14) and (2.17) yields

(x − x0 )2 + (y − y0 )2 = rL2 (2.17)

which is an equation of circle with radius rL and center lying at (x0 , y0 ). Thus,
particles in this configuration have a constant velocity along the field line and time
varying circular projection in perpendicular x − y plane, i.e. a helical path along
the field line is formed as shown in Fig. 2.2, where pitch of this helical motion is
denoted by ∆z, i.e. the distance between two consecutive wingdings and can be
calculated as

2πrL
∆z = z(t + T ) − z(t), T =
v⊥
   
2πrL

= z0 + vk t + − z0 + vk t
v⊥
(2.18)
2πrL vk
=
v⊥
2πvk
=
ωc
Therefor, an isolated charge particles in a constant, homogeneous magnetic field
moves on a helical path of constant radius, frequency and pitch.
26 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.3: Trajectory of electrons and ions in a uniform magnetic field, having
different gyro radii and cyclotron frequencies.

Another important point to note here is that averaging over one full gyration or any
integer number of gyro periods - using the fact that hSin(ωc t)i = hCos(ωc t)i = 0 -
will provide
hx − xgc i = 0, hvx i = 0, (From Eq. 2.14)
hy − ygc i = 0, hvy i = 0, (From Eq. 2.16) (2.19)
=⇒ h~v i = ~vgc = vk k̂
which means that guiding center moves only along the field line. In the cases, when
there are additional forces in the system the guiding center also drifts away in the
perpendicular direction, which is called the guiding center drift motion and will be
discussed in the subsequent sections.

Comparison of ion and electron trajectories

As the cyclotron frequency is inversely proportional to the mass, and me  mi ,
hence ωce  ωci . Whereas the Larmor radius has exactly opposite trend, i.e. re  ri .
Thus, during their helical motion a heavy ion will have a larger radius than a lighter
electrons at the same v⊥ , whereas an electron will possess a much higher angular
frequency than the heavy ion as shown in Fig. 2.3. Also the cyclotron frequency is
proportional to the strength of magnetic field, which is physically obvious, stronger
the field is stronger is the corresponding force and resulting in higher values of ωc .

Plasmas are diamagnetic

Figure 2.4 depicts the circular trajectory of plasma species, i.e. ion and electrons
in the x − y plane under the action of a uniform magnetic field directed along z
2.1. UNIFORM FORCE FIELDS 27

Figure 2.4: Projection of ion and electron motion on the xy plane to highlight the
diamagnetic nature of a plasma [Harms 2000].

direction. It turns out that plasmas are diamagnetic in nature, which means that
the direction of gyration will always be such that the self-generated B-field points in
the opposite direction to the external magnetic field. A useful rule of thumb is that
with B pointing into the plane of this page, an ion circles counter-clockwise while
electrons move clockwise.

2.1.3 Combine electric and magnetic fields

If electric field is directed along the magnetic field then equation of motion reads

mv̇k = qEk

which means that charge particles will have a constant acceleration parallel/anti-
parallel to the magnetic field. That will cause positive and negative charged species
to move in opposite direction, as a result of which there is a generation of internal
electric field which is in opposite direction to the external counterpart. Thus, it
causes a very rapid cancellation of large-scale parallel electric field in plasmas.

Consider the situation as depicted in Fig. 2.5, where the magnetic field is directed
along z-axis and the electric field is chosen to have both components, i.e. perpendic-
ular (x-directed) as well as parallel (z-directed). We could have chosen y-component
as well, however due to symmetry of the system and mathematical convenience we
adopted for only one perpendicular component. As stated earlier, if there is any
force (other than the one due to inform magnetic field) we expect a drift of the GC
in a perpendicular (to B) direction, here the force (F~ = q E)~ can produce a drift.
28 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.5: Orientation of uniform electric and magnetic fields [Harms 2000].

Equation of motion for this configuration reads

d~v 
~ + ~v × B
~

m =q E
dt
d~v q ~ ~

= E + ~v × B
dt m (2.20)
q
v̇x x̂ + v̇y ŷ + v̇z ẑ = [Ex x̂ + Ez ẑ + vy B x̂ − vx B ŷ]
m
q
= [(Ex + vy B)x̂ − vx B ŷ + Ez ẑ]
m

Upon equating the coefficients of x̂, ŷ and ẑ, we find

q q
v̇x = [(Ex + vy B)] = Ex ± ωc vy , (2.21)
m m

v̇y = ∓ωc vx (2.22)

and, respectively

qEz qEz qEz

v̇z = =⇒ vz = t + vz0 = vk + t
m m m (2.23)
1
z = z0 + vk t + az t2
2

Equations (2.21) and (2.22) are coupled, which can be decoupled by taking the time
2.1. UNIFORM FORCE FIELDS 29

derivatives as follows.

v̈x = ±ωc v̇y = −ωc2 vx ,

v̈y = ∓ωc v̇x
q 
= ∓ωc Ex ± ωc vy
m  (2.24)
2 Ex
= −ωc + vy
B
d2 d2 Ex
     
Ex 2 Ex
vy + = −ωc vy + ∵ 2 =0
dt2 B B dt B

Above equations are exactly the same as Eqs. (2.10) except for the factor Ex /B in
y-component, thus we can write the solution as

vx = v⊥ eiωc t
Ex (2.25)
vy = ±iv⊥ eiωc t −
B
Here the first terms indicate the usual Larmor gyrations (if there is no electric field),
whereas the second term denotes the superimposed GC drift in negative y-direction
(for Ex > 0) and we can write

Ex ~ ×B
E ~
~vgc = ~vE = − = (2.26)
B B2
Hence, the guiding center’s drift is in the direction perpendicular to both the B
and E-fields and, in contrast to the preceding case, ion and electron motion across
the field lines now occur. Additionally, the pitch increases with time as seen in
Eq.(2.23), hence the trajectory resembles a slanted helix with increasing pitch.
Furthermore, we note that E ~ ×B ~ velocity is independent of the charge and mass
of the species, i.e. both electrons and ions will drift exactly in the same direction
as shown in Fig. 2.6. Physically: during the gyro motion a charge particle gains
energy (from E) in first half cycle and thus v⊥ and thus the Larmor radius increase.
Whereas in the next half cycle rL decreases, this difference in gyro radius on left
and right hand side of the orbit causes a drift of the guiding center.

Drift velocity for a force F~

Now we consider a generalization of the drift velocity as caused by an arbitrary force
F~ on a charged q moving in a uniform B-field, for that consider

~ ×B
E ~ q F~E × B~
~vE = × = (2.27)
B2 q qB 2

and for any force we can write

F~ × B~
~vF = (2.28)
qB 2
30 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.6: Ion and electron drift in a combined uniform magnetic and electric fields
[Harms 2000].

Figure 2.7: A positive test charge motion as subjected to a uniform magnetic field
and a perpendicular force F~ , clearly demonstrating the cross field GC drift.
2.2. NON-UNIFORM MAGNETIC FIELDS 31

~ drift, we can also derive the above expression

Instead of deriving it indirectly from E
for a general case of a uniform F~ (⊥ B) ~ acting on a positive charge q. For that
consider the geometry in Fig. 2.7, for which the equation of motion reads

d~v  
~ + F~
m = q ~v × B (2.29)
dt
Note that at the bottom of its path, i.e. at point P1 both the forces act in upward
direction and resulting in an enhanced normal acceleration - which in turn - makes
the trajectory more sharply bent than it would have for F~ = 0 case. Whereas,
when the particle is at point P2 the Lorentz force and F~ are in opposite direction,
thus effect of Lorentz force is diluted and thereby causing the trajectory to be less
sharply bent as depicted in the same figure. This difference causes a GC drift in a
direction perpendicular to both force fields.

We can write the perpendicular component of velocity as a superposition ~v = ~vc +~vgc ,

where the former corresponds to circular motion, whereas the later is associated
with GC drift motion and is expected to be a uniform, i.e. ~v̇gc = 0. Thus the
corresponding force equation reads

 
~ + F~ = 0
q ~vgc × B
 
~
q ~vgc × B × B ~ + F~ × B
~ =0
(2.30)
F~ × B~
~ = 0)
=⇒ ~vgc = (∵ ~vgc · B
qB 2
Hence - due to charge dependence in above relation - we expect positive and negative
particles to drift in opposite directions if the force does not depend on the charge
(as in the case of a gravitational force see Problem 2.4).

2.2 Non-uniform magnetic fields

In all the preceding cases the magnetic field was taken to be uniform throughout the
space of interest. Such an idealization is, in practice, impossible although it can be
approached to some degree by using sufficiently large magnet or coil arrangements.
Now we consider the motion of plasma particles in a spatially varying magnetic
configurations. The associated equation of motion is then written as
d~v h
~
i
m = q ~v × B(r) (2.31)
dt
In this regard new drifts are expected to appear, however the things become compli-
cated mathematically for which various approximations are employed. In particular,
one expands different equations in small parameter (rL /L), where L denotes the scale
length of in-homogeneity.

In the subsequent chapters, we will discover that fusion devices which utilize mag-
netic fields for confinement of the reacting ions involve very complex magnetic field
32 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.8: Illustration of a spatially varying magnetic field B directed along z-axis
and having a gradient in y direction [Harms 2000].

topologies. Not only are there electric fields and spatially varying magnetic fields
present - thus giving rise to the drift velocities as we just discussed - but there are
additional complexities due to curvature in the magnetic field lines and gradients
along these field lines. We thus now formulate expressions for the effective forces
and resulting drift velocities due to such in-homogeneities in the B-fields.

2.2.1 ~ drift
Grad-B (⊥ to B)
We consider a dominant magnetic field in the z-direction characterized by an in-
creasing strength in the y-direction, that is we have a constant non-zero gradient of
the magnetic field strength as shown in Fig. 2.8.

As gyro radius is inversely proportional to the magnetic field strength, thus a charge
particle is expected to have larger values of rL in the bottom part of figure in
comparison with the upper part. Again this difference in Larmor radii in upper and
lower regions lead to the drift of guiding center, i.e. on top of particle gyration there
is an average velocity in a particular direction. Any motion in the z-direction will
be additive and depends upon the initial vk component. For the reasons of algebraic
simplification and pictorial clarity, we may therefore take vk = 0. The equation of
motion in this case reads
h i
F~ = q ~v × B(r)
~ = q (vy B x̂ − vx B ŷ) =⇒ Fy = −qvx B(y) (2.32)

For the considered weak perturbations we can approximate y from the respective
undisturbed orbit calculations, i.e. y = y0 ± rL Cos(ωc t) to write

Fy = −qv⊥ Cos(ωc t)B [y0 ± rL Cos(ωc t)] (vx = v⊥ Cos(ωc t)) (2.33)

Assuming a weak magnetic field in-homogeneity over distances corresponding to a

gyro radius, i.e. |∇B|/B  1/rL , which corresponds to the GC approximation,
namely rL  L. Thus, we can write Taylor series expansion for Bz (y) and by
2.2. NON-UNIFORM MAGNETIC FIELDS 33

retaining the first two terms as follows

 
∂B
Fy ≈ −qv⊥ Cos(ωc t) B(y0 ) ± rL Cos(ωc t)
∂y
  (2.34)
∂B
= −qv⊥ Cos(ωc t) B0 ± rL Cos(ωc t)
∂y

Now gyro averaging, over the gyro-angle θ(= ωc t) yields

:0 : 1/2
∂B
2
hFy i = −qv⊥ B0
hCosθi
∓ qv⊥ rL Cos
 θ

∂y (2.35)
1 ∂B
= ∓ qv⊥ rL
2 ∂y

Note that if there is no gradient then ∂B/∂y → 0, and thus there is no extra force
acting on the system. The force in above equation will eventually result in the GC
drift velocity as given by

F~ × B~ hFy i × B ~
~vgc = =
qB 2 qB 2 (2.36)
v⊥ rL ∂B
=∓ x̂ (ŷ × ẑ = x̂)
2B ∂y

By similar analysis we find that hFy i = 0 (see Problem 2.5). In general, we can
write the GC drift velocity associated with the grad-B drift as

1 ~ × ∇B
B 2 ~
1 v⊥ B × ∇B
~v∇B = ± v ⊥ rL = ± (2.37)
2 B2 2 ωc B2
Here ± signs are used for ions and electrons, respectively. Thus the ∇B drift velocity,
in contrast to vE , is in opposite direction for electrons and ions as shown in Fig. 2.9.
Which means that it causes a separation of positive and negative charges and thus
can result in an electric field, which generates E ~ ×B ~ drift of all charged species in
the same direction.

2.2.2 Curvature drift

It is quite often, in particular for toroidal fusion devices, that there is a curvature
in magnetic field lines, lets assume such configuration with Rc denoting the radius
of curvature as shown in Fig. 2.10. When the charge particles gyrate around such
curved files lines, they actually follow a circular path. And hence are associated
with a centripetal force as given by

mvk2
F~c = r̂
Rc
(2.38)
mvk2
= ~c
R
Rc2
34 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.9: Motion of plasma charged particles in a spatially varying magnetic field
[Harms 2000].

Figure 2.10: Topology of the curved magnetic field configuration.

2.2. NON-UNIFORM MAGNETIC FIELDS 35

Figure 2.11: Illustration of curvature drift of a positive ion moving along the curved
B-field line [Harms 2000].

The corresponding drift velocity can be written as

F~c × B~
~vR =
qB 2
(2.39)
mvk2 R~c × B
~
=
qB 2 Rc2

which is also dependent on the sign of the charge. Figure 2.11 illustrates the cur-
vature drift of an ion spiralling around a curved magnetic field line. Furthermore,
there is an intrinsic gradient (in magnetic field) associated with curved field lines,
and thus there must also be a grad-B drift accompanied by the curvature drift.
Thus both specific drift velocities components add to yield the resultant drift. Let
us calculate the corresponding grad-B drift by writing the Maxwell’s equation and
using the cylindrical coordinates

0
0 ~
~ = µ0 J ∂ E 7

∇×B
~ + µ0 0  = 0



   ∂t   (2.40)
1 ∂Bz ∂Bθ ∂Br ∂Bz 1 ∂(rBθ ) ∂Br
− r̂+ − θ̂ + − ẑ = 0
r ∂θ ∂z ∂z ∂r r ∂r ∂θ
36 CHAPTER 2. PLASMA ORBIT THEORY

~ = Bθ (r), hence all

The curved filed lines are assumed to be along θ-direction, i.e. B
the terms in above equation vanish except for the second last, which provides
1 d
(rBθ ) = 0
r dr
d
(rBθ ) = 0 (2.41)
dr
1 r̂
=⇒ Bθ ∝ and ∇Bθ ∝ − 2
r r
Hence the Maxwell’s equation imply that a curved magnetic field can not be uniform.
In our notation, we may write

1 R̂c
Bθ ∝ and ∇Bθ ∝ −
Rc Rc2
∇Bθ −Rc R̂c
=
Bθ Rc2
(2.42)
∇Bθ ~c
R
=− 2
Bθ Rc
R~c
=⇒ ∇Bθ = −Bθ 2
Rc
Thus, the corresponding grad-B velocity can be evaluated from Eq.(2.37) as
2 ~
1 v⊥ B × ∇B
~v∇B = ±
2 ωc B2
1 v 2 −B B~ ×R ~c
=± ⊥
2 ωc Rc2 B 2
(2.43)
1 v⊥ 2
BR ~c × B~
=±
2 qB/m Rc2 B 2
2 ~ ~
1 mv⊥ Rc × B
=±
2 q Rc2 B 2
Hence, the resultant drift velocity – due to curved field lines – can be written as a
superposition
~vcurve = ~vR + ~v∇B
mR ~c × B
~  2
v⊥

(2.44)
2
=± vk +
q Rc2 B 2 2
which is in opposite direction for ions and electrons.

2.3 ~ field variations: Magnetic mirror

Axial B
We have already considered the cases where ∇B ⊥ B ~ and found the related grad-B
drift. Now, we take into account the scenario in which gradient is along the field
 ~ FIELD VARIATIONS: MAGNETIC MIRROR
2.3. AXIAL B 37

~
Figure 2.12: The magnetic configuration for a system with ∇B k B.

~ Figure 2.12 depicts a situation in which plasma particles see

lines, i.e. ∇B k B.
changing magnetic field strength as it moves along the field line, at point P it is in
a region of weak magnetic field and as it moves towards point Q the magnitude of
B~ increases. Let the magnetic field is directed along z-axis and is axi-symmetric,
i.e. Bθ = 0 as well as ∂/∂θ → 0, thus we have

~ = Bz ẑ + Br r̂
B (2.45)

Writing the Maxwell equation ∇ · B = 0, in cylindrical coordinates as

1 ∂ 1 ∂Bθ ∂Bz
(rBr ) + + =0
r ∂r r ∂θ ∂z (2.46)
d ∂Bz
(rBr ) = −r
dr ∂z

However, the exact dependence of magnetic field is not known and hence ∂Bz /∂z
can not be evaluated. For the mathematical simplicity we assume that the B-field is
slowly changing and we can write ∂Bz /∂z = Constant. This assumption simplifies
above equation as

 ˆ r
∂Bz
rBr = rdr
∂z 0
∂Bz r2
 
rBr = (2.47)
∂z 2
 
r ∂Bz
=⇒ Br = −
2 ∂z

The Lorentz force acting on a charge q in this magnetic field (Br ) can be evaluated
38 CHAPTER 2. PLASMA ORBIT THEORY

as
 
~ ~
F = q ~v × B = (vθ Bz )r̂ − (vr Bz − vz Br )θ̂ − (vθ Br )ẑ
=⇒ Fz = −qvθ Br
  
r ∂Bz (2.48)
= −qvθ −
2 ∂z
1 ∂Bz
= qvθ r
2 ∂z
The average force acting on the particle during one gyro cycle can be calculated by
taking r = rL , and vθ = −v⊥ , where negative sign takes into account the direction
of motion. Thus, we have
 
1 ∂Bz
hFz i = − qv⊥ rL
2 ∂z
2
 
1 v⊥ ∂Bz
=− q (2.49)
2 ωc ∂z
2
 
1 mv⊥ ∂Bz
=−
2 B ∂z

which in a generalized form can be written as

hFz i = −µ∇B (2.50)

Here µ is the magnetic moment and is given by

2
mv⊥ perpendicular kinetic energy
µ= = (2.51)
2B magnetic field

2.3.1 Invariance of magnetic moment

The magnetic moment associated with a charge particle moving in an orbit of radius
rL and generating a current I is defined as

µ = IA
q
= × πrL2
τ
2
qωc πv⊥
= × 2
2π ωc
2
q v (2.52)
= × ⊥
2 ωc
2
q mv⊥
= ×
2 qB
2
mv⊥
=
2B
 ~ FIELD VARIATIONS: MAGNETIC MIRROR
2.3. AXIAL B 39

If we choose a line element ds along the field line then Eq.(2.50) yields
 
dvk dB vk
m = −µ ×
dt ds vk
dvk dB ds
mvk = −µ (2.53)
dt ! ds dt
2
d mvk dB
= −µ
dt 2 dt

Only electric fields can do work and change the energy of the particle. Kinetic
energy is conserved, if only magnetic fields are present (see Problem 2.23), i.e.
 
d 1 2 1 2
mv + mv =0
dt 2 k 2 ⊥
2
! (2.54)
d mvk d
=⇒ = − (µB)
dt 2 dt

Now comparing Eqs. (2.53) and (2.54) to write

d dB
(µB) = µ
dt dt
dB dµ dB
µ +B =µ (2.55)
dt dt dt
dµ
=⇒ =0 (∵ B 6= 0)
dt
Hence, as the individual isolated charged particle moves in the (slowly) spatially
varying magnetic field, the magnetic moment does not vary with time, i.e. it is an
adiabatic invariant.

2.3.2 The magnetic mirror effect

The practical consequence of the conservation of energy and magnetic moment is
related with the so-called magnetic mirror effect. Consider Fig. 2.13, as the value
of magnetic field increases toward the throat of the mirror region (P to R), v⊥ must
increase and by energy conservation vk decreases; thus, ions tend to decelerate along
the axial direction as they move into the higher magnetic field of the mirror throat
where vk can even become zero, if B is high enough. Note that Fk , always points
opposite to ∇B, is still acting on the particles thus causing a reflection at both ends
of magnetic bottle, i.e. at points Q and R.

2.3.3 The loss cone

The condition for reflection at points Q and R can be written as
 
vk B≤Bmax = 0 (2.56)

Conservation of kinetic energy demands

40 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.13: The mirror effect on charge particle motion in a magnetic configuration
for a system with ∇B k B.~

1
E0 = m vk2 max + µBmin ,


(2.57)
2
or equivalently we can write
1
E0 = m vk2 min + µBmax


2
=⇒ E0 ≤ 0 + µBmax (2.58)
E0 ≤ µBmax

Comparing above equation with Eq.(2.57), we find

1
m vk2 max + µBmin ≤ µBmax


(2.59)
2
The above condition holds for maximum vk2 value for trapping, upon re-arranging
we can write
 
1/2 mvk2 Bmax
max
≤ −1
µBmin Bmin
  (2.60)
1/2 mvk2 Bmax
max
=⇒ 2
≤ −1
1/2 (mv⊥ )min Bmin
Thus, for the trapping domain we must have
!
vk2 Bmax
2
≤ −1 (2.61)
v⊥ Bmin
MP
 ~ FIELD VARIATIONS: MAGNETIC MIRROR
2.3. AXIAL B 41

Figure 2.14: The velocity vector of a particles at mid-plane in a mirror configuration.

and the loss domain is defined by the following condition at the mid-plane (MP)
!
vk2 Bmax
2
> −1 (2.62)
v⊥ Bmin
MP

The velocity vector of a particle at the MP of a magnetic mirror can be decomposed

into its components along and perpendicular to the field line as depicted in Fig.
2.14, and to write the trapping condition as follows
vk v⊥
cos θ = , sin θ =
v v
2
 2
cos θ vk
=⇒ 2 =
sin θ v⊥
 2 (2.63)
1 vk
= +1
sin2 θ v⊥
1 Bmax
=⇒ 2 ≤
sin θ Bmin
Thus, the trapping condition can be written in terms of the angle that velocity
vector makes with the magnetic field line at mid-plane as
Bmin
sin θ ≥ (Trapping condition) (2.64)
Bmax
Whereas, all the other particles are lost which are within the so-called Loss cone
with θ ≤ θ0 , with r r
B min 1
θ0 = sin−1 = sin−1 , (2.65)
Bmax Rm
with Rm = (Bmax /Bmin ) is denoting the mirror ratio. Thus, all the particles in the
loss cone are eventually lost from the system. One notes that the plasma particles
can be lost from either end of the cone, i.e. we have a double cone as shown in Fig.
2.15.

2.3.4 Calculation of lost fraction for an isotropic plasma

Consider the case of a plasma having an isotropic VDF, which means that, in spher-
ical coordinates, it is independent of angular coordinates (θ, ϕ) with 0 ≤ θ ≤ π and
42 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.15: The loss cone geometry.

0 ≤ ϕ ≤ 2π. Note that there are actually two values of loss angles related with the
two opposite loss cones and accounting for the motions parallel and anti-parallel to
~ The fraction of loss particles can be evaluated as
B.
Number of particles in double cone
fL =
Total number of particles

´
f (v)d3 v
= ´
doublecone
f (v)d3 v

´2π ´  ´θ0 ´π
" #

dϕ
   2f (v)v dv sin θdθ sin θdθ
0
 0 π−θ0 (2.66)
= ´2π ´ 
2 ´ π 
dϕ
  f(v)v dv t sin θdθ 0
0


cos θ|0θ0 − cos θ|π−θ

π
0

=
cos θ|0π
1 − cos θ0 + cos(π − θ0 ) − cos π
=
cos 0 − cos π
= 1 − cos θ0

Therefor, the trapped fraction is given by

p
fT = 1 − fL = cos θ0 = 1 − sin2 θ0
(2.67)
r r
Bmin 1
= 1− = 1−
Bmax Rm

Analysis of fL :
(i) In the limit when there is no gradient of magnetic field, i.e. Bmax = Bmin , we
find fT = 0, i.e. there is no trapping (or mirror effect). In this case a charge par-
 ~ FIELD VARIATIONS: MAGNETIC MIRROR
2.3. AXIAL B 43

Figure 2.16: A magnetic dipolar field with a belt of trapped charged particles
[Chaudhry 1998], and the schematics of Van Allen belts.

ticle will just follow a helical path along the magnetic field lines out of the open ends.

(ii) For a magnetic bottle with a large mirror ratio, Rm (= Bmax /Bmin )  1, in that
case we have a perfect mirror action and all the particles are trapped, i.e. fT → 1.

2.3.5 Magnetic mirror and Van Allen belt

Figure 2.16 depicts the magnetic field configuration outside a spherical dipole, where
filed lines are closer together near the poles. Thus there are field variations along
the field line, and we can expect the mirror effete at the poles. Hence the region
between the points P and Q can be thought of as a curved magnetic bottle and
charges can remain trap there while moving back and forth between the reflection
points (polar regions). But this is not the whole story. Due to curvature of these
field lines there is an associated curvature drift, which is always accompanied by
grad-B drift. Both these cross field drifts will produce a GC drift in the azimuthal
(ϕ) direction. Hence the guiding center, while moving to and fro between the polar
regions, also goes around the polar axis of the spherical body.

Apart from giving an idea as to how the results of the GC theory can be put together,
the above discussion should make it clear that it is possible for charged particles
to remain tapped in the Earth’s magnetosphere. Such a region of trapped ions was
discovered by Van Allen with instruments borne on spacecrafts. This belt of charged
particles is known as the Van Allen belt. As the gradient and the curvature drifts of
electrons and ions are in opposite directions. This gives rise to an azimuthal current
(through this belt) which is called the ring current that play an important role in
communications that take place via the ionosphere. Figure 2.17 shows the resultant
motion of plasma charge particles in the magnetic field of earth, here the trapping
is due to the mirror effect, whereas the azimuthal drifts are caused by grad-B and
curved nature of magnetic field lines.
44 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.17: Effect of various drifts in the magnetic field of earth.

2.4 ~ finite Larmor radius ef-

Spatially varying E:
fect
We have shown that a uniform electric field causes the so-called E ~ ×B ~ drift of
the guiding center across magnetic field, and the drift velocity ~vE is independent of
charge and mass of the species. Now we consider a sinusoidal spatial dependence,
~
i.e. E(x) ~ = B ẑ. Thus, equation of motion reads
= E0 Cos(kx)x̂ and B
d~v q h~ ~
i
= E(x) + ~v × B (2.68)
dt m
which provides the following horizontal and vertical components
q qB
v̇x = Ex + vy
m m
q
= Ex ± ωc vy
m
q (2.69)
=⇒ v̈x = Ėx ± ωc v˙y
m
Ėx
= ±ωc + ±ωc v˙y
B
and, respectively
qB
v̇y = − vx
m
= ∓ωc vx (2.70)
=⇒ v̈y = ∓ωc v̇x

Above equations are coupled and can be un-coupled to write

Ėx
v̈x = −ωc2 vx ± ωc
B (2.71)
E x
v̈y = −ωc2 vy − ωc2
B
 ~ FINITE LARMOR RADIUS EFFECT
2.4. SPATIALLY VARYING E: 45

We know from our unperturbed orbit calculations that x = x0 + rL Sin(ωc t), which
yields
ωc2
v̈y = −ωc2 vy − E0 Cos [k(x0 + rL Sinωc t)]
B
2 ωc2
= −ωc vy − E0 Cos [kx0 + krL Sinωc t)] (2.72)
B
ω2
= −ωc2 vy − c E0 [Cos(kx0 )Cos(krL Sinωc t) − Sin(kx0 )Sin(krL Sinωc t)]
B
Now performing averaging over the gyro-angle (θ = ωc t) and using - as we expect -
a steady drift velocity, i.e. < v̈y >= 0 to write

ωc2 E0
0 = −ωc2 hvy i − [Coskx0 hCos(krL Sinωc t)i − Sinkx0 hSin(krL Sinωc t)i]
B
E0
hvy i = − [Coskx0 hCos(krL Sinωc t)i − Sinkx0 hSin(krL Sinωc t)i]
B
(2.73)
The small gyro-radius, in comparison with any scale length, approximation yields

rL  λ(= 1/k) =⇒ krL  1

and using the following expansions of trigonometric functions

k 2 rL2
cos(krL ) ≈ 1 − , Sin(krL ) ≈ krL
2
to evaluate different terms in Eq.(2.73) as follows
k 2 rL2

Sin2 ωc t

hCos(krL Sinωc t)i ≈ 1 −
2
k 2 rL2
=1− (2.74)
4
hSin(krL Sinωc t)i ≈ krL hSinωc ti
=0
Thus, one finds
k 2 rL2
   
E0
hvy i = − Coskx0 1 − −0
B 4
k 2 rL2
 
E0 Coskx0
=− 1−
B 4
 2 2
 (2.75)
Ex k rL
=− 1−
B 4
k 2 rL2
 
E×B
= 1−
B2 4
Hence, a slight decrease in the usual vE is observed, this is due to the fact that a
particle at larger values of E spends a good amount of time in the weak electric field
46 CHAPTER 2. PLASMA ORBIT THEORY

regions as well. For an arbitrary electric field dependence, one can write E in its
Fourier components such that

E ∝ eikx , =⇒ ∇E = ikE and ∇2 E = (ik)2 E = −k 2 E (2.76)

Therefore, we finally write

  ~ ~
1 E×B
hvy i = 1 + rL2 ∇2 , (2.77)
4 B2
where we note that the second term vanishes if we take the limit rL = 0, hence this
term is called the finite Larmor radius effect. Moreover, above equation depicts that
the GC drift depends on charge species via rL and thus causes a charge separation
in the system. For a constant electric field the second terms vanishes and we get
the usual E~ ×B ~ drift velocity.

2.5 ~ Polarization drift

Time varying E:
As a last example, we consider an oscillatory time dependence in electric field having
a frequency ω and is directed along x-axis, i.e.
~ = E0 eiωt x̂
E (2.78)
~
Ė = iω E~ (2.79)
~ = −ω 2 E
Ë ~ (2.80)

The respective components, using Eq.(2.71), read

Ėx
v̈x = ±ωc − ωc2 vx
B (2.81)
E x
v̈y = −ωc2 − ωc2 vy
B
Let us denote any time varying quantities A by the overhead symbol à to write
!
ω c ω ω ˜x
E
v̈x = −ωc2 vx ± i E˜x = −ωc2 vx ∓ i
B ωc B
! (2.82)
E ˜x E˜x
v̈y = −ωc2 vy − ωc2 = −ωc2 vy +
B B

Here we define two oscillatory drift velocities as

E˜x
ṽE = − (2.83)
B
ω E˜x
ṽp = i (2.84)
ωc B
m dE ~
=± 2 (2.85)
qB dt
2.6. PROBLEMS 47

which transform Eqs.(2.82) as follows

v̈x = −ωc2 (vx − ṽp )

(2.86)
v̈y = −ωc2 (vy − ṽE )

Above equations look similar to the one we have solved for the unperturbed case
except for the terms ṽE and ṽp , thus we propose a solution of the form

vx = v⊥ eiωc t + ṽp
(2.87)
vy = ±iv⊥ eiωc t + ṽE

To check the validity of our solutions, lets take the time derivatives of above equa-
tions, which yield

v˙x = iωc v⊥ eiωc t + iωṽp

=⇒ v¨x = −ωc2 v⊥ eiωc t − ω 2 ṽp ± ωc2 ṽp
(2.88)
= ωc2 v⊥ eiωc t + ṽp + ṽp ωc2 − ω 2

= ωc2 vx + ṽp ωc2 − ω 2

Similarly, for vy one finds

v¨x = ωc2 vy + ṽE ωc2 − ω 2

(2.89)

Note that above equations are not the same as Eq.(2.86) unless ω 2  ωc2 . Thus, our
solutions are applicable only when the oscillating frequency of electric field is much
smaller than the gyro frequency. In the above expression ṽE is the usual E ~ ×B~
drift for time varying electric field, whereas ṽp is defined as the polarization drift
and is linked with changing electric field. Here we note that in contrast to ṽE , the
polarization drift velocity depends on charge and mass of the species, moreover it
is directed along the electric field. The corresponding polarization current density -
due to opposite moving electrons and ions - can be evaluated as

Jp = ne (ṽip − ṽep )
~
ne(mi + me ) dE
= (2.90)
eB 2 dt
~
dE
= ρm ,
dt
where ρm denotes the total mass density of plasma composed of electrons and ions.

2.6 Problems
P(2.1) A uniform electric field is directed to the right and a uniform magnetic field
is applied in the direction perpendicular to the electric field (into the page). If a
positive test charge q is moving upward, the magnetic force is to the left and the
electric force is to the right. Calculate the velocity of the particle for which it moves
48 CHAPTER 2. PLASMA ORBIT THEORY

in a straight vertical line (undeflected) through the region of the fields. This is the
main principle of velocity selectors.
(Ans: v = E/B)

P(2.2) A beam of ions first passes through a velocity selector having perpendicular
fields E x̂, B0 ẑ and then enters a second uniform magnetic field B1 ẑ. Where it is
deflected and moves in a semicircle of radius r before striking a detector.
(a) If the charge is positive, in what direction it is deflected?
(b) Show that charge to mass ratio can be written as
q rB0 B1
= (2.91)
m E
P(2.3) Suppose that there is an electric field along z-axis (upward) and the magnetic
field points in the x-direction (out of the page). A positive test particle starts at
rest at the origin, show that it will follow a cycloid motion with its guiding center
having a drift velocity perpendicular to both the fields.

P(2.4) Starting with Eq.(2.28), calculate the corresponding gravitational drift ve-
locity and show that ions and electrons drift in opposite directions. Then derive the
following expression for total current density due to this drift

~g × B~
J = n(me + mi ) (2.92)
B2
P(2.5) In the context of grad-B drift we have derived the y-component of force in
Eq.(2.35), perform similar analysis to show that hFx i = 0.

P(2.6) From Eq.(2.44) show that the curvature drift depends on the charge but not
on the mass of species by showing that
2kB T
~vcurve = ± ŷ (2.93)
qmRc
P(2.7) Using the fact that µ is an adiabatic invariant, show that the magnetic flux
through the Larmor orbit is a constant.
(Ans: Φm = 2πmµ/q 2 )

P(2.8) Using the following hints show that for a (slowly) time varying magnetic field
the magnetic moment is an adiabatic invariant.
~
Hints: (a) The changing magnetic field will induced an E-field, for which the equa-
tion of motion provides the following relation
 
1 2 ~
~ · dl
∆ mv = qE (2.94)
2 ⊥
(b) Integrate over the gyro period and use Faraday’s law to write
  ˆ ˆ
1 2 ~
~ · ds,
∆ mv = −q B ds = πrL2 (2.95)
2 ⊥
2.6. PROBLEMS 49

(c) Eventually use the following steps

  !
1 2 2π Ḃ
∆ mv⊥ = µ = µ∆B, ∆B = τ (∂B/∂t)
2 ωc (2.96)
=⇒ ∆(µB) = µ∆B =⇒ µ = Constant.

P(2.9) In the context of mirror effect, suppose that an electron is located at z = 0

with initial velocity v0 and an initial pitch angle θ. The mirror field is given by

Bz = B0 1 + α2 z 2

(a) Using the conservation of energy and magnetic moment calculate the reflection
(or turning) point zr .
(b) Calculate the trapped particle fraction in this configuration.
1
(Ans: zr = αtanθ )
P(2.10) Repeat the previous problem for

Bz = B0 1 + α4 z 4

1
(Ans: zr = √
α×tanθ
)

P(2.11) (a) For the Fig. 2.18 the time dependence of the position of the electron in
the first part is expressed as x(t) = 0.125t4 + π. The electron remains in the 1st part
for one second. Calculate the magnitude of the velocity that the electron acquires
at the end of the first part.
(Ans: 0.5m/s)
(b) After that, it enters the second part, where a transverseE of strength 10−10 V /m
acts on it. This field is generated by the plates of a capacitor with the length
d = 1m. What is the vertical displacement of the electron with respect to the
starting position at the end of the second part?
(Ans: 35.2m)
(c) Finally, the electron enters a homogeneous magnetic field of the magnitude 20.6×
10−6 T . Calculate the Larmor radius, cyclotron frequency and the magnitude of the
magnetic moment of the rotating electron.
(Ans: 9.72 × 10−6m, 3.6M rad/s, 2.7 × 10−23 A · m2 )
P(2.12) Consider a Deutron (with mass 1875M eV /c2 ) in large mirror machine having
energy 30keV , it has a pitch angle of 45o at mid-plane, where B = 0.9T . Compute
its Larmor radius.
(Ans: 0.03m.)

P(2.13) Consider the Fig. 2.19 where an electron beam with ne = 1016 m−3 and
radius a = 1 cm flows along a magnetic field B = B0 ẑ where B0 = 2 T. Calculate
the magnitude and direction of vE at r = a.
50 CHAPTER 2. PLASMA ORBIT THEORY

Figure 2.18: Problem (2.11)

Figure 2.19: Problem (2.13).

(Ans: 4.52 × 103 m/s. )

P(2.14) A 10 keV proton with vk = 0 in a uniform magnetic field of 1 T, is accelerated
as B is slowly increased to 10 T. It then makes an elastic collision such that vk = v⊥ .
Finally, the magnetic field is slowly decreased back to 1 T. What is the proton’s
energy now?
(Ans: 5.5keV )

(b) A magnetic mirror configuration has√ a mirror ratio Rm , show that the Larmor
radius of the particle at the center is Rm times the Larmor radius at the point of
maximum magnetic field.

P(2.15) In a fusion device, Deutron with energy 200keV enters the machine and
will be considered confined if the corresponding Larmor radius is less than radius of
device a = 0.6 m. Compute the maximum value of rL in a magnetic field of strength
5 T to see if the particles remained confined.
(Ans: 0.018m)

P(2.16) For a cylindrical plasma with a uniform magnetic field of strength 1T and
Ti = Te = 0.2eV . The density profile as found experimentally to be described by
the relation
−r 2 /a2 −1
n(r) = n0 ee (2.97)
Assuming that the density obey the Boltzmann relation.
~ ×B
(a) Calculate the maximum value of E ~ speed at a = 1cm.
(b) Calculate the corresponding gravitational drift speed.
2.6. PROBLEMS 51

(c) Calculate the value of magnetic field at which ions of potassium have a rL = a.

P(2.17) Suppose that earth’s magnetic field varies as B = B0 /r3 , where B0 =

3 × 10−5 T . If there are protons with 1eV and electrons with 30keV energies. Such
that np = ne = 107 /m3 at 5 earth radii.
(a) Compute the ion and electron drift velocities, i.e. both magnitude and direction.

(Ans: 0.39m/s toward west, 1.17 × 104 m/s towards east)

(b) With this speed how long does it take for an electron to encircle the earth?
(Ans: 4.8hours )
(c) Calculate the corresponding ring current density.
(Ans: 1.87h × 10−8 A/m2 )

P(2.18) A plasma with an isotropic distribution is placed in a mirror configuration

with Rm = 4. Assuming there are no collisions in the system. Calculate the fraction
of trapped particles.
√
(Ans: 3/2)

P(2.19) The polarization drift velocity can also be derived from energy conservation.
If E ~ is oscillating, the E
~ ×B~ drift also oscillates, and there is an energy 0.5mv 2
E
associated with the guiding center motion. Since energy can be gained from an E
field only be motion along E, ~ there must be a drift vp in the electric field direction.
By equating the rate of change of energy gain from ~vp · E,~ find the required value of
~vp .

P(2.20) A Hydrogen plasma is heated by applying a radio frequency wave with

~ ⊥B
E ~ and ω = 109 rad/s. The confining magnetic field is 1 T. The motion of a
species is considered to be adiabatic if its frequency is much greater than ω. Is the
motion of (a) the electrons and (b) the ions in response to this wave adiabatic?
(Ans: ωe = 1.76 × 1011 rad/s, ωi = 9.58 × 107 rad/s)

P(2.21) The mirror field is given by B(z) = B0 (1 + γ 2 z 2 ). Assume that at z = 0

2
v 2 = 3vk2 = 1.5v⊥ . Calculate the value of z at which particle gets reflected. Write
the equation of motion of the guiding center for the parallel direction. And show
that the motion is sinusoidal.
√ p 
(Ans: zr = ±1/ 2α, z = zr Sin 2µ0 B0 /mαt

P(2.22) Consider the configuration in which B ~ = B0 ẑ, ∇B = B 0 x̂, E

~ = E0 ŷ and
~g = g ŷ. Calculate the magnitude of the net drift velocity.

P(2.23) Show that Kinetic energy is conserved, if only magnetic fields are present.
Hints: Write down the equation of motion for a charged particle in an electro-
magnetic field, which contains only Lorentz force. Then take a dot product of the
equation with ~v to write  
d 1 2 ~ · ~v
mv = q E
dt 2
52 CHAPTER 2. PLASMA ORBIT THEORY

which clearly shows that only electric fields can do work and change the energy of
the particle. Kinetic energy is conserved, if only magnetic fields are present.
Chapter 3

Fluid theory of plasmas

Plasma is a media consisting of interspersed charged particles which are affected by

short and long range forces. The cumulative effect of these forces combined with
the intractability of an analytical description suggests that various approaches be
used in the determination of the macroscopic behavior of an ensemble of moving
and colliding particles. Where electric and magnetic fields are not prescribed but
are determined by the motion of charged particles themselves, these fields in turn
affect the plasma particle’s motion. For which one needs to consider the equation
of motion for each particle i, i.e.
dvi 
~ + ~v × B

~ ,
mi = qi E i=1→N (3.1)
dt
Hence, if we get (somehow) the electric and magnetic fields we can calculate the
particle dynamics, these fields could be external (as was the case in previous chapter)
or internal (to be calculated by the particle’s motion). For which one needs to solve
the Maxwell’s equation using bunch of free and bound charge information. Thus,
in principle a self-consistent set of equations must be solved for the complete and
accurate information about the plasma dynamics, the work flow of such calculations
is depicted in Fig. 3.1.
However, in reality, for typical plasma densities of 1012 − 1018 m−3 we need to follow
huge number of particles to calculate the charge and current densities to be used in
Maxwell’s equations for evaluation of fields and continue repeating the calculations
in a cycle. Which seems to be hopeless task as N is huge, even with with most
powerful super-computers people are not able to follow 1019 particles. Hence we can
not easily view the plasma as a bunch of individual particles and add them up, i.e.
we do not do adopt the microscopic approach and are driven to a so-called fluid (or
macroscopic) description where the identity of individual particle is not considered.
Luckily, many of the plasma phenomena can be explained by fluid model, where the
motion of fluid elements is taken into account. However, in contrast to the ordinary
fluid where collisions between the particles keep the fluid moving, the plasmas con-
tain charged particle as its constituents and thus electromagnetic forces are taken
into account.

As an alternative to using an exceedingly large number of equations each contain-

ing many terms, it has been found that other approaches which are mathematically

53
54 CHAPTER 3. FLUID THEORY OF PLASMAS

Figure 3.1: A schematic diagram of the calculations involved in various computa-

tional codes.

tractable provide, in selected applications, satisfactory agreement with experiments

and a useful predictive quality. One frequently used conceptual approach is to con-
sider a plasma as a multi-component interpenetrating low-density fluid and then
employ suitable continuum mechanics methods. Another approach is to use proba-
bilistic considerations for each group of particle species and then perform an analysis
based on methods of sampling statistical physics. While no one approach is gen-
erally applicable to all cases of conceivable interest, a careful choice of conceptual
constructs and methodologies can lead to descriptions which are remarkably useful
in characterizing selected space-velocity-time aspects of a particular species. It is
important therefore to develop a good understanding of the imposed assumptions
in order to recognize important restrictions of a particular conceptual development
and its consequent mathematical description.
In plasma physics we generally work with Maxwell’s equations in a vacuum in which
the charge and current densities include, respectively all the charges and currents,
both external and internal.

3.1 Classical magnetic materials v/s plasma

Lets defined µj to be the magnetic moment associated with a gyrating particle j,
then the magnetization M is defined as the total magnetic moment per unit volume,
3.2. CLASSICAL TREATMENT OF DIELECTRIC AND PLASMA 55
P
i.e. M = j µj /V , from which one can evaluate the bound current density as

J~b = ∇
~ ×M
~ (3.2)

The Maxwell’s equation reads

~
 
~ ×B
∇ ~ = µ0 J~b + J~f + 0 Ė
~
~ × B =∇
∇ ~ ×M ~
~ + J~f + 0 Ė (3.3)
µ0
∇~ ×H ~
~ = J~f + 0 Ė,

where

~
~ = B −M
H ~ (3.4)
µ0
For a linear media, the magnetic susceptibility χm is defined through the relation
~ ∝H
M ~
(3.5)
~ = χm H
M ~

Thus, Eq. (3.4) can be written as

 
~ ~ ~
B = µ0 H + χm H
~
= µ0 (1 + χm ) H (3.6)
= µm H~

Here µm = µ0 (1 + χm ) is the magnetic permeability of the system.

However,we have noted that in a plasma µi ∝ 1/B, which also implies that M ∝
1/B, i.e. it is not a linear media and the analysis presented above can not be applied
to it. Hence we can not write B ~ = µm H~ for a plasma system. Therefore, it is not
useful to consider the plasma as a magnetic medium.

3.2 Classical treatment of dielectric and plasma

In theory of dielectric we define defined pj as the electric dipole moment, then the
polarization
P P~ is defined as the total dipole moment per unit volume, i.e. P~ =
j pj /V , from which one can evaluate the bound charge density as

~ · P~
ρ b = −∇ (3.7)

The Gauss’s law can be written as

56 CHAPTER 3. FLUID THEORY OF PLASMAS

~ ·E
0 ∇ ~ = ρb + ρf
 
~ · 0 E
∇ ~ = −∇ ~ · P~ + ρf (3.8)
~ ·D
=⇒ ∇ ~ = ρf

where

~ = 0 E
D ~ + P~ (3.9)
For a linear media, the electrical susceptibility χe is defined through the following
relation
P~ ∝ E
~
(3.10)
P~ = 0 χe E
~

Thus, Eq. (3.9) can be written as

~ = 0 E
D ~ + 0 χe E
~
~
= 0 (1 + χe ) E (3.11)
~
= E
Here  = 0 (1 + χe ) is the electric permittivity of the system.

Again, in a plasma the linear relation has not been reported. Hence above treatment
can not be used directly for  in a plasma system.To conclude the quantities  and
µm are not straight forward and can not by using Maxwell’s equation in a medium
approach, rather they need to be evaluated by other schemes.

3.3 Dielectric constant for plasma in low frequency

limit
We have already noted in previous chapter that for E = E(t) we get the polarization
drift and hence a polarization current density Jp . Then the corresponding charge
density ρp can be computed by using the continuity equation, namely

∂ρp ~ ~
+ ∇ · Jp = 0 (3.12)
∂t
As a result, the Ampere’s law takes the form
~
 
~ ~ ~ ~
∇ × B = µ0 Jp + Jf + 0 Ė
   
~ Jp ~
= µ0 Jf + 0 + Ė (3.13)
Ė
~
 
= µ0 J~f + Ė
3.4. THE FLUID EQUATIONS 57

with
Jp
 = 0 +
Ė
ρm Ė/B 2 (3.14)
= 0 +
Ė
ρm
= 0 + 2
B
which provides the desired dielectric constant R as
 ρm
R = =1+
0 0 B 2
(3.15)
ρm µ0 c2
=1+
B2
√
Here we have used the relation for speed of light in vacuum, i.e. c = 1/ 0 µ0 as
well as for the polarization current density for the limit when ω  ωc . Thus, the
above relation provides a low frequency plasma dielectric constant for the transverse
motion (x̂).
(i) If ρm → 0, means no plasma is there then from Eq. (3.15) we find R → 1, which
is true for the vacuum case.
(ii) If there is very strong magnetic field, i.e. B → ∞, still we get the vacuum
scenario, namely R → 1. This is due to the fact that in that case there is no
polarization drift velocity. However, in typical plasma systems the second term is
usually much greater than unity, which means that the internal electric field of the
plasma greatly alters the external E(t). A plasma system with larger values of
dielectric constant shields out alternating fields.

3.4 The Fluid equations

In the fluid description one considers the plasma to be composed of two (or more)
interpenetrating fluids. For the simplest case when plasma is considered to be a
single fluid (known as the magneto hydrodynamic (MHD) model), we shall need
one equation of motion. Likewise for an electron ion plasma (fully ionized) there
are two equations required, one for each species. While for partially ionized case -
involving neutrals as well - an additional equation is needed for the neutrals as well.

3.4.1 Convective derivative

There are also alternate names for this quantity, e.g. substantial, Lagrangian and
material derivative. It refers to the total change in a quantity as observed by a
moving observer. The two observers, one of them is stationary and the other one
is moving with a velocity v, at a given position and time should record the same
value of an observable, i.e. their measurements are equivalent but the corresponding
interpretations are different.
58 CHAPTER 3. FLUID THEORY OF PLASMAS

Figure 3.2: The fluid element and changes as measured by a moving and stationary
observer.

Consider the Fig. 3.2 in a plane containing some fluid on the inside, let us focus on
a fluid particle that can move around in the system. It can change it shape but it
remains the same otherwise. At the time t = 0 particle (fluid particle) is at initial
point r0 . At any later time it is at position r which is function of initial coordinate
and time, i.e. r = r(r0 , t), if we are interested to measure some concentration field
C, then the two measurements - by an stationary observer and the one moving with
the fluid element - must give the same result independent of the coordinate system.
Thus, any field in that fluid can be represented as

C(~r, t) = C(~r0 , t) (3.16)

3.4. THE FLUID EQUATIONS 59

C(~r, t) = C (~r(~r0 , t), t)

d ∂C ∂~r ∂C
C (~r(~r0 , t), t) |r0 =con. = + (3.17)
dt ∂t ∂t ∂r
∂C  ~ 
= + ~v · ∇ C,
∂t
where ~v denotes the particle velocity. Here the first term corresponds to the changes
(in C) observed at fixed position, whereas the second term tells about the changes
when we are moving with a velocity ~v . Thus, we can write the material derivative
in the form

d D ∂ ~ f ixed−t ,
= = |f ixed−x + ~v · ∇| (3.18)
dt Dt ∂t
Just an illustrative example is to consider a person who is interested to calculate the
changes in temperature T as it moves through a cave or a tunnel. As he/she moves
into the cave it will experience a change in T due to its motion (~v ), i.e. the second
term in above equation. If at the instant he steps in, someone throws a snowball at
him, again the change in temperature is observed at a fixed spatial coordinate, this
temporal change corresponds to the first term in above expression.

For any macroscopic quantity Q(~r, t) we can write the convective derivative as

D d ∂Q(~r, t)  ~ 
Q(~r, t) = Q(~r, t) = + ~v · ∇ Q(~r, t) (3.19)
Dt dt ∂t

3.4.2 Continuity equation

The continuity equation is actually the mass (or particle) conservation equation.
Consider an arbitrary volume V containing a time-varying particle populations N (t)
such that the volume surface is to be non re-entrant and having the total surface
area A. The quantity N (t) can be calculated from the number density n(~r, t) as
follows

ˆ
N (t) = n(~r, t)d3 r (3.20)

Now the question is by what mechanism the population can change? There are
various types of gain and loss reactions possible in the system with reaction rates
R±j , other than that N can change due to inflow and outflow rates F±j across the
boundary as shown in Fig. (4.8). Hence a general rate equation reads

!
dN X X
= R+jk − R−jl + (F+j − F−j ) , (3.21)
dt k l
60 CHAPTER 3. FLUID THEORY OF PLASMAS

Figure 3.3: The number of particles in a volume element can change due to inward
and outward flux as well due to various reactions.

where the subscripts k and l enumerating different reaction types. We restrict

ourselves to the case for which the reaction contributions are zero or, equivalently,
the reaction gains are exactly canceled by the reaction losses, additionally we refer
to the condition F+j = 0, which means there is no fueling by injection or other
mechanisms. Whereas, the outflow rate across the surface, accounting for global
particle leakage, can be evaluated via the local particle current density vector as

‹
F−j = ~ r, t) · dA,
J(~ ~
ˆ ‹
∂ 3 ~ r, t) · dA,
~
=⇒ n(~r, t)d r = J(~
∂t
ˆ ˆ (3.22)
∂ 3 ~ · J(~
~ r, t)d3 r
n(~r, t)d r = − ∇ (Gauss’s theorem)
∂t
∂ ~ · J(~
~ r, t) = 0
=⇒ n(~r, t) + ∇
∂t
Or, in a more general case it can be written for any jth species in the form

∂ ~ · J~j (~r, t) = 0
nj (~r, t) + ∇
∂t (3.23)
∂ ~ · (nj ~uj ) = 0
nj (~r, t) + ∇
∂t
Here ~uj is the corresponding fluid velocity, the above relation is known as the conti-
nuity equation and must hold everywhere in the volume of interest and for all times
of relevance providing the assumptions imposed here hold. If, however, there were
sources and sinks for the considered species j in the volume of interest – i.e., reac-
tions which produce or consume particles – or particle injection, the corresponding
3.4. THE FLUID EQUATIONS 61

gain and loss rate densities would appear on the right hand side of it.

Although continuity equation is compact, however suffers from a severe shortcoming:

it represents only one equation containing two unknown functions, the scalar particle
density nj (r, t) and the vector ~uj , which has to be determined from the force balance
equation.

3.4.3 Momentum (Force) balance equation

As stated in previous subsection that one needs to calculate the vector ~uj in order
to solve the continuity equation for which we have to write equation of motion
for the jth fluid species. If the fluid is composed of various type particles having
their respective densities nj and fluid velocity ~uj then the corresponding equation
of motion yields

d 
~ + ~uj × B

~ + Other forces
mj nj ~uj = nj qj E (3.24)
dt

Now for a symbolic convenience let us omit the subscript j, and from Eq.(3.19) with
Q(~r, t) = ~u, we can write the above equation in the form

  
∂ 
~

~ ~

mn ~u + ~u · ∇ ~u = nq E + ~u × B + Other forces (3.25)
∂t

Here we have only included the Lorentz force, the other forces may include pressure
gradient force, gravitational and/or collisional forces etc.

3.4.4 Pressure gradient force

If there are more particles in one region of space than the other, i.e. there is a
concentration difference between two regions, then a pressure gradient would arise
from low to high concentration regions as shown in Fig. (3.4). The corresponding
pressure gradient force density is then given by

F~p = −∇P
~ (3.26)

For P one needs to write an equation of state, as discussed in the following.

3.4.5 Equation of state

A relation which connects the pressure P of the system with density is termed as
equation of state, namely P = P (ρ). For example consider the following thermody-
namic equation of state
62 CHAPTER 3. FLUID THEORY OF PLASMAS

Figure 3.4: The pressure gradient force.

P = cργ , (γ = Cp /Cv )
~ = cγρ ∇ρ
∇P γ−1 ~

~
∇ρ
=γ
ρ (3.27)
~
∇P ~
∇ρ
=⇒ =γ
P ρ
~
∇n
=γ
n
Here γ is called the polytropic coefficient and for an isothermal process we can write

P = nkB T
~ = kB T ∇n
∇P ~
(3.28)
~
∇P ~
∇n
=⇒ =
P n
Comparison of Eqs.(3.27) and (3.28) yields γ = 1 for an isothermal process.

However, for an adiabatic process with d degrees of freedom it can be shown that
3.5. FLUID DRIFT PERPENDICULAR TO MAGNETIC FIELD 63

2+d
γ= (3.29)
d

3.4.6 Collision (with neutrals) and momentum change

In a partially ionized plasma the charged fluid can exchange momentum through
collisions with neutrals. If ~u0 is the velocity of neutral fluid then momentum loss
per collision will be mnνc (~u −~u0 ). Here νc is the collision frequency and the negative
sign corresponds to a loss of momentum, of the charged fluid, due to collisions with
neutrals. The collision with charged particles is also possible and is modeled by the
Fokker-Planck collision term, however in this chapter we only consider the former.
Hence by including pressure gradient, gravitational and collisional forces Eq. (3.25)
can be written in the form
  
∂ 
~

~ ~

~ − mnνc (~u − ~u0 )
mn ~u + ~u · ∇ ~u = nq E + ~u × B + mn~g − ∇P (3.30)
∂t

3.5 Fluid drift perpendicular to magnetic field

Similar to the guiding center drift, as discussed in previous chapter, there are also
drifts associated with a plasma fluid. This make sense as the fluid itself is composed
of many individual charged particles. As we have seen that there is a pressure
gradient force present in the fluids, and thus one should expect a corresponding
drift. Consider the equation of motion in the presence of Lorentz and ∇P ~ forces,
i.e.

d 
~ + ~u × B

~ − ∇P
~
mn ~u = nq E (3.31)
dt
Assuming an equilibrium situation (or a case where the drifts are slow in compar-
ison with time scale of ωc ) and thus left hand side of the above equation vanishes.
Moreover, writing the velocity vector as sum of its parallel and perpendicular com-
ponents, i.e. ~u = ~uk + ~u⊥ and taking the cross product with magnetic field vector
to write
h   i
~ ~ ~
0 = nq E + ~u × B − ∇P × B ~
h    i
~ ×B
= nq E ~ + ~u⊥ × B)~ ×B ~ − ∇P
~ ×B ~ (3.32)
h   i
~ ×B
= nq E ~ − u⊥ B 2 − ∇P~ ×B ~ ~ = 0)
(∵ u⊥ · B

which can be solved for the perpendicular drift velocity to write

~ ×B
E ~ ~ ×B
∇P ~
u⊥ = −
B2 qnB 2 (3.33)
= ~uE + ~uD
64 CHAPTER 3. FLUID THEORY OF PLASMAS

Figure 3.5: The cylindrical geometry and corresponding diamagnetic drift velocities
for electrons and ion.

~ ×B
Here the first term represents the usual E ~ drift velocity which has same value for
electrons and ions, whereas the second term is called the diamagnetic drift velocity
and is associated with pressure gradient force in fluids. From Eq. (3.27) we have

~
~ = γ P ∇n
∇P
n
~ ×B ~ = γP ~ ×B ~
∇P ∇n
n
γP B ~ ~ = B ẑ)
= ∇n × ẑ (∵ B (3.34)
n
1 γP B ~
=⇒ ~uD = ± ẑ × ∇n
qnB 2 n
~
γkB T ẑ × ∇n
~uD = ± (∵ P = nkB T )
qB n

For an isothermal process γ = 1, and for a cylindrical geometry as shown in Fig. 3.5
~ = r̂∂n/∂r = n0 r̂. For any confined plasma we expect a higher pressure at
we have ∇n
~ = n0 ẑ×r̂ = n0 θ̂,
the center than at the boundaries and thus n0 < 0. Moreover, ẑ× ∇n
thus the drift velocities for ions and electrons are in opposite direction and can be
written as

kB Ti n0
~uDi = θ̂
eB n (3.35)
kB Te n0
~uDe =− θ̂
eB n
3.6. TWO MAIN LIMITS OR APPROXIMATIONS 65

Figure 3.6: A density gradient results in a net upward drift of the fluid.

If we define |n0 /n| = 1/Λ then the magnitude of the diamagnetic drift speed takes
the form

T [eV ]
uD [m/s] = (3.36)
Λ[m]B[T ]
The said drift is a result of gradient in concentration of particles, which means that
if there are less particles on the left hand side than right side in Fig. (3.6) and in any
small element (shaded region) there are more ions moving up than moving down.
As a result there is a net fluid drift upward.

3.6 Two main limits or approximations

Let us write the momentum equation along the field line for an electrostatic case

d ~ kP
mn uk = nqEk − ∇
dt (3.37)
~ kφ − ∇
= −nq ∇ ~ kP

Now there are two main scenarios that are generally considered in plasma physics,
namely adiabatic response and fluid response.

3.6.1 Fluid response

This notion corresponds to a fast process in which the frequency of interest (for a
given process) is much larger than the corresponding frequency related with particle
dynamics, i.e.

ω  [Something related with particle dynamics],

where the term in [ ] can be, for example, cyclotron frequency or thermal vibrations.
Any such process is considered to be fast and if the above condition is satisfied we
66 CHAPTER 3. FLUID THEORY OF PLASMAS

refer it to the fluid response of the system.

Now let us consider Eq. (3.37) and look for fluid response approximation, in that
case we are talking about a fast process, i.e. ω  vther /Lk =⇒ d/dt  vther /Lk .
Thus, the last term which involving thermal speed and parallel gradient length can
be ignored in comparison with frequency of interest and we may write

d ~ kφ
mn uk = −nq ∇ (3.38)
dt
~
For the perturbations of the form ei(k·~r−ωt ), i.e.

uk , φ ∝ ei(kk z−ωt )

which leads to

−iωmnuk = −ikk nqφ

kk qφ (3.39)
=⇒ uk =
mω

3.6.2 Adiabatic response

If we are interested in a phenomenon which has a characteristic frequency ω, such
that
ω  [Something related with particle dynamics],
where the term in [ ] can be, for example, cyclotron frequency or thermal vibrations.
Any such process is considered to be a slow one and if the above condition is satisfied
we refer it to an adiabatic response of the system. Consider an example of such slow
process in which we are interested in wave mode that has low frequency compared
to how fast particles of the system oscillate back and forth, i.e.

ω  vther /Lk (≈ vk kk )

which implies that left hand side of Eq. (3.37) vanishes, i.e. d/dt ∝ ω → 0.
Therefore we have

~ kφ − ∇
0 = −nq ∇ ~ kP
~
~ k qφ − ∇k n
0 = −∇
kB T n
~
∇k n (3.40)
=⇒ ~ k qφ
= −∇
n kB T
 
~ ~ qφ ∂ 1 ∂A
∇k lnn = −∇k lnA =
kB T ∂z A ∂z

Upon integrating we find

3.7. PLASMA APPROXIMATION 67

 
n qφ
ln =−
n0 kB T (3.41)
− kqφT
=⇒ n = n0 e B

which is the usual Boltzmann relation (or Gibbs distribution) corresponding to an

adiabatic response of the system and provides an expression for the density distri-
bution for a given potential. The electron in the system - due to smaller mass -
accelerates to higher energies very quickly (for a non-zero force) and leave behind
a large positive charge. The resulting electrostatic force when balances with the
pressure gradient force results in the Boltzmann density distribution.

It is worth mentioning here that it is possible that different plasma species response
differently, e.g. electrons are adiabatic and ions are fluid. Likewise both species are
adiabatic (or fluid) for a given scenario.

3.7 Plasma approximation

For plasma electrons we expect an adiabatic response, and thus to follow the Boltz-
mann distribution, they move faster from a region of space under the action of a
force, leaving behind ions. Thus, there are more ions in that particular space, which
results in build up of potential φ that holds the electron back. We can say that
plasma wants to stay quasi-neutral (as we have already proved in first chapter),
via this induced potential, i.e. ni ≈ ne , which implies that the total charge density
ρ → 0, in this case we don’t use Gauss’s law to determined electric field or potential.
Because we generally use the Gauss’s law for a given ρ, i.e. E = E(ρ). However,
in plasma opposite is done, namely (a) we calculate electric field from momentum
balance equation and (b) then density is evaluated via the Poisson’s equation, i.e.
ρ = ρ(E).

The electric field must adjust itself such that orbits of electrons and ions preserve
~ ·E
neutrality, i.e. ni ≈ ne and it is still possible that ∇ ~ 6= 0.Thus, in plasma theory
we try not to use the Poisson’s equation to obtain electric field unless unavoidable.
The plasma approximation, i.e. ni ≈ ne is a mathematical shortcut used to study
various plasma waves. For slow enough phenomenon such that electrons and ions
have time to response (move) we can replace Poisson’s equation approach to plasma
approximation approach.

3.8 Stress tensor (optional)

While writing the equation of motion we introduced a pressure gradient force by
writing F~p = −∇P
~ and treated pressure as a scalar quantity. However, in general
pressure is a tensor, to be exact it is a tensor of rank 2 and is called the stress
tensor. For a scalar (tensor of rank 0) you only need one information, i.e. a number
68 CHAPTER 3. FLUID THEORY OF PLASMAS

or magnitude, for a vector one requires two things magnitude + direction and is
termed as tensor of rank 1. Whereas for a second rank tensor it requires three things
to describe it completely, namely magnitude + direction + plane of application. Let
us write the stress tensor S in a matrix form as
 
Sxx Sxy Sxz
S =  Syx Syy Syz  , (3.42)
Szx Szy Szz
where each of the element represents stress in a particular plane. We can equally
write the above relation in terms of diagonal (normal) and off-diagonal (shear) ele-
ments, as

Normal stress, i = j
Sij =
Shear stress, i 6= j
and hence we have 6 shear stresses and 3 normal stresses. We generally write the
normal stress in σ notation as Sxx = σx and similarly for other two parts. Thus we
have
 
σx Sxy Sxz
S =  Syx σy Syz  (3.43)
Szx Szy σz
In the symbol Sij the first index in subscript refers to the name of the plane on which
stress is acting, whereas the second index provides the direction in which stress acts.
Thus, σx = Sxx corresponds to the stress acting on x-plane and directed along x-axis
which means a normal (to the plane) stress as shown in Fig. (3.7), likewise by Sxz
means stress on x-plane which is directed along z-axis (i.e. tangential to the plane)
and is one of the shear stress.
For a general case we will have an expression of the type F~p = −∇ ~ · S, here we note
that the scalar product always reduces the rank of a tensor by 1, that is why we get
a force (tensor of rank 1) by taking dot product of shear (rank 2 tensor). It is to
mention here that the pressure or stress tensor can always be written as a sum of
isotropic and an-isotropic parts as

P = pI + Π,

where I is the unit dyad, i.e. a 3 × 3 unit matrix and appears in the isotropic part
of pressure tensor, whereas Π is denoting the so-called viscosity tensor and is the
an-isotropic part.

To conclude this chapter we might state that the fluid analysis of plasma leads to
a compact space-time description for the particle or mass density. Of particular
interest to us now is a characterization of collective motion in a plasma. In such
a description, the identity of individual particles is put aside and the plasma is
characterized by the space-time changes of macroscopic variables such as bulk speed,
temperature, and pressure defined in a fluidic context.
3.9. PROBLEMS 69

Figure 3.7: Shear and normal stresses presented for a cubical geometry.

3.9 Problems
P(3.1) Show that diamagnetic current density is given by

~ ×B
∇n ~
J~D = − (kB Ti + kB Te ) (3.44)
B2
P(3.2) For a cylindrical plasma in a uniform magnetic field can be modeled by the
radial density distributions given as
2 /a2
n(r) = n0 e−r (3.45)

Moreover, lets assume that both ions and electrons satisfy the Gibbs relation

ni = ne = n0 eeφ/kB T (3.46)

Show that vE and vDe are equal and opposite. Also show that the plasma rotates
as a solid body.

P(3.3) Starting with the equation P (x) = n(x)kB T (x) show that
 
dP dn n dT /dx
= γkB T where γ = 1 + (3.47)
dx dx T dn/dx
Here γ is called the polytropic coefficient. Also show that it γ = 1 for isothermal
process.

P(3.4) Consider a plasma confined between the planes x = ±d in a magnetic field

B ẑ with a density distribution given below. Assuming the plasma to be isothermal
70 CHAPTER 3. FLUID THEORY OF PLASMAS

calculate the electron diamagnetic drift as a function of x.

x2
 
n(x) = n0 1 − 2 , c = constant (3.48)
c
Chapter 4

Plasma waves: A fluid description

Whenever you perturb a plasma in some way, it reacts by producing various type
of waves. The plasma is a medium that - depending on different perturbation
conditions - can support a huge number of waves. Here we discuss some of the
basic wave modes and their propagation in plasmas in various approximations. For
that the the linear plasma theory is used, where only the perturbations upto the
first order are considered.

4.1 Review of waves and wave solutions

Waves are all around us, e.g. sound waves, electromagnetic waves, water waves
etc. Any wave can be represented in the form f (x − vt), which describes a moving
structure with speed v. We generally present waves by oscillatory functions of the
form

ψ(x, t) = Aei(kx−ωt) , (4.1)

where A is denoting the amplitude, k(= 2π/λ) is the wave number with λ denot-
ing the wave-length of the considered mode. Any periodic motion of fluid can be
decomposed, by Fourier analysis, into superposition of sinusoidal oscillations with
various ω 0 s and k 0 s, a simple wave is one of them, i.e.
X
ψ(x, t) = Aj ei(kj x−ωj t) (4.2)
j=1

4.1.1 Phase and group velocities

Consider a ball of mass m moving in a given direction with a velocity v, then all parts
of the ball move with same speed and there is no uncertainty in its position x(t) and
velocity v(t). Similarly for a bus, the driver, passengers, windows and all other parts
move with same speed. However, a quantum particle - introduced by De Broglie - is
represented by a group of waves, which interfere in a way that constructive interfer-
ence occur only at one place. This group of waves is called a wave packet or envelop
as shown in Fig. 4.1. The resulting amplitude of the wave packet gives us the
probability Pr of finding the particles, which means that Pr1 < Pr2 < Pr3 .... < Pr6 .

71
72 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.1: An envelop or a wave packet.

The velocity of each individual component is called the phase velocity vφ = ωj /kj ,
whereas the velocity of the envelop refers to the group velocity vg = dω/dk and
in general vφ 6= vg . For example a group of honey bees is flying over you then the
speed of each bee (phase velocity) could be different than the group (group velocity).

Let us consider two waves of same amplitude A and slightly difference ω and k, i.e.
for dk  k, dω  ω, we have

ψ1 (x, t) = ACos(kx − ωt),

(4.3)
ψ2 (x, t) = ACos [(k + dk)x + (ω + dω)t] ,

with the resultant profile given by a superposition of the form

ψ = ψ1 + ψ2
= ACos(kx − ωt) + ACos [(k + dk)x + (ω + dω)t]
0 0
 
= 2ACos (k +  dk/2)x + (ω + 
>
 dω/2)t Cos [(dk/2)x + (dω/2)t]
>

(4.4)
 
dk dω
= 2ACos x+ t Cos(kx − ωt)
2 2 | {z }
| {z } Wave like part
Modulated amplitude

Now the phase velocity - that is related with the individual component - can be
calculated by setting the phase kx − ωt equal to a constant, i.e.
4.2. ELECTRON PLASMA FREQUENCY 73

d
(kx − ωt) = 0
dt
kdx − ωdt = 0
dx ω (4.5)
=
dt k
ω
=⇒ vφ =
k
Likewise for the group velocity we have

 
d dk dω
x+ t =0
dt 2 2
dkdx − dωdt = 0
dx dω (4.6)
=
dt dk
dω
=⇒ vg =
dk

4.2 Electron plasma frequency

We have already introduced the concept of plasma frequency ωp in first chapter, here
we will explicitly derive the relation for it. For that let use consider the following
assumptions;
(i) Unmagnetized system, B = 0.
(ii) Cold plasma, i.e. no thermal motions are involved T → 0.
(iii) Ions - due to their heavy mass - are considered to be fixed, i.e. ni = n0 . This
assumption is valid for all those cases - like the one we are dealing here of electron
oscillations - which occur on a fast time scale such that ions do not have enough
time to move.
(iv) Infinite 1D plasma is considered, i.e. there are no boundary effects.
(v) Electrostatic case, E~ = −∇φ.

Equation of motion and the continuity relation for electrons read

 
∂~ve ~
mne + (~ve · ∇)~ve = −ene E
∂t
  (4.7)
∂~ve ~
m + (~ve · ∇)~ve = −eE
∂t
and, respectively
∂ne
+ ∇ · (ne~ve ) = 0 (4.8)
∂t
The Gauss’s law can be written as
74 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

~ = e(ni − ne )
0 ∇ · E
~
∂E (4.9)
0 = e(ni − ne )
∂x
We have three coupled equations, which are difficult and complicated mostly because
of the non linear terms. However, we can get a tractable problem by linearizing.
That means, take some known equilibrium solution and suppose the deviation (per-
turbation) from it is small such that we can retain only the 1st linear terms and not
the others.

4.2.1 Linear theory

Let us write any fluid quantity Q as a superposition of corresponding equilibrium
value and multiples of perturbation parameter λ  1 as

Q(r, t) = Q0 + λQ1 + λ2 Q2 + . . .
|{z} |{z} | {z }
Equilibrium value Q1 Q2
(4.10)

≈ Q0 (r) + Q1 (r, t) (Upto the first order term)

Here due to smallness of perturbation parameter Q1  Q0 and we have ignored the

higher order terms. In the light of above relation we can write

n(r, t) = n0 (r) + n1 (r, t),

v(r, t) = v0 (r) +v1 (r, t),
| {z }
=0
φ(r, t) = φ0 (r) +φ1 (r, t) (4.11)
| {z }
=0
E(r, t) = E0 (r) +E1 (r, t)
| {z }
=0

Here the equilibrium velocity, electric field and potential are chosen to be zero as a
reference.

As ∂n0 /∂t = 0 and using a uniform (homogeneous) plasma, i.e. ∂n0 /∂x = 0, we can
write Eq. (4.7) in the form

 
∂~v1 ~1
m + (~v1 · ∇)~v1  = −eE
∂t | {z }
2nd order
(4.12)

∂~v1 ~1
=⇒ m = −eE
∂t
4.2. ELECTRON PLASMA FREQUENCY 75

Similarly, the continuity equation and Gauss’s law, in linear analysis, read

∂n1
+ ∇ · [(n0 + n1 )~v1 ] = 0
∂t
∂n1
+ ∇ · [n0~v1 ] = 0
∂t (4.13)
∂n1 *0
+ n0 ∇ · ~v1 + ~v1 · 
∇n 0 = 0

∂t
∂n1
+ n0 ∇ · ~v1 = 0
∂t
and, respectively

~ = e(ni − ne )
0 ∇ · E
∂E1 :0 (4.14)
ni0− ne0 − n1 )

0 = e(  
∂x
Here we have made use of the equilibrium quasi-neutrality condition. Now intro-
ducing all the perturbed quantities, which have an oscillatory nature, in the form
Q1 ∝ ei(kx−ωt) , which is equivalent of replacing ∂/∂t by −iω and ∇ = ∂/∂x by ik.
Thus, Eqs. (4.12)- (4.14) yield

−iωmv1 = −eE1 , (4.15)

−iωn1 = −ikn0 v1 (4.16)

and

ik0 E1 = −en1
=⇒ E1 = ien k0
1
(4.17)

which can be easily solved for the oscillation frequency ω, substitute the value of v1
from the second into the first equation to write

 
iωn1
−iωm = −eE1
n0 ik
(4.18)
−mω 2 n1
=⇒ E1 =
n0 ike

Upon comparing above equation with expression in Eq. (4.17), we have

−mω 2 n1 ien1
=
n0 ike k0
(4.19)
n0 e2
=⇒ ω 2 = ≡ ωp
0 m
76 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Here ωp is known as the electron plasma frequency and describes one of the funda-
mental oscillations in plasmas. Using the values of fundamental constants we can
write

s
n0 e2 √
ωp [rad/s] = ≈ 56 ne
0 m
(4.20)
ωp √
fp [Hz] = ≈ 9 ne
2π

Physical interpretation of ωp
When an electron is, somehow, displaced away from the ions then - due to charge
separation - there is a built up of an electric field in such a way to restore the neu-
trality of plasma. Thus, the electron is pulled back toward their original position.
Because of their inertia this electron overshoots the equilibrium point and moves to
the other side, again E field is developed to pull them back and this cycle contin-
ues. Hence the electrons vibrate around equilibrium positions with a characteristic
plasma frequency ωp .
(i) The frequency is inversely proportional to mass, which is physical obvious be-
cause the heavier the specie is the more difficult it is to cause oscillations.
(ii) As n increase so we expect that electric field should also enhances, resulting in
a stronger electric force on electrons. Thus, faster oscillations are generated.
(iii) As ωp does not depend on the wave vector k, therefore the group velocity
vg = dω/dk = 0. Which implies that its not a wave but only oscillations that does
not propagate. We can think of different electrons as independent harmonic oscilla-
tors having a frequency ωp that all oscillate by themselves. In the next section we
well see that by including the thermal effects these oscillators become coupled and
can results in a wave.

4.2.2 Collisional damping of plasma oscillations

Once we include the collision term in equation of motion to account for collisions
between electrons and neutrals we expect that the electron plasma waves to get
damped. For any oscillating quantity Q ∼ ei(kx−ωt) , the time dependent part is
Q ∼ e−iωt and if frequency is a complex number, i.e. ω = ωr + iωi we can write

! 

Q∝ eiωr t
|{z}
 eωi t
|{z}
 (4.21)
Oscillations Growth/damping

Hence we can write a criteria for damping or growth of wave as


> 0, Wave grows
ωi = .
< 0, Wave damping
4.3. LANGMUIR WAVES: ELECTRON PLASMA WAVES 77

For simplicity let us assume that pressure gradient term is small and can be ignored
thus we would have plasma oscillations that can be damped due to the collision
term. Let us rewrite Eq.(4.27)

−iωmn0 v1 = −en0 E1 − mn0 νc v1

−1
(4.22)

−ieE1 iνc
v1 = 1+
mω ω
Now substituting the values of E1 and v1 interms of n1 from Eqs. (4.15)to write
above relation in the form

 
iνc ieE1
v1 1 + =−
ω mω
   
ω
n1 iνc −ie ie n1
1+ =
kn0 ω mω k0
n0 e2 (4.23)
ω + iνc = = ωp2
0 mω
ω 2 + iνc ω = ωp2
p 2
−iνc 4ωp − νc2
=⇒ ω = ±
2 2
In the limit when 4ωp2 − νc2 > 0, the imaginary part of frequency is negative, i.e.
Im(ω) = −νc /2, which corresponds to the damping oscillations.

4.3 Langmuir waves: electron plasma waves

In the last section we noted that if there are no thermal motions of electrons then
they perform the so-called plasma oscillations with a frequency ωp and it does not
constitute any waves. Let us include electron-thermal motions by including the grad-
p term in equation of motion, whereas rest of the equations remain the same. For
that we consider the ideal gas equation for electrons in the isothermal and adiabatic
limit to write

Pe = γnkB Te
∇Pe = γkB Te ∇ne
= γkB Te ∇(n0 + n1 ) (4.24)
= γkB Te ∇n1 (∇n0 = 0)
∂n1
= 3kB Te x̂ (γ = 3)
∂x
which for sinusoidal perturbation becomes

∇Pe = 3kB Te (ikn1 )

(4.25)
= 3ikkB Te n1
78 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Now rewriting the linearized version equation of motion (4.12) in the presence of
thermal motions to write

∂~v1 ~ 1 − 3ikkB Te n1
mn0 = −en0 E
∂t (4.26)
~ 1 + 3ikkB Te n1
iωmn0 v1 = en0 E
Using the values of E1 and v1 interms of n1 from Eqs. (4.15)to write above relation
in the form
   
ωn1 ien1
iωmn0 = en0 + 3ikkB Te n1
kn0 k0
 2
ω n0 e
m = + 3kkB Te
k k0 (4.27)
2
n 0 e k B Te
ω2 = + 3k 2
m0 m
3
ω 2 = ωp2 + k 2 vt2
2
which is the required dispersion relation.

Analysis of dispersion relation

(a) If there are no thermal motions, i.e. vt = 0 then we have the usual oscillations
with plasma frequency.
(b) The corresponding group velocity can be calculated by differentiating above
equation to write

3
2ωdω = 0 + vt2 2kdk
2
dω 3 2ω
= vt
dk 2 k
3 1 (4.28)
vg = vt2
2 vφ
3
vg vφ = vt2  c
2
(c) Equation (4.27) highlights a parabolic dispersion relation with its minimum
lying at ω = ωp . That implies that no wave propagation is possible below plasma
frequency.
(d) The asymptotic behavior can be seen in the limit k → ∞ and one can ignore
the first term to write

3
ω 2 = k 2 vt2
2
r !
3 (4.29)
ω= vt k (y = mx + c)
2
4.4. ACOUSTIC AND ION-ACOUSTIC WAVES 79

Figure 4.2: The dispersion curve for electron plasma waves.

p
it is an equation of straight line
p with slope given by 3/2vt , which also the phase
and group velocity vφ = vg = 3/2vt as shown in Fig. 4.2.
Thus at small wave length information travels essentially with thermal velocity,
whereas for large wavelength it travels more slowly than thermal velocity.
(e) At any point A on the dispersion curve, the phase speed vφ = ω/k or, ω = vφ k
again a straight line with slope vφ . Whereas for the group velocity we need to
draw a tangent to the curve at point A and to measure the corresponding slop, i.e.
vg = dω/dk.

4.4 Acoustic and ion-acoustic waves

Sound waves

The waves associated with ion motion have similar characteristics as that of sound
waves, that’s why they are also refer to as ion-acoustic waves. Hence before we derive
the dispersion relation for ion waves we present a brief review of the acoustic waves,
which are pressure waves propagating from one layer to the next via compression and
rarefaction. However, as they are produced due to the motion of neutral particles,
thus equation of motion and continuity relation takes the form
 
∂~v −γP ∇ρm
ρm + (~v · ∇) ~v = −∇P = (4.30)
∂t ρm
and, respectively

∂ρm
+ ∇ · (ρm~v ) = 0 (4.31)
∂t
80 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

After linearization and using sinusoidal approximation we can write equation of

motion as

∂~v1 −γP0 ∇ρm1

ρm0 = (∵ ∇ρm0 = 0)
∂t ρm0
−γP0 ikρm1
−iωρm0 v1 = (4.32)
ρm0
γP0 kρm1
v1 =
ρ2m0 ω

Likewise, from continuity equation we have

∂ρm1
+ ρm0 ∇ · ~v1 = 0
∂t
−iωρm1 = −ρm0 ikv1 (∵ ~k k ~v1 ) (4.33)
rhom0 kv1
ρm1 =
ω
Using Eq.(4.33) in Eq.(4.32) yields

γP0 k rhom0 kv1

v1 = ×
ρ2m0 ω ω
γP0
ω2 = k2
ρ
s m0 r (4.34)
ω γP0 γkB T
= = (P = nkB T, ρm0 = M n)
k ρm0 M
r
ω γkB T
=⇒ = cs ≡
k M
Here cs is the sound speed. Note that for sound wave vφ = vg = cs , all sound waves
(neglecting viscosity) propagate at the same speed cs .

Ion acoustic waves

In plasma we have mostly charged particles, however an analogous phenomenon
occur and is refer to as ion acoustic waves that are caused by the intermediary
electric field. As for the electric field both species are critical, thus we expect that
both electrons and ions to play a role in the propagation of ion waves, moreover
these waves - due to heavy ion motion - are low-frequency waves in comparison with
Langmuir waves. Consider the ion’s momentum balance equation

 
∂~vi ~ − ∇Pi
Mn + (~vi · ∇) ~vi = enE (4.35)
∂t
4.4. ACOUSTIC AND ION-ACOUSTIC WAVES 81

~ = ∇φ,
Now applying the linearization, sinusoidal perturbations and using E ∇P =
γkB T ∇n to write

∂~vi1
M n0 = −en0 ∇φ1 − γi kB Ti ∇ni1
∂t
−iωM n0~vi1 = −iken0 φ1 − ikγi kB Ti ni1 (4.36)
ωM n0~vi1 = ken0 φ1 + ikγi kB Ti ni1
The corresponding ion continuity equation can be used to calculate perturbed den-
sity as
∂ni1
+ n0 ∇ · ~vi1 = 0
∂t
−iωni1 + ikn0 vi1 = 0 (4.37)
kn0
=⇒ ni1 = vi1
ω
For electrons we can choose the Gibbs distribution and apply small perturbation
limit to write

eφ1
ne = n0 e kB Te
 
eφ1
= n0 1 + + ...
kB Te

eφ1
 (4.38)
= n0 + n0
k T
 B e
eφ1
=⇒ ne1 = n0 (∵ n = n0 + n1 )
kB Te
Now using the quasi-neutrality condition, i.e. equating Eqs.(4.37) and (4.38) to find
perturbed ion velocity in terms of potential as
 
kn0 eφ1
vi1 = n0
ω k T
 B e (4.39)
ω eφ1
vi1 =
k kB Te
Using above equations in Eq.(4.36) yields
   
ω eφ1 eφ1
ωM n0 = ken0 φ1 + n0 γi kkB Ti
k kB Te kB Te
2
Mω γi kkB Ti
=k+
kkB Te k T
 B e  (4.40)
2 2 kB Te γi kB Ti
ω =k +
M M
r
ω kB Te + γi kB Ti
= ≡ vs
k M
82 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Here vs is the speed of sound in plasma and, in general, is given by

r
γe kB Te + γi kB Ti
vs = (4.41)
M
(i) Electrons are the fast moving particles and get thermalize very quickly, and thus
are treated as isothermal for which γe = 1.
(ii) For a 1D motion of ions, they can be treated adiabatic for which γi = 3.
(iii) If there are no thermal motions in system, i.e. Ti = Te = 0 then vs = 0.
However, if ions are cold Ti = 0 the ions waves, in contrast to the neutral ordinary
air case, still exist with
r
γe kB Te
vs =
M
which is governed by temperature of electrons and inertia (mass) of ions.
(iv) The ion acoustic waves are constant velocity waves.

4.5 Plasma approximation

For plasma electrons we expect an adiabatic response, and to follow the Boltzmann
distribution. They move faster from a region of space under the action of a force,
leaving behind ions. Thus, there are more ions in that particular space, which results
in build up of potential φ that holds the electron back. We can say that plasma
wants to stay quasi-neutral (as we have already proved in first chapter), via this
induced potential, i.e. ni ≈ ne , which implies that the total charge density ρ → 0
and in this case we don’t use Gauss’s law to determined electric field or potential.
Because we generally use the Gauss’s law for a given ρ, i.e E = E(ρ). However, in
plasma opposite is done, namely (a) we calculate electric field from momentum bal-
ance equation and (b) then density is evaluated via the Poisson’s equation ρ = ρ(E).

The electric field must adjust itself such that orbits of electrons and ions preserve
~ 6= 0. Thus, in plasma theory
neutrality, i.e. ni ≈ ne and it is still possible that ∇ · E
we try not to use the Poisson equation to obtain electric field unless unavoidable. The
plasma approximation, i.e. ni ≈ ne is a mathematical shortcut used to study various
plasma waves. For slow enough phenomenon such that electrons and ions have time
to response (move) we can replace Poisson approach to plasma approximation.

4.5.1 Validity of plasma approximation

As we know that to solve for fluid density we need to solve the continuity equation,
which however involves the fluid velocity term and has to be evaluated from equation
of motion that, along with other parameters, requires the knowledge the knowledge
of potential distribution. The potential and number density are connected via the
Poisson’s equation that is needed to be solved. However, as in the ion acoustic
waves, it is nor always necessary to solve the Poisson’s equation, instead we used
4.5. PLASMA APPROXIMATION 83

the realtion ni = ne (φ) and to find n(φ). Now let us use the Poisson’s equation and
to see what is new and under what condition the plasma approximation is valid?
Consider the Gauss’s law

~1 = ρ1
∇·E = (ni1 − ne1 )
0
e
∇2 φ1 = (ni1 − ne1 )
0
0 k φ1 = e ni1 − n0 eeφ1 /kB Te
2



n0 e 2
 (4.42)
2
0 φ1 k + = eni1
0 kB Te
0 φ1 1 + k 2 λ2D = eni1 λ2D

eni1 λ2D
φ1 =
0 (1 + k 2 λ2D )

Similarly continuity equation provides the expression for perturbed number density
in terms of velocity as

kn0
ni1 = vi1 (4.43)
ω

Using Eqs.(4.42) and (4.43) in Eq.(4.36) yields

eλ2D kn0 kn0

ωM n0 vi1 = ken0 2 2
vi1 + γi kkB Ti vi1
0 (1 + k λD ) ω ω
k2 n0 e2 λ2D
 
2
ω = + γi kkB Ti
M 0 (1 + k 2 λ2D ) (4.44)
v
ω u u kB Te 1 γi kkB Ti
=u 2 2 +
k t M 1 + k λD M
| {z }
Correction

Hence if we use the plasma approximation - instead of solving the Poisson’s equation
- then the error is 1/(1 + k 2 λ2D ) and for the limiting case k 2 λ2D → 0 the plasma
approximation provides the same results as the more accurate counterpart. We can
also write

k 2 λ2D → 0
2
(4.45)

2πλD
→0
λ

In many of the wave analysis λD /λ → 0, thus plasma approximation is valid for all
but for short wave length (high frequency) waves, e.g. Langmuir waves.
84 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.3: The dispersion curve for ion and Langmuir waves,
www2.mps.mpg.de/solar − system − school/lectures.

Comparison of ion and electron waves

(a) In the limit Te → 0, from Eq.(4.27) the electrons waves are constant frequency
oscillations. Whereas the ions waves for Ti → 0 are constant velocity waves.
(b) For short wavelength (large k) in Eq.(4.27), we canpignore the first term in
comparison with the second, and find that vφ = ω/k = 3/2vt = Constant, i.e.
the Langmuir waves also become constant velocity waves. (c) For Ti → 0 and very
high frequency (short wavelength) limit, i.e. k 2 λ2D  1 we have

kB Te ne2 2 ne2
 
2 2 kB Te
ω =k = ω =
M k 2 λ2D M 0 kB T λ2D M 0

Thus for high frequency limit the ion acoustic waves turn into constant frequency
waves.

To conclude, the electron waves in the limit Te = 0 are constant frequency waves and
become constant velocity waves for short wavelength. Whereas for Ti → 0 the ion
acoustic waves are constant velocity waves and becomes constant frequency waves
in the short wavelength limit as shown in Fig. 4.3, which also highlights that no
electrostatic wave can propagate between ωpe and ωpi in an unmagnetized plasma.
 ~ IN A COLD PLASMA85
4.6. ELECTROSTATIC ELECTRON OSCILLATIONS ⊥ TO B

4.6 ~ in a
Electrostatic electron Oscillations ⊥ to B
cold plasma
So far we have considered the unmagnetized plasma waves, the inclusion of magnetic
field results in many other types of waves, however the analytical treatment of the
problem becomes more complicated.

Notations/Terminologies

1. Parallel/Perpendicular: it is used to describe the direction of wave vector ~k

with respect to the unperturbed magnetic field B~0 .

2. Longitudinal/Transverse: this notions provides direction of ~k with respect to

perturbed electric field E~1 .
~ 1 = 0, which means
3. Electrostatic waves corresponds to the situation when B
~ ~ ~ ~
that ∇ × E1 = −∂ B1 /∂t = 0 and thus E = −∇φ. ~

~1 =
4. Electromagnetic waves are the more general description when B 6 0.

Equivalently, for sinusoidal perturbations we can write

~
~ ×E
∇ ~ 1 = − ∂ B1
∂t
~ ~ ~1 (4.46)
ik × E1 = −(−iω)B
~k × E~1 = ω B
~1

Hence the longitudinal waves, i.e. ~k k E~1 ⇒ ~k × E~1 → 0 and B

~ 1 = 0, corresponds to
the electrostatic modes. Likewise for a transverse wave, namely ~k ⊥ E~1 ⇒ B ~ 1 6= 0,
we have electromagnetic scenario. Whereas for all other cases a mixture of both
modes can exist.

Now consider the following assumptions

• the waves have high frequency such that the electrons can follow it, whereas
the massive ions can be treated stationary,
~ = −∇φ,
• the waves are electrostatic in nature, i.e. E ~

~ and
• electrons oscillations are perpendicular to B

• no thermal motion exist, namely KB Te → 0.

In this section we are considering the fast oscillation modes, for which the momentum
balance equation – in its linear form – for electrons reads
∂ v~e1 
~ + ~ve1 × B
~0

m = −e E (4.47)
∂t
86 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

The corresponding continuity equation and Gauss’s law takes the form
∂ne1 ~ · ~ve1 = 0
+ n0 ∇ (4.48)
∂t
and, respectively
~ · E~1 = −ene1
0 ∇ (4.49)
For the analysis of purely longitudinal waves (~k k E~1 ) let us chose the geometry as
ky = kz = Ey = Ez = 0 and ~k = kx̂, E ~ 1 = E1 x̂. For a mathematical convenience
lets drops the subscripts (e and 1) and using the usual sinusoidal perturbations to
write Eqs.(4.47) to (4.49) as

iωm~v = −e (E x̂ + vy B0 x̂ − vx B0 ŷ) (4.50)

which can be written in its Cartesian components as follows

iωm~vx = −eE − evy B0 (4.51)

ieB0
iωmvy = evx B0 =⇒ vy = vx (4.52)
mω
iωmvz = 0 (4.53)
Using Eq.(4.52) in Eq.(4.51) to write
 
ieB0
−iωmvx = −eE − eB0 vx
mω
eE eB0 iωc
vx = + vx
imω m ω
(4.54)
ωc2
 
eE
vx 1 − 2 =
ω imω
(eE/imω)
vx =
1 − ωc2 /ω 2
Note that if ω → ωc ⇒ vx → ∞ (cyclotron resonance), however obviously its not
possible to get an infinite speed, as then the non-linear effects become significant
that we have ignored altogether in deriving above expressions. Now from Eq.(4.48)

−iωn1 = −n0 i~k · vx

n0 kvx (4.55)
n1 =
ω
Equation (4.49) can be simplified as following
n0 kvx
ik0 E = −en1 = −e
ω
en0 eE
i0 E =− (using value of vx )
ωimω(1 − ωc2 /ω 2 ) (4.56)
ωc2 ωp2
⇒1− = 2
ω ω
ω2 = ωp2 + ωc2
 ~ IN A COLD PLASMA87
4.6. ELECTROSTATIC ELECTRON OSCILLATIONS ⊥ TO B

which represents the non-propagating fixed frequency oscillations. For the unmag-
netized case, i.e. B → 0, we get the usually plasma oscillations as ω 2 = ωp2 . Thus
they are only oscillations, i.e. ω 6= ω(k) if no thermal motions are considered and
hence vg = 0.

Upper-Hybrid Frequency

It is defined as the square sum of electron plasma frequency (which corresponds to

longitudinal oscillations) and electron cyclotron frequency, i.e.

ωh2 ≡ ωp2 + ωc2 (4.57)

(1) If there is no plasma, i.e. n → 0, which means ωp → 0 ⇒ ω = ωc . Then we have

only particle gyrations as caused by a constant magnetic field.
(2) If there is no magnetic field then we get only plasma oscillations, i.e. ω = ωp .
(3) The upper-hybrid frequency has been verified by various experimental observa-
tions. For that we look for the transmission of microwaves across a magnetic field,
when n is changed the transmission takes a dip at a density n0 at which ωext = ωh .
This indicates that the upper-hybrid oscillations are excited and - due to the reso-
nance with external applied frequency - energy is absorbed by plasma. Lets rewrite
above equation in the form

ωc2 = ω 2 − ωp2
ωc2 ωp2
=1− (4.58)
ω2 ω2
ωc2 n0 e 2
=1−
ω2 0 mω 2

which represent a linear relation between left hand side and number density such
that the curve has a negative slop, and agrees with the experimental findings of
Harps et al. as shown in Fig. 4.4.

Dispersion relation for electron waves propagating at an arbitrary angle

we can extend the above calculation for an arbitrary propagation, i.e. at angle θ
~ 0 . For that consider the Fig. 4.5, for which we write
relative to B

~k = kx x̂ + kz ẑ (4.59)

~ = Ex x̂ + Ez ẑ,
E (4.60)

where ~k and E
~ lie in the xz plan, having the components are Ex = Esinθ, Ez =
Ecosθ and likewise kx = ksinθ, kz = kcosθ. The corresponding Gauss’s law then
reads
88 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.4: The linear relation as shown by Eq.(4.58) that agrees with experimental
data. The plot is produced for the resonance condition that corresponds to maximum
absorption [From Chen].

Figure 4.5: The propagation of electrostatic electrons waves at an arbitrary angle.

 ~ IN A COLD PLASMA89
4.6. ELECTROSTATIC ELECTRON OSCILLATIONS ⊥ TO B

∇·E ~ = ρ
0
i (kx Ex + kz Ez ) = −en1 0
(4.61)
kE cos2 θ + sin2 θ = ien1 0

ien1
kE =
0

Equation (4.55) for this particular case transform to the form

−iωn1 + in0 (kx vx + kz vz )

n0 (4.62)
n1 = (kx vx + kz vz )
ω
Upon using Eq.(4.62) in Eq.(4.61) yields

ien0
kE = (kx vx + kz vz )
0 ω
(4.63)
ien0
= (ksinθvx + kcosθvz )
0 ω

The various velocity components in equation of motion can be written as

−ie iωc
vx = Ex − vy , (4.64)
mω ω

iωc
vy = vx (4.65)
ω
and

−ie −ie
vz = Ez = Esinθ (4.66)
mω mω
To uncouple x and y components lets use Eq.(4.65) in (4.64) to write

−ie iωc iωc

vx = Ex − vx
mω ω ω
−ie iωc2
= Ex 2 vx
mω ω
−ie

ωc2
−1 (4.67)
= Ex 1 − 2
mω ω
−1
ωc2

−ie
= Esinθ 1 − 2
mω ω

Substituting these values of vx and vz in Eq.(4.63) yields

90 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

ωc2
   
ien0 −ie −ie
kE = ksinθ Esinθ 1 − 2 + kcosθ Esinθ
0 ω mω ω mω
" −1 #
n0 e 2 ωc2

2
kE = 2
kEsin θ 1 − 2 + kEcos2 θ
0 mω ω
ωp2 ω2
 
2 2
1 = 2 sin θ 2 + cos θ
ω ω − ωc2
ω2 ω2 ω2
 
2 (4.68)
2
= 2 2
+ cos θ 1 − 2 2
ωp ω − ωc ω − ωc
ω2 2 2

2 2 2
ω − ωc = ω − ωc cos θ
ωp2
ω 2 ω 2 − ωc2 + ωp2 ωc2 cos2 θ − ω 2 ωp2 = 0

ω2 ω 2 − ωc2 − ωp2 + ωp2 ωc2 cos2 θ = 0

ω 2 ω 2 − ωh2 + ωp2 ωc2 cos2 θ = 0

Note that in the limit when θ → π/2 we get the upper-hybrid oscillations, i.e.

ω 2 = ωp2 + ωc2 = ωh2

4.7 ~
Electrostatic ion waves ⊥ to B
In the study of plasma waves we generally use the two main scenarios. First, the
electron oscillations are so rapid that the heavier ions are unaffected and are taken
to provide a uniform background. Secondly, the ionic oscillations are so slow that
electronic density has its equilibrium value all the times, which is generally take to
be Boltzmannian. Assume the following simplifications;

~ 0 = 0, ~v0 = E
1. Infinite plasma in equilibrium with ∇n ~ = φ0 = 0.

2. Cold ions are considered, i.e. Ti → 0.

3. Electrostatic waves are being studied, which implies that ~k × E

~ = 0 and
~ = −∇φ.
E ~

The equation of motion for plasma ions - in linearize form - reads

∂~vi1 ~ 1 + e~vi1 × B ~0
M = −e∇φ
∂t
−iωM~vi1 = −eikφ1 x̂ + e(viy B0 x̂ − vix B0 ŷ)

with corresponding x and y components given by

−iωM vix = −eikφ1 + e(viy B0 ) (4.69)

 ~
4.7. ELECTROSTATIC ION WAVES ⊥ TO B 91

and, respectively
iωM viy = evix B0 (4.70)
Using Eq.(4.69) in Eq.(4.70) yields

evix B0
−iωM vix = −iekφ1 + eB0
iωM
 2 2 (4.71)
e B0 vix M
ωM vix = ekφ1 + eB0
M2 ω
which upon using the definition of ion cyclotron frequency, i.e. Ωc = eB0 /M becomes
vix M
ωM vix − Ω2c = ekφ1
ω

Ω2c
 
ek
⇒ vix 1 − 2 = φ1
ω Mω
−1
Ω2c

ek
vix = 1− 2 φ1 (4.72)
Mω ω
The linearized version of ion continuity equation can be written as
∂ni1 ~ · (~vi1 ) = 0
+ n0 ∇
∂t

−iωni1 = −n0 ikvi1

 
k
ni1 = n0 vix (4.73)
ω
For electrons we can use Boltzmann density distribution and in the lowest order it
can be written as
 
eφ1
ne1 = n0
kB Te
using the plasma approximation to write above relation in the form
 
eφ1
ni1 = n0 (4.74)
kB Te
Now comparing Eqs.(4.73) and (4.74) yields
k eφ1
vix =
ω kB Te
Substituting for the perturbed ion density from Eq.(4.72) to write
−1
Ω2c

k ek eφ1
1− 2 φ1 =
M Mω ω kB Te
92 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

k 2 kB Te Ω2c
2
= 1 −
 ω M ω2
(4.75)
kB Te
k2 = ω 2 − Ω2c
M

Or, equivalently we can write

 
2 kB Te
ω = +k Ω2c 2
M (4.76)
2 2 2
= Ωc + k vs ,

where the ion acoustic speed in cold ion limit (Ti → 0) is given by
r
kB Te
vs =
M
Equation (4.76) is the dispersion relation for electrostatics ion cyclotron waves and
shows a propagating structure, i.e. vg 6= 0. The first term here represents the
ion cyclotron gyrations due to B~ and in the limit when B ~ → 0, Ωc → 0 and only
second term - representing the ion acoustic waves - contribute as observe for the
unmagnetized case. For Te → 0 the second term vanishes in the dispersion relation
and there is no propagation of wave but only gyro motion of ions with frequency Ωc .

The lower-hybrid frequency

In the derivation of last dispersion relation we have employed Boltzmann density
distribution for electrons, however instead if we use the corresponding fluid equations
we can proceed as follows. As for ions we had
−1
Ω2c

ek
vix = 1− 2 φ1 , (4.77)
Mω ω

likewise for plasma electrons one will get similar relation with the replacements;
M → m , e → −e and Ω2c → ωc2 to write
−1
ωc2

ek
vex =− 1− 2 φ1 (4.78)
mω ω

Similarly, continuity equations for the two species read

 
k
ni1 = n0 vix (4.79)
ω

and  
k
ne1 = n0 vex (4.80)
ω
which can be use in the plasma approximation ni1 = ne1 to write
4.8. ELECTROMAGNETIC WAVES IN FIELD FREE COLD PLASMA 93

vix = vex (4.81)

Putting the values of perturbed velocity components yields
−1 −1
Ω2c ωc2
 
ek ek
1− 2 =− 1− 2
Mω ω mω ω
 2
ω − ωc2 ω 2 − Ω2c
  
−m =M
ω2 ω2
ω 2 (M + m) = mωc2 + M Ω2c (4.82)
 2 2  2 2
eB eB
=m + M
m2 M2
e2 B 2 (M + m)
 
1 1
= e2 B 2 + =
m M mM
which upon rearranging can be written in the form
  
2 eB eB
ω = = ωc Ωc
m M
p
=⇒ ω = ωc Ωc ≡ ωl (4.83)
which is defined as the lower-hybrid frequency.

4.8 Electromagnetic waves in field free cold plasma

So for we have been discussing electrostatic waves for which following equations hold
true
~
~ ×E
∇ ~ 1 = ∂ B1 ,
∂t
~ ~
∇ × E1 = 0,
~k × E
~1 = 0

Lets introduce the next order complexity, namely the cases when B ~ 1 6= 0 having a
i(kz−wt)
sinusoidal behavior, i.e. B1 , E1 ∝ e and consider cold plasma model, namely
Te = Ti = 0. This corresponds to transverse electromagnetic waves in our system of
interest.

Brief review of electromagnetic waves in vacuum

The Maxwell’s equations in free space, i.e. J = 0 and ρ = 0 read

~
∇ ~ 1 = − ∂ B1
~ ×E (4.84)
∂t
~
∇ ~ 1 = 1 ∂ E1
~ ×B (4.85)
c2 ∂t
94 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Taking the curl of second equation and using the result of first equation to write

~ × (∇
∇ ~ 1 ) = 1 ∂ (∇
~ ×B ~ ×E~ 1)
c2 ∂t
*0
!
 1 ∂ −∂ ~1
B
~ 
∇(∇~ ~ − ∇2 B

· B) ~1 =
c2 ∂t ∂t
1 ∂ 2B~1
2~
∇ B1 = 2
c ∂t2
1
k 2 B1 = 2 (ω 2 B1 )
c
ω 2 = c2 k 2 =⇒ ω = ck

from which we find that vφ = vg = c = 3 × 108 m/s. Thus, electromagnetic waves in

vacuum travel with a constant velocity, i.e. speed of light in free space.

In plasma with B~0 = 0

In contrast to the waves in free space, for a plasma the Ampere’s law must also
contain J1 term as its not a vacuum anymore. Thus we can write
~
~ ×B
∇ ~ 1 = µ0 J~1 + 1 ∂ E1
c2 ∂t (4.86)
~ = µ ∂ J~1 + 1 ∂ E
2~
~ × Ḃ 1
∇ 1 o 2 2
∂t c ∂t
The Faraday’s law of electromagnetic induction reads
~
~ ×E
∇ ~ 1 = − ∂ B1
∂t
=⇒ ∇ ~ × (∇~ ×E~ 1) = ∇
~ × Ḃ~
1

~ ∇
∇( ~ · E)
~ − ∇2 E~ =∇ ~ × Ḃ~
1 1
2
−~k(~k · E ~ 1 = iωµo J~1 + ω E
~ 1) + k2E ~1
c2

For transverse electromagnetic waves ~k · E

~ 1 = 0, thus we find

~ 1 = iω J~1 c2 µo + ω 2 E
c2 k 2 E ~1
~ (4.87)
~ 1 (ω 2 − c2 k 2 ) = −iω J1
E
εo

If J~1 = 0, we get ω 2 = c2 k 2 which are the corresponding dispersion relation for free
space. However, for plasma we need J1 (E1 ), to evaluate the dispersion relation in
Eq.(4.87). For that we consider that ions are fixed vi1 = 0 on the time scale of fast
oscillations. Hence the current is carried - mainly by - electrons with current density
given by
J~1 v −n0 e~ve1 (4.88)
4.8. ELECTROMAGNETIC WAVES IN FIELD FREE COLD PLASMA 95

To calculate the ~ve1 , we consider momentum balance equation for electrons with
Te → 0, to write
 
∂~ve1 ~ ~ + ~v × B)~ − ∇p
~ e
m + (~ve1 .∇)~ve1 = −e(E
∂t
∂~ve1 ~1
m = −eE (Neglecting non-linear terms)
∂t
iωm~ve1 = eE ~1 (Sinusoidal perturbations)
eE~1
~ve1 =
iωm
Thus by using above result in Eq.(4.88) we find the current density in terms of
electric field as given by
!
e ~1
E
J~1 = −no e
iωm
no e2 ~
=i E1
mω
From Eq.(4.87) we find
no e2 ~
 
~ 2 2 2 ω
E1 (ω − c k ) = −i i E1
εo mω
no e2
ω 2 − c2 k 2 = = ωp2
εo m
ω 2 = ωp2 + c2 k 2 (4.89)
is the desired dispersion relation for electromagnetic waves propagating in a plasma
with no DC magnetic component.

Analysis of dispersion relation

(1) For a free space, i.e. if there is no plasma meaning thereby ωp → 0, one finds
ω = ck, which agrees with our previous findings.
2 ω2 2
ωp2
(2) From Eq.(4.89) we can write the phase speed of the wave as vφ = 2 = c + 2 >
k k
c2 ⇒ vφ > c. Hence in plasma it is possible for the phase speed to be greater than
speed of light in vacuum.
(3) From the dispersion relation the group velocity can be evaluated as

ω 2 = ωp2 + c2 k 2
2ωdω = c2 2kdk
dω c2
=
dk vφ
c2
=⇒ vg = r  ω 2 < c (4.90)
2 p
c +
k
96 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

In contrast to the phase velocity, the group velocity - with which information travels
- is always less than speed of light in free air.

(4) For the dispersion relation in Eq.(4.89) we note that ωmin → ωp and the asymp-
totically, i.e. k → ∞ when the first term → 0, one finds ω = ck representing a
straight line with slope given by c.

4.8.1 Cut off

If one sends a signal of frequency ω into a plasma - as discussed above - its wavelength
λ = 2π/k will be described by the dispersion relation of Eq(4.89). If the plasma
density (and thus ωp ) increases it would result in degradation of wave-number, i.e.
k 2 . It is because of the fact that ω = constant. Thus the corresponding wave length
becomes longer. And finally a density can be reached such that k 2 → 0 and wave
cannot propagate. This cutoff condition occurs at a critical density nc , which can
be calculated by setting n → nc ⇒ k 2 → 0 and ω = ωp , namely

nc e2
= ω2
εo m
εo mω 2
=⇒ nc =
e2

4.8.2 Index of refraction

The refractive index nR for a wave is defined as nR ≡ c/vφ , which means that for
ck c
vacuum it is unity and for our plasma system we can write nR = = < 1.
ω vφ
Which is responsible for the diverging nature of a plasma lens.

4.8.3 Skin depth

If we choose an oscillator which generates waves of frequency less than that of plasma
frequency, i.e. ω < ωpe then

c2 k 2 = ω 2 − ωp2e = −|ω 2 − ωp2e |,

q
ck = ±i |ω 2 − ωp2e |,

where k is a complex variable now. Thus for perturbed electric field of sinusoidal
form we can write
E1 ∝ eikx = e−x(Im(k)) = e−x/δ
with the skin depth δ is defined as
1 c
δ≡ =p 2
Imk ωpe − ω 2

and in the limit ω  ωpe we get

4.8. ELECTROMAGNETIC WAVES IN FIELD FREE COLD PLASMA 97

c
δ≈
wpe

It is the electromagnetic skin depth, and for typical parameters ne = 1019 /m3 then
δ turns out to be 2mm.

4.8.4 Applications
Measurement of plasma density
As propagation of waves in plasma is dependent upon plasma density (via ωp ), you
can measure the plasma density by varying the frequency of an externally produced
electromagnetic wave. For ω < ωpe waves do not get transmitted. In this regard, one
needs to use a variable frequency transmitter and observe the wave on other side.
If the wave cease to appear (at the moment it stops) this would mean that ⇒ ω =
ωpe . From the knowledge of plasma frequency we can compute the corresponding
electronic density.

Ionosphere
As stated in the first chapter that the ionosphere is a plasma and plays important
role in various communications. The electromagnetic waves can get reflected of
the ionosphere, i.e. these waves cannot propagate. Typically in this region we
√
have n0 v 1012 /m3 and accordingly ωpe ≈ 56 ne = 6 × 107 rad/s or equivalently
fpe = 10M Hz, which is a radio frequency. Condition for wave reflection from the
ionosphere reads f < fpe . The schematic of an electromagnetic wave, generated at
point A and getting reflected to another point B, is shown in Fig. 4.6.
Similar phenomenon occurs for transmission of radio waves. In Fig. 4.7 we present,
schematically, a radio station that generates corresponding signals which gets re-
flected from ionosphere to reach us (there are also surface waves but are not dis-
cussed here). Why radios sound better at night? It is due to the fact that the
ionosphere layer shrinks - due to less ionization rate as caused by the sun light - and
reflection becomes better. The long λ waves reflect from near layers of ionosphere
where as short λ are reflected from for away regions (E layers). Whereas for the
TV stations, the sound waves get reflected from the ionosphere but the light waves
from a height of 36000 km.

Satellites around earth

There are various satellites that are shot around the earth and on their return they
create a nice plasma around them and it can cause the so-called communication black
out when satellite enters earth atmosphere. To be more accurate the communications
blackouts that affect spacecraft re-entering the earth’s atmosphere, which are also
known as radio blackouts, ionization blackouts, or reentry blackouts, are caused by
an envelope of ionized air around the craft, created by the heat from the compression
of the atmosphere by the craft.
98 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.6: Traveling of an electromagnetic signal from A to B via ionosphere.

Figure 4.7: Radio and TV signals transmissions.

 ~0
4.9. ELECTROMAGNETIC WAVES ⊥ TO B 99

4.9 ~0
Electromagnetic waves ⊥ to B
~ 1 = 0, k ⊥ B
This scenario, in our terminology, refers to the case when B ~ 0 and
k⊥E ~ 1 . For which there are following two possibilities:

4.9.1 ~1 k B
Ordinary waves: E ~0
Lets assume the geometry in which B0 → ẑ, E~1 → ẑ and ~k → x̂. Upon using the
usual linearizations, namely
~ =B
B ~0 + B ~1
~ =0+E
E ~1
~v = 0 + ~v1
n = n0 + n1

we can write the Ampere’s Law in the following form

~
∇ ~ 1 = µ0 J~1 + 1 ∂ E1
~ ×B (4.91)
c2 ∂t
with the time derivative yielding
2 ~
~ × Ḃ~1 = µ0 J~˙1 + 1 ∂ E1
∇ (4.92)
c2 ∂t2
The Faraday’s Law of electromagnetic induction reads
~
~ × E~1 = − ∂ B1
∇
∂t
Now using Eq.(4.88) and taking curl of the above equation to write
2 ~
~ ∇
∇( ~ · E~1 ) − ∇2 E1 = µ0 J~˙1 + 1 ∂ E1
c2 ∂t2

c2 k 2 E~1 = iωc2 µ0 J~1 + ω 2 E~1

ω in0 eω
E~1 (ω 2 − c2 k 2 ) = −i J~1 = ~ve1 (4.93)
ε0 ε0
As the perturbed electric field vector is directed along z direction, i.e. E1 → ẑ on
LHS of above expression and hence we also need to write ve1 → ẑ on RHS. For that
we may write parallel equation of motion as

∂~ve1 ~
m = −eE
∂t
iωmvez = eEz
e
=⇒ vez = Ez
imw
100 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.8: Dispersion curve for O-wave depicting the forbidden and cutoff regions.

which is exactly the same as our previous calculation with B~0 and therefore we will
get the same results as that of field free case, namely

ω 2 = ωp2 + c2 k 2

Since the scenario E~1 k B~0 does not bring anything new to the previous case, the
wave is refer to as an ordinary wave (O-wave) and is not dependent on magnetic
field. Thus, for an O-wave we can write

ω 2 = ωp2 + c2 k 2
ωp2 c2 k 2
1= + 2
ω2 ω
c 2 ωp2
n2R = 2 = 1 − 2
vφ ω
1 1
Y = 2 = ≥0
nR ωp2
1− 2
ω

The function Y = 1/n2R is plotted against the frequency ω in Fig. 4.8, where the
regions Y < 1 is the forbidden region, which represent a frequency range 0 ≤ ω < ωp
for which the wave can not propagate in the system. At the plasma frequency there
is a singularity in the plot and depicts the cutoff for O-wave. Whereas the resonance
Y → 0 is not achieved for any finite ω.
 ~0
4.9. ELECTROMAGNETIC WAVES ⊥ TO B 101

4.9.2 Extra ordinary wave: E~1 ⊥ B

~0
In this case the electronic contribution term J~1 will be changed and we expect a
modified dispersion relation that shall include the effects of B ~ 1 . For that consider
the following configuration;
(1) E1 → ŷ, ~k → x̂ and as usual B~0 → ẑ,
(2) it turns out that such waves with E~1 ⊥ B~0 (without a proof) tend to be elliptically
polarized, means that when this wave travel in plasma it develops a component Ex
(along ~k) as well, i.e. E → Ex x̂ + Ey ŷ, hence we let E~1 = Ex x̂ + Ey ŷ (mode is
elliptically polarized w.r.t ⊥ B~0 ). The momentum balance equation for electron
reads
∂ v~e
m 1 = −e(Ex x̂ + Ey ŷ + vey B0 x̂ − vey B0 ŷ)
∂t
for short-hand notation lets omit the subscript e, 1 and consider the x component
to write
iωmvx = e(Ey + vy B0 )
ie
vx = − (Ex + vy B0 ) (4.94)
mω
Likewise, the y component yields
ie
vy = − (Ey + vx B0 ) (4.95)
mω
Using Eq.(4.95) in Eq.(4.94) to write
e h  e i
vx = −i Ex + B0 −i (Ey + vx B0 )
mω  mω
  
e ieB0 e −ieB0
= −i Ex − Ey − i (−vx B0 )
mω mω mω mω
e h ω i ω2
= −iEx − Ey + c2 vx
 mω h ω ω
ωc2

e ω i
vx 1 − 2 = −iEx − Ey
ω mω ω
−1
ωc2

e h ω i
vx = −iEx − Ey 1 − 2 (4.96)
mω ω ω

Similar analysis for y-component provides the following

e h  e  i
vy = −i Ey − B0 −i (Ex + vy B0 )
mω mω
e h ωc i ωc2
vy = −i Ey + i Ex + 2 vy B0
mω ω c
2 −1
 
ω e  ωc 
vy 1 − c2 = −Ey + i Ex
ω mω ω
−1
ωc2

e  ωc 
vy = −Ey + i Ex 1− 2 (4.97)
mω ω ω
102 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

and −1
ωc2

e h ω i
vx = −iEx − Ey 1 − 2 (4.98)
mω ω ω
Now consider the following expression

2 ~
~ ∇
∇( ~ 1 ) − ∇2 E1 = µ0 J~˙1 + 1 ∂ E1
~ ·E
c2 ∂t2
which, upon using the sinusoidal perturbations, yields

in0 eω
(ω 2 − c2 k 2 )Ex + c2 kEx~k = ~ve1
ε0

The x-component of above relation provides

in0 eω
(ω 2 − c2 k 2 )Ex + c2 k 2 Ex = ~ve1
ε0

Now using Eq.(4.98), one finds

−1
ωc2

2 n0 eω e  ωc 
ω Ex = −i iEx + Ey 1− 2 (4.99)
ε0 mω ω ω

and, respectively
−1
ωc2

2 2 2 n0 eω e  ωc 
(ω − c k )Ey = −i iEy − Ex 1− 2 (4.100)
ε0 mω ω ω
" −1 # −1
ω2 ωp2 ωc ωc2
 
Ex ω − 2
ωp2 1 − c2 + i 2 Ey 1 − 2 =0
ω ω ω
 2 
ω2 ωp ωc
   
Ex ω 2
1 − c2 − ωp2 + Ey i 2 =0 (4.101)
ω ω
Similarly, we find

ωp2 ωc ωc2
     
2 2 2 2
Ex i 2 + Ey (ω − c k ) 1 − 2 − ωp = 0 (4.102)
ω ω

Above equations are of the form

AEx + BEy = 0

CEx + DEy = 0
for which the non trivial solution is obtained by imposing the condition

AD − BC = 0
 ~0
4.9. ELECTROMAGNETIC WAVES ⊥ TO B 103

Thus, in our case we find

  2 
ωc2 ωc2 ωp ωc ωp2 ωc
      
2 2 2 2 2 2
ω 1 − 2 − ωp (ω − c k ) 1 − 2 − ωp = i 2 −i 2
ω ω ω ω
2
ω2 ωp2 ωc
    
(ω 2 − ωh2 ) ω 2 − ωh2 − c2 k 2 1 − c2 =
ω ω
2
c2 k 2 2 ωp2 ωc
  
2 2 2 2 2
(ω − ωh ) ω − ωh − 2 (ω − ωc ) =
ω ω

2
2 2 c2 k 2 2 2
ωp2 ωc
ω − ωh − 2 (ω − ωc ) = 2 2
ω ω (ω − ωh2 )

c2 k 2 (ω 2 − ωh2 ) − β/(ω 2 − ωh2 )

=
ω2 (ω 2 − ωc2 )
As the upper hybrid frequency is given by ωh2 = ωc2 + ωp2 , hence one can write

c2 k 2 (ω 2 − ωc2 − ωh2 ) − β/(ω 2 − ωh2 )

=
ω2 (ω 2 − ωc2 )
ωp2 − β/(ω 2 − ωh2 )
=1−
(ω 2 − ωc2 )
ωp2 (ω 2 − ωh2 ) − β
=1− 2
(ω − ωc2 )(ω 2 − ωh2 )
ωp2 (ω 2 − ωh2 ) − (ωp2 ωc2 )/ω 2
=
(ω 2 − ωc2 )(ω 2 − ωh2 )
 
ωp2 ω 2 (ω 2 − ωh2 ) + ωp2 ωc2
=1−
ω 2 (ω 2 − ωc2 )(ω 2 − ωh2 )
ωp2 ω 2 (ω 2 − ωp2 − ωc2 ) + ωp2 ωc2
 
=1− 2
ω (ω 2 − ωc2 )(ω 2 − ωh2 )
ωp2 ω 2 (ω 2 − ωc2 ) − ωp2 (ω 2 − ωc2 )
 
=1− 2
ω (ω 2 − ωc2 )(ω 2 − ωh2 )
c2 k 2 ωp2 ω 2 − ωp2
 
=1− 2
ω2 ω ω 2 − ωh2

c2 ωp2 ω 2 − ωp2
 
=1− 2 (4.103)
vφ2 ω ω 2 − ωh2
The above dispersion relation corresponds to the so-called extra ordinary waves with
2
ω

~1 ⊥ B
E ~ 0 = 0, ωc → 0, ωh → ωp and hence c2 = v 2 1 − p . And if
~ 0 . Check : if B
φ
ω2
there is no plasma (vacuum), i.e. ωp → 0 which provides the usual electromagnetic
waves in free space with vφ = c.

Whenever the square of refractive index is negative, i.e. n2R < 0, it implies that nR
is imaginary for which k is also imaginary. As a consequence of which the space
104 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

dependence (in perturbations) is not oscillatory anymore and the wave is said to be
evanescent, that is it does not propagate through the medium, but just damped as
follows

eikx = ei(ikx) ∼ e±|k|x

Analysis of X-wave

Starting with the dispersion relation

c2 c2 k 2 ωp2 ω 2 − ωp2
 
n2R ≡ 2 = 2 =1− 2 (4.104)
vφ ω ω ω 2 − ωh2

for which the cutoff is defined by the limit: nR → 0, or k → 0, whereas the

resonance is obtained by using nR → ∞. A wave is reflected at cutoff (because it
cannot propagate and reflection occurs) and absorbed at the resonance frequency.

Resonance: to find the resonance conditions for X-wave we need to use |k| → ∞
in Eq.(4.104) which implies

ω 2 − ωp2
→∞
ω 2 − ωh2

⇒ ω 2 = ωh2

ω 2 = ωp2 + ωc2

which states that the resonance occur when the frequency of the wave (square)
is equal to the sum of (squares) of plasma and cyclotron (due to magnetic field)
frequencies. Note that if there is no plasma then ωp = 0 then ω = ωc , i.e. we have
a cyclotron resonance. While in the other case if there is no magnetic field then we
find the so-called plasma resonance. Thus, these findings are consistent with our
previous calculations.

Cutoff: this represents a scenario in which there is no propagation of wave through

 ~0
4.10. ELECTROMAGNETIC WAVES k TO B 105

the plasma, namely when k → 0, for which Eq.(4.104) yields

ωp2 ω 2 − ωp2
 
0=1− 2
ω ω 2 − ωh2
ωp2 ω 2 − ωp2
 
1= 2
ω ω 2 − ωh2
 

ωp2 ω 2 − ωp2 ωp2 

 
1 
1= 2 = 2
 
2 2
ω (ω − ωp ) − ωc 2 ω  ωc2

1−

ω 2 − ωp2
−1
ωp2 ω2

= 1− 2 c 2
ω2 ω − ωp
ωc2 /ω 2
1 − ωp2 /ω 2 =
1 − ωp2 /ω 2
2
ωp2

2 2
ωc /ω = 1 − 2
ω
2 2
ω ωp
± c2 = 1 − 2
ω ω
Multiplying by ω 2 and arranging it to write

ω 2 ∓ ωc ω − ωp2 = 0

1 q 
ω=±ωc + ωc2 + 4ωp2 (4.105)
2
Here the right and left cutoff limits are characterized by choosing the ± signs,
respectively as
1 p

Right hand cutoff: ωR = ωc + ωc2 + 4ωp2 and, respectively
2
1 p

Left hand cutoff: ωL = −ωc + ωc2 + 4ωp2 .
2
Therefore for waves with frequency ωR ≤ ω ≤ ωL can not propagate in plasmas.
Figure 4.9 depicts the dispersion curve for X-wave, we note that at ω = ωp the
second term in dispersion relation Eq.(4.104) vanishes as a result we get unity, i.e.
nR → 1.

4.10 ~0
Electromagnetic Waves k to B
The present scenario can be described by the geometry in which B ~ 1 6= 0, ~k = kẑ
and B0 ẑ. Moreover, we let the electric field to have both transverse components,
~ 1 = Ex x̂ + Ey ŷ. Thus, the calculations are similar to X-wave calculations
i.e. E
presented in the previous section, namely we find

~ 1 + c2 kEx~k = i n0 eω ~ve1
(ω 2 − c2 k 2 )E (4.106)
ε0
106 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.9: Reciprocal of index of refraction (squared) as a function of frequency for

for X-wave [Chen].

However, in this case ~k = kẑ and consequently one finds (see Problem 4.8)

ωp2  ωc 
(ω 2 − c2 k 2 )Ex = Ex − iEy (4.107)
1 − ωc2 /ω 2 ω
ωp2  ωc 
(ω 2 − c2 k 2 )Ey = Ey + i Ex (4.108)
1 − ωc2 /ω 2 ω

Let us define for a mathematical convenience

ωp2
α≡
1 − ωc2 /ω 2

to write
ωc
(ω 2 − c2 k 2 − α)Ex + iα Ey = 0 (4.109)
ω

ωc
(ω 2 − c2 k 2 − α)Ey + iα Ex = 0 (4.110)
ω
 ~0
4.10. ELECTROMAGNETIC WAVES k TO B 107

For a non-trivial solution we must have

 αω 2
c
(ω 2 − c2 k 2 − α)2 =
ω
αωc
ω 2 − c2 k 2 − α = ±
ω
 ωc  ωp2  ωc 
ω 2 − c2 k 2 = α 1 ± = 1 ±
ω 1 − ωc2 /ω 2 ω
2
(1 ± ωc /ω)ωp
=
(1 + ωc /ω)(1 − ωc /ω)
2 2
ck ωp2 /ω 2
1− 2 =
ω 1 ∓ ωc2 /ω
2 c2 k 2 ωp2 /ω 2
nR = 2 = 1 −
ω 1 ∓ ωc /ω
~ 0 , namely
The above equation has two solution presenting two different waves along B
(1) The R-wave denoting a right hand circularly polarized mode with
ωp2 /ω 2
n2R =1−
1 − ωc /ω
(2) L-wave presenting a left hand circularly polarized component, namely
ωp2 /ω 2
n2L =1−
1 + ωc /ω

Analysis of R and L waves

1
R-wave: (1) For the resonance we have to take nr → ∞, or equivalently →0
nR
which occurs at ω = ωc as well as at ω = 0. Thus these two frequencies present a
situation for which the R-wave cannot propagate.
1
(2) For cutoff, namely when → ∞ we find
nR
ωp2 /ω 2
1= ,
1 − ωc /ω
ωp2 ωc
= 1 − ,
ω2 ω
After multiplying by ω 2 we find

ω 2 − ωc ω − ωp2 = 0,
1 q 
ω= ωc ± ωc2 + 4ωp2
2
Lets choose +ve sign for frequency to write
1 q 
ω= ωc + ωc2 + 4ωp2 ≡ ωR
2
108 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.10: The cutoff and resonance for L and R waves [Chen].

which presents the desired cutoff condition for R-wave.

L-waves: (1)We note that there is no Resonance frequency possible for the L-wave
as there is no such point where nL → ∞.
1
(2) The cutoff is obtained by the condition ⇒ nR → 0, → ∞. For that we find
nR

ωp2 /ω 2
1=
1 + ωc /ω

1 q 
⇒ω= −ωc + ωc2 + 4ωp2 ≡ ωL
2
. Notes: The dispersion curve for L and R waves is presented in Fig. 4.10, from
which we note the following

• L-wave has a forbidden band (stop band) for the range ω < ωL .

• No resonance corresponding to L-wave. Actually if we had included ions in

our calculation then we would have got a resonance at ω = Ωc .

• L-waves looks similar to O wave except that cutoff ocurs at ωL and not ωp .

• R-wave has a stop band for frequency range ωc < ω < ωR , moreover it has
two bands of propagation; (1) 0 < ω < ωc and, respectively (2) ω > ωR for
0 < ω < ωc .
4.11. MHD WAVES IN A UNIFORM PLASMA 109

Figure 4.11: The pictorial presentation of magnetic field shear in Alfvén waves.

4.11 MHD waves in a uniform plasma

We are now interested in low frequency wave modes in the presence of an ambient
magnetic field B0 , and hence the ions dynamics will also be included in our calcu-
lations. Here by low frequency we mean that the range of oscillations below the
ion cyclotron frequency, i.e. ω  Ωc . Thus, the present section is an extension of
last topic where ion motion is also included in respective fluid equations. In this
regard - of the many possible modes - we present the analysis of two most impor-
tant hydro-magnetic modes, namely the Alfvén (~k k B ~ 0 ) and magnetosonic waves
(~k ⊥ B
~ 0 ).

4.11.1 Alfvén Waves

In the magnetic hydro-dynamic description of plasma the magnetic field lines behave
as mass-loaded string under tension. The MHD fluid feels a magnetic tension B02 /µ0
along the field lines. Hence - by analogy with transverse vibration of elastic strings
- we also expect that if the conducting fluid is slightly disturbed from equilibrium
conditions, the magnetic field lines will perform transverse vibrations and the dis-
turbance would propagate along the field lines with the so-called Alfvén velocity as
given by
s
tension
vA =
density

In early 1940’s Alfvén first pointed out the existence of this type of low frequency
wave motion in a conducting magnetized fluid. There is no density or pressure
fluctuations associated with this wave and it is often called the torsional or shear
Alfvén. The fluid motion and magnetic perturbations are all perpendicular to the
field lines. Figure 4.11 gives a schematic representation of shear Alfvén waves in a
MHD fluid.

In such magnetic configuration there are also longitudinal oscillations (along the
magnetic field line) in a compressible conducting MHD fluid. For motion of particles
and propagation of waves along the magnetic field lines there will be no magnetic
field perturbation along the magnetic field lines since the particles are forced to
move in this direction. The corresponding wave is an ordinary longitudinal sound
waves as depicted in Fig. 4.12.
110 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.12: The longitudinal waves along the field line [Gosh].

Figure 4.13: The geometry used for description of Alfvén waves [Chen].

For such waves in a plane geometry consider Fig. 4.13 we have the following config-
uration
(1) ~k k B~0
(2) E~ 1 and J~1 perpendicular to B
~0
(3) ~v1 and B~ 0 perpendicular to both B
~ 0 and E
~0

The Faraday’s and Ampere’s Law thus read

~
∇ ~ 1 = µ0 J~1 + 1 ∂ E1 ,
~ ×B (4.111)
c2 ∂t
4.11. MHD WAVES IN A UNIFORM PLASMA 111

and, respectively

~
~ ×E
∇ ~ 1 = − ∂ B1 (4.112)
∂t !
~1
~ × (∇
⇒∇ ~ ×E ~ 1) = − ∂
~ 1 ) = − ∂ (∇ × B µ0 J~1 +
1 ∂E
∂t ∂t c2 ∂t
2~
~ ∇
∇( ~ ·E ~ 1 = −µ0 J~˙1 − 1 ∂ Eo
~ 1 ) − ∇2 E
c2 ∂t2
2
−~k(~k · E
~ 1) + k2E~1 = ω E ~ 1 + iω J~1 (4.113)
c2 ε o c2

Note that for the geometry we have chosen only x-component of above equation
exists.
As we are considering low frequency oscillations, thus both electrons and ion con-
tributes in J~1 and we find

ω2 ~ ω
k E1 = 2 E1 + i 2 (J~1e + J~1i )
2~
c εo c
0 (ω − c k )E1 = −iω(J1e + J~1i )
2 2 2 ~
0 (ω 2 − c2 k 2 )E1 = −iωn0 e(vix − vex ) (4.114)

The equation of motion for ion, in the absence of thermal motion T → 0, reads

∂~v1i ~0
M = +eE1 + e~v1i × B
∂t
with respective x and y components given by

−iωM vix = eE1 + eviy B0 (4.115)

and

−iωM viy = evix B0 (4.116)

Upon linearizing and using the sinusoidal perturbations, one can evaluate the per-
turbed horizontal velocity component as
 
eB0
−iωM vix = eE1 + eB0 vix
iωM
ieE1 e2 B 2
vix = + 2 02 vix
Mω M ω
Ω2c
 
ieE1
vix 1 − 2 =
ω Mω
−1
Ω2c

ie
vix = 1− 2 E1 (4.117)
Mω ω
112 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Likewise, for y-component of velocity one finds

 
eB0 eE1 + eviy B0
viy =
iM ω −iωM
2
e B0
= (eE1 + eB0 viy )
M 2ω2
e2 B0 e2 B02
= E1 + 2 2 viy
M 2ω2 M ω
Ω2
 
eΩc
viy 1 − 2c =
ω M ω2
−1
Ω2c

eΩc
viy = 1− 2 E1 (4.118)
M ω2 ω

The force balance equation for electrons can be simplified in the same manner as for
ions by replacing M → m, e → −e and Ωc → −ωc . Also using the fact that we are
looking for a slow frequency wave modes for which ωc2  ω 2 , thus the corresponding
velocity components for electrons can be found as

−1
ωc2

ie
vex =− 1+ 2 E1
mω ω

In the limit ωc2  ω 2 the term in parentheses vanishes and thus we find vex → 0.
Whereas for y component of perturbed electronic velocity, one finds

−1
ωc2

eωc
viy = 1+ 2 E1
mω 2 ω
e ωc ω 2 e em
=− 2 2
E1 = − E1 = − E1
m ω ωc mωc meB0
E1
vey = − (4.119)
B0

Upon using these velocities in Eq. (4.114) yields

"  #
2 −1

ie Ω
0 (ω 2 − c2 k 2 )E1 = −iωn0 e 1 − 2c E1
Mω ω
−1
no e2 Ω2c

2 2 2
ω −c k = 1− 2
0 M ω
2 −1
 
2 2 2 2 Ωc
ω − c k = Ωp 1 − 2
ω

For the considered hydro-magnetic waves with ω 2 < Ω2c , one can ignore the factor 1
4.11. MHD WAVES IN A UNIFORM PLASMA 113

in parenthesis and above relation takes the form

ω2 Ω2p
 
ω − c k = Ωp − 2 = −ω 2 2
2 2 2 2
Ωc Ωc
2 2
n0 e M
= −ω 2
0 M e2 B02
ω2 ρm
ω 2 − c2 k 2 = − 2
(M no ) = −ω 2
0 B0 0 B02
 
ρm
c2 k 2 = ω 2 1 +
0 B02
ω2 c2 c2
=   =  
k2 ρm ρ m µ0 2
1+ 1+ c
0 B02 B02

ω2 c2
=   (4.120)
k2 ρ m µ0 2
1+ c
B02
which is like the phase velocity of a wave in a dielectric medium, i.e.
ω c c
= 1/2
=√
k (R µR ) R

with µR = 1. For most of the laboratory plasmas we can ignore 1 in the denominator
of Eq.(4.120) to write
ω c B0
=r =√ ≡ vA
k ρ m µ0 ρ m µ0
c
B02

~ 0 with a velocity vA , which

• Thus, these hydro-magnetic waves move along B
depends on mass density ρm and the ambient magnetic field strength B0 .

• In contrast to most of our previous calculations, regardless of the frequency of

the wave the phase speed vφ is going to be same.

• Not only does the plasma moves but it moves in synchronization with the
magnetic field.

• As stated earlier the magnetic field lines are like plucked strings, and plasma
moves with it.

Discussion
In all of previous examples the magnetic field lines were unaffected by plasma. For
Alfvén waves plasma and the field lines move together in a ⊥ direction to direction
~ 1 ⊥ k̂. Further, note that for the electromagnetic waves
of the waves, i.e. B

~
~ 1 = − ∂ B1 ,
~ ×E
∇
∂t
114 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.14: The schematics of magnetic field variations and fluid drift in Alfvén
waves [Chen].

~k × E
~ 1 = −iω B
~1
Equating the coefficients of ŷ yields

kEx = ωBy
ω 
Ex = By
k
which depicts the electromagnetic nature of our waves. Further we have
~ =B
B ~0 + B
~ 1 = B0 ẑ + B1 ŷ

which highlights a sinusoidal ripple in original (unperturbed) magnetic field B ~ 0 . As

~ 1 ×B
By 6= 0, Ex 6= 0, B0 → ẑ, thus the resulting fluid drift can be along E ~ 0 ⇒ x̂× ẑ =
−ŷ-direction. In other words plasma is drifted along ŷ just like the component B ~ 1.
In this regard both ions and electrons are subjected to the drift in same direction
and the fluid moves up and down in ŷ direction (having a magnitude Ex /B) as
shown in Fig. 4.14. Frome where we can also write the vertical component of the
magnetic field ripples as

vripple = vφ sinα (4.121)

 
By
vripple = vφ (4.122)
B0
Now in a fluid description, we can write
ω
Ex = By
k
Ex ωBy
=
B0 kB0
⇒ vf luid = vripple
4.11. MHD WAVES IN A UNIFORM PLASMA 115

Figure 4.15: Perpendicular propagation of magnetosonic waves [Gosh].

which shows that the fluid (via drift) and field lines (via their vφ ) oscillate together,
as if the particles were stuck to the field lines. Hence, the lines of force act as if they
are mass loaded strings under tension and Alfvén wave propagate through them
when string is plucked.

4.11.2 Magnetosonic waves

In the perpendicular direction, a different type of longitudinal wave motion is pos-
sible, which involves compression and rarefaction of both the lines of force and the
conducting fluid as shown in Fig. 4.15. It is known as magnetosonic, magnetoacous-
tic or simply compressional wave. It is known as magnetosonic, magneto-acoustic
or simply compressional wave.
In the plasma approximation one ignores the pressure term in momentum balance
equation and it is valid as long as the phase velocity of the wave is much greater
than the thermal velocities, namely
ω
 vth , (4.123)
k
or, equivalently
k⊥ rL  1 (4.124)
However, a great variety of waves in plasma may not satisfy above condition, in
particular the ion acoustic waves and ion cyclotron waves for which we find ω < kvth .
And thus we need to keep the ∇P in force balance equation.
Before we start our detailed calculations let us review few important relations:
eB0 eB0
1: ωc = and Ωc =
m M
116 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

B0
2: vA = √ , ρm = M no = Mass density
µ0 ρ m
γe kB Te + γi kB Ti
3: Ion-sound speed, vs2 =
M
ω2
4: Dispersion relation for ion-acoustic speed ω 2 = k 2 vs2 ⇒ = vs2
k2
ω2
5: Dispersion relation for Alfvé-waves, = vA2
k2
For our analysis of low frequency electromagnetic waves across the ambient magnetic
field, consider the geometry in which B ~ 1 = E1 x̂, ~k = kŷ. Note that
~ 0 = B0 ẑ, E
E~1 × B~ 0 → ŷ, i.e. during the oscillations plasma (due to electric drift) will compress
and relax along direction of propagation. Equation of motion for ions reads
∂~vi1 ~ + ~vi1 × B
~ 0 ) − γi kB Ti ∇ni1
M no = eno (E
∂t
~ + eno (viy B0 x̂ − vix B0 ŷ) − i~kγi kB Ti ni1
−iωM no~vi1 = eno E

with respective x and y components given as

−iωM vix = eno Ex + eno viy B0

ie
vix = (Ex + viy B0 ) (4.125)
Mω
and, respectively

−iωM no viy = −eno vix B0 − iγi kB Ti ni1

−eno vix B0 ikγi kB Ti ni1
viy = −
−iωM no −iωM no
ie kγi kB Ti ni1
= (−vix B0 ) + (4.126)
Mω ωM no
Likewise, the ion continuity equation yields
∂ni1
+ ∇.(no~vi1 ) = 0
∂t
−iωni1 + ino~k.~vi1 = 0
ωni1 − no kviy = 0
 
ni1 k
⇒ = viy (4.127)
no ω
Using above relation in Eq.(4.126) to write

k2
 
ie γi kB Ti
viy = − (vix B0 ) + 2 viy ,
Mω ω M
Ωc
⇒ viy (1 − A) = −i vix , (4.128)
ω
4.11. MHD WAVES IN A UNIFORM PLASMA 117

where
k2
 
γi kB Ti
A≡ 2
ω M
Equations (4.125) and (4.128) are coupled and to solve them lets use the later in
former to write
 
ieEx ie Ωc
vix = + B0 −i vix (1 − A)−1
Mω Mω ω
2
ieEx Ω
= + 2 c vix
Mω ω (1 − A)
which simplifies to the form
Ω2c /ω 2
 
ieEx
vix 1 − = (4.129)
1−A Mω
Now the following Maxwell’s equations
~
~ 1 = µ0 J~1 + 1 ∂ E1
~ ×B
∇
c2 ∂t
~
~ 1 = − ∂ B1
~ ×E
∇
∂t
can be written, after linearization and using sinusoidal perturbations, in the form
~ 2~
~ ∇
∇( ~ ·E~ 1 ) − ∇2 E ~ 1 = −µ0 ∂ J1 − 1 ∂ E1
∂t c2 ∂t2
2
⇒ k2E ~ 1 = iω J~1 + ω E ~1
c2 o c2
~ 1 = − iω J~1
(ω 2 − k 2 c2 )E (4.130)
o
2 2 2
0 (ω − c k )Ex = −iωno e(vix − vex ) (4.131)
Note that we have already evaluated vix in Eq. (4.129) and for corresponding elec-
tronic component, i.e. vex we have to do the following replacement: e → −e,
Ωc → −ωc and Ti → Te , also lets approximate (due to m  M ) ω 2  ωc2 and
ω 2  k 2 vT2 e . All of these substitutions yield
−1
ωc2 /ω 2
  
ieEx ieEx 1 − Ae
vex = − 1− =
mω 1 − Ae mω ωc2 /ω 2
ie ω 2 k 2 γe kB Te
 
= 1 − Ex
mw ωc2 ω2 m
ik 2 e γe kB Te
 
=− 2 Ex
ωωc m2
ik 2 em2 γe kB Te
 
=− 2 2 Ex
ωe B0 m2
ik 2 γe kB Te
 
=⇒ vex = − Ex (4.132)
ωB02 e
118 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Upon substituting vix and vex in Eq.(4.131) one finds

ik 2 M γe kB Te
   
2 2 2 ie 1−A
o (ω − c k )Ex = −iωno e + Ex
Mω 1 − A − Ω2c /ω 2 ωB02 M e

which in the limit ω 2  Ω2c yields

iωno e ik 2 γe kB Te
 
2 ωno e ie
2 2 1 − Ai
ω −c k =− − M
o M ω −Ω2c /ω o ωB02 M e
ω2 ρµ0 γe kB Te
= −Ω2p 2 (1 − Ai ) + k 2 2 c2
Ωc B0 M

Ω2p
   
2 2 2γe kB Te 2 2 γi kB Ti
ω −c k 1+ + 2 ω −k =0
vA2 Ωc M
Now consider
Ω2p no e2 M 2 M n o c2 c2
= = µ0 =
Ω2c o M e2 B02 B02 vA2
Thus, we can solve for the dispersion relation for the magnetosonic wave as

c2 vs2
     
2 2 2 γe kB Te + γi kB Ti 2 2
ω 1+ 2 =c k 1+ =c k 1+ 2
vA vA2 M vA

−1
ω2 vs2 c2
 
2
=c 1+ 2 1+ 2
k2 vA vA
 2 2
vA2
 
2 vA + vs
=c
vA2 vA2 + c2
ω2 v 2 + vA2
 
=⇒ 2 = c2 2s (4.133)
k c + vA2

This wave can be considered as an acoustic wave, where the compression and rar-
efaction are produced not only by motion along electric field but also by the electric
~ For the unmagnetized case, the Alfvén velocity vanishes and the
drift across E.
magnetoacoustic wave turns into an ordinary ion acoustic mode. On the other hand
in the absence of any thermal motion, vs → 0 and the wave becomes a modified
Alfvén wave. For a very strongly magnetized plasma the wave travels with the
Alfvén speed, i.e. vφ → vA . The phase velocity of the magnetosonic wave is al-
ways larger than corresponding Alfvén velocity and for this it is often called the fast
hydro-magnetic wave.

In the above discussions we have neglected the effects of displacement current. This
approximation is valid only for highly conducting fluids at relatively low frequencies
well below the ion cyclotron frequency. We have assumed the fluid to be non-
viscous and perfectly conducting. If viscous effects are present and the fluid has a
finite conductivity the MHD waves are found to be damped.
4.12. PROBLEMS 119

4.12 Problems
P(4.1) How long does it take for a thermal electrons to cross one Debye length?

(Ans: √1 )
2ωp

P(4.2) To demonstrate that the electron plasma oscillations are fast, consider a
plasma with density 1018 /m3 and in a strong magnetic field of 0.3T . Calculate the
ratio of plasma frequency to cyclotron frequency (which is a very high value in such
strong magnetic field configuration).

(Ans: 1.07)

P(4.3) (a)For electron plasma oscillations we did not consider the ion motion and
assumed that they are stationary to show that ω = ωp . Show that when we in-
clude ion motion and include n1 term in Poisson’s equation the resulting frequency
is the superposition of electron and ion plasma frequencies, i.e. ω 2 = ωp + Ωp with
Ωp = n0 e2 /0 M .
(b) Calculate the ration between electron and ion plasma frequencies.

~ = 0 then show that

P(4.4) In the context of electron plasma oscillations, if ∇ · (E)
2 2
the dielectric constant  is given by  = 1 − ωp /ω . Here we note that if there is no
plasma then we get the usual unity for .

P(4.5) Plasma in the so-called E-layer of ionosphere has ne = 1011 /m3 and is located
at altitude of 100km. Then All electromagnetic waves with frequency lower than
the plasma frequency will get reflected from this layer.
(a) Find the upper most frequency of a wave that can pass through it.
(Ans: 2.8 MHz)
(b) Calculate the dielectric constant of plasma for the waves with frequencies of 100
MHz and 1000 Hz. Explain what does it mean to have a negative value of ?
(Ans: 0.9991, -8064037)
(c) Define the skin depth δs , and calculate
p 2 its 2value for the wave with frequency of
1000 Hz by using the relation δs = c/ ωp − ω .
(Ans:16.8 m)

P(4.6) In Eq.(4.68) we have derived the dispersion relation for propagation of elec-
tron waves at an arbitrary angle. For which there are two possible solutions.
(a) Calculate the two solutions of this equation for ω2 and check your calculations
in the limit θ → 0 and θ → π/2 to reproduce the previous results. Moreover, also
show that one of the root is not physical.
(b) Show that it represents an equation of ellipse - which we expect for charge par-
ticle motion under the combine effect of electric and magnetic fields - by obtaining
the relation
x2
(y − 1)2 + 2 = 1,
a
120 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

ω2 2
where x = cosθ, a = 2ωchωp and y = 2ω2 .
ωh
(c) If ωc > ωp then show that the lower root is always less than ωp for all positive
angles and the upper root lies between ωc and ωh .
(d) Whereas for ωp > ωc prove that the lower root lies below ωc whereas the upper
root is between ωp and ωh .

P(4.7) While deriving the relation for lower-hybrid frequency Eq.(??) we employed
the plasma approximation. Repeat the same calculation by using the Gauss’s law
and show
n0 e2
 
1 1 1 2
= + Ωp = (4.134)
ωl2 ωc Ωc Ω2c M
2
Ω2p = nM
0e
which shows that in low density plasma the second term can-not be ne-
glected. Hence the plasma approximation not valid at such high frequencies.

P(4.8) Derive the results as presented in Eq.(4.107) by using the following hints.Hints:

Start with −1

ωc2

e  ωc 
vx = −iEx − Ey 1− 2 (4.135)
mω ω ω
And use the Maxwell’s equation with ~k → kẑ to write

~ 1 + c2 k 2 Ex ẑ = i n0 eω vx
(ω 2 − c2 k 2 )E
ε0
As a result the x-component yields
n0 eω
⇒ (ω 2 − c2 k 2 )Ex = i vx
ε0
which upon using the value of vx provides the desired relation as
−1
ωc2

2 2 2 n0 eω e  ωc 
(ω − c k )Ex = i −iEx − Ey 1− 2
ε0 mω ω ω
−1
ω2
   ωc 
= ωp2 1 − c2 Ex − i Ey
ω ω
2
ωp  ωc 
(ω 2 − c2 k 2 )Ex = 2 2
Ex − i Ey
1 − ωc /ω ω
A similar mathematical analysis can be performed for y-component to derive the
second equation.
P(4.9) Consider the dispersion curve as shown in Fig. 4.16 and is given by the
dispersion relation
αωp2 βωp2 γωp2
   
2
nR = 1 − 2 1− 2 1− 2
ω ω ω

(a) What are the values of α, β and γ?

(b) Show the cutoff and resonance point on the plot.
4.12. PROBLEMS 121

Figure 4.16: Problem 4.9.

122 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION

Figure 4.17: Problem 4.11.

(c) If there is a wave whose frequency is in the range 1 < ω < 1.5, what do you
expect about the propagation of such wave mode?
P(4.10) Starting with the following dispersion relation of L-wave, namely

c2 k 2 ωp2 /ω 2
=1−
ω2 1 + ωc /ω

Show that the density at which we get the cutoff can e written in the following form
 
m0 2 eB
n0 = 2 (2πf ) + 2πf ,
e m
where f represents the frequency in Hz.
P(4.11) For the dispersion relation ω 2 = ωp2 + ωc2 and the experimental data as pre-
sented in Fig. 4.17. Calculate the value of applied frequency ω from the graph.

P(4.12) For the given dispersion relation

c2 k 2 ωp2 ω 2 − ωp2
 
= 1 −
ω2 ω2 ω 2 − ωh2
4.12. PROBLEMS 123

Show that the corresponding group velocity is given by

kc2
vg =   2 
ωc ωp
ω 1+ 2
ω 2 −ωh

P(4.13) Show that that there is no real k values for which the Langmuir and ion
acoustic branches cross each other.
P(4.14) Show that for Alfvén wave motion the magnetic energy density of the wave
motion is equal to the time-averaged kinetic energy density, i.e.

B12 1
= ρm v12
2µ0 2
Hints: Start with the following equation

∂B~1  
~0
= ∇ × ~v1 × B
∂t
to derive the following
~1 = B0 B0
B ~v1 = ~v1
(ω/k) vA
124 CHAPTER 4. PLASMA WAVES: A FLUID DESCRIPTION